diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,124439 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 76.60623229461757,
+  "eval_steps": 250,
+  "global_step": 6750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00026303675938712433,
+      "grad_norm": 174.60673820638053,
+      "learning_rate": 5e-06,
+      "loss": 6.7232,
+      "num_input_tokens_seen": 172192,
+      "step": 1
+    },
+    {
+      "epoch": 0.00026303675938712433,
+      "loss": 6.8635101318359375,
+      "loss_ce": 5.2287445068359375,
+      "loss_iou": 0.81640625,
+      "loss_num": 0.328125,
+      "loss_xval": 1.6328125,
+      "num_input_tokens_seen": 172192,
+      "step": 1
+    },
+    {
+      "epoch": 0.0005260735187742487,
+      "grad_norm": 113.61145210636417,
+      "learning_rate": 5e-06,
+      "loss": 5.2198,
+      "num_input_tokens_seen": 344520,
+      "step": 2
+    },
+    {
+      "epoch": 0.0005260735187742487,
+      "loss": 5.116145133972168,
+      "loss_ce": 3.611750364303589,
+      "loss_iou": 0.0,
+      "loss_num": 0.30078125,
+      "loss_xval": 1.5078125,
+      "num_input_tokens_seen": 344520,
+      "step": 2
+    },
+    {
+      "epoch": 0.000789110278161373,
+      "grad_norm": 181.6487958512906,
+      "learning_rate": 5e-06,
+      "loss": 3.6895,
+      "num_input_tokens_seen": 516868,
+      "step": 3
+    },
+    {
+      "epoch": 0.000789110278161373,
+      "loss": 3.7434756755828857,
+      "loss_ce": 2.4915225505828857,
+      "loss_iou": 0.0,
+      "loss_num": 0.25,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 516868,
+      "step": 3
+    },
+    {
+      "epoch": 0.0010521470375484973,
+      "grad_norm": 136.74708656913646,
+      "learning_rate": 5e-06,
+      "loss": 3.5593,
+      "num_input_tokens_seen": 687240,
+      "step": 4
+    },
+    {
+      "epoch": 0.0010521470375484973,
+      "loss": 3.5931553840637207,
+      "loss_ce": 1.3441319465637207,
+      "loss_iou": 0.80859375,
+      "loss_num": 0.44921875,
+      "loss_xval": 2.25,
+      "num_input_tokens_seen": 687240,
+      "step": 4
+    },
+    {
+      "epoch": 0.0013151837969356218,
+      "grad_norm": 79.92211667172693,
+      "learning_rate": 5e-06,
+      "loss": 2.8187,
+      "num_input_tokens_seen": 859388,
+      "step": 5
+    },
+    {
+      "epoch": 0.0013151837969356218,
+      "loss": 2.8585305213928223,
+      "loss_ce": 1.6251322031021118,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.24609375,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 859388,
+      "step": 5
+    },
+    {
+      "epoch": 0.001578220556322746,
+      "grad_norm": 51.97557929429645,
+      "learning_rate": 5e-06,
+      "loss": 2.1961,
+      "num_input_tokens_seen": 1029916,
+      "step": 6
+    },
+    {
+      "epoch": 0.001578220556322746,
+      "loss": 2.261239528656006,
+      "loss_ce": 1.1806731224060059,
+      "loss_iou": 0.0,
+      "loss_num": 0.2158203125,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 1029916,
+      "step": 6
+    },
+    {
+      "epoch": 0.0018412573157098704,
+      "grad_norm": 60.93429618456285,
+      "learning_rate": 5e-06,
+      "loss": 2.066,
+      "num_input_tokens_seen": 1202056,
+      "step": 7
+    },
+    {
+      "epoch": 0.0018412573157098704,
+      "loss": 1.9779821634292603,
+      "loss_ce": 1.0946813821792603,
+      "loss_iou": 0.0966796875,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 1202056,
+      "step": 7
+    },
+    {
+      "epoch": 0.0021042940750969946,
+      "grad_norm": 42.670837197481326,
+      "learning_rate": 5e-06,
+      "loss": 2.0543,
+      "num_input_tokens_seen": 1373904,
+      "step": 8
+    },
+    {
+      "epoch": 0.0021042940750969946,
+      "loss": 2.022350311279297,
+      "loss_ce": 0.9168814420700073,
+      "loss_iou": 0.0,
+      "loss_num": 0.220703125,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 1373904,
+      "step": 8
+    },
+    {
+      "epoch": 0.0023673308344841193,
+      "grad_norm": 48.53251052050697,
+      "learning_rate": 5e-06,
+      "loss": 2.4818,
+      "num_input_tokens_seen": 1546124,
+      "step": 9
+    },
+    {
+      "epoch": 0.0023673308344841193,
+      "loss": 2.4849071502685547,
+      "loss_ce": 0.8374460935592651,
+      "loss_iou": 0.01202392578125,
+      "loss_num": 0.330078125,
+      "loss_xval": 1.6484375,
+      "num_input_tokens_seen": 1546124,
+      "step": 9
+    },
+    {
+      "epoch": 0.0026303675938712436,
+      "grad_norm": 30.87810784426679,
+      "learning_rate": 5e-06,
+      "loss": 1.8016,
+      "num_input_tokens_seen": 1715836,
+      "step": 10
+    },
+    {
+      "epoch": 0.0026303675938712436,
+      "loss": 1.7729861736297607,
+      "loss_ce": 0.8613650798797607,
+      "loss_iou": 0.09326171875,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.91015625,
+      "num_input_tokens_seen": 1715836,
+      "step": 10
+    },
+    {
+      "epoch": 0.002893404353258368,
+      "grad_norm": 25.204935627992594,
+      "learning_rate": 5e-06,
+      "loss": 1.6242,
+      "num_input_tokens_seen": 1887800,
+      "step": 11
+    },
+    {
+      "epoch": 0.002893404353258368,
+      "loss": 1.8032605648040771,
+      "loss_ce": 0.7612683176994324,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.2080078125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 1887800,
+      "step": 11
+    },
+    {
+      "epoch": 0.003156441112645492,
+      "grad_norm": 27.719293877780373,
+      "learning_rate": 5e-06,
+      "loss": 1.5953,
+      "num_input_tokens_seen": 2058072,
+      "step": 12
+    },
+    {
+      "epoch": 0.003156441112645492,
+      "loss": 1.530227541923523,
+      "loss_ce": 0.660110354423523,
+      "loss_iou": 0.017822265625,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 2058072,
+      "step": 12
+    },
+    {
+      "epoch": 0.0034194778720326164,
+      "grad_norm": 77.78462332066468,
+      "learning_rate": 5e-06,
+      "loss": 1.9432,
+      "num_input_tokens_seen": 2230244,
+      "step": 13
+    },
+    {
+      "epoch": 0.0034194778720326164,
+      "loss": 2.0266342163085938,
+      "loss_ce": 0.7092512845993042,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.263671875,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 2230244,
+      "step": 13
+    },
+    {
+      "epoch": 0.0036825146314197407,
+      "grad_norm": 51.375242508900634,
+      "learning_rate": 5e-06,
+      "loss": 1.9389,
+      "num_input_tokens_seen": 2402352,
+      "step": 14
+    },
+    {
+      "epoch": 0.0036825146314197407,
+      "loss": 1.8962193727493286,
+      "loss_ce": 0.9186803102493286,
+      "loss_iou": 0.037841796875,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 2402352,
+      "step": 14
+    },
+    {
+      "epoch": 0.0039455513908068654,
+      "grad_norm": 29.266495242442826,
+      "learning_rate": 5e-06,
+      "loss": 1.6229,
+      "num_input_tokens_seen": 2574284,
+      "step": 15
+    },
+    {
+      "epoch": 0.0039455513908068654,
+      "loss": 1.7341835498809814,
+      "loss_ce": 0.7971718311309814,
+      "loss_iou": 0.052734375,
+      "loss_num": 0.1875,
+      "loss_xval": 0.9375,
+      "num_input_tokens_seen": 2574284,
+      "step": 15
+    },
+    {
+      "epoch": 0.004208588150193989,
+      "grad_norm": 22.43237476170095,
+      "learning_rate": 5e-06,
+      "loss": 1.4957,
+      "num_input_tokens_seen": 2746500,
+      "step": 16
+    },
+    {
+      "epoch": 0.004208588150193989,
+      "loss": 1.5246176719665527,
+      "loss_ce": 0.6452231407165527,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 2746500,
+      "step": 16
+    },
+    {
+      "epoch": 0.004471624909581114,
+      "grad_norm": 20.976636929732265,
+      "learning_rate": 5e-06,
+      "loss": 1.3572,
+      "num_input_tokens_seen": 2918736,
+      "step": 17
+    },
+    {
+      "epoch": 0.004471624909581114,
+      "loss": 1.3845367431640625,
+      "loss_ce": 0.6897125244140625,
+      "loss_iou": 0.0140380859375,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 2918736,
+      "step": 17
+    },
+    {
+      "epoch": 0.004734661668968239,
+      "grad_norm": 23.059953533674786,
+      "learning_rate": 5e-06,
+      "loss": 1.3283,
+      "num_input_tokens_seen": 3091200,
+      "step": 18
+    },
+    {
+      "epoch": 0.004734661668968239,
+      "loss": 1.3037350177764893,
+      "loss_ce": 0.5959713459014893,
+      "loss_iou": 0.052490234375,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 3091200,
+      "step": 18
+    },
+    {
+      "epoch": 0.0049976984283553625,
+      "grad_norm": 28.375886109478266,
+      "learning_rate": 5e-06,
+      "loss": 1.4794,
+      "num_input_tokens_seen": 3263136,
+      "step": 19
+    },
+    {
+      "epoch": 0.0049976984283553625,
+      "loss": 1.406355381011963,
+      "loss_ce": 0.6358475685119629,
+      "loss_iou": 0.03173828125,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 3263136,
+      "step": 19
+    },
+    {
+      "epoch": 0.005260735187742487,
+      "grad_norm": 25.383372657888618,
+      "learning_rate": 5e-06,
+      "loss": 1.4853,
+      "num_input_tokens_seen": 3435304,
+      "step": 20
+    },
+    {
+      "epoch": 0.005260735187742487,
+      "loss": 1.4931797981262207,
+      "loss_ce": 0.6591953635215759,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 3435304,
+      "step": 20
+    },
+    {
+      "epoch": 0.005523771947129611,
+      "grad_norm": 20.893788220271592,
+      "learning_rate": 5e-06,
+      "loss": 1.2467,
+      "num_input_tokens_seen": 3604172,
+      "step": 21
+    },
+    {
+      "epoch": 0.005523771947129611,
+      "loss": 1.2556164264678955,
+      "loss_ce": 0.5534679889678955,
+      "loss_iou": NaN,
+      "loss_num": 0.140625,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 3604172,
+      "step": 21
+    },
+    {
+      "epoch": 0.005786808706516736,
+      "grad_norm": 16.968829307639602,
+      "learning_rate": 5e-06,
+      "loss": 1.331,
+      "num_input_tokens_seen": 3776444,
+      "step": 22
+    },
+    {
+      "epoch": 0.005786808706516736,
+      "loss": 1.4187769889831543,
+      "loss_ce": 0.5535425543785095,
+      "loss_iou": 0.076171875,
+      "loss_num": 0.1728515625,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 3776444,
+      "step": 22
+    },
+    {
+      "epoch": 0.00604984546590386,
+      "grad_norm": 17.03009551099335,
+      "learning_rate": 5e-06,
+      "loss": 1.2354,
+      "num_input_tokens_seen": 3948760,
+      "step": 23
+    },
+    {
+      "epoch": 0.00604984546590386,
+      "loss": 1.2475543022155762,
+      "loss_ce": 0.5319781303405762,
+      "loss_iou": 0.07275390625,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 3948760,
+      "step": 23
+    },
+    {
+      "epoch": 0.006312882225290984,
+      "grad_norm": 17.413119153715428,
+      "learning_rate": 5e-06,
+      "loss": 1.2253,
+      "num_input_tokens_seen": 4120896,
+      "step": 24
+    },
+    {
+      "epoch": 0.006312882225290984,
+      "loss": 1.2492257356643677,
+      "loss_ce": 0.6032296419143677,
+      "loss_iou": 0.0400390625,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 4120896,
+      "step": 24
+    },
+    {
+      "epoch": 0.006575918984678109,
+      "grad_norm": 15.646317268104498,
+      "learning_rate": 5e-06,
+      "loss": 1.1599,
+      "num_input_tokens_seen": 4293068,
+      "step": 25
+    },
+    {
+      "epoch": 0.006575918984678109,
+      "loss": 1.221449613571167,
+      "loss_ce": 0.590346097946167,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 4293068,
+      "step": 25
+    },
+    {
+      "epoch": 0.006838955744065233,
+      "grad_norm": 16.80260761287207,
+      "learning_rate": 5e-06,
+      "loss": 1.1163,
+      "num_input_tokens_seen": 4465316,
+      "step": 26
+    },
+    {
+      "epoch": 0.006838955744065233,
+      "loss": 1.0530353784561157,
+      "loss_ce": 0.5227619409561157,
+      "loss_iou": 0.08447265625,
+      "loss_num": 0.10595703125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 4465316,
+      "step": 26
+    },
+    {
+      "epoch": 0.007101992503452358,
+      "grad_norm": 15.596544427195678,
+      "learning_rate": 5e-06,
+      "loss": 1.0849,
+      "num_input_tokens_seen": 4636968,
+      "step": 27
+    },
+    {
+      "epoch": 0.007101992503452358,
+      "loss": 1.1465504169464111,
+      "loss_ce": 0.5161793231964111,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 4636968,
+      "step": 27
+    },
+    {
+      "epoch": 0.0073650292628394814,
+      "grad_norm": 17.97152237441673,
+      "learning_rate": 5e-06,
+      "loss": 1.1255,
+      "num_input_tokens_seen": 4808916,
+      "step": 28
+    },
+    {
+      "epoch": 0.0073650292628394814,
+      "loss": 1.0834856033325195,
+      "loss_ce": 0.5703020095825195,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 4808916,
+      "step": 28
+    },
+    {
+      "epoch": 0.007628066022226606,
+      "grad_norm": 26.650637151013097,
+      "learning_rate": 5e-06,
+      "loss": 1.0819,
+      "num_input_tokens_seen": 4979072,
+      "step": 29
+    },
+    {
+      "epoch": 0.007628066022226606,
+      "loss": 1.083469271659851,
+      "loss_ce": 0.49557873606681824,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.11767578125,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 4979072,
+      "step": 29
+    },
+    {
+      "epoch": 0.007891102781613731,
+      "grad_norm": 47.98868961106415,
+      "learning_rate": 5e-06,
+      "loss": 1.5465,
+      "num_input_tokens_seen": 5150996,
+      "step": 30
+    },
+    {
+      "epoch": 0.007891102781613731,
+      "loss": 1.4472854137420654,
+      "loss_ce": 0.4790237545967102,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.96875,
+      "num_input_tokens_seen": 5150996,
+      "step": 30
+    },
+    {
+      "epoch": 0.008154139541000855,
+      "grad_norm": 36.03247925799523,
+      "learning_rate": 5e-06,
+      "loss": 1.479,
+      "num_input_tokens_seen": 5321588,
+      "step": 31
+    },
+    {
+      "epoch": 0.008154139541000855,
+      "loss": 1.3790192604064941,
+      "loss_ce": 0.5357575416564941,
+      "loss_iou": 0.0703125,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 5321588,
+      "step": 31
+    },
+    {
+      "epoch": 0.008417176300387979,
+      "grad_norm": 18.14805166699206,
+      "learning_rate": 5e-06,
+      "loss": 1.1322,
+      "num_input_tokens_seen": 5493708,
+      "step": 32
+    },
+    {
+      "epoch": 0.008417176300387979,
+      "loss": 1.069289207458496,
+      "loss_ce": 0.5192403793334961,
+      "loss_iou": 0.0654296875,
+      "loss_num": 0.10986328125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 5493708,
+      "step": 32
+    },
+    {
+      "epoch": 0.008680213059775104,
+      "grad_norm": 17.23835135853453,
+      "learning_rate": 5e-06,
+      "loss": 1.0735,
+      "num_input_tokens_seen": 5665524,
+      "step": 33
+    },
+    {
+      "epoch": 0.008680213059775104,
+      "loss": 1.063108205795288,
+      "loss_ce": 0.4869362711906433,
+      "loss_iou": 0.0556640625,
+      "loss_num": 0.115234375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 5665524,
+      "step": 33
+    },
+    {
+      "epoch": 0.008943249819162228,
+      "grad_norm": 15.02530944865542,
+      "learning_rate": 5e-06,
+      "loss": 1.0261,
+      "num_input_tokens_seen": 5837680,
+      "step": 34
+    },
+    {
+      "epoch": 0.008943249819162228,
+      "loss": 1.00763738155365,
+      "loss_ce": 0.4573444128036499,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.1103515625,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 5837680,
+      "step": 34
+    },
+    {
+      "epoch": 0.009206286578549352,
+      "grad_norm": 16.61205610634839,
+      "learning_rate": 5e-06,
+      "loss": 1.0035,
+      "num_input_tokens_seen": 6009848,
+      "step": 35
+    },
+    {
+      "epoch": 0.009206286578549352,
+      "loss": 0.9831359386444092,
+      "loss_ce": 0.46995237469673157,
+      "loss_iou": 0.123046875,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 6009848,
+      "step": 35
+    },
+    {
+      "epoch": 0.009469323337936477,
+      "grad_norm": 15.512257057890888,
+      "learning_rate": 5e-06,
+      "loss": 1.0223,
+      "num_input_tokens_seen": 6182164,
+      "step": 36
+    },
+    {
+      "epoch": 0.009469323337936477,
+      "loss": 1.0216686725616455,
+      "loss_ce": 0.5036022663116455,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.103515625,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 6182164,
+      "step": 36
+    },
+    {
+      "epoch": 0.009732360097323601,
+      "grad_norm": 19.305295502162483,
+      "learning_rate": 5e-06,
+      "loss": 1.0144,
+      "num_input_tokens_seen": 6352672,
+      "step": 37
+    },
+    {
+      "epoch": 0.009732360097323601,
+      "loss": 1.0505774021148682,
+      "loss_ce": 0.4739171862602234,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.115234375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 6352672,
+      "step": 37
+    },
+    {
+      "epoch": 0.009995396856710725,
+      "grad_norm": 20.69114777569909,
+      "learning_rate": 5e-06,
+      "loss": 1.0293,
+      "num_input_tokens_seen": 6525264,
+      "step": 38
+    },
+    {
+      "epoch": 0.009995396856710725,
+      "loss": 1.069566249847412,
+      "loss_ce": 0.4638533592224121,
+      "loss_iou": 0.02099609375,
+      "loss_num": 0.12109375,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 6525264,
+      "step": 38
+    },
+    {
+      "epoch": 0.010258433616097849,
+      "grad_norm": 32.793827182045035,
+      "learning_rate": 5e-06,
+      "loss": 1.0754,
+      "num_input_tokens_seen": 6697524,
+      "step": 39
+    },
+    {
+      "epoch": 0.010258433616097849,
+      "loss": 1.0720198154449463,
+      "loss_ce": 0.5224591493606567,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.10986328125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 6697524,
+      "step": 39
+    },
+    {
+      "epoch": 0.010521470375484974,
+      "grad_norm": 22.436128254245773,
+      "learning_rate": 5e-06,
+      "loss": 1.125,
+      "num_input_tokens_seen": 6869752,
+      "step": 40
+    },
+    {
+      "epoch": 0.010521470375484974,
+      "loss": 1.1199672222137451,
+      "loss_ce": 0.5101040601730347,
+      "loss_iou": NaN,
+      "loss_num": 0.1220703125,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 6869752,
+      "step": 40
+    },
+    {
+      "epoch": 0.010784507134872098,
+      "grad_norm": 17.66019658365854,
+      "learning_rate": 5e-06,
+      "loss": 0.9975,
+      "num_input_tokens_seen": 7041884,
+      "step": 41
+    },
+    {
+      "epoch": 0.010784507134872098,
+      "loss": 0.9673187732696533,
+      "loss_ce": 0.37991636991500854,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.11767578125,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 7041884,
+      "step": 41
+    },
+    {
+      "epoch": 0.011047543894259222,
+      "grad_norm": 17.977306189337394,
+      "learning_rate": 5e-06,
+      "loss": 0.9918,
+      "num_input_tokens_seen": 7214400,
+      "step": 42
+    },
+    {
+      "epoch": 0.011047543894259222,
+      "loss": 0.9905650615692139,
+      "loss_ce": 0.42928582429885864,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.1123046875,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 7214400,
+      "step": 42
+    },
+    {
+      "epoch": 0.011310580653646348,
+      "grad_norm": 16.758234031981292,
+      "learning_rate": 5e-06,
+      "loss": 1.0186,
+      "num_input_tokens_seen": 7386432,
+      "step": 43
+    },
+    {
+      "epoch": 0.011310580653646348,
+      "loss": 1.0381113290786743,
+      "loss_ce": 0.47536715865135193,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.1123046875,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 7386432,
+      "step": 43
+    },
+    {
+      "epoch": 0.011573617413033472,
+      "grad_norm": 19.20192611310373,
+      "learning_rate": 5e-06,
+      "loss": 0.9474,
+      "num_input_tokens_seen": 7558604,
+      "step": 44
+    },
+    {
+      "epoch": 0.011573617413033472,
+      "loss": 0.8546841144561768,
+      "loss_ce": 0.45356106758117676,
+      "loss_iou": 0.0634765625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 7558604,
+      "step": 44
+    },
+    {
+      "epoch": 0.011836654172420595,
+      "grad_norm": 24.07755280606732,
+      "learning_rate": 5e-06,
+      "loss": 1.0256,
+      "num_input_tokens_seen": 7730896,
+      "step": 45
+    },
+    {
+      "epoch": 0.011836654172420595,
+      "loss": 0.960330605506897,
+      "loss_ce": 0.48523297905921936,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.09521484375,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 7730896,
+      "step": 45
+    },
+    {
+      "epoch": 0.01209969093180772,
+      "grad_norm": 21.918622706923347,
+      "learning_rate": 5e-06,
+      "loss": 0.9858,
+      "num_input_tokens_seen": 7903036,
+      "step": 46
+    },
+    {
+      "epoch": 0.01209969093180772,
+      "loss": 1.0328285694122314,
+      "loss_ce": 0.41710585355758667,
+      "loss_iou": 0.119140625,
+      "loss_num": 0.123046875,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 7903036,
+      "step": 46
+    },
+    {
+      "epoch": 0.012362727691194845,
+      "grad_norm": 17.78591800932899,
+      "learning_rate": 5e-06,
+      "loss": 0.9217,
+      "num_input_tokens_seen": 8075568,
+      "step": 47
+    },
+    {
+      "epoch": 0.012362727691194845,
+      "loss": 0.9050750136375427,
+      "loss_ce": 0.4551238417625427,
+      "loss_iou": 0.21875,
+      "loss_num": 0.08984375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 8075568,
+      "step": 47
+    },
+    {
+      "epoch": 0.012625764450581969,
+      "grad_norm": 18.144694112865135,
+      "learning_rate": 5e-06,
+      "loss": 0.8763,
+      "num_input_tokens_seen": 8247888,
+      "step": 48
+    },
+    {
+      "epoch": 0.012625764450581969,
+      "loss": 0.8872632384300232,
+      "loss_ce": 0.3972730040550232,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.09765625,
+      "loss_xval": 0.490234375,
+      "num_input_tokens_seen": 8247888,
+      "step": 48
+    },
+    {
+      "epoch": 0.012888801209969093,
+      "grad_norm": 18.397929867872243,
+      "learning_rate": 5e-06,
+      "loss": 0.8479,
+      "num_input_tokens_seen": 8419884,
+      "step": 49
+    },
+    {
+      "epoch": 0.012888801209969093,
+      "loss": 0.8515866994857788,
+      "loss_ce": 0.3969968557357788,
+      "loss_iou": 0.109375,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 8419884,
+      "step": 49
+    },
+    {
+      "epoch": 0.013151837969356218,
+      "grad_norm": 16.926260406149144,
+      "learning_rate": 5e-06,
+      "loss": 0.8125,
+      "num_input_tokens_seen": 8592264,
+      "step": 50
+    },
+    {
+      "epoch": 0.013151837969356218,
+      "loss": 0.8142160177230835,
+      "loss_ce": 0.3877023756504059,
+      "loss_iou": 0.1123046875,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 8592264,
+      "step": 50
+    },
+    {
+      "epoch": 0.013414874728743342,
+      "grad_norm": 24.314274074580883,
+      "learning_rate": 5e-06,
+      "loss": 0.8943,
+      "num_input_tokens_seen": 8764528,
+      "step": 51
+    },
+    {
+      "epoch": 0.013414874728743342,
+      "loss": 0.8517386317253113,
+      "loss_ce": 0.43084025382995605,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 8764528,
+      "step": 51
+    },
+    {
+      "epoch": 0.013677911488130466,
+      "grad_norm": 26.15955176323275,
+      "learning_rate": 5e-06,
+      "loss": 0.9129,
+      "num_input_tokens_seen": 8936892,
+      "step": 52
+    },
+    {
+      "epoch": 0.013677911488130466,
+      "loss": 0.9079768657684326,
+      "loss_ce": 0.3640315532684326,
+      "loss_iou": 0.0654296875,
+      "loss_num": 0.10888671875,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 8936892,
+      "step": 52
+    },
+    {
+      "epoch": 0.013940948247517591,
+      "grad_norm": 29.358592778676385,
+      "learning_rate": 5e-06,
+      "loss": 0.9409,
+      "num_input_tokens_seen": 9108972,
+      "step": 53
+    },
+    {
+      "epoch": 0.013940948247517591,
+      "loss": 0.8754456043243408,
+      "loss_ce": 0.4201233685016632,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 9108972,
+      "step": 53
+    },
+    {
+      "epoch": 0.014203985006904715,
+      "grad_norm": 20.238375239505068,
+      "learning_rate": 5e-06,
+      "loss": 0.8928,
+      "num_input_tokens_seen": 9281248,
+      "step": 54
+    },
+    {
+      "epoch": 0.014203985006904715,
+      "loss": 0.8624146580696106,
+      "loss_ce": 0.4217408299446106,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.087890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 9281248,
+      "step": 54
+    },
+    {
+      "epoch": 0.014467021766291839,
+      "grad_norm": 21.380412065088443,
+      "learning_rate": 5e-06,
+      "loss": 0.8996,
+      "num_input_tokens_seen": 9453248,
+      "step": 55
+    },
+    {
+      "epoch": 0.014467021766291839,
+      "loss": 0.8669720888137817,
+      "loss_ce": 0.3911420404911041,
+      "loss_iou": 0.095703125,
+      "loss_num": 0.09521484375,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 9453248,
+      "step": 55
+    },
+    {
+      "epoch": 0.014730058525678963,
+      "grad_norm": 21.398016209346718,
+      "learning_rate": 5e-06,
+      "loss": 0.8695,
+      "num_input_tokens_seen": 9625308,
+      "step": 56
+    },
+    {
+      "epoch": 0.014730058525678963,
+      "loss": 0.7916536331176758,
+      "loss_ce": 0.3697786033153534,
+      "loss_iou": 0.265625,
+      "loss_num": 0.08447265625,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 9625308,
+      "step": 56
+    },
+    {
+      "epoch": 0.014993095285066088,
+      "grad_norm": 22.596856129715338,
+      "learning_rate": 5e-06,
+      "loss": 0.8385,
+      "num_input_tokens_seen": 9797472,
+      "step": 57
+    },
+    {
+      "epoch": 0.014993095285066088,
+      "loss": 0.8159988522529602,
+      "loss_ce": 0.3955886960029602,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 9797472,
+      "step": 57
+    },
+    {
+      "epoch": 0.015256132044453212,
+      "grad_norm": 19.06235083584469,
+      "learning_rate": 5e-06,
+      "loss": 0.791,
+      "num_input_tokens_seen": 9969708,
+      "step": 58
+    },
+    {
+      "epoch": 0.015256132044453212,
+      "loss": 0.8669772148132324,
+      "loss_ce": 0.3650240898132324,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.1005859375,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 9969708,
+      "step": 58
+    },
+    {
+      "epoch": 0.015519168803840336,
+      "grad_norm": 19.2496987382769,
+      "learning_rate": 5e-06,
+      "loss": 0.8251,
+      "num_input_tokens_seen": 10141772,
+      "step": 59
+    },
+    {
+      "epoch": 0.015519168803840336,
+      "loss": 0.8268835544586182,
+      "loss_ce": 0.39744019508361816,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 10141772,
+      "step": 59
+    },
+    {
+      "epoch": 0.015782205563227462,
+      "grad_norm": 18.91781603668572,
+      "learning_rate": 5e-06,
+      "loss": 0.8407,
+      "num_input_tokens_seen": 10312188,
+      "step": 60
+    },
+    {
+      "epoch": 0.015782205563227462,
+      "loss": 0.7802078723907471,
+      "loss_ce": 0.35906529426574707,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 10312188,
+      "step": 60
+    },
+    {
+      "epoch": 0.016045242322614586,
+      "grad_norm": 17.697618089112353,
+      "learning_rate": 5e-06,
+      "loss": 0.7755,
+      "num_input_tokens_seen": 10484428,
+      "step": 61
+    },
+    {
+      "epoch": 0.016045242322614586,
+      "loss": 0.8741220235824585,
+      "loss_ce": 0.3609383702278137,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 10484428,
+      "step": 61
+    },
+    {
+      "epoch": 0.01630827908200171,
+      "grad_norm": 19.734011329204773,
+      "learning_rate": 5e-06,
+      "loss": 0.769,
+      "num_input_tokens_seen": 10656840,
+      "step": 62
+    },
+    {
+      "epoch": 0.01630827908200171,
+      "loss": 0.7350368499755859,
+      "loss_ce": 0.36150169372558594,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 10656840,
+      "step": 62
+    },
+    {
+      "epoch": 0.016571315841388833,
+      "grad_norm": 24.730627091015997,
+      "learning_rate": 5e-06,
+      "loss": 0.7931,
+      "num_input_tokens_seen": 10828884,
+      "step": 63
+    },
+    {
+      "epoch": 0.016571315841388833,
+      "loss": 0.7593971490859985,
+      "loss_ce": 0.36462172865867615,
+      "loss_iou": NaN,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 10828884,
+      "step": 63
+    },
+    {
+      "epoch": 0.016834352600775957,
+      "grad_norm": 18.667625860875532,
+      "learning_rate": 5e-06,
+      "loss": 0.8089,
+      "num_input_tokens_seen": 11001164,
+      "step": 64
+    },
+    {
+      "epoch": 0.016834352600775957,
+      "loss": 0.8844671845436096,
+      "loss_ce": 0.3607855439186096,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.10498046875,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 11001164,
+      "step": 64
+    },
+    {
+      "epoch": 0.017097389360163084,
+      "grad_norm": 23.265809096889907,
+      "learning_rate": 5e-06,
+      "loss": 0.7793,
+      "num_input_tokens_seen": 11173188,
+      "step": 65
+    },
+    {
+      "epoch": 0.017097389360163084,
+      "loss": 0.7827451229095459,
+      "loss_ce": 0.3672178089618683,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 11173188,
+      "step": 65
+    },
+    {
+      "epoch": 0.017360426119550208,
+      "grad_norm": 24.005431476496838,
+      "learning_rate": 5e-06,
+      "loss": 0.8061,
+      "num_input_tokens_seen": 11345216,
+      "step": 66
+    },
+    {
+      "epoch": 0.017360426119550208,
+      "loss": 0.7301403284072876,
+      "loss_ce": 0.3616100549697876,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 11345216,
+      "step": 66
+    },
+    {
+      "epoch": 0.017623462878937332,
+      "grad_norm": 30.491458847817103,
+      "learning_rate": 5e-06,
+      "loss": 0.8474,
+      "num_input_tokens_seen": 11517388,
+      "step": 67
+    },
+    {
+      "epoch": 0.017623462878937332,
+      "loss": 0.8404921889305115,
+      "loss_ce": 0.3583144545555115,
+      "loss_iou": 0.07861328125,
+      "loss_num": 0.0966796875,
+      "loss_xval": 0.482421875,
+      "num_input_tokens_seen": 11517388,
+      "step": 67
+    },
+    {
+      "epoch": 0.017886499638324456,
+      "grad_norm": 17.38452589704735,
+      "learning_rate": 5e-06,
+      "loss": 0.8004,
+      "num_input_tokens_seen": 11689284,
+      "step": 68
+    },
+    {
+      "epoch": 0.017886499638324456,
+      "loss": 0.8011830449104309,
+      "loss_ce": 0.3470814824104309,
+      "loss_iou": 0.095703125,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 11689284,
+      "step": 68
+    },
+    {
+      "epoch": 0.01814953639771158,
+      "grad_norm": 22.0592753440465,
+      "learning_rate": 5e-06,
+      "loss": 0.7808,
+      "num_input_tokens_seen": 11861484,
+      "step": 69
+    },
+    {
+      "epoch": 0.01814953639771158,
+      "loss": 0.8063881397247314,
+      "loss_ce": 0.38475728034973145,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.08447265625,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 11861484,
+      "step": 69
+    },
+    {
+      "epoch": 0.018412573157098704,
+      "grad_norm": 18.31950796902093,
+      "learning_rate": 5e-06,
+      "loss": 0.766,
+      "num_input_tokens_seen": 12033752,
+      "step": 70
+    },
+    {
+      "epoch": 0.018412573157098704,
+      "loss": 0.7670217156410217,
+      "loss_ce": 0.31560570001602173,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.09033203125,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 12033752,
+      "step": 70
+    },
+    {
+      "epoch": 0.018675609916485827,
+      "grad_norm": 18.14951438740657,
+      "learning_rate": 5e-06,
+      "loss": 0.6971,
+      "num_input_tokens_seen": 12205604,
+      "step": 71
+    },
+    {
+      "epoch": 0.018675609916485827,
+      "loss": 0.6512900590896606,
+      "loss_ce": 0.31205666065216064,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 12205604,
+      "step": 71
+    },
+    {
+      "epoch": 0.018938646675872955,
+      "grad_norm": 18.910816353150064,
+      "learning_rate": 5e-06,
+      "loss": 0.7748,
+      "num_input_tokens_seen": 12377968,
+      "step": 72
+    },
+    {
+      "epoch": 0.018938646675872955,
+      "loss": 0.8064651489257812,
+      "loss_ce": 0.3355178236961365,
+      "loss_iou": 0.171875,
+      "loss_num": 0.09423828125,
+      "loss_xval": 0.470703125,
+      "num_input_tokens_seen": 12377968,
+      "step": 72
+    },
+    {
+      "epoch": 0.01920168343526008,
+      "grad_norm": 18.381185745083314,
+      "learning_rate": 5e-06,
+      "loss": 0.7552,
+      "num_input_tokens_seen": 12549896,
+      "step": 73
+    },
+    {
+      "epoch": 0.01920168343526008,
+      "loss": 0.817658007144928,
+      "loss_ce": 0.28372251987457275,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.10693359375,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 12549896,
+      "step": 73
+    },
+    {
+      "epoch": 0.019464720194647202,
+      "grad_norm": 16.482511811798446,
+      "learning_rate": 5e-06,
+      "loss": 0.6763,
+      "num_input_tokens_seen": 12721856,
+      "step": 74
+    },
+    {
+      "epoch": 0.019464720194647202,
+      "loss": 0.5998687744140625,
+      "loss_ce": 0.2707671821117401,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 12721856,
+      "step": 74
+    },
+    {
+      "epoch": 0.019727756954034326,
+      "grad_norm": 15.444850813034535,
+      "learning_rate": 5e-06,
+      "loss": 0.6806,
+      "num_input_tokens_seen": 12894040,
+      "step": 75
+    },
+    {
+      "epoch": 0.019727756954034326,
+      "loss": 0.7105993032455444,
+      "loss_ce": 0.27493035793304443,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.08740234375,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 12894040,
+      "step": 75
+    },
+    {
+      "epoch": 0.01999079371342145,
+      "grad_norm": 20.590812596799903,
+      "learning_rate": 5e-06,
+      "loss": 0.7622,
+      "num_input_tokens_seen": 13064296,
+      "step": 76
+    },
+    {
+      "epoch": 0.01999079371342145,
+      "loss": 0.6805918216705322,
+      "loss_ce": 0.3353770077228546,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 13064296,
+      "step": 76
+    },
+    {
+      "epoch": 0.020253830472808574,
+      "grad_norm": 20.47406440355872,
+      "learning_rate": 5e-06,
+      "loss": 0.74,
+      "num_input_tokens_seen": 13233888,
+      "step": 77
+    },
+    {
+      "epoch": 0.020253830472808574,
+      "loss": 0.7564910650253296,
+      "loss_ce": 0.3099578022956848,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.08935546875,
+      "loss_xval": 0.447265625,
+      "num_input_tokens_seen": 13233888,
+      "step": 77
+    },
+    {
+      "epoch": 0.020516867232195698,
+      "grad_norm": 19.122832804118445,
+      "learning_rate": 5e-06,
+      "loss": 0.741,
+      "num_input_tokens_seen": 13405900,
+      "step": 78
+    },
+    {
+      "epoch": 0.020516867232195698,
+      "loss": 0.6968704462051392,
+      "loss_ce": 0.26059114933013916,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.08740234375,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 13405900,
+      "step": 78
+    },
+    {
+      "epoch": 0.020779903991582825,
+      "grad_norm": 16.68220779277857,
+      "learning_rate": 5e-06,
+      "loss": 0.6822,
+      "num_input_tokens_seen": 13578336,
+      "step": 79
+    },
+    {
+      "epoch": 0.020779903991582825,
+      "loss": 0.6721813082695007,
+      "loss_ce": 0.26910513639450073,
+      "loss_iou": 0.125,
+      "loss_num": 0.08056640625,
+      "loss_xval": 0.40234375,
+      "num_input_tokens_seen": 13578336,
+      "step": 79
+    },
+    {
+      "epoch": 0.02104294075096995,
+      "grad_norm": 15.592038335799979,
+      "learning_rate": 5e-06,
+      "loss": 0.6458,
+      "num_input_tokens_seen": 13750256,
+      "step": 80
+    },
+    {
+      "epoch": 0.02104294075096995,
+      "loss": 0.604525625705719,
+      "loss_ce": 0.253695547580719,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 13750256,
+      "step": 80
+    },
+    {
+      "epoch": 0.021305977510357073,
+      "grad_norm": 19.123452214665015,
+      "learning_rate": 5e-06,
+      "loss": 0.7034,
+      "num_input_tokens_seen": 13922220,
+      "step": 81
+    },
+    {
+      "epoch": 0.021305977510357073,
+      "loss": 0.7461546659469604,
+      "loss_ce": 0.28350815176963806,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0927734375,
+      "loss_xval": 0.462890625,
+      "num_input_tokens_seen": 13922220,
+      "step": 81
+    },
+    {
+      "epoch": 0.021569014269744197,
+      "grad_norm": 14.596994780035258,
+      "learning_rate": 5e-06,
+      "loss": 0.6357,
+      "num_input_tokens_seen": 14094312,
+      "step": 82
+    },
+    {
+      "epoch": 0.021569014269744197,
+      "loss": 0.602331280708313,
+      "loss_ce": 0.2593136727809906,
+      "loss_iou": 0.328125,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 14094312,
+      "step": 82
+    },
+    {
+      "epoch": 0.02183205102913132,
+      "grad_norm": 16.4283520636599,
+      "learning_rate": 5e-06,
+      "loss": 0.5782,
+      "num_input_tokens_seen": 14264984,
+      "step": 83
+    },
+    {
+      "epoch": 0.02183205102913132,
+      "loss": 0.5699411630630493,
+      "loss_ce": 0.2635447084903717,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 14264984,
+      "step": 83
+    },
+    {
+      "epoch": 0.022095087788518444,
+      "grad_norm": 22.311971257546926,
+      "learning_rate": 5e-06,
+      "loss": 0.6398,
+      "num_input_tokens_seen": 14437168,
+      "step": 84
+    },
+    {
+      "epoch": 0.022095087788518444,
+      "loss": 0.6384698748588562,
+      "loss_ce": 0.240032359957695,
+      "loss_iou": 0.21875,
+      "loss_num": 0.07958984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 14437168,
+      "step": 84
+    },
+    {
+      "epoch": 0.022358124547905568,
+      "grad_norm": 26.609904157941,
+      "learning_rate": 5e-06,
+      "loss": 0.8114,
+      "num_input_tokens_seen": 14609504,
+      "step": 85
+    },
+    {
+      "epoch": 0.022358124547905568,
+      "loss": 0.7600446939468384,
+      "loss_ce": 0.23172441124916077,
+      "loss_iou": 0.076171875,
+      "loss_num": 0.10546875,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 14609504,
+      "step": 85
+    },
+    {
+      "epoch": 0.022621161307292696,
+      "grad_norm": 28.311475657648764,
+      "learning_rate": 5e-06,
+      "loss": 0.8067,
+      "num_input_tokens_seen": 14781812,
+      "step": 86
+    },
+    {
+      "epoch": 0.022621161307292696,
+      "loss": 0.839857816696167,
+      "loss_ce": 0.2500140368938446,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.1181640625,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 14781812,
+      "step": 86
+    },
+    {
+      "epoch": 0.02288419806667982,
+      "grad_norm": 24.92836303632298,
+      "learning_rate": 5e-06,
+      "loss": 0.731,
+      "num_input_tokens_seen": 14954408,
+      "step": 87
+    },
+    {
+      "epoch": 0.02288419806667982,
+      "loss": 0.7506657838821411,
+      "loss_ce": 0.2797185182571411,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.09423828125,
+      "loss_xval": 0.470703125,
+      "num_input_tokens_seen": 14954408,
+      "step": 87
+    },
+    {
+      "epoch": 0.023147234826066943,
+      "grad_norm": 26.5172487427058,
+      "learning_rate": 5e-06,
+      "loss": 0.8202,
+      "num_input_tokens_seen": 15123552,
+      "step": 88
+    },
+    {
+      "epoch": 0.023147234826066943,
+      "loss": 0.808368980884552,
+      "loss_ce": 0.243183434009552,
+      "loss_iou": 0.234375,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 15123552,
+      "step": 88
+    },
+    {
+      "epoch": 0.023410271585454067,
+      "grad_norm": 17.71060921250277,
+      "learning_rate": 5e-06,
+      "loss": 0.6546,
+      "num_input_tokens_seen": 15295844,
+      "step": 89
+    },
+    {
+      "epoch": 0.023410271585454067,
+      "loss": 0.6125390529632568,
+      "loss_ce": 0.24852542579174042,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 15295844,
+      "step": 89
+    },
+    {
+      "epoch": 0.02367330834484119,
+      "grad_norm": 15.735800845783544,
+      "learning_rate": 5e-06,
+      "loss": 0.5838,
+      "num_input_tokens_seen": 15467856,
+      "step": 90
+    },
+    {
+      "epoch": 0.02367330834484119,
+      "loss": 0.5133854150772095,
+      "loss_ce": 0.22456704080104828,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 15467856,
+      "step": 90
+    },
+    {
+      "epoch": 0.023936345104228315,
+      "grad_norm": 16.198479692260427,
+      "learning_rate": 5e-06,
+      "loss": 0.6229,
+      "num_input_tokens_seen": 15640280,
+      "step": 91
+    },
+    {
+      "epoch": 0.023936345104228315,
+      "loss": 0.5497957468032837,
+      "loss_ce": 0.23717370629310608,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 15640280,
+      "step": 91
+    },
+    {
+      "epoch": 0.02419938186361544,
+      "grad_norm": 15.820711258384152,
+      "learning_rate": 5e-06,
+      "loss": 0.5852,
+      "num_input_tokens_seen": 15812532,
+      "step": 92
+    },
+    {
+      "epoch": 0.02419938186361544,
+      "loss": 0.5448044538497925,
+      "loss_ce": 0.22046364843845367,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 15812532,
+      "step": 92
+    },
+    {
+      "epoch": 0.024462418623002566,
+      "grad_norm": 19.666745517026683,
+      "learning_rate": 5e-06,
+      "loss": 0.5956,
+      "num_input_tokens_seen": 15984492,
+      "step": 93
+    },
+    {
+      "epoch": 0.024462418623002566,
+      "loss": 0.6414846181869507,
+      "loss_ce": 0.22668972611427307,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 15984492,
+      "step": 93
+    },
+    {
+      "epoch": 0.02472545538238969,
+      "grad_norm": 18.823515656198328,
+      "learning_rate": 5e-06,
+      "loss": 0.5443,
+      "num_input_tokens_seen": 16156556,
+      "step": 94
+    },
+    {
+      "epoch": 0.02472545538238969,
+      "loss": 0.507426381111145,
+      "loss_ce": 0.2339888960123062,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 16156556,
+      "step": 94
+    },
+    {
+      "epoch": 0.024988492141776814,
+      "grad_norm": 14.625660314629584,
+      "learning_rate": 5e-06,
+      "loss": 0.5402,
+      "num_input_tokens_seen": 16326928,
+      "step": 95
+    },
+    {
+      "epoch": 0.024988492141776814,
+      "loss": 0.5663172006607056,
+      "loss_ce": 0.20572152733802795,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 16326928,
+      "step": 95
+    },
+    {
+      "epoch": 0.025251528901163937,
+      "grad_norm": 15.783911320035779,
+      "learning_rate": 5e-06,
+      "loss": 0.5712,
+      "num_input_tokens_seen": 16499268,
+      "step": 96
+    },
+    {
+      "epoch": 0.025251528901163937,
+      "loss": 0.5024785399436951,
+      "loss_ce": 0.19632619619369507,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 16499268,
+      "step": 96
+    },
+    {
+      "epoch": 0.02551456566055106,
+      "grad_norm": 16.72236617723868,
+      "learning_rate": 5e-06,
+      "loss": 0.6175,
+      "num_input_tokens_seen": 16668388,
+      "step": 97
+    },
+    {
+      "epoch": 0.02551456566055106,
+      "loss": 0.6046057939529419,
+      "loss_ce": 0.19554820656776428,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 16668388,
+      "step": 97
+    },
+    {
+      "epoch": 0.025777602419938185,
+      "grad_norm": 19.132328111043208,
+      "learning_rate": 5e-06,
+      "loss": 0.6948,
+      "num_input_tokens_seen": 16840748,
+      "step": 98
+    },
+    {
+      "epoch": 0.025777602419938185,
+      "loss": 0.635749340057373,
+      "loss_ce": 0.22290757298469543,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.08251953125,
+      "loss_xval": 0.412109375,
+      "num_input_tokens_seen": 16840748,
+      "step": 98
+    },
+    {
+      "epoch": 0.026040639179325312,
+      "grad_norm": 17.546404729768323,
+      "learning_rate": 5e-06,
+      "loss": 0.6342,
+      "num_input_tokens_seen": 17012828,
+      "step": 99
+    },
+    {
+      "epoch": 0.026040639179325312,
+      "loss": 0.6548875570297241,
+      "loss_ce": 0.23105943202972412,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0849609375,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 17012828,
+      "step": 99
+    },
+    {
+      "epoch": 0.026303675938712436,
+      "grad_norm": 18.526246284946534,
+      "learning_rate": 5e-06,
+      "loss": 0.6155,
+      "num_input_tokens_seen": 17185216,
+      "step": 100
+    },
+    {
+      "epoch": 0.026303675938712436,
+      "loss": 0.6633030772209167,
+      "loss_ce": 0.20383042097091675,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.091796875,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 17185216,
+      "step": 100
+    },
+    {
+      "epoch": 0.02656671269809956,
+      "grad_norm": 19.37711860101834,
+      "learning_rate": 5e-06,
+      "loss": 0.6252,
+      "num_input_tokens_seen": 17355688,
+      "step": 101
+    },
+    {
+      "epoch": 0.02656671269809956,
+      "loss": 0.6485173106193542,
+      "loss_ce": 0.20686691999435425,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.08837890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 17355688,
+      "step": 101
+    },
+    {
+      "epoch": 0.026829749457486684,
+      "grad_norm": 24.272978763244147,
+      "learning_rate": 5e-06,
+      "loss": 0.6144,
+      "num_input_tokens_seen": 17527784,
+      "step": 102
+    },
+    {
+      "epoch": 0.026829749457486684,
+      "loss": 0.6552723050117493,
+      "loss_ce": 0.20190313458442688,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 17527784,
+      "step": 102
+    },
+    {
+      "epoch": 0.027092786216873808,
+      "grad_norm": 28.829409367834085,
+      "learning_rate": 5e-06,
+      "loss": 0.6606,
+      "num_input_tokens_seen": 17700208,
+      "step": 103
+    },
+    {
+      "epoch": 0.027092786216873808,
+      "loss": 0.7094471454620361,
+      "loss_ce": 0.19260142743587494,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.10302734375,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 17700208,
+      "step": 103
+    },
+    {
+      "epoch": 0.02735582297626093,
+      "grad_norm": 23.160720035944347,
+      "learning_rate": 5e-06,
+      "loss": 0.7452,
+      "num_input_tokens_seen": 17872308,
+      "step": 104
+    },
+    {
+      "epoch": 0.02735582297626093,
+      "loss": 0.7341784238815308,
+      "loss_ce": 0.19169792532920837,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.1083984375,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 17872308,
+      "step": 104
+    },
+    {
+      "epoch": 0.027618859735648055,
+      "grad_norm": 17.418429269221175,
+      "learning_rate": 5e-06,
+      "loss": 0.5803,
+      "num_input_tokens_seen": 18044728,
+      "step": 105
+    },
+    {
+      "epoch": 0.027618859735648055,
+      "loss": 0.5607779026031494,
+      "loss_ce": 0.20787259936332703,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 18044728,
+      "step": 105
+    },
+    {
+      "epoch": 0.027881896495035183,
+      "grad_norm": 17.11728575546532,
+      "learning_rate": 5e-06,
+      "loss": 0.5934,
+      "num_input_tokens_seen": 18215172,
+      "step": 106
+    },
+    {
+      "epoch": 0.027881896495035183,
+      "loss": 0.609626293182373,
+      "loss_ce": 0.17749738693237305,
+      "loss_iou": 0.09228515625,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 18215172,
+      "step": 106
+    },
+    {
+      "epoch": 0.028144933254422307,
+      "grad_norm": 17.86366659461009,
+      "learning_rate": 5e-06,
+      "loss": 0.5335,
+      "num_input_tokens_seen": 18387084,
+      "step": 107
+    },
+    {
+      "epoch": 0.028144933254422307,
+      "loss": 0.48998841643333435,
+      "loss_ce": 0.19213685393333435,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 18387084,
+      "step": 107
+    },
+    {
+      "epoch": 0.02840797001380943,
+      "grad_norm": 20.662496273440333,
+      "learning_rate": 5e-06,
+      "loss": 0.5971,
+      "num_input_tokens_seen": 18559272,
+      "step": 108
+    },
+    {
+      "epoch": 0.02840797001380943,
+      "loss": 0.6421129703521729,
+      "loss_ce": 0.18044306337833405,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.09228515625,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 18559272,
+      "step": 108
+    },
+    {
+      "epoch": 0.028671006773196554,
+      "grad_norm": 21.435817028210696,
+      "learning_rate": 5e-06,
+      "loss": 0.5635,
+      "num_input_tokens_seen": 18731380,
+      "step": 109
+    },
+    {
+      "epoch": 0.028671006773196554,
+      "loss": 0.5063576698303223,
+      "loss_ce": 0.18104028701782227,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 18731380,
+      "step": 109
+    },
+    {
+      "epoch": 0.028934043532583678,
+      "grad_norm": 18.8758769532993,
+      "learning_rate": 5e-06,
+      "loss": 0.6115,
+      "num_input_tokens_seen": 18901956,
+      "step": 110
+    },
+    {
+      "epoch": 0.028934043532583678,
+      "loss": 0.6692255139350891,
+      "loss_ce": 0.17020206153392792,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.099609375,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 18901956,
+      "step": 110
+    },
+    {
+      "epoch": 0.029197080291970802,
+      "grad_norm": 17.651564073843637,
+      "learning_rate": 5e-06,
+      "loss": 0.5336,
+      "num_input_tokens_seen": 19072248,
+      "step": 111
+    },
+    {
+      "epoch": 0.029197080291970802,
+      "loss": 0.4951699376106262,
+      "loss_ce": 0.17558985948562622,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 19072248,
+      "step": 111
+    },
+    {
+      "epoch": 0.029460117051357926,
+      "grad_norm": 16.158969266118735,
+      "learning_rate": 5e-06,
+      "loss": 0.4952,
+      "num_input_tokens_seen": 19244772,
+      "step": 112
+    },
+    {
+      "epoch": 0.029460117051357926,
+      "loss": 0.4770987629890442,
+      "loss_ce": 0.1765616536140442,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 19244772,
+      "step": 112
+    },
+    {
+      "epoch": 0.029723153810745053,
+      "grad_norm": 14.450830992407502,
+      "learning_rate": 5e-06,
+      "loss": 0.5026,
+      "num_input_tokens_seen": 19416868,
+      "step": 113
+    },
+    {
+      "epoch": 0.029723153810745053,
+      "loss": 0.4577527642250061,
+      "loss_ce": 0.1758924424648285,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 19416868,
+      "step": 113
+    },
+    {
+      "epoch": 0.029986190570132177,
+      "grad_norm": 14.560235178346835,
+      "learning_rate": 5e-06,
+      "loss": 0.4865,
+      "num_input_tokens_seen": 19589016,
+      "step": 114
+    },
+    {
+      "epoch": 0.029986190570132177,
+      "loss": 0.5253910422325134,
+      "loss_ce": 0.15832561254501343,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 19589016,
+      "step": 114
+    },
+    {
+      "epoch": 0.0302492273295193,
+      "grad_norm": 18.605711889816632,
+      "learning_rate": 5e-06,
+      "loss": 0.49,
+      "num_input_tokens_seen": 19760904,
+      "step": 115
+    },
+    {
+      "epoch": 0.0302492273295193,
+      "loss": 0.4801591634750366,
+      "loss_ce": 0.17290815711021423,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 19760904,
+      "step": 115
+    },
+    {
+      "epoch": 0.030512264088906425,
+      "grad_norm": 20.839979972241576,
+      "learning_rate": 5e-06,
+      "loss": 0.5137,
+      "num_input_tokens_seen": 19932724,
+      "step": 116
+    },
+    {
+      "epoch": 0.030512264088906425,
+      "loss": 0.4843531847000122,
+      "loss_ce": 0.14585217833518982,
+      "loss_iou": NaN,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 19932724,
+      "step": 116
+    },
+    {
+      "epoch": 0.03077530084829355,
+      "grad_norm": 20.407619855634906,
+      "learning_rate": 5e-06,
+      "loss": 0.5188,
+      "num_input_tokens_seen": 20105196,
+      "step": 117
+    },
+    {
+      "epoch": 0.03077530084829355,
+      "loss": 0.521455705165863,
+      "loss_ce": 0.17148011922836304,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 20105196,
+      "step": 117
+    },
+    {
+      "epoch": 0.031038337607680672,
+      "grad_norm": 17.136634668569023,
+      "learning_rate": 5e-06,
+      "loss": 0.529,
+      "num_input_tokens_seen": 20277468,
+      "step": 118
+    },
+    {
+      "epoch": 0.031038337607680672,
+      "loss": 0.5619306564331055,
+      "loss_ce": 0.15323926508426666,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 20277468,
+      "step": 118
+    },
+    {
+      "epoch": 0.0313013743670678,
+      "grad_norm": 15.540166390852267,
+      "learning_rate": 5e-06,
+      "loss": 0.5196,
+      "num_input_tokens_seen": 20448108,
+      "step": 119
+    },
+    {
+      "epoch": 0.0313013743670678,
+      "loss": 0.5409432649612427,
+      "loss_ce": 0.15117278695106506,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.078125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 20448108,
+      "step": 119
+    },
+    {
+      "epoch": 0.031564411126454923,
+      "grad_norm": 12.306363086115368,
+      "learning_rate": 5e-06,
+      "loss": 0.4996,
+      "num_input_tokens_seen": 20620368,
+      "step": 120
+    },
+    {
+      "epoch": 0.031564411126454923,
+      "loss": 0.49895310401916504,
+      "loss_ce": 0.13860151171684265,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 20620368,
+      "step": 120
+    },
+    {
+      "epoch": 0.03182744788584205,
+      "grad_norm": 13.090987000723873,
+      "learning_rate": 5e-06,
+      "loss": 0.4537,
+      "num_input_tokens_seen": 20792584,
+      "step": 121
+    },
+    {
+      "epoch": 0.03182744788584205,
+      "loss": 0.47374969720840454,
+      "loss_ce": 0.14452606439590454,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 20792584,
+      "step": 121
+    },
+    {
+      "epoch": 0.03209048464522917,
+      "grad_norm": 12.574059354034341,
+      "learning_rate": 5e-06,
+      "loss": 0.4245,
+      "num_input_tokens_seen": 20964948,
+      "step": 122
+    },
+    {
+      "epoch": 0.03209048464522917,
+      "loss": 0.4678102135658264,
+      "loss_ce": 0.12357192486524582,
+      "loss_iou": 0.10400390625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 20964948,
+      "step": 122
+    },
+    {
+      "epoch": 0.032353521404616295,
+      "grad_norm": 34.33651980128175,
+      "learning_rate": 5e-06,
+      "loss": 0.5624,
+      "num_input_tokens_seen": 21134744,
+      "step": 123
+    },
+    {
+      "epoch": 0.032353521404616295,
+      "loss": 0.4556346535682678,
+      "loss_ce": 0.13812974095344543,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 21134744,
+      "step": 123
+    },
+    {
+      "epoch": 0.03261655816400342,
+      "grad_norm": 22.118966434798295,
+      "learning_rate": 5e-06,
+      "loss": 0.6017,
+      "num_input_tokens_seen": 21305320,
+      "step": 124
+    },
+    {
+      "epoch": 0.03261655816400342,
+      "loss": 0.5379438996315002,
+      "loss_ce": 0.14316853880882263,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 21305320,
+      "step": 124
+    },
+    {
+      "epoch": 0.03287959492339054,
+      "grad_norm": 22.565129824621987,
+      "learning_rate": 5e-06,
+      "loss": 0.6452,
+      "num_input_tokens_seen": 21477544,
+      "step": 125
+    },
+    {
+      "epoch": 0.03287959492339054,
+      "loss": 0.5944963693618774,
+      "loss_ce": 0.13990655541419983,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 21477544,
+      "step": 125
+    },
+    {
+      "epoch": 0.033142631682777667,
+      "grad_norm": 13.89281994742959,
+      "learning_rate": 5e-06,
+      "loss": 0.5755,
+      "num_input_tokens_seen": 21649600,
+      "step": 126
+    },
+    {
+      "epoch": 0.033142631682777667,
+      "loss": 0.5376471877098083,
+      "loss_ce": 0.12211985141038895,
+      "loss_iou": 0.10888671875,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 21649600,
+      "step": 126
+    },
+    {
+      "epoch": 0.03340566844216479,
+      "grad_norm": 15.070169412653112,
+      "learning_rate": 5e-06,
+      "loss": 0.4704,
+      "num_input_tokens_seen": 21821564,
+      "step": 127
+    },
+    {
+      "epoch": 0.03340566844216479,
+      "loss": 0.4526183605194092,
+      "loss_ce": 0.12595820426940918,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 21821564,
+      "step": 127
+    },
+    {
+      "epoch": 0.033668705201551914,
+      "grad_norm": 14.62625838728594,
+      "learning_rate": 5e-06,
+      "loss": 0.4885,
+      "num_input_tokens_seen": 21993864,
+      "step": 128
+    },
+    {
+      "epoch": 0.033668705201551914,
+      "loss": 0.5165751576423645,
+      "loss_ce": 0.12289837747812271,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 21993864,
+      "step": 128
+    },
+    {
+      "epoch": 0.03393174196093904,
+      "grad_norm": 20.05433315099477,
+      "learning_rate": 5e-06,
+      "loss": 0.4754,
+      "num_input_tokens_seen": 22162596,
+      "step": 129
+    },
+    {
+      "epoch": 0.03393174196093904,
+      "loss": 0.4400935173034668,
+      "loss_ce": 0.11538645625114441,
+      "loss_iou": 0.59375,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 22162596,
+      "step": 129
+    },
+    {
+      "epoch": 0.03419477872032617,
+      "grad_norm": 17.28797678972647,
+      "learning_rate": 5e-06,
+      "loss": 0.5272,
+      "num_input_tokens_seen": 22334624,
+      "step": 130
+    },
+    {
+      "epoch": 0.03419477872032617,
+      "loss": 0.5726144313812256,
+      "loss_ce": 0.1124093234539032,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.091796875,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 22334624,
+      "step": 130
+    },
+    {
+      "epoch": 0.03445781547971329,
+      "grad_norm": 19.611364973231527,
+      "learning_rate": 5e-06,
+      "loss": 0.5629,
+      "num_input_tokens_seen": 22506552,
+      "step": 131
+    },
+    {
+      "epoch": 0.03445781547971329,
+      "loss": 0.5982115268707275,
+      "loss_ce": 0.10993030667304993,
+      "loss_iou": 0.08642578125,
+      "loss_num": 0.09765625,
+      "loss_xval": 0.48828125,
+      "num_input_tokens_seen": 22506552,
+      "step": 131
+    },
+    {
+      "epoch": 0.034720852239100417,
+      "grad_norm": 16.543739789854843,
+      "learning_rate": 5e-06,
+      "loss": 0.5381,
+      "num_input_tokens_seen": 22678808,
+      "step": 132
+    },
+    {
+      "epoch": 0.034720852239100417,
+      "loss": 0.45885854959487915,
+      "loss_ce": 0.11523060500621796,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 22678808,
+      "step": 132
+    },
+    {
+      "epoch": 0.03498388899848754,
+      "grad_norm": 17.674510611725847,
+      "learning_rate": 5e-06,
+      "loss": 0.4525,
+      "num_input_tokens_seen": 22850736,
+      "step": 133
+    },
+    {
+      "epoch": 0.03498388899848754,
+      "loss": 0.45272764563560486,
+      "loss_ce": 0.12191709131002426,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 22850736,
+      "step": 133
+    },
+    {
+      "epoch": 0.035246925757874664,
+      "grad_norm": 15.972341632067714,
+      "learning_rate": 5e-06,
+      "loss": 0.4719,
+      "num_input_tokens_seen": 23020040,
+      "step": 134
+    },
+    {
+      "epoch": 0.035246925757874664,
+      "loss": 0.5267431139945984,
+      "loss_ce": 0.11499997228384018,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.08251953125,
+      "loss_xval": 0.412109375,
+      "num_input_tokens_seen": 23020040,
+      "step": 134
+    },
+    {
+      "epoch": 0.03550996251726179,
+      "grad_norm": 18.16452757572112,
+      "learning_rate": 5e-06,
+      "loss": 0.4878,
+      "num_input_tokens_seen": 23192048,
+      "step": 135
+    },
+    {
+      "epoch": 0.03550996251726179,
+      "loss": 0.42464134097099304,
+      "loss_ce": 0.10701439529657364,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 23192048,
+      "step": 135
+    },
+    {
+      "epoch": 0.03577299927664891,
+      "grad_norm": 16.237572483673375,
+      "learning_rate": 5e-06,
+      "loss": 0.4644,
+      "num_input_tokens_seen": 23360448,
+      "step": 136
+    },
+    {
+      "epoch": 0.03577299927664891,
+      "loss": 0.4710727334022522,
+      "loss_ce": 0.0985141396522522,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 23360448,
+      "step": 136
+    },
+    {
+      "epoch": 0.036036036036036036,
+      "grad_norm": 18.804237772436053,
+      "learning_rate": 5e-06,
+      "loss": 0.4291,
+      "num_input_tokens_seen": 23530300,
+      "step": 137
+    },
+    {
+      "epoch": 0.036036036036036036,
+      "loss": 0.47930601239204407,
+      "loss_ce": 0.10406187921762466,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 23530300,
+      "step": 137
+    },
+    {
+      "epoch": 0.03629907279542316,
+      "grad_norm": 17.355021474463314,
+      "learning_rate": 5e-06,
+      "loss": 0.5374,
+      "num_input_tokens_seen": 23702304,
+      "step": 138
+    },
+    {
+      "epoch": 0.03629907279542316,
+      "loss": 0.6615355014801025,
+      "loss_ce": 0.10416243970394135,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.111328125,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 23702304,
+      "step": 138
+    },
+    {
+      "epoch": 0.03656210955481028,
+      "grad_norm": 14.642948575679453,
+      "learning_rate": 5e-06,
+      "loss": 0.4417,
+      "num_input_tokens_seen": 23872132,
+      "step": 139
+    },
+    {
+      "epoch": 0.03656210955481028,
+      "loss": 0.44469529390335083,
+      "loss_ce": 0.10735400021076202,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 23872132,
+      "step": 139
+    },
+    {
+      "epoch": 0.03682514631419741,
+      "grad_norm": 14.422554847110444,
+      "learning_rate": 5e-06,
+      "loss": 0.4228,
+      "num_input_tokens_seen": 24044200,
+      "step": 140
+    },
+    {
+      "epoch": 0.03682514631419741,
+      "loss": 0.4664979577064514,
+      "loss_ce": 0.10651260614395142,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 24044200,
+      "step": 140
+    },
+    {
+      "epoch": 0.03708818307358453,
+      "grad_norm": 27.027123187239088,
+      "learning_rate": 5e-06,
+      "loss": 0.4106,
+      "num_input_tokens_seen": 24216464,
+      "step": 141
+    },
+    {
+      "epoch": 0.03708818307358453,
+      "loss": 0.4140710234642029,
+      "loss_ce": 0.09876340627670288,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 24216464,
+      "step": 141
+    },
+    {
+      "epoch": 0.037351219832971655,
+      "grad_norm": 20.30988259642017,
+      "learning_rate": 5e-06,
+      "loss": 0.5663,
+      "num_input_tokens_seen": 24386876,
+      "step": 142
+    },
+    {
+      "epoch": 0.037351219832971655,
+      "loss": 0.500027596950531,
+      "loss_ce": 0.091580331325531,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 24386876,
+      "step": 142
+    },
+    {
+      "epoch": 0.03761425659235878,
+      "grad_norm": 17.151019381080523,
+      "learning_rate": 5e-06,
+      "loss": 0.4943,
+      "num_input_tokens_seen": 24559004,
+      "step": 143
+    },
+    {
+      "epoch": 0.03761425659235878,
+      "loss": 0.5077698826789856,
+      "loss_ce": 0.08748182654380798,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 24559004,
+      "step": 143
+    },
+    {
+      "epoch": 0.03787729335174591,
+      "grad_norm": 13.449145519268138,
+      "learning_rate": 5e-06,
+      "loss": 0.4628,
+      "num_input_tokens_seen": 24729228,
+      "step": 144
+    },
+    {
+      "epoch": 0.03787729335174591,
+      "loss": 0.477780282497406,
+      "loss_ce": 0.0906953439116478,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 24729228,
+      "step": 144
+    },
+    {
+      "epoch": 0.03814033011113303,
+      "grad_norm": 15.658548235445116,
+      "learning_rate": 5e-06,
+      "loss": 0.4482,
+      "num_input_tokens_seen": 24901324,
+      "step": 145
+    },
+    {
+      "epoch": 0.03814033011113303,
+      "loss": 0.460429847240448,
+      "loss_ce": 0.086406409740448,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 24901324,
+      "step": 145
+    },
+    {
+      "epoch": 0.03840336687052016,
+      "grad_norm": 14.557303161182968,
+      "learning_rate": 5e-06,
+      "loss": 0.5115,
+      "num_input_tokens_seen": 25073508,
+      "step": 146
+    },
+    {
+      "epoch": 0.03840336687052016,
+      "loss": 0.6442508697509766,
+      "loss_ce": 0.07833293080329895,
+      "loss_iou": 0.11376953125,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 25073508,
+      "step": 146
+    },
+    {
+      "epoch": 0.03866640362990728,
+      "grad_norm": 17.777843374612736,
+      "learning_rate": 5e-06,
+      "loss": 0.4213,
+      "num_input_tokens_seen": 25245752,
+      "step": 147
+    },
+    {
+      "epoch": 0.03866640362990728,
+      "loss": 0.40006011724472046,
+      "loss_ce": 0.08243316411972046,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 25245752,
+      "step": 147
+    },
+    {
+      "epoch": 0.038929440389294405,
+      "grad_norm": 13.44125897811313,
+      "learning_rate": 5e-06,
+      "loss": 0.4791,
+      "num_input_tokens_seen": 25418156,
+      "step": 148
+    },
+    {
+      "epoch": 0.038929440389294405,
+      "loss": 0.5981014966964722,
+      "loss_ce": 0.08491791784763336,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 25418156,
+      "step": 148
+    },
+    {
+      "epoch": 0.03919247714868153,
+      "grad_norm": 11.76514004625091,
+      "learning_rate": 5e-06,
+      "loss": 0.4001,
+      "num_input_tokens_seen": 25589904,
+      "step": 149
+    },
+    {
+      "epoch": 0.03919247714868153,
+      "loss": 0.35078275203704834,
+      "loss_ce": 0.08393705636262894,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 25589904,
+      "step": 149
+    },
+    {
+      "epoch": 0.03945551390806865,
+      "grad_norm": 12.03724933893627,
+      "learning_rate": 5e-06,
+      "loss": 0.4147,
+      "num_input_tokens_seen": 25762000,
+      "step": 150
+    },
+    {
+      "epoch": 0.03945551390806865,
+      "loss": 0.4596315622329712,
+      "loss_ce": 0.0690065547823906,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.078125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 25762000,
+      "step": 150
+    },
+    {
+      "epoch": 0.039718550667455776,
+      "grad_norm": 13.05163620933256,
+      "learning_rate": 5e-06,
+      "loss": 0.398,
+      "num_input_tokens_seen": 25934528,
+      "step": 151
+    },
+    {
+      "epoch": 0.039718550667455776,
+      "loss": 0.38974958658218384,
+      "loss_ce": 0.07322127372026443,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 25934528,
+      "step": 151
+    },
+    {
+      "epoch": 0.0399815874268429,
+      "grad_norm": 23.58601457584165,
+      "learning_rate": 5e-06,
+      "loss": 0.3897,
+      "num_input_tokens_seen": 26104148,
+      "step": 152
+    },
+    {
+      "epoch": 0.0399815874268429,
+      "loss": 0.3437004089355469,
+      "loss_ce": 0.07172775268554688,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 26104148,
+      "step": 152
+    },
+    {
+      "epoch": 0.040244624186230024,
+      "grad_norm": 18.61300216202182,
+      "learning_rate": 5e-06,
+      "loss": 0.4682,
+      "num_input_tokens_seen": 26276256,
+      "step": 153
+    },
+    {
+      "epoch": 0.040244624186230024,
+      "loss": 0.43069130182266235,
+      "loss_ce": 0.06960733234882355,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 26276256,
+      "step": 153
+    },
+    {
+      "epoch": 0.04050766094561715,
+      "grad_norm": 14.116759080889954,
+      "learning_rate": 5e-06,
+      "loss": 0.4744,
+      "num_input_tokens_seen": 26448288,
+      "step": 154
+    },
+    {
+      "epoch": 0.04050766094561715,
+      "loss": 0.5761303901672363,
+      "loss_ce": 0.0805249810218811,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.09912109375,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 26448288,
+      "step": 154
+    },
+    {
+      "epoch": 0.04077069770500427,
+      "grad_norm": 10.146662886533779,
+      "learning_rate": 5e-06,
+      "loss": 0.3974,
+      "num_input_tokens_seen": 26620604,
+      "step": 155
+    },
+    {
+      "epoch": 0.04077069770500427,
+      "loss": 0.3559998869895935,
+      "loss_ce": 0.07047741115093231,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 26620604,
+      "step": 155
+    },
+    {
+      "epoch": 0.041033734464391396,
+      "grad_norm": 11.74557828439257,
+      "learning_rate": 5e-06,
+      "loss": 0.3616,
+      "num_input_tokens_seen": 26789104,
+      "step": 156
+    },
+    {
+      "epoch": 0.041033734464391396,
+      "loss": 0.3762925863265991,
+      "loss_ce": 0.06867540627717972,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 26789104,
+      "step": 156
+    },
+    {
+      "epoch": 0.041296771223778526,
+      "grad_norm": 11.967339752616043,
+      "learning_rate": 5e-06,
+      "loss": 0.3655,
+      "num_input_tokens_seen": 26961408,
+      "step": 157
+    },
+    {
+      "epoch": 0.041296771223778526,
+      "loss": 0.36129921674728394,
+      "loss_ce": 0.060029659420251846,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 26961408,
+      "step": 157
+    },
+    {
+      "epoch": 0.04155980798316565,
+      "grad_norm": 16.462145017062344,
+      "learning_rate": 5e-06,
+      "loss": 0.3599,
+      "num_input_tokens_seen": 27133600,
+      "step": 158
+    },
+    {
+      "epoch": 0.04155980798316565,
+      "loss": 0.4060816764831543,
+      "loss_ce": 0.06208755075931549,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 27133600,
+      "step": 158
+    },
+    {
+      "epoch": 0.041822844742552774,
+      "grad_norm": 18.2471111237373,
+      "learning_rate": 5e-06,
+      "loss": 0.4173,
+      "num_input_tokens_seen": 27305676,
+      "step": 159
+    },
+    {
+      "epoch": 0.041822844742552774,
+      "loss": 0.3555372357368469,
+      "loss_ce": 0.07025889307260513,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 27305676,
+      "step": 159
+    },
+    {
+      "epoch": 0.0420858815019399,
+      "grad_norm": 22.735392677918405,
+      "learning_rate": 5e-06,
+      "loss": 0.5962,
+      "num_input_tokens_seen": 27477824,
+      "step": 160
+    },
+    {
+      "epoch": 0.0420858815019399,
+      "loss": 0.6341134309768677,
+      "loss_ce": 0.06184776872396469,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.1142578125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 27477824,
+      "step": 160
+    },
+    {
+      "epoch": 0.04234891826132702,
+      "grad_norm": 13.104760953447814,
+      "learning_rate": 5e-06,
+      "loss": 0.4091,
+      "num_input_tokens_seen": 27650024,
+      "step": 161
+    },
+    {
+      "epoch": 0.04234891826132702,
+      "loss": 0.4603307843208313,
+      "loss_ce": 0.059085663408041,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 27650024,
+      "step": 161
+    },
+    {
+      "epoch": 0.042611955020714146,
+      "grad_norm": 15.303759630511669,
+      "learning_rate": 5e-06,
+      "loss": 0.4433,
+      "num_input_tokens_seen": 27822408,
+      "step": 162
+    },
+    {
+      "epoch": 0.042611955020714146,
+      "loss": 0.43349525332450867,
+      "loss_ce": 0.06215735524892807,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 27822408,
+      "step": 162
+    },
+    {
+      "epoch": 0.04287499178010127,
+      "grad_norm": 14.84522191071012,
+      "learning_rate": 5e-06,
+      "loss": 0.3769,
+      "num_input_tokens_seen": 27992440,
+      "step": 163
+    },
+    {
+      "epoch": 0.04287499178010127,
+      "loss": 0.34604543447494507,
+      "loss_ce": 0.05673878639936447,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 27992440,
+      "step": 163
+    },
+    {
+      "epoch": 0.04313802853948839,
+      "grad_norm": 17.265704948492278,
+      "learning_rate": 5e-06,
+      "loss": 0.4298,
+      "num_input_tokens_seen": 28164464,
+      "step": 164
+    },
+    {
+      "epoch": 0.04313802853948839,
+      "loss": 0.48720821738243103,
+      "loss_ce": 0.05849727243185043,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 28164464,
+      "step": 164
+    },
+    {
+      "epoch": 0.04340106529887552,
+      "grad_norm": 13.53959312060916,
+      "learning_rate": 5e-06,
+      "loss": 0.3991,
+      "num_input_tokens_seen": 28334436,
+      "step": 165
+    },
+    {
+      "epoch": 0.04340106529887552,
+      "loss": 0.3757469058036804,
+      "loss_ce": 0.05201642960309982,
+      "loss_iou": 0.1220703125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 28334436,
+      "step": 165
+    },
+    {
+      "epoch": 0.04366410205826264,
+      "grad_norm": 12.759789778648042,
+      "learning_rate": 5e-06,
+      "loss": 0.3563,
+      "num_input_tokens_seen": 28506772,
+      "step": 166
+    },
+    {
+      "epoch": 0.04366410205826264,
+      "loss": 0.3605605363845825,
+      "loss_ce": 0.05233298987150192,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 28506772,
+      "step": 166
+    },
+    {
+      "epoch": 0.043927138817649765,
+      "grad_norm": 14.640887630603752,
+      "learning_rate": 5e-06,
+      "loss": 0.3881,
+      "num_input_tokens_seen": 28676980,
+      "step": 167
+    },
+    {
+      "epoch": 0.043927138817649765,
+      "loss": 0.3610857427120209,
+      "loss_ce": 0.056642383337020874,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 28676980,
+      "step": 167
+    },
+    {
+      "epoch": 0.04419017557703689,
+      "grad_norm": 12.095247845748437,
+      "learning_rate": 5e-06,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 28849236,
+      "step": 168
+    },
+    {
+      "epoch": 0.04419017557703689,
+      "loss": 0.40532949566841125,
+      "loss_ce": 0.05388907343149185,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 28849236,
+      "step": 168
+    },
+    {
+      "epoch": 0.04445321233642401,
+      "grad_norm": 10.33587361024043,
+      "learning_rate": 5e-06,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 29021180,
+      "step": 169
+    },
+    {
+      "epoch": 0.04445321233642401,
+      "loss": 0.3248461186885834,
+      "loss_ce": 0.046464771032333374,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 29021180,
+      "step": 169
+    },
+    {
+      "epoch": 0.044716249095811136,
+      "grad_norm": 10.708625450816182,
+      "learning_rate": 5e-06,
+      "loss": 0.3951,
+      "num_input_tokens_seen": 29191532,
+      "step": 170
+    },
+    {
+      "epoch": 0.044716249095811136,
+      "loss": 0.4214698076248169,
+      "loss_ce": 0.052573323249816895,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 29191532,
+      "step": 170
+    },
+    {
+      "epoch": 0.04497928585519827,
+      "grad_norm": 14.817164885231227,
+      "learning_rate": 5e-06,
+      "loss": 0.3936,
+      "num_input_tokens_seen": 29363536,
+      "step": 171
+    },
+    {
+      "epoch": 0.04497928585519827,
+      "loss": 0.40210169553756714,
+      "loss_ce": 0.04602260887622833,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 29363536,
+      "step": 171
+    },
+    {
+      "epoch": 0.04524232261458539,
+      "grad_norm": 11.284174351804506,
+      "learning_rate": 5e-06,
+      "loss": 0.3753,
+      "num_input_tokens_seen": 29533848,
+      "step": 172
+    },
+    {
+      "epoch": 0.04524232261458539,
+      "loss": 0.3555999994277954,
+      "loss_ce": 0.04358828812837601,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 29533848,
+      "step": 172
+    },
+    {
+      "epoch": 0.045505359373972515,
+      "grad_norm": 9.66534144364822,
+      "learning_rate": 5e-06,
+      "loss": 0.307,
+      "num_input_tokens_seen": 29706424,
+      "step": 173
+    },
+    {
+      "epoch": 0.045505359373972515,
+      "loss": 0.3183665871620178,
+      "loss_ce": 0.04297598451375961,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 29706424,
+      "step": 173
+    },
+    {
+      "epoch": 0.04576839613335964,
+      "grad_norm": 15.933225853148851,
+      "learning_rate": 5e-06,
+      "loss": 0.3645,
+      "num_input_tokens_seen": 29878348,
+      "step": 174
+    },
+    {
+      "epoch": 0.04576839613335964,
+      "loss": 0.24233956634998322,
+      "loss_ce": 0.051421597599983215,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 29878348,
+      "step": 174
+    },
+    {
+      "epoch": 0.04603143289274676,
+      "grad_norm": 15.32149954029387,
+      "learning_rate": 5e-06,
+      "loss": 0.403,
+      "num_input_tokens_seen": 30050700,
+      "step": 175
+    },
+    {
+      "epoch": 0.04603143289274676,
+      "loss": 0.41682887077331543,
+      "loss_ce": 0.047810301184654236,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 30050700,
+      "step": 175
+    },
+    {
+      "epoch": 0.046294469652133886,
+      "grad_norm": 11.301125755600708,
+      "learning_rate": 5e-06,
+      "loss": 0.3852,
+      "num_input_tokens_seen": 30222872,
+      "step": 176
+    },
+    {
+      "epoch": 0.046294469652133886,
+      "loss": 0.35361334681510925,
+      "loss_ce": 0.04599615931510925,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 30222872,
+      "step": 176
+    },
+    {
+      "epoch": 0.04655750641152101,
+      "grad_norm": 15.04642729797989,
+      "learning_rate": 5e-06,
+      "loss": 0.3569,
+      "num_input_tokens_seen": 30394820,
+      "step": 177
+    },
+    {
+      "epoch": 0.04655750641152101,
+      "loss": 0.31377077102661133,
+      "loss_ce": 0.04106569290161133,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 30394820,
+      "step": 177
+    },
+    {
+      "epoch": 0.046820543170908134,
+      "grad_norm": 14.180764275493368,
+      "learning_rate": 5e-06,
+      "loss": 0.3856,
+      "num_input_tokens_seen": 30567088,
+      "step": 178
+    },
+    {
+      "epoch": 0.046820543170908134,
+      "loss": 0.3383968770503998,
+      "loss_ce": 0.04188808798789978,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.059326171875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 30567088,
+      "step": 178
+    },
+    {
+      "epoch": 0.04708357993029526,
+      "grad_norm": 11.43960593758222,
+      "learning_rate": 5e-06,
+      "loss": 0.3723,
+      "num_input_tokens_seen": 30739392,
+      "step": 179
+    },
+    {
+      "epoch": 0.04708357993029526,
+      "loss": 0.4008994996547699,
+      "loss_ce": 0.0434776172041893,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 30739392,
+      "step": 179
+    },
+    {
+      "epoch": 0.04734661668968238,
+      "grad_norm": 10.118856123139876,
+      "learning_rate": 5e-06,
+      "loss": 0.3842,
+      "num_input_tokens_seen": 30911592,
+      "step": 180
+    },
+    {
+      "epoch": 0.04734661668968238,
+      "loss": 0.33974575996398926,
+      "loss_ce": 0.03798792511224747,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 30911592,
+      "step": 180
+    },
+    {
+      "epoch": 0.047609653449069506,
+      "grad_norm": 8.662737873806904,
+      "learning_rate": 5e-06,
+      "loss": 0.2718,
+      "num_input_tokens_seen": 31083320,
+      "step": 181
+    },
+    {
+      "epoch": 0.047609653449069506,
+      "loss": 0.24805772304534912,
+      "loss_ce": 0.034434687346220016,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 31083320,
+      "step": 181
+    },
+    {
+      "epoch": 0.04787269020845663,
+      "grad_norm": 9.949547683003779,
+      "learning_rate": 5e-06,
+      "loss": 0.3404,
+      "num_input_tokens_seen": 31255136,
+      "step": 182
+    },
+    {
+      "epoch": 0.04787269020845663,
+      "loss": 0.3085269331932068,
+      "loss_ce": 0.03093905746936798,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 31255136,
+      "step": 182
+    },
+    {
+      "epoch": 0.04813572696784375,
+      "grad_norm": 11.304350377916183,
+      "learning_rate": 5e-06,
+      "loss": 0.3478,
+      "num_input_tokens_seen": 31427380,
+      "step": 183
+    },
+    {
+      "epoch": 0.04813572696784375,
+      "loss": 0.2798244059085846,
+      "loss_ce": 0.037881046533584595,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 31427380,
+      "step": 183
+    },
+    {
+      "epoch": 0.04839876372723088,
+      "grad_norm": 20.65800125552813,
+      "learning_rate": 5e-06,
+      "loss": 0.4261,
+      "num_input_tokens_seen": 31599344,
+      "step": 184
+    },
+    {
+      "epoch": 0.04839876372723088,
+      "loss": 0.37513959407806396,
+      "loss_ce": 0.03566203638911247,
+      "loss_iou": 0.59375,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 31599344,
+      "step": 184
+    },
+    {
+      "epoch": 0.04866180048661801,
+      "grad_norm": 21.767539584052432,
+      "learning_rate": 5e-06,
+      "loss": 0.4585,
+      "num_input_tokens_seen": 31771536,
+      "step": 185
+    },
+    {
+      "epoch": 0.04866180048661801,
+      "loss": 0.437938392162323,
+      "loss_ce": 0.031200092285871506,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 31771536,
+      "step": 185
+    },
+    {
+      "epoch": 0.04892483724600513,
+      "grad_norm": 19.478046485138204,
+      "learning_rate": 5e-06,
+      "loss": 0.5036,
+      "num_input_tokens_seen": 31943688,
+      "step": 186
+    },
+    {
+      "epoch": 0.04892483724600513,
+      "loss": 0.4687976837158203,
+      "loss_ce": 0.03593636304140091,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 31943688,
+      "step": 186
+    },
+    {
+      "epoch": 0.049187874005392256,
+      "grad_norm": 11.185610516547799,
+      "learning_rate": 5e-06,
+      "loss": 0.3628,
+      "num_input_tokens_seen": 32116040,
+      "step": 187
+    },
+    {
+      "epoch": 0.049187874005392256,
+      "loss": 0.39327770471572876,
+      "loss_ce": 0.03439096361398697,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 32116040,
+      "step": 187
+    },
+    {
+      "epoch": 0.04945091076477938,
+      "grad_norm": 11.155969289477222,
+      "learning_rate": 5e-06,
+      "loss": 0.2974,
+      "num_input_tokens_seen": 32288500,
+      "step": 188
+    },
+    {
+      "epoch": 0.04945091076477938,
+      "loss": 0.31713372468948364,
+      "loss_ce": 0.028681576251983643,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 32288500,
+      "step": 188
+    },
+    {
+      "epoch": 0.0497139475241665,
+      "grad_norm": 11.834615664464208,
+      "learning_rate": 5e-06,
+      "loss": 0.3369,
+      "num_input_tokens_seen": 32460772,
+      "step": 189
+    },
+    {
+      "epoch": 0.0497139475241665,
+      "loss": 0.3497753441333771,
+      "loss_ce": 0.02921871840953827,
+      "loss_iou": 0.484375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 32460772,
+      "step": 189
+    },
+    {
+      "epoch": 0.04997698428355363,
+      "grad_norm": 10.434886006122214,
+      "learning_rate": 5e-06,
+      "loss": 0.3387,
+      "num_input_tokens_seen": 32632848,
+      "step": 190
+    },
+    {
+      "epoch": 0.04997698428355363,
+      "loss": 0.36610347032546997,
+      "loss_ce": 0.029921812936663628,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 32632848,
+      "step": 190
+    },
+    {
+      "epoch": 0.05024002104294075,
+      "grad_norm": 9.800996766479896,
+      "learning_rate": 5e-06,
+      "loss": 0.3599,
+      "num_input_tokens_seen": 32805200,
+      "step": 191
+    },
+    {
+      "epoch": 0.05024002104294075,
+      "loss": 0.3484704792499542,
+      "loss_ce": 0.032674580812454224,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 32805200,
+      "step": 191
+    },
+    {
+      "epoch": 0.050503057802327875,
+      "grad_norm": 9.1652653754744,
+      "learning_rate": 5e-06,
+      "loss": 0.3086,
+      "num_input_tokens_seen": 32977420,
+      "step": 192
+    },
+    {
+      "epoch": 0.050503057802327875,
+      "loss": 0.3598284125328064,
+      "loss_ce": 0.027797123417258263,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 32977420,
+      "step": 192
+    },
+    {
+      "epoch": 0.050766094561715,
+      "grad_norm": 11.082061385505076,
+      "learning_rate": 5e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 33149792,
+      "step": 193
+    },
+    {
+      "epoch": 0.050766094561715,
+      "loss": 0.3082242012023926,
+      "loss_ce": 0.035763248801231384,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 33149792,
+      "step": 193
+    },
+    {
+      "epoch": 0.05102913132110212,
+      "grad_norm": 10.154212185692936,
+      "learning_rate": 5e-06,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 33321580,
+      "step": 194
+    },
+    {
+      "epoch": 0.05102913132110212,
+      "loss": 0.3002238869667053,
+      "loss_ce": 0.02568773366510868,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 33321580,
+      "step": 194
+    },
+    {
+      "epoch": 0.051292168080489246,
+      "grad_norm": 12.486303050942563,
+      "learning_rate": 5e-06,
+      "loss": 0.3875,
+      "num_input_tokens_seen": 33493632,
+      "step": 195
+    },
+    {
+      "epoch": 0.051292168080489246,
+      "loss": 0.35659241676330566,
+      "loss_ce": 0.030542613938450813,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 33493632,
+      "step": 195
+    },
+    {
+      "epoch": 0.05155520483987637,
+      "grad_norm": 13.41419397842875,
+      "learning_rate": 5e-06,
+      "loss": 0.3335,
+      "num_input_tokens_seen": 33665544,
+      "step": 196
+    },
+    {
+      "epoch": 0.05155520483987637,
+      "loss": 0.36018121242523193,
+      "loss_ce": 0.02436576411128044,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 33665544,
+      "step": 196
+    },
+    {
+      "epoch": 0.051818241599263494,
+      "grad_norm": 19.34380060773709,
+      "learning_rate": 5e-06,
+      "loss": 0.3472,
+      "num_input_tokens_seen": 33837492,
+      "step": 197
+    },
+    {
+      "epoch": 0.051818241599263494,
+      "loss": 0.37059885263442993,
+      "loss_ce": 0.024651601910591125,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 33837492,
+      "step": 197
+    },
+    {
+      "epoch": 0.052081278358650625,
+      "grad_norm": 12.626529129648052,
+      "learning_rate": 5e-06,
+      "loss": 0.3657,
+      "num_input_tokens_seen": 34009512,
+      "step": 198
+    },
+    {
+      "epoch": 0.052081278358650625,
+      "loss": 0.29348307847976685,
+      "loss_ce": 0.028102193027734756,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 34009512,
+      "step": 198
+    },
+    {
+      "epoch": 0.05234431511803775,
+      "grad_norm": 10.467016024509215,
+      "learning_rate": 5e-06,
+      "loss": 0.3464,
+      "num_input_tokens_seen": 34181600,
+      "step": 199
+    },
+    {
+      "epoch": 0.05234431511803775,
+      "loss": 0.30455371737480164,
+      "loss_ce": 0.022815439850091934,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 34181600,
+      "step": 199
+    },
+    {
+      "epoch": 0.05260735187742487,
+      "grad_norm": 13.453464722997529,
+      "learning_rate": 5e-06,
+      "loss": 0.3248,
+      "num_input_tokens_seen": 34352348,
+      "step": 200
+    },
+    {
+      "epoch": 0.05260735187742487,
+      "loss": 0.45307034254074097,
+      "loss_ce": 0.020819369703531265,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 34352348,
+      "step": 200
+    },
+    {
+      "epoch": 0.052870388636811996,
+      "grad_norm": 20.491451966511576,
+      "learning_rate": 5e-06,
+      "loss": 0.4317,
+      "num_input_tokens_seen": 34524600,
+      "step": 201
+    },
+    {
+      "epoch": 0.052870388636811996,
+      "loss": 0.41993263363838196,
+      "loss_ce": 0.024668946862220764,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 34524600,
+      "step": 201
+    },
+    {
+      "epoch": 0.05313342539619912,
+      "grad_norm": 17.643454599965338,
+      "learning_rate": 5e-06,
+      "loss": 0.4565,
+      "num_input_tokens_seen": 34696600,
+      "step": 202
+    },
+    {
+      "epoch": 0.05313342539619912,
+      "loss": 0.5106714963912964,
+      "loss_ce": 0.022634411230683327,
+      "loss_iou": 0.0888671875,
+      "loss_num": 0.09765625,
+      "loss_xval": 0.48828125,
+      "num_input_tokens_seen": 34696600,
+      "step": 202
+    },
+    {
+      "epoch": 0.053396462155586244,
+      "grad_norm": 13.599251013665107,
+      "learning_rate": 5e-06,
+      "loss": 0.3165,
+      "num_input_tokens_seen": 34868596,
+      "step": 203
+    },
+    {
+      "epoch": 0.053396462155586244,
+      "loss": 0.3383103609085083,
+      "loss_ce": 0.026054508984088898,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 34868596,
+      "step": 203
+    },
+    {
+      "epoch": 0.05365949891497337,
+      "grad_norm": 15.289442638831803,
+      "learning_rate": 5e-06,
+      "loss": 0.3217,
+      "num_input_tokens_seen": 35038040,
+      "step": 204
+    },
+    {
+      "epoch": 0.05365949891497337,
+      "loss": 0.306024968624115,
+      "loss_ce": 0.022089410573244095,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 35038040,
+      "step": 204
+    },
+    {
+      "epoch": 0.05392253567436049,
+      "grad_norm": 10.407634168683858,
+      "learning_rate": 5e-06,
+      "loss": 0.3039,
+      "num_input_tokens_seen": 35208292,
+      "step": 205
+    },
+    {
+      "epoch": 0.05392253567436049,
+      "loss": 0.35352200269699097,
+      "loss_ce": 0.026007331907749176,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 35208292,
+      "step": 205
+    },
+    {
+      "epoch": 0.054185572433747616,
+      "grad_norm": 11.313536084612993,
+      "learning_rate": 5e-06,
+      "loss": 0.3141,
+      "num_input_tokens_seen": 35377388,
+      "step": 206
+    },
+    {
+      "epoch": 0.054185572433747616,
+      "loss": 0.27950865030288696,
+      "loss_ce": 0.02438168227672577,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 35377388,
+      "step": 206
+    },
+    {
+      "epoch": 0.05444860919313474,
+      "grad_norm": 11.696274899578064,
+      "learning_rate": 5e-06,
+      "loss": 0.3578,
+      "num_input_tokens_seen": 35547812,
+      "step": 207
+    },
+    {
+      "epoch": 0.05444860919313474,
+      "loss": 0.3501763641834259,
+      "loss_ce": 0.0216851644217968,
+      "loss_iou": 0.53125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 35547812,
+      "step": 207
+    },
+    {
+      "epoch": 0.05471164595252186,
+      "grad_norm": 10.19535036839742,
+      "learning_rate": 5e-06,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 35720256,
+      "step": 208
+    },
+    {
+      "epoch": 0.05471164595252186,
+      "loss": 0.28618597984313965,
+      "loss_ce": 0.023246534168720245,
+      "loss_iou": 0.421875,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 35720256,
+      "step": 208
+    },
+    {
+      "epoch": 0.05497468271190899,
+      "grad_norm": 10.108238117808526,
+      "learning_rate": 5e-06,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 35892100,
+      "step": 209
+    },
+    {
+      "epoch": 0.05497468271190899,
+      "loss": 0.28424978256225586,
+      "loss_ce": 0.025460712611675262,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 35892100,
+      "step": 209
+    },
+    {
+      "epoch": 0.05523771947129611,
+      "grad_norm": 9.765870248571328,
+      "learning_rate": 5e-06,
+      "loss": 0.3151,
+      "num_input_tokens_seen": 36064296,
+      "step": 210
+    },
+    {
+      "epoch": 0.05523771947129611,
+      "loss": 0.30667591094970703,
+      "loss_ce": 0.02457140013575554,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 36064296,
+      "step": 210
+    },
+    {
+      "epoch": 0.055500756230683235,
+      "grad_norm": 8.667238391078952,
+      "learning_rate": 5e-06,
+      "loss": 0.3062,
+      "num_input_tokens_seen": 36236356,
+      "step": 211
+    },
+    {
+      "epoch": 0.055500756230683235,
+      "loss": 0.32360944151878357,
+      "loss_ce": 0.021241270005702972,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 36236356,
+      "step": 211
+    },
+    {
+      "epoch": 0.055763792990070365,
+      "grad_norm": 7.623662373992362,
+      "learning_rate": 5e-06,
+      "loss": 0.2441,
+      "num_input_tokens_seen": 36406508,
+      "step": 212
+    },
+    {
+      "epoch": 0.055763792990070365,
+      "loss": 0.2571391761302948,
+      "loss_ce": 0.022825222462415695,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 36406508,
+      "step": 212
+    },
+    {
+      "epoch": 0.05602682974945749,
+      "grad_norm": 9.522758823829298,
+      "learning_rate": 5e-06,
+      "loss": 0.263,
+      "num_input_tokens_seen": 36578384,
+      "step": 213
+    },
+    {
+      "epoch": 0.05602682974945749,
+      "loss": 0.20427684485912323,
+      "loss_ce": 0.022819336503744125,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 36578384,
+      "step": 213
+    },
+    {
+      "epoch": 0.05628986650884461,
+      "grad_norm": 14.400118985818274,
+      "learning_rate": 5e-06,
+      "loss": 0.293,
+      "num_input_tokens_seen": 36750552,
+      "step": 214
+    },
+    {
+      "epoch": 0.05628986650884461,
+      "loss": 0.29619529843330383,
+      "loss_ce": 0.021537089720368385,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 36750552,
+      "step": 214
+    },
+    {
+      "epoch": 0.05655290326823174,
+      "grad_norm": 16.206900517822692,
+      "learning_rate": 5e-06,
+      "loss": 0.3487,
+      "num_input_tokens_seen": 36922680,
+      "step": 215
+    },
+    {
+      "epoch": 0.05655290326823174,
+      "loss": 0.37616318464279175,
+      "loss_ce": 0.018497148528695107,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 36922680,
+      "step": 215
+    },
+    {
+      "epoch": 0.05681594002761886,
+      "grad_norm": 16.214529098872383,
+      "learning_rate": 5e-06,
+      "loss": 0.3418,
+      "num_input_tokens_seen": 37095056,
+      "step": 216
+    },
+    {
+      "epoch": 0.05681594002761886,
+      "loss": 0.37693583965301514,
+      "loss_ce": 0.018171211704611778,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 37095056,
+      "step": 216
+    },
+    {
+      "epoch": 0.057078976787005985,
+      "grad_norm": 11.604435704258858,
+      "learning_rate": 5e-06,
+      "loss": 0.3517,
+      "num_input_tokens_seen": 37267308,
+      "step": 217
+    },
+    {
+      "epoch": 0.057078976787005985,
+      "loss": 0.37936773896217346,
+      "loss_ce": 0.022922419011592865,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 37267308,
+      "step": 217
+    },
+    {
+      "epoch": 0.05734201354639311,
+      "grad_norm": 25.63677528330297,
+      "learning_rate": 5e-06,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 37434780,
+      "step": 218
+    },
+    {
+      "epoch": 0.05734201354639311,
+      "loss": 0.31407007575035095,
+      "loss_ce": 0.02378687635064125,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 37434780,
+      "step": 218
+    },
+    {
+      "epoch": 0.05760505030578023,
+      "grad_norm": 11.86583190083979,
+      "learning_rate": 5e-06,
+      "loss": 0.3346,
+      "num_input_tokens_seen": 37605076,
+      "step": 219
+    },
+    {
+      "epoch": 0.05760505030578023,
+      "loss": 0.31814104318618774,
+      "loss_ce": 0.02236468717455864,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 37605076,
+      "step": 219
+    },
+    {
+      "epoch": 0.057868087065167356,
+      "grad_norm": 14.940327825772103,
+      "learning_rate": 5e-06,
+      "loss": 0.3733,
+      "num_input_tokens_seen": 37777704,
+      "step": 220
+    },
+    {
+      "epoch": 0.057868087065167356,
+      "loss": 0.37386685609817505,
+      "loss_ce": 0.020107077434659004,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 37777704,
+      "step": 220
+    },
+    {
+      "epoch": 0.05813112382455448,
+      "grad_norm": 13.873681072861537,
+      "learning_rate": 5e-06,
+      "loss": 0.3326,
+      "num_input_tokens_seen": 37948064,
+      "step": 221
+    },
+    {
+      "epoch": 0.05813112382455448,
+      "loss": 0.3946155905723572,
+      "loss_ce": 0.019127311185002327,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 37948064,
+      "step": 221
+    },
+    {
+      "epoch": 0.058394160583941604,
+      "grad_norm": 13.253297702701516,
+      "learning_rate": 5e-06,
+      "loss": 0.3428,
+      "num_input_tokens_seen": 38118984,
+      "step": 222
+    },
+    {
+      "epoch": 0.058394160583941604,
+      "loss": 0.3087061643600464,
+      "loss_ce": 0.018300898373126984,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 38118984,
+      "step": 222
+    },
+    {
+      "epoch": 0.05865719734332873,
+      "grad_norm": 11.082820014401106,
+      "learning_rate": 5e-06,
+      "loss": 0.2927,
+      "num_input_tokens_seen": 38291092,
+      "step": 223
+    },
+    {
+      "epoch": 0.05865719734332873,
+      "loss": 0.2926178574562073,
+      "loss_ce": 0.017837589606642723,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 38291092,
+      "step": 223
+    },
+    {
+      "epoch": 0.05892023410271585,
+      "grad_norm": 13.606311852622206,
+      "learning_rate": 5e-06,
+      "loss": 0.2877,
+      "num_input_tokens_seen": 38461712,
+      "step": 224
+    },
+    {
+      "epoch": 0.05892023410271585,
+      "loss": 0.2914125323295593,
+      "loss_ce": 0.016632266342639923,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 38461712,
+      "step": 224
+    },
+    {
+      "epoch": 0.059183270862102975,
+      "grad_norm": 12.858292254216362,
+      "learning_rate": 5e-06,
+      "loss": 0.3576,
+      "num_input_tokens_seen": 38633936,
+      "step": 225
+    },
+    {
+      "epoch": 0.059183270862102975,
+      "loss": 0.36546608805656433,
+      "loss_ce": 0.015124273486435413,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 38633936,
+      "step": 225
+    },
+    {
+      "epoch": 0.059446307621490106,
+      "grad_norm": 15.356311927838101,
+      "learning_rate": 5e-06,
+      "loss": 0.2591,
+      "num_input_tokens_seen": 38802732,
+      "step": 226
+    },
+    {
+      "epoch": 0.059446307621490106,
+      "loss": 0.24970154464244843,
+      "loss_ce": 0.01520447339862585,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 38802732,
+      "step": 226
+    },
+    {
+      "epoch": 0.05970934438087723,
+      "grad_norm": 33.523690387446095,
+      "learning_rate": 5e-06,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 38975160,
+      "step": 227
+    },
+    {
+      "epoch": 0.05970934438087723,
+      "loss": 0.3422040641307831,
+      "loss_ce": 0.015910113230347633,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 38975160,
+      "step": 227
+    },
+    {
+      "epoch": 0.059972381140264354,
+      "grad_norm": 8.426822348321474,
+      "learning_rate": 5e-06,
+      "loss": 0.2888,
+      "num_input_tokens_seen": 39147352,
+      "step": 228
+    },
+    {
+      "epoch": 0.059972381140264354,
+      "loss": 0.29814714193344116,
+      "loss_ce": 0.014455747790634632,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 39147352,
+      "step": 228
+    },
+    {
+      "epoch": 0.06023541789965148,
+      "grad_norm": 9.384703696891318,
+      "learning_rate": 5e-06,
+      "loss": 0.2861,
+      "num_input_tokens_seen": 39319476,
+      "step": 229
+    },
+    {
+      "epoch": 0.06023541789965148,
+      "loss": 0.2668268382549286,
+      "loss_ce": 0.020000681281089783,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 39319476,
+      "step": 229
+    },
+    {
+      "epoch": 0.0604984546590386,
+      "grad_norm": 18.427622180632213,
+      "learning_rate": 5e-06,
+      "loss": 0.3917,
+      "num_input_tokens_seen": 39491664,
+      "step": 230
+    },
+    {
+      "epoch": 0.0604984546590386,
+      "loss": 0.390484094619751,
+      "loss_ce": 0.015484098345041275,
+      "loss_iou": 0.875,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 39491664,
+      "step": 230
+    },
+    {
+      "epoch": 0.060761491418425725,
+      "grad_norm": 10.977429362928614,
+      "learning_rate": 5e-06,
+      "loss": 0.3949,
+      "num_input_tokens_seen": 39663560,
+      "step": 231
+    },
+    {
+      "epoch": 0.060761491418425725,
+      "loss": 0.334339439868927,
+      "loss_ce": 0.015735914930701256,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 39663560,
+      "step": 231
+    },
+    {
+      "epoch": 0.06102452817781285,
+      "grad_norm": 7.478546882380691,
+      "learning_rate": 5e-06,
+      "loss": 0.2993,
+      "num_input_tokens_seen": 39835708,
+      "step": 232
+    },
+    {
+      "epoch": 0.06102452817781285,
+      "loss": 0.2514447867870331,
+      "loss_ce": 0.014872531406581402,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 39835708,
+      "step": 232
+    },
+    {
+      "epoch": 0.06128756493719997,
+      "grad_norm": 7.3196146291196635,
+      "learning_rate": 5e-06,
+      "loss": 0.2693,
+      "num_input_tokens_seen": 40008008,
+      "step": 233
+    },
+    {
+      "epoch": 0.06128756493719997,
+      "loss": 0.3245670795440674,
+      "loss_ce": 0.01768231764435768,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 40008008,
+      "step": 233
+    },
+    {
+      "epoch": 0.0615506016965871,
+      "grad_norm": 10.123913601017227,
+      "learning_rate": 5e-06,
+      "loss": 0.2834,
+      "num_input_tokens_seen": 40179816,
+      "step": 234
+    },
+    {
+      "epoch": 0.0615506016965871,
+      "loss": 0.2691105008125305,
+      "loss_ce": 0.01740150898694992,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 40179816,
+      "step": 234
+    },
+    {
+      "epoch": 0.06181363845597422,
+      "grad_norm": 16.840337823170596,
+      "learning_rate": 5e-06,
+      "loss": 0.3163,
+      "num_input_tokens_seen": 40352180,
+      "step": 235
+    },
+    {
+      "epoch": 0.06181363845597422,
+      "loss": 0.3705546259880066,
+      "loss_ce": 0.015330012887716293,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 40352180,
+      "step": 235
+    },
+    {
+      "epoch": 0.062076675215361345,
+      "grad_norm": 15.151708802383407,
+      "learning_rate": 5e-06,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 40524204,
+      "step": 236
+    },
+    {
+      "epoch": 0.062076675215361345,
+      "loss": 0.342043399810791,
+      "loss_ce": 0.015383241698145866,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 40524204,
+      "step": 236
+    },
+    {
+      "epoch": 0.06233971197474847,
+      "grad_norm": 23.279462740978527,
+      "learning_rate": 5e-06,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 40694412,
+      "step": 237
+    },
+    {
+      "epoch": 0.06233971197474847,
+      "loss": 0.3619577884674072,
+      "loss_ce": 0.014667754992842674,
+      "loss_iou": NaN,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 40694412,
+      "step": 237
+    },
+    {
+      "epoch": 0.0626027487341356,
+      "grad_norm": 7.854110724383197,
+      "learning_rate": 5e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 40866332,
+      "step": 238
+    },
+    {
+      "epoch": 0.0626027487341356,
+      "loss": 0.31817951798439026,
+      "loss_ce": 0.013125804252922535,
+      "loss_iou": 0.390625,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 40866332,
+      "step": 238
+    },
+    {
+      "epoch": 0.06286578549352272,
+      "grad_norm": 10.293296601258389,
+      "learning_rate": 5e-06,
+      "loss": 0.2924,
+      "num_input_tokens_seen": 41036884,
+      "step": 239
+    },
+    {
+      "epoch": 0.06286578549352272,
+      "loss": 0.20027026534080505,
+      "loss_ce": 0.012953377328813076,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 41036884,
+      "step": 239
+    },
+    {
+      "epoch": 0.06312882225290985,
+      "grad_norm": 27.954530904789568,
+      "learning_rate": 5e-06,
+      "loss": 0.3242,
+      "num_input_tokens_seen": 41209080,
+      "step": 240
+    },
+    {
+      "epoch": 0.06312882225290985,
+      "loss": 0.3325929641723633,
+      "loss_ce": 0.015088059939444065,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 41209080,
+      "step": 240
+    },
+    {
+      "epoch": 0.06339185901229696,
+      "grad_norm": 10.084557723509068,
+      "learning_rate": 5e-06,
+      "loss": 0.4017,
+      "num_input_tokens_seen": 41378712,
+      "step": 241
+    },
+    {
+      "epoch": 0.06339185901229696,
+      "loss": 0.42962974309921265,
+      "loss_ce": 0.01459068525582552,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 41378712,
+      "step": 241
+    },
+    {
+      "epoch": 0.0636548957716841,
+      "grad_norm": 7.845640861327211,
+      "learning_rate": 5e-06,
+      "loss": 0.3082,
+      "num_input_tokens_seen": 41550924,
+      "step": 242
+    },
+    {
+      "epoch": 0.0636548957716841,
+      "loss": 0.3617081940174103,
+      "loss_ce": 0.013441601768136024,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 41550924,
+      "step": 242
+    },
+    {
+      "epoch": 0.06391793253107121,
+      "grad_norm": 7.755635758991768,
+      "learning_rate": 5e-06,
+      "loss": 0.2863,
+      "num_input_tokens_seen": 41723448,
+      "step": 243
+    },
+    {
+      "epoch": 0.06391793253107121,
+      "loss": 0.31799250841140747,
+      "loss_ce": 0.01220636535435915,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 41723448,
+      "step": 243
+    },
+    {
+      "epoch": 0.06418096929045834,
+      "grad_norm": 10.028128094591828,
+      "learning_rate": 5e-06,
+      "loss": 0.3455,
+      "num_input_tokens_seen": 41895440,
+      "step": 244
+    },
+    {
+      "epoch": 0.06418096929045834,
+      "loss": 0.2664147913455963,
+      "loss_ce": 0.015926510095596313,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 41895440,
+      "step": 244
+    },
+    {
+      "epoch": 0.06444400604984546,
+      "grad_norm": 7.455815676512238,
+      "learning_rate": 5e-06,
+      "loss": 0.3158,
+      "num_input_tokens_seen": 42067812,
+      "step": 245
+    },
+    {
+      "epoch": 0.06444400604984546,
+      "loss": 0.315701961517334,
+      "loss_ce": 0.012784458696842194,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 42067812,
+      "step": 245
+    },
+    {
+      "epoch": 0.06470704280923259,
+      "grad_norm": 7.183484053450612,
+      "learning_rate": 5e-06,
+      "loss": 0.2873,
+      "num_input_tokens_seen": 42238248,
+      "step": 246
+    },
+    {
+      "epoch": 0.06470704280923259,
+      "loss": 0.408550500869751,
+      "loss_ce": 0.015972375869750977,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 42238248,
+      "step": 246
+    },
+    {
+      "epoch": 0.06497007956861972,
+      "grad_norm": 11.483582566544728,
+      "learning_rate": 5e-06,
+      "loss": 0.2422,
+      "num_input_tokens_seen": 42410540,
+      "step": 247
+    },
+    {
+      "epoch": 0.06497007956861972,
+      "loss": 0.24055451154708862,
+      "loss_ce": 0.012405097484588623,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 42410540,
+      "step": 247
+    },
+    {
+      "epoch": 0.06523311632800684,
+      "grad_norm": 9.093470574393532,
+      "learning_rate": 5e-06,
+      "loss": 0.2447,
+      "num_input_tokens_seen": 42582896,
+      "step": 248
+    },
+    {
+      "epoch": 0.06523311632800684,
+      "loss": 0.2380836009979248,
+      "loss_ce": 0.009445905685424805,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 42582896,
+      "step": 248
+    },
+    {
+      "epoch": 0.06549615308739397,
+      "grad_norm": 11.56824851448475,
+      "learning_rate": 5e-06,
+      "loss": 0.3024,
+      "num_input_tokens_seen": 42755248,
+      "step": 249
+    },
+    {
+      "epoch": 0.06549615308739397,
+      "loss": 0.3298885226249695,
+      "loss_ce": 0.012871915474534035,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 42755248,
+      "step": 249
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "grad_norm": 12.53262678976841,
+      "learning_rate": 5e-06,
+      "loss": 0.3279,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_websight_new_CIoU": 0.6498270332813263,
+      "eval_websight_new_GIoU": 0.6454348266124725,
+      "eval_websight_new_IoU": 0.6616384983062744,
+      "eval_websight_new_MAE_all": 0.05562719702720642,
+      "eval_websight_new_MAE_h": 0.049970587715506554,
+      "eval_websight_new_MAE_w": 0.07883360609412193,
+      "eval_websight_new_MAE_x": 0.06017959490418434,
+      "eval_websight_new_MAE_y": 0.03352500405162573,
+      "eval_websight_new_NUM_probability": 0.9025295078754425,
+      "eval_websight_new_inside_bbox": 0.9253472089767456,
+      "eval_websight_new_loss": 0.27164769172668457,
+      "eval_websight_new_loss_ce": 0.01081773592159152,
+      "eval_websight_new_loss_iou": 0.693359375,
+      "eval_websight_new_loss_num": 0.049407958984375,
+      "eval_websight_new_loss_xval": 0.24688720703125,
+      "eval_websight_new_runtime": 55.9304,
+      "eval_websight_new_samples_per_second": 0.894,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_seeclick_CIoU": 0.3509342074394226,
+      "eval_seeclick_GIoU": 0.3347575068473816,
+      "eval_seeclick_IoU": 0.3994097113609314,
+      "eval_seeclick_MAE_all": 0.0986204668879509,
+      "eval_seeclick_MAE_h": 0.0804138220846653,
+      "eval_seeclick_MAE_w": 0.13582541793584824,
+      "eval_seeclick_MAE_x": 0.12244484201073647,
+      "eval_seeclick_MAE_y": 0.05579778365790844,
+      "eval_seeclick_NUM_probability": 0.898758739233017,
+      "eval_seeclick_inside_bbox": 0.47727273404598236,
+      "eval_seeclick_loss": 0.4158971905708313,
+      "eval_seeclick_loss_ce": 0.027117961086332798,
+      "eval_seeclick_loss_iou": 0.66552734375,
+      "eval_seeclick_loss_num": 0.0771331787109375,
+      "eval_seeclick_loss_xval": 0.3856201171875,
+      "eval_seeclick_runtime": 77.1195,
+      "eval_seeclick_samples_per_second": 0.558,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_icons_CIoU": 0.533334881067276,
+      "eval_icons_GIoU": 0.525450587272644,
+      "eval_icons_IoU": 0.5554244518280029,
+      "eval_icons_MAE_all": 0.06896939501166344,
+      "eval_icons_MAE_h": 0.07008247822523117,
+      "eval_icons_MAE_w": 0.07993372157216072,
+      "eval_icons_MAE_x": 0.06088143028318882,
+      "eval_icons_MAE_y": 0.06497994437813759,
+      "eval_icons_NUM_probability": 0.9118345677852631,
+      "eval_icons_inside_bbox": 0.8038194477558136,
+      "eval_icons_loss": 0.238239586353302,
+      "eval_icons_loss_ce": 0.01238260930404067,
+      "eval_icons_loss_iou": 0.45098876953125,
+      "eval_icons_loss_num": 0.04291534423828125,
+      "eval_icons_loss_xval": 0.2144775390625,
+      "eval_icons_runtime": 89.0702,
+      "eval_icons_samples_per_second": 0.561,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_screenspot_CIoU": 0.40743913253148395,
+      "eval_screenspot_GIoU": 0.38262539108594257,
+      "eval_screenspot_IoU": 0.4510061542193095,
+      "eval_screenspot_MAE_all": 0.10948735972245534,
+      "eval_screenspot_MAE_h": 0.07818744828303655,
+      "eval_screenspot_MAE_w": 0.15290210396051407,
+      "eval_screenspot_MAE_x": 0.12599809964497885,
+      "eval_screenspot_MAE_y": 0.08086179196834564,
+      "eval_screenspot_NUM_probability": 0.9086714188257853,
+      "eval_screenspot_inside_bbox": 0.7354166706403097,
+      "eval_screenspot_loss": 0.9193825125694275,
+      "eval_screenspot_loss_ce": 0.48994149764378864,
+      "eval_screenspot_loss_iou": 0.5732421875,
+      "eval_screenspot_loss_num": 0.08390299479166667,
+      "eval_screenspot_loss_xval": 0.4192708333333333,
+      "eval_screenspot_runtime": 138.8534,
+      "eval_screenspot_samples_per_second": 0.641,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "loss": 0.8642065525054932,
+      "loss_ce": 0.45453858375549316,
+      "loss_iou": 0.53125,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06602222660616822,
+      "grad_norm": 16.162110660395133,
+      "learning_rate": 5e-06,
+      "loss": 0.3389,
+      "num_input_tokens_seen": 43099272,
+      "step": 251
+    },
+    {
+      "epoch": 0.06602222660616822,
+      "loss": 0.3324888348579407,
+      "loss_ce": 0.009490801021456718,
+      "loss_iou": 0.484375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 43099272,
+      "step": 251
+    },
+    {
+      "epoch": 0.06628526336555533,
+      "grad_norm": 15.788401372124632,
+      "learning_rate": 5e-06,
+      "loss": 0.3827,
+      "num_input_tokens_seen": 43271440,
+      "step": 252
+    },
+    {
+      "epoch": 0.06628526336555533,
+      "loss": 0.3184235990047455,
+      "loss_ce": 0.009219500236213207,
+      "loss_iou": 0.453125,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 43271440,
+      "step": 252
+    },
+    {
+      "epoch": 0.06654830012494246,
+      "grad_norm": 19.21368677855186,
+      "learning_rate": 5e-06,
+      "loss": 0.4031,
+      "num_input_tokens_seen": 43443704,
+      "step": 253
+    },
+    {
+      "epoch": 0.06654830012494246,
+      "loss": 0.3337656557559967,
+      "loss_ce": 0.009546896442770958,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 43443704,
+      "step": 253
+    },
+    {
+      "epoch": 0.06681133688432958,
+      "grad_norm": 9.350851106906616,
+      "learning_rate": 5e-06,
+      "loss": 0.2501,
+      "num_input_tokens_seen": 43615820,
+      "step": 254
+    },
+    {
+      "epoch": 0.06681133688432958,
+      "loss": 0.22743698954582214,
+      "loss_ce": 0.009236322715878487,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 43615820,
+      "step": 254
+    },
+    {
+      "epoch": 0.06707437364371671,
+      "grad_norm": 12.730744024793406,
+      "learning_rate": 5e-06,
+      "loss": 0.3033,
+      "num_input_tokens_seen": 43787664,
+      "step": 255
+    },
+    {
+      "epoch": 0.06707437364371671,
+      "loss": 0.3385705351829529,
+      "loss_ce": 0.009468959644436836,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 43787664,
+      "step": 255
+    },
+    {
+      "epoch": 0.06733741040310383,
+      "grad_norm": 20.602499715875325,
+      "learning_rate": 5e-06,
+      "loss": 0.335,
+      "num_input_tokens_seen": 43956324,
+      "step": 256
+    },
+    {
+      "epoch": 0.06733741040310383,
+      "loss": 0.33848732709884644,
+      "loss_ce": 0.012559601105749607,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 43956324,
+      "step": 256
+    },
+    {
+      "epoch": 0.06760044716249096,
+      "grad_norm": 11.274485245872459,
+      "learning_rate": 5e-06,
+      "loss": 0.241,
+      "num_input_tokens_seen": 44128740,
+      "step": 257
+    },
+    {
+      "epoch": 0.06760044716249096,
+      "loss": 0.24955862760543823,
+      "loss_ce": 0.011765659786760807,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 44128740,
+      "step": 257
+    },
+    {
+      "epoch": 0.06786348392187808,
+      "grad_norm": 12.119376554642344,
+      "learning_rate": 5e-06,
+      "loss": 0.359,
+      "num_input_tokens_seen": 44299372,
+      "step": 258
+    },
+    {
+      "epoch": 0.06786348392187808,
+      "loss": 0.26774948835372925,
+      "loss_ce": 0.01604049652814865,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 44299372,
+      "step": 258
+    },
+    {
+      "epoch": 0.0681265206812652,
+      "grad_norm": 15.925957923618926,
+      "learning_rate": 5e-06,
+      "loss": 0.3045,
+      "num_input_tokens_seen": 44469736,
+      "step": 259
+    },
+    {
+      "epoch": 0.0681265206812652,
+      "loss": 0.35633599758148193,
+      "loss_ce": 0.01502738706767559,
+      "loss_iou": 0.515625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 44469736,
+      "step": 259
+    },
+    {
+      "epoch": 0.06838955744065234,
+      "grad_norm": 13.189087784364153,
+      "learning_rate": 5e-06,
+      "loss": 0.328,
+      "num_input_tokens_seen": 44642020,
+      "step": 260
+    },
+    {
+      "epoch": 0.06838955744065234,
+      "loss": 0.4303101599216461,
+      "loss_ce": 0.01173105463385582,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.08349609375,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 44642020,
+      "step": 260
+    },
+    {
+      "epoch": 0.06865259420003945,
+      "grad_norm": 11.878594650017698,
+      "learning_rate": 5e-06,
+      "loss": 0.2752,
+      "num_input_tokens_seen": 44814452,
+      "step": 261
+    },
+    {
+      "epoch": 0.06865259420003945,
+      "loss": 0.28906819224357605,
+      "loss_ce": 0.0090388935059309,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 44814452,
+      "step": 261
+    },
+    {
+      "epoch": 0.06891563095942659,
+      "grad_norm": 8.62440603357029,
+      "learning_rate": 5e-06,
+      "loss": 0.2926,
+      "num_input_tokens_seen": 44986656,
+      "step": 262
+    },
+    {
+      "epoch": 0.06891563095942659,
+      "loss": 0.24431782960891724,
+      "loss_ce": 0.008111760020256042,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 44986656,
+      "step": 262
+    },
+    {
+      "epoch": 0.0691786677188137,
+      "grad_norm": 7.7035777517167645,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 45158540,
+      "step": 263
+    },
+    {
+      "epoch": 0.0691786677188137,
+      "loss": 0.23680636286735535,
+      "loss_ce": 0.013417699374258518,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 45158540,
+      "step": 263
+    },
+    {
+      "epoch": 0.06944170447820083,
+      "grad_norm": 11.11690642897216,
+      "learning_rate": 5e-06,
+      "loss": 0.2503,
+      "num_input_tokens_seen": 45330356,
+      "step": 264
+    },
+    {
+      "epoch": 0.06944170447820083,
+      "loss": 0.25701650977134705,
+      "loss_ce": 0.01019032672047615,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 45330356,
+      "step": 264
+    },
+    {
+      "epoch": 0.06970474123758795,
+      "grad_norm": 16.27774203292456,
+      "learning_rate": 5e-06,
+      "loss": 0.327,
+      "num_input_tokens_seen": 45502460,
+      "step": 265
+    },
+    {
+      "epoch": 0.06970474123758795,
+      "loss": 0.3339051902294159,
+      "loss_ce": 0.011639568954706192,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 45502460,
+      "step": 265
+    },
+    {
+      "epoch": 0.06996777799697508,
+      "grad_norm": 11.740826510166434,
+      "learning_rate": 5e-06,
+      "loss": 0.3816,
+      "num_input_tokens_seen": 45674688,
+      "step": 266
+    },
+    {
+      "epoch": 0.06996777799697508,
+      "loss": 0.30319204926490784,
+      "loss_ce": 0.016082679852843285,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 45674688,
+      "step": 266
+    },
+    {
+      "epoch": 0.0702308147563622,
+      "grad_norm": 7.300788157576167,
+      "learning_rate": 5e-06,
+      "loss": 0.29,
+      "num_input_tokens_seen": 45847124,
+      "step": 267
+    },
+    {
+      "epoch": 0.0702308147563622,
+      "loss": 0.3247534930706024,
+      "loss_ce": 0.009140691719949245,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 45847124,
+      "step": 267
+    },
+    {
+      "epoch": 0.07049385151574933,
+      "grad_norm": 7.309781555609638,
+      "learning_rate": 5e-06,
+      "loss": 0.2756,
+      "num_input_tokens_seen": 46019340,
+      "step": 268
+    },
+    {
+      "epoch": 0.07049385151574933,
+      "loss": 0.2533281445503235,
+      "loss_ce": 0.008699209429323673,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 46019340,
+      "step": 268
+    },
+    {
+      "epoch": 0.07075688827513645,
+      "grad_norm": 13.089258421868523,
+      "learning_rate": 5e-06,
+      "loss": 0.2701,
+      "num_input_tokens_seen": 46189784,
+      "step": 269
+    },
+    {
+      "epoch": 0.07075688827513645,
+      "loss": 0.2624807357788086,
+      "loss_ce": 0.01223657839000225,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 46189784,
+      "step": 269
+    },
+    {
+      "epoch": 0.07101992503452358,
+      "grad_norm": 6.658182143864948,
+      "learning_rate": 5e-06,
+      "loss": 0.2118,
+      "num_input_tokens_seen": 46361728,
+      "step": 270
+    },
+    {
+      "epoch": 0.07101992503452358,
+      "loss": 0.2072766274213791,
+      "loss_ce": 0.007569607347249985,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 46361728,
+      "step": 270
+    },
+    {
+      "epoch": 0.07128296179391069,
+      "grad_norm": 9.079661259155806,
+      "learning_rate": 5e-06,
+      "loss": 0.2305,
+      "num_input_tokens_seen": 46534100,
+      "step": 271
+    },
+    {
+      "epoch": 0.07128296179391069,
+      "loss": 0.21106520295143127,
+      "loss_ce": 0.008977807126939297,
+      "loss_iou": 0.609375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 46534100,
+      "step": 271
+    },
+    {
+      "epoch": 0.07154599855329782,
+      "grad_norm": 9.513732560495592,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "num_input_tokens_seen": 46704648,
+      "step": 272
+    },
+    {
+      "epoch": 0.07154599855329782,
+      "loss": 0.2694496214389801,
+      "loss_ce": 0.007120497524738312,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 46704648,
+      "step": 272
+    },
+    {
+      "epoch": 0.07180903531268495,
+      "grad_norm": 15.714221513863984,
+      "learning_rate": 5e-06,
+      "loss": 0.3078,
+      "num_input_tokens_seen": 46876784,
+      "step": 273
+    },
+    {
+      "epoch": 0.07180903531268495,
+      "loss": 0.39289578795433044,
+      "loss_ce": 0.006909476593136787,
+      "loss_iou": 0.5,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 46876784,
+      "step": 273
+    },
+    {
+      "epoch": 0.07207207207207207,
+      "grad_norm": 8.093127002316251,
+      "learning_rate": 5e-06,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 47048876,
+      "step": 274
+    },
+    {
+      "epoch": 0.07207207207207207,
+      "loss": 0.25725314021110535,
+      "loss_ce": 0.011525607667863369,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 47048876,
+      "step": 274
+    },
+    {
+      "epoch": 0.0723351088314592,
+      "grad_norm": 8.569390693529225,
+      "learning_rate": 5e-06,
+      "loss": 0.297,
+      "num_input_tokens_seen": 47221068,
+      "step": 275
+    },
+    {
+      "epoch": 0.0723351088314592,
+      "loss": 0.2686833143234253,
+      "loss_ce": 0.006720416247844696,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 47221068,
+      "step": 275
+    },
+    {
+      "epoch": 0.07259814559084632,
+      "grad_norm": 10.067643264055917,
+      "learning_rate": 5e-06,
+      "loss": 0.2437,
+      "num_input_tokens_seen": 47391436,
+      "step": 276
+    },
+    {
+      "epoch": 0.07259814559084632,
+      "loss": 0.2530132830142975,
+      "loss_ce": 0.007896100170910358,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 47391436,
+      "step": 276
+    },
+    {
+      "epoch": 0.07286118235023345,
+      "grad_norm": 13.479763232881819,
+      "learning_rate": 5e-06,
+      "loss": 0.3643,
+      "num_input_tokens_seen": 47563512,
+      "step": 277
+    },
+    {
+      "epoch": 0.07286118235023345,
+      "loss": 0.2713325619697571,
+      "loss_ce": 0.013642151840031147,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 47563512,
+      "step": 277
+    },
+    {
+      "epoch": 0.07312421910962057,
+      "grad_norm": 16.755961470396134,
+      "learning_rate": 5e-06,
+      "loss": 0.3279,
+      "num_input_tokens_seen": 47735372,
+      "step": 278
+    },
+    {
+      "epoch": 0.07312421910962057,
+      "loss": 0.42755797505378723,
+      "loss_ce": 0.008124373853206635,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 47735372,
+      "step": 278
+    },
+    {
+      "epoch": 0.0733872558690077,
+      "grad_norm": 13.89589851066108,
+      "learning_rate": 5e-06,
+      "loss": 0.3883,
+      "num_input_tokens_seen": 47907704,
+      "step": 279
+    },
+    {
+      "epoch": 0.0733872558690077,
+      "loss": 0.41463613510131836,
+      "loss_ce": 0.013879301026463509,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 47907704,
+      "step": 279
+    },
+    {
+      "epoch": 0.07365029262839481,
+      "grad_norm": 13.52575149586993,
+      "learning_rate": 5e-06,
+      "loss": 0.2798,
+      "num_input_tokens_seen": 48079692,
+      "step": 280
+    },
+    {
+      "epoch": 0.07365029262839481,
+      "loss": 0.2906692624092102,
+      "loss_ce": 0.007099920883774757,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 48079692,
+      "step": 280
+    },
+    {
+      "epoch": 0.07391332938778195,
+      "grad_norm": 11.637653668969445,
+      "learning_rate": 5e-06,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 48251572,
+      "step": 281
+    },
+    {
+      "epoch": 0.07391332938778195,
+      "loss": 0.2844490110874176,
+      "loss_ce": 0.010157023556530476,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 48251572,
+      "step": 281
+    },
+    {
+      "epoch": 0.07417636614716906,
+      "grad_norm": 9.21180215619434,
+      "learning_rate": 5e-06,
+      "loss": 0.2978,
+      "num_input_tokens_seen": 48423560,
+      "step": 282
+    },
+    {
+      "epoch": 0.07417636614716906,
+      "loss": 0.32979628443717957,
+      "loss_ce": 0.012047262862324715,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 48423560,
+      "step": 282
+    },
+    {
+      "epoch": 0.07443940290655619,
+      "grad_norm": 7.167038204924764,
+      "learning_rate": 5e-06,
+      "loss": 0.256,
+      "num_input_tokens_seen": 48595680,
+      "step": 283
+    },
+    {
+      "epoch": 0.07443940290655619,
+      "loss": 0.25096797943115234,
+      "loss_ce": 0.007925992831587791,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 48595680,
+      "step": 283
+    },
+    {
+      "epoch": 0.07470243966594331,
+      "grad_norm": 11.715910802137108,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "num_input_tokens_seen": 48766076,
+      "step": 284
+    },
+    {
+      "epoch": 0.07470243966594331,
+      "loss": 0.19452951848506927,
+      "loss_ce": 0.005930891260504723,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 48766076,
+      "step": 284
+    },
+    {
+      "epoch": 0.07496547642533044,
+      "grad_norm": 11.971840696280033,
+      "learning_rate": 5e-06,
+      "loss": 0.3413,
+      "num_input_tokens_seen": 48937948,
+      "step": 285
+    },
+    {
+      "epoch": 0.07496547642533044,
+      "loss": 0.349312424659729,
+      "loss_ce": 0.00910246279090643,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 48937948,
+      "step": 285
+    },
+    {
+      "epoch": 0.07522851318471756,
+      "grad_norm": 9.505447479710712,
+      "learning_rate": 5e-06,
+      "loss": 0.2998,
+      "num_input_tokens_seen": 49110048,
+      "step": 286
+    },
+    {
+      "epoch": 0.07522851318471756,
+      "loss": 0.3620803654193878,
+      "loss_ce": 0.0063674794510006905,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 49110048,
+      "step": 286
+    },
+    {
+      "epoch": 0.07549154994410469,
+      "grad_norm": 9.524937593510838,
+      "learning_rate": 5e-06,
+      "loss": 0.2473,
+      "num_input_tokens_seen": 49281924,
+      "step": 287
+    },
+    {
+      "epoch": 0.07549154994410469,
+      "loss": 0.21198034286499023,
+      "loss_ce": 0.0065360115841031075,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 49281924,
+      "step": 287
+    },
+    {
+      "epoch": 0.07575458670349182,
+      "grad_norm": 9.713489351185899,
+      "learning_rate": 5e-06,
+      "loss": 0.2911,
+      "num_input_tokens_seen": 49454288,
+      "step": 288
+    },
+    {
+      "epoch": 0.07575458670349182,
+      "loss": 0.35456106066703796,
+      "loss_ce": 0.01032277476042509,
+      "loss_iou": 0.4375,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 49454288,
+      "step": 288
+    },
+    {
+      "epoch": 0.07601762346287894,
+      "grad_norm": 8.377848047343468,
+      "learning_rate": 5e-06,
+      "loss": 0.2711,
+      "num_input_tokens_seen": 49626816,
+      "step": 289
+    },
+    {
+      "epoch": 0.07601762346287894,
+      "loss": 0.2953585684299469,
+      "loss_ce": 0.005685708485543728,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 49626816,
+      "step": 289
+    },
+    {
+      "epoch": 0.07628066022226607,
+      "grad_norm": 10.429005548611427,
+      "learning_rate": 5e-06,
+      "loss": 0.2446,
+      "num_input_tokens_seen": 49798788,
+      "step": 290
+    },
+    {
+      "epoch": 0.07628066022226607,
+      "loss": 0.19721034169197083,
+      "loss_ce": 0.007268924731761217,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 49798788,
+      "step": 290
+    },
+    {
+      "epoch": 0.07654369698165318,
+      "grad_norm": 13.641317737771786,
+      "learning_rate": 5e-06,
+      "loss": 0.3162,
+      "num_input_tokens_seen": 49970924,
+      "step": 291
+    },
+    {
+      "epoch": 0.07654369698165318,
+      "loss": 0.3052144944667816,
+      "loss_ce": 0.00528772734105587,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 49970924,
+      "step": 291
+    },
+    {
+      "epoch": 0.07680673374104031,
+      "grad_norm": 13.717892281601975,
+      "learning_rate": 5e-06,
+      "loss": 0.2925,
+      "num_input_tokens_seen": 50143128,
+      "step": 292
+    },
+    {
+      "epoch": 0.07680673374104031,
+      "loss": 0.3288165032863617,
+      "loss_ce": 0.006795027758926153,
+      "loss_iou": 0.34375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 50143128,
+      "step": 292
+    },
+    {
+      "epoch": 0.07706977050042743,
+      "grad_norm": 9.087941964788197,
+      "learning_rate": 5e-06,
+      "loss": 0.2844,
+      "num_input_tokens_seen": 50315356,
+      "step": 293
+    },
+    {
+      "epoch": 0.07706977050042743,
+      "loss": 0.30062055587768555,
+      "loss_ce": 0.006186962127685547,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 50315356,
+      "step": 293
+    },
+    {
+      "epoch": 0.07733280725981456,
+      "grad_norm": 8.950504335091761,
+      "learning_rate": 5e-06,
+      "loss": 0.2474,
+      "num_input_tokens_seen": 50487452,
+      "step": 294
+    },
+    {
+      "epoch": 0.07733280725981456,
+      "loss": 0.20027770102024078,
+      "loss_ce": 0.007284537889063358,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 50487452,
+      "step": 294
+    },
+    {
+      "epoch": 0.07759584401920168,
+      "grad_norm": 12.276590112874837,
+      "learning_rate": 5e-06,
+      "loss": 0.2962,
+      "num_input_tokens_seen": 50659340,
+      "step": 295
+    },
+    {
+      "epoch": 0.07759584401920168,
+      "loss": 0.3079564869403839,
+      "loss_ce": 0.008151799440383911,
+      "loss_iou": 0.5625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 50659340,
+      "step": 295
+    },
+    {
+      "epoch": 0.07785888077858881,
+      "grad_norm": 10.920612112253043,
+      "learning_rate": 5e-06,
+      "loss": 0.2779,
+      "num_input_tokens_seen": 50831668,
+      "step": 296
+    },
+    {
+      "epoch": 0.07785888077858881,
+      "loss": 0.3526954650878906,
+      "loss_ce": 0.01395035907626152,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 50831668,
+      "step": 296
+    },
+    {
+      "epoch": 0.07812191753797593,
+      "grad_norm": 14.39030370540137,
+      "learning_rate": 5e-06,
+      "loss": 0.317,
+      "num_input_tokens_seen": 51004036,
+      "step": 297
+    },
+    {
+      "epoch": 0.07812191753797593,
+      "loss": 0.3473885655403137,
+      "loss_ce": 0.01047449093312025,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 51004036,
+      "step": 297
+    },
+    {
+      "epoch": 0.07838495429736306,
+      "grad_norm": 10.790411077010845,
+      "learning_rate": 5e-06,
+      "loss": 0.2641,
+      "num_input_tokens_seen": 51173952,
+      "step": 298
+    },
+    {
+      "epoch": 0.07838495429736306,
+      "loss": 0.24334201216697693,
+      "loss_ce": 0.0049387002363801,
+      "loss_iou": 0.453125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 51173952,
+      "step": 298
+    },
+    {
+      "epoch": 0.07864799105675017,
+      "grad_norm": 12.163313302369062,
+      "learning_rate": 5e-06,
+      "loss": 0.277,
+      "num_input_tokens_seen": 51346208,
+      "step": 299
+    },
+    {
+      "epoch": 0.07864799105675017,
+      "loss": 0.33479589223861694,
+      "loss_ce": 0.005816406104713678,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 51346208,
+      "step": 299
+    },
+    {
+      "epoch": 0.0789110278161373,
+      "grad_norm": 13.775522127007955,
+      "learning_rate": 5e-06,
+      "loss": 0.3084,
+      "num_input_tokens_seen": 51516552,
+      "step": 300
+    },
+    {
+      "epoch": 0.0789110278161373,
+      "loss": 0.2704133987426758,
+      "loss_ce": 0.004422190133482218,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 51516552,
+      "step": 300
+    },
+    {
+      "epoch": 0.07917406457552444,
+      "grad_norm": 11.14460650886844,
+      "learning_rate": 5e-06,
+      "loss": 0.2851,
+      "num_input_tokens_seen": 51688716,
+      "step": 301
+    },
+    {
+      "epoch": 0.07917406457552444,
+      "loss": 0.2599673867225647,
+      "loss_ce": 0.008197364397346973,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 51688716,
+      "step": 301
+    },
+    {
+      "epoch": 0.07943710133491155,
+      "grad_norm": 14.089462259095386,
+      "learning_rate": 5e-06,
+      "loss": 0.3106,
+      "num_input_tokens_seen": 51861284,
+      "step": 302
+    },
+    {
+      "epoch": 0.07943710133491155,
+      "loss": 0.25006186962127686,
+      "loss_ce": 0.004822590388357639,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 51861284,
+      "step": 302
+    },
+    {
+      "epoch": 0.07970013809429868,
+      "grad_norm": 11.560263696218914,
+      "learning_rate": 5e-06,
+      "loss": 0.2256,
+      "num_input_tokens_seen": 52033296,
+      "step": 303
+    },
+    {
+      "epoch": 0.07970013809429868,
+      "loss": 0.2260829657316208,
+      "loss_ce": 0.005379845853894949,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 52033296,
+      "step": 303
+    },
+    {
+      "epoch": 0.0799631748536858,
+      "grad_norm": 11.109765340752924,
+      "learning_rate": 5e-06,
+      "loss": 0.2728,
+      "num_input_tokens_seen": 52205488,
+      "step": 304
+    },
+    {
+      "epoch": 0.0799631748536858,
+      "loss": 0.26471367478370667,
+      "loss_ce": 0.0049480353482067585,
+      "loss_iou": 0.296875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 52205488,
+      "step": 304
+    },
+    {
+      "epoch": 0.08022621161307293,
+      "grad_norm": 9.561565409079957,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 52377756,
+      "step": 305
+    },
+    {
+      "epoch": 0.08022621161307293,
+      "loss": 0.18537692725658417,
+      "loss_ce": 0.0050790803506970406,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 52377756,
+      "step": 305
+    },
+    {
+      "epoch": 0.08048924837246005,
+      "grad_norm": 11.190229795777121,
+      "learning_rate": 5e-06,
+      "loss": 0.2198,
+      "num_input_tokens_seen": 52549784,
+      "step": 306
+    },
+    {
+      "epoch": 0.08048924837246005,
+      "loss": 0.1936880648136139,
+      "loss_ce": 0.009972257539629936,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 52549784,
+      "step": 306
+    },
+    {
+      "epoch": 0.08075228513184718,
+      "grad_norm": 9.790391883995309,
+      "learning_rate": 5e-06,
+      "loss": 0.2383,
+      "num_input_tokens_seen": 52719856,
+      "step": 307
+    },
+    {
+      "epoch": 0.08075228513184718,
+      "loss": 0.16280654072761536,
+      "loss_ce": 0.007838279940187931,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 52719856,
+      "step": 307
+    },
+    {
+      "epoch": 0.0810153218912343,
+      "grad_norm": 9.839075915450485,
+      "learning_rate": 5e-06,
+      "loss": 0.2743,
+      "num_input_tokens_seen": 52892084,
+      "step": 308
+    },
+    {
+      "epoch": 0.0810153218912343,
+      "loss": 0.15204857289791107,
+      "loss_ce": 0.004587644245475531,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 52892084,
+      "step": 308
+    },
+    {
+      "epoch": 0.08127835865062143,
+      "grad_norm": 12.7104246981192,
+      "learning_rate": 5e-06,
+      "loss": 0.2624,
+      "num_input_tokens_seen": 53064144,
+      "step": 309
+    },
+    {
+      "epoch": 0.08127835865062143,
+      "loss": 0.25361326336860657,
+      "loss_ce": 0.009228497743606567,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 53064144,
+      "step": 309
+    },
+    {
+      "epoch": 0.08154139541000854,
+      "grad_norm": 9.929561281282899,
+      "learning_rate": 5e-06,
+      "loss": 0.3246,
+      "num_input_tokens_seen": 53236468,
+      "step": 310
+    },
+    {
+      "epoch": 0.08154139541000854,
+      "loss": 0.39354461431503296,
+      "loss_ce": 0.005116886459290981,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 53236468,
+      "step": 310
+    },
+    {
+      "epoch": 0.08180443216939567,
+      "grad_norm": 7.973294747772443,
+      "learning_rate": 5e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 53408452,
+      "step": 311
+    },
+    {
+      "epoch": 0.08180443216939567,
+      "loss": 0.2905076742172241,
+      "loss_ce": 0.005351424217224121,
+      "loss_iou": NaN,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 53408452,
+      "step": 311
+    },
+    {
+      "epoch": 0.08206746892878279,
+      "grad_norm": 7.623528366567178,
+      "learning_rate": 5e-06,
+      "loss": 0.2153,
+      "num_input_tokens_seen": 53580848,
+      "step": 312
+    },
+    {
+      "epoch": 0.08206746892878279,
+      "loss": 0.19026613235473633,
+      "loss_ce": 0.005817888304591179,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 53580848,
+      "step": 312
+    },
+    {
+      "epoch": 0.08233050568816992,
+      "grad_norm": 8.591099989866738,
+      "learning_rate": 5e-06,
+      "loss": 0.2192,
+      "num_input_tokens_seen": 53751420,
+      "step": 313
+    },
+    {
+      "epoch": 0.08233050568816992,
+      "loss": 0.16798478364944458,
+      "loss_ce": 0.005326096434146166,
+      "loss_iou": 0.515625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 53751420,
+      "step": 313
+    },
+    {
+      "epoch": 0.08259354244755705,
+      "grad_norm": 8.168930137137453,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 53923692,
+      "step": 314
+    },
+    {
+      "epoch": 0.08259354244755705,
+      "loss": 0.1808127462863922,
+      "loss_ce": 0.006557377986609936,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 53923692,
+      "step": 314
+    },
+    {
+      "epoch": 0.08285657920694417,
+      "grad_norm": 12.69374602286009,
+      "learning_rate": 5e-06,
+      "loss": 0.255,
+      "num_input_tokens_seen": 54095596,
+      "step": 315
+    },
+    {
+      "epoch": 0.08285657920694417,
+      "loss": 0.2128537893295288,
+      "loss_ce": 0.00447976216673851,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 54095596,
+      "step": 315
+    },
+    {
+      "epoch": 0.0831196159663313,
+      "grad_norm": 15.86710980709006,
+      "learning_rate": 5e-06,
+      "loss": 0.3231,
+      "num_input_tokens_seen": 54267660,
+      "step": 316
+    },
+    {
+      "epoch": 0.0831196159663313,
+      "loss": 0.29302194714546204,
+      "loss_ce": 0.004081526771187782,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 54267660,
+      "step": 316
+    },
+    {
+      "epoch": 0.08338265272571842,
+      "grad_norm": 15.523568167979095,
+      "learning_rate": 5e-06,
+      "loss": 0.2565,
+      "num_input_tokens_seen": 54439776,
+      "step": 317
+    },
+    {
+      "epoch": 0.08338265272571842,
+      "loss": 0.45379549264907837,
+      "loss_ce": 0.00494293263182044,
+      "loss_iou": 0.421875,
+      "loss_num": 0.08984375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 54439776,
+      "step": 317
+    },
+    {
+      "epoch": 0.08364568948510555,
+      "grad_norm": 8.655289602311944,
+      "learning_rate": 5e-06,
+      "loss": 0.2346,
+      "num_input_tokens_seen": 54612224,
+      "step": 318
+    },
+    {
+      "epoch": 0.08364568948510555,
+      "loss": 0.22014674544334412,
+      "loss_ce": 0.008476818911731243,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 54612224,
+      "step": 318
+    },
+    {
+      "epoch": 0.08390872624449267,
+      "grad_norm": 10.678383550359573,
+      "learning_rate": 5e-06,
+      "loss": 0.2569,
+      "num_input_tokens_seen": 54782956,
+      "step": 319
+    },
+    {
+      "epoch": 0.08390872624449267,
+      "loss": 0.2195930778980255,
+      "loss_ce": 0.004505200777202845,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 54782956,
+      "step": 319
+    },
+    {
+      "epoch": 0.0841717630038798,
+      "grad_norm": 9.930533969178247,
+      "learning_rate": 5e-06,
+      "loss": 0.2305,
+      "num_input_tokens_seen": 54955104,
+      "step": 320
+    },
+    {
+      "epoch": 0.0841717630038798,
+      "loss": 0.24099573493003845,
+      "loss_ce": 0.008909543044865131,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 54955104,
+      "step": 320
+    },
+    {
+      "epoch": 0.08443479976326691,
+      "grad_norm": 12.929508923145086,
+      "learning_rate": 5e-06,
+      "loss": 0.2998,
+      "num_input_tokens_seen": 55127092,
+      "step": 321
+    },
+    {
+      "epoch": 0.08443479976326691,
+      "loss": 0.3190174698829651,
+      "loss_ce": 0.008958851918578148,
+      "loss_iou": 0.53125,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 55127092,
+      "step": 321
+    },
+    {
+      "epoch": 0.08469783652265404,
+      "grad_norm": 8.336572461824584,
+      "learning_rate": 5e-06,
+      "loss": 0.2781,
+      "num_input_tokens_seen": 55299364,
+      "step": 322
+    },
+    {
+      "epoch": 0.08469783652265404,
+      "loss": 0.21446217596530914,
+      "loss_ce": 0.005966084077954292,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 55299364,
+      "step": 322
+    },
+    {
+      "epoch": 0.08496087328204116,
+      "grad_norm": 9.09772392048227,
+      "learning_rate": 5e-06,
+      "loss": 0.2929,
+      "num_input_tokens_seen": 55468252,
+      "step": 323
+    },
+    {
+      "epoch": 0.08496087328204116,
+      "loss": 0.28219783306121826,
+      "loss_ce": 0.007356537505984306,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 55468252,
+      "step": 323
+    },
+    {
+      "epoch": 0.08522391004142829,
+      "grad_norm": 9.926795422915726,
+      "learning_rate": 5e-06,
+      "loss": 0.2412,
+      "num_input_tokens_seen": 55640432,
+      "step": 324
+    },
+    {
+      "epoch": 0.08522391004142829,
+      "loss": 0.2594638466835022,
+      "loss_ce": 0.006412106566131115,
+      "loss_iou": 0.59375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 55640432,
+      "step": 324
+    },
+    {
+      "epoch": 0.08548694680081541,
+      "grad_norm": 13.161460539721551,
+      "learning_rate": 5e-06,
+      "loss": 0.3056,
+      "num_input_tokens_seen": 55812284,
+      "step": 325
+    },
+    {
+      "epoch": 0.08548694680081541,
+      "loss": 0.30750948190689087,
+      "loss_ce": 0.003920634277164936,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 55812284,
+      "step": 325
+    },
+    {
+      "epoch": 0.08574998356020254,
+      "grad_norm": 12.20729496362805,
+      "learning_rate": 5e-06,
+      "loss": 0.2492,
+      "num_input_tokens_seen": 55982812,
+      "step": 326
+    },
+    {
+      "epoch": 0.08574998356020254,
+      "loss": 0.25009769201278687,
+      "loss_ce": 0.003637729212641716,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 55982812,
+      "step": 326
+    },
+    {
+      "epoch": 0.08601302031958966,
+      "grad_norm": 8.486549179327342,
+      "learning_rate": 5e-06,
+      "loss": 0.2612,
+      "num_input_tokens_seen": 56154880,
+      "step": 327
+    },
+    {
+      "epoch": 0.08601302031958966,
+      "loss": 0.24035008251667023,
+      "loss_ce": 0.006341288331896067,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 56154880,
+      "step": 327
+    },
+    {
+      "epoch": 0.08627605707897679,
+      "grad_norm": 12.368715754730403,
+      "learning_rate": 5e-06,
+      "loss": 0.2435,
+      "num_input_tokens_seen": 56327104,
+      "step": 328
+    },
+    {
+      "epoch": 0.08627605707897679,
+      "loss": 0.2434852123260498,
+      "loss_ce": 0.003983248956501484,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 56327104,
+      "step": 328
+    },
+    {
+      "epoch": 0.08653909383836392,
+      "grad_norm": 16.286201391843196,
+      "learning_rate": 5e-06,
+      "loss": 0.3578,
+      "num_input_tokens_seen": 56498760,
+      "step": 329
+    },
+    {
+      "epoch": 0.08653909383836392,
+      "loss": 0.3906838297843933,
+      "loss_ce": 0.00481957383453846,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 56498760,
+      "step": 329
+    },
+    {
+      "epoch": 0.08680213059775103,
+      "grad_norm": 9.709126690100668,
+      "learning_rate": 5e-06,
+      "loss": 0.2709,
+      "num_input_tokens_seen": 56669204,
+      "step": 330
+    },
+    {
+      "epoch": 0.08680213059775103,
+      "loss": 0.2887915074825287,
+      "loss_ce": 0.006198735907673836,
+      "loss_iou": 0.359375,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 56669204,
+      "step": 330
+    },
+    {
+      "epoch": 0.08706516735713817,
+      "grad_norm": 9.72107226431463,
+      "learning_rate": 5e-06,
+      "loss": 0.2546,
+      "num_input_tokens_seen": 56841324,
+      "step": 331
+    },
+    {
+      "epoch": 0.08706516735713817,
+      "loss": 0.17188376188278198,
+      "loss_ce": 0.005013656336814165,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 56841324,
+      "step": 331
+    },
+    {
+      "epoch": 0.08732820411652528,
+      "grad_norm": 14.410704147758572,
+      "learning_rate": 5e-06,
+      "loss": 0.2888,
+      "num_input_tokens_seen": 57013308,
+      "step": 332
+    },
+    {
+      "epoch": 0.08732820411652528,
+      "loss": 0.2731274664402008,
+      "loss_ce": 0.007868663407862186,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 57013308,
+      "step": 332
+    },
+    {
+      "epoch": 0.08759124087591241,
+      "grad_norm": 9.729446702170758,
+      "learning_rate": 5e-06,
+      "loss": 0.2747,
+      "num_input_tokens_seen": 57185320,
+      "step": 333
+    },
+    {
+      "epoch": 0.08759124087591241,
+      "loss": 0.18412570655345917,
+      "loss_ce": 0.004499247297644615,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 57185320,
+      "step": 333
+    },
+    {
+      "epoch": 0.08785427763529953,
+      "grad_norm": 13.290037619570391,
+      "learning_rate": 5e-06,
+      "loss": 0.2065,
+      "num_input_tokens_seen": 57355852,
+      "step": 334
+    },
+    {
+      "epoch": 0.08785427763529953,
+      "loss": 0.17899873852729797,
+      "loss_ce": 0.007001678459346294,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 57355852,
+      "step": 334
+    },
+    {
+      "epoch": 0.08811731439468666,
+      "grad_norm": 13.54518630333599,
+      "learning_rate": 5e-06,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 57526160,
+      "step": 335
+    },
+    {
+      "epoch": 0.08811731439468666,
+      "loss": 0.42548078298568726,
+      "loss_ce": 0.004582356195896864,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 57526160,
+      "step": 335
+    },
+    {
+      "epoch": 0.08838035115407378,
+      "grad_norm": 12.0949131727458,
+      "learning_rate": 5e-06,
+      "loss": 0.3092,
+      "num_input_tokens_seen": 57698240,
+      "step": 336
+    },
+    {
+      "epoch": 0.08838035115407378,
+      "loss": 0.2829555869102478,
+      "loss_ce": 0.005245603621006012,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 57698240,
+      "step": 336
+    },
+    {
+      "epoch": 0.08864338791346091,
+      "grad_norm": 11.320262534807219,
+      "learning_rate": 5e-06,
+      "loss": 0.2301,
+      "num_input_tokens_seen": 57870476,
+      "step": 337
+    },
+    {
+      "epoch": 0.08864338791346091,
+      "loss": 0.18072998523712158,
+      "loss_ce": 0.004704595077782869,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 57870476,
+      "step": 337
+    },
+    {
+      "epoch": 0.08890642467284803,
+      "grad_norm": 7.708963725187913,
+      "learning_rate": 5e-06,
+      "loss": 0.2643,
+      "num_input_tokens_seen": 58042960,
+      "step": 338
+    },
+    {
+      "epoch": 0.08890642467284803,
+      "loss": 0.2719469666481018,
+      "loss_ce": 0.004002647008746862,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 58042960,
+      "step": 338
+    },
+    {
+      "epoch": 0.08916946143223516,
+      "grad_norm": 8.11559752489591,
+      "learning_rate": 5e-06,
+      "loss": 0.285,
+      "num_input_tokens_seen": 58215276,
+      "step": 339
+    },
+    {
+      "epoch": 0.08916946143223516,
+      "loss": 0.2689915895462036,
+      "loss_ce": 0.004709342960268259,
+      "loss_iou": 0.578125,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 58215276,
+      "step": 339
+    },
+    {
+      "epoch": 0.08943249819162227,
+      "grad_norm": 16.72537995408633,
+      "learning_rate": 5e-06,
+      "loss": 0.2206,
+      "num_input_tokens_seen": 58387140,
+      "step": 340
+    },
+    {
+      "epoch": 0.08943249819162227,
+      "loss": 0.19433817267417908,
+      "loss_ce": 0.007021272089332342,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 58387140,
+      "step": 340
+    },
+    {
+      "epoch": 0.0896955349510094,
+      "grad_norm": 10.068757947032363,
+      "learning_rate": 5e-06,
+      "loss": 0.2708,
+      "num_input_tokens_seen": 58559132,
+      "step": 341
+    },
+    {
+      "epoch": 0.0896955349510094,
+      "loss": 0.19732418656349182,
+      "loss_ce": 0.004819315858185291,
+      "loss_iou": 0.4375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 58559132,
+      "step": 341
+    },
+    {
+      "epoch": 0.08995857171039653,
+      "grad_norm": 15.667162296382404,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 58729752,
+      "step": 342
+    },
+    {
+      "epoch": 0.08995857171039653,
+      "loss": 0.251327782869339,
+      "loss_ce": 0.006515774410218,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 58729752,
+      "step": 342
+    },
+    {
+      "epoch": 0.09022160846978365,
+      "grad_norm": 10.706496762114405,
+      "learning_rate": 5e-06,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 58901996,
+      "step": 343
+    },
+    {
+      "epoch": 0.09022160846978365,
+      "loss": 0.2672620415687561,
+      "loss_ce": 0.003956370986998081,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 58901996,
+      "step": 343
+    },
+    {
+      "epoch": 0.09048464522917078,
+      "grad_norm": 8.994760858015587,
+      "learning_rate": 5e-06,
+      "loss": 0.2729,
+      "num_input_tokens_seen": 59071632,
+      "step": 344
+    },
+    {
+      "epoch": 0.09048464522917078,
+      "loss": 0.2578084170818329,
+      "loss_ce": 0.0040242355316877365,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 59071632,
+      "step": 344
+    },
+    {
+      "epoch": 0.0907476819885579,
+      "grad_norm": 10.48877247334524,
+      "learning_rate": 5e-06,
+      "loss": 0.1963,
+      "num_input_tokens_seen": 59240464,
+      "step": 345
+    },
+    {
+      "epoch": 0.0907476819885579,
+      "loss": 0.20404496788978577,
+      "loss_ce": 0.0031172330491244793,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 59240464,
+      "step": 345
+    },
+    {
+      "epoch": 0.09101071874794503,
+      "grad_norm": 11.358390932213611,
+      "learning_rate": 5e-06,
+      "loss": 0.2309,
+      "num_input_tokens_seen": 59411172,
+      "step": 346
+    },
+    {
+      "epoch": 0.09101071874794503,
+      "loss": 0.23203009366989136,
+      "loss_ce": 0.013524244539439678,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 59411172,
+      "step": 346
+    },
+    {
+      "epoch": 0.09127375550733215,
+      "grad_norm": 11.905025621505947,
+      "learning_rate": 5e-06,
+      "loss": 0.2869,
+      "num_input_tokens_seen": 59581764,
+      "step": 347
+    },
+    {
+      "epoch": 0.09127375550733215,
+      "loss": 0.34626448154449463,
+      "loss_ce": 0.005199995823204517,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 59581764,
+      "step": 347
+    },
+    {
+      "epoch": 0.09153679226671928,
+      "grad_norm": 12.735636545678894,
+      "learning_rate": 5e-06,
+      "loss": 0.3255,
+      "num_input_tokens_seen": 59754336,
+      "step": 348
+    },
+    {
+      "epoch": 0.09153679226671928,
+      "loss": 0.395874559879303,
+      "loss_ce": 0.005859900265932083,
+      "loss_iou": 0.3125,
+      "loss_num": 0.078125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 59754336,
+      "step": 348
+    },
+    {
+      "epoch": 0.0917998290261064,
+      "grad_norm": 12.687289585677311,
+      "learning_rate": 5e-06,
+      "loss": 0.3009,
+      "num_input_tokens_seen": 59926836,
+      "step": 349
+    },
+    {
+      "epoch": 0.0917998290261064,
+      "loss": 0.25403833389282227,
+      "loss_ce": 0.0031838202849030495,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 59926836,
+      "step": 349
+    },
+    {
+      "epoch": 0.09206286578549353,
+      "grad_norm": 7.2501298492412145,
+      "learning_rate": 5e-06,
+      "loss": 0.193,
+      "num_input_tokens_seen": 60099096,
+      "step": 350
+    },
+    {
+      "epoch": 0.09206286578549353,
+      "loss": 0.25642409920692444,
+      "loss_ce": 0.005325470119714737,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 60099096,
+      "step": 350
+    },
+    {
+      "epoch": 0.09232590254488064,
+      "grad_norm": 7.233766618859882,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 60271456,
+      "step": 351
+    },
+    {
+      "epoch": 0.09232590254488064,
+      "loss": 0.23899057507514954,
+      "loss_ce": 0.003883154597133398,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 60271456,
+      "step": 351
+    },
+    {
+      "epoch": 0.09258893930426777,
+      "grad_norm": 10.116609125862757,
+      "learning_rate": 5e-06,
+      "loss": 0.2124,
+      "num_input_tokens_seen": 60443440,
+      "step": 352
+    },
+    {
+      "epoch": 0.09258893930426777,
+      "loss": 0.2204253077507019,
+      "loss_ce": 0.005032242741435766,
+      "loss_iou": 0.53125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 60443440,
+      "step": 352
+    },
+    {
+      "epoch": 0.09285197606365489,
+      "grad_norm": 11.612967488254396,
+      "learning_rate": 5e-06,
+      "loss": 0.2054,
+      "num_input_tokens_seen": 60615884,
+      "step": 353
+    },
+    {
+      "epoch": 0.09285197606365489,
+      "loss": 0.20439094305038452,
+      "loss_ce": 0.005050132982432842,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 60615884,
+      "step": 353
+    },
+    {
+      "epoch": 0.09311501282304202,
+      "grad_norm": 14.924807522940846,
+      "learning_rate": 5e-06,
+      "loss": 0.2902,
+      "num_input_tokens_seen": 60787976,
+      "step": 354
+    },
+    {
+      "epoch": 0.09311501282304202,
+      "loss": 0.35237956047058105,
+      "loss_ce": 0.006432283669710159,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 60787976,
+      "step": 354
+    },
+    {
+      "epoch": 0.09337804958242915,
+      "grad_norm": 14.749952758612022,
+      "learning_rate": 5e-06,
+      "loss": 0.2866,
+      "num_input_tokens_seen": 60960464,
+      "step": 355
+    },
+    {
+      "epoch": 0.09337804958242915,
+      "loss": 0.27518266439437866,
+      "loss_ce": 0.008092833682894707,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 60960464,
+      "step": 355
+    },
+    {
+      "epoch": 0.09364108634181627,
+      "grad_norm": 8.028682980536022,
+      "learning_rate": 5e-06,
+      "loss": 0.2222,
+      "num_input_tokens_seen": 61130588,
+      "step": 356
+    },
+    {
+      "epoch": 0.09364108634181627,
+      "loss": 0.2519834041595459,
+      "loss_ce": 0.004424803890287876,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 61130588,
+      "step": 356
+    },
+    {
+      "epoch": 0.0939041231012034,
+      "grad_norm": 9.199423751114487,
+      "learning_rate": 5e-06,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 61302652,
+      "step": 357
+    },
+    {
+      "epoch": 0.0939041231012034,
+      "loss": 0.39324474334716797,
+      "loss_ce": 0.012019152753055096,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.380859375,
+      "num_input_tokens_seen": 61302652,
+      "step": 357
+    },
+    {
+      "epoch": 0.09416715986059052,
+      "grad_norm": 8.678464430801892,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 61473012,
+      "step": 358
+    },
+    {
+      "epoch": 0.09416715986059052,
+      "loss": 0.20840191841125488,
+      "loss_ce": 0.00448344461619854,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 61473012,
+      "step": 358
+    },
+    {
+      "epoch": 0.09443019661997765,
+      "grad_norm": 13.785653955051378,
+      "learning_rate": 5e-06,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 61645272,
+      "step": 359
+    },
+    {
+      "epoch": 0.09443019661997765,
+      "loss": 0.19086723029613495,
+      "loss_ce": 0.0067241680808365345,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 61645272,
+      "step": 359
+    },
+    {
+      "epoch": 0.09469323337936476,
+      "grad_norm": 7.119442902133065,
+      "learning_rate": 5e-06,
+      "loss": 0.2102,
+      "num_input_tokens_seen": 61817436,
+      "step": 360
+    },
+    {
+      "epoch": 0.09469323337936476,
+      "loss": 0.21873927116394043,
+      "loss_ce": 0.0036513670347630978,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 61817436,
+      "step": 360
+    },
+    {
+      "epoch": 0.0949562701387519,
+      "grad_norm": 10.914204757351408,
+      "learning_rate": 5e-06,
+      "loss": 0.2247,
+      "num_input_tokens_seen": 61987764,
+      "step": 361
+    },
+    {
+      "epoch": 0.0949562701387519,
+      "loss": 0.2072867900133133,
+      "loss_ce": 0.0077018230222165585,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 61987764,
+      "step": 361
+    },
+    {
+      "epoch": 0.09521930689813901,
+      "grad_norm": 10.090324495449488,
+      "learning_rate": 5e-06,
+      "loss": 0.2923,
+      "num_input_tokens_seen": 62158356,
+      "step": 362
+    },
+    {
+      "epoch": 0.09521930689813901,
+      "loss": 0.2174176275730133,
+      "loss_ce": 0.007944983430206776,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 62158356,
+      "step": 362
+    },
+    {
+      "epoch": 0.09548234365752614,
+      "grad_norm": 8.545687476622918,
+      "learning_rate": 5e-06,
+      "loss": 0.2599,
+      "num_input_tokens_seen": 62330412,
+      "step": 363
+    },
+    {
+      "epoch": 0.09548234365752614,
+      "loss": 0.2114790380001068,
+      "loss_ce": 0.004081577528268099,
+      "loss_iou": 0.546875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 62330412,
+      "step": 363
+    },
+    {
+      "epoch": 0.09574538041691326,
+      "grad_norm": 9.203465723081402,
+      "learning_rate": 5e-06,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 62502416,
+      "step": 364
+    },
+    {
+      "epoch": 0.09574538041691326,
+      "loss": 0.20543332397937775,
+      "loss_ce": 0.004505585879087448,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 62502416,
+      "step": 364
+    },
+    {
+      "epoch": 0.09600841717630039,
+      "grad_norm": 11.930210195365788,
+      "learning_rate": 5e-06,
+      "loss": 0.3152,
+      "num_input_tokens_seen": 62674708,
+      "step": 365
+    },
+    {
+      "epoch": 0.09600841717630039,
+      "loss": 0.3104754686355591,
+      "loss_ce": 0.006398319266736507,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 62674708,
+      "step": 365
+    },
+    {
+      "epoch": 0.0962714539356875,
+      "grad_norm": 9.520122083335602,
+      "learning_rate": 5e-06,
+      "loss": 0.2852,
+      "num_input_tokens_seen": 62846712,
+      "step": 366
+    },
+    {
+      "epoch": 0.0962714539356875,
+      "loss": 0.30926454067230225,
+      "loss_ce": 0.008239164017140865,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 62846712,
+      "step": 366
+    },
+    {
+      "epoch": 0.09653449069507464,
+      "grad_norm": 7.113583060734605,
+      "learning_rate": 5e-06,
+      "loss": 0.2553,
+      "num_input_tokens_seen": 63018700,
+      "step": 367
+    },
+    {
+      "epoch": 0.09653449069507464,
+      "loss": 0.27072834968566895,
+      "loss_ce": 0.006201992742717266,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 63018700,
+      "step": 367
+    },
+    {
+      "epoch": 0.09679752745446175,
+      "grad_norm": 7.613528329197277,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 63188284,
+      "step": 368
+    },
+    {
+      "epoch": 0.09679752745446175,
+      "loss": 0.19734525680541992,
+      "loss_ce": 0.0055117676965892315,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 63188284,
+      "step": 368
+    },
+    {
+      "epoch": 0.09706056421384888,
+      "grad_norm": 7.59501735101136,
+      "learning_rate": 5e-06,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 63360584,
+      "step": 369
+    },
+    {
+      "epoch": 0.09706056421384888,
+      "loss": 0.23239563405513763,
+      "loss_ce": 0.003757936879992485,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 63360584,
+      "step": 369
+    },
+    {
+      "epoch": 0.09732360097323602,
+      "grad_norm": 9.071713140748203,
+      "learning_rate": 5e-06,
+      "loss": 0.2417,
+      "num_input_tokens_seen": 63530348,
+      "step": 370
+    },
+    {
+      "epoch": 0.09732360097323602,
+      "loss": 0.3295820653438568,
+      "loss_ce": 0.004997124895453453,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 63530348,
+      "step": 370
+    },
+    {
+      "epoch": 0.09758663773262313,
+      "grad_norm": 18.945858878396198,
+      "learning_rate": 5e-06,
+      "loss": 0.2916,
+      "num_input_tokens_seen": 63702772,
+      "step": 371
+    },
+    {
+      "epoch": 0.09758663773262313,
+      "loss": 0.38020581007003784,
+      "loss_ce": 0.007525136228650808,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 63702772,
+      "step": 371
+    },
+    {
+      "epoch": 0.09784967449201026,
+      "grad_norm": 10.004542997696944,
+      "learning_rate": 5e-06,
+      "loss": 0.2707,
+      "num_input_tokens_seen": 63874988,
+      "step": 372
+    },
+    {
+      "epoch": 0.09784967449201026,
+      "loss": 0.21838681399822235,
+      "loss_ce": 0.005313089117407799,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 63874988,
+      "step": 372
+    },
+    {
+      "epoch": 0.09811271125139738,
+      "grad_norm": 12.569402693685879,
+      "learning_rate": 5e-06,
+      "loss": 0.3095,
+      "num_input_tokens_seen": 64047192,
+      "step": 373
+    },
+    {
+      "epoch": 0.09811271125139738,
+      "loss": 0.23138675093650818,
+      "loss_ce": 0.0027490435168147087,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 64047192,
+      "step": 373
+    },
+    {
+      "epoch": 0.09837574801078451,
+      "grad_norm": 7.761350865641741,
+      "learning_rate": 5e-06,
+      "loss": 0.2808,
+      "num_input_tokens_seen": 64217628,
+      "step": 374
+    },
+    {
+      "epoch": 0.09837574801078451,
+      "loss": 0.22036589682102203,
+      "loss_ce": 0.005583181045949459,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 64217628,
+      "step": 374
+    },
+    {
+      "epoch": 0.09863878477017163,
+      "grad_norm": 15.752581561948926,
+      "learning_rate": 5e-06,
+      "loss": 0.2404,
+      "num_input_tokens_seen": 64389824,
+      "step": 375
+    },
+    {
+      "epoch": 0.09863878477017163,
+      "loss": 0.31320562958717346,
+      "loss_ce": 0.0048560285940766335,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 64389824,
+      "step": 375
+    },
+    {
+      "epoch": 0.09890182152955876,
+      "grad_norm": 8.987788273052438,
+      "learning_rate": 5e-06,
+      "loss": 0.2167,
+      "num_input_tokens_seen": 64557964,
+      "step": 376
+    },
+    {
+      "epoch": 0.09890182152955876,
+      "loss": 0.15088969469070435,
+      "loss_ce": 0.0028794521931558847,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 64557964,
+      "step": 376
+    },
+    {
+      "epoch": 0.09916485828894588,
+      "grad_norm": 14.201905900297023,
+      "learning_rate": 5e-06,
+      "loss": 0.282,
+      "num_input_tokens_seen": 64730276,
+      "step": 377
+    },
+    {
+      "epoch": 0.09916485828894588,
+      "loss": 0.2574768662452698,
+      "loss_ce": 0.004425112158060074,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 64730276,
+      "step": 377
+    },
+    {
+      "epoch": 0.099427895048333,
+      "grad_norm": 10.355784479360198,
+      "learning_rate": 5e-06,
+      "loss": 0.2585,
+      "num_input_tokens_seen": 64902196,
+      "step": 378
+    },
+    {
+      "epoch": 0.099427895048333,
+      "loss": 0.38254526257514954,
+      "loss_ce": 0.00461557786911726,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 64902196,
+      "step": 378
+    },
+    {
+      "epoch": 0.09969093180772012,
+      "grad_norm": 8.387859091991706,
+      "learning_rate": 5e-06,
+      "loss": 0.266,
+      "num_input_tokens_seen": 65074196,
+      "step": 379
+    },
+    {
+      "epoch": 0.09969093180772012,
+      "loss": 0.3536115884780884,
+      "loss_ce": 0.007664328906685114,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 65074196,
+      "step": 379
+    },
+    {
+      "epoch": 0.09995396856710725,
+      "grad_norm": 6.106776880682037,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 65245768,
+      "step": 380
+    },
+    {
+      "epoch": 0.09995396856710725,
+      "loss": 0.16343596577644348,
+      "loss_ce": 0.003645919729024172,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 65245768,
+      "step": 380
+    },
+    {
+      "epoch": 0.10021700532649437,
+      "grad_norm": 13.780341497897325,
+      "learning_rate": 5e-06,
+      "loss": 0.203,
+      "num_input_tokens_seen": 65418016,
+      "step": 381
+    },
+    {
+      "epoch": 0.10021700532649437,
+      "loss": 0.21609731018543243,
+      "loss_ce": 0.004000143148005009,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 65418016,
+      "step": 381
+    },
+    {
+      "epoch": 0.1004800420858815,
+      "grad_norm": 9.32261949794866,
+      "learning_rate": 5e-06,
+      "loss": 0.2265,
+      "num_input_tokens_seen": 65590312,
+      "step": 382
+    },
+    {
+      "epoch": 0.1004800420858815,
+      "loss": 0.22935181856155396,
+      "loss_ce": 0.00425416324287653,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 65590312,
+      "step": 382
+    },
+    {
+      "epoch": 0.10074307884526863,
+      "grad_norm": 8.048917738163752,
+      "learning_rate": 5e-06,
+      "loss": 0.2741,
+      "num_input_tokens_seen": 65762432,
+      "step": 383
+    },
+    {
+      "epoch": 0.10074307884526863,
+      "loss": 0.3397749662399292,
+      "loss_ce": 0.0026167738251388073,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 65762432,
+      "step": 383
+    },
+    {
+      "epoch": 0.10100611560465575,
+      "grad_norm": 10.637781772264608,
+      "learning_rate": 5e-06,
+      "loss": 0.2218,
+      "num_input_tokens_seen": 65934468,
+      "step": 384
+    },
+    {
+      "epoch": 0.10100611560465575,
+      "loss": 0.2347402572631836,
+      "loss_ce": 0.01049709226936102,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 65934468,
+      "step": 384
+    },
+    {
+      "epoch": 0.10126915236404288,
+      "grad_norm": 8.002274766525902,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 66106856,
+      "step": 385
+    },
+    {
+      "epoch": 0.10126915236404288,
+      "loss": 0.2602759301662445,
+      "loss_ce": 0.005881410092115402,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 66106856,
+      "step": 385
+    },
+    {
+      "epoch": 0.10153218912343,
+      "grad_norm": 5.854621538534226,
+      "learning_rate": 5e-06,
+      "loss": 0.2458,
+      "num_input_tokens_seen": 66278808,
+      "step": 386
+    },
+    {
+      "epoch": 0.10153218912343,
+      "loss": 0.24760979413986206,
+      "loss_ce": 0.008718185126781464,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 66278808,
+      "step": 386
+    },
+    {
+      "epoch": 0.10179522588281713,
+      "grad_norm": 8.25201471581488,
+      "learning_rate": 5e-06,
+      "loss": 0.2036,
+      "num_input_tokens_seen": 66449192,
+      "step": 387
+    },
+    {
+      "epoch": 0.10179522588281713,
+      "loss": 0.18327301740646362,
+      "loss_ce": 0.007003485690802336,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 66449192,
+      "step": 387
+    },
+    {
+      "epoch": 0.10205826264220424,
+      "grad_norm": 7.445331676424441,
+      "learning_rate": 5e-06,
+      "loss": 0.2392,
+      "num_input_tokens_seen": 66619484,
+      "step": 388
+    },
+    {
+      "epoch": 0.10205826264220424,
+      "loss": 0.19971191883087158,
+      "loss_ce": 0.0065356409177184105,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 66619484,
+      "step": 388
+    },
+    {
+      "epoch": 0.10232129940159138,
+      "grad_norm": 6.590904938513716,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 66791680,
+      "step": 389
+    },
+    {
+      "epoch": 0.10232129940159138,
+      "loss": 0.186821848154068,
+      "loss_ce": 0.0036553400568664074,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 66791680,
+      "step": 389
+    },
+    {
+      "epoch": 0.10258433616097849,
+      "grad_norm": 7.115969301930752,
+      "learning_rate": 5e-06,
+      "loss": 0.2258,
+      "num_input_tokens_seen": 66963964,
+      "step": 390
+    },
+    {
+      "epoch": 0.10258433616097849,
+      "loss": 0.21056599915027618,
+      "loss_ce": 0.005732023622840643,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 66963964,
+      "step": 390
+    },
+    {
+      "epoch": 0.10284737292036562,
+      "grad_norm": 10.39999100726405,
+      "learning_rate": 5e-06,
+      "loss": 0.2688,
+      "num_input_tokens_seen": 67136264,
+      "step": 391
+    },
+    {
+      "epoch": 0.10284737292036562,
+      "loss": 0.32519546151161194,
+      "loss_ce": 0.0023194823879748583,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 67136264,
+      "step": 391
+    },
+    {
+      "epoch": 0.10311040967975274,
+      "grad_norm": 10.853923665788189,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 67308332,
+      "step": 392
+    },
+    {
+      "epoch": 0.10311040967975274,
+      "loss": 0.22598545253276825,
+      "loss_ce": 0.004671982489526272,
+      "loss_iou": 0.578125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 67308332,
+      "step": 392
+    },
+    {
+      "epoch": 0.10337344643913987,
+      "grad_norm": 12.644758786705145,
+      "learning_rate": 5e-06,
+      "loss": 0.2568,
+      "num_input_tokens_seen": 67480732,
+      "step": 393
+    },
+    {
+      "epoch": 0.10337344643913987,
+      "loss": 0.22954288125038147,
+      "loss_ce": 0.0029803775250911713,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 67480732,
+      "step": 393
+    },
+    {
+      "epoch": 0.10363648319852699,
+      "grad_norm": 9.802676625201595,
+      "learning_rate": 5e-06,
+      "loss": 0.2089,
+      "num_input_tokens_seen": 67652736,
+      "step": 394
+    },
+    {
+      "epoch": 0.10363648319852699,
+      "loss": 0.2672483026981354,
+      "loss_ce": 0.006017843261361122,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 67652736,
+      "step": 394
+    },
+    {
+      "epoch": 0.10389951995791412,
+      "grad_norm": 7.134550479825786,
+      "learning_rate": 5e-06,
+      "loss": 0.2498,
+      "num_input_tokens_seen": 67825356,
+      "step": 395
+    },
+    {
+      "epoch": 0.10389951995791412,
+      "loss": 0.2973036766052246,
+      "loss_ce": 0.004334905184805393,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 67825356,
+      "step": 395
+    },
+    {
+      "epoch": 0.10416255671730125,
+      "grad_norm": 9.967886960744174,
+      "learning_rate": 5e-06,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 67997632,
+      "step": 396
+    },
+    {
+      "epoch": 0.10416255671730125,
+      "loss": 0.16415753960609436,
+      "loss_ce": 0.0022923052310943604,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 67997632,
+      "step": 396
+    },
+    {
+      "epoch": 0.10442559347668837,
+      "grad_norm": 10.176506863521315,
+      "learning_rate": 5e-06,
+      "loss": 0.2534,
+      "num_input_tokens_seen": 68168484,
+      "step": 397
+    },
+    {
+      "epoch": 0.10442559347668837,
+      "loss": 0.2146688550710678,
+      "loss_ce": 0.002388589084148407,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 68168484,
+      "step": 397
+    },
+    {
+      "epoch": 0.1046886302360755,
+      "grad_norm": 12.732415116106893,
+      "learning_rate": 5e-06,
+      "loss": 0.2673,
+      "num_input_tokens_seen": 68337496,
+      "step": 398
+    },
+    {
+      "epoch": 0.1046886302360755,
+      "loss": 0.28035295009613037,
+      "loss_ce": 0.003985744901001453,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 68337496,
+      "step": 398
+    },
+    {
+      "epoch": 0.10495166699546261,
+      "grad_norm": 8.052010611714167,
+      "learning_rate": 5e-06,
+      "loss": 0.278,
+      "num_input_tokens_seen": 68509860,
+      "step": 399
+    },
+    {
+      "epoch": 0.10495166699546261,
+      "loss": 0.36081743240356445,
+      "loss_ce": 0.003395556937903166,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 68509860,
+      "step": 399
+    },
+    {
+      "epoch": 0.10521470375484974,
+      "grad_norm": 7.830511214158693,
+      "learning_rate": 5e-06,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 68681940,
+      "step": 400
+    },
+    {
+      "epoch": 0.10521470375484974,
+      "loss": 0.2878537178039551,
+      "loss_ce": 0.005016806535422802,
+      "loss_iou": 0.40625,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 68681940,
+      "step": 400
+    },
+    {
+      "epoch": 0.10547774051423686,
+      "grad_norm": 8.196278314466337,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 68854332,
+      "step": 401
+    },
+    {
+      "epoch": 0.10547774051423686,
+      "loss": 0.27157318592071533,
+      "loss_ce": 0.004483355674892664,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 68854332,
+      "step": 401
+    },
+    {
+      "epoch": 0.10574077727362399,
+      "grad_norm": 8.502085361700777,
+      "learning_rate": 5e-06,
+      "loss": 0.2004,
+      "num_input_tokens_seen": 69024884,
+      "step": 402
+    },
+    {
+      "epoch": 0.10574077727362399,
+      "loss": 0.20401804149150848,
+      "loss_ce": 0.002235804684460163,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 69024884,
+      "step": 402
+    },
+    {
+      "epoch": 0.10600381403301111,
+      "grad_norm": 12.566157085058673,
+      "learning_rate": 5e-06,
+      "loss": 0.2292,
+      "num_input_tokens_seen": 69197236,
+      "step": 403
+    },
+    {
+      "epoch": 0.10600381403301111,
+      "loss": 0.24189046025276184,
+      "loss_ce": 0.0023274626582860947,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 69197236,
+      "step": 403
+    },
+    {
+      "epoch": 0.10626685079239824,
+      "grad_norm": 8.163593280105717,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 69369436,
+      "step": 404
+    },
+    {
+      "epoch": 0.10626685079239824,
+      "loss": 0.17368575930595398,
+      "loss_ce": 0.005472864024341106,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 69369436,
+      "step": 404
+    },
+    {
+      "epoch": 0.10652988755178536,
+      "grad_norm": 7.3996201552939596,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 69540036,
+      "step": 405
+    },
+    {
+      "epoch": 0.10652988755178536,
+      "loss": 0.1399962604045868,
+      "loss_ce": 0.004925462882965803,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 69540036,
+      "step": 405
+    },
+    {
+      "epoch": 0.10679292431117249,
+      "grad_norm": 10.197562859133864,
+      "learning_rate": 5e-06,
+      "loss": 0.2373,
+      "num_input_tokens_seen": 69712000,
+      "step": 406
+    },
+    {
+      "epoch": 0.10679292431117249,
+      "loss": 0.19172173738479614,
+      "loss_ce": 0.0054424479603767395,
+      "loss_iou": 0.71875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 69712000,
+      "step": 406
+    },
+    {
+      "epoch": 0.1070559610705596,
+      "grad_norm": 13.16161294170375,
+      "learning_rate": 5e-06,
+      "loss": 0.2782,
+      "num_input_tokens_seen": 69883872,
+      "step": 407
+    },
+    {
+      "epoch": 0.1070559610705596,
+      "loss": 0.28615695238113403,
+      "loss_ce": 0.002465539611876011,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 69883872,
+      "step": 407
+    },
+    {
+      "epoch": 0.10731899782994674,
+      "grad_norm": 8.88729554885214,
+      "learning_rate": 5e-06,
+      "loss": 0.2054,
+      "num_input_tokens_seen": 70056060,
+      "step": 408
+    },
+    {
+      "epoch": 0.10731899782994674,
+      "loss": 0.19157525897026062,
+      "loss_ce": 0.0030986934434622526,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 70056060,
+      "step": 408
+    },
+    {
+      "epoch": 0.10758203458933385,
+      "grad_norm": 10.615365655634278,
+      "learning_rate": 5e-06,
+      "loss": 0.2465,
+      "num_input_tokens_seen": 70228104,
+      "step": 409
+    },
+    {
+      "epoch": 0.10758203458933385,
+      "loss": 0.315712034702301,
+      "loss_ce": 0.003150993725284934,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 70228104,
+      "step": 409
+    },
+    {
+      "epoch": 0.10784507134872098,
+      "grad_norm": 8.946094916164988,
+      "learning_rate": 5e-06,
+      "loss": 0.2711,
+      "num_input_tokens_seen": 70398676,
+      "step": 410
+    },
+    {
+      "epoch": 0.10784507134872098,
+      "loss": 0.37502604722976685,
+      "loss_ce": 0.010768221691250801,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 70398676,
+      "step": 410
+    },
+    {
+      "epoch": 0.10810810810810811,
+      "grad_norm": 10.043204254830277,
+      "learning_rate": 5e-06,
+      "loss": 0.2838,
+      "num_input_tokens_seen": 70570976,
+      "step": 411
+    },
+    {
+      "epoch": 0.10810810810810811,
+      "loss": 0.23639510571956635,
+      "loss_ce": 0.004095299169421196,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 70570976,
+      "step": 411
+    },
+    {
+      "epoch": 0.10837114486749523,
+      "grad_norm": 6.784469948331844,
+      "learning_rate": 5e-06,
+      "loss": 0.2968,
+      "num_input_tokens_seen": 70741228,
+      "step": 412
+    },
+    {
+      "epoch": 0.10837114486749523,
+      "loss": 0.3139989376068115,
+      "loss_ce": 0.001865162281319499,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 70741228,
+      "step": 412
+    },
+    {
+      "epoch": 0.10863418162688236,
+      "grad_norm": 6.343923096243914,
+      "learning_rate": 5e-06,
+      "loss": 0.2107,
+      "num_input_tokens_seen": 70913516,
+      "step": 413
+    },
+    {
+      "epoch": 0.10863418162688236,
+      "loss": 0.19345833361148834,
+      "loss_ce": 0.0018079333240166306,
+      "loss_iou": 0.5,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 70913516,
+      "step": 413
+    },
+    {
+      "epoch": 0.10889721838626948,
+      "grad_norm": 7.92695485009758,
+      "learning_rate": 5e-06,
+      "loss": 0.1941,
+      "num_input_tokens_seen": 71085736,
+      "step": 414
+    },
+    {
+      "epoch": 0.10889721838626948,
+      "loss": 0.18501858413219452,
+      "loss_ce": 0.001791047165170312,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 71085736,
+      "step": 414
+    },
+    {
+      "epoch": 0.10916025514565661,
+      "grad_norm": 8.754602822026959,
+      "learning_rate": 5e-06,
+      "loss": 0.2194,
+      "num_input_tokens_seen": 71257772,
+      "step": 415
+    },
+    {
+      "epoch": 0.10916025514565661,
+      "loss": 0.2221953421831131,
+      "loss_ce": 0.0017973824869841337,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 71257772,
+      "step": 415
+    },
+    {
+      "epoch": 0.10942329190504373,
+      "grad_norm": 7.459161826900908,
+      "learning_rate": 5e-06,
+      "loss": 0.2477,
+      "num_input_tokens_seen": 71430004,
+      "step": 416
+    },
+    {
+      "epoch": 0.10942329190504373,
+      "loss": 0.2557048797607422,
+      "loss_ce": 0.0026531266048550606,
+      "loss_iou": 0.578125,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 71430004,
+      "step": 416
+    },
+    {
+      "epoch": 0.10968632866443086,
+      "grad_norm": 6.337226894093766,
+      "learning_rate": 5e-06,
+      "loss": 0.236,
+      "num_input_tokens_seen": 71602160,
+      "step": 417
+    },
+    {
+      "epoch": 0.10968632866443086,
+      "loss": 0.1684013307094574,
+      "loss_ce": 0.0018363934941589832,
+      "loss_iou": 0.59375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 71602160,
+      "step": 417
+    },
+    {
+      "epoch": 0.10994936542381797,
+      "grad_norm": 10.092397777907028,
+      "learning_rate": 5e-06,
+      "loss": 0.1998,
+      "num_input_tokens_seen": 71774264,
+      "step": 418
+    },
+    {
+      "epoch": 0.10994936542381797,
+      "loss": 0.17219412326812744,
+      "loss_ce": 0.0024553609546273947,
+      "loss_iou": 0.53125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 71774264,
+      "step": 418
+    },
+    {
+      "epoch": 0.1102124021832051,
+      "grad_norm": 10.933777042764003,
+      "learning_rate": 5e-06,
+      "loss": 0.2549,
+      "num_input_tokens_seen": 71943760,
+      "step": 419
+    },
+    {
+      "epoch": 0.1102124021832051,
+      "loss": 0.2607925236225128,
+      "loss_ce": 0.0052993567660450935,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 71943760,
+      "step": 419
+    },
+    {
+      "epoch": 0.11047543894259222,
+      "grad_norm": 18.099442946915016,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 72114360,
+      "step": 420
+    },
+    {
+      "epoch": 0.11047543894259222,
+      "loss": 0.2702986001968384,
+      "loss_ce": 0.0022321869619190693,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 72114360,
+      "step": 420
+    },
+    {
+      "epoch": 0.11073847570197935,
+      "grad_norm": 6.645880081378423,
+      "learning_rate": 5e-06,
+      "loss": 0.2374,
+      "num_input_tokens_seen": 72285984,
+      "step": 421
+    },
+    {
+      "epoch": 0.11073847570197935,
+      "loss": 0.17263615131378174,
+      "loss_ce": 0.0023480583913624287,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 72285984,
+      "step": 421
+    },
+    {
+      "epoch": 0.11100151246136647,
+      "grad_norm": 10.576055281968134,
+      "learning_rate": 5e-06,
+      "loss": 0.1819,
+      "num_input_tokens_seen": 72458472,
+      "step": 422
+    },
+    {
+      "epoch": 0.11100151246136647,
+      "loss": 0.2114211916923523,
+      "loss_ce": 0.005244437139481306,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 72458472,
+      "step": 422
+    },
+    {
+      "epoch": 0.1112645492207536,
+      "grad_norm": 12.150497670240854,
+      "learning_rate": 5e-06,
+      "loss": 0.2879,
+      "num_input_tokens_seen": 72630848,
+      "step": 423
+    },
+    {
+      "epoch": 0.1112645492207536,
+      "loss": 0.195449560880661,
+      "loss_ce": 0.0027005516458302736,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 72630848,
+      "step": 423
+    },
+    {
+      "epoch": 0.11152758598014073,
+      "grad_norm": 9.206681239637351,
+      "learning_rate": 5e-06,
+      "loss": 0.1882,
+      "num_input_tokens_seen": 72802948,
+      "step": 424
+    },
+    {
+      "epoch": 0.11152758598014073,
+      "loss": 0.20556196570396423,
+      "loss_ce": 0.003230433911085129,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 72802948,
+      "step": 424
+    },
+    {
+      "epoch": 0.11179062273952785,
+      "grad_norm": 7.914405274492032,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 72973304,
+      "step": 425
+    },
+    {
+      "epoch": 0.11179062273952785,
+      "loss": 0.24065472185611725,
+      "loss_ce": 0.002800729824230075,
+      "loss_iou": 0.46875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 72973304,
+      "step": 425
+    },
+    {
+      "epoch": 0.11205365949891498,
+      "grad_norm": 8.079757468011927,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 73145520,
+      "step": 426
+    },
+    {
+      "epoch": 0.11205365949891498,
+      "loss": 0.17487749457359314,
+      "loss_ce": 0.007641167379915714,
+      "loss_iou": 0.515625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 73145520,
+      "step": 426
+    },
+    {
+      "epoch": 0.1123166962583021,
+      "grad_norm": 26.685292496143536,
+      "learning_rate": 5e-06,
+      "loss": 0.2674,
+      "num_input_tokens_seen": 73317728,
+      "step": 427
+    },
+    {
+      "epoch": 0.1123166962583021,
+      "loss": 0.24447987973690033,
+      "loss_ce": 0.003635148983448744,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 73317728,
+      "step": 427
+    },
+    {
+      "epoch": 0.11257973301768923,
+      "grad_norm": 10.587399579059058,
+      "learning_rate": 5e-06,
+      "loss": 0.2517,
+      "num_input_tokens_seen": 73490108,
+      "step": 428
+    },
+    {
+      "epoch": 0.11257973301768923,
+      "loss": 0.24800382554531097,
+      "loss_ce": 0.002154202200472355,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 73490108,
+      "step": 428
+    },
+    {
+      "epoch": 0.11284276977707634,
+      "grad_norm": 6.6654701515468515,
+      "learning_rate": 5e-06,
+      "loss": 0.2127,
+      "num_input_tokens_seen": 73662104,
+      "step": 429
+    },
+    {
+      "epoch": 0.11284276977707634,
+      "loss": 0.35326629877090454,
+      "loss_ce": 0.004389348905533552,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 73662104,
+      "step": 429
+    },
+    {
+      "epoch": 0.11310580653646347,
+      "grad_norm": 9.359106186429873,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 73833940,
+      "step": 430
+    },
+    {
+      "epoch": 0.11310580653646347,
+      "loss": 0.19281955063343048,
+      "loss_ce": 0.0022678023669868708,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 73833940,
+      "step": 430
+    },
+    {
+      "epoch": 0.11336884329585059,
+      "grad_norm": 10.912300657166487,
+      "learning_rate": 5e-06,
+      "loss": 0.2706,
+      "num_input_tokens_seen": 74006248,
+      "step": 431
+    },
+    {
+      "epoch": 0.11336884329585059,
+      "loss": 0.270729124546051,
+      "loss_ce": 0.0032730703242123127,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 74006248,
+      "step": 431
+    },
+    {
+      "epoch": 0.11363188005523772,
+      "grad_norm": 7.80923377535479,
+      "learning_rate": 5e-06,
+      "loss": 0.2294,
+      "num_input_tokens_seen": 74178292,
+      "step": 432
+    },
+    {
+      "epoch": 0.11363188005523772,
+      "loss": 0.21906697750091553,
+      "loss_ce": 0.002758371876552701,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 74178292,
+      "step": 432
+    },
+    {
+      "epoch": 0.11389491681462484,
+      "grad_norm": 7.423483614252486,
+      "learning_rate": 5e-06,
+      "loss": 0.2228,
+      "num_input_tokens_seen": 74350588,
+      "step": 433
+    },
+    {
+      "epoch": 0.11389491681462484,
+      "loss": 0.17986077070236206,
+      "loss_ce": 0.0017601896543055773,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 74350588,
+      "step": 433
+    },
+    {
+      "epoch": 0.11415795357401197,
+      "grad_norm": 8.108512259957333,
+      "learning_rate": 5e-06,
+      "loss": 0.226,
+      "num_input_tokens_seen": 74520920,
+      "step": 434
+    },
+    {
+      "epoch": 0.11415795357401197,
+      "loss": 0.21928107738494873,
+      "loss_ce": 0.00455939956009388,
+      "loss_iou": 0.640625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 74520920,
+      "step": 434
+    },
+    {
+      "epoch": 0.11442099033339909,
+      "grad_norm": 19.006805430751477,
+      "learning_rate": 5e-06,
+      "loss": 0.2661,
+      "num_input_tokens_seen": 74693212,
+      "step": 435
+    },
+    {
+      "epoch": 0.11442099033339909,
+      "loss": 0.3151628077030182,
+      "loss_ce": 0.0064469738863408566,
+      "loss_iou": 0.625,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 74693212,
+      "step": 435
+    },
+    {
+      "epoch": 0.11468402709278622,
+      "grad_norm": 8.863661674496596,
+      "learning_rate": 5e-06,
+      "loss": 0.2843,
+      "num_input_tokens_seen": 74865396,
+      "step": 436
+    },
+    {
+      "epoch": 0.11468402709278622,
+      "loss": 0.280520498752594,
+      "loss_ce": 0.006594708655029535,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 74865396,
+      "step": 436
+    },
+    {
+      "epoch": 0.11494706385217335,
+      "grad_norm": 7.497393138459489,
+      "learning_rate": 5e-06,
+      "loss": 0.1834,
+      "num_input_tokens_seen": 75037856,
+      "step": 437
+    },
+    {
+      "epoch": 0.11494706385217335,
+      "loss": 0.18996562063694,
+      "loss_ce": 0.0047849551774561405,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 75037856,
+      "step": 437
+    },
+    {
+      "epoch": 0.11521010061156046,
+      "grad_norm": 12.688606481249035,
+      "learning_rate": 5e-06,
+      "loss": 0.2195,
+      "num_input_tokens_seen": 75210192,
+      "step": 438
+    },
+    {
+      "epoch": 0.11521010061156046,
+      "loss": 0.20892465114593506,
+      "loss_ce": 0.006348971277475357,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 75210192,
+      "step": 438
+    },
+    {
+      "epoch": 0.1154731373709476,
+      "grad_norm": 8.208628036837384,
+      "learning_rate": 5e-06,
+      "loss": 0.2418,
+      "num_input_tokens_seen": 75382312,
+      "step": 439
+    },
+    {
+      "epoch": 0.1154731373709476,
+      "loss": 0.20543652772903442,
+      "loss_ce": 0.001701178727671504,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 75382312,
+      "step": 439
+    },
+    {
+      "epoch": 0.11573617413033471,
+      "grad_norm": 9.785336279235814,
+      "learning_rate": 5e-06,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 75554688,
+      "step": 440
+    },
+    {
+      "epoch": 0.11573617413033471,
+      "loss": 0.26715782284736633,
+      "loss_ce": 0.0069039189256727695,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 75554688,
+      "step": 440
+    },
+    {
+      "epoch": 0.11599921088972184,
+      "grad_norm": 23.031133350657914,
+      "learning_rate": 5e-06,
+      "loss": 0.2357,
+      "num_input_tokens_seen": 75726964,
+      "step": 441
+    },
+    {
+      "epoch": 0.11599921088972184,
+      "loss": 0.3105895519256592,
+      "loss_ce": 0.004559269640594721,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 75726964,
+      "step": 441
+    },
+    {
+      "epoch": 0.11626224764910896,
+      "grad_norm": 8.337171777387358,
+      "learning_rate": 5e-06,
+      "loss": 0.239,
+      "num_input_tokens_seen": 75899032,
+      "step": 442
+    },
+    {
+      "epoch": 0.11626224764910896,
+      "loss": 0.20333924889564514,
+      "loss_ce": 0.002655645599588752,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 75899032,
+      "step": 442
+    },
+    {
+      "epoch": 0.11652528440849609,
+      "grad_norm": 4.896294530096544,
+      "learning_rate": 5e-06,
+      "loss": 0.1725,
+      "num_input_tokens_seen": 76071304,
+      "step": 443
+    },
+    {
+      "epoch": 0.11652528440849609,
+      "loss": 0.20664632320404053,
+      "loss_ce": 0.0032161371782422066,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 76071304,
+      "step": 443
+    },
+    {
+      "epoch": 0.11678832116788321,
+      "grad_norm": 7.74925868300208,
+      "learning_rate": 5e-06,
+      "loss": 0.2369,
+      "num_input_tokens_seen": 76241732,
+      "step": 444
+    },
+    {
+      "epoch": 0.11678832116788321,
+      "loss": 0.2300896942615509,
+      "loss_ce": 0.0016351052327081561,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 76241732,
+      "step": 444
+    },
+    {
+      "epoch": 0.11705135792727034,
+      "grad_norm": 8.593064888198587,
+      "learning_rate": 5e-06,
+      "loss": 0.2023,
+      "num_input_tokens_seen": 76414004,
+      "step": 445
+    },
+    {
+      "epoch": 0.11705135792727034,
+      "loss": 0.14537394046783447,
+      "loss_ce": 0.0013919961638748646,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 76414004,
+      "step": 445
+    },
+    {
+      "epoch": 0.11731439468665746,
+      "grad_norm": 10.869613464978638,
+      "learning_rate": 5e-06,
+      "loss": 0.2657,
+      "num_input_tokens_seen": 76586248,
+      "step": 446
+    },
+    {
+      "epoch": 0.11731439468665746,
+      "loss": 0.2349330186843872,
+      "loss_ce": 0.004464263096451759,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 76586248,
+      "step": 446
+    },
+    {
+      "epoch": 0.11757743144604459,
+      "grad_norm": 12.757935439615991,
+      "learning_rate": 5e-06,
+      "loss": 0.2717,
+      "num_input_tokens_seen": 76758328,
+      "step": 447
+    },
+    {
+      "epoch": 0.11757743144604459,
+      "loss": 0.331451952457428,
+      "loss_ce": 0.0016179666854441166,
+      "loss_iou": 0.359375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 76758328,
+      "step": 447
+    },
+    {
+      "epoch": 0.1178404682054317,
+      "grad_norm": 11.375965574655721,
+      "learning_rate": 5e-06,
+      "loss": 0.2032,
+      "num_input_tokens_seen": 76930392,
+      "step": 448
+    },
+    {
+      "epoch": 0.1178404682054317,
+      "loss": 0.21405665576457977,
+      "loss_ce": 0.004584001377224922,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 76930392,
+      "step": 448
+    },
+    {
+      "epoch": 0.11810350496481883,
+      "grad_norm": 7.336838831766789,
+      "learning_rate": 5e-06,
+      "loss": 0.206,
+      "num_input_tokens_seen": 77102556,
+      "step": 449
+    },
+    {
+      "epoch": 0.11810350496481883,
+      "loss": 0.2166377753019333,
+      "loss_ce": 0.005700268317013979,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 77102556,
+      "step": 449
+    },
+    {
+      "epoch": 0.11836654172420595,
+      "grad_norm": 11.990630578782412,
+      "learning_rate": 5e-06,
+      "loss": 0.2454,
+      "num_input_tokens_seen": 77274600,
+      "step": 450
+    },
+    {
+      "epoch": 0.11836654172420595,
+      "loss": 0.24853403866291046,
+      "loss_ce": 0.00378305627964437,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 77274600,
+      "step": 450
+    },
+    {
+      "epoch": 0.11862957848359308,
+      "grad_norm": 9.139541774912184,
+      "learning_rate": 5e-06,
+      "loss": 0.2682,
+      "num_input_tokens_seen": 77446596,
+      "step": 451
+    },
+    {
+      "epoch": 0.11862957848359308,
+      "loss": 0.2904722988605499,
+      "loss_ce": 0.002020149724557996,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 77446596,
+      "step": 451
+    },
+    {
+      "epoch": 0.11889261524298021,
+      "grad_norm": 7.052664643006658,
+      "learning_rate": 5e-06,
+      "loss": 0.199,
+      "num_input_tokens_seen": 77618608,
+      "step": 452
+    },
+    {
+      "epoch": 0.11889261524298021,
+      "loss": 0.23526260256767273,
+      "loss_ce": 0.004305572714656591,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 77618608,
+      "step": 452
+    },
+    {
+      "epoch": 0.11915565200236733,
+      "grad_norm": 8.748276102929202,
+      "learning_rate": 5e-06,
+      "loss": 0.2101,
+      "num_input_tokens_seen": 77790812,
+      "step": 453
+    },
+    {
+      "epoch": 0.11915565200236733,
+      "loss": 0.2472519874572754,
+      "loss_ce": 0.0025620569940656424,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 77790812,
+      "step": 453
+    },
+    {
+      "epoch": 0.11941868876175446,
+      "grad_norm": 5.877127785016851,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 77962644,
+      "step": 454
+    },
+    {
+      "epoch": 0.11941868876175446,
+      "loss": 0.2044929563999176,
+      "loss_ce": 0.005640420597046614,
+      "loss_iou": 0.5625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 77962644,
+      "step": 454
+    },
+    {
+      "epoch": 0.11968172552114158,
+      "grad_norm": 12.849038288174842,
+      "learning_rate": 5e-06,
+      "loss": 0.1941,
+      "num_input_tokens_seen": 78134616,
+      "step": 455
+    },
+    {
+      "epoch": 0.11968172552114158,
+      "loss": 0.20492224395275116,
+      "loss_ce": 0.002163449302315712,
+      "loss_iou": 0.640625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 78134616,
+      "step": 455
+    },
+    {
+      "epoch": 0.11994476228052871,
+      "grad_norm": 9.804877236674342,
+      "learning_rate": 5e-06,
+      "loss": 0.2629,
+      "num_input_tokens_seen": 78303656,
+      "step": 456
+    },
+    {
+      "epoch": 0.11994476228052871,
+      "loss": 0.23393824696540833,
+      "loss_ce": 0.0058498685248196125,
+      "loss_iou": NaN,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 78303656,
+      "step": 456
+    },
+    {
+      "epoch": 0.12020779903991582,
+      "grad_norm": 23.369495819583523,
+      "learning_rate": 5e-06,
+      "loss": 0.2449,
+      "num_input_tokens_seen": 78475936,
+      "step": 457
+    },
+    {
+      "epoch": 0.12020779903991582,
+      "loss": 0.2364155352115631,
+      "loss_ce": 0.009792003780603409,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 78475936,
+      "step": 457
+    },
+    {
+      "epoch": 0.12047083579930296,
+      "grad_norm": 9.860704537193882,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 78648124,
+      "step": 458
+    },
+    {
+      "epoch": 0.12047083579930296,
+      "loss": 0.1498676836490631,
+      "loss_ce": 0.002711937762796879,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 78648124,
+      "step": 458
+    },
+    {
+      "epoch": 0.12073387255869007,
+      "grad_norm": 8.562392622535716,
+      "learning_rate": 5e-06,
+      "loss": 0.234,
+      "num_input_tokens_seen": 78820104,
+      "step": 459
+    },
+    {
+      "epoch": 0.12073387255869007,
+      "loss": 0.23113086819648743,
+      "loss_ce": 0.0024931649677455425,
+      "loss_iou": 0.484375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 78820104,
+      "step": 459
+    },
+    {
+      "epoch": 0.1209969093180772,
+      "grad_norm": 6.866997211199077,
+      "learning_rate": 5e-06,
+      "loss": 0.2359,
+      "num_input_tokens_seen": 78992332,
+      "step": 460
+    },
+    {
+      "epoch": 0.1209969093180772,
+      "loss": 0.15971623361110687,
+      "loss_ce": 0.0017572464421391487,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 78992332,
+      "step": 460
+    },
+    {
+      "epoch": 0.12125994607746432,
+      "grad_norm": 6.404476112237903,
+      "learning_rate": 5e-06,
+      "loss": 0.2015,
+      "num_input_tokens_seen": 79164376,
+      "step": 461
+    },
+    {
+      "epoch": 0.12125994607746432,
+      "loss": 0.2323172241449356,
+      "loss_ce": 0.005937827751040459,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 79164376,
+      "step": 461
+    },
+    {
+      "epoch": 0.12152298283685145,
+      "grad_norm": 9.290291620775372,
+      "learning_rate": 5e-06,
+      "loss": 0.1959,
+      "num_input_tokens_seen": 79336652,
+      "step": 462
+    },
+    {
+      "epoch": 0.12152298283685145,
+      "loss": 0.18938115239143372,
+      "loss_ce": 0.002125292085111141,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 79336652,
+      "step": 462
+    },
+    {
+      "epoch": 0.12178601959623857,
+      "grad_norm": 11.240674872659573,
+      "learning_rate": 5e-06,
+      "loss": 0.1971,
+      "num_input_tokens_seen": 79508724,
+      "step": 463
+    },
+    {
+      "epoch": 0.12178601959623857,
+      "loss": 0.2277367115020752,
+      "loss_ce": 0.001418360392563045,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 79508724,
+      "step": 463
+    },
+    {
+      "epoch": 0.1220490563556257,
+      "grad_norm": 13.32960488624403,
+      "learning_rate": 5e-06,
+      "loss": 0.1817,
+      "num_input_tokens_seen": 79678784,
+      "step": 464
+    },
+    {
+      "epoch": 0.1220490563556257,
+      "loss": 0.21636496484279633,
+      "loss_ce": 0.00182638771366328,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 79678784,
+      "step": 464
+    },
+    {
+      "epoch": 0.12231209311501283,
+      "grad_norm": 12.994343511615464,
+      "learning_rate": 5e-06,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 79850852,
+      "step": 465
+    },
+    {
+      "epoch": 0.12231209311501283,
+      "loss": 0.24236971139907837,
+      "loss_ce": 0.0015249918214976788,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 79850852,
+      "step": 465
+    },
+    {
+      "epoch": 0.12257512987439995,
+      "grad_norm": 10.746140563102669,
+      "learning_rate": 5e-06,
+      "loss": 0.2493,
+      "num_input_tokens_seen": 80021192,
+      "step": 466
+    },
+    {
+      "epoch": 0.12257512987439995,
+      "loss": 0.1711360514163971,
+      "loss_ce": 0.001702459529042244,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 80021192,
+      "step": 466
+    },
+    {
+      "epoch": 0.12283816663378708,
+      "grad_norm": 20.444744144323252,
+      "learning_rate": 5e-06,
+      "loss": 0.2662,
+      "num_input_tokens_seen": 80190112,
+      "step": 467
+    },
+    {
+      "epoch": 0.12283816663378708,
+      "loss": 0.21875979006290436,
+      "loss_ce": 0.0026953346095979214,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 80190112,
+      "step": 467
+    },
+    {
+      "epoch": 0.1231012033931742,
+      "grad_norm": 12.488827753013481,
+      "learning_rate": 5e-06,
+      "loss": 0.2328,
+      "num_input_tokens_seen": 80362892,
+      "step": 468
+    },
+    {
+      "epoch": 0.1231012033931742,
+      "loss": 0.17285287380218506,
+      "loss_ce": 0.002381683327257633,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 80362892,
+      "step": 468
+    },
+    {
+      "epoch": 0.12336424015256132,
+      "grad_norm": 9.176610530631967,
+      "learning_rate": 5e-06,
+      "loss": 0.2275,
+      "num_input_tokens_seen": 80534972,
+      "step": 469
+    },
+    {
+      "epoch": 0.12336424015256132,
+      "loss": 0.27365219593048096,
+      "loss_ce": 0.0032664609607309103,
+      "loss_iou": 0.421875,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 80534972,
+      "step": 469
+    },
+    {
+      "epoch": 0.12362727691194844,
+      "grad_norm": 33.167673801928274,
+      "learning_rate": 5e-06,
+      "loss": 0.2499,
+      "num_input_tokens_seen": 80705332,
+      "step": 470
+    },
+    {
+      "epoch": 0.12362727691194844,
+      "loss": 0.23707842826843262,
+      "loss_ce": 0.0022151488810777664,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 80705332,
+      "step": 470
+    },
+    {
+      "epoch": 0.12389031367133557,
+      "grad_norm": 7.668114037963123,
+      "learning_rate": 5e-06,
+      "loss": 0.2394,
+      "num_input_tokens_seen": 80874564,
+      "step": 471
+    },
+    {
+      "epoch": 0.12389031367133557,
+      "loss": 0.23063993453979492,
+      "loss_ce": 0.006457816809415817,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 80874564,
+      "step": 471
+    },
+    {
+      "epoch": 0.12415335043072269,
+      "grad_norm": 10.510373606639076,
+      "learning_rate": 5e-06,
+      "loss": 0.2114,
+      "num_input_tokens_seen": 81046768,
+      "step": 472
+    },
+    {
+      "epoch": 0.12415335043072269,
+      "loss": 0.21059830486774445,
+      "loss_ce": 0.002834630198776722,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 81046768,
+      "step": 472
+    },
+    {
+      "epoch": 0.12441638719010982,
+      "grad_norm": 11.032749822897834,
+      "learning_rate": 5e-06,
+      "loss": 0.2413,
+      "num_input_tokens_seen": 81218788,
+      "step": 473
+    },
+    {
+      "epoch": 0.12441638719010982,
+      "loss": 0.2567683458328247,
+      "loss_ce": 0.004815223626792431,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 81218788,
+      "step": 473
+    },
+    {
+      "epoch": 0.12467942394949694,
+      "grad_norm": 8.640049648990784,
+      "learning_rate": 5e-06,
+      "loss": 0.2498,
+      "num_input_tokens_seen": 81390764,
+      "step": 474
+    },
+    {
+      "epoch": 0.12467942394949694,
+      "loss": 0.3335922062397003,
+      "loss_ce": 0.0035140730906277895,
+      "loss_iou": 0.40625,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 81390764,
+      "step": 474
+    },
+    {
+      "epoch": 0.12494246070888407,
+      "grad_norm": 6.133084134287981,
+      "learning_rate": 5e-06,
+      "loss": 0.1862,
+      "num_input_tokens_seen": 81561192,
+      "step": 475
+    },
+    {
+      "epoch": 0.12494246070888407,
+      "loss": 0.19583408534526825,
+      "loss_ce": 0.002474710112437606,
+      "loss_iou": 0.28125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 81561192,
+      "step": 475
+    },
+    {
+      "epoch": 0.1252054974682712,
+      "grad_norm": 15.123599266995042,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "num_input_tokens_seen": 81731608,
+      "step": 476
+    },
+    {
+      "epoch": 0.1252054974682712,
+      "loss": 0.2318619191646576,
+      "loss_ce": 0.0012100562453269958,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 81731608,
+      "step": 476
+    },
+    {
+      "epoch": 0.1254685342276583,
+      "grad_norm": 10.43813000825477,
+      "learning_rate": 5e-06,
+      "loss": 0.1887,
+      "num_input_tokens_seen": 81902388,
+      "step": 477
+    },
+    {
+      "epoch": 0.1254685342276583,
+      "loss": 0.1947634220123291,
+      "loss_ce": 0.0017092193011194468,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 81902388,
+      "step": 477
+    },
+    {
+      "epoch": 0.12573157098704543,
+      "grad_norm": 15.053311018918661,
+      "learning_rate": 5e-06,
+      "loss": 0.2559,
+      "num_input_tokens_seen": 82074796,
+      "step": 478
+    },
+    {
+      "epoch": 0.12573157098704543,
+      "loss": 0.23443953692913055,
+      "loss_ce": 0.002872154116630554,
+      "loss_iou": 0.59375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 82074796,
+      "step": 478
+    },
+    {
+      "epoch": 0.12599460774643256,
+      "grad_norm": 8.071545338749708,
+      "learning_rate": 5e-06,
+      "loss": 0.2732,
+      "num_input_tokens_seen": 82246976,
+      "step": 479
+    },
+    {
+      "epoch": 0.12599460774643256,
+      "loss": 0.22861449420452118,
+      "loss_ce": 0.001929928082972765,
+      "loss_iou": NaN,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 82246976,
+      "step": 479
+    },
+    {
+      "epoch": 0.1262576445058197,
+      "grad_norm": 13.388933170286325,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 82419244,
+      "step": 480
+    },
+    {
+      "epoch": 0.1262576445058197,
+      "loss": 0.24052694439888,
+      "loss_ce": 0.002611914649605751,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 82419244,
+      "step": 480
+    },
+    {
+      "epoch": 0.12652068126520682,
+      "grad_norm": 10.19375568056882,
+      "learning_rate": 5e-06,
+      "loss": 0.2132,
+      "num_input_tokens_seen": 82591568,
+      "step": 481
+    },
+    {
+      "epoch": 0.12652068126520682,
+      "loss": 0.27330607175827026,
+      "loss_ce": 0.007314843591302633,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 82591568,
+      "step": 481
+    },
+    {
+      "epoch": 0.12678371802459393,
+      "grad_norm": 8.166078619911394,
+      "learning_rate": 5e-06,
+      "loss": 0.21,
+      "num_input_tokens_seen": 82760532,
+      "step": 482
+    },
+    {
+      "epoch": 0.12678371802459393,
+      "loss": 0.19459792971611023,
+      "loss_ce": 0.003313753753900528,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 82760532,
+      "step": 482
+    },
+    {
+      "epoch": 0.12704675478398106,
+      "grad_norm": 8.980724720396978,
+      "learning_rate": 5e-06,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 82932700,
+      "step": 483
+    },
+    {
+      "epoch": 0.12704675478398106,
+      "loss": 0.2369249314069748,
+      "loss_ce": 0.0062120286747813225,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 82932700,
+      "step": 483
+    },
+    {
+      "epoch": 0.1273097915433682,
+      "grad_norm": 7.529582302992287,
+      "learning_rate": 5e-06,
+      "loss": 0.2337,
+      "num_input_tokens_seen": 83104784,
+      "step": 484
+    },
+    {
+      "epoch": 0.1273097915433682,
+      "loss": 0.27052199840545654,
+      "loss_ce": 0.0020893928594887257,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 83104784,
+      "step": 484
+    },
+    {
+      "epoch": 0.12757282830275532,
+      "grad_norm": 9.051361983377177,
+      "learning_rate": 5e-06,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 83276660,
+      "step": 485
+    },
+    {
+      "epoch": 0.12757282830275532,
+      "loss": 0.1893678605556488,
+      "loss_ce": 0.004553401842713356,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 83276660,
+      "step": 485
+    },
+    {
+      "epoch": 0.12783586506214242,
+      "grad_norm": 7.363403269312881,
+      "learning_rate": 5e-06,
+      "loss": 0.2164,
+      "num_input_tokens_seen": 83448804,
+      "step": 486
+    },
+    {
+      "epoch": 0.12783586506214242,
+      "loss": 0.18258926272392273,
+      "loss_ce": 0.004244527779519558,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 83448804,
+      "step": 486
+    },
+    {
+      "epoch": 0.12809890182152955,
+      "grad_norm": 6.462059422866227,
+      "learning_rate": 5e-06,
+      "loss": 0.1922,
+      "num_input_tokens_seen": 83621024,
+      "step": 487
+    },
+    {
+      "epoch": 0.12809890182152955,
+      "loss": 0.14729665219783783,
+      "loss_ce": 0.003986096940934658,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 83621024,
+      "step": 487
+    },
+    {
+      "epoch": 0.12836193858091668,
+      "grad_norm": 9.164596498872053,
+      "learning_rate": 5e-06,
+      "loss": 0.2078,
+      "num_input_tokens_seen": 83793612,
+      "step": 488
+    },
+    {
+      "epoch": 0.12836193858091668,
+      "loss": 0.20358332991600037,
+      "loss_ce": 0.0016790404915809631,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 83793612,
+      "step": 488
+    },
+    {
+      "epoch": 0.12862497534030382,
+      "grad_norm": 13.35296525183839,
+      "learning_rate": 5e-06,
+      "loss": 0.3196,
+      "num_input_tokens_seen": 83965664,
+      "step": 489
+    },
+    {
+      "epoch": 0.12862497534030382,
+      "loss": 0.36003273725509644,
+      "loss_ce": 0.007981948554515839,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 83965664,
+      "step": 489
+    },
+    {
+      "epoch": 0.12888801209969092,
+      "grad_norm": 8.219901608770293,
+      "learning_rate": 5e-06,
+      "loss": 0.2425,
+      "num_input_tokens_seen": 84137656,
+      "step": 490
+    },
+    {
+      "epoch": 0.12888801209969092,
+      "loss": 0.21684233844280243,
+      "loss_ce": 0.0033413656055927277,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 84137656,
+      "step": 490
+    },
+    {
+      "epoch": 0.12915104885907805,
+      "grad_norm": 5.772697796397355,
+      "learning_rate": 5e-06,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 84308240,
+      "step": 491
+    },
+    {
+      "epoch": 0.12915104885907805,
+      "loss": 0.1743691861629486,
+      "loss_ce": 0.0028604045510292053,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 84308240,
+      "step": 491
+    },
+    {
+      "epoch": 0.12941408561846518,
+      "grad_norm": 7.154674159301149,
+      "learning_rate": 5e-06,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 84478680,
+      "step": 492
+    },
+    {
+      "epoch": 0.12941408561846518,
+      "loss": 0.15929880738258362,
+      "loss_ce": 0.003598117269575596,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 84478680,
+      "step": 492
+    },
+    {
+      "epoch": 0.1296771223778523,
+      "grad_norm": 10.931778081568359,
+      "learning_rate": 5e-06,
+      "loss": 0.2397,
+      "num_input_tokens_seen": 84648900,
+      "step": 493
+    },
+    {
+      "epoch": 0.1296771223778523,
+      "loss": 0.24825721979141235,
+      "loss_ce": 0.002285533118993044,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 84648900,
+      "step": 493
+    },
+    {
+      "epoch": 0.12994015913723944,
+      "grad_norm": 10.142675625135299,
+      "learning_rate": 5e-06,
+      "loss": 0.2203,
+      "num_input_tokens_seen": 84820972,
+      "step": 494
+    },
+    {
+      "epoch": 0.12994015913723944,
+      "loss": 0.23026269674301147,
+      "loss_ce": 0.0015639647608622909,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 84820972,
+      "step": 494
+    },
+    {
+      "epoch": 0.13020319589662654,
+      "grad_norm": 10.196346960471569,
+      "learning_rate": 5e-06,
+      "loss": 0.197,
+      "num_input_tokens_seen": 84993508,
+      "step": 495
+    },
+    {
+      "epoch": 0.13020319589662654,
+      "loss": 0.21560978889465332,
+      "loss_ce": 0.001620523864403367,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 84993508,
+      "step": 495
+    },
+    {
+      "epoch": 0.13046623265601368,
+      "grad_norm": 8.523677693002021,
+      "learning_rate": 5e-06,
+      "loss": 0.2357,
+      "num_input_tokens_seen": 85165596,
+      "step": 496
+    },
+    {
+      "epoch": 0.13046623265601368,
+      "loss": 0.17495451867580414,
+      "loss_ce": 0.003750909585505724,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 85165596,
+      "step": 496
+    },
+    {
+      "epoch": 0.1307292694154008,
+      "grad_norm": 5.883749010160293,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 85337976,
+      "step": 497
+    },
+    {
+      "epoch": 0.1307292694154008,
+      "loss": 0.22726929187774658,
+      "loss_ce": 0.009862057864665985,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 85337976,
+      "step": 497
+    },
+    {
+      "epoch": 0.13099230617478794,
+      "grad_norm": 7.408260148240015,
+      "learning_rate": 5e-06,
+      "loss": 0.2231,
+      "num_input_tokens_seen": 85509860,
+      "step": 498
+    },
+    {
+      "epoch": 0.13099230617478794,
+      "loss": 0.1337101012468338,
+      "loss_ce": 0.004010388161987066,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 85509860,
+      "step": 498
+    },
+    {
+      "epoch": 0.13125534293417504,
+      "grad_norm": 15.7224480316305,
+      "learning_rate": 5e-06,
+      "loss": 0.2184,
+      "num_input_tokens_seen": 85680488,
+      "step": 499
+    },
+    {
+      "epoch": 0.13125534293417504,
+      "loss": 0.3409336507320404,
+      "loss_ce": 0.002554745879024267,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 85680488,
+      "step": 499
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "grad_norm": 9.533912931076111,
+      "learning_rate": 5e-06,
+      "loss": 0.2073,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_websight_new_CIoU": 0.7392345666885376,
+      "eval_websight_new_GIoU": 0.7380270659923553,
+      "eval_websight_new_IoU": 0.7466294467449188,
+      "eval_websight_new_MAE_all": 0.04153955727815628,
+      "eval_websight_new_MAE_h": 0.03558222949504852,
+      "eval_websight_new_MAE_w": 0.06108394265174866,
+      "eval_websight_new_MAE_x": 0.04947785474359989,
+      "eval_websight_new_MAE_y": 0.020014189183712006,
+      "eval_websight_new_NUM_probability": 0.9849532246589661,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.20474952459335327,
+      "eval_websight_new_loss_ce": 0.0016258999821729958,
+      "eval_websight_new_loss_iou": 0.635986328125,
+      "eval_websight_new_loss_num": 0.037567138671875,
+      "eval_websight_new_loss_xval": 0.187774658203125,
+      "eval_websight_new_runtime": 54.6509,
+      "eval_websight_new_samples_per_second": 0.915,
+      "eval_websight_new_steps_per_second": 0.037,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_seeclick_CIoU": 0.41250014305114746,
+      "eval_seeclick_GIoU": 0.40925678610801697,
+      "eval_seeclick_IoU": 0.4613874703645706,
+      "eval_seeclick_MAE_all": 0.08603048324584961,
+      "eval_seeclick_MAE_h": 0.05444946512579918,
+      "eval_seeclick_MAE_w": 0.12106707319617271,
+      "eval_seeclick_MAE_x": 0.12660933285951614,
+      "eval_seeclick_MAE_y": 0.04199606738984585,
+      "eval_seeclick_NUM_probability": 0.9906161725521088,
+      "eval_seeclick_inside_bbox": 0.7698863744735718,
+      "eval_seeclick_loss": 0.3508862257003784,
+      "eval_seeclick_loss_ce": 0.013088095001876354,
+      "eval_seeclick_loss_iou": 0.609375,
+      "eval_seeclick_loss_num": 0.0647735595703125,
+      "eval_seeclick_loss_xval": 0.323974609375,
+      "eval_seeclick_runtime": 71.4374,
+      "eval_seeclick_samples_per_second": 0.602,
+      "eval_seeclick_steps_per_second": 0.028,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_icons_CIoU": 0.7077827751636505,
+      "eval_icons_GIoU": 0.7009360492229462,
+      "eval_icons_IoU": 0.7175993025302887,
+      "eval_icons_MAE_all": 0.041869472712278366,
+      "eval_icons_MAE_h": 0.04292410984635353,
+      "eval_icons_MAE_w": 0.04752085544168949,
+      "eval_icons_MAE_x": 0.038647109642624855,
+      "eval_icons_MAE_y": 0.03838581405580044,
+      "eval_icons_NUM_probability": 0.9924971163272858,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.13132929801940918,
+      "eval_icons_loss_ce": 0.003774530749069527,
+      "eval_icons_loss_iou": 0.590576171875,
+      "eval_icons_loss_num": 0.024749755859375,
+      "eval_icons_loss_xval": 0.12384033203125,
+      "eval_icons_runtime": 78.9038,
+      "eval_icons_samples_per_second": 0.634,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_screenspot_CIoU": 0.4913978377978007,
+      "eval_screenspot_GIoU": 0.47330527504285175,
+      "eval_screenspot_IoU": 0.5310356616973877,
+      "eval_screenspot_MAE_all": 0.09648379683494568,
+      "eval_screenspot_MAE_h": 0.061782063295443855,
+      "eval_screenspot_MAE_w": 0.14932986597220102,
+      "eval_screenspot_MAE_x": 0.11150848865509033,
+      "eval_screenspot_MAE_y": 0.0633147731423378,
+      "eval_screenspot_NUM_probability": 0.9926036596298218,
+      "eval_screenspot_inside_bbox": 0.8454166650772095,
+      "eval_screenspot_loss": 0.8485715389251709,
+      "eval_screenspot_loss_ce": 0.47645074129104614,
+      "eval_screenspot_loss_iou": 0.5421142578125,
+      "eval_screenspot_loss_num": 0.07304890950520833,
+      "eval_screenspot_loss_xval": 0.3654378255208333,
+      "eval_screenspot_runtime": 144.4943,
+      "eval_screenspot_samples_per_second": 0.616,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "loss": 0.8175798058509827,
+      "loss_ce": 0.44428879022598267,
+      "loss_iou": 0.5,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.1317814164529493,
+      "grad_norm": 9.07691940041295,
+      "learning_rate": 5e-06,
+      "loss": 0.1533,
+      "num_input_tokens_seen": 86021700,
+      "step": 501
+    },
+    {
+      "epoch": 0.1317814164529493,
+      "loss": 0.15382635593414307,
+      "loss_ce": 0.003191583789885044,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 86021700,
+      "step": 501
+    },
+    {
+      "epoch": 0.13204445321233643,
+      "grad_norm": 8.240988876623113,
+      "learning_rate": 5e-06,
+      "loss": 0.2297,
+      "num_input_tokens_seen": 86193792,
+      "step": 502
+    },
+    {
+      "epoch": 0.13204445321233643,
+      "loss": 0.23657214641571045,
+      "loss_ce": 0.003112667240202427,
+      "loss_iou": 0.625,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 86193792,
+      "step": 502
+    },
+    {
+      "epoch": 0.13230748997172354,
+      "grad_norm": 8.766196459329715,
+      "learning_rate": 5e-06,
+      "loss": 0.2049,
+      "num_input_tokens_seen": 86365784,
+      "step": 503
+    },
+    {
+      "epoch": 0.13230748997172354,
+      "loss": 0.23612971603870392,
+      "loss_ce": 0.004379219841212034,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 86365784,
+      "step": 503
+    },
+    {
+      "epoch": 0.13257052673111067,
+      "grad_norm": 5.336005731654041,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 86538044,
+      "step": 504
+    },
+    {
+      "epoch": 0.13257052673111067,
+      "loss": 0.15480023622512817,
+      "loss_ce": 0.0017545849550515413,
+      "loss_iou": 0.625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 86538044,
+      "step": 504
+    },
+    {
+      "epoch": 0.1328335634904978,
+      "grad_norm": 15.391291602176388,
+      "learning_rate": 5e-06,
+      "loss": 0.2019,
+      "num_input_tokens_seen": 86710012,
+      "step": 505
+    },
+    {
+      "epoch": 0.1328335634904978,
+      "loss": 0.16446326673030853,
+      "loss_ce": 0.0012552611296996474,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 86710012,
+      "step": 505
+    },
+    {
+      "epoch": 0.13309660024988493,
+      "grad_norm": 7.414971095415403,
+      "learning_rate": 5e-06,
+      "loss": 0.2046,
+      "num_input_tokens_seen": 86882600,
+      "step": 506
+    },
+    {
+      "epoch": 0.13309660024988493,
+      "loss": 0.18114808201789856,
+      "loss_ce": 0.0028033575508743525,
+      "loss_iou": 0.625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 86882600,
+      "step": 506
+    },
+    {
+      "epoch": 0.13335963700927206,
+      "grad_norm": 7.149578381431456,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 87055160,
+      "step": 507
+    },
+    {
+      "epoch": 0.13335963700927206,
+      "loss": 0.185621976852417,
+      "loss_ce": 0.0038592712953686714,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 87055160,
+      "step": 507
+    },
+    {
+      "epoch": 0.13362267376865916,
+      "grad_norm": 7.2770390495694235,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 87227176,
+      "step": 508
+    },
+    {
+      "epoch": 0.13362267376865916,
+      "loss": 0.12990637123584747,
+      "loss_ce": 0.0011221927125006914,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 87227176,
+      "step": 508
+    },
+    {
+      "epoch": 0.1338857105280463,
+      "grad_norm": 13.071840767784375,
+      "learning_rate": 5e-06,
+      "loss": 0.2921,
+      "num_input_tokens_seen": 87397716,
+      "step": 509
+    },
+    {
+      "epoch": 0.1338857105280463,
+      "loss": 0.31984156370162964,
+      "loss_ce": 0.0027639116160571575,
+      "loss_iou": NaN,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 87397716,
+      "step": 509
+    },
+    {
+      "epoch": 0.13414874728743342,
+      "grad_norm": 9.630666522841075,
+      "learning_rate": 5e-06,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 87570180,
+      "step": 510
+    },
+    {
+      "epoch": 0.13414874728743342,
+      "loss": 0.14270807802677155,
+      "loss_ce": 0.002327217720448971,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 87570180,
+      "step": 510
+    },
+    {
+      "epoch": 0.13441178404682055,
+      "grad_norm": 7.92909505971618,
+      "learning_rate": 5e-06,
+      "loss": 0.2147,
+      "num_input_tokens_seen": 87742132,
+      "step": 511
+    },
+    {
+      "epoch": 0.13441178404682055,
+      "loss": 0.24002233147621155,
+      "loss_ce": 0.0020462563261389732,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 87742132,
+      "step": 511
+    },
+    {
+      "epoch": 0.13467482080620766,
+      "grad_norm": 11.73787088896753,
+      "learning_rate": 5e-06,
+      "loss": 0.2146,
+      "num_input_tokens_seen": 87914144,
+      "step": 512
+    },
+    {
+      "epoch": 0.13467482080620766,
+      "loss": 0.2217179834842682,
+      "loss_ce": 0.00376143422909081,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 87914144,
+      "step": 512
+    },
+    {
+      "epoch": 0.1349378575655948,
+      "grad_norm": 14.481870714144165,
+      "learning_rate": 5e-06,
+      "loss": 0.2105,
+      "num_input_tokens_seen": 88086392,
+      "step": 513
+    },
+    {
+      "epoch": 0.1349378575655948,
+      "loss": 0.16927534341812134,
+      "loss_ce": 0.0027714259922504425,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 88086392,
+      "step": 513
+    },
+    {
+      "epoch": 0.13520089432498192,
+      "grad_norm": 11.732408058803708,
+      "learning_rate": 5e-06,
+      "loss": 0.2117,
+      "num_input_tokens_seen": 88258824,
+      "step": 514
+    },
+    {
+      "epoch": 0.13520089432498192,
+      "loss": 0.21276208758354187,
+      "loss_ce": 0.0047542620450258255,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 88258824,
+      "step": 514
+    },
+    {
+      "epoch": 0.13546393108436905,
+      "grad_norm": 8.699627697080732,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 88431280,
+      "step": 515
+    },
+    {
+      "epoch": 0.13546393108436905,
+      "loss": 0.13956406712532043,
+      "loss_ce": 0.005469819065183401,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 88431280,
+      "step": 515
+    },
+    {
+      "epoch": 0.13572696784375615,
+      "grad_norm": 9.441247877196542,
+      "learning_rate": 5e-06,
+      "loss": 0.2418,
+      "num_input_tokens_seen": 88603308,
+      "step": 516
+    },
+    {
+      "epoch": 0.13572696784375615,
+      "loss": 0.16743244230747223,
+      "loss_ce": 0.00141681800596416,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 88603308,
+      "step": 516
+    },
+    {
+      "epoch": 0.13599000460314328,
+      "grad_norm": 11.156675067329255,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 88775492,
+      "step": 517
+    },
+    {
+      "epoch": 0.13599000460314328,
+      "loss": 0.167589008808136,
+      "loss_ce": 0.0012681989464908838,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 88775492,
+      "step": 517
+    },
+    {
+      "epoch": 0.1362530413625304,
+      "grad_norm": 9.64852272360873,
+      "learning_rate": 5e-06,
+      "loss": 0.2377,
+      "num_input_tokens_seen": 88946204,
+      "step": 518
+    },
+    {
+      "epoch": 0.1362530413625304,
+      "loss": 0.2998642921447754,
+      "loss_ce": 0.004698258824646473,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 88946204,
+      "step": 518
+    },
+    {
+      "epoch": 0.13651607812191754,
+      "grad_norm": 8.486100540746042,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 89116056,
+      "step": 519
+    },
+    {
+      "epoch": 0.13651607812191754,
+      "loss": 0.21377842128276825,
+      "loss_ce": 0.004488877020776272,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 89116056,
+      "step": 519
+    },
+    {
+      "epoch": 0.13677911488130468,
+      "grad_norm": 7.241270611453955,
+      "learning_rate": 5e-06,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 89287012,
+      "step": 520
+    },
+    {
+      "epoch": 0.13677911488130468,
+      "loss": 0.23230193555355072,
+      "loss_ce": 0.0012228279374539852,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 89287012,
+      "step": 520
+    },
+    {
+      "epoch": 0.13704215164069178,
+      "grad_norm": 9.086385631838745,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 89457780,
+      "step": 521
+    },
+    {
+      "epoch": 0.13704215164069178,
+      "loss": 0.16174045205116272,
+      "loss_ce": 0.0012180046178400517,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 89457780,
+      "step": 521
+    },
+    {
+      "epoch": 0.1373051884000789,
+      "grad_norm": 9.260504659218878,
+      "learning_rate": 5e-06,
+      "loss": 0.1871,
+      "num_input_tokens_seen": 89628244,
+      "step": 522
+    },
+    {
+      "epoch": 0.1373051884000789,
+      "loss": 0.18322458863258362,
+      "loss_ce": 0.005673316773027182,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 89628244,
+      "step": 522
+    },
+    {
+      "epoch": 0.13756822515946604,
+      "grad_norm": 10.862554096761864,
+      "learning_rate": 5e-06,
+      "loss": 0.1938,
+      "num_input_tokens_seen": 89798512,
+      "step": 523
+    },
+    {
+      "epoch": 0.13756822515946604,
+      "loss": 0.18914146721363068,
+      "loss_ce": 0.002434919821098447,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 89798512,
+      "step": 523
+    },
+    {
+      "epoch": 0.13783126191885317,
+      "grad_norm": 8.527732112130064,
+      "learning_rate": 5e-06,
+      "loss": 0.2597,
+      "num_input_tokens_seen": 89968992,
+      "step": 524
+    },
+    {
+      "epoch": 0.13783126191885317,
+      "loss": 0.2936267554759979,
+      "loss_ce": 0.0038318424485623837,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 89968992,
+      "step": 524
+    },
+    {
+      "epoch": 0.13809429867824027,
+      "grad_norm": 8.96210431629978,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 90140828,
+      "step": 525
+    },
+    {
+      "epoch": 0.13809429867824027,
+      "loss": 0.22792883217334747,
+      "loss_ce": 0.001427364069968462,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 90140828,
+      "step": 525
+    },
+    {
+      "epoch": 0.1383573354376274,
+      "grad_norm": 10.303553365642298,
+      "learning_rate": 5e-06,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 90311476,
+      "step": 526
+    },
+    {
+      "epoch": 0.1383573354376274,
+      "loss": 0.09766636043787003,
+      "loss_ce": 0.00202426896430552,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 90311476,
+      "step": 526
+    },
+    {
+      "epoch": 0.13862037219701454,
+      "grad_norm": 7.578553389392675,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 90483668,
+      "step": 527
+    },
+    {
+      "epoch": 0.13862037219701454,
+      "loss": 0.2255394458770752,
+      "loss_ce": 0.0017845738912001252,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 90483668,
+      "step": 527
+    },
+    {
+      "epoch": 0.13888340895640167,
+      "grad_norm": 11.866590519507064,
+      "learning_rate": 5e-06,
+      "loss": 0.2463,
+      "num_input_tokens_seen": 90655996,
+      "step": 528
+    },
+    {
+      "epoch": 0.13888340895640167,
+      "loss": 0.21356430649757385,
+      "loss_ce": 0.004030614625662565,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 90655996,
+      "step": 528
+    },
+    {
+      "epoch": 0.13914644571578877,
+      "grad_norm": 9.66204006000912,
+      "learning_rate": 5e-06,
+      "loss": 0.2311,
+      "num_input_tokens_seen": 90828348,
+      "step": 529
+    },
+    {
+      "epoch": 0.13914644571578877,
+      "loss": 0.21380871534347534,
+      "loss_ce": 0.0014063662383705378,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 90828348,
+      "step": 529
+    },
+    {
+      "epoch": 0.1394094824751759,
+      "grad_norm": 7.3801351915919975,
+      "learning_rate": 5e-06,
+      "loss": 0.2048,
+      "num_input_tokens_seen": 91000476,
+      "step": 530
+    },
+    {
+      "epoch": 0.1394094824751759,
+      "loss": 0.19109413027763367,
+      "loss_ce": 0.0008780673379078507,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 91000476,
+      "step": 530
+    },
+    {
+      "epoch": 0.13967251923456303,
+      "grad_norm": 6.737214273696564,
+      "learning_rate": 5e-06,
+      "loss": 0.2037,
+      "num_input_tokens_seen": 91173056,
+      "step": 531
+    },
+    {
+      "epoch": 0.13967251923456303,
+      "loss": 0.26445770263671875,
+      "loss_ce": 0.0033187787048518658,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 91173056,
+      "step": 531
+    },
+    {
+      "epoch": 0.13993555599395016,
+      "grad_norm": 5.9400720051741835,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 91345516,
+      "step": 532
+    },
+    {
+      "epoch": 0.13993555599395016,
+      "loss": 0.12527181208133698,
+      "loss_ce": 0.001858725561760366,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 91345516,
+      "step": 532
+    },
+    {
+      "epoch": 0.1401985927533373,
+      "grad_norm": 8.591575042379741,
+      "learning_rate": 5e-06,
+      "loss": 0.2351,
+      "num_input_tokens_seen": 91516156,
+      "step": 533
+    },
+    {
+      "epoch": 0.1401985927533373,
+      "loss": 0.22694742679595947,
+      "loss_ce": 0.005511872004717588,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 91516156,
+      "step": 533
+    },
+    {
+      "epoch": 0.1404616295127244,
+      "grad_norm": 26.674816384255838,
+      "learning_rate": 5e-06,
+      "loss": 0.2705,
+      "num_input_tokens_seen": 91685124,
+      "step": 534
+    },
+    {
+      "epoch": 0.1404616295127244,
+      "loss": 0.2157442569732666,
+      "loss_ce": 0.005356077570468187,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 91685124,
+      "step": 534
+    },
+    {
+      "epoch": 0.14072466627211153,
+      "grad_norm": 7.992225607802382,
+      "learning_rate": 5e-06,
+      "loss": 0.2194,
+      "num_input_tokens_seen": 91857436,
+      "step": 535
+    },
+    {
+      "epoch": 0.14072466627211153,
+      "loss": 0.18514756858348846,
+      "loss_ce": 0.0036290136631578207,
+      "loss_iou": 0.625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 91857436,
+      "step": 535
+    },
+    {
+      "epoch": 0.14098770303149866,
+      "grad_norm": 7.005269973220872,
+      "learning_rate": 5e-06,
+      "loss": 0.2265,
+      "num_input_tokens_seen": 92029236,
+      "step": 536
+    },
+    {
+      "epoch": 0.14098770303149866,
+      "loss": 0.23437106609344482,
+      "loss_ce": 0.0035971456672996283,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 92029236,
+      "step": 536
+    },
+    {
+      "epoch": 0.1412507397908858,
+      "grad_norm": 13.679910251313718,
+      "learning_rate": 5e-06,
+      "loss": 0.2237,
+      "num_input_tokens_seen": 92201248,
+      "step": 537
+    },
+    {
+      "epoch": 0.1412507397908858,
+      "loss": 0.26536300778388977,
+      "loss_ce": 0.006451865192502737,
+      "loss_iou": 0.625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 92201248,
+      "step": 537
+    },
+    {
+      "epoch": 0.1415137765502729,
+      "grad_norm": 15.448137214976848,
+      "learning_rate": 5e-06,
+      "loss": 0.2463,
+      "num_input_tokens_seen": 92373276,
+      "step": 538
+    },
+    {
+      "epoch": 0.1415137765502729,
+      "loss": 0.2220609188079834,
+      "loss_ce": 0.0029447050765156746,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 92373276,
+      "step": 538
+    },
+    {
+      "epoch": 0.14177681330966002,
+      "grad_norm": 11.236165761153213,
+      "learning_rate": 5e-06,
+      "loss": 0.2009,
+      "num_input_tokens_seen": 92545880,
+      "step": 539
+    },
+    {
+      "epoch": 0.14177681330966002,
+      "loss": 0.21155700087547302,
+      "loss_ce": 0.006112661678344011,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 92545880,
+      "step": 539
+    },
+    {
+      "epoch": 0.14203985006904715,
+      "grad_norm": 7.303097321727043,
+      "learning_rate": 5e-06,
+      "loss": 0.2029,
+      "num_input_tokens_seen": 92717956,
+      "step": 540
+    },
+    {
+      "epoch": 0.14203985006904715,
+      "loss": 0.2111251950263977,
+      "loss_ce": 0.0041549778543412685,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 92717956,
+      "step": 540
+    },
+    {
+      "epoch": 0.14230288682843428,
+      "grad_norm": 28.317923328050057,
+      "learning_rate": 5e-06,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 92890260,
+      "step": 541
+    },
+    {
+      "epoch": 0.14230288682843428,
+      "loss": 0.18131288886070251,
+      "loss_ce": 0.004371959716081619,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 92890260,
+      "step": 541
+    },
+    {
+      "epoch": 0.14256592358782139,
+      "grad_norm": 12.97806529461824,
+      "learning_rate": 5e-06,
+      "loss": 0.2534,
+      "num_input_tokens_seen": 93062192,
+      "step": 542
+    },
+    {
+      "epoch": 0.14256592358782139,
+      "loss": 0.30919933319091797,
+      "loss_ce": 0.003779401071369648,
+      "loss_iou": 0.640625,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 93062192,
+      "step": 542
+    },
+    {
+      "epoch": 0.14282896034720852,
+      "grad_norm": 7.8709246328059725,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 93234480,
+      "step": 543
+    },
+    {
+      "epoch": 0.14282896034720852,
+      "loss": 0.2343926727771759,
+      "loss_ce": 0.002581145381554961,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 93234480,
+      "step": 543
+    },
+    {
+      "epoch": 0.14309199710659565,
+      "grad_norm": 8.436976597382053,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 93406784,
+      "step": 544
+    },
+    {
+      "epoch": 0.14309199710659565,
+      "loss": 0.20959031581878662,
+      "loss_ce": 0.0013383585028350353,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 93406784,
+      "step": 544
+    },
+    {
+      "epoch": 0.14335503386598278,
+      "grad_norm": 12.053836172217155,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 93577272,
+      "step": 545
+    },
+    {
+      "epoch": 0.14335503386598278,
+      "loss": 0.18772834539413452,
+      "loss_ce": 0.002242510672658682,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 93577272,
+      "step": 545
+    },
+    {
+      "epoch": 0.1436180706253699,
+      "grad_norm": 9.099796427619822,
+      "learning_rate": 5e-06,
+      "loss": 0.1713,
+      "num_input_tokens_seen": 93749292,
+      "step": 546
+    },
+    {
+      "epoch": 0.1436180706253699,
+      "loss": 0.2121119648218155,
+      "loss_ce": 0.00813247635960579,
+      "loss_iou": 0.671875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 93749292,
+      "step": 546
+    },
+    {
+      "epoch": 0.143881107384757,
+      "grad_norm": 13.08678717007557,
+      "learning_rate": 5e-06,
+      "loss": 0.2241,
+      "num_input_tokens_seen": 93921812,
+      "step": 547
+    },
+    {
+      "epoch": 0.143881107384757,
+      "loss": 0.22192896902561188,
+      "loss_ce": 0.002934828167781234,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 93921812,
+      "step": 547
+    },
+    {
+      "epoch": 0.14414414414414414,
+      "grad_norm": 8.230228011363112,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 94093976,
+      "step": 548
+    },
+    {
+      "epoch": 0.14414414414414414,
+      "loss": 0.21945567429065704,
+      "loss_ce": 0.0016822349280118942,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 94093976,
+      "step": 548
+    },
+    {
+      "epoch": 0.14440718090353127,
+      "grad_norm": 6.524669746029216,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 94262972,
+      "step": 549
+    },
+    {
+      "epoch": 0.14440718090353127,
+      "loss": 0.13413353264331818,
+      "loss_ce": 0.0032741604372859,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 94262972,
+      "step": 549
+    },
+    {
+      "epoch": 0.1446702176629184,
+      "grad_norm": 7.614359717596038,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 94435240,
+      "step": 550
+    },
+    {
+      "epoch": 0.1446702176629184,
+      "loss": 0.22648407518863678,
+      "loss_ce": 0.0034616070333868265,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 94435240,
+      "step": 550
+    },
+    {
+      "epoch": 0.1449332544223055,
+      "grad_norm": 7.224416897933664,
+      "learning_rate": 5e-06,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 94607488,
+      "step": 551
+    },
+    {
+      "epoch": 0.1449332544223055,
+      "loss": 0.18614572286605835,
+      "loss_ce": 0.0017585159512236714,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 94607488,
+      "step": 551
+    },
+    {
+      "epoch": 0.14519629118169264,
+      "grad_norm": 7.882321147436634,
+      "learning_rate": 5e-06,
+      "loss": 0.1957,
+      "num_input_tokens_seen": 94776340,
+      "step": 552
+    },
+    {
+      "epoch": 0.14519629118169264,
+      "loss": 0.22142915427684784,
+      "loss_ce": 0.003350543323904276,
+      "loss_iou": 0.53125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 94776340,
+      "step": 552
+    },
+    {
+      "epoch": 0.14545932794107977,
+      "grad_norm": 8.491750307418846,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 94948568,
+      "step": 553
+    },
+    {
+      "epoch": 0.14545932794107977,
+      "loss": 0.17424368858337402,
+      "loss_ce": 0.0024297323543578386,
+      "loss_iou": 0.578125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 94948568,
+      "step": 553
+    },
+    {
+      "epoch": 0.1457223647004669,
+      "grad_norm": 9.839418631011416,
+      "learning_rate": 5e-06,
+      "loss": 0.2278,
+      "num_input_tokens_seen": 95120868,
+      "step": 554
+    },
+    {
+      "epoch": 0.1457223647004669,
+      "loss": 0.23207631707191467,
+      "loss_ce": 0.001241360092535615,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 95120868,
+      "step": 554
+    },
+    {
+      "epoch": 0.145985401459854,
+      "grad_norm": 10.298790259293808,
+      "learning_rate": 5e-06,
+      "loss": 0.2316,
+      "num_input_tokens_seen": 95293380,
+      "step": 555
+    },
+    {
+      "epoch": 0.145985401459854,
+      "loss": 0.2527633607387543,
+      "loss_ce": 0.0020309346728026867,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 95293380,
+      "step": 555
+    },
+    {
+      "epoch": 0.14624843821924113,
+      "grad_norm": 11.750914646944318,
+      "learning_rate": 5e-06,
+      "loss": 0.1353,
+      "num_input_tokens_seen": 95465572,
+      "step": 556
+    },
+    {
+      "epoch": 0.14624843821924113,
+      "loss": 0.14519110321998596,
+      "loss_ce": 0.0012702068779617548,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 95465572,
+      "step": 556
+    },
+    {
+      "epoch": 0.14651147497862826,
+      "grad_norm": 11.674444875910362,
+      "learning_rate": 5e-06,
+      "loss": 0.248,
+      "num_input_tokens_seen": 95636384,
+      "step": 557
+    },
+    {
+      "epoch": 0.14651147497862826,
+      "loss": 0.2826082706451416,
+      "loss_ce": 0.0019686208106577396,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 95636384,
+      "step": 557
+    },
+    {
+      "epoch": 0.1467745117380154,
+      "grad_norm": 9.602535161901319,
+      "learning_rate": 5e-06,
+      "loss": 0.2109,
+      "num_input_tokens_seen": 95808820,
+      "step": 558
+    },
+    {
+      "epoch": 0.1467745117380154,
+      "loss": 0.1991540640592575,
+      "loss_ce": 0.0036584637127816677,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 95808820,
+      "step": 558
+    },
+    {
+      "epoch": 0.1470375484974025,
+      "grad_norm": 8.512482701290667,
+      "learning_rate": 5e-06,
+      "loss": 0.2626,
+      "num_input_tokens_seen": 95980876,
+      "step": 559
+    },
+    {
+      "epoch": 0.1470375484974025,
+      "loss": 0.3269794285297394,
+      "loss_ce": 0.0010516871698200703,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 95980876,
+      "step": 559
+    },
+    {
+      "epoch": 0.14730058525678963,
+      "grad_norm": 8.300386324136479,
+      "learning_rate": 5e-06,
+      "loss": 0.178,
+      "num_input_tokens_seen": 96153036,
+      "step": 560
+    },
+    {
+      "epoch": 0.14730058525678963,
+      "loss": 0.2191367745399475,
+      "loss_ce": 0.006795480381697416,
+      "loss_iou": 0.5625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 96153036,
+      "step": 560
+    },
+    {
+      "epoch": 0.14756362201617676,
+      "grad_norm": 5.775876970805374,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 96322532,
+      "step": 561
+    },
+    {
+      "epoch": 0.14756362201617676,
+      "loss": 0.1977005898952484,
+      "loss_ce": 0.006996248383074999,
+      "loss_iou": 0.625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 96322532,
+      "step": 561
+    },
+    {
+      "epoch": 0.1478266587755639,
+      "grad_norm": 6.533356383107408,
+      "learning_rate": 5e-06,
+      "loss": 0.2025,
+      "num_input_tokens_seen": 96494568,
+      "step": 562
+    },
+    {
+      "epoch": 0.1478266587755639,
+      "loss": 0.15407304465770721,
+      "loss_ce": 0.0012410087510943413,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 96494568,
+      "step": 562
+    },
+    {
+      "epoch": 0.14808969553495102,
+      "grad_norm": 9.772718457216582,
+      "learning_rate": 5e-06,
+      "loss": 0.2334,
+      "num_input_tokens_seen": 96666892,
+      "step": 563
+    },
+    {
+      "epoch": 0.14808969553495102,
+      "loss": 0.18993595242500305,
+      "loss_ce": 0.002985279308632016,
+      "loss_iou": 0.75,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 96666892,
+      "step": 563
+    },
+    {
+      "epoch": 0.14835273229433812,
+      "grad_norm": 10.172183960721854,
+      "learning_rate": 5e-06,
+      "loss": 0.2474,
+      "num_input_tokens_seen": 96837620,
+      "step": 564
+    },
+    {
+      "epoch": 0.14835273229433812,
+      "loss": 0.20410403609275818,
+      "loss_ce": 0.004641146864742041,
+      "loss_iou": 0.5,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 96837620,
+      "step": 564
+    },
+    {
+      "epoch": 0.14861576905372526,
+      "grad_norm": 9.445540103601473,
+      "learning_rate": 5e-06,
+      "loss": 0.2275,
+      "num_input_tokens_seen": 97009692,
+      "step": 565
+    },
+    {
+      "epoch": 0.14861576905372526,
+      "loss": 0.13070067763328552,
+      "loss_ce": 0.0018554661655798554,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 97009692,
+      "step": 565
+    },
+    {
+      "epoch": 0.14887880581311239,
+      "grad_norm": 7.521927615990519,
+      "learning_rate": 5e-06,
+      "loss": 0.2008,
+      "num_input_tokens_seen": 97182076,
+      "step": 566
+    },
+    {
+      "epoch": 0.14887880581311239,
+      "loss": 0.26979702711105347,
+      "loss_ce": 0.009176918305456638,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 97182076,
+      "step": 566
+    },
+    {
+      "epoch": 0.14914184257249952,
+      "grad_norm": 6.945071393253576,
+      "learning_rate": 5e-06,
+      "loss": 0.2175,
+      "num_input_tokens_seen": 97352348,
+      "step": 567
+    },
+    {
+      "epoch": 0.14914184257249952,
+      "loss": 0.2486119419336319,
+      "loss_ce": 0.004898556973785162,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 97352348,
+      "step": 567
+    },
+    {
+      "epoch": 0.14940487933188662,
+      "grad_norm": 19.761163824718725,
+      "learning_rate": 5e-06,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 97524808,
+      "step": 568
+    },
+    {
+      "epoch": 0.14940487933188662,
+      "loss": 0.2611408531665802,
+      "loss_ce": 0.0016193758929148316,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 97524808,
+      "step": 568
+    },
+    {
+      "epoch": 0.14966791609127375,
+      "grad_norm": 11.05044000141412,
+      "learning_rate": 5e-06,
+      "loss": 0.2071,
+      "num_input_tokens_seen": 97696860,
+      "step": 569
+    },
+    {
+      "epoch": 0.14966791609127375,
+      "loss": 0.17859557271003723,
+      "loss_ce": 0.002020859392359853,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 97696860,
+      "step": 569
+    },
+    {
+      "epoch": 0.14993095285066088,
+      "grad_norm": 10.09349160636077,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 97869048,
+      "step": 570
+    },
+    {
+      "epoch": 0.14993095285066088,
+      "loss": 0.1531415581703186,
+      "loss_ce": 0.001164011424407363,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 97869048,
+      "step": 570
+    },
+    {
+      "epoch": 0.150193989610048,
+      "grad_norm": 6.128271963212333,
+      "learning_rate": 5e-06,
+      "loss": 0.1797,
+      "num_input_tokens_seen": 98041152,
+      "step": 571
+    },
+    {
+      "epoch": 0.150193989610048,
+      "loss": 0.18703126907348633,
+      "loss_ce": 0.003956317901611328,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 98041152,
+      "step": 571
+    },
+    {
+      "epoch": 0.15045702636943512,
+      "grad_norm": 9.404399204091265,
+      "learning_rate": 5e-06,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 98213508,
+      "step": 572
+    },
+    {
+      "epoch": 0.15045702636943512,
+      "loss": 0.1673622578382492,
+      "loss_ce": 0.011722613126039505,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 98213508,
+      "step": 572
+    },
+    {
+      "epoch": 0.15072006312882225,
+      "grad_norm": 8.281687271797518,
+      "learning_rate": 5e-06,
+      "loss": 0.2554,
+      "num_input_tokens_seen": 98385768,
+      "step": 573
+    },
+    {
+      "epoch": 0.15072006312882225,
+      "loss": 0.29049456119537354,
+      "loss_ce": 0.007413491606712341,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 98385768,
+      "step": 573
+    },
+    {
+      "epoch": 0.15098309988820938,
+      "grad_norm": 9.329556868568911,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 98555404,
+      "step": 574
+    },
+    {
+      "epoch": 0.15098309988820938,
+      "loss": 0.17510266602039337,
+      "loss_ce": 0.006096326746046543,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 98555404,
+      "step": 574
+    },
+    {
+      "epoch": 0.1512461366475965,
+      "grad_norm": 6.666929195510284,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 98727452,
+      "step": 575
+    },
+    {
+      "epoch": 0.1512461366475965,
+      "loss": 0.2249300628900528,
+      "loss_ce": 0.008560429327189922,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 98727452,
+      "step": 575
+    },
+    {
+      "epoch": 0.15150917340698364,
+      "grad_norm": 12.576347626904536,
+      "learning_rate": 5e-06,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 98899788,
+      "step": 576
+    },
+    {
+      "epoch": 0.15150917340698364,
+      "loss": 0.21397414803504944,
+      "loss_ce": 0.004837184213101864,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 98899788,
+      "step": 576
+    },
+    {
+      "epoch": 0.15177221016637074,
+      "grad_norm": 9.891929331498387,
+      "learning_rate": 5e-06,
+      "loss": 0.2031,
+      "num_input_tokens_seen": 99072416,
+      "step": 577
+    },
+    {
+      "epoch": 0.15177221016637074,
+      "loss": 0.24624097347259521,
+      "loss_ce": 0.011255611665546894,
+      "loss_iou": 0.546875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 99072416,
+      "step": 577
+    },
+    {
+      "epoch": 0.15203524692575787,
+      "grad_norm": 12.530200633920343,
+      "learning_rate": 5e-06,
+      "loss": 0.2216,
+      "num_input_tokens_seen": 99244280,
+      "step": 578
+    },
+    {
+      "epoch": 0.15203524692575787,
+      "loss": 0.19744133949279785,
+      "loss_ce": 0.001030205050483346,
+      "loss_iou": 0.671875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 99244280,
+      "step": 578
+    },
+    {
+      "epoch": 0.152298283685145,
+      "grad_norm": 10.630444163918117,
+      "learning_rate": 5e-06,
+      "loss": 0.2121,
+      "num_input_tokens_seen": 99416184,
+      "step": 579
+    },
+    {
+      "epoch": 0.152298283685145,
+      "loss": 0.2667901813983917,
+      "loss_ce": 0.0020807269029319286,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 99416184,
+      "step": 579
+    },
+    {
+      "epoch": 0.15256132044453213,
+      "grad_norm": 10.945427939330838,
+      "learning_rate": 5e-06,
+      "loss": 0.1943,
+      "num_input_tokens_seen": 99588500,
+      "step": 580
+    },
+    {
+      "epoch": 0.15256132044453213,
+      "loss": 0.19293344020843506,
+      "loss_ce": 0.005006188526749611,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 99588500,
+      "step": 580
+    },
+    {
+      "epoch": 0.15282435720391924,
+      "grad_norm": 10.349037140854193,
+      "learning_rate": 5e-06,
+      "loss": 0.1856,
+      "num_input_tokens_seen": 99760948,
+      "step": 581
+    },
+    {
+      "epoch": 0.15282435720391924,
+      "loss": 0.1601850688457489,
+      "loss_ce": 0.0018598883179947734,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 99760948,
+      "step": 581
+    },
+    {
+      "epoch": 0.15308739396330637,
+      "grad_norm": 13.119978211180065,
+      "learning_rate": 5e-06,
+      "loss": 0.1813,
+      "num_input_tokens_seen": 99932992,
+      "step": 582
+    },
+    {
+      "epoch": 0.15308739396330637,
+      "loss": 0.24380066990852356,
+      "loss_ce": 0.002528695622459054,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 99932992,
+      "step": 582
+    },
+    {
+      "epoch": 0.1533504307226935,
+      "grad_norm": 8.165624923829593,
+      "learning_rate": 5e-06,
+      "loss": 0.1637,
+      "num_input_tokens_seen": 100104992,
+      "step": 583
+    },
+    {
+      "epoch": 0.1533504307226935,
+      "loss": 0.18439146876335144,
+      "loss_ce": 0.0024456623941659927,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 100104992,
+      "step": 583
+    },
+    {
+      "epoch": 0.15361346748208063,
+      "grad_norm": 9.086411500582741,
+      "learning_rate": 5e-06,
+      "loss": 0.1853,
+      "num_input_tokens_seen": 100277144,
+      "step": 584
+    },
+    {
+      "epoch": 0.15361346748208063,
+      "loss": 0.17580674588680267,
+      "loss_ce": 0.0032603610306978226,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 100277144,
+      "step": 584
+    },
+    {
+      "epoch": 0.15387650424146773,
+      "grad_norm": 11.772959826434088,
+      "learning_rate": 5e-06,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 100449660,
+      "step": 585
+    },
+    {
+      "epoch": 0.15387650424146773,
+      "loss": 0.2301916778087616,
+      "loss_ce": 0.0029577831737697124,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 100449660,
+      "step": 585
+    },
+    {
+      "epoch": 0.15413954100085486,
+      "grad_norm": 8.411861716804005,
+      "learning_rate": 5e-06,
+      "loss": 0.2131,
+      "num_input_tokens_seen": 100622028,
+      "step": 586
+    },
+    {
+      "epoch": 0.15413954100085486,
+      "loss": 0.183104008436203,
+      "loss_ce": 0.0029892674647271633,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 100622028,
+      "step": 586
+    },
+    {
+      "epoch": 0.154402577760242,
+      "grad_norm": 8.219849696412924,
+      "learning_rate": 5e-06,
+      "loss": 0.2095,
+      "num_input_tokens_seen": 100792792,
+      "step": 587
+    },
+    {
+      "epoch": 0.154402577760242,
+      "loss": 0.23411154747009277,
+      "loss_ce": 0.0037648691795766354,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 100792792,
+      "step": 587
+    },
+    {
+      "epoch": 0.15466561451962912,
+      "grad_norm": 7.806947443925601,
+      "learning_rate": 5e-06,
+      "loss": 0.1858,
+      "num_input_tokens_seen": 100964900,
+      "step": 588
+    },
+    {
+      "epoch": 0.15466561451962912,
+      "loss": 0.15801170468330383,
+      "loss_ce": 0.002433100016787648,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 100964900,
+      "step": 588
+    },
+    {
+      "epoch": 0.15492865127901626,
+      "grad_norm": 9.15652521254843,
+      "learning_rate": 5e-06,
+      "loss": 0.1918,
+      "num_input_tokens_seen": 101137068,
+      "step": 589
+    },
+    {
+      "epoch": 0.15492865127901626,
+      "loss": 0.14395104348659515,
+      "loss_ce": 0.0018611999694257975,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 101137068,
+      "step": 589
+    },
+    {
+      "epoch": 0.15519168803840336,
+      "grad_norm": 18.2857885682754,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 101309424,
+      "step": 590
+    },
+    {
+      "epoch": 0.15519168803840336,
+      "loss": 0.19831930100917816,
+      "loss_ce": 0.0009926356142386794,
+      "loss_iou": 0.546875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 101309424,
+      "step": 590
+    },
+    {
+      "epoch": 0.1554547247977905,
+      "grad_norm": 7.944412820463009,
+      "learning_rate": 5e-06,
+      "loss": 0.205,
+      "num_input_tokens_seen": 101478352,
+      "step": 591
+    },
+    {
+      "epoch": 0.1554547247977905,
+      "loss": 0.21606285870075226,
+      "loss_ce": 0.00683434447273612,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 101478352,
+      "step": 591
+    },
+    {
+      "epoch": 0.15571776155717762,
+      "grad_norm": 6.618575384086146,
+      "learning_rate": 5e-06,
+      "loss": 0.1821,
+      "num_input_tokens_seen": 101650324,
+      "step": 592
+    },
+    {
+      "epoch": 0.15571776155717762,
+      "loss": 0.20387296378612518,
+      "loss_ce": 0.0040438538417220116,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 101650324,
+      "step": 592
+    },
+    {
+      "epoch": 0.15598079831656475,
+      "grad_norm": 13.644949403576716,
+      "learning_rate": 5e-06,
+      "loss": 0.1791,
+      "num_input_tokens_seen": 101822564,
+      "step": 593
+    },
+    {
+      "epoch": 0.15598079831656475,
+      "loss": 0.12713779509067535,
+      "loss_ce": 0.004823335446417332,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 101822564,
+      "step": 593
+    },
+    {
+      "epoch": 0.15624383507595185,
+      "grad_norm": 11.597792275294081,
+      "learning_rate": 5e-06,
+      "loss": 0.2129,
+      "num_input_tokens_seen": 101994900,
+      "step": 594
+    },
+    {
+      "epoch": 0.15624383507595185,
+      "loss": 0.17677326500415802,
+      "loss_ce": 0.0048372335731983185,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 101994900,
+      "step": 594
+    },
+    {
+      "epoch": 0.15650687183533898,
+      "grad_norm": 7.4047027755949175,
+      "learning_rate": 5e-06,
+      "loss": 0.1867,
+      "num_input_tokens_seen": 102165488,
+      "step": 595
+    },
+    {
+      "epoch": 0.15650687183533898,
+      "loss": 0.1512867510318756,
+      "loss_ce": 0.0011402517557144165,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 102165488,
+      "step": 595
+    },
+    {
+      "epoch": 0.15676990859472612,
+      "grad_norm": 7.866562092019479,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 102337576,
+      "step": 596
+    },
+    {
+      "epoch": 0.15676990859472612,
+      "loss": 0.11417586356401443,
+      "loss_ce": 0.0016270325286313891,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 102337576,
+      "step": 596
+    },
+    {
+      "epoch": 0.15703294535411325,
+      "grad_norm": 14.250796858508032,
+      "learning_rate": 5e-06,
+      "loss": 0.2038,
+      "num_input_tokens_seen": 102509976,
+      "step": 597
+    },
+    {
+      "epoch": 0.15703294535411325,
+      "loss": 0.20375049114227295,
+      "loss_ce": 0.003677244298160076,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 102509976,
+      "step": 597
+    },
+    {
+      "epoch": 0.15729598211350035,
+      "grad_norm": 5.485478227776666,
+      "learning_rate": 5e-06,
+      "loss": 0.1839,
+      "num_input_tokens_seen": 102682292,
+      "step": 598
+    },
+    {
+      "epoch": 0.15729598211350035,
+      "loss": 0.1260145604610443,
+      "loss_ce": 0.002143711317330599,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 102682292,
+      "step": 598
+    },
+    {
+      "epoch": 0.15755901887288748,
+      "grad_norm": 16.90449659888107,
+      "learning_rate": 5e-06,
+      "loss": 0.1869,
+      "num_input_tokens_seen": 102854396,
+      "step": 599
+    },
+    {
+      "epoch": 0.15755901887288748,
+      "loss": 0.18277683854103088,
+      "loss_ce": 0.002356911078095436,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 102854396,
+      "step": 599
+    },
+    {
+      "epoch": 0.1578220556322746,
+      "grad_norm": 8.022384772643738,
+      "learning_rate": 5e-06,
+      "loss": 0.2261,
+      "num_input_tokens_seen": 103024668,
+      "step": 600
+    },
+    {
+      "epoch": 0.1578220556322746,
+      "loss": 0.2725946605205536,
+      "loss_ce": 0.0058710225857794285,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 103024668,
+      "step": 600
+    },
+    {
+      "epoch": 0.15808509239166174,
+      "grad_norm": 6.502303135299271,
+      "learning_rate": 5e-06,
+      "loss": 0.2163,
+      "num_input_tokens_seen": 103195160,
+      "step": 601
+    },
+    {
+      "epoch": 0.15808509239166174,
+      "loss": 0.16184811294078827,
+      "loss_ce": 0.003950160928070545,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 103195160,
+      "step": 601
+    },
+    {
+      "epoch": 0.15834812915104887,
+      "grad_norm": 11.6613104395809,
+      "learning_rate": 5e-06,
+      "loss": 0.1783,
+      "num_input_tokens_seen": 103365588,
+      "step": 602
+    },
+    {
+      "epoch": 0.15834812915104887,
+      "loss": 0.27217578887939453,
+      "loss_ce": 0.006367700640112162,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 103365588,
+      "step": 602
+    },
+    {
+      "epoch": 0.15861116591043598,
+      "grad_norm": 6.846414395262611,
+      "learning_rate": 5e-06,
+      "loss": 0.2057,
+      "num_input_tokens_seen": 103537444,
+      "step": 603
+    },
+    {
+      "epoch": 0.15861116591043598,
+      "loss": 0.24833053350448608,
+      "loss_ce": 0.004678180906921625,
+      "loss_iou": 0.625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 103537444,
+      "step": 603
+    },
+    {
+      "epoch": 0.1588742026698231,
+      "grad_norm": 5.663069385724008,
+      "learning_rate": 5e-06,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 103709932,
+      "step": 604
+    },
+    {
+      "epoch": 0.1588742026698231,
+      "loss": 0.19428852200508118,
+      "loss_ce": 0.000929144793190062,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 103709932,
+      "step": 604
+    },
+    {
+      "epoch": 0.15913723942921024,
+      "grad_norm": 6.358230492219544,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 103882084,
+      "step": 605
+    },
+    {
+      "epoch": 0.15913723942921024,
+      "loss": 0.19055569171905518,
+      "loss_ce": 0.005802282597869635,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 103882084,
+      "step": 605
+    },
+    {
+      "epoch": 0.15940027618859737,
+      "grad_norm": 8.229813023952058,
+      "learning_rate": 5e-06,
+      "loss": 0.194,
+      "num_input_tokens_seen": 104054236,
+      "step": 606
+    },
+    {
+      "epoch": 0.15940027618859737,
+      "loss": 0.14631760120391846,
+      "loss_ce": 0.0008098002290353179,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 104054236,
+      "step": 606
+    },
+    {
+      "epoch": 0.15966331294798447,
+      "grad_norm": 8.649522876210598,
+      "learning_rate": 5e-06,
+      "loss": 0.1982,
+      "num_input_tokens_seen": 104224580,
+      "step": 607
+    },
+    {
+      "epoch": 0.15966331294798447,
+      "loss": 0.2813430428504944,
+      "loss_ce": 0.0039382753893733025,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 104224580,
+      "step": 607
+    },
+    {
+      "epoch": 0.1599263497073716,
+      "grad_norm": 9.061064934100147,
+      "learning_rate": 5e-06,
+      "loss": 0.2389,
+      "num_input_tokens_seen": 104396656,
+      "step": 608
+    },
+    {
+      "epoch": 0.1599263497073716,
+      "loss": 0.25585615634918213,
+      "loss_ce": 0.005581488832831383,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 104396656,
+      "step": 608
+    },
+    {
+      "epoch": 0.16018938646675873,
+      "grad_norm": 10.782439869738488,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "num_input_tokens_seen": 104568804,
+      "step": 609
+    },
+    {
+      "epoch": 0.16018938646675873,
+      "loss": 0.22171396017074585,
+      "loss_ce": 0.0010718655539676547,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 104568804,
+      "step": 609
+    },
+    {
+      "epoch": 0.16045242322614586,
+      "grad_norm": 7.912260836349583,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 104740976,
+      "step": 610
+    },
+    {
+      "epoch": 0.16045242322614586,
+      "loss": 0.20210719108581543,
+      "loss_ce": 0.002094991272315383,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 104740976,
+      "step": 610
+    },
+    {
+      "epoch": 0.16071545998553297,
+      "grad_norm": 7.072100604468161,
+      "learning_rate": 5e-06,
+      "loss": 0.1877,
+      "num_input_tokens_seen": 104913120,
+      "step": 611
+    },
+    {
+      "epoch": 0.16071545998553297,
+      "loss": 0.24669376015663147,
+      "loss_ce": 0.005971122998744249,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 104913120,
+      "step": 611
+    },
+    {
+      "epoch": 0.1609784967449201,
+      "grad_norm": 6.775986925165555,
+      "learning_rate": 5e-06,
+      "loss": 0.2033,
+      "num_input_tokens_seen": 105085268,
+      "step": 612
+    },
+    {
+      "epoch": 0.1609784967449201,
+      "loss": 0.17514100670814514,
+      "loss_ce": 0.005036028102040291,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 105085268,
+      "step": 612
+    },
+    {
+      "epoch": 0.16124153350430723,
+      "grad_norm": 13.436082189650097,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 105255536,
+      "step": 613
+    },
+    {
+      "epoch": 0.16124153350430723,
+      "loss": 0.1662948876619339,
+      "loss_ce": 0.004856906831264496,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 105255536,
+      "step": 613
+    },
+    {
+      "epoch": 0.16150457026369436,
+      "grad_norm": 15.212823710443937,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 105426276,
+      "step": 614
+    },
+    {
+      "epoch": 0.16150457026369436,
+      "loss": 0.21024103462696075,
+      "loss_ce": 0.000707346829585731,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 105426276,
+      "step": 614
+    },
+    {
+      "epoch": 0.1617676070230815,
+      "grad_norm": 5.811817664832679,
+      "learning_rate": 5e-06,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 105598528,
+      "step": 615
+    },
+    {
+      "epoch": 0.1617676070230815,
+      "loss": 0.17331555485725403,
+      "loss_ce": 0.0019593401812016964,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 105598528,
+      "step": 615
+    },
+    {
+      "epoch": 0.1620306437824686,
+      "grad_norm": 7.929569482892307,
+      "learning_rate": 5e-06,
+      "loss": 0.1901,
+      "num_input_tokens_seen": 105768684,
+      "step": 616
+    },
+    {
+      "epoch": 0.1620306437824686,
+      "loss": 0.22191157937049866,
+      "loss_ce": 0.0013305249158293009,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 105768684,
+      "step": 616
+    },
+    {
+      "epoch": 0.16229368054185572,
+      "grad_norm": 9.722668759038529,
+      "learning_rate": 5e-06,
+      "loss": 0.238,
+      "num_input_tokens_seen": 105938984,
+      "step": 617
+    },
+    {
+      "epoch": 0.16229368054185572,
+      "loss": 0.2918306887149811,
+      "loss_ce": 0.004660272039473057,
+      "loss_iou": 0.76953125,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 105938984,
+      "step": 617
+    },
+    {
+      "epoch": 0.16255671730124285,
+      "grad_norm": 8.495376084254536,
+      "learning_rate": 5e-06,
+      "loss": 0.2003,
+      "num_input_tokens_seen": 106111216,
+      "step": 618
+    },
+    {
+      "epoch": 0.16255671730124285,
+      "loss": 0.13662417232990265,
+      "loss_ce": 0.0008209550869651139,
+      "loss_iou": 0.46875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 106111216,
+      "step": 618
+    },
+    {
+      "epoch": 0.16281975406062998,
+      "grad_norm": 10.294809858036002,
+      "learning_rate": 5e-06,
+      "loss": 0.2276,
+      "num_input_tokens_seen": 106283444,
+      "step": 619
+    },
+    {
+      "epoch": 0.16281975406062998,
+      "loss": 0.23945499956607819,
+      "loss_ce": 0.0016620358219370246,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 106283444,
+      "step": 619
+    },
+    {
+      "epoch": 0.1630827908200171,
+      "grad_norm": 8.591329712303546,
+      "learning_rate": 5e-06,
+      "loss": 0.1832,
+      "num_input_tokens_seen": 106455556,
+      "step": 620
+    },
+    {
+      "epoch": 0.1630827908200171,
+      "loss": 0.24216794967651367,
+      "loss_ce": 0.004680164158344269,
+      "loss_iou": 0.703125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 106455556,
+      "step": 620
+    },
+    {
+      "epoch": 0.16334582757940422,
+      "grad_norm": 11.580746408707443,
+      "learning_rate": 5e-06,
+      "loss": 0.222,
+      "num_input_tokens_seen": 106625564,
+      "step": 621
+    },
+    {
+      "epoch": 0.16334582757940422,
+      "loss": 0.27062344551086426,
+      "loss_ce": 0.0013363163452595472,
+      "loss_iou": 0.625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 106625564,
+      "step": 621
+    },
+    {
+      "epoch": 0.16360886433879135,
+      "grad_norm": 9.075656365353165,
+      "learning_rate": 5e-06,
+      "loss": 0.2241,
+      "num_input_tokens_seen": 106797500,
+      "step": 622
+    },
+    {
+      "epoch": 0.16360886433879135,
+      "loss": 0.25023964047431946,
+      "loss_ce": 0.002986219245940447,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 106797500,
+      "step": 622
+    },
+    {
+      "epoch": 0.16387190109817848,
+      "grad_norm": 7.173557460385501,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 106969624,
+      "step": 623
+    },
+    {
+      "epoch": 0.16387190109817848,
+      "loss": 0.15543386340141296,
+      "loss_ce": 0.004310814663767815,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 106969624,
+      "step": 623
+    },
+    {
+      "epoch": 0.16413493785756558,
+      "grad_norm": 7.146771956956073,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 107139936,
+      "step": 624
+    },
+    {
+      "epoch": 0.16413493785756558,
+      "loss": 0.2210107445716858,
+      "loss_ce": 0.0020776439923793077,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 107139936,
+      "step": 624
+    },
+    {
+      "epoch": 0.1643979746169527,
+      "grad_norm": 7.784232141951156,
+      "learning_rate": 5e-06,
+      "loss": 0.1852,
+      "num_input_tokens_seen": 107312016,
+      "step": 625
+    },
+    {
+      "epoch": 0.1643979746169527,
+      "loss": 0.16655078530311584,
+      "loss_ce": 0.00200000312179327,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 107312016,
+      "step": 625
+    },
+    {
+      "epoch": 0.16466101137633984,
+      "grad_norm": 9.666585010057217,
+      "learning_rate": 5e-06,
+      "loss": 0.2351,
+      "num_input_tokens_seen": 107484484,
+      "step": 626
+    },
+    {
+      "epoch": 0.16466101137633984,
+      "loss": 0.17668788135051727,
+      "loss_ce": 0.0009676595800556242,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 107484484,
+      "step": 626
+    },
+    {
+      "epoch": 0.16492404813572697,
+      "grad_norm": 15.599284961255806,
+      "learning_rate": 5e-06,
+      "loss": 0.1943,
+      "num_input_tokens_seen": 107656148,
+      "step": 627
+    },
+    {
+      "epoch": 0.16492404813572697,
+      "loss": 0.2105821818113327,
+      "loss_ce": 0.0011095235822722316,
+      "loss_iou": 0.5,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 107656148,
+      "step": 627
+    },
+    {
+      "epoch": 0.1651870848951141,
+      "grad_norm": 8.248204010218137,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 107828348,
+      "step": 628
+    },
+    {
+      "epoch": 0.1651870848951141,
+      "loss": 0.1219111904501915,
+      "loss_ce": 0.009545465931296349,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 107828348,
+      "step": 628
+    },
+    {
+      "epoch": 0.1654501216545012,
+      "grad_norm": 10.643869540803188,
+      "learning_rate": 5e-06,
+      "loss": 0.191,
+      "num_input_tokens_seen": 108000340,
+      "step": 629
+    },
+    {
+      "epoch": 0.1654501216545012,
+      "loss": 0.21501825749874115,
+      "loss_ce": 0.0015172738349065185,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 108000340,
+      "step": 629
+    },
+    {
+      "epoch": 0.16571315841388834,
+      "grad_norm": 6.957892002207251,
+      "learning_rate": 5e-06,
+      "loss": 0.1723,
+      "num_input_tokens_seen": 108172648,
+      "step": 630
+    },
+    {
+      "epoch": 0.16571315841388834,
+      "loss": 0.17599767446517944,
+      "loss_ce": 0.0010709069902077317,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 108172648,
+      "step": 630
+    },
+    {
+      "epoch": 0.16597619517327547,
+      "grad_norm": 8.642857572938437,
+      "learning_rate": 5e-06,
+      "loss": 0.2342,
+      "num_input_tokens_seen": 108343044,
+      "step": 631
+    },
+    {
+      "epoch": 0.16597619517327547,
+      "loss": 0.18351054191589355,
+      "loss_ce": 0.00296853668987751,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 108343044,
+      "step": 631
+    },
+    {
+      "epoch": 0.1662392319326626,
+      "grad_norm": 7.980653444631839,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 108515048,
+      "step": 632
+    },
+    {
+      "epoch": 0.1662392319326626,
+      "loss": 0.1818259358406067,
+      "loss_ce": 0.008302995935082436,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 108515048,
+      "step": 632
+    },
+    {
+      "epoch": 0.1665022686920497,
+      "grad_norm": 7.668823525125559,
+      "learning_rate": 5e-06,
+      "loss": 0.1603,
+      "num_input_tokens_seen": 108687320,
+      "step": 633
+    },
+    {
+      "epoch": 0.1665022686920497,
+      "loss": 0.2121184766292572,
+      "loss_ce": 0.0024016951210796833,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 108687320,
+      "step": 633
+    },
+    {
+      "epoch": 0.16676530545143683,
+      "grad_norm": 8.952178505337718,
+      "learning_rate": 5e-06,
+      "loss": 0.2244,
+      "num_input_tokens_seen": 108859852,
+      "step": 634
+    },
+    {
+      "epoch": 0.16676530545143683,
+      "loss": 0.24805203080177307,
+      "loss_ce": 0.004399674944579601,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 108859852,
+      "step": 634
+    },
+    {
+      "epoch": 0.16702834221082397,
+      "grad_norm": 8.269015730220213,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 109031824,
+      "step": 635
+    },
+    {
+      "epoch": 0.16702834221082397,
+      "loss": 0.3403623104095459,
+      "loss_ce": 0.01095556654036045,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 109031824,
+      "step": 635
+    },
+    {
+      "epoch": 0.1672913789702111,
+      "grad_norm": 11.588758914669935,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 109204100,
+      "step": 636
+    },
+    {
+      "epoch": 0.1672913789702111,
+      "loss": 0.25584501028060913,
+      "loss_ce": 0.007065705489367247,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 109204100,
+      "step": 636
+    },
+    {
+      "epoch": 0.1675544157295982,
+      "grad_norm": 9.273803296838299,
+      "learning_rate": 5e-06,
+      "loss": 0.203,
+      "num_input_tokens_seen": 109376120,
+      "step": 637
+    },
+    {
+      "epoch": 0.1675544157295982,
+      "loss": 0.24991419911384583,
+      "loss_ce": 0.0015011176001280546,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 109376120,
+      "step": 637
+    },
+    {
+      "epoch": 0.16781745248898533,
+      "grad_norm": 5.629607113211884,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 109548116,
+      "step": 638
+    },
+    {
+      "epoch": 0.16781745248898533,
+      "loss": 0.22593827545642853,
+      "loss_ce": 0.005753945559263229,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 109548116,
+      "step": 638
+    },
+    {
+      "epoch": 0.16808048924837246,
+      "grad_norm": 10.827873954115349,
+      "learning_rate": 5e-06,
+      "loss": 0.154,
+      "num_input_tokens_seen": 109720360,
+      "step": 639
+    },
+    {
+      "epoch": 0.16808048924837246,
+      "loss": 0.10586154460906982,
+      "loss_ce": 0.0011862462852150202,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 109720360,
+      "step": 639
+    },
+    {
+      "epoch": 0.1683435260077596,
+      "grad_norm": 6.979104215801615,
+      "learning_rate": 5e-06,
+      "loss": 0.1811,
+      "num_input_tokens_seen": 109892700,
+      "step": 640
+    },
+    {
+      "epoch": 0.1683435260077596,
+      "loss": 0.21827656030654907,
+      "loss_ce": 0.0010524489916861057,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 109892700,
+      "step": 640
+    },
+    {
+      "epoch": 0.1686065627671467,
+      "grad_norm": 12.485157254065346,
+      "learning_rate": 5e-06,
+      "loss": 0.2113,
+      "num_input_tokens_seen": 110064992,
+      "step": 641
+    },
+    {
+      "epoch": 0.1686065627671467,
+      "loss": 0.26375001668930054,
+      "loss_ce": 0.0014819505158811808,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 110064992,
+      "step": 641
+    },
+    {
+      "epoch": 0.16886959952653383,
+      "grad_norm": 9.158584740167319,
+      "learning_rate": 5e-06,
+      "loss": 0.1418,
+      "num_input_tokens_seen": 110237352,
+      "step": 642
+    },
+    {
+      "epoch": 0.16886959952653383,
+      "loss": 0.12861773371696472,
+      "loss_ce": 0.006120163947343826,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 110237352,
+      "step": 642
+    },
+    {
+      "epoch": 0.16913263628592096,
+      "grad_norm": 12.50685540012725,
+      "learning_rate": 5e-06,
+      "loss": 0.226,
+      "num_input_tokens_seen": 110409596,
+      "step": 643
+    },
+    {
+      "epoch": 0.16913263628592096,
+      "loss": 0.29422521591186523,
+      "loss_ce": 0.006322397850453854,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 110409596,
+      "step": 643
+    },
+    {
+      "epoch": 0.1693956730453081,
+      "grad_norm": 8.91219200815607,
+      "learning_rate": 5e-06,
+      "loss": 0.1801,
+      "num_input_tokens_seen": 110581872,
+      "step": 644
+    },
+    {
+      "epoch": 0.1693956730453081,
+      "loss": 0.19362246990203857,
+      "loss_ce": 0.0011175863910466433,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 110581872,
+      "step": 644
+    },
+    {
+      "epoch": 0.16965870980469522,
+      "grad_norm": 6.648045150838095,
+      "learning_rate": 5e-06,
+      "loss": 0.1817,
+      "num_input_tokens_seen": 110754112,
+      "step": 645
+    },
+    {
+      "epoch": 0.16965870980469522,
+      "loss": 0.22110968828201294,
+      "loss_ce": 0.0007727851625531912,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 110754112,
+      "step": 645
+    },
+    {
+      "epoch": 0.16992174656408232,
+      "grad_norm": 5.685997850783031,
+      "learning_rate": 5e-06,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 110926320,
+      "step": 646
+    },
+    {
+      "epoch": 0.16992174656408232,
+      "loss": 0.16139136254787445,
+      "loss_ce": 0.0022421882022172213,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 110926320,
+      "step": 646
+    },
+    {
+      "epoch": 0.17018478332346945,
+      "grad_norm": 5.619576982998229,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 111098480,
+      "step": 647
+    },
+    {
+      "epoch": 0.17018478332346945,
+      "loss": 0.07204495370388031,
+      "loss_ce": 0.0005727877141907811,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 111098480,
+      "step": 647
+    },
+    {
+      "epoch": 0.17044782008285658,
+      "grad_norm": 7.873961542866977,
+      "learning_rate": 5e-06,
+      "loss": 0.2025,
+      "num_input_tokens_seen": 111270660,
+      "step": 648
+    },
+    {
+      "epoch": 0.17044782008285658,
+      "loss": 0.2468957006931305,
+      "loss_ce": 0.001473332871682942,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 111270660,
+      "step": 648
+    },
+    {
+      "epoch": 0.1707108568422437,
+      "grad_norm": 10.015757077433259,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 111442504,
+      "step": 649
+    },
+    {
+      "epoch": 0.1707108568422437,
+      "loss": 0.14880256354808807,
+      "loss_ce": 0.0109546585008502,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 111442504,
+      "step": 649
+    },
+    {
+      "epoch": 0.17097389360163082,
+      "grad_norm": 7.899725217893395,
+      "learning_rate": 5e-06,
+      "loss": 0.2213,
+      "num_input_tokens_seen": 111614612,
+      "step": 650
+    },
+    {
+      "epoch": 0.17097389360163082,
+      "loss": 0.31617793440818787,
+      "loss_ce": 0.005386924371123314,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 111614612,
+      "step": 650
+    },
+    {
+      "epoch": 0.17123693036101795,
+      "grad_norm": 11.203783050737329,
+      "learning_rate": 5e-06,
+      "loss": 0.2011,
+      "num_input_tokens_seen": 111786832,
+      "step": 651
+    },
+    {
+      "epoch": 0.17123693036101795,
+      "loss": 0.16059955954551697,
+      "loss_ce": 0.003617130685597658,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 111786832,
+      "step": 651
+    },
+    {
+      "epoch": 0.17149996712040508,
+      "grad_norm": 10.553074580421196,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 111959112,
+      "step": 652
+    },
+    {
+      "epoch": 0.17149996712040508,
+      "loss": 0.10537019371986389,
+      "loss_ce": 0.0024038811679929495,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 111959112,
+      "step": 652
+    },
+    {
+      "epoch": 0.1717630038797922,
+      "grad_norm": 7.878994667708907,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 112129472,
+      "step": 653
+    },
+    {
+      "epoch": 0.1717630038797922,
+      "loss": 0.14291326701641083,
+      "loss_ce": 0.0017999822739511728,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 112129472,
+      "step": 653
+    },
+    {
+      "epoch": 0.1720260406391793,
+      "grad_norm": 8.502469919546648,
+      "learning_rate": 5e-06,
+      "loss": 0.2067,
+      "num_input_tokens_seen": 112301764,
+      "step": 654
+    },
+    {
+      "epoch": 0.1720260406391793,
+      "loss": 0.14102406799793243,
+      "loss_ce": 0.0035118628293275833,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 112301764,
+      "step": 654
+    },
+    {
+      "epoch": 0.17228907739856644,
+      "grad_norm": 7.248102860819298,
+      "learning_rate": 5e-06,
+      "loss": 0.1947,
+      "num_input_tokens_seen": 112473728,
+      "step": 655
+    },
+    {
+      "epoch": 0.17228907739856644,
+      "loss": 0.3125525116920471,
+      "loss_ce": 0.002799113281071186,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 112473728,
+      "step": 655
+    },
+    {
+      "epoch": 0.17255211415795357,
+      "grad_norm": 7.514690707459716,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 112646020,
+      "step": 656
+    },
+    {
+      "epoch": 0.17255211415795357,
+      "loss": 0.19242502748966217,
+      "loss_ce": 0.0032160417176783085,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 112646020,
+      "step": 656
+    },
+    {
+      "epoch": 0.1728151509173407,
+      "grad_norm": 12.157018929724885,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 112818344,
+      "step": 657
+    },
+    {
+      "epoch": 0.1728151509173407,
+      "loss": 0.17007869482040405,
+      "loss_ce": 0.0014995899982750416,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 112818344,
+      "step": 657
+    },
+    {
+      "epoch": 0.17307818767672783,
+      "grad_norm": 12.713612456980975,
+      "learning_rate": 5e-06,
+      "loss": 0.1854,
+      "num_input_tokens_seen": 112984384,
+      "step": 658
+    },
+    {
+      "epoch": 0.17307818767672783,
+      "loss": 0.2311791479587555,
+      "loss_ce": 0.003090762998908758,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 112984384,
+      "step": 658
+    },
+    {
+      "epoch": 0.17334122443611494,
+      "grad_norm": 7.9869346345765235,
+      "learning_rate": 5e-06,
+      "loss": 0.1714,
+      "num_input_tokens_seen": 113156440,
+      "step": 659
+    },
+    {
+      "epoch": 0.17334122443611494,
+      "loss": 0.14270631968975067,
+      "loss_ce": 0.007757591083645821,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 113156440,
+      "step": 659
+    },
+    {
+      "epoch": 0.17360426119550207,
+      "grad_norm": 19.43755709327552,
+      "learning_rate": 5e-06,
+      "loss": 0.2275,
+      "num_input_tokens_seen": 113328684,
+      "step": 660
+    },
+    {
+      "epoch": 0.17360426119550207,
+      "loss": 0.19741018116474152,
+      "loss_ce": 0.023887230083346367,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 113328684,
+      "step": 660
+    },
+    {
+      "epoch": 0.1738672979548892,
+      "grad_norm": 12.233621879837852,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 113499120,
+      "step": 661
+    },
+    {
+      "epoch": 0.1738672979548892,
+      "loss": 0.17290905117988586,
+      "loss_ce": 0.03179576247930527,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 113499120,
+      "step": 661
+    },
+    {
+      "epoch": 0.17413033471427633,
+      "grad_norm": 8.034436568495584,
+      "learning_rate": 5e-06,
+      "loss": 0.2288,
+      "num_input_tokens_seen": 113670976,
+      "step": 662
+    },
+    {
+      "epoch": 0.17413033471427633,
+      "loss": 0.18118935823440552,
+      "loss_ce": 0.0038822239730507135,
+      "loss_iou": 0.5625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 113670976,
+      "step": 662
+    },
+    {
+      "epoch": 0.17439337147366343,
+      "grad_norm": 6.838416843124344,
+      "learning_rate": 5e-06,
+      "loss": 0.211,
+      "num_input_tokens_seen": 113841352,
+      "step": 663
+    },
+    {
+      "epoch": 0.17439337147366343,
+      "loss": 0.16460734605789185,
+      "loss_ce": 0.0039017903618514538,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 113841352,
+      "step": 663
+    },
+    {
+      "epoch": 0.17465640823305056,
+      "grad_norm": 8.007803887726045,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 114013720,
+      "step": 664
+    },
+    {
+      "epoch": 0.17465640823305056,
+      "loss": 0.19987425208091736,
+      "loss_ce": 0.004256566055119038,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 114013720,
+      "step": 664
+    },
+    {
+      "epoch": 0.1749194449924377,
+      "grad_norm": 15.416941151295424,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 114185964,
+      "step": 665
+    },
+    {
+      "epoch": 0.1749194449924377,
+      "loss": 0.12474516034126282,
+      "loss_ce": 0.0013320783618837595,
+      "loss_iou": 0.53125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 114185964,
+      "step": 665
+    },
+    {
+      "epoch": 0.17518248175182483,
+      "grad_norm": 7.3780372763976105,
+      "learning_rate": 5e-06,
+      "loss": 0.1978,
+      "num_input_tokens_seen": 114354848,
+      "step": 666
+    },
+    {
+      "epoch": 0.17518248175182483,
+      "loss": 0.22682063281536102,
+      "loss_ce": 0.002394365146756172,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 114354848,
+      "step": 666
+    },
+    {
+      "epoch": 0.17544551851121193,
+      "grad_norm": 6.805100830892847,
+      "learning_rate": 5e-06,
+      "loss": 0.1893,
+      "num_input_tokens_seen": 114526920,
+      "step": 667
+    },
+    {
+      "epoch": 0.17544551851121193,
+      "loss": 0.16719527542591095,
+      "loss_ce": 0.003987260162830353,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 114526920,
+      "step": 667
+    },
+    {
+      "epoch": 0.17570855527059906,
+      "grad_norm": 13.214372039286259,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 114698988,
+      "step": 668
+    },
+    {
+      "epoch": 0.17570855527059906,
+      "loss": 0.11197628825902939,
+      "loss_ce": 0.0018078316934406757,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 114698988,
+      "step": 668
+    },
+    {
+      "epoch": 0.1759715920299862,
+      "grad_norm": 11.38353136981813,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 114871164,
+      "step": 669
+    },
+    {
+      "epoch": 0.1759715920299862,
+      "loss": 0.13217589259147644,
+      "loss_ce": 0.0022930747363716364,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 114871164,
+      "step": 669
+    },
+    {
+      "epoch": 0.17623462878937332,
+      "grad_norm": 6.3416006325736705,
+      "learning_rate": 5e-06,
+      "loss": 0.1994,
+      "num_input_tokens_seen": 115043272,
+      "step": 670
+    },
+    {
+      "epoch": 0.17623462878937332,
+      "loss": 0.2425597459077835,
+      "loss_ce": 0.00726923206821084,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 115043272,
+      "step": 670
+    },
+    {
+      "epoch": 0.17649766554876045,
+      "grad_norm": 7.208802786478081,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 115215332,
+      "step": 671
+    },
+    {
+      "epoch": 0.17649766554876045,
+      "loss": 0.2342541515827179,
+      "loss_ce": 0.006531976629048586,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 115215332,
+      "step": 671
+    },
+    {
+      "epoch": 0.17676070230814755,
+      "grad_norm": 5.677644761077757,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 115387420,
+      "step": 672
+    },
+    {
+      "epoch": 0.17676070230814755,
+      "loss": 0.1173291727900505,
+      "loss_ce": 0.0031934345606714487,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 115387420,
+      "step": 672
+    },
+    {
+      "epoch": 0.17702373906753469,
+      "grad_norm": 13.804747880548247,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 115559764,
+      "step": 673
+    },
+    {
+      "epoch": 0.17702373906753469,
+      "loss": 0.11581701785326004,
+      "loss_ce": 0.0009488522773608565,
+      "loss_iou": 0.546875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 115559764,
+      "step": 673
+    },
+    {
+      "epoch": 0.17728677582692182,
+      "grad_norm": 7.7796454748869435,
+      "learning_rate": 5e-06,
+      "loss": 0.2202,
+      "num_input_tokens_seen": 115731572,
+      "step": 674
+    },
+    {
+      "epoch": 0.17728677582692182,
+      "loss": 0.22158196568489075,
+      "loss_ce": 0.00228265137411654,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 115731572,
+      "step": 674
+    },
+    {
+      "epoch": 0.17754981258630895,
+      "grad_norm": 7.463691885246712,
+      "learning_rate": 5e-06,
+      "loss": 0.1882,
+      "num_input_tokens_seen": 115903680,
+      "step": 675
+    },
+    {
+      "epoch": 0.17754981258630895,
+      "loss": 0.22185131907463074,
+      "loss_ce": 0.002002692548558116,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 115903680,
+      "step": 675
+    },
+    {
+      "epoch": 0.17781284934569605,
+      "grad_norm": 8.426002860661324,
+      "learning_rate": 5e-06,
+      "loss": 0.1793,
+      "num_input_tokens_seen": 116075520,
+      "step": 676
+    },
+    {
+      "epoch": 0.17781284934569605,
+      "loss": 0.1606462597846985,
+      "loss_ce": 0.0009782931301742792,
+      "loss_iou": 0.5625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 116075520,
+      "step": 676
+    },
+    {
+      "epoch": 0.17807588610508318,
+      "grad_norm": 8.301724485985723,
+      "learning_rate": 5e-06,
+      "loss": 0.2,
+      "num_input_tokens_seen": 116243620,
+      "step": 677
+    },
+    {
+      "epoch": 0.17807588610508318,
+      "loss": 0.23937323689460754,
+      "loss_ce": 0.002251650206744671,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 116243620,
+      "step": 677
+    },
+    {
+      "epoch": 0.1783389228644703,
+      "grad_norm": 8.16746221218087,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 116414372,
+      "step": 678
+    },
+    {
+      "epoch": 0.1783389228644703,
+      "loss": 0.2846035361289978,
+      "loss_ce": 0.003506140550598502,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 116414372,
+      "step": 678
+    },
+    {
+      "epoch": 0.17860195962385744,
+      "grad_norm": 11.233813688882323,
+      "learning_rate": 5e-06,
+      "loss": 0.2052,
+      "num_input_tokens_seen": 116586316,
+      "step": 679
+    },
+    {
+      "epoch": 0.17860195962385744,
+      "loss": 0.19775235652923584,
+      "loss_ce": 0.0026840060018002987,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 116586316,
+      "step": 679
+    },
+    {
+      "epoch": 0.17886499638324455,
+      "grad_norm": 7.134240923845786,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 116758540,
+      "step": 680
+    },
+    {
+      "epoch": 0.17886499638324455,
+      "loss": 0.17349669337272644,
+      "loss_ce": 0.0008892616024240851,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 116758540,
+      "step": 680
+    },
+    {
+      "epoch": 0.17912803314263168,
+      "grad_norm": 8.457982820816902,
+      "learning_rate": 5e-06,
+      "loss": 0.2133,
+      "num_input_tokens_seen": 116930944,
+      "step": 681
+    },
+    {
+      "epoch": 0.17912803314263168,
+      "loss": 0.2098885327577591,
+      "loss_ce": 0.0027962373569607735,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 116930944,
+      "step": 681
+    },
+    {
+      "epoch": 0.1793910699020188,
+      "grad_norm": 8.07197838986281,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 117102756,
+      "step": 682
+    },
+    {
+      "epoch": 0.1793910699020188,
+      "loss": 0.18535807728767395,
+      "loss_ce": 0.003778480924665928,
+      "loss_iou": 0.578125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 117102756,
+      "step": 682
+    },
+    {
+      "epoch": 0.17965410666140594,
+      "grad_norm": 7.799168173540665,
+      "learning_rate": 5e-06,
+      "loss": 0.2208,
+      "num_input_tokens_seen": 117274704,
+      "step": 683
+    },
+    {
+      "epoch": 0.17965410666140594,
+      "loss": 0.22785347700119019,
+      "loss_ce": 0.0029999692924320698,
+      "loss_iou": 0.3125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 117274704,
+      "step": 683
+    },
+    {
+      "epoch": 0.17991714342079307,
+      "grad_norm": 6.357635509161791,
+      "learning_rate": 5e-06,
+      "loss": 0.1508,
+      "num_input_tokens_seen": 117446896,
+      "step": 684
+    },
+    {
+      "epoch": 0.17991714342079307,
+      "loss": 0.2038782835006714,
+      "loss_ce": 0.010396835394203663,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 117446896,
+      "step": 684
+    },
+    {
+      "epoch": 0.18018018018018017,
+      "grad_norm": 7.961998959961417,
+      "learning_rate": 5e-06,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 117618944,
+      "step": 685
+    },
+    {
+      "epoch": 0.18018018018018017,
+      "loss": 0.18519067764282227,
+      "loss_ce": 0.0017190101789310575,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 117618944,
+      "step": 685
+    },
+    {
+      "epoch": 0.1804432169395673,
+      "grad_norm": 9.08254127907995,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 117791316,
+      "step": 686
+    },
+    {
+      "epoch": 0.1804432169395673,
+      "loss": 0.20686465501785278,
+      "loss_ce": 0.0025189572479575872,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 117791316,
+      "step": 686
+    },
+    {
+      "epoch": 0.18070625369895443,
+      "grad_norm": 10.367656958651155,
+      "learning_rate": 5e-06,
+      "loss": 0.1855,
+      "num_input_tokens_seen": 117963308,
+      "step": 687
+    },
+    {
+      "epoch": 0.18070625369895443,
+      "loss": 0.13359057903289795,
+      "loss_ce": 0.0041044931858778,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 117963308,
+      "step": 687
+    },
+    {
+      "epoch": 0.18096929045834156,
+      "grad_norm": 7.549954240364692,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 118135556,
+      "step": 688
+    },
+    {
+      "epoch": 0.18096929045834156,
+      "loss": 0.14778929948806763,
+      "loss_ce": 0.0011828583665192127,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 118135556,
+      "step": 688
+    },
+    {
+      "epoch": 0.18123232721772867,
+      "grad_norm": 9.741300529319401,
+      "learning_rate": 5e-06,
+      "loss": 0.166,
+      "num_input_tokens_seen": 118305884,
+      "step": 689
+    },
+    {
+      "epoch": 0.18123232721772867,
+      "loss": 0.1753019541501999,
+      "loss_ce": 0.003182805608958006,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 118305884,
+      "step": 689
+    },
+    {
+      "epoch": 0.1814953639771158,
+      "grad_norm": 11.31734919561229,
+      "learning_rate": 5e-06,
+      "loss": 0.1811,
+      "num_input_tokens_seen": 118478000,
+      "step": 690
+    },
+    {
+      "epoch": 0.1814953639771158,
+      "loss": 0.189756840467453,
+      "loss_ce": 0.0007919906638562679,
+      "loss_iou": 0.77734375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 118478000,
+      "step": 690
+    },
+    {
+      "epoch": 0.18175840073650293,
+      "grad_norm": 9.003341647966222,
+      "learning_rate": 5e-06,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 118650436,
+      "step": 691
+    },
+    {
+      "epoch": 0.18175840073650293,
+      "loss": 0.11313121020793915,
+      "loss_ce": 0.0006434204406104982,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 118650436,
+      "step": 691
+    },
+    {
+      "epoch": 0.18202143749589006,
+      "grad_norm": 15.98200660549263,
+      "learning_rate": 5e-06,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 118822416,
+      "step": 692
+    },
+    {
+      "epoch": 0.18202143749589006,
+      "loss": 0.14253322780132294,
+      "loss_ce": 0.001908229780383408,
+      "loss_iou": 0.453125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 118822416,
+      "step": 692
+    },
+    {
+      "epoch": 0.18228447425527716,
+      "grad_norm": 5.776705438700134,
+      "learning_rate": 5e-06,
+      "loss": 0.1588,
+      "num_input_tokens_seen": 118994492,
+      "step": 693
+    },
+    {
+      "epoch": 0.18228447425527716,
+      "loss": 0.15580043196678162,
+      "loss_ce": 0.0014730504481121898,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 118994492,
+      "step": 693
+    },
+    {
+      "epoch": 0.1825475110146643,
+      "grad_norm": 13.97214493017941,
+      "learning_rate": 5e-06,
+      "loss": 0.1825,
+      "num_input_tokens_seen": 119166828,
+      "step": 694
+    },
+    {
+      "epoch": 0.1825475110146643,
+      "loss": 0.14117339253425598,
+      "loss_ce": 0.005309135653078556,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 119166828,
+      "step": 694
+    },
+    {
+      "epoch": 0.18281054777405142,
+      "grad_norm": 9.340996582360178,
+      "learning_rate": 5e-06,
+      "loss": 0.2039,
+      "num_input_tokens_seen": 119339040,
+      "step": 695
+    },
+    {
+      "epoch": 0.18281054777405142,
+      "loss": 0.21793845295906067,
+      "loss_ce": 0.005658184178173542,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 119339040,
+      "step": 695
+    },
+    {
+      "epoch": 0.18307358453343855,
+      "grad_norm": 6.679514738557093,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 119511024,
+      "step": 696
+    },
+    {
+      "epoch": 0.18307358453343855,
+      "loss": 0.18582630157470703,
+      "loss_ce": 0.0012559981551021338,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 119511024,
+      "step": 696
+    },
+    {
+      "epoch": 0.18333662129282569,
+      "grad_norm": 5.326914153992354,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 119683564,
+      "step": 697
+    },
+    {
+      "epoch": 0.18333662129282569,
+      "loss": 0.15467330813407898,
+      "loss_ce": 0.0042826831340789795,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 119683564,
+      "step": 697
+    },
+    {
+      "epoch": 0.1835996580522128,
+      "grad_norm": 8.515559517433303,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 119856000,
+      "step": 698
+    },
+    {
+      "epoch": 0.1835996580522128,
+      "loss": 0.12481513619422913,
+      "loss_ce": 0.0006696260534226894,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 119856000,
+      "step": 698
+    },
+    {
+      "epoch": 0.18386269481159992,
+      "grad_norm": 9.420262409376758,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 120028676,
+      "step": 699
+    },
+    {
+      "epoch": 0.18386269481159992,
+      "loss": 0.22470733523368835,
+      "loss_ce": 0.001868000952526927,
+      "loss_iou": 0.6875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 120028676,
+      "step": 699
+    },
+    {
+      "epoch": 0.18412573157098705,
+      "grad_norm": 7.530169678535974,
+      "learning_rate": 5e-06,
+      "loss": 0.1624,
+      "num_input_tokens_seen": 120201208,
+      "step": 700
+    },
+    {
+      "epoch": 0.18412573157098705,
+      "loss": 0.10793297737836838,
+      "loss_ce": 0.0036849307361990213,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 120201208,
+      "step": 700
+    },
+    {
+      "epoch": 0.18438876833037418,
+      "grad_norm": 8.205415154202601,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 120373444,
+      "step": 701
+    },
+    {
+      "epoch": 0.18438876833037418,
+      "loss": 0.1353437304496765,
+      "loss_ce": 0.001066376455128193,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 120373444,
+      "step": 701
+    },
+    {
+      "epoch": 0.18465180508976128,
+      "grad_norm": 8.319297966322562,
+      "learning_rate": 5e-06,
+      "loss": 0.1785,
+      "num_input_tokens_seen": 120545552,
+      "step": 702
+    },
+    {
+      "epoch": 0.18465180508976128,
+      "loss": 0.18953613936901093,
+      "loss_ce": 0.0019751053769141436,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 120545552,
+      "step": 702
+    },
+    {
+      "epoch": 0.18491484184914841,
+      "grad_norm": 8.464289102867136,
+      "learning_rate": 5e-06,
+      "loss": 0.219,
+      "num_input_tokens_seen": 120717584,
+      "step": 703
+    },
+    {
+      "epoch": 0.18491484184914841,
+      "loss": 0.24215811491012573,
+      "loss_ce": 0.004975516349077225,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 120717584,
+      "step": 703
+    },
+    {
+      "epoch": 0.18517787860853555,
+      "grad_norm": 12.962080293446185,
+      "learning_rate": 5e-06,
+      "loss": 0.2162,
+      "num_input_tokens_seen": 120889756,
+      "step": 704
+    },
+    {
+      "epoch": 0.18517787860853555,
+      "loss": 0.17355692386627197,
+      "loss_ce": 0.0025364109314978123,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 120889756,
+      "step": 704
+    },
+    {
+      "epoch": 0.18544091536792268,
+      "grad_norm": 13.968822655265907,
+      "learning_rate": 5e-06,
+      "loss": 0.2237,
+      "num_input_tokens_seen": 121062200,
+      "step": 705
+    },
+    {
+      "epoch": 0.18544091536792268,
+      "loss": 0.2510673403739929,
+      "loss_ce": 0.005400843918323517,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 121062200,
+      "step": 705
+    },
+    {
+      "epoch": 0.18570395212730978,
+      "grad_norm": 15.418397137980255,
+      "learning_rate": 5e-06,
+      "loss": 0.2126,
+      "num_input_tokens_seen": 121234300,
+      "step": 706
+    },
+    {
+      "epoch": 0.18570395212730978,
+      "loss": 0.239446759223938,
+      "loss_ce": 0.0006772410124540329,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 121234300,
+      "step": 706
+    },
+    {
+      "epoch": 0.1859669888866969,
+      "grad_norm": 4.8632608475065355,
+      "learning_rate": 5e-06,
+      "loss": 0.1269,
+      "num_input_tokens_seen": 121406408,
+      "step": 707
+    },
+    {
+      "epoch": 0.1859669888866969,
+      "loss": 0.10976609587669373,
+      "loss_ce": 0.003137679770588875,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 121406408,
+      "step": 707
+    },
+    {
+      "epoch": 0.18623002564608404,
+      "grad_norm": 9.856831060146035,
+      "learning_rate": 5e-06,
+      "loss": 0.1699,
+      "num_input_tokens_seen": 121578364,
+      "step": 708
+    },
+    {
+      "epoch": 0.18623002564608404,
+      "loss": 0.06338398158550262,
+      "loss_ce": 0.0025319333653897047,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 121578364,
+      "step": 708
+    },
+    {
+      "epoch": 0.18649306240547117,
+      "grad_norm": 12.596787814338684,
+      "learning_rate": 5e-06,
+      "loss": 0.1839,
+      "num_input_tokens_seen": 121750612,
+      "step": 709
+    },
+    {
+      "epoch": 0.18649306240547117,
+      "loss": 0.2992492616176605,
+      "loss_ce": 0.0012756290379911661,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 121750612,
+      "step": 709
+    },
+    {
+      "epoch": 0.1867560991648583,
+      "grad_norm": 7.147545819350773,
+      "learning_rate": 5e-06,
+      "loss": 0.1776,
+      "num_input_tokens_seen": 121922804,
+      "step": 710
+    },
+    {
+      "epoch": 0.1867560991648583,
+      "loss": 0.18950411677360535,
+      "loss_ce": 0.0032858517952263355,
+      "loss_iou": 0.609375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 121922804,
+      "step": 710
+    },
+    {
+      "epoch": 0.1870191359242454,
+      "grad_norm": 22.63136110983146,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 122094924,
+      "step": 711
+    },
+    {
+      "epoch": 0.1870191359242454,
+      "loss": 0.12849926948547363,
+      "loss_ce": 0.0014851100277155638,
+      "loss_iou": 0.625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 122094924,
+      "step": 711
+    },
+    {
+      "epoch": 0.18728217268363254,
+      "grad_norm": 6.766417510968721,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 122265668,
+      "step": 712
+    },
+    {
+      "epoch": 0.18728217268363254,
+      "loss": 0.15207967162132263,
+      "loss_ce": 0.006968589033931494,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 122265668,
+      "step": 712
+    },
+    {
+      "epoch": 0.18754520944301967,
+      "grad_norm": 7.923158875388113,
+      "learning_rate": 5e-06,
+      "loss": 0.1904,
+      "num_input_tokens_seen": 122437796,
+      "step": 713
+    },
+    {
+      "epoch": 0.18754520944301967,
+      "loss": 0.20568042993545532,
+      "loss_ce": 0.002189208287745714,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 122437796,
+      "step": 713
+    },
+    {
+      "epoch": 0.1878082462024068,
+      "grad_norm": 8.667331104283747,
+      "learning_rate": 5e-06,
+      "loss": 0.1887,
+      "num_input_tokens_seen": 122610336,
+      "step": 714
+    },
+    {
+      "epoch": 0.1878082462024068,
+      "loss": 0.1953202784061432,
+      "loss_ce": 0.0012895169202238321,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 122610336,
+      "step": 714
+    },
+    {
+      "epoch": 0.1880712829617939,
+      "grad_norm": 12.96410805672465,
+      "learning_rate": 5e-06,
+      "loss": 0.2143,
+      "num_input_tokens_seen": 122779392,
+      "step": 715
+    },
+    {
+      "epoch": 0.1880712829617939,
+      "loss": 0.21114467084407806,
+      "loss_ce": 0.001916159177199006,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 122779392,
+      "step": 715
+    },
+    {
+      "epoch": 0.18833431972118103,
+      "grad_norm": 6.980680552365313,
+      "learning_rate": 5e-06,
+      "loss": 0.1831,
+      "num_input_tokens_seen": 122951724,
+      "step": 716
+    },
+    {
+      "epoch": 0.18833431972118103,
+      "loss": 0.24306175112724304,
+      "loss_ce": 0.006245340220630169,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 122951724,
+      "step": 716
+    },
+    {
+      "epoch": 0.18859735648056816,
+      "grad_norm": 12.855577653554155,
+      "learning_rate": 5e-06,
+      "loss": 0.1591,
+      "num_input_tokens_seen": 123122088,
+      "step": 717
+    },
+    {
+      "epoch": 0.18859735648056816,
+      "loss": 0.1986149549484253,
+      "loss_ce": 0.0012882874580100179,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 123122088,
+      "step": 717
+    },
+    {
+      "epoch": 0.1888603932399553,
+      "grad_norm": 6.930918072961039,
+      "learning_rate": 5e-06,
+      "loss": 0.1899,
+      "num_input_tokens_seen": 123294160,
+      "step": 718
+    },
+    {
+      "epoch": 0.1888603932399553,
+      "loss": 0.19642534852027893,
+      "loss_ce": 0.0014790646964684129,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 123294160,
+      "step": 718
+    },
+    {
+      "epoch": 0.1891234299993424,
+      "grad_norm": 6.929009893187008,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 123466060,
+      "step": 719
+    },
+    {
+      "epoch": 0.1891234299993424,
+      "loss": 0.1527690589427948,
+      "loss_ce": 0.005094507243484259,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 123466060,
+      "step": 719
+    },
+    {
+      "epoch": 0.18938646675872953,
+      "grad_norm": 10.25908293873023,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 123636388,
+      "step": 720
+    },
+    {
+      "epoch": 0.18938646675872953,
+      "loss": 0.18421480059623718,
+      "loss_ce": 0.0032455746550112963,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 123636388,
+      "step": 720
+    },
+    {
+      "epoch": 0.18964950351811666,
+      "grad_norm": 7.318122077183922,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 123808316,
+      "step": 721
+    },
+    {
+      "epoch": 0.18964950351811666,
+      "loss": 0.1352817267179489,
+      "loss_ce": 0.0030185491777956486,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 123808316,
+      "step": 721
+    },
+    {
+      "epoch": 0.1899125402775038,
+      "grad_norm": 7.382171237495957,
+      "learning_rate": 5e-06,
+      "loss": 0.2283,
+      "num_input_tokens_seen": 123980672,
+      "step": 722
+    },
+    {
+      "epoch": 0.1899125402775038,
+      "loss": 0.15074753761291504,
+      "loss_ce": 0.0036528080236166716,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 123980672,
+      "step": 722
+    },
+    {
+      "epoch": 0.19017557703689092,
+      "grad_norm": 5.982120923855094,
+      "learning_rate": 5e-06,
+      "loss": 0.1499,
+      "num_input_tokens_seen": 124150936,
+      "step": 723
+    },
+    {
+      "epoch": 0.19017557703689092,
+      "loss": 0.10008575022220612,
+      "loss_ce": 0.0014224194455891848,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 124150936,
+      "step": 723
+    },
+    {
+      "epoch": 0.19043861379627802,
+      "grad_norm": 8.136060280729641,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 124323280,
+      "step": 724
+    },
+    {
+      "epoch": 0.19043861379627802,
+      "loss": 0.12422403693199158,
+      "loss_ce": 0.0017264705384150147,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 124323280,
+      "step": 724
+    },
+    {
+      "epoch": 0.19070165055566515,
+      "grad_norm": 8.724019912821047,
+      "learning_rate": 5e-06,
+      "loss": 0.2613,
+      "num_input_tokens_seen": 124495636,
+      "step": 725
+    },
+    {
+      "epoch": 0.19070165055566515,
+      "loss": 0.13688622415065765,
+      "loss_ce": 0.0040737236849963665,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 124495636,
+      "step": 725
+    },
+    {
+      "epoch": 0.19096468731505228,
+      "grad_norm": 7.423558318920191,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 124667644,
+      "step": 726
+    },
+    {
+      "epoch": 0.19096468731505228,
+      "loss": 0.2597463130950928,
+      "loss_ce": 0.0010182850528508425,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 124667644,
+      "step": 726
+    },
+    {
+      "epoch": 0.19122772407443941,
+      "grad_norm": 7.827506457517143,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 124839252,
+      "step": 727
+    },
+    {
+      "epoch": 0.19122772407443941,
+      "loss": 0.1310674101114273,
+      "loss_ce": 0.003076688153669238,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 124839252,
+      "step": 727
+    },
+    {
+      "epoch": 0.19149076083382652,
+      "grad_norm": 9.183421040665964,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 125011496,
+      "step": 728
+    },
+    {
+      "epoch": 0.19149076083382652,
+      "loss": 0.11314553767442703,
+      "loss_ce": 0.0003525639185681939,
+      "loss_iou": 0.625,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 125011496,
+      "step": 728
+    },
+    {
+      "epoch": 0.19175379759321365,
+      "grad_norm": 10.185706416909431,
+      "learning_rate": 5e-06,
+      "loss": 0.2047,
+      "num_input_tokens_seen": 125179336,
+      "step": 729
+    },
+    {
+      "epoch": 0.19175379759321365,
+      "loss": 0.1552383303642273,
+      "loss_ce": 0.005152884405106306,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 125179336,
+      "step": 729
+    },
+    {
+      "epoch": 0.19201683435260078,
+      "grad_norm": 8.517948232569886,
+      "learning_rate": 5e-06,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 125351404,
+      "step": 730
+    },
+    {
+      "epoch": 0.19201683435260078,
+      "loss": 0.23154987394809723,
+      "loss_ce": 0.0012031885562464595,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 125351404,
+      "step": 730
+    },
+    {
+      "epoch": 0.1922798711119879,
+      "grad_norm": 8.61986864690339,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 125523692,
+      "step": 731
+    },
+    {
+      "epoch": 0.1922798711119879,
+      "loss": 0.14681357145309448,
+      "loss_ce": 0.0021907794289290905,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 125523692,
+      "step": 731
+    },
+    {
+      "epoch": 0.192542907871375,
+      "grad_norm": 5.859065568818718,
+      "learning_rate": 5e-06,
+      "loss": 0.178,
+      "num_input_tokens_seen": 125693908,
+      "step": 732
+    },
+    {
+      "epoch": 0.192542907871375,
+      "loss": 0.15787754952907562,
+      "loss_ce": 0.0012613451108336449,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 125693908,
+      "step": 732
+    },
+    {
+      "epoch": 0.19280594463076214,
+      "grad_norm": 10.469269184120652,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 125865868,
+      "step": 733
+    },
+    {
+      "epoch": 0.19280594463076214,
+      "loss": 0.13217297196388245,
+      "loss_ce": 0.0014051578473299742,
+      "loss_iou": 0.578125,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 125865868,
+      "step": 733
+    },
+    {
+      "epoch": 0.19306898139014927,
+      "grad_norm": 9.209135916468261,
+      "learning_rate": 5e-06,
+      "loss": 0.1745,
+      "num_input_tokens_seen": 126037852,
+      "step": 734
+    },
+    {
+      "epoch": 0.19306898139014927,
+      "loss": 0.17599225044250488,
+      "loss_ce": 0.0004551436868496239,
+      "loss_iou": 0.65625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 126037852,
+      "step": 734
+    },
+    {
+      "epoch": 0.1933320181495364,
+      "grad_norm": 8.89690790064976,
+      "learning_rate": 5e-06,
+      "loss": 0.2197,
+      "num_input_tokens_seen": 126209984,
+      "step": 735
+    },
+    {
+      "epoch": 0.1933320181495364,
+      "loss": 0.1378391683101654,
+      "loss_ce": 0.0020359433256089687,
+      "loss_iou": 0.6875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 126209984,
+      "step": 735
+    },
+    {
+      "epoch": 0.1935950549089235,
+      "grad_norm": 9.250244451235272,
+      "learning_rate": 5e-06,
+      "loss": 0.1999,
+      "num_input_tokens_seen": 126382140,
+      "step": 736
+    },
+    {
+      "epoch": 0.1935950549089235,
+      "loss": 0.24104052782058716,
+      "loss_ce": 0.0028813518583774567,
+      "loss_iou": 0.671875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 126382140,
+      "step": 736
+    },
+    {
+      "epoch": 0.19385809166831064,
+      "grad_norm": 6.538014726838841,
+      "learning_rate": 5e-06,
+      "loss": 0.1574,
+      "num_input_tokens_seen": 126554124,
+      "step": 737
+    },
+    {
+      "epoch": 0.19385809166831064,
+      "loss": 0.19325746595859528,
+      "loss_ce": 0.0034991574939340353,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 126554124,
+      "step": 737
+    },
+    {
+      "epoch": 0.19412112842769777,
+      "grad_norm": 6.359733816292311,
+      "learning_rate": 5e-06,
+      "loss": 0.1727,
+      "num_input_tokens_seen": 126726352,
+      "step": 738
+    },
+    {
+      "epoch": 0.19412112842769777,
+      "loss": 0.1890466809272766,
+      "loss_ce": 0.003866007784381509,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 126726352,
+      "step": 738
+    },
+    {
+      "epoch": 0.1943841651870849,
+      "grad_norm": 6.737716233597974,
+      "learning_rate": 5e-06,
+      "loss": 0.168,
+      "num_input_tokens_seen": 126898504,
+      "step": 739
+    },
+    {
+      "epoch": 0.1943841651870849,
+      "loss": 0.22711391746997833,
+      "loss_ce": 0.0020162612199783325,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 126898504,
+      "step": 739
+    },
+    {
+      "epoch": 0.19464720194647203,
+      "grad_norm": 10.63587672601612,
+      "learning_rate": 5e-06,
+      "loss": 0.2406,
+      "num_input_tokens_seen": 127070440,
+      "step": 740
+    },
+    {
+      "epoch": 0.19464720194647203,
+      "loss": 0.2545185983181,
+      "loss_ce": 0.0035420400090515614,
+      "loss_iou": 0.53125,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 127070440,
+      "step": 740
+    },
+    {
+      "epoch": 0.19491023870585913,
+      "grad_norm": 10.573425928841477,
+      "learning_rate": 5e-06,
+      "loss": 0.1461,
+      "num_input_tokens_seen": 127242884,
+      "step": 741
+    },
+    {
+      "epoch": 0.19491023870585913,
+      "loss": 0.07598280906677246,
+      "loss_ce": 0.0007874930743128061,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 127242884,
+      "step": 741
+    },
+    {
+      "epoch": 0.19517327546524627,
+      "grad_norm": 5.624166188859615,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 127413548,
+      "step": 742
+    },
+    {
+      "epoch": 0.19517327546524627,
+      "loss": 0.12612518668174744,
+      "loss_ce": 0.0008200095035135746,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 127413548,
+      "step": 742
+    },
+    {
+      "epoch": 0.1954363122246334,
+      "grad_norm": 5.416712373816477,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 127583504,
+      "step": 743
+    },
+    {
+      "epoch": 0.1954363122246334,
+      "loss": 0.14368192851543427,
+      "loss_ce": 0.003209515009075403,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 127583504,
+      "step": 743
+    },
+    {
+      "epoch": 0.19569934898402053,
+      "grad_norm": 7.55903022235747,
+      "learning_rate": 5e-06,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 127753924,
+      "step": 744
+    },
+    {
+      "epoch": 0.19569934898402053,
+      "loss": 0.26765167713165283,
+      "loss_ce": 0.009717106819152832,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 127753924,
+      "step": 744
+    },
+    {
+      "epoch": 0.19596238574340763,
+      "grad_norm": 6.325854098350077,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 127925840,
+      "step": 745
+    },
+    {
+      "epoch": 0.19596238574340763,
+      "loss": 0.13880833983421326,
+      "loss_ce": 0.0011435477063059807,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 127925840,
+      "step": 745
+    },
+    {
+      "epoch": 0.19622542250279476,
+      "grad_norm": 7.469182967125795,
+      "learning_rate": 5e-06,
+      "loss": 0.1661,
+      "num_input_tokens_seen": 128098212,
+      "step": 746
+    },
+    {
+      "epoch": 0.19622542250279476,
+      "loss": 0.22049343585968018,
+      "loss_ce": 0.002109651220962405,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 128098212,
+      "step": 746
+    },
+    {
+      "epoch": 0.1964884592621819,
+      "grad_norm": 8.506872378366532,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 128270372,
+      "step": 747
+    },
+    {
+      "epoch": 0.1964884592621819,
+      "loss": 0.1590556651353836,
+      "loss_ce": 0.0010356476996093988,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 128270372,
+      "step": 747
+    },
+    {
+      "epoch": 0.19675149602156902,
+      "grad_norm": 7.2518370649313075,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 128439832,
+      "step": 748
+    },
+    {
+      "epoch": 0.19675149602156902,
+      "loss": 0.16326385736465454,
+      "loss_ce": 0.0021310443989932537,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 128439832,
+      "step": 748
+    },
+    {
+      "epoch": 0.19701453278095613,
+      "grad_norm": 9.929926428237398,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 128612032,
+      "step": 749
+    },
+    {
+      "epoch": 0.19701453278095613,
+      "loss": 0.09666653722524643,
+      "loss_ce": 0.002489290665835142,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 128612032,
+      "step": 749
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "grad_norm": 8.129979685164507,
+      "learning_rate": 5e-06,
+      "loss": 0.144,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_websight_new_CIoU": 0.7427884042263031,
+      "eval_websight_new_GIoU": 0.7464376986026764,
+      "eval_websight_new_IoU": 0.7504112422466278,
+      "eval_websight_new_MAE_all": 0.03649342246353626,
+      "eval_websight_new_MAE_h": 0.03233582433313131,
+      "eval_websight_new_MAE_w": 0.056174855679273605,
+      "eval_websight_new_MAE_x": 0.046134982258081436,
+      "eval_websight_new_MAE_y": 0.01132803549990058,
+      "eval_websight_new_NUM_probability": 0.9972327351570129,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.19190411269664764,
+      "eval_websight_new_loss_ce": 0.00034623414103407413,
+      "eval_websight_new_loss_iou": 0.777099609375,
+      "eval_websight_new_loss_num": 0.03619384765625,
+      "eval_websight_new_loss_xval": 0.18096923828125,
+      "eval_websight_new_runtime": 55.1355,
+      "eval_websight_new_samples_per_second": 0.907,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_seeclick_CIoU": 0.4580978453159332,
+      "eval_seeclick_GIoU": 0.44802993535995483,
+      "eval_seeclick_IoU": 0.4842703342437744,
+      "eval_seeclick_MAE_all": 0.05876399576663971,
+      "eval_seeclick_MAE_h": 0.0604591965675354,
+      "eval_seeclick_MAE_w": 0.06378005631268024,
+      "eval_seeclick_MAE_x": 0.07332894578576088,
+      "eval_seeclick_MAE_y": 0.03748778998851776,
+      "eval_seeclick_NUM_probability": 0.9986195266246796,
+      "eval_seeclick_inside_bbox": 0.9375,
+      "eval_seeclick_loss": 0.3042093813419342,
+      "eval_seeclick_loss_ce": 0.012256910093128681,
+      "eval_seeclick_loss_iou": 0.6614990234375,
+      "eval_seeclick_loss_num": 0.055084228515625,
+      "eval_seeclick_loss_xval": 0.27545166015625,
+      "eval_seeclick_runtime": 69.7623,
+      "eval_seeclick_samples_per_second": 0.616,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_icons_CIoU": 0.7633987963199615,
+      "eval_icons_GIoU": 0.7561427056789398,
+      "eval_icons_IoU": 0.7705403864383698,
+      "eval_icons_MAE_all": 0.029183855280280113,
+      "eval_icons_MAE_h": 0.035788778215646744,
+      "eval_icons_MAE_w": 0.03255164809525013,
+      "eval_icons_MAE_x": 0.02722846996039152,
+      "eval_icons_MAE_y": 0.02116652298718691,
+      "eval_icons_NUM_probability": 0.9982486963272095,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.13746696710586548,
+      "eval_icons_loss_ce": 0.0027548681828193367,
+      "eval_icons_loss_iou": 0.7633056640625,
+      "eval_icons_loss_num": 0.02550506591796875,
+      "eval_icons_loss_xval": 0.127532958984375,
+      "eval_icons_runtime": 87.2396,
+      "eval_icons_samples_per_second": 0.573,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_screenspot_CIoU": 0.4680892725785573,
+      "eval_screenspot_GIoU": 0.45600322882334393,
+      "eval_screenspot_IoU": 0.507049967845281,
+      "eval_screenspot_MAE_all": 0.09560441970825195,
+      "eval_screenspot_MAE_h": 0.08166227489709854,
+      "eval_screenspot_MAE_w": 0.14845576385656992,
+      "eval_screenspot_MAE_x": 0.09815465907255809,
+      "eval_screenspot_MAE_y": 0.054144968589146934,
+      "eval_screenspot_NUM_probability": 0.9986165563265482,
+      "eval_screenspot_inside_bbox": 0.8558333317438761,
+      "eval_screenspot_loss": 0.805972158908844,
+      "eval_screenspot_loss_ce": 0.38150885701179504,
+      "eval_screenspot_loss_iou": 0.580078125,
+      "eval_screenspot_loss_num": 0.0841064453125,
+      "eval_screenspot_loss_xval": 0.4202473958333333,
+      "eval_screenspot_runtime": 148.1708,
+      "eval_screenspot_samples_per_second": 0.601,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "loss": 0.7833826541900635,
+      "loss_ce": 0.36297255754470825,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.1975406062997304,
+      "grad_norm": 6.107960207443286,
+      "learning_rate": 5e-06,
+      "loss": 0.2391,
+      "num_input_tokens_seen": 128956536,
+      "step": 751
+    },
+    {
+      "epoch": 0.1975406062997304,
+      "loss": 0.27341228723526,
+      "loss_ce": 0.0016227375017479062,
+      "loss_iou": 0.734375,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 128956536,
+      "step": 751
+    },
+    {
+      "epoch": 0.19780364305911752,
+      "grad_norm": 12.720415598286513,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 129128836,
+      "step": 752
+    },
+    {
+      "epoch": 0.19780364305911752,
+      "loss": 0.12072408944368362,
+      "loss_ce": 0.002621056977659464,
+      "loss_iou": 0.515625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 129128836,
+      "step": 752
+    },
+    {
+      "epoch": 0.19806667981850465,
+      "grad_norm": 14.921842726715303,
+      "learning_rate": 5e-06,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 129300844,
+      "step": 753
+    },
+    {
+      "epoch": 0.19806667981850465,
+      "loss": 0.1643597036600113,
+      "loss_ce": 0.0029217104893177748,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 129300844,
+      "step": 753
+    },
+    {
+      "epoch": 0.19832971657789175,
+      "grad_norm": 6.005245677684448,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 129471448,
+      "step": 754
+    },
+    {
+      "epoch": 0.19832971657789175,
+      "loss": 0.09330937266349792,
+      "loss_ce": 0.0012683530803769827,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 129471448,
+      "step": 754
+    },
+    {
+      "epoch": 0.19859275333727888,
+      "grad_norm": 10.390450693463213,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 129643296,
+      "step": 755
+    },
+    {
+      "epoch": 0.19859275333727888,
+      "loss": 0.19149892032146454,
+      "loss_ce": 0.000733550579752773,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 129643296,
+      "step": 755
+    },
+    {
+      "epoch": 0.198855790096666,
+      "grad_norm": 12.828044454318112,
+      "learning_rate": 5e-06,
+      "loss": 0.1805,
+      "num_input_tokens_seen": 129815492,
+      "step": 756
+    },
+    {
+      "epoch": 0.198855790096666,
+      "loss": 0.17852596938610077,
+      "loss_ce": 0.003294031834229827,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 129815492,
+      "step": 756
+    },
+    {
+      "epoch": 0.19911882685605314,
+      "grad_norm": 8.380872387581206,
+      "learning_rate": 5e-06,
+      "loss": 0.2269,
+      "num_input_tokens_seen": 129987692,
+      "step": 757
+    },
+    {
+      "epoch": 0.19911882685605314,
+      "loss": 0.20438869297504425,
+      "loss_ce": 0.00040920061292126775,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 129987692,
+      "step": 757
+    },
+    {
+      "epoch": 0.19938186361544025,
+      "grad_norm": 11.044900252894326,
+      "learning_rate": 5e-06,
+      "loss": 0.1723,
+      "num_input_tokens_seen": 130159828,
+      "step": 758
+    },
+    {
+      "epoch": 0.19938186361544025,
+      "loss": 0.1886797845363617,
+      "loss_ce": 0.0009661591611802578,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 130159828,
+      "step": 758
+    },
+    {
+      "epoch": 0.19964490037482738,
+      "grad_norm": 8.638981565005851,
+      "learning_rate": 5e-06,
+      "loss": 0.1719,
+      "num_input_tokens_seen": 130331756,
+      "step": 759
+    },
+    {
+      "epoch": 0.19964490037482738,
+      "loss": 0.1088617593050003,
+      "loss_ce": 0.0055597638711333275,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 130331756,
+      "step": 759
+    },
+    {
+      "epoch": 0.1999079371342145,
+      "grad_norm": 16.96292619187744,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 130503960,
+      "step": 760
+    },
+    {
+      "epoch": 0.1999079371342145,
+      "loss": 0.1997271627187729,
+      "loss_ce": 0.001668088138103485,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 130503960,
+      "step": 760
+    },
+    {
+      "epoch": 0.20017097389360164,
+      "grad_norm": 18.254879073962975,
+      "learning_rate": 5e-06,
+      "loss": 0.209,
+      "num_input_tokens_seen": 130676124,
+      "step": 761
+    },
+    {
+      "epoch": 0.20017097389360164,
+      "loss": 0.22872218489646912,
+      "loss_ce": 0.002464861376211047,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 130676124,
+      "step": 761
+    },
+    {
+      "epoch": 0.20043401065298874,
+      "grad_norm": 16.873564483656185,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 130848448,
+      "step": 762
+    },
+    {
+      "epoch": 0.20043401065298874,
+      "loss": 0.11374461650848389,
+      "loss_ce": 0.001867176266387105,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 130848448,
+      "step": 762
+    },
+    {
+      "epoch": 0.20069704741237587,
+      "grad_norm": 8.061510131196302,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 131020600,
+      "step": 763
+    },
+    {
+      "epoch": 0.20069704741237587,
+      "loss": 0.24045339226722717,
+      "loss_ce": 0.003575955517590046,
+      "loss_iou": 0.59375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 131020600,
+      "step": 763
+    },
+    {
+      "epoch": 0.200960084171763,
+      "grad_norm": 7.1675953763190865,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 131191064,
+      "step": 764
+    },
+    {
+      "epoch": 0.200960084171763,
+      "loss": 0.1495012640953064,
+      "loss_ce": 0.004115516785532236,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 131191064,
+      "step": 764
+    },
+    {
+      "epoch": 0.20122312093115013,
+      "grad_norm": 5.20536952677903,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 131363148,
+      "step": 765
+    },
+    {
+      "epoch": 0.20122312093115013,
+      "loss": 0.23652556538581848,
+      "loss_ce": 0.009566339664161205,
+      "loss_iou": 0.546875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 131363148,
+      "step": 765
+    },
+    {
+      "epoch": 0.20148615769053727,
+      "grad_norm": 8.038569612667256,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 131535488,
+      "step": 766
+    },
+    {
+      "epoch": 0.20148615769053727,
+      "loss": 0.24296867847442627,
+      "loss_ce": 0.0015136117581278086,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 131535488,
+      "step": 766
+    },
+    {
+      "epoch": 0.20174919444992437,
+      "grad_norm": 6.190176430910417,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 131707808,
+      "step": 767
+    },
+    {
+      "epoch": 0.20174919444992437,
+      "loss": 0.24412932991981506,
+      "loss_ce": 0.00218596076592803,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 131707808,
+      "step": 767
+    },
+    {
+      "epoch": 0.2020122312093115,
+      "grad_norm": 6.201791565366413,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 131879928,
+      "step": 768
+    },
+    {
+      "epoch": 0.2020122312093115,
+      "loss": 0.09396322071552277,
+      "loss_ce": 0.002288414863869548,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 131879928,
+      "step": 768
+    },
+    {
+      "epoch": 0.20227526796869863,
+      "grad_norm": 7.34120818873771,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 132052340,
+      "step": 769
+    },
+    {
+      "epoch": 0.20227526796869863,
+      "loss": 0.22239628434181213,
+      "loss_ce": 0.005477334372699261,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 132052340,
+      "step": 769
+    },
+    {
+      "epoch": 0.20253830472808576,
+      "grad_norm": 6.124473321987556,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 132224712,
+      "step": 770
+    },
+    {
+      "epoch": 0.20253830472808576,
+      "loss": 0.17558854818344116,
+      "loss_ce": 0.0018824923317879438,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 132224712,
+      "step": 770
+    },
+    {
+      "epoch": 0.20280134148747286,
+      "grad_norm": 9.284836415516965,
+      "learning_rate": 5e-06,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 132396772,
+      "step": 771
+    },
+    {
+      "epoch": 0.20280134148747286,
+      "loss": 0.15597118437290192,
+      "loss_ce": 0.0003925706841982901,
+      "loss_iou": 0.734375,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 132396772,
+      "step": 771
+    },
+    {
+      "epoch": 0.20306437824686,
+      "grad_norm": 7.764344629279706,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 132568908,
+      "step": 772
+    },
+    {
+      "epoch": 0.20306437824686,
+      "loss": 0.18731489777565002,
+      "loss_ce": 0.0028056304436177015,
+      "loss_iou": 0.53125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 132568908,
+      "step": 772
+    },
+    {
+      "epoch": 0.20332741500624713,
+      "grad_norm": 10.387981116967136,
+      "learning_rate": 5e-06,
+      "loss": 0.2105,
+      "num_input_tokens_seen": 132741140,
+      "step": 773
+    },
+    {
+      "epoch": 0.20332741500624713,
+      "loss": 0.2155558466911316,
+      "loss_ce": 0.00147504813503474,
+      "loss_iou": 0.5625,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 132741140,
+      "step": 773
+    },
+    {
+      "epoch": 0.20359045176563426,
+      "grad_norm": 11.81609487087241,
+      "learning_rate": 5e-06,
+      "loss": 0.1972,
+      "num_input_tokens_seen": 132913412,
+      "step": 774
+    },
+    {
+      "epoch": 0.20359045176563426,
+      "loss": 0.2562327980995178,
+      "loss_ce": 0.0022044687066227198,
+      "loss_iou": 0.625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 132913412,
+      "step": 774
+    },
+    {
+      "epoch": 0.20385348852502136,
+      "grad_norm": 11.070638283175898,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 133085388,
+      "step": 775
+    },
+    {
+      "epoch": 0.20385348852502136,
+      "loss": 0.1850009262561798,
+      "loss_ce": 0.0011630415683612227,
+      "loss_iou": 0.671875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 133085388,
+      "step": 775
+    },
+    {
+      "epoch": 0.2041165252844085,
+      "grad_norm": 8.169381502927404,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 133257492,
+      "step": 776
+    },
+    {
+      "epoch": 0.2041165252844085,
+      "loss": 0.17405013740062714,
+      "loss_ce": 0.0007713312515988946,
+      "loss_iou": 0.609375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 133257492,
+      "step": 776
+    },
+    {
+      "epoch": 0.20437956204379562,
+      "grad_norm": 7.483205340151747,
+      "learning_rate": 5e-06,
+      "loss": 0.1812,
+      "num_input_tokens_seen": 133429516,
+      "step": 777
+    },
+    {
+      "epoch": 0.20437956204379562,
+      "loss": 0.21947714686393738,
+      "loss_ce": 0.0011543984292075038,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 133429516,
+      "step": 777
+    },
+    {
+      "epoch": 0.20464259880318275,
+      "grad_norm": 15.835699409497824,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 133599692,
+      "step": 778
+    },
+    {
+      "epoch": 0.20464259880318275,
+      "loss": 0.1631580889225006,
+      "loss_ce": 0.0015064696781337261,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 133599692,
+      "step": 778
+    },
+    {
+      "epoch": 0.20490563556256988,
+      "grad_norm": 6.9792797252935,
+      "learning_rate": 5e-06,
+      "loss": 0.1661,
+      "num_input_tokens_seen": 133771756,
+      "step": 779
+    },
+    {
+      "epoch": 0.20490563556256988,
+      "loss": 0.16169646382331848,
+      "loss_ce": 0.0007467527757398784,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 133771756,
+      "step": 779
+    },
+    {
+      "epoch": 0.20516867232195699,
+      "grad_norm": 12.331720736489249,
+      "learning_rate": 5e-06,
+      "loss": 0.1981,
+      "num_input_tokens_seen": 133944348,
+      "step": 780
+    },
+    {
+      "epoch": 0.20516867232195699,
+      "loss": 0.13550271093845367,
+      "loss_ce": 0.0025071091949939728,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 133944348,
+      "step": 780
+    },
+    {
+      "epoch": 0.20543170908134412,
+      "grad_norm": 5.7846193156700725,
+      "learning_rate": 5e-06,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 134116796,
+      "step": 781
+    },
+    {
+      "epoch": 0.20543170908134412,
+      "loss": 0.09005297720432281,
+      "loss_ce": 0.0008500947151333094,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 134116796,
+      "step": 781
+    },
+    {
+      "epoch": 0.20569474584073125,
+      "grad_norm": 20.38915176858911,
+      "learning_rate": 5e-06,
+      "loss": 0.2145,
+      "num_input_tokens_seen": 134285564,
+      "step": 782
+    },
+    {
+      "epoch": 0.20569474584073125,
+      "loss": 0.2173474133014679,
+      "loss_ce": 0.0005505430162884295,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 134285564,
+      "step": 782
+    },
+    {
+      "epoch": 0.20595778260011838,
+      "grad_norm": 13.943592884184664,
+      "learning_rate": 5e-06,
+      "loss": 0.2368,
+      "num_input_tokens_seen": 134457600,
+      "step": 783
+    },
+    {
+      "epoch": 0.20595778260011838,
+      "loss": 0.23010152578353882,
+      "loss_ce": 0.00341693963855505,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 134457600,
+      "step": 783
+    },
+    {
+      "epoch": 0.20622081935950548,
+      "grad_norm": 7.47232647542354,
+      "learning_rate": 5e-06,
+      "loss": 0.1884,
+      "num_input_tokens_seen": 134627836,
+      "step": 784
+    },
+    {
+      "epoch": 0.20622081935950548,
+      "loss": 0.31371766328811646,
+      "loss_ce": 0.0035369964316487312,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 134627836,
+      "step": 784
+    },
+    {
+      "epoch": 0.2064838561188926,
+      "grad_norm": 7.561434198006136,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 134800024,
+      "step": 785
+    },
+    {
+      "epoch": 0.2064838561188926,
+      "loss": 0.25232362747192383,
+      "loss_ce": 0.003910548985004425,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 134800024,
+      "step": 785
+    },
+    {
+      "epoch": 0.20674689287827974,
+      "grad_norm": 7.384265604656037,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 134972276,
+      "step": 786
+    },
+    {
+      "epoch": 0.20674689287827974,
+      "loss": 0.22025543451309204,
+      "loss_ce": 0.0012307591969147325,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 134972276,
+      "step": 786
+    },
+    {
+      "epoch": 0.20700992963766687,
+      "grad_norm": 6.5497695680105705,
+      "learning_rate": 5e-06,
+      "loss": 0.1985,
+      "num_input_tokens_seen": 135144300,
+      "step": 787
+    },
+    {
+      "epoch": 0.20700992963766687,
+      "loss": 0.24491435289382935,
+      "loss_ce": 0.0007126981508918107,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 135144300,
+      "step": 787
+    },
+    {
+      "epoch": 0.20727296639705398,
+      "grad_norm": 7.321479905780818,
+      "learning_rate": 5e-06,
+      "loss": 0.1764,
+      "num_input_tokens_seen": 135316344,
+      "step": 788
+    },
+    {
+      "epoch": 0.20727296639705398,
+      "loss": 0.19332991540431976,
+      "loss_ce": 0.0017405691323801875,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 135316344,
+      "step": 788
+    },
+    {
+      "epoch": 0.2075360031564411,
+      "grad_norm": 5.255608280002089,
+      "learning_rate": 5e-06,
+      "loss": 0.1867,
+      "num_input_tokens_seen": 135488352,
+      "step": 789
+    },
+    {
+      "epoch": 0.2075360031564411,
+      "loss": 0.18961402773857117,
+      "loss_ce": 0.0032126582227647305,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 135488352,
+      "step": 789
+    },
+    {
+      "epoch": 0.20779903991582824,
+      "grad_norm": 6.817599663694252,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 135660716,
+      "step": 790
+    },
+    {
+      "epoch": 0.20779903991582824,
+      "loss": 0.11433705687522888,
+      "loss_ce": 0.0007506305119022727,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 135660716,
+      "step": 790
+    },
+    {
+      "epoch": 0.20806207667521537,
+      "grad_norm": 7.049616575343475,
+      "learning_rate": 5e-06,
+      "loss": 0.2041,
+      "num_input_tokens_seen": 135833108,
+      "step": 791
+    },
+    {
+      "epoch": 0.20806207667521537,
+      "loss": 0.16387248039245605,
+      "loss_ce": 0.005669359117746353,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 135833108,
+      "step": 791
+    },
+    {
+      "epoch": 0.2083251134346025,
+      "grad_norm": 9.83283000996709,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 136001408,
+      "step": 792
+    },
+    {
+      "epoch": 0.2083251134346025,
+      "loss": 0.13693515956401825,
+      "loss_ce": 0.0015286724083125591,
+      "loss_iou": 0.546875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 136001408,
+      "step": 792
+    },
+    {
+      "epoch": 0.2085881501939896,
+      "grad_norm": 18.56546458626088,
+      "learning_rate": 5e-06,
+      "loss": 0.2126,
+      "num_input_tokens_seen": 136173768,
+      "step": 793
+    },
+    {
+      "epoch": 0.2085881501939896,
+      "loss": 0.2671317458152771,
+      "loss_ce": 0.004436435177922249,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 136173768,
+      "step": 793
+    },
+    {
+      "epoch": 0.20885118695337673,
+      "grad_norm": 7.460522492414201,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 136345864,
+      "step": 794
+    },
+    {
+      "epoch": 0.20885118695337673,
+      "loss": 0.19496268033981323,
+      "loss_ce": 0.003495402168482542,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 136345864,
+      "step": 794
+    },
+    {
+      "epoch": 0.20911422371276386,
+      "grad_norm": 6.238640410439663,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 136518076,
+      "step": 795
+    },
+    {
+      "epoch": 0.20911422371276386,
+      "loss": 0.1869560331106186,
+      "loss_ce": 0.0006157027091830969,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 136518076,
+      "step": 795
+    },
+    {
+      "epoch": 0.209377260472151,
+      "grad_norm": 4.750721959517019,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 136690192,
+      "step": 796
+    },
+    {
+      "epoch": 0.209377260472151,
+      "loss": 0.10885806381702423,
+      "loss_ce": 0.0011920429533347487,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 136690192,
+      "step": 796
+    },
+    {
+      "epoch": 0.2096402972315381,
+      "grad_norm": 11.655608393949823,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 136862596,
+      "step": 797
+    },
+    {
+      "epoch": 0.2096402972315381,
+      "loss": 0.13095784187316895,
+      "loss_ce": 0.0011970889754593372,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 136862596,
+      "step": 797
+    },
+    {
+      "epoch": 0.20990333399092523,
+      "grad_norm": 7.6147795291746645,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 137033096,
+      "step": 798
+    },
+    {
+      "epoch": 0.20990333399092523,
+      "loss": 0.14115872979164124,
+      "loss_ce": 0.002456323243677616,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 137033096,
+      "step": 798
+    },
+    {
+      "epoch": 0.21016637075031236,
+      "grad_norm": 6.436556075292691,
+      "learning_rate": 5e-06,
+      "loss": 0.164,
+      "num_input_tokens_seen": 137203708,
+      "step": 799
+    },
+    {
+      "epoch": 0.21016637075031236,
+      "loss": 0.15657231211662292,
+      "loss_ce": 0.0031909646932035685,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 137203708,
+      "step": 799
+    },
+    {
+      "epoch": 0.2104294075096995,
+      "grad_norm": 5.272376224873815,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 137374352,
+      "step": 800
+    },
+    {
+      "epoch": 0.2104294075096995,
+      "loss": 0.19644752144813538,
+      "loss_ce": 0.003332281718030572,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 137374352,
+      "step": 800
+    },
+    {
+      "epoch": 0.2106924442690866,
+      "grad_norm": 9.274938800244652,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 137546372,
+      "step": 801
+    },
+    {
+      "epoch": 0.2106924442690866,
+      "loss": 0.23770156502723694,
+      "loss_ce": 0.0010682701831683517,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 137546372,
+      "step": 801
+    },
+    {
+      "epoch": 0.21095548102847372,
+      "grad_norm": 12.494816629710325,
+      "learning_rate": 5e-06,
+      "loss": 0.1484,
+      "num_input_tokens_seen": 137718496,
+      "step": 802
+    },
+    {
+      "epoch": 0.21095548102847372,
+      "loss": 0.13832518458366394,
+      "loss_ce": 0.00380369508638978,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 137718496,
+      "step": 802
+    },
+    {
+      "epoch": 0.21121851778786085,
+      "grad_norm": 11.143811760293481,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 137890492,
+      "step": 803
+    },
+    {
+      "epoch": 0.21121851778786085,
+      "loss": 0.17278538644313812,
+      "loss_ce": 0.0006662444211542606,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 137890492,
+      "step": 803
+    },
+    {
+      "epoch": 0.21148155454724799,
+      "grad_norm": 6.694127616757882,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 138062576,
+      "step": 804
+    },
+    {
+      "epoch": 0.21148155454724799,
+      "loss": 0.08578141778707504,
+      "loss_ce": 0.0013697945978492498,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 138062576,
+      "step": 804
+    },
+    {
+      "epoch": 0.21174459130663512,
+      "grad_norm": 6.879766720907239,
+      "learning_rate": 5e-06,
+      "loss": 0.2407,
+      "num_input_tokens_seen": 138234992,
+      "step": 805
+    },
+    {
+      "epoch": 0.21174459130663512,
+      "loss": 0.26608556509017944,
+      "loss_ce": 0.0011929699685424566,
+      "loss_iou": 0.671875,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 138234992,
+      "step": 805
+    },
+    {
+      "epoch": 0.21200762806602222,
+      "grad_norm": 10.68020187128554,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 138407296,
+      "step": 806
+    },
+    {
+      "epoch": 0.21200762806602222,
+      "loss": 0.16970132291316986,
+      "loss_ce": 0.003929831553250551,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 138407296,
+      "step": 806
+    },
+    {
+      "epoch": 0.21227066482540935,
+      "grad_norm": 6.8835699764215,
+      "learning_rate": 5e-06,
+      "loss": 0.2142,
+      "num_input_tokens_seen": 138579820,
+      "step": 807
+    },
+    {
+      "epoch": 0.21227066482540935,
+      "loss": 0.20114630460739136,
+      "loss_ce": 0.0011341023491695523,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 138579820,
+      "step": 807
+    },
+    {
+      "epoch": 0.21253370158479648,
+      "grad_norm": 5.0783612426941795,
+      "learning_rate": 5e-06,
+      "loss": 0.1658,
+      "num_input_tokens_seen": 138752112,
+      "step": 808
+    },
+    {
+      "epoch": 0.21253370158479648,
+      "loss": 0.22068345546722412,
+      "loss_ce": 0.004802103620022535,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 138752112,
+      "step": 808
+    },
+    {
+      "epoch": 0.2127967383441836,
+      "grad_norm": 6.76267182747039,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 138924516,
+      "step": 809
+    },
+    {
+      "epoch": 0.2127967383441836,
+      "loss": 0.16195307672023773,
+      "loss_ce": 0.0010644117137417197,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 138924516,
+      "step": 809
+    },
+    {
+      "epoch": 0.21305977510357071,
+      "grad_norm": 12.983359860046427,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 139096976,
+      "step": 810
+    },
+    {
+      "epoch": 0.21305977510357071,
+      "loss": 0.14437143504619598,
+      "loss_ce": 0.0023426164407283068,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 139096976,
+      "step": 810
+    },
+    {
+      "epoch": 0.21332281186295785,
+      "grad_norm": 5.900519173332491,
+      "learning_rate": 5e-06,
+      "loss": 0.1796,
+      "num_input_tokens_seen": 139269224,
+      "step": 811
+    },
+    {
+      "epoch": 0.21332281186295785,
+      "loss": 0.11451567709445953,
+      "loss_ce": 0.0014175281394273043,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 139269224,
+      "step": 811
+    },
+    {
+      "epoch": 0.21358584862234498,
+      "grad_norm": 11.704739365440266,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 139441048,
+      "step": 812
+    },
+    {
+      "epoch": 0.21358584862234498,
+      "loss": 0.06064599007368088,
+      "loss_ce": 0.0007399858441203833,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 139441048,
+      "step": 812
+    },
+    {
+      "epoch": 0.2138488853817321,
+      "grad_norm": 5.765338635306649,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 139613180,
+      "step": 813
+    },
+    {
+      "epoch": 0.2138488853817321,
+      "loss": 0.1868004947900772,
+      "loss_ce": 0.001986048649996519,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 139613180,
+      "step": 813
+    },
+    {
+      "epoch": 0.2141119221411192,
+      "grad_norm": 8.301852392840289,
+      "learning_rate": 5e-06,
+      "loss": 0.1731,
+      "num_input_tokens_seen": 139785340,
+      "step": 814
+    },
+    {
+      "epoch": 0.2141119221411192,
+      "loss": 0.21647384762763977,
+      "loss_ce": 0.002240448724478483,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 139785340,
+      "step": 814
+    },
+    {
+      "epoch": 0.21437495890050634,
+      "grad_norm": 10.75059655667213,
+      "learning_rate": 5e-06,
+      "loss": 0.2066,
+      "num_input_tokens_seen": 139957612,
+      "step": 815
+    },
+    {
+      "epoch": 0.21437495890050634,
+      "loss": 0.2657305598258972,
+      "loss_ce": 0.007979076355695724,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 139957612,
+      "step": 815
+    },
+    {
+      "epoch": 0.21463799565989347,
+      "grad_norm": 8.1554830347417,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 140129832,
+      "step": 816
+    },
+    {
+      "epoch": 0.21463799565989347,
+      "loss": 0.21090401709079742,
+      "loss_ce": 0.0022858483716845512,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 140129832,
+      "step": 816
+    },
+    {
+      "epoch": 0.2149010324192806,
+      "grad_norm": 6.39307582100075,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 140302204,
+      "step": 817
+    },
+    {
+      "epoch": 0.2149010324192806,
+      "loss": 0.12444409728050232,
+      "loss_ce": 0.0020075817592442036,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 140302204,
+      "step": 817
+    },
+    {
+      "epoch": 0.2151640691786677,
+      "grad_norm": 11.031920247903537,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 140470936,
+      "step": 818
+    },
+    {
+      "epoch": 0.2151640691786677,
+      "loss": 0.24400946497917175,
+      "loss_ce": 0.00112004519905895,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 140470936,
+      "step": 818
+    },
+    {
+      "epoch": 0.21542710593805484,
+      "grad_norm": 5.897881271346391,
+      "learning_rate": 5e-06,
+      "loss": 0.194,
+      "num_input_tokens_seen": 140643120,
+      "step": 819
+    },
+    {
+      "epoch": 0.21542710593805484,
+      "loss": 0.12999695539474487,
+      "loss_ce": 0.001182260224595666,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 140643120,
+      "step": 819
+    },
+    {
+      "epoch": 0.21569014269744197,
+      "grad_norm": 13.442111711356437,
+      "learning_rate": 5e-06,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 140815608,
+      "step": 820
+    },
+    {
+      "epoch": 0.21569014269744197,
+      "loss": 0.1816408634185791,
+      "loss_ce": 0.0014040416572242975,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 140815608,
+      "step": 820
+    },
+    {
+      "epoch": 0.2159531794568291,
+      "grad_norm": 13.297750331398602,
+      "learning_rate": 5e-06,
+      "loss": 0.1926,
+      "num_input_tokens_seen": 140987824,
+      "step": 821
+    },
+    {
+      "epoch": 0.2159531794568291,
+      "loss": 0.29018890857696533,
+      "loss_ce": 0.002347112400457263,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 140987824,
+      "step": 821
+    },
+    {
+      "epoch": 0.21621621621621623,
+      "grad_norm": 18.74775436478718,
+      "learning_rate": 5e-06,
+      "loss": 0.2948,
+      "num_input_tokens_seen": 141160200,
+      "step": 822
+    },
+    {
+      "epoch": 0.21621621621621623,
+      "loss": 0.2370918244123459,
+      "loss_ce": 0.1260078400373459,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 141160200,
+      "step": 822
+    },
+    {
+      "epoch": 0.21647925297560333,
+      "grad_norm": 19.14528392541103,
+      "learning_rate": 5e-06,
+      "loss": 0.1977,
+      "num_input_tokens_seen": 141329928,
+      "step": 823
+    },
+    {
+      "epoch": 0.21647925297560333,
+      "loss": 0.1949683427810669,
+      "loss_ce": 0.0917884111404419,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 141329928,
+      "step": 823
+    },
+    {
+      "epoch": 0.21674228973499046,
+      "grad_norm": 6.8551193847587735,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 141502236,
+      "step": 824
+    },
+    {
+      "epoch": 0.21674228973499046,
+      "loss": 0.09739409387111664,
+      "loss_ce": 0.007184132467955351,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 141502236,
+      "step": 824
+    },
+    {
+      "epoch": 0.2170053264943776,
+      "grad_norm": 7.934353513499139,
+      "learning_rate": 5e-06,
+      "loss": 0.2174,
+      "num_input_tokens_seen": 141674272,
+      "step": 825
+    },
+    {
+      "epoch": 0.2170053264943776,
+      "loss": 0.3033770024776459,
+      "loss_ce": 0.026491012424230576,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 141674272,
+      "step": 825
+    },
+    {
+      "epoch": 0.21726836325376472,
+      "grad_norm": 6.296485584445855,
+      "learning_rate": 5e-06,
+      "loss": 0.2048,
+      "num_input_tokens_seen": 141846556,
+      "step": 826
+    },
+    {
+      "epoch": 0.21726836325376472,
+      "loss": 0.16873130202293396,
+      "loss_ce": 0.0013118635397404432,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 141846556,
+      "step": 826
+    },
+    {
+      "epoch": 0.21753140001315183,
+      "grad_norm": 10.428529158159186,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 142018940,
+      "step": 827
+    },
+    {
+      "epoch": 0.21753140001315183,
+      "loss": 0.15984642505645752,
+      "loss_ce": 0.0018874472007155418,
+      "loss_iou": 0.421875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 142018940,
+      "step": 827
+    },
+    {
+      "epoch": 0.21779443677253896,
+      "grad_norm": 7.702760290749295,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 142189520,
+      "step": 828
+    },
+    {
+      "epoch": 0.21779443677253896,
+      "loss": 0.17661917209625244,
+      "loss_ce": 0.007002475671470165,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 142189520,
+      "step": 828
+    },
+    {
+      "epoch": 0.2180574735319261,
+      "grad_norm": 5.154197949486892,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 142359752,
+      "step": 829
+    },
+    {
+      "epoch": 0.2180574735319261,
+      "loss": 0.11216248571872711,
+      "loss_ce": 0.0025128289125859737,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 142359752,
+      "step": 829
+    },
+    {
+      "epoch": 0.21832051029131322,
+      "grad_norm": 9.212654853971248,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 142530028,
+      "step": 830
+    },
+    {
+      "epoch": 0.21832051029131322,
+      "loss": 0.17883381247520447,
+      "loss_ce": 0.006104309111833572,
+      "loss_iou": 0.515625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 142530028,
+      "step": 830
+    },
+    {
+      "epoch": 0.21858354705070032,
+      "grad_norm": 22.355687744069783,
+      "learning_rate": 5e-06,
+      "loss": 0.2118,
+      "num_input_tokens_seen": 142698896,
+      "step": 831
+    },
+    {
+      "epoch": 0.21858354705070032,
+      "loss": 0.2873075604438782,
+      "loss_ce": 0.00788860023021698,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 142698896,
+      "step": 831
+    },
+    {
+      "epoch": 0.21884658381008745,
+      "grad_norm": 7.646461001716363,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 142869236,
+      "step": 832
+    },
+    {
+      "epoch": 0.21884658381008745,
+      "loss": 0.19723272323608398,
+      "loss_ce": 0.00063848658464849,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 142869236,
+      "step": 832
+    },
+    {
+      "epoch": 0.21910962056947458,
+      "grad_norm": 5.709950173979812,
+      "learning_rate": 5e-06,
+      "loss": 0.1462,
+      "num_input_tokens_seen": 143041432,
+      "step": 833
+    },
+    {
+      "epoch": 0.21910962056947458,
+      "loss": 0.13923318684101105,
+      "loss_ce": 0.006512241438031197,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 143041432,
+      "step": 833
+    },
+    {
+      "epoch": 0.21937265732886171,
+      "grad_norm": 7.759034787872298,
+      "learning_rate": 5e-06,
+      "loss": 0.1411,
+      "num_input_tokens_seen": 143213276,
+      "step": 834
+    },
+    {
+      "epoch": 0.21937265732886171,
+      "loss": 0.13741275668144226,
+      "loss_ce": 0.00820133276283741,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 143213276,
+      "step": 834
+    },
+    {
+      "epoch": 0.21963569408824885,
+      "grad_norm": 14.10397350971688,
+      "learning_rate": 5e-06,
+      "loss": 0.1819,
+      "num_input_tokens_seen": 143385916,
+      "step": 835
+    },
+    {
+      "epoch": 0.21963569408824885,
+      "loss": 0.17109227180480957,
+      "loss_ce": 0.009959458373486996,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 143385916,
+      "step": 835
+    },
+    {
+      "epoch": 0.21989873084763595,
+      "grad_norm": 10.610026241323633,
+      "learning_rate": 5e-06,
+      "loss": 0.1959,
+      "num_input_tokens_seen": 143558108,
+      "step": 836
+    },
+    {
+      "epoch": 0.21989873084763595,
+      "loss": 0.1955014169216156,
+      "loss_ce": 0.006353452801704407,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 143558108,
+      "step": 836
+    },
+    {
+      "epoch": 0.22016176760702308,
+      "grad_norm": 6.884069616265665,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 143730560,
+      "step": 837
+    },
+    {
+      "epoch": 0.22016176760702308,
+      "loss": 0.1851910948753357,
+      "loss_ce": 0.001475287601351738,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 143730560,
+      "step": 837
+    },
+    {
+      "epoch": 0.2204248043664102,
+      "grad_norm": 8.581569398296145,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 143901168,
+      "step": 838
+    },
+    {
+      "epoch": 0.2204248043664102,
+      "loss": 0.10049735009670258,
+      "loss_ce": 0.00043021421879529953,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 143901168,
+      "step": 838
+    },
+    {
+      "epoch": 0.22068784112579734,
+      "grad_norm": 8.473859363920706,
+      "learning_rate": 5e-06,
+      "loss": 0.1656,
+      "num_input_tokens_seen": 144073300,
+      "step": 839
+    },
+    {
+      "epoch": 0.22068784112579734,
+      "loss": 0.18692679703235626,
+      "loss_ce": 0.001440959284082055,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 144073300,
+      "step": 839
+    },
+    {
+      "epoch": 0.22095087788518444,
+      "grad_norm": 4.5516671382314495,
+      "learning_rate": 5e-06,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 144245656,
+      "step": 840
+    },
+    {
+      "epoch": 0.22095087788518444,
+      "loss": 0.1278304159641266,
+      "loss_ce": 0.003227140521630645,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 144245656,
+      "step": 840
+    },
+    {
+      "epoch": 0.22121391464457157,
+      "grad_norm": 8.854170908003372,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 144414684,
+      "step": 841
+    },
+    {
+      "epoch": 0.22121391464457157,
+      "loss": 0.11698315292596817,
+      "loss_ce": 0.003732412587851286,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 144414684,
+      "step": 841
+    },
+    {
+      "epoch": 0.2214769514039587,
+      "grad_norm": 6.139198378486474,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 144586900,
+      "step": 842
+    },
+    {
+      "epoch": 0.2214769514039587,
+      "loss": 0.16039735078811646,
+      "loss_ce": 0.001461806707084179,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 144586900,
+      "step": 842
+    },
+    {
+      "epoch": 0.22173998816334584,
+      "grad_norm": 10.644206851478819,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 144759108,
+      "step": 843
+    },
+    {
+      "epoch": 0.22173998816334584,
+      "loss": 0.14353252947330475,
+      "loss_ce": 0.0032432209700345993,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 144759108,
+      "step": 843
+    },
+    {
+      "epoch": 0.22200302492273294,
+      "grad_norm": 6.739417345524847,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 144931444,
+      "step": 844
+    },
+    {
+      "epoch": 0.22200302492273294,
+      "loss": 0.09944656491279602,
+      "loss_ce": 0.0004170280881226063,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 144931444,
+      "step": 844
+    },
+    {
+      "epoch": 0.22226606168212007,
+      "grad_norm": 8.110959144365463,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 145102036,
+      "step": 845
+    },
+    {
+      "epoch": 0.22226606168212007,
+      "loss": 0.2286926507949829,
+      "loss_ce": 0.001336704008281231,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 145102036,
+      "step": 845
+    },
+    {
+      "epoch": 0.2225290984415072,
+      "grad_norm": 10.034770268919976,
+      "learning_rate": 5e-06,
+      "loss": 0.2222,
+      "num_input_tokens_seen": 145274324,
+      "step": 846
+    },
+    {
+      "epoch": 0.2225290984415072,
+      "loss": 0.23924441635608673,
+      "loss_ce": 0.004075955133885145,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 145274324,
+      "step": 846
+    },
+    {
+      "epoch": 0.22279213520089433,
+      "grad_norm": 6.615833632469255,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 145446460,
+      "step": 847
+    },
+    {
+      "epoch": 0.22279213520089433,
+      "loss": 0.1300104260444641,
+      "loss_ce": 0.001287288498133421,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 145446460,
+      "step": 847
+    },
+    {
+      "epoch": 0.22305517196028146,
+      "grad_norm": 6.286344612415352,
+      "learning_rate": 5e-06,
+      "loss": 0.2284,
+      "num_input_tokens_seen": 145618532,
+      "step": 848
+    },
+    {
+      "epoch": 0.22305517196028146,
+      "loss": 0.18182075023651123,
+      "loss_ce": 0.0016144568799063563,
+      "loss_iou": NaN,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 145618532,
+      "step": 848
+    },
+    {
+      "epoch": 0.22331820871966857,
+      "grad_norm": 15.090592675837648,
+      "learning_rate": 5e-06,
+      "loss": 0.1974,
+      "num_input_tokens_seen": 145788852,
+      "step": 849
+    },
+    {
+      "epoch": 0.22331820871966857,
+      "loss": 0.20251962542533875,
+      "loss_ce": 0.0014087767340242863,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 145788852,
+      "step": 849
+    },
+    {
+      "epoch": 0.2235812454790557,
+      "grad_norm": 6.5100345965661806,
+      "learning_rate": 5e-06,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 145961180,
+      "step": 850
+    },
+    {
+      "epoch": 0.2235812454790557,
+      "loss": 0.15530481934547424,
+      "loss_ce": 0.002533819992095232,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 145961180,
+      "step": 850
+    },
+    {
+      "epoch": 0.22384428223844283,
+      "grad_norm": 6.940039828670214,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 146133828,
+      "step": 851
+    },
+    {
+      "epoch": 0.22384428223844283,
+      "loss": 0.1814190298318863,
+      "loss_ce": 0.001731535536237061,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 146133828,
+      "step": 851
+    },
+    {
+      "epoch": 0.22410731899782996,
+      "grad_norm": 8.305849772963127,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 146306088,
+      "step": 852
+    },
+    {
+      "epoch": 0.22410731899782996,
+      "loss": 0.18216609954833984,
+      "loss_ce": 0.002936376491561532,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 146306088,
+      "step": 852
+    },
+    {
+      "epoch": 0.22437035575721706,
+      "grad_norm": 7.934420093916033,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 146478248,
+      "step": 853
+    },
+    {
+      "epoch": 0.22437035575721706,
+      "loss": 0.1829485148191452,
+      "loss_ce": 0.0025896350853145123,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 146478248,
+      "step": 853
+    },
+    {
+      "epoch": 0.2246333925166042,
+      "grad_norm": 8.916802424159366,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 146650532,
+      "step": 854
+    },
+    {
+      "epoch": 0.2246333925166042,
+      "loss": 0.13404083251953125,
+      "loss_ce": 0.0006179830525070429,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 146650532,
+      "step": 854
+    },
+    {
+      "epoch": 0.22489642927599132,
+      "grad_norm": 16.03094644250741,
+      "learning_rate": 5e-06,
+      "loss": 0.1819,
+      "num_input_tokens_seen": 146822816,
+      "step": 855
+    },
+    {
+      "epoch": 0.22489642927599132,
+      "loss": 0.19942086935043335,
+      "loss_ce": 0.007404262199997902,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 146822816,
+      "step": 855
+    },
+    {
+      "epoch": 0.22515946603537845,
+      "grad_norm": 10.96099831988347,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 146994984,
+      "step": 856
+    },
+    {
+      "epoch": 0.22515946603537845,
+      "loss": 0.1878020018339157,
+      "loss_ce": 0.0012785641010850668,
+      "loss_iou": 0.546875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 146994984,
+      "step": 856
+    },
+    {
+      "epoch": 0.22542250279476556,
+      "grad_norm": 5.009376224527154,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 147167200,
+      "step": 857
+    },
+    {
+      "epoch": 0.22542250279476556,
+      "loss": 0.16484007239341736,
+      "loss_ce": 0.005507797468453646,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 147167200,
+      "step": 857
+    },
+    {
+      "epoch": 0.2256855395541527,
+      "grad_norm": 6.031288325368203,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 147337356,
+      "step": 858
+    },
+    {
+      "epoch": 0.2256855395541527,
+      "loss": 0.14030741155147552,
+      "loss_ce": 0.0010862206108868122,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 147337356,
+      "step": 858
+    },
+    {
+      "epoch": 0.22594857631353982,
+      "grad_norm": 8.953664182124317,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 147509464,
+      "step": 859
+    },
+    {
+      "epoch": 0.22594857631353982,
+      "loss": 0.19507214426994324,
+      "loss_ce": 0.0016822540201246738,
+      "loss_iou": 0.359375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 147509464,
+      "step": 859
+    },
+    {
+      "epoch": 0.22621161307292695,
+      "grad_norm": 5.535666138038005,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 147681864,
+      "step": 860
+    },
+    {
+      "epoch": 0.22621161307292695,
+      "loss": 0.194808691740036,
+      "loss_ce": 0.002822606358677149,
+      "loss_iou": 0.5,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 147681864,
+      "step": 860
+    },
+    {
+      "epoch": 0.22647464983231408,
+      "grad_norm": 7.1540218345582,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 147854184,
+      "step": 861
+    },
+    {
+      "epoch": 0.22647464983231408,
+      "loss": 0.18444868922233582,
+      "loss_ce": 0.0037846285849809647,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 147854184,
+      "step": 861
+    },
+    {
+      "epoch": 0.22673768659170118,
+      "grad_norm": 7.954229564657017,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 148026492,
+      "step": 862
+    },
+    {
+      "epoch": 0.22673768659170118,
+      "loss": 0.1205034852027893,
+      "loss_ce": 0.0003863019519485533,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 148026492,
+      "step": 862
+    },
+    {
+      "epoch": 0.2270007233510883,
+      "grad_norm": 10.790250508426157,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 148198492,
+      "step": 863
+    },
+    {
+      "epoch": 0.2270007233510883,
+      "loss": 0.08790126442909241,
+      "loss_ce": 0.000956688541918993,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 148198492,
+      "step": 863
+    },
+    {
+      "epoch": 0.22726376011047544,
+      "grad_norm": 7.413582271316403,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 148370744,
+      "step": 864
+    },
+    {
+      "epoch": 0.22726376011047544,
+      "loss": 0.11584703624248505,
+      "loss_ce": 0.0007652430795133114,
+      "loss_iou": 0.625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 148370744,
+      "step": 864
+    },
+    {
+      "epoch": 0.22752679686986257,
+      "grad_norm": 6.505762518470224,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 148541368,
+      "step": 865
+    },
+    {
+      "epoch": 0.22752679686986257,
+      "loss": 0.12583567202091217,
+      "loss_ce": 0.0008661894826218486,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 148541368,
+      "step": 865
+    },
+    {
+      "epoch": 0.22778983362924968,
+      "grad_norm": 11.31233359994311,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 148713652,
+      "step": 866
+    },
+    {
+      "epoch": 0.22778983362924968,
+      "loss": 0.09625812619924545,
+      "loss_ce": 0.0005244807107374072,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 148713652,
+      "step": 866
+    },
+    {
+      "epoch": 0.2280528703886368,
+      "grad_norm": 8.070061922275455,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 148885724,
+      "step": 867
+    },
+    {
+      "epoch": 0.2280528703886368,
+      "loss": 0.12818799912929535,
+      "loss_ce": 0.003035409841686487,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 148885724,
+      "step": 867
+    },
+    {
+      "epoch": 0.22831590714802394,
+      "grad_norm": 8.83029409823491,
+      "learning_rate": 5e-06,
+      "loss": 0.2104,
+      "num_input_tokens_seen": 149057764,
+      "step": 868
+    },
+    {
+      "epoch": 0.22831590714802394,
+      "loss": 0.2020527571439743,
+      "loss_ce": 0.00414624810218811,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 149057764,
+      "step": 868
+    },
+    {
+      "epoch": 0.22857894390741107,
+      "grad_norm": 5.578690860970403,
+      "learning_rate": 5e-06,
+      "loss": 0.1492,
+      "num_input_tokens_seen": 149230132,
+      "step": 869
+    },
+    {
+      "epoch": 0.22857894390741107,
+      "loss": 0.16226467490196228,
+      "loss_ce": 0.0007046046666800976,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 149230132,
+      "step": 869
+    },
+    {
+      "epoch": 0.22884198066679817,
+      "grad_norm": 8.388877746304392,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 149402312,
+      "step": 870
+    },
+    {
+      "epoch": 0.22884198066679817,
+      "loss": 0.13199341297149658,
+      "loss_ce": 0.001866456470452249,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 149402312,
+      "step": 870
+    },
+    {
+      "epoch": 0.2291050174261853,
+      "grad_norm": 13.402550744349123,
+      "learning_rate": 5e-06,
+      "loss": 0.1888,
+      "num_input_tokens_seen": 149572632,
+      "step": 871
+    },
+    {
+      "epoch": 0.2291050174261853,
+      "loss": 0.20936883985996246,
+      "loss_ce": 0.004016047343611717,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 149572632,
+      "step": 871
+    },
+    {
+      "epoch": 0.22936805418557243,
+      "grad_norm": 5.899366870528114,
+      "learning_rate": 5e-06,
+      "loss": 0.1651,
+      "num_input_tokens_seen": 149744908,
+      "step": 872
+    },
+    {
+      "epoch": 0.22936805418557243,
+      "loss": 0.1176171749830246,
+      "loss_ce": 0.001223139464855194,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 149744908,
+      "step": 872
+    },
+    {
+      "epoch": 0.22963109094495956,
+      "grad_norm": 8.241998846381511,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 149917100,
+      "step": 873
+    },
+    {
+      "epoch": 0.22963109094495956,
+      "loss": 0.1720806509256363,
+      "loss_ce": 0.004966393578797579,
+      "loss_iou": 0.546875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 149917100,
+      "step": 873
+    },
+    {
+      "epoch": 0.2298941277043467,
+      "grad_norm": 9.153818862978659,
+      "learning_rate": 5e-06,
+      "loss": 0.1555,
+      "num_input_tokens_seen": 150087628,
+      "step": 874
+    },
+    {
+      "epoch": 0.2298941277043467,
+      "loss": 0.14027956128120422,
+      "loss_ce": 0.002584239235147834,
+      "loss_iou": 0.5625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 150087628,
+      "step": 874
+    },
+    {
+      "epoch": 0.2301571644637338,
+      "grad_norm": 25.107851811290338,
+      "learning_rate": 5e-06,
+      "loss": 0.1805,
+      "num_input_tokens_seen": 150258428,
+      "step": 875
+    },
+    {
+      "epoch": 0.2301571644637338,
+      "loss": 0.17914238572120667,
+      "loss_ce": 0.0014690514653921127,
+      "loss_iou": 0.5625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 150258428,
+      "step": 875
+    },
+    {
+      "epoch": 0.23042020122312093,
+      "grad_norm": 6.938403780407342,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 150430692,
+      "step": 876
+    },
+    {
+      "epoch": 0.23042020122312093,
+      "loss": 0.13967271149158478,
+      "loss_ce": 0.002038437407463789,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 150430692,
+      "step": 876
+    },
+    {
+      "epoch": 0.23068323798250806,
+      "grad_norm": 8.594672662971634,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 150603108,
+      "step": 877
+    },
+    {
+      "epoch": 0.23068323798250806,
+      "loss": 0.3066813349723816,
+      "loss_ce": 0.0005289965192787349,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 150603108,
+      "step": 877
+    },
+    {
+      "epoch": 0.2309462747418952,
+      "grad_norm": 5.613988390531258,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 150775124,
+      "step": 878
+    },
+    {
+      "epoch": 0.2309462747418952,
+      "loss": 0.16111034154891968,
+      "loss_ce": 0.008644518442451954,
+      "loss_iou": 0.640625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 150775124,
+      "step": 878
+    },
+    {
+      "epoch": 0.2312093115012823,
+      "grad_norm": 34.35737633666629,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 150947036,
+      "step": 879
+    },
+    {
+      "epoch": 0.2312093115012823,
+      "loss": 0.15406344830989838,
+      "loss_ce": 0.0015213302103802562,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 150947036,
+      "step": 879
+    },
+    {
+      "epoch": 0.23147234826066942,
+      "grad_norm": 6.1651043542261466,
+      "learning_rate": 5e-06,
+      "loss": 0.1981,
+      "num_input_tokens_seen": 151119388,
+      "step": 880
+    },
+    {
+      "epoch": 0.23147234826066942,
+      "loss": 0.29976093769073486,
+      "loss_ce": 0.00825704075396061,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 151119388,
+      "step": 880
+    },
+    {
+      "epoch": 0.23173538502005656,
+      "grad_norm": 8.982832830215536,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 151290092,
+      "step": 881
+    },
+    {
+      "epoch": 0.23173538502005656,
+      "loss": 0.18080484867095947,
+      "loss_ce": 0.0034977139439433813,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 151290092,
+      "step": 881
+    },
+    {
+      "epoch": 0.2319984217794437,
+      "grad_norm": 13.943571194131867,
+      "learning_rate": 5e-06,
+      "loss": 0.214,
+      "num_input_tokens_seen": 151462268,
+      "step": 882
+    },
+    {
+      "epoch": 0.2319984217794437,
+      "loss": 0.16014625132083893,
+      "loss_ce": 0.004170912317931652,
+      "loss_iou": 0.625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 151462268,
+      "step": 882
+    },
+    {
+      "epoch": 0.2322614585388308,
+      "grad_norm": 7.043443131491897,
+      "learning_rate": 5e-06,
+      "loss": 0.1524,
+      "num_input_tokens_seen": 151634180,
+      "step": 883
+    },
+    {
+      "epoch": 0.2322614585388308,
+      "loss": 0.12871429324150085,
+      "loss_ce": 0.0008456383948214352,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 151634180,
+      "step": 883
+    },
+    {
+      "epoch": 0.23252449529821792,
+      "grad_norm": 6.396513380644998,
+      "learning_rate": 5e-06,
+      "loss": 0.1833,
+      "num_input_tokens_seen": 151806384,
+      "step": 884
+    },
+    {
+      "epoch": 0.23252449529821792,
+      "loss": 0.18081963062286377,
+      "loss_ce": 0.002169727347791195,
+      "loss_iou": 0.421875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 151806384,
+      "step": 884
+    },
+    {
+      "epoch": 0.23278753205760505,
+      "grad_norm": 13.149227804701054,
+      "learning_rate": 5e-06,
+      "loss": 0.152,
+      "num_input_tokens_seen": 151978708,
+      "step": 885
+    },
+    {
+      "epoch": 0.23278753205760505,
+      "loss": 0.1582297682762146,
+      "loss_ce": 0.007228789385408163,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 151978708,
+      "step": 885
+    },
+    {
+      "epoch": 0.23305056881699218,
+      "grad_norm": 6.784444096578918,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 152150848,
+      "step": 886
+    },
+    {
+      "epoch": 0.23305056881699218,
+      "loss": 0.19164127111434937,
+      "loss_ce": 0.0013946772087365389,
+      "loss_iou": NaN,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 152150848,
+      "step": 886
+    },
+    {
+      "epoch": 0.2333136055763793,
+      "grad_norm": 6.295673229022739,
+      "learning_rate": 5e-06,
+      "loss": 0.1772,
+      "num_input_tokens_seen": 152323052,
+      "step": 887
+    },
+    {
+      "epoch": 0.2333136055763793,
+      "loss": 0.22069396078586578,
+      "loss_ce": 0.0015167115489020944,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 152323052,
+      "step": 887
+    },
+    {
+      "epoch": 0.23357664233576642,
+      "grad_norm": 11.781517950000717,
+      "learning_rate": 5e-06,
+      "loss": 0.1798,
+      "num_input_tokens_seen": 152493800,
+      "step": 888
+    },
+    {
+      "epoch": 0.23357664233576642,
+      "loss": 0.1682368516921997,
+      "loss_ce": 0.0009394832304678857,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 152493800,
+      "step": 888
+    },
+    {
+      "epoch": 0.23383967909515355,
+      "grad_norm": 11.368208703500017,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 152665776,
+      "step": 889
+    },
+    {
+      "epoch": 0.23383967909515355,
+      "loss": 0.2839386761188507,
+      "loss_ce": 0.003970403224229813,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 152665776,
+      "step": 889
+    },
+    {
+      "epoch": 0.23410271585454068,
+      "grad_norm": 8.093455275529028,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 152834764,
+      "step": 890
+    },
+    {
+      "epoch": 0.23410271585454068,
+      "loss": 0.1241624653339386,
+      "loss_ce": 0.004472525790333748,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 152834764,
+      "step": 890
+    },
+    {
+      "epoch": 0.2343657526139278,
+      "grad_norm": 12.52857780858346,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 153007304,
+      "step": 891
+    },
+    {
+      "epoch": 0.2343657526139278,
+      "loss": 0.21788766980171204,
+      "loss_ce": 0.002006314927712083,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 153007304,
+      "step": 891
+    },
+    {
+      "epoch": 0.2346287893733149,
+      "grad_norm": 7.50091412461157,
+      "learning_rate": 5e-06,
+      "loss": 0.1804,
+      "num_input_tokens_seen": 153179168,
+      "step": 892
+    },
+    {
+      "epoch": 0.2346287893733149,
+      "loss": 0.17867043614387512,
+      "loss_ce": 0.0008750315755605698,
+      "loss_iou": NaN,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 153179168,
+      "step": 892
+    },
+    {
+      "epoch": 0.23489182613270204,
+      "grad_norm": 6.297998473336444,
+      "learning_rate": 5e-06,
+      "loss": 0.1679,
+      "num_input_tokens_seen": 153351220,
+      "step": 893
+    },
+    {
+      "epoch": 0.23489182613270204,
+      "loss": 0.13165241479873657,
+      "loss_ce": 0.001311830012127757,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 153351220,
+      "step": 893
+    },
+    {
+      "epoch": 0.23515486289208917,
+      "grad_norm": 7.909655629561448,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 153523232,
+      "step": 894
+    },
+    {
+      "epoch": 0.23515486289208917,
+      "loss": 0.13778507709503174,
+      "loss_ce": 0.002287032548338175,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 153523232,
+      "step": 894
+    },
+    {
+      "epoch": 0.2354178996514763,
+      "grad_norm": 5.883815728394397,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 153692208,
+      "step": 895
+    },
+    {
+      "epoch": 0.2354178996514763,
+      "loss": 0.1446894109249115,
+      "loss_ce": 0.001439890475012362,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 153692208,
+      "step": 895
+    },
+    {
+      "epoch": 0.2356809364108634,
+      "grad_norm": 15.735436674991021,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 153862388,
+      "step": 896
+    },
+    {
+      "epoch": 0.2356809364108634,
+      "loss": 0.16044028103351593,
+      "loss_ce": 0.00043661610106937587,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 153862388,
+      "step": 896
+    },
+    {
+      "epoch": 0.23594397317025054,
+      "grad_norm": 5.835513921908954,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 154034480,
+      "step": 897
+    },
+    {
+      "epoch": 0.23594397317025054,
+      "loss": 0.14927011728286743,
+      "loss_ce": 0.0002833124599419534,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 154034480,
+      "step": 897
+    },
+    {
+      "epoch": 0.23620700992963767,
+      "grad_norm": 4.347875981694168,
+      "learning_rate": 5e-06,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 154203000,
+      "step": 898
+    },
+    {
+      "epoch": 0.23620700992963767,
+      "loss": 0.1722668707370758,
+      "loss_ce": 0.0003613463486544788,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 154203000,
+      "step": 898
+    },
+    {
+      "epoch": 0.2364700466890248,
+      "grad_norm": 6.555211118822418,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 154375292,
+      "step": 899
+    },
+    {
+      "epoch": 0.2364700466890248,
+      "loss": 0.12109389901161194,
+      "loss_ce": 0.001983799273148179,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 154375292,
+      "step": 899
+    },
+    {
+      "epoch": 0.2367330834484119,
+      "grad_norm": 4.963219405700268,
+      "learning_rate": 5e-06,
+      "loss": 0.1507,
+      "num_input_tokens_seen": 154547472,
+      "step": 900
+    },
+    {
+      "epoch": 0.2367330834484119,
+      "loss": 0.1948363184928894,
+      "loss_ce": 0.001843146630562842,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 154547472,
+      "step": 900
+    },
+    {
+      "epoch": 0.23699612020779903,
+      "grad_norm": 9.476812377662082,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 154719756,
+      "step": 901
+    },
+    {
+      "epoch": 0.23699612020779903,
+      "loss": 0.1722353994846344,
+      "loss_ce": 0.0028628362342715263,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 154719756,
+      "step": 901
+    },
+    {
+      "epoch": 0.23725915696718616,
+      "grad_norm": 8.867901615298983,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 154891816,
+      "step": 902
+    },
+    {
+      "epoch": 0.23725915696718616,
+      "loss": 0.13942725956439972,
+      "loss_ce": 0.0014267791993916035,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 154891816,
+      "step": 902
+    },
+    {
+      "epoch": 0.2375221937265733,
+      "grad_norm": 7.06026317137913,
+      "learning_rate": 5e-06,
+      "loss": 0.1247,
+      "num_input_tokens_seen": 155064164,
+      "step": 903
+    },
+    {
+      "epoch": 0.2375221937265733,
+      "loss": 0.13094615936279297,
+      "loss_ce": 0.000575068814214319,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 155064164,
+      "step": 903
+    },
+    {
+      "epoch": 0.23778523048596042,
+      "grad_norm": 6.065414379643311,
+      "learning_rate": 5e-06,
+      "loss": 0.1882,
+      "num_input_tokens_seen": 155236136,
+      "step": 904
+    },
+    {
+      "epoch": 0.23778523048596042,
+      "loss": 0.17110927402973175,
+      "loss_ce": 0.0006380859995260835,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 155236136,
+      "step": 904
+    },
+    {
+      "epoch": 0.23804826724534753,
+      "grad_norm": 6.4154554676892275,
+      "learning_rate": 5e-06,
+      "loss": 0.1664,
+      "num_input_tokens_seen": 155408400,
+      "step": 905
+    },
+    {
+      "epoch": 0.23804826724534753,
+      "loss": 0.18800213932991028,
+      "loss_ce": 0.005415464285761118,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 155408400,
+      "step": 905
+    },
+    {
+      "epoch": 0.23831130400473466,
+      "grad_norm": 6.061588736644807,
+      "learning_rate": 5e-06,
+      "loss": 0.1656,
+      "num_input_tokens_seen": 155578772,
+      "step": 906
+    },
+    {
+      "epoch": 0.23831130400473466,
+      "loss": 0.11921393871307373,
+      "loss_ce": 0.0009278038050979376,
+      "loss_iou": 0.609375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 155578772,
+      "step": 906
+    },
+    {
+      "epoch": 0.2385743407641218,
+      "grad_norm": 9.748504628242028,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 155751088,
+      "step": 907
+    },
+    {
+      "epoch": 0.2385743407641218,
+      "loss": 0.13661867380142212,
+      "loss_ce": 0.0004492364823818207,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 155751088,
+      "step": 907
+    },
+    {
+      "epoch": 0.23883737752350892,
+      "grad_norm": 11.132080080363805,
+      "learning_rate": 5e-06,
+      "loss": 0.2072,
+      "num_input_tokens_seen": 155923228,
+      "step": 908
+    },
+    {
+      "epoch": 0.23883737752350892,
+      "loss": 0.2608073353767395,
+      "loss_ce": 0.006229718215763569,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 155923228,
+      "step": 908
+    },
+    {
+      "epoch": 0.23910041428289602,
+      "grad_norm": 6.802731098529186,
+      "learning_rate": 5e-06,
+      "loss": 0.1071,
+      "num_input_tokens_seen": 156095144,
+      "step": 909
+    },
+    {
+      "epoch": 0.23910041428289602,
+      "loss": 0.12534289062023163,
+      "loss_ce": 0.0022044701036065817,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 156095144,
+      "step": 909
+    },
+    {
+      "epoch": 0.23936345104228315,
+      "grad_norm": 5.86214467113572,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 156267348,
+      "step": 910
+    },
+    {
+      "epoch": 0.23936345104228315,
+      "loss": 0.1707063615322113,
+      "loss_ce": 0.0013948287814855576,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 156267348,
+      "step": 910
+    },
+    {
+      "epoch": 0.23962648780167028,
+      "grad_norm": 5.562832964795879,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 156439364,
+      "step": 911
+    },
+    {
+      "epoch": 0.23962648780167028,
+      "loss": 0.12454073876142502,
+      "loss_ce": 0.0003036795533262193,
+      "loss_iou": 0.578125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 156439364,
+      "step": 911
+    },
+    {
+      "epoch": 0.23988952456105742,
+      "grad_norm": 13.223359809657884,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 156609552,
+      "step": 912
+    },
+    {
+      "epoch": 0.23988952456105742,
+      "loss": 0.15059047937393188,
+      "loss_ce": 0.00239713117480278,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 156609552,
+      "step": 912
+    },
+    {
+      "epoch": 0.24015256132044452,
+      "grad_norm": 6.777584209416996,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 156781920,
+      "step": 913
+    },
+    {
+      "epoch": 0.24015256132044452,
+      "loss": 0.13880465924739838,
+      "loss_ce": 0.0020859187934547663,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 156781920,
+      "step": 913
+    },
+    {
+      "epoch": 0.24041559807983165,
+      "grad_norm": 11.055320488873154,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 156954340,
+      "step": 914
+    },
+    {
+      "epoch": 0.24041559807983165,
+      "loss": 0.2452932596206665,
+      "loss_ce": 0.002312319353222847,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 156954340,
+      "step": 914
+    },
+    {
+      "epoch": 0.24067863483921878,
+      "grad_norm": 4.458423775022664,
+      "learning_rate": 5e-06,
+      "loss": 0.1784,
+      "num_input_tokens_seen": 157126652,
+      "step": 915
+    },
+    {
+      "epoch": 0.24067863483921878,
+      "loss": 0.13856951892375946,
+      "loss_ce": 0.0020949181634932756,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 157126652,
+      "step": 915
+    },
+    {
+      "epoch": 0.2409416715986059,
+      "grad_norm": 5.681838115677692,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 157298968,
+      "step": 916
+    },
+    {
+      "epoch": 0.2409416715986059,
+      "loss": 0.1539350152015686,
+      "loss_ce": 0.00638251006603241,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 157298968,
+      "step": 916
+    },
+    {
+      "epoch": 0.24120470835799304,
+      "grad_norm": 5.180460717860643,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 157471004,
+      "step": 917
+    },
+    {
+      "epoch": 0.24120470835799304,
+      "loss": 0.10441954433917999,
+      "loss_ce": 0.0008428902365267277,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 157471004,
+      "step": 917
+    },
+    {
+      "epoch": 0.24146774511738014,
+      "grad_norm": 10.247033875410487,
+      "learning_rate": 5e-06,
+      "loss": 0.161,
+      "num_input_tokens_seen": 157643232,
+      "step": 918
+    },
+    {
+      "epoch": 0.24146774511738014,
+      "loss": 0.18021947145462036,
+      "loss_ce": 0.002607175149023533,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 157643232,
+      "step": 918
+    },
+    {
+      "epoch": 0.24173078187676728,
+      "grad_norm": 6.8962184523908725,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 157815292,
+      "step": 919
+    },
+    {
+      "epoch": 0.24173078187676728,
+      "loss": 0.1455521434545517,
+      "loss_ce": 0.0008988262270577252,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 157815292,
+      "step": 919
+    },
+    {
+      "epoch": 0.2419938186361544,
+      "grad_norm": 6.7883399023716775,
+      "learning_rate": 5e-06,
+      "loss": 0.1916,
+      "num_input_tokens_seen": 157987716,
+      "step": 920
+    },
+    {
+      "epoch": 0.2419938186361544,
+      "loss": 0.19279745221138,
+      "loss_ce": 0.0026729374658316374,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 157987716,
+      "step": 920
+    },
+    {
+      "epoch": 0.24225685539554154,
+      "grad_norm": 7.426866121442803,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 158160224,
+      "step": 921
+    },
+    {
+      "epoch": 0.24225685539554154,
+      "loss": 0.1912505030632019,
+      "loss_ce": 0.001980474451556802,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 158160224,
+      "step": 921
+    },
+    {
+      "epoch": 0.24251989215492864,
+      "grad_norm": 7.602353481412061,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 158332464,
+      "step": 922
+    },
+    {
+      "epoch": 0.24251989215492864,
+      "loss": 0.1421346664428711,
+      "loss_ce": 0.003951081074774265,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 158332464,
+      "step": 922
+    },
+    {
+      "epoch": 0.24278292891431577,
+      "grad_norm": 9.798997838296735,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 158504560,
+      "step": 923
+    },
+    {
+      "epoch": 0.24278292891431577,
+      "loss": 0.15015605092048645,
+      "loss_ce": 0.0013523304369300604,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 158504560,
+      "step": 923
+    },
+    {
+      "epoch": 0.2430459656737029,
+      "grad_norm": 9.276091083653826,
+      "learning_rate": 5e-06,
+      "loss": 0.1582,
+      "num_input_tokens_seen": 158676676,
+      "step": 924
+    },
+    {
+      "epoch": 0.2430459656737029,
+      "loss": 0.13057658076286316,
+      "loss_ce": 0.0024943118914961815,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 158676676,
+      "step": 924
+    },
+    {
+      "epoch": 0.24330900243309003,
+      "grad_norm": 7.79676627099927,
+      "learning_rate": 5e-06,
+      "loss": 0.1547,
+      "num_input_tokens_seen": 158849084,
+      "step": 925
+    },
+    {
+      "epoch": 0.24330900243309003,
+      "loss": 0.1701420098543167,
+      "loss_ce": 0.004248456098139286,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 158849084,
+      "step": 925
+    },
+    {
+      "epoch": 0.24357203919247714,
+      "grad_norm": 7.627935616299721,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 159019648,
+      "step": 926
+    },
+    {
+      "epoch": 0.24357203919247714,
+      "loss": 0.09866867959499359,
+      "loss_ce": 0.002080547623336315,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 159019648,
+      "step": 926
+    },
+    {
+      "epoch": 0.24383507595186427,
+      "grad_norm": 7.674777640749283,
+      "learning_rate": 5e-06,
+      "loss": 0.1837,
+      "num_input_tokens_seen": 159191744,
+      "step": 927
+    },
+    {
+      "epoch": 0.24383507595186427,
+      "loss": 0.09513729810714722,
+      "loss_ce": 0.0016314350068569183,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 159191744,
+      "step": 927
+    },
+    {
+      "epoch": 0.2440981127112514,
+      "grad_norm": 8.029174499906352,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 159364172,
+      "step": 928
+    },
+    {
+      "epoch": 0.2440981127112514,
+      "loss": 0.2412642240524292,
+      "loss_ce": 0.0012739873491227627,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 159364172,
+      "step": 928
+    },
+    {
+      "epoch": 0.24436114947063853,
+      "grad_norm": 3.8948353119563537,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 159534608,
+      "step": 929
+    },
+    {
+      "epoch": 0.24436114947063853,
+      "loss": 0.11882825195789337,
+      "loss_ce": 0.0011219491716474295,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 159534608,
+      "step": 929
+    },
+    {
+      "epoch": 0.24462418623002566,
+      "grad_norm": 12.725190215625348,
+      "learning_rate": 5e-06,
+      "loss": 0.1728,
+      "num_input_tokens_seen": 159705228,
+      "step": 930
+    },
+    {
+      "epoch": 0.24462418623002566,
+      "loss": 0.15004633367061615,
+      "loss_ce": 0.000662794045638293,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 159705228,
+      "step": 930
+    },
+    {
+      "epoch": 0.24488722298941276,
+      "grad_norm": 8.38127835969171,
+      "learning_rate": 5e-06,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 159875324,
+      "step": 931
+    },
+    {
+      "epoch": 0.24488722298941276,
+      "loss": 0.1881195604801178,
+      "loss_ce": 0.0022675050422549248,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 159875324,
+      "step": 931
+    },
+    {
+      "epoch": 0.2451502597487999,
+      "grad_norm": 8.09047633001587,
+      "learning_rate": 5e-06,
+      "loss": 0.1537,
+      "num_input_tokens_seen": 160045932,
+      "step": 932
+    },
+    {
+      "epoch": 0.2451502597487999,
+      "loss": 0.23839473724365234,
+      "loss_ce": 0.0010290088830515742,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 160045932,
+      "step": 932
+    },
+    {
+      "epoch": 0.24541329650818702,
+      "grad_norm": 14.720243529942747,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 160218252,
+      "step": 933
+    },
+    {
+      "epoch": 0.24541329650818702,
+      "loss": 0.12284128367900848,
+      "loss_ce": 0.0022663308773189783,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 160218252,
+      "step": 933
+    },
+    {
+      "epoch": 0.24567633326757415,
+      "grad_norm": 6.031479541696206,
+      "learning_rate": 5e-06,
+      "loss": 0.1669,
+      "num_input_tokens_seen": 160390560,
+      "step": 934
+    },
+    {
+      "epoch": 0.24567633326757415,
+      "loss": 0.1301022320985794,
+      "loss_ce": 0.0005245967186056077,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 160390560,
+      "step": 934
+    },
+    {
+      "epoch": 0.24593937002696126,
+      "grad_norm": 6.383554643597018,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 160562528,
+      "step": 935
+    },
+    {
+      "epoch": 0.24593937002696126,
+      "loss": 0.17644909024238586,
+      "loss_ce": 0.0030787207651883364,
+      "loss_iou": 0.375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 160562528,
+      "step": 935
+    },
+    {
+      "epoch": 0.2462024067863484,
+      "grad_norm": 20.862114065213355,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 160734384,
+      "step": 936
+    },
+    {
+      "epoch": 0.2462024067863484,
+      "loss": 0.11419504880905151,
+      "loss_ce": 0.0004255172680132091,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 160734384,
+      "step": 936
+    },
+    {
+      "epoch": 0.24646544354573552,
+      "grad_norm": 8.0386150468435,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 160906448,
+      "step": 937
+    },
+    {
+      "epoch": 0.24646544354573552,
+      "loss": 0.14023496210575104,
+      "loss_ce": 0.0016241249395534396,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 160906448,
+      "step": 937
+    },
+    {
+      "epoch": 0.24672848030512265,
+      "grad_norm": 7.276863008633557,
+      "learning_rate": 5e-06,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 161078828,
+      "step": 938
+    },
+    {
+      "epoch": 0.24672848030512265,
+      "loss": 0.21741217374801636,
+      "loss_ce": 0.004765682853758335,
+      "loss_iou": 0.7265625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 161078828,
+      "step": 938
+    },
+    {
+      "epoch": 0.24699151706450975,
+      "grad_norm": 10.285417073408015,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 161249228,
+      "step": 939
+    },
+    {
+      "epoch": 0.24699151706450975,
+      "loss": 0.2533302903175354,
+      "loss_ce": 0.004825636278837919,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 161249228,
+      "step": 939
+    },
+    {
+      "epoch": 0.24725455382389688,
+      "grad_norm": 7.503087196122763,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 161421624,
+      "step": 940
+    },
+    {
+      "epoch": 0.24725455382389688,
+      "loss": 0.1497778743505478,
+      "loss_ce": 0.003537639044225216,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 161421624,
+      "step": 940
+    },
+    {
+      "epoch": 0.24751759058328401,
+      "grad_norm": 6.078271645066026,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 161593500,
+      "step": 941
+    },
+    {
+      "epoch": 0.24751759058328401,
+      "loss": 0.15894815325737,
+      "loss_ce": 0.0006839816924184561,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 161593500,
+      "step": 941
+    },
+    {
+      "epoch": 0.24778062734267114,
+      "grad_norm": 16.470075984430842,
+      "learning_rate": 5e-06,
+      "loss": 0.1725,
+      "num_input_tokens_seen": 161764108,
+      "step": 942
+    },
+    {
+      "epoch": 0.24778062734267114,
+      "loss": 0.2659192383289337,
+      "loss_ce": 0.003468066919595003,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 161764108,
+      "step": 942
+    },
+    {
+      "epoch": 0.24804366410205828,
+      "grad_norm": 18.319111638013048,
+      "learning_rate": 5e-06,
+      "loss": 0.1876,
+      "num_input_tokens_seen": 161936252,
+      "step": 943
+    },
+    {
+      "epoch": 0.24804366410205828,
+      "loss": 0.23192915320396423,
+      "loss_ce": 0.004817330744117498,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 161936252,
+      "step": 943
+    },
+    {
+      "epoch": 0.24830670086144538,
+      "grad_norm": 11.296239885199183,
+      "learning_rate": 5e-06,
+      "loss": 0.1547,
+      "num_input_tokens_seen": 162108564,
+      "step": 944
+    },
+    {
+      "epoch": 0.24830670086144538,
+      "loss": 0.13432571291923523,
+      "loss_ce": 0.002459259470924735,
+      "loss_iou": 0.7578125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 162108564,
+      "step": 944
+    },
+    {
+      "epoch": 0.2485697376208325,
+      "grad_norm": 5.511055668564898,
+      "learning_rate": 5e-06,
+      "loss": 0.1393,
+      "num_input_tokens_seen": 162280792,
+      "step": 945
+    },
+    {
+      "epoch": 0.2485697376208325,
+      "loss": 0.13674385845661163,
+      "loss_ce": 0.0015509906224906445,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 162280792,
+      "step": 945
+    },
+    {
+      "epoch": 0.24883277438021964,
+      "grad_norm": 10.355301687463253,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 162449528,
+      "step": 946
+    },
+    {
+      "epoch": 0.24883277438021964,
+      "loss": 0.2201877236366272,
+      "loss_ce": 0.0006442689918912947,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 162449528,
+      "step": 946
+    },
+    {
+      "epoch": 0.24909581113960677,
+      "grad_norm": 11.25377814108321,
+      "learning_rate": 5e-06,
+      "loss": 0.1493,
+      "num_input_tokens_seen": 162617256,
+      "step": 947
+    },
+    {
+      "epoch": 0.24909581113960677,
+      "loss": 0.188047856092453,
+      "loss_ce": 0.0031418518628925085,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 162617256,
+      "step": 947
+    },
+    {
+      "epoch": 0.24935884789899387,
+      "grad_norm": 12.235451694711202,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 162789400,
+      "step": 948
+    },
+    {
+      "epoch": 0.24935884789899387,
+      "loss": 0.15606345236301422,
+      "loss_ce": 0.005215056240558624,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 162789400,
+      "step": 948
+    },
+    {
+      "epoch": 0.249621884658381,
+      "grad_norm": 7.1804139994510034,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 162961432,
+      "step": 949
+    },
+    {
+      "epoch": 0.249621884658381,
+      "loss": 0.10785458981990814,
+      "loss_ce": 0.002507905475795269,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 162961432,
+      "step": 949
+    },
+    {
+      "epoch": 0.24988492141776814,
+      "grad_norm": 14.388122342778688,
+      "learning_rate": 5e-06,
+      "loss": 0.1883,
+      "num_input_tokens_seen": 163133524,
+      "step": 950
+    },
+    {
+      "epoch": 0.24988492141776814,
+      "loss": 0.16639769077301025,
+      "loss_ce": 0.001511220121756196,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 163133524,
+      "step": 950
+    },
+    {
+      "epoch": 0.25014795817715524,
+      "grad_norm": 10.331941667190199,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 163305816,
+      "step": 951
+    },
+    {
+      "epoch": 0.25014795817715524,
+      "loss": 0.1127045676112175,
+      "loss_ce": 0.0008271271362900734,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 163305816,
+      "step": 951
+    },
+    {
+      "epoch": 0.2504109949365424,
+      "grad_norm": 6.829599375925184,
+      "learning_rate": 5e-06,
+      "loss": 0.154,
+      "num_input_tokens_seen": 163478368,
+      "step": 952
+    },
+    {
+      "epoch": 0.2504109949365424,
+      "loss": 0.1521347612142563,
+      "loss_ce": 0.0029038134962320328,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 163478368,
+      "step": 952
+    },
+    {
+      "epoch": 0.2506740316959295,
+      "grad_norm": 5.382923356179618,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 163650500,
+      "step": 953
+    },
+    {
+      "epoch": 0.2506740316959295,
+      "loss": 0.14390447735786438,
+      "loss_ce": 0.00224187970161438,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 163650500,
+      "step": 953
+    },
+    {
+      "epoch": 0.2509370684553166,
+      "grad_norm": 4.693926744893699,
+      "learning_rate": 5e-06,
+      "loss": 0.1612,
+      "num_input_tokens_seen": 163822252,
+      "step": 954
+    },
+    {
+      "epoch": 0.2509370684553166,
+      "loss": 0.1426747441291809,
+      "loss_ce": 0.001286811544559896,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 163822252,
+      "step": 954
+    },
+    {
+      "epoch": 0.25120010521470376,
+      "grad_norm": 6.416178956242753,
+      "learning_rate": 5e-06,
+      "loss": 0.1743,
+      "num_input_tokens_seen": 163992628,
+      "step": 955
+    },
+    {
+      "epoch": 0.25120010521470376,
+      "loss": 0.2167130708694458,
+      "loss_ce": 0.0013200179673731327,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 163992628,
+      "step": 955
+    },
+    {
+      "epoch": 0.25146314197409086,
+      "grad_norm": 8.38564274924814,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 164164972,
+      "step": 956
+    },
+    {
+      "epoch": 0.25146314197409086,
+      "loss": 0.14087224006652832,
+      "loss_ce": 0.0018951823003590107,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 164164972,
+      "step": 956
+    },
+    {
+      "epoch": 0.251726178733478,
+      "grad_norm": 10.557767516476979,
+      "learning_rate": 5e-06,
+      "loss": 0.1952,
+      "num_input_tokens_seen": 164334056,
+      "step": 957
+    },
+    {
+      "epoch": 0.251726178733478,
+      "loss": 0.10351097583770752,
+      "loss_ce": 0.003108144039288163,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 164334056,
+      "step": 957
+    },
+    {
+      "epoch": 0.2519892154928651,
+      "grad_norm": 6.187664908064733,
+      "learning_rate": 5e-06,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 164506496,
+      "step": 958
+    },
+    {
+      "epoch": 0.2519892154928651,
+      "loss": 0.15677396953105927,
+      "loss_ce": 0.005193163640797138,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 164506496,
+      "step": 958
+    },
+    {
+      "epoch": 0.25225225225225223,
+      "grad_norm": 9.49095916832686,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 164678644,
+      "step": 959
+    },
+    {
+      "epoch": 0.25225225225225223,
+      "loss": 0.18635977804660797,
+      "loss_ce": 0.002003092784434557,
+      "loss_iou": 0.5,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 164678644,
+      "step": 959
+    },
+    {
+      "epoch": 0.2525152890116394,
+      "grad_norm": 6.024170228931713,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 164851184,
+      "step": 960
+    },
+    {
+      "epoch": 0.2525152890116394,
+      "loss": 0.13998761773109436,
+      "loss_ce": 0.003940259106457233,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 164851184,
+      "step": 960
+    },
+    {
+      "epoch": 0.2527783257710265,
+      "grad_norm": 6.120337923811076,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 165023544,
+      "step": 961
+    },
+    {
+      "epoch": 0.2527783257710265,
+      "loss": 0.09865675866603851,
+      "loss_ce": 0.003258807584643364,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 165023544,
+      "step": 961
+    },
+    {
+      "epoch": 0.25304136253041365,
+      "grad_norm": 9.066155915123039,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 165193832,
+      "step": 962
+    },
+    {
+      "epoch": 0.25304136253041365,
+      "loss": 0.1392088085412979,
+      "loss_ce": 0.0009946945356205106,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 165193832,
+      "step": 962
+    },
+    {
+      "epoch": 0.25330439928980075,
+      "grad_norm": 7.472497977314338,
+      "learning_rate": 5e-06,
+      "loss": 0.1704,
+      "num_input_tokens_seen": 165365892,
+      "step": 963
+    },
+    {
+      "epoch": 0.25330439928980075,
+      "loss": 0.23057040572166443,
+      "loss_ce": 0.002054777694866061,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 165365892,
+      "step": 963
+    },
+    {
+      "epoch": 0.25356743604918786,
+      "grad_norm": 8.732932808526876,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 165538272,
+      "step": 964
+    },
+    {
+      "epoch": 0.25356743604918786,
+      "loss": 0.16193270683288574,
+      "loss_ce": 0.0018985318019986153,
+      "loss_iou": 0.640625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 165538272,
+      "step": 964
+    },
+    {
+      "epoch": 0.253830472808575,
+      "grad_norm": 7.508690685861332,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 165710728,
+      "step": 965
+    },
+    {
+      "epoch": 0.253830472808575,
+      "loss": 0.16999930143356323,
+      "loss_ce": 0.002335726749151945,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 165710728,
+      "step": 965
+    },
+    {
+      "epoch": 0.2540935095679621,
+      "grad_norm": 8.520838856927949,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 165882912,
+      "step": 966
+    },
+    {
+      "epoch": 0.2540935095679621,
+      "loss": 0.1505521535873413,
+      "loss_ce": 0.0006498107686638832,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 165882912,
+      "step": 966
+    },
+    {
+      "epoch": 0.2543565463273492,
+      "grad_norm": 7.07336887298293,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 166053292,
+      "step": 967
+    },
+    {
+      "epoch": 0.2543565463273492,
+      "loss": 0.13186028599739075,
+      "loss_ce": 0.002404727740213275,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 166053292,
+      "step": 967
+    },
+    {
+      "epoch": 0.2546195830867364,
+      "grad_norm": 5.861693184502666,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 166222324,
+      "step": 968
+    },
+    {
+      "epoch": 0.2546195830867364,
+      "loss": 0.12424527108669281,
+      "loss_ce": 0.0015340839745476842,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 166222324,
+      "step": 968
+    },
+    {
+      "epoch": 0.2548826198461235,
+      "grad_norm": 4.8001560190338335,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 166394732,
+      "step": 969
+    },
+    {
+      "epoch": 0.2548826198461235,
+      "loss": 0.13953326642513275,
+      "loss_ce": 0.0008613896206952631,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 166394732,
+      "step": 969
+    },
+    {
+      "epoch": 0.25514565660551064,
+      "grad_norm": 4.5364064648643065,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 166566920,
+      "step": 970
+    },
+    {
+      "epoch": 0.25514565660551064,
+      "loss": 0.15965688228607178,
+      "loss_ce": 0.0016368532087653875,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 166566920,
+      "step": 970
+    },
+    {
+      "epoch": 0.25540869336489774,
+      "grad_norm": 6.371608621088164,
+      "learning_rate": 5e-06,
+      "loss": 0.1801,
+      "num_input_tokens_seen": 166739024,
+      "step": 971
+    },
+    {
+      "epoch": 0.25540869336489774,
+      "loss": 0.2314717024564743,
+      "loss_ce": 0.003017107956111431,
+      "loss_iou": 0.5625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 166739024,
+      "step": 971
+    },
+    {
+      "epoch": 0.25567173012428485,
+      "grad_norm": 10.601164768004658,
+      "learning_rate": 5e-06,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 166911376,
+      "step": 972
+    },
+    {
+      "epoch": 0.25567173012428485,
+      "loss": 0.16122400760650635,
+      "loss_ce": 0.001586547470651567,
+      "loss_iou": 0.75390625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 166911376,
+      "step": 972
+    },
+    {
+      "epoch": 0.255934766883672,
+      "grad_norm": 7.052937396329928,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 167083504,
+      "step": 973
+    },
+    {
+      "epoch": 0.255934766883672,
+      "loss": 0.09488484263420105,
+      "loss_ce": 0.0012263880344107747,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 167083504,
+      "step": 973
+    },
+    {
+      "epoch": 0.2561978036430591,
+      "grad_norm": 11.500260657406463,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 167255708,
+      "step": 974
+    },
+    {
+      "epoch": 0.2561978036430591,
+      "loss": 0.08826225996017456,
+      "loss_ce": 0.0011956070084124804,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 167255708,
+      "step": 974
+    },
+    {
+      "epoch": 0.25646084040244627,
+      "grad_norm": 15.92469225824967,
+      "learning_rate": 5e-06,
+      "loss": 0.1805,
+      "num_input_tokens_seen": 167428028,
+      "step": 975
+    },
+    {
+      "epoch": 0.25646084040244627,
+      "loss": 0.2100502997636795,
+      "loss_ce": 0.0013405811041593552,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 167428028,
+      "step": 975
+    },
+    {
+      "epoch": 0.25672387716183337,
+      "grad_norm": 8.58380851703317,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 167600008,
+      "step": 976
+    },
+    {
+      "epoch": 0.25672387716183337,
+      "loss": 0.16451242566108704,
+      "loss_ce": 0.0012128613889217377,
+      "loss_iou": 0.73046875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 167600008,
+      "step": 976
+    },
+    {
+      "epoch": 0.2569869139212205,
+      "grad_norm": 18.1690929466902,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 167772344,
+      "step": 977
+    },
+    {
+      "epoch": 0.2569869139212205,
+      "loss": 0.1349577009677887,
+      "loss_ce": 0.0059293946251273155,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 167772344,
+      "step": 977
+    },
+    {
+      "epoch": 0.25724995068060763,
+      "grad_norm": 19.89834413612689,
+      "learning_rate": 5e-06,
+      "loss": 0.1866,
+      "num_input_tokens_seen": 167944388,
+      "step": 978
+    },
+    {
+      "epoch": 0.25724995068060763,
+      "loss": 0.2309381365776062,
+      "loss_ce": 0.0008356063044629991,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 167944388,
+      "step": 978
+    },
+    {
+      "epoch": 0.25751298743999473,
+      "grad_norm": 9.124352830191828,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 168116168,
+      "step": 979
+    },
+    {
+      "epoch": 0.25751298743999473,
+      "loss": 0.11332334578037262,
+      "loss_ce": 0.0035516121424734592,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 168116168,
+      "step": 979
+    },
+    {
+      "epoch": 0.25777602419938184,
+      "grad_norm": 13.324870310905945,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 168288220,
+      "step": 980
+    },
+    {
+      "epoch": 0.25777602419938184,
+      "loss": 0.14565590023994446,
+      "loss_ce": 0.0014908593147993088,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 168288220,
+      "step": 980
+    },
+    {
+      "epoch": 0.258039060958769,
+      "grad_norm": 12.888295473836926,
+      "learning_rate": 5e-06,
+      "loss": 0.147,
+      "num_input_tokens_seen": 168460592,
+      "step": 981
+    },
+    {
+      "epoch": 0.258039060958769,
+      "loss": 0.13789984583854675,
+      "loss_ce": 0.0010895461309701204,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 168460592,
+      "step": 981
+    },
+    {
+      "epoch": 0.2583020977181561,
+      "grad_norm": 7.142842896707461,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 168632604,
+      "step": 982
+    },
+    {
+      "epoch": 0.2583020977181561,
+      "loss": 0.2077009379863739,
+      "loss_ce": 0.0013105443213135004,
+      "loss_iou": 0.59375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 168632604,
+      "step": 982
+    },
+    {
+      "epoch": 0.25856513447754326,
+      "grad_norm": 5.171300059065281,
+      "learning_rate": 5e-06,
+      "loss": 0.194,
+      "num_input_tokens_seen": 168804468,
+      "step": 983
+    },
+    {
+      "epoch": 0.25856513447754326,
+      "loss": 0.15598775446414948,
+      "loss_ce": 0.005291945766657591,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 168804468,
+      "step": 983
+    },
+    {
+      "epoch": 0.25882817123693036,
+      "grad_norm": 11.204489011375072,
+      "learning_rate": 5e-06,
+      "loss": 0.1807,
+      "num_input_tokens_seen": 168976856,
+      "step": 984
+    },
+    {
+      "epoch": 0.25882817123693036,
+      "loss": 0.19184689223766327,
+      "loss_ce": 0.0026989425532519817,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 168976856,
+      "step": 984
+    },
+    {
+      "epoch": 0.25909120799631746,
+      "grad_norm": 6.6174997360939205,
+      "learning_rate": 5e-06,
+      "loss": 0.1608,
+      "num_input_tokens_seen": 169149104,
+      "step": 985
+    },
+    {
+      "epoch": 0.25909120799631746,
+      "loss": 0.23251253366470337,
+      "loss_ce": 0.0004568799340631813,
+      "loss_iou": 0.578125,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 169149104,
+      "step": 985
+    },
+    {
+      "epoch": 0.2593542447557046,
+      "grad_norm": 7.169144370545132,
+      "learning_rate": 5e-06,
+      "loss": 0.1498,
+      "num_input_tokens_seen": 169321296,
+      "step": 986
+    },
+    {
+      "epoch": 0.2593542447557046,
+      "loss": 0.12904971837997437,
+      "loss_ce": 0.0024627982638776302,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 169321296,
+      "step": 986
+    },
+    {
+      "epoch": 0.2596172815150917,
+      "grad_norm": 17.414290331962896,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 169493420,
+      "step": 987
+    },
+    {
+      "epoch": 0.2596172815150917,
+      "loss": 0.13824069499969482,
+      "loss_ce": 0.0014914304483681917,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 169493420,
+      "step": 987
+    },
+    {
+      "epoch": 0.2598803182744789,
+      "grad_norm": 5.807379438415879,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 169664140,
+      "step": 988
+    },
+    {
+      "epoch": 0.2598803182744789,
+      "loss": 0.15334829688072205,
+      "loss_ce": 0.001828520093113184,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 169664140,
+      "step": 988
+    },
+    {
+      "epoch": 0.260143355033866,
+      "grad_norm": 6.246751757081568,
+      "learning_rate": 5e-06,
+      "loss": 0.1808,
+      "num_input_tokens_seen": 169836220,
+      "step": 989
+    },
+    {
+      "epoch": 0.260143355033866,
+      "loss": 0.25014275312423706,
+      "loss_ce": 0.003926943056285381,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 169836220,
+      "step": 989
+    },
+    {
+      "epoch": 0.2604063917932531,
+      "grad_norm": 11.305993296500802,
+      "learning_rate": 5e-06,
+      "loss": 0.1627,
+      "num_input_tokens_seen": 170008204,
+      "step": 990
+    },
+    {
+      "epoch": 0.2604063917932531,
+      "loss": 0.19628843665122986,
+      "loss_ce": 0.00878843106329441,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 170008204,
+      "step": 990
+    },
+    {
+      "epoch": 0.26066942855264025,
+      "grad_norm": 6.544136822322436,
+      "learning_rate": 5e-06,
+      "loss": 0.1516,
+      "num_input_tokens_seen": 170178608,
+      "step": 991
+    },
+    {
+      "epoch": 0.26066942855264025,
+      "loss": 0.23060224950313568,
+      "loss_ce": 0.004283890128135681,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 170178608,
+      "step": 991
+    },
+    {
+      "epoch": 0.26093246531202735,
+      "grad_norm": 6.798376445965723,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 170350580,
+      "step": 992
+    },
+    {
+      "epoch": 0.26093246531202735,
+      "loss": 0.15799343585968018,
+      "loss_ce": 0.0012551653198897839,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 170350580,
+      "step": 992
+    },
+    {
+      "epoch": 0.26119550207141445,
+      "grad_norm": 6.259630629604519,
+      "learning_rate": 5e-06,
+      "loss": 0.1793,
+      "num_input_tokens_seen": 170522692,
+      "step": 993
+    },
+    {
+      "epoch": 0.26119550207141445,
+      "loss": 0.2015601247549057,
+      "loss_ce": 0.000510324549395591,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 170522692,
+      "step": 993
+    },
+    {
+      "epoch": 0.2614585388308016,
+      "grad_norm": 6.902940616756998,
+      "learning_rate": 5e-06,
+      "loss": 0.2081,
+      "num_input_tokens_seen": 170695152,
+      "step": 994
+    },
+    {
+      "epoch": 0.2614585388308016,
+      "loss": 0.15382197499275208,
+      "loss_ce": 0.0003795886295847595,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 170695152,
+      "step": 994
+    },
+    {
+      "epoch": 0.2617215755901887,
+      "grad_norm": 4.496994630933735,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 170867532,
+      "step": 995
+    },
+    {
+      "epoch": 0.2617215755901887,
+      "loss": 0.15487955510616302,
+      "loss_ce": 0.005709626711905003,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 170867532,
+      "step": 995
+    },
+    {
+      "epoch": 0.2619846123495759,
+      "grad_norm": 5.819986109817203,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 171039632,
+      "step": 996
+    },
+    {
+      "epoch": 0.2619846123495759,
+      "loss": 0.1341342031955719,
+      "loss_ce": 0.0008944571018218994,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 171039632,
+      "step": 996
+    },
+    {
+      "epoch": 0.262247649108963,
+      "grad_norm": 4.6990222732792075,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 171211920,
+      "step": 997
+    },
+    {
+      "epoch": 0.262247649108963,
+      "loss": 0.2527538239955902,
+      "loss_ce": 0.0029674398247152567,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 171211920,
+      "step": 997
+    },
+    {
+      "epoch": 0.2625106858683501,
+      "grad_norm": 6.547042177560564,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 171384024,
+      "step": 998
+    },
+    {
+      "epoch": 0.2625106858683501,
+      "loss": 0.18961063027381897,
+      "loss_ce": 0.0035754733253270388,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 171384024,
+      "step": 998
+    },
+    {
+      "epoch": 0.26277372262773724,
+      "grad_norm": 17.093722901281232,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 171556336,
+      "step": 999
+    },
+    {
+      "epoch": 0.26277372262773724,
+      "loss": 0.18390598893165588,
+      "loss_ce": 0.0009531003306619823,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 171556336,
+      "step": 999
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "grad_norm": 5.862402162816912,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_websight_new_CIoU": 0.7987666130065918,
+      "eval_websight_new_GIoU": 0.7927780747413635,
+      "eval_websight_new_IoU": 0.8096525371074677,
+      "eval_websight_new_MAE_all": 0.03391252178698778,
+      "eval_websight_new_MAE_h": 0.024881365709006786,
+      "eval_websight_new_MAE_w": 0.04275708086788654,
+      "eval_websight_new_MAE_x": 0.046329958364367485,
+      "eval_websight_new_MAE_y": 0.021681691519916058,
+      "eval_websight_new_NUM_probability": 0.9994822144508362,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.15745767951011658,
+      "eval_websight_new_loss_ce": 9.287914144806564e-05,
+      "eval_websight_new_loss_iou": 0.457275390625,
+      "eval_websight_new_loss_num": 0.027721405029296875,
+      "eval_websight_new_loss_xval": 0.1386260986328125,
+      "eval_websight_new_runtime": 55.0835,
+      "eval_websight_new_samples_per_second": 0.908,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_seeclick_CIoU": 0.5508884787559509,
+      "eval_seeclick_GIoU": 0.5434170663356781,
+      "eval_seeclick_IoU": 0.5745402276515961,
+      "eval_seeclick_MAE_all": 0.057029979303479195,
+      "eval_seeclick_MAE_h": 0.03887217864394188,
+      "eval_seeclick_MAE_w": 0.08262282982468605,
+      "eval_seeclick_MAE_x": 0.07410039007663727,
+      "eval_seeclick_MAE_y": 0.032524523325264454,
+      "eval_seeclick_NUM_probability": 0.9997861981391907,
+      "eval_seeclick_inside_bbox": 0.9076704680919647,
+      "eval_seeclick_loss": 0.24120275676250458,
+      "eval_seeclick_loss_ce": 0.00980278616771102,
+      "eval_seeclick_loss_iou": 0.60888671875,
+      "eval_seeclick_loss_num": 0.0429229736328125,
+      "eval_seeclick_loss_xval": 0.214630126953125,
+      "eval_seeclick_runtime": 69.8263,
+      "eval_seeclick_samples_per_second": 0.616,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_icons_CIoU": 0.8137890696525574,
+      "eval_icons_GIoU": 0.8057061433792114,
+      "eval_icons_IoU": 0.8213592171669006,
+      "eval_icons_MAE_all": 0.024967025965452194,
+      "eval_icons_MAE_h": 0.028143037110567093,
+      "eval_icons_MAE_w": 0.026135658845305443,
+      "eval_icons_MAE_x": 0.022455199621617794,
+      "eval_icons_MAE_y": 0.023134205490350723,
+      "eval_icons_NUM_probability": 0.9995008409023285,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.09549810737371445,
+      "eval_icons_loss_ce": 0.0016497992401127703,
+      "eval_icons_loss_iou": 0.6103515625,
+      "eval_icons_loss_num": 0.01790618896484375,
+      "eval_icons_loss_xval": 0.0895233154296875,
+      "eval_icons_runtime": 88.8842,
+      "eval_icons_samples_per_second": 0.563,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_screenspot_CIoU": 0.5512810548146566,
+      "eval_screenspot_GIoU": 0.5352131823698679,
+      "eval_screenspot_IoU": 0.5900407036145529,
+      "eval_screenspot_MAE_all": 0.08661519487698872,
+      "eval_screenspot_MAE_h": 0.056614277263482414,
+      "eval_screenspot_MAE_w": 0.13663912812868753,
+      "eval_screenspot_MAE_x": 0.10082270950078964,
+      "eval_screenspot_MAE_y": 0.052384667098522186,
+      "eval_screenspot_NUM_probability": 0.9995323220888773,
+      "eval_screenspot_inside_bbox": 0.8454166650772095,
+      "eval_screenspot_loss": 0.7656806111335754,
+      "eval_screenspot_loss_ce": 0.42391865452130634,
+      "eval_screenspot_loss_iou": 0.4834391276041667,
+      "eval_screenspot_loss_num": 0.0673370361328125,
+      "eval_screenspot_loss_xval": 0.3365885416666667,
+      "eval_screenspot_runtime": 148.461,
+      "eval_screenspot_samples_per_second": 0.599,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "loss": 0.7618361711502075,
+      "loss_ce": 0.4089309275150299,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2632997961465115,
+      "grad_norm": 12.369274544442106,
+      "learning_rate": 5e-06,
+      "loss": 0.1395,
+      "num_input_tokens_seen": 171897180,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2632997961465115,
+      "loss": 0.23002395033836365,
+      "loss_ce": 0.0021186815574765205,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 171897180,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2635628329058986,
+      "grad_norm": 6.891119328212443,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 172069384,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2635628329058986,
+      "loss": 0.1037362664937973,
+      "loss_ce": 0.0008004722185432911,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 172069384,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2638258696652857,
+      "grad_norm": 5.298705978749544,
+      "learning_rate": 5e-06,
+      "loss": 0.1493,
+      "num_input_tokens_seen": 172239736,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2638258696652857,
+      "loss": 0.16244152188301086,
+      "loss_ce": 0.0035059780348092318,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 172239736,
+      "step": 1003
+    },
+    {
+      "epoch": 0.26408890642467286,
+      "grad_norm": 28.95390611927557,
+      "learning_rate": 5e-06,
+      "loss": 0.1674,
+      "num_input_tokens_seen": 172409356,
+      "step": 1004
+    },
+    {
+      "epoch": 0.26408890642467286,
+      "loss": 0.16002172231674194,
+      "loss_ce": 0.001452386612072587,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 172409356,
+      "step": 1004
+    },
+    {
+      "epoch": 0.26435194318405997,
+      "grad_norm": 6.613551299237438,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 172579624,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26435194318405997,
+      "loss": 0.11175885051488876,
+      "loss_ce": 0.0038792139384895563,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 172579624,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26461497994344707,
+      "grad_norm": 5.235806011716001,
+      "learning_rate": 5e-06,
+      "loss": 0.181,
+      "num_input_tokens_seen": 172752052,
+      "step": 1006
+    },
+    {
+      "epoch": 0.26461497994344707,
+      "loss": 0.19632884860038757,
+      "loss_ce": 0.001321525895036757,
+      "loss_iou": 0.75390625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 172752052,
+      "step": 1006
+    },
+    {
+      "epoch": 0.26487801670283423,
+      "grad_norm": 11.226831704177643,
+      "learning_rate": 5e-06,
+      "loss": 0.164,
+      "num_input_tokens_seen": 172924420,
+      "step": 1007
+    },
+    {
+      "epoch": 0.26487801670283423,
+      "loss": 0.16282187402248383,
+      "loss_ce": 0.0012007836485281587,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 172924420,
+      "step": 1007
+    },
+    {
+      "epoch": 0.26514105346222133,
+      "grad_norm": 5.82233219860515,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "num_input_tokens_seen": 173096824,
+      "step": 1008
+    },
+    {
+      "epoch": 0.26514105346222133,
+      "loss": 0.17045088112354279,
+      "loss_ce": 0.005320262163877487,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 173096824,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2654040902216085,
+      "grad_norm": 5.52283167756611,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 173268948,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2654040902216085,
+      "loss": 0.17849504947662354,
+      "loss_ce": 0.004483824595808983,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 173268948,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2656671269809956,
+      "grad_norm": 5.624587832123806,
+      "learning_rate": 5e-06,
+      "loss": 0.1733,
+      "num_input_tokens_seen": 173441324,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2656671269809956,
+      "loss": 0.15812638401985168,
+      "loss_ce": 0.0008387943962588906,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 173441324,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2659301637403827,
+      "grad_norm": 8.141965504781345,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 173611452,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2659301637403827,
+      "loss": 0.2730504870414734,
+      "loss_ce": 0.004678931087255478,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 173611452,
+      "step": 1011
+    },
+    {
+      "epoch": 0.26619320049976986,
+      "grad_norm": 9.595372252411392,
+      "learning_rate": 5e-06,
+      "loss": 0.1596,
+      "num_input_tokens_seen": 173783736,
+      "step": 1012
+    },
+    {
+      "epoch": 0.26619320049976986,
+      "loss": 0.23428162932395935,
+      "loss_ce": 0.0011273245327174664,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 173783736,
+      "step": 1012
+    },
+    {
+      "epoch": 0.26645623725915696,
+      "grad_norm": 8.106802909743756,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 173955656,
+      "step": 1013
+    },
+    {
+      "epoch": 0.26645623725915696,
+      "loss": 0.15512457489967346,
+      "loss_ce": 0.003147047944366932,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 173955656,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2667192740185441,
+      "grad_norm": 5.365195100132044,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 174126100,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2667192740185441,
+      "loss": 0.17733250558376312,
+      "loss_ce": 0.004175758454948664,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 174126100,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2669823107779312,
+      "grad_norm": 6.156957339464345,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 174298328,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2669823107779312,
+      "loss": 0.1237088292837143,
+      "loss_ce": 0.001119712833315134,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 174298328,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2672453475373183,
+      "grad_norm": 5.364063512164187,
+      "learning_rate": 5e-06,
+      "loss": 0.1536,
+      "num_input_tokens_seen": 174470324,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2672453475373183,
+      "loss": 0.2085983008146286,
+      "loss_ce": 0.003947417717427015,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 174470324,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2675083842967055,
+      "grad_norm": 26.166059849551687,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 174642656,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2675083842967055,
+      "loss": 0.0957513153553009,
+      "loss_ce": 0.0004143980913795531,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 174642656,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2677714210560926,
+      "grad_norm": 13.570714399931546,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 174813336,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2677714210560926,
+      "loss": 0.08781825006008148,
+      "loss_ce": 0.0015755778877064586,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 174813336,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2680344578154797,
+      "grad_norm": 5.193615864048545,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 174983476,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2680344578154797,
+      "loss": 0.09173595905303955,
+      "loss_ce": 0.0038148202002048492,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 174983476,
+      "step": 1019
+    },
+    {
+      "epoch": 0.26829749457486685,
+      "grad_norm": 6.860708381616737,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 175155688,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26829749457486685,
+      "loss": 0.1804433912038803,
+      "loss_ce": 0.002403826452791691,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 175155688,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26856053133425395,
+      "grad_norm": 4.544920825241194,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 175327636,
+      "step": 1021
+    },
+    {
+      "epoch": 0.26856053133425395,
+      "loss": 0.14717841148376465,
+      "loss_ce": 0.0010602545225992799,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 175327636,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2688235680936411,
+      "grad_norm": 8.925039478850847,
+      "learning_rate": 5e-06,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 175499748,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2688235680936411,
+      "loss": 0.1853310763835907,
+      "loss_ce": 0.0006386763998307288,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 175499748,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2690866048530282,
+      "grad_norm": 5.74787431130144,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 175671808,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2690866048530282,
+      "loss": 0.14083652198314667,
+      "loss_ce": 0.0009744655108079314,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 175671808,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2693496416124153,
+      "grad_norm": 6.6733989699604725,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 175844188,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2693496416124153,
+      "loss": 0.1698358803987503,
+      "loss_ce": 0.0007990067824721336,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 175844188,
+      "step": 1024
+    },
+    {
+      "epoch": 0.26961267837180247,
+      "grad_norm": 8.873866851189364,
+      "learning_rate": 5e-06,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 176016264,
+      "step": 1025
+    },
+    {
+      "epoch": 0.26961267837180247,
+      "loss": 0.28822407126426697,
+      "loss_ce": 0.006241639144718647,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 176016264,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2698757151311896,
+      "grad_norm": 5.745558264753103,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 176188840,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2698757151311896,
+      "loss": 0.20782078802585602,
+      "loss_ce": 0.003353000618517399,
+      "loss_iou": 0.53125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 176188840,
+      "step": 1026
+    },
+    {
+      "epoch": 0.27013875189057673,
+      "grad_norm": 6.250713326486415,
+      "learning_rate": 5e-06,
+      "loss": 0.1462,
+      "num_input_tokens_seen": 176360864,
+      "step": 1027
+    },
+    {
+      "epoch": 0.27013875189057673,
+      "loss": 0.17586824297904968,
+      "loss_ce": 0.0016433752607554197,
+      "loss_iou": 0.59375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 176360864,
+      "step": 1027
+    },
+    {
+      "epoch": 0.27040178864996384,
+      "grad_norm": 9.18813465584604,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 176533304,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27040178864996384,
+      "loss": 0.1184663325548172,
+      "loss_ce": 0.0005769361741840839,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 176533304,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27066482540935094,
+      "grad_norm": 6.1602828275187,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 176705536,
+      "step": 1029
+    },
+    {
+      "epoch": 0.27066482540935094,
+      "loss": 0.20641186833381653,
+      "loss_ce": 0.0027985800988972187,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 176705536,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2709278621687381,
+      "grad_norm": 6.699314102286077,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 176877744,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2709278621687381,
+      "loss": 0.12170865386724472,
+      "loss_ce": 0.003544584382325411,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 176877744,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2711908989281252,
+      "grad_norm": 7.305231502769662,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 177050116,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2711908989281252,
+      "loss": 0.1749892234802246,
+      "loss_ce": 0.001527311746031046,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 177050116,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2714539356875123,
+      "grad_norm": 23.98337624564377,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 177222180,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2714539356875123,
+      "loss": 0.11504107713699341,
+      "loss_ce": 0.0008137800614349544,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 177222180,
+      "step": 1032
+    },
+    {
+      "epoch": 0.27171697244689946,
+      "grad_norm": 8.870505808545992,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 177394132,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27171697244689946,
+      "loss": 0.1380763053894043,
+      "loss_ce": 0.001357543864287436,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 177394132,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27198000920628657,
+      "grad_norm": 6.7102870745927,
+      "learning_rate": 5e-06,
+      "loss": 0.152,
+      "num_input_tokens_seen": 177564532,
+      "step": 1034
+    },
+    {
+      "epoch": 0.27198000920628657,
+      "loss": 0.1144593358039856,
+      "loss_ce": 0.0005372193409129977,
+      "loss_iou": 0.625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 177564532,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2722430459656737,
+      "grad_norm": 7.475407895210686,
+      "learning_rate": 5e-06,
+      "loss": 0.1639,
+      "num_input_tokens_seen": 177735192,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2722430459656737,
+      "loss": 0.1714543104171753,
+      "loss_ce": 0.001501921215094626,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 177735192,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2725060827250608,
+      "grad_norm": 7.019122274754133,
+      "learning_rate": 5e-06,
+      "loss": 0.183,
+      "num_input_tokens_seen": 177905512,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2725060827250608,
+      "loss": 0.23724797368049622,
+      "loss_ce": 0.0011640018783509731,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 177905512,
+      "step": 1036
+    },
+    {
+      "epoch": 0.27276911948444793,
+      "grad_norm": 6.609200156734422,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 178077776,
+      "step": 1037
+    },
+    {
+      "epoch": 0.27276911948444793,
+      "loss": 0.1989010125398636,
+      "loss_ce": 0.002367813140153885,
+      "loss_iou": 0.421875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 178077776,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2730321562438351,
+      "grad_norm": 7.044877026833013,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 178250200,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2730321562438351,
+      "loss": 0.1487899273633957,
+      "loss_ce": 0.0012069200165569782,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 178250200,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2732951930032222,
+      "grad_norm": 7.367246122761307,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 178420744,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2732951930032222,
+      "loss": 0.1740129590034485,
+      "loss_ce": 0.002565213944762945,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 178420744,
+      "step": 1039
+    },
+    {
+      "epoch": 0.27355822976260935,
+      "grad_norm": 6.283509699437948,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 178592888,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27355822976260935,
+      "loss": 0.09046860039234161,
+      "loss_ce": 0.0011436456115916371,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 178592888,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27382126652199645,
+      "grad_norm": 8.68361428045609,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 178763324,
+      "step": 1041
+    },
+    {
+      "epoch": 0.27382126652199645,
+      "loss": 0.10273198038339615,
+      "loss_ce": 0.0014746561646461487,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 178763324,
+      "step": 1041
+    },
+    {
+      "epoch": 0.27408430328138356,
+      "grad_norm": 5.923468447654299,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 178935692,
+      "step": 1042
+    },
+    {
+      "epoch": 0.27408430328138356,
+      "loss": 0.11228330433368683,
+      "loss_ce": 0.0015655276365578175,
+      "loss_iou": 0.7265625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 178935692,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2743473400407707,
+      "grad_norm": 12.335079487643208,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 179105416,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2743473400407707,
+      "loss": 0.13459762930870056,
+      "loss_ce": 0.0016020219773054123,
+      "loss_iou": 0.59375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 179105416,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2746103768001578,
+      "grad_norm": 5.798275831622124,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 179277628,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2746103768001578,
+      "loss": 0.07603298872709274,
+      "loss_ce": 0.0007766383932903409,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 179277628,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2748734135595449,
+      "grad_norm": 23.44861216824249,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 179447304,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2748734135595449,
+      "loss": 0.143830344080925,
+      "loss_ce": 0.0031137943733483553,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 179447304,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2751364503189321,
+      "grad_norm": 5.56459204907325,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 179619344,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2751364503189321,
+      "loss": 0.1308884471654892,
+      "loss_ce": 0.0006089094094932079,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 179619344,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2753994870783192,
+      "grad_norm": 7.96724794117892,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 179791548,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2753994870783192,
+      "loss": 0.1258929818868637,
+      "loss_ce": 0.006264072842895985,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 179791548,
+      "step": 1047
+    },
+    {
+      "epoch": 0.27566252383770634,
+      "grad_norm": 5.070360731345708,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 179963676,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27566252383770634,
+      "loss": 0.11563927680253983,
+      "loss_ce": 0.00031335209496319294,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 179963676,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27592556059709344,
+      "grad_norm": 7.345106018369934,
+      "learning_rate": 5e-06,
+      "loss": 0.1455,
+      "num_input_tokens_seen": 180135984,
+      "step": 1049
+    },
+    {
+      "epoch": 0.27592556059709344,
+      "loss": 0.1866682916879654,
+      "loss_ce": 0.0012129689566791058,
+      "loss_iou": 0.609375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 180135984,
+      "step": 1049
+    },
+    {
+      "epoch": 0.27618859735648055,
+      "grad_norm": 9.442805115405356,
+      "learning_rate": 5e-06,
+      "loss": 0.2447,
+      "num_input_tokens_seen": 180308156,
+      "step": 1050
+    },
+    {
+      "epoch": 0.27618859735648055,
+      "loss": 0.23040437698364258,
+      "loss_ce": 0.001797212054952979,
+      "loss_iou": 0.578125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 180308156,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2764516341158677,
+      "grad_norm": 10.313405249427115,
+      "learning_rate": 5e-06,
+      "loss": 0.1419,
+      "num_input_tokens_seen": 180480268,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2764516341158677,
+      "loss": 0.1318507045507431,
+      "loss_ce": 0.003188594477251172,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 180480268,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2767146708752548,
+      "grad_norm": 12.938419489278349,
+      "learning_rate": 5e-06,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 180652480,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2767146708752548,
+      "loss": 0.1456303596496582,
+      "loss_ce": 0.0045475889928638935,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 180652480,
+      "step": 1052
+    },
+    {
+      "epoch": 0.27697770763464197,
+      "grad_norm": 6.129454843218688,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 180824868,
+      "step": 1053
+    },
+    {
+      "epoch": 0.27697770763464197,
+      "loss": 0.10654839873313904,
+      "loss_ce": 0.0010186205618083477,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 180824868,
+      "step": 1053
+    },
+    {
+      "epoch": 0.27724074439402907,
+      "grad_norm": 6.137928561662,
+      "learning_rate": 5e-06,
+      "loss": 0.1994,
+      "num_input_tokens_seen": 180996888,
+      "step": 1054
+    },
+    {
+      "epoch": 0.27724074439402907,
+      "loss": 0.21584706008434296,
+      "loss_ce": 0.0034447195939719677,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 180996888,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2775037811534162,
+      "grad_norm": 4.330377156785293,
+      "learning_rate": 5e-06,
+      "loss": 0.098,
+      "num_input_tokens_seen": 181168924,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2775037811534162,
+      "loss": 0.08097569644451141,
+      "loss_ce": 0.0013553331373259425,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 181168924,
+      "step": 1055
+    },
+    {
+      "epoch": 0.27776681791280333,
+      "grad_norm": 6.97802036205834,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 181341092,
+      "step": 1056
+    },
+    {
+      "epoch": 0.27776681791280333,
+      "loss": 0.1544983983039856,
+      "loss_ce": 0.0006592837744392455,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 181341092,
+      "step": 1056
+    },
+    {
+      "epoch": 0.27802985467219044,
+      "grad_norm": 5.494026690913253,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 181513136,
+      "step": 1057
+    },
+    {
+      "epoch": 0.27802985467219044,
+      "loss": 0.09461110830307007,
+      "loss_ce": 0.004279076587408781,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 181513136,
+      "step": 1057
+    },
+    {
+      "epoch": 0.27829289143157754,
+      "grad_norm": 8.725452399730845,
+      "learning_rate": 5e-06,
+      "loss": 0.1293,
+      "num_input_tokens_seen": 181684960,
+      "step": 1058
+    },
+    {
+      "epoch": 0.27829289143157754,
+      "loss": 0.14191409945487976,
+      "loss_ce": 0.004157752729952335,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 181684960,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2785559281909647,
+      "grad_norm": 5.556456214125238,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 181857208,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2785559281909647,
+      "loss": 0.14501094818115234,
+      "loss_ce": 0.0029821395874023438,
+      "loss_iou": 0.5,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 181857208,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2788189649503518,
+      "grad_norm": 5.694123648635864,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 182029272,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2788189649503518,
+      "loss": 0.1433970034122467,
+      "loss_ce": 0.0002085179730784148,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 182029272,
+      "step": 1060
+    },
+    {
+      "epoch": 0.27908200170973896,
+      "grad_norm": 11.17547001843261,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 182201700,
+      "step": 1061
+    },
+    {
+      "epoch": 0.27908200170973896,
+      "loss": 0.2361234724521637,
+      "loss_ce": 0.0002225826756330207,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 182201700,
+      "step": 1061
+    },
+    {
+      "epoch": 0.27934503846912606,
+      "grad_norm": 7.952701295024384,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 182373760,
+      "step": 1062
+    },
+    {
+      "epoch": 0.27934503846912606,
+      "loss": 0.12168803811073303,
+      "loss_ce": 0.00483623007312417,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 182373760,
+      "step": 1062
+    },
+    {
+      "epoch": 0.27960807522851316,
+      "grad_norm": 8.381017586610616,
+      "learning_rate": 5e-06,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 182544140,
+      "step": 1063
+    },
+    {
+      "epoch": 0.27960807522851316,
+      "loss": 0.18032154440879822,
+      "loss_ce": 0.003350101877003908,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 182544140,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2798711119879003,
+      "grad_norm": 7.079859253709746,
+      "learning_rate": 5e-06,
+      "loss": 0.1588,
+      "num_input_tokens_seen": 182716400,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2798711119879003,
+      "loss": 0.2143479734659195,
+      "loss_ce": 0.005302563309669495,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 182716400,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2801341487472874,
+      "grad_norm": 5.981555911348935,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 182888924,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2801341487472874,
+      "loss": 0.07972423732280731,
+      "loss_ce": 0.0004090492147952318,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 182888924,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2803971855066746,
+      "grad_norm": 6.957965061807421,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 183061040,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2803971855066746,
+      "loss": 0.18109014630317688,
+      "loss_ce": 0.0029285247437655926,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 183061040,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2806602222660617,
+      "grad_norm": 7.354674178304409,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 183233464,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2806602222660617,
+      "loss": 0.19740189611911774,
+      "loss_ce": 0.0007771397940814495,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 183233464,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2809232590254488,
+      "grad_norm": 6.666818027916791,
+      "learning_rate": 5e-06,
+      "loss": 0.1868,
+      "num_input_tokens_seen": 183405764,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2809232590254488,
+      "loss": 0.15091687440872192,
+      "loss_ce": 0.0006788407335989177,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 183405764,
+      "step": 1068
+    },
+    {
+      "epoch": 0.28118629578483595,
+      "grad_norm": 5.646804925247109,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 183578300,
+      "step": 1069
+    },
+    {
+      "epoch": 0.28118629578483595,
+      "loss": 0.15913698077201843,
+      "loss_ce": 0.0021545523777604103,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 183578300,
+      "step": 1069
+    },
+    {
+      "epoch": 0.28144933254422305,
+      "grad_norm": 20.5753376454494,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 183750176,
+      "step": 1070
+    },
+    {
+      "epoch": 0.28144933254422305,
+      "loss": 0.11271088570356369,
+      "loss_ce": 0.0023593269288539886,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 183750176,
+      "step": 1070
+    },
+    {
+      "epoch": 0.28171236930361016,
+      "grad_norm": 4.644203324007545,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 183922404,
+      "step": 1071
+    },
+    {
+      "epoch": 0.28171236930361016,
+      "loss": 0.12499827146530151,
+      "loss_ce": 0.0020124230068176985,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 183922404,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2819754060629973,
+      "grad_norm": 5.233661334274966,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 184094188,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2819754060629973,
+      "loss": 0.1377188265323639,
+      "loss_ce": 0.0002676558797247708,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 184094188,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2822384428223844,
+      "grad_norm": 7.522178114598793,
+      "learning_rate": 5e-06,
+      "loss": 0.1911,
+      "num_input_tokens_seen": 184262844,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2822384428223844,
+      "loss": 0.2804732322692871,
+      "loss_ce": 0.0030379469972103834,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 184262844,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2825014795817716,
+      "grad_norm": 10.418392232208523,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 184431224,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2825014795817716,
+      "loss": 0.17388112843036652,
+      "loss_ce": 0.002219748916104436,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 184431224,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2827645163411587,
+      "grad_norm": 7.018369739955214,
+      "learning_rate": 5e-06,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 184603440,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2827645163411587,
+      "loss": 0.24287168681621552,
+      "loss_ce": 0.0036749078426510096,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 184603440,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2830275531005458,
+      "grad_norm": 8.441416409211774,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 184775664,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2830275531005458,
+      "loss": 0.253373384475708,
+      "loss_ce": 0.0034344326704740524,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 184775664,
+      "step": 1076
+    },
+    {
+      "epoch": 0.28329058985993294,
+      "grad_norm": 5.273714948615874,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 184947684,
+      "step": 1077
+    },
+    {
+      "epoch": 0.28329058985993294,
+      "loss": 0.12758958339691162,
+      "loss_ce": 0.0030168381053954363,
+      "loss_iou": 0.59375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 184947684,
+      "step": 1077
+    },
+    {
+      "epoch": 0.28355362661932004,
+      "grad_norm": 5.622485240246158,
+      "learning_rate": 5e-06,
+      "loss": 0.1707,
+      "num_input_tokens_seen": 185119764,
+      "step": 1078
+    },
+    {
+      "epoch": 0.28355362661932004,
+      "loss": 0.15722918510437012,
+      "loss_ce": 0.0001552198955323547,
+      "loss_iou": 0.546875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 185119764,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2838166633787072,
+      "grad_norm": 6.064053191448847,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 185291940,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2838166633787072,
+      "loss": 0.12457242608070374,
+      "loss_ce": 0.0010983101092278957,
+      "loss_iou": 0.53125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 185291940,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2840797001380943,
+      "grad_norm": 8.508529448099232,
+      "learning_rate": 5e-06,
+      "loss": 0.1667,
+      "num_input_tokens_seen": 185460372,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2840797001380943,
+      "loss": 0.2045111060142517,
+      "loss_ce": 0.0020574983209371567,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 185460372,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2843427368974814,
+      "grad_norm": 5.269932611835083,
+      "learning_rate": 5e-06,
+      "loss": 0.1797,
+      "num_input_tokens_seen": 185632912,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2843427368974814,
+      "loss": 0.13937309384346008,
+      "loss_ce": 0.009490270167589188,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 185632912,
+      "step": 1081
+    },
+    {
+      "epoch": 0.28460577365686857,
+      "grad_norm": 8.826575998844994,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 185804888,
+      "step": 1082
+    },
+    {
+      "epoch": 0.28460577365686857,
+      "loss": 0.099105603992939,
+      "loss_ce": 0.0018155663274228573,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 185804888,
+      "step": 1082
+    },
+    {
+      "epoch": 0.28486881041625567,
+      "grad_norm": 5.9573043544472535,
+      "learning_rate": 5e-06,
+      "loss": 0.1276,
+      "num_input_tokens_seen": 185977324,
+      "step": 1083
+    },
+    {
+      "epoch": 0.28486881041625567,
+      "loss": 0.17646890878677368,
+      "loss_ce": 0.0035868186969310045,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 185977324,
+      "step": 1083
+    },
+    {
+      "epoch": 0.28513184717564277,
+      "grad_norm": 5.170909045758908,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 186149548,
+      "step": 1084
+    },
+    {
+      "epoch": 0.28513184717564277,
+      "loss": 0.19262221455574036,
+      "loss_ce": 0.000941307342145592,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 186149548,
+      "step": 1084
+    },
+    {
+      "epoch": 0.28539488393502993,
+      "grad_norm": 16.76166177176129,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 186321516,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28539488393502993,
+      "loss": 0.13177794218063354,
+      "loss_ce": 0.0014678854495286942,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 186321516,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28565792069441703,
+      "grad_norm": 13.385145963732795,
+      "learning_rate": 5e-06,
+      "loss": 0.1565,
+      "num_input_tokens_seen": 186493748,
+      "step": 1086
+    },
+    {
+      "epoch": 0.28565792069441703,
+      "loss": 0.24282805621623993,
+      "loss_ce": 0.0057369922287762165,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 186493748,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2859209574538042,
+      "grad_norm": 7.0564105719656025,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 186666020,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2859209574538042,
+      "loss": 0.10800403356552124,
+      "loss_ce": 0.000704226375091821,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 186666020,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2861839942131913,
+      "grad_norm": 5.57534276048312,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 186834708,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2861839942131913,
+      "loss": 0.1970679759979248,
+      "loss_ce": 0.0005652993568219244,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 186834708,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2864470309725784,
+      "grad_norm": 12.382219859522896,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 187006628,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2864470309725784,
+      "loss": 0.13300946354866028,
+      "loss_ce": 0.0005631742533296347,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 187006628,
+      "step": 1089
+    },
+    {
+      "epoch": 0.28671006773196556,
+      "grad_norm": 6.678243304603748,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 187178940,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28671006773196556,
+      "loss": 0.16839508712291718,
+      "loss_ce": 0.004485180135816336,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 187178940,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28697310449135266,
+      "grad_norm": 17.285977120564166,
+      "learning_rate": 5e-06,
+      "loss": 0.1817,
+      "num_input_tokens_seen": 187351348,
+      "step": 1091
+    },
+    {
+      "epoch": 0.28697310449135266,
+      "loss": 0.24864555895328522,
+      "loss_ce": 0.0034063111525028944,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 187351348,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2872361412507398,
+      "grad_norm": 5.362480990027185,
+      "learning_rate": 5e-06,
+      "loss": 0.1333,
+      "num_input_tokens_seen": 187523476,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2872361412507398,
+      "loss": 0.19447633624076843,
+      "loss_ce": 0.005084256641566753,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 187523476,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2874991780101269,
+      "grad_norm": 9.00142705860072,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 187695596,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2874991780101269,
+      "loss": 0.13948455452919006,
+      "loss_ce": 0.0006295705679804087,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 187695596,
+      "step": 1093
+    },
+    {
+      "epoch": 0.287762214769514,
+      "grad_norm": 11.592832787644355,
+      "learning_rate": 5e-06,
+      "loss": 0.1673,
+      "num_input_tokens_seen": 187867844,
+      "step": 1094
+    },
+    {
+      "epoch": 0.287762214769514,
+      "loss": 0.1788289099931717,
+      "loss_ce": 0.0009724590927362442,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 187867844,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2880252515289012,
+      "grad_norm": 5.4984009631677,
+      "learning_rate": 5e-06,
+      "loss": 0.1593,
+      "num_input_tokens_seen": 188039960,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2880252515289012,
+      "loss": 0.2135852873325348,
+      "loss_ce": 0.0043262611143291,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 188039960,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2882882882882883,
+      "grad_norm": 6.039228998020695,
+      "learning_rate": 5e-06,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 188212228,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2882882882882883,
+      "loss": 0.10800454020500183,
+      "loss_ce": 0.0013761227019131184,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 188212228,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2885513250476754,
+      "grad_norm": 6.337937600162702,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 188384548,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2885513250476754,
+      "loss": 0.2018011212348938,
+      "loss_ce": 0.0031927230302244425,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 188384548,
+      "step": 1097
+    },
+    {
+      "epoch": 0.28881436180706255,
+      "grad_norm": 6.282757808284281,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 188556612,
+      "step": 1098
+    },
+    {
+      "epoch": 0.28881436180706255,
+      "loss": 0.07978774607181549,
+      "loss_ce": 0.0007472233846783638,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 188556612,
+      "step": 1098
+    },
+    {
+      "epoch": 0.28907739856644965,
+      "grad_norm": 5.5362247942440135,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 188727124,
+      "step": 1099
+    },
+    {
+      "epoch": 0.28907739856644965,
+      "loss": 0.18471282720565796,
+      "loss_ce": 0.0071005141362547874,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 188727124,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2893404353258368,
+      "grad_norm": 5.455389129921704,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 188899384,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2893404353258368,
+      "loss": 0.08721458911895752,
+      "loss_ce": 0.0006362219573929906,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 188899384,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2896034720852239,
+      "grad_norm": 6.057305949672798,
+      "learning_rate": 5e-06,
+      "loss": 0.1669,
+      "num_input_tokens_seen": 189068024,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2896034720852239,
+      "loss": 0.1826099157333374,
+      "loss_ce": 0.00026738218730315566,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 189068024,
+      "step": 1101
+    },
+    {
+      "epoch": 0.289866508844611,
+      "grad_norm": 10.886874649593773,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 189240248,
+      "step": 1102
+    },
+    {
+      "epoch": 0.289866508844611,
+      "loss": 0.1780916154384613,
+      "loss_ce": 0.0009065555641427636,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 189240248,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2901295456039982,
+      "grad_norm": 8.357621038563343,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 189412208,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2901295456039982,
+      "loss": 0.13078002631664276,
+      "loss_ce": 0.001751709496602416,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 189412208,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2903925823633853,
+      "grad_norm": 13.284219513588615,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 189584480,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2903925823633853,
+      "loss": 0.23028159141540527,
+      "loss_ce": 0.003474962431937456,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 189584480,
+      "step": 1104
+    },
+    {
+      "epoch": 0.29065561912277244,
+      "grad_norm": 6.716057098151779,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 189755004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29065561912277244,
+      "loss": 0.12404203414916992,
+      "loss_ce": 0.002643106272444129,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 189755004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29091865588215954,
+      "grad_norm": 13.264130378049543,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 189927084,
+      "step": 1106
+    },
+    {
+      "epoch": 0.29091865588215954,
+      "loss": 0.10699759423732758,
+      "loss_ce": 0.0014678104780614376,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 189927084,
+      "step": 1106
+    },
+    {
+      "epoch": 0.29118169264154664,
+      "grad_norm": 19.87883744683546,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 190097456,
+      "step": 1107
+    },
+    {
+      "epoch": 0.29118169264154664,
+      "loss": 0.19540926814079285,
+      "loss_ce": 0.0016531546134501696,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 190097456,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2914447294009338,
+      "grad_norm": 7.462386686254148,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 190269508,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2914447294009338,
+      "loss": 0.08214541524648666,
+      "loss_ce": 0.004813872277736664,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 190269508,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2917077661603209,
+      "grad_norm": 10.674953183668224,
+      "learning_rate": 5e-06,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 190441836,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2917077661603209,
+      "loss": 0.1546817272901535,
+      "loss_ce": 0.006824057083576918,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 190441836,
+      "step": 1109
+    },
+    {
+      "epoch": 0.291970802919708,
+      "grad_norm": 6.315839345670245,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 190614052,
+      "step": 1110
+    },
+    {
+      "epoch": 0.291970802919708,
+      "loss": 0.15810704231262207,
+      "loss_ce": 0.0053970692679286,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 190614052,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29223383967909516,
+      "grad_norm": 6.217305097644484,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 190786084,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29223383967909516,
+      "loss": 0.07959192991256714,
+      "loss_ce": 0.0006734670605510473,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 190786084,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29249687643848227,
+      "grad_norm": 5.126517505767914,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 190956588,
+      "step": 1112
+    },
+    {
+      "epoch": 0.29249687643848227,
+      "loss": 0.13071726262569427,
+      "loss_ce": 0.003184308996424079,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 190956588,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2927599131978694,
+      "grad_norm": 8.996821720570772,
+      "learning_rate": 5e-06,
+      "loss": 0.1638,
+      "num_input_tokens_seen": 191126948,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2927599131978694,
+      "loss": 0.07688204199075699,
+      "loss_ce": 0.0015036254189908504,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 191126948,
+      "step": 1113
+    },
+    {
+      "epoch": 0.29302294995725653,
+      "grad_norm": 5.558729986145218,
+      "learning_rate": 5e-06,
+      "loss": 0.1861,
+      "num_input_tokens_seen": 191299288,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29302294995725653,
+      "loss": 0.15988363325595856,
+      "loss_ce": 0.0014058587839826941,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 191299288,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29328598671664363,
+      "grad_norm": 6.8874530652883825,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 191471672,
+      "step": 1115
+    },
+    {
+      "epoch": 0.29328598671664363,
+      "loss": 0.09309347718954086,
+      "loss_ce": 0.0007167698349803686,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 191471672,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2935490234760308,
+      "grad_norm": 13.843828225173697,
+      "learning_rate": 5e-06,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 191643552,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2935490234760308,
+      "loss": 0.08135861903429031,
+      "loss_ce": 0.00021237613691482693,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 191643552,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2938120602354179,
+      "grad_norm": 5.436785087484111,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 191815984,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2938120602354179,
+      "loss": 0.10953962057828903,
+      "loss_ce": 0.0004392758710309863,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 191815984,
+      "step": 1117
+    },
+    {
+      "epoch": 0.294075096994805,
+      "grad_norm": 11.598374668500005,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 191988036,
+      "step": 1118
+    },
+    {
+      "epoch": 0.294075096994805,
+      "loss": 0.0961490124464035,
+      "loss_ce": 0.0009952039690688252,
+      "loss_iou": 0.65625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 191988036,
+      "step": 1118
+    },
+    {
+      "epoch": 0.29433813375419215,
+      "grad_norm": 7.498841309335166,
+      "learning_rate": 5e-06,
+      "loss": 0.1726,
+      "num_input_tokens_seen": 192160116,
+      "step": 1119
+    },
+    {
+      "epoch": 0.29433813375419215,
+      "loss": 0.25876477360725403,
+      "loss_ce": 0.004339736420661211,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 192160116,
+      "step": 1119
+    },
+    {
+      "epoch": 0.29460117051357926,
+      "grad_norm": 5.645236594614784,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 192332060,
+      "step": 1120
+    },
+    {
+      "epoch": 0.29460117051357926,
+      "loss": 0.07815182209014893,
+      "loss_ce": 0.0011254575802013278,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 192332060,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2948642072729664,
+      "grad_norm": 5.473412167736241,
+      "learning_rate": 5e-06,
+      "loss": 0.1499,
+      "num_input_tokens_seen": 192504064,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2948642072729664,
+      "loss": 0.19324743747711182,
+      "loss_ce": 0.0004373906413093209,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 192504064,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2951272440323535,
+      "grad_norm": 15.77937603528804,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 192676312,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2951272440323535,
+      "loss": 0.11738383769989014,
+      "loss_ce": 0.0015696310438215733,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 192676312,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2953902807917406,
+      "grad_norm": 6.595809745274096,
+      "learning_rate": 5e-06,
+      "loss": 0.1605,
+      "num_input_tokens_seen": 192848608,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2953902807917406,
+      "loss": 0.1380428671836853,
+      "loss_ce": 0.0005306481616571546,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 192848608,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2956533175511278,
+      "grad_norm": 5.143986978274753,
+      "learning_rate": 5e-06,
+      "loss": 0.2009,
+      "num_input_tokens_seen": 193021044,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2956533175511278,
+      "loss": 0.2367524653673172,
+      "loss_ce": 0.00335403298959136,
+      "loss_iou": 0.796875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 193021044,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2959163543105149,
+      "grad_norm": 6.676850446443053,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 193193124,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2959163543105149,
+      "loss": 0.1376284509897232,
+      "loss_ce": 0.002099899807944894,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 193193124,
+      "step": 1125
+    },
+    {
+      "epoch": 0.29617939106990204,
+      "grad_norm": 6.051359778802083,
+      "learning_rate": 5e-06,
+      "loss": 0.168,
+      "num_input_tokens_seen": 193365612,
+      "step": 1126
+    },
+    {
+      "epoch": 0.29617939106990204,
+      "loss": 0.11157628893852234,
+      "loss_ce": 0.001163685112260282,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 193365612,
+      "step": 1126
+    },
+    {
+      "epoch": 0.29644242782928915,
+      "grad_norm": 9.91415103149531,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 193537580,
+      "step": 1127
+    },
+    {
+      "epoch": 0.29644242782928915,
+      "loss": 0.06942566484212875,
+      "loss_ce": 0.004575815983116627,
+      "loss_iou": 0.625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 193537580,
+      "step": 1127
+    },
+    {
+      "epoch": 0.29670546458867625,
+      "grad_norm": 8.171761917591171,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 193709920,
+      "step": 1128
+    },
+    {
+      "epoch": 0.29670546458867625,
+      "loss": 0.15288731455802917,
+      "loss_ce": 0.0006046106573194265,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 193709920,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2969685013480634,
+      "grad_norm": 4.591948701112417,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 193878608,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2969685013480634,
+      "loss": 0.15515002608299255,
+      "loss_ce": 0.005400286056101322,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 193878608,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2972315381074505,
+      "grad_norm": 9.7802730005981,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 194051024,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2972315381074505,
+      "loss": 0.13672733306884766,
+      "loss_ce": 0.0005579069838859141,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 194051024,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2974945748668376,
+      "grad_norm": 17.109119761118713,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 194223696,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2974945748668376,
+      "loss": 0.0957454964518547,
+      "loss_ce": 0.0012325569987297058,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 194223696,
+      "step": 1131
+    },
+    {
+      "epoch": 0.29775761162622477,
+      "grad_norm": 4.844704657006506,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 194393332,
+      "step": 1132
+    },
+    {
+      "epoch": 0.29775761162622477,
+      "loss": 0.08335787057876587,
+      "loss_ce": 0.000533167680259794,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 194393332,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2980206483856119,
+      "grad_norm": 12.589562992323502,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 194563660,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2980206483856119,
+      "loss": 0.10209088772535324,
+      "loss_ce": 0.003366521093994379,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 194563660,
+      "step": 1133
+    },
+    {
+      "epoch": 0.29828368514499903,
+      "grad_norm": 5.818392431333322,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 194735892,
+      "step": 1134
+    },
+    {
+      "epoch": 0.29828368514499903,
+      "loss": 0.09164222329854965,
+      "loss_ce": 0.004392467439174652,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 194735892,
+      "step": 1134
+    },
+    {
+      "epoch": 0.29854672190438614,
+      "grad_norm": 6.569465101392607,
+      "learning_rate": 5e-06,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 194907844,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29854672190438614,
+      "loss": 0.12394984811544418,
+      "loss_ce": 0.0028560941573232412,
+      "loss_iou": 0.390625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 194907844,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29880975866377324,
+      "grad_norm": 5.132755422401755,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 195079704,
+      "step": 1136
+    },
+    {
+      "epoch": 0.29880975866377324,
+      "loss": 0.11163240671157837,
+      "loss_ce": 0.001341882161796093,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 195079704,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2990727954231604,
+      "grad_norm": 12.81251876887425,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 195251904,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2990727954231604,
+      "loss": 0.18034929037094116,
+      "loss_ce": 0.0009364524157717824,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 195251904,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2993358321825475,
+      "grad_norm": 6.432718409068677,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 195424180,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2993358321825475,
+      "loss": 0.08383812010288239,
+      "loss_ce": 0.0012575555592775345,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 195424180,
+      "step": 1138
+    },
+    {
+      "epoch": 0.29959886894193466,
+      "grad_norm": 11.352446611671667,
+      "learning_rate": 5e-06,
+      "loss": 0.1314,
+      "num_input_tokens_seen": 195596468,
+      "step": 1139
+    },
+    {
+      "epoch": 0.29959886894193466,
+      "loss": 0.1503646820783615,
+      "loss_ce": 0.0016220146790146828,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 195596468,
+      "step": 1139
+    },
+    {
+      "epoch": 0.29986190570132176,
+      "grad_norm": 7.444433348783749,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 195768360,
+      "step": 1140
+    },
+    {
+      "epoch": 0.29986190570132176,
+      "loss": 0.12922053039073944,
+      "loss_ce": 0.0006804917939007282,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 195768360,
+      "step": 1140
+    },
+    {
+      "epoch": 0.30012494246070887,
+      "grad_norm": 6.813355666039827,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 195940504,
+      "step": 1141
+    },
+    {
+      "epoch": 0.30012494246070887,
+      "loss": 0.14036604762077332,
+      "loss_ce": 0.002609711140394211,
+      "loss_iou": 0.515625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 195940504,
+      "step": 1141
+    },
+    {
+      "epoch": 0.300387979220096,
+      "grad_norm": 9.605056179900458,
+      "learning_rate": 5e-06,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 196112784,
+      "step": 1142
+    },
+    {
+      "epoch": 0.300387979220096,
+      "loss": 0.14964430034160614,
+      "loss_ce": 0.0015119716990739107,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 196112784,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3006510159794831,
+      "grad_norm": 7.338861721143607,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 196285072,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3006510159794831,
+      "loss": 0.13087643682956696,
+      "loss_ce": 0.0021532890386879444,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 196285072,
+      "step": 1143
+    },
+    {
+      "epoch": 0.30091405273887023,
+      "grad_norm": 11.553065995211972,
+      "learning_rate": 5e-06,
+      "loss": 0.1705,
+      "num_input_tokens_seen": 196457252,
+      "step": 1144
+    },
+    {
+      "epoch": 0.30091405273887023,
+      "loss": 0.20345042645931244,
+      "loss_ce": 0.0005085308803245425,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 196457252,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3011770894982574,
+      "grad_norm": 7.552187403058781,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 196629168,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3011770894982574,
+      "loss": 0.17339852452278137,
+      "loss_ce": 0.004331144969910383,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 196629168,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3014401262576445,
+      "grad_norm": 5.762343246857412,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 196801432,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3014401262576445,
+      "loss": 0.07828192412853241,
+      "loss_ce": 0.001377625041641295,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 196801432,
+      "step": 1146
+    },
+    {
+      "epoch": 0.30170316301703165,
+      "grad_norm": 5.174420810424456,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 196973908,
+      "step": 1147
+    },
+    {
+      "epoch": 0.30170316301703165,
+      "loss": 0.22339066863059998,
+      "loss_ce": 0.004701712634414434,
+      "loss_iou": 0.53125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 196973908,
+      "step": 1147
+    },
+    {
+      "epoch": 0.30196619977641875,
+      "grad_norm": 10.875286083758624,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 197146088,
+      "step": 1148
+    },
+    {
+      "epoch": 0.30196619977641875,
+      "loss": 0.1432519108057022,
+      "loss_ce": 0.002657424658536911,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 197146088,
+      "step": 1148
+    },
+    {
+      "epoch": 0.30222923653580586,
+      "grad_norm": 5.705651681197872,
+      "learning_rate": 5e-06,
+      "loss": 0.1464,
+      "num_input_tokens_seen": 197318460,
+      "step": 1149
+    },
+    {
+      "epoch": 0.30222923653580586,
+      "loss": 0.12248589098453522,
+      "loss_ce": 0.002002496039494872,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 197318460,
+      "step": 1149
+    },
+    {
+      "epoch": 0.302492273295193,
+      "grad_norm": 6.2074966030457706,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 197490428,
+      "step": 1150
+    },
+    {
+      "epoch": 0.302492273295193,
+      "loss": 0.1659487783908844,
+      "loss_ce": 0.0009402353898622096,
+      "loss_iou": 0.5,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 197490428,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3027553100545801,
+      "grad_norm": 11.278645202454504,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 197662972,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3027553100545801,
+      "loss": 0.12700702250003815,
+      "loss_ce": 0.0012135641882196069,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 197662972,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3030183468139673,
+      "grad_norm": 5.7121477930459665,
+      "learning_rate": 5e-06,
+      "loss": 0.1739,
+      "num_input_tokens_seen": 197835168,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3030183468139673,
+      "loss": 0.10988777130842209,
+      "loss_ce": 0.004174881149083376,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 197835168,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3032813835733544,
+      "grad_norm": 12.425343043979453,
+      "learning_rate": 5e-06,
+      "loss": 0.1541,
+      "num_input_tokens_seen": 198007440,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3032813835733544,
+      "loss": 0.18807393312454224,
+      "loss_ce": 0.0031679358799010515,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 198007440,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3035444203327415,
+      "grad_norm": 5.925789792578691,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 198179896,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3035444203327415,
+      "loss": 0.07772859930992126,
+      "loss_ce": 0.0008548187324777246,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 198179896,
+      "step": 1154
+    },
+    {
+      "epoch": 0.30380745709212864,
+      "grad_norm": 5.901916078113717,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 198352320,
+      "step": 1155
+    },
+    {
+      "epoch": 0.30380745709212864,
+      "loss": 0.17940585315227509,
+      "loss_ce": 0.0011221629101783037,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 198352320,
+      "step": 1155
+    },
+    {
+      "epoch": 0.30407049385151574,
+      "grad_norm": 10.483911413706299,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 198524512,
+      "step": 1156
+    },
+    {
+      "epoch": 0.30407049385151574,
+      "loss": 0.12064538896083832,
+      "loss_ce": 0.0006197594339028001,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 198524512,
+      "step": 1156
+    },
+    {
+      "epoch": 0.30433353061090285,
+      "grad_norm": 4.522315194840346,
+      "learning_rate": 5e-06,
+      "loss": 0.1278,
+      "num_input_tokens_seen": 198696612,
+      "step": 1157
+    },
+    {
+      "epoch": 0.30433353061090285,
+      "loss": 0.09706555306911469,
+      "loss_ce": 0.0013929473934695125,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 198696612,
+      "step": 1157
+    },
+    {
+      "epoch": 0.30459656737029,
+      "grad_norm": 28.08652601411638,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "num_input_tokens_seen": 198868856,
+      "step": 1158
+    },
+    {
+      "epoch": 0.30459656737029,
+      "loss": 0.052179381251335144,
+      "loss_ce": 0.00020794683950953186,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 198868856,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3048596041296771,
+      "grad_norm": 7.240097871891992,
+      "learning_rate": 5e-06,
+      "loss": 0.1803,
+      "num_input_tokens_seen": 199041064,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3048596041296771,
+      "loss": 0.18336477875709534,
+      "loss_ce": 0.0020293283741921186,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 199041064,
+      "step": 1159
+    },
+    {
+      "epoch": 0.30512264088906427,
+      "grad_norm": 5.628334621930633,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 199213476,
+      "step": 1160
+    },
+    {
+      "epoch": 0.30512264088906427,
+      "loss": 0.12429136037826538,
+      "loss_ce": 0.0002068809699267149,
+      "loss_iou": 0.79296875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 199213476,
+      "step": 1160
+    },
+    {
+      "epoch": 0.30538567764845137,
+      "grad_norm": 5.043014865834959,
+      "learning_rate": 5e-06,
+      "loss": 0.1778,
+      "num_input_tokens_seen": 199385580,
+      "step": 1161
+    },
+    {
+      "epoch": 0.30538567764845137,
+      "loss": 0.09477389603853226,
+      "loss_ce": 0.001115447492338717,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 199385580,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3056487144078385,
+      "grad_norm": 5.156573422162299,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 199556156,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3056487144078385,
+      "loss": 0.14625222980976105,
+      "loss_ce": 0.00046976495650596917,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 199556156,
+      "step": 1162
+    },
+    {
+      "epoch": 0.30591175116722563,
+      "grad_norm": 4.702903740874693,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 199728484,
+      "step": 1163
+    },
+    {
+      "epoch": 0.30591175116722563,
+      "loss": 0.156254380941391,
+      "loss_ce": 0.001774407341144979,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 199728484,
+      "step": 1163
+    },
+    {
+      "epoch": 0.30617478792661273,
+      "grad_norm": 5.380502082020364,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 199900748,
+      "step": 1164
+    },
+    {
+      "epoch": 0.30617478792661273,
+      "loss": 0.16096284985542297,
+      "loss_ce": 0.0019052416319027543,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 199900748,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3064378246859999,
+      "grad_norm": 5.7025282283031595,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 200072488,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3064378246859999,
+      "loss": 0.17774316668510437,
+      "loss_ce": 0.00037499924656003714,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 200072488,
+      "step": 1165
+    },
+    {
+      "epoch": 0.306700861445387,
+      "grad_norm": 5.532941140209234,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 200244696,
+      "step": 1166
+    },
+    {
+      "epoch": 0.306700861445387,
+      "loss": 0.12291580438613892,
+      "loss_ce": 0.0016694690566509962,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 200244696,
+      "step": 1166
+    },
+    {
+      "epoch": 0.3069638982047741,
+      "grad_norm": 5.5856010120270705,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 200417156,
+      "step": 1167
+    },
+    {
+      "epoch": 0.3069638982047741,
+      "loss": 0.12629887461662292,
+      "loss_ce": 0.0023059630766510963,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 200417156,
+      "step": 1167
+    },
+    {
+      "epoch": 0.30722693496416126,
+      "grad_norm": 4.725090769923517,
+      "learning_rate": 5e-06,
+      "loss": 0.1267,
+      "num_input_tokens_seen": 200587780,
+      "step": 1168
+    },
+    {
+      "epoch": 0.30722693496416126,
+      "loss": 0.1289183497428894,
+      "loss_ce": 0.0026366179808974266,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 200587780,
+      "step": 1168
+    },
+    {
+      "epoch": 0.30748997172354836,
+      "grad_norm": 4.307353086128186,
+      "learning_rate": 5e-06,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 200759832,
+      "step": 1169
+    },
+    {
+      "epoch": 0.30748997172354836,
+      "loss": 0.16026800870895386,
+      "loss_ce": 0.0027362804394215345,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 200759832,
+      "step": 1169
+    },
+    {
+      "epoch": 0.30775300848293546,
+      "grad_norm": 11.530918199945434,
+      "learning_rate": 5e-06,
+      "loss": 0.1914,
+      "num_input_tokens_seen": 200929932,
+      "step": 1170
+    },
+    {
+      "epoch": 0.30775300848293546,
+      "loss": 0.19007167220115662,
+      "loss_ce": 0.0034261636901646852,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 200929932,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3080160452423226,
+      "grad_norm": 7.858471531155482,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 201100520,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3080160452423226,
+      "loss": 0.204483300447464,
+      "loss_ce": 0.0005648470250889659,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 201100520,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3082790820017097,
+      "grad_norm": 5.1657164225988,
+      "learning_rate": 5e-06,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 201272808,
+      "step": 1172
+    },
+    {
+      "epoch": 0.3082790820017097,
+      "loss": 0.12782040238380432,
+      "loss_ce": 0.006909756921231747,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 201272808,
+      "step": 1172
+    },
+    {
+      "epoch": 0.3085421187610969,
+      "grad_norm": 5.7855382812859295,
+      "learning_rate": 5e-06,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 201445044,
+      "step": 1173
+    },
+    {
+      "epoch": 0.3085421187610969,
+      "loss": 0.22742098569869995,
+      "loss_ce": 0.0040017981082201,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 201445044,
+      "step": 1173
+    },
+    {
+      "epoch": 0.308805155520484,
+      "grad_norm": 19.043518960426944,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 201617324,
+      "step": 1174
+    },
+    {
+      "epoch": 0.308805155520484,
+      "loss": 0.09914430975914001,
+      "loss_ce": 0.001243914244696498,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 201617324,
+      "step": 1174
+    },
+    {
+      "epoch": 0.3090681922798711,
+      "grad_norm": 7.337870221119852,
+      "learning_rate": 5e-06,
+      "loss": 0.1898,
+      "num_input_tokens_seen": 201789648,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3090681922798711,
+      "loss": 0.2523178160190582,
+      "loss_ce": 0.0009750226745381951,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 201789648,
+      "step": 1175
+    },
+    {
+      "epoch": 0.30933122903925825,
+      "grad_norm": 18.238078934317997,
+      "learning_rate": 5e-06,
+      "loss": 0.1987,
+      "num_input_tokens_seen": 201961724,
+      "step": 1176
+    },
+    {
+      "epoch": 0.30933122903925825,
+      "loss": 0.17623552680015564,
+      "loss_ce": 0.003994307480752468,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 201961724,
+      "step": 1176
+    },
+    {
+      "epoch": 0.30959426579864535,
+      "grad_norm": 4.51493162048169,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 202133700,
+      "step": 1177
+    },
+    {
+      "epoch": 0.30959426579864535,
+      "loss": 0.1072007492184639,
+      "loss_ce": 0.005058412905782461,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 202133700,
+      "step": 1177
+    },
+    {
+      "epoch": 0.3098573025580325,
+      "grad_norm": 5.072583382632679,
+      "learning_rate": 5e-06,
+      "loss": 0.1459,
+      "num_input_tokens_seen": 202304256,
+      "step": 1178
+    },
+    {
+      "epoch": 0.3098573025580325,
+      "loss": 0.14473965764045715,
+      "loss_ce": 0.0024361968971788883,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 202304256,
+      "step": 1178
+    },
+    {
+      "epoch": 0.3101203393174196,
+      "grad_norm": 7.387732741099245,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 202476488,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3101203393174196,
+      "loss": 0.25686854124069214,
+      "loss_ce": 0.005464731715619564,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 202476488,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3103833760768067,
+      "grad_norm": 7.564357354372493,
+      "learning_rate": 5e-06,
+      "loss": 0.1721,
+      "num_input_tokens_seen": 202648844,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3103833760768067,
+      "loss": 0.2612742781639099,
+      "loss_ce": 0.00047106179408729076,
+      "loss_iou": 0.546875,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 202648844,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3106464128361939,
+      "grad_norm": 12.304004570686834,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 202821176,
+      "step": 1181
+    },
+    {
+      "epoch": 0.3106464128361939,
+      "loss": 0.2418098747730255,
+      "loss_ce": 0.0016365369083359838,
+      "loss_iou": 0.453125,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 202821176,
+      "step": 1181
+    },
+    {
+      "epoch": 0.310909449595581,
+      "grad_norm": 4.510810550512272,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 202993428,
+      "step": 1182
+    },
+    {
+      "epoch": 0.310909449595581,
+      "loss": 0.058775611221790314,
+      "loss_ce": 0.0007006583036854863,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 202993428,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3111724863549681,
+      "grad_norm": 44.4539532395822,
+      "learning_rate": 5e-06,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 203164156,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3111724863549681,
+      "loss": 0.16502994298934937,
+      "loss_ce": 0.0006927890353836119,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 203164156,
+      "step": 1183
+    },
+    {
+      "epoch": 0.31143552311435524,
+      "grad_norm": 7.885064131627441,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 203336328,
+      "step": 1184
+    },
+    {
+      "epoch": 0.31143552311435524,
+      "loss": 0.07413887977600098,
+      "loss_ce": 0.00013374855916481465,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 203336328,
+      "step": 1184
+    },
+    {
+      "epoch": 0.31169855987374234,
+      "grad_norm": 5.056026310722222,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 203508700,
+      "step": 1185
+    },
+    {
+      "epoch": 0.31169855987374234,
+      "loss": 0.07386209070682526,
+      "loss_ce": 0.0002842045505531132,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 203508700,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3119615966331295,
+      "grad_norm": 9.611820717208333,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 203680940,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3119615966331295,
+      "loss": 0.13154730200767517,
+      "loss_ce": 0.0037701984401792288,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 203680940,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3122246333925166,
+      "grad_norm": 10.99160414493074,
+      "learning_rate": 5e-06,
+      "loss": 0.1515,
+      "num_input_tokens_seen": 203852944,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3122246333925166,
+      "loss": 0.17437157034873962,
+      "loss_ce": 0.0008486199658364058,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 203852944,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3124876701519037,
+      "grad_norm": 7.566892022382808,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 204025224,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3124876701519037,
+      "loss": 0.16989970207214355,
+      "loss_ce": 0.004982716403901577,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 204025224,
+      "step": 1188
+    },
+    {
+      "epoch": 0.31275070691129087,
+      "grad_norm": 4.336576845925468,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 204197436,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31275070691129087,
+      "loss": 0.30051514506340027,
+      "loss_ce": 0.0009546162909828126,
+      "loss_iou": 0.390625,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 204197436,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31301374367067797,
+      "grad_norm": 8.328730704134305,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 204369676,
+      "step": 1190
+    },
+    {
+      "epoch": 0.31301374367067797,
+      "loss": 0.18648235499858856,
+      "loss_ce": 0.0020035963971167803,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 204369676,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3132767804300651,
+      "grad_norm": 6.248602390066733,
+      "learning_rate": 5e-06,
+      "loss": 0.2073,
+      "num_input_tokens_seen": 204541920,
+      "step": 1191
+    },
+    {
+      "epoch": 0.3132767804300651,
+      "loss": 0.12927605211734772,
+      "loss_ce": 0.0012548138620331883,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 204541920,
+      "step": 1191
+    },
+    {
+      "epoch": 0.31353981718945223,
+      "grad_norm": 13.468629090291936,
+      "learning_rate": 5e-06,
+      "loss": 0.1227,
+      "num_input_tokens_seen": 204714352,
+      "step": 1192
+    },
+    {
+      "epoch": 0.31353981718945223,
+      "loss": 0.13916926085948944,
+      "loss_ce": 0.002206363482400775,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 204714352,
+      "step": 1192
+    },
+    {
+      "epoch": 0.31380285394883933,
+      "grad_norm": 14.97613941577105,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 204886516,
+      "step": 1193
+    },
+    {
+      "epoch": 0.31380285394883933,
+      "loss": 0.07034310698509216,
+      "loss_ce": 0.002716155955567956,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 204886516,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3140658907082265,
+      "grad_norm": 7.0534203975755085,
+      "learning_rate": 5e-06,
+      "loss": 0.1572,
+      "num_input_tokens_seen": 205058780,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3140658907082265,
+      "loss": 0.1390216201543808,
+      "loss_ce": 0.00144837680272758,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 205058780,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3143289274676136,
+      "grad_norm": 8.447198281833328,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 205231072,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3143289274676136,
+      "loss": 0.13909873366355896,
+      "loss_ce": 0.00030477988184429705,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 205231072,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3145919642270007,
+      "grad_norm": 11.546456728325778,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 205403380,
+      "step": 1196
+    },
+    {
+      "epoch": 0.3145919642270007,
+      "loss": 0.21954891085624695,
+      "loss_ce": 0.0075127785094082355,
+      "loss_iou": NaN,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 205403380,
+      "step": 1196
+    },
+    {
+      "epoch": 0.31485500098638786,
+      "grad_norm": 22.283163028622614,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 205575536,
+      "step": 1197
+    },
+    {
+      "epoch": 0.31485500098638786,
+      "loss": 0.10035932064056396,
+      "loss_ce": 0.0016349535435438156,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 205575536,
+      "step": 1197
+    },
+    {
+      "epoch": 0.31511803774577496,
+      "grad_norm": 13.092900629924303,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 205747664,
+      "step": 1198
+    },
+    {
+      "epoch": 0.31511803774577496,
+      "loss": 0.1355750560760498,
+      "loss_ce": 0.0009009751374833286,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 205747664,
+      "step": 1198
+    },
+    {
+      "epoch": 0.3153810745051621,
+      "grad_norm": 4.560396330595116,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 205919612,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3153810745051621,
+      "loss": 0.085871621966362,
+      "loss_ce": 0.0012463756138458848,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 205919612,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3156441112645492,
+      "grad_norm": 7.441372297845623,
+      "learning_rate": 5e-06,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 206092012,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3156441112645492,
+      "loss": 0.13713139295578003,
+      "loss_ce": 0.004868209362030029,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 206092012,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3159071480239363,
+      "grad_norm": 5.3666102379931795,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 206262016,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3159071480239363,
+      "loss": 0.14059683680534363,
+      "loss_ce": 0.001070460770279169,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 206262016,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3161701847833235,
+      "grad_norm": 8.935322327098595,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 206434192,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3161701847833235,
+      "loss": 0.13417284190654755,
+      "loss_ce": 0.0015129297971725464,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 206434192,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3164332215427106,
+      "grad_norm": 10.825802927092472,
+      "learning_rate": 5e-06,
+      "loss": 0.1619,
+      "num_input_tokens_seen": 206604572,
+      "step": 1203
+    },
+    {
+      "epoch": 0.3164332215427106,
+      "loss": 0.1651817262172699,
+      "loss_ce": 0.002462008036673069,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 206604572,
+      "step": 1203
+    },
+    {
+      "epoch": 0.31669625830209774,
+      "grad_norm": 14.238886956988445,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 206776604,
+      "step": 1204
+    },
+    {
+      "epoch": 0.31669625830209774,
+      "loss": 0.1644083857536316,
+      "loss_ce": 0.0012003772426396608,
+      "loss_iou": 0.75390625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 206776604,
+      "step": 1204
+    },
+    {
+      "epoch": 0.31695929506148485,
+      "grad_norm": 7.214768469241478,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 206948516,
+      "step": 1205
+    },
+    {
+      "epoch": 0.31695929506148485,
+      "loss": 0.15767714381217957,
+      "loss_ce": 0.0004200635012239218,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 206948516,
+      "step": 1205
+    },
+    {
+      "epoch": 0.31722233182087195,
+      "grad_norm": 8.494853848101869,
+      "learning_rate": 5e-06,
+      "loss": 0.1633,
+      "num_input_tokens_seen": 207120552,
+      "step": 1206
+    },
+    {
+      "epoch": 0.31722233182087195,
+      "loss": 0.20658773183822632,
+      "loss_ce": 0.007216397672891617,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 207120552,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3174853685802591,
+      "grad_norm": 8.328419666401985,
+      "learning_rate": 5e-06,
+      "loss": 0.1285,
+      "num_input_tokens_seen": 207292524,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3174853685802591,
+      "loss": 0.1329333782196045,
+      "loss_ce": 0.0009753549238666892,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 207292524,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3177484053396462,
+      "grad_norm": 10.047313768362223,
+      "learning_rate": 5e-06,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 207464680,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3177484053396462,
+      "loss": 0.12349092215299606,
+      "loss_ce": 0.0011764641385525465,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 207464680,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3180114420990333,
+      "grad_norm": 12.242484724411801,
+      "learning_rate": 5e-06,
+      "loss": 0.1693,
+      "num_input_tokens_seen": 207634788,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3180114420990333,
+      "loss": 0.21163830161094666,
+      "loss_ce": 0.0032947922591120005,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 207634788,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3182744788584205,
+      "grad_norm": 5.397067809441159,
+      "learning_rate": 5e-06,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 207806604,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3182744788584205,
+      "loss": 0.14931055903434753,
+      "loss_ce": 0.00026271765818819404,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 207806604,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3185375156178076,
+      "grad_norm": 5.849344287815652,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 207979236,
+      "step": 1211
+    },
+    {
+      "epoch": 0.3185375156178076,
+      "loss": 0.16523879766464233,
+      "loss_ce": 0.000810077937785536,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 207979236,
+      "step": 1211
+    },
+    {
+      "epoch": 0.31880055237719473,
+      "grad_norm": 6.099253346488696,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 208151368,
+      "step": 1212
+    },
+    {
+      "epoch": 0.31880055237719473,
+      "loss": 0.12017640471458435,
+      "loss_ce": 0.00015077157877385616,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 208151368,
+      "step": 1212
+    },
+    {
+      "epoch": 0.31906358913658184,
+      "grad_norm": 6.749805208741857,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 208323600,
+      "step": 1213
+    },
+    {
+      "epoch": 0.31906358913658184,
+      "loss": 0.1262407749891281,
+      "loss_ce": 0.0037432105746120214,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 208323600,
+      "step": 1213
+    },
+    {
+      "epoch": 0.31932662589596894,
+      "grad_norm": 7.672226851345336,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 208495836,
+      "step": 1214
+    },
+    {
+      "epoch": 0.31932662589596894,
+      "loss": 0.09719178080558777,
+      "loss_ce": 0.0004205434233881533,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 208495836,
+      "step": 1214
+    },
+    {
+      "epoch": 0.3195896626553561,
+      "grad_norm": 7.992027772652848,
+      "learning_rate": 5e-06,
+      "loss": 0.1721,
+      "num_input_tokens_seen": 208667840,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3195896626553561,
+      "loss": 0.1424046754837036,
+      "loss_ce": 0.00037586723919957876,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 208667840,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3198526994147432,
+      "grad_norm": 6.160546703869267,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 208840040,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3198526994147432,
+      "loss": 0.11277418583631516,
+      "loss_ce": 0.0004084610554855317,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 208840040,
+      "step": 1216
+    },
+    {
+      "epoch": 0.32011573617413036,
+      "grad_norm": 14.924491605710916,
+      "learning_rate": 5e-06,
+      "loss": 0.1657,
+      "num_input_tokens_seen": 209010176,
+      "step": 1217
+    },
+    {
+      "epoch": 0.32011573617413036,
+      "loss": 0.1102285385131836,
+      "loss_ce": 0.0007619824027642608,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 209010176,
+      "step": 1217
+    },
+    {
+      "epoch": 0.32037877293351746,
+      "grad_norm": 13.936435144745488,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 209182304,
+      "step": 1218
+    },
+    {
+      "epoch": 0.32037877293351746,
+      "loss": 0.12400620430707932,
+      "loss_ce": 0.0005168293137103319,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 209182304,
+      "step": 1218
+    },
+    {
+      "epoch": 0.32064180969290457,
+      "grad_norm": 5.31537857924368,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 209354400,
+      "step": 1219
+    },
+    {
+      "epoch": 0.32064180969290457,
+      "loss": 0.15276062488555908,
+      "loss_ce": 0.0029498354997485876,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 209354400,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3209048464522917,
+      "grad_norm": 8.044699509860337,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 209526592,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3209048464522917,
+      "loss": 0.13095101714134216,
+      "loss_ce": 0.00137336365878582,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 209526592,
+      "step": 1220
+    },
+    {
+      "epoch": 0.32116788321167883,
+      "grad_norm": 6.178496640552883,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 209699004,
+      "step": 1221
+    },
+    {
+      "epoch": 0.32116788321167883,
+      "loss": 0.11922457814216614,
+      "loss_ce": 0.00017550383927300572,
+      "loss_iou": 0.625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 209699004,
+      "step": 1221
+    },
+    {
+      "epoch": 0.32143091997106593,
+      "grad_norm": 5.682094913930693,
+      "learning_rate": 5e-06,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 209871116,
+      "step": 1222
+    },
+    {
+      "epoch": 0.32143091997106593,
+      "loss": 0.14516758918762207,
+      "loss_ce": 0.004481561481952667,
+      "loss_iou": 0.328125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 209871116,
+      "step": 1222
+    },
+    {
+      "epoch": 0.3216939567304531,
+      "grad_norm": 7.037588469004313,
+      "learning_rate": 5e-06,
+      "loss": 0.1381,
+      "num_input_tokens_seen": 210041224,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3216939567304531,
+      "loss": 0.1431758999824524,
+      "loss_ce": 0.002764530945569277,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 210041224,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3219569934898402,
+      "grad_norm": 23.622534135841367,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 210211380,
+      "step": 1224
+    },
+    {
+      "epoch": 0.3219569934898402,
+      "loss": 0.1730581820011139,
+      "loss_ce": 0.0024649298284202814,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 210211380,
+      "step": 1224
+    },
+    {
+      "epoch": 0.32222003024922735,
+      "grad_norm": 12.04380926493331,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 210383088,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32222003024922735,
+      "loss": 0.23281848430633545,
+      "loss_ce": 0.00115956028457731,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 210383088,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32248306700861445,
+      "grad_norm": 4.756101930631895,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 210555072,
+      "step": 1226
+    },
+    {
+      "epoch": 0.32248306700861445,
+      "loss": 0.06910556554794312,
+      "loss_ce": 0.003004492959007621,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 210555072,
+      "step": 1226
+    },
+    {
+      "epoch": 0.32274610376800156,
+      "grad_norm": 3.8368982745320284,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 210727236,
+      "step": 1227
+    },
+    {
+      "epoch": 0.32274610376800156,
+      "loss": 0.10896629840135574,
+      "loss_ce": 0.0005983852897770703,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 210727236,
+      "step": 1227
+    },
+    {
+      "epoch": 0.3230091405273887,
+      "grad_norm": 5.290425948494437,
+      "learning_rate": 5e-06,
+      "loss": 0.1453,
+      "num_input_tokens_seen": 210899444,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3230091405273887,
+      "loss": 0.12627889215946198,
+      "loss_ce": 8.869695011526346e-05,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 210899444,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3232721772867758,
+      "grad_norm": 5.380666634187743,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 211071748,
+      "step": 1229
+    },
+    {
+      "epoch": 0.3232721772867758,
+      "loss": 0.16187983751296997,
+      "loss_ce": 0.003463074564933777,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 211071748,
+      "step": 1229
+    },
+    {
+      "epoch": 0.323535214046163,
+      "grad_norm": 6.929606907390503,
+      "learning_rate": 5e-06,
+      "loss": 0.1638,
+      "num_input_tokens_seen": 211243884,
+      "step": 1230
+    },
+    {
+      "epoch": 0.323535214046163,
+      "loss": 0.16732358932495117,
+      "loss_ce": 0.003932466730475426,
+      "loss_iou": 0.76171875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 211243884,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3237982508055501,
+      "grad_norm": 4.103469407229988,
+      "learning_rate": 5e-06,
+      "loss": 0.1433,
+      "num_input_tokens_seen": 211416144,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3237982508055501,
+      "loss": 0.16871199011802673,
+      "loss_ce": 0.0031236184295266867,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 211416144,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3240612875649372,
+      "grad_norm": 4.57933905462206,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 211588296,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3240612875649372,
+      "loss": 0.16614994406700134,
+      "loss_ce": 0.0009277852368541062,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 211588296,
+      "step": 1232
+    },
+    {
+      "epoch": 0.32432432432432434,
+      "grad_norm": 4.940862588040882,
+      "learning_rate": 5e-06,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 211758972,
+      "step": 1233
+    },
+    {
+      "epoch": 0.32432432432432434,
+      "loss": 0.14131012558937073,
+      "loss_ce": 0.0015701348893344402,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 211758972,
+      "step": 1233
+    },
+    {
+      "epoch": 0.32458736108371145,
+      "grad_norm": 11.781672019450719,
+      "learning_rate": 5e-06,
+      "loss": 0.1631,
+      "num_input_tokens_seen": 211928944,
+      "step": 1234
+    },
+    {
+      "epoch": 0.32458736108371145,
+      "loss": 0.14883792400360107,
+      "loss_ce": 0.0010107720736414194,
+      "loss_iou": 0.6875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 211928944,
+      "step": 1234
+    },
+    {
+      "epoch": 0.32485039784309855,
+      "grad_norm": 19.10070261855633,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 212100832,
+      "step": 1235
+    },
+    {
+      "epoch": 0.32485039784309855,
+      "loss": 0.07846543192863464,
+      "loss_ce": 0.0002793997118715197,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 212100832,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3251134346024857,
+      "grad_norm": 15.821804140128867,
+      "learning_rate": 5e-06,
+      "loss": 0.1606,
+      "num_input_tokens_seen": 212271444,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3251134346024857,
+      "loss": 0.1032799631357193,
+      "loss_ce": 0.004799742251634598,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 212271444,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3253764713618728,
+      "grad_norm": 5.984115475857608,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 212443976,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3253764713618728,
+      "loss": 0.06768647581338882,
+      "loss_ce": 0.0013717777328565717,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 212443976,
+      "step": 1237
+    },
+    {
+      "epoch": 0.32563950812125997,
+      "grad_norm": 7.104938203697124,
+      "learning_rate": 5e-06,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 212612720,
+      "step": 1238
+    },
+    {
+      "epoch": 0.32563950812125997,
+      "loss": 0.10599420964717865,
+      "loss_ce": 0.002387029118835926,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 212612720,
+      "step": 1238
+    },
+    {
+      "epoch": 0.32590254488064707,
+      "grad_norm": 5.6715016587403655,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 212784924,
+      "step": 1239
+    },
+    {
+      "epoch": 0.32590254488064707,
+      "loss": 0.07988797873258591,
+      "loss_ce": 0.00020658349967561662,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 212784924,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3261655816400342,
+      "grad_norm": 10.066783726153659,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 212957204,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3261655816400342,
+      "loss": 0.24682722985744476,
+      "loss_ce": 0.0004893409786745906,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 212957204,
+      "step": 1240
+    },
+    {
+      "epoch": 0.32642861839942133,
+      "grad_norm": 6.605871983356305,
+      "learning_rate": 5e-06,
+      "loss": 0.1476,
+      "num_input_tokens_seen": 213129636,
+      "step": 1241
+    },
+    {
+      "epoch": 0.32642861839942133,
+      "loss": 0.14497268199920654,
+      "loss_ce": 0.008375998586416245,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 213129636,
+      "step": 1241
+    },
+    {
+      "epoch": 0.32669165515880844,
+      "grad_norm": 6.65301903793999,
+      "learning_rate": 5e-06,
+      "loss": 0.1883,
+      "num_input_tokens_seen": 213302188,
+      "step": 1242
+    },
+    {
+      "epoch": 0.32669165515880844,
+      "loss": 0.25356027483940125,
+      "loss_ce": 0.002705805469304323,
+      "loss_iou": 0.71875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 213302188,
+      "step": 1242
+    },
+    {
+      "epoch": 0.3269546919181956,
+      "grad_norm": 6.148620904940904,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 213474112,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3269546919181956,
+      "loss": 0.10272787511348724,
+      "loss_ce": 0.00427817041054368,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 213474112,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3272177286775827,
+      "grad_norm": 8.559691577151979,
+      "learning_rate": 5e-06,
+      "loss": 0.214,
+      "num_input_tokens_seen": 213646008,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3272177286775827,
+      "loss": 0.2430175393819809,
+      "loss_ce": 0.0016845206264406443,
+      "loss_iou": 0.5625,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 213646008,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3274807654369698,
+      "grad_norm": 4.550203288561425,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 213816528,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3274807654369698,
+      "loss": 0.10727906972169876,
+      "loss_ce": 0.001383074326440692,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 213816528,
+      "step": 1245
+    },
+    {
+      "epoch": 0.32774380219635696,
+      "grad_norm": 8.680127486966335,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 213988932,
+      "step": 1246
+    },
+    {
+      "epoch": 0.32774380219635696,
+      "loss": 0.16796328127384186,
+      "loss_ce": 0.0009405763121321797,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 213988932,
+      "step": 1246
+    },
+    {
+      "epoch": 0.32800683895574406,
+      "grad_norm": 9.402943236610666,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 214161184,
+      "step": 1247
+    },
+    {
+      "epoch": 0.32800683895574406,
+      "loss": 0.18202053010463715,
+      "loss_ce": 0.008070331066846848,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 214161184,
+      "step": 1247
+    },
+    {
+      "epoch": 0.32826987571513117,
+      "grad_norm": 8.949806900643816,
+      "learning_rate": 5e-06,
+      "loss": 0.1733,
+      "num_input_tokens_seen": 214333220,
+      "step": 1248
+    },
+    {
+      "epoch": 0.32826987571513117,
+      "loss": 0.18508628010749817,
+      "loss_ce": 0.001980806002393365,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 214333220,
+      "step": 1248
+    },
+    {
+      "epoch": 0.3285329124745183,
+      "grad_norm": 9.623428359724258,
+      "learning_rate": 5e-06,
+      "loss": 0.1899,
+      "num_input_tokens_seen": 214501988,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3285329124745183,
+      "loss": 0.18457330763339996,
+      "loss_ce": 0.002108700107783079,
+      "loss_iou": 0.46875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 214501988,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "grad_norm": 5.549112096762312,
+      "learning_rate": 5e-06,
+      "loss": 0.1748,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_websight_new_CIoU": 0.8321745097637177,
+      "eval_websight_new_GIoU": 0.8282029330730438,
+      "eval_websight_new_IoU": 0.8425185084342957,
+      "eval_websight_new_MAE_all": 0.030896120704710484,
+      "eval_websight_new_MAE_h": 0.020350518636405468,
+      "eval_websight_new_MAE_w": 0.041066043078899384,
+      "eval_websight_new_MAE_x": 0.04287236928939819,
+      "eval_websight_new_MAE_y": 0.019295550882816315,
+      "eval_websight_new_NUM_probability": 0.9998577535152435,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.13457995653152466,
+      "eval_websight_new_loss_ce": 2.8165732146590017e-05,
+      "eval_websight_new_loss_iou": 0.39599609375,
+      "eval_websight_new_loss_num": 0.022901535034179688,
+      "eval_websight_new_loss_xval": 0.1145477294921875,
+      "eval_websight_new_runtime": 54.6461,
+      "eval_websight_new_samples_per_second": 0.915,
+      "eval_websight_new_steps_per_second": 0.037,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_seeclick_CIoU": 0.5724749565124512,
+      "eval_seeclick_GIoU": 0.5666466653347015,
+      "eval_seeclick_IoU": 0.5994701087474823,
+      "eval_seeclick_MAE_all": 0.054681919515132904,
+      "eval_seeclick_MAE_h": 0.03272205591201782,
+      "eval_seeclick_MAE_w": 0.07547581382095814,
+      "eval_seeclick_MAE_x": 0.07945681735873222,
+      "eval_seeclick_MAE_y": 0.031073003076016903,
+      "eval_seeclick_NUM_probability": 0.9999328255653381,
+      "eval_seeclick_inside_bbox": 0.8465909063816071,
+      "eval_seeclick_loss": 0.2362295389175415,
+      "eval_seeclick_loss_ce": 0.008988222572952509,
+      "eval_seeclick_loss_iou": 0.5384521484375,
+      "eval_seeclick_loss_num": 0.04405975341796875,
+      "eval_seeclick_loss_xval": 0.22039794921875,
+      "eval_seeclick_runtime": 77.1136,
+      "eval_seeclick_samples_per_second": 0.558,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_icons_CIoU": 0.8087750673294067,
+      "eval_icons_GIoU": 0.7993076145648956,
+      "eval_icons_IoU": 0.8191726505756378,
+      "eval_icons_MAE_all": 0.02884785458445549,
+      "eval_icons_MAE_h": 0.033479243516922,
+      "eval_icons_MAE_w": 0.024947408586740494,
+      "eval_icons_MAE_x": 0.02466664183884859,
+      "eval_icons_MAE_y": 0.0322981309145689,
+      "eval_icons_NUM_probability": 0.9998048841953278,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.08432088792324066,
+      "eval_icons_loss_ce": 7.355650814133696e-05,
+      "eval_icons_loss_iou": 0.567626953125,
+      "eval_icons_loss_num": 0.015069961547851562,
+      "eval_icons_loss_xval": 0.0753631591796875,
+      "eval_icons_runtime": 78.2995,
+      "eval_icons_samples_per_second": 0.639,
+      "eval_icons_steps_per_second": 0.026,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_screenspot_CIoU": 0.5492339730262756,
+      "eval_screenspot_GIoU": 0.5352775057156881,
+      "eval_screenspot_IoU": 0.5879749854405721,
+      "eval_screenspot_MAE_all": 0.0870671272277832,
+      "eval_screenspot_MAE_h": 0.051297743494311966,
+      "eval_screenspot_MAE_w": 0.14652628699938455,
+      "eval_screenspot_MAE_x": 0.10099601248900096,
+      "eval_screenspot_MAE_y": 0.04944847462077936,
+      "eval_screenspot_NUM_probability": 0.999727189540863,
+      "eval_screenspot_inside_bbox": 0.8529166579246521,
+      "eval_screenspot_loss": 0.8003170490264893,
+      "eval_screenspot_loss_ce": 0.4581688741842906,
+      "eval_screenspot_loss_iou": 0.4616292317708333,
+      "eval_screenspot_loss_num": 0.06711324055989583,
+      "eval_screenspot_loss_xval": 0.3355305989583333,
+      "eval_screenspot_runtime": 146.7221,
+      "eval_screenspot_samples_per_second": 0.607,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "loss": 0.7908815741539001,
+      "loss_ce": 0.44920679926872253,
+      "loss_iou": 0.390625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3290589859932926,
+      "grad_norm": 5.65145087220912,
+      "learning_rate": 5e-06,
+      "loss": 0.1612,
+      "num_input_tokens_seen": 214840640,
+      "step": 1251
+    },
+    {
+      "epoch": 0.3290589859932926,
+      "loss": 0.17782355844974518,
+      "loss_ce": 0.0008826321572996676,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 214840640,
+      "step": 1251
+    },
+    {
+      "epoch": 0.3293220227526797,
+      "grad_norm": 7.22202274640801,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 215011212,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3293220227526797,
+      "loss": 0.13640594482421875,
+      "loss_ce": 0.0011520386906340718,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 215011212,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3295850595120668,
+      "grad_norm": 5.1577086011143685,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 215183424,
+      "step": 1253
+    },
+    {
+      "epoch": 0.3295850595120668,
+      "loss": 0.12353098392486572,
+      "loss_ce": 0.0006061755702830851,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 215183424,
+      "step": 1253
+    },
+    {
+      "epoch": 0.32984809627145395,
+      "grad_norm": 7.368992653841811,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 215355224,
+      "step": 1254
+    },
+    {
+      "epoch": 0.32984809627145395,
+      "loss": 0.15012162923812866,
+      "loss_ce": 0.0011348105035722256,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 215355224,
+      "step": 1254
+    },
+    {
+      "epoch": 0.33011113303084105,
+      "grad_norm": 5.485255366846693,
+      "learning_rate": 5e-06,
+      "loss": 0.1751,
+      "num_input_tokens_seen": 215525468,
+      "step": 1255
+    },
+    {
+      "epoch": 0.33011113303084105,
+      "loss": 0.14393854141235352,
+      "loss_ce": 0.0008110922062769532,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 215525468,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3303741697902282,
+      "grad_norm": 5.150585875058386,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 215697664,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3303741697902282,
+      "loss": 0.1360258162021637,
+      "loss_ce": 0.0006498318398371339,
+      "loss_iou": 0.6875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 215697664,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3306372065496153,
+      "grad_norm": 4.902248025801288,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 215869772,
+      "step": 1257
+    },
+    {
+      "epoch": 0.3306372065496153,
+      "loss": 0.08430365473031998,
+      "loss_ce": 0.0006854891544207931,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 215869772,
+      "step": 1257
+    },
+    {
+      "epoch": 0.3309002433090024,
+      "grad_norm": 8.067058733106478,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 216041972,
+      "step": 1258
+    },
+    {
+      "epoch": 0.3309002433090024,
+      "loss": 0.12436328083276749,
+      "loss_ce": 0.0012248535640537739,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 216041972,
+      "step": 1258
+    },
+    {
+      "epoch": 0.3311632800683896,
+      "grad_norm": 10.281119628319207,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 216213968,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3311632800683896,
+      "loss": 0.1294836401939392,
+      "loss_ce": 0.002225350122898817,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 216213968,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3314263168277767,
+      "grad_norm": 9.188895459960923,
+      "learning_rate": 5e-06,
+      "loss": 0.1358,
+      "num_input_tokens_seen": 216384484,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3314263168277767,
+      "loss": 0.17113396525382996,
+      "loss_ce": 0.0021886550821363926,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 216384484,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3316893535871638,
+      "grad_norm": 9.858408287132631,
+      "learning_rate": 5e-06,
+      "loss": 0.1691,
+      "num_input_tokens_seen": 216556712,
+      "step": 1261
+    },
+    {
+      "epoch": 0.3316893535871638,
+      "loss": 0.09802193194627762,
+      "loss_ce": 0.004027791786938906,
+      "loss_iou": 0.53125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 216556712,
+      "step": 1261
+    },
+    {
+      "epoch": 0.33195239034655094,
+      "grad_norm": 13.43147706837513,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 216727532,
+      "step": 1262
+    },
+    {
+      "epoch": 0.33195239034655094,
+      "loss": 0.12088888883590698,
+      "loss_ce": 0.001778791076503694,
+      "loss_iou": NaN,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 216727532,
+      "step": 1262
+    },
+    {
+      "epoch": 0.33221542710593804,
+      "grad_norm": 22.307350229032924,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 216899424,
+      "step": 1263
+    },
+    {
+      "epoch": 0.33221542710593804,
+      "loss": 0.13408984243869781,
+      "loss_ce": 0.00048388654249720275,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 216899424,
+      "step": 1263
+    },
+    {
+      "epoch": 0.3324784638653252,
+      "grad_norm": 12.430687637818028,
+      "learning_rate": 5e-06,
+      "loss": 0.1269,
+      "num_input_tokens_seen": 217071512,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3324784638653252,
+      "loss": 0.12219381332397461,
+      "loss_ce": 0.0014357574982568622,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 217071512,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3327415006247123,
+      "grad_norm": 5.892819252370443,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 217241728,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3327415006247123,
+      "loss": 0.07451170682907104,
+      "loss_ce": 0.0026122929994016886,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 217241728,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3330045373840994,
+      "grad_norm": 7.045469734572638,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 217413824,
+      "step": 1266
+    },
+    {
+      "epoch": 0.3330045373840994,
+      "loss": 0.19595244526863098,
+      "loss_ce": 0.0024709957651793957,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 217413824,
+      "step": 1266
+    },
+    {
+      "epoch": 0.33326757414348657,
+      "grad_norm": 10.385925957202762,
+      "learning_rate": 5e-06,
+      "loss": 0.1572,
+      "num_input_tokens_seen": 217585864,
+      "step": 1267
+    },
+    {
+      "epoch": 0.33326757414348657,
+      "loss": 0.14605620503425598,
+      "loss_ce": 0.0005483938148245215,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 217585864,
+      "step": 1267
+    },
+    {
+      "epoch": 0.33353061090287367,
+      "grad_norm": 6.514979263584694,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 217757720,
+      "step": 1268
+    },
+    {
+      "epoch": 0.33353061090287367,
+      "loss": 0.1336522251367569,
+      "loss_ce": 0.000595581834204495,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 217757720,
+      "step": 1268
+    },
+    {
+      "epoch": 0.33379364766226083,
+      "grad_norm": 7.782042489694296,
+      "learning_rate": 5e-06,
+      "loss": 0.1537,
+      "num_input_tokens_seen": 217929844,
+      "step": 1269
+    },
+    {
+      "epoch": 0.33379364766226083,
+      "loss": 0.14830312132835388,
+      "loss_ce": 0.0005370010621845722,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 217929844,
+      "step": 1269
+    },
+    {
+      "epoch": 0.33405668442164793,
+      "grad_norm": 37.57598415614969,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 218102180,
+      "step": 1270
+    },
+    {
+      "epoch": 0.33405668442164793,
+      "loss": 0.125936821103096,
+      "loss_ce": 0.000570611678995192,
+      "loss_iou": 0.77734375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 218102180,
+      "step": 1270
+    },
+    {
+      "epoch": 0.33431972118103503,
+      "grad_norm": 11.611735941773203,
+      "learning_rate": 5e-06,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 218272528,
+      "step": 1271
+    },
+    {
+      "epoch": 0.33431972118103503,
+      "loss": 0.15356029570102692,
+      "loss_ce": 0.0005146406474523246,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 218272528,
+      "step": 1271
+    },
+    {
+      "epoch": 0.3345827579404222,
+      "grad_norm": 8.349189669659024,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 218445008,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3345827579404222,
+      "loss": 0.14823183417320251,
+      "loss_ce": 0.0009540055179968476,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 218445008,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3348457946998093,
+      "grad_norm": 8.794519146282962,
+      "learning_rate": 5e-06,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 218617480,
+      "step": 1273
+    },
+    {
+      "epoch": 0.3348457946998093,
+      "loss": 0.10918127745389938,
+      "loss_ce": 0.0011185340117663145,
+      "loss_iou": 0.625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 218617480,
+      "step": 1273
+    },
+    {
+      "epoch": 0.3351088314591964,
+      "grad_norm": 5.0624867047638045,
+      "learning_rate": 5e-06,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 218789900,
+      "step": 1274
+    },
+    {
+      "epoch": 0.3351088314591964,
+      "loss": 0.13429242372512817,
+      "loss_ce": 0.00411969143897295,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 218789900,
+      "step": 1274
+    },
+    {
+      "epoch": 0.33537186821858356,
+      "grad_norm": 5.751729189048832,
+      "learning_rate": 5e-06,
+      "loss": 0.1612,
+      "num_input_tokens_seen": 218962156,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33537186821858356,
+      "loss": 0.11978072673082352,
+      "loss_ce": 0.00350876129232347,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 218962156,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33563490497797066,
+      "grad_norm": 6.311955311958974,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 219132288,
+      "step": 1276
+    },
+    {
+      "epoch": 0.33563490497797066,
+      "loss": 0.1355704814195633,
+      "loss_ce": 0.0025443662889301777,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 219132288,
+      "step": 1276
+    },
+    {
+      "epoch": 0.3358979417373578,
+      "grad_norm": 7.099991467306234,
+      "learning_rate": 5e-06,
+      "loss": 0.1531,
+      "num_input_tokens_seen": 219304616,
+      "step": 1277
+    },
+    {
+      "epoch": 0.3358979417373578,
+      "loss": 0.13075271248817444,
+      "loss_ce": 0.0017243996262550354,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 219304616,
+      "step": 1277
+    },
+    {
+      "epoch": 0.3361609784967449,
+      "grad_norm": 7.34332951683246,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 219476876,
+      "step": 1278
+    },
+    {
+      "epoch": 0.3361609784967449,
+      "loss": 0.08354561030864716,
+      "loss_ce": 0.0010565929114818573,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 219476876,
+      "step": 1278
+    },
+    {
+      "epoch": 0.336424015256132,
+      "grad_norm": 20.864457223612828,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 219649440,
+      "step": 1279
+    },
+    {
+      "epoch": 0.336424015256132,
+      "loss": 0.14009949564933777,
+      "loss_ce": 0.00029846589313820004,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 219649440,
+      "step": 1279
+    },
+    {
+      "epoch": 0.3366870520155192,
+      "grad_norm": 6.27874029788255,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 219821712,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3366870520155192,
+      "loss": 0.14031162858009338,
+      "loss_ce": 0.0007242212886922061,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 219821712,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3369500887749063,
+      "grad_norm": 4.5668271174612025,
+      "learning_rate": 5e-06,
+      "loss": 0.1628,
+      "num_input_tokens_seen": 219993864,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3369500887749063,
+      "loss": 0.10837851464748383,
+      "loss_ce": 0.0014449162408709526,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 219993864,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3372131255342934,
+      "grad_norm": 4.389742759586719,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 220166248,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3372131255342934,
+      "loss": 0.23685070872306824,
+      "loss_ce": 0.0005836054333485663,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 220166248,
+      "step": 1282
+    },
+    {
+      "epoch": 0.33747616229368055,
+      "grad_norm": 5.570317525274513,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 220338428,
+      "step": 1283
+    },
+    {
+      "epoch": 0.33747616229368055,
+      "loss": 0.13890241086483002,
+      "loss_ce": 0.0007493281736969948,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 220338428,
+      "step": 1283
+    },
+    {
+      "epoch": 0.33773919905306765,
+      "grad_norm": 10.985772181850571,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 220510776,
+      "step": 1284
+    },
+    {
+      "epoch": 0.33773919905306765,
+      "loss": 0.12931521236896515,
+      "loss_ce": 0.0006531005492433906,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 220510776,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3380022358124548,
+      "grad_norm": 5.926973995328122,
+      "learning_rate": 5e-06,
+      "loss": 0.139,
+      "num_input_tokens_seen": 220682848,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3380022358124548,
+      "loss": 0.21676884591579437,
+      "loss_ce": 0.007204629480838776,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 220682848,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3382652725718419,
+      "grad_norm": 7.996976963449769,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 220855140,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3382652725718419,
+      "loss": 0.08844804763793945,
+      "loss_ce": 0.0003895749687217176,
+      "loss_iou": 0.59375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 220855140,
+      "step": 1286
+    },
+    {
+      "epoch": 0.338528309331229,
+      "grad_norm": 8.499510082866491,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 221027260,
+      "step": 1287
+    },
+    {
+      "epoch": 0.338528309331229,
+      "loss": 0.12149707973003387,
+      "loss_ce": 0.0004948831629008055,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 221027260,
+      "step": 1287
+    },
+    {
+      "epoch": 0.3387913460906162,
+      "grad_norm": 5.973131676418204,
+      "learning_rate": 5e-06,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 221199416,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3387913460906162,
+      "loss": 0.09466279298067093,
+      "loss_ce": 0.00011933030327782035,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 221199416,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3390543828500033,
+      "grad_norm": 4.808893821005273,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 221371720,
+      "step": 1289
+    },
+    {
+      "epoch": 0.3390543828500033,
+      "loss": 0.09180793166160583,
+      "loss_ce": 0.0001941679511219263,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 221371720,
+      "step": 1289
+    },
+    {
+      "epoch": 0.33931741960939044,
+      "grad_norm": 7.016843255216574,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 221542276,
+      "step": 1290
+    },
+    {
+      "epoch": 0.33931741960939044,
+      "loss": 0.19763490557670593,
+      "loss_ce": 0.0007354922126978636,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 221542276,
+      "step": 1290
+    },
+    {
+      "epoch": 0.33958045636877754,
+      "grad_norm": 6.06110322351886,
+      "learning_rate": 5e-06,
+      "loss": 0.1227,
+      "num_input_tokens_seen": 221714344,
+      "step": 1291
+    },
+    {
+      "epoch": 0.33958045636877754,
+      "loss": 0.15020573139190674,
+      "loss_ce": 0.0027447929605841637,
+      "loss_iou": 0.4375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 221714344,
+      "step": 1291
+    },
+    {
+      "epoch": 0.33984349312816464,
+      "grad_norm": 8.158154012506673,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 221886284,
+      "step": 1292
+    },
+    {
+      "epoch": 0.33984349312816464,
+      "loss": 0.06007716804742813,
+      "loss_ce": 0.0002016789367189631,
+      "loss_iou": 0.609375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 221886284,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3401065298875518,
+      "grad_norm": 4.887380704800873,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 222058520,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3401065298875518,
+      "loss": 0.09021516144275665,
+      "loss_ce": 0.0008291734848171473,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 222058520,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3403695666469389,
+      "grad_norm": 13.907525065592614,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 222231156,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3403695666469389,
+      "loss": 0.22907251119613647,
+      "loss_ce": 0.0007704915478825569,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 222231156,
+      "step": 1294
+    },
+    {
+      "epoch": 0.340632603406326,
+      "grad_norm": 7.536500056913184,
+      "learning_rate": 5e-06,
+      "loss": 0.1822,
+      "num_input_tokens_seen": 222403476,
+      "step": 1295
+    },
+    {
+      "epoch": 0.340632603406326,
+      "loss": 0.1321713924407959,
+      "loss_ce": 0.002853148616850376,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 222403476,
+      "step": 1295
+    },
+    {
+      "epoch": 0.34089564016571317,
+      "grad_norm": 8.538688840874034,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 222575540,
+      "step": 1296
+    },
+    {
+      "epoch": 0.34089564016571317,
+      "loss": 0.09548459947109222,
+      "loss_ce": 0.0024365070275962353,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 222575540,
+      "step": 1296
+    },
+    {
+      "epoch": 0.34115867692510027,
+      "grad_norm": 4.787366743041949,
+      "learning_rate": 5e-06,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 222747704,
+      "step": 1297
+    },
+    {
+      "epoch": 0.34115867692510027,
+      "loss": 0.1637537181377411,
+      "loss_ce": 0.0010034843580797315,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 222747704,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3414217136844874,
+      "grad_norm": 6.917043775641942,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 222919896,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3414217136844874,
+      "loss": 0.05339755862951279,
+      "loss_ce": 0.001883886638097465,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 222919896,
+      "step": 1298
+    },
+    {
+      "epoch": 0.34168475044387453,
+      "grad_norm": 8.403015648704343,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 223092124,
+      "step": 1299
+    },
+    {
+      "epoch": 0.34168475044387453,
+      "loss": 0.15119820833206177,
+      "loss_ce": 0.0011737870518118143,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 223092124,
+      "step": 1299
+    },
+    {
+      "epoch": 0.34194778720326163,
+      "grad_norm": 7.291522615294142,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 223264436,
+      "step": 1300
+    },
+    {
+      "epoch": 0.34194778720326163,
+      "loss": 0.14489704370498657,
+      "loss_ce": 0.0020442616660147905,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 223264436,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3422108239626488,
+      "grad_norm": 8.380731335959293,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 223436588,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3422108239626488,
+      "loss": 0.15774638950824738,
+      "loss_ce": 0.0007334585534408689,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 223436588,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3424738607220359,
+      "grad_norm": 4.869592726119341,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 223607388,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3424738607220359,
+      "loss": 0.15213216841220856,
+      "loss_ce": 0.0007344604237005115,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 223607388,
+      "step": 1302
+    },
+    {
+      "epoch": 0.34273689748142305,
+      "grad_norm": 5.639489902327201,
+      "learning_rate": 5e-06,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 223779372,
+      "step": 1303
+    },
+    {
+      "epoch": 0.34273689748142305,
+      "loss": 0.17259347438812256,
+      "loss_ce": 0.0032819565385580063,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 223779372,
+      "step": 1303
+    },
+    {
+      "epoch": 0.34299993424081016,
+      "grad_norm": 9.093938521490475,
+      "learning_rate": 5e-06,
+      "loss": 0.2003,
+      "num_input_tokens_seen": 223951388,
+      "step": 1304
+    },
+    {
+      "epoch": 0.34299993424081016,
+      "loss": 0.22136257588863373,
+      "loss_ce": 0.0015749745070934296,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 223951388,
+      "step": 1304
+    },
+    {
+      "epoch": 0.34326297100019726,
+      "grad_norm": 6.510612178547868,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 224123456,
+      "step": 1305
+    },
+    {
+      "epoch": 0.34326297100019726,
+      "loss": 0.14626947045326233,
+      "loss_ce": 0.0030199564062058926,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 224123456,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3435260077595844,
+      "grad_norm": 6.4702586677675855,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "num_input_tokens_seen": 224295676,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3435260077595844,
+      "loss": 0.20608918368816376,
+      "loss_ce": 0.0004007022944279015,
+      "loss_iou": 0.671875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 224295676,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3437890445189715,
+      "grad_norm": 24.89374059454028,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 224467880,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3437890445189715,
+      "loss": 0.12941348552703857,
+      "loss_ce": 0.000507236341945827,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 224467880,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3440520812783586,
+      "grad_norm": 8.792041343116132,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 224640076,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3440520812783586,
+      "loss": 0.18377459049224854,
+      "loss_ce": 0.0007301591685973108,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 224640076,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3443151180377458,
+      "grad_norm": 7.075463244307207,
+      "learning_rate": 5e-06,
+      "loss": 0.1801,
+      "num_input_tokens_seen": 224812516,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3443151180377458,
+      "loss": 0.10930690169334412,
+      "loss_ce": 0.0017934793140739202,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 224812516,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3445781547971329,
+      "grad_norm": 5.622949079750027,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 224985136,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3445781547971329,
+      "loss": 0.13441142439842224,
+      "loss_ce": 0.0033689369447529316,
+      "loss_iou": 0.625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 224985136,
+      "step": 1310
+    },
+    {
+      "epoch": 0.34484119155652004,
+      "grad_norm": 6.126844030056675,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 225157120,
+      "step": 1311
+    },
+    {
+      "epoch": 0.34484119155652004,
+      "loss": 0.1415342092514038,
+      "loss_ce": 0.000787146098446101,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 225157120,
+      "step": 1311
+    },
+    {
+      "epoch": 0.34510422831590715,
+      "grad_norm": 5.277811585596711,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 225329012,
+      "step": 1312
+    },
+    {
+      "epoch": 0.34510422831590715,
+      "loss": 0.08680924028158188,
+      "loss_ce": 0.006761634722352028,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 225329012,
+      "step": 1312
+    },
+    {
+      "epoch": 0.34536726507529425,
+      "grad_norm": 4.953551061650772,
+      "learning_rate": 5e-06,
+      "loss": 0.0972,
+      "num_input_tokens_seen": 225501536,
+      "step": 1313
+    },
+    {
+      "epoch": 0.34536726507529425,
+      "loss": 0.10339295864105225,
+      "loss_ce": 0.0035394439473748207,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 225501536,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3456303018346814,
+      "grad_norm": 14.376989767483847,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 225673976,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3456303018346814,
+      "loss": 0.13013647496700287,
+      "loss_ce": 0.0019016144797205925,
+      "loss_iou": 0.76953125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 225673976,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3458933385940685,
+      "grad_norm": 4.351015648559307,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 225844284,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3458933385940685,
+      "loss": 0.07338554412126541,
+      "loss_ce": 0.0006926720961928368,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 225844284,
+      "step": 1315
+    },
+    {
+      "epoch": 0.34615637535345567,
+      "grad_norm": 8.704702003668412,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 226014724,
+      "step": 1316
+    },
+    {
+      "epoch": 0.34615637535345567,
+      "loss": 0.13227002322673798,
+      "loss_ce": 0.00031201643287204206,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 226014724,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3464194121128428,
+      "grad_norm": 9.402885662503637,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 226186864,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3464194121128428,
+      "loss": 0.13149945437908173,
+      "loss_ce": 0.0032035536132752895,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 226186864,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3466824488722299,
+      "grad_norm": 5.4441598572448475,
+      "learning_rate": 5e-06,
+      "loss": 0.1247,
+      "num_input_tokens_seen": 226359284,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3466824488722299,
+      "loss": 0.09241662174463272,
+      "loss_ce": 0.00010094831668538973,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 226359284,
+      "step": 1318
+    },
+    {
+      "epoch": 0.34694548563161703,
+      "grad_norm": 5.838383036724073,
+      "learning_rate": 5e-06,
+      "loss": 0.1544,
+      "num_input_tokens_seen": 226531468,
+      "step": 1319
+    },
+    {
+      "epoch": 0.34694548563161703,
+      "loss": 0.10588014125823975,
+      "loss_ce": 0.00016725034220144153,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 226531468,
+      "step": 1319
+    },
+    {
+      "epoch": 0.34720852239100414,
+      "grad_norm": 6.350689517126913,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 226703648,
+      "step": 1320
+    },
+    {
+      "epoch": 0.34720852239100414,
+      "loss": 0.11672288179397583,
+      "loss_ce": 0.0005119539564475417,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 226703648,
+      "step": 1320
+    },
+    {
+      "epoch": 0.34747155915039124,
+      "grad_norm": 13.847941425419814,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 226873276,
+      "step": 1321
+    },
+    {
+      "epoch": 0.34747155915039124,
+      "loss": 0.1297488510608673,
+      "loss_ce": 0.0002322565414942801,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 226873276,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3477345959097784,
+      "grad_norm": 23.595553939708278,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 227045376,
+      "step": 1322
+    },
+    {
+      "epoch": 0.3477345959097784,
+      "loss": 0.1232631504535675,
+      "loss_ce": 0.00033834436908364296,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 227045376,
+      "step": 1322
+    },
+    {
+      "epoch": 0.3479976326691655,
+      "grad_norm": 44.36951009678579,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 227215620,
+      "step": 1323
+    },
+    {
+      "epoch": 0.3479976326691655,
+      "loss": 0.07254654914140701,
+      "loss_ce": 0.000280924781691283,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 227215620,
+      "step": 1323
+    },
+    {
+      "epoch": 0.34826066942855266,
+      "grad_norm": 5.242558618670207,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 227387716,
+      "step": 1324
+    },
+    {
+      "epoch": 0.34826066942855266,
+      "loss": 0.11591322720050812,
+      "loss_ce": 0.0015028227353468537,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 227387716,
+      "step": 1324
+    },
+    {
+      "epoch": 0.34852370618793976,
+      "grad_norm": 6.076654000097945,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 227560176,
+      "step": 1325
+    },
+    {
+      "epoch": 0.34852370618793976,
+      "loss": 0.09558144956827164,
+      "loss_ce": 0.0009464399190619588,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 227560176,
+      "step": 1325
+    },
+    {
+      "epoch": 0.34878674294732687,
+      "grad_norm": 8.445480824601002,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 227732380,
+      "step": 1326
+    },
+    {
+      "epoch": 0.34878674294732687,
+      "loss": 0.18825021386146545,
+      "loss_ce": 0.0008112426148727536,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 227732380,
+      "step": 1326
+    },
+    {
+      "epoch": 0.349049779706714,
+      "grad_norm": 7.063227159799308,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 227904472,
+      "step": 1327
+    },
+    {
+      "epoch": 0.349049779706714,
+      "loss": 0.23714260756969452,
+      "loss_ce": 0.0035610701888799667,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 227904472,
+      "step": 1327
+    },
+    {
+      "epoch": 0.34931281646610113,
+      "grad_norm": 5.912334076761878,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 228076948,
+      "step": 1328
+    },
+    {
+      "epoch": 0.34931281646610113,
+      "loss": 0.14180362224578857,
+      "loss_ce": 0.0037115837913006544,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 228076948,
+      "step": 1328
+    },
+    {
+      "epoch": 0.3495758532254883,
+      "grad_norm": 6.00754661823903,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 228248972,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3495758532254883,
+      "loss": 0.14186972379684448,
+      "loss_ce": 0.0010616088984534144,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 228248972,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3498388899848754,
+      "grad_norm": 5.090693648053783,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 228421168,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3498388899848754,
+      "loss": 0.08804985880851746,
+      "loss_ce": 0.0005559585988521576,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 228421168,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3501019267442625,
+      "grad_norm": 5.2360127706674335,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 228593364,
+      "step": 1331
+    },
+    {
+      "epoch": 0.3501019267442625,
+      "loss": 0.09556721895933151,
+      "loss_ce": 0.0012373843928799033,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 228593364,
+      "step": 1331
+    },
+    {
+      "epoch": 0.35036496350364965,
+      "grad_norm": 7.817751449255831,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 228765480,
+      "step": 1332
+    },
+    {
+      "epoch": 0.35036496350364965,
+      "loss": 0.19872622191905975,
+      "loss_ce": 0.0040851132944226265,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 228765480,
+      "step": 1332
+    },
+    {
+      "epoch": 0.35062800026303675,
+      "grad_norm": 6.154736113026185,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 228938028,
+      "step": 1333
+    },
+    {
+      "epoch": 0.35062800026303675,
+      "loss": 0.08339493721723557,
+      "loss_ce": 0.0010737692937254906,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 228938028,
+      "step": 1333
+    },
+    {
+      "epoch": 0.35089103702242386,
+      "grad_norm": 9.10695022243035,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 229110180,
+      "step": 1334
+    },
+    {
+      "epoch": 0.35089103702242386,
+      "loss": 0.17051678895950317,
+      "loss_ce": 0.0020597607363015413,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 229110180,
+      "step": 1334
+    },
+    {
+      "epoch": 0.351154073781811,
+      "grad_norm": 6.861938259215169,
+      "learning_rate": 5e-06,
+      "loss": 0.192,
+      "num_input_tokens_seen": 229282592,
+      "step": 1335
+    },
+    {
+      "epoch": 0.351154073781811,
+      "loss": 0.20712369680404663,
+      "loss_ce": 0.003968170844018459,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 229282592,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3514171105411981,
+      "grad_norm": 6.702358306191994,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 229454940,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3514171105411981,
+      "loss": 0.1241491287946701,
+      "loss_ce": 0.002719694282859564,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 229454940,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3516801473005853,
+      "grad_norm": 25.54665143162104,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 229626740,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3516801473005853,
+      "loss": 0.13124999403953552,
+      "loss_ce": 0.0006042490131221712,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 229626740,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3519431840599724,
+      "grad_norm": 10.986092691383933,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 229795396,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3519431840599724,
+      "loss": 0.07223416119813919,
+      "loss_ce": 0.00028897292213514447,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 229795396,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3522062208193595,
+      "grad_norm": 9.146823823601064,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 229967592,
+      "step": 1339
+    },
+    {
+      "epoch": 0.3522062208193595,
+      "loss": 0.14658141136169434,
+      "loss_ce": 0.00021910574287176132,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 229967592,
+      "step": 1339
+    },
+    {
+      "epoch": 0.35246925757874664,
+      "grad_norm": 12.31823820919867,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 230139960,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35246925757874664,
+      "loss": 0.07799485325813293,
+      "loss_ce": 0.0002360617509111762,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 230139960,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35273229433813375,
+      "grad_norm": 6.277536451715721,
+      "learning_rate": 5e-06,
+      "loss": 0.1745,
+      "num_input_tokens_seen": 230312536,
+      "step": 1341
+    },
+    {
+      "epoch": 0.35273229433813375,
+      "loss": 0.15891912579536438,
+      "loss_ce": 0.0026691171806305647,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 230312536,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3529953310975209,
+      "grad_norm": 10.816707100229003,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 230482960,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3529953310975209,
+      "loss": 0.10626394301652908,
+      "loss_ce": 0.0003069115919061005,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 230482960,
+      "step": 1342
+    },
+    {
+      "epoch": 0.353258367856908,
+      "grad_norm": 7.461204110486792,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 230655564,
+      "step": 1343
+    },
+    {
+      "epoch": 0.353258367856908,
+      "loss": 0.14832650125026703,
+      "loss_ce": 0.005412681493908167,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 230655564,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3535214046162951,
+      "grad_norm": 3.829230892926663,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 230827776,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3535214046162951,
+      "loss": 0.11698116362094879,
+      "loss_ce": 0.0007397143635898829,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 230827776,
+      "step": 1344
+    },
+    {
+      "epoch": 0.35378444137568227,
+      "grad_norm": 7.33327936841338,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 230999976,
+      "step": 1345
+    },
+    {
+      "epoch": 0.35378444137568227,
+      "loss": 0.18063044548034668,
+      "loss_ce": 0.005978355184197426,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 230999976,
+      "step": 1345
+    },
+    {
+      "epoch": 0.35404747813506937,
+      "grad_norm": 4.588965012769037,
+      "learning_rate": 5e-06,
+      "loss": 0.1907,
+      "num_input_tokens_seen": 231171944,
+      "step": 1346
+    },
+    {
+      "epoch": 0.35404747813506937,
+      "loss": 0.19280412793159485,
+      "loss_ce": 0.0011537342797964811,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 231171944,
+      "step": 1346
+    },
+    {
+      "epoch": 0.3543105148944565,
+      "grad_norm": 15.486355760387976,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 231342336,
+      "step": 1347
+    },
+    {
+      "epoch": 0.3543105148944565,
+      "loss": 0.11234519630670547,
+      "loss_ce": 0.0008644815534353256,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 231342336,
+      "step": 1347
+    },
+    {
+      "epoch": 0.35457355165384363,
+      "grad_norm": 17.3485969876628,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 231514328,
+      "step": 1348
+    },
+    {
+      "epoch": 0.35457355165384363,
+      "loss": 0.12869834899902344,
+      "loss_ce": 0.0009517711587250233,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 231514328,
+      "step": 1348
+    },
+    {
+      "epoch": 0.35483658841323074,
+      "grad_norm": 19.41812876236682,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 231686780,
+      "step": 1349
+    },
+    {
+      "epoch": 0.35483658841323074,
+      "loss": 0.13225057721138,
+      "loss_ce": 0.0008724014624021947,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 231686780,
+      "step": 1349
+    },
+    {
+      "epoch": 0.3550996251726179,
+      "grad_norm": 10.841969185593292,
+      "learning_rate": 5e-06,
+      "loss": 0.1386,
+      "num_input_tokens_seen": 231858892,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3550996251726179,
+      "loss": 0.1675529032945633,
+      "loss_ce": 0.00019450299441814423,
+      "loss_iou": 0.6875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 231858892,
+      "step": 1350
+    },
+    {
+      "epoch": 0.355362661932005,
+      "grad_norm": 9.874209565962254,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 232029240,
+      "step": 1351
+    },
+    {
+      "epoch": 0.355362661932005,
+      "loss": 0.13038085401058197,
+      "loss_ce": 0.00037596753099933267,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 232029240,
+      "step": 1351
+    },
+    {
+      "epoch": 0.3556256986913921,
+      "grad_norm": 6.572064767991026,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 232199320,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3556256986913921,
+      "loss": 0.11705964803695679,
+      "loss_ce": 0.00029938769876025617,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 232199320,
+      "step": 1352
+    },
+    {
+      "epoch": 0.35588873545077926,
+      "grad_norm": 19.449170559601413,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 232371336,
+      "step": 1353
+    },
+    {
+      "epoch": 0.35588873545077926,
+      "loss": 0.10365074872970581,
+      "loss_ce": 0.00291221821680665,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 232371336,
+      "step": 1353
+    },
+    {
+      "epoch": 0.35615177221016636,
+      "grad_norm": 6.060818585296291,
+      "learning_rate": 5e-06,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 232543572,
+      "step": 1354
+    },
+    {
+      "epoch": 0.35615177221016636,
+      "loss": 0.22082458436489105,
+      "loss_ce": 0.005858768709003925,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 232543572,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3564148089695535,
+      "grad_norm": 5.2405872305191386,
+      "learning_rate": 5e-06,
+      "loss": 0.1508,
+      "num_input_tokens_seen": 232715652,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3564148089695535,
+      "loss": 0.1176377683877945,
+      "loss_ce": 0.001854072092100978,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 232715652,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3566778457289406,
+      "grad_norm": 9.250810729782357,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 232888076,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3566778457289406,
+      "loss": 0.2215210646390915,
+      "loss_ce": 0.0008179338765330613,
+      "loss_iou": 0.59375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 232888076,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3569408824883277,
+      "grad_norm": 5.562200706936175,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 233060156,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3569408824883277,
+      "loss": 0.15043510496616364,
+      "loss_ce": 0.0006548258243128657,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 233060156,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3572039192477149,
+      "grad_norm": 5.308030458771483,
+      "learning_rate": 5e-06,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 233232200,
+      "step": 1358
+    },
+    {
+      "epoch": 0.3572039192477149,
+      "loss": 0.09737985581159592,
+      "loss_ce": 0.0021955338306725025,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 233232200,
+      "step": 1358
+    },
+    {
+      "epoch": 0.357466956007102,
+      "grad_norm": 22.688298388253326,
+      "learning_rate": 5e-06,
+      "loss": 0.1826,
+      "num_input_tokens_seen": 233402524,
+      "step": 1359
+    },
+    {
+      "epoch": 0.357466956007102,
+      "loss": 0.13730812072753906,
+      "loss_ce": 0.000497827713843435,
+      "loss_iou": 0.578125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 233402524,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3577299927664891,
+      "grad_norm": 7.149196234866028,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 233574352,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3577299927664891,
+      "loss": 0.13957476615905762,
+      "loss_ce": 0.002062564715743065,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 233574352,
+      "step": 1360
+    },
+    {
+      "epoch": 0.35799302952587625,
+      "grad_norm": 4.747458347619557,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 233746476,
+      "step": 1361
+    },
+    {
+      "epoch": 0.35799302952587625,
+      "loss": 0.09390648454427719,
+      "loss_ce": 0.001377185108140111,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 233746476,
+      "step": 1361
+    },
+    {
+      "epoch": 0.35825606628526335,
+      "grad_norm": 4.140560580446189,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 233918696,
+      "step": 1362
+    },
+    {
+      "epoch": 0.35825606628526335,
+      "loss": 0.10030673444271088,
+      "loss_ce": 0.0024216112215071917,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 233918696,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3585191030446505,
+      "grad_norm": 5.484628612516693,
+      "learning_rate": 5e-06,
+      "loss": 0.1618,
+      "num_input_tokens_seen": 234090660,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3585191030446505,
+      "loss": 0.1829943060874939,
+      "loss_ce": 0.001201094826683402,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 234090660,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3587821398040376,
+      "grad_norm": 7.875545161330649,
+      "learning_rate": 5e-06,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 234262444,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3587821398040376,
+      "loss": 0.10195118188858032,
+      "loss_ce": 0.00035817097523249686,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 234262444,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3590451765634247,
+      "grad_norm": 5.112293535327179,
+      "learning_rate": 5e-06,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 234434852,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3590451765634247,
+      "loss": 0.12388080358505249,
+      "loss_ce": 0.0015663461526855826,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 234434852,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3593082133228119,
+      "grad_norm": 7.038268029638918,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 234607044,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3593082133228119,
+      "loss": 0.10762982070446014,
+      "loss_ce": 0.0011539864353835583,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 234607044,
+      "step": 1366
+    },
+    {
+      "epoch": 0.359571250082199,
+      "grad_norm": 23.600297531220633,
+      "learning_rate": 5e-06,
+      "loss": 0.1188,
+      "num_input_tokens_seen": 234779280,
+      "step": 1367
+    },
+    {
+      "epoch": 0.359571250082199,
+      "loss": 0.06237838417291641,
+      "loss_ce": 0.0010838248999789357,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 234779280,
+      "step": 1367
+    },
+    {
+      "epoch": 0.35983428684158614,
+      "grad_norm": 3.7489621345618547,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 234951544,
+      "step": 1368
+    },
+    {
+      "epoch": 0.35983428684158614,
+      "loss": 0.09055154025554657,
+      "loss_ce": 0.0044309417717158794,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 234951544,
+      "step": 1368
+    },
+    {
+      "epoch": 0.36009732360097324,
+      "grad_norm": 11.577809346393625,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 235124044,
+      "step": 1369
+    },
+    {
+      "epoch": 0.36009732360097324,
+      "loss": 0.10378709435462952,
+      "loss_ce": 0.0021635598968714476,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 235124044,
+      "step": 1369
+    },
+    {
+      "epoch": 0.36036036036036034,
+      "grad_norm": 12.169579645431945,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 235295776,
+      "step": 1370
+    },
+    {
+      "epoch": 0.36036036036036034,
+      "loss": 0.26028677821159363,
+      "loss_ce": 0.00123831897508353,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 235295776,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3606233971197475,
+      "grad_norm": 6.068163006201439,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 235467876,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3606233971197475,
+      "loss": 0.12217633426189423,
+      "loss_ce": 0.0007163715199567378,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 235467876,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3608864338791346,
+      "grad_norm": 5.0641588944194575,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 235640260,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3608864338791346,
+      "loss": 0.13001395761966705,
+      "loss_ce": 0.003243940882384777,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 235640260,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3611494706385217,
+      "grad_norm": 5.364003970603183,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 235812276,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3611494706385217,
+      "loss": 0.1422898769378662,
+      "loss_ce": 0.002214185893535614,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 235812276,
+      "step": 1373
+    },
+    {
+      "epoch": 0.36141250739790887,
+      "grad_norm": 20.072012728606875,
+      "learning_rate": 5e-06,
+      "loss": 0.1562,
+      "num_input_tokens_seen": 235984384,
+      "step": 1374
+    },
+    {
+      "epoch": 0.36141250739790887,
+      "loss": 0.1751733273267746,
+      "loss_ce": 0.0007043338264338672,
+      "loss_iou": NaN,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 235984384,
+      "step": 1374
+    },
+    {
+      "epoch": 0.36167554415729597,
+      "grad_norm": 9.286393703464922,
+      "learning_rate": 5e-06,
+      "loss": 0.1541,
+      "num_input_tokens_seen": 236156884,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36167554415729597,
+      "loss": 0.21126675605773926,
+      "loss_ce": 0.0010006362572312355,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 236156884,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36193858091668313,
+      "grad_norm": 11.903565521275382,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 236328968,
+      "step": 1376
+    },
+    {
+      "epoch": 0.36193858091668313,
+      "loss": 0.12723658978939056,
+      "loss_ce": 0.0002834574261214584,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 236328968,
+      "step": 1376
+    },
+    {
+      "epoch": 0.36220161767607023,
+      "grad_norm": 16.599820050329747,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 236501060,
+      "step": 1377
+    },
+    {
+      "epoch": 0.36220161767607023,
+      "loss": 0.11763446778059006,
+      "loss_ce": 0.0014235277194529772,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 236501060,
+      "step": 1377
+    },
+    {
+      "epoch": 0.36246465443545733,
+      "grad_norm": 12.84169775199896,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 236673348,
+      "step": 1378
+    },
+    {
+      "epoch": 0.36246465443545733,
+      "loss": 0.1577835977077484,
+      "loss_ce": 0.00424966961145401,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 236673348,
+      "step": 1378
+    },
+    {
+      "epoch": 0.3627276911948445,
+      "grad_norm": 4.530082641105361,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 236845952,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3627276911948445,
+      "loss": 0.16972720623016357,
+      "loss_ce": 0.0046881334856152534,
+      "loss_iou": 0.546875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 236845952,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3629907279542316,
+      "grad_norm": 4.453547029103764,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 237016412,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3629907279542316,
+      "loss": 0.1337898075580597,
+      "loss_ce": 0.0024726560804992914,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 237016412,
+      "step": 1380
+    },
+    {
+      "epoch": 0.36325376471361875,
+      "grad_norm": 14.926786019286814,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 237188648,
+      "step": 1381
+    },
+    {
+      "epoch": 0.36325376471361875,
+      "loss": 0.09095098078250885,
+      "loss_ce": 0.0004968784051015973,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 237188648,
+      "step": 1381
+    },
+    {
+      "epoch": 0.36351680147300586,
+      "grad_norm": 5.283936723228349,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 237360968,
+      "step": 1382
+    },
+    {
+      "epoch": 0.36351680147300586,
+      "loss": 0.10726694762706757,
+      "loss_ce": 0.002011828124523163,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 237360968,
+      "step": 1382
+    },
+    {
+      "epoch": 0.36377983823239296,
+      "grad_norm": 6.9353605673209175,
+      "learning_rate": 5e-06,
+      "loss": 0.1229,
+      "num_input_tokens_seen": 237533240,
+      "step": 1383
+    },
+    {
+      "epoch": 0.36377983823239296,
+      "loss": 0.12320241332054138,
+      "loss_ce": 0.0009337374940514565,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 237533240,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3640428749917801,
+      "grad_norm": 7.872299034496559,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 237705412,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3640428749917801,
+      "loss": 0.10494999587535858,
+      "loss_ce": 0.0007019541808404028,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 237705412,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3643059117511672,
+      "grad_norm": 9.299219065335537,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 237875984,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3643059117511672,
+      "loss": 0.09333762526512146,
+      "loss_ce": 0.0001979749504243955,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 237875984,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3645689485105543,
+      "grad_norm": 4.989443397895941,
+      "learning_rate": 5e-06,
+      "loss": 0.1464,
+      "num_input_tokens_seen": 238048268,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3645689485105543,
+      "loss": 0.14278042316436768,
+      "loss_ce": 0.0023690357338637114,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 238048268,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3648319852699415,
+      "grad_norm": 12.398705819599233,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 238218704,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3648319852699415,
+      "loss": 0.18032394349575043,
+      "loss_ce": 0.000819545122794807,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 238218704,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3650950220293286,
+      "grad_norm": 6.66114485908042,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 238390452,
+      "step": 1388
+    },
+    {
+      "epoch": 0.3650950220293286,
+      "loss": 0.23675096035003662,
+      "loss_ce": 0.0027726897969841957,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 238390452,
+      "step": 1388
+    },
+    {
+      "epoch": 0.36535805878871574,
+      "grad_norm": 9.52863817159708,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 238562708,
+      "step": 1389
+    },
+    {
+      "epoch": 0.36535805878871574,
+      "loss": 0.10956455767154694,
+      "loss_ce": 0.0007693836814723909,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 238562708,
+      "step": 1389
+    },
+    {
+      "epoch": 0.36562109554810285,
+      "grad_norm": 14.33207869070322,
+      "learning_rate": 5e-06,
+      "loss": 0.1554,
+      "num_input_tokens_seen": 238734768,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36562109554810285,
+      "loss": 0.11770792305469513,
+      "loss_ce": 0.00018472480587661266,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 238734768,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36588413230748995,
+      "grad_norm": 6.704869940759653,
+      "learning_rate": 5e-06,
+      "loss": 0.1267,
+      "num_input_tokens_seen": 238906944,
+      "step": 1391
+    },
+    {
+      "epoch": 0.36588413230748995,
+      "loss": 0.12555649876594543,
+      "loss_ce": 0.0015330680180341005,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 238906944,
+      "step": 1391
+    },
+    {
+      "epoch": 0.3661471690668771,
+      "grad_norm": 7.947535402036419,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 239078728,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3661471690668771,
+      "loss": 0.12712188065052032,
+      "loss_ce": 0.0004739244468510151,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 239078728,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3664102058262642,
+      "grad_norm": 21.67277442987669,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 239251004,
+      "step": 1393
+    },
+    {
+      "epoch": 0.3664102058262642,
+      "loss": 0.16826726496219635,
+      "loss_ce": 0.0013361112214624882,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 239251004,
+      "step": 1393
+    },
+    {
+      "epoch": 0.36667324258565137,
+      "grad_norm": 10.958283524277078,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 239418772,
+      "step": 1394
+    },
+    {
+      "epoch": 0.36667324258565137,
+      "loss": 0.05772838741540909,
+      "loss_ce": 0.0009962135227397084,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 239418772,
+      "step": 1394
+    },
+    {
+      "epoch": 0.3669362793450385,
+      "grad_norm": 6.917710662760069,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 239591128,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3669362793450385,
+      "loss": 0.09961553663015366,
+      "loss_ce": 0.001104797120206058,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 239591128,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3671993161044256,
+      "grad_norm": 15.902726176045867,
+      "learning_rate": 5e-06,
+      "loss": 0.1675,
+      "num_input_tokens_seen": 239763344,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3671993161044256,
+      "loss": 0.32926326990127563,
+      "loss_ce": 0.011209084652364254,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 239763344,
+      "step": 1396
+    },
+    {
+      "epoch": 0.36746235286381274,
+      "grad_norm": 13.284413053795395,
+      "learning_rate": 5e-06,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 239933592,
+      "step": 1397
+    },
+    {
+      "epoch": 0.36746235286381274,
+      "loss": 0.11202233284711838,
+      "loss_ce": 0.004936150275170803,
+      "loss_iou": 0.5625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 239933592,
+      "step": 1397
+    },
+    {
+      "epoch": 0.36772538962319984,
+      "grad_norm": 4.642604596899854,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 240106096,
+      "step": 1398
+    },
+    {
+      "epoch": 0.36772538962319984,
+      "loss": 0.12524864077568054,
+      "loss_ce": 0.0025679690297693014,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 240106096,
+      "step": 1398
+    },
+    {
+      "epoch": 0.36798842638258694,
+      "grad_norm": 6.5389674494027,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 240278216,
+      "step": 1399
+    },
+    {
+      "epoch": 0.36798842638258694,
+      "loss": 0.17912393808364868,
+      "loss_ce": 0.002244053641334176,
+      "loss_iou": 0.53125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 240278216,
+      "step": 1399
+    },
+    {
+      "epoch": 0.3682514631419741,
+      "grad_norm": 6.524251878336125,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 240450288,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3682514631419741,
+      "loss": 0.1508733034133911,
+      "loss_ce": 0.00200856477022171,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 240450288,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3685144999013612,
+      "grad_norm": 7.825263953291389,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 240619860,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3685144999013612,
+      "loss": 0.21069373190402985,
+      "loss_ce": 0.002014526631683111,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 240619860,
+      "step": 1401
+    },
+    {
+      "epoch": 0.36877753666074836,
+      "grad_norm": 21.746555075327876,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 240792004,
+      "step": 1402
+    },
+    {
+      "epoch": 0.36877753666074836,
+      "loss": 0.10193300247192383,
+      "loss_ce": 0.003086569719016552,
+      "loss_iou": 0.546875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 240792004,
+      "step": 1402
+    },
+    {
+      "epoch": 0.36904057342013546,
+      "grad_norm": 13.416067582793307,
+      "learning_rate": 5e-06,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 240963952,
+      "step": 1403
+    },
+    {
+      "epoch": 0.36904057342013546,
+      "loss": 0.14882555603981018,
+      "loss_ce": 0.0034703421406447887,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 240963952,
+      "step": 1403
+    },
+    {
+      "epoch": 0.36930361017952257,
+      "grad_norm": 7.0351108134980604,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 241135992,
+      "step": 1404
+    },
+    {
+      "epoch": 0.36930361017952257,
+      "loss": 0.16983582079410553,
+      "loss_ce": 0.004186400678008795,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 241135992,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3695666469389097,
+      "grad_norm": 10.37111011550267,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 241306532,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3695666469389097,
+      "loss": 0.13351476192474365,
+      "loss_ce": 0.0033267755061388016,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 241306532,
+      "step": 1405
+    },
+    {
+      "epoch": 0.36982968369829683,
+      "grad_norm": 4.1471191240598,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 241478648,
+      "step": 1406
+    },
+    {
+      "epoch": 0.36982968369829683,
+      "loss": 0.08705037832260132,
+      "loss_ce": 0.0007466700626537204,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 241478648,
+      "step": 1406
+    },
+    {
+      "epoch": 0.370092720457684,
+      "grad_norm": 4.611801393612808,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 241650796,
+      "step": 1407
+    },
+    {
+      "epoch": 0.370092720457684,
+      "loss": 0.22668591141700745,
+      "loss_ce": 0.0017713564448058605,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 241650796,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3703557572170711,
+      "grad_norm": 8.388794391473104,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 241822848,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3703557572170711,
+      "loss": 0.11565081030130386,
+      "loss_ce": 0.006123221945017576,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 241822848,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3706187939764582,
+      "grad_norm": 4.295439998788119,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 241994956,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3706187939764582,
+      "loss": 0.19000545144081116,
+      "loss_ce": 0.0015594041906297207,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 241994956,
+      "step": 1409
+    },
+    {
+      "epoch": 0.37088183073584535,
+      "grad_norm": 5.005427734613651,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 242167240,
+      "step": 1410
+    },
+    {
+      "epoch": 0.37088183073584535,
+      "loss": 0.10752324759960175,
+      "loss_ce": 0.00040654875920154154,
+      "loss_iou": 0.609375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 242167240,
+      "step": 1410
+    },
+    {
+      "epoch": 0.37114486749523246,
+      "grad_norm": 5.984568060178908,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 242339532,
+      "step": 1411
+    },
+    {
+      "epoch": 0.37114486749523246,
+      "loss": 0.14311346411705017,
+      "loss_ce": 0.0001386186049785465,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 242339532,
+      "step": 1411
+    },
+    {
+      "epoch": 0.37140790425461956,
+      "grad_norm": 43.56689122924667,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 242511708,
+      "step": 1412
+    },
+    {
+      "epoch": 0.37140790425461956,
+      "loss": 0.127020463347435,
+      "loss_ce": 0.0014406184200197458,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 242511708,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3716709410140067,
+      "grad_norm": 8.230300180002782,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 242684000,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3716709410140067,
+      "loss": 0.1775522530078888,
+      "loss_ce": 0.000489264028146863,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 242684000,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3719339777733938,
+      "grad_norm": 7.623956423887346,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 242856400,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3719339777733938,
+      "loss": 0.172633096575737,
+      "loss_ce": 0.0017346586100757122,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 242856400,
+      "step": 1414
+    },
+    {
+      "epoch": 0.372197014532781,
+      "grad_norm": 5.086262256381093,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 243028760,
+      "step": 1415
+    },
+    {
+      "epoch": 0.372197014532781,
+      "loss": 0.11333785206079483,
+      "loss_ce": 0.0045731994323432446,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 243028760,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3724600512921681,
+      "grad_norm": 7.319953499179629,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 243200888,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3724600512921681,
+      "loss": 0.12842552363872528,
+      "loss_ce": 0.0017775753512978554,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 243200888,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3727230880515552,
+      "grad_norm": 7.14806250434601,
+      "learning_rate": 5e-06,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 243373224,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3727230880515552,
+      "loss": 0.17957568168640137,
+      "loss_ce": 0.004038581624627113,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 243373224,
+      "step": 1417
+    },
+    {
+      "epoch": 0.37298612481094234,
+      "grad_norm": 10.549476863868529,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 243545548,
+      "step": 1418
+    },
+    {
+      "epoch": 0.37298612481094234,
+      "loss": 0.14556309580802917,
+      "loss_ce": 0.0017947773449122906,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 243545548,
+      "step": 1418
+    },
+    {
+      "epoch": 0.37324916157032945,
+      "grad_norm": 10.0643806609024,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 243715796,
+      "step": 1419
+    },
+    {
+      "epoch": 0.37324916157032945,
+      "loss": 0.13023152947425842,
+      "loss_ce": 0.0014473494375124574,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 243715796,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3735121983297166,
+      "grad_norm": 6.667777106999036,
+      "learning_rate": 5e-06,
+      "loss": 0.1642,
+      "num_input_tokens_seen": 243886244,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3735121983297166,
+      "loss": 0.21719014644622803,
+      "loss_ce": 0.0024379536043852568,
+      "loss_iou": 0.40625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 243886244,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3737752350891037,
+      "grad_norm": 6.203781395805212,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 244058256,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3737752350891037,
+      "loss": 0.14644312858581543,
+      "loss_ce": 0.003773454576730728,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 244058256,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3740382718484908,
+      "grad_norm": 11.121295779991378,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 244230348,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3740382718484908,
+      "loss": 0.13860073685646057,
+      "loss_ce": 0.004567527212202549,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 244230348,
+      "step": 1422
+    },
+    {
+      "epoch": 0.37430130860787797,
+      "grad_norm": 8.262433746875137,
+      "learning_rate": 5e-06,
+      "loss": 0.1552,
+      "num_input_tokens_seen": 244402756,
+      "step": 1423
+    },
+    {
+      "epoch": 0.37430130860787797,
+      "loss": 0.17389166355133057,
+      "loss_ce": 0.0005212996620684862,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 244402756,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3745643453672651,
+      "grad_norm": 4.574844176213277,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 244575028,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3745643453672651,
+      "loss": 0.2532828748226166,
+      "loss_ce": 0.002657280070707202,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 244575028,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3748273821266522,
+      "grad_norm": 4.771481627209834,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 244745344,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3748273821266522,
+      "loss": 0.1295078694820404,
+      "loss_ce": 0.004813054576516151,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 244745344,
+      "step": 1425
+    },
+    {
+      "epoch": 0.37509041888603933,
+      "grad_norm": 9.163217114560078,
+      "learning_rate": 5e-06,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 244917724,
+      "step": 1426
+    },
+    {
+      "epoch": 0.37509041888603933,
+      "loss": 0.08564335107803345,
+      "loss_ce": 0.004375034011900425,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 244917724,
+      "step": 1426
+    },
+    {
+      "epoch": 0.37535345564542644,
+      "grad_norm": 4.314222759697504,
+      "learning_rate": 5e-06,
+      "loss": 0.1353,
+      "num_input_tokens_seen": 245089960,
+      "step": 1427
+    },
+    {
+      "epoch": 0.37535345564542644,
+      "loss": 0.1637798547744751,
+      "loss_ce": 0.004539141897112131,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 245089960,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3756164924048136,
+      "grad_norm": 4.131046733957264,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 245261980,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3756164924048136,
+      "loss": 0.07748128473758698,
+      "loss_ce": 0.0020113117061555386,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 245261980,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3758795291642007,
+      "grad_norm": 5.689319334005767,
+      "learning_rate": 5e-06,
+      "loss": 0.1536,
+      "num_input_tokens_seen": 245434416,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3758795291642007,
+      "loss": 0.11119981110095978,
+      "loss_ce": 0.0007566966232843697,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 245434416,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3761425659235878,
+      "grad_norm": 20.10795595759354,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 245606636,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3761425659235878,
+      "loss": 0.12385141104459763,
+      "loss_ce": 0.005015961825847626,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 245606636,
+      "step": 1430
+    },
+    {
+      "epoch": 0.37640560268297496,
+      "grad_norm": 5.445619300828039,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 245778556,
+      "step": 1431
+    },
+    {
+      "epoch": 0.37640560268297496,
+      "loss": 0.09182294458150864,
+      "loss_ce": 0.0022538499906659126,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 245778556,
+      "step": 1431
+    },
+    {
+      "epoch": 0.37666863944236206,
+      "grad_norm": 6.5911007721042045,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 245950596,
+      "step": 1432
+    },
+    {
+      "epoch": 0.37666863944236206,
+      "loss": 0.11425422877073288,
+      "loss_ce": 0.002651446033269167,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 245950596,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3769316762017492,
+      "grad_norm": 5.727028044767281,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 246120388,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3769316762017492,
+      "loss": 0.10625661909580231,
+      "loss_ce": 0.0015508129727095366,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 246120388,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3771947129611363,
+      "grad_norm": 11.215881906654301,
+      "learning_rate": 5e-06,
+      "loss": 0.138,
+      "num_input_tokens_seen": 246292408,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3771947129611363,
+      "loss": 0.1351088285446167,
+      "loss_ce": 0.0014418261125683784,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 246292408,
+      "step": 1434
+    },
+    {
+      "epoch": 0.37745774972052343,
+      "grad_norm": 3.9647228124096485,
+      "learning_rate": 5e-06,
+      "loss": 0.1515,
+      "num_input_tokens_seen": 246464396,
+      "step": 1435
+    },
+    {
+      "epoch": 0.37745774972052343,
+      "loss": 0.12623311579227448,
+      "loss_ce": 0.00034810492070391774,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 246464396,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3777207864799106,
+      "grad_norm": 3.981981545629712,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 246636468,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3777207864799106,
+      "loss": 0.08713729679584503,
+      "loss_ce": 0.0009556564036756754,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 246636468,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3779838232392977,
+      "grad_norm": 8.169176906981303,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 246808440,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3779838232392977,
+      "loss": 0.10194505006074905,
+      "loss_ce": 0.005204326473176479,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 246808440,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3782468599986848,
+      "grad_norm": 11.188980116706098,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 246980788,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3782468599986848,
+      "loss": 0.19040054082870483,
+      "loss_ce": 0.0052198669873178005,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 246980788,
+      "step": 1438
+    },
+    {
+      "epoch": 0.37850989675807195,
+      "grad_norm": 7.739931791211071,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 247152940,
+      "step": 1439
+    },
+    {
+      "epoch": 0.37850989675807195,
+      "loss": 0.057150378823280334,
+      "loss_ce": 0.00032664957689121366,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 247152940,
+      "step": 1439
+    },
+    {
+      "epoch": 0.37877293351745905,
+      "grad_norm": 4.659888569440988,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 247321728,
+      "step": 1440
+    },
+    {
+      "epoch": 0.37877293351745905,
+      "loss": 0.23644393682479858,
+      "loss_ce": 0.0026182467117905617,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 247321728,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3790359702768462,
+      "grad_norm": 4.432660256180253,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 247493788,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3790359702768462,
+      "loss": 0.13106387853622437,
+      "loss_ce": 0.0005707137170247734,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 247493788,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3792990070362333,
+      "grad_norm": 14.394067386487718,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 247666148,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3792990070362333,
+      "loss": 0.06981781125068665,
+      "loss_ce": 0.0019467112142592669,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 247666148,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3795620437956204,
+      "grad_norm": 7.3199746686785785,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 247838148,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3795620437956204,
+      "loss": 0.12706537544727325,
+      "loss_ce": 0.001332960557192564,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 247838148,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3798250805550076,
+      "grad_norm": 9.946074142043194,
+      "learning_rate": 5e-06,
+      "loss": 0.1498,
+      "num_input_tokens_seen": 248008104,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3798250805550076,
+      "loss": 0.08655127882957458,
+      "loss_ce": 0.002902593230828643,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 248008104,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3800881173143947,
+      "grad_norm": 6.769456974696547,
+      "learning_rate": 5e-06,
+      "loss": 0.1537,
+      "num_input_tokens_seen": 248178340,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3800881173143947,
+      "loss": 0.12096725404262543,
+      "loss_ce": 0.0002702331403270364,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 248178340,
+      "step": 1445
+    },
+    {
+      "epoch": 0.38035115407378184,
+      "grad_norm": 8.140876700736197,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 248350556,
+      "step": 1446
+    },
+    {
+      "epoch": 0.38035115407378184,
+      "loss": 0.09277988225221634,
+      "loss_ce": 0.004370463080704212,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 248350556,
+      "step": 1446
+    },
+    {
+      "epoch": 0.38061419083316894,
+      "grad_norm": 5.023139574097896,
+      "learning_rate": 5e-06,
+      "loss": 0.1556,
+      "num_input_tokens_seen": 248522684,
+      "step": 1447
+    },
+    {
+      "epoch": 0.38061419083316894,
+      "loss": 0.19670161604881287,
+      "loss_ce": 0.002487761899828911,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 248522684,
+      "step": 1447
+    },
+    {
+      "epoch": 0.38087722759255604,
+      "grad_norm": 10.119019719981985,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 248692816,
+      "step": 1448
+    },
+    {
+      "epoch": 0.38087722759255604,
+      "loss": 0.14071118831634521,
+      "loss_ce": 0.001856213086284697,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 248692816,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3811402643519432,
+      "grad_norm": 4.80565522402478,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 248865220,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3811402643519432,
+      "loss": 0.138558492064476,
+      "loss_ce": 0.005471333395689726,
+      "loss_iou": 0.546875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 248865220,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3814033011113303,
+      "grad_norm": 5.799079918279409,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 249037228,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3814033011113303,
+      "loss": 0.1300484836101532,
+      "loss_ce": 0.0007149941520765424,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 249037228,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3816663378707174,
+      "grad_norm": 5.998961200928992,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 249209488,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3816663378707174,
+      "loss": 0.15324485301971436,
+      "loss_ce": 0.0011757557513192296,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 249209488,
+      "step": 1451
+    },
+    {
+      "epoch": 0.38192937463010457,
+      "grad_norm": 6.571427699631083,
+      "learning_rate": 5e-06,
+      "loss": 0.1472,
+      "num_input_tokens_seen": 249381876,
+      "step": 1452
+    },
+    {
+      "epoch": 0.38192937463010457,
+      "loss": 0.09512484073638916,
+      "loss_ce": 0.001863121404312551,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 249381876,
+      "step": 1452
+    },
+    {
+      "epoch": 0.38219241138949167,
+      "grad_norm": 5.686083518821853,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 249554160,
+      "step": 1453
+    },
+    {
+      "epoch": 0.38219241138949167,
+      "loss": 0.12024913728237152,
+      "loss_ce": 0.002786980476230383,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 249554160,
+      "step": 1453
+    },
+    {
+      "epoch": 0.38245544814887883,
+      "grad_norm": 5.286753977297326,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 249726068,
+      "step": 1454
+    },
+    {
+      "epoch": 0.38245544814887883,
+      "loss": 0.15219104290008545,
+      "loss_ce": 0.001312148873694241,
+      "loss_iou": NaN,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 249726068,
+      "step": 1454
+    },
+    {
+      "epoch": 0.38271848490826593,
+      "grad_norm": 6.1144167801042295,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 249898264,
+      "step": 1455
+    },
+    {
+      "epoch": 0.38271848490826593,
+      "loss": 0.20618750154972076,
+      "loss_ce": 0.001353507163003087,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 249898264,
+      "step": 1455
+    },
+    {
+      "epoch": 0.38298152166765304,
+      "grad_norm": 5.254827116017921,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 250070312,
+      "step": 1456
+    },
+    {
+      "epoch": 0.38298152166765304,
+      "loss": 0.09227914363145828,
+      "loss_ce": 0.00014657669817097485,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 250070312,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3832445584270402,
+      "grad_norm": 6.7921237774310095,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 250242620,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3832445584270402,
+      "loss": 0.16699600219726562,
+      "loss_ce": 0.0013771126978099346,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 250242620,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3835075951864273,
+      "grad_norm": 5.478639252002706,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 250414736,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3835075951864273,
+      "loss": 0.05368629842996597,
+      "loss_ce": 0.00015846370661165565,
+      "loss_iou": 0.734375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 250414736,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3837706319458144,
+      "grad_norm": 5.108471243837392,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 250586900,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3837706319458144,
+      "loss": 0.12694557011127472,
+      "loss_ce": 0.0016709101619198918,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 250586900,
+      "step": 1459
+    },
+    {
+      "epoch": 0.38403366870520156,
+      "grad_norm": 5.673015930654883,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 250757288,
+      "step": 1460
+    },
+    {
+      "epoch": 0.38403366870520156,
+      "loss": 0.13443070650100708,
+      "loss_ce": 0.0023811361752450466,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 250757288,
+      "step": 1460
+    },
+    {
+      "epoch": 0.38429670546458866,
+      "grad_norm": 8.46180134502284,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 250929524,
+      "step": 1461
+    },
+    {
+      "epoch": 0.38429670546458866,
+      "loss": 0.10377843677997589,
+      "loss_ce": 0.0022159344516694546,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 250929524,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3845597422239758,
+      "grad_norm": 19.056127756550847,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 251100012,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3845597422239758,
+      "loss": 0.13582435250282288,
+      "loss_ce": 0.0002042302949121222,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 251100012,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3848227789833629,
+      "grad_norm": 13.96527055803463,
+      "learning_rate": 5e-06,
+      "loss": 0.1363,
+      "num_input_tokens_seen": 251272524,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3848227789833629,
+      "loss": 0.09230969846248627,
+      "loss_ce": 0.0005128234624862671,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 251272524,
+      "step": 1463
+    },
+    {
+      "epoch": 0.38508581574275,
+      "grad_norm": 8.704363897996222,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 251444684,
+      "step": 1464
+    },
+    {
+      "epoch": 0.38508581574275,
+      "loss": 0.06702134013175964,
+      "loss_ce": 0.0022020034957677126,
+      "loss_iou": 0.59375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 251444684,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3853488525021372,
+      "grad_norm": 5.865897344217803,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 251616872,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3853488525021372,
+      "loss": 0.0932367742061615,
+      "loss_ce": 0.0029200036078691483,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 251616872,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3856118892615243,
+      "grad_norm": 7.51772092325757,
+      "learning_rate": 5e-06,
+      "loss": 0.1453,
+      "num_input_tokens_seen": 251787464,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3856118892615243,
+      "loss": 0.08029404282569885,
+      "loss_ce": 0.0014366218820214272,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 251787464,
+      "step": 1466
+    },
+    {
+      "epoch": 0.38587492602091145,
+      "grad_norm": 5.797632938807012,
+      "learning_rate": 5e-06,
+      "loss": 0.1405,
+      "num_input_tokens_seen": 251959420,
+      "step": 1467
+    },
+    {
+      "epoch": 0.38587492602091145,
+      "loss": 0.1458974927663803,
+      "loss_ce": 0.0016409052768722177,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 251959420,
+      "step": 1467
+    },
+    {
+      "epoch": 0.38613796278029855,
+      "grad_norm": 11.05258372992866,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 252130168,
+      "step": 1468
+    },
+    {
+      "epoch": 0.38613796278029855,
+      "loss": 0.13372868299484253,
+      "loss_ce": 0.002472587861120701,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 252130168,
+      "step": 1468
+    },
+    {
+      "epoch": 0.38640099953968565,
+      "grad_norm": 14.608414515106098,
+      "learning_rate": 5e-06,
+      "loss": 0.2148,
+      "num_input_tokens_seen": 252302400,
+      "step": 1469
+    },
+    {
+      "epoch": 0.38640099953968565,
+      "loss": 0.17314574122428894,
+      "loss_ce": 0.0009960737079381943,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 252302400,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3866640362990728,
+      "grad_norm": 5.302717622816423,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 252474472,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3866640362990728,
+      "loss": 0.11253952234983444,
+      "loss_ce": 0.0038511687889695168,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 252474472,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3869270730584599,
+      "grad_norm": 6.326570096654801,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 252646620,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3869270730584599,
+      "loss": 0.12959828972816467,
+      "loss_ce": 0.0016685951268300414,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 252646620,
+      "step": 1471
+    },
+    {
+      "epoch": 0.387190109817847,
+      "grad_norm": 4.47840778836987,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 252818676,
+      "step": 1472
+    },
+    {
+      "epoch": 0.387190109817847,
+      "loss": 0.09814205765724182,
+      "loss_ce": 0.0010656400118023157,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 252818676,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3874531465772342,
+      "grad_norm": 8.330831236084894,
+      "learning_rate": 5e-06,
+      "loss": 0.1907,
+      "num_input_tokens_seen": 252989028,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3874531465772342,
+      "loss": 0.13005918264389038,
+      "loss_ce": 0.0018853526562452316,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 252989028,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3877161833366213,
+      "grad_norm": 5.836741339054395,
+      "learning_rate": 5e-06,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 253160776,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3877161833366213,
+      "loss": 0.16334399580955505,
+      "loss_ce": 0.0019365199841558933,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 253160776,
+      "step": 1474
+    },
+    {
+      "epoch": 0.38797922009600844,
+      "grad_norm": 6.563945205492379,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 253333080,
+      "step": 1475
+    },
+    {
+      "epoch": 0.38797922009600844,
+      "loss": 0.11816126108169556,
+      "loss_ce": 0.0005465176654979587,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 253333080,
+      "step": 1475
+    },
+    {
+      "epoch": 0.38824225685539554,
+      "grad_norm": 5.424235438541153,
+      "learning_rate": 5e-06,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 253504996,
+      "step": 1476
+    },
+    {
+      "epoch": 0.38824225685539554,
+      "loss": 0.2769371271133423,
+      "loss_ce": 0.0038658371195197105,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 253504996,
+      "step": 1476
+    },
+    {
+      "epoch": 0.38850529361478264,
+      "grad_norm": 8.554483113755165,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 253677224,
+      "step": 1477
+    },
+    {
+      "epoch": 0.38850529361478264,
+      "loss": 0.2999889850616455,
+      "loss_ce": 0.0013744828756898642,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 253677224,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3887683303741698,
+      "grad_norm": 9.64593506404694,
+      "learning_rate": 5e-06,
+      "loss": 0.1791,
+      "num_input_tokens_seen": 253849140,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3887683303741698,
+      "loss": 0.10786094516515732,
+      "loss_ce": 0.0013545926194638014,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 253849140,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3890313671335569,
+      "grad_norm": 4.897950380873555,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 254021328,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3890313671335569,
+      "loss": 0.09722624719142914,
+      "loss_ce": 0.0006075926939956844,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 254021328,
+      "step": 1479
+    },
+    {
+      "epoch": 0.38929440389294406,
+      "grad_norm": 15.722256371647902,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 254191780,
+      "step": 1480
+    },
+    {
+      "epoch": 0.38929440389294406,
+      "loss": 0.24498049914836884,
+      "loss_ce": 0.003861122764647007,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 254191780,
+      "step": 1480
+    },
+    {
+      "epoch": 0.38955744065233117,
+      "grad_norm": 8.008352967265257,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 254363968,
+      "step": 1481
+    },
+    {
+      "epoch": 0.38955744065233117,
+      "loss": 0.11913852393627167,
+      "loss_ce": 0.004117771051824093,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 254363968,
+      "step": 1481
+    },
+    {
+      "epoch": 0.38982047741171827,
+      "grad_norm": 49.00606130732605,
+      "learning_rate": 5e-06,
+      "loss": 0.1557,
+      "num_input_tokens_seen": 254534476,
+      "step": 1482
+    },
+    {
+      "epoch": 0.38982047741171827,
+      "loss": 0.10790035128593445,
+      "loss_ce": 0.001271925400942564,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 254534476,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3900835141711054,
+      "grad_norm": 7.1385674525154705,
+      "learning_rate": 5e-06,
+      "loss": 0.1465,
+      "num_input_tokens_seen": 254706676,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3900835141711054,
+      "loss": 0.21054460108280182,
+      "loss_ce": 0.0007057388429529965,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 254706676,
+      "step": 1483
+    },
+    {
+      "epoch": 0.39034655093049253,
+      "grad_norm": 11.652107001526453,
+      "learning_rate": 5e-06,
+      "loss": 0.1735,
+      "num_input_tokens_seen": 254877300,
+      "step": 1484
+    },
+    {
+      "epoch": 0.39034655093049253,
+      "loss": 0.1516016721725464,
+      "loss_ce": 0.0008448172593489289,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 254877300,
+      "step": 1484
+    },
+    {
+      "epoch": 0.39060958768987963,
+      "grad_norm": 6.761059905321281,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 255049560,
+      "step": 1485
+    },
+    {
+      "epoch": 0.39060958768987963,
+      "loss": 0.17154067754745483,
+      "loss_ce": 0.00045912445057183504,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 255049560,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3908726244492668,
+      "grad_norm": 4.6049295798328655,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 255221840,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3908726244492668,
+      "loss": 0.12419994175434113,
+      "loss_ce": 0.0011530672200024128,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 255221840,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3911356612086539,
+      "grad_norm": 9.631901215468224,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 255393804,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3911356612086539,
+      "loss": 0.14977988600730896,
+      "loss_ce": 0.006835547741502523,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 255393804,
+      "step": 1487
+    },
+    {
+      "epoch": 0.39139869796804105,
+      "grad_norm": 6.350539398563022,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 255565956,
+      "step": 1488
+    },
+    {
+      "epoch": 0.39139869796804105,
+      "loss": 0.08242587745189667,
+      "loss_ce": 0.0012491194065660238,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 255565956,
+      "step": 1488
+    },
+    {
+      "epoch": 0.39166173472742816,
+      "grad_norm": 8.23209899989984,
+      "learning_rate": 5e-06,
+      "loss": 0.1259,
+      "num_input_tokens_seen": 255738272,
+      "step": 1489
+    },
+    {
+      "epoch": 0.39166173472742816,
+      "loss": 0.11557637155056,
+      "loss_ce": 0.00025044637732207775,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 255738272,
+      "step": 1489
+    },
+    {
+      "epoch": 0.39192477148681526,
+      "grad_norm": 6.0709262055147555,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 255910628,
+      "step": 1490
+    },
+    {
+      "epoch": 0.39192477148681526,
+      "loss": 0.06516115367412567,
+      "loss_ce": 0.0002197464054916054,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 255910628,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3921878082462024,
+      "grad_norm": 4.913795659722062,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 256079800,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3921878082462024,
+      "loss": 0.12981277704238892,
+      "loss_ce": 0.0006013567326590419,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 256079800,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3924508450055895,
+      "grad_norm": 10.053329401258154,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 256251908,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3924508450055895,
+      "loss": 0.11420565843582153,
+      "loss_ce": 0.0003750904288608581,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 256251908,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3927138817649767,
+      "grad_norm": 9.9668682397801,
+      "learning_rate": 5e-06,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 256423824,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3927138817649767,
+      "loss": 0.1758217215538025,
+      "loss_ce": 0.0006203037919476628,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 256423824,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3929769185243638,
+      "grad_norm": 11.611915023620941,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 256596028,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3929769185243638,
+      "loss": 0.1324601024389267,
+      "loss_ce": 0.0007157221552915871,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 256596028,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3932399552837509,
+      "grad_norm": 8.374710474816515,
+      "learning_rate": 5e-06,
+      "loss": 0.1707,
+      "num_input_tokens_seen": 256764620,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3932399552837509,
+      "loss": 0.14535440504550934,
+      "loss_ce": 0.00076211744453758,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 256764620,
+      "step": 1495
+    },
+    {
+      "epoch": 0.39350299204313804,
+      "grad_norm": 3.7888421024340295,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 256936860,
+      "step": 1496
+    },
+    {
+      "epoch": 0.39350299204313804,
+      "loss": 0.11324183642864227,
+      "loss_ce": 0.0021883677691221237,
+      "loss_iou": 0.75,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 256936860,
+      "step": 1496
+    },
+    {
+      "epoch": 0.39376602880252515,
+      "grad_norm": 6.4208864756609865,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 257108880,
+      "step": 1497
+    },
+    {
+      "epoch": 0.39376602880252515,
+      "loss": 0.14706888794898987,
+      "loss_ce": 0.0031480020843446255,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 257108880,
+      "step": 1497
+    },
+    {
+      "epoch": 0.39402906556191225,
+      "grad_norm": 7.914747299122871,
+      "learning_rate": 5e-06,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 257280748,
+      "step": 1498
+    },
+    {
+      "epoch": 0.39402906556191225,
+      "loss": 0.12499146163463593,
+      "loss_ce": 0.0012732032919302583,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 257280748,
+      "step": 1498
+    },
+    {
+      "epoch": 0.3942921023212994,
+      "grad_norm": 6.479265884809814,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 257452832,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3942921023212994,
+      "loss": 0.15066663920879364,
+      "loss_ce": 0.000764301570598036,
+      "loss_iou": 0.515625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 257452832,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "grad_norm": 30.54873849483752,
+      "learning_rate": 5e-06,
+      "loss": 0.1459,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_websight_new_CIoU": 0.8511096835136414,
+      "eval_websight_new_GIoU": 0.8473265767097473,
+      "eval_websight_new_IoU": 0.8611634373664856,
+      "eval_websight_new_MAE_all": 0.029499702155590057,
+      "eval_websight_new_MAE_h": 0.019038498401641846,
+      "eval_websight_new_MAE_w": 0.04285556077957153,
+      "eval_websight_new_MAE_x": 0.04122600890696049,
+      "eval_websight_new_MAE_y": 0.014878739370033145,
+      "eval_websight_new_NUM_probability": 0.9999458193778992,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.13057953119277954,
+      "eval_websight_new_loss_ce": 1.3184878298488911e-05,
+      "eval_websight_new_loss_iou": 0.335205078125,
+      "eval_websight_new_loss_num": 0.022043228149414062,
+      "eval_websight_new_loss_xval": 0.1102447509765625,
+      "eval_websight_new_runtime": 55.84,
+      "eval_websight_new_samples_per_second": 0.895,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_seeclick_CIoU": 0.6286773383617401,
+      "eval_seeclick_GIoU": 0.627009391784668,
+      "eval_seeclick_IoU": 0.646778017282486,
+      "eval_seeclick_MAE_all": 0.04383368603885174,
+      "eval_seeclick_MAE_h": 0.028551836498081684,
+      "eval_seeclick_MAE_w": 0.05717572197318077,
+      "eval_seeclick_MAE_x": 0.06311946921050549,
+      "eval_seeclick_MAE_y": 0.02648772206157446,
+      "eval_seeclick_NUM_probability": 0.9999659061431885,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.205887109041214,
+      "eval_seeclick_loss_ce": 0.008159147575497627,
+      "eval_seeclick_loss_iou": 0.525634765625,
+      "eval_seeclick_loss_num": 0.0388031005859375,
+      "eval_seeclick_loss_xval": 0.1939697265625,
+      "eval_seeclick_runtime": 77.0722,
+      "eval_seeclick_samples_per_second": 0.558,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_icons_CIoU": 0.8122206926345825,
+      "eval_icons_GIoU": 0.8060621917247772,
+      "eval_icons_IoU": 0.8242988586425781,
+      "eval_icons_MAE_all": 0.029898496810346842,
+      "eval_icons_MAE_h": 0.02304963255301118,
+      "eval_icons_MAE_w": 0.03834102302789688,
+      "eval_icons_MAE_x": 0.035982510074973106,
+      "eval_icons_MAE_y": 0.022220822051167488,
+      "eval_icons_NUM_probability": 0.9998922646045685,
+      "eval_icons_inside_bbox": 0.9288194477558136,
+      "eval_icons_loss": 0.08346885442733765,
+      "eval_icons_loss_ce": 3.6747020203620195e-05,
+      "eval_icons_loss_iou": 0.5107421875,
+      "eval_icons_loss_num": 0.014827728271484375,
+      "eval_icons_loss_xval": 0.0741424560546875,
+      "eval_icons_runtime": 80.4282,
+      "eval_icons_samples_per_second": 0.622,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_screenspot_CIoU": 0.5216498871644338,
+      "eval_screenspot_GIoU": 0.5137112041314443,
+      "eval_screenspot_IoU": 0.5683091680208842,
+      "eval_screenspot_MAE_all": 0.09167192876338959,
+      "eval_screenspot_MAE_h": 0.04900683710972468,
+      "eval_screenspot_MAE_w": 0.15865817666053772,
+      "eval_screenspot_MAE_x": 0.1116797278324763,
+      "eval_screenspot_MAE_y": 0.04734297779699167,
+      "eval_screenspot_NUM_probability": 0.9997473557790121,
+      "eval_screenspot_inside_bbox": 0.8291666706403097,
+      "eval_screenspot_loss": 0.8493114709854126,
+      "eval_screenspot_loss_ce": 0.5014328161875407,
+      "eval_screenspot_loss_iou": 0.4333089192708333,
+      "eval_screenspot_loss_num": 0.06844584147135417,
+      "eval_screenspot_loss_xval": 0.3421223958333333,
+      "eval_screenspot_runtime": 144.4244,
+      "eval_screenspot_samples_per_second": 0.616,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "loss": 0.8482523560523987,
+      "loss_ce": 0.4914408326148987,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.39481817584007367,
+      "grad_norm": 4.597601910727454,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 257797176,
+      "step": 1501
+    },
+    {
+      "epoch": 0.39481817584007367,
+      "loss": 0.11587628722190857,
+      "loss_ce": 0.0010386451613157988,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 257797176,
+      "step": 1501
+    },
+    {
+      "epoch": 0.3950812125994608,
+      "grad_norm": 20.41167402936065,
+      "learning_rate": 5e-06,
+      "loss": 0.1429,
+      "num_input_tokens_seen": 257969472,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3950812125994608,
+      "loss": 0.13194513320922852,
+      "loss_ce": 0.00023126379528548568,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 257969472,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3953442493588479,
+      "grad_norm": 5.303324872084742,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 258141772,
+      "step": 1503
+    },
+    {
+      "epoch": 0.3953442493588479,
+      "loss": 0.26048120856285095,
+      "loss_ce": 0.0004714562091976404,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 258141772,
+      "step": 1503
+    },
+    {
+      "epoch": 0.39560728611823504,
+      "grad_norm": 8.26410044022957,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 258312152,
+      "step": 1504
+    },
+    {
+      "epoch": 0.39560728611823504,
+      "loss": 0.11661653220653534,
+      "loss_ce": 0.0012600838672369719,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 258312152,
+      "step": 1504
+    },
+    {
+      "epoch": 0.39587032287762214,
+      "grad_norm": 5.069526368720703,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 258484140,
+      "step": 1505
+    },
+    {
+      "epoch": 0.39587032287762214,
+      "loss": 0.15614590048789978,
+      "loss_ce": 0.002184713026508689,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 258484140,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3961333596370093,
+      "grad_norm": 8.21868694477306,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 258656432,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3961333596370093,
+      "loss": 0.15648871660232544,
+      "loss_ce": 0.006220155395567417,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 258656432,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3963963963963964,
+      "grad_norm": 4.840531799006556,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 258828268,
+      "step": 1507
+    },
+    {
+      "epoch": 0.3963963963963964,
+      "loss": 0.16232186555862427,
+      "loss_ce": 0.0027149375528097153,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 258828268,
+      "step": 1507
+    },
+    {
+      "epoch": 0.3966594331557835,
+      "grad_norm": 13.037528909207806,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 259000404,
+      "step": 1508
+    },
+    {
+      "epoch": 0.3966594331557835,
+      "loss": 0.14472803473472595,
+      "loss_ce": 0.0009902361780405045,
+      "loss_iou": 0.625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 259000404,
+      "step": 1508
+    },
+    {
+      "epoch": 0.39692246991517066,
+      "grad_norm": 5.085713777784683,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 259172564,
+      "step": 1509
+    },
+    {
+      "epoch": 0.39692246991517066,
+      "loss": 0.07742594182491302,
+      "loss_ce": 0.001498206052929163,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 259172564,
+      "step": 1509
+    },
+    {
+      "epoch": 0.39718550667455776,
+      "grad_norm": 11.735468019604104,
+      "learning_rate": 5e-06,
+      "loss": 0.1383,
+      "num_input_tokens_seen": 259344984,
+      "step": 1510
+    },
+    {
+      "epoch": 0.39718550667455776,
+      "loss": 0.19085130095481873,
+      "loss_ce": 0.006021593697369099,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 259344984,
+      "step": 1510
+    },
+    {
+      "epoch": 0.39744854343394487,
+      "grad_norm": 5.2454273554423505,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 259517348,
+      "step": 1511
+    },
+    {
+      "epoch": 0.39744854343394487,
+      "loss": 0.18470171093940735,
+      "loss_ce": 0.0007722551235929132,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 259517348,
+      "step": 1511
+    },
+    {
+      "epoch": 0.397711580193332,
+      "grad_norm": 5.93272222700996,
+      "learning_rate": 5e-06,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 259689536,
+      "step": 1512
+    },
+    {
+      "epoch": 0.397711580193332,
+      "loss": 0.08698225021362305,
+      "loss_ce": 9.870098438113928e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 259689536,
+      "step": 1512
+    },
+    {
+      "epoch": 0.39797461695271913,
+      "grad_norm": 6.869955142591404,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 259861960,
+      "step": 1513
+    },
+    {
+      "epoch": 0.39797461695271913,
+      "loss": 0.17443957924842834,
+      "loss_ce": 0.0013743957970291376,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 259861960,
+      "step": 1513
+    },
+    {
+      "epoch": 0.3982376537121063,
+      "grad_norm": 5.07838683649017,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 260034008,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3982376537121063,
+      "loss": 0.11854679882526398,
+      "loss_ce": 0.0005353257874958217,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 260034008,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3985006904714934,
+      "grad_norm": 4.287082894022155,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 260206128,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3985006904714934,
+      "loss": 0.13691899180412292,
+      "loss_ce": 0.004594762809574604,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 260206128,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3987637272308805,
+      "grad_norm": 11.905443274772528,
+      "learning_rate": 5e-06,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 260377024,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3987637272308805,
+      "loss": 0.2244867980480194,
+      "loss_ce": 0.004058347083628178,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 260377024,
+      "step": 1516
+    },
+    {
+      "epoch": 0.39902676399026765,
+      "grad_norm": 5.162416897168446,
+      "learning_rate": 5e-06,
+      "loss": 0.1643,
+      "num_input_tokens_seen": 260549044,
+      "step": 1517
+    },
+    {
+      "epoch": 0.39902676399026765,
+      "loss": 0.1554432064294815,
+      "loss_ce": 0.006700540892779827,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 260549044,
+      "step": 1517
+    },
+    {
+      "epoch": 0.39928980074965476,
+      "grad_norm": 10.132521694727306,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 260721092,
+      "step": 1518
+    },
+    {
+      "epoch": 0.39928980074965476,
+      "loss": 0.10422030091285706,
+      "loss_ce": 0.0018338197842240334,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 260721092,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3995528375090419,
+      "grad_norm": 5.444540349408582,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 260893156,
+      "step": 1519
+    },
+    {
+      "epoch": 0.3995528375090419,
+      "loss": 0.151461660861969,
+      "loss_ce": 0.0018339705420657992,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 260893156,
+      "step": 1519
+    },
+    {
+      "epoch": 0.399815874268429,
+      "grad_norm": 11.977186984155779,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 261065364,
+      "step": 1520
+    },
+    {
+      "epoch": 0.399815874268429,
+      "loss": 0.10812121629714966,
+      "loss_ce": 0.0015843516448512673,
+      "loss_iou": 0.53125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 261065364,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4000789110278161,
+      "grad_norm": 18.63106485848457,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 261237376,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4000789110278161,
+      "loss": 0.16819404065608978,
+      "loss_ce": 0.0020105685107409954,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 261237376,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4003419477872033,
+      "grad_norm": 10.102453678806217,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 261409528,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4003419477872033,
+      "loss": 0.16048389673233032,
+      "loss_ce": 0.0018229965353384614,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 261409528,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4006049845465904,
+      "grad_norm": 4.815296030108717,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 261581516,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4006049845465904,
+      "loss": 0.16827590763568878,
+      "loss_ce": 0.0006428433116525412,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 261581516,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4008680213059775,
+      "grad_norm": 7.523079284642639,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 261753732,
+      "step": 1524
+    },
+    {
+      "epoch": 0.4008680213059775,
+      "loss": 0.07389776408672333,
+      "loss_ce": 0.0012354037025943398,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 261753732,
+      "step": 1524
+    },
+    {
+      "epoch": 0.40113105806536464,
+      "grad_norm": 5.8778602893728955,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 261925788,
+      "step": 1525
+    },
+    {
+      "epoch": 0.40113105806536464,
+      "loss": 0.1510400027036667,
+      "loss_ce": 0.00025264715077355504,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 261925788,
+      "step": 1525
+    },
+    {
+      "epoch": 0.40139409482475175,
+      "grad_norm": 5.850697556727792,
+      "learning_rate": 5e-06,
+      "loss": 0.1457,
+      "num_input_tokens_seen": 262097916,
+      "step": 1526
+    },
+    {
+      "epoch": 0.40139409482475175,
+      "loss": 0.1881968379020691,
+      "loss_ce": 0.002436349866911769,
+      "loss_iou": 0.53125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 262097916,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4016571315841389,
+      "grad_norm": 6.537837429507002,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 262269964,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4016571315841389,
+      "loss": 0.09432707726955414,
+      "loss_ce": 0.004178154282271862,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 262269964,
+      "step": 1527
+    },
+    {
+      "epoch": 0.401920168343526,
+      "grad_norm": 16.05135902232025,
+      "learning_rate": 5e-06,
+      "loss": 0.148,
+      "num_input_tokens_seen": 262442240,
+      "step": 1528
+    },
+    {
+      "epoch": 0.401920168343526,
+      "loss": 0.22578378021717072,
+      "loss_ce": 0.0013269821647554636,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 262442240,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4021832051029131,
+      "grad_norm": 23.071196756112307,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 262612560,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4021832051029131,
+      "loss": 0.1546820104122162,
+      "loss_ce": 0.0018499757861718535,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 262612560,
+      "step": 1529
+    },
+    {
+      "epoch": 0.40244624186230027,
+      "grad_norm": 15.456725226696141,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 262784424,
+      "step": 1530
+    },
+    {
+      "epoch": 0.40244624186230027,
+      "loss": 0.09581418335437775,
+      "loss_ce": 0.00661130016669631,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 262784424,
+      "step": 1530
+    },
+    {
+      "epoch": 0.40270927862168737,
+      "grad_norm": 4.753085917564063,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 262956820,
+      "step": 1531
+    },
+    {
+      "epoch": 0.40270927862168737,
+      "loss": 0.17357602715492249,
+      "loss_ce": 0.000693951384164393,
+      "loss_iou": 0.484375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 262956820,
+      "step": 1531
+    },
+    {
+      "epoch": 0.40297231538107453,
+      "grad_norm": 8.23993450342922,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 263129000,
+      "step": 1532
+    },
+    {
+      "epoch": 0.40297231538107453,
+      "loss": 0.104616180062294,
+      "loss_ce": 0.002107633277773857,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 263129000,
+      "step": 1532
+    },
+    {
+      "epoch": 0.40323535214046163,
+      "grad_norm": 5.28557757850573,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 263301064,
+      "step": 1533
+    },
+    {
+      "epoch": 0.40323535214046163,
+      "loss": 0.1304527074098587,
+      "loss_ce": 0.004811838734894991,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 263301064,
+      "step": 1533
+    },
+    {
+      "epoch": 0.40349838889984874,
+      "grad_norm": 6.190146085552892,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 263473624,
+      "step": 1534
+    },
+    {
+      "epoch": 0.40349838889984874,
+      "loss": 0.18483060598373413,
+      "loss_ce": 0.00016873711138032377,
+      "loss_iou": 0.609375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 263473624,
+      "step": 1534
+    },
+    {
+      "epoch": 0.4037614256592359,
+      "grad_norm": 7.04287854016945,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 263645984,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4037614256592359,
+      "loss": 0.2618888020515442,
+      "loss_ce": 0.003191267838701606,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 263645984,
+      "step": 1535
+    },
+    {
+      "epoch": 0.404024462418623,
+      "grad_norm": 12.524958974577965,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 263815796,
+      "step": 1536
+    },
+    {
+      "epoch": 0.404024462418623,
+      "loss": 0.1076509952545166,
+      "loss_ce": 0.0004732571542263031,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 263815796,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4042874991780101,
+      "grad_norm": 11.13747389496862,
+      "learning_rate": 5e-06,
+      "loss": 0.1302,
+      "num_input_tokens_seen": 263985512,
+      "step": 1537
+    },
+    {
+      "epoch": 0.4042874991780101,
+      "loss": 0.12391631305217743,
+      "loss_ce": 0.004134822636842728,
+      "loss_iou": 0.484375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 263985512,
+      "step": 1537
+    },
+    {
+      "epoch": 0.40455053593739726,
+      "grad_norm": 13.524733657823086,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 264157652,
+      "step": 1538
+    },
+    {
+      "epoch": 0.40455053593739726,
+      "loss": 0.1372789740562439,
+      "loss_ce": 0.00034661110839806497,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 264157652,
+      "step": 1538
+    },
+    {
+      "epoch": 0.40481357269678436,
+      "grad_norm": 10.591750756623403,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 264328132,
+      "step": 1539
+    },
+    {
+      "epoch": 0.40481357269678436,
+      "loss": 0.11795895546674728,
+      "loss_ce": 0.0021752638276666403,
+      "loss_iou": 0.484375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 264328132,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4050766094561715,
+      "grad_norm": 13.38043942604694,
+      "learning_rate": 5e-06,
+      "loss": 0.159,
+      "num_input_tokens_seen": 264500388,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4050766094561715,
+      "loss": 0.19694536924362183,
+      "loss_ce": 0.00218218588270247,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 264500388,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4053396462155586,
+      "grad_norm": 4.192366808850045,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 264672276,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4053396462155586,
+      "loss": 0.07517112791538239,
+      "loss_ce": 0.00021995243150740862,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 264672276,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4056026829749457,
+      "grad_norm": 5.297574673677269,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 264844464,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4056026829749457,
+      "loss": 0.12331673502922058,
+      "loss_ce": 0.00359627278521657,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 264844464,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4058657197343329,
+      "grad_norm": 5.716341192152297,
+      "learning_rate": 5e-06,
+      "loss": 0.1333,
+      "num_input_tokens_seen": 265016664,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4058657197343329,
+      "loss": 0.16498282551765442,
+      "loss_ce": 0.001957924338057637,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 265016664,
+      "step": 1543
+    },
+    {
+      "epoch": 0.40612875649372,
+      "grad_norm": 6.6104135503037895,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 265188924,
+      "step": 1544
+    },
+    {
+      "epoch": 0.40612875649372,
+      "loss": 0.15974299609661102,
+      "loss_ce": 0.0041338615119457245,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 265188924,
+      "step": 1544
+    },
+    {
+      "epoch": 0.40639179325310715,
+      "grad_norm": 5.4155340578643365,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 265359312,
+      "step": 1545
+    },
+    {
+      "epoch": 0.40639179325310715,
+      "loss": 0.21989840269088745,
+      "loss_ce": 0.0024301379453390837,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 265359312,
+      "step": 1545
+    },
+    {
+      "epoch": 0.40665483001249425,
+      "grad_norm": 4.753477134622953,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 265531700,
+      "step": 1546
+    },
+    {
+      "epoch": 0.40665483001249425,
+      "loss": 0.05390855669975281,
+      "loss_ce": 0.0012352181365713477,
+      "loss_iou": 0.734375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 265531700,
+      "step": 1546
+    },
+    {
+      "epoch": 0.40691786677188135,
+      "grad_norm": 6.220827805530868,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 265700312,
+      "step": 1547
+    },
+    {
+      "epoch": 0.40691786677188135,
+      "loss": 0.12480812519788742,
+      "loss_ce": 0.0009983108611777425,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 265700312,
+      "step": 1547
+    },
+    {
+      "epoch": 0.4071809035312685,
+      "grad_norm": 6.131445725549716,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 265872224,
+      "step": 1548
+    },
+    {
+      "epoch": 0.4071809035312685,
+      "loss": 0.12791498005390167,
+      "loss_ce": 0.0017858227947726846,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 265872224,
+      "step": 1548
+    },
+    {
+      "epoch": 0.4074439402906556,
+      "grad_norm": 6.062694054094362,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 266044368,
+      "step": 1549
+    },
+    {
+      "epoch": 0.4074439402906556,
+      "loss": 0.09559094905853271,
+      "loss_ce": 0.00013196848158258945,
+      "loss_iou": 0.625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 266044368,
+      "step": 1549
+    },
+    {
+      "epoch": 0.4077069770500427,
+      "grad_norm": 6.379902327089613,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 266216648,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4077069770500427,
+      "loss": 0.26357126235961914,
+      "loss_ce": 0.007589830085635185,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 266216648,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4079700138094299,
+      "grad_norm": 5.16918021793661,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 266387044,
+      "step": 1551
+    },
+    {
+      "epoch": 0.4079700138094299,
+      "loss": 0.0737496167421341,
+      "loss_ce": 0.00215538265183568,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 266387044,
+      "step": 1551
+    },
+    {
+      "epoch": 0.408233050568817,
+      "grad_norm": 36.40974082123275,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 266557400,
+      "step": 1552
+    },
+    {
+      "epoch": 0.408233050568817,
+      "loss": 0.13131964206695557,
+      "loss_ce": 0.0013757951091974974,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 266557400,
+      "step": 1552
+    },
+    {
+      "epoch": 0.40849608732820414,
+      "grad_norm": 4.785820854925791,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 266729564,
+      "step": 1553
+    },
+    {
+      "epoch": 0.40849608732820414,
+      "loss": 0.12293117493391037,
+      "loss_ce": 0.0009524148190394044,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 266729564,
+      "step": 1553
+    },
+    {
+      "epoch": 0.40875912408759124,
+      "grad_norm": 19.710178190555972,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 266901912,
+      "step": 1554
+    },
+    {
+      "epoch": 0.40875912408759124,
+      "loss": 0.10157528519630432,
+      "loss_ce": 0.000501063244882971,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 266901912,
+      "step": 1554
+    },
+    {
+      "epoch": 0.40902216084697834,
+      "grad_norm": 8.53788862222279,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 267074204,
+      "step": 1555
+    },
+    {
+      "epoch": 0.40902216084697834,
+      "loss": 0.18818530440330505,
+      "loss_ce": 0.0011735922889783978,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 267074204,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4092851976063655,
+      "grad_norm": 9.427604183463405,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 267246448,
+      "step": 1556
+    },
+    {
+      "epoch": 0.4092851976063655,
+      "loss": 0.11521363258361816,
+      "loss_ce": 0.0002539134002290666,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 267246448,
+      "step": 1556
+    },
+    {
+      "epoch": 0.4095482343657526,
+      "grad_norm": 6.05814112958034,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 267418524,
+      "step": 1557
+    },
+    {
+      "epoch": 0.4095482343657526,
+      "loss": 0.17979669570922852,
+      "loss_ce": 0.002214906271547079,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 267418524,
+      "step": 1557
+    },
+    {
+      "epoch": 0.40981127112513976,
+      "grad_norm": 5.372569051634822,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 267590932,
+      "step": 1558
+    },
+    {
+      "epoch": 0.40981127112513976,
+      "loss": 0.08670764416456223,
+      "loss_ce": 0.0017467074794694781,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 267590932,
+      "step": 1558
+    },
+    {
+      "epoch": 0.41007430788452687,
+      "grad_norm": 7.2829791835725635,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 267763292,
+      "step": 1559
+    },
+    {
+      "epoch": 0.41007430788452687,
+      "loss": 0.1519029438495636,
+      "loss_ce": 0.002458356786519289,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 267763292,
+      "step": 1559
+    },
+    {
+      "epoch": 0.41033734464391397,
+      "grad_norm": 5.098616866005624,
+      "learning_rate": 5e-06,
+      "loss": 0.1,
+      "num_input_tokens_seen": 267935712,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41033734464391397,
+      "loss": 0.11065052449703217,
+      "loss_ce": 0.0015807072632014751,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 267935712,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41060038140330113,
+      "grad_norm": 10.199812853245577,
+      "learning_rate": 5e-06,
+      "loss": 0.147,
+      "num_input_tokens_seen": 268108048,
+      "step": 1561
+    },
+    {
+      "epoch": 0.41060038140330113,
+      "loss": 0.09520716965198517,
+      "loss_ce": 0.003898570779711008,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 268108048,
+      "step": 1561
+    },
+    {
+      "epoch": 0.41086341816268823,
+      "grad_norm": 40.24459139135473,
+      "learning_rate": 5e-06,
+      "loss": 0.131,
+      "num_input_tokens_seen": 268280092,
+      "step": 1562
+    },
+    {
+      "epoch": 0.41086341816268823,
+      "loss": 0.07629628479480743,
+      "loss_ce": 0.0009788942988961935,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 268280092,
+      "step": 1562
+    },
+    {
+      "epoch": 0.41112645492207534,
+      "grad_norm": 10.392982307740652,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 268450660,
+      "step": 1563
+    },
+    {
+      "epoch": 0.41112645492207534,
+      "loss": 0.122515007853508,
+      "loss_ce": 0.0010550380684435368,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 268450660,
+      "step": 1563
+    },
+    {
+      "epoch": 0.4113894916814625,
+      "grad_norm": 9.96483413417414,
+      "learning_rate": 5e-06,
+      "loss": 0.1875,
+      "num_input_tokens_seen": 268622864,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4113894916814625,
+      "loss": 0.22188733518123627,
+      "loss_ce": 0.002923724940046668,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 268622864,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4116525284408496,
+      "grad_norm": 4.677477301563895,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 268793508,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4116525284408496,
+      "loss": 0.08342467993497849,
+      "loss_ce": 0.00023376270837616175,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 268793508,
+      "step": 1565
+    },
+    {
+      "epoch": 0.41191556520023676,
+      "grad_norm": 5.461599242953511,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 268965604,
+      "step": 1566
+    },
+    {
+      "epoch": 0.41191556520023676,
+      "loss": 0.21633076667785645,
+      "loss_ce": 0.003531699301674962,
+      "loss_iou": 0.71875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 268965604,
+      "step": 1566
+    },
+    {
+      "epoch": 0.41217860195962386,
+      "grad_norm": 4.406500798076577,
+      "learning_rate": 5e-06,
+      "loss": 0.1875,
+      "num_input_tokens_seen": 269137692,
+      "step": 1567
+    },
+    {
+      "epoch": 0.41217860195962386,
+      "loss": 0.1873759627342224,
+      "loss_ce": 0.0010966623667627573,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 269137692,
+      "step": 1567
+    },
+    {
+      "epoch": 0.41244163871901096,
+      "grad_norm": 5.2924295964415595,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 269309988,
+      "step": 1568
+    },
+    {
+      "epoch": 0.41244163871901096,
+      "loss": 0.20892329514026642,
+      "loss_ce": 0.0004729711217805743,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 269309988,
+      "step": 1568
+    },
+    {
+      "epoch": 0.4127046754783981,
+      "grad_norm": 5.459543507128776,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 269482356,
+      "step": 1569
+    },
+    {
+      "epoch": 0.4127046754783981,
+      "loss": 0.18314355611801147,
+      "loss_ce": 0.005287110339850187,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 269482356,
+      "step": 1569
+    },
+    {
+      "epoch": 0.4129677122377852,
+      "grad_norm": 7.266131528232208,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 269654572,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4129677122377852,
+      "loss": 0.1481064409017563,
+      "loss_ce": 0.002049315720796585,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 269654572,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4132307489971724,
+      "grad_norm": 9.868014321011369,
+      "learning_rate": 5e-06,
+      "loss": 0.169,
+      "num_input_tokens_seen": 269826912,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4132307489971724,
+      "loss": 0.21563705801963806,
+      "loss_ce": 0.0007627883460372686,
+      "loss_iou": 0.390625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 269826912,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4134937857565595,
+      "grad_norm": 6.475065616065912,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 269998912,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4134937857565595,
+      "loss": 0.20219947397708893,
+      "loss_ce": 0.0007529358845204115,
+      "loss_iou": 0.609375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 269998912,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4137568225159466,
+      "grad_norm": 5.376660767415141,
+      "learning_rate": 5e-06,
+      "loss": 0.1247,
+      "num_input_tokens_seen": 270171036,
+      "step": 1573
+    },
+    {
+      "epoch": 0.4137568225159466,
+      "loss": 0.2056845724582672,
+      "loss_ce": 0.0031394038815051317,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 270171036,
+      "step": 1573
+    },
+    {
+      "epoch": 0.41401985927533375,
+      "grad_norm": 7.42827277488161,
+      "learning_rate": 5e-06,
+      "loss": 0.1727,
+      "num_input_tokens_seen": 270342784,
+      "step": 1574
+    },
+    {
+      "epoch": 0.41401985927533375,
+      "loss": 0.15383180975914001,
+      "loss_ce": 0.00045046067680232227,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 270342784,
+      "step": 1574
+    },
+    {
+      "epoch": 0.41428289603472085,
+      "grad_norm": 5.130198059184452,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 270514876,
+      "step": 1575
+    },
+    {
+      "epoch": 0.41428289603472085,
+      "loss": 0.17609894275665283,
+      "loss_ce": 0.001187438378110528,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 270514876,
+      "step": 1575
+    },
+    {
+      "epoch": 0.41454593279410795,
+      "grad_norm": 7.421258492917825,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 270687292,
+      "step": 1576
+    },
+    {
+      "epoch": 0.41454593279410795,
+      "loss": 0.07864829152822495,
+      "loss_ce": 0.0015761489048600197,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 270687292,
+      "step": 1576
+    },
+    {
+      "epoch": 0.4148089695534951,
+      "grad_norm": 6.818009590404441,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 270859520,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4148089695534951,
+      "loss": 0.18715041875839233,
+      "loss_ce": 0.0005964583833701909,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 270859520,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4150720063128822,
+      "grad_norm": 16.40610165209138,
+      "learning_rate": 5e-06,
+      "loss": 0.1347,
+      "num_input_tokens_seen": 271029200,
+      "step": 1578
+    },
+    {
+      "epoch": 0.4150720063128822,
+      "loss": 0.15075725317001343,
+      "loss_ce": 0.0008091325289569795,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 271029200,
+      "step": 1578
+    },
+    {
+      "epoch": 0.41533504307226937,
+      "grad_norm": 5.724561524768373,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 271201432,
+      "step": 1579
+    },
+    {
+      "epoch": 0.41533504307226937,
+      "loss": 0.18409447371959686,
+      "loss_ce": 0.0010500368662178516,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 271201432,
+      "step": 1579
+    },
+    {
+      "epoch": 0.4155980798316565,
+      "grad_norm": 5.423586189008012,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 271373604,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4155980798316565,
+      "loss": 0.10582901537418365,
+      "loss_ce": 0.0017030383460223675,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 271373604,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4158611165910436,
+      "grad_norm": 4.639007188262618,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 271544156,
+      "step": 1581
+    },
+    {
+      "epoch": 0.4158611165910436,
+      "loss": 0.08775737881660461,
+      "loss_ce": 0.00032452167943120003,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 271544156,
+      "step": 1581
+    },
+    {
+      "epoch": 0.41612415335043074,
+      "grad_norm": 4.194479063501224,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 271716244,
+      "step": 1582
+    },
+    {
+      "epoch": 0.41612415335043074,
+      "loss": 0.07552362233400345,
+      "loss_ce": 5.3650168410968035e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 271716244,
+      "step": 1582
+    },
+    {
+      "epoch": 0.41638719010981784,
+      "grad_norm": 6.1414517921780325,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 271888460,
+      "step": 1583
+    },
+    {
+      "epoch": 0.41638719010981784,
+      "loss": 0.09023694694042206,
+      "loss_ce": 0.004726693499833345,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 271888460,
+      "step": 1583
+    },
+    {
+      "epoch": 0.416650226869205,
+      "grad_norm": 4.588576414677694,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 272060548,
+      "step": 1584
+    },
+    {
+      "epoch": 0.416650226869205,
+      "loss": 0.06312213093042374,
+      "loss_ce": 0.004131653346121311,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 272060548,
+      "step": 1584
+    },
+    {
+      "epoch": 0.4169132636285921,
+      "grad_norm": 6.028144781636759,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 272232820,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4169132636285921,
+      "loss": 0.0890943706035614,
+      "loss_ce": 0.0011732284910976887,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 272232820,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4171763003879792,
+      "grad_norm": 4.472287089666441,
+      "learning_rate": 5e-06,
+      "loss": 0.1729,
+      "num_input_tokens_seen": 272405268,
+      "step": 1586
+    },
+    {
+      "epoch": 0.4171763003879792,
+      "loss": 0.2705861032009125,
+      "loss_ce": 0.00334367249161005,
+      "loss_iou": 0.5625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 272405268,
+      "step": 1586
+    },
+    {
+      "epoch": 0.41743933714736636,
+      "grad_norm": 3.852608585701711,
+      "learning_rate": 5e-06,
+      "loss": 0.1019,
+      "num_input_tokens_seen": 272577520,
+      "step": 1587
+    },
+    {
+      "epoch": 0.41743933714736636,
+      "loss": 0.1504625827074051,
+      "loss_ce": 0.00028558107442222536,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 272577520,
+      "step": 1587
+    },
+    {
+      "epoch": 0.41770237390675347,
+      "grad_norm": 5.376381690191942,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 272747276,
+      "step": 1588
+    },
+    {
+      "epoch": 0.41770237390675347,
+      "loss": 0.15104855597019196,
+      "loss_ce": 0.005449185613542795,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 272747276,
+      "step": 1588
+    },
+    {
+      "epoch": 0.41796541066614057,
+      "grad_norm": 6.161958958679319,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 272919388,
+      "step": 1589
+    },
+    {
+      "epoch": 0.41796541066614057,
+      "loss": 0.19656141102313995,
+      "loss_ce": 0.0006995859439484775,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 272919388,
+      "step": 1589
+    },
+    {
+      "epoch": 0.4182284474255277,
+      "grad_norm": 9.65888325757511,
+      "learning_rate": 5e-06,
+      "loss": 0.1406,
+      "num_input_tokens_seen": 273091656,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4182284474255277,
+      "loss": 0.21912047266960144,
+      "loss_ce": 0.0045361267402768135,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 273091656,
+      "step": 1590
+    },
+    {
+      "epoch": 0.41849148418491483,
+      "grad_norm": 8.813988618374006,
+      "learning_rate": 5e-06,
+      "loss": 0.1735,
+      "num_input_tokens_seen": 273263768,
+      "step": 1591
+    },
+    {
+      "epoch": 0.41849148418491483,
+      "loss": 0.23192375898361206,
+      "loss_ce": 0.00888605136424303,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 273263768,
+      "step": 1591
+    },
+    {
+      "epoch": 0.418754520944302,
+      "grad_norm": 7.699871319865347,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 273435676,
+      "step": 1592
+    },
+    {
+      "epoch": 0.418754520944302,
+      "loss": 0.10040029883384705,
+      "loss_ce": 0.0014928288292139769,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 273435676,
+      "step": 1592
+    },
+    {
+      "epoch": 0.4190175577036891,
+      "grad_norm": 5.319980033934329,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 273607832,
+      "step": 1593
+    },
+    {
+      "epoch": 0.4190175577036891,
+      "loss": 0.149379163980484,
+      "loss_ce": 0.004207056015729904,
+      "loss_iou": 0.59375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 273607832,
+      "step": 1593
+    },
+    {
+      "epoch": 0.4192805944630762,
+      "grad_norm": 9.352989040804356,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 273778508,
+      "step": 1594
+    },
+    {
+      "epoch": 0.4192805944630762,
+      "loss": 0.12792231142520905,
+      "loss_ce": 0.0009081543539650738,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 273778508,
+      "step": 1594
+    },
+    {
+      "epoch": 0.41954363122246335,
+      "grad_norm": 10.015857880890572,
+      "learning_rate": 5e-06,
+      "loss": 0.1659,
+      "num_input_tokens_seen": 273950492,
+      "step": 1595
+    },
+    {
+      "epoch": 0.41954363122246335,
+      "loss": 0.1797780990600586,
+      "loss_ce": 0.004515648819506168,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 273950492,
+      "step": 1595
+    },
+    {
+      "epoch": 0.41980666798185046,
+      "grad_norm": 5.437610936264711,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 274122788,
+      "step": 1596
+    },
+    {
+      "epoch": 0.41980666798185046,
+      "loss": 0.18884079158306122,
+      "loss_ce": 0.0007914789603091776,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 274122788,
+      "step": 1596
+    },
+    {
+      "epoch": 0.4200697047412376,
+      "grad_norm": 7.822775629663786,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 274294960,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4200697047412376,
+      "loss": 0.10433492064476013,
+      "loss_ce": 0.000544635346159339,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 274294960,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4203327415006247,
+      "grad_norm": 5.742663276174034,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 274466828,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4203327415006247,
+      "loss": 0.21367287635803223,
+      "loss_ce": 0.0007822535699233413,
+      "loss_iou": 0.546875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 274466828,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4205957782600118,
+      "grad_norm": 7.632689703873666,
+      "learning_rate": 5e-06,
+      "loss": 0.1806,
+      "num_input_tokens_seen": 274639136,
+      "step": 1599
+    },
+    {
+      "epoch": 0.4205957782600118,
+      "loss": 0.15357787907123566,
+      "loss_ce": 0.0009594644652679563,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 274639136,
+      "step": 1599
+    },
+    {
+      "epoch": 0.420858815019399,
+      "grad_norm": 6.551977275944205,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 274811300,
+      "step": 1600
+    },
+    {
+      "epoch": 0.420858815019399,
+      "loss": 0.10377545654773712,
+      "loss_ce": 0.005798771977424622,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 274811300,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4211218517787861,
+      "grad_norm": 3.634688713329233,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 274983232,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4211218517787861,
+      "loss": 0.0826454609632492,
+      "loss_ce": 0.0008583518210798502,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 274983232,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4213848885381732,
+      "grad_norm": 6.513487522029516,
+      "learning_rate": 5e-06,
+      "loss": 0.1857,
+      "num_input_tokens_seen": 275155252,
+      "step": 1602
+    },
+    {
+      "epoch": 0.4213848885381732,
+      "loss": 0.22003334760665894,
+      "loss_ce": 0.0016495751915499568,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 275155252,
+      "step": 1602
+    },
+    {
+      "epoch": 0.42164792529756034,
+      "grad_norm": 7.844073689048222,
+      "learning_rate": 5e-06,
+      "loss": 0.164,
+      "num_input_tokens_seen": 275327452,
+      "step": 1603
+    },
+    {
+      "epoch": 0.42164792529756034,
+      "loss": 0.21324840188026428,
+      "loss_ce": 0.001090196194127202,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 275327452,
+      "step": 1603
+    },
+    {
+      "epoch": 0.42191096205694745,
+      "grad_norm": 5.702970135636054,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 275499556,
+      "step": 1604
+    },
+    {
+      "epoch": 0.42191096205694745,
+      "loss": 0.12330840528011322,
+      "loss_ce": 0.00029205146711319685,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 275499556,
+      "step": 1604
+    },
+    {
+      "epoch": 0.4221739988163346,
+      "grad_norm": 7.295315362378517,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 275671424,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4221739988163346,
+      "loss": 0.09813028573989868,
+      "loss_ce": 0.0004130033776164055,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 275671424,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4224370355757217,
+      "grad_norm": 9.326948217180478,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 275843716,
+      "step": 1606
+    },
+    {
+      "epoch": 0.4224370355757217,
+      "loss": 0.08426269888877869,
+      "loss_ce": 0.00183471804484725,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 275843716,
+      "step": 1606
+    },
+    {
+      "epoch": 0.4227000723351088,
+      "grad_norm": 5.932370446925655,
+      "learning_rate": 5e-06,
+      "loss": 0.1547,
+      "num_input_tokens_seen": 276016052,
+      "step": 1607
+    },
+    {
+      "epoch": 0.4227000723351088,
+      "loss": 0.1150435283780098,
+      "loss_ce": 0.0007857157033868134,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 276016052,
+      "step": 1607
+    },
+    {
+      "epoch": 0.42296310909449597,
+      "grad_norm": 14.783104322463815,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 276188232,
+      "step": 1608
+    },
+    {
+      "epoch": 0.42296310909449597,
+      "loss": 0.2393971085548401,
+      "loss_ce": 0.005510389804840088,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 276188232,
+      "step": 1608
+    },
+    {
+      "epoch": 0.4232261458538831,
+      "grad_norm": 37.72538153737276,
+      "learning_rate": 5e-06,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 276360224,
+      "step": 1609
+    },
+    {
+      "epoch": 0.4232261458538831,
+      "loss": 0.10361947864294052,
+      "loss_ce": 0.001171966316178441,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 276360224,
+      "step": 1609
+    },
+    {
+      "epoch": 0.42348918261327023,
+      "grad_norm": 5.929115604459541,
+      "learning_rate": 5e-06,
+      "loss": 0.185,
+      "num_input_tokens_seen": 276532388,
+      "step": 1610
+    },
+    {
+      "epoch": 0.42348918261327023,
+      "loss": 0.2490251511335373,
+      "loss_ce": 0.001466565066948533,
+      "loss_iou": 0.515625,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 276532388,
+      "step": 1610
+    },
+    {
+      "epoch": 0.42375221937265733,
+      "grad_norm": 4.065855918819384,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 276704452,
+      "step": 1611
+    },
+    {
+      "epoch": 0.42375221937265733,
+      "loss": 0.08056612312793732,
+      "loss_ce": 0.0004574810154736042,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 276704452,
+      "step": 1611
+    },
+    {
+      "epoch": 0.42401525613204444,
+      "grad_norm": 5.144248302598616,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 276876660,
+      "step": 1612
+    },
+    {
+      "epoch": 0.42401525613204444,
+      "loss": 0.1256069540977478,
+      "loss_ce": 0.003719748929142952,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 276876660,
+      "step": 1612
+    },
+    {
+      "epoch": 0.4242782928914316,
+      "grad_norm": 6.484383508592383,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 277048524,
+      "step": 1613
+    },
+    {
+      "epoch": 0.4242782928914316,
+      "loss": 0.08503228425979614,
+      "loss_ce": 0.0004375518183223903,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 277048524,
+      "step": 1613
+    },
+    {
+      "epoch": 0.4245413296508187,
+      "grad_norm": 4.995173925299756,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 277220588,
+      "step": 1614
+    },
+    {
+      "epoch": 0.4245413296508187,
+      "loss": 0.17099690437316895,
+      "loss_ce": 0.0004646642482839525,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 277220588,
+      "step": 1614
+    },
+    {
+      "epoch": 0.4248043664102058,
+      "grad_norm": 8.497364636941706,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 277392824,
+      "step": 1615
+    },
+    {
+      "epoch": 0.4248043664102058,
+      "loss": 0.13706764578819275,
+      "loss_ce": 0.0016916776075959206,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 277392824,
+      "step": 1615
+    },
+    {
+      "epoch": 0.42506740316959296,
+      "grad_norm": 6.760370062216128,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 277565104,
+      "step": 1616
+    },
+    {
+      "epoch": 0.42506740316959296,
+      "loss": 0.10927695780992508,
+      "loss_ce": 0.00048179191071540117,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 277565104,
+      "step": 1616
+    },
+    {
+      "epoch": 0.42533043992898006,
+      "grad_norm": 5.035401913498838,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 277737224,
+      "step": 1617
+    },
+    {
+      "epoch": 0.42533043992898006,
+      "loss": 0.13474591076374054,
+      "loss_ce": 0.002299622166901827,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 277737224,
+      "step": 1617
+    },
+    {
+      "epoch": 0.4255934766883672,
+      "grad_norm": 6.777922664584483,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 277907488,
+      "step": 1618
+    },
+    {
+      "epoch": 0.4255934766883672,
+      "loss": 0.11205422878265381,
+      "loss_ce": 0.0009702442912384868,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 277907488,
+      "step": 1618
+    },
+    {
+      "epoch": 0.4258565134477543,
+      "grad_norm": 7.0419776595399695,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 278079908,
+      "step": 1619
+    },
+    {
+      "epoch": 0.4258565134477543,
+      "loss": 0.1803514063358307,
+      "loss_ce": 0.0011216606944799423,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 278079908,
+      "step": 1619
+    },
+    {
+      "epoch": 0.42611955020714143,
+      "grad_norm": 4.869531542128126,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 278252060,
+      "step": 1620
+    },
+    {
+      "epoch": 0.42611955020714143,
+      "loss": 0.17065666615962982,
+      "loss_ce": 0.0023522234987467527,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 278252060,
+      "step": 1620
+    },
+    {
+      "epoch": 0.4263825869665286,
+      "grad_norm": 15.025432063122246,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 278424152,
+      "step": 1621
+    },
+    {
+      "epoch": 0.4263825869665286,
+      "loss": 0.16370511054992676,
+      "loss_ce": 0.0011379658244550228,
+      "loss_iou": 0.53125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 278424152,
+      "step": 1621
+    },
+    {
+      "epoch": 0.4266456237259157,
+      "grad_norm": 7.359642440972258,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 278596208,
+      "step": 1622
+    },
+    {
+      "epoch": 0.4266456237259157,
+      "loss": 0.18550482392311096,
+      "loss_ce": 0.0022162585519254208,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 278596208,
+      "step": 1622
+    },
+    {
+      "epoch": 0.42690866048530285,
+      "grad_norm": 3.859276017343743,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 278768348,
+      "step": 1623
+    },
+    {
+      "epoch": 0.42690866048530285,
+      "loss": 0.08928422629833221,
+      "loss_ce": 0.003163617569953203,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 278768348,
+      "step": 1623
+    },
+    {
+      "epoch": 0.42717169724468995,
+      "grad_norm": 3.491408872838399,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 278940432,
+      "step": 1624
+    },
+    {
+      "epoch": 0.42717169724468995,
+      "loss": 0.10469355434179306,
+      "loss_ce": 0.004763749893754721,
+      "loss_iou": 0.609375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 278940432,
+      "step": 1624
+    },
+    {
+      "epoch": 0.42743473400407705,
+      "grad_norm": 8.925089197668054,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 279112480,
+      "step": 1625
+    },
+    {
+      "epoch": 0.42743473400407705,
+      "loss": 0.15482491254806519,
+      "loss_ce": 0.0005127866170369089,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 279112480,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4276977707634642,
+      "grad_norm": 5.017252116965269,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 279284544,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4276977707634642,
+      "loss": 0.11934304237365723,
+      "loss_ce": 0.0036509071942418814,
+      "loss_iou": 0.80859375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 279284544,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4279608075228513,
+      "grad_norm": 11.28367561887442,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 279456684,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4279608075228513,
+      "loss": 0.08618146926164627,
+      "loss_ce": 0.0002134529349859804,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 279456684,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4282238442822384,
+      "grad_norm": 14.334013868271164,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 279629144,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4282238442822384,
+      "loss": 0.21622659265995026,
+      "loss_ce": 0.0035800987388938665,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 279629144,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4284868810416256,
+      "grad_norm": 5.177758249324022,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 279800920,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4284868810416256,
+      "loss": 0.11388795077800751,
+      "loss_ce": 0.0024377545341849327,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 279800920,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4287499178010127,
+      "grad_norm": 5.002716354322989,
+      "learning_rate": 5e-06,
+      "loss": 0.16,
+      "num_input_tokens_seen": 279973164,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4287499178010127,
+      "loss": 0.11221545934677124,
+      "loss_ce": 0.0021690744906663895,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 279973164,
+      "step": 1630
+    },
+    {
+      "epoch": 0.42901295456039984,
+      "grad_norm": 11.888987422058314,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 280144932,
+      "step": 1631
+    },
+    {
+      "epoch": 0.42901295456039984,
+      "loss": 0.10366816818714142,
+      "loss_ce": 0.0032348139211535454,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 280144932,
+      "step": 1631
+    },
+    {
+      "epoch": 0.42927599131978694,
+      "grad_norm": 5.9683348593169105,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 280316892,
+      "step": 1632
+    },
+    {
+      "epoch": 0.42927599131978694,
+      "loss": 0.1221093013882637,
+      "loss_ce": 0.0030907560139894485,
+      "loss_iou": 0.734375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 280316892,
+      "step": 1632
+    },
+    {
+      "epoch": 0.42953902807917405,
+      "grad_norm": 4.306403695098148,
+      "learning_rate": 5e-06,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 280488680,
+      "step": 1633
+    },
+    {
+      "epoch": 0.42953902807917405,
+      "loss": 0.16619864106178284,
+      "loss_ce": 0.0007323222234845161,
+      "loss_iou": 0.421875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 280488680,
+      "step": 1633
+    },
+    {
+      "epoch": 0.4298020648385612,
+      "grad_norm": 6.542037684380333,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 280658812,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4298020648385612,
+      "loss": 0.1335218995809555,
+      "loss_ce": 0.0008009535376913846,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 280658812,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4300651015979483,
+      "grad_norm": 5.780919966084494,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 280829020,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4300651015979483,
+      "loss": 0.13609978556632996,
+      "loss_ce": 0.0019139924552291632,
+      "loss_iou": 0.375,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 280829020,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4303281383573354,
+      "grad_norm": 4.625756427735958,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 280999460,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4303281383573354,
+      "loss": 0.05524425953626633,
+      "loss_ce": 0.0004957281635142863,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 280999460,
+      "step": 1636
+    },
+    {
+      "epoch": 0.43059117511672257,
+      "grad_norm": 4.597155715023034,
+      "learning_rate": 5e-06,
+      "loss": 0.1455,
+      "num_input_tokens_seen": 281169544,
+      "step": 1637
+    },
+    {
+      "epoch": 0.43059117511672257,
+      "loss": 0.09542928636074066,
+      "loss_ce": 0.002106534782797098,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 281169544,
+      "step": 1637
+    },
+    {
+      "epoch": 0.43085421187610967,
+      "grad_norm": 6.599089127666852,
+      "learning_rate": 5e-06,
+      "loss": 0.1655,
+      "num_input_tokens_seen": 281341764,
+      "step": 1638
+    },
+    {
+      "epoch": 0.43085421187610967,
+      "loss": 0.15369150042533875,
+      "loss_ce": 0.0017444868572056293,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 281341764,
+      "step": 1638
+    },
+    {
+      "epoch": 0.43111724863549683,
+      "grad_norm": 10.985722993191203,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 281514480,
+      "step": 1639
+    },
+    {
+      "epoch": 0.43111724863549683,
+      "loss": 0.12693974375724792,
+      "loss_ce": 0.004442187491804361,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 281514480,
+      "step": 1639
+    },
+    {
+      "epoch": 0.43138028539488393,
+      "grad_norm": 5.112052305423048,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 281686636,
+      "step": 1640
+    },
+    {
+      "epoch": 0.43138028539488393,
+      "loss": 0.13622896373271942,
+      "loss_ce": 0.001615930232219398,
+      "loss_iou": 0.6875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 281686636,
+      "step": 1640
+    },
+    {
+      "epoch": 0.43164332215427104,
+      "grad_norm": 4.445172786952326,
+      "learning_rate": 5e-06,
+      "loss": 0.1298,
+      "num_input_tokens_seen": 281859068,
+      "step": 1641
+    },
+    {
+      "epoch": 0.43164332215427104,
+      "loss": 0.16077426075935364,
+      "loss_ce": 0.0008926668670028448,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 281859068,
+      "step": 1641
+    },
+    {
+      "epoch": 0.4319063589136582,
+      "grad_norm": 18.643962143254182,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 282031296,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4319063589136582,
+      "loss": 0.11041103303432465,
+      "loss_ce": 0.00216518621891737,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 282031296,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4321693956730453,
+      "grad_norm": 14.94723205702143,
+      "learning_rate": 5e-06,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 282203456,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4321693956730453,
+      "loss": 0.10057017207145691,
+      "loss_ce": 0.002517198445275426,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 282203456,
+      "step": 1643
+    },
+    {
+      "epoch": 0.43243243243243246,
+      "grad_norm": 3.591864332881924,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 282374020,
+      "step": 1644
+    },
+    {
+      "epoch": 0.43243243243243246,
+      "loss": 0.08621848374605179,
+      "loss_ce": 0.00044883223017677665,
+      "loss_iou": 0.625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 282374020,
+      "step": 1644
+    },
+    {
+      "epoch": 0.43269546919181956,
+      "grad_norm": 6.406579991141497,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 282544200,
+      "step": 1645
+    },
+    {
+      "epoch": 0.43269546919181956,
+      "loss": 0.10012087225914001,
+      "loss_ce": 0.001396512147039175,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 282544200,
+      "step": 1645
+    },
+    {
+      "epoch": 0.43295850595120666,
+      "grad_norm": 6.633263160453237,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 282716276,
+      "step": 1646
+    },
+    {
+      "epoch": 0.43295850595120666,
+      "loss": 0.09250755608081818,
+      "loss_ce": 0.0007411979604512453,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 282716276,
+      "step": 1646
+    },
+    {
+      "epoch": 0.4332215427105938,
+      "grad_norm": 11.88641544130966,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 282888756,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4332215427105938,
+      "loss": 0.11388491094112396,
+      "loss_ce": 0.0009088441729545593,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 282888756,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4334845794699809,
+      "grad_norm": 5.478375817282589,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 283060788,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4334845794699809,
+      "loss": 0.08237907290458679,
+      "loss_ce": 0.003521653823554516,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 283060788,
+      "step": 1648
+    },
+    {
+      "epoch": 0.433747616229368,
+      "grad_norm": 7.7141358569237015,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 283232744,
+      "step": 1649
+    },
+    {
+      "epoch": 0.433747616229368,
+      "loss": 0.16650542616844177,
+      "loss_ce": 0.001252750400453806,
+      "loss_iou": 0.515625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 283232744,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4340106529887552,
+      "grad_norm": 4.902197150094181,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 283404860,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4340106529887552,
+      "loss": 0.11392060667276382,
+      "loss_ce": 0.0009140149923041463,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 283404860,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4342736897481423,
+      "grad_norm": 6.611385564549315,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 283577340,
+      "step": 1651
+    },
+    {
+      "epoch": 0.4342736897481423,
+      "loss": 0.14234262704849243,
+      "loss_ce": 0.0011072808410972357,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 283577340,
+      "step": 1651
+    },
+    {
+      "epoch": 0.43453672650752945,
+      "grad_norm": 5.270564238834203,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 283747872,
+      "step": 1652
+    },
+    {
+      "epoch": 0.43453672650752945,
+      "loss": 0.08932007849216461,
+      "loss_ce": 0.0014904842246323824,
+      "loss_iou": 0.515625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 283747872,
+      "step": 1652
+    },
+    {
+      "epoch": 0.43479976326691655,
+      "grad_norm": 5.150524639297202,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 283920212,
+      "step": 1653
+    },
+    {
+      "epoch": 0.43479976326691655,
+      "loss": 0.08878134936094284,
+      "loss_ce": 0.00037192486342974007,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 283920212,
+      "step": 1653
+    },
+    {
+      "epoch": 0.43506280002630365,
+      "grad_norm": 5.389390494681365,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 284090700,
+      "step": 1654
+    },
+    {
+      "epoch": 0.43506280002630365,
+      "loss": 0.09808811545372009,
+      "loss_ce": 0.0017746419180184603,
+      "loss_iou": 0.59375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 284090700,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4353258367856908,
+      "grad_norm": 5.2829708598355625,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 284262932,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4353258367856908,
+      "loss": 0.24406926333904266,
+      "loss_ce": 0.0009052163222804666,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 284262932,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4355888735450779,
+      "grad_norm": 4.415055027745429,
+      "learning_rate": 5e-06,
+      "loss": 0.1467,
+      "num_input_tokens_seen": 284433136,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4355888735450779,
+      "loss": 0.14210422337055206,
+      "loss_ce": 0.0005942133138887584,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 284433136,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4358519103044651,
+      "grad_norm": 11.10469204899232,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 284604980,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4358519103044651,
+      "loss": 0.06616829335689545,
+      "loss_ce": 0.0029969080351293087,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 284604980,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4361149470638522,
+      "grad_norm": 37.454091592129345,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 284777024,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4361149470638522,
+      "loss": 0.09428656101226807,
+      "loss_ce": 0.00023138479446060956,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 284777024,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4363779838232393,
+      "grad_norm": 6.943748076830087,
+      "learning_rate": 5e-06,
+      "loss": 0.1641,
+      "num_input_tokens_seen": 284948692,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4363779838232393,
+      "loss": 0.12960584461688995,
+      "loss_ce": 0.0008827036363072693,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 284948692,
+      "step": 1659
+    },
+    {
+      "epoch": 0.43664102058262644,
+      "grad_norm": 6.322360699878544,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 285120852,
+      "step": 1660
+    },
+    {
+      "epoch": 0.43664102058262644,
+      "loss": 0.08696160465478897,
+      "loss_ce": 0.0016649758908897638,
+      "loss_iou": 0.609375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 285120852,
+      "step": 1660
+    },
+    {
+      "epoch": 0.43690405734201354,
+      "grad_norm": 4.9857397500362,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 285293104,
+      "step": 1661
+    },
+    {
+      "epoch": 0.43690405734201354,
+      "loss": 0.1620713770389557,
+      "loss_ce": 0.0027085873298346996,
+      "loss_iou": 0.359375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 285293104,
+      "step": 1661
+    },
+    {
+      "epoch": 0.43716709410140064,
+      "grad_norm": 6.167930639257179,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 285465116,
+      "step": 1662
+    },
+    {
+      "epoch": 0.43716709410140064,
+      "loss": 0.13285255432128906,
+      "loss_ce": 0.002481454983353615,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 285465116,
+      "step": 1662
+    },
+    {
+      "epoch": 0.4374301308607878,
+      "grad_norm": 4.796617637287243,
+      "learning_rate": 5e-06,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 285637712,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4374301308607878,
+      "loss": 0.0941682979464531,
+      "loss_ce": 0.0005708856624551117,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 285637712,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4376931676201749,
+      "grad_norm": 6.365145441420231,
+      "learning_rate": 5e-06,
+      "loss": 0.1524,
+      "num_input_tokens_seen": 285808140,
+      "step": 1664
+    },
+    {
+      "epoch": 0.4376931676201749,
+      "loss": 0.08691577613353729,
+      "loss_ce": 0.0016801799647510052,
+      "loss_iou": 0.703125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 285808140,
+      "step": 1664
+    },
+    {
+      "epoch": 0.43795620437956206,
+      "grad_norm": 9.14556589502857,
+      "learning_rate": 5e-06,
+      "loss": 0.1419,
+      "num_input_tokens_seen": 285980020,
+      "step": 1665
+    },
+    {
+      "epoch": 0.43795620437956206,
+      "loss": 0.11803478002548218,
+      "loss_ce": 0.0006641793297603726,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 285980020,
+      "step": 1665
+    },
+    {
+      "epoch": 0.43821924113894917,
+      "grad_norm": 8.36736614358982,
+      "learning_rate": 5e-06,
+      "loss": 0.1476,
+      "num_input_tokens_seen": 286152224,
+      "step": 1666
+    },
+    {
+      "epoch": 0.43821924113894917,
+      "loss": 0.17869716882705688,
+      "loss_ce": 0.002091944683343172,
+      "loss_iou": 0.578125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 286152224,
+      "step": 1666
+    },
+    {
+      "epoch": 0.43848227789833627,
+      "grad_norm": 5.977656991369799,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 286324392,
+      "step": 1667
+    },
+    {
+      "epoch": 0.43848227789833627,
+      "loss": 0.07165973633527756,
+      "loss_ce": 0.0008894691127352417,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 286324392,
+      "step": 1667
+    },
+    {
+      "epoch": 0.43874531465772343,
+      "grad_norm": 4.564829435087821,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 286496724,
+      "step": 1668
+    },
+    {
+      "epoch": 0.43874531465772343,
+      "loss": 0.10358568280935287,
+      "loss_ce": 0.0014128325274214149,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 286496724,
+      "step": 1668
+    },
+    {
+      "epoch": 0.43900835141711053,
+      "grad_norm": 5.942642735170272,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 286669072,
+      "step": 1669
+    },
+    {
+      "epoch": 0.43900835141711053,
+      "loss": 0.15151304006576538,
+      "loss_ce": 0.00020689686061814427,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 286669072,
+      "step": 1669
+    },
+    {
+      "epoch": 0.4392713881764977,
+      "grad_norm": 10.62471442505103,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 286839656,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4392713881764977,
+      "loss": 0.07143907248973846,
+      "loss_ce": 0.0015233027515932918,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 286839656,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4395344249358848,
+      "grad_norm": 10.099626860743896,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 287009348,
+      "step": 1671
+    },
+    {
+      "epoch": 0.4395344249358848,
+      "loss": 0.07566662132740021,
+      "loss_ce": 0.0009290680172853172,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 287009348,
+      "step": 1671
+    },
+    {
+      "epoch": 0.4397974616952719,
+      "grad_norm": 5.653901499064607,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 287181772,
+      "step": 1672
+    },
+    {
+      "epoch": 0.4397974616952719,
+      "loss": 0.13149940967559814,
+      "loss_ce": 0.003600239520892501,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 287181772,
+      "step": 1672
+    },
+    {
+      "epoch": 0.44006049845465905,
+      "grad_norm": 5.524031672248575,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 287353976,
+      "step": 1673
+    },
+    {
+      "epoch": 0.44006049845465905,
+      "loss": 0.09840566664934158,
+      "loss_ce": 0.0025804713368415833,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 287353976,
+      "step": 1673
+    },
+    {
+      "epoch": 0.44032353521404616,
+      "grad_norm": 9.273568707994169,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 287526172,
+      "step": 1674
+    },
+    {
+      "epoch": 0.44032353521404616,
+      "loss": 0.06107574701309204,
+      "loss_ce": 0.0005288777174428105,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 287526172,
+      "step": 1674
+    },
+    {
+      "epoch": 0.44058657197343326,
+      "grad_norm": 4.970980583351768,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 287698712,
+      "step": 1675
+    },
+    {
+      "epoch": 0.44058657197343326,
+      "loss": 0.15458138287067413,
+      "loss_ce": 0.004831631202250719,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 287698712,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4408496087328204,
+      "grad_norm": 5.818694779717612,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 287871108,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4408496087328204,
+      "loss": 0.09261530637741089,
+      "loss_ce": 0.0006353222415782511,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 287871108,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4411126454922075,
+      "grad_norm": 13.742421763597438,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 288043424,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4411126454922075,
+      "loss": 0.0773499608039856,
+      "loss_ce": 0.0007813603151589632,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 288043424,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4413756822515947,
+      "grad_norm": 4.64225389501199,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 288215812,
+      "step": 1678
+    },
+    {
+      "epoch": 0.4413756822515947,
+      "loss": 0.1821221113204956,
+      "loss_ce": 0.0025872092228382826,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 288215812,
+      "step": 1678
+    },
+    {
+      "epoch": 0.4416387190109818,
+      "grad_norm": 3.729084500729301,
+      "learning_rate": 5e-06,
+      "loss": 0.1359,
+      "num_input_tokens_seen": 288388096,
+      "step": 1679
+    },
+    {
+      "epoch": 0.4416387190109818,
+      "loss": 0.07683113217353821,
+      "loss_ce": 0.000735556473955512,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 288388096,
+      "step": 1679
+    },
+    {
+      "epoch": 0.4419017557703689,
+      "grad_norm": 3.7239911787320956,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 288560596,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4419017557703689,
+      "loss": 0.09967576712369919,
+      "loss_ce": 0.002248398493975401,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 288560596,
+      "step": 1680
+    },
+    {
+      "epoch": 0.44216479252975605,
+      "grad_norm": 7.1544540740612845,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 288730844,
+      "step": 1681
+    },
+    {
+      "epoch": 0.44216479252975605,
+      "loss": 0.14395672082901,
+      "loss_ce": 0.0017753278370946646,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 288730844,
+      "step": 1681
+    },
+    {
+      "epoch": 0.44242782928914315,
+      "grad_norm": 6.963458789097864,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 288901440,
+      "step": 1682
+    },
+    {
+      "epoch": 0.44242782928914315,
+      "loss": 0.12861037254333496,
+      "loss_ce": 0.0015962182078510523,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 288901440,
+      "step": 1682
+    },
+    {
+      "epoch": 0.4426908660485303,
+      "grad_norm": 6.124914951048281,
+      "learning_rate": 5e-06,
+      "loss": 0.2107,
+      "num_input_tokens_seen": 289073420,
+      "step": 1683
+    },
+    {
+      "epoch": 0.4426908660485303,
+      "loss": 0.19420379400253296,
+      "loss_ce": 0.0013021675404161215,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 289073420,
+      "step": 1683
+    },
+    {
+      "epoch": 0.4429539028079174,
+      "grad_norm": 9.860837139100692,
+      "learning_rate": 5e-06,
+      "loss": 0.172,
+      "num_input_tokens_seen": 289245604,
+      "step": 1684
+    },
+    {
+      "epoch": 0.4429539028079174,
+      "loss": 0.18933850526809692,
+      "loss_ce": 0.002204704098403454,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 289245604,
+      "step": 1684
+    },
+    {
+      "epoch": 0.4432169395673045,
+      "grad_norm": 7.0676385774671875,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 289417760,
+      "step": 1685
+    },
+    {
+      "epoch": 0.4432169395673045,
+      "loss": 0.16348493099212646,
+      "loss_ce": 0.0013145222328603268,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 289417760,
+      "step": 1685
+    },
+    {
+      "epoch": 0.44347997632669167,
+      "grad_norm": 5.183221235705912,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 289590036,
+      "step": 1686
+    },
+    {
+      "epoch": 0.44347997632669167,
+      "loss": 0.07584193348884583,
+      "loss_ce": 0.003545790910720825,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 289590036,
+      "step": 1686
+    },
+    {
+      "epoch": 0.4437430130860788,
+      "grad_norm": 4.680777365130206,
+      "learning_rate": 5e-06,
+      "loss": 0.13,
+      "num_input_tokens_seen": 289762148,
+      "step": 1687
+    },
+    {
+      "epoch": 0.4437430130860788,
+      "loss": 0.15222427248954773,
+      "loss_ce": 0.000765529228374362,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 289762148,
+      "step": 1687
+    },
+    {
+      "epoch": 0.4440060498454659,
+      "grad_norm": 8.359052775831522,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 289932260,
+      "step": 1688
+    },
+    {
+      "epoch": 0.4440060498454659,
+      "loss": 0.15807722508907318,
+      "loss_ce": 0.003322579897940159,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 289932260,
+      "step": 1688
+    },
+    {
+      "epoch": 0.44426908660485304,
+      "grad_norm": 12.300759876906785,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 290101064,
+      "step": 1689
+    },
+    {
+      "epoch": 0.44426908660485304,
+      "loss": 0.1050846129655838,
+      "loss_ce": 0.00016517633048351854,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 290101064,
+      "step": 1689
+    },
+    {
+      "epoch": 0.44453212336424014,
+      "grad_norm": 7.817323041044055,
+      "learning_rate": 5e-06,
+      "loss": 0.153,
+      "num_input_tokens_seen": 290273392,
+      "step": 1690
+    },
+    {
+      "epoch": 0.44453212336424014,
+      "loss": 0.10968184471130371,
+      "loss_ce": 0.004731892608106136,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 290273392,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4447951601236273,
+      "grad_norm": 10.723057383347166,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 290445728,
+      "step": 1691
+    },
+    {
+      "epoch": 0.4447951601236273,
+      "loss": 0.11715184152126312,
+      "loss_ce": 0.0027719633653759956,
+      "loss_iou": 0.5,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 290445728,
+      "step": 1691
+    },
+    {
+      "epoch": 0.4450581968830144,
+      "grad_norm": 5.5921534938980795,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 290618040,
+      "step": 1692
+    },
+    {
+      "epoch": 0.4450581968830144,
+      "loss": 0.11884011328220367,
+      "loss_ce": 0.0018509816145524383,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 290618040,
+      "step": 1692
+    },
+    {
+      "epoch": 0.4453212336424015,
+      "grad_norm": 14.56261412387689,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 290790260,
+      "step": 1693
+    },
+    {
+      "epoch": 0.4453212336424015,
+      "loss": 0.12701714038848877,
+      "loss_ce": 0.0006743660196661949,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 290790260,
+      "step": 1693
+    },
+    {
+      "epoch": 0.44558427040178866,
+      "grad_norm": 18.011685132647553,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 290960676,
+      "step": 1694
+    },
+    {
+      "epoch": 0.44558427040178866,
+      "loss": 0.17821697890758514,
+      "loss_ce": 0.00457195146009326,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 290960676,
+      "step": 1694
+    },
+    {
+      "epoch": 0.44584730716117577,
+      "grad_norm": 5.536946199556455,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 291132980,
+      "step": 1695
+    },
+    {
+      "epoch": 0.44584730716117577,
+      "loss": 0.10020774602890015,
+      "loss_ce": 0.0019716662354767323,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 291132980,
+      "step": 1695
+    },
+    {
+      "epoch": 0.4461103439205629,
+      "grad_norm": 14.998078131446578,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 291305020,
+      "step": 1696
+    },
+    {
+      "epoch": 0.4461103439205629,
+      "loss": 0.26434189081192017,
+      "loss_ce": 0.003111420664936304,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 291305020,
+      "step": 1696
+    },
+    {
+      "epoch": 0.44637338067995,
+      "grad_norm": 5.503328090464836,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 291477360,
+      "step": 1697
+    },
+    {
+      "epoch": 0.44637338067995,
+      "loss": 0.08737830072641373,
+      "loss_ce": 0.0004947560373693705,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 291477360,
+      "step": 1697
+    },
+    {
+      "epoch": 0.44663641743933713,
+      "grad_norm": 4.950750612606285,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 291647704,
+      "step": 1698
+    },
+    {
+      "epoch": 0.44663641743933713,
+      "loss": 0.09089531749486923,
+      "loss_ce": 0.00031914000282995403,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 291647704,
+      "step": 1698
+    },
+    {
+      "epoch": 0.4468994541987243,
+      "grad_norm": 7.442760799509784,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 291820320,
+      "step": 1699
+    },
+    {
+      "epoch": 0.4468994541987243,
+      "loss": 0.1341613531112671,
+      "loss_ce": 0.004461641423404217,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 291820320,
+      "step": 1699
+    },
+    {
+      "epoch": 0.4471624909581114,
+      "grad_norm": 12.789945030223832,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 291992692,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4471624909581114,
+      "loss": 0.12477381527423859,
+      "loss_ce": 0.002901863306760788,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 291992692,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4474255277174985,
+      "grad_norm": 4.791603869867936,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 292164820,
+      "step": 1701
+    },
+    {
+      "epoch": 0.4474255277174985,
+      "loss": 0.1102890819311142,
+      "loss_ce": 0.002928241156041622,
+      "loss_iou": 0.421875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 292164820,
+      "step": 1701
+    },
+    {
+      "epoch": 0.44768856447688565,
+      "grad_norm": 4.5597072248390145,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 292336840,
+      "step": 1702
+    },
+    {
+      "epoch": 0.44768856447688565,
+      "loss": 0.13155938684940338,
+      "loss_ce": 0.0015850251074880362,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 292336840,
+      "step": 1702
+    },
+    {
+      "epoch": 0.44795160123627276,
+      "grad_norm": 4.229506785629678,
+      "learning_rate": 5e-06,
+      "loss": 0.1794,
+      "num_input_tokens_seen": 292507240,
+      "step": 1703
+    },
+    {
+      "epoch": 0.44795160123627276,
+      "loss": 0.12941154837608337,
+      "loss_ce": 0.0014818647177889943,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 292507240,
+      "step": 1703
+    },
+    {
+      "epoch": 0.4482146379956599,
+      "grad_norm": 5.641270696096721,
+      "learning_rate": 5e-06,
+      "loss": 0.1276,
+      "num_input_tokens_seen": 292679176,
+      "step": 1704
+    },
+    {
+      "epoch": 0.4482146379956599,
+      "loss": 0.12304902821779251,
+      "loss_ce": 0.002321491949260235,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 292679176,
+      "step": 1704
+    },
+    {
+      "epoch": 0.448477674755047,
+      "grad_norm": 4.862520419399453,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 292851500,
+      "step": 1705
+    },
+    {
+      "epoch": 0.448477674755047,
+      "loss": 0.089045949280262,
+      "loss_ce": 0.00478691840544343,
+      "loss_iou": 0.546875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 292851500,
+      "step": 1705
+    },
+    {
+      "epoch": 0.4487407115144341,
+      "grad_norm": 5.2865399556662975,
+      "learning_rate": 5e-06,
+      "loss": 0.1533,
+      "num_input_tokens_seen": 293020992,
+      "step": 1706
+    },
+    {
+      "epoch": 0.4487407115144341,
+      "loss": 0.20740769803524017,
+      "loss_ce": 0.0010478447657078505,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 293020992,
+      "step": 1706
+    },
+    {
+      "epoch": 0.4490037482738213,
+      "grad_norm": 4.239149151423396,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 293192840,
+      "step": 1707
+    },
+    {
+      "epoch": 0.4490037482738213,
+      "loss": 0.1247292309999466,
+      "loss_ce": 0.002048558322712779,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 293192840,
+      "step": 1707
+    },
+    {
+      "epoch": 0.4492667850332084,
+      "grad_norm": 14.697036954882604,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 293359152,
+      "step": 1708
+    },
+    {
+      "epoch": 0.4492667850332084,
+      "loss": 0.17449304461479187,
+      "loss_ce": 0.000573370314668864,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 293359152,
+      "step": 1708
+    },
+    {
+      "epoch": 0.44952982179259554,
+      "grad_norm": 6.428150274293997,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 293531168,
+      "step": 1709
+    },
+    {
+      "epoch": 0.44952982179259554,
+      "loss": 0.0824984684586525,
+      "loss_ce": 0.00025359162827953696,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 293531168,
+      "step": 1709
+    },
+    {
+      "epoch": 0.44979285855198264,
+      "grad_norm": 12.082892344047602,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 293701596,
+      "step": 1710
+    },
+    {
+      "epoch": 0.44979285855198264,
+      "loss": 0.075187087059021,
+      "loss_ce": 0.00011384247045498341,
+      "loss_iou": 0.77734375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 293701596,
+      "step": 1710
+    },
+    {
+      "epoch": 0.45005589531136975,
+      "grad_norm": 6.447369973784934,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 293874048,
+      "step": 1711
+    },
+    {
+      "epoch": 0.45005589531136975,
+      "loss": 0.12107305228710175,
+      "loss_ce": 0.0001013779838103801,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 293874048,
+      "step": 1711
+    },
+    {
+      "epoch": 0.4503189320707569,
+      "grad_norm": 4.920690909069883,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 294046324,
+      "step": 1712
+    },
+    {
+      "epoch": 0.4503189320707569,
+      "loss": 0.09293599426746368,
+      "loss_ce": 0.0002541054564062506,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 294046324,
+      "step": 1712
+    },
+    {
+      "epoch": 0.450581968830144,
+      "grad_norm": 5.111263917035011,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 294218352,
+      "step": 1713
+    },
+    {
+      "epoch": 0.450581968830144,
+      "loss": 0.08807346224784851,
+      "loss_ce": 0.005675997585058212,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 294218352,
+      "step": 1713
+    },
+    {
+      "epoch": 0.4508450055895311,
+      "grad_norm": 22.02715708529402,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 294390700,
+      "step": 1714
+    },
+    {
+      "epoch": 0.4508450055895311,
+      "loss": 0.16420426964759827,
+      "loss_ce": 0.00023331816191785038,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 294390700,
+      "step": 1714
+    },
+    {
+      "epoch": 0.45110804234891827,
+      "grad_norm": 36.948349834543606,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 294562520,
+      "step": 1715
+    },
+    {
+      "epoch": 0.45110804234891827,
+      "loss": 0.20613673329353333,
+      "loss_ce": 0.0074672941118478775,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 294562520,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4513710791083054,
+      "grad_norm": 11.731491784871416,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 294732980,
+      "step": 1716
+    },
+    {
+      "epoch": 0.4513710791083054,
+      "loss": 0.2512318789958954,
+      "loss_ce": 0.0024831017944961786,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 294732980,
+      "step": 1716
+    },
+    {
+      "epoch": 0.45163411586769253,
+      "grad_norm": 14.33875314671704,
+      "learning_rate": 5e-06,
+      "loss": 0.146,
+      "num_input_tokens_seen": 294904972,
+      "step": 1717
+    },
+    {
+      "epoch": 0.45163411586769253,
+      "loss": 0.1545875370502472,
+      "loss_ce": 0.0034644976258277893,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 294904972,
+      "step": 1717
+    },
+    {
+      "epoch": 0.45189715262707963,
+      "grad_norm": 4.667158870565546,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 295077136,
+      "step": 1718
+    },
+    {
+      "epoch": 0.45189715262707963,
+      "loss": 0.10374893248081207,
+      "loss_ce": 0.0009657314512878656,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 295077136,
+      "step": 1718
+    },
+    {
+      "epoch": 0.45216018938646674,
+      "grad_norm": 4.671478799286235,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 295249644,
+      "step": 1719
+    },
+    {
+      "epoch": 0.45216018938646674,
+      "loss": 0.12534737586975098,
+      "loss_ce": 0.004924997687339783,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 295249644,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4524232261458539,
+      "grad_norm": 9.77050555348203,
+      "learning_rate": 5e-06,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 295421628,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4524232261458539,
+      "loss": 0.10829440504312515,
+      "loss_ce": 0.0021237479522824287,
+      "loss_iou": 0.65625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 295421628,
+      "step": 1720
+    },
+    {
+      "epoch": 0.452686262905241,
+      "grad_norm": 6.10598748249797,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 295593664,
+      "step": 1721
+    },
+    {
+      "epoch": 0.452686262905241,
+      "loss": 0.05939865857362747,
+      "loss_ce": 0.0012016374384984374,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 295593664,
+      "step": 1721
+    },
+    {
+      "epoch": 0.45294929966462816,
+      "grad_norm": 10.371953857475111,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 295765768,
+      "step": 1722
+    },
+    {
+      "epoch": 0.45294929966462816,
+      "loss": 0.10428635776042938,
+      "loss_ce": 0.0005265921936370432,
+      "loss_iou": 0.76171875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 295765768,
+      "step": 1722
+    },
+    {
+      "epoch": 0.45321233642401526,
+      "grad_norm": 6.589623724090896,
+      "learning_rate": 5e-06,
+      "loss": 0.1679,
+      "num_input_tokens_seen": 295937984,
+      "step": 1723
+    },
+    {
+      "epoch": 0.45321233642401526,
+      "loss": 0.12776082754135132,
+      "loss_ce": 0.0015095948474481702,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 295937984,
+      "step": 1723
+    },
+    {
+      "epoch": 0.45347537318340236,
+      "grad_norm": 5.38306851263364,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 296109952,
+      "step": 1724
+    },
+    {
+      "epoch": 0.45347537318340236,
+      "loss": 0.10374290496110916,
+      "loss_ce": 0.0004714199749287218,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 296109952,
+      "step": 1724
+    },
+    {
+      "epoch": 0.4537384099427895,
+      "grad_norm": 16.069341006935545,
+      "learning_rate": 5e-06,
+      "loss": 0.1861,
+      "num_input_tokens_seen": 296282320,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4537384099427895,
+      "loss": 0.3360915184020996,
+      "loss_ce": 0.003205763641744852,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 296282320,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4540014467021766,
+      "grad_norm": 7.355868991768278,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 296454412,
+      "step": 1726
+    },
+    {
+      "epoch": 0.4540014467021766,
+      "loss": 0.11106541752815247,
+      "loss_ce": 0.0014462803956121206,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 296454412,
+      "step": 1726
+    },
+    {
+      "epoch": 0.45426448346156373,
+      "grad_norm": 4.637757034723655,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 296626392,
+      "step": 1727
+    },
+    {
+      "epoch": 0.45426448346156373,
+      "loss": 0.07483752816915512,
+      "loss_ce": 0.00046619208296760917,
+      "loss_iou": 0.359375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 296626392,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4545275202209509,
+      "grad_norm": 4.094172069183009,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 296795792,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4545275202209509,
+      "loss": 0.1533362716436386,
+      "loss_ce": 0.00154183991253376,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 296795792,
+      "step": 1728
+    },
+    {
+      "epoch": 0.454790556980338,
+      "grad_norm": 4.550319533659712,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 296966316,
+      "step": 1729
+    },
+    {
+      "epoch": 0.454790556980338,
+      "loss": 0.25365394353866577,
+      "loss_ce": 0.002402704209089279,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 296966316,
+      "step": 1729
+    },
+    {
+      "epoch": 0.45505359373972515,
+      "grad_norm": 5.052226011800126,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 297136040,
+      "step": 1730
+    },
+    {
+      "epoch": 0.45505359373972515,
+      "loss": 0.1388048231601715,
+      "loss_ce": 0.0004991517635062337,
+      "loss_iou": 0.640625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 297136040,
+      "step": 1730
+    },
+    {
+      "epoch": 0.45531663049911225,
+      "grad_norm": 9.365703053895057,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 297306228,
+      "step": 1731
+    },
+    {
+      "epoch": 0.45531663049911225,
+      "loss": 0.09631586819887161,
+      "loss_ce": 0.0004296356055419892,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 297306228,
+      "step": 1731
+    },
+    {
+      "epoch": 0.45557966725849935,
+      "grad_norm": 5.623520624492047,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 297478380,
+      "step": 1732
+    },
+    {
+      "epoch": 0.45557966725849935,
+      "loss": 0.21242645382881165,
+      "loss_ce": 0.0005734282894991338,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 297478380,
+      "step": 1732
+    },
+    {
+      "epoch": 0.4558427040178865,
+      "grad_norm": 22.29793682222897,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 297650336,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4558427040178865,
+      "loss": 0.12880732119083405,
+      "loss_ce": 0.0012133296113461256,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 297650336,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4561057407772736,
+      "grad_norm": 19.5277076823797,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 297822376,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4561057407772736,
+      "loss": 0.10357876121997833,
+      "loss_ce": 0.00030727204284630716,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 297822376,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4563687775366608,
+      "grad_norm": 4.597687331185091,
+      "learning_rate": 5e-06,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 297994524,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4563687775366608,
+      "loss": 0.11487319320440292,
+      "loss_ce": 0.0003712356265168637,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 297994524,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4566318142960479,
+      "grad_norm": 17.34014327670223,
+      "learning_rate": 5e-06,
+      "loss": 0.1904,
+      "num_input_tokens_seen": 298164216,
+      "step": 1736
+    },
+    {
+      "epoch": 0.4566318142960479,
+      "loss": 0.21512141823768616,
+      "loss_ce": 0.0003387040051165968,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 298164216,
+      "step": 1736
+    },
+    {
+      "epoch": 0.456894851055435,
+      "grad_norm": 5.201392853210846,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 298336416,
+      "step": 1737
+    },
+    {
+      "epoch": 0.456894851055435,
+      "loss": 0.14383375644683838,
+      "loss_ce": 0.0030866768211126328,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 298336416,
+      "step": 1737
+    },
+    {
+      "epoch": 0.45715788781482214,
+      "grad_norm": 4.466741788292063,
+      "learning_rate": 5e-06,
+      "loss": 0.1267,
+      "num_input_tokens_seen": 298508336,
+      "step": 1738
+    },
+    {
+      "epoch": 0.45715788781482214,
+      "loss": 0.08519221842288971,
+      "loss_ce": 0.001879227813333273,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 298508336,
+      "step": 1738
+    },
+    {
+      "epoch": 0.45742092457420924,
+      "grad_norm": 5.105148151648125,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 298680480,
+      "step": 1739
+    },
+    {
+      "epoch": 0.45742092457420924,
+      "loss": 0.08455468714237213,
+      "loss_ce": 0.00038720344309695065,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 298680480,
+      "step": 1739
+    },
+    {
+      "epoch": 0.45768396133359635,
+      "grad_norm": 4.88765152218918,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 298852576,
+      "step": 1740
+    },
+    {
+      "epoch": 0.45768396133359635,
+      "loss": 0.06836723536252975,
+      "loss_ce": 0.0006487306673079729,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 298852576,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4579469980929835,
+      "grad_norm": 11.438189750287325,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 299025188,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4579469980929835,
+      "loss": 0.2103656679391861,
+      "loss_ce": 0.0009235285106115043,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 299025188,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4582100348523706,
+      "grad_norm": 17.844442720051195,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 299195652,
+      "step": 1742
+    },
+    {
+      "epoch": 0.4582100348523706,
+      "loss": 0.0902654230594635,
+      "loss_ce": 0.0008184025646187365,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 299195652,
+      "step": 1742
+    },
+    {
+      "epoch": 0.45847307161175777,
+      "grad_norm": 5.593927162724121,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 299367692,
+      "step": 1743
+    },
+    {
+      "epoch": 0.45847307161175777,
+      "loss": 0.0877409279346466,
+      "loss_ce": 0.004031214863061905,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 299367692,
+      "step": 1743
+    },
+    {
+      "epoch": 0.45873610837114487,
+      "grad_norm": 6.084084258485456,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 299539820,
+      "step": 1744
+    },
+    {
+      "epoch": 0.45873610837114487,
+      "loss": 0.17818066477775574,
+      "loss_ce": 0.0005988804623484612,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 299539820,
+      "step": 1744
+    },
+    {
+      "epoch": 0.45899914513053197,
+      "grad_norm": 4.3161100884144785,
+      "learning_rate": 5e-06,
+      "loss": 0.0857,
+      "num_input_tokens_seen": 299710180,
+      "step": 1745
+    },
+    {
+      "epoch": 0.45899914513053197,
+      "loss": 0.11408813297748566,
+      "loss_ce": 0.005079346243292093,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 299710180,
+      "step": 1745
+    },
+    {
+      "epoch": 0.45926218188991913,
+      "grad_norm": 9.432280511482096,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "num_input_tokens_seen": 299882192,
+      "step": 1746
+    },
+    {
+      "epoch": 0.45926218188991913,
+      "loss": 0.28233322501182556,
+      "loss_ce": 0.002944799605756998,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 299882192,
+      "step": 1746
+    },
+    {
+      "epoch": 0.45952521864930623,
+      "grad_norm": 4.947993402328502,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 300054328,
+      "step": 1747
+    },
+    {
+      "epoch": 0.45952521864930623,
+      "loss": 0.12365365773439407,
+      "loss_ce": 0.00454354751855135,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 300054328,
+      "step": 1747
+    },
+    {
+      "epoch": 0.4597882554086934,
+      "grad_norm": 29.241280731418787,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 300226268,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4597882554086934,
+      "loss": 0.09467601031064987,
+      "loss_ce": 0.002909653354436159,
+      "loss_iou": 0.546875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 300226268,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4600512921680805,
+      "grad_norm": 40.89605718973821,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 300398324,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4600512921680805,
+      "loss": 0.18484918773174286,
+      "loss_ce": 0.002537175314500928,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 300398324,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "grad_norm": 6.154438338656609,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_websight_new_CIoU": 0.8543897271156311,
+      "eval_websight_new_GIoU": 0.8555817008018494,
+      "eval_websight_new_IoU": 0.8602511882781982,
+      "eval_websight_new_MAE_all": 0.022463313303887844,
+      "eval_websight_new_MAE_h": 0.007148948730900884,
+      "eval_websight_new_MAE_w": 0.03532572276890278,
+      "eval_websight_new_MAE_x": 0.03696838486939669,
+      "eval_websight_new_MAE_y": 0.010410191491246223,
+      "eval_websight_new_NUM_probability": 0.9999746978282928,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.11025163531303406,
+      "eval_websight_new_loss_ce": 7.8859661698516e-06,
+      "eval_websight_new_loss_iou": 0.3748779296875,
+      "eval_websight_new_loss_num": 0.019195556640625,
+      "eval_websight_new_loss_xval": 0.09600830078125,
+      "eval_websight_new_runtime": 55.1926,
+      "eval_websight_new_samples_per_second": 0.906,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_seeclick_CIoU": 0.6177069246768951,
+      "eval_seeclick_GIoU": 0.6163533926010132,
+      "eval_seeclick_IoU": 0.6425465941429138,
+      "eval_seeclick_MAE_all": 0.04845697060227394,
+      "eval_seeclick_MAE_h": 0.026225415989756584,
+      "eval_seeclick_MAE_w": 0.06570588797330856,
+      "eval_seeclick_MAE_x": 0.07359151728451252,
+      "eval_seeclick_MAE_y": 0.0283050537109375,
+      "eval_seeclick_NUM_probability": 0.9999766051769257,
+      "eval_seeclick_inside_bbox": 0.890625,
+      "eval_seeclick_loss": 0.23021526634693146,
+      "eval_seeclick_loss_ce": 0.009315645787864923,
+      "eval_seeclick_loss_iou": 0.5054931640625,
+      "eval_seeclick_loss_num": 0.04460906982421875,
+      "eval_seeclick_loss_xval": 0.22314453125,
+      "eval_seeclick_runtime": 72.9378,
+      "eval_seeclick_samples_per_second": 0.59,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_icons_CIoU": 0.8409464359283447,
+      "eval_icons_GIoU": 0.8350943326950073,
+      "eval_icons_IoU": 0.847510576248169,
+      "eval_icons_MAE_all": 0.022730856202542782,
+      "eval_icons_MAE_h": 0.02164691872894764,
+      "eval_icons_MAE_w": 0.023992713540792465,
+      "eval_icons_MAE_x": 0.025108729489147663,
+      "eval_icons_MAE_y": 0.02017505932599306,
+      "eval_icons_NUM_probability": 0.9999510943889618,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.079879030585289,
+      "eval_icons_loss_ce": 2.25404792217887e-05,
+      "eval_icons_loss_iou": 0.5426025390625,
+      "eval_icons_loss_num": 0.014867782592773438,
+      "eval_icons_loss_xval": 0.0743560791015625,
+      "eval_icons_runtime": 87.1455,
+      "eval_icons_samples_per_second": 0.574,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_screenspot_CIoU": 0.5540184179941813,
+      "eval_screenspot_GIoU": 0.543454110622406,
+      "eval_screenspot_IoU": 0.5961946249008179,
+      "eval_screenspot_MAE_all": 0.08684368679920833,
+      "eval_screenspot_MAE_h": 0.04899499130745729,
+      "eval_screenspot_MAE_w": 0.1521986499428749,
+      "eval_screenspot_MAE_x": 0.09981551021337509,
+      "eval_screenspot_MAE_y": 0.046365607529878616,
+      "eval_screenspot_NUM_probability": 0.9998787045478821,
+      "eval_screenspot_inside_bbox": 0.8737499912579855,
+      "eval_screenspot_loss": 0.8499577641487122,
+      "eval_screenspot_loss_ce": 0.5143006145954132,
+      "eval_screenspot_loss_iou": 0.4506022135416667,
+      "eval_screenspot_loss_num": 0.06610107421875,
+      "eval_screenspot_loss_xval": 0.3305257161458333,
+      "eval_screenspot_runtime": 149.7903,
+      "eval_screenspot_samples_per_second": 0.594,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "loss": 0.855069637298584,
+      "loss_ce": 0.507047176361084,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46057736568685476,
+      "grad_norm": 6.10997370548099,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 300742608,
+      "step": 1751
+    },
+    {
+      "epoch": 0.46057736568685476,
+      "loss": 0.11830765753984451,
+      "loss_ce": 0.0002961806021630764,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 300742608,
+      "step": 1751
+    },
+    {
+      "epoch": 0.46084040244624186,
+      "grad_norm": 4.63619140673612,
+      "learning_rate": 5e-06,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 300914728,
+      "step": 1752
+    },
+    {
+      "epoch": 0.46084040244624186,
+      "loss": 0.14693626761436462,
+      "loss_ce": 0.0036867514718323946,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 300914728,
+      "step": 1752
+    },
+    {
+      "epoch": 0.46110343920562896,
+      "grad_norm": 11.051347194862576,
+      "learning_rate": 5e-06,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 301087180,
+      "step": 1753
+    },
+    {
+      "epoch": 0.46110343920562896,
+      "loss": 0.2443966418504715,
+      "loss_ce": 0.0005001651006750762,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 301087180,
+      "step": 1753
+    },
+    {
+      "epoch": 0.4613664759650161,
+      "grad_norm": 9.081946267145833,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 301259116,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4613664759650161,
+      "loss": 0.09555494785308838,
+      "loss_ce": 0.002583135850727558,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 301259116,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4616295127244032,
+      "grad_norm": 7.9541431928925395,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 301427764,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4616295127244032,
+      "loss": 0.12469062209129333,
+      "loss_ce": 0.0007892490248195827,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 301427764,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4618925494837904,
+      "grad_norm": 4.563316138635991,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 301599772,
+      "step": 1756
+    },
+    {
+      "epoch": 0.4618925494837904,
+      "loss": 0.0649976134300232,
+      "loss_ce": 0.0009717366192489862,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 301599772,
+      "step": 1756
+    },
+    {
+      "epoch": 0.4621555862431775,
+      "grad_norm": 13.734783204569586,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 301771888,
+      "step": 1757
+    },
+    {
+      "epoch": 0.4621555862431775,
+      "loss": 0.2562709152698517,
+      "loss_ce": 0.0031581264920532703,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 301771888,
+      "step": 1757
+    },
+    {
+      "epoch": 0.4624186230025646,
+      "grad_norm": 8.489244421344079,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 301944272,
+      "step": 1758
+    },
+    {
+      "epoch": 0.4624186230025646,
+      "loss": 0.08469030261039734,
+      "loss_ce": 0.00088903569849208,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 301944272,
+      "step": 1758
+    },
+    {
+      "epoch": 0.46268165976195175,
+      "grad_norm": 5.029084121420607,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 302116376,
+      "step": 1759
+    },
+    {
+      "epoch": 0.46268165976195175,
+      "loss": 0.12341859936714172,
+      "loss_ce": 0.0011651779059320688,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 302116376,
+      "step": 1759
+    },
+    {
+      "epoch": 0.46294469652133885,
+      "grad_norm": 20.972527246553025,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 302288436,
+      "step": 1760
+    },
+    {
+      "epoch": 0.46294469652133885,
+      "loss": 0.0873933807015419,
+      "loss_ce": 0.0027071028016507626,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 302288436,
+      "step": 1760
+    },
+    {
+      "epoch": 0.463207733280726,
+      "grad_norm": 4.449125782082312,
+      "learning_rate": 5e-06,
+      "loss": 0.1406,
+      "num_input_tokens_seen": 302460504,
+      "step": 1761
+    },
+    {
+      "epoch": 0.463207733280726,
+      "loss": 0.10758376121520996,
+      "loss_ce": 0.0050141820684075356,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 302460504,
+      "step": 1761
+    },
+    {
+      "epoch": 0.4634707700401131,
+      "grad_norm": 6.144634607422675,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 302632396,
+      "step": 1762
+    },
+    {
+      "epoch": 0.4634707700401131,
+      "loss": 0.19673708081245422,
+      "loss_ce": 0.0027063230518251657,
+      "loss_iou": 0.484375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 302632396,
+      "step": 1762
+    },
+    {
+      "epoch": 0.4637338067995002,
+      "grad_norm": 4.45571483495035,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 302804644,
+      "step": 1763
+    },
+    {
+      "epoch": 0.4637338067995002,
+      "loss": 0.0787121132016182,
+      "loss_ce": 0.005683551542460918,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 302804644,
+      "step": 1763
+    },
+    {
+      "epoch": 0.4639968435588874,
+      "grad_norm": 20.80296883365536,
+      "learning_rate": 5e-06,
+      "loss": 0.1304,
+      "num_input_tokens_seen": 302976856,
+      "step": 1764
+    },
+    {
+      "epoch": 0.4639968435588874,
+      "loss": 0.14017972350120544,
+      "loss_ce": 0.0037661464884877205,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 302976856,
+      "step": 1764
+    },
+    {
+      "epoch": 0.4642598803182745,
+      "grad_norm": 5.519977681881848,
+      "learning_rate": 5e-06,
+      "loss": 0.1241,
+      "num_input_tokens_seen": 303146760,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4642598803182745,
+      "loss": 0.12335249036550522,
+      "loss_ce": 0.002991169923916459,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 303146760,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4645229170776616,
+      "grad_norm": 6.31203658122147,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 303318968,
+      "step": 1766
+    },
+    {
+      "epoch": 0.4645229170776616,
+      "loss": 0.08828569203615189,
+      "loss_ce": 0.002073533833026886,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 303318968,
+      "step": 1766
+    },
+    {
+      "epoch": 0.46478595383704874,
+      "grad_norm": 5.046640432518341,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 303491036,
+      "step": 1767
+    },
+    {
+      "epoch": 0.46478595383704874,
+      "loss": 0.08855307102203369,
+      "loss_ce": 0.002066256944090128,
+      "loss_iou": 0.640625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 303491036,
+      "step": 1767
+    },
+    {
+      "epoch": 0.46504899059643584,
+      "grad_norm": 8.451371161082179,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 303663404,
+      "step": 1768
+    },
+    {
+      "epoch": 0.46504899059643584,
+      "loss": 0.16699054837226868,
+      "loss_ce": 0.002805991331115365,
+      "loss_iou": 0.640625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 303663404,
+      "step": 1768
+    },
+    {
+      "epoch": 0.465312027355823,
+      "grad_norm": 9.603701773353785,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 303832160,
+      "step": 1769
+    },
+    {
+      "epoch": 0.465312027355823,
+      "loss": 0.11529731005430222,
+      "loss_ce": 0.0019550304859876633,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 303832160,
+      "step": 1769
+    },
+    {
+      "epoch": 0.4655750641152101,
+      "grad_norm": 4.893083096864921,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 304004460,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4655750641152101,
+      "loss": 0.15418250858783722,
+      "loss_ce": 0.0016556488117203116,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 304004460,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4658381008745972,
+      "grad_norm": 5.299927971595323,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 304176564,
+      "step": 1771
+    },
+    {
+      "epoch": 0.4658381008745972,
+      "loss": 0.11469468474388123,
+      "loss_ce": 0.004251571837812662,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 304176564,
+      "step": 1771
+    },
+    {
+      "epoch": 0.46610113763398436,
+      "grad_norm": 4.786837798902794,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 304349064,
+      "step": 1772
+    },
+    {
+      "epoch": 0.46610113763398436,
+      "loss": 0.13777077198028564,
+      "loss_ce": 0.002303245011717081,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 304349064,
+      "step": 1772
+    },
+    {
+      "epoch": 0.46636417439337147,
+      "grad_norm": 7.199109826840915,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 304521336,
+      "step": 1773
+    },
+    {
+      "epoch": 0.46636417439337147,
+      "loss": 0.13359083235263824,
+      "loss_ce": 0.0007783286855556071,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 304521336,
+      "step": 1773
+    },
+    {
+      "epoch": 0.4666272111527586,
+      "grad_norm": 5.7773703638814355,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 304693604,
+      "step": 1774
+    },
+    {
+      "epoch": 0.4666272111527586,
+      "loss": 0.09553907811641693,
+      "loss_ce": 0.0015449414495378733,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 304693604,
+      "step": 1774
+    },
+    {
+      "epoch": 0.46689024791214573,
+      "grad_norm": 4.7750261071300475,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 304865896,
+      "step": 1775
+    },
+    {
+      "epoch": 0.46689024791214573,
+      "loss": 0.17456401884555817,
+      "loss_ce": 0.0003696825006045401,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 304865896,
+      "step": 1775
+    },
+    {
+      "epoch": 0.46715328467153283,
+      "grad_norm": 4.560916465958671,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 305038112,
+      "step": 1776
+    },
+    {
+      "epoch": 0.46715328467153283,
+      "loss": 0.16979777812957764,
+      "loss_ce": 0.002805587835609913,
+      "loss_iou": 0.6875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 305038112,
+      "step": 1776
+    },
+    {
+      "epoch": 0.46741632143092,
+      "grad_norm": 9.580731840657464,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 305210520,
+      "step": 1777
+    },
+    {
+      "epoch": 0.46741632143092,
+      "loss": 0.1295488327741623,
+      "loss_ce": 0.0010087917326018214,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 305210520,
+      "step": 1777
+    },
+    {
+      "epoch": 0.4676793581903071,
+      "grad_norm": 12.328201105743592,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 305382852,
+      "step": 1778
+    },
+    {
+      "epoch": 0.4676793581903071,
+      "loss": 0.10579667240381241,
+      "loss_ce": 0.0018843174912035465,
+      "loss_iou": 0.5625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 305382852,
+      "step": 1778
+    },
+    {
+      "epoch": 0.4679423949496942,
+      "grad_norm": 7.809054499493732,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 305554812,
+      "step": 1779
+    },
+    {
+      "epoch": 0.4679423949496942,
+      "loss": 0.12438150495290756,
+      "loss_ce": 0.003974398132413626,
+      "loss_iou": 0.59375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 305554812,
+      "step": 1779
+    },
+    {
+      "epoch": 0.46820543170908135,
+      "grad_norm": 8.973200505315825,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 305727124,
+      "step": 1780
+    },
+    {
+      "epoch": 0.46820543170908135,
+      "loss": 0.08462625741958618,
+      "loss_ce": 0.0006571417325176299,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 305727124,
+      "step": 1780
+    },
+    {
+      "epoch": 0.46846846846846846,
+      "grad_norm": 5.227611768036767,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 305899256,
+      "step": 1781
+    },
+    {
+      "epoch": 0.46846846846846846,
+      "loss": 0.1279701292514801,
+      "loss_ce": 0.0015358042437583208,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 305899256,
+      "step": 1781
+    },
+    {
+      "epoch": 0.4687315052278556,
+      "grad_norm": 3.599452107221254,
+      "learning_rate": 5e-06,
+      "loss": 0.109,
+      "num_input_tokens_seen": 306069952,
+      "step": 1782
+    },
+    {
+      "epoch": 0.4687315052278556,
+      "loss": 0.1316523402929306,
+      "loss_ce": 0.002959707286208868,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 306069952,
+      "step": 1782
+    },
+    {
+      "epoch": 0.4689945419872427,
+      "grad_norm": 6.0196869037820235,
+      "learning_rate": 5e-06,
+      "loss": 0.1144,
+      "num_input_tokens_seen": 306242268,
+      "step": 1783
+    },
+    {
+      "epoch": 0.4689945419872427,
+      "loss": 0.0880986899137497,
+      "loss_ce": 0.0006353051285259426,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 306242268,
+      "step": 1783
+    },
+    {
+      "epoch": 0.4692575787466298,
+      "grad_norm": 5.3230563681220735,
+      "learning_rate": 5e-06,
+      "loss": 0.1359,
+      "num_input_tokens_seen": 306414220,
+      "step": 1784
+    },
+    {
+      "epoch": 0.4692575787466298,
+      "loss": 0.08738180994987488,
+      "loss_ce": 0.0023750958498567343,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 306414220,
+      "step": 1784
+    },
+    {
+      "epoch": 0.469520615506017,
+      "grad_norm": 4.772051394078166,
+      "learning_rate": 5e-06,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 306586436,
+      "step": 1785
+    },
+    {
+      "epoch": 0.469520615506017,
+      "loss": 0.17895328998565674,
+      "loss_ce": 0.000791671103797853,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 306586436,
+      "step": 1785
+    },
+    {
+      "epoch": 0.4697836522654041,
+      "grad_norm": 5.377031057320062,
+      "learning_rate": 5e-06,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 306758712,
+      "step": 1786
+    },
+    {
+      "epoch": 0.4697836522654041,
+      "loss": 0.201407790184021,
+      "loss_ce": 0.004172691144049168,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 306758712,
+      "step": 1786
+    },
+    {
+      "epoch": 0.47004668902479124,
+      "grad_norm": 4.6699721270287275,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 306931112,
+      "step": 1787
+    },
+    {
+      "epoch": 0.47004668902479124,
+      "loss": 0.10016533732414246,
+      "loss_ce": 0.00031182204838842154,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 306931112,
+      "step": 1787
+    },
+    {
+      "epoch": 0.47030972578417835,
+      "grad_norm": 8.618398116569972,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 307103388,
+      "step": 1788
+    },
+    {
+      "epoch": 0.47030972578417835,
+      "loss": 0.27383407950401306,
+      "loss_ce": 0.0001524553372291848,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 307103388,
+      "step": 1788
+    },
+    {
+      "epoch": 0.47057276254356545,
+      "grad_norm": 11.779661781344625,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 307275620,
+      "step": 1789
+    },
+    {
+      "epoch": 0.47057276254356545,
+      "loss": 0.08500531315803528,
+      "loss_ce": 0.00019696829258464277,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 307275620,
+      "step": 1789
+    },
+    {
+      "epoch": 0.4708357993029526,
+      "grad_norm": 23.708154257705488,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 307448180,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4708357993029526,
+      "loss": 0.17157645523548126,
+      "loss_ce": 0.002493813633918762,
+      "loss_iou": 0.453125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 307448180,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4710988360623397,
+      "grad_norm": 4.7495128426496,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 307620540,
+      "step": 1791
+    },
+    {
+      "epoch": 0.4710988360623397,
+      "loss": 0.23302927613258362,
+      "loss_ce": 0.0009736126521602273,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 307620540,
+      "step": 1791
+    },
+    {
+      "epoch": 0.4713618728217268,
+      "grad_norm": 7.18577313542124,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 307792704,
+      "step": 1792
+    },
+    {
+      "epoch": 0.4713618728217268,
+      "loss": 0.13438743352890015,
+      "loss_ce": 0.0005068117170594633,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 307792704,
+      "step": 1792
+    },
+    {
+      "epoch": 0.47162490958111397,
+      "grad_norm": 7.179208918692874,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 307965184,
+      "step": 1793
+    },
+    {
+      "epoch": 0.47162490958111397,
+      "loss": 0.15465694665908813,
+      "loss_ce": 0.00011593455565162003,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 307965184,
+      "step": 1793
+    },
+    {
+      "epoch": 0.4718879463405011,
+      "grad_norm": 6.391344075166243,
+      "learning_rate": 5e-06,
+      "loss": 0.1742,
+      "num_input_tokens_seen": 308135620,
+      "step": 1794
+    },
+    {
+      "epoch": 0.4718879463405011,
+      "loss": 0.23051750659942627,
+      "loss_ce": 0.00010977771307807416,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 308135620,
+      "step": 1794
+    },
+    {
+      "epoch": 0.47215098309988823,
+      "grad_norm": 8.145404749300642,
+      "learning_rate": 5e-06,
+      "loss": 0.1465,
+      "num_input_tokens_seen": 308308152,
+      "step": 1795
+    },
+    {
+      "epoch": 0.47215098309988823,
+      "loss": 0.1917172074317932,
+      "loss_ce": 0.008871147409081459,
+      "loss_iou": 0.640625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 308308152,
+      "step": 1795
+    },
+    {
+      "epoch": 0.47241401985927534,
+      "grad_norm": 5.398862877955347,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 308477996,
+      "step": 1796
+    },
+    {
+      "epoch": 0.47241401985927534,
+      "loss": 0.08200335502624512,
+      "loss_ce": 0.0003993565624114126,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 308477996,
+      "step": 1796
+    },
+    {
+      "epoch": 0.47267705661866244,
+      "grad_norm": 9.91724255704017,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 308648424,
+      "step": 1797
+    },
+    {
+      "epoch": 0.47267705661866244,
+      "loss": 0.12875403463840485,
+      "loss_ce": 0.0009769393363967538,
+      "loss_iou": 0.515625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 308648424,
+      "step": 1797
+    },
+    {
+      "epoch": 0.4729400933780496,
+      "grad_norm": 6.138930079819964,
+      "learning_rate": 5e-06,
+      "loss": 0.1842,
+      "num_input_tokens_seen": 308818888,
+      "step": 1798
+    },
+    {
+      "epoch": 0.4729400933780496,
+      "loss": 0.16927096247673035,
+      "loss_ce": 0.0021567128133028746,
+      "loss_iou": 0.73046875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 308818888,
+      "step": 1798
+    },
+    {
+      "epoch": 0.4732031301374367,
+      "grad_norm": 19.105244313122192,
+      "learning_rate": 5e-06,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 308991284,
+      "step": 1799
+    },
+    {
+      "epoch": 0.4732031301374367,
+      "loss": 0.21268007159233093,
+      "loss_ce": 9.461388253839687e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 308991284,
+      "step": 1799
+    },
+    {
+      "epoch": 0.4734661668968238,
+      "grad_norm": 6.0514901455791215,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 309163244,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4734661668968238,
+      "loss": 0.11982670426368713,
+      "loss_ce": 0.00047245476162061095,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 309163244,
+      "step": 1800
+    },
+    {
+      "epoch": 0.47372920365621096,
+      "grad_norm": 4.529009692158789,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 309335260,
+      "step": 1801
+    },
+    {
+      "epoch": 0.47372920365621096,
+      "loss": 0.16322672367095947,
+      "loss_ce": 0.000812180747743696,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 309335260,
+      "step": 1801
+    },
+    {
+      "epoch": 0.47399224041559807,
+      "grad_norm": 4.778436741374609,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 309507516,
+      "step": 1802
+    },
+    {
+      "epoch": 0.47399224041559807,
+      "loss": 0.09654629230499268,
+      "loss_ce": 0.0012398946564644575,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 309507516,
+      "step": 1802
+    },
+    {
+      "epoch": 0.4742552771749852,
+      "grad_norm": 4.954945704123918,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 309677988,
+      "step": 1803
+    },
+    {
+      "epoch": 0.4742552771749852,
+      "loss": 0.0787278264760971,
+      "loss_ce": 0.00014505814760923386,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 309677988,
+      "step": 1803
+    },
+    {
+      "epoch": 0.4745183139343723,
+      "grad_norm": 12.56847461072702,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 309850180,
+      "step": 1804
+    },
+    {
+      "epoch": 0.4745183139343723,
+      "loss": 0.13278400897979736,
+      "loss_ce": 0.00015460627037100494,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 309850180,
+      "step": 1804
+    },
+    {
+      "epoch": 0.47478135069375943,
+      "grad_norm": 15.493537781093892,
+      "learning_rate": 5e-06,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 310019768,
+      "step": 1805
+    },
+    {
+      "epoch": 0.47478135069375943,
+      "loss": 0.05194393917918205,
+      "loss_ce": 0.0007049225969240069,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 310019768,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4750443874531466,
+      "grad_norm": 6.1982861025667795,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 310191904,
+      "step": 1806
+    },
+    {
+      "epoch": 0.4750443874531466,
+      "loss": 0.09039468318223953,
+      "loss_ce": 0.00045937972026877105,
+      "loss_iou": 0.7265625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 310191904,
+      "step": 1806
+    },
+    {
+      "epoch": 0.4753074242125337,
+      "grad_norm": 8.01906133906792,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 310364020,
+      "step": 1807
+    },
+    {
+      "epoch": 0.4753074242125337,
+      "loss": 0.1431303471326828,
+      "loss_ce": 0.002688447944819927,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 310364020,
+      "step": 1807
+    },
+    {
+      "epoch": 0.47557046097192085,
+      "grad_norm": 8.410195178652696,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 310536388,
+      "step": 1808
+    },
+    {
+      "epoch": 0.47557046097192085,
+      "loss": 0.1313067376613617,
+      "loss_ce": 0.002064801286906004,
+      "loss_iou": 0.73046875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 310536388,
+      "step": 1808
+    },
+    {
+      "epoch": 0.47583349773130795,
+      "grad_norm": 5.758329237623371,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 310706940,
+      "step": 1809
+    },
+    {
+      "epoch": 0.47583349773130795,
+      "loss": 0.0813891738653183,
+      "loss_ce": 0.0004565550771076232,
+      "loss_iou": 0.484375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 310706940,
+      "step": 1809
+    },
+    {
+      "epoch": 0.47609653449069506,
+      "grad_norm": 38.441949828692486,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 310879292,
+      "step": 1810
+    },
+    {
+      "epoch": 0.47609653449069506,
+      "loss": 0.06832106411457062,
+      "loss_ce": 0.007591082248836756,
+      "loss_iou": 0.75,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 310879292,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4763595712500822,
+      "grad_norm": 7.22086526843683,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 311051572,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4763595712500822,
+      "loss": 0.0902928039431572,
+      "loss_ce": 0.00047957096830941737,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 311051572,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4766226080094693,
+      "grad_norm": 11.063199574808042,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 311222148,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4766226080094693,
+      "loss": 0.2204355001449585,
+      "loss_ce": 0.001777050900273025,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 311222148,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4768856447688564,
+      "grad_norm": 10.035743209995298,
+      "learning_rate": 5e-06,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 311394364,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4768856447688564,
+      "loss": 0.07709920406341553,
+      "loss_ce": 0.003307701088488102,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 311394364,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4771486815282436,
+      "grad_norm": 8.932384890899103,
+      "learning_rate": 5e-06,
+      "loss": 0.1796,
+      "num_input_tokens_seen": 311566412,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4771486815282436,
+      "loss": 0.18021875619888306,
+      "loss_ce": 0.002743777120485902,
+      "loss_iou": 0.4375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 311566412,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4774117182876307,
+      "grad_norm": 7.670958022785577,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 311738816,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4774117182876307,
+      "loss": 0.06826284527778625,
+      "loss_ce": 0.002039696555584669,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 311738816,
+      "step": 1815
+    },
+    {
+      "epoch": 0.47767475504701784,
+      "grad_norm": 9.178964340109715,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 311910812,
+      "step": 1816
+    },
+    {
+      "epoch": 0.47767475504701784,
+      "loss": 0.15946456789970398,
+      "loss_ce": 0.0008036750950850546,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 311910812,
+      "step": 1816
+    },
+    {
+      "epoch": 0.47793779180640494,
+      "grad_norm": 8.864036975405911,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 312082820,
+      "step": 1817
+    },
+    {
+      "epoch": 0.47793779180640494,
+      "loss": 0.0714251697063446,
+      "loss_ce": 0.002028202638030052,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 312082820,
+      "step": 1817
+    },
+    {
+      "epoch": 0.47820082856579205,
+      "grad_norm": 4.5785017279302185,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 312255312,
+      "step": 1818
+    },
+    {
+      "epoch": 0.47820082856579205,
+      "loss": 0.1785389930009842,
+      "loss_ce": 0.0035817159805446863,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 312255312,
+      "step": 1818
+    },
+    {
+      "epoch": 0.4784638653251792,
+      "grad_norm": 9.103884045463392,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 312427552,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4784638653251792,
+      "loss": 0.09079764038324356,
+      "loss_ce": 0.0007707877666689456,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 312427552,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4787269020845663,
+      "grad_norm": 3.611814351166419,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 312599880,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4787269020845663,
+      "loss": 0.16918551921844482,
+      "loss_ce": 0.0004843563656322658,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 312599880,
+      "step": 1820
+    },
+    {
+      "epoch": 0.47898993884395347,
+      "grad_norm": 6.886471414344964,
+      "learning_rate": 5e-06,
+      "loss": 0.144,
+      "num_input_tokens_seen": 312771960,
+      "step": 1821
+    },
+    {
+      "epoch": 0.47898993884395347,
+      "loss": 0.1718224436044693,
+      "loss_ce": 0.0013817725703120232,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 312771960,
+      "step": 1821
+    },
+    {
+      "epoch": 0.47925297560334057,
+      "grad_norm": 12.670187653553288,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 312943884,
+      "step": 1822
+    },
+    {
+      "epoch": 0.47925297560334057,
+      "loss": 0.17194947600364685,
+      "loss_ce": 0.0042553916573524475,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 312943884,
+      "step": 1822
+    },
+    {
+      "epoch": 0.4795160123627277,
+      "grad_norm": 5.735564443277563,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 313115988,
+      "step": 1823
+    },
+    {
+      "epoch": 0.4795160123627277,
+      "loss": 0.11030334234237671,
+      "loss_ce": 0.00010437482706038281,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 313115988,
+      "step": 1823
+    },
+    {
+      "epoch": 0.47977904912211483,
+      "grad_norm": 5.327091023011001,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 313288456,
+      "step": 1824
+    },
+    {
+      "epoch": 0.47977904912211483,
+      "loss": 0.2672034204006195,
+      "loss_ce": 0.0009680833900347352,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 313288456,
+      "step": 1824
+    },
+    {
+      "epoch": 0.48004208588150193,
+      "grad_norm": 8.698833105847095,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 313460532,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48004208588150193,
+      "loss": 0.11867256462574005,
+      "loss_ce": 0.0026752520352602005,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 313460532,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48030512264088904,
+      "grad_norm": 4.784883256464839,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 313630464,
+      "step": 1826
+    },
+    {
+      "epoch": 0.48030512264088904,
+      "loss": 0.18028897047042847,
+      "loss_ce": 0.0021883829031139612,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 313630464,
+      "step": 1826
+    },
+    {
+      "epoch": 0.4805681594002762,
+      "grad_norm": 7.070087319121284,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 313802864,
+      "step": 1827
+    },
+    {
+      "epoch": 0.4805681594002762,
+      "loss": 0.19596196711063385,
+      "loss_ce": 0.0022058698814362288,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 313802864,
+      "step": 1827
+    },
+    {
+      "epoch": 0.4808311961596633,
+      "grad_norm": 6.746240646813248,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 313975036,
+      "step": 1828
+    },
+    {
+      "epoch": 0.4808311961596633,
+      "loss": 0.12373416870832443,
+      "loss_ce": 0.0009009129134938121,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 313975036,
+      "step": 1828
+    },
+    {
+      "epoch": 0.48109423291905046,
+      "grad_norm": 12.282763523774303,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 314147252,
+      "step": 1829
+    },
+    {
+      "epoch": 0.48109423291905046,
+      "loss": 0.09647711366415024,
+      "loss_ce": 0.0027576321735978127,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 314147252,
+      "step": 1829
+    },
+    {
+      "epoch": 0.48135726967843756,
+      "grad_norm": 6.5229587096602915,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 314317636,
+      "step": 1830
+    },
+    {
+      "epoch": 0.48135726967843756,
+      "loss": 0.11963652074337006,
+      "loss_ce": 0.004478443879634142,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 314317636,
+      "step": 1830
+    },
+    {
+      "epoch": 0.48162030643782466,
+      "grad_norm": 10.474914605426608,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 314489804,
+      "step": 1831
+    },
+    {
+      "epoch": 0.48162030643782466,
+      "loss": 0.08893167972564697,
+      "loss_ce": 0.0019260660046711564,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 314489804,
+      "step": 1831
+    },
+    {
+      "epoch": 0.4818833431972118,
+      "grad_norm": 4.796918355311461,
+      "learning_rate": 5e-06,
+      "loss": 0.1701,
+      "num_input_tokens_seen": 314661804,
+      "step": 1832
+    },
+    {
+      "epoch": 0.4818833431972118,
+      "loss": 0.18161356449127197,
+      "loss_ce": 0.0023533080238848925,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 314661804,
+      "step": 1832
+    },
+    {
+      "epoch": 0.4821463799565989,
+      "grad_norm": 6.244958331343755,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 314834036,
+      "step": 1833
+    },
+    {
+      "epoch": 0.4821463799565989,
+      "loss": 0.13272178173065186,
+      "loss_ce": 0.003785028588026762,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 314834036,
+      "step": 1833
+    },
+    {
+      "epoch": 0.4824094167159861,
+      "grad_norm": 7.896413991695191,
+      "learning_rate": 5e-06,
+      "loss": 0.1255,
+      "num_input_tokens_seen": 315004484,
+      "step": 1834
+    },
+    {
+      "epoch": 0.4824094167159861,
+      "loss": 0.12204363942146301,
+      "loss_ce": 0.0003090191457886249,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 315004484,
+      "step": 1834
+    },
+    {
+      "epoch": 0.4826724534753732,
+      "grad_norm": 7.031269313570851,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 315176780,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4826724534753732,
+      "loss": 0.12622271478176117,
+      "loss_ce": 0.002534976229071617,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 315176780,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4829354902347603,
+      "grad_norm": 6.072447688187852,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 315348656,
+      "step": 1836
+    },
+    {
+      "epoch": 0.4829354902347603,
+      "loss": 0.08488506823778152,
+      "loss_ce": 0.0010227651800960302,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 315348656,
+      "step": 1836
+    },
+    {
+      "epoch": 0.48319852699414745,
+      "grad_norm": 14.427685690814952,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 315520828,
+      "step": 1837
+    },
+    {
+      "epoch": 0.48319852699414745,
+      "loss": 0.0883752852678299,
+      "loss_ce": 0.00042362496606074274,
+      "loss_iou": 0.5625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 315520828,
+      "step": 1837
+    },
+    {
+      "epoch": 0.48346156375353455,
+      "grad_norm": 17.485755335788554,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 315692964,
+      "step": 1838
+    },
+    {
+      "epoch": 0.48346156375353455,
+      "loss": 0.10190844535827637,
+      "loss_ce": 0.0014140586135908961,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 315692964,
+      "step": 1838
+    },
+    {
+      "epoch": 0.48372460051292165,
+      "grad_norm": 9.721032690775148,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 315863376,
+      "step": 1839
+    },
+    {
+      "epoch": 0.48372460051292165,
+      "loss": 0.18402375280857086,
+      "loss_ce": 0.0028408921789377928,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 315863376,
+      "step": 1839
+    },
+    {
+      "epoch": 0.4839876372723088,
+      "grad_norm": 7.726124987543399,
+      "learning_rate": 5e-06,
+      "loss": 0.1728,
+      "num_input_tokens_seen": 316035520,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4839876372723088,
+      "loss": 0.13577872514724731,
+      "loss_ce": 0.00031119072809815407,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 316035520,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4842506740316959,
+      "grad_norm": 7.757576022533387,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 316207616,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4842506740316959,
+      "loss": 0.2208271026611328,
+      "loss_ce": 0.0022144389804452658,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 316207616,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4845137107910831,
+      "grad_norm": 11.993527938590784,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 316379548,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4845137107910831,
+      "loss": 0.17184340953826904,
+      "loss_ce": 0.002074118936434388,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 316379548,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4847767475504702,
+      "grad_norm": 5.880659481834313,
+      "learning_rate": 5e-06,
+      "loss": 0.1596,
+      "num_input_tokens_seen": 316551724,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4847767475504702,
+      "loss": 0.30887043476104736,
+      "loss_ce": 0.0007344337645918131,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 316551724,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4850397843098573,
+      "grad_norm": 9.164014647902235,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 316724140,
+      "step": 1844
+    },
+    {
+      "epoch": 0.4850397843098573,
+      "loss": 0.1765921413898468,
+      "loss_ce": 0.0012076160637661815,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 316724140,
+      "step": 1844
+    },
+    {
+      "epoch": 0.48530282106924444,
+      "grad_norm": 25.18318601638769,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 316896084,
+      "step": 1845
+    },
+    {
+      "epoch": 0.48530282106924444,
+      "loss": 0.14695365726947784,
+      "loss_ce": 0.0009270399459637702,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 316896084,
+      "step": 1845
+    },
+    {
+      "epoch": 0.48556585782863154,
+      "grad_norm": 10.73033101211756,
+      "learning_rate": 5e-06,
+      "loss": 0.183,
+      "num_input_tokens_seen": 317068004,
+      "step": 1846
+    },
+    {
+      "epoch": 0.48556585782863154,
+      "loss": 0.3143799304962158,
+      "loss_ce": 0.004962218925356865,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 317068004,
+      "step": 1846
+    },
+    {
+      "epoch": 0.4858288945880187,
+      "grad_norm": 14.757236176384808,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 317240256,
+      "step": 1847
+    },
+    {
+      "epoch": 0.4858288945880187,
+      "loss": 0.074017733335495,
+      "loss_ce": 0.0028812657110393047,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 317240256,
+      "step": 1847
+    },
+    {
+      "epoch": 0.4860919313474058,
+      "grad_norm": 5.470394923350598,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 317412376,
+      "step": 1848
+    },
+    {
+      "epoch": 0.4860919313474058,
+      "loss": 0.13736534118652344,
+      "loss_ce": 0.0006160617922432721,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 317412376,
+      "step": 1848
+    },
+    {
+      "epoch": 0.4863549681067929,
+      "grad_norm": 11.769346294638343,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 317584644,
+      "step": 1849
+    },
+    {
+      "epoch": 0.4863549681067929,
+      "loss": 0.07671768963336945,
+      "loss_ce": 0.002804110525175929,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 317584644,
+      "step": 1849
+    },
+    {
+      "epoch": 0.48661800486618007,
+      "grad_norm": 7.934603221234343,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 317756784,
+      "step": 1850
+    },
+    {
+      "epoch": 0.48661800486618007,
+      "loss": 0.1591799259185791,
+      "loss_ce": 0.0034792337100952864,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 317756784,
+      "step": 1850
+    },
+    {
+      "epoch": 0.48688104162556717,
+      "grad_norm": 5.949457674338729,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 317929000,
+      "step": 1851
+    },
+    {
+      "epoch": 0.48688104162556717,
+      "loss": 0.15429449081420898,
+      "loss_ce": 0.0003638358903117478,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 317929000,
+      "step": 1851
+    },
+    {
+      "epoch": 0.48714407838495427,
+      "grad_norm": 4.538326131208253,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 318101400,
+      "step": 1852
+    },
+    {
+      "epoch": 0.48714407838495427,
+      "loss": 0.14405781030654907,
+      "loss_ce": 0.004745060577988625,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 318101400,
+      "step": 1852
+    },
+    {
+      "epoch": 0.48740711514434143,
+      "grad_norm": 11.796493242549497,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 318273392,
+      "step": 1853
+    },
+    {
+      "epoch": 0.48740711514434143,
+      "loss": 0.05749022588133812,
+      "loss_ce": 0.0008801189833320677,
+      "loss_iou": 0.453125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 318273392,
+      "step": 1853
+    },
+    {
+      "epoch": 0.48767015190372853,
+      "grad_norm": 4.839316076441186,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 318444936,
+      "step": 1854
+    },
+    {
+      "epoch": 0.48767015190372853,
+      "loss": 0.15674933791160583,
+      "loss_ce": 0.0008960673003457487,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 318444936,
+      "step": 1854
+    },
+    {
+      "epoch": 0.4879331886631157,
+      "grad_norm": 4.489339979618492,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 318617236,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4879331886631157,
+      "loss": 0.17843472957611084,
+      "loss_ce": 0.001768482499755919,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 318617236,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4881962254225028,
+      "grad_norm": 5.676162725297575,
+      "learning_rate": 5e-06,
+      "loss": 0.178,
+      "num_input_tokens_seen": 318787468,
+      "step": 1856
+    },
+    {
+      "epoch": 0.4881962254225028,
+      "loss": 0.25485190749168396,
+      "loss_ce": 0.0009456594125367701,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 318787468,
+      "step": 1856
+    },
+    {
+      "epoch": 0.4884592621818899,
+      "grad_norm": 4.081201767827966,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 318957884,
+      "step": 1857
+    },
+    {
+      "epoch": 0.4884592621818899,
+      "loss": 0.0997210294008255,
+      "loss_ce": 0.00020320963812991977,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 318957884,
+      "step": 1857
+    },
+    {
+      "epoch": 0.48872229894127706,
+      "grad_norm": 23.818400579938295,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 319129856,
+      "step": 1858
+    },
+    {
+      "epoch": 0.48872229894127706,
+      "loss": 0.07231907546520233,
+      "loss_ce": 0.00011447950237197801,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 319129856,
+      "step": 1858
+    },
+    {
+      "epoch": 0.48898533570066416,
+      "grad_norm": 6.068112059865649,
+      "learning_rate": 5e-06,
+      "loss": 0.1742,
+      "num_input_tokens_seen": 319302108,
+      "step": 1859
+    },
+    {
+      "epoch": 0.48898533570066416,
+      "loss": 0.1806957721710205,
+      "loss_ce": 0.00039790940354578197,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 319302108,
+      "step": 1859
+    },
+    {
+      "epoch": 0.4892483724600513,
+      "grad_norm": 4.764918083593623,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 319473076,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4892483724600513,
+      "loss": 0.056661054491996765,
+      "loss_ce": 0.0034689174499362707,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 319473076,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4895114092194384,
+      "grad_norm": 4.591820064091911,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 319645292,
+      "step": 1861
+    },
+    {
+      "epoch": 0.4895114092194384,
+      "loss": 0.10820820182561874,
+      "loss_ce": 8.442218677373603e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 319645292,
+      "step": 1861
+    },
+    {
+      "epoch": 0.4897744459788255,
+      "grad_norm": 11.395874847747328,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 319817560,
+      "step": 1862
+    },
+    {
+      "epoch": 0.4897744459788255,
+      "loss": 0.1541745364665985,
+      "loss_ce": 0.0011288827518001199,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 319817560,
+      "step": 1862
+    },
+    {
+      "epoch": 0.4900374827382127,
+      "grad_norm": 8.74594499498662,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 319990008,
+      "step": 1863
+    },
+    {
+      "epoch": 0.4900374827382127,
+      "loss": 0.1368078887462616,
+      "loss_ce": 0.0013403687626123428,
+      "loss_iou": 0.5625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 319990008,
+      "step": 1863
+    },
+    {
+      "epoch": 0.4903005194975998,
+      "grad_norm": 4.454006905645775,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 320162308,
+      "step": 1864
+    },
+    {
+      "epoch": 0.4903005194975998,
+      "loss": 0.09877588599920273,
+      "loss_ce": 0.0007534276228398085,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 320162308,
+      "step": 1864
+    },
+    {
+      "epoch": 0.4905635562569869,
+      "grad_norm": 9.654102944086466,
+      "learning_rate": 5e-06,
+      "loss": 0.1351,
+      "num_input_tokens_seen": 320334456,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4905635562569869,
+      "loss": 0.09071889519691467,
+      "loss_ce": 0.0011192907113581896,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 320334456,
+      "step": 1865
+    },
+    {
+      "epoch": 0.49082659301637405,
+      "grad_norm": 7.49930478522843,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 320506784,
+      "step": 1866
+    },
+    {
+      "epoch": 0.49082659301637405,
+      "loss": 0.1436455398797989,
+      "loss_ce": 0.0007622435805387795,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 320506784,
+      "step": 1866
+    },
+    {
+      "epoch": 0.49108962977576115,
+      "grad_norm": 4.213229463422797,
+      "learning_rate": 5e-06,
+      "loss": 0.1405,
+      "num_input_tokens_seen": 320678800,
+      "step": 1867
+    },
+    {
+      "epoch": 0.49108962977576115,
+      "loss": 0.19276383519172668,
+      "loss_ce": 0.0017237972933799028,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 320678800,
+      "step": 1867
+    },
+    {
+      "epoch": 0.4913526665351483,
+      "grad_norm": 13.139928650685244,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 320851152,
+      "step": 1868
+    },
+    {
+      "epoch": 0.4913526665351483,
+      "loss": 0.13707411289215088,
+      "loss_ce": 0.0001722496235743165,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 320851152,
+      "step": 1868
+    },
+    {
+      "epoch": 0.4916157032945354,
+      "grad_norm": 10.621311625988106,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 321023324,
+      "step": 1869
+    },
+    {
+      "epoch": 0.4916157032945354,
+      "loss": 0.17078933119773865,
+      "loss_ce": 0.0011115875095129013,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 321023324,
+      "step": 1869
+    },
+    {
+      "epoch": 0.4918787400539225,
+      "grad_norm": 9.794878479481557,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 321195396,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4918787400539225,
+      "loss": 0.1828850954771042,
+      "loss_ce": 0.0016412028344348073,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 321195396,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4921417768133097,
+      "grad_norm": 3.6593406817873193,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 321367508,
+      "step": 1871
+    },
+    {
+      "epoch": 0.4921417768133097,
+      "loss": 0.07714089751243591,
+      "loss_ce": 0.0005722964997403324,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 321367508,
+      "step": 1871
+    },
+    {
+      "epoch": 0.4924048135726968,
+      "grad_norm": 3.9995708653462874,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 321538180,
+      "step": 1872
+    },
+    {
+      "epoch": 0.4924048135726968,
+      "loss": 0.10363311320543289,
+      "loss_ce": 0.004054257180541754,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 321538180,
+      "step": 1872
+    },
+    {
+      "epoch": 0.49266785033208393,
+      "grad_norm": 8.331617815221565,
+      "learning_rate": 5e-06,
+      "loss": 0.1423,
+      "num_input_tokens_seen": 321710632,
+      "step": 1873
+    },
+    {
+      "epoch": 0.49266785033208393,
+      "loss": 0.13812920451164246,
+      "loss_ce": 0.0021428640466183424,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 321710632,
+      "step": 1873
+    },
+    {
+      "epoch": 0.49293088709147104,
+      "grad_norm": 5.452515467611325,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 321882748,
+      "step": 1874
+    },
+    {
+      "epoch": 0.49293088709147104,
+      "loss": 0.14979197084903717,
+      "loss_ce": 0.002147932071238756,
+      "loss_iou": NaN,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 321882748,
+      "step": 1874
+    },
+    {
+      "epoch": 0.49319392385085814,
+      "grad_norm": 14.062855136219687,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 322054892,
+      "step": 1875
+    },
+    {
+      "epoch": 0.49319392385085814,
+      "loss": 0.24241477251052856,
+      "loss_ce": 0.0016921274363994598,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 322054892,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4934569606102453,
+      "grad_norm": 3.3198848239673966,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 322226992,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4934569606102453,
+      "loss": 0.06338231265544891,
+      "loss_ce": 0.0020724977366626263,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 322226992,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4937199973696324,
+      "grad_norm": 5.974832705883753,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 322399220,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4937199973696324,
+      "loss": 0.20875243842601776,
+      "loss_ce": 0.004009997006505728,
+      "loss_iou": 0.53125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 322399220,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4939830341290195,
+      "grad_norm": 11.103490307412729,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 322569372,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4939830341290195,
+      "loss": 0.10719159990549088,
+      "loss_ce": 0.006331001408398151,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 322569372,
+      "step": 1878
+    },
+    {
+      "epoch": 0.49424607088840666,
+      "grad_norm": 4.615361697814784,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 322741596,
+      "step": 1879
+    },
+    {
+      "epoch": 0.49424607088840666,
+      "loss": 0.13756033778190613,
+      "loss_ce": 0.002703155390918255,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 322741596,
+      "step": 1879
+    },
+    {
+      "epoch": 0.49450910764779377,
+      "grad_norm": 5.99195178330634,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 322913828,
+      "step": 1880
+    },
+    {
+      "epoch": 0.49450910764779377,
+      "loss": 0.20525991916656494,
+      "loss_ce": 0.0015550723765045404,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 322913828,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4947721444071809,
+      "grad_norm": 8.544161694628455,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 323084272,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4947721444071809,
+      "loss": 0.13442979753017426,
+      "loss_ce": 0.0013731509679928422,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 323084272,
+      "step": 1881
+    },
+    {
+      "epoch": 0.49503518116656803,
+      "grad_norm": 12.157441672743493,
+      "learning_rate": 5e-06,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 323254564,
+      "step": 1882
+    },
+    {
+      "epoch": 0.49503518116656803,
+      "loss": 0.06910552829504013,
+      "loss_ce": 0.00015105513739399612,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 323254564,
+      "step": 1882
+    },
+    {
+      "epoch": 0.49529821792595513,
+      "grad_norm": 7.758308533551838,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 323426936,
+      "step": 1883
+    },
+    {
+      "epoch": 0.49529821792595513,
+      "loss": 0.07692838460206985,
+      "loss_ce": 0.000985392602160573,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 323426936,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4955612546853423,
+      "grad_norm": 10.56929332370978,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 323599244,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4955612546853423,
+      "loss": 0.17598523199558258,
+      "loss_ce": 0.0006922531756572425,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 323599244,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4958242914447294,
+      "grad_norm": 4.870883244723278,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 323771396,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4958242914447294,
+      "loss": 0.08608455955982208,
+      "loss_ce": 0.0016424173954874277,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 323771396,
+      "step": 1885
+    },
+    {
+      "epoch": 0.49608732820411655,
+      "grad_norm": 9.112063415949194,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 323943464,
+      "step": 1886
+    },
+    {
+      "epoch": 0.49608732820411655,
+      "loss": 0.05756930261850357,
+      "loss_ce": 0.00025729500339366496,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 323943464,
+      "step": 1886
+    },
+    {
+      "epoch": 0.49635036496350365,
+      "grad_norm": 11.77050689912044,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 324115664,
+      "step": 1887
+    },
+    {
+      "epoch": 0.49635036496350365,
+      "loss": 0.13590523600578308,
+      "loss_ce": 0.0011243472108617425,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 324115664,
+      "step": 1887
+    },
+    {
+      "epoch": 0.49661340172289076,
+      "grad_norm": 19.144215192825804,
+      "learning_rate": 5e-06,
+      "loss": 0.154,
+      "num_input_tokens_seen": 324288076,
+      "step": 1888
+    },
+    {
+      "epoch": 0.49661340172289076,
+      "loss": 0.1540539562702179,
+      "loss_ce": 0.00036742445081472397,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 324288076,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4968764384822779,
+      "grad_norm": 4.245370979797126,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 324460724,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4968764384822779,
+      "loss": 0.08321575820446014,
+      "loss_ce": 0.0006657101330347359,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 324460724,
+      "step": 1889
+    },
+    {
+      "epoch": 0.497139475241665,
+      "grad_norm": 4.913753500635943,
+      "learning_rate": 5e-06,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 324633088,
+      "step": 1890
+    },
+    {
+      "epoch": 0.497139475241665,
+      "loss": 0.13971787691116333,
+      "loss_ce": 0.001625841949135065,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 324633088,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4974025120010521,
+      "grad_norm": 4.7392923084206275,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 324805420,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4974025120010521,
+      "loss": 0.10387594997882843,
+      "loss_ce": 0.001977758714929223,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 324805420,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4976655487604393,
+      "grad_norm": 3.934998355466595,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 324977628,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4976655487604393,
+      "loss": 0.08378120511770248,
+      "loss_ce": 0.0015668454580008984,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 324977628,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4979285855198264,
+      "grad_norm": 16.6043348209483,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 325148040,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4979285855198264,
+      "loss": 0.11267328262329102,
+      "loss_ce": 0.0008110918570309877,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 325148040,
+      "step": 1893
+    },
+    {
+      "epoch": 0.49819162227921354,
+      "grad_norm": 4.246925011279612,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 325317336,
+      "step": 1894
+    },
+    {
+      "epoch": 0.49819162227921354,
+      "loss": 0.20916813611984253,
+      "loss_ce": 0.000855154765304178,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 325317336,
+      "step": 1894
+    },
+    {
+      "epoch": 0.49845465903860064,
+      "grad_norm": 22.177021376541713,
+      "learning_rate": 5e-06,
+      "loss": 0.1381,
+      "num_input_tokens_seen": 325489652,
+      "step": 1895
+    },
+    {
+      "epoch": 0.49845465903860064,
+      "loss": 0.15358038246631622,
+      "loss_ce": 0.0006873153615742922,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 325489652,
+      "step": 1895
+    },
+    {
+      "epoch": 0.49871769579798775,
+      "grad_norm": 4.369246978330159,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 325661796,
+      "step": 1896
+    },
+    {
+      "epoch": 0.49871769579798775,
+      "loss": 0.11801205575466156,
+      "loss_ce": 0.0016180112725123763,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 325661796,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4989807325573749,
+      "grad_norm": 4.8736707793033975,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 325834028,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4989807325573749,
+      "loss": 0.09736193716526031,
+      "loss_ce": 0.0003770706243813038,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 325834028,
+      "step": 1897
+    },
+    {
+      "epoch": 0.499243769316762,
+      "grad_norm": 13.34167032829397,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 326006352,
+      "step": 1898
+    },
+    {
+      "epoch": 0.499243769316762,
+      "loss": 0.1529536247253418,
+      "loss_ce": 0.0012812747154384851,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 326006352,
+      "step": 1898
+    },
+    {
+      "epoch": 0.49950680607614917,
+      "grad_norm": 8.275559071074456,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 326178328,
+      "step": 1899
+    },
+    {
+      "epoch": 0.49950680607614917,
+      "loss": 0.040668413043022156,
+      "loss_ce": 0.0004920191713608801,
+      "loss_iou": 0.515625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 326178328,
+      "step": 1899
+    },
+    {
+      "epoch": 0.49976984283553627,
+      "grad_norm": 6.7636763663201105,
+      "learning_rate": 5e-06,
+      "loss": 0.146,
+      "num_input_tokens_seen": 326347792,
+      "step": 1900
+    },
+    {
+      "epoch": 0.49976984283553627,
+      "loss": 0.12708882987499237,
+      "loss_ce": 0.0006239861249923706,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 326347792,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5000328795949234,
+      "grad_norm": 11.91106437107254,
+      "learning_rate": 5e-06,
+      "loss": 0.1809,
+      "num_input_tokens_seen": 326520084,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5000328795949234,
+      "loss": 0.24568378925323486,
+      "loss_ce": 0.0030690436251461506,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 326520084,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5002959163543105,
+      "grad_norm": 15.380203467807723,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 326691988,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5002959163543105,
+      "loss": 0.1489952951669693,
+      "loss_ce": 0.00034418603172525764,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 326691988,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5005589531136977,
+      "grad_norm": 6.172818867405067,
+      "learning_rate": 5e-06,
+      "loss": 0.1834,
+      "num_input_tokens_seen": 326864500,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5005589531136977,
+      "loss": 0.08618461340665817,
+      "loss_ce": 0.0021086866036057472,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 326864500,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5008219898730848,
+      "grad_norm": 6.280435924544313,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 327036512,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5008219898730848,
+      "loss": 0.16495370864868164,
+      "loss_ce": 0.0016236326191574335,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 327036512,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5010850266324719,
+      "grad_norm": 5.592537029701005,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 327208668,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5010850266324719,
+      "loss": 0.09431658685207367,
+      "loss_ce": 0.0009938328294083476,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 327208668,
+      "step": 1905
+    },
+    {
+      "epoch": 0.501348063391859,
+      "grad_norm": 6.449445852698605,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 327378984,
+      "step": 1906
+    },
+    {
+      "epoch": 0.501348063391859,
+      "loss": 0.1146991178393364,
+      "loss_ce": 0.0006091539980843663,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 327378984,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5016111001512461,
+      "grad_norm": 4.708352961414117,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 327551376,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5016111001512461,
+      "loss": 0.15458180010318756,
+      "loss_ce": 0.0006816480308771133,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 327551376,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5018741369106332,
+      "grad_norm": 11.511613556933392,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 327724012,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5018741369106332,
+      "loss": 0.20126180350780487,
+      "loss_ce": 0.0008528655744157732,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 327724012,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5021371736700204,
+      "grad_norm": 5.154439626897641,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 327896420,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5021371736700204,
+      "loss": 0.13825711607933044,
+      "loss_ce": 0.0017519897082820535,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 327896420,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5024002104294075,
+      "grad_norm": 9.76818168107923,
+      "learning_rate": 5e-06,
+      "loss": 0.1211,
+      "num_input_tokens_seen": 328066120,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5024002104294075,
+      "loss": 0.20892378687858582,
+      "loss_ce": 0.00096173956990242,
+      "loss_iou": 0.609375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 328066120,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5026632471887946,
+      "grad_norm": 6.89128626408132,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 328238388,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5026632471887946,
+      "loss": 0.19616608321666718,
+      "loss_ce": 0.001982737798243761,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 328238388,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5029262839481817,
+      "grad_norm": 4.852078459217102,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 328410228,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5029262839481817,
+      "loss": 0.13402841985225677,
+      "loss_ce": 0.0019178204238414764,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 328410228,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5031893207075688,
+      "grad_norm": 8.495391184677569,
+      "learning_rate": 5e-06,
+      "loss": 0.1507,
+      "num_input_tokens_seen": 328582660,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5031893207075688,
+      "loss": 0.14262062311172485,
+      "loss_ce": 0.004253931809216738,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 328582660,
+      "step": 1913
+    },
+    {
+      "epoch": 0.503452357466956,
+      "grad_norm": 7.2140273397639385,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 328754772,
+      "step": 1914
+    },
+    {
+      "epoch": 0.503452357466956,
+      "loss": 0.09518692642450333,
+      "loss_ce": 0.003878333605825901,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 328754772,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5037153942263431,
+      "grad_norm": 4.268432766820409,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 328926900,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5037153942263431,
+      "loss": 0.11751651018857956,
+      "loss_ce": 0.0052423360757529736,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 328926900,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5039784309857303,
+      "grad_norm": 3.9588210706780522,
+      "learning_rate": 5e-06,
+      "loss": 0.1548,
+      "num_input_tokens_seen": 329096944,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5039784309857303,
+      "loss": 0.1656898707151413,
+      "loss_ce": 0.005258964374661446,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 329096944,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5042414677451174,
+      "grad_norm": 5.63809501756183,
+      "learning_rate": 5e-06,
+      "loss": 0.0966,
+      "num_input_tokens_seen": 329268892,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5042414677451174,
+      "loss": 0.07544635236263275,
+      "loss_ce": 0.000571468030102551,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 329268892,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5045045045045045,
+      "grad_norm": 7.590332140913489,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 329441344,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5045045045045045,
+      "loss": 0.15229275822639465,
+      "loss_ce": 0.006144077517092228,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 329441344,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5047675412638917,
+      "grad_norm": 9.59054902064166,
+      "learning_rate": 5e-06,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 329613564,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5047675412638917,
+      "loss": 0.08887225389480591,
+      "loss_ce": 0.001500429236330092,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 329613564,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5050305780232788,
+      "grad_norm": 4.210072070601249,
+      "learning_rate": 5e-06,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 329785680,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5050305780232788,
+      "loss": 0.2718814015388489,
+      "loss_ce": 0.002228060271590948,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 329785680,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5052936147826659,
+      "grad_norm": 8.380948003304292,
+      "learning_rate": 5e-06,
+      "loss": 0.1587,
+      "num_input_tokens_seen": 329957636,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5052936147826659,
+      "loss": 0.1278029978275299,
+      "loss_ce": 0.00286403251811862,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 329957636,
+      "step": 1921
+    },
+    {
+      "epoch": 0.505556651542053,
+      "grad_norm": 7.7625200359697715,
+      "learning_rate": 5e-06,
+      "loss": 0.1491,
+      "num_input_tokens_seen": 330129632,
+      "step": 1922
+    },
+    {
+      "epoch": 0.505556651542053,
+      "loss": 0.11747082322835922,
+      "loss_ce": 0.004799924790859222,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 330129632,
+      "step": 1922
+    },
+    {
+      "epoch": 0.5058196883014401,
+      "grad_norm": 6.9606844299684205,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 330301512,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5058196883014401,
+      "loss": 0.09333023428916931,
+      "loss_ce": 0.0011671524262055755,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 330301512,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5060827250608273,
+      "grad_norm": 13.729941886848659,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 330473972,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5060827250608273,
+      "loss": 0.09531684219837189,
+      "loss_ce": 0.000803898845333606,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 330473972,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5063457618202144,
+      "grad_norm": 6.989119579903773,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 330644424,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5063457618202144,
+      "loss": 0.17603036761283875,
+      "loss_ce": 0.0006458393763750792,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 330644424,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5066087985796015,
+      "grad_norm": 33.383727313004705,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 330816512,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5066087985796015,
+      "loss": 0.07775846868753433,
+      "loss_ce": 0.0003048558428417891,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 330816512,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5068718353389886,
+      "grad_norm": 7.427483033873037,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 330988952,
+      "step": 1927
+    },
+    {
+      "epoch": 0.5068718353389886,
+      "loss": 0.16052240133285522,
+      "loss_ce": 0.000579762679990381,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 330988952,
+      "step": 1927
+    },
+    {
+      "epoch": 0.5071348720983757,
+      "grad_norm": 8.482091734395144,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 331161228,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5071348720983757,
+      "loss": 0.14751845598220825,
+      "loss_ce": 0.000576329359319061,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 331161228,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5073979088577629,
+      "grad_norm": 6.087066534105735,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 331329832,
+      "step": 1929
+    },
+    {
+      "epoch": 0.5073979088577629,
+      "loss": 0.09639698266983032,
+      "loss_ce": 0.002372324001044035,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 331329832,
+      "step": 1929
+    },
+    {
+      "epoch": 0.50766094561715,
+      "grad_norm": 6.499679773970415,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 331502148,
+      "step": 1930
+    },
+    {
+      "epoch": 0.50766094561715,
+      "loss": 0.11911525577306747,
+      "loss_ce": 0.0002492967469152063,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 331502148,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5079239823765371,
+      "grad_norm": 16.477307771999516,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 331674540,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5079239823765371,
+      "loss": 0.14295902848243713,
+      "loss_ce": 0.0038904245011508465,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 331674540,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5081870191359242,
+      "grad_norm": 4.977000748222245,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 331846988,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5081870191359242,
+      "loss": 0.1609978973865509,
+      "loss_ce": 0.00042966773617081344,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 331846988,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5084500558953113,
+      "grad_norm": 4.68697406244425,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 332019216,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5084500558953113,
+      "loss": 0.12378443777561188,
+      "loss_ce": 0.0009511768585070968,
+      "loss_iou": 0.5,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 332019216,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5087130926546984,
+      "grad_norm": 5.061790101979302,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 332191420,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5087130926546984,
+      "loss": 0.09300635755062103,
+      "loss_ce": 0.0008737844182178378,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 332191420,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5089761294140857,
+      "grad_norm": 18.6525779104533,
+      "learning_rate": 5e-06,
+      "loss": 0.1468,
+      "num_input_tokens_seen": 332363456,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5089761294140857,
+      "loss": 0.13239170610904694,
+      "loss_ce": 0.0003726637805812061,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 332363456,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5092391661734728,
+      "grad_norm": 16.50490516998537,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 332535912,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5092391661734728,
+      "loss": 0.08137423545122147,
+      "loss_ce": 0.0015249941498041153,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 332535912,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5095022029328599,
+      "grad_norm": 7.868678191164257,
+      "learning_rate": 5e-06,
+      "loss": 0.1762,
+      "num_input_tokens_seen": 332708064,
+      "step": 1937
+    },
+    {
+      "epoch": 0.5095022029328599,
+      "loss": 0.2602325677871704,
+      "loss_ce": 0.005715976003557444,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 332708064,
+      "step": 1937
+    },
+    {
+      "epoch": 0.509765239692247,
+      "grad_norm": 4.650960242488344,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 332880260,
+      "step": 1938
+    },
+    {
+      "epoch": 0.509765239692247,
+      "loss": 0.08695125579833984,
+      "loss_ce": 0.0018987648654729128,
+      "loss_iou": 0.7578125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 332880260,
+      "step": 1938
+    },
+    {
+      "epoch": 0.5100282764516341,
+      "grad_norm": 5.979900826127572,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 333052492,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5100282764516341,
+      "loss": 0.11689235270023346,
+      "loss_ce": 0.0012612489517778158,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 333052492,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5102913132110213,
+      "grad_norm": 15.330918281866088,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 333222132,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5102913132110213,
+      "loss": 0.16124433279037476,
+      "loss_ce": 0.0012101602042093873,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 333222132,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5105543499704084,
+      "grad_norm": 5.298876176442514,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 333394244,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5105543499704084,
+      "loss": 0.09125322848558426,
+      "loss_ce": 0.0003718816442415118,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 333394244,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5108173867297955,
+      "grad_norm": 25.903681591063247,
+      "learning_rate": 5e-06,
+      "loss": 0.0986,
+      "num_input_tokens_seen": 333564616,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5108173867297955,
+      "loss": 0.0830872505903244,
+      "loss_ce": 0.003009122796356678,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 333564616,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5110804234891826,
+      "grad_norm": 7.412064312390374,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 333736828,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5110804234891826,
+      "loss": 0.13648918271064758,
+      "loss_ce": 0.0012352685444056988,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 333736828,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5113434602485697,
+      "grad_norm": 5.823332421421152,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 333909000,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5113434602485697,
+      "loss": 0.08375194668769836,
+      "loss_ce": 0.00024059813586063683,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 333909000,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5116064970079569,
+      "grad_norm": 5.7512491483697685,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 334081116,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5116064970079569,
+      "loss": 0.11631600558757782,
+      "loss_ce": 0.0020276757422834635,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 334081116,
+      "step": 1945
+    },
+    {
+      "epoch": 0.511869533767344,
+      "grad_norm": 6.061348306489617,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 334252872,
+      "step": 1946
+    },
+    {
+      "epoch": 0.511869533767344,
+      "loss": 0.1273353099822998,
+      "loss_ce": 0.0007484056986868382,
+      "loss_iou": 0.671875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 334252872,
+      "step": 1946
+    },
+    {
+      "epoch": 0.5121325705267311,
+      "grad_norm": 4.656625920812538,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 334421696,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5121325705267311,
+      "loss": 0.07545529305934906,
+      "loss_ce": 0.002487759804353118,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 334421696,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5123956072861182,
+      "grad_norm": 11.2997609374861,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 334593968,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5123956072861182,
+      "loss": 0.15095758438110352,
+      "loss_ce": 0.00047542020911350846,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 334593968,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5126586440455053,
+      "grad_norm": 4.740639931376309,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 334766256,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5126586440455053,
+      "loss": 0.0767994076013565,
+      "loss_ce": 0.00020028470316901803,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 334766256,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5129216808048925,
+      "grad_norm": 15.829229511064487,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 334938720,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5129216808048925,
+      "loss": 0.10283501446247101,
+      "loss_ce": 0.003225642256438732,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 334938720,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5131847175642796,
+      "grad_norm": 4.547176342272478,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 335108948,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5131847175642796,
+      "loss": 0.07713186740875244,
+      "loss_ce": 0.00047171738697215915,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 335108948,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5134477543236667,
+      "grad_norm": 13.66162052286571,
+      "learning_rate": 5e-06,
+      "loss": 0.1628,
+      "num_input_tokens_seen": 335281148,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5134477543236667,
+      "loss": 0.2089971899986267,
+      "loss_ce": 0.0048346007242798805,
+      "loss_iou": 0.390625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 335281148,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5137107910830538,
+      "grad_norm": 4.53549366428838,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 335453172,
+      "step": 1953
+    },
+    {
+      "epoch": 0.5137107910830538,
+      "loss": 0.0756625235080719,
+      "loss_ce": 0.0019930913113057613,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 335453172,
+      "step": 1953
+    },
+    {
+      "epoch": 0.513973827842441,
+      "grad_norm": 5.655145923562407,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 335623636,
+      "step": 1954
+    },
+    {
+      "epoch": 0.513973827842441,
+      "loss": 0.14977125823497772,
+      "loss_ce": 0.0032868883572518826,
+      "loss_iou": NaN,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 335623636,
+      "step": 1954
+    },
+    {
+      "epoch": 0.5142368646018282,
+      "grad_norm": 9.198223458900488,
+      "learning_rate": 5e-06,
+      "loss": 0.158,
+      "num_input_tokens_seen": 335793500,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5142368646018282,
+      "loss": 0.13681824505329132,
+      "loss_ce": 0.0008624300826340914,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 335793500,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5144999013612153,
+      "grad_norm": 14.576489713233544,
+      "learning_rate": 5e-06,
+      "loss": 0.1376,
+      "num_input_tokens_seen": 335965788,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5144999013612153,
+      "loss": 0.14106330275535583,
+      "loss_ce": 0.0012928071664646268,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 335965788,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5147629381206024,
+      "grad_norm": 12.321724088847931,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 336138060,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5147629381206024,
+      "loss": 0.1251417100429535,
+      "loss_ce": 0.0011182638118043542,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 336138060,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5150259748799895,
+      "grad_norm": 8.213839245926183,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 336308692,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5150259748799895,
+      "loss": 0.06178643926978111,
+      "loss_ce": 0.00011041228572139516,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 336308692,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5152890116393766,
+      "grad_norm": 4.514500271556744,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 336481248,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5152890116393766,
+      "loss": 0.15207350254058838,
+      "loss_ce": 0.0027662513311952353,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 336481248,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5155520483987637,
+      "grad_norm": 3.270020202020867,
+      "learning_rate": 5e-06,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 336653556,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5155520483987637,
+      "loss": 0.07474862039089203,
+      "loss_ce": 0.00031624053372070193,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 336653556,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5158150851581509,
+      "grad_norm": 5.036795139632983,
+      "learning_rate": 5e-06,
+      "loss": 0.1524,
+      "num_input_tokens_seen": 336825632,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5158150851581509,
+      "loss": 0.0949145182967186,
+      "loss_ce": 0.004765587393194437,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 336825632,
+      "step": 1961
+    },
+    {
+      "epoch": 0.516078121917538,
+      "grad_norm": 5.18788380960706,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 336997908,
+      "step": 1962
+    },
+    {
+      "epoch": 0.516078121917538,
+      "loss": 0.1347174346446991,
+      "loss_ce": 0.0003027569910045713,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 336997908,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5163411586769251,
+      "grad_norm": 19.722687183501755,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 337170032,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5163411586769251,
+      "loss": 0.15522822737693787,
+      "loss_ce": 0.004196731373667717,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 337170032,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5166041954363122,
+      "grad_norm": 5.156355195655944,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 337342000,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5166041954363122,
+      "loss": 0.0796535462141037,
+      "loss_ce": 0.0009487088536843657,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 337342000,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5168672321956993,
+      "grad_norm": 4.599075113473689,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 337514164,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5168672321956993,
+      "loss": 0.21942217648029327,
+      "loss_ce": 0.0008857909124344587,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 337514164,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5171302689550865,
+      "grad_norm": 8.512554957251284,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 337686236,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5171302689550865,
+      "loss": 0.08479119837284088,
+      "loss_ce": 0.001386661664582789,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 337686236,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5173933057144736,
+      "grad_norm": 5.381969767631673,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 337858240,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5173933057144736,
+      "loss": 0.15276584029197693,
+      "loss_ce": 0.004938698373734951,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 337858240,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5176563424738607,
+      "grad_norm": 12.700271834998917,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 338030560,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5176563424738607,
+      "loss": 0.15190255641937256,
+      "loss_ce": 0.0028852252289652824,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 338030560,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5179193792332478,
+      "grad_norm": 21.983637705034123,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 338202560,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5179193792332478,
+      "loss": 0.0981290340423584,
+      "loss_ce": 0.0023496169596910477,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 338202560,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5181824159926349,
+      "grad_norm": 10.469128023533157,
+      "learning_rate": 5e-06,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 338374836,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5181824159926349,
+      "loss": 0.16515450179576874,
+      "loss_ce": 0.0016413143603131175,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 338374836,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5184454527520221,
+      "grad_norm": 5.186485351415196,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 338546996,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5184454527520221,
+      "loss": 0.24401625990867615,
+      "loss_ce": 0.004056546837091446,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 338546996,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5187084895114092,
+      "grad_norm": 6.802546863240928,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 338716756,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5187084895114092,
+      "loss": 0.0805911123752594,
+      "loss_ce": 0.0030764644034206867,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 338716756,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5189715262707963,
+      "grad_norm": 11.859766109061349,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 338886988,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5189715262707963,
+      "loss": 0.10678447037935257,
+      "loss_ce": 0.0013767611235380173,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 338886988,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5192345630301834,
+      "grad_norm": 7.006309240426411,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 339059260,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5192345630301834,
+      "loss": 0.1808294951915741,
+      "loss_ce": 0.00034853501711040735,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 339059260,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5194975997895706,
+      "grad_norm": 21.985452701611596,
+      "learning_rate": 5e-06,
+      "loss": 0.1751,
+      "num_input_tokens_seen": 339231652,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5194975997895706,
+      "loss": 0.16675767302513123,
+      "loss_ce": 0.0018406773451715708,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 339231652,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5197606365489578,
+      "grad_norm": 6.071372766680396,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 339403696,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5197606365489578,
+      "loss": 0.08192337304353714,
+      "loss_ce": 0.0025166317354887724,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 339403696,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5200236733083449,
+      "grad_norm": 7.303457699958576,
+      "learning_rate": 5e-06,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 339576232,
+      "step": 1977
+    },
+    {
+      "epoch": 0.5200236733083449,
+      "loss": 0.06348910182714462,
+      "loss_ce": 0.0037356873508542776,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 339576232,
+      "step": 1977
+    },
+    {
+      "epoch": 0.520286710067732,
+      "grad_norm": 4.058844588183022,
+      "learning_rate": 5e-06,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 339748468,
+      "step": 1978
+    },
+    {
+      "epoch": 0.520286710067732,
+      "loss": 0.07935698330402374,
+      "loss_ce": 0.00019439200696069747,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 339748468,
+      "step": 1978
+    },
+    {
+      "epoch": 0.5205497468271191,
+      "grad_norm": 9.520705989391612,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 339920428,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5205497468271191,
+      "loss": 0.10209144651889801,
+      "loss_ce": 0.002482067793607712,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 339920428,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5208127835865062,
+      "grad_norm": 22.26578813252007,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 340092516,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5208127835865062,
+      "loss": 0.19446319341659546,
+      "loss_ce": 0.006719036493450403,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 340092516,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5210758203458934,
+      "grad_norm": 9.547756324115959,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 340265052,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5210758203458934,
+      "loss": 0.14365576207637787,
+      "loss_ce": 0.0007572055910713971,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 340265052,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5213388571052805,
+      "grad_norm": 7.840208454239884,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 340436896,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5213388571052805,
+      "loss": 0.11067777872085571,
+      "loss_ce": 0.0016995080513879657,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 340436896,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5216018938646676,
+      "grad_norm": 24.43574592347015,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 340609344,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5216018938646676,
+      "loss": 0.10227973759174347,
+      "loss_ce": 0.0038605397567152977,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 340609344,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5218649306240547,
+      "grad_norm": 6.254699637052052,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 340779700,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5218649306240547,
+      "loss": 0.1373625099658966,
+      "loss_ce": 0.0002012598270084709,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 340779700,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5221279673834418,
+      "grad_norm": 4.970529369812158,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 340952012,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5221279673834418,
+      "loss": 0.07744970917701721,
+      "loss_ce": 0.00027074594981968403,
+      "loss_iou": 0.625,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 340952012,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5223910041428289,
+      "grad_norm": 9.386513844488379,
+      "learning_rate": 5e-06,
+      "loss": 0.1525,
+      "num_input_tokens_seen": 341124172,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5223910041428289,
+      "loss": 0.16602635383605957,
+      "loss_ce": 0.0037033448461443186,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 341124172,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5226540409022161,
+      "grad_norm": 75.36425054276123,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 341296144,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5226540409022161,
+      "loss": 0.08925444632768631,
+      "loss_ce": 0.0014248627703636885,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 341296144,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5229170776616032,
+      "grad_norm": 6.0244434485136455,
+      "learning_rate": 5e-06,
+      "loss": 0.1624,
+      "num_input_tokens_seen": 341468128,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5229170776616032,
+      "loss": 0.18709853291511536,
+      "loss_ce": 0.0007887266110628843,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 341468128,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5231801144209903,
+      "grad_norm": 6.46446584669901,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 341640532,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5231801144209903,
+      "loss": 0.13015246391296387,
+      "loss_ce": 0.00030016410164535046,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 341640532,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5234431511803774,
+      "grad_norm": 12.345485884753447,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 341812700,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5234431511803774,
+      "loss": 0.10668568313121796,
+      "loss_ce": 0.00027087965281680226,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 341812700,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5237061879397645,
+      "grad_norm": 3.3770749440530894,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 341984700,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5237061879397645,
+      "loss": 0.1165972501039505,
+      "loss_ce": 0.0006304577691480517,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 341984700,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5239692246991517,
+      "grad_norm": 9.677414053666478,
+      "learning_rate": 5e-06,
+      "loss": 0.1048,
+      "num_input_tokens_seen": 342156884,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5239692246991517,
+      "loss": 0.1423652172088623,
+      "loss_ce": 0.0010383009212091565,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 342156884,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5242322614585389,
+      "grad_norm": 3.6750881151911643,
+      "learning_rate": 5e-06,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 342329140,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5242322614585389,
+      "loss": 0.18173760175704956,
+      "loss_ce": 0.009221725165843964,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 342329140,
+      "step": 1993
+    },
+    {
+      "epoch": 0.524495298217926,
+      "grad_norm": 19.882541327036233,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 342501472,
+      "step": 1994
+    },
+    {
+      "epoch": 0.524495298217926,
+      "loss": 0.0589841827750206,
+      "loss_ce": 0.001641655690036714,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 342501472,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5247583349773131,
+      "grad_norm": 16.32555909861509,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 342673528,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5247583349773131,
+      "loss": 0.18063244223594666,
+      "loss_ce": 0.0004871786804869771,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 342673528,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5250213717367002,
+      "grad_norm": 9.845315436294685,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 342845892,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5250213717367002,
+      "loss": 0.17416182160377502,
+      "loss_ce": 0.0006388599867932498,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 342845892,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5252844084960874,
+      "grad_norm": 4.531599732042389,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 343018040,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5252844084960874,
+      "loss": 0.15178070962429047,
+      "loss_ce": 0.0012069816002622247,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 343018040,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5255474452554745,
+      "grad_norm": 15.562861804573469,
+      "learning_rate": 5e-06,
+      "loss": 0.1677,
+      "num_input_tokens_seen": 343189936,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5255474452554745,
+      "loss": 0.1337474286556244,
+      "loss_ce": 0.0038340911269187927,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 343189936,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5258104820148616,
+      "grad_norm": 6.338186207451289,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 343361876,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5258104820148616,
+      "loss": 0.08655121922492981,
+      "loss_ce": 0.0022006274666637182,
+      "loss_iou": 0.5,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 343361876,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "grad_norm": 36.314036588641784,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_websight_new_CIoU": 0.8649432361125946,
+      "eval_websight_new_GIoU": 0.8667054772377014,
+      "eval_websight_new_IoU": 0.8704231679439545,
+      "eval_websight_new_MAE_all": 0.021235700696706772,
+      "eval_websight_new_MAE_h": 0.007067237980663776,
+      "eval_websight_new_MAE_w": 0.03388772998005152,
+      "eval_websight_new_MAE_x": 0.03554858546704054,
+      "eval_websight_new_MAE_y": 0.008439254947006702,
+      "eval_websight_new_NUM_probability": 0.999984085559845,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.10596824437379837,
+      "eval_websight_new_loss_ce": 6.8271494910732144e-06,
+      "eval_websight_new_loss_iou": 0.36328125,
+      "eval_websight_new_loss_num": 0.018640518188476562,
+      "eval_websight_new_loss_xval": 0.09316253662109375,
+      "eval_websight_new_runtime": 54.898,
+      "eval_websight_new_samples_per_second": 0.911,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_seeclick_CIoU": 0.6263805329799652,
+      "eval_seeclick_GIoU": 0.627049595117569,
+      "eval_seeclick_IoU": 0.6531890332698822,
+      "eval_seeclick_MAE_all": 0.050390077754855156,
+      "eval_seeclick_MAE_h": 0.02558732032775879,
+      "eval_seeclick_MAE_w": 0.07135490141808987,
+      "eval_seeclick_MAE_x": 0.07951905764639378,
+      "eval_seeclick_MAE_y": 0.025099032558500767,
+      "eval_seeclick_NUM_probability": 0.9999794960021973,
+      "eval_seeclick_inside_bbox": 0.921875,
+      "eval_seeclick_loss": 0.22126971185207367,
+      "eval_seeclick_loss_ce": 0.008990719448775053,
+      "eval_seeclick_loss_iou": 0.47021484375,
+      "eval_seeclick_loss_num": 0.04229736328125,
+      "eval_seeclick_loss_xval": 0.211334228515625,
+      "eval_seeclick_runtime": 78.1311,
+      "eval_seeclick_samples_per_second": 0.55,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_icons_CIoU": 0.8664080798625946,
+      "eval_icons_GIoU": 0.8631992936134338,
+      "eval_icons_IoU": 0.8702020049095154,
+      "eval_icons_MAE_all": 0.0180177534930408,
+      "eval_icons_MAE_h": 0.016653602942824364,
+      "eval_icons_MAE_w": 0.018555423244833946,
+      "eval_icons_MAE_x": 0.01882947515696287,
+      "eval_icons_MAE_y": 0.018032516352832317,
+      "eval_icons_NUM_probability": 0.9999754428863525,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.059605177491903305,
+      "eval_icons_loss_ce": 1.412479377904674e-05,
+      "eval_icons_loss_iou": 0.5333251953125,
+      "eval_icons_loss_num": 0.011335372924804688,
+      "eval_icons_loss_xval": 0.05667877197265625,
+      "eval_icons_runtime": 80.1161,
+      "eval_icons_samples_per_second": 0.624,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_screenspot_CIoU": 0.5386523008346558,
+      "eval_screenspot_GIoU": 0.5300994714101156,
+      "eval_screenspot_IoU": 0.5833619435628256,
+      "eval_screenspot_MAE_all": 0.08941345165173213,
+      "eval_screenspot_MAE_h": 0.04660519336660703,
+      "eval_screenspot_MAE_w": 0.1580625375111898,
+      "eval_screenspot_MAE_x": 0.10776859025160472,
+      "eval_screenspot_MAE_y": 0.04521748423576355,
+      "eval_screenspot_NUM_probability": 0.9995922644933065,
+      "eval_screenspot_inside_bbox": 0.850000003973643,
+      "eval_screenspot_loss": 0.8645088076591492,
+      "eval_screenspot_loss_ce": 0.5230478445688883,
+      "eval_screenspot_loss_iou": 0.4464925130208333,
+      "eval_screenspot_loss_num": 0.06682078043619792,
+      "eval_screenspot_loss_xval": 0.3342692057291667,
+      "eval_screenspot_runtime": 148.9183,
+      "eval_screenspot_samples_per_second": 0.598,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "loss": 0.8609392046928406,
+      "loss_ce": 0.5118181109428406,
+      "loss_iou": 0.390625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5263365555336358,
+      "grad_norm": 8.429330373903323,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 343704524,
+      "step": 2001
+    },
+    {
+      "epoch": 0.5263365555336358,
+      "loss": 0.21621274948120117,
+      "loss_ce": 0.0006976150907576084,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 343704524,
+      "step": 2001
+    },
+    {
+      "epoch": 0.526599592293023,
+      "grad_norm": 16.618545138422856,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 343876532,
+      "step": 2002
+    },
+    {
+      "epoch": 0.526599592293023,
+      "loss": 0.14034898579120636,
+      "loss_ce": 0.0025926402304321527,
+      "loss_iou": 0.515625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 343876532,
+      "step": 2002
+    },
+    {
+      "epoch": 0.5268626290524101,
+      "grad_norm": 7.347451448031887,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 344048440,
+      "step": 2003
+    },
+    {
+      "epoch": 0.5268626290524101,
+      "loss": 0.11765069514513016,
+      "loss_ce": 0.0015160476323217154,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 344048440,
+      "step": 2003
+    },
+    {
+      "epoch": 0.5271256658117972,
+      "grad_norm": 11.90854860345847,
+      "learning_rate": 5e-06,
+      "loss": 0.1657,
+      "num_input_tokens_seen": 344220524,
+      "step": 2004
+    },
+    {
+      "epoch": 0.5271256658117972,
+      "loss": 0.09579437971115112,
+      "loss_ce": 0.0003353926877025515,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 344220524,
+      "step": 2004
+    },
+    {
+      "epoch": 0.5273887025711843,
+      "grad_norm": 15.406217883818895,
+      "learning_rate": 5e-06,
+      "loss": 0.1745,
+      "num_input_tokens_seen": 344391120,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5273887025711843,
+      "loss": 0.08335284888744354,
+      "loss_ce": 0.005258369259536266,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 344391120,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5276517393305714,
+      "grad_norm": 6.625644981750321,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 344563368,
+      "step": 2006
+    },
+    {
+      "epoch": 0.5276517393305714,
+      "loss": 0.1281585693359375,
+      "loss_ce": 0.0008697626180946827,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 344563368,
+      "step": 2006
+    },
+    {
+      "epoch": 0.5279147760899586,
+      "grad_norm": 5.116207808263436,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 344735272,
+      "step": 2007
+    },
+    {
+      "epoch": 0.5279147760899586,
+      "loss": 0.10428653657436371,
+      "loss_ce": 0.00014529118197970092,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 344735272,
+      "step": 2007
+    },
+    {
+      "epoch": 0.5281778128493457,
+      "grad_norm": 5.758545412441202,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 344907448,
+      "step": 2008
+    },
+    {
+      "epoch": 0.5281778128493457,
+      "loss": 0.07528108358383179,
+      "loss_ce": 0.0017337151803076267,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 344907448,
+      "step": 2008
+    },
+    {
+      "epoch": 0.5284408496087328,
+      "grad_norm": 5.079743263348732,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 345080020,
+      "step": 2009
+    },
+    {
+      "epoch": 0.5284408496087328,
+      "loss": 0.06423554569482803,
+      "loss_ce": 0.0006674337200820446,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 345080020,
+      "step": 2009
+    },
+    {
+      "epoch": 0.5287038863681199,
+      "grad_norm": 6.791724647885975,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 345251684,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5287038863681199,
+      "loss": 0.13502314686775208,
+      "loss_ce": 0.0006847689510323107,
+      "loss_iou": 0.484375,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 345251684,
+      "step": 2010
+    },
+    {
+      "epoch": 0.528966923127507,
+      "grad_norm": 7.5731259754899245,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 345421972,
+      "step": 2011
+    },
+    {
+      "epoch": 0.528966923127507,
+      "loss": 0.10425636172294617,
+      "loss_ce": 0.0018088552169501781,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 345421972,
+      "step": 2011
+    },
+    {
+      "epoch": 0.5292299598868941,
+      "grad_norm": 4.024834656974236,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 345594140,
+      "step": 2012
+    },
+    {
+      "epoch": 0.5292299598868941,
+      "loss": 0.05266657471656799,
+      "loss_ce": 0.00011530861956998706,
+      "loss_iou": 0.53125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 345594140,
+      "step": 2012
+    },
+    {
+      "epoch": 0.5294929966462814,
+      "grad_norm": 4.913874816205936,
+      "learning_rate": 5e-06,
+      "loss": 0.1334,
+      "num_input_tokens_seen": 345766220,
+      "step": 2013
+    },
+    {
+      "epoch": 0.5294929966462814,
+      "loss": 0.08730873465538025,
+      "loss_ce": 0.0039957487024366856,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 345766220,
+      "step": 2013
+    },
+    {
+      "epoch": 0.5297560334056685,
+      "grad_norm": 5.534989831922425,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 345938392,
+      "step": 2014
+    },
+    {
+      "epoch": 0.5297560334056685,
+      "loss": 0.11161148548126221,
+      "loss_ce": 0.0032130456529557705,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 345938392,
+      "step": 2014
+    },
+    {
+      "epoch": 0.5300190701650556,
+      "grad_norm": 7.209865250975092,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 346110656,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5300190701650556,
+      "loss": 0.12729424238204956,
+      "loss_ce": 0.0035759946331381798,
+      "loss_iou": 0.625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 346110656,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5302821069244427,
+      "grad_norm": 5.146240130208517,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 346283032,
+      "step": 2016
+    },
+    {
+      "epoch": 0.5302821069244427,
+      "loss": 0.12445038557052612,
+      "loss_ce": 0.0015866123139858246,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 346283032,
+      "step": 2016
+    },
+    {
+      "epoch": 0.5305451436838298,
+      "grad_norm": 4.097776262787885,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 346455396,
+      "step": 2017
+    },
+    {
+      "epoch": 0.5305451436838298,
+      "loss": 0.08194537460803986,
+      "loss_ce": 0.0027369949966669083,
+      "loss_iou": 0.46875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 346455396,
+      "step": 2017
+    },
+    {
+      "epoch": 0.530808180443217,
+      "grad_norm": 8.807636765638556,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 346627308,
+      "step": 2018
+    },
+    {
+      "epoch": 0.530808180443217,
+      "loss": 0.16153287887573242,
+      "loss_ce": 0.0020022375974804163,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 346627308,
+      "step": 2018
+    },
+    {
+      "epoch": 0.5310712172026041,
+      "grad_norm": 10.071725544370869,
+      "learning_rate": 5e-06,
+      "loss": 0.1096,
+      "num_input_tokens_seen": 346799504,
+      "step": 2019
+    },
+    {
+      "epoch": 0.5310712172026041,
+      "loss": 0.10446594655513763,
+      "loss_ce": 0.0008282391354441643,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 346799504,
+      "step": 2019
+    },
+    {
+      "epoch": 0.5313342539619912,
+      "grad_norm": 5.791284152439371,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 346971328,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5313342539619912,
+      "loss": 0.21368807554244995,
+      "loss_ce": 0.002475916873663664,
+      "loss_iou": 0.578125,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 346971328,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5315972907213783,
+      "grad_norm": 9.778332313218632,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 347139872,
+      "step": 2021
+    },
+    {
+      "epoch": 0.5315972907213783,
+      "loss": 0.2882145047187805,
+      "loss_ce": 0.0021427052561193705,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 347139872,
+      "step": 2021
+    },
+    {
+      "epoch": 0.5318603274807654,
+      "grad_norm": 7.217119694281503,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 347312176,
+      "step": 2022
+    },
+    {
+      "epoch": 0.5318603274807654,
+      "loss": 0.14617526531219482,
+      "loss_ce": 0.0012472879607230425,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 347312176,
+      "step": 2022
+    },
+    {
+      "epoch": 0.5321233642401526,
+      "grad_norm": 5.444211224879485,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 347483864,
+      "step": 2023
+    },
+    {
+      "epoch": 0.5321233642401526,
+      "loss": 0.15533965826034546,
+      "loss_ce": 0.0009817371610552073,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 347483864,
+      "step": 2023
+    },
+    {
+      "epoch": 0.5323864009995397,
+      "grad_norm": 5.88671307787278,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 347654444,
+      "step": 2024
+    },
+    {
+      "epoch": 0.5323864009995397,
+      "loss": 0.10597267001867294,
+      "loss_ce": 0.0017551433993503451,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 347654444,
+      "step": 2024
+    },
+    {
+      "epoch": 0.5326494377589268,
+      "grad_norm": 4.540449525857699,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 347826732,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5326494377589268,
+      "loss": 0.10671254992485046,
+      "loss_ce": 0.002159323776140809,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 347826732,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5329124745183139,
+      "grad_norm": 12.403111361726802,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 347995312,
+      "step": 2026
+    },
+    {
+      "epoch": 0.5329124745183139,
+      "loss": 0.1710810512304306,
+      "loss_ce": 0.00019787647761404514,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 347995312,
+      "step": 2026
+    },
+    {
+      "epoch": 0.533175511277701,
+      "grad_norm": 4.826031974791996,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 348167268,
+      "step": 2027
+    },
+    {
+      "epoch": 0.533175511277701,
+      "loss": 0.14301443099975586,
+      "loss_ce": 0.004235736560076475,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 348167268,
+      "step": 2027
+    },
+    {
+      "epoch": 0.5334385480370882,
+      "grad_norm": 5.858834044093049,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 348339324,
+      "step": 2028
+    },
+    {
+      "epoch": 0.5334385480370882,
+      "loss": 0.13567912578582764,
+      "loss_ce": 0.0013865029904991388,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 348339324,
+      "step": 2028
+    },
+    {
+      "epoch": 0.5337015847964753,
+      "grad_norm": 18.687645390034255,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 348511688,
+      "step": 2029
+    },
+    {
+      "epoch": 0.5337015847964753,
+      "loss": 0.09527582675218582,
+      "loss_ce": 0.0008391792071051896,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 348511688,
+      "step": 2029
+    },
+    {
+      "epoch": 0.5339646215558624,
+      "grad_norm": 13.127056473400154,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 348682072,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5339646215558624,
+      "loss": 0.10760138183832169,
+      "loss_ce": 0.003078682580962777,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 348682072,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5342276583152495,
+      "grad_norm": 4.612341549056293,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 348854384,
+      "step": 2031
+    },
+    {
+      "epoch": 0.5342276583152495,
+      "loss": 0.12830308079719543,
+      "loss_ce": 0.004981548525393009,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 348854384,
+      "step": 2031
+    },
+    {
+      "epoch": 0.5344906950746366,
+      "grad_norm": 8.941916447828152,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 349026540,
+      "step": 2032
+    },
+    {
+      "epoch": 0.5344906950746366,
+      "loss": 0.11764685809612274,
+      "loss_ce": 0.0009781570406630635,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 349026540,
+      "step": 2032
+    },
+    {
+      "epoch": 0.5347537318340239,
+      "grad_norm": 3.1856540562472104,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 349195704,
+      "step": 2033
+    },
+    {
+      "epoch": 0.5347537318340239,
+      "loss": 0.15431983768939972,
+      "loss_ce": 0.001899797236546874,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 349195704,
+      "step": 2033
+    },
+    {
+      "epoch": 0.535016768593411,
+      "grad_norm": 6.665681138831115,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 349367820,
+      "step": 2034
+    },
+    {
+      "epoch": 0.535016768593411,
+      "loss": 0.17121072113513947,
+      "loss_ce": 0.001441433560103178,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 349367820,
+      "step": 2034
+    },
+    {
+      "epoch": 0.5352798053527981,
+      "grad_norm": 5.921520862992091,
+      "learning_rate": 5e-06,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 349540064,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5352798053527981,
+      "loss": 0.102360799908638,
+      "loss_ce": 0.0002947567554656416,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 349540064,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5355428421121852,
+      "grad_norm": 5.3538979358096785,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 349710408,
+      "step": 2036
+    },
+    {
+      "epoch": 0.5355428421121852,
+      "loss": 0.06293447315692902,
+      "loss_ce": 0.0006481010350398719,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 349710408,
+      "step": 2036
+    },
+    {
+      "epoch": 0.5358058788715723,
+      "grad_norm": 11.23909243130062,
+      "learning_rate": 5e-06,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 349880916,
+      "step": 2037
+    },
+    {
+      "epoch": 0.5358058788715723,
+      "loss": 0.1736185997724533,
+      "loss_ce": 0.003116899635642767,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 349880916,
+      "step": 2037
+    },
+    {
+      "epoch": 0.5360689156309594,
+      "grad_norm": 11.327258827718145,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 350053024,
+      "step": 2038
+    },
+    {
+      "epoch": 0.5360689156309594,
+      "loss": 0.14642956852912903,
+      "loss_ce": 0.0015015878016129136,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 350053024,
+      "step": 2038
+    },
+    {
+      "epoch": 0.5363319523903466,
+      "grad_norm": 10.572751348187758,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 350225056,
+      "step": 2039
+    },
+    {
+      "epoch": 0.5363319523903466,
+      "loss": 0.08779959380626678,
+      "loss_ce": 0.0008397561614401639,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 350225056,
+      "step": 2039
+    },
+    {
+      "epoch": 0.5365949891497337,
+      "grad_norm": 4.830573829484519,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 350395628,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5365949891497337,
+      "loss": 0.15380313992500305,
+      "loss_ce": 0.0012152513954788446,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 350395628,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5368580259091208,
+      "grad_norm": 6.867013774487234,
+      "learning_rate": 5e-06,
+      "loss": 0.1852,
+      "num_input_tokens_seen": 350567756,
+      "step": 2041
+    },
+    {
+      "epoch": 0.5368580259091208,
+      "loss": 0.1892320066690445,
+      "loss_ce": 0.00420394167304039,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 350567756,
+      "step": 2041
+    },
+    {
+      "epoch": 0.5371210626685079,
+      "grad_norm": 9.243159752879412,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 350739840,
+      "step": 2042
+    },
+    {
+      "epoch": 0.5371210626685079,
+      "loss": 0.1004796102643013,
+      "loss_ce": 0.00013781688176095486,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 350739840,
+      "step": 2042
+    },
+    {
+      "epoch": 0.537384099427895,
+      "grad_norm": 11.965655673255789,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 350912084,
+      "step": 2043
+    },
+    {
+      "epoch": 0.537384099427895,
+      "loss": 0.20747891068458557,
+      "loss_ce": 0.0016378372674807906,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 350912084,
+      "step": 2043
+    },
+    {
+      "epoch": 0.5376471361872822,
+      "grad_norm": 3.1288535063063327,
+      "learning_rate": 5e-06,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 351081480,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5376471361872822,
+      "loss": 0.1123102456331253,
+      "loss_ce": 0.001989200245589018,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 351081480,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5379101729466693,
+      "grad_norm": 4.103876123324951,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 351253596,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5379101729466693,
+      "loss": 0.11436055600643158,
+      "loss_ce": 0.0011403337121009827,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 351253596,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5381732097060564,
+      "grad_norm": 5.468839567129726,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 351425656,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5381732097060564,
+      "loss": 0.0928923487663269,
+      "loss_ce": 0.0006071930401958525,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 351425656,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5384362464654435,
+      "grad_norm": 14.247562489878419,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 351597816,
+      "step": 2047
+    },
+    {
+      "epoch": 0.5384362464654435,
+      "loss": 0.2033635675907135,
+      "loss_ce": 0.0019475510343909264,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 351597816,
+      "step": 2047
+    },
+    {
+      "epoch": 0.5386992832248306,
+      "grad_norm": 4.3738903937509885,
+      "learning_rate": 5e-06,
+      "loss": 0.1305,
+      "num_input_tokens_seen": 351769980,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5386992832248306,
+      "loss": 0.059591565281152725,
+      "loss_ce": 8.22915681055747e-05,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 351769980,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5389623199842178,
+      "grad_norm": 4.415343795853702,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 351939320,
+      "step": 2049
+    },
+    {
+      "epoch": 0.5389623199842178,
+      "loss": 0.24034851789474487,
+      "loss_ce": 0.0008465623832307756,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 351939320,
+      "step": 2049
+    },
+    {
+      "epoch": 0.5392253567436049,
+      "grad_norm": 16.797461240634142,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 352111376,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5392253567436049,
+      "loss": 0.14554640650749207,
+      "loss_ce": 0.0012287711724638939,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 352111376,
+      "step": 2050
+    },
+    {
+      "epoch": 0.539488393502992,
+      "grad_norm": 7.010827810107144,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 352283872,
+      "step": 2051
+    },
+    {
+      "epoch": 0.539488393502992,
+      "loss": 0.1082100197672844,
+      "loss_ce": 0.0016121190274134278,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 352283872,
+      "step": 2051
+    },
+    {
+      "epoch": 0.5397514302623792,
+      "grad_norm": 5.086953515153334,
+      "learning_rate": 5e-06,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 352456124,
+      "step": 2052
+    },
+    {
+      "epoch": 0.5397514302623792,
+      "loss": 0.05017915368080139,
+      "loss_ce": 0.0004965342814102769,
+      "loss_iou": 0.53125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 352456124,
+      "step": 2052
+    },
+    {
+      "epoch": 0.5400144670217663,
+      "grad_norm": 9.025520856951152,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 352626592,
+      "step": 2053
+    },
+    {
+      "epoch": 0.5400144670217663,
+      "loss": 0.16316679120063782,
+      "loss_ce": 0.0005996549734845757,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 352626592,
+      "step": 2053
+    },
+    {
+      "epoch": 0.5402775037811535,
+      "grad_norm": 6.628974819075429,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 352798532,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5402775037811535,
+      "loss": 0.10173699259757996,
+      "loss_ce": 0.002814263803884387,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 352798532,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 4.016489890807633,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 352970876,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "loss": 0.12183534353971481,
+      "loss_ce": 0.00010072031000163406,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 352970876,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5408035772999277,
+      "grad_norm": 10.022531221280579,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 353143148,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5408035772999277,
+      "loss": 0.11547866463661194,
+      "loss_ce": 0.0016328342026099563,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 353143148,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5410666140593148,
+      "grad_norm": 4.0813858316372205,
+      "learning_rate": 5e-06,
+      "loss": 0.1298,
+      "num_input_tokens_seen": 353315092,
+      "step": 2057
+    },
+    {
+      "epoch": 0.5410666140593148,
+      "loss": 0.2096938192844391,
+      "loss_ce": 0.005348118022084236,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 353315092,
+      "step": 2057
+    },
+    {
+      "epoch": 0.5413296508187019,
+      "grad_norm": 6.060936634204627,
+      "learning_rate": 5e-06,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 353487316,
+      "step": 2058
+    },
+    {
+      "epoch": 0.5413296508187019,
+      "loss": 0.0933343917131424,
+      "loss_ce": 0.0019037279998883605,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 353487316,
+      "step": 2058
+    },
+    {
+      "epoch": 0.541592687578089,
+      "grad_norm": 5.183981355840974,
+      "learning_rate": 5e-06,
+      "loss": 0.1516,
+      "num_input_tokens_seen": 353659672,
+      "step": 2059
+    },
+    {
+      "epoch": 0.541592687578089,
+      "loss": 0.1340794712305069,
+      "loss_ce": 0.001663701143115759,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 353659672,
+      "step": 2059
+    },
+    {
+      "epoch": 0.5418557243374762,
+      "grad_norm": 4.266501842679085,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 353829984,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5418557243374762,
+      "loss": 0.16655325889587402,
+      "loss_ce": 0.0011174663668498397,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 353829984,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5421187610968633,
+      "grad_norm": 5.044214417820774,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 354002080,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5421187610968633,
+      "loss": 0.11388491839170456,
+      "loss_ce": 0.0009240994695574045,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 354002080,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5423817978562504,
+      "grad_norm": 5.933256816179162,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 354174264,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5423817978562504,
+      "loss": 0.07782945036888123,
+      "loss_ce": 0.00034532046993263066,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 354174264,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5426448346156375,
+      "grad_norm": 6.3010547634359915,
+      "learning_rate": 5e-06,
+      "loss": 0.1807,
+      "num_input_tokens_seen": 354346228,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5426448346156375,
+      "loss": 0.2516539692878723,
+      "loss_ce": 0.0021422426216304302,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 354346228,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5429078713750246,
+      "grad_norm": 6.370139521144386,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 354516236,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5429078713750246,
+      "loss": 0.1860380321741104,
+      "loss_ce": 0.0007352972170338035,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 354516236,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5431709081344118,
+      "grad_norm": 5.277591681669997,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 354688188,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5431709081344118,
+      "loss": 0.14346206188201904,
+      "loss_ce": 0.0007008376996964216,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 354688188,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5434339448937989,
+      "grad_norm": 4.479693164703334,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 354860592,
+      "step": 2066
+    },
+    {
+      "epoch": 0.5434339448937989,
+      "loss": 0.11085185408592224,
+      "loss_ce": 7.304361497517675e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 354860592,
+      "step": 2066
+    },
+    {
+      "epoch": 0.543696981653186,
+      "grad_norm": 4.919682261722731,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 355033176,
+      "step": 2067
+    },
+    {
+      "epoch": 0.543696981653186,
+      "loss": 0.16546472907066345,
+      "loss_ce": 0.001188602764159441,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 355033176,
+      "step": 2067
+    },
+    {
+      "epoch": 0.5439600184125731,
+      "grad_norm": 13.936900281461735,
+      "learning_rate": 5e-06,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 355205536,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5439600184125731,
+      "loss": 0.08491555601358414,
+      "loss_ce": 1.5654470189474523e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 355205536,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5442230551719602,
+      "grad_norm": 7.496864557637642,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 355375248,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5442230551719602,
+      "loss": 0.11474957317113876,
+      "loss_ce": 0.005954409018158913,
+      "loss_iou": 0.671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 355375248,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5444860919313474,
+      "grad_norm": 7.5192560384682325,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 355545380,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5444860919313474,
+      "loss": 0.156136155128479,
+      "loss_ce": 0.0023580677807331085,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 355545380,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5447491286907346,
+      "grad_norm": 5.871220250123853,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 355717808,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5447491286907346,
+      "loss": 0.1303853988647461,
+      "loss_ce": 0.0030660659540444613,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 355717808,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5450121654501217,
+      "grad_norm": 5.18139773459563,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 355889920,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5450121654501217,
+      "loss": 0.08777904510498047,
+      "loss_ce": 0.0009260187507607043,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 355889920,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5452752022095088,
+      "grad_norm": 5.534523394507571,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 356062284,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5452752022095088,
+      "loss": 0.07954730838537216,
+      "loss_ce": 0.0016664512222632766,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 356062284,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5455382389688959,
+      "grad_norm": 5.798318301378571,
+      "learning_rate": 5e-06,
+      "loss": 0.13,
+      "num_input_tokens_seen": 356234572,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5455382389688959,
+      "loss": 0.11368724703788757,
+      "loss_ce": 0.0001923761737998575,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 356234572,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5458012757282831,
+      "grad_norm": 7.153408449848725,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 356406460,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5458012757282831,
+      "loss": 0.12134350836277008,
+      "loss_ce": 0.0004023421206511557,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 356406460,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5460643124876702,
+      "grad_norm": 5.616257612175736,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 356578828,
+      "step": 2076
+    },
+    {
+      "epoch": 0.5460643124876702,
+      "loss": 0.15952152013778687,
+      "loss_ce": 0.007452425081282854,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 356578828,
+      "step": 2076
+    },
+    {
+      "epoch": 0.5463273492470573,
+      "grad_norm": 4.268860828505962,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 356750940,
+      "step": 2077
+    },
+    {
+      "epoch": 0.5463273492470573,
+      "loss": 0.12928339838981628,
+      "loss_ce": 0.0006517980364151299,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 356750940,
+      "step": 2077
+    },
+    {
+      "epoch": 0.5465903860064444,
+      "grad_norm": 5.706661955848177,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 356923100,
+      "step": 2078
+    },
+    {
+      "epoch": 0.5465903860064444,
+      "loss": 0.07689663022756577,
+      "loss_ce": 0.002494774293154478,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 356923100,
+      "step": 2078
+    },
+    {
+      "epoch": 0.5468534227658315,
+      "grad_norm": 6.622312056596725,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 357095216,
+      "step": 2079
+    },
+    {
+      "epoch": 0.5468534227658315,
+      "loss": 0.2388056218624115,
+      "loss_ce": 0.0006769794854335487,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 357095216,
+      "step": 2079
+    },
+    {
+      "epoch": 0.5471164595252187,
+      "grad_norm": 5.101694254730695,
+      "learning_rate": 5e-06,
+      "loss": 0.131,
+      "num_input_tokens_seen": 357265716,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5471164595252187,
+      "loss": 0.14891289174556732,
+      "loss_ce": 0.0022454019635915756,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 357265716,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5473794962846058,
+      "grad_norm": 7.015564745743221,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 357437784,
+      "step": 2081
+    },
+    {
+      "epoch": 0.5473794962846058,
+      "loss": 0.09505030512809753,
+      "loss_ce": 0.003283948404714465,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 357437784,
+      "step": 2081
+    },
+    {
+      "epoch": 0.5476425330439929,
+      "grad_norm": 6.361890201898857,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 357609848,
+      "step": 2082
+    },
+    {
+      "epoch": 0.5476425330439929,
+      "loss": 0.11097072064876556,
+      "loss_ce": 0.0005886423168703914,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 357609848,
+      "step": 2082
+    },
+    {
+      "epoch": 0.54790556980338,
+      "grad_norm": 11.534449289054661,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 357782120,
+      "step": 2083
+    },
+    {
+      "epoch": 0.54790556980338,
+      "loss": 0.20264874398708344,
+      "loss_ce": 0.0027280959766358137,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 357782120,
+      "step": 2083
+    },
+    {
+      "epoch": 0.5481686065627671,
+      "grad_norm": 5.295463538809882,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 357954088,
+      "step": 2084
+    },
+    {
+      "epoch": 0.5481686065627671,
+      "loss": 0.13939827680587769,
+      "loss_ce": 0.0004822692717425525,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 357954088,
+      "step": 2084
+    },
+    {
+      "epoch": 0.5484316433221542,
+      "grad_norm": 5.108081687062298,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 358123060,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5484316433221542,
+      "loss": 0.11673710495233536,
+      "loss_ce": 0.002631876850500703,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 358123060,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5486946800815414,
+      "grad_norm": 8.727300119230549,
+      "learning_rate": 5e-06,
+      "loss": 0.1803,
+      "num_input_tokens_seen": 358295276,
+      "step": 2086
+    },
+    {
+      "epoch": 0.5486946800815414,
+      "loss": 0.21835477650165558,
+      "loss_ce": 0.0030837799422442913,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 358295276,
+      "step": 2086
+    },
+    {
+      "epoch": 0.5489577168409285,
+      "grad_norm": 11.066658379743997,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 358467252,
+      "step": 2087
+    },
+    {
+      "epoch": 0.5489577168409285,
+      "loss": 0.10189958661794662,
+      "loss_ce": 0.00464006420224905,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 358467252,
+      "step": 2087
+    },
+    {
+      "epoch": 0.5492207536003156,
+      "grad_norm": 6.149852742286746,
+      "learning_rate": 5e-06,
+      "loss": 0.1068,
+      "num_input_tokens_seen": 358639392,
+      "step": 2088
+    },
+    {
+      "epoch": 0.5492207536003156,
+      "loss": 0.11351503431797028,
+      "loss_ce": 0.0006915500853210688,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 358639392,
+      "step": 2088
+    },
+    {
+      "epoch": 0.5494837903597027,
+      "grad_norm": 4.11715937749224,
+      "learning_rate": 5e-06,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 358811532,
+      "step": 2089
+    },
+    {
+      "epoch": 0.5494837903597027,
+      "loss": 0.06601230055093765,
+      "loss_ce": 0.0033291929867118597,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 358811532,
+      "step": 2089
+    },
+    {
+      "epoch": 0.5497468271190898,
+      "grad_norm": 5.926803480110229,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 358982120,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5497468271190898,
+      "loss": 0.165444478392601,
+      "loss_ce": 0.0021449108608067036,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 358982120,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5500098638784771,
+      "grad_norm": 3.8279286211407215,
+      "learning_rate": 5e-06,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 359154336,
+      "step": 2091
+    },
+    {
+      "epoch": 0.5500098638784771,
+      "loss": 0.1499071568250656,
+      "loss_ce": 0.0012255202746018767,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 359154336,
+      "step": 2091
+    },
+    {
+      "epoch": 0.5502729006378642,
+      "grad_norm": 11.488893840776536,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 359326324,
+      "step": 2092
+    },
+    {
+      "epoch": 0.5502729006378642,
+      "loss": 0.169376939535141,
+      "loss_ce": 0.0019880137406289577,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 359326324,
+      "step": 2092
+    },
+    {
+      "epoch": 0.5505359373972513,
+      "grad_norm": 7.053437919021253,
+      "learning_rate": 5e-06,
+      "loss": 0.1738,
+      "num_input_tokens_seen": 359498532,
+      "step": 2093
+    },
+    {
+      "epoch": 0.5505359373972513,
+      "loss": 0.17925216257572174,
+      "loss_ce": 0.003562463214620948,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 359498532,
+      "step": 2093
+    },
+    {
+      "epoch": 0.5507989741566384,
+      "grad_norm": 4.495803041591062,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 359668724,
+      "step": 2094
+    },
+    {
+      "epoch": 0.5507989741566384,
+      "loss": 0.16115409135818481,
+      "loss_ce": 0.001760771730914712,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 359668724,
+      "step": 2094
+    },
+    {
+      "epoch": 0.5510620109160255,
+      "grad_norm": 4.2291818189369135,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 359841036,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5510620109160255,
+      "loss": 0.04658431187272072,
+      "loss_ce": 0.000533288111910224,
+      "loss_iou": NaN,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 359841036,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5513250476754127,
+      "grad_norm": 14.952283770001413,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 360013076,
+      "step": 2096
+    },
+    {
+      "epoch": 0.5513250476754127,
+      "loss": 0.056577593088150024,
+      "loss_ce": 0.00405684020370245,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 360013076,
+      "step": 2096
+    },
+    {
+      "epoch": 0.5515880844347998,
+      "grad_norm": 6.919816354273125,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 360181792,
+      "step": 2097
+    },
+    {
+      "epoch": 0.5515880844347998,
+      "loss": 0.08817453682422638,
+      "loss_ce": 0.0007874465081840754,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 360181792,
+      "step": 2097
+    },
+    {
+      "epoch": 0.5518511211941869,
+      "grad_norm": 4.617511415402225,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 360353784,
+      "step": 2098
+    },
+    {
+      "epoch": 0.5518511211941869,
+      "loss": 0.10340078175067902,
+      "loss_ce": 0.0006938728038221598,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 360353784,
+      "step": 2098
+    },
+    {
+      "epoch": 0.552114157953574,
+      "grad_norm": 21.83477585251536,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 360526236,
+      "step": 2099
+    },
+    {
+      "epoch": 0.552114157953574,
+      "loss": 0.09873877465724945,
+      "loss_ce": 0.005629643332213163,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 360526236,
+      "step": 2099
+    },
+    {
+      "epoch": 0.5523771947129611,
+      "grad_norm": 22.010760760138638,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 360698124,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5523771947129611,
+      "loss": 0.1067737340927124,
+      "loss_ce": 0.0001147918519563973,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 360698124,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5526402314723483,
+      "grad_norm": 8.309454312868692,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 360867340,
+      "step": 2101
+    },
+    {
+      "epoch": 0.5526402314723483,
+      "loss": 0.1460711508989334,
+      "loss_ce": 0.002302848733961582,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 360867340,
+      "step": 2101
+    },
+    {
+      "epoch": 0.5529032682317354,
+      "grad_norm": 15.176993944828276,
+      "learning_rate": 5e-06,
+      "loss": 0.1412,
+      "num_input_tokens_seen": 361039524,
+      "step": 2102
+    },
+    {
+      "epoch": 0.5529032682317354,
+      "loss": 0.18546344339847565,
+      "loss_ce": 0.003029361367225647,
+      "loss_iou": 0.546875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 361039524,
+      "step": 2102
+    },
+    {
+      "epoch": 0.5531663049911225,
+      "grad_norm": 7.240375850988684,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 361212152,
+      "step": 2103
+    },
+    {
+      "epoch": 0.5531663049911225,
+      "loss": 0.1453348845243454,
+      "loss_ce": 0.0008341491920873523,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 361212152,
+      "step": 2103
+    },
+    {
+      "epoch": 0.5534293417505096,
+      "grad_norm": 4.504467098918756,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 361384636,
+      "step": 2104
+    },
+    {
+      "epoch": 0.5534293417505096,
+      "loss": 0.14879915118217468,
+      "loss_ce": 0.0031692716293036938,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 361384636,
+      "step": 2104
+    },
+    {
+      "epoch": 0.5536923785098967,
+      "grad_norm": 14.969295535193908,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 361556736,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5536923785098967,
+      "loss": 0.14452342689037323,
+      "loss_ce": 0.0027692681178450584,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 361556736,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5539554152692839,
+      "grad_norm": 4.430127573397192,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 361726936,
+      "step": 2106
+    },
+    {
+      "epoch": 0.5539554152692839,
+      "loss": 0.22009092569351196,
+      "loss_ce": 0.0009747114963829517,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 361726936,
+      "step": 2106
+    },
+    {
+      "epoch": 0.554218452028671,
+      "grad_norm": 4.147747665163274,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 361899184,
+      "step": 2107
+    },
+    {
+      "epoch": 0.554218452028671,
+      "loss": 0.13676050305366516,
+      "loss_ce": 0.0002858861698769033,
+      "loss_iou": 0.703125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 361899184,
+      "step": 2107
+    },
+    {
+      "epoch": 0.5544814887880581,
+      "grad_norm": 3.7276098432513174,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 362071768,
+      "step": 2108
+    },
+    {
+      "epoch": 0.5544814887880581,
+      "loss": 0.15069580078125,
+      "loss_ce": 0.0011291508562862873,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 362071768,
+      "step": 2108
+    },
+    {
+      "epoch": 0.5547445255474452,
+      "grad_norm": 5.320538695832165,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 362240216,
+      "step": 2109
+    },
+    {
+      "epoch": 0.5547445255474452,
+      "loss": 0.09280645847320557,
+      "loss_ce": 0.001528381835669279,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 362240216,
+      "step": 2109
+    },
+    {
+      "epoch": 0.5550075623068323,
+      "grad_norm": 11.977017579320059,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 362412692,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5550075623068323,
+      "loss": 0.09514741599559784,
+      "loss_ce": 0.005608838051557541,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 362412692,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5552705990662195,
+      "grad_norm": 5.133455963412773,
+      "learning_rate": 5e-06,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 362583480,
+      "step": 2111
+    },
+    {
+      "epoch": 0.5552705990662195,
+      "loss": 0.20054732263088226,
+      "loss_ce": 0.0018321146490052342,
+      "loss_iou": NaN,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 362583480,
+      "step": 2111
+    },
+    {
+      "epoch": 0.5555336358256067,
+      "grad_norm": 13.046079025621871,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 362755416,
+      "step": 2112
+    },
+    {
+      "epoch": 0.5555336358256067,
+      "loss": 0.09682411700487137,
+      "loss_ce": 0.005820699501782656,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 362755416,
+      "step": 2112
+    },
+    {
+      "epoch": 0.5557966725849938,
+      "grad_norm": 5.170449556654708,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 362927780,
+      "step": 2113
+    },
+    {
+      "epoch": 0.5557966725849938,
+      "loss": 0.1499776542186737,
+      "loss_ce": 0.0003499832237139344,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 362927780,
+      "step": 2113
+    },
+    {
+      "epoch": 0.5560597093443809,
+      "grad_norm": 4.492587389090409,
+      "learning_rate": 5e-06,
+      "loss": 0.1634,
+      "num_input_tokens_seen": 363100176,
+      "step": 2114
+    },
+    {
+      "epoch": 0.5560597093443809,
+      "loss": 0.12822586297988892,
+      "loss_ce": 0.001638943562284112,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 363100176,
+      "step": 2114
+    },
+    {
+      "epoch": 0.556322746103768,
+      "grad_norm": 6.53280654225724,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 363272032,
+      "step": 2115
+    },
+    {
+      "epoch": 0.556322746103768,
+      "loss": 0.09251531958580017,
+      "loss_ce": 0.0013898293254896998,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 363272032,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5565857828631551,
+      "grad_norm": 6.772073322495271,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 363444356,
+      "step": 2116
+    },
+    {
+      "epoch": 0.5565857828631551,
+      "loss": 0.22190499305725098,
+      "loss_ce": 0.00468086265027523,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 363444356,
+      "step": 2116
+    },
+    {
+      "epoch": 0.5568488196225423,
+      "grad_norm": 4.259888500977366,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 363616480,
+      "step": 2117
+    },
+    {
+      "epoch": 0.5568488196225423,
+      "loss": 0.11629009246826172,
+      "loss_ce": 0.0017881433013826609,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 363616480,
+      "step": 2117
+    },
+    {
+      "epoch": 0.5571118563819294,
+      "grad_norm": 5.727482587569898,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 363788792,
+      "step": 2118
+    },
+    {
+      "epoch": 0.5571118563819294,
+      "loss": 0.14177094399929047,
+      "loss_ce": 0.0018478452693670988,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 363788792,
+      "step": 2118
+    },
+    {
+      "epoch": 0.5573748931413165,
+      "grad_norm": 10.912411227157135,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 363960764,
+      "step": 2119
+    },
+    {
+      "epoch": 0.5573748931413165,
+      "loss": 0.1305292397737503,
+      "loss_ce": 0.0011652277316898108,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 363960764,
+      "step": 2119
+    },
+    {
+      "epoch": 0.5576379299007036,
+      "grad_norm": 3.94400025085083,
+      "learning_rate": 5e-06,
+      "loss": 0.1587,
+      "num_input_tokens_seen": 364133156,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5576379299007036,
+      "loss": 0.19977153837680817,
+      "loss_ce": 0.008945131674408913,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 364133156,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5579009666600907,
+      "grad_norm": 3.6301701938092217,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 364305580,
+      "step": 2121
+    },
+    {
+      "epoch": 0.5579009666600907,
+      "loss": 0.1176564022898674,
+      "loss_ce": 0.0034901422914117575,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 364305580,
+      "step": 2121
+    },
+    {
+      "epoch": 0.5581640034194779,
+      "grad_norm": 19.838531274898244,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 364478040,
+      "step": 2122
+    },
+    {
+      "epoch": 0.5581640034194779,
+      "loss": 0.06633633375167847,
+      "loss_ce": 0.0024020099081099033,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 364478040,
+      "step": 2122
+    },
+    {
+      "epoch": 0.558427040178865,
+      "grad_norm": 9.488127206942975,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 364647728,
+      "step": 2123
+    },
+    {
+      "epoch": 0.558427040178865,
+      "loss": 0.07145293056964874,
+      "loss_ce": 0.00010283520532539114,
+      "loss_iou": 0.5,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 364647728,
+      "step": 2123
+    },
+    {
+      "epoch": 0.5586900769382521,
+      "grad_norm": 4.8177202903523115,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 364816920,
+      "step": 2124
+    },
+    {
+      "epoch": 0.5586900769382521,
+      "loss": 0.13919737935066223,
+      "loss_ce": 0.002829591976478696,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 364816920,
+      "step": 2124
+    },
+    {
+      "epoch": 0.5589531136976392,
+      "grad_norm": 3.304918735754454,
+      "learning_rate": 5e-06,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 364989156,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5589531136976392,
+      "loss": 0.062035560607910156,
+      "loss_ce": 0.002953530289232731,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 364989156,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5592161504570263,
+      "grad_norm": 2.856854904917954,
+      "learning_rate": 5e-06,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 365161052,
+      "step": 2126
+    },
+    {
+      "epoch": 0.5592161504570263,
+      "loss": 0.11198246479034424,
+      "loss_ce": 0.007825973443686962,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 365161052,
+      "step": 2126
+    },
+    {
+      "epoch": 0.5594791872164135,
+      "grad_norm": 3.111770745456122,
+      "learning_rate": 5e-06,
+      "loss": 0.196,
+      "num_input_tokens_seen": 365333284,
+      "step": 2127
+    },
+    {
+      "epoch": 0.5594791872164135,
+      "loss": 0.0993409976363182,
+      "loss_ce": 0.00046404742170125246,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 365333284,
+      "step": 2127
+    },
+    {
+      "epoch": 0.5597422239758006,
+      "grad_norm": 4.17183040056909,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 365505612,
+      "step": 2128
+    },
+    {
+      "epoch": 0.5597422239758006,
+      "loss": 0.08107022941112518,
+      "loss_ce": 0.000320716411806643,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 365505612,
+      "step": 2128
+    },
+    {
+      "epoch": 0.5600052607351877,
+      "grad_norm": 5.3777331039695175,
+      "learning_rate": 5e-06,
+      "loss": 0.1018,
+      "num_input_tokens_seen": 365677880,
+      "step": 2129
+    },
+    {
+      "epoch": 0.5600052607351877,
+      "loss": 0.10484224557876587,
+      "loss_ce": 0.0005636783316731453,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 365677880,
+      "step": 2129
+    },
+    {
+      "epoch": 0.5602682974945749,
+      "grad_norm": 13.479247744568044,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 365849928,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5602682974945749,
+      "loss": 0.1259067952632904,
+      "loss_ce": 0.0024021633435040712,
+      "loss_iou": 0.484375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 365849928,
+      "step": 2130
+    },
+    {
+      "epoch": 0.560531334253962,
+      "grad_norm": 20.004883443678533,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 366021916,
+      "step": 2131
+    },
+    {
+      "epoch": 0.560531334253962,
+      "loss": 0.11340343207120895,
+      "loss_ce": 0.0057374173775315285,
+      "loss_iou": 0.515625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 366021916,
+      "step": 2131
+    },
+    {
+      "epoch": 0.5607943710133492,
+      "grad_norm": 11.042521647809682,
+      "learning_rate": 5e-06,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 366194008,
+      "step": 2132
+    },
+    {
+      "epoch": 0.5607943710133492,
+      "loss": 0.09386501461267471,
+      "loss_ce": 0.00017604799359105527,
+      "loss_iou": 0.640625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 366194008,
+      "step": 2132
+    },
+    {
+      "epoch": 0.5610574077727363,
+      "grad_norm": 5.842914268629563,
+      "learning_rate": 5e-06,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 366366212,
+      "step": 2133
+    },
+    {
+      "epoch": 0.5610574077727363,
+      "loss": 0.11772558093070984,
+      "loss_ce": 0.0009958385489881039,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 366366212,
+      "step": 2133
+    },
+    {
+      "epoch": 0.5613204445321234,
+      "grad_norm": 5.960122855306223,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 366538220,
+      "step": 2134
+    },
+    {
+      "epoch": 0.5613204445321234,
+      "loss": 0.08179056644439697,
+      "loss_ce": 0.003818158758804202,
+      "loss_iou": 0.703125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 366538220,
+      "step": 2134
+    },
+    {
+      "epoch": 0.5615834812915105,
+      "grad_norm": 5.389091778733193,
+      "learning_rate": 5e-06,
+      "loss": 0.1578,
+      "num_input_tokens_seen": 366710096,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5615834812915105,
+      "loss": 0.18556632101535797,
+      "loss_ce": 0.0002635964483488351,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 366710096,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5618465180508976,
+      "grad_norm": 7.320192738875152,
+      "learning_rate": 5e-06,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 366880456,
+      "step": 2136
+    },
+    {
+      "epoch": 0.5618465180508976,
+      "loss": 0.13861671090126038,
+      "loss_ce": 0.003179695922881365,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 366880456,
+      "step": 2136
+    },
+    {
+      "epoch": 0.5621095548102847,
+      "grad_norm": 14.114626386817225,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 367052752,
+      "step": 2137
+    },
+    {
+      "epoch": 0.5621095548102847,
+      "loss": 0.12559227645397186,
+      "loss_ce": 0.001477292738854885,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 367052752,
+      "step": 2137
+    },
+    {
+      "epoch": 0.5623725915696719,
+      "grad_norm": 3.498748268755153,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 367225060,
+      "step": 2138
+    },
+    {
+      "epoch": 0.5623725915696719,
+      "loss": 0.08549857884645462,
+      "loss_ce": 0.000232468664762564,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 367225060,
+      "step": 2138
+    },
+    {
+      "epoch": 0.562635628329059,
+      "grad_norm": 20.701513936136166,
+      "learning_rate": 5e-06,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 367397332,
+      "step": 2139
+    },
+    {
+      "epoch": 0.562635628329059,
+      "loss": 0.08525611460208893,
+      "loss_ce": 0.004125134088099003,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 367397332,
+      "step": 2139
+    },
+    {
+      "epoch": 0.5628986650884461,
+      "grad_norm": 11.983965441231234,
+      "learning_rate": 5e-06,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 367569560,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5628986650884461,
+      "loss": 0.06727585196495056,
+      "loss_ce": 0.00010666107118595392,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 367569560,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5631617018478332,
+      "grad_norm": 6.398433590682968,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 367741696,
+      "step": 2141
+    },
+    {
+      "epoch": 0.5631617018478332,
+      "loss": 0.16246706247329712,
+      "loss_ce": 0.0022497763857245445,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 367741696,
+      "step": 2141
+    },
+    {
+      "epoch": 0.5634247386072203,
+      "grad_norm": 4.385698804822032,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 367913912,
+      "step": 2142
+    },
+    {
+      "epoch": 0.5634247386072203,
+      "loss": 0.10307023674249649,
+      "loss_ce": 0.0009889386128634214,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 367913912,
+      "step": 2142
+    },
+    {
+      "epoch": 0.5636877753666075,
+      "grad_norm": 7.856910300327451,
+      "learning_rate": 5e-06,
+      "loss": 0.1906,
+      "num_input_tokens_seen": 368086176,
+      "step": 2143
+    },
+    {
+      "epoch": 0.5636877753666075,
+      "loss": 0.19684657454490662,
+      "loss_ce": 0.005714981816709042,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 368086176,
+      "step": 2143
+    },
+    {
+      "epoch": 0.5639508121259946,
+      "grad_norm": 6.368187003490122,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 368258440,
+      "step": 2144
+    },
+    {
+      "epoch": 0.5639508121259946,
+      "loss": 0.1019875556230545,
+      "loss_ce": 0.0008217811118811369,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 368258440,
+      "step": 2144
+    },
+    {
+      "epoch": 0.5642138488853817,
+      "grad_norm": 7.190126438403998,
+      "learning_rate": 5e-06,
+      "loss": 0.1332,
+      "num_input_tokens_seen": 368430620,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5642138488853817,
+      "loss": 0.1344844102859497,
+      "loss_ce": 0.00020707116345874965,
+      "loss_iou": 0.546875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 368430620,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5644768856447688,
+      "grad_norm": 6.755563239238797,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 368603024,
+      "step": 2146
+    },
+    {
+      "epoch": 0.5644768856447688,
+      "loss": 0.09109672158956528,
+      "loss_ce": 0.0006883963942527771,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 368603024,
+      "step": 2146
+    },
+    {
+      "epoch": 0.5647399224041559,
+      "grad_norm": 4.616140839020028,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 368775624,
+      "step": 2147
+    },
+    {
+      "epoch": 0.5647399224041559,
+      "loss": 0.19518432021141052,
+      "loss_ce": 0.002435298403725028,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 368775624,
+      "step": 2147
+    },
+    {
+      "epoch": 0.5650029591635432,
+      "grad_norm": 11.734928019848066,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 368945448,
+      "step": 2148
+    },
+    {
+      "epoch": 0.5650029591635432,
+      "loss": 0.17952315509319305,
+      "loss_ce": 0.009143512696027756,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 368945448,
+      "step": 2148
+    },
+    {
+      "epoch": 0.5652659959229303,
+      "grad_norm": 6.710624849462096,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 369115952,
+      "step": 2149
+    },
+    {
+      "epoch": 0.5652659959229303,
+      "loss": 0.13233953714370728,
+      "loss_ce": 0.0019989716820418835,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 369115952,
+      "step": 2149
+    },
+    {
+      "epoch": 0.5655290326823174,
+      "grad_norm": 6.669120586380705,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 369288028,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5655290326823174,
+      "loss": 0.08679264783859253,
+      "loss_ce": 0.00695866858586669,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 369288028,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5657920694417045,
+      "grad_norm": 4.502790600863043,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 369460112,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5657920694417045,
+      "loss": 0.09304537624120712,
+      "loss_ce": 0.002102992497384548,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 369460112,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5660551062010916,
+      "grad_norm": 5.258977022311985,
+      "learning_rate": 5e-06,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 369632404,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5660551062010916,
+      "loss": 0.08912669122219086,
+      "loss_ce": 0.0003358002286404371,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 369632404,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5663181429604788,
+      "grad_norm": 5.721568404254781,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 369804776,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5663181429604788,
+      "loss": 0.10320156812667847,
+      "loss_ce": 0.0007235408993437886,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 369804776,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5665811797198659,
+      "grad_norm": 4.65695630930855,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 369976952,
+      "step": 2154
+    },
+    {
+      "epoch": 0.5665811797198659,
+      "loss": 0.08642168343067169,
+      "loss_ce": 0.0003621142532210797,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 369976952,
+      "step": 2154
+    },
+    {
+      "epoch": 0.566844216479253,
+      "grad_norm": 2.957349659939758,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 370149132,
+      "step": 2155
+    },
+    {
+      "epoch": 0.566844216479253,
+      "loss": 0.11206680536270142,
+      "loss_ce": 0.0026002456434071064,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 370149132,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5671072532386401,
+      "grad_norm": 13.220367815915187,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 370321528,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5671072532386401,
+      "loss": 0.13273131847381592,
+      "loss_ce": 0.0014446950517594814,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 370321528,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5673702899980272,
+      "grad_norm": 6.5167958115597955,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 370493692,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5673702899980272,
+      "loss": 0.0516238659620285,
+      "loss_ce": 0.00018648749392013997,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 370493692,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5676333267574144,
+      "grad_norm": 9.055547000690966,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 370666128,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5676333267574144,
+      "loss": 0.20648962259292603,
+      "loss_ce": 0.0006485594203695655,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 370666128,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5678963635168015,
+      "grad_norm": 20.191615671139235,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 370838096,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5678963635168015,
+      "loss": 0.1389261931180954,
+      "loss_ce": 0.0004679340636357665,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 370838096,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5681594002761886,
+      "grad_norm": 16.724015569984758,
+      "learning_rate": 5e-06,
+      "loss": 0.0926,
+      "num_input_tokens_seen": 371010276,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5681594002761886,
+      "loss": 0.09113931655883789,
+      "loss_ce": 0.004194741137325764,
+      "loss_iou": 0.53125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 371010276,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5684224370355757,
+      "grad_norm": 9.65595197930286,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 371180756,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5684224370355757,
+      "loss": 0.10710924118757248,
+      "loss_ce": 0.0031358497217297554,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 371180756,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5686854737949628,
+      "grad_norm": 4.559805216337552,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 371351216,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5686854737949628,
+      "loss": 0.18528233468532562,
+      "loss_ce": 0.0012918633874505758,
+      "loss_iou": 0.421875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 371351216,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5689485105543499,
+      "grad_norm": 3.9132748752015707,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 371521504,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5689485105543499,
+      "loss": 0.09377571940422058,
+      "loss_ce": 0.00017830087745096534,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 371521504,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5692115473137371,
+      "grad_norm": 6.221489940852924,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 371693620,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5692115473137371,
+      "loss": 0.08065352588891983,
+      "loss_ce": 0.000270225660642609,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 371693620,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5694745840731242,
+      "grad_norm": 5.565730297773139,
+      "learning_rate": 5e-06,
+      "loss": 0.1544,
+      "num_input_tokens_seen": 371864020,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5694745840731242,
+      "loss": 0.14530430734157562,
+      "loss_ce": 0.0009256468038074672,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 371864020,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5697376208325113,
+      "grad_norm": 7.807853954323307,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 372036364,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5697376208325113,
+      "loss": 0.1258363127708435,
+      "loss_ce": 0.00034803448943421245,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 372036364,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5700006575918984,
+      "grad_norm": 6.119338737471246,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 372208520,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5700006575918984,
+      "loss": 0.183029443025589,
+      "loss_ce": 0.0006258888752199709,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 372208520,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5702636943512855,
+      "grad_norm": 4.796742469743409,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 372380908,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5702636943512855,
+      "loss": 0.24278424680233002,
+      "loss_ce": 0.006517150904983282,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 372380908,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5705267311106728,
+      "grad_norm": 7.9441210817884444,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 372553152,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5705267311106728,
+      "loss": 0.1403554081916809,
+      "loss_ce": 0.0024769881274551153,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 372553152,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5707897678700599,
+      "grad_norm": 8.945699343509036,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 372722892,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5707897678700599,
+      "loss": 0.17014455795288086,
+      "loss_ce": 0.0013518218183889985,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 372722892,
+      "step": 2170
+    },
+    {
+      "epoch": 0.571052804629447,
+      "grad_norm": 7.4926547587717405,
+      "learning_rate": 5e-06,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 372895068,
+      "step": 2171
+    },
+    {
+      "epoch": 0.571052804629447,
+      "loss": 0.2696492373943329,
+      "loss_ce": 0.0004536675405688584,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 372895068,
+      "step": 2171
+    },
+    {
+      "epoch": 0.5713158413888341,
+      "grad_norm": 17.833109235832993,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 373067312,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5713158413888341,
+      "loss": 0.09599291533231735,
+      "loss_ce": 0.0006254853797145188,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 373067312,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5715788781482212,
+      "grad_norm": 14.290103080283112,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 373239588,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5715788781482212,
+      "loss": 0.15090827643871307,
+      "loss_ce": 0.002836985979229212,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 373239588,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5718419149076084,
+      "grad_norm": 4.534353132879591,
+      "learning_rate": 5e-06,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 373411840,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5718419149076084,
+      "loss": 0.07687939703464508,
+      "loss_ce": 0.0007075219764374197,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 373411840,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5721049516669955,
+      "grad_norm": 12.524713753292392,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 373582408,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5721049516669955,
+      "loss": 0.17097817361354828,
+      "loss_ce": 0.0006900950102135539,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 373582408,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5723679884263826,
+      "grad_norm": 18.760082445449868,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 373754416,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5723679884263826,
+      "loss": 0.11672262102365494,
+      "loss_ce": 0.0011525547597557306,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 373754416,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5726310251857697,
+      "grad_norm": 8.516129726413212,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 373922312,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5726310251857697,
+      "loss": 0.19126161932945251,
+      "loss_ce": 0.0003436502593103796,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 373922312,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5728940619451568,
+      "grad_norm": 6.898009634423141,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 374092720,
+      "step": 2178
+    },
+    {
+      "epoch": 0.5728940619451568,
+      "loss": 0.15485724806785583,
+      "loss_ce": 0.0002551883808337152,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 374092720,
+      "step": 2178
+    },
+    {
+      "epoch": 0.573157098704544,
+      "grad_norm": 6.043872383339532,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 374264864,
+      "step": 2179
+    },
+    {
+      "epoch": 0.573157098704544,
+      "loss": 0.12720485031604767,
+      "loss_ce": 0.0004958686186000705,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 374264864,
+      "step": 2179
+    },
+    {
+      "epoch": 0.5734201354639311,
+      "grad_norm": 5.166426865666374,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 374436928,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5734201354639311,
+      "loss": 0.10286220163106918,
+      "loss_ce": 0.0005062465788796544,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 374436928,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5736831722233182,
+      "grad_norm": 3.8181355710391034,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 374608984,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5736831722233182,
+      "loss": 0.1574546992778778,
+      "loss_ce": 0.00025864943745546043,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 374608984,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5739462089827053,
+      "grad_norm": 9.672382405473702,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 374779732,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5739462089827053,
+      "loss": 0.17890840768814087,
+      "loss_ce": 0.007033395115286112,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 374779732,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5742092457420924,
+      "grad_norm": 5.19568946256922,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 374951880,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5742092457420924,
+      "loss": 0.11335025727748871,
+      "loss_ce": 0.0024798910599201918,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 374951880,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5744722825014796,
+      "grad_norm": 4.272972418213817,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 375120908,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5744722825014796,
+      "loss": 0.16440820693969727,
+      "loss_ce": 0.000330454291542992,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 375120908,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5747353192608667,
+      "grad_norm": 4.877678921853992,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 375292940,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5747353192608667,
+      "loss": 0.19540463387966156,
+      "loss_ce": 0.0004278157721273601,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 375292940,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5749983560202538,
+      "grad_norm": 9.795881894231353,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 375463220,
+      "step": 2186
+    },
+    {
+      "epoch": 0.5749983560202538,
+      "loss": 0.18603767454624176,
+      "loss_ce": 0.0036951417569071054,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 375463220,
+      "step": 2186
+    },
+    {
+      "epoch": 0.575261392779641,
+      "grad_norm": 14.3298059229615,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 375633332,
+      "step": 2187
+    },
+    {
+      "epoch": 0.575261392779641,
+      "loss": 0.11824968457221985,
+      "loss_ce": 0.0010011474369093776,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 375633332,
+      "step": 2187
+    },
+    {
+      "epoch": 0.575524429539028,
+      "grad_norm": 5.9170529044976305,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 375803896,
+      "step": 2188
+    },
+    {
+      "epoch": 0.575524429539028,
+      "loss": 0.09120282530784607,
+      "loss_ce": 0.0025797830894589424,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 375803896,
+      "step": 2188
+    },
+    {
+      "epoch": 0.5757874662984152,
+      "grad_norm": 4.8523353127793944,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 375976212,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5757874662984152,
+      "loss": 0.11096520721912384,
+      "loss_ce": 0.0007662302814424038,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 375976212,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5760505030578024,
+      "grad_norm": 3.7437825333865065,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 376148372,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5760505030578024,
+      "loss": 0.1311783641576767,
+      "loss_ce": 0.0013108099810779095,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 376148372,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5763135398171895,
+      "grad_norm": 3.8863610288761765,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 376320448,
+      "step": 2191
+    },
+    {
+      "epoch": 0.5763135398171895,
+      "loss": 0.08886295557022095,
+      "loss_ce": 0.00014836144691798836,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 376320448,
+      "step": 2191
+    },
+    {
+      "epoch": 0.5765765765765766,
+      "grad_norm": 26.72560789877691,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 376492652,
+      "step": 2192
+    },
+    {
+      "epoch": 0.5765765765765766,
+      "loss": 0.09363338351249695,
+      "loss_ce": 0.002172202803194523,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 376492652,
+      "step": 2192
+    },
+    {
+      "epoch": 0.5768396133359637,
+      "grad_norm": 3.4367691751013334,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 376665088,
+      "step": 2193
+    },
+    {
+      "epoch": 0.5768396133359637,
+      "loss": 0.05514270067214966,
+      "loss_ce": 0.0010502950754016638,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 376665088,
+      "step": 2193
+    },
+    {
+      "epoch": 0.5771026500953508,
+      "grad_norm": 9.788333451803501,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 376837384,
+      "step": 2194
+    },
+    {
+      "epoch": 0.5771026500953508,
+      "loss": 0.09362407773733139,
+      "loss_ce": 0.0004081379738636315,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 376837384,
+      "step": 2194
+    },
+    {
+      "epoch": 0.577365686854738,
+      "grad_norm": 5.1293333388841935,
+      "learning_rate": 5e-06,
+      "loss": 0.1096,
+      "num_input_tokens_seen": 377006200,
+      "step": 2195
+    },
+    {
+      "epoch": 0.577365686854738,
+      "loss": 0.14207029342651367,
+      "loss_ce": 0.0032763422932475805,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 377006200,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5776287236141251,
+      "grad_norm": 9.582310699127605,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 377178552,
+      "step": 2196
+    },
+    {
+      "epoch": 0.5776287236141251,
+      "loss": 0.14796333014965057,
+      "loss_ce": 0.0009296314674429595,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 377178552,
+      "step": 2196
+    },
+    {
+      "epoch": 0.5778917603735122,
+      "grad_norm": 5.51238614508857,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 377350508,
+      "step": 2197
+    },
+    {
+      "epoch": 0.5778917603735122,
+      "loss": 0.12102293223142624,
+      "loss_ce": 0.011342758312821388,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 377350508,
+      "step": 2197
+    },
+    {
+      "epoch": 0.5781547971328993,
+      "grad_norm": 12.700643820920646,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 377522740,
+      "step": 2198
+    },
+    {
+      "epoch": 0.5781547971328993,
+      "loss": 0.13567548990249634,
+      "loss_ce": 0.004617748782038689,
+      "loss_iou": 0.53125,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 377522740,
+      "step": 2198
+    },
+    {
+      "epoch": 0.5784178338922864,
+      "grad_norm": 10.703599236021498,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 377694584,
+      "step": 2199
+    },
+    {
+      "epoch": 0.5784178338922864,
+      "loss": 0.17266914248466492,
+      "loss_ce": 0.0024420833215117455,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 377694584,
+      "step": 2199
+    },
+    {
+      "epoch": 0.5786808706516736,
+      "grad_norm": 3.7187673646063684,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 377866484,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5786808706516736,
+      "loss": 0.04466433823108673,
+      "loss_ce": 0.0001697081606835127,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 377866484,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5789439074110607,
+      "grad_norm": 6.267393061703399,
+      "learning_rate": 5e-06,
+      "loss": 0.1317,
+      "num_input_tokens_seen": 378038564,
+      "step": 2201
+    },
+    {
+      "epoch": 0.5789439074110607,
+      "loss": 0.18358194828033447,
+      "loss_ce": 0.0036503085866570473,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 378038564,
+      "step": 2201
+    },
+    {
+      "epoch": 0.5792069441704478,
+      "grad_norm": 8.846753851466788,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 378211032,
+      "step": 2202
+    },
+    {
+      "epoch": 0.5792069441704478,
+      "loss": 0.14636895060539246,
+      "loss_ce": 0.0015783084090799093,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 378211032,
+      "step": 2202
+    },
+    {
+      "epoch": 0.5794699809298349,
+      "grad_norm": 9.056222556282368,
+      "learning_rate": 5e-06,
+      "loss": 0.1862,
+      "num_input_tokens_seen": 378383136,
+      "step": 2203
+    },
+    {
+      "epoch": 0.5794699809298349,
+      "loss": 0.10566692054271698,
+      "loss_ce": 0.003677169792354107,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 378383136,
+      "step": 2203
+    },
+    {
+      "epoch": 0.579733017689222,
+      "grad_norm": 4.9933922483079725,
+      "learning_rate": 5e-06,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 378553532,
+      "step": 2204
+    },
+    {
+      "epoch": 0.579733017689222,
+      "loss": 0.10826604068279266,
+      "loss_ce": 0.001820725854486227,
+      "loss_iou": 0.74609375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 378553532,
+      "step": 2204
+    },
+    {
+      "epoch": 0.5799960544486092,
+      "grad_norm": 15.853504685750568,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 378725804,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5799960544486092,
+      "loss": 0.05737042799592018,
+      "loss_ce": 0.0001499689242336899,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 378725804,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5802590912079963,
+      "grad_norm": 10.934142417787772,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 378897924,
+      "step": 2206
+    },
+    {
+      "epoch": 0.5802590912079963,
+      "loss": 0.08355730026960373,
+      "loss_ce": 0.004913499113172293,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 378897924,
+      "step": 2206
+    },
+    {
+      "epoch": 0.5805221279673835,
+      "grad_norm": 12.227334484970761,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 379070348,
+      "step": 2207
+    },
+    {
+      "epoch": 0.5805221279673835,
+      "loss": 0.07382334768772125,
+      "loss_ce": 0.00047435000305995345,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 379070348,
+      "step": 2207
+    },
+    {
+      "epoch": 0.5807851647267706,
+      "grad_norm": 5.142567427993278,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 379242648,
+      "step": 2208
+    },
+    {
+      "epoch": 0.5807851647267706,
+      "loss": 0.06755711138248444,
+      "loss_ce": 0.0022494932636618614,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 379242648,
+      "step": 2208
+    },
+    {
+      "epoch": 0.5810482014861577,
+      "grad_norm": 4.5313412775127055,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 379414684,
+      "step": 2209
+    },
+    {
+      "epoch": 0.5810482014861577,
+      "loss": 0.050227776169776917,
+      "loss_ce": 0.0009724035626277328,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 379414684,
+      "step": 2209
+    },
+    {
+      "epoch": 0.5813112382455449,
+      "grad_norm": 51.12006755064005,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 379582192,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5813112382455449,
+      "loss": 0.11350201815366745,
+      "loss_ce": 0.004676333162933588,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 379582192,
+      "step": 2210
+    },
+    {
+      "epoch": 0.581574275004932,
+      "grad_norm": 18.708708783651804,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 379754536,
+      "step": 2211
+    },
+    {
+      "epoch": 0.581574275004932,
+      "loss": 0.09766032546758652,
+      "loss_ce": 0.0012552967527881265,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 379754536,
+      "step": 2211
+    },
+    {
+      "epoch": 0.5818373117643191,
+      "grad_norm": 5.58601503702725,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 379926528,
+      "step": 2212
+    },
+    {
+      "epoch": 0.5818373117643191,
+      "loss": 0.18606778979301453,
+      "loss_ce": 0.0023214598186314106,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 379926528,
+      "step": 2212
+    },
+    {
+      "epoch": 0.5821003485237062,
+      "grad_norm": 3.8208876250172965,
+      "learning_rate": 5e-06,
+      "loss": 0.1334,
+      "num_input_tokens_seen": 380096704,
+      "step": 2213
+    },
+    {
+      "epoch": 0.5821003485237062,
+      "loss": 0.18421456217765808,
+      "loss_ce": 0.001994105987250805,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 380096704,
+      "step": 2213
+    },
+    {
+      "epoch": 0.5823633852830933,
+      "grad_norm": 4.7316491932517595,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 380269136,
+      "step": 2214
+    },
+    {
+      "epoch": 0.5823633852830933,
+      "loss": 0.23052500188350677,
+      "loss_ce": 0.0009107402293011546,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 380269136,
+      "step": 2214
+    },
+    {
+      "epoch": 0.5826264220424804,
+      "grad_norm": 5.131239937970846,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 380441472,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5826264220424804,
+      "loss": 0.09817831218242645,
+      "loss_ce": 0.0009187856921926141,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 380441472,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5828894588018676,
+      "grad_norm": 5.520590188036205,
+      "learning_rate": 5e-06,
+      "loss": 0.1,
+      "num_input_tokens_seen": 380613832,
+      "step": 2216
+    },
+    {
+      "epoch": 0.5828894588018676,
+      "loss": 0.09423954784870148,
+      "loss_ce": 0.0019696487579494715,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 380613832,
+      "step": 2216
+    },
+    {
+      "epoch": 0.5831524955612547,
+      "grad_norm": 18.191266422169875,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 380785988,
+      "step": 2217
+    },
+    {
+      "epoch": 0.5831524955612547,
+      "loss": 0.135514497756958,
+      "loss_ce": 0.0037701106630265713,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 380785988,
+      "step": 2217
+    },
+    {
+      "epoch": 0.5834155323206418,
+      "grad_norm": 13.995891706301725,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 380958116,
+      "step": 2218
+    },
+    {
+      "epoch": 0.5834155323206418,
+      "loss": 0.07710295170545578,
+      "loss_ce": 0.00120573490858078,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 380958116,
+      "step": 2218
+    },
+    {
+      "epoch": 0.5836785690800289,
+      "grad_norm": 3.998007795342675,
+      "learning_rate": 5e-06,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 381130168,
+      "step": 2219
+    },
+    {
+      "epoch": 0.5836785690800289,
+      "loss": 0.17625044286251068,
+      "loss_ce": 0.0008964374428614974,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 381130168,
+      "step": 2219
+    },
+    {
+      "epoch": 0.583941605839416,
+      "grad_norm": 8.315710539211903,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 381302476,
+      "step": 2220
+    },
+    {
+      "epoch": 0.583941605839416,
+      "loss": 0.20635367929935455,
+      "loss_ce": 0.002496248111128807,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 381302476,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5842046425988032,
+      "grad_norm": 9.769478710683881,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 381474700,
+      "step": 2221
+    },
+    {
+      "epoch": 0.5842046425988032,
+      "loss": 0.1044168546795845,
+      "loss_ce": 0.0017252071993425488,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 381474700,
+      "step": 2221
+    },
+    {
+      "epoch": 0.5844676793581903,
+      "grad_norm": 5.959703884673753,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 381646792,
+      "step": 2222
+    },
+    {
+      "epoch": 0.5844676793581903,
+      "loss": 0.20204247534275055,
+      "loss_ce": 0.0024575116112828255,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 381646792,
+      "step": 2222
+    },
+    {
+      "epoch": 0.5847307161175774,
+      "grad_norm": 4.639177291482811,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 381818872,
+      "step": 2223
+    },
+    {
+      "epoch": 0.5847307161175774,
+      "loss": 0.1277788281440735,
+      "loss_ce": 0.007478540297597647,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 381818872,
+      "step": 2223
+    },
+    {
+      "epoch": 0.5849937528769645,
+      "grad_norm": 4.556750582512925,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 381991284,
+      "step": 2224
+    },
+    {
+      "epoch": 0.5849937528769645,
+      "loss": 0.06165578216314316,
+      "loss_ce": 0.003062034724280238,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 381991284,
+      "step": 2224
+    },
+    {
+      "epoch": 0.5852567896363516,
+      "grad_norm": 5.717259207619562,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 382163040,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5852567896363516,
+      "loss": 0.08200475573539734,
+      "loss_ce": 0.0018045613542199135,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 382163040,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5855198263957389,
+      "grad_norm": 4.765363459193876,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 382335172,
+      "step": 2226
+    },
+    {
+      "epoch": 0.5855198263957389,
+      "loss": 0.08970290422439575,
+      "loss_ce": 0.001568139297887683,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 382335172,
+      "step": 2226
+    },
+    {
+      "epoch": 0.585782863155126,
+      "grad_norm": 5.040130370486427,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 382507360,
+      "step": 2227
+    },
+    {
+      "epoch": 0.585782863155126,
+      "loss": 0.06643694639205933,
+      "loss_ce": 0.0008241523755714297,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 382507360,
+      "step": 2227
+    },
+    {
+      "epoch": 0.5860458999145131,
+      "grad_norm": 27.12917591646992,
+      "learning_rate": 5e-06,
+      "loss": 0.1669,
+      "num_input_tokens_seen": 382677836,
+      "step": 2228
+    },
+    {
+      "epoch": 0.5860458999145131,
+      "loss": 0.11577419936656952,
+      "loss_ce": 0.0032864054664969444,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 382677836,
+      "step": 2228
+    },
+    {
+      "epoch": 0.5863089366739002,
+      "grad_norm": 5.353068404352155,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 382849968,
+      "step": 2229
+    },
+    {
+      "epoch": 0.5863089366739002,
+      "loss": 0.11902253329753876,
+      "loss_ce": 0.0011636477429419756,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 382849968,
+      "step": 2229
+    },
+    {
+      "epoch": 0.5865719734332873,
+      "grad_norm": 44.6262248448857,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 383022244,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5865719734332873,
+      "loss": 0.08623628318309784,
+      "loss_ce": 0.0015957842115312815,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 383022244,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5868350101926745,
+      "grad_norm": 11.243592990037332,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 383190620,
+      "step": 2231
+    },
+    {
+      "epoch": 0.5868350101926745,
+      "loss": 0.11198394745588303,
+      "loss_ce": 0.0013882413040846586,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 383190620,
+      "step": 2231
+    },
+    {
+      "epoch": 0.5870980469520616,
+      "grad_norm": 4.97022834965809,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 383362980,
+      "step": 2232
+    },
+    {
+      "epoch": 0.5870980469520616,
+      "loss": 0.17192208766937256,
+      "loss_ce": 0.0035566147416830063,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 383362980,
+      "step": 2232
+    },
+    {
+      "epoch": 0.5873610837114487,
+      "grad_norm": 4.21141022600957,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 383535200,
+      "step": 2233
+    },
+    {
+      "epoch": 0.5873610837114487,
+      "loss": 0.1112288236618042,
+      "loss_ce": 0.00690448796376586,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 383535200,
+      "step": 2233
+    },
+    {
+      "epoch": 0.5876241204708358,
+      "grad_norm": 4.036382922773338,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 383705348,
+      "step": 2234
+    },
+    {
+      "epoch": 0.5876241204708358,
+      "loss": 0.05898036062717438,
+      "loss_ce": 0.001027482096105814,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 383705348,
+      "step": 2234
+    },
+    {
+      "epoch": 0.5878871572302229,
+      "grad_norm": 7.244981585553582,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 383877692,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5878871572302229,
+      "loss": 0.10776747018098831,
+      "loss_ce": 0.0007423229981213808,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 383877692,
+      "step": 2235
+    },
+    {
+      "epoch": 0.58815019398961,
+      "grad_norm": 3.396542457402831,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "num_input_tokens_seen": 384049540,
+      "step": 2236
+    },
+    {
+      "epoch": 0.58815019398961,
+      "loss": 0.1896773874759674,
+      "loss_ce": 0.0011398009955883026,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 384049540,
+      "step": 2236
+    },
+    {
+      "epoch": 0.5884132307489972,
+      "grad_norm": 8.64901320262372,
+      "learning_rate": 5e-06,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 384221584,
+      "step": 2237
+    },
+    {
+      "epoch": 0.5884132307489972,
+      "loss": 0.12090113013982773,
+      "loss_ce": 0.0009365270379930735,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 384221584,
+      "step": 2237
+    },
+    {
+      "epoch": 0.5886762675083843,
+      "grad_norm": 4.107798805009285,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 384393708,
+      "step": 2238
+    },
+    {
+      "epoch": 0.5886762675083843,
+      "loss": 0.10467529296875,
+      "loss_ce": 0.0007019043550826609,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 384393708,
+      "step": 2238
+    },
+    {
+      "epoch": 0.5889393042677714,
+      "grad_norm": 5.687252203871965,
+      "learning_rate": 5e-06,
+      "loss": 0.1541,
+      "num_input_tokens_seen": 384566100,
+      "step": 2239
+    },
+    {
+      "epoch": 0.5889393042677714,
+      "loss": 0.2502034306526184,
+      "loss_ce": 0.004079162143170834,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 384566100,
+      "step": 2239
+    },
+    {
+      "epoch": 0.5892023410271585,
+      "grad_norm": 5.391300577620961,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 384738000,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5892023410271585,
+      "loss": 0.09789521992206573,
+      "loss_ce": 0.0031381379812955856,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 384738000,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5894653777865456,
+      "grad_norm": 2.700202309873461,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 384910408,
+      "step": 2241
+    },
+    {
+      "epoch": 0.5894653777865456,
+      "loss": 0.18347935378551483,
+      "loss_ce": 0.00031284932629205287,
+      "loss_iou": 0.484375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 384910408,
+      "step": 2241
+    },
+    {
+      "epoch": 0.5897284145459328,
+      "grad_norm": 8.687970034600534,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 385082740,
+      "step": 2242
+    },
+    {
+      "epoch": 0.5897284145459328,
+      "loss": 0.11214028298854828,
+      "loss_ce": 0.0005375072360038757,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 385082740,
+      "step": 2242
+    },
+    {
+      "epoch": 0.5899914513053199,
+      "grad_norm": 7.347975569465696,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 385254812,
+      "step": 2243
+    },
+    {
+      "epoch": 0.5899914513053199,
+      "loss": 0.10931709408760071,
+      "loss_ce": 0.000491409155074507,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 385254812,
+      "step": 2243
+    },
+    {
+      "epoch": 0.590254488064707,
+      "grad_norm": 5.972916509227761,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 385427116,
+      "step": 2244
+    },
+    {
+      "epoch": 0.590254488064707,
+      "loss": 0.06860466301441193,
+      "loss_ce": 0.0034191168379038572,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 385427116,
+      "step": 2244
+    },
+    {
+      "epoch": 0.5905175248240941,
+      "grad_norm": 5.792932058485695,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 385599308,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5905175248240941,
+      "loss": 0.09140485525131226,
+      "loss_ce": 0.0006760837859474123,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 385599308,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5907805615834812,
+      "grad_norm": 6.999898072921272,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 385771596,
+      "step": 2246
+    },
+    {
+      "epoch": 0.5907805615834812,
+      "loss": 0.18632760643959045,
+      "loss_ce": 0.0021235125605016947,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 385771596,
+      "step": 2246
+    },
+    {
+      "epoch": 0.5910435983428685,
+      "grad_norm": 18.38237494107746,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 385940788,
+      "step": 2247
+    },
+    {
+      "epoch": 0.5910435983428685,
+      "loss": 0.16689589619636536,
+      "loss_ce": 0.004466078244149685,
+      "loss_iou": 0.40625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 385940788,
+      "step": 2247
+    },
+    {
+      "epoch": 0.5913066351022556,
+      "grad_norm": 7.020065723323422,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 386112940,
+      "step": 2248
+    },
+    {
+      "epoch": 0.5913066351022556,
+      "loss": 0.06020001322031021,
+      "loss_ce": 0.0016672981437295675,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 386112940,
+      "step": 2248
+    },
+    {
+      "epoch": 0.5915696718616427,
+      "grad_norm": 4.523572034689062,
+      "learning_rate": 5e-06,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 386285088,
+      "step": 2249
+    },
+    {
+      "epoch": 0.5915696718616427,
+      "loss": 0.10929292440414429,
+      "loss_ce": 0.002771313302218914,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 386285088,
+      "step": 2249
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "grad_norm": 4.838564447785546,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_websight_new_CIoU": 0.8620143532752991,
+      "eval_websight_new_GIoU": 0.8631013035774231,
+      "eval_websight_new_IoU": 0.8667041063308716,
+      "eval_websight_new_MAE_all": 0.020488929003477097,
+      "eval_websight_new_MAE_h": 0.008965343236923218,
+      "eval_websight_new_MAE_w": 0.0330337006598711,
+      "eval_websight_new_MAE_x": 0.03212242014706135,
+      "eval_websight_new_MAE_y": 0.007834249641746283,
+      "eval_websight_new_NUM_probability": 0.9999847710132599,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.10749460011720657,
+      "eval_websight_new_loss_ce": 1.4298896530817728e-05,
+      "eval_websight_new_loss_iou": 0.3743896484375,
+      "eval_websight_new_loss_num": 0.019153594970703125,
+      "eval_websight_new_loss_xval": 0.095703125,
+      "eval_websight_new_runtime": 58.289,
+      "eval_websight_new_samples_per_second": 0.858,
+      "eval_websight_new_steps_per_second": 0.034,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_seeclick_CIoU": 0.6236494481563568,
+      "eval_seeclick_GIoU": 0.6230664253234863,
+      "eval_seeclick_IoU": 0.6453896760940552,
+      "eval_seeclick_MAE_all": 0.046443790197372437,
+      "eval_seeclick_MAE_h": 0.026547173038125038,
+      "eval_seeclick_MAE_w": 0.06541823036968708,
+      "eval_seeclick_MAE_x": 0.06924234330654144,
+      "eval_seeclick_MAE_y": 0.024567410349845886,
+      "eval_seeclick_NUM_probability": 0.9999750256538391,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.21426968276500702,
+      "eval_seeclick_loss_ce": 0.009121979121118784,
+      "eval_seeclick_loss_iou": 0.506591796875,
+      "eval_seeclick_loss_num": 0.039905548095703125,
+      "eval_seeclick_loss_xval": 0.1995849609375,
+      "eval_seeclick_runtime": 71.9379,
+      "eval_seeclick_samples_per_second": 0.598,
+      "eval_seeclick_steps_per_second": 0.028,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_icons_CIoU": 0.834777295589447,
+      "eval_icons_GIoU": 0.8241135478019714,
+      "eval_icons_IoU": 0.8441117405891418,
+      "eval_icons_MAE_all": 0.02446969971060753,
+      "eval_icons_MAE_h": 0.024498000741004944,
+      "eval_icons_MAE_w": 0.02540498599410057,
+      "eval_icons_MAE_x": 0.02417761366814375,
+      "eval_icons_MAE_y": 0.02379819191992283,
+      "eval_icons_NUM_probability": 0.9999534487724304,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.0800996944308281,
+      "eval_icons_loss_ce": 2.8939639378222637e-05,
+      "eval_icons_loss_iou": 0.520263671875,
+      "eval_icons_loss_num": 0.014501571655273438,
+      "eval_icons_loss_xval": 0.072509765625,
+      "eval_icons_runtime": 87.9614,
+      "eval_icons_samples_per_second": 0.568,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_screenspot_CIoU": 0.5634338855743408,
+      "eval_screenspot_GIoU": 0.5513045191764832,
+      "eval_screenspot_IoU": 0.6039714018503824,
+      "eval_screenspot_MAE_all": 0.08249951650698979,
+      "eval_screenspot_MAE_h": 0.04705421378215154,
+      "eval_screenspot_MAE_w": 0.14340341091156006,
+      "eval_screenspot_MAE_x": 0.09280380109945933,
+      "eval_screenspot_MAE_y": 0.04673664582272371,
+      "eval_screenspot_NUM_probability": 0.99980628490448,
+      "eval_screenspot_inside_bbox": 0.8454166650772095,
+      "eval_screenspot_loss": 0.8701639175415039,
+      "eval_screenspot_loss_ce": 0.543925940990448,
+      "eval_screenspot_loss_iou": 0.45556640625,
+      "eval_screenspot_loss_num": 0.06413777669270833,
+      "eval_screenspot_loss_xval": 0.3208414713541667,
+      "eval_screenspot_runtime": 147.2039,
+      "eval_screenspot_samples_per_second": 0.605,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "loss": 0.8634153604507446,
+      "loss_ce": 0.5327268838882446,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5920957453804169,
+      "grad_norm": 17.382874089820607,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 386629432,
+      "step": 2251
+    },
+    {
+      "epoch": 0.5920957453804169,
+      "loss": 0.06886275112628937,
+      "loss_ce": 0.0024259830825030804,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 386629432,
+      "step": 2251
+    },
+    {
+      "epoch": 0.5923587821398041,
+      "grad_norm": 4.874671437046262,
+      "learning_rate": 5e-06,
+      "loss": 0.1568,
+      "num_input_tokens_seen": 386801572,
+      "step": 2252
+    },
+    {
+      "epoch": 0.5923587821398041,
+      "loss": 0.14957007765769958,
+      "loss_ce": 0.0020786237437278032,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 386801572,
+      "step": 2252
+    },
+    {
+      "epoch": 0.5926218188991912,
+      "grad_norm": 5.10088801695758,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 386973592,
+      "step": 2253
+    },
+    {
+      "epoch": 0.5926218188991912,
+      "loss": 0.14472725987434387,
+      "loss_ce": 0.00031808449421077967,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 386973592,
+      "step": 2253
+    },
+    {
+      "epoch": 0.5928848556585783,
+      "grad_norm": 7.206762089035192,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 387143876,
+      "step": 2254
+    },
+    {
+      "epoch": 0.5928848556585783,
+      "loss": 0.1427307277917862,
+      "loss_ce": 0.0004577827639877796,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 387143876,
+      "step": 2254
+    },
+    {
+      "epoch": 0.5931478924179654,
+      "grad_norm": 28.833326639866428,
+      "learning_rate": 5e-06,
+      "loss": 0.1461,
+      "num_input_tokens_seen": 387316360,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5931478924179654,
+      "loss": 0.10336272418498993,
+      "loss_ce": 0.0003353758074808866,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 387316360,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5934109291773525,
+      "grad_norm": 11.869740305960864,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 387488944,
+      "step": 2256
+    },
+    {
+      "epoch": 0.5934109291773525,
+      "loss": 0.0730450302362442,
+      "loss_ce": 0.0006573314312845469,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 387488944,
+      "step": 2256
+    },
+    {
+      "epoch": 0.5936739659367397,
+      "grad_norm": 8.676646075770014,
+      "learning_rate": 5e-06,
+      "loss": 0.1525,
+      "num_input_tokens_seen": 387661216,
+      "step": 2257
+    },
+    {
+      "epoch": 0.5936739659367397,
+      "loss": 0.09287041425704956,
+      "loss_ce": 0.00037164040259085596,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 387661216,
+      "step": 2257
+    },
+    {
+      "epoch": 0.5939370026961268,
+      "grad_norm": 6.091102593612703,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 387833316,
+      "step": 2258
+    },
+    {
+      "epoch": 0.5939370026961268,
+      "loss": 0.047980114817619324,
+      "loss_ce": 0.002722549019381404,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 387833316,
+      "step": 2258
+    },
+    {
+      "epoch": 0.5942000394555139,
+      "grad_norm": 13.12094289217443,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 388005488,
+      "step": 2259
+    },
+    {
+      "epoch": 0.5942000394555139,
+      "loss": 0.1262204349040985,
+      "loss_ce": 0.006103249732404947,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 388005488,
+      "step": 2259
+    },
+    {
+      "epoch": 0.594463076214901,
+      "grad_norm": 6.977408677449423,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 388178100,
+      "step": 2260
+    },
+    {
+      "epoch": 0.594463076214901,
+      "loss": 0.09303668141365051,
+      "loss_ce": 0.0011177423875778913,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 388178100,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5947261129742881,
+      "grad_norm": 5.548896564721941,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 388349964,
+      "step": 2261
+    },
+    {
+      "epoch": 0.5947261129742881,
+      "loss": 0.07538396120071411,
+      "loss_ce": 0.0003412406367715448,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 388349964,
+      "step": 2261
+    },
+    {
+      "epoch": 0.5949891497336752,
+      "grad_norm": 3.9845795265650694,
+      "learning_rate": 5e-06,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 388520668,
+      "step": 2262
+    },
+    {
+      "epoch": 0.5949891497336752,
+      "loss": 0.18895787000656128,
+      "loss_ce": 0.00042028201278299093,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 388520668,
+      "step": 2262
+    },
+    {
+      "epoch": 0.5952521864930624,
+      "grad_norm": 5.603595618443302,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 388692900,
+      "step": 2263
+    },
+    {
+      "epoch": 0.5952521864930624,
+      "loss": 0.08959123492240906,
+      "loss_ce": 0.0002662862534634769,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 388692900,
+      "step": 2263
+    },
+    {
+      "epoch": 0.5955152232524495,
+      "grad_norm": 5.071553765162839,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 388865328,
+      "step": 2264
+    },
+    {
+      "epoch": 0.5955152232524495,
+      "loss": 0.07327542454004288,
+      "loss_ce": 0.0008877270738594234,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 388865328,
+      "step": 2264
+    },
+    {
+      "epoch": 0.5957782600118366,
+      "grad_norm": 7.1017098248073385,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 389037536,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5957782600118366,
+      "loss": 0.15171518921852112,
+      "loss_ce": 0.0005311004933901131,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 389037536,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5960412967712237,
+      "grad_norm": 6.446588247534734,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 389209660,
+      "step": 2266
+    },
+    {
+      "epoch": 0.5960412967712237,
+      "loss": 0.11115900427103043,
+      "loss_ce": 0.00028864690102636814,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 389209660,
+      "step": 2266
+    },
+    {
+      "epoch": 0.5963043335306109,
+      "grad_norm": 9.702541385516984,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 389381624,
+      "step": 2267
+    },
+    {
+      "epoch": 0.5963043335306109,
+      "loss": 0.13201884925365448,
+      "loss_ce": 0.00581339979544282,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 389381624,
+      "step": 2267
+    },
+    {
+      "epoch": 0.5965673702899981,
+      "grad_norm": 4.325039584368215,
+      "learning_rate": 5e-06,
+      "loss": 0.1332,
+      "num_input_tokens_seen": 389553348,
+      "step": 2268
+    },
+    {
+      "epoch": 0.5965673702899981,
+      "loss": 0.09481080621480942,
+      "loss_ce": 0.0006488211220130324,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 389553348,
+      "step": 2268
+    },
+    {
+      "epoch": 0.5968304070493852,
+      "grad_norm": 4.06473977313393,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 389723828,
+      "step": 2269
+    },
+    {
+      "epoch": 0.5968304070493852,
+      "loss": 0.130482017993927,
+      "loss_ce": 0.00023298643645830452,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 389723828,
+      "step": 2269
+    },
+    {
+      "epoch": 0.5970934438087723,
+      "grad_norm": 11.83991543858794,
+      "learning_rate": 5e-06,
+      "loss": 0.1278,
+      "num_input_tokens_seen": 389896036,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5970934438087723,
+      "loss": 0.1708485186100006,
+      "loss_ce": 0.0017506256699562073,
+      "loss_iou": 0.5625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 389896036,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5973564805681594,
+      "grad_norm": 4.637610311727533,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 390068288,
+      "step": 2271
+    },
+    {
+      "epoch": 0.5973564805681594,
+      "loss": 0.12343515455722809,
+      "loss_ce": 0.0004340623854659498,
+      "loss_iou": NaN,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 390068288,
+      "step": 2271
+    },
+    {
+      "epoch": 0.5976195173275465,
+      "grad_norm": 4.239687367925931,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 390240724,
+      "step": 2272
+    },
+    {
+      "epoch": 0.5976195173275465,
+      "loss": 0.09358179569244385,
+      "loss_ce": 0.0032497686333954334,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 390240724,
+      "step": 2272
+    },
+    {
+      "epoch": 0.5978825540869337,
+      "grad_norm": 3.650780325093881,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 390413092,
+      "step": 2273
+    },
+    {
+      "epoch": 0.5978825540869337,
+      "loss": 0.0876360684633255,
+      "loss_ce": 0.00032527127768844366,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 390413092,
+      "step": 2273
+    },
+    {
+      "epoch": 0.5981455908463208,
+      "grad_norm": 4.543660924331624,
+      "learning_rate": 5e-06,
+      "loss": 0.0989,
+      "num_input_tokens_seen": 390585320,
+      "step": 2274
+    },
+    {
+      "epoch": 0.5981455908463208,
+      "loss": 0.1153046190738678,
+      "loss_ce": 0.0005890398169867694,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 390585320,
+      "step": 2274
+    },
+    {
+      "epoch": 0.5984086276057079,
+      "grad_norm": 5.356363884549334,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 390757376,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5984086276057079,
+      "loss": 0.1397184282541275,
+      "loss_ce": 0.0057157427072525024,
+      "loss_iou": 0.5,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 390757376,
+      "step": 2275
+    },
+    {
+      "epoch": 0.598671664365095,
+      "grad_norm": 3.3713877207341088,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 390929460,
+      "step": 2276
+    },
+    {
+      "epoch": 0.598671664365095,
+      "loss": 0.0523032546043396,
+      "loss_ce": 0.000316562014631927,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 390929460,
+      "step": 2276
+    },
+    {
+      "epoch": 0.5989347011244821,
+      "grad_norm": 10.008368190926095,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 391100108,
+      "step": 2277
+    },
+    {
+      "epoch": 0.5989347011244821,
+      "loss": 0.08954879641532898,
+      "loss_ce": 0.0027873138897120953,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 391100108,
+      "step": 2277
+    },
+    {
+      "epoch": 0.5991977378838693,
+      "grad_norm": 13.178397333981023,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 391272352,
+      "step": 2278
+    },
+    {
+      "epoch": 0.5991977378838693,
+      "loss": 0.1727224737405777,
+      "loss_ce": 0.00011504795111250132,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 391272352,
+      "step": 2278
+    },
+    {
+      "epoch": 0.5994607746432564,
+      "grad_norm": 5.1375780540443134,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 391444648,
+      "step": 2279
+    },
+    {
+      "epoch": 0.5994607746432564,
+      "loss": 0.05273713544011116,
+      "loss_ce": 0.0022915778681635857,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 391444648,
+      "step": 2279
+    },
+    {
+      "epoch": 0.5997238114026435,
+      "grad_norm": 25.459943093434337,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 391616960,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5997238114026435,
+      "loss": 0.1544933021068573,
+      "loss_ce": 0.0004405686049722135,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 391616960,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5999868481620306,
+      "grad_norm": 4.8890917043699185,
+      "learning_rate": 5e-06,
+      "loss": 0.1464,
+      "num_input_tokens_seen": 391788924,
+      "step": 2281
+    },
+    {
+      "epoch": 0.5999868481620306,
+      "loss": 0.06912894546985626,
+      "loss_ce": 0.006689978763461113,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 391788924,
+      "step": 2281
+    },
+    {
+      "epoch": 0.6002498849214177,
+      "grad_norm": 6.219770384955685,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 391961072,
+      "step": 2282
+    },
+    {
+      "epoch": 0.6002498849214177,
+      "loss": 0.07950527220964432,
+      "loss_ce": 0.00041896995389834046,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 391961072,
+      "step": 2282
+    },
+    {
+      "epoch": 0.600512921680805,
+      "grad_norm": 3.971504685954048,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 392130392,
+      "step": 2283
+    },
+    {
+      "epoch": 0.600512921680805,
+      "loss": 0.11902518570423126,
+      "loss_ce": 0.0007390595856122673,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 392130392,
+      "step": 2283
+    },
+    {
+      "epoch": 0.600775958440192,
+      "grad_norm": 3.36927604682008,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 392300644,
+      "step": 2284
+    },
+    {
+      "epoch": 0.600775958440192,
+      "loss": 0.09242402017116547,
+      "loss_ce": 0.0041824462823569775,
+      "loss_iou": 0.375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 392300644,
+      "step": 2284
+    },
+    {
+      "epoch": 0.6010389951995792,
+      "grad_norm": 5.15292132366606,
+      "learning_rate": 5e-06,
+      "loss": 0.1317,
+      "num_input_tokens_seen": 392472968,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6010389951995792,
+      "loss": 0.14593744277954102,
+      "loss_ce": 0.0034508705139160156,
+      "loss_iou": 0.484375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 392472968,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6013020319589663,
+      "grad_norm": 4.237274665038102,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 392644756,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6013020319589663,
+      "loss": 0.07083064317703247,
+      "loss_ce": 0.00170833186712116,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 392644756,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6015650687183534,
+      "grad_norm": 43.77503497663098,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 392816816,
+      "step": 2287
+    },
+    {
+      "epoch": 0.6015650687183534,
+      "loss": 0.09309446811676025,
+      "loss_ce": 0.00016844802303239703,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 392816816,
+      "step": 2287
+    },
+    {
+      "epoch": 0.6018281054777405,
+      "grad_norm": 8.046683810113267,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 392987164,
+      "step": 2288
+    },
+    {
+      "epoch": 0.6018281054777405,
+      "loss": 0.12998059391975403,
+      "loss_ce": 0.0008912362391129136,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 392987164,
+      "step": 2288
+    },
+    {
+      "epoch": 0.6020911422371277,
+      "grad_norm": 6.745453284904858,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 393159584,
+      "step": 2289
+    },
+    {
+      "epoch": 0.6020911422371277,
+      "loss": 0.18720246851444244,
+      "loss_ce": 0.0019302507862448692,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 393159584,
+      "step": 2289
+    },
+    {
+      "epoch": 0.6023541789965148,
+      "grad_norm": 9.388482886571397,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 393331696,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6023541789965148,
+      "loss": 0.153926283121109,
+      "loss_ce": 0.005275162868201733,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 393331696,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6026172157559019,
+      "grad_norm": 5.436488955385426,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 393504044,
+      "step": 2291
+    },
+    {
+      "epoch": 0.6026172157559019,
+      "loss": 0.11098843812942505,
+      "loss_ce": 0.00027066541952081025,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 393504044,
+      "step": 2291
+    },
+    {
+      "epoch": 0.602880252515289,
+      "grad_norm": 8.104274681671972,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 393676272,
+      "step": 2292
+    },
+    {
+      "epoch": 0.602880252515289,
+      "loss": 0.16863158345222473,
+      "loss_ce": 0.003592532593756914,
+      "loss_iou": 0.453125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 393676272,
+      "step": 2292
+    },
+    {
+      "epoch": 0.6031432892746761,
+      "grad_norm": 12.32428248921536,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 393848392,
+      "step": 2293
+    },
+    {
+      "epoch": 0.6031432892746761,
+      "loss": 0.08580140769481659,
+      "loss_ce": 0.0013287551701068878,
+      "loss_iou": 0.53125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 393848392,
+      "step": 2293
+    },
+    {
+      "epoch": 0.6034063260340633,
+      "grad_norm": 10.386325441280649,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 394020680,
+      "step": 2294
+    },
+    {
+      "epoch": 0.6034063260340633,
+      "loss": 0.07794995605945587,
+      "loss_ce": 0.003273440757766366,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 394020680,
+      "step": 2294
+    },
+    {
+      "epoch": 0.6036693627934504,
+      "grad_norm": 5.0087549483878355,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 394192740,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6036693627934504,
+      "loss": 0.12549570202827454,
+      "loss_ce": 0.0021436563692986965,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 394192740,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6039323995528375,
+      "grad_norm": 12.109968647066866,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 394362948,
+      "step": 2296
+    },
+    {
+      "epoch": 0.6039323995528375,
+      "loss": 0.1552933156490326,
+      "loss_ce": 0.0013626604340970516,
+      "loss_iou": 0.625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 394362948,
+      "step": 2296
+    },
+    {
+      "epoch": 0.6041954363122246,
+      "grad_norm": 3.064699260610535,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 394534856,
+      "step": 2297
+    },
+    {
+      "epoch": 0.6041954363122246,
+      "loss": 0.13219855725765228,
+      "loss_ce": 0.0020563420839607716,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 394534856,
+      "step": 2297
+    },
+    {
+      "epoch": 0.6044584730716117,
+      "grad_norm": 9.354536255922673,
+      "learning_rate": 5e-06,
+      "loss": 0.1776,
+      "num_input_tokens_seen": 394707060,
+      "step": 2298
+    },
+    {
+      "epoch": 0.6044584730716117,
+      "loss": 0.16126899421215057,
+      "loss_ce": 0.003599932650104165,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 394707060,
+      "step": 2298
+    },
+    {
+      "epoch": 0.6047215098309989,
+      "grad_norm": 4.517992664125833,
+      "learning_rate": 5e-06,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 394879016,
+      "step": 2299
+    },
+    {
+      "epoch": 0.6047215098309989,
+      "loss": 0.0760730504989624,
+      "loss_ce": 0.002098444849252701,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 394879016,
+      "step": 2299
+    },
+    {
+      "epoch": 0.604984546590386,
+      "grad_norm": 4.501760214176101,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 395051188,
+      "step": 2300
+    },
+    {
+      "epoch": 0.604984546590386,
+      "loss": 0.15630951523780823,
+      "loss_ce": 0.0008682362968102098,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 395051188,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6052475833497731,
+      "grad_norm": 4.195253270921506,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 395223544,
+      "step": 2301
+    },
+    {
+      "epoch": 0.6052475833497731,
+      "loss": 0.08342162519693375,
+      "loss_ce": 0.0002612252428662032,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 395223544,
+      "step": 2301
+    },
+    {
+      "epoch": 0.6055106201091602,
+      "grad_norm": 5.118374892921478,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 395395376,
+      "step": 2302
+    },
+    {
+      "epoch": 0.6055106201091602,
+      "loss": 0.11348429322242737,
+      "loss_ce": 0.0011490845354273915,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 395395376,
+      "step": 2302
+    },
+    {
+      "epoch": 0.6057736568685473,
+      "grad_norm": 4.210270169640029,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 395567140,
+      "step": 2303
+    },
+    {
+      "epoch": 0.6057736568685473,
+      "loss": 0.08656048029661179,
+      "loss_ce": 0.0008671237155795097,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 395567140,
+      "step": 2303
+    },
+    {
+      "epoch": 0.6060366936279346,
+      "grad_norm": 4.272171126326938,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 395739120,
+      "step": 2304
+    },
+    {
+      "epoch": 0.6060366936279346,
+      "loss": 0.14756399393081665,
+      "loss_ce": 0.002758089918643236,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 395739120,
+      "step": 2304
+    },
+    {
+      "epoch": 0.6062997303873217,
+      "grad_norm": 8.535390245901846,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 395911508,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6062997303873217,
+      "loss": 0.1481376737356186,
+      "loss_ce": 0.002080546924844384,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 395911508,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6065627671467088,
+      "grad_norm": 6.5128397067856785,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 396083668,
+      "step": 2306
+    },
+    {
+      "epoch": 0.6065627671467088,
+      "loss": 0.05626985430717468,
+      "loss_ce": 0.0008499314426444471,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 396083668,
+      "step": 2306
+    },
+    {
+      "epoch": 0.6068258039060959,
+      "grad_norm": 4.785548469928152,
+      "learning_rate": 5e-06,
+      "loss": 0.0998,
+      "num_input_tokens_seen": 396255888,
+      "step": 2307
+    },
+    {
+      "epoch": 0.6068258039060959,
+      "loss": 0.08620062470436096,
+      "loss_ce": 0.002124701626598835,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 396255888,
+      "step": 2307
+    },
+    {
+      "epoch": 0.607088840665483,
+      "grad_norm": 11.035467862846048,
+      "learning_rate": 5e-06,
+      "loss": 0.1858,
+      "num_input_tokens_seen": 396427920,
+      "step": 2308
+    },
+    {
+      "epoch": 0.607088840665483,
+      "loss": 0.19853177666664124,
+      "loss_ce": 0.0004726996412500739,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 396427920,
+      "step": 2308
+    },
+    {
+      "epoch": 0.6073518774248702,
+      "grad_norm": 5.6025345838645055,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 396599820,
+      "step": 2309
+    },
+    {
+      "epoch": 0.6073518774248702,
+      "loss": 0.09216836839914322,
+      "loss_ce": 0.0037742014974355698,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 396599820,
+      "step": 2309
+    },
+    {
+      "epoch": 0.6076149141842573,
+      "grad_norm": 21.234701292276423,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 396771948,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6076149141842573,
+      "loss": 0.08635897189378738,
+      "loss_ce": 0.000482511764857918,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 396771948,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6078779509436444,
+      "grad_norm": 4.7419638393344545,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 396943768,
+      "step": 2311
+    },
+    {
+      "epoch": 0.6078779509436444,
+      "loss": 0.12838855385780334,
+      "loss_ce": 0.002534065628424287,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 396943768,
+      "step": 2311
+    },
+    {
+      "epoch": 0.6081409877030315,
+      "grad_norm": 5.73379136559562,
+      "learning_rate": 5e-06,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 397116008,
+      "step": 2312
+    },
+    {
+      "epoch": 0.6081409877030315,
+      "loss": 0.17652729153633118,
+      "loss_ce": 0.0020277751609683037,
+      "loss_iou": NaN,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 397116008,
+      "step": 2312
+    },
+    {
+      "epoch": 0.6084040244624186,
+      "grad_norm": 4.062322434152511,
+      "learning_rate": 5e-06,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 397288120,
+      "step": 2313
+    },
+    {
+      "epoch": 0.6084040244624186,
+      "loss": 0.09052719175815582,
+      "loss_ce": 0.0007139619556255639,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 397288120,
+      "step": 2313
+    },
+    {
+      "epoch": 0.6086670612218057,
+      "grad_norm": 16.02293214120212,
+      "learning_rate": 5e-06,
+      "loss": 0.1668,
+      "num_input_tokens_seen": 397460304,
+      "step": 2314
+    },
+    {
+      "epoch": 0.6086670612218057,
+      "loss": 0.13694977760314941,
+      "loss_ce": 0.0005667208461090922,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 397460304,
+      "step": 2314
+    },
+    {
+      "epoch": 0.6089300979811929,
+      "grad_norm": 7.987103586052065,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 397632608,
+      "step": 2315
+    },
+    {
+      "epoch": 0.6089300979811929,
+      "loss": 0.06060004234313965,
+      "loss_ce": 0.002006293274462223,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 397632608,
+      "step": 2315
+    },
+    {
+      "epoch": 0.60919313474058,
+      "grad_norm": 7.316466982553544,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 397804820,
+      "step": 2316
+    },
+    {
+      "epoch": 0.60919313474058,
+      "loss": 0.11626386642456055,
+      "loss_ce": 0.004111772403120995,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 397804820,
+      "step": 2316
+    },
+    {
+      "epoch": 0.6094561714999671,
+      "grad_norm": 5.539376425734617,
+      "learning_rate": 5e-06,
+      "loss": 0.1638,
+      "num_input_tokens_seen": 397976940,
+      "step": 2317
+    },
+    {
+      "epoch": 0.6094561714999671,
+      "loss": 0.16956084966659546,
+      "loss_ce": 0.004979567602276802,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 397976940,
+      "step": 2317
+    },
+    {
+      "epoch": 0.6097192082593542,
+      "grad_norm": 24.75685247660265,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 398147040,
+      "step": 2318
+    },
+    {
+      "epoch": 0.6097192082593542,
+      "loss": 0.18709902465343475,
+      "loss_ce": 0.0012164636282250285,
+      "loss_iou": NaN,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 398147040,
+      "step": 2318
+    },
+    {
+      "epoch": 0.6099822450187413,
+      "grad_norm": 4.315961063244564,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 398318996,
+      "step": 2319
+    },
+    {
+      "epoch": 0.6099822450187413,
+      "loss": 0.18130794167518616,
+      "loss_ce": 0.0009795635705813766,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 398318996,
+      "step": 2319
+    },
+    {
+      "epoch": 0.6102452817781285,
+      "grad_norm": 4.640931709162671,
+      "learning_rate": 5e-06,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 398489328,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6102452817781285,
+      "loss": 0.09983004629611969,
+      "loss_ce": 0.0005563638987950981,
+      "loss_iou": 0.546875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 398489328,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6105083185375156,
+      "grad_norm": 23.53542790698555,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 398661592,
+      "step": 2321
+    },
+    {
+      "epoch": 0.6105083185375156,
+      "loss": 0.14297714829444885,
+      "loss_ce": 0.0037559503689408302,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 398661592,
+      "step": 2321
+    },
+    {
+      "epoch": 0.6107713552969027,
+      "grad_norm": 4.191481208630257,
+      "learning_rate": 5e-06,
+      "loss": 0.1143,
+      "num_input_tokens_seen": 398833940,
+      "step": 2322
+    },
+    {
+      "epoch": 0.6107713552969027,
+      "loss": 0.1795070767402649,
+      "loss_ce": 0.003298588562756777,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 398833940,
+      "step": 2322
+    },
+    {
+      "epoch": 0.6110343920562898,
+      "grad_norm": 4.031146460659383,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 399006244,
+      "step": 2323
+    },
+    {
+      "epoch": 0.6110343920562898,
+      "loss": 0.0546613447368145,
+      "loss_ce": 0.0011029954766854644,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 399006244,
+      "step": 2323
+    },
+    {
+      "epoch": 0.611297428815677,
+      "grad_norm": 4.121680425084659,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 399178456,
+      "step": 2324
+    },
+    {
+      "epoch": 0.611297428815677,
+      "loss": 0.16250503063201904,
+      "loss_ce": 0.002928605070337653,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 399178456,
+      "step": 2324
+    },
+    {
+      "epoch": 0.6115604655750642,
+      "grad_norm": 4.598379596612616,
+      "learning_rate": 5e-06,
+      "loss": 0.0851,
+      "num_input_tokens_seen": 399350964,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6115604655750642,
+      "loss": 0.06914816796779633,
+      "loss_ce": 0.002833473263308406,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 399350964,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6118235023344513,
+      "grad_norm": 7.270975374787245,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 399523228,
+      "step": 2326
+    },
+    {
+      "epoch": 0.6118235023344513,
+      "loss": 0.08900976926088333,
+      "loss_ce": 0.0011801763903349638,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 399523228,
+      "step": 2326
+    },
+    {
+      "epoch": 0.6120865390938384,
+      "grad_norm": 4.064818082785728,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 399691856,
+      "step": 2327
+    },
+    {
+      "epoch": 0.6120865390938384,
+      "loss": 0.10691290348768234,
+      "loss_ce": 0.00043707285658456385,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 399691856,
+      "step": 2327
+    },
+    {
+      "epoch": 0.6123495758532255,
+      "grad_norm": 7.343987420034391,
+      "learning_rate": 5e-06,
+      "loss": 0.1249,
+      "num_input_tokens_seen": 399864108,
+      "step": 2328
+    },
+    {
+      "epoch": 0.6123495758532255,
+      "loss": 0.09687530994415283,
+      "loss_ce": 0.001996160950511694,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 399864108,
+      "step": 2328
+    },
+    {
+      "epoch": 0.6126126126126126,
+      "grad_norm": 5.275060955551622,
+      "learning_rate": 5e-06,
+      "loss": 0.0841,
+      "num_input_tokens_seen": 400036500,
+      "step": 2329
+    },
+    {
+      "epoch": 0.6126126126126126,
+      "loss": 0.08277605473995209,
+      "loss_ce": 0.004162768833339214,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 400036500,
+      "step": 2329
+    },
+    {
+      "epoch": 0.6128756493719998,
+      "grad_norm": 4.648207674990067,
+      "learning_rate": 5e-06,
+      "loss": 0.1259,
+      "num_input_tokens_seen": 400206892,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6128756493719998,
+      "loss": 0.19954022765159607,
+      "loss_ce": 0.0007487052353098989,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 400206892,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6131386861313869,
+      "grad_norm": 6.976991385409509,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 400377220,
+      "step": 2331
+    },
+    {
+      "epoch": 0.6131386861313869,
+      "loss": 0.10119281709194183,
+      "loss_ce": 0.0006984363426454365,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 400377220,
+      "step": 2331
+    },
+    {
+      "epoch": 0.613401722890774,
+      "grad_norm": 12.605542023426503,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 400549468,
+      "step": 2332
+    },
+    {
+      "epoch": 0.613401722890774,
+      "loss": 0.20965467393398285,
+      "loss_ce": 0.009276244789361954,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 400549468,
+      "step": 2332
+    },
+    {
+      "epoch": 0.6136647596501611,
+      "grad_norm": 4.707098406894473,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 400721520,
+      "step": 2333
+    },
+    {
+      "epoch": 0.6136647596501611,
+      "loss": 0.11055370420217514,
+      "loss_ce": 0.0006599072366952896,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 400721520,
+      "step": 2333
+    },
+    {
+      "epoch": 0.6139277964095482,
+      "grad_norm": 14.198987666898462,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 400893600,
+      "step": 2334
+    },
+    {
+      "epoch": 0.6139277964095482,
+      "loss": 0.11723913997411728,
+      "loss_ce": 0.0025235607754439116,
+      "loss_iou": NaN,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 400893600,
+      "step": 2334
+    },
+    {
+      "epoch": 0.6141908331689354,
+      "grad_norm": 4.053613863943027,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 401065668,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6141908331689354,
+      "loss": 0.07270236313343048,
+      "loss_ce": 0.0009250181610696018,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 401065668,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6144538699283225,
+      "grad_norm": 5.7996514772347005,
+      "learning_rate": 5e-06,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 401237708,
+      "step": 2336
+    },
+    {
+      "epoch": 0.6144538699283225,
+      "loss": 0.07430876046419144,
+      "loss_ce": 0.0035384970251470804,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 401237708,
+      "step": 2336
+    },
+    {
+      "epoch": 0.6147169066877096,
+      "grad_norm": 4.799102165669682,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 401410012,
+      "step": 2337
+    },
+    {
+      "epoch": 0.6147169066877096,
+      "loss": 0.0582679845392704,
+      "loss_ce": 0.0014747708337381482,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 401410012,
+      "step": 2337
+    },
+    {
+      "epoch": 0.6149799434470967,
+      "grad_norm": 4.382135516303559,
+      "learning_rate": 5e-06,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 401582352,
+      "step": 2338
+    },
+    {
+      "epoch": 0.6149799434470967,
+      "loss": 0.09339425712823868,
+      "loss_ce": 0.00275705405510962,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 401582352,
+      "step": 2338
+    },
+    {
+      "epoch": 0.6152429802064838,
+      "grad_norm": 4.340567932001983,
+      "learning_rate": 5e-06,
+      "loss": 0.1484,
+      "num_input_tokens_seen": 401754308,
+      "step": 2339
+    },
+    {
+      "epoch": 0.6152429802064838,
+      "loss": 0.14638805389404297,
+      "loss_ce": 0.0007886901148594916,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 401754308,
+      "step": 2339
+    },
+    {
+      "epoch": 0.6155060169658709,
+      "grad_norm": 6.569770077552981,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 401926216,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6155060169658709,
+      "loss": 0.12163828313350677,
+      "loss_ce": 0.004069316200911999,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 401926216,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6157690537252581,
+      "grad_norm": 4.067934681272699,
+      "learning_rate": 5e-06,
+      "loss": 0.096,
+      "num_input_tokens_seen": 402098264,
+      "step": 2341
+    },
+    {
+      "epoch": 0.6157690537252581,
+      "loss": 0.0871957540512085,
+      "loss_ce": 0.0004342720494605601,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 402098264,
+      "step": 2341
+    },
+    {
+      "epoch": 0.6160320904846452,
+      "grad_norm": 4.130435258439036,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 402270372,
+      "step": 2342
+    },
+    {
+      "epoch": 0.6160320904846452,
+      "loss": 0.046382177621126175,
+      "loss_ce": 7.175222708610818e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 402270372,
+      "step": 2342
+    },
+    {
+      "epoch": 0.6162951272440323,
+      "grad_norm": 5.637480244426784,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 402440632,
+      "step": 2343
+    },
+    {
+      "epoch": 0.6162951272440323,
+      "loss": 0.07675184309482574,
+      "loss_ce": 0.0019837813451886177,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 402440632,
+      "step": 2343
+    },
+    {
+      "epoch": 0.6165581640034195,
+      "grad_norm": 8.043736182852212,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 402612920,
+      "step": 2344
+    },
+    {
+      "epoch": 0.6165581640034195,
+      "loss": 0.09050323814153671,
+      "loss_ce": 0.00018646713579073548,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 402612920,
+      "step": 2344
+    },
+    {
+      "epoch": 0.6168212007628066,
+      "grad_norm": 6.221194498282131,
+      "learning_rate": 5e-06,
+      "loss": 0.1581,
+      "num_input_tokens_seen": 402785192,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6168212007628066,
+      "loss": 0.14238294959068298,
+      "loss_ce": 0.0007356047863140702,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 402785192,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6170842375221938,
+      "grad_norm": 4.044951640988235,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 402955576,
+      "step": 2346
+    },
+    {
+      "epoch": 0.6170842375221938,
+      "loss": 0.0908626914024353,
+      "loss_ce": 0.0020260235760360956,
+      "loss_iou": 0.5,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 402955576,
+      "step": 2346
+    },
+    {
+      "epoch": 0.6173472742815809,
+      "grad_norm": 6.79401496913443,
+      "learning_rate": 5e-06,
+      "loss": 0.1167,
+      "num_input_tokens_seen": 403125596,
+      "step": 2347
+    },
+    {
+      "epoch": 0.6173472742815809,
+      "loss": 0.19389519095420837,
+      "loss_ce": 0.0025499900802969933,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 403125596,
+      "step": 2347
+    },
+    {
+      "epoch": 0.617610311040968,
+      "grad_norm": 4.6407931564353255,
+      "learning_rate": 5e-06,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 403295848,
+      "step": 2348
+    },
+    {
+      "epoch": 0.617610311040968,
+      "loss": 0.12875889241695404,
+      "loss_ce": 0.0007681695278733969,
+      "loss_iou": 0.671875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 403295848,
+      "step": 2348
+    },
+    {
+      "epoch": 0.6178733478003551,
+      "grad_norm": 15.448434585562659,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 403468128,
+      "step": 2349
+    },
+    {
+      "epoch": 0.6178733478003551,
+      "loss": 0.12984147667884827,
+      "loss_ce": 0.00026383629301562905,
+      "loss_iou": 0.578125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 403468128,
+      "step": 2349
+    },
+    {
+      "epoch": 0.6181363845597422,
+      "grad_norm": 6.470906735340733,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 403640328,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6181363845597422,
+      "loss": 0.08621760457754135,
+      "loss_ce": 0.002935132011771202,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 403640328,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6183994213191294,
+      "grad_norm": 3.904089267434168,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 403812552,
+      "step": 2351
+    },
+    {
+      "epoch": 0.6183994213191294,
+      "loss": 0.09386193752288818,
+      "loss_ce": 0.00026452430756762624,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 403812552,
+      "step": 2351
+    },
+    {
+      "epoch": 0.6186624580785165,
+      "grad_norm": 6.850647458248975,
+      "learning_rate": 5e-06,
+      "loss": 0.1608,
+      "num_input_tokens_seen": 403984536,
+      "step": 2352
+    },
+    {
+      "epoch": 0.6186624580785165,
+      "loss": 0.10526977479457855,
+      "loss_ce": 0.001601562718860805,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 403984536,
+      "step": 2352
+    },
+    {
+      "epoch": 0.6189254948379036,
+      "grad_norm": 6.380076499827068,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 404156840,
+      "step": 2353
+    },
+    {
+      "epoch": 0.6189254948379036,
+      "loss": 0.10814794898033142,
+      "loss_ce": 0.00020728506206069142,
+      "loss_iou": 0.5,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 404156840,
+      "step": 2353
+    },
+    {
+      "epoch": 0.6191885315972907,
+      "grad_norm": 16.46365513417012,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 404328912,
+      "step": 2354
+    },
+    {
+      "epoch": 0.6191885315972907,
+      "loss": 0.17106536030769348,
+      "loss_ce": 0.00019742565928027034,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 404328912,
+      "step": 2354
+    },
+    {
+      "epoch": 0.6194515683566778,
+      "grad_norm": 12.873967018770987,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 404501048,
+      "step": 2355
+    },
+    {
+      "epoch": 0.6194515683566778,
+      "loss": 0.1692376732826233,
+      "loss_ce": 0.0014825284015387297,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 404501048,
+      "step": 2355
+    },
+    {
+      "epoch": 0.619714605116065,
+      "grad_norm": 3.2257038395599706,
+      "learning_rate": 5e-06,
+      "loss": 0.1655,
+      "num_input_tokens_seen": 404673104,
+      "step": 2356
+    },
+    {
+      "epoch": 0.619714605116065,
+      "loss": 0.23891915380954742,
+      "loss_ce": 0.007901079021394253,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 404673104,
+      "step": 2356
+    },
+    {
+      "epoch": 0.6199776418754521,
+      "grad_norm": 7.80566516619346,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 404845264,
+      "step": 2357
+    },
+    {
+      "epoch": 0.6199776418754521,
+      "loss": 0.17429864406585693,
+      "loss_ce": 0.005078674294054508,
+      "loss_iou": 0.4375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 404845264,
+      "step": 2357
+    },
+    {
+      "epoch": 0.6202406786348392,
+      "grad_norm": 4.957580646805035,
+      "learning_rate": 5e-06,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 405017460,
+      "step": 2358
+    },
+    {
+      "epoch": 0.6202406786348392,
+      "loss": 0.05168257653713226,
+      "loss_ce": 0.003480062121525407,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 405017460,
+      "step": 2358
+    },
+    {
+      "epoch": 0.6205037153942263,
+      "grad_norm": 4.916881514910221,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 405189524,
+      "step": 2359
+    },
+    {
+      "epoch": 0.6205037153942263,
+      "loss": 0.06677095592021942,
+      "loss_ce": 0.001280229538679123,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 405189524,
+      "step": 2359
+    },
+    {
+      "epoch": 0.6207667521536134,
+      "grad_norm": 3.691480970561855,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 405362028,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6207667521536134,
+      "loss": 0.09511809051036835,
+      "loss_ce": 5.582500307355076e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 405362028,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6210297889130006,
+      "grad_norm": 23.640084314089293,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 405534164,
+      "step": 2361
+    },
+    {
+      "epoch": 0.6210297889130006,
+      "loss": 0.1902218908071518,
+      "loss_ce": 0.00021944480249658227,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 405534164,
+      "step": 2361
+    },
+    {
+      "epoch": 0.6212928256723877,
+      "grad_norm": 3.82381362472469,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 405706288,
+      "step": 2362
+    },
+    {
+      "epoch": 0.6212928256723877,
+      "loss": 0.07140006124973297,
+      "loss_ce": 0.0007060917560011148,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 405706288,
+      "step": 2362
+    },
+    {
+      "epoch": 0.6215558624317749,
+      "grad_norm": 3.7402655746033293,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 405878688,
+      "step": 2363
+    },
+    {
+      "epoch": 0.6215558624317749,
+      "loss": 0.09456443786621094,
+      "loss_ce": 0.0030269669368863106,
+      "loss_iou": 0.390625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 405878688,
+      "step": 2363
+    },
+    {
+      "epoch": 0.621818899191162,
+      "grad_norm": 5.548477550947741,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 406050532,
+      "step": 2364
+    },
+    {
+      "epoch": 0.621818899191162,
+      "loss": 0.12644195556640625,
+      "loss_ce": 0.005317692644894123,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 406050532,
+      "step": 2364
+    },
+    {
+      "epoch": 0.6220819359505491,
+      "grad_norm": 8.39488287912259,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 406220760,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6220819359505491,
+      "loss": 0.13592961430549622,
+      "loss_ce": 0.0007520002545788884,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 406220760,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6223449727099362,
+      "grad_norm": 4.84850760753409,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 406393068,
+      "step": 2366
+    },
+    {
+      "epoch": 0.6223449727099362,
+      "loss": 0.09754068404436111,
+      "loss_ce": 0.0005253083654679358,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 406393068,
+      "step": 2366
+    },
+    {
+      "epoch": 0.6226080094693234,
+      "grad_norm": 4.613478147053686,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 406565304,
+      "step": 2367
+    },
+    {
+      "epoch": 0.6226080094693234,
+      "loss": 0.1131492406129837,
+      "loss_ce": 0.002370435046032071,
+      "loss_iou": 0.625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 406565304,
+      "step": 2367
+    },
+    {
+      "epoch": 0.6228710462287105,
+      "grad_norm": 4.098258220864737,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 406737308,
+      "step": 2368
+    },
+    {
+      "epoch": 0.6228710462287105,
+      "loss": 0.09665839374065399,
+      "loss_ce": 0.0019318348495289683,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 406737308,
+      "step": 2368
+    },
+    {
+      "epoch": 0.6231340829880976,
+      "grad_norm": 10.800256206470614,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 406909296,
+      "step": 2369
+    },
+    {
+      "epoch": 0.6231340829880976,
+      "loss": 0.08466358482837677,
+      "loss_ce": 0.0036088963970541954,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 406909296,
+      "step": 2369
+    },
+    {
+      "epoch": 0.6233971197474847,
+      "grad_norm": 4.097548305836531,
+      "learning_rate": 5e-06,
+      "loss": 0.133,
+      "num_input_tokens_seen": 407078532,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6233971197474847,
+      "loss": 0.14105001091957092,
+      "loss_ce": 0.0007301777368411422,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 407078532,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6236601565068718,
+      "grad_norm": 4.885935118310897,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 407250520,
+      "step": 2371
+    },
+    {
+      "epoch": 0.6236601565068718,
+      "loss": 0.185832679271698,
+      "loss_ce": 0.001598058152012527,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 407250520,
+      "step": 2371
+    },
+    {
+      "epoch": 0.623923193266259,
+      "grad_norm": 4.243940025684864,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 407420132,
+      "step": 2372
+    },
+    {
+      "epoch": 0.623923193266259,
+      "loss": 0.17022864520549774,
+      "loss_ce": 0.0003678113571368158,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 407420132,
+      "step": 2372
+    },
+    {
+      "epoch": 0.6241862300256461,
+      "grad_norm": 6.84932788040415,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 407592472,
+      "step": 2373
+    },
+    {
+      "epoch": 0.6241862300256461,
+      "loss": 0.07611523568630219,
+      "loss_ce": 0.0008588911150582135,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 407592472,
+      "step": 2373
+    },
+    {
+      "epoch": 0.6244492667850332,
+      "grad_norm": 5.11826467429585,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 407764812,
+      "step": 2374
+    },
+    {
+      "epoch": 0.6244492667850332,
+      "loss": 0.22299307584762573,
+      "loss_ce": 0.001649091369472444,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 407764812,
+      "step": 2374
+    },
+    {
+      "epoch": 0.6247123035444203,
+      "grad_norm": 8.470257803757693,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 407937232,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6247123035444203,
+      "loss": 0.12985503673553467,
+      "loss_ce": 0.0015591441188007593,
+      "loss_iou": 0.5625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 407937232,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6249753403038074,
+      "grad_norm": 6.1769339682539135,
+      "learning_rate": 5e-06,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 408109660,
+      "step": 2376
+    },
+    {
+      "epoch": 0.6249753403038074,
+      "loss": 0.1395704746246338,
+      "loss_ce": 0.0009596287272870541,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 408109660,
+      "step": 2376
+    },
+    {
+      "epoch": 0.6252383770631946,
+      "grad_norm": 16.39235046187932,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 408281820,
+      "step": 2377
+    },
+    {
+      "epoch": 0.6252383770631946,
+      "loss": 0.12146437168121338,
+      "loss_ce": 0.0002790701691992581,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 408281820,
+      "step": 2377
+    },
+    {
+      "epoch": 0.6255014138225817,
+      "grad_norm": 7.613001701595713,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 408453984,
+      "step": 2378
+    },
+    {
+      "epoch": 0.6255014138225817,
+      "loss": 0.120358906686306,
+      "loss_ce": 0.002347429981455207,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 408453984,
+      "step": 2378
+    },
+    {
+      "epoch": 0.6257644505819688,
+      "grad_norm": 4.614954338110094,
+      "learning_rate": 5e-06,
+      "loss": 0.1215,
+      "num_input_tokens_seen": 408625956,
+      "step": 2379
+    },
+    {
+      "epoch": 0.6257644505819688,
+      "loss": 0.12903685867786407,
+      "loss_ce": 0.003334960900247097,
+      "loss_iou": 0.640625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 408625956,
+      "step": 2379
+    },
+    {
+      "epoch": 0.6260274873413559,
+      "grad_norm": 4.387875886048538,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 408797808,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6260274873413559,
+      "loss": 0.1543307602405548,
+      "loss_ce": 0.0021701250225305557,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 408797808,
+      "step": 2380
+    },
+    {
+      "epoch": 0.626290524100743,
+      "grad_norm": 5.128859786695588,
+      "learning_rate": 5e-06,
+      "loss": 0.1393,
+      "num_input_tokens_seen": 408969768,
+      "step": 2381
+    },
+    {
+      "epoch": 0.626290524100743,
+      "loss": 0.1084834560751915,
+      "loss_ce": 0.0040522972121834755,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 408969768,
+      "step": 2381
+    },
+    {
+      "epoch": 0.6265535608601303,
+      "grad_norm": 9.647359334804406,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 409141828,
+      "step": 2382
+    },
+    {
+      "epoch": 0.6265535608601303,
+      "loss": 0.253137469291687,
+      "loss_ce": 0.00036035641096532345,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 409141828,
+      "step": 2382
+    },
+    {
+      "epoch": 0.6268165976195174,
+      "grad_norm": 7.279560183961185,
+      "learning_rate": 5e-06,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 409313972,
+      "step": 2383
+    },
+    {
+      "epoch": 0.6268165976195174,
+      "loss": 0.07071413099765778,
+      "loss_ce": 0.00015749200247228146,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 409313972,
+      "step": 2383
+    },
+    {
+      "epoch": 0.6270796343789045,
+      "grad_norm": 34.71304497396249,
+      "learning_rate": 5e-06,
+      "loss": 0.1461,
+      "num_input_tokens_seen": 409482936,
+      "step": 2384
+    },
+    {
+      "epoch": 0.6270796343789045,
+      "loss": 0.07908271253108978,
+      "loss_ce": 0.0020258277654647827,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 409482936,
+      "step": 2384
+    },
+    {
+      "epoch": 0.6273426711382916,
+      "grad_norm": 3.2691371809714322,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 409655000,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6273426711382916,
+      "loss": 0.06744687259197235,
+      "loss_ce": 0.00012509411317296326,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 409655000,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6276057078976787,
+      "grad_norm": 3.653366346571365,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 409826932,
+      "step": 2386
+    },
+    {
+      "epoch": 0.6276057078976787,
+      "loss": 0.11388548463582993,
+      "loss_ce": 0.0006347582675516605,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 409826932,
+      "step": 2386
+    },
+    {
+      "epoch": 0.6278687446570659,
+      "grad_norm": 4.394076994308062,
+      "learning_rate": 5e-06,
+      "loss": 0.126,
+      "num_input_tokens_seen": 409997112,
+      "step": 2387
+    },
+    {
+      "epoch": 0.6278687446570659,
+      "loss": 0.07691079378128052,
+      "loss_ce": 0.0022647997830063105,
+      "loss_iou": 0.625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 409997112,
+      "step": 2387
+    },
+    {
+      "epoch": 0.628131781416453,
+      "grad_norm": 5.885043334484188,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 410169412,
+      "step": 2388
+    },
+    {
+      "epoch": 0.628131781416453,
+      "loss": 0.09735430032014847,
+      "loss_ce": 0.0005830569425597787,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 410169412,
+      "step": 2388
+    },
+    {
+      "epoch": 0.6283948181758401,
+      "grad_norm": 9.328795985316807,
+      "learning_rate": 5e-06,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 410341700,
+      "step": 2389
+    },
+    {
+      "epoch": 0.6283948181758401,
+      "loss": 0.08082857728004456,
+      "loss_ce": 0.0010098508791998029,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 410341700,
+      "step": 2389
+    },
+    {
+      "epoch": 0.6286578549352272,
+      "grad_norm": 4.002799196322825,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 410513660,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6286578549352272,
+      "loss": 0.054968155920505524,
+      "loss_ce": 0.000341689505148679,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 410513660,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6289208916946143,
+      "grad_norm": 6.9030334525022505,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 410685956,
+      "step": 2391
+    },
+    {
+      "epoch": 0.6289208916946143,
+      "loss": 0.15183863043785095,
+      "loss_ce": 0.0031264659482985735,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 410685956,
+      "step": 2391
+    },
+    {
+      "epoch": 0.6291839284540014,
+      "grad_norm": 5.941678610362804,
+      "learning_rate": 5e-06,
+      "loss": 0.0866,
+      "num_input_tokens_seen": 410856348,
+      "step": 2392
+    },
+    {
+      "epoch": 0.6291839284540014,
+      "loss": 0.1089685708284378,
+      "loss_ce": 0.0001734074903652072,
+      "loss_iou": 0.671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 410856348,
+      "step": 2392
+    },
+    {
+      "epoch": 0.6294469652133886,
+      "grad_norm": 7.460369582099088,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 411028268,
+      "step": 2393
+    },
+    {
+      "epoch": 0.6294469652133886,
+      "loss": 0.17290328443050385,
+      "loss_ce": 0.0006010266370140016,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 411028268,
+      "step": 2393
+    },
+    {
+      "epoch": 0.6297100019727757,
+      "grad_norm": 4.673085403235698,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 411200296,
+      "step": 2394
+    },
+    {
+      "epoch": 0.6297100019727757,
+      "loss": 0.11547736823558807,
+      "loss_ce": 0.0025928488466888666,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 411200296,
+      "step": 2394
+    },
+    {
+      "epoch": 0.6299730387321628,
+      "grad_norm": 4.914810997482069,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 411372592,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6299730387321628,
+      "loss": 0.1332027018070221,
+      "loss_ce": 0.0009089965024031699,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 411372592,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6302360754915499,
+      "grad_norm": 10.387773947226163,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 411544920,
+      "step": 2396
+    },
+    {
+      "epoch": 0.6302360754915499,
+      "loss": 0.12052314728498459,
+      "loss_ce": 0.0002533740480430424,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 411544920,
+      "step": 2396
+    },
+    {
+      "epoch": 0.630499112250937,
+      "grad_norm": 8.18207434885096,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 411717416,
+      "step": 2397
+    },
+    {
+      "epoch": 0.630499112250937,
+      "loss": 0.09294469654560089,
+      "loss_ce": 0.0004459216434042901,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 411717416,
+      "step": 2397
+    },
+    {
+      "epoch": 0.6307621490103242,
+      "grad_norm": 9.784216604991045,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 411889656,
+      "step": 2398
+    },
+    {
+      "epoch": 0.6307621490103242,
+      "loss": 0.06576241552829742,
+      "loss_ce": 0.00011910500325029716,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 411889656,
+      "step": 2398
+    },
+    {
+      "epoch": 0.6310251857697113,
+      "grad_norm": 9.394179875352377,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 412061964,
+      "step": 2399
+    },
+    {
+      "epoch": 0.6310251857697113,
+      "loss": 0.2204100638628006,
+      "loss_ce": 0.003613188164308667,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 412061964,
+      "step": 2399
+    },
+    {
+      "epoch": 0.6312882225290984,
+      "grad_norm": 5.421055277942169,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 412233844,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6312882225290984,
+      "loss": 0.058117613196372986,
+      "loss_ce": 0.00016473224968649447,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 412233844,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6315512592884855,
+      "grad_norm": 32.78384103492781,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 412404404,
+      "step": 2401
+    },
+    {
+      "epoch": 0.6315512592884855,
+      "loss": 0.11002543568611145,
+      "loss_ce": 0.0008640556479804218,
+      "loss_iou": NaN,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 412404404,
+      "step": 2401
+    },
+    {
+      "epoch": 0.6318142960478726,
+      "grad_norm": 5.144111967585496,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 412576424,
+      "step": 2402
+    },
+    {
+      "epoch": 0.6318142960478726,
+      "loss": 0.10595827549695969,
+      "loss_ce": 0.0008252161205746233,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 412576424,
+      "step": 2402
+    },
+    {
+      "epoch": 0.6320773328072599,
+      "grad_norm": 9.535369648896554,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 412748540,
+      "step": 2403
+    },
+    {
+      "epoch": 0.6320773328072599,
+      "loss": 0.08199536800384521,
+      "loss_ce": 0.0007880894117988646,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 412748540,
+      "step": 2403
+    },
+    {
+      "epoch": 0.632340369566647,
+      "grad_norm": 3.8619766281957557,
+      "learning_rate": 5e-06,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 412919044,
+      "step": 2404
+    },
+    {
+      "epoch": 0.632340369566647,
+      "loss": 0.07876091450452805,
+      "loss_ce": 0.0004375489370431751,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 412919044,
+      "step": 2404
+    },
+    {
+      "epoch": 0.6326034063260341,
+      "grad_norm": 4.690903318512814,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 413089276,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6326034063260341,
+      "loss": 0.12442415952682495,
+      "loss_ce": 0.0037271445617079735,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 413089276,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6328664430854212,
+      "grad_norm": 5.509443001792319,
+      "learning_rate": 5e-06,
+      "loss": 0.1865,
+      "num_input_tokens_seen": 413261512,
+      "step": 2406
+    },
+    {
+      "epoch": 0.6328664430854212,
+      "loss": 0.2314736247062683,
+      "loss_ce": 0.0039345622062683105,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 413261512,
+      "step": 2406
+    },
+    {
+      "epoch": 0.6331294798448083,
+      "grad_norm": 4.789366484519657,
+      "learning_rate": 5e-06,
+      "loss": 0.1568,
+      "num_input_tokens_seen": 413433928,
+      "step": 2407
+    },
+    {
+      "epoch": 0.6331294798448083,
+      "loss": 0.10793605446815491,
+      "loss_ce": 0.002009541727602482,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 413433928,
+      "step": 2407
+    },
+    {
+      "epoch": 0.6333925166041955,
+      "grad_norm": 8.59758393749405,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 413606128,
+      "step": 2408
+    },
+    {
+      "epoch": 0.6333925166041955,
+      "loss": 0.0926143079996109,
+      "loss_ce": 0.001000535092316568,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 413606128,
+      "step": 2408
+    },
+    {
+      "epoch": 0.6336555533635826,
+      "grad_norm": 9.998355206402028,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 413778176,
+      "step": 2409
+    },
+    {
+      "epoch": 0.6336555533635826,
+      "loss": 0.12268656492233276,
+      "loss_ce": 0.0002958154655061662,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 413778176,
+      "step": 2409
+    },
+    {
+      "epoch": 0.6339185901229697,
+      "grad_norm": 4.79036194916934,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 413950292,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6339185901229697,
+      "loss": 0.22198227047920227,
+      "loss_ce": 0.0022557121701538563,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 413950292,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6341816268823568,
+      "grad_norm": 21.148690076997823,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 414119124,
+      "step": 2411
+    },
+    {
+      "epoch": 0.6341816268823568,
+      "loss": 0.08586390316486359,
+      "loss_ce": 0.000368903303751722,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 414119124,
+      "step": 2411
+    },
+    {
+      "epoch": 0.6344446636417439,
+      "grad_norm": 7.240778669769743,
+      "learning_rate": 5e-06,
+      "loss": 0.1765,
+      "num_input_tokens_seen": 414291560,
+      "step": 2412
+    },
+    {
+      "epoch": 0.6344446636417439,
+      "loss": 0.24769088625907898,
+      "loss_ce": 0.0022685134317725897,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 414291560,
+      "step": 2412
+    },
+    {
+      "epoch": 0.634707700401131,
+      "grad_norm": 15.982142868420826,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 414463784,
+      "step": 2413
+    },
+    {
+      "epoch": 0.634707700401131,
+      "loss": 0.22224080562591553,
+      "loss_ce": 0.0012630214914679527,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 414463784,
+      "step": 2413
+    },
+    {
+      "epoch": 0.6349707371605182,
+      "grad_norm": 5.563886751421209,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 414635968,
+      "step": 2414
+    },
+    {
+      "epoch": 0.6349707371605182,
+      "loss": 0.12378741800785065,
+      "loss_ce": 0.0024190132971853018,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 414635968,
+      "step": 2414
+    },
+    {
+      "epoch": 0.6352337739199053,
+      "grad_norm": 6.971489879379984,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 414808184,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6352337739199053,
+      "loss": 0.09163232147693634,
+      "loss_ce": 0.00017114286310970783,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 414808184,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6354968106792924,
+      "grad_norm": 10.763166082287821,
+      "learning_rate": 5e-06,
+      "loss": 0.1798,
+      "num_input_tokens_seen": 414978292,
+      "step": 2416
+    },
+    {
+      "epoch": 0.6354968106792924,
+      "loss": 0.1381605565547943,
+      "loss_ce": 0.00016007423982955515,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 414978292,
+      "step": 2416
+    },
+    {
+      "epoch": 0.6357598474386795,
+      "grad_norm": 9.836473738616426,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 415150344,
+      "step": 2417
+    },
+    {
+      "epoch": 0.6357598474386795,
+      "loss": 0.1071944609284401,
+      "loss_ce": 0.0013594944030046463,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 415150344,
+      "step": 2417
+    },
+    {
+      "epoch": 0.6360228841980666,
+      "grad_norm": 3.9009374145565308,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 415322328,
+      "step": 2418
+    },
+    {
+      "epoch": 0.6360228841980666,
+      "loss": 0.08769555389881134,
+      "loss_ce": 0.00026269443333148956,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 415322328,
+      "step": 2418
+    },
+    {
+      "epoch": 0.6362859209574538,
+      "grad_norm": 12.748558785828688,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 415494296,
+      "step": 2419
+    },
+    {
+      "epoch": 0.6362859209574538,
+      "loss": 0.09499529004096985,
+      "loss_ce": 0.004266531206667423,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 415494296,
+      "step": 2419
+    },
+    {
+      "epoch": 0.636548957716841,
+      "grad_norm": 19.009247758406477,
+      "learning_rate": 5e-06,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 415666184,
+      "step": 2420
+    },
+    {
+      "epoch": 0.636548957716841,
+      "loss": 0.07810772955417633,
+      "loss_ce": 0.0017832687590271235,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 415666184,
+      "step": 2420
+    },
+    {
+      "epoch": 0.636811994476228,
+      "grad_norm": 4.624128366794573,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 415838228,
+      "step": 2421
+    },
+    {
+      "epoch": 0.636811994476228,
+      "loss": 0.11569841206073761,
+      "loss_ce": 0.0010743860621005297,
+      "loss_iou": 0.546875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 415838228,
+      "step": 2421
+    },
+    {
+      "epoch": 0.6370750312356152,
+      "grad_norm": 4.489659891906354,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 416010568,
+      "step": 2422
+    },
+    {
+      "epoch": 0.6370750312356152,
+      "loss": 0.09839779138565063,
+      "loss_ce": 0.0043120919726789,
+      "loss_iou": 0.578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 416010568,
+      "step": 2422
+    },
+    {
+      "epoch": 0.6373380679950023,
+      "grad_norm": 5.451345385819082,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 416182976,
+      "step": 2423
+    },
+    {
+      "epoch": 0.6373380679950023,
+      "loss": 0.11042429506778717,
+      "loss_ce": 0.005230204667896032,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 416182976,
+      "step": 2423
+    },
+    {
+      "epoch": 0.6376011047543895,
+      "grad_norm": 4.648386667387077,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 416354968,
+      "step": 2424
+    },
+    {
+      "epoch": 0.6376011047543895,
+      "loss": 0.08212631195783615,
+      "loss_ce": 0.002170257270336151,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 416354968,
+      "step": 2424
+    },
+    {
+      "epoch": 0.6378641415137766,
+      "grad_norm": 7.161469269147777,
+      "learning_rate": 5e-06,
+      "loss": 0.1544,
+      "num_input_tokens_seen": 416527268,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6378641415137766,
+      "loss": 0.1374281942844391,
+      "loss_ce": 0.0008009903831407428,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 416527268,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6381271782731637,
+      "grad_norm": 7.707879726473491,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 416699532,
+      "step": 2426
+    },
+    {
+      "epoch": 0.6381271782731637,
+      "loss": 0.06318493187427521,
+      "loss_ce": 0.00016612766194157302,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 416699532,
+      "step": 2426
+    },
+    {
+      "epoch": 0.6383902150325508,
+      "grad_norm": 4.53622350388229,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 416871872,
+      "step": 2427
+    },
+    {
+      "epoch": 0.6383902150325508,
+      "loss": 0.14566369354724884,
+      "loss_ce": 0.0013460592599585652,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 416871872,
+      "step": 2427
+    },
+    {
+      "epoch": 0.6386532517919379,
+      "grad_norm": 5.815980535010829,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 417044020,
+      "step": 2428
+    },
+    {
+      "epoch": 0.6386532517919379,
+      "loss": 0.12992171943187714,
+      "loss_ce": 0.0009391760104335845,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 417044020,
+      "step": 2428
+    },
+    {
+      "epoch": 0.6389162885513251,
+      "grad_norm": 4.632913290565498,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 417215912,
+      "step": 2429
+    },
+    {
+      "epoch": 0.6389162885513251,
+      "loss": 0.10175777971744537,
+      "loss_ce": 0.0023772907443344593,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 417215912,
+      "step": 2429
+    },
+    {
+      "epoch": 0.6391793253107122,
+      "grad_norm": 3.31675640465626,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 417388272,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6391793253107122,
+      "loss": 0.13262677192687988,
+      "loss_ce": 0.0031712136697024107,
+      "loss_iou": 0.46875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 417388272,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6394423620700993,
+      "grad_norm": 9.57220985801255,
+      "learning_rate": 5e-06,
+      "loss": 0.0935,
+      "num_input_tokens_seen": 417560304,
+      "step": 2431
+    },
+    {
+      "epoch": 0.6394423620700993,
+      "loss": 0.08074182271957397,
+      "loss_ce": 0.00218957313336432,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 417560304,
+      "step": 2431
+    },
+    {
+      "epoch": 0.6397053988294864,
+      "grad_norm": 5.444532606716201,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 417732592,
+      "step": 2432
+    },
+    {
+      "epoch": 0.6397053988294864,
+      "loss": 0.23035961389541626,
+      "loss_ce": 0.0006538145244121552,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 417732592,
+      "step": 2432
+    },
+    {
+      "epoch": 0.6399684355888735,
+      "grad_norm": 10.22830520629982,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 417904600,
+      "step": 2433
+    },
+    {
+      "epoch": 0.6399684355888735,
+      "loss": 0.08408404886722565,
+      "loss_ce": 0.0004964005202054977,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 417904600,
+      "step": 2433
+    },
+    {
+      "epoch": 0.6402314723482607,
+      "grad_norm": 4.834822071481386,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 418076896,
+      "step": 2434
+    },
+    {
+      "epoch": 0.6402314723482607,
+      "loss": 0.07859447598457336,
+      "loss_ce": 0.0008967254543676972,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 418076896,
+      "step": 2434
+    },
+    {
+      "epoch": 0.6404945091076478,
+      "grad_norm": 11.222393651192586,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 418249128,
+      "step": 2435
+    },
+    {
+      "epoch": 0.6404945091076478,
+      "loss": 0.17654769122600555,
+      "loss_ce": 0.0008885157876648009,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 418249128,
+      "step": 2435
+    },
+    {
+      "epoch": 0.6407575458670349,
+      "grad_norm": 4.617257720330074,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 418421340,
+      "step": 2436
+    },
+    {
+      "epoch": 0.6407575458670349,
+      "loss": 0.15517690777778625,
+      "loss_ce": 0.0020702139008790255,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 418421340,
+      "step": 2436
+    },
+    {
+      "epoch": 0.641020582626422,
+      "grad_norm": 4.8286585158927,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 418591736,
+      "step": 2437
+    },
+    {
+      "epoch": 0.641020582626422,
+      "loss": 0.17714877426624298,
+      "loss_ce": 0.003656333312392235,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 418591736,
+      "step": 2437
+    },
+    {
+      "epoch": 0.6412836193858091,
+      "grad_norm": 13.06198419822853,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 418763952,
+      "step": 2438
+    },
+    {
+      "epoch": 0.6412836193858091,
+      "loss": 0.1256633698940277,
+      "loss_ce": 0.0007243968429975212,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 418763952,
+      "step": 2438
+    },
+    {
+      "epoch": 0.6415466561451962,
+      "grad_norm": 18.947469777145443,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 418936088,
+      "step": 2439
+    },
+    {
+      "epoch": 0.6415466561451962,
+      "loss": 0.09127810597419739,
+      "loss_ce": 0.003967312164604664,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 418936088,
+      "step": 2439
+    },
+    {
+      "epoch": 0.6418096929045835,
+      "grad_norm": 10.72734062355858,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 419108496,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6418096929045835,
+      "loss": 0.16916052997112274,
+      "loss_ce": 0.0024124737828969955,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 419108496,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6420727296639706,
+      "grad_norm": 9.781068157842057,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 419280412,
+      "step": 2441
+    },
+    {
+      "epoch": 0.6420727296639706,
+      "loss": 0.06414149701595306,
+      "loss_ce": 0.0008175191469490528,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 419280412,
+      "step": 2441
+    },
+    {
+      "epoch": 0.6423357664233577,
+      "grad_norm": 4.632619855595305,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 419452340,
+      "step": 2442
+    },
+    {
+      "epoch": 0.6423357664233577,
+      "loss": 0.11532483249902725,
+      "loss_ce": 0.0007618411909788847,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 419452340,
+      "step": 2442
+    },
+    {
+      "epoch": 0.6425988031827448,
+      "grad_norm": 6.28676819593018,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 419624260,
+      "step": 2443
+    },
+    {
+      "epoch": 0.6425988031827448,
+      "loss": 0.08432676643133163,
+      "loss_ce": 0.0007086057448759675,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 419624260,
+      "step": 2443
+    },
+    {
+      "epoch": 0.6428618399421319,
+      "grad_norm": 4.106105061012571,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 419796396,
+      "step": 2444
+    },
+    {
+      "epoch": 0.6428618399421319,
+      "loss": 0.08845219761133194,
+      "loss_ce": 0.00016484873776789755,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 419796396,
+      "step": 2444
+    },
+    {
+      "epoch": 0.6431248767015191,
+      "grad_norm": 5.5585844707251555,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 419965620,
+      "step": 2445
+    },
+    {
+      "epoch": 0.6431248767015191,
+      "loss": 0.08765023946762085,
+      "loss_ce": 0.0027198141906410456,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 419965620,
+      "step": 2445
+    },
+    {
+      "epoch": 0.6433879134609062,
+      "grad_norm": 6.941283237766657,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 420137700,
+      "step": 2446
+    },
+    {
+      "epoch": 0.6433879134609062,
+      "loss": 0.15698650479316711,
+      "loss_ce": 0.0023844558745622635,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 420137700,
+      "step": 2446
+    },
+    {
+      "epoch": 0.6436509502202933,
+      "grad_norm": 4.724784845199971,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 420309964,
+      "step": 2447
+    },
+    {
+      "epoch": 0.6436509502202933,
+      "loss": 0.05271516367793083,
+      "loss_ce": 0.0011099397670477629,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 420309964,
+      "step": 2447
+    },
+    {
+      "epoch": 0.6439139869796804,
+      "grad_norm": 6.277321029907694,
+      "learning_rate": 5e-06,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 420482028,
+      "step": 2448
+    },
+    {
+      "epoch": 0.6439139869796804,
+      "loss": 0.19532331824302673,
+      "loss_ce": 0.0010789325460791588,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 420482028,
+      "step": 2448
+    },
+    {
+      "epoch": 0.6441770237390675,
+      "grad_norm": 13.80293628596545,
+      "learning_rate": 5e-06,
+      "loss": 0.1117,
+      "num_input_tokens_seen": 420654152,
+      "step": 2449
+    },
+    {
+      "epoch": 0.6441770237390675,
+      "loss": 0.07553352415561676,
+      "loss_ce": 0.00017035921337082982,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 420654152,
+      "step": 2449
+    },
+    {
+      "epoch": 0.6444400604984547,
+      "grad_norm": 10.20337023035234,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 420823676,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6444400604984547,
+      "loss": 0.1215822771191597,
+      "loss_ce": 0.0013124945107847452,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 420823676,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6447030972578418,
+      "grad_norm": 3.8758961183069034,
+      "learning_rate": 5e-06,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 420991548,
+      "step": 2451
+    },
+    {
+      "epoch": 0.6447030972578418,
+      "loss": 0.04900962486863136,
+      "loss_ce": 0.0008071088814176619,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 420991548,
+      "step": 2451
+    },
+    {
+      "epoch": 0.6449661340172289,
+      "grad_norm": 17.020894064566225,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 421163940,
+      "step": 2452
+    },
+    {
+      "epoch": 0.6449661340172289,
+      "loss": 0.11804142594337463,
+      "loss_ce": 6.047027636668645e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 421163940,
+      "step": 2452
+    },
+    {
+      "epoch": 0.645229170776616,
+      "grad_norm": 9.681794920323762,
+      "learning_rate": 5e-06,
+      "loss": 0.0851,
+      "num_input_tokens_seen": 421333576,
+      "step": 2453
+    },
+    {
+      "epoch": 0.645229170776616,
+      "loss": 0.1210954412817955,
+      "loss_ce": 0.0014970521442592144,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 421333576,
+      "step": 2453
+    },
+    {
+      "epoch": 0.6454922075360031,
+      "grad_norm": 4.642104923921384,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 421505944,
+      "step": 2454
+    },
+    {
+      "epoch": 0.6454922075360031,
+      "loss": 0.1038353219628334,
+      "loss_ce": 0.0009910848457366228,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 421505944,
+      "step": 2454
+    },
+    {
+      "epoch": 0.6457552442953903,
+      "grad_norm": 9.523149673908406,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 421677944,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6457552442953903,
+      "loss": 0.15976807475090027,
+      "loss_ce": 0.0006189080304466188,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 421677944,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6460182810547774,
+      "grad_norm": 5.022843580524971,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 421850200,
+      "step": 2456
+    },
+    {
+      "epoch": 0.6460182810547774,
+      "loss": 0.11521396040916443,
+      "loss_ce": 0.0040994551964104176,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 421850200,
+      "step": 2456
+    },
+    {
+      "epoch": 0.6462813178141645,
+      "grad_norm": 18.53576034320469,
+      "learning_rate": 5e-06,
+      "loss": 0.1467,
+      "num_input_tokens_seen": 422020768,
+      "step": 2457
+    },
+    {
+      "epoch": 0.6462813178141645,
+      "loss": 0.145416259765625,
+      "loss_ce": 0.002319341991096735,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 422020768,
+      "step": 2457
+    },
+    {
+      "epoch": 0.6465443545735516,
+      "grad_norm": 3.5647933372156486,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 422188996,
+      "step": 2458
+    },
+    {
+      "epoch": 0.6465443545735516,
+      "loss": 0.09572234004735947,
+      "loss_ce": 0.002552174963057041,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 422188996,
+      "step": 2458
+    },
+    {
+      "epoch": 0.6468073913329387,
+      "grad_norm": 7.264219107970771,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 422361088,
+      "step": 2459
+    },
+    {
+      "epoch": 0.6468073913329387,
+      "loss": 0.25412267446517944,
+      "loss_ce": 0.002886736299842596,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 422361088,
+      "step": 2459
+    },
+    {
+      "epoch": 0.647070428092326,
+      "grad_norm": 4.6344065191429635,
+      "learning_rate": 5e-06,
+      "loss": 0.1305,
+      "num_input_tokens_seen": 422533008,
+      "step": 2460
+    },
+    {
+      "epoch": 0.647070428092326,
+      "loss": 0.17847508192062378,
+      "loss_ce": 0.00795811414718628,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 422533008,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6473334648517131,
+      "grad_norm": 8.368380580785693,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 422705020,
+      "step": 2461
+    },
+    {
+      "epoch": 0.6473334648517131,
+      "loss": 0.11911526322364807,
+      "loss_ce": 0.0007375775021500885,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 422705020,
+      "step": 2461
+    },
+    {
+      "epoch": 0.6475965016111002,
+      "grad_norm": 14.530786026480682,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 422876816,
+      "step": 2462
+    },
+    {
+      "epoch": 0.6475965016111002,
+      "loss": 0.1273796260356903,
+      "loss_ce": 0.0006401161663234234,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 422876816,
+      "step": 2462
+    },
+    {
+      "epoch": 0.6478595383704873,
+      "grad_norm": 5.386199695848856,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 423048624,
+      "step": 2463
+    },
+    {
+      "epoch": 0.6478595383704873,
+      "loss": 0.07592228055000305,
+      "loss_ce": 0.00014713153359480202,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 423048624,
+      "step": 2463
+    },
+    {
+      "epoch": 0.6481225751298744,
+      "grad_norm": 4.520647486495735,
+      "learning_rate": 5e-06,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 423220672,
+      "step": 2464
+    },
+    {
+      "epoch": 0.6481225751298744,
+      "loss": 0.0708494782447815,
+      "loss_ce": 0.0003843960876110941,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 423220672,
+      "step": 2464
+    },
+    {
+      "epoch": 0.6483856118892615,
+      "grad_norm": 6.550669243795055,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 423392552,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6483856118892615,
+      "loss": 0.07499829679727554,
+      "loss_ce": 0.0005048871971666813,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 423392552,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6486486486486487,
+      "grad_norm": 19.46883155452454,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 423564752,
+      "step": 2466
+    },
+    {
+      "epoch": 0.6486486486486487,
+      "loss": 0.20438973605632782,
+      "loss_ce": 0.000684903294313699,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 423564752,
+      "step": 2466
+    },
+    {
+      "epoch": 0.6489116854080358,
+      "grad_norm": 5.889523777651774,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 423737504,
+      "step": 2467
+    },
+    {
+      "epoch": 0.6489116854080358,
+      "loss": 0.1293552964925766,
+      "loss_ce": 0.002432688605040312,
+      "loss_iou": 0.609375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 423737504,
+      "step": 2467
+    },
+    {
+      "epoch": 0.6491747221674229,
+      "grad_norm": 7.016832670115666,
+      "learning_rate": 5e-06,
+      "loss": 0.1505,
+      "num_input_tokens_seen": 423909608,
+      "step": 2468
+    },
+    {
+      "epoch": 0.6491747221674229,
+      "loss": 0.05709705874323845,
+      "loss_ce": 0.0013414426939561963,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 423909608,
+      "step": 2468
+    },
+    {
+      "epoch": 0.64943775892681,
+      "grad_norm": 3.9924309473660817,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 424081612,
+      "step": 2469
+    },
+    {
+      "epoch": 0.64943775892681,
+      "loss": 0.21573078632354736,
+      "loss_ce": 0.0024129238445311785,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 424081612,
+      "step": 2469
+    },
+    {
+      "epoch": 0.6497007956861971,
+      "grad_norm": 8.786316595444035,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 424253804,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6497007956861971,
+      "loss": 0.05929354950785637,
+      "loss_ce": 0.00044039852218702435,
+      "loss_iou": NaN,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 424253804,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6499638324455843,
+      "grad_norm": 15.002230358263791,
+      "learning_rate": 5e-06,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 424425936,
+      "step": 2471
+    },
+    {
+      "epoch": 0.6499638324455843,
+      "loss": 0.13272526860237122,
+      "loss_ce": 0.004475146532058716,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 424425936,
+      "step": 2471
+    },
+    {
+      "epoch": 0.6502268692049714,
+      "grad_norm": 5.3031138296122675,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 424596308,
+      "step": 2472
+    },
+    {
+      "epoch": 0.6502268692049714,
+      "loss": 0.06613775342702866,
+      "loss_ce": 0.005651914514601231,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 424596308,
+      "step": 2472
+    },
+    {
+      "epoch": 0.6504899059643585,
+      "grad_norm": 17.448583587295317,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 424768552,
+      "step": 2473
+    },
+    {
+      "epoch": 0.6504899059643585,
+      "loss": 0.08512883633375168,
+      "loss_ce": 0.0031891404651105404,
+      "loss_iou": 0.5625,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 424768552,
+      "step": 2473
+    },
+    {
+      "epoch": 0.6507529427237456,
+      "grad_norm": 6.87156581112055,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 424940764,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6507529427237456,
+      "loss": 0.07900265604257584,
+      "loss_ce": 0.0035021707881242037,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 424940764,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6510159794831327,
+      "grad_norm": 7.4967444521140765,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 425112860,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6510159794831327,
+      "loss": 0.1566176563501358,
+      "loss_ce": 0.0004592128098011017,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 425112860,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6512790162425199,
+      "grad_norm": 7.198491382802353,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 425284996,
+      "step": 2476
+    },
+    {
+      "epoch": 0.6512790162425199,
+      "loss": 0.12442326545715332,
+      "loss_ce": 0.0022308877669274807,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 425284996,
+      "step": 2476
+    },
+    {
+      "epoch": 0.651542053001907,
+      "grad_norm": 12.660738408941835,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 425457444,
+      "step": 2477
+    },
+    {
+      "epoch": 0.651542053001907,
+      "loss": 0.139640212059021,
+      "loss_ce": 0.0008462676778435707,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 425457444,
+      "step": 2477
+    },
+    {
+      "epoch": 0.6518050897612941,
+      "grad_norm": 4.788908959245028,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 425629356,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6518050897612941,
+      "loss": 0.11505892872810364,
+      "loss_ce": 0.0043716710060834885,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 425629356,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6520681265206812,
+      "grad_norm": 14.713914944582257,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 425801412,
+      "step": 2479
+    },
+    {
+      "epoch": 0.6520681265206812,
+      "loss": 0.09329426288604736,
+      "loss_ce": 0.0003987499512732029,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 425801412,
+      "step": 2479
+    },
+    {
+      "epoch": 0.6523311632800683,
+      "grad_norm": 11.370966467516885,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 425973396,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6523311632800683,
+      "loss": 0.12143571674823761,
+      "loss_ce": 0.0008607673225924373,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 425973396,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6525942000394556,
+      "grad_norm": 16.96417675320112,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 426142788,
+      "step": 2481
+    },
+    {
+      "epoch": 0.6525942000394556,
+      "loss": 0.058773696422576904,
+      "loss_ce": 0.00017994196969084442,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 426142788,
+      "step": 2481
+    },
+    {
+      "epoch": 0.6528572367988427,
+      "grad_norm": 3.267968495608878,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 426314852,
+      "step": 2482
+    },
+    {
+      "epoch": 0.6528572367988427,
+      "loss": 0.14266598224639893,
+      "loss_ce": 0.0022698603570461273,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 426314852,
+      "step": 2482
+    },
+    {
+      "epoch": 0.6531202735582298,
+      "grad_norm": 12.820952382520073,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 426487324,
+      "step": 2483
+    },
+    {
+      "epoch": 0.6531202735582298,
+      "loss": 0.09693928062915802,
+      "loss_ce": 0.0016328811179846525,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 426487324,
+      "step": 2483
+    },
+    {
+      "epoch": 0.6533833103176169,
+      "grad_norm": 10.800205106839552,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 426659468,
+      "step": 2484
+    },
+    {
+      "epoch": 0.6533833103176169,
+      "loss": 0.07587607949972153,
+      "loss_ce": 0.00039084581658244133,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 426659468,
+      "step": 2484
+    },
+    {
+      "epoch": 0.653646347077004,
+      "grad_norm": 7.9171669810832395,
+      "learning_rate": 5e-06,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 426831496,
+      "step": 2485
+    },
+    {
+      "epoch": 0.653646347077004,
+      "loss": 0.07471035420894623,
+      "loss_ce": 0.0035586238373070955,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 426831496,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6539093838363912,
+      "grad_norm": 8.230428013130412,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 427003612,
+      "step": 2486
+    },
+    {
+      "epoch": 0.6539093838363912,
+      "loss": 0.08275149762630463,
+      "loss_ce": 0.0038025237154215574,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 427003612,
+      "step": 2486
+    },
+    {
+      "epoch": 0.6541724205957783,
+      "grad_norm": 3.632885838865945,
+      "learning_rate": 5e-06,
+      "loss": 0.1363,
+      "num_input_tokens_seen": 427175916,
+      "step": 2487
+    },
+    {
+      "epoch": 0.6541724205957783,
+      "loss": 0.1029161587357521,
+      "loss_ce": 0.000987447565421462,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 427175916,
+      "step": 2487
+    },
+    {
+      "epoch": 0.6544354573551654,
+      "grad_norm": 3.5649981841807588,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 427348060,
+      "step": 2488
+    },
+    {
+      "epoch": 0.6544354573551654,
+      "loss": 0.07192617654800415,
+      "loss_ce": 0.003597316797822714,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 427348060,
+      "step": 2488
+    },
+    {
+      "epoch": 0.6546984941145525,
+      "grad_norm": 5.397602518661847,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 427520416,
+      "step": 2489
+    },
+    {
+      "epoch": 0.6546984941145525,
+      "loss": 0.08971969783306122,
+      "loss_ce": 0.0012797524686902761,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 427520416,
+      "step": 2489
+    },
+    {
+      "epoch": 0.6549615308739396,
+      "grad_norm": 2.996287220023088,
+      "learning_rate": 5e-06,
+      "loss": 0.0974,
+      "num_input_tokens_seen": 427692328,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6549615308739396,
+      "loss": 0.11320722848176956,
+      "loss_ce": 0.0006584011716768146,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 427692328,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6552245676333267,
+      "grad_norm": 9.620885086024206,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 427864376,
+      "step": 2491
+    },
+    {
+      "epoch": 0.6552245676333267,
+      "loss": 0.10166727006435394,
+      "loss_ce": 0.0020884163677692413,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 427864376,
+      "step": 2491
+    },
+    {
+      "epoch": 0.6554876043927139,
+      "grad_norm": 38.296683588407596,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 428036792,
+      "step": 2492
+    },
+    {
+      "epoch": 0.6554876043927139,
+      "loss": 0.06815087795257568,
+      "loss_ce": 6.61654194118455e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 428036792,
+      "step": 2492
+    },
+    {
+      "epoch": 0.655750641152101,
+      "grad_norm": 29.525923288124442,
+      "learning_rate": 5e-06,
+      "loss": 0.1634,
+      "num_input_tokens_seen": 428207124,
+      "step": 2493
+    },
+    {
+      "epoch": 0.655750641152101,
+      "loss": 0.23705750703811646,
+      "loss_ce": 0.0035522649995982647,
+      "loss_iou": 0.59375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 428207124,
+      "step": 2493
+    },
+    {
+      "epoch": 0.6560136779114881,
+      "grad_norm": 5.543812647744825,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 428379688,
+      "step": 2494
+    },
+    {
+      "epoch": 0.6560136779114881,
+      "loss": 0.09441094100475311,
+      "loss_ce": 0.0024462228175252676,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 428379688,
+      "step": 2494
+    },
+    {
+      "epoch": 0.6562767146708752,
+      "grad_norm": 32.70215077495664,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 428551908,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6562767146708752,
+      "loss": 0.05353452265262604,
+      "loss_ce": 0.00015927490312606096,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 428551908,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6565397514302623,
+      "grad_norm": 9.54910159759898,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 428724396,
+      "step": 2496
+    },
+    {
+      "epoch": 0.6565397514302623,
+      "loss": 0.1285240650177002,
+      "loss_ce": 0.00035023505915887654,
+      "loss_iou": 0.609375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 428724396,
+      "step": 2496
+    },
+    {
+      "epoch": 0.6568027881896495,
+      "grad_norm": 22.277656875348782,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 428891900,
+      "step": 2497
+    },
+    {
+      "epoch": 0.6568027881896495,
+      "loss": 0.08849343657493591,
+      "loss_ce": 0.0024643833748996258,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 428891900,
+      "step": 2497
+    },
+    {
+      "epoch": 0.6570658249490366,
+      "grad_norm": 6.04979449937967,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 429064016,
+      "step": 2498
+    },
+    {
+      "epoch": 0.6570658249490366,
+      "loss": 0.04533851146697998,
+      "loss_ce": 0.000935432268306613,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 429064016,
+      "step": 2498
+    },
+    {
+      "epoch": 0.6573288617084238,
+      "grad_norm": 13.279231347366395,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 429236096,
+      "step": 2499
+    },
+    {
+      "epoch": 0.6573288617084238,
+      "loss": 0.09342057257890701,
+      "loss_ce": 0.0018373207421973348,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 429236096,
+      "step": 2499
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "grad_norm": 9.388446901609708,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_websight_new_CIoU": 0.8709481358528137,
+      "eval_websight_new_GIoU": 0.8742890954017639,
+      "eval_websight_new_IoU": 0.8760707080364227,
+      "eval_websight_new_MAE_all": 0.01990661583840847,
+      "eval_websight_new_MAE_h": 0.01017875224351883,
+      "eval_websight_new_MAE_w": 0.03395752049982548,
+      "eval_websight_new_MAE_x": 0.03019585181027651,
+      "eval_websight_new_MAE_y": 0.005294335773214698,
+      "eval_websight_new_NUM_probability": 0.9999906122684479,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.09985960274934769,
+      "eval_websight_new_loss_ce": 2.431606571917655e-05,
+      "eval_websight_new_loss_iou": 0.4124755859375,
+      "eval_websight_new_loss_num": 0.017595291137695312,
+      "eval_websight_new_loss_xval": 0.08795166015625,
+      "eval_websight_new_runtime": 55.684,
+      "eval_websight_new_samples_per_second": 0.898,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_seeclick_CIoU": 0.6180358529090881,
+      "eval_seeclick_GIoU": 0.6200732290744781,
+      "eval_seeclick_IoU": 0.6430684626102448,
+      "eval_seeclick_MAE_all": 0.04816816933453083,
+      "eval_seeclick_MAE_h": 0.03338594362139702,
+      "eval_seeclick_MAE_w": 0.06272775307297707,
+      "eval_seeclick_MAE_x": 0.06945410370826721,
+      "eval_seeclick_MAE_y": 0.027104882523417473,
+      "eval_seeclick_NUM_probability": 0.9999706149101257,
+      "eval_seeclick_inside_bbox": 0.8764204680919647,
+      "eval_seeclick_loss": 0.22014649212360382,
+      "eval_seeclick_loss_ce": 0.008939406834542751,
+      "eval_seeclick_loss_iou": 0.5093994140625,
+      "eval_seeclick_loss_num": 0.040313720703125,
+      "eval_seeclick_loss_xval": 0.201568603515625,
+      "eval_seeclick_runtime": 68.7244,
+      "eval_seeclick_samples_per_second": 0.626,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_icons_CIoU": 0.8709467053413391,
+      "eval_icons_GIoU": 0.8682527244091034,
+      "eval_icons_IoU": 0.874538779258728,
+      "eval_icons_MAE_all": 0.01756941620260477,
+      "eval_icons_MAE_h": 0.02226724848151207,
+      "eval_icons_MAE_w": 0.020997921004891396,
+      "eval_icons_MAE_x": 0.0130357148591429,
+      "eval_icons_MAE_y": 0.013976779766380787,
+      "eval_icons_NUM_probability": 0.9999793469905853,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.06909541040658951,
+      "eval_icons_loss_ce": 1.1558900041563902e-05,
+      "eval_icons_loss_iou": 0.66064453125,
+      "eval_icons_loss_num": 0.01297760009765625,
+      "eval_icons_loss_xval": 0.0648956298828125,
+      "eval_icons_runtime": 83.8008,
+      "eval_icons_samples_per_second": 0.597,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_screenspot_CIoU": 0.5514570474624634,
+      "eval_screenspot_GIoU": 0.5452684958775839,
+      "eval_screenspot_IoU": 0.5914793411890665,
+      "eval_screenspot_MAE_all": 0.08535195142030716,
+      "eval_screenspot_MAE_h": 0.06157554934422175,
+      "eval_screenspot_MAE_w": 0.14490507543087006,
+      "eval_screenspot_MAE_x": 0.08271919315059979,
+      "eval_screenspot_MAE_y": 0.05220799893140793,
+      "eval_screenspot_NUM_probability": 0.9994663198788961,
+      "eval_screenspot_inside_bbox": 0.850000003973643,
+      "eval_screenspot_loss": 0.9172521233558655,
+      "eval_screenspot_loss_ce": 0.5530519783496857,
+      "eval_screenspot_loss_iou": 0.529052734375,
+      "eval_screenspot_loss_num": 0.07132466634114583,
+      "eval_screenspot_loss_xval": 0.3565266927083333,
+      "eval_screenspot_runtime": 139.3325,
+      "eval_screenspot_samples_per_second": 0.639,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.657854935227198,
+      "grad_norm": 4.075831219377808,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 429580660,
+      "step": 2501
+    },
+    {
+      "epoch": 0.657854935227198,
+      "loss": 0.13817401230335236,
+      "loss_ce": 0.0033473544754087925,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 429580660,
+      "step": 2501
+    },
+    {
+      "epoch": 0.6581179719865852,
+      "grad_norm": 7.116588034171532,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 429752720,
+      "step": 2502
+    },
+    {
+      "epoch": 0.6581179719865852,
+      "loss": 0.1050395742058754,
+      "loss_ce": 0.0004253170336596668,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 429752720,
+      "step": 2502
+    },
+    {
+      "epoch": 0.6583810087459723,
+      "grad_norm": 4.475079180174651,
+      "learning_rate": 5e-06,
+      "loss": 0.0863,
+      "num_input_tokens_seen": 429924908,
+      "step": 2503
+    },
+    {
+      "epoch": 0.6583810087459723,
+      "loss": 0.06600432842969894,
+      "loss_ce": 0.0007577461656183004,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 429924908,
+      "step": 2503
+    },
+    {
+      "epoch": 0.6586440455053594,
+      "grad_norm": 4.723970217436198,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 430097000,
+      "step": 2504
+    },
+    {
+      "epoch": 0.6586440455053594,
+      "loss": 0.07487765699625015,
+      "loss_ce": 0.0005978714907541871,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 430097000,
+      "step": 2504
+    },
+    {
+      "epoch": 0.6589070822647465,
+      "grad_norm": 14.464350947162439,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 430266684,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6589070822647465,
+      "loss": 0.24609506130218506,
+      "loss_ce": 0.00015389968757517636,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 430266684,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6591701190241336,
+      "grad_norm": 4.076117698308253,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 430436940,
+      "step": 2506
+    },
+    {
+      "epoch": 0.6591701190241336,
+      "loss": 0.109318308532238,
+      "loss_ce": 0.002522045513615012,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 430436940,
+      "step": 2506
+    },
+    {
+      "epoch": 0.6594331557835208,
+      "grad_norm": 5.795601298430443,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 430608572,
+      "step": 2507
+    },
+    {
+      "epoch": 0.6594331557835208,
+      "loss": 0.07380083203315735,
+      "loss_ce": 0.00034502719063311815,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 430608572,
+      "step": 2507
+    },
+    {
+      "epoch": 0.6596961925429079,
+      "grad_norm": 3.8858425270409165,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 430778744,
+      "step": 2508
+    },
+    {
+      "epoch": 0.6596961925429079,
+      "loss": 0.1448889970779419,
+      "loss_ce": 0.00018989352975040674,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 430778744,
+      "step": 2508
+    },
+    {
+      "epoch": 0.659959229302295,
+      "grad_norm": 9.24454413883364,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 430950904,
+      "step": 2509
+    },
+    {
+      "epoch": 0.659959229302295,
+      "loss": 0.09062638133764267,
+      "loss_ce": 0.0013014276046305895,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 430950904,
+      "step": 2509
+    },
+    {
+      "epoch": 0.6602222660616821,
+      "grad_norm": 5.68970165561044,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 431121056,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6602222660616821,
+      "loss": 0.09496060013771057,
+      "loss_ce": 0.0006155127775855362,
+      "loss_iou": NaN,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 431121056,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6604853028210692,
+      "grad_norm": 3.2016449128749023,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 431293012,
+      "step": 2511
+    },
+    {
+      "epoch": 0.6604853028210692,
+      "loss": 0.03823622688651085,
+      "loss_ce": 1.2961418178747408e-05,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 431293012,
+      "step": 2511
+    },
+    {
+      "epoch": 0.6607483395804564,
+      "grad_norm": 7.269277911217335,
+      "learning_rate": 5e-06,
+      "loss": 0.145,
+      "num_input_tokens_seen": 431461708,
+      "step": 2512
+    },
+    {
+      "epoch": 0.6607483395804564,
+      "loss": 0.27740049362182617,
+      "loss_ce": 0.004237642977386713,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 431461708,
+      "step": 2512
+    },
+    {
+      "epoch": 0.6610113763398435,
+      "grad_norm": 3.005900285482203,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 431633828,
+      "step": 2513
+    },
+    {
+      "epoch": 0.6610113763398435,
+      "loss": 0.0913277417421341,
+      "loss_ce": 0.0005684680072590709,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 431633828,
+      "step": 2513
+    },
+    {
+      "epoch": 0.6612744130992306,
+      "grad_norm": 3.8195415395135934,
+      "learning_rate": 5e-06,
+      "loss": 0.098,
+      "num_input_tokens_seen": 431805968,
+      "step": 2514
+    },
+    {
+      "epoch": 0.6612744130992306,
+      "loss": 0.04755711182951927,
+      "loss_ce": 0.0009872873779386282,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 431805968,
+      "step": 2514
+    },
+    {
+      "epoch": 0.6615374498586177,
+      "grad_norm": 4.443605112638486,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 431977936,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6615374498586177,
+      "loss": 0.0916135311126709,
+      "loss_ce": 0.0008542468422092497,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 431977936,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6618004866180048,
+      "grad_norm": 5.518155601836215,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 432150116,
+      "step": 2516
+    },
+    {
+      "epoch": 0.6618004866180048,
+      "loss": 0.14211586117744446,
+      "loss_ce": 0.0029557030647993088,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 432150116,
+      "step": 2516
+    },
+    {
+      "epoch": 0.6620635233773919,
+      "grad_norm": 5.250876023224286,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 432322696,
+      "step": 2517
+    },
+    {
+      "epoch": 0.6620635233773919,
+      "loss": 0.07852162420749664,
+      "loss_ce": 0.00102223118301481,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 432322696,
+      "step": 2517
+    },
+    {
+      "epoch": 0.6623265601367792,
+      "grad_norm": 4.979602548630015,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 432490748,
+      "step": 2518
+    },
+    {
+      "epoch": 0.6623265601367792,
+      "loss": 0.07791407406330109,
+      "loss_ce": 0.0011623608879745007,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 432490748,
+      "step": 2518
+    },
+    {
+      "epoch": 0.6625895968961663,
+      "grad_norm": 7.14143158314321,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 432662924,
+      "step": 2519
+    },
+    {
+      "epoch": 0.6625895968961663,
+      "loss": 0.168125718832016,
+      "loss_ce": 0.000858865212649107,
+      "loss_iou": 0.515625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 432662924,
+      "step": 2519
+    },
+    {
+      "epoch": 0.6628526336555534,
+      "grad_norm": 3.8877457754066387,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 432835068,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6628526336555534,
+      "loss": 0.12645836174488068,
+      "loss_ce": 0.0006496440037153661,
+      "loss_iou": 0.546875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 432835068,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6631156704149405,
+      "grad_norm": 4.906018490784193,
+      "learning_rate": 5e-06,
+      "loss": 0.1167,
+      "num_input_tokens_seen": 433007048,
+      "step": 2521
+    },
+    {
+      "epoch": 0.6631156704149405,
+      "loss": 0.08021432906389236,
+      "loss_ce": 0.000868627626914531,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 433007048,
+      "step": 2521
+    },
+    {
+      "epoch": 0.6633787071743276,
+      "grad_norm": 4.283295443334077,
+      "learning_rate": 5e-06,
+      "loss": 0.158,
+      "num_input_tokens_seen": 433177068,
+      "step": 2522
+    },
+    {
+      "epoch": 0.6633787071743276,
+      "loss": 0.11892493069171906,
+      "loss_ce": 0.00060828379355371,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 433177068,
+      "step": 2522
+    },
+    {
+      "epoch": 0.6636417439337148,
+      "grad_norm": 4.831641088226549,
+      "learning_rate": 5e-06,
+      "loss": 0.0736,
+      "num_input_tokens_seen": 433349504,
+      "step": 2523
+    },
+    {
+      "epoch": 0.6636417439337148,
+      "loss": 0.06323867291212082,
+      "loss_ce": 0.00031142536317929626,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 433349504,
+      "step": 2523
+    },
+    {
+      "epoch": 0.6639047806931019,
+      "grad_norm": 24.62858299579612,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 433521616,
+      "step": 2524
+    },
+    {
+      "epoch": 0.6639047806931019,
+      "loss": 0.12700864672660828,
+      "loss_ce": 0.003351423656567931,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 433521616,
+      "step": 2524
+    },
+    {
+      "epoch": 0.664167817452489,
+      "grad_norm": 6.377416561201896,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 433692012,
+      "step": 2525
+    },
+    {
+      "epoch": 0.664167817452489,
+      "loss": 0.06224376708269119,
+      "loss_ce": 0.003238032106310129,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 433692012,
+      "step": 2525
+    },
+    {
+      "epoch": 0.6644308542118761,
+      "grad_norm": 7.501801025322401,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 433864040,
+      "step": 2526
+    },
+    {
+      "epoch": 0.6644308542118761,
+      "loss": 0.09891114383935928,
+      "loss_ce": 0.002109395107254386,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 433864040,
+      "step": 2526
+    },
+    {
+      "epoch": 0.6646938909712632,
+      "grad_norm": 4.527105351225178,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 434034556,
+      "step": 2527
+    },
+    {
+      "epoch": 0.6646938909712632,
+      "loss": 0.14892783761024475,
+      "loss_ce": 0.0010244110599160194,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 434034556,
+      "step": 2527
+    },
+    {
+      "epoch": 0.6649569277306504,
+      "grad_norm": 31.749018149089892,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 434206672,
+      "step": 2528
+    },
+    {
+      "epoch": 0.6649569277306504,
+      "loss": 0.12175662815570831,
+      "loss_ce": 0.0015021114377304912,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 434206672,
+      "step": 2528
+    },
+    {
+      "epoch": 0.6652199644900375,
+      "grad_norm": 4.8576669690091885,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 434378728,
+      "step": 2529
+    },
+    {
+      "epoch": 0.6652199644900375,
+      "loss": 0.07344581931829453,
+      "loss_ce": 0.006276628468185663,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 434378728,
+      "step": 2529
+    },
+    {
+      "epoch": 0.6654830012494246,
+      "grad_norm": 4.9340551800619,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 434550708,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6654830012494246,
+      "loss": 0.0640825480222702,
+      "loss_ce": 0.0013689253246411681,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 434550708,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6657460380088117,
+      "grad_norm": 9.63881283843686,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 434723100,
+      "step": 2531
+    },
+    {
+      "epoch": 0.6657460380088117,
+      "loss": 0.050916872918605804,
+      "loss_ce": 0.0004713151138275862,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 434723100,
+      "step": 2531
+    },
+    {
+      "epoch": 0.6660090747681988,
+      "grad_norm": 5.34856475630482,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 434893672,
+      "step": 2532
+    },
+    {
+      "epoch": 0.6660090747681988,
+      "loss": 0.11488111317157745,
+      "loss_ce": 0.00108106411062181,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 434893672,
+      "step": 2532
+    },
+    {
+      "epoch": 0.666272111527586,
+      "grad_norm": 6.482198264697821,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 435065996,
+      "step": 2533
+    },
+    {
+      "epoch": 0.666272111527586,
+      "loss": 0.1057998389005661,
+      "loss_ce": 0.00013272129581309855,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 435065996,
+      "step": 2533
+    },
+    {
+      "epoch": 0.6665351482869731,
+      "grad_norm": 15.608471694941006,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 435236164,
+      "step": 2534
+    },
+    {
+      "epoch": 0.6665351482869731,
+      "loss": 0.08966468274593353,
+      "loss_ce": 0.004398562014102936,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 435236164,
+      "step": 2534
+    },
+    {
+      "epoch": 0.6667981850463602,
+      "grad_norm": 3.821066216794497,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 435408552,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6667981850463602,
+      "loss": 0.13420158624649048,
+      "loss_ce": 0.0008855484193190932,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 435408552,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6670612218057473,
+      "grad_norm": 7.414329883383495,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 435580704,
+      "step": 2536
+    },
+    {
+      "epoch": 0.6670612218057473,
+      "loss": 0.09095387160778046,
+      "loss_ce": 0.0024223732762038708,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 435580704,
+      "step": 2536
+    },
+    {
+      "epoch": 0.6673242585651344,
+      "grad_norm": 33.27432314606939,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 435752624,
+      "step": 2537
+    },
+    {
+      "epoch": 0.6673242585651344,
+      "loss": 0.15344524383544922,
+      "loss_ce": 0.0014066637959331274,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 435752624,
+      "step": 2537
+    },
+    {
+      "epoch": 0.6675872953245217,
+      "grad_norm": 7.327533892405677,
+      "learning_rate": 5e-06,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 435924672,
+      "step": 2538
+    },
+    {
+      "epoch": 0.6675872953245217,
+      "loss": 0.11101265996694565,
+      "loss_ce": 0.000935757125262171,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 435924672,
+      "step": 2538
+    },
+    {
+      "epoch": 0.6678503320839088,
+      "grad_norm": 7.348552482036463,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 436096756,
+      "step": 2539
+    },
+    {
+      "epoch": 0.6678503320839088,
+      "loss": 0.1398707926273346,
+      "loss_ce": 0.0033656705636531115,
+      "loss_iou": NaN,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 436096756,
+      "step": 2539
+    },
+    {
+      "epoch": 0.6681133688432959,
+      "grad_norm": 20.204014590374456,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 436267320,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6681133688432959,
+      "loss": 0.11994585394859314,
+      "loss_ce": 0.0005916071822866797,
+      "loss_iou": 0.4375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 436267320,
+      "step": 2540
+    },
+    {
+      "epoch": 0.668376405602683,
+      "grad_norm": 8.407025005461401,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 436439328,
+      "step": 2541
+    },
+    {
+      "epoch": 0.668376405602683,
+      "loss": 0.11520107090473175,
+      "loss_ce": 0.003811910282820463,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 436439328,
+      "step": 2541
+    },
+    {
+      "epoch": 0.6686394423620701,
+      "grad_norm": 5.145720476296494,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 436609828,
+      "step": 2542
+    },
+    {
+      "epoch": 0.6686394423620701,
+      "loss": 0.06836480647325516,
+      "loss_ce": 0.00037164578679949045,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 436609828,
+      "step": 2542
+    },
+    {
+      "epoch": 0.6689024791214572,
+      "grad_norm": 8.206215780846446,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 436781764,
+      "step": 2543
+    },
+    {
+      "epoch": 0.6689024791214572,
+      "loss": 0.107704758644104,
+      "loss_ce": 0.002266527386382222,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 436781764,
+      "step": 2543
+    },
+    {
+      "epoch": 0.6691655158808444,
+      "grad_norm": 7.4768704044685865,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 436953548,
+      "step": 2544
+    },
+    {
+      "epoch": 0.6691655158808444,
+      "loss": 0.17240890860557556,
+      "loss_ce": 0.0010527035919949412,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 436953548,
+      "step": 2544
+    },
+    {
+      "epoch": 0.6694285526402315,
+      "grad_norm": 9.203598751541529,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 437125648,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6694285526402315,
+      "loss": 0.0696527361869812,
+      "loss_ce": 0.0004388623929116875,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 437125648,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6696915893996186,
+      "grad_norm": 4.3933164564180816,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 437297756,
+      "step": 2546
+    },
+    {
+      "epoch": 0.6696915893996186,
+      "loss": 0.12276500463485718,
+      "loss_ce": 0.004127927124500275,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 437297756,
+      "step": 2546
+    },
+    {
+      "epoch": 0.6699546261590057,
+      "grad_norm": 4.6906533688912795,
+      "learning_rate": 5e-06,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 437470168,
+      "step": 2547
+    },
+    {
+      "epoch": 0.6699546261590057,
+      "loss": 0.16925577819347382,
+      "loss_ce": 0.00012736135977320373,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 437470168,
+      "step": 2547
+    },
+    {
+      "epoch": 0.6702176629183928,
+      "grad_norm": 5.700480300794898,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 437641860,
+      "step": 2548
+    },
+    {
+      "epoch": 0.6702176629183928,
+      "loss": 0.08314710855484009,
+      "loss_ce": 0.0011463778791949153,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 437641860,
+      "step": 2548
+    },
+    {
+      "epoch": 0.67048069967778,
+      "grad_norm": 4.582897287065671,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 437814364,
+      "step": 2549
+    },
+    {
+      "epoch": 0.67048069967778,
+      "loss": 0.1702890694141388,
+      "loss_ce": 0.0015268486458808184,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 437814364,
+      "step": 2549
+    },
+    {
+      "epoch": 0.6707437364371671,
+      "grad_norm": 5.613370832938644,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 437986344,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6707437364371671,
+      "loss": 0.13855043053627014,
+      "loss_ce": 0.0012823636643588543,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 437986344,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6710067731965542,
+      "grad_norm": 17.044687935022132,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 438158712,
+      "step": 2551
+    },
+    {
+      "epoch": 0.6710067731965542,
+      "loss": 0.08713387697935104,
+      "loss_ce": 0.00023507134756073356,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 438158712,
+      "step": 2551
+    },
+    {
+      "epoch": 0.6712698099559413,
+      "grad_norm": 6.906118924960931,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 438330820,
+      "step": 2552
+    },
+    {
+      "epoch": 0.6712698099559413,
+      "loss": 0.13604578375816345,
+      "loss_ce": 0.0038131249602884054,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 438330820,
+      "step": 2552
+    },
+    {
+      "epoch": 0.6715328467153284,
+      "grad_norm": 5.001208896831707,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 438503128,
+      "step": 2553
+    },
+    {
+      "epoch": 0.6715328467153284,
+      "loss": 0.11509568989276886,
+      "loss_ce": 0.00016648891323711723,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 438503128,
+      "step": 2553
+    },
+    {
+      "epoch": 0.6717958834747156,
+      "grad_norm": 5.88413772044494,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 438675408,
+      "step": 2554
+    },
+    {
+      "epoch": 0.6717958834747156,
+      "loss": 0.06504229456186295,
+      "loss_ce": 0.000772272062022239,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 438675408,
+      "step": 2554
+    },
+    {
+      "epoch": 0.6720589202341027,
+      "grad_norm": 5.286645052918952,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 438847472,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6720589202341027,
+      "loss": 0.11925958842039108,
+      "loss_ce": 0.00024103187024593353,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 438847472,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6723219569934898,
+      "grad_norm": 7.503881873733868,
+      "learning_rate": 5e-06,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 439019864,
+      "step": 2556
+    },
+    {
+      "epoch": 0.6723219569934898,
+      "loss": 0.1399090439081192,
+      "loss_ce": 0.0017254444537684321,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 439019864,
+      "step": 2556
+    },
+    {
+      "epoch": 0.672584993752877,
+      "grad_norm": 4.642912313294466,
+      "learning_rate": 5e-06,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 439191776,
+      "step": 2557
+    },
+    {
+      "epoch": 0.672584993752877,
+      "loss": 0.12639446556568146,
+      "loss_ce": 0.0015470522921532393,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 439191776,
+      "step": 2557
+    },
+    {
+      "epoch": 0.672848030512264,
+      "grad_norm": 5.916383494046343,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 439360860,
+      "step": 2558
+    },
+    {
+      "epoch": 0.672848030512264,
+      "loss": 0.09076009690761566,
+      "loss_ce": 0.0006722048274241388,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 439360860,
+      "step": 2558
+    },
+    {
+      "epoch": 0.6731110672716513,
+      "grad_norm": 20.566659262884578,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 439532496,
+      "step": 2559
+    },
+    {
+      "epoch": 0.6731110672716513,
+      "loss": 0.10060098767280579,
+      "loss_ce": 0.003646642668172717,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 439532496,
+      "step": 2559
+    },
+    {
+      "epoch": 0.6733741040310384,
+      "grad_norm": 4.011859025446864,
+      "learning_rate": 5e-06,
+      "loss": 0.1167,
+      "num_input_tokens_seen": 439704520,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6733741040310384,
+      "loss": 0.0790782943367958,
+      "loss_ce": 0.0015483875758945942,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 439704520,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6736371407904255,
+      "grad_norm": 7.247619498755374,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 439877048,
+      "step": 2561
+    },
+    {
+      "epoch": 0.6736371407904255,
+      "loss": 0.09221772849559784,
+      "loss_ce": 0.0026181198190897703,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 439877048,
+      "step": 2561
+    },
+    {
+      "epoch": 0.6739001775498126,
+      "grad_norm": 6.385962315237775,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 440049300,
+      "step": 2562
+    },
+    {
+      "epoch": 0.6739001775498126,
+      "loss": 0.041686464101076126,
+      "loss_ce": 0.0010523094097152352,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 440049300,
+      "step": 2562
+    },
+    {
+      "epoch": 0.6741632143091997,
+      "grad_norm": 26.7431708970354,
+      "learning_rate": 5e-06,
+      "loss": 0.1921,
+      "num_input_tokens_seen": 440221368,
+      "step": 2563
+    },
+    {
+      "epoch": 0.6741632143091997,
+      "loss": 0.18678849935531616,
+      "loss_ce": 0.002889578230679035,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 440221368,
+      "step": 2563
+    },
+    {
+      "epoch": 0.6744262510685868,
+      "grad_norm": 3.4897355882240686,
+      "learning_rate": 5e-06,
+      "loss": 0.131,
+      "num_input_tokens_seen": 440393588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.6744262510685868,
+      "loss": 0.13501334190368652,
+      "loss_ce": 0.0007817824953235686,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 440393588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.674689287827974,
+      "grad_norm": 5.796770664321287,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 440565712,
+      "step": 2565
+    },
+    {
+      "epoch": 0.674689287827974,
+      "loss": 0.06512662768363953,
+      "loss_ce": 0.0005819504731334746,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 440565712,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6749523245873611,
+      "grad_norm": 6.5511648198704755,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 440737988,
+      "step": 2566
+    },
+    {
+      "epoch": 0.6749523245873611,
+      "loss": 0.11750826984643936,
+      "loss_ce": 0.0036166671197861433,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 440737988,
+      "step": 2566
+    },
+    {
+      "epoch": 0.6752153613467482,
+      "grad_norm": 21.358635365800044,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 440910068,
+      "step": 2567
+    },
+    {
+      "epoch": 0.6752153613467482,
+      "loss": 0.10725726187229156,
+      "loss_ce": 0.0005067750462330878,
+      "loss_iou": 0.6875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 440910068,
+      "step": 2567
+    },
+    {
+      "epoch": 0.6754783981061353,
+      "grad_norm": 4.519751532821699,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 441080800,
+      "step": 2568
+    },
+    {
+      "epoch": 0.6754783981061353,
+      "loss": 0.0696103498339653,
+      "loss_ce": 0.00015234279271680862,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 441080800,
+      "step": 2568
+    },
+    {
+      "epoch": 0.6757414348655224,
+      "grad_norm": 4.051067943325395,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 441251032,
+      "step": 2569
+    },
+    {
+      "epoch": 0.6757414348655224,
+      "loss": 0.09424732625484467,
+      "loss_ce": 0.00022266953601501882,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 441251032,
+      "step": 2569
+    },
+    {
+      "epoch": 0.6760044716249096,
+      "grad_norm": 6.209787964250504,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 441422784,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6760044716249096,
+      "loss": 0.0798967033624649,
+      "loss_ce": 0.00027634453726932406,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 441422784,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6762675083842967,
+      "grad_norm": 7.3672437629155505,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 441593108,
+      "step": 2571
+    },
+    {
+      "epoch": 0.6762675083842967,
+      "loss": 0.12310583889484406,
+      "loss_ce": 0.0010050098644569516,
+      "loss_iou": NaN,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 441593108,
+      "step": 2571
+    },
+    {
+      "epoch": 0.6765305451436838,
+      "grad_norm": 5.196433267413594,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 441765456,
+      "step": 2572
+    },
+    {
+      "epoch": 0.6765305451436838,
+      "loss": 0.1754513680934906,
+      "loss_ce": 0.0007077160989865661,
+      "loss_iou": NaN,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 441765456,
+      "step": 2572
+    },
+    {
+      "epoch": 0.6767935819030709,
+      "grad_norm": 4.136115815472187,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 441937580,
+      "step": 2573
+    },
+    {
+      "epoch": 0.6767935819030709,
+      "loss": 0.13903826475143433,
+      "loss_ce": 0.0002443119592498988,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 441937580,
+      "step": 2573
+    },
+    {
+      "epoch": 0.677056618662458,
+      "grad_norm": 18.488728586585154,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 442108052,
+      "step": 2574
+    },
+    {
+      "epoch": 0.677056618662458,
+      "loss": 0.12112629413604736,
+      "loss_ce": 0.0002309026604052633,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 442108052,
+      "step": 2574
+    },
+    {
+      "epoch": 0.6773196554218452,
+      "grad_norm": 7.779321814196593,
+      "learning_rate": 5e-06,
+      "loss": 0.1347,
+      "num_input_tokens_seen": 442279576,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6773196554218452,
+      "loss": 0.10174375027418137,
+      "loss_ce": 0.0021038581617176533,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 442279576,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6775826921812323,
+      "grad_norm": 4.027957209739208,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 442447800,
+      "step": 2576
+    },
+    {
+      "epoch": 0.6775826921812323,
+      "loss": 0.08731138706207275,
+      "loss_ce": 0.0017401032382622361,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 442447800,
+      "step": 2576
+    },
+    {
+      "epoch": 0.6778457289406195,
+      "grad_norm": 3.5976030894091644,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 442620136,
+      "step": 2577
+    },
+    {
+      "epoch": 0.6778457289406195,
+      "loss": 0.11548551917076111,
+      "loss_ce": 0.0009377849055454135,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 442620136,
+      "step": 2577
+    },
+    {
+      "epoch": 0.6781087657000066,
+      "grad_norm": 14.27185242111941,
+      "learning_rate": 5e-06,
+      "loss": 0.1706,
+      "num_input_tokens_seen": 442792328,
+      "step": 2578
+    },
+    {
+      "epoch": 0.6781087657000066,
+      "loss": 0.20107831060886383,
+      "loss_ce": 0.01119793951511383,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 442792328,
+      "step": 2578
+    },
+    {
+      "epoch": 0.6783718024593937,
+      "grad_norm": 2.6862819825925905,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 442964732,
+      "step": 2579
+    },
+    {
+      "epoch": 0.6783718024593937,
+      "loss": 0.12557528913021088,
+      "loss_ce": 0.001063565374352038,
+      "loss_iou": 0.328125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 442964732,
+      "step": 2579
+    },
+    {
+      "epoch": 0.6786348392187809,
+      "grad_norm": 19.12719362396153,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 443136988,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6786348392187809,
+      "loss": 0.1754310131072998,
+      "loss_ce": 0.001114598591811955,
+      "loss_iou": 0.578125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 443136988,
+      "step": 2580
+    },
+    {
+      "epoch": 0.678897875978168,
+      "grad_norm": 4.296169271228249,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 443309004,
+      "step": 2581
+    },
+    {
+      "epoch": 0.678897875978168,
+      "loss": 0.056013718247413635,
+      "loss_ce": 0.0018907939083874226,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 443309004,
+      "step": 2581
+    },
+    {
+      "epoch": 0.6791609127375551,
+      "grad_norm": 14.58212290872784,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 443481132,
+      "step": 2582
+    },
+    {
+      "epoch": 0.6791609127375551,
+      "loss": 0.10740009695291519,
+      "loss_ce": 0.00046649359865114093,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 443481132,
+      "step": 2582
+    },
+    {
+      "epoch": 0.6794239494969422,
+      "grad_norm": 5.824835527761782,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 443651636,
+      "step": 2583
+    },
+    {
+      "epoch": 0.6794239494969422,
+      "loss": 0.13724330067634583,
+      "loss_ce": 0.0017452588072046638,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 443651636,
+      "step": 2583
+    },
+    {
+      "epoch": 0.6796869862563293,
+      "grad_norm": 6.27799845618246,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 443823548,
+      "step": 2584
+    },
+    {
+      "epoch": 0.6796869862563293,
+      "loss": 0.1675836145877838,
+      "loss_ce": 0.002330929273739457,
+      "loss_iou": 0.390625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 443823548,
+      "step": 2584
+    },
+    {
+      "epoch": 0.6799500230157165,
+      "grad_norm": 5.283309975586119,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 443994196,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6799500230157165,
+      "loss": 0.07297757267951965,
+      "loss_ce": 7.10797612555325e-05,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 443994196,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6802130597751036,
+      "grad_norm": 4.83126163860268,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 444166260,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6802130597751036,
+      "loss": 0.15566733479499817,
+      "loss_ce": 0.0010347592178732157,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 444166260,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6804760965344907,
+      "grad_norm": 5.935537777211587,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 444338748,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6804760965344907,
+      "loss": 0.07495879381895065,
+      "loss_ce": 0.0001907299447339028,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 444338748,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6807391332938778,
+      "grad_norm": 4.981341494504371,
+      "learning_rate": 5e-06,
+      "loss": 0.1501,
+      "num_input_tokens_seen": 444510996,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6807391332938778,
+      "loss": 0.2452811598777771,
+      "loss_ce": 0.002849509473890066,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 444510996,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6810021700532649,
+      "grad_norm": 34.8294057916369,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 444682880,
+      "step": 2589
+    },
+    {
+      "epoch": 0.6810021700532649,
+      "loss": 0.20022635161876678,
+      "loss_ce": 0.002899688435718417,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 444682880,
+      "step": 2589
+    },
+    {
+      "epoch": 0.681265206812652,
+      "grad_norm": 6.783440844262528,
+      "learning_rate": 5e-06,
+      "loss": 0.1355,
+      "num_input_tokens_seen": 444855080,
+      "step": 2590
+    },
+    {
+      "epoch": 0.681265206812652,
+      "loss": 0.17689135670661926,
+      "loss_ce": 0.0014458001824095845,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 444855080,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6815282435720392,
+      "grad_norm": 6.398055261239124,
+      "learning_rate": 5e-06,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 445027648,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6815282435720392,
+      "loss": 0.06686560064554214,
+      "loss_ce": 0.003968872129917145,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 445027648,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6817912803314263,
+      "grad_norm": 9.754653714308724,
+      "learning_rate": 5e-06,
+      "loss": 0.158,
+      "num_input_tokens_seen": 445198060,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6817912803314263,
+      "loss": 0.0956597551703453,
+      "loss_ce": 0.001345182885415852,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 445198060,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6820543170908134,
+      "grad_norm": 11.093960051238781,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 445370376,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6820543170908134,
+      "loss": 0.14656749367713928,
+      "loss_ce": 0.001456425990909338,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 445370376,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6823173538502005,
+      "grad_norm": 7.226137026161222,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 445542828,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6823173538502005,
+      "loss": 0.06742848455905914,
+      "loss_ce": 0.005309954285621643,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 445542828,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6825803906095876,
+      "grad_norm": 4.617442760822597,
+      "learning_rate": 5e-06,
+      "loss": 0.137,
+      "num_input_tokens_seen": 445715128,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6825803906095876,
+      "loss": 0.11120542138814926,
+      "loss_ce": 0.0006707610446028411,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 445715128,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6828434273689749,
+      "grad_norm": 8.070925307261565,
+      "learning_rate": 5e-06,
+      "loss": 0.1584,
+      "num_input_tokens_seen": 445887032,
+      "step": 2596
+    },
+    {
+      "epoch": 0.6828434273689749,
+      "loss": 0.12233078479766846,
+      "loss_ce": 0.0008708295645192266,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 445887032,
+      "step": 2596
+    },
+    {
+      "epoch": 0.683106464128362,
+      "grad_norm": 4.079666869126499,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 446059408,
+      "step": 2597
+    },
+    {
+      "epoch": 0.683106464128362,
+      "loss": 0.1118651032447815,
+      "loss_ce": 0.00023180160496849567,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 446059408,
+      "step": 2597
+    },
+    {
+      "epoch": 0.6833695008877491,
+      "grad_norm": 6.132037428307681,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 446229696,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6833695008877491,
+      "loss": 0.13406622409820557,
+      "loss_ce": 0.0008264797506853938,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 446229696,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6836325376471362,
+      "grad_norm": 3.96696464215725,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 446401824,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6836325376471362,
+      "loss": 0.07421835511922836,
+      "loss_ce": 0.0017085927538573742,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 446401824,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6838955744065233,
+      "grad_norm": 5.0831174679987035,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 446573796,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6838955744065233,
+      "loss": 0.10569039732217789,
+      "loss_ce": 0.0010151021415367723,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 446573796,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6841586111659105,
+      "grad_norm": 15.882119320832834,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 446746000,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6841586111659105,
+      "loss": 0.1552843451499939,
+      "loss_ce": 0.000499195302836597,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 446746000,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6844216479252976,
+      "grad_norm": 3.8491365485511118,
+      "learning_rate": 5e-06,
+      "loss": 0.0928,
+      "num_input_tokens_seen": 446918100,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6844216479252976,
+      "loss": 0.050599753856658936,
+      "loss_ce": 0.0002762702642939985,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 446918100,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6846846846846847,
+      "grad_norm": 4.058506484572602,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 447088492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6846846846846847,
+      "loss": 0.043845463544130325,
+      "loss_ce": 0.0001748104114085436,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 447088492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6849477214440718,
+      "grad_norm": 4.841823848779041,
+      "learning_rate": 5e-06,
+      "loss": 0.079,
+      "num_input_tokens_seen": 447260652,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6849477214440718,
+      "loss": 0.08405909687280655,
+      "loss_ce": 0.004438734147697687,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 447260652,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6852107582034589,
+      "grad_norm": 9.044057313780108,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 447432936,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6852107582034589,
+      "loss": 0.07848000526428223,
+      "loss_ce": 0.0012094933772459626,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 447432936,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6854737949628461,
+      "grad_norm": 7.595343434642406,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 447605384,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6854737949628461,
+      "loss": 0.12424921244382858,
+      "loss_ce": 0.0026061516255140305,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 447605384,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6857368317222332,
+      "grad_norm": 5.0763063086969415,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 447777576,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6857368317222332,
+      "loss": 0.11916627734899521,
+      "loss_ce": 0.002253434620797634,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 447777576,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6859998684816203,
+      "grad_norm": 11.527878652419272,
+      "learning_rate": 5e-06,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 447949852,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6859998684816203,
+      "loss": 0.0797332376241684,
+      "loss_ce": 0.0030425682198256254,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 447949852,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6862629052410074,
+      "grad_norm": 6.796020734872217,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 448122008,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6862629052410074,
+      "loss": 0.09912531077861786,
+      "loss_ce": 0.0006145666702650487,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 448122008,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6865259420003945,
+      "grad_norm": 53.2300135535103,
+      "learning_rate": 5e-06,
+      "loss": 0.1731,
+      "num_input_tokens_seen": 448294084,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6865259420003945,
+      "loss": 0.14799347519874573,
+      "loss_ce": 0.0005325321108102798,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 448294084,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6867889787597817,
+      "grad_norm": 8.384011107333327,
+      "learning_rate": 5e-06,
+      "loss": 0.1662,
+      "num_input_tokens_seen": 448466064,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6867889787597817,
+      "loss": 0.16361187398433685,
+      "loss_ce": 0.0061564212664961815,
+      "loss_iou": 0.40625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 448466064,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6870520155191688,
+      "grad_norm": 58.16654404486195,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 448638396,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6870520155191688,
+      "loss": 0.11319980025291443,
+      "loss_ce": 0.008921236731112003,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 448638396,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6873150522785559,
+      "grad_norm": 4.791126855568047,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 448808820,
+      "step": 2613
+    },
+    {
+      "epoch": 0.6873150522785559,
+      "loss": 0.08992376923561096,
+      "loss_ce": 0.0010565832490101457,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 448808820,
+      "step": 2613
+    },
+    {
+      "epoch": 0.687578089037943,
+      "grad_norm": 4.623158042374579,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 448980968,
+      "step": 2614
+    },
+    {
+      "epoch": 0.687578089037943,
+      "loss": 0.10407891124486923,
+      "loss_ce": 0.0005632878746837378,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 448980968,
+      "step": 2614
+    },
+    {
+      "epoch": 0.6878411257973301,
+      "grad_norm": 16.128815011570154,
+      "learning_rate": 5e-06,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 449153104,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6878411257973301,
+      "loss": 0.05162462592124939,
+      "loss_ce": 0.0013011416886001825,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 449153104,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6881041625567172,
+      "grad_norm": 12.10527552944659,
+      "learning_rate": 5e-06,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 449325208,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6881041625567172,
+      "loss": 0.056678079068660736,
+      "loss_ce": 0.0005867721047252417,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 449325208,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6883671993161045,
+      "grad_norm": 4.0544613329310675,
+      "learning_rate": 5e-06,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 449497440,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6883671993161045,
+      "loss": 0.12780970335006714,
+      "loss_ce": 0.0006887409836053848,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 449497440,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6886302360754916,
+      "grad_norm": 6.140781979511808,
+      "learning_rate": 5e-06,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 449669860,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6886302360754916,
+      "loss": 0.12178224325180054,
+      "loss_ce": 0.0038012792356312275,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 449669860,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6888932728348787,
+      "grad_norm": 4.582153125692399,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 449842236,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6888932728348787,
+      "loss": 0.11555735766887665,
+      "loss_ce": 0.003313705325126648,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 449842236,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6891563095942658,
+      "grad_norm": 10.075980623616992,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 450014444,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6891563095942658,
+      "loss": 0.20255392789840698,
+      "loss_ce": 0.004891556687653065,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 450014444,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6894193463536529,
+      "grad_norm": 16.09024773291766,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 450186632,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6894193463536529,
+      "loss": 0.15711162984371185,
+      "loss_ce": 0.001548278727568686,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 450186632,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6896823831130401,
+      "grad_norm": 4.077753866376941,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 450358652,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6896823831130401,
+      "loss": 0.17466840147972107,
+      "loss_ce": 0.0030528109055012465,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 450358652,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6899454198724272,
+      "grad_norm": 5.782402643209976,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 450530992,
+      "step": 2623
+    },
+    {
+      "epoch": 0.6899454198724272,
+      "loss": 0.09183860570192337,
+      "loss_ce": 0.000987776555120945,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 450530992,
+      "step": 2623
+    },
+    {
+      "epoch": 0.6902084566318143,
+      "grad_norm": 4.962596372966127,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 450702972,
+      "step": 2624
+    },
+    {
+      "epoch": 0.6902084566318143,
+      "loss": 0.16587726771831512,
+      "loss_ce": 0.004210404586046934,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 450702972,
+      "step": 2624
+    },
+    {
+      "epoch": 0.6904714933912014,
+      "grad_norm": 4.41836936316546,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 450871644,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6904714933912014,
+      "loss": 0.08390143513679504,
+      "loss_ce": 0.00022223126143217087,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 450871644,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6907345301505885,
+      "grad_norm": 6.24899387116621,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 451043764,
+      "step": 2626
+    },
+    {
+      "epoch": 0.6907345301505885,
+      "loss": 0.07276784628629684,
+      "loss_ce": 0.0041185528971254826,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 451043764,
+      "step": 2626
+    },
+    {
+      "epoch": 0.6909975669099757,
+      "grad_norm": 3.855416348391677,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 451215860,
+      "step": 2627
+    },
+    {
+      "epoch": 0.6909975669099757,
+      "loss": 0.0781233012676239,
+      "loss_ce": 0.00033398933010175824,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 451215860,
+      "step": 2627
+    },
+    {
+      "epoch": 0.6912606036693628,
+      "grad_norm": 9.945431727329952,
+      "learning_rate": 5e-06,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 451387900,
+      "step": 2628
+    },
+    {
+      "epoch": 0.6912606036693628,
+      "loss": 0.11348669975996017,
+      "loss_ce": 0.0029215135145932436,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 451387900,
+      "step": 2628
+    },
+    {
+      "epoch": 0.6915236404287499,
+      "grad_norm": 4.281056455844802,
+      "learning_rate": 5e-06,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 451558132,
+      "step": 2629
+    },
+    {
+      "epoch": 0.6915236404287499,
+      "loss": 0.11048734933137894,
+      "loss_ce": 0.00626982469111681,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 451558132,
+      "step": 2629
+    },
+    {
+      "epoch": 0.691786677188137,
+      "grad_norm": 4.21783711884085,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 451730300,
+      "step": 2630
+    },
+    {
+      "epoch": 0.691786677188137,
+      "loss": 0.06679116189479828,
+      "loss_ce": 0.0006290507735684514,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 451730300,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6920497139475241,
+      "grad_norm": 3.6368581797967257,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 451902464,
+      "step": 2631
+    },
+    {
+      "epoch": 0.6920497139475241,
+      "loss": 0.07377897202968597,
+      "loss_ce": 0.002199993235990405,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 451902464,
+      "step": 2631
+    },
+    {
+      "epoch": 0.6923127507069113,
+      "grad_norm": 4.812739476349578,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 452074432,
+      "step": 2632
+    },
+    {
+      "epoch": 0.6923127507069113,
+      "loss": 0.08376286178827286,
+      "loss_ce": 0.0004193550848867744,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 452074432,
+      "step": 2632
+    },
+    {
+      "epoch": 0.6925757874662984,
+      "grad_norm": 17.745450709557872,
+      "learning_rate": 5e-06,
+      "loss": 0.1677,
+      "num_input_tokens_seen": 452246508,
+      "step": 2633
+    },
+    {
+      "epoch": 0.6925757874662984,
+      "loss": 0.12932631373405457,
+      "loss_ce": 0.00045057572424411774,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 452246508,
+      "step": 2633
+    },
+    {
+      "epoch": 0.6928388242256855,
+      "grad_norm": 3.8241822612681404,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 452418684,
+      "step": 2634
+    },
+    {
+      "epoch": 0.6928388242256855,
+      "loss": 0.07180548459291458,
+      "loss_ce": 0.000638492638245225,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 452418684,
+      "step": 2634
+    },
+    {
+      "epoch": 0.6931018609850726,
+      "grad_norm": 5.017758815401084,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 452591112,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6931018609850726,
+      "loss": 0.14692719280719757,
+      "loss_ce": 0.001221017329953611,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 452591112,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6933648977444598,
+      "grad_norm": 4.688279396436893,
+      "learning_rate": 5e-06,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 452763100,
+      "step": 2636
+    },
+    {
+      "epoch": 0.6933648977444598,
+      "loss": 0.11540480703115463,
+      "loss_ce": 0.0031611607410013676,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 452763100,
+      "step": 2636
+    },
+    {
+      "epoch": 0.693627934503847,
+      "grad_norm": 5.800484869059427,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 452932760,
+      "step": 2637
+    },
+    {
+      "epoch": 0.693627934503847,
+      "loss": 0.20021981000900269,
+      "loss_ce": 0.002526947297155857,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 452932760,
+      "step": 2637
+    },
+    {
+      "epoch": 0.6938909712632341,
+      "grad_norm": 6.943406352493944,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 453104824,
+      "step": 2638
+    },
+    {
+      "epoch": 0.6938909712632341,
+      "loss": 0.11978557705879211,
+      "loss_ce": 0.003910328261554241,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 453104824,
+      "step": 2638
+    },
+    {
+      "epoch": 0.6941540080226212,
+      "grad_norm": 8.392991417214587,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 453276756,
+      "step": 2639
+    },
+    {
+      "epoch": 0.6941540080226212,
+      "loss": 0.14925961196422577,
+      "loss_ce": 0.00796323362737894,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 453276756,
+      "step": 2639
+    },
+    {
+      "epoch": 0.6944170447820083,
+      "grad_norm": 7.152620797713759,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 453447368,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6944170447820083,
+      "loss": 0.12115476280450821,
+      "loss_ce": 0.002624482847750187,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 453447368,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6946800815413954,
+      "grad_norm": 9.172566148062016,
+      "learning_rate": 5e-06,
+      "loss": 0.1376,
+      "num_input_tokens_seen": 453619160,
+      "step": 2641
+    },
+    {
+      "epoch": 0.6946800815413954,
+      "loss": 0.09918803721666336,
+      "loss_ce": 0.0022947255056351423,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 453619160,
+      "step": 2641
+    },
+    {
+      "epoch": 0.6949431183007825,
+      "grad_norm": 4.761554471447507,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 453791024,
+      "step": 2642
+    },
+    {
+      "epoch": 0.6949431183007825,
+      "loss": 0.12317492812871933,
+      "loss_ce": 0.004400510806590319,
+      "loss_iou": 0.5,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 453791024,
+      "step": 2642
+    },
+    {
+      "epoch": 0.6952061550601697,
+      "grad_norm": 6.764608063652922,
+      "learning_rate": 5e-06,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 453963268,
+      "step": 2643
+    },
+    {
+      "epoch": 0.6952061550601697,
+      "loss": 0.17664819955825806,
+      "loss_ce": 0.0005617668502964079,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 453963268,
+      "step": 2643
+    },
+    {
+      "epoch": 0.6954691918195568,
+      "grad_norm": 8.472259366872775,
+      "learning_rate": 5e-06,
+      "loss": 0.1241,
+      "num_input_tokens_seen": 454133580,
+      "step": 2644
+    },
+    {
+      "epoch": 0.6954691918195568,
+      "loss": 0.05588904023170471,
+      "loss_ce": 0.0003470498777460307,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 454133580,
+      "step": 2644
+    },
+    {
+      "epoch": 0.6957322285789439,
+      "grad_norm": 9.45776397859236,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 454305316,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6957322285789439,
+      "loss": 0.11683906614780426,
+      "loss_ce": 0.002215046202763915,
+      "loss_iou": NaN,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 454305316,
+      "step": 2645
+    },
+    {
+      "epoch": 0.695995265338331,
+      "grad_norm": 12.21950005723,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 454477460,
+      "step": 2646
+    },
+    {
+      "epoch": 0.695995265338331,
+      "loss": 0.19950228929519653,
+      "loss_ce": 0.00556308263912797,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 454477460,
+      "step": 2646
+    },
+    {
+      "epoch": 0.6962583020977181,
+      "grad_norm": 4.539781489505394,
+      "learning_rate": 5e-06,
+      "loss": 0.1468,
+      "num_input_tokens_seen": 454649576,
+      "step": 2647
+    },
+    {
+      "epoch": 0.6962583020977181,
+      "loss": 0.053970806300640106,
+      "loss_ce": 0.0009465104667469859,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 454649576,
+      "step": 2647
+    },
+    {
+      "epoch": 0.6965213388571053,
+      "grad_norm": 14.32877869742435,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 454821436,
+      "step": 2648
+    },
+    {
+      "epoch": 0.6965213388571053,
+      "loss": 0.10093516111373901,
+      "loss_ce": 0.00645274156704545,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 454821436,
+      "step": 2648
+    },
+    {
+      "epoch": 0.6967843756164924,
+      "grad_norm": 7.804241962530211,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 454993508,
+      "step": 2649
+    },
+    {
+      "epoch": 0.6967843756164924,
+      "loss": 0.1862741857767105,
+      "loss_ce": 0.0016580985393375158,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 454993508,
+      "step": 2649
+    },
+    {
+      "epoch": 0.6970474123758795,
+      "grad_norm": 16.275796095386262,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 455165748,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6970474123758795,
+      "loss": 0.07620816677808762,
+      "loss_ce": 0.0016232050256803632,
+      "loss_iou": 0.5625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 455165748,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6973104491352666,
+      "grad_norm": 4.357121015686003,
+      "learning_rate": 5e-06,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 455338084,
+      "step": 2651
+    },
+    {
+      "epoch": 0.6973104491352666,
+      "loss": 0.13134872913360596,
+      "loss_ce": 0.0034037926234304905,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 455338084,
+      "step": 2651
+    },
+    {
+      "epoch": 0.6975734858946537,
+      "grad_norm": 5.3899957417482725,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 455509908,
+      "step": 2652
+    },
+    {
+      "epoch": 0.6975734858946537,
+      "loss": 0.14220395684242249,
+      "loss_ce": 0.0009075828129425645,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 455509908,
+      "step": 2652
+    },
+    {
+      "epoch": 0.697836522654041,
+      "grad_norm": 6.823610210106841,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 455681908,
+      "step": 2653
+    },
+    {
+      "epoch": 0.697836522654041,
+      "loss": 0.05261433497071266,
+      "loss_ce": 0.0018330833408981562,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 455681908,
+      "step": 2653
+    },
+    {
+      "epoch": 0.698099559413428,
+      "grad_norm": 6.597254450850545,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 455853952,
+      "step": 2654
+    },
+    {
+      "epoch": 0.698099559413428,
+      "loss": 0.147089421749115,
+      "loss_ce": 0.00023882777895778418,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 455853952,
+      "step": 2654
+    },
+    {
+      "epoch": 0.6983625961728152,
+      "grad_norm": 6.378423815524031,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 456026208,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6983625961728152,
+      "loss": 0.18122422695159912,
+      "loss_ce": 0.0033067562617361546,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 456026208,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6986256329322023,
+      "grad_norm": 25.554315055519154,
+      "learning_rate": 5e-06,
+      "loss": 0.1181,
+      "num_input_tokens_seen": 456198288,
+      "step": 2656
+    },
+    {
+      "epoch": 0.6986256329322023,
+      "loss": 0.10486012697219849,
+      "loss_ce": 0.0011003611143678427,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 456198288,
+      "step": 2656
+    },
+    {
+      "epoch": 0.6988886696915894,
+      "grad_norm": 4.576359717222489,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 456365496,
+      "step": 2657
+    },
+    {
+      "epoch": 0.6988886696915894,
+      "loss": 0.13455136120319366,
+      "loss_ce": 0.0007317845011129975,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 456365496,
+      "step": 2657
+    },
+    {
+      "epoch": 0.6991517064509766,
+      "grad_norm": 5.013450140740463,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 456537676,
+      "step": 2658
+    },
+    {
+      "epoch": 0.6991517064509766,
+      "loss": 0.08290005475282669,
+      "loss_ce": 0.003767975838854909,
+      "loss_iou": 0.5,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 456537676,
+      "step": 2658
+    },
+    {
+      "epoch": 0.6994147432103637,
+      "grad_norm": 4.084720539079834,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 456710364,
+      "step": 2659
+    },
+    {
+      "epoch": 0.6994147432103637,
+      "loss": 0.06024003401398659,
+      "loss_ce": 0.0006849807105027139,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 456710364,
+      "step": 2659
+    },
+    {
+      "epoch": 0.6996777799697508,
+      "grad_norm": 3.7990429726373636,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 456880760,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6996777799697508,
+      "loss": 0.09679127484560013,
+      "loss_ce": 0.0005083205760456622,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 456880760,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6999408167291379,
+      "grad_norm": 5.055246469854178,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 457051432,
+      "step": 2661
+    },
+    {
+      "epoch": 0.6999408167291379,
+      "loss": 0.21587547659873962,
+      "loss_ce": 0.0004213774227537215,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 457051432,
+      "step": 2661
+    },
+    {
+      "epoch": 0.700203853488525,
+      "grad_norm": 3.8688477802914774,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 457223592,
+      "step": 2662
+    },
+    {
+      "epoch": 0.700203853488525,
+      "loss": 0.14515820145606995,
+      "loss_ce": 0.0040449099615216255,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 457223592,
+      "step": 2662
+    },
+    {
+      "epoch": 0.7004668902479122,
+      "grad_norm": 10.967803606252325,
+      "learning_rate": 5e-06,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 457396048,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7004668902479122,
+      "loss": 0.13383157551288605,
+      "loss_ce": 0.0020261560566723347,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 457396048,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7007299270072993,
+      "grad_norm": 3.339673678629444,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 457566272,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7007299270072993,
+      "loss": 0.20303812623023987,
+      "loss_ce": 0.005711473990231752,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 457566272,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7009929637666864,
+      "grad_norm": 10.341407204229276,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 457738384,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7009929637666864,
+      "loss": 0.14979971945285797,
+      "loss_ce": 0.0018810234032571316,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 457738384,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7012560005260735,
+      "grad_norm": 3.460962528007649,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 457910472,
+      "step": 2666
+    },
+    {
+      "epoch": 0.7012560005260735,
+      "loss": 0.12169472873210907,
+      "loss_ce": 0.004110502544790506,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 457910472,
+      "step": 2666
+    },
+    {
+      "epoch": 0.7015190372854606,
+      "grad_norm": 5.718591075026739,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 458082324,
+      "step": 2667
+    },
+    {
+      "epoch": 0.7015190372854606,
+      "loss": 0.1776396483182907,
+      "loss_ce": 0.0005613988032564521,
+      "loss_iou": 0.359375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 458082324,
+      "step": 2667
+    },
+    {
+      "epoch": 0.7017820740448477,
+      "grad_norm": 5.655030135877534,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 458254304,
+      "step": 2668
+    },
+    {
+      "epoch": 0.7017820740448477,
+      "loss": 0.12232168763875961,
+      "loss_ce": 0.0004344757762737572,
+      "loss_iou": 0.375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 458254304,
+      "step": 2668
+    },
+    {
+      "epoch": 0.7020451108042349,
+      "grad_norm": 5.899988129903735,
+      "learning_rate": 5e-06,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 458426532,
+      "step": 2669
+    },
+    {
+      "epoch": 0.7020451108042349,
+      "loss": 0.09821672737598419,
+      "loss_ce": 0.0016896221786737442,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 458426532,
+      "step": 2669
+    },
+    {
+      "epoch": 0.702308147563622,
+      "grad_norm": 5.5784599020012715,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 458596896,
+      "step": 2670
+    },
+    {
+      "epoch": 0.702308147563622,
+      "loss": 0.16502083837985992,
+      "loss_ce": 0.0003174682497046888,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 458596896,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7025711843230091,
+      "grad_norm": 14.038116606042868,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 458769240,
+      "step": 2671
+    },
+    {
+      "epoch": 0.7025711843230091,
+      "loss": 0.15062229335308075,
+      "loss_ce": 0.0042905076406896114,
+      "loss_iou": 0.53125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 458769240,
+      "step": 2671
+    },
+    {
+      "epoch": 0.7028342210823962,
+      "grad_norm": 4.120047179172289,
+      "learning_rate": 5e-06,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 458941376,
+      "step": 2672
+    },
+    {
+      "epoch": 0.7028342210823962,
+      "loss": 0.08248385787010193,
+      "loss_ce": 0.0018869286868721247,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 458941376,
+      "step": 2672
+    },
+    {
+      "epoch": 0.7030972578417833,
+      "grad_norm": 4.682779609797298,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 459113732,
+      "step": 2673
+    },
+    {
+      "epoch": 0.7030972578417833,
+      "loss": 0.08461406081914902,
+      "loss_ce": 0.0013010749826207757,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 459113732,
+      "step": 2673
+    },
+    {
+      "epoch": 0.7033602946011706,
+      "grad_norm": 6.053537160791169,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 459283668,
+      "step": 2674
+    },
+    {
+      "epoch": 0.7033602946011706,
+      "loss": 0.21815051138401031,
+      "loss_ce": 0.0011094921501353383,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 459283668,
+      "step": 2674
+    },
+    {
+      "epoch": 0.7036233313605577,
+      "grad_norm": 3.9631091480761715,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 459455956,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7036233313605577,
+      "loss": 0.13024334609508514,
+      "loss_ce": 0.00025372387608513236,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 459455956,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7038863681199448,
+      "grad_norm": 5.129114238432721,
+      "learning_rate": 5e-06,
+      "loss": 0.137,
+      "num_input_tokens_seen": 459627968,
+      "step": 2676
+    },
+    {
+      "epoch": 0.7038863681199448,
+      "loss": 0.08658237755298615,
+      "loss_ce": 0.002750593703240156,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 459627968,
+      "step": 2676
+    },
+    {
+      "epoch": 0.7041494048793319,
+      "grad_norm": 20.187762197991425,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 459800260,
+      "step": 2677
+    },
+    {
+      "epoch": 0.7041494048793319,
+      "loss": 0.08818955719470978,
+      "loss_ce": 0.0012297153007239103,
+      "loss_iou": 0.53125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 459800260,
+      "step": 2677
+    },
+    {
+      "epoch": 0.704412441638719,
+      "grad_norm": 7.902489579853604,
+      "learning_rate": 5e-06,
+      "loss": 0.1564,
+      "num_input_tokens_seen": 459971088,
+      "step": 2678
+    },
+    {
+      "epoch": 0.704412441638719,
+      "loss": 0.12185804545879364,
+      "loss_ce": 0.00024549951194785535,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 459971088,
+      "step": 2678
+    },
+    {
+      "epoch": 0.7046754783981062,
+      "grad_norm": 27.40955645070263,
+      "learning_rate": 5e-06,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 460143240,
+      "step": 2679
+    },
+    {
+      "epoch": 0.7046754783981062,
+      "loss": 0.06965695321559906,
+      "loss_ce": 0.0006261939415708184,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 460143240,
+      "step": 2679
+    },
+    {
+      "epoch": 0.7049385151574933,
+      "grad_norm": 8.476577616174554,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 460315544,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7049385151574933,
+      "loss": 0.14567114412784576,
+      "loss_ce": 0.0017502475529909134,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 460315544,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7052015519168804,
+      "grad_norm": 11.423647115492507,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 460487612,
+      "step": 2681
+    },
+    {
+      "epoch": 0.7052015519168804,
+      "loss": 0.09613794833421707,
+      "loss_ce": 0.002235355554148555,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 460487612,
+      "step": 2681
+    },
+    {
+      "epoch": 0.7054645886762675,
+      "grad_norm": 13.90990204859974,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 460655964,
+      "step": 2682
+    },
+    {
+      "epoch": 0.7054645886762675,
+      "loss": 0.06838610768318176,
+      "loss_ce": 0.00024035980459302664,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 460655964,
+      "step": 2682
+    },
+    {
+      "epoch": 0.7057276254356546,
+      "grad_norm": 14.957706126877364,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 460827816,
+      "step": 2683
+    },
+    {
+      "epoch": 0.7057276254356546,
+      "loss": 0.1538136899471283,
+      "loss_ce": 0.0006154490984044969,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 460827816,
+      "step": 2683
+    },
+    {
+      "epoch": 0.7059906621950418,
+      "grad_norm": 6.337066775015052,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 461000128,
+      "step": 2684
+    },
+    {
+      "epoch": 0.7059906621950418,
+      "loss": 0.18371257185935974,
+      "loss_ce": 0.0005155415856279433,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 461000128,
+      "step": 2684
+    },
+    {
+      "epoch": 0.7062536989544289,
+      "grad_norm": 4.77556589512149,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 461172480,
+      "step": 2685
+    },
+    {
+      "epoch": 0.7062536989544289,
+      "loss": 0.06664656102657318,
+      "loss_ce": 0.001003247918561101,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 461172480,
+      "step": 2685
+    },
+    {
+      "epoch": 0.706516735713816,
+      "grad_norm": 4.934194738374163,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 461344704,
+      "step": 2686
+    },
+    {
+      "epoch": 0.706516735713816,
+      "loss": 0.07681739330291748,
+      "loss_ce": 0.002476579276844859,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 461344704,
+      "step": 2686
+    },
+    {
+      "epoch": 0.7067797724732031,
+      "grad_norm": 5.14148867298625,
+      "learning_rate": 5e-06,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 461517244,
+      "step": 2687
+    },
+    {
+      "epoch": 0.7067797724732031,
+      "loss": 0.08841484785079956,
+      "loss_ce": 9.696922643342987e-05,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 461517244,
+      "step": 2687
+    },
+    {
+      "epoch": 0.7070428092325902,
+      "grad_norm": 4.290637783097341,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 461689392,
+      "step": 2688
+    },
+    {
+      "epoch": 0.7070428092325902,
+      "loss": 0.12317334860563278,
+      "loss_ce": 0.0057722218334674835,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 461689392,
+      "step": 2688
+    },
+    {
+      "epoch": 0.7073058459919774,
+      "grad_norm": 18.724820792441665,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 461861916,
+      "step": 2689
+    },
+    {
+      "epoch": 0.7073058459919774,
+      "loss": 0.05893798917531967,
+      "loss_ce": 0.00014587045006919652,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 461861916,
+      "step": 2689
+    },
+    {
+      "epoch": 0.7075688827513645,
+      "grad_norm": 6.318558285374775,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 462033832,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7075688827513645,
+      "loss": 0.129222571849823,
+      "loss_ce": 0.00034684882848523557,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 462033832,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7078319195107516,
+      "grad_norm": 5.104970581967641,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 462206008,
+      "step": 2691
+    },
+    {
+      "epoch": 0.7078319195107516,
+      "loss": 0.06862036883831024,
+      "loss_ce": 0.00010840100003406405,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 462206008,
+      "step": 2691
+    },
+    {
+      "epoch": 0.7080949562701387,
+      "grad_norm": 7.0445063257027885,
+      "learning_rate": 5e-06,
+      "loss": 0.1429,
+      "num_input_tokens_seen": 462377836,
+      "step": 2692
+    },
+    {
+      "epoch": 0.7080949562701387,
+      "loss": 0.16992726922035217,
+      "loss_ce": 0.014318134635686874,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 462377836,
+      "step": 2692
+    },
+    {
+      "epoch": 0.7083579930295258,
+      "grad_norm": 3.568328460902499,
+      "learning_rate": 5e-06,
+      "loss": 0.0725,
+      "num_input_tokens_seen": 462550184,
+      "step": 2693
+    },
+    {
+      "epoch": 0.7083579930295258,
+      "loss": 0.058605365455150604,
+      "loss_ce": 0.0025903512723743916,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 462550184,
+      "step": 2693
+    },
+    {
+      "epoch": 0.708621029788913,
+      "grad_norm": 16.451151717262515,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 462722228,
+      "step": 2694
+    },
+    {
+      "epoch": 0.708621029788913,
+      "loss": 0.15793108940124512,
+      "loss_ce": 0.0002467722224537283,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 462722228,
+      "step": 2694
+    },
+    {
+      "epoch": 0.7088840665483002,
+      "grad_norm": 18.144741325116385,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 462892644,
+      "step": 2695
+    },
+    {
+      "epoch": 0.7088840665483002,
+      "loss": 0.18394407629966736,
+      "loss_ce": 0.005751936696469784,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 462892644,
+      "step": 2695
+    },
+    {
+      "epoch": 0.7091471033076873,
+      "grad_norm": 4.177749037836763,
+      "learning_rate": 5e-06,
+      "loss": 0.0915,
+      "num_input_tokens_seen": 463059460,
+      "step": 2696
+    },
+    {
+      "epoch": 0.7091471033076873,
+      "loss": 0.08263557404279709,
+      "loss_ce": 0.0037171156145632267,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 463059460,
+      "step": 2696
+    },
+    {
+      "epoch": 0.7094101400670744,
+      "grad_norm": 24.286466900686108,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 463231680,
+      "step": 2697
+    },
+    {
+      "epoch": 0.7094101400670744,
+      "loss": 0.1826072484254837,
+      "loss_ce": 0.005239082965999842,
+      "loss_iou": 0.40625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 463231680,
+      "step": 2697
+    },
+    {
+      "epoch": 0.7096731768264615,
+      "grad_norm": 3.90367546541105,
+      "learning_rate": 5e-06,
+      "loss": 0.1316,
+      "num_input_tokens_seen": 463403632,
+      "step": 2698
+    },
+    {
+      "epoch": 0.7096731768264615,
+      "loss": 0.19678181409835815,
+      "loss_ce": 0.0011641355231404305,
+      "loss_iou": NaN,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 463403632,
+      "step": 2698
+    },
+    {
+      "epoch": 0.7099362135858486,
+      "grad_norm": 4.6638342162135,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 463575764,
+      "step": 2699
+    },
+    {
+      "epoch": 0.7099362135858486,
+      "loss": 0.05382794886827469,
+      "loss_ce": 0.0004832194827031344,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 463575764,
+      "step": 2699
+    },
+    {
+      "epoch": 0.7101992503452358,
+      "grad_norm": 10.964970610276819,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 463747960,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7101992503452358,
+      "loss": 0.10617360472679138,
+      "loss_ce": 0.002444351091980934,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 463747960,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7104622871046229,
+      "grad_norm": 4.313980855592432,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 463920292,
+      "step": 2701
+    },
+    {
+      "epoch": 0.7104622871046229,
+      "loss": 0.06053323671221733,
+      "loss_ce": 0.0002305020607309416,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 463920292,
+      "step": 2701
+    },
+    {
+      "epoch": 0.71072532386401,
+      "grad_norm": 13.238355947380045,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 464092396,
+      "step": 2702
+    },
+    {
+      "epoch": 0.71072532386401,
+      "loss": 0.09292985498905182,
+      "loss_ce": 0.002384199295192957,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 464092396,
+      "step": 2702
+    },
+    {
+      "epoch": 0.7109883606233971,
+      "grad_norm": 3.301662541659157,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 464264724,
+      "step": 2703
+    },
+    {
+      "epoch": 0.7109883606233971,
+      "loss": 0.05558721721172333,
+      "loss_ce": 7.5738578743767e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 464264724,
+      "step": 2703
+    },
+    {
+      "epoch": 0.7112513973827842,
+      "grad_norm": 3.9814210468792854,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 464436668,
+      "step": 2704
+    },
+    {
+      "epoch": 0.7112513973827842,
+      "loss": 0.0715593621134758,
+      "loss_ce": 0.004069737158715725,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 464436668,
+      "step": 2704
+    },
+    {
+      "epoch": 0.7115144341421714,
+      "grad_norm": 24.903703628485385,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 464607064,
+      "step": 2705
+    },
+    {
+      "epoch": 0.7115144341421714,
+      "loss": 0.13525709509849548,
+      "loss_ce": 0.0011323521612212062,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 464607064,
+      "step": 2705
+    },
+    {
+      "epoch": 0.7117774709015585,
+      "grad_norm": 8.733454229975411,
+      "learning_rate": 5e-06,
+      "loss": 0.099,
+      "num_input_tokens_seen": 464779372,
+      "step": 2706
+    },
+    {
+      "epoch": 0.7117774709015585,
+      "loss": 0.1117292046546936,
+      "loss_ce": 0.002689904533326626,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 464779372,
+      "step": 2706
+    },
+    {
+      "epoch": 0.7120405076609456,
+      "grad_norm": 4.906686826576695,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 464951196,
+      "step": 2707
+    },
+    {
+      "epoch": 0.7120405076609456,
+      "loss": 0.09186941385269165,
+      "loss_ce": 0.00013357413990888745,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 464951196,
+      "step": 2707
+    },
+    {
+      "epoch": 0.7123035444203327,
+      "grad_norm": 3.5122712023777214,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 465119688,
+      "step": 2708
+    },
+    {
+      "epoch": 0.7123035444203327,
+      "loss": 0.17270034551620483,
+      "loss_ce": 0.004762125201523304,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 465119688,
+      "step": 2708
+    },
+    {
+      "epoch": 0.7125665811797198,
+      "grad_norm": 4.667301843217583,
+      "learning_rate": 5e-06,
+      "loss": 0.1511,
+      "num_input_tokens_seen": 465291884,
+      "step": 2709
+    },
+    {
+      "epoch": 0.7125665811797198,
+      "loss": 0.24567091464996338,
+      "loss_ce": 0.0022016754373908043,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 465291884,
+      "step": 2709
+    },
+    {
+      "epoch": 0.712829617939107,
+      "grad_norm": 4.677081009771049,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 465463732,
+      "step": 2710
+    },
+    {
+      "epoch": 0.712829617939107,
+      "loss": 0.10853127390146255,
+      "loss_ce": 0.0006211129948496819,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 465463732,
+      "step": 2710
+    },
+    {
+      "epoch": 0.7130926546984941,
+      "grad_norm": 9.121056040645339,
+      "learning_rate": 5e-06,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 465635964,
+      "step": 2711
+    },
+    {
+      "epoch": 0.7130926546984941,
+      "loss": 0.05089259892702103,
+      "loss_ce": 0.0005080772680230439,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 465635964,
+      "step": 2711
+    },
+    {
+      "epoch": 0.7133556914578812,
+      "grad_norm": 5.554084070604222,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 465808444,
+      "step": 2712
+    },
+    {
+      "epoch": 0.7133556914578812,
+      "loss": 0.10150805115699768,
+      "loss_ce": 0.0013188383309170604,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 465808444,
+      "step": 2712
+    },
+    {
+      "epoch": 0.7136187282172683,
+      "grad_norm": 7.721557509382751,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 465980716,
+      "step": 2713
+    },
+    {
+      "epoch": 0.7136187282172683,
+      "loss": 0.1155381053686142,
+      "loss_ce": 0.0032639428973197937,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 465980716,
+      "step": 2713
+    },
+    {
+      "epoch": 0.7138817649766555,
+      "grad_norm": 4.213278179953207,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 466153208,
+      "step": 2714
+    },
+    {
+      "epoch": 0.7138817649766555,
+      "loss": 0.1079680472612381,
+      "loss_ce": 0.002407738706097007,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 466153208,
+      "step": 2714
+    },
+    {
+      "epoch": 0.7141448017360427,
+      "grad_norm": 4.8611508745546725,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 466325716,
+      "step": 2715
+    },
+    {
+      "epoch": 0.7141448017360427,
+      "loss": 0.07399383187294006,
+      "loss_ce": 0.0005075072403997183,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 466325716,
+      "step": 2715
+    },
+    {
+      "epoch": 0.7144078384954298,
+      "grad_norm": 34.00071276149313,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 466497816,
+      "step": 2716
+    },
+    {
+      "epoch": 0.7144078384954298,
+      "loss": 0.11154329776763916,
+      "loss_ce": 0.00032198382541537285,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 466497816,
+      "step": 2716
+    },
+    {
+      "epoch": 0.7146708752548169,
+      "grad_norm": 4.289011543265615,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 466666960,
+      "step": 2717
+    },
+    {
+      "epoch": 0.7146708752548169,
+      "loss": 0.10135161876678467,
+      "loss_ce": 0.000498652458190918,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 466666960,
+      "step": 2717
+    },
+    {
+      "epoch": 0.714933912014204,
+      "grad_norm": 5.080575214282277,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 466839356,
+      "step": 2718
+    },
+    {
+      "epoch": 0.714933912014204,
+      "loss": 0.04582955688238144,
+      "loss_ce": 9.896683332044631e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 466839356,
+      "step": 2718
+    },
+    {
+      "epoch": 0.7151969487735911,
+      "grad_norm": 5.237294473459814,
+      "learning_rate": 5e-06,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 467011304,
+      "step": 2719
+    },
+    {
+      "epoch": 0.7151969487735911,
+      "loss": 0.0945780873298645,
+      "loss_ce": 0.0005534276133403182,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 467011304,
+      "step": 2719
+    },
+    {
+      "epoch": 0.7154599855329782,
+      "grad_norm": 3.135766341307675,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 467180536,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7154599855329782,
+      "loss": 0.042348556220531464,
+      "loss_ce": 0.0003716244827955961,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 467180536,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7157230222923654,
+      "grad_norm": 3.5558668353050815,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 467352776,
+      "step": 2721
+    },
+    {
+      "epoch": 0.7157230222923654,
+      "loss": 0.175160214304924,
+      "loss_ce": 0.0027969309594482183,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 467352776,
+      "step": 2721
+    },
+    {
+      "epoch": 0.7159860590517525,
+      "grad_norm": 27.92814548048443,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 467524800,
+      "step": 2722
+    },
+    {
+      "epoch": 0.7159860590517525,
+      "loss": 0.11873021721839905,
+      "loss_ce": 0.004441884811967611,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 467524800,
+      "step": 2722
+    },
+    {
+      "epoch": 0.7162490958111396,
+      "grad_norm": 4.250603227803627,
+      "learning_rate": 5e-06,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 467693664,
+      "step": 2723
+    },
+    {
+      "epoch": 0.7162490958111396,
+      "loss": 0.07381434738636017,
+      "loss_ce": 0.0007857821765355766,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 467693664,
+      "step": 2723
+    },
+    {
+      "epoch": 0.7165121325705267,
+      "grad_norm": 29.613017586366666,
+      "learning_rate": 5e-06,
+      "loss": 0.1783,
+      "num_input_tokens_seen": 467865648,
+      "step": 2724
+    },
+    {
+      "epoch": 0.7165121325705267,
+      "loss": 0.06583578884601593,
+      "loss_ce": 0.0012300718808546662,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 467865648,
+      "step": 2724
+    },
+    {
+      "epoch": 0.7167751693299138,
+      "grad_norm": 9.322275949588843,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 468035300,
+      "step": 2725
+    },
+    {
+      "epoch": 0.7167751693299138,
+      "loss": 0.11070144176483154,
+      "loss_ce": 0.00022780938888899982,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 468035300,
+      "step": 2725
+    },
+    {
+      "epoch": 0.717038206089301,
+      "grad_norm": 7.146952822401602,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 468207524,
+      "step": 2726
+    },
+    {
+      "epoch": 0.717038206089301,
+      "loss": 0.11441102623939514,
+      "loss_ce": 0.0002447651932016015,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 468207524,
+      "step": 2726
+    },
+    {
+      "epoch": 0.7173012428486881,
+      "grad_norm": 11.196754628011416,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 468379608,
+      "step": 2727
+    },
+    {
+      "epoch": 0.7173012428486881,
+      "loss": 0.16638273000717163,
+      "loss_ce": 0.005860275588929653,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 468379608,
+      "step": 2727
+    },
+    {
+      "epoch": 0.7175642796080752,
+      "grad_norm": 10.776354846598704,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 468551976,
+      "step": 2728
+    },
+    {
+      "epoch": 0.7175642796080752,
+      "loss": 0.11593279242515564,
+      "loss_ce": 0.002559985499829054,
+      "loss_iou": 0.5625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 468551976,
+      "step": 2728
+    },
+    {
+      "epoch": 0.7178273163674623,
+      "grad_norm": 5.353937326922854,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 468724232,
+      "step": 2729
+    },
+    {
+      "epoch": 0.7178273163674623,
+      "loss": 0.09276724606752396,
+      "loss_ce": 0.00023794792650733143,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 468724232,
+      "step": 2729
+    },
+    {
+      "epoch": 0.7180903531268494,
+      "grad_norm": 4.435607019386441,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 468896348,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7180903531268494,
+      "loss": 0.14431005716323853,
+      "loss_ce": 0.001487781759351492,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 468896348,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7183533898862366,
+      "grad_norm": 5.860440789547134,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 469068240,
+      "step": 2731
+    },
+    {
+      "epoch": 0.7183533898862366,
+      "loss": 0.11086121201515198,
+      "loss_ce": 0.003713999642059207,
+      "loss_iou": 0.5,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 469068240,
+      "step": 2731
+    },
+    {
+      "epoch": 0.7186164266456238,
+      "grad_norm": 11.64009264254108,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 469238748,
+      "step": 2732
+    },
+    {
+      "epoch": 0.7186164266456238,
+      "loss": 0.14171399176120758,
+      "loss_ce": 0.0019129666034132242,
+      "loss_iou": NaN,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 469238748,
+      "step": 2732
+    },
+    {
+      "epoch": 0.7188794634050109,
+      "grad_norm": 19.26931594616413,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 469409356,
+      "step": 2733
+    },
+    {
+      "epoch": 0.7188794634050109,
+      "loss": 0.16873466968536377,
+      "loss_ce": 0.0015593739226460457,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 469409356,
+      "step": 2733
+    },
+    {
+      "epoch": 0.719142500164398,
+      "grad_norm": 39.88591044746162,
+      "learning_rate": 5e-06,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 469581536,
+      "step": 2734
+    },
+    {
+      "epoch": 0.719142500164398,
+      "loss": 0.13763278722763062,
+      "loss_ce": 0.0018295738846063614,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 469581536,
+      "step": 2734
+    },
+    {
+      "epoch": 0.7194055369237851,
+      "grad_norm": 5.070448909039763,
+      "learning_rate": 5e-06,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 469753888,
+      "step": 2735
+    },
+    {
+      "epoch": 0.7194055369237851,
+      "loss": 0.11928269267082214,
+      "loss_ce": 0.0068864524364471436,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 469753888,
+      "step": 2735
+    },
+    {
+      "epoch": 0.7196685736831723,
+      "grad_norm": 4.73412496708124,
+      "learning_rate": 5e-06,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 469926224,
+      "step": 2736
+    },
+    {
+      "epoch": 0.7196685736831723,
+      "loss": 0.10194897651672363,
+      "loss_ce": 0.000539067608769983,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 469926224,
+      "step": 2736
+    },
+    {
+      "epoch": 0.7199316104425594,
+      "grad_norm": 14.69635519193442,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 470096688,
+      "step": 2737
+    },
+    {
+      "epoch": 0.7199316104425594,
+      "loss": 0.14469808340072632,
+      "loss_ce": 0.0047444626688957214,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 470096688,
+      "step": 2737
+    },
+    {
+      "epoch": 0.7201946472019465,
+      "grad_norm": 5.497552383274807,
+      "learning_rate": 5e-06,
+      "loss": 0.1507,
+      "num_input_tokens_seen": 470268692,
+      "step": 2738
+    },
+    {
+      "epoch": 0.7201946472019465,
+      "loss": 0.07712777704000473,
+      "loss_ce": 0.0003760677354875952,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 470268692,
+      "step": 2738
+    },
+    {
+      "epoch": 0.7204576839613336,
+      "grad_norm": 4.073372808076834,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 470440752,
+      "step": 2739
+    },
+    {
+      "epoch": 0.7204576839613336,
+      "loss": 0.1480931043624878,
+      "loss_ce": 0.001303559634834528,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 470440752,
+      "step": 2739
+    },
+    {
+      "epoch": 0.7207207207207207,
+      "grad_norm": 5.675528284561051,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 470612568,
+      "step": 2740
+    },
+    {
+      "epoch": 0.7207207207207207,
+      "loss": 0.07299304753541946,
+      "loss_ce": 0.004481084644794464,
+      "loss_iou": 0.625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 470612568,
+      "step": 2740
+    },
+    {
+      "epoch": 0.7209837574801078,
+      "grad_norm": 20.879698464276387,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 470784724,
+      "step": 2741
+    },
+    {
+      "epoch": 0.7209837574801078,
+      "loss": 0.1316141039133072,
+      "loss_ce": 0.001822838094085455,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 470784724,
+      "step": 2741
+    },
+    {
+      "epoch": 0.721246794239495,
+      "grad_norm": 10.194443634028673,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 470957128,
+      "step": 2742
+    },
+    {
+      "epoch": 0.721246794239495,
+      "loss": 0.2153215855360031,
+      "loss_ce": 0.0009050846565514803,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 470957128,
+      "step": 2742
+    },
+    {
+      "epoch": 0.7215098309988821,
+      "grad_norm": 4.046503425790058,
+      "learning_rate": 5e-06,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 471129076,
+      "step": 2743
+    },
+    {
+      "epoch": 0.7215098309988821,
+      "loss": 0.17502932250499725,
+      "loss_ce": 0.005137969274073839,
+      "loss_iou": NaN,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 471129076,
+      "step": 2743
+    },
+    {
+      "epoch": 0.7217728677582692,
+      "grad_norm": 16.718028178353897,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 471301596,
+      "step": 2744
+    },
+    {
+      "epoch": 0.7217728677582692,
+      "loss": 0.1273680329322815,
+      "loss_ce": 0.0013304388849064708,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 471301596,
+      "step": 2744
+    },
+    {
+      "epoch": 0.7220359045176563,
+      "grad_norm": 19.351969856870078,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 471473704,
+      "step": 2745
+    },
+    {
+      "epoch": 0.7220359045176563,
+      "loss": 0.08187679946422577,
+      "loss_ce": 4.3911892134929076e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 471473704,
+      "step": 2745
+    },
+    {
+      "epoch": 0.7222989412770434,
+      "grad_norm": 3.5611349833279236,
+      "learning_rate": 5e-06,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 471644084,
+      "step": 2746
+    },
+    {
+      "epoch": 0.7222989412770434,
+      "loss": 0.04295755550265312,
+      "loss_ce": 0.00018716827617026865,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 471644084,
+      "step": 2746
+    },
+    {
+      "epoch": 0.7225619780364306,
+      "grad_norm": 4.500519410155881,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 471814452,
+      "step": 2747
+    },
+    {
+      "epoch": 0.7225619780364306,
+      "loss": 0.15782231092453003,
+      "loss_ce": 0.0029913773760199547,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 471814452,
+      "step": 2747
+    },
+    {
+      "epoch": 0.7228250147958177,
+      "grad_norm": 5.3088355134433645,
+      "learning_rate": 5e-06,
+      "loss": 0.1564,
+      "num_input_tokens_seen": 471980364,
+      "step": 2748
+    },
+    {
+      "epoch": 0.7228250147958177,
+      "loss": 0.16449454426765442,
+      "loss_ce": 0.0011339561315253377,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 471980364,
+      "step": 2748
+    },
+    {
+      "epoch": 0.7230880515552048,
+      "grad_norm": 8.162974487273814,
+      "learning_rate": 5e-06,
+      "loss": 0.1117,
+      "num_input_tokens_seen": 472152464,
+      "step": 2749
+    },
+    {
+      "epoch": 0.7230880515552048,
+      "loss": 0.08619838953018188,
+      "loss_ce": 0.0012984833447262645,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 472152464,
+      "step": 2749
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "grad_norm": 6.354241375539321,
+      "learning_rate": 5e-06,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_websight_new_CIoU": 0.8880318701267242,
+      "eval_websight_new_GIoU": 0.8903599679470062,
+      "eval_websight_new_IoU": 0.8917953372001648,
+      "eval_websight_new_MAE_all": 0.014968848787248135,
+      "eval_websight_new_MAE_h": 0.008249826729297638,
+      "eval_websight_new_MAE_w": 0.021866537630558014,
+      "eval_websight_new_MAE_x": 0.02344994805753231,
+      "eval_websight_new_MAE_y": 0.006309080636128783,
+      "eval_websight_new_NUM_probability": 0.9999924898147583,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.0760061964392662,
+      "eval_websight_new_loss_ce": 5.840479570906609e-05,
+      "eval_websight_new_loss_iou": 0.35406494140625,
+      "eval_websight_new_loss_num": 0.013774871826171875,
+      "eval_websight_new_loss_xval": 0.0689239501953125,
+      "eval_websight_new_runtime": 59.6817,
+      "eval_websight_new_samples_per_second": 0.838,
+      "eval_websight_new_steps_per_second": 0.034,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_seeclick_CIoU": 0.6159887313842773,
+      "eval_seeclick_GIoU": 0.6205049157142639,
+      "eval_seeclick_IoU": 0.6426993608474731,
+      "eval_seeclick_MAE_all": 0.048651453107595444,
+      "eval_seeclick_MAE_h": 0.030230149626731873,
+      "eval_seeclick_MAE_w": 0.06347078271210194,
+      "eval_seeclick_MAE_x": 0.0767427384853363,
+      "eval_seeclick_MAE_y": 0.024162148125469685,
+      "eval_seeclick_NUM_probability": 0.9999720454216003,
+      "eval_seeclick_inside_bbox": 0.8920454680919647,
+      "eval_seeclick_loss": 0.22562426328659058,
+      "eval_seeclick_loss_ce": 0.008924027904868126,
+      "eval_seeclick_loss_iou": 0.5213623046875,
+      "eval_seeclick_loss_num": 0.04332733154296875,
+      "eval_seeclick_loss_xval": 0.2166900634765625,
+      "eval_seeclick_runtime": 76.0092,
+      "eval_seeclick_samples_per_second": 0.566,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_icons_CIoU": 0.8715368807315826,
+      "eval_icons_GIoU": 0.8687921464443207,
+      "eval_icons_IoU": 0.8755140602588654,
+      "eval_icons_MAE_all": 0.015967791434377432,
+      "eval_icons_MAE_h": 0.016439005732536316,
+      "eval_icons_MAE_w": 0.015302729327231646,
+      "eval_icons_MAE_x": 0.015842752531170845,
+      "eval_icons_MAE_y": 0.016286680474877357,
+      "eval_icons_NUM_probability": 0.9999882578849792,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.057806555181741714,
+      "eval_icons_loss_ce": 1.1600203379202867e-05,
+      "eval_icons_loss_iou": 0.617919921875,
+      "eval_icons_loss_num": 0.010974884033203125,
+      "eval_icons_loss_xval": 0.0548553466796875,
+      "eval_icons_runtime": 81.9252,
+      "eval_icons_samples_per_second": 0.61,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_screenspot_CIoU": 0.5503915150960287,
+      "eval_screenspot_GIoU": 0.5448275804519653,
+      "eval_screenspot_IoU": 0.5854077339172363,
+      "eval_screenspot_MAE_all": 0.0787569632132848,
+      "eval_screenspot_MAE_h": 0.05612564583619436,
+      "eval_screenspot_MAE_w": 0.1299388830860456,
+      "eval_screenspot_MAE_x": 0.07724836965401967,
+      "eval_screenspot_MAE_y": 0.05171496793627739,
+      "eval_screenspot_NUM_probability": 0.9999065001805624,
+      "eval_screenspot_inside_bbox": 0.8291666706403097,
+      "eval_screenspot_loss": 0.9146350622177124,
+      "eval_screenspot_loss_ce": 0.567759374777476,
+      "eval_screenspot_loss_iou": 0.5614013671875,
+      "eval_screenspot_loss_num": 0.06831868489583333,
+      "eval_screenspot_loss_xval": 0.3416341145833333,
+      "eval_screenspot_runtime": 140.0636,
+      "eval_screenspot_samples_per_second": 0.635,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "loss": 0.9040678143501282,
+      "loss_ce": 0.5593412518501282,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.723614125073979,
+      "grad_norm": 4.425480410673166,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 472496880,
+      "step": 2751
+    },
+    {
+      "epoch": 0.723614125073979,
+      "loss": 0.10025835037231445,
+      "loss_ce": 0.0005421665264293551,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 472496880,
+      "step": 2751
+    },
+    {
+      "epoch": 0.7238771618333663,
+      "grad_norm": 4.0784104980939135,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 472668928,
+      "step": 2752
+    },
+    {
+      "epoch": 0.7238771618333663,
+      "loss": 0.0833350419998169,
+      "loss_ce": 0.00038827050593681633,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 472668928,
+      "step": 2752
+    },
+    {
+      "epoch": 0.7241401985927534,
+      "grad_norm": 6.523452376364869,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 472841040,
+      "step": 2753
+    },
+    {
+      "epoch": 0.7241401985927534,
+      "loss": 0.13216346502304077,
+      "loss_ce": 0.00023597091785632074,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 472841040,
+      "step": 2753
+    },
+    {
+      "epoch": 0.7244032353521405,
+      "grad_norm": 13.511689037708722,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 473013468,
+      "step": 2754
+    },
+    {
+      "epoch": 0.7244032353521405,
+      "loss": 0.0918908640742302,
+      "loss_ce": 0.001101069850847125,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 473013468,
+      "step": 2754
+    },
+    {
+      "epoch": 0.7246662721115276,
+      "grad_norm": 4.609556566608224,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 473180556,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7246662721115276,
+      "loss": 0.2066127210855484,
+      "loss_ce": 0.0006495795678347349,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 473180556,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7249293088709147,
+      "grad_norm": 17.674016284792703,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 473352636,
+      "step": 2756
+    },
+    {
+      "epoch": 0.7249293088709147,
+      "loss": 0.13866102695465088,
+      "loss_ce": 0.0018507244531065226,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 473352636,
+      "step": 2756
+    },
+    {
+      "epoch": 0.7251923456303019,
+      "grad_norm": 12.378222072828299,
+      "learning_rate": 5e-06,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 473524848,
+      "step": 2757
+    },
+    {
+      "epoch": 0.7251923456303019,
+      "loss": 0.059792935848236084,
+      "loss_ce": 0.0014433301985263824,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 473524848,
+      "step": 2757
+    },
+    {
+      "epoch": 0.725455382389689,
+      "grad_norm": 4.317210813813628,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 473697016,
+      "step": 2758
+    },
+    {
+      "epoch": 0.725455382389689,
+      "loss": 0.1696222722530365,
+      "loss_ce": 0.0004328044014982879,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 473697016,
+      "step": 2758
+    },
+    {
+      "epoch": 0.7257184191490761,
+      "grad_norm": 14.320905421992641,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 473868992,
+      "step": 2759
+    },
+    {
+      "epoch": 0.7257184191490761,
+      "loss": 0.07304464280605316,
+      "loss_ce": 0.0016335132531821728,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 473868992,
+      "step": 2759
+    },
+    {
+      "epoch": 0.7259814559084632,
+      "grad_norm": 3.9251643218505756,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 474041228,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7259814559084632,
+      "loss": 0.09449034929275513,
+      "loss_ce": 0.0015948471846058965,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 474041228,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7262444926678503,
+      "grad_norm": 4.80079690317671,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 474213616,
+      "step": 2761
+    },
+    {
+      "epoch": 0.7262444926678503,
+      "loss": 0.14180995523929596,
+      "loss_ce": 0.001276510301977396,
+      "loss_iou": 0.53125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 474213616,
+      "step": 2761
+    },
+    {
+      "epoch": 0.7265075294272375,
+      "grad_norm": 5.815165285313656,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 474385980,
+      "step": 2762
+    },
+    {
+      "epoch": 0.7265075294272375,
+      "loss": 0.14968647062778473,
+      "loss_ce": 0.0027748444117605686,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 474385980,
+      "step": 2762
+    },
+    {
+      "epoch": 0.7267705661866246,
+      "grad_norm": 18.689053225654995,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 474558212,
+      "step": 2763
+    },
+    {
+      "epoch": 0.7267705661866246,
+      "loss": 0.07762917876243591,
+      "loss_ce": 0.0015488516073673964,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 474558212,
+      "step": 2763
+    },
+    {
+      "epoch": 0.7270336029460117,
+      "grad_norm": 26.06170268014936,
+      "learning_rate": 5e-06,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 474730824,
+      "step": 2764
+    },
+    {
+      "epoch": 0.7270336029460117,
+      "loss": 0.2366107702255249,
+      "loss_ce": 0.0011676568537950516,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 474730824,
+      "step": 2764
+    },
+    {
+      "epoch": 0.7272966397053988,
+      "grad_norm": 4.246687258335021,
+      "learning_rate": 5e-06,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 474901372,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7272966397053988,
+      "loss": 0.05860138311982155,
+      "loss_ce": 0.0005874672788195312,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 474901372,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7275596764647859,
+      "grad_norm": 10.116948882200285,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 475070388,
+      "step": 2766
+    },
+    {
+      "epoch": 0.7275596764647859,
+      "loss": 0.14292961359024048,
+      "loss_ce": 0.0015111502725630999,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 475070388,
+      "step": 2766
+    },
+    {
+      "epoch": 0.727822713224173,
+      "grad_norm": 5.556641114443334,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 475242528,
+      "step": 2767
+    },
+    {
+      "epoch": 0.727822713224173,
+      "loss": 0.05033715069293976,
+      "loss_ce": 0.00010521705553401262,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 475242528,
+      "step": 2767
+    },
+    {
+      "epoch": 0.7280857499835602,
+      "grad_norm": 7.020407355867949,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 475414560,
+      "step": 2768
+    },
+    {
+      "epoch": 0.7280857499835602,
+      "loss": 0.14076019823551178,
+      "loss_ce": 0.003125916002318263,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 475414560,
+      "step": 2768
+    },
+    {
+      "epoch": 0.7283487867429473,
+      "grad_norm": 3.869087312985774,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 475586900,
+      "step": 2769
+    },
+    {
+      "epoch": 0.7283487867429473,
+      "loss": 0.1432042121887207,
+      "loss_ce": 0.001144890207797289,
+      "loss_iou": NaN,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 475586900,
+      "step": 2769
+    },
+    {
+      "epoch": 0.7286118235023344,
+      "grad_norm": 5.203369255610556,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 475759012,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7286118235023344,
+      "loss": 0.09897112846374512,
+      "loss_ce": 0.000979190575890243,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 475759012,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7288748602617215,
+      "grad_norm": 5.470559222873129,
+      "learning_rate": 5e-06,
+      "loss": 0.1425,
+      "num_input_tokens_seen": 475931112,
+      "step": 2771
+    },
+    {
+      "epoch": 0.7288748602617215,
+      "loss": 0.057690735906362534,
+      "loss_ce": 0.00324737885966897,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 475931112,
+      "step": 2771
+    },
+    {
+      "epoch": 0.7291378970211086,
+      "grad_norm": 18.974366841188825,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 476100764,
+      "step": 2772
+    },
+    {
+      "epoch": 0.7291378970211086,
+      "loss": 0.15548212826251984,
+      "loss_ce": 0.0019024083158001304,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 476100764,
+      "step": 2772
+    },
+    {
+      "epoch": 0.7294009337804959,
+      "grad_norm": 5.671161770383539,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 476272960,
+      "step": 2773
+    },
+    {
+      "epoch": 0.7294009337804959,
+      "loss": 0.0945616215467453,
+      "loss_ce": 0.00023179112758953124,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 476272960,
+      "step": 2773
+    },
+    {
+      "epoch": 0.729663970539883,
+      "grad_norm": 34.00148217519992,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 476445160,
+      "step": 2774
+    },
+    {
+      "epoch": 0.729663970539883,
+      "loss": 0.2345729023218155,
+      "loss_ce": 0.0021205078810453415,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 476445160,
+      "step": 2774
+    },
+    {
+      "epoch": 0.7299270072992701,
+      "grad_norm": 4.401043478623018,
+      "learning_rate": 5e-06,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 476617452,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7299270072992701,
+      "loss": 0.07482883334159851,
+      "loss_ce": 0.00036594344419427216,
+      "loss_iou": 0.671875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 476617452,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7301900440586572,
+      "grad_norm": 4.444423155502646,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 476789580,
+      "step": 2776
+    },
+    {
+      "epoch": 0.7301900440586572,
+      "loss": 0.08125682920217514,
+      "loss_ce": 0.0011787032708525658,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 476789580,
+      "step": 2776
+    },
+    {
+      "epoch": 0.7304530808180443,
+      "grad_norm": 9.388146192273584,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 476961420,
+      "step": 2777
+    },
+    {
+      "epoch": 0.7304530808180443,
+      "loss": 0.09009505808353424,
+      "loss_ce": 0.0032420377247035503,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 476961420,
+      "step": 2777
+    },
+    {
+      "epoch": 0.7307161175774315,
+      "grad_norm": 13.509674001993902,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 477133352,
+      "step": 2778
+    },
+    {
+      "epoch": 0.7307161175774315,
+      "loss": 0.15684369206428528,
+      "loss_ce": 0.0034623502288013697,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 477133352,
+      "step": 2778
+    },
+    {
+      "epoch": 0.7309791543368186,
+      "grad_norm": 10.371891827363712,
+      "learning_rate": 5e-06,
+      "loss": 0.1536,
+      "num_input_tokens_seen": 477305700,
+      "step": 2779
+    },
+    {
+      "epoch": 0.7309791543368186,
+      "loss": 0.1920810043811798,
+      "loss_ce": 0.003238243516534567,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 477305700,
+      "step": 2779
+    },
+    {
+      "epoch": 0.7312421910962057,
+      "grad_norm": 4.473462123830256,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 477476192,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7312421910962057,
+      "loss": 0.12351857125759125,
+      "loss_ce": 0.003828628221526742,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 477476192,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7315052278555928,
+      "grad_norm": 8.785650264431224,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 477648396,
+      "step": 2781
+    },
+    {
+      "epoch": 0.7315052278555928,
+      "loss": 0.10561161488294601,
+      "loss_ce": 0.00282841082662344,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 477648396,
+      "step": 2781
+    },
+    {
+      "epoch": 0.7317682646149799,
+      "grad_norm": 4.9317310113146835,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 477820536,
+      "step": 2782
+    },
+    {
+      "epoch": 0.7317682646149799,
+      "loss": 0.15500710904598236,
+      "loss_ce": 0.0015952409012243152,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 477820536,
+      "step": 2782
+    },
+    {
+      "epoch": 0.7320313013743671,
+      "grad_norm": 14.38075821387521,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 477992760,
+      "step": 2783
+    },
+    {
+      "epoch": 0.7320313013743671,
+      "loss": 0.09361109137535095,
+      "loss_ce": 0.0013869699323549867,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 477992760,
+      "step": 2783
+    },
+    {
+      "epoch": 0.7322943381337542,
+      "grad_norm": 3.3741559107447294,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 478164884,
+      "step": 2784
+    },
+    {
+      "epoch": 0.7322943381337542,
+      "loss": 0.06834319978952408,
+      "loss_ce": 0.00392058864235878,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 478164884,
+      "step": 2784
+    },
+    {
+      "epoch": 0.7325573748931413,
+      "grad_norm": 4.560868774415877,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 478337276,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7325573748931413,
+      "loss": 0.08397236466407776,
+      "loss_ce": 0.0004152356996200979,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 478337276,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7328204116525284,
+      "grad_norm": 17.451789372881052,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 478509332,
+      "step": 2786
+    },
+    {
+      "epoch": 0.7328204116525284,
+      "loss": 0.1723887324333191,
+      "loss_ce": 0.00039166733040474355,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 478509332,
+      "step": 2786
+    },
+    {
+      "epoch": 0.7330834484119155,
+      "grad_norm": 6.047350371102969,
+      "learning_rate": 5e-06,
+      "loss": 0.1491,
+      "num_input_tokens_seen": 478681352,
+      "step": 2787
+    },
+    {
+      "epoch": 0.7330834484119155,
+      "loss": 0.04348166286945343,
+      "loss_ce": 0.0019167213467881083,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 478681352,
+      "step": 2787
+    },
+    {
+      "epoch": 0.7333464851713027,
+      "grad_norm": 4.3763950262189235,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 478853448,
+      "step": 2788
+    },
+    {
+      "epoch": 0.7333464851713027,
+      "loss": 0.11647917330265045,
+      "loss_ce": 0.0007565193227492273,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 478853448,
+      "step": 2788
+    },
+    {
+      "epoch": 0.7336095219306898,
+      "grad_norm": 3.397549309274833,
+      "learning_rate": 5e-06,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 479025420,
+      "step": 2789
+    },
+    {
+      "epoch": 0.7336095219306898,
+      "loss": 0.04994508996605873,
+      "loss_ce": 0.0003540250181686133,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 479025420,
+      "step": 2789
+    },
+    {
+      "epoch": 0.733872558690077,
+      "grad_norm": 6.639247678980395,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 479197512,
+      "step": 2790
+    },
+    {
+      "epoch": 0.733872558690077,
+      "loss": 0.0962657481431961,
+      "loss_ce": 0.0033092054072767496,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 479197512,
+      "step": 2790
+    },
+    {
+      "epoch": 0.734135595449464,
+      "grad_norm": 70.18841715820902,
+      "learning_rate": 5e-06,
+      "loss": 0.121,
+      "num_input_tokens_seen": 479369608,
+      "step": 2791
+    },
+    {
+      "epoch": 0.734135595449464,
+      "loss": 0.057409316301345825,
+      "loss_ce": 0.000661880592815578,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 479369608,
+      "step": 2791
+    },
+    {
+      "epoch": 0.7343986322088512,
+      "grad_norm": 8.444086759452258,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 479541840,
+      "step": 2792
+    },
+    {
+      "epoch": 0.7343986322088512,
+      "loss": 0.18742145597934723,
+      "loss_ce": 0.0015999219613149762,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 479541840,
+      "step": 2792
+    },
+    {
+      "epoch": 0.7346616689682383,
+      "grad_norm": 5.294583182815855,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 479714196,
+      "step": 2793
+    },
+    {
+      "epoch": 0.7346616689682383,
+      "loss": 0.1654970794916153,
+      "loss_ce": 0.002441658638417721,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 479714196,
+      "step": 2793
+    },
+    {
+      "epoch": 0.7349247057276255,
+      "grad_norm": 4.160027277201995,
+      "learning_rate": 5e-06,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 479886580,
+      "step": 2794
+    },
+    {
+      "epoch": 0.7349247057276255,
+      "loss": 0.11283927410840988,
+      "loss_ce": 0.0021825337316840887,
+      "loss_iou": 0.5,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 479886580,
+      "step": 2794
+    },
+    {
+      "epoch": 0.7351877424870126,
+      "grad_norm": 8.261194775106992,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 480058736,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7351877424870126,
+      "loss": 0.12811481952667236,
+      "loss_ce": 0.0015584270004183054,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 480058736,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7354507792463997,
+      "grad_norm": 3.186827546083555,
+      "learning_rate": 5e-06,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 480230860,
+      "step": 2796
+    },
+    {
+      "epoch": 0.7354507792463997,
+      "loss": 0.12094112485647202,
+      "loss_ce": 0.00286861858330667,
+      "loss_iou": 0.296875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 480230860,
+      "step": 2796
+    },
+    {
+      "epoch": 0.7357138160057868,
+      "grad_norm": 3.6199423783668534,
+      "learning_rate": 5e-06,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 480403260,
+      "step": 2797
+    },
+    {
+      "epoch": 0.7357138160057868,
+      "loss": 0.0642661452293396,
+      "loss_ce": 0.0005149244680069387,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 480403260,
+      "step": 2797
+    },
+    {
+      "epoch": 0.7359768527651739,
+      "grad_norm": 4.853254066183438,
+      "learning_rate": 5e-06,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 480575740,
+      "step": 2798
+    },
+    {
+      "epoch": 0.7359768527651739,
+      "loss": 0.1287141740322113,
+      "loss_ce": 0.000998095260001719,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 480575740,
+      "step": 2798
+    },
+    {
+      "epoch": 0.7362398895245611,
+      "grad_norm": 6.187299771807659,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 480745892,
+      "step": 2799
+    },
+    {
+      "epoch": 0.7362398895245611,
+      "loss": 0.11742156744003296,
+      "loss_ce": 0.0004476910980883986,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 480745892,
+      "step": 2799
+    },
+    {
+      "epoch": 0.7365029262839482,
+      "grad_norm": 3.2510976930909194,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 480916480,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7365029262839482,
+      "loss": 0.14714287221431732,
+      "loss_ce": 7.865649240557104e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 480916480,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7367659630433353,
+      "grad_norm": 4.500033616411093,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 481088428,
+      "step": 2801
+    },
+    {
+      "epoch": 0.7367659630433353,
+      "loss": 0.07280252873897552,
+      "loss_ce": 0.0012845833553001285,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 481088428,
+      "step": 2801
+    },
+    {
+      "epoch": 0.7370289998027224,
+      "grad_norm": 6.733873919958713,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 481260648,
+      "step": 2802
+    },
+    {
+      "epoch": 0.7370289998027224,
+      "loss": 0.13488659262657166,
+      "loss_ce": 0.002318233484402299,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 481260648,
+      "step": 2802
+    },
+    {
+      "epoch": 0.7372920365621095,
+      "grad_norm": 4.443731950557251,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 481432928,
+      "step": 2803
+    },
+    {
+      "epoch": 0.7372920365621095,
+      "loss": 0.05556986480951309,
+      "loss_ce": 0.0003940859460271895,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 481432928,
+      "step": 2803
+    },
+    {
+      "epoch": 0.7375550733214967,
+      "grad_norm": 3.9159514379676663,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 481603740,
+      "step": 2804
+    },
+    {
+      "epoch": 0.7375550733214967,
+      "loss": 0.07777837663888931,
+      "loss_ce": 5.010394670534879e-05,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 481603740,
+      "step": 2804
+    },
+    {
+      "epoch": 0.7378181100808838,
+      "grad_norm": 3.806670191888079,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 481774292,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7378181100808838,
+      "loss": 0.046146463602781296,
+      "loss_ce": 0.00412375945597887,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 481774292,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7380811468402709,
+      "grad_norm": 18.2473830072325,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 481946452,
+      "step": 2806
+    },
+    {
+      "epoch": 0.7380811468402709,
+      "loss": 0.15219584107398987,
+      "loss_ce": 0.0003708918229676783,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 481946452,
+      "step": 2806
+    },
+    {
+      "epoch": 0.738344183599658,
+      "grad_norm": 5.953555323593533,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 482118496,
+      "step": 2807
+    },
+    {
+      "epoch": 0.738344183599658,
+      "loss": 0.14251753687858582,
+      "loss_ce": 0.00039717700565233827,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 482118496,
+      "step": 2807
+    },
+    {
+      "epoch": 0.7386072203590451,
+      "grad_norm": 7.279415163791317,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 482287100,
+      "step": 2808
+    },
+    {
+      "epoch": 0.7386072203590451,
+      "loss": 0.11856916546821594,
+      "loss_ce": 0.0012290815357118845,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 482287100,
+      "step": 2808
+    },
+    {
+      "epoch": 0.7388702571184323,
+      "grad_norm": 60.57211297247697,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 482459096,
+      "step": 2809
+    },
+    {
+      "epoch": 0.7388702571184323,
+      "loss": 0.14476290345191956,
+      "loss_ce": 0.003466519294306636,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 482459096,
+      "step": 2809
+    },
+    {
+      "epoch": 0.7391332938778195,
+      "grad_norm": 4.216089199918953,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 482631768,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7391332938778195,
+      "loss": 0.1267136037349701,
+      "loss_ce": 0.0018051671795547009,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 482631768,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7393963306372066,
+      "grad_norm": 5.529488884431239,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 482803968,
+      "step": 2811
+    },
+    {
+      "epoch": 0.7393963306372066,
+      "loss": 0.12075556814670563,
+      "loss_ce": 0.0023778879549354315,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 482803968,
+      "step": 2811
+    },
+    {
+      "epoch": 0.7396593673965937,
+      "grad_norm": 14.970093925458164,
+      "learning_rate": 5e-06,
+      "loss": 0.089,
+      "num_input_tokens_seen": 482976316,
+      "step": 2812
+    },
+    {
+      "epoch": 0.7396593673965937,
+      "loss": 0.11577010154724121,
+      "loss_ce": 0.00230574794113636,
+      "loss_iou": 0.53125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 482976316,
+      "step": 2812
+    },
+    {
+      "epoch": 0.7399224041559808,
+      "grad_norm": 16.71817355365124,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 483148324,
+      "step": 2813
+    },
+    {
+      "epoch": 0.7399224041559808,
+      "loss": 0.12452364712953568,
+      "loss_ce": 0.0020718637388199568,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 483148324,
+      "step": 2813
+    },
+    {
+      "epoch": 0.740185440915368,
+      "grad_norm": 6.067200171837272,
+      "learning_rate": 5e-06,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 483320396,
+      "step": 2814
+    },
+    {
+      "epoch": 0.740185440915368,
+      "loss": 0.08812370151281357,
+      "loss_ce": 0.003498460166156292,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 483320396,
+      "step": 2814
+    },
+    {
+      "epoch": 0.7404484776747551,
+      "grad_norm": 4.557027585995325,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 483492288,
+      "step": 2815
+    },
+    {
+      "epoch": 0.7404484776747551,
+      "loss": 0.11246176064014435,
+      "loss_ce": 0.0007674178341403604,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 483492288,
+      "step": 2815
+    },
+    {
+      "epoch": 0.7407115144341422,
+      "grad_norm": 4.433500401049952,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 483664284,
+      "step": 2816
+    },
+    {
+      "epoch": 0.7407115144341422,
+      "loss": 0.11581188440322876,
+      "loss_ce": 0.0005164705216884613,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 483664284,
+      "step": 2816
+    },
+    {
+      "epoch": 0.7409745511935293,
+      "grad_norm": 4.273944166752719,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 483836152,
+      "step": 2817
+    },
+    {
+      "epoch": 0.7409745511935293,
+      "loss": 0.1244652196764946,
+      "loss_ce": 0.0025780070573091507,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 483836152,
+      "step": 2817
+    },
+    {
+      "epoch": 0.7412375879529164,
+      "grad_norm": 7.341058016704505,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 484005324,
+      "step": 2818
+    },
+    {
+      "epoch": 0.7412375879529164,
+      "loss": 0.09527404606342316,
+      "loss_ce": 0.0004711919464170933,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 484005324,
+      "step": 2818
+    },
+    {
+      "epoch": 0.7415006247123035,
+      "grad_norm": 5.580265155977227,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 484175008,
+      "step": 2819
+    },
+    {
+      "epoch": 0.7415006247123035,
+      "loss": 0.061688005924224854,
+      "loss_ce": 0.000408706720918417,
+      "loss_iou": 0.8203125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 484175008,
+      "step": 2819
+    },
+    {
+      "epoch": 0.7417636614716907,
+      "grad_norm": 10.569140888378845,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 484346776,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7417636614716907,
+      "loss": 0.1439221203327179,
+      "loss_ce": 0.00351074174977839,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 484346776,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7420266982310778,
+      "grad_norm": 4.104540427593442,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 484518992,
+      "step": 2821
+    },
+    {
+      "epoch": 0.7420266982310778,
+      "loss": 0.09965825080871582,
+      "loss_ce": 0.000567668757867068,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 484518992,
+      "step": 2821
+    },
+    {
+      "epoch": 0.7422897349904649,
+      "grad_norm": 5.580054688233454,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 484691244,
+      "step": 2822
+    },
+    {
+      "epoch": 0.7422897349904649,
+      "loss": 0.17825458943843842,
+      "loss_ce": 0.002351263538002968,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 484691244,
+      "step": 2822
+    },
+    {
+      "epoch": 0.742552771749852,
+      "grad_norm": 3.6849080723348067,
+      "learning_rate": 5e-06,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 484861364,
+      "step": 2823
+    },
+    {
+      "epoch": 0.742552771749852,
+      "loss": 0.12127910554409027,
+      "loss_ce": 0.0025046863593161106,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 484861364,
+      "step": 2823
+    },
+    {
+      "epoch": 0.7428158085092391,
+      "grad_norm": 13.128530967150855,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 485032096,
+      "step": 2824
+    },
+    {
+      "epoch": 0.7428158085092391,
+      "loss": 0.044097013771533966,
+      "loss_ce": 0.0010977452620863914,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 485032096,
+      "step": 2824
+    },
+    {
+      "epoch": 0.7430788452686263,
+      "grad_norm": 29.97671813936596,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 485203848,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7430788452686263,
+      "loss": 0.1745963990688324,
+      "loss_ce": 0.0014701783657073975,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 485203848,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7433418820280134,
+      "grad_norm": 3.457370742890275,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 485376152,
+      "step": 2826
+    },
+    {
+      "epoch": 0.7433418820280134,
+      "loss": 0.10546036064624786,
+      "loss_ce": 0.0021430959459394217,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 485376152,
+      "step": 2826
+    },
+    {
+      "epoch": 0.7436049187874005,
+      "grad_norm": 14.472339934978308,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 485546204,
+      "step": 2827
+    },
+    {
+      "epoch": 0.7436049187874005,
+      "loss": 0.1524585783481598,
+      "loss_ce": 0.0006031189695931971,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 485546204,
+      "step": 2827
+    },
+    {
+      "epoch": 0.7438679555467876,
+      "grad_norm": 15.092363857837377,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 485718156,
+      "step": 2828
+    },
+    {
+      "epoch": 0.7438679555467876,
+      "loss": 0.11204935610294342,
+      "loss_ce": 0.00023295017308555543,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 485718156,
+      "step": 2828
+    },
+    {
+      "epoch": 0.7441309923061747,
+      "grad_norm": 3.949937010975672,
+      "learning_rate": 5e-06,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 485889984,
+      "step": 2829
+    },
+    {
+      "epoch": 0.7441309923061747,
+      "loss": 0.08257782459259033,
+      "loss_ce": 0.002652282826602459,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 485889984,
+      "step": 2829
+    },
+    {
+      "epoch": 0.744394029065562,
+      "grad_norm": 5.74645673965003,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 486062200,
+      "step": 2830
+    },
+    {
+      "epoch": 0.744394029065562,
+      "loss": 0.10757631063461304,
+      "loss_ce": 0.0008258260786533356,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 486062200,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7446570658249491,
+      "grad_norm": 5.133478707353028,
+      "learning_rate": 5e-06,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 486234156,
+      "step": 2831
+    },
+    {
+      "epoch": 0.7446570658249491,
+      "loss": 0.05313008278608322,
+      "loss_ce": 0.0007771779783070087,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 486234156,
+      "step": 2831
+    },
+    {
+      "epoch": 0.7449201025843362,
+      "grad_norm": 3.5956313939936164,
+      "learning_rate": 5e-06,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 486406012,
+      "step": 2832
+    },
+    {
+      "epoch": 0.7449201025843362,
+      "loss": 0.041897065937519073,
+      "loss_ce": 0.0002405716950306669,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 486406012,
+      "step": 2832
+    },
+    {
+      "epoch": 0.7451831393437233,
+      "grad_norm": 14.659992900472892,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 486577696,
+      "step": 2833
+    },
+    {
+      "epoch": 0.7451831393437233,
+      "loss": 0.10231424868106842,
+      "loss_ce": 0.00011088042083429173,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 486577696,
+      "step": 2833
+    },
+    {
+      "epoch": 0.7454461761031104,
+      "grad_norm": 4.0898533446969365,
+      "learning_rate": 5e-06,
+      "loss": 0.132,
+      "num_input_tokens_seen": 486749788,
+      "step": 2834
+    },
+    {
+      "epoch": 0.7454461761031104,
+      "loss": 0.16116830706596375,
+      "loss_ce": 0.002553200349211693,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 486749788,
+      "step": 2834
+    },
+    {
+      "epoch": 0.7457092128624976,
+      "grad_norm": 7.24387104067779,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 486921812,
+      "step": 2835
+    },
+    {
+      "epoch": 0.7457092128624976,
+      "loss": 0.15365542471408844,
+      "loss_ce": 0.0005182233871892095,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 486921812,
+      "step": 2835
+    },
+    {
+      "epoch": 0.7459722496218847,
+      "grad_norm": 4.544468769186057,
+      "learning_rate": 5e-06,
+      "loss": 0.1593,
+      "num_input_tokens_seen": 487092260,
+      "step": 2836
+    },
+    {
+      "epoch": 0.7459722496218847,
+      "loss": 0.13337098062038422,
+      "loss_ce": 0.0010772723471745849,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 487092260,
+      "step": 2836
+    },
+    {
+      "epoch": 0.7462352863812718,
+      "grad_norm": 25.066833727752858,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 487264416,
+      "step": 2837
+    },
+    {
+      "epoch": 0.7462352863812718,
+      "loss": 0.053436558693647385,
+      "loss_ce": 0.0022585804108530283,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 487264416,
+      "step": 2837
+    },
+    {
+      "epoch": 0.7464983231406589,
+      "grad_norm": 5.520845699973876,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 487436620,
+      "step": 2838
+    },
+    {
+      "epoch": 0.7464983231406589,
+      "loss": 0.16054442524909973,
+      "loss_ce": 0.002860102104023099,
+      "loss_iou": 0.59375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 487436620,
+      "step": 2838
+    },
+    {
+      "epoch": 0.746761359900046,
+      "grad_norm": 3.6663304276460456,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 487608856,
+      "step": 2839
+    },
+    {
+      "epoch": 0.746761359900046,
+      "loss": 0.07470214366912842,
+      "loss_ce": 0.0005749509437009692,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 487608856,
+      "step": 2839
+    },
+    {
+      "epoch": 0.7470243966594332,
+      "grad_norm": 3.816698805705744,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 487779452,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7470243966594332,
+      "loss": 0.08027718961238861,
+      "loss_ce": 0.00013802893226966262,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 487779452,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7472874334188203,
+      "grad_norm": 5.187350725945626,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 487951444,
+      "step": 2841
+    },
+    {
+      "epoch": 0.7472874334188203,
+      "loss": 0.13180628418922424,
+      "loss_ce": 0.0002907742455136031,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 487951444,
+      "step": 2841
+    },
+    {
+      "epoch": 0.7475504701782074,
+      "grad_norm": 17.82443490036777,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 488123816,
+      "step": 2842
+    },
+    {
+      "epoch": 0.7475504701782074,
+      "loss": 0.10174276679754257,
+      "loss_ce": 0.00042440436664037406,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 488123816,
+      "step": 2842
+    },
+    {
+      "epoch": 0.7478135069375945,
+      "grad_norm": 5.349643556816576,
+      "learning_rate": 5e-06,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 488294544,
+      "step": 2843
+    },
+    {
+      "epoch": 0.7478135069375945,
+      "loss": 0.06341783702373505,
+      "loss_ce": 6.33449453744106e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 488294544,
+      "step": 2843
+    },
+    {
+      "epoch": 0.7480765436969816,
+      "grad_norm": 5.881783273596092,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 488466512,
+      "step": 2844
+    },
+    {
+      "epoch": 0.7480765436969816,
+      "loss": 0.07953569293022156,
+      "loss_ce": 0.0013496556784957647,
+      "loss_iou": 0.515625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 488466512,
+      "step": 2844
+    },
+    {
+      "epoch": 0.7483395804563687,
+      "grad_norm": 4.967245481323238,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 488638580,
+      "step": 2845
+    },
+    {
+      "epoch": 0.7483395804563687,
+      "loss": 0.13291889429092407,
+      "loss_ce": 0.0028682297561317682,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 488638580,
+      "step": 2845
+    },
+    {
+      "epoch": 0.7486026172157559,
+      "grad_norm": 26.635770395926077,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 488810652,
+      "step": 2846
+    },
+    {
+      "epoch": 0.7486026172157559,
+      "loss": 0.13681207597255707,
+      "loss_ce": 0.006761421915143728,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 488810652,
+      "step": 2846
+    },
+    {
+      "epoch": 0.748865653975143,
+      "grad_norm": 3.755973604918146,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 488982904,
+      "step": 2847
+    },
+    {
+      "epoch": 0.748865653975143,
+      "loss": 0.08354687690734863,
+      "loss_ce": 0.0004475130117498338,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 488982904,
+      "step": 2847
+    },
+    {
+      "epoch": 0.7491286907345301,
+      "grad_norm": 3.8356014362129516,
+      "learning_rate": 5e-06,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 489153060,
+      "step": 2848
+    },
+    {
+      "epoch": 0.7491286907345301,
+      "loss": 0.1231696754693985,
+      "loss_ce": 0.0056769950315356255,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 489153060,
+      "step": 2848
+    },
+    {
+      "epoch": 0.7493917274939172,
+      "grad_norm": 6.542346686389534,
+      "learning_rate": 5e-06,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 489324920,
+      "step": 2849
+    },
+    {
+      "epoch": 0.7493917274939172,
+      "loss": 0.10556487739086151,
+      "loss_ce": 0.00360565772280097,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 489324920,
+      "step": 2849
+    },
+    {
+      "epoch": 0.7496547642533044,
+      "grad_norm": 3.9950005088023786,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 489497376,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7496547642533044,
+      "loss": 0.11913494765758514,
+      "loss_ce": 0.001535467803478241,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 489497376,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7499178010126916,
+      "grad_norm": 3.9324452711240707,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 489669656,
+      "step": 2851
+    },
+    {
+      "epoch": 0.7499178010126916,
+      "loss": 0.14491230249404907,
+      "loss_ce": 0.0019679656252264977,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 489669656,
+      "step": 2851
+    },
+    {
+      "epoch": 0.7501808377720787,
+      "grad_norm": 4.883057174767028,
+      "learning_rate": 5e-06,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 489841572,
+      "step": 2852
+    },
+    {
+      "epoch": 0.7501808377720787,
+      "loss": 0.06473391503095627,
+      "loss_ce": 0.000845368776936084,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 489841572,
+      "step": 2852
+    },
+    {
+      "epoch": 0.7504438745314658,
+      "grad_norm": 5.5775463930736695,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 490010444,
+      "step": 2853
+    },
+    {
+      "epoch": 0.7504438745314658,
+      "loss": 0.2125670313835144,
+      "loss_ce": 0.0003172852157149464,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 490010444,
+      "step": 2853
+    },
+    {
+      "epoch": 0.7507069112908529,
+      "grad_norm": 12.138601127345398,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 490182752,
+      "step": 2854
+    },
+    {
+      "epoch": 0.7507069112908529,
+      "loss": 0.05896022543311119,
+      "loss_ce": 0.0001223331200890243,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 490182752,
+      "step": 2854
+    },
+    {
+      "epoch": 0.75096994805024,
+      "grad_norm": 7.597618581630176,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 490354572,
+      "step": 2855
+    },
+    {
+      "epoch": 0.75096994805024,
+      "loss": 0.1745096743106842,
+      "loss_ce": 0.0004984359256923199,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 490354572,
+      "step": 2855
+    },
+    {
+      "epoch": 0.7512329848096272,
+      "grad_norm": 4.639912180988158,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 490524680,
+      "step": 2856
+    },
+    {
+      "epoch": 0.7512329848096272,
+      "loss": 0.16067692637443542,
+      "loss_ce": 0.006074874196201563,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 490524680,
+      "step": 2856
+    },
+    {
+      "epoch": 0.7514960215690143,
+      "grad_norm": 14.921784803655473,
+      "learning_rate": 5e-06,
+      "loss": 0.1355,
+      "num_input_tokens_seen": 490696768,
+      "step": 2857
+    },
+    {
+      "epoch": 0.7514960215690143,
+      "loss": 0.1124132052063942,
+      "loss_ce": 0.0032213088124990463,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 490696768,
+      "step": 2857
+    },
+    {
+      "epoch": 0.7517590583284014,
+      "grad_norm": 5.608340999176847,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 490869040,
+      "step": 2858
+    },
+    {
+      "epoch": 0.7517590583284014,
+      "loss": 0.05557282269001007,
+      "loss_ce": 0.003662426257506013,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 490869040,
+      "step": 2858
+    },
+    {
+      "epoch": 0.7520220950877885,
+      "grad_norm": 4.649441181141725,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 491041492,
+      "step": 2859
+    },
+    {
+      "epoch": 0.7520220950877885,
+      "loss": 0.08483953773975372,
+      "loss_ce": 0.00024481338914483786,
+      "loss_iou": 0.625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 491041492,
+      "step": 2859
+    },
+    {
+      "epoch": 0.7522851318471756,
+      "grad_norm": 5.629942269274102,
+      "learning_rate": 5e-06,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 491213328,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7522851318471756,
+      "loss": 0.06876988708972931,
+      "loss_ce": 0.0007156922947615385,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 491213328,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7525481686065628,
+      "grad_norm": 5.7661162016412355,
+      "learning_rate": 5e-06,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 491384112,
+      "step": 2861
+    },
+    {
+      "epoch": 0.7525481686065628,
+      "loss": 0.12756821513175964,
+      "loss_ce": 0.0015000998973846436,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 491384112,
+      "step": 2861
+    },
+    {
+      "epoch": 0.7528112053659499,
+      "grad_norm": 4.585306018678188,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 491556404,
+      "step": 2862
+    },
+    {
+      "epoch": 0.7528112053659499,
+      "loss": 0.1506051868200302,
+      "loss_ce": 0.00195405725389719,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 491556404,
+      "step": 2862
+    },
+    {
+      "epoch": 0.753074242125337,
+      "grad_norm": 5.018961759147868,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 491728560,
+      "step": 2863
+    },
+    {
+      "epoch": 0.753074242125337,
+      "loss": 0.08526686578989029,
+      "loss_ce": 0.0047004627995193005,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 491728560,
+      "step": 2863
+    },
+    {
+      "epoch": 0.7533372788847241,
+      "grad_norm": 5.649735345560609,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 491900292,
+      "step": 2864
+    },
+    {
+      "epoch": 0.7533372788847241,
+      "loss": 0.07598397135734558,
+      "loss_ce": 0.0024213448632508516,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 491900292,
+      "step": 2864
+    },
+    {
+      "epoch": 0.7536003156441112,
+      "grad_norm": 5.741199989182661,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 492072492,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7536003156441112,
+      "loss": 0.17299330234527588,
+      "loss_ce": 0.00398694584146142,
+      "loss_iou": 0.296875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 492072492,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7538633524034984,
+      "grad_norm": 6.545995942105509,
+      "learning_rate": 5e-06,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 492244560,
+      "step": 2866
+    },
+    {
+      "epoch": 0.7538633524034984,
+      "loss": 0.11732736229896545,
+      "loss_ce": 0.00035348787787370384,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 492244560,
+      "step": 2866
+    },
+    {
+      "epoch": 0.7541263891628855,
+      "grad_norm": 7.690174312653901,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 492416460,
+      "step": 2867
+    },
+    {
+      "epoch": 0.7541263891628855,
+      "loss": 0.11988115310668945,
+      "loss_ce": 0.0037617662455886602,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 492416460,
+      "step": 2867
+    },
+    {
+      "epoch": 0.7543894259222726,
+      "grad_norm": 4.911561927440868,
+      "learning_rate": 5e-06,
+      "loss": 0.1762,
+      "num_input_tokens_seen": 492582560,
+      "step": 2868
+    },
+    {
+      "epoch": 0.7543894259222726,
+      "loss": 0.17355158925056458,
+      "loss_ce": 0.0006237310590222478,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 492582560,
+      "step": 2868
+    },
+    {
+      "epoch": 0.7546524626816598,
+      "grad_norm": 9.965344498896014,
+      "learning_rate": 5e-06,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 492754468,
+      "step": 2869
+    },
+    {
+      "epoch": 0.7546524626816598,
+      "loss": 0.12515220046043396,
+      "loss_ce": 0.0038448250852525234,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 492754468,
+      "step": 2869
+    },
+    {
+      "epoch": 0.7549154994410469,
+      "grad_norm": 7.897878518361561,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 492926916,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7549154994410469,
+      "loss": 0.07539217174053192,
+      "loss_ce": 0.002699300181120634,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 492926916,
+      "step": 2870
+    },
+    {
+      "epoch": 0.755178536200434,
+      "grad_norm": 8.119348564456503,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 493099340,
+      "step": 2871
+    },
+    {
+      "epoch": 0.755178536200434,
+      "loss": 0.10948731005191803,
+      "loss_ce": 0.002187497215345502,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 493099340,
+      "step": 2871
+    },
+    {
+      "epoch": 0.7554415729598212,
+      "grad_norm": 4.655104945227204,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 493269864,
+      "step": 2872
+    },
+    {
+      "epoch": 0.7554415729598212,
+      "loss": 0.1148996502161026,
+      "loss_ce": 9.252215386368334e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 493269864,
+      "step": 2872
+    },
+    {
+      "epoch": 0.7557046097192083,
+      "grad_norm": 7.2487773846350985,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 493441844,
+      "step": 2873
+    },
+    {
+      "epoch": 0.7557046097192083,
+      "loss": 0.22252312302589417,
+      "loss_ce": 0.00032462860690429807,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 493441844,
+      "step": 2873
+    },
+    {
+      "epoch": 0.7559676464785954,
+      "grad_norm": 8.669322747332805,
+      "learning_rate": 5e-06,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 493612416,
+      "step": 2874
+    },
+    {
+      "epoch": 0.7559676464785954,
+      "loss": 0.1249040961265564,
+      "loss_ce": 0.0034746630117297173,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 493612416,
+      "step": 2874
+    },
+    {
+      "epoch": 0.7562306832379825,
+      "grad_norm": 16.40313430717725,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 493782864,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7562306832379825,
+      "loss": 0.09373937547206879,
+      "loss_ce": 0.0021256012842059135,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 493782864,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7564937199973696,
+      "grad_norm": 6.017272387510744,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 493955332,
+      "step": 2876
+    },
+    {
+      "epoch": 0.7564937199973696,
+      "loss": 0.1287456452846527,
+      "loss_ce": 0.0006633760058321059,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 493955332,
+      "step": 2876
+    },
+    {
+      "epoch": 0.7567567567567568,
+      "grad_norm": 5.38013008566038,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 494127604,
+      "step": 2877
+    },
+    {
+      "epoch": 0.7567567567567568,
+      "loss": 0.10688350349664688,
+      "loss_ce": 0.0010485434904694557,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 494127604,
+      "step": 2877
+    },
+    {
+      "epoch": 0.7570197935161439,
+      "grad_norm": 16.650882691357587,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 494298004,
+      "step": 2878
+    },
+    {
+      "epoch": 0.7570197935161439,
+      "loss": 0.1486469805240631,
+      "loss_ce": 0.0029255489353090525,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 494298004,
+      "step": 2878
+    },
+    {
+      "epoch": 0.757282830275531,
+      "grad_norm": 6.750510186398287,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 494470108,
+      "step": 2879
+    },
+    {
+      "epoch": 0.757282830275531,
+      "loss": 0.1393243670463562,
+      "loss_ce": 0.0016443128697574139,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 494470108,
+      "step": 2879
+    },
+    {
+      "epoch": 0.7575458670349181,
+      "grad_norm": 18.523627538074937,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 494640736,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7575458670349181,
+      "loss": 0.11819358170032501,
+      "loss_ce": 0.0007009088294580579,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 494640736,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7578089037943052,
+      "grad_norm": 6.843824113206992,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 494812784,
+      "step": 2881
+    },
+    {
+      "epoch": 0.7578089037943052,
+      "loss": 0.18498176336288452,
+      "loss_ce": 0.001769477385096252,
+      "loss_iou": 0.390625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 494812784,
+      "step": 2881
+    },
+    {
+      "epoch": 0.7580719405536924,
+      "grad_norm": 4.288128333077997,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 494984900,
+      "step": 2882
+    },
+    {
+      "epoch": 0.7580719405536924,
+      "loss": 0.11615432053804398,
+      "loss_ce": 0.0020796118769794703,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 494984900,
+      "step": 2882
+    },
+    {
+      "epoch": 0.7583349773130795,
+      "grad_norm": 7.805094359622811,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 495157204,
+      "step": 2883
+    },
+    {
+      "epoch": 0.7583349773130795,
+      "loss": 0.15978480875492096,
+      "loss_ce": 0.002222547074779868,
+      "loss_iou": 0.578125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 495157204,
+      "step": 2883
+    },
+    {
+      "epoch": 0.7585980140724666,
+      "grad_norm": 9.068753339594036,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 495325872,
+      "step": 2884
+    },
+    {
+      "epoch": 0.7585980140724666,
+      "loss": 0.06683582067489624,
+      "loss_ce": 0.001055177883245051,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 495325872,
+      "step": 2884
+    },
+    {
+      "epoch": 0.7588610508318537,
+      "grad_norm": 5.303316590010937,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 495497944,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7588610508318537,
+      "loss": 0.09937077760696411,
+      "loss_ce": 0.00031071933335624635,
+      "loss_iou": 0.5625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 495497944,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7591240875912408,
+      "grad_norm": 5.954668834746019,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 495670336,
+      "step": 2886
+    },
+    {
+      "epoch": 0.7591240875912408,
+      "loss": 0.04735005646944046,
+      "loss_ce": 0.0002461753028910607,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 495670336,
+      "step": 2886
+    },
+    {
+      "epoch": 0.759387124350628,
+      "grad_norm": 9.86702776219205,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 495842556,
+      "step": 2887
+    },
+    {
+      "epoch": 0.759387124350628,
+      "loss": 0.09923535585403442,
+      "loss_ce": 0.0010908262338489294,
+      "loss_iou": NaN,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 495842556,
+      "step": 2887
+    },
+    {
+      "epoch": 0.7596501611100152,
+      "grad_norm": 11.076727887110565,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 496012000,
+      "step": 2888
+    },
+    {
+      "epoch": 0.7596501611100152,
+      "loss": 0.07854500412940979,
+      "loss_ce": 0.0005115569802001119,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 496012000,
+      "step": 2888
+    },
+    {
+      "epoch": 0.7599131978694023,
+      "grad_norm": 4.344396012390115,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 496181744,
+      "step": 2889
+    },
+    {
+      "epoch": 0.7599131978694023,
+      "loss": 0.07574538886547089,
+      "loss_ce": 0.0021980288438498974,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 496181744,
+      "step": 2889
+    },
+    {
+      "epoch": 0.7601762346287894,
+      "grad_norm": 4.8762260883901964,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 496353892,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7601762346287894,
+      "loss": 0.1910967230796814,
+      "loss_ce": 0.0009111673571169376,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 496353892,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7604392713881765,
+      "grad_norm": 31.258153024005672,
+      "learning_rate": 5e-06,
+      "loss": 0.1188,
+      "num_input_tokens_seen": 496525916,
+      "step": 2891
+    },
+    {
+      "epoch": 0.7604392713881765,
+      "loss": 0.07003885507583618,
+      "loss_ce": 0.0008249912643805146,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 496525916,
+      "step": 2891
+    },
+    {
+      "epoch": 0.7607023081475637,
+      "grad_norm": 3.99666824293353,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 496698052,
+      "step": 2892
+    },
+    {
+      "epoch": 0.7607023081475637,
+      "loss": 0.09317123889923096,
+      "loss_ce": 0.0031443799380213022,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 496698052,
+      "step": 2892
+    },
+    {
+      "epoch": 0.7609653449069508,
+      "grad_norm": 3.917894736012383,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 496868580,
+      "step": 2893
+    },
+    {
+      "epoch": 0.7609653449069508,
+      "loss": 0.1594741940498352,
+      "loss_ce": 0.0014236548449844122,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 496868580,
+      "step": 2893
+    },
+    {
+      "epoch": 0.7612283816663379,
+      "grad_norm": 6.805778622837066,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 497040720,
+      "step": 2894
+    },
+    {
+      "epoch": 0.7612283816663379,
+      "loss": 0.04234718531370163,
+      "loss_ce": 0.0003244808176532388,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 497040720,
+      "step": 2894
+    },
+    {
+      "epoch": 0.761491418425725,
+      "grad_norm": 8.055134529886692,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 497213204,
+      "step": 2895
+    },
+    {
+      "epoch": 0.761491418425725,
+      "loss": 0.06772617995738983,
+      "loss_ce": 0.0006180237978696823,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 497213204,
+      "step": 2895
+    },
+    {
+      "epoch": 0.7617544551851121,
+      "grad_norm": 20.74511641540883,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 497382348,
+      "step": 2896
+    },
+    {
+      "epoch": 0.7617544551851121,
+      "loss": 0.1839737594127655,
+      "loss_ce": 0.006239374168217182,
+      "loss_iou": 0.5,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 497382348,
+      "step": 2896
+    },
+    {
+      "epoch": 0.7620174919444992,
+      "grad_norm": 3.827051566364027,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 497554288,
+      "step": 2897
+    },
+    {
+      "epoch": 0.7620174919444992,
+      "loss": 0.1408492773771286,
+      "loss_ce": 0.0003158297040499747,
+      "loss_iou": 0.59375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 497554288,
+      "step": 2897
+    },
+    {
+      "epoch": 0.7622805287038864,
+      "grad_norm": 4.251155438400414,
+      "learning_rate": 5e-06,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 497726624,
+      "step": 2898
+    },
+    {
+      "epoch": 0.7622805287038864,
+      "loss": 0.08204406499862671,
+      "loss_ce": 0.0024694851599633694,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 497726624,
+      "step": 2898
+    },
+    {
+      "epoch": 0.7625435654632735,
+      "grad_norm": 25.921624546006864,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 497899052,
+      "step": 2899
+    },
+    {
+      "epoch": 0.7625435654632735,
+      "loss": 0.125847727060318,
+      "loss_ce": 5.42689704161603e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 497899052,
+      "step": 2899
+    },
+    {
+      "epoch": 0.7628066022226606,
+      "grad_norm": 3.225720909448473,
+      "learning_rate": 5e-06,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 498071164,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7628066022226606,
+      "loss": 0.13742178678512573,
+      "loss_ce": 0.002015294972807169,
+      "loss_iou": 0.5625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 498071164,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7630696389820477,
+      "grad_norm": 10.436595772958983,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 498243380,
+      "step": 2901
+    },
+    {
+      "epoch": 0.7630696389820477,
+      "loss": 0.08406674116849899,
+      "loss_ce": 0.0019744576420634985,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 498243380,
+      "step": 2901
+    },
+    {
+      "epoch": 0.7633326757414348,
+      "grad_norm": 5.876814424855767,
+      "learning_rate": 5e-06,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 498415740,
+      "step": 2902
+    },
+    {
+      "epoch": 0.7633326757414348,
+      "loss": 0.12014832347631454,
+      "loss_ce": 0.0023199557326734066,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 498415740,
+      "step": 2902
+    },
+    {
+      "epoch": 0.763595712500822,
+      "grad_norm": 6.217199501402136,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 498587740,
+      "step": 2903
+    },
+    {
+      "epoch": 0.763595712500822,
+      "loss": 0.0937931090593338,
+      "loss_ce": 0.0013553638709709048,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 498587740,
+      "step": 2903
+    },
+    {
+      "epoch": 0.7638587492602091,
+      "grad_norm": 4.965193192238494,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 498759856,
+      "step": 2904
+    },
+    {
+      "epoch": 0.7638587492602091,
+      "loss": 0.0735275000333786,
+      "loss_ce": 0.00013272266369313002,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 498759856,
+      "step": 2904
+    },
+    {
+      "epoch": 0.7641217860195962,
+      "grad_norm": 36.02324764024362,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 498929524,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7641217860195962,
+      "loss": 0.13745470345020294,
+      "loss_ce": 0.002353382296860218,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 498929524,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7643848227789833,
+      "grad_norm": 4.064476858285287,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 499099092,
+      "step": 2906
+    },
+    {
+      "epoch": 0.7643848227789833,
+      "loss": 0.12520155310630798,
+      "loss_ce": 0.0020325970835983753,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 499099092,
+      "step": 2906
+    },
+    {
+      "epoch": 0.7646478595383704,
+      "grad_norm": 9.564182625696946,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 499271132,
+      "step": 2907
+    },
+    {
+      "epoch": 0.7646478595383704,
+      "loss": 0.10383596271276474,
+      "loss_ce": 0.0036772743333131075,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 499271132,
+      "step": 2907
+    },
+    {
+      "epoch": 0.7649108962977577,
+      "grad_norm": 7.38548887746826,
+      "learning_rate": 5e-06,
+      "loss": 0.1407,
+      "num_input_tokens_seen": 499443276,
+      "step": 2908
+    },
+    {
+      "epoch": 0.7649108962977577,
+      "loss": 0.07420553267002106,
+      "loss_ce": 0.006151327397674322,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 499443276,
+      "step": 2908
+    },
+    {
+      "epoch": 0.7651739330571448,
+      "grad_norm": 3.357102180261463,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 499615436,
+      "step": 2909
+    },
+    {
+      "epoch": 0.7651739330571448,
+      "loss": 0.13687880337238312,
+      "loss_ce": 0.0005110005149617791,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 499615436,
+      "step": 2909
+    },
+    {
+      "epoch": 0.7654369698165319,
+      "grad_norm": 5.532042225189138,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 499787568,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7654369698165319,
+      "loss": 0.10780411958694458,
+      "loss_ce": 0.0017250193050131202,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 499787568,
+      "step": 2910
+    },
+    {
+      "epoch": 0.765700006575919,
+      "grad_norm": 9.098827733621626,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 499959744,
+      "step": 2911
+    },
+    {
+      "epoch": 0.765700006575919,
+      "loss": 0.11246542632579803,
+      "loss_ce": 0.0012135956203565001,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 499959744,
+      "step": 2911
+    },
+    {
+      "epoch": 0.7659630433353061,
+      "grad_norm": 4.671256681679922,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 500132012,
+      "step": 2912
+    },
+    {
+      "epoch": 0.7659630433353061,
+      "loss": 0.07265074551105499,
+      "loss_ce": 0.003986193798482418,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 500132012,
+      "step": 2912
+    },
+    {
+      "epoch": 0.7662260800946933,
+      "grad_norm": 9.318330323174848,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 500304120,
+      "step": 2913
+    },
+    {
+      "epoch": 0.7662260800946933,
+      "loss": 0.10342703014612198,
+      "loss_ce": 0.0003996905288659036,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 500304120,
+      "step": 2913
+    },
+    {
+      "epoch": 0.7664891168540804,
+      "grad_norm": 4.11205791514663,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 500476220,
+      "step": 2914
+    },
+    {
+      "epoch": 0.7664891168540804,
+      "loss": 0.16668415069580078,
+      "loss_ce": 0.00170612963847816,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 500476220,
+      "step": 2914
+    },
+    {
+      "epoch": 0.7667521536134675,
+      "grad_norm": 7.921166949882202,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 500648456,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7667521536134675,
+      "loss": 0.14688915014266968,
+      "loss_ce": 0.0016559937503188848,
+      "loss_iou": 0.546875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 500648456,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7670151903728546,
+      "grad_norm": 7.89559416185462,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 500817440,
+      "step": 2916
+    },
+    {
+      "epoch": 0.7670151903728546,
+      "loss": 0.13112960755825043,
+      "loss_ce": 0.0047868345864117146,
+      "loss_iou": 0.5,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 500817440,
+      "step": 2916
+    },
+    {
+      "epoch": 0.7672782271322417,
+      "grad_norm": 7.07373988410314,
+      "learning_rate": 5e-06,
+      "loss": 0.1543,
+      "num_input_tokens_seen": 500989576,
+      "step": 2917
+    },
+    {
+      "epoch": 0.7672782271322417,
+      "loss": 0.17218878865242004,
+      "loss_ce": 0.0034876265563070774,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 500989576,
+      "step": 2917
+    },
+    {
+      "epoch": 0.7675412638916288,
+      "grad_norm": 4.658693255259618,
+      "learning_rate": 5e-06,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 501161844,
+      "step": 2918
+    },
+    {
+      "epoch": 0.7675412638916288,
+      "loss": 0.0911635234951973,
+      "loss_ce": 0.0017775364685803652,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 501161844,
+      "step": 2918
+    },
+    {
+      "epoch": 0.767804300651016,
+      "grad_norm": 5.6482830730287485,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 501334012,
+      "step": 2919
+    },
+    {
+      "epoch": 0.767804300651016,
+      "loss": 0.12131966650485992,
+      "loss_ce": 0.00028695265064015985,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 501334012,
+      "step": 2919
+    },
+    {
+      "epoch": 0.7680673374104031,
+      "grad_norm": 6.055907471900818,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 501504632,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7680673374104031,
+      "loss": 0.06848346441984177,
+      "loss_ce": 0.0003071928513236344,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 501504632,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7683303741697902,
+      "grad_norm": 4.128716866112718,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 501676892,
+      "step": 2921
+    },
+    {
+      "epoch": 0.7683303741697902,
+      "loss": 0.12142758071422577,
+      "loss_ce": 0.000974692520685494,
+      "loss_iou": 0.5,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 501676892,
+      "step": 2921
+    },
+    {
+      "epoch": 0.7685934109291773,
+      "grad_norm": 6.285708685708194,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 501846408,
+      "step": 2922
+    },
+    {
+      "epoch": 0.7685934109291773,
+      "loss": 0.13839051127433777,
+      "loss_ce": 0.0014276191359385848,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 501846408,
+      "step": 2922
+    },
+    {
+      "epoch": 0.7688564476885644,
+      "grad_norm": 5.199523762158551,
+      "learning_rate": 5e-06,
+      "loss": 0.0935,
+      "num_input_tokens_seen": 502018884,
+      "step": 2923
+    },
+    {
+      "epoch": 0.7688564476885644,
+      "loss": 0.0823674127459526,
+      "loss_ce": 0.0034794718958437443,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 502018884,
+      "step": 2923
+    },
+    {
+      "epoch": 0.7691194844479516,
+      "grad_norm": 7.865534449035968,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 502189840,
+      "step": 2924
+    },
+    {
+      "epoch": 0.7691194844479516,
+      "loss": 0.09236955642700195,
+      "loss_ce": 0.0008473452762700617,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 502189840,
+      "step": 2924
+    },
+    {
+      "epoch": 0.7693825212073387,
+      "grad_norm": 4.464952808674884,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 502361908,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7693825212073387,
+      "loss": 0.09519391506910324,
+      "loss_ce": 0.002862985013052821,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 502361908,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7696455579667258,
+      "grad_norm": 3.137863532599948,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 502533852,
+      "step": 2926
+    },
+    {
+      "epoch": 0.7696455579667258,
+      "loss": 0.062405504286289215,
+      "loss_ce": 0.000225941272219643,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 502533852,
+      "step": 2926
+    },
+    {
+      "epoch": 0.769908594726113,
+      "grad_norm": 5.424523366119041,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 502706068,
+      "step": 2927
+    },
+    {
+      "epoch": 0.769908594726113,
+      "loss": 0.15939806401729584,
+      "loss_ce": 0.0009050165535882115,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 502706068,
+      "step": 2927
+    },
+    {
+      "epoch": 0.7701716314855,
+      "grad_norm": 21.020644002444403,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 502878204,
+      "step": 2928
+    },
+    {
+      "epoch": 0.7701716314855,
+      "loss": 0.1498870849609375,
+      "loss_ce": 0.00108338613063097,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 502878204,
+      "step": 2928
+    },
+    {
+      "epoch": 0.7704346682448873,
+      "grad_norm": 6.513721562381314,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 503050516,
+      "step": 2929
+    },
+    {
+      "epoch": 0.7704346682448873,
+      "loss": 0.14237374067306519,
+      "loss_ce": 0.0013520093634724617,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 503050516,
+      "step": 2929
+    },
+    {
+      "epoch": 0.7706977050042744,
+      "grad_norm": 21.68669265107994,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 503222664,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7706977050042744,
+      "loss": 0.09872519969940186,
+      "loss_ce": 0.0024269861169159412,
+      "loss_iou": 0.53125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 503222664,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7709607417636615,
+      "grad_norm": 23.71932442538902,
+      "learning_rate": 5e-06,
+      "loss": 0.1131,
+      "num_input_tokens_seen": 503394820,
+      "step": 2931
+    },
+    {
+      "epoch": 0.7709607417636615,
+      "loss": 0.14681634306907654,
+      "loss_ce": 0.0005455805221572518,
+      "loss_iou": 0.359375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 503394820,
+      "step": 2931
+    },
+    {
+      "epoch": 0.7712237785230486,
+      "grad_norm": 14.0356485916023,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 503567064,
+      "step": 2932
+    },
+    {
+      "epoch": 0.7712237785230486,
+      "loss": 0.1981443464756012,
+      "loss_ce": 0.002374083735048771,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 503567064,
+      "step": 2932
+    },
+    {
+      "epoch": 0.7714868152824357,
+      "grad_norm": 4.952386040850401,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 503739480,
+      "step": 2933
+    },
+    {
+      "epoch": 0.7714868152824357,
+      "loss": 0.2314581573009491,
+      "loss_ce": 0.002728914376348257,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 503739480,
+      "step": 2933
+    },
+    {
+      "epoch": 0.7717498520418229,
+      "grad_norm": 5.454737692407838,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 503911536,
+      "step": 2934
+    },
+    {
+      "epoch": 0.7717498520418229,
+      "loss": 0.1048421710729599,
+      "loss_ce": 0.004591919481754303,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 503911536,
+      "step": 2934
+    },
+    {
+      "epoch": 0.77201288880121,
+      "grad_norm": 7.9773349981616795,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 504082084,
+      "step": 2935
+    },
+    {
+      "epoch": 0.77201288880121,
+      "loss": 0.1008715033531189,
+      "loss_ce": 0.0018419669941067696,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 504082084,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7722759255605971,
+      "grad_norm": 5.256239273173906,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 504254328,
+      "step": 2936
+    },
+    {
+      "epoch": 0.7722759255605971,
+      "loss": 0.10146059095859528,
+      "loss_ce": 0.0031024364288896322,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 504254328,
+      "step": 2936
+    },
+    {
+      "epoch": 0.7725389623199842,
+      "grad_norm": 7.796343823545596,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 504426408,
+      "step": 2937
+    },
+    {
+      "epoch": 0.7725389623199842,
+      "loss": 0.09895452111959457,
+      "loss_ce": 0.00047429182450287044,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 504426408,
+      "step": 2937
+    },
+    {
+      "epoch": 0.7728019990793713,
+      "grad_norm": 4.506271114804364,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 504599160,
+      "step": 2938
+    },
+    {
+      "epoch": 0.7728019990793713,
+      "loss": 0.10439816117286682,
+      "loss_ce": 0.0011114203371107578,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 504599160,
+      "step": 2938
+    },
+    {
+      "epoch": 0.7730650358387585,
+      "grad_norm": 7.051793538200423,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 504771620,
+      "step": 2939
+    },
+    {
+      "epoch": 0.7730650358387585,
+      "loss": 0.04662296548485756,
+      "loss_ce": 0.001548499334603548,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 504771620,
+      "step": 2939
+    },
+    {
+      "epoch": 0.7733280725981456,
+      "grad_norm": 4.48321731819358,
+      "learning_rate": 5e-06,
+      "loss": 0.1049,
+      "num_input_tokens_seen": 504942212,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7733280725981456,
+      "loss": 0.1206519603729248,
+      "loss_ce": 0.0022132450249046087,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 504942212,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7735911093575327,
+      "grad_norm": 3.542637672706516,
+      "learning_rate": 5e-06,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 505114396,
+      "step": 2941
+    },
+    {
+      "epoch": 0.7735911093575327,
+      "loss": 0.07642598450183868,
+      "loss_ce": 0.002970176050439477,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 505114396,
+      "step": 2941
+    },
+    {
+      "epoch": 0.7738541461169198,
+      "grad_norm": 13.31557941839502,
+      "learning_rate": 5e-06,
+      "loss": 0.152,
+      "num_input_tokens_seen": 505286572,
+      "step": 2942
+    },
+    {
+      "epoch": 0.7738541461169198,
+      "loss": 0.11694711446762085,
+      "loss_ce": 0.000492042163386941,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 505286572,
+      "step": 2942
+    },
+    {
+      "epoch": 0.7741171828763069,
+      "grad_norm": 4.217985208788173,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 505458580,
+      "step": 2943
+    },
+    {
+      "epoch": 0.7741171828763069,
+      "loss": 0.12103226035833359,
+      "loss_ce": 0.00256301905028522,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 505458580,
+      "step": 2943
+    },
+    {
+      "epoch": 0.774380219635694,
+      "grad_norm": 5.305953482185517,
+      "learning_rate": 5e-06,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 505630628,
+      "step": 2944
+    },
+    {
+      "epoch": 0.774380219635694,
+      "loss": 0.13166974484920502,
+      "loss_ce": 0.0009019209537655115,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 505630628,
+      "step": 2944
+    },
+    {
+      "epoch": 0.7746432563950812,
+      "grad_norm": 3.7265892335604103,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 505802760,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7746432563950812,
+      "loss": 0.174719899892807,
+      "loss_ce": 0.005622013006359339,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 505802760,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7749062931544684,
+      "grad_norm": 3.906958326198367,
+      "learning_rate": 5e-06,
+      "loss": 0.0915,
+      "num_input_tokens_seen": 505975008,
+      "step": 2946
+    },
+    {
+      "epoch": 0.7749062931544684,
+      "loss": 0.15144148468971252,
+      "loss_ce": 0.003522793762385845,
+      "loss_iou": 0.609375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 505975008,
+      "step": 2946
+    },
+    {
+      "epoch": 0.7751693299138555,
+      "grad_norm": 30.273164796131727,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 506146864,
+      "step": 2947
+    },
+    {
+      "epoch": 0.7751693299138555,
+      "loss": 0.12878431379795074,
+      "loss_ce": 0.00238050683401525,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 506146864,
+      "step": 2947
+    },
+    {
+      "epoch": 0.7754323666732426,
+      "grad_norm": 9.326627239945015,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 506317520,
+      "step": 2948
+    },
+    {
+      "epoch": 0.7754323666732426,
+      "loss": 0.13574595749378204,
+      "loss_ce": 0.001163432258181274,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 506317520,
+      "step": 2948
+    },
+    {
+      "epoch": 0.7756954034326297,
+      "grad_norm": 4.291282220054208,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 506489644,
+      "step": 2949
+    },
+    {
+      "epoch": 0.7756954034326297,
+      "loss": 0.07531201094388962,
+      "loss_ce": 0.0006965312641113997,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 506489644,
+      "step": 2949
+    },
+    {
+      "epoch": 0.7759584401920169,
+      "grad_norm": 5.132046879043553,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 506661552,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7759584401920169,
+      "loss": 0.12264753133058548,
+      "loss_ce": 0.003171210875734687,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 506661552,
+      "step": 2950
+    },
+    {
+      "epoch": 0.776221476951404,
+      "grad_norm": 18.605800930237596,
+      "learning_rate": 5e-06,
+      "loss": 0.1066,
+      "num_input_tokens_seen": 506832080,
+      "step": 2951
+    },
+    {
+      "epoch": 0.776221476951404,
+      "loss": 0.08513291925191879,
+      "loss_ce": 0.0006144904182292521,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 506832080,
+      "step": 2951
+    },
+    {
+      "epoch": 0.7764845137107911,
+      "grad_norm": 13.809771584503252,
+      "learning_rate": 5e-06,
+      "loss": 0.148,
+      "num_input_tokens_seen": 507004396,
+      "step": 2952
+    },
+    {
+      "epoch": 0.7764845137107911,
+      "loss": 0.16220027208328247,
+      "loss_ce": 0.007384595461189747,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 507004396,
+      "step": 2952
+    },
+    {
+      "epoch": 0.7767475504701782,
+      "grad_norm": 6.052098852686667,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 507176636,
+      "step": 2953
+    },
+    {
+      "epoch": 0.7767475504701782,
+      "loss": 0.10348343849182129,
+      "loss_ce": 7.461909262929112e-05,
+      "loss_iou": 0.640625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 507176636,
+      "step": 2953
+    },
+    {
+      "epoch": 0.7770105872295653,
+      "grad_norm": 6.115673418463162,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 507348812,
+      "step": 2954
+    },
+    {
+      "epoch": 0.7770105872295653,
+      "loss": 0.17799662053585052,
+      "loss_ce": 0.0009031177032738924,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 507348812,
+      "step": 2954
+    },
+    {
+      "epoch": 0.7772736239889525,
+      "grad_norm": 4.920564715708625,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 507521260,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7772736239889525,
+      "loss": 0.12127618491649628,
+      "loss_ce": 0.002562812063843012,
+      "loss_iou": 0.625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 507521260,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7775366607483396,
+      "grad_norm": 4.1269903282175875,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 507693580,
+      "step": 2956
+    },
+    {
+      "epoch": 0.7775366607483396,
+      "loss": 0.1181151419878006,
+      "loss_ce": 0.0023619618732482195,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 507693580,
+      "step": 2956
+    },
+    {
+      "epoch": 0.7777996975077267,
+      "grad_norm": 6.530908453055091,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 507866000,
+      "step": 2957
+    },
+    {
+      "epoch": 0.7777996975077267,
+      "loss": 0.09897395223379135,
+      "loss_ce": 0.001989088486880064,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 507866000,
+      "step": 2957
+    },
+    {
+      "epoch": 0.7780627342671138,
+      "grad_norm": 13.799197287571243,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 508036520,
+      "step": 2958
+    },
+    {
+      "epoch": 0.7780627342671138,
+      "loss": 0.12385988235473633,
+      "loss_ce": 0.0020031901076436043,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 508036520,
+      "step": 2958
+    },
+    {
+      "epoch": 0.7783257710265009,
+      "grad_norm": 6.771630034008941,
+      "learning_rate": 5e-06,
+      "loss": 0.1293,
+      "num_input_tokens_seen": 508208512,
+      "step": 2959
+    },
+    {
+      "epoch": 0.7783257710265009,
+      "loss": 0.09063053131103516,
+      "loss_ce": 0.001976970350369811,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 508208512,
+      "step": 2959
+    },
+    {
+      "epoch": 0.7785888077858881,
+      "grad_norm": 17.947272506590085,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 508380916,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7785888077858881,
+      "loss": 0.18887397646903992,
+      "loss_ce": 0.000855189049616456,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 508380916,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7788518445452752,
+      "grad_norm": 8.543500304163382,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 508551252,
+      "step": 2961
+    },
+    {
+      "epoch": 0.7788518445452752,
+      "loss": 0.10420133173465729,
+      "loss_ce": 0.000197422195924446,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 508551252,
+      "step": 2961
+    },
+    {
+      "epoch": 0.7791148813046623,
+      "grad_norm": 11.527993211943581,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 508723096,
+      "step": 2962
+    },
+    {
+      "epoch": 0.7791148813046623,
+      "loss": 0.13255223631858826,
+      "loss_ce": 0.0016013117274269462,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 508723096,
+      "step": 2962
+    },
+    {
+      "epoch": 0.7793779180640494,
+      "grad_norm": 3.765008399602552,
+      "learning_rate": 5e-06,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 508895336,
+      "step": 2963
+    },
+    {
+      "epoch": 0.7793779180640494,
+      "loss": 0.15525312721729279,
+      "loss_ce": 0.00026197341503575444,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 508895336,
+      "step": 2963
+    },
+    {
+      "epoch": 0.7796409548234365,
+      "grad_norm": 4.0681392149062905,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 509065748,
+      "step": 2964
+    },
+    {
+      "epoch": 0.7796409548234365,
+      "loss": 0.103541798889637,
+      "loss_ce": 0.0006670450093224645,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 509065748,
+      "step": 2964
+    },
+    {
+      "epoch": 0.7799039915828238,
+      "grad_norm": 5.764259729112096,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 509237988,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7799039915828238,
+      "loss": 0.09293576329946518,
+      "loss_ce": 0.0004980218946002424,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 509237988,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7801670283422109,
+      "grad_norm": 6.653297635090324,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 509407480,
+      "step": 2966
+    },
+    {
+      "epoch": 0.7801670283422109,
+      "loss": 0.10752134025096893,
+      "loss_ce": 0.0008471491746604443,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 509407480,
+      "step": 2966
+    },
+    {
+      "epoch": 0.780430065101598,
+      "grad_norm": 5.8594293697060476,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 509579708,
+      "step": 2967
+    },
+    {
+      "epoch": 0.780430065101598,
+      "loss": 0.19974008202552795,
+      "loss_ce": 0.005846647545695305,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 509579708,
+      "step": 2967
+    },
+    {
+      "epoch": 0.7806931018609851,
+      "grad_norm": 4.074243325830813,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 509750260,
+      "step": 2968
+    },
+    {
+      "epoch": 0.7806931018609851,
+      "loss": 0.09352241456508636,
+      "loss_ce": 0.00045905529987066984,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 509750260,
+      "step": 2968
+    },
+    {
+      "epoch": 0.7809561386203722,
+      "grad_norm": 18.473086839056062,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 509922416,
+      "step": 2969
+    },
+    {
+      "epoch": 0.7809561386203722,
+      "loss": 0.10857398062944412,
+      "loss_ce": 0.002586429938673973,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 509922416,
+      "step": 2969
+    },
+    {
+      "epoch": 0.7812191753797593,
+      "grad_norm": 4.297341068075648,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 510094352,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7812191753797593,
+      "loss": 0.10554330050945282,
+      "loss_ce": 0.002088708570227027,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 510094352,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7814822121391465,
+      "grad_norm": 4.523551186515909,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 510266568,
+      "step": 2971
+    },
+    {
+      "epoch": 0.7814822121391465,
+      "loss": 0.1601797491312027,
+      "loss_ce": 0.000939027639105916,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 510266568,
+      "step": 2971
+    },
+    {
+      "epoch": 0.7817452488985336,
+      "grad_norm": 11.257967210341226,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 510438880,
+      "step": 2972
+    },
+    {
+      "epoch": 0.7817452488985336,
+      "loss": 0.08428631722927094,
+      "loss_ce": 0.0006986761000007391,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 510438880,
+      "step": 2972
+    },
+    {
+      "epoch": 0.7820082856579207,
+      "grad_norm": 4.879983737803821,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 510610888,
+      "step": 2973
+    },
+    {
+      "epoch": 0.7820082856579207,
+      "loss": 0.061606645584106445,
+      "loss_ce": 0.0010292520746588707,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 510610888,
+      "step": 2973
+    },
+    {
+      "epoch": 0.7822713224173078,
+      "grad_norm": 6.784338572567158,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 510780736,
+      "step": 2974
+    },
+    {
+      "epoch": 0.7822713224173078,
+      "loss": 0.09628412127494812,
+      "loss_ce": 0.0025646386202424765,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 510780736,
+      "step": 2974
+    },
+    {
+      "epoch": 0.7825343591766949,
+      "grad_norm": 2.4966016721195827,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 510952740,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7825343591766949,
+      "loss": 0.11784331500530243,
+      "loss_ce": 0.002517384709790349,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 510952740,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7827973959360821,
+      "grad_norm": 9.752032559631301,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 511124984,
+      "step": 2976
+    },
+    {
+      "epoch": 0.7827973959360821,
+      "loss": 0.10057765245437622,
+      "loss_ce": 0.001853286987170577,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 511124984,
+      "step": 2976
+    },
+    {
+      "epoch": 0.7830604326954692,
+      "grad_norm": 4.047284415442861,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 511297176,
+      "step": 2977
+    },
+    {
+      "epoch": 0.7830604326954692,
+      "loss": 0.1439676731824875,
+      "loss_ce": 0.004929587244987488,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 511297176,
+      "step": 2977
+    },
+    {
+      "epoch": 0.7833234694548563,
+      "grad_norm": 4.111909259977495,
+      "learning_rate": 5e-06,
+      "loss": 0.121,
+      "num_input_tokens_seen": 511469216,
+      "step": 2978
+    },
+    {
+      "epoch": 0.7833234694548563,
+      "loss": 0.1689942479133606,
+      "loss_ce": 0.008380233310163021,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 511469216,
+      "step": 2978
+    },
+    {
+      "epoch": 0.7835865062142434,
+      "grad_norm": 3.736811899980652,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 511641400,
+      "step": 2979
+    },
+    {
+      "epoch": 0.7835865062142434,
+      "loss": 0.09928886592388153,
+      "loss_ce": 0.0024871169589459896,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 511641400,
+      "step": 2979
+    },
+    {
+      "epoch": 0.7838495429736305,
+      "grad_norm": 3.8885212513343204,
+      "learning_rate": 5e-06,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 511813644,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7838495429736305,
+      "loss": 0.08123628050088882,
+      "loss_ce": 0.0019821308087557554,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 511813644,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7841125797330177,
+      "grad_norm": 3.6947801269712643,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 511985744,
+      "step": 2981
+    },
+    {
+      "epoch": 0.7841125797330177,
+      "loss": 0.07377283275127411,
+      "loss_ce": 0.0037197312340140343,
+      "loss_iou": 0.71875,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 511985744,
+      "step": 2981
+    },
+    {
+      "epoch": 0.7843756164924048,
+      "grad_norm": 4.052383876786155,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 512157708,
+      "step": 2982
+    },
+    {
+      "epoch": 0.7843756164924048,
+      "loss": 0.18616530299186707,
+      "loss_ce": 0.00116774532943964,
+      "loss_iou": 0.59375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 512157708,
+      "step": 2982
+    },
+    {
+      "epoch": 0.7846386532517919,
+      "grad_norm": 3.4825813224277185,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 512330084,
+      "step": 2983
+    },
+    {
+      "epoch": 0.7846386532517919,
+      "loss": 0.1472734808921814,
+      "loss_ce": 0.0032915552146732807,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 512330084,
+      "step": 2983
+    },
+    {
+      "epoch": 0.784901690011179,
+      "grad_norm": 3.5815743222985947,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 512502076,
+      "step": 2984
+    },
+    {
+      "epoch": 0.784901690011179,
+      "loss": 0.10481996834278107,
+      "loss_ce": 0.0021588318049907684,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 512502076,
+      "step": 2984
+    },
+    {
+      "epoch": 0.7851647267705661,
+      "grad_norm": 7.905852445879054,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 512674288,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7851647267705661,
+      "loss": 0.12316185235977173,
+      "loss_ce": 0.0030751884914934635,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 512674288,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7854277635299534,
+      "grad_norm": 5.926087069922328,
+      "learning_rate": 5e-06,
+      "loss": 0.0863,
+      "num_input_tokens_seen": 512846624,
+      "step": 2986
+    },
+    {
+      "epoch": 0.7854277635299534,
+      "loss": 0.06602786481380463,
+      "loss_ce": 0.0011474882485345006,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 512846624,
+      "step": 2986
+    },
+    {
+      "epoch": 0.7856908002893405,
+      "grad_norm": 5.039950797702919,
+      "learning_rate": 5e-06,
+      "loss": 0.1472,
+      "num_input_tokens_seen": 513018668,
+      "step": 2987
+    },
+    {
+      "epoch": 0.7856908002893405,
+      "loss": 0.10930870473384857,
+      "loss_ce": 0.004847035743296146,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 513018668,
+      "step": 2987
+    },
+    {
+      "epoch": 0.7859538370487276,
+      "grad_norm": 4.823970003989147,
+      "learning_rate": 5e-06,
+      "loss": 0.1425,
+      "num_input_tokens_seen": 513190872,
+      "step": 2988
+    },
+    {
+      "epoch": 0.7859538370487276,
+      "loss": 0.15449807047843933,
+      "loss_ce": 0.0005368961137719452,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 513190872,
+      "step": 2988
+    },
+    {
+      "epoch": 0.7862168738081147,
+      "grad_norm": 11.156095172455718,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 513363216,
+      "step": 2989
+    },
+    {
+      "epoch": 0.7862168738081147,
+      "loss": 0.07052050530910492,
+      "loss_ce": 0.002100098878145218,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 513363216,
+      "step": 2989
+    },
+    {
+      "epoch": 0.7864799105675018,
+      "grad_norm": 6.545014028470699,
+      "learning_rate": 5e-06,
+      "loss": 0.1922,
+      "num_input_tokens_seen": 513532780,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7864799105675018,
+      "loss": 0.20626530051231384,
+      "loss_ce": 0.004162629134953022,
+      "loss_iou": 0.515625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 513532780,
+      "step": 2990
+    },
+    {
+      "epoch": 0.786742947326889,
+      "grad_norm": 5.675250290858523,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 513705312,
+      "step": 2991
+    },
+    {
+      "epoch": 0.786742947326889,
+      "loss": 0.10429085791110992,
+      "loss_ce": 0.002728358842432499,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 513705312,
+      "step": 2991
+    },
+    {
+      "epoch": 0.7870059840862761,
+      "grad_norm": 3.771208751532015,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 513877348,
+      "step": 2992
+    },
+    {
+      "epoch": 0.7870059840862761,
+      "loss": 0.09906225651502609,
+      "loss_ce": 0.0011008285218849778,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 513877348,
+      "step": 2992
+    },
+    {
+      "epoch": 0.7872690208456632,
+      "grad_norm": 8.702146962798412,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 514046516,
+      "step": 2993
+    },
+    {
+      "epoch": 0.7872690208456632,
+      "loss": 0.11204151809215546,
+      "loss_ce": 0.004131356719881296,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 514046516,
+      "step": 2993
+    },
+    {
+      "epoch": 0.7875320576050503,
+      "grad_norm": 6.0868763702009465,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 514218536,
+      "step": 2994
+    },
+    {
+      "epoch": 0.7875320576050503,
+      "loss": 0.15223419666290283,
+      "loss_ce": 0.0040408410131931305,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 514218536,
+      "step": 2994
+    },
+    {
+      "epoch": 0.7877950943644374,
+      "grad_norm": 4.672234975964948,
+      "learning_rate": 5e-06,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 514391068,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7877950943644374,
+      "loss": 0.05085252225399017,
+      "loss_ce": 0.0038707097992300987,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 514391068,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7880581311238245,
+      "grad_norm": 6.661101757033409,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 514561472,
+      "step": 2996
+    },
+    {
+      "epoch": 0.7880581311238245,
+      "loss": 0.20267510414123535,
+      "loss_ce": 0.000404604768846184,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 514561472,
+      "step": 2996
+    },
+    {
+      "epoch": 0.7883211678832117,
+      "grad_norm": 9.694053172636844,
+      "learning_rate": 5e-06,
+      "loss": 0.148,
+      "num_input_tokens_seen": 514733680,
+      "step": 2997
+    },
+    {
+      "epoch": 0.7883211678832117,
+      "loss": 0.13084860146045685,
+      "loss_ce": 0.002278049010783434,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 514733680,
+      "step": 2997
+    },
+    {
+      "epoch": 0.7885842046425988,
+      "grad_norm": 69.63701857559735,
+      "learning_rate": 5e-06,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 514905800,
+      "step": 2998
+    },
+    {
+      "epoch": 0.7885842046425988,
+      "loss": 0.13946260511875153,
+      "loss_ce": 0.004315503872931004,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 514905800,
+      "step": 2998
+    },
+    {
+      "epoch": 0.7888472414019859,
+      "grad_norm": 16.568333279972496,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 515077960,
+      "step": 2999
+    },
+    {
+      "epoch": 0.7888472414019859,
+      "loss": 0.10318569839000702,
+      "loss_ce": 0.004888580180704594,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 515077960,
+      "step": 2999
+    },
+    {
+      "epoch": 0.789110278161373,
+      "grad_norm": 5.577264845510526,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_websight_new_CIoU": 0.8789950311183929,
+      "eval_websight_new_GIoU": 0.881831705570221,
+      "eval_websight_new_IoU": 0.8831256628036499,
+      "eval_websight_new_MAE_all": 0.016948864795267582,
+      "eval_websight_new_MAE_h": 0.00989671004936099,
+      "eval_websight_new_MAE_w": 0.027918956242501736,
+      "eval_websight_new_MAE_x": 0.024959519505500793,
+      "eval_websight_new_MAE_y": 0.005020270356908441,
+      "eval_websight_new_NUM_probability": 0.9999921023845673,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.08870729804039001,
+      "eval_websight_new_loss_ce": 6.795420176786138e-06,
+      "eval_websight_new_loss_iou": 0.3958740234375,
+      "eval_websight_new_loss_num": 0.016002655029296875,
+      "eval_websight_new_loss_xval": 0.0800323486328125,
+      "eval_websight_new_runtime": 55.9259,
+      "eval_websight_new_samples_per_second": 0.894,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_seeclick_CIoU": 0.6246416866779327,
+      "eval_seeclick_GIoU": 0.6264897286891937,
+      "eval_seeclick_IoU": 0.650597870349884,
+      "eval_seeclick_MAE_all": 0.04822420887649059,
+      "eval_seeclick_MAE_h": 0.030418247915804386,
+      "eval_seeclick_MAE_w": 0.0654730387032032,
+      "eval_seeclick_MAE_x": 0.07548732310533524,
+      "eval_seeclick_MAE_y": 0.0215182239189744,
+      "eval_seeclick_NUM_probability": 0.9999599456787109,
+      "eval_seeclick_inside_bbox": 0.9375,
+      "eval_seeclick_loss": 0.2215959131717682,
+      "eval_seeclick_loss_ce": 0.008739723358303308,
+      "eval_seeclick_loss_iou": 0.4912109375,
+      "eval_seeclick_loss_num": 0.043231964111328125,
+      "eval_seeclick_loss_xval": 0.216217041015625,
+      "eval_seeclick_runtime": 91.1896,
+      "eval_seeclick_samples_per_second": 0.472,
+      "eval_seeclick_steps_per_second": 0.022,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_icons_CIoU": 0.8647165596485138,
+      "eval_icons_GIoU": 0.8629220426082611,
+      "eval_icons_IoU": 0.8704103231430054,
+      "eval_icons_MAE_all": 0.018641653936356306,
+      "eval_icons_MAE_h": 0.02465621568262577,
+      "eval_icons_MAE_w": 0.017738113179802895,
+      "eval_icons_MAE_x": 0.013989617582410574,
+      "eval_icons_MAE_y": 0.018182669766247272,
+      "eval_icons_NUM_probability": 0.999986469745636,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.06894619762897491,
+      "eval_icons_loss_ce": 9.26887514651753e-06,
+      "eval_icons_loss_iou": 0.6270751953125,
+      "eval_icons_loss_num": 0.012828826904296875,
+      "eval_icons_loss_xval": 0.0641326904296875,
+      "eval_icons_runtime": 101.3174,
+      "eval_icons_samples_per_second": 0.493,
+      "eval_icons_steps_per_second": 0.02,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_screenspot_CIoU": 0.5550089081128439,
+      "eval_screenspot_GIoU": 0.5516955653826395,
+      "eval_screenspot_IoU": 0.5956698457400004,
+      "eval_screenspot_MAE_all": 0.08370348066091537,
+      "eval_screenspot_MAE_h": 0.056737360854943596,
+      "eval_screenspot_MAE_w": 0.1454519679148992,
+      "eval_screenspot_MAE_x": 0.08442502965529759,
+      "eval_screenspot_MAE_y": 0.04819955242176851,
+      "eval_screenspot_NUM_probability": 0.9998689492543539,
+      "eval_screenspot_inside_bbox": 0.8395833373069763,
+      "eval_screenspot_loss": 0.9374791979789734,
+      "eval_screenspot_loss_ce": 0.5804212689399719,
+      "eval_screenspot_loss_iou": 0.5504557291666666,
+      "eval_screenspot_loss_num": 0.069854736328125,
+      "eval_screenspot_loss_xval": 0.3492228190104167,
+      "eval_screenspot_runtime": 149.2861,
+      "eval_screenspot_samples_per_second": 0.596,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "loss": 0.9090390205383301,
+      "loss_ce": 0.5727353096008301,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7893733149207601,
+      "grad_norm": 4.193864370890535,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 515420392,
+      "step": 3001
+    },
+    {
+      "epoch": 0.7893733149207601,
+      "loss": 0.12226281315088272,
+      "loss_ce": 0.0011690594255924225,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 515420392,
+      "step": 3001
+    },
+    {
+      "epoch": 0.7896363516801473,
+      "grad_norm": 4.537019888829979,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 515590772,
+      "step": 3002
+    },
+    {
+      "epoch": 0.7896363516801473,
+      "loss": 0.14288941025733948,
+      "loss_ce": 0.0013793996768072248,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 515590772,
+      "step": 3002
+    },
+    {
+      "epoch": 0.7898993884395344,
+      "grad_norm": 7.837104532573747,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 515763064,
+      "step": 3003
+    },
+    {
+      "epoch": 0.7898993884395344,
+      "loss": 0.20384354889392853,
+      "loss_ce": 0.0005659655435010791,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 515763064,
+      "step": 3003
+    },
+    {
+      "epoch": 0.7901624251989215,
+      "grad_norm": 9.871988037860998,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 515935288,
+      "step": 3004
+    },
+    {
+      "epoch": 0.7901624251989215,
+      "loss": 0.1931522786617279,
+      "loss_ce": 0.004248465411365032,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 515935288,
+      "step": 3004
+    },
+    {
+      "epoch": 0.7904254619583087,
+      "grad_norm": 6.705662570329979,
+      "learning_rate": 5e-06,
+      "loss": 0.1395,
+      "num_input_tokens_seen": 516107836,
+      "step": 3005
+    },
+    {
+      "epoch": 0.7904254619583087,
+      "loss": 0.24680155515670776,
+      "loss_ce": 0.0018979848828166723,
+      "loss_iou": 0.53125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 516107836,
+      "step": 3005
+    },
+    {
+      "epoch": 0.7906884987176958,
+      "grad_norm": 3.468178023705032,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 516280260,
+      "step": 3006
+    },
+    {
+      "epoch": 0.7906884987176958,
+      "loss": 0.12856581807136536,
+      "loss_ce": 0.0021620113402605057,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 516280260,
+      "step": 3006
+    },
+    {
+      "epoch": 0.790951535477083,
+      "grad_norm": 2.516769311233636,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 516452472,
+      "step": 3007
+    },
+    {
+      "epoch": 0.790951535477083,
+      "loss": 0.03532446175813675,
+      "loss_ce": 7.665850716875866e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 516452472,
+      "step": 3007
+    },
+    {
+      "epoch": 0.7912145722364701,
+      "grad_norm": 2.8390538522305993,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 516624540,
+      "step": 3008
+    },
+    {
+      "epoch": 0.7912145722364701,
+      "loss": 0.10718082636594772,
+      "loss_ce": 0.0042908103205263615,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 516624540,
+      "step": 3008
+    },
+    {
+      "epoch": 0.7914776089958572,
+      "grad_norm": 4.2650017166958545,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 516796808,
+      "step": 3009
+    },
+    {
+      "epoch": 0.7914776089958572,
+      "loss": 0.17158488929271698,
+      "loss_ce": 0.0024106951896101236,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 516796808,
+      "step": 3009
+    },
+    {
+      "epoch": 0.7917406457552443,
+      "grad_norm": 6.803158022173843,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 516969144,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7917406457552443,
+      "loss": 0.12062282115221024,
+      "loss_ce": 0.0013601221144199371,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 516969144,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7920036825146314,
+      "grad_norm": 5.268394185167123,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 517139672,
+      "step": 3011
+    },
+    {
+      "epoch": 0.7920036825146314,
+      "loss": 0.10529518872499466,
+      "loss_ce": 0.0033054398372769356,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 517139672,
+      "step": 3011
+    },
+    {
+      "epoch": 0.7922667192740186,
+      "grad_norm": 4.897299268033889,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 517311812,
+      "step": 3012
+    },
+    {
+      "epoch": 0.7922667192740186,
+      "loss": 0.15520727634429932,
+      "loss_ce": 0.0010019636247307062,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 517311812,
+      "step": 3012
+    },
+    {
+      "epoch": 0.7925297560334057,
+      "grad_norm": 6.998508129858562,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 517483804,
+      "step": 3013
+    },
+    {
+      "epoch": 0.7925297560334057,
+      "loss": 0.14621217548847198,
+      "loss_ce": 0.004305441863834858,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 517483804,
+      "step": 3013
+    },
+    {
+      "epoch": 0.7927927927927928,
+      "grad_norm": 8.855328154069232,
+      "learning_rate": 5e-06,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 517656016,
+      "step": 3014
+    },
+    {
+      "epoch": 0.7927927927927928,
+      "loss": 0.07204173505306244,
+      "loss_ce": 0.003957017324864864,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 517656016,
+      "step": 3014
+    },
+    {
+      "epoch": 0.7930558295521799,
+      "grad_norm": 26.2413884381383,
+      "learning_rate": 5e-06,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 517828272,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7930558295521799,
+      "loss": 0.11117606610059738,
+      "loss_ce": 0.0008550205966457725,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 517828272,
+      "step": 3015
+    },
+    {
+      "epoch": 0.793318866311567,
+      "grad_norm": 4.148118560901186,
+      "learning_rate": 5e-06,
+      "loss": 0.1143,
+      "num_input_tokens_seen": 518000272,
+      "step": 3016
+    },
+    {
+      "epoch": 0.793318866311567,
+      "loss": 0.06484581530094147,
+      "loss_ce": 0.0005757926846854389,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 518000272,
+      "step": 3016
+    },
+    {
+      "epoch": 0.7935819030709542,
+      "grad_norm": 13.9691023081111,
+      "learning_rate": 5e-06,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 518172412,
+      "step": 3017
+    },
+    {
+      "epoch": 0.7935819030709542,
+      "loss": 0.10305923223495483,
+      "loss_ce": 0.002427519764751196,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 518172412,
+      "step": 3017
+    },
+    {
+      "epoch": 0.7938449398303413,
+      "grad_norm": 4.54961216940048,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 518344668,
+      "step": 3018
+    },
+    {
+      "epoch": 0.7938449398303413,
+      "loss": 0.14651378989219666,
+      "loss_ce": 0.0014942658599466085,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 518344668,
+      "step": 3018
+    },
+    {
+      "epoch": 0.7941079765897284,
+      "grad_norm": 4.200570019762456,
+      "learning_rate": 5e-06,
+      "loss": 0.107,
+      "num_input_tokens_seen": 518517120,
+      "step": 3019
+    },
+    {
+      "epoch": 0.7941079765897284,
+      "loss": 0.11832761764526367,
+      "loss_ce": 0.00037718465318903327,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 518517120,
+      "step": 3019
+    },
+    {
+      "epoch": 0.7943710133491155,
+      "grad_norm": 7.5294315936950005,
+      "learning_rate": 5e-06,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 518689296,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7943710133491155,
+      "loss": 0.06614542007446289,
+      "loss_ce": 0.00021218777692411095,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 518689296,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7946340501085026,
+      "grad_norm": 5.635502401126071,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 518859544,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7946340501085026,
+      "loss": 0.12032654881477356,
+      "loss_ce": 0.0027118013240396976,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 518859544,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7948970868678897,
+      "grad_norm": 6.009060825525287,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 519031676,
+      "step": 3022
+    },
+    {
+      "epoch": 0.7948970868678897,
+      "loss": 0.1436431109905243,
+      "loss_ce": 0.0033232811838388443,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 519031676,
+      "step": 3022
+    },
+    {
+      "epoch": 0.795160123627277,
+      "grad_norm": 4.029495813443641,
+      "learning_rate": 5e-06,
+      "loss": 0.1666,
+      "num_input_tokens_seen": 519203700,
+      "step": 3023
+    },
+    {
+      "epoch": 0.795160123627277,
+      "loss": 0.15425805747509003,
+      "loss_ce": 0.0003884279867634177,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 519203700,
+      "step": 3023
+    },
+    {
+      "epoch": 0.795423160386664,
+      "grad_norm": 37.082876522501756,
+      "learning_rate": 5e-06,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 519375616,
+      "step": 3024
+    },
+    {
+      "epoch": 0.795423160386664,
+      "loss": 0.1391124576330185,
+      "loss_ce": 0.00012015047832392156,
+      "loss_iou": 0.5625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 519375616,
+      "step": 3024
+    },
+    {
+      "epoch": 0.7956861971460512,
+      "grad_norm": 3.823112661234258,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 519547736,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7956861971460512,
+      "loss": 0.09849868714809418,
+      "loss_ce": 0.0015290760202333331,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 519547736,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7959492339054383,
+      "grad_norm": 4.930287612034227,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 519720084,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7959492339054383,
+      "loss": 0.1211109384894371,
+      "loss_ce": 0.005083112046122551,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 519720084,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7962122706648254,
+      "grad_norm": 13.986605867255715,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 519892180,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7962122706648254,
+      "loss": 0.16327086091041565,
+      "loss_ce": 0.0026263254694640636,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 519892180,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7964753074242126,
+      "grad_norm": 12.476632712407037,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 520064636,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7964753074242126,
+      "loss": 0.11255937814712524,
+      "loss_ce": 0.000788743665907532,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 520064636,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7967383441835997,
+      "grad_norm": 3.6081873902324326,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 520236756,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7967383441835997,
+      "loss": 0.12160242348909378,
+      "loss_ce": 0.0018361852271482348,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 520236756,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7970013809429868,
+      "grad_norm": 8.030386724109855,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 520407308,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7970013809429868,
+      "loss": 0.18061389029026031,
+      "loss_ce": 0.0010331911034882069,
+      "loss_iou": 0.5625,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 520407308,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7972644177023739,
+      "grad_norm": 11.21273309596263,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 520579744,
+      "step": 3031
+    },
+    {
+      "epoch": 0.7972644177023739,
+      "loss": 0.082832470536232,
+      "loss_ce": 0.0002824235416483134,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 520579744,
+      "step": 3031
+    },
+    {
+      "epoch": 0.797527454461761,
+      "grad_norm": 20.85679503542732,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 520752148,
+      "step": 3032
+    },
+    {
+      "epoch": 0.797527454461761,
+      "loss": 0.12713350355625153,
+      "loss_ce": 0.0008517719688825309,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 520752148,
+      "step": 3032
+    },
+    {
+      "epoch": 0.7977904912211482,
+      "grad_norm": 37.09104960564763,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 520924220,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7977904912211482,
+      "loss": 0.10619133710861206,
+      "loss_ce": 0.0030724371317774057,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 520924220,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7980535279805353,
+      "grad_norm": 16.059073148748364,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 521093348,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7980535279805353,
+      "loss": 0.06166623532772064,
+      "loss_ce": 0.0008294428698718548,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 521093348,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7983165647399224,
+      "grad_norm": 8.368290368776519,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 521265688,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7983165647399224,
+      "loss": 0.14087893068790436,
+      "loss_ce": 0.004129666369408369,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 521265688,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7985796014993095,
+      "grad_norm": 25.178404936799044,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 521437496,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7985796014993095,
+      "loss": 0.29297274351119995,
+      "loss_ce": 0.002842147834599018,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 521437496,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7988426382586966,
+      "grad_norm": 3.991516418389156,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 521609940,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7988426382586966,
+      "loss": 0.09374190121889114,
+      "loss_ce": 0.0009379457915201783,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 521609940,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7991056750180838,
+      "grad_norm": 4.808453824486521,
+      "learning_rate": 5e-06,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 521781892,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7991056750180838,
+      "loss": 0.07893365621566772,
+      "loss_ce": 0.001052800682373345,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 521781892,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7993687117774709,
+      "grad_norm": 12.778560985993714,
+      "learning_rate": 5e-06,
+      "loss": 0.109,
+      "num_input_tokens_seen": 521954108,
+      "step": 3039
+    },
+    {
+      "epoch": 0.7993687117774709,
+      "loss": 0.12000415474176407,
+      "loss_ce": 0.0017790585989132524,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 521954108,
+      "step": 3039
+    },
+    {
+      "epoch": 0.799631748536858,
+      "grad_norm": 5.33906426825674,
+      "learning_rate": 5e-06,
+      "loss": 0.1944,
+      "num_input_tokens_seen": 522126236,
+      "step": 3040
+    },
+    {
+      "epoch": 0.799631748536858,
+      "loss": 0.2638484239578247,
+      "loss_ce": 0.00042068029870279133,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 522126236,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7998947852962451,
+      "grad_norm": 4.623536872405031,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 522298380,
+      "step": 3041
+    },
+    {
+      "epoch": 0.7998947852962451,
+      "loss": 0.12680000066757202,
+      "loss_ce": 0.0012506938073784113,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 522298380,
+      "step": 3041
+    },
+    {
+      "epoch": 0.8001578220556322,
+      "grad_norm": 4.125445081698177,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 522470360,
+      "step": 3042
+    },
+    {
+      "epoch": 0.8001578220556322,
+      "loss": 0.15849211812019348,
+      "loss_ce": 0.0013113392051309347,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 522470360,
+      "step": 3042
+    },
+    {
+      "epoch": 0.8004208588150195,
+      "grad_norm": 11.789889533203668,
+      "learning_rate": 5e-06,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 522642472,
+      "step": 3043
+    },
+    {
+      "epoch": 0.8004208588150195,
+      "loss": 0.09324462711811066,
+      "loss_ce": 0.0006237818161025643,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 522642472,
+      "step": 3043
+    },
+    {
+      "epoch": 0.8006838955744066,
+      "grad_norm": 4.414508918444819,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 522814808,
+      "step": 3044
+    },
+    {
+      "epoch": 0.8006838955744066,
+      "loss": 0.07435610890388489,
+      "loss_ce": 0.0009460713481530547,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 522814808,
+      "step": 3044
+    },
+    {
+      "epoch": 0.8009469323337937,
+      "grad_norm": 6.1075510398524395,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 522986892,
+      "step": 3045
+    },
+    {
+      "epoch": 0.8009469323337937,
+      "loss": 0.08977600932121277,
+      "loss_ce": 0.0010614084312692285,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 522986892,
+      "step": 3045
+    },
+    {
+      "epoch": 0.8012099690931808,
+      "grad_norm": 6.86235832016632,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 523159044,
+      "step": 3046
+    },
+    {
+      "epoch": 0.8012099690931808,
+      "loss": 0.10283628106117249,
+      "loss_ce": 0.003226900240406394,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 523159044,
+      "step": 3046
+    },
+    {
+      "epoch": 0.8014730058525679,
+      "grad_norm": 8.620167546317841,
+      "learning_rate": 5e-06,
+      "loss": 0.1508,
+      "num_input_tokens_seen": 523331372,
+      "step": 3047
+    },
+    {
+      "epoch": 0.8014730058525679,
+      "loss": 0.17512959241867065,
+      "loss_ce": 0.0035597749520093203,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 523331372,
+      "step": 3047
+    },
+    {
+      "epoch": 0.801736042611955,
+      "grad_norm": 4.823577960221348,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 523504004,
+      "step": 3048
+    },
+    {
+      "epoch": 0.801736042611955,
+      "loss": 0.07766547054052353,
+      "loss_ce": 0.0020429138094186783,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 523504004,
+      "step": 3048
+    },
+    {
+      "epoch": 0.8019990793713422,
+      "grad_norm": 11.00975772697334,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 523675952,
+      "step": 3049
+    },
+    {
+      "epoch": 0.8019990793713422,
+      "loss": 0.11927802860736847,
+      "loss_ce": 0.001342848176136613,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 523675952,
+      "step": 3049
+    },
+    {
+      "epoch": 0.8022621161307293,
+      "grad_norm": 10.685552832243063,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 523848152,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8022621161307293,
+      "loss": 0.0827643871307373,
+      "loss_ce": 0.002106426050886512,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 523848152,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8025251528901164,
+      "grad_norm": 24.17010500449297,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 524020708,
+      "step": 3051
+    },
+    {
+      "epoch": 0.8025251528901164,
+      "loss": 0.10048617422580719,
+      "loss_ce": 0.000510584854055196,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 524020708,
+      "step": 3051
+    },
+    {
+      "epoch": 0.8027881896495035,
+      "grad_norm": 4.992845761930844,
+      "learning_rate": 5e-06,
+      "loss": 0.1304,
+      "num_input_tokens_seen": 524191372,
+      "step": 3052
+    },
+    {
+      "epoch": 0.8027881896495035,
+      "loss": 0.15961629152297974,
+      "loss_ce": 0.006387531757354736,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 524191372,
+      "step": 3052
+    },
+    {
+      "epoch": 0.8030512264088906,
+      "grad_norm": 2.943381415031694,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 524363912,
+      "step": 3053
+    },
+    {
+      "epoch": 0.8030512264088906,
+      "loss": 0.15981021523475647,
+      "loss_ce": 0.0005695016006939113,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 524363912,
+      "step": 3053
+    },
+    {
+      "epoch": 0.8033142631682778,
+      "grad_norm": 5.648094984205903,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 524536316,
+      "step": 3054
+    },
+    {
+      "epoch": 0.8033142631682778,
+      "loss": 0.1740710288286209,
+      "loss_ce": 0.0022723155561834574,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 524536316,
+      "step": 3054
+    },
+    {
+      "epoch": 0.8035772999276649,
+      "grad_norm": 7.984973488378181,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 524708444,
+      "step": 3055
+    },
+    {
+      "epoch": 0.8035772999276649,
+      "loss": 0.07127489894628525,
+      "loss_ce": 0.000809812976513058,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 524708444,
+      "step": 3055
+    },
+    {
+      "epoch": 0.803840336687052,
+      "grad_norm": 6.031677657650838,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 524880960,
+      "step": 3056
+    },
+    {
+      "epoch": 0.803840336687052,
+      "loss": 0.0638066828250885,
+      "loss_ce": 0.0005437473300844431,
+      "loss_iou": 0.703125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 524880960,
+      "step": 3056
+    },
+    {
+      "epoch": 0.8041033734464391,
+      "grad_norm": 5.599082172393745,
+      "learning_rate": 5e-06,
+      "loss": 0.086,
+      "num_input_tokens_seen": 525053308,
+      "step": 3057
+    },
+    {
+      "epoch": 0.8041033734464391,
+      "loss": 0.07932358235120773,
+      "loss_ce": 0.0024650623090565205,
+      "loss_iou": 0.546875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 525053308,
+      "step": 3057
+    },
+    {
+      "epoch": 0.8043664102058262,
+      "grad_norm": 28.17799645696428,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 525225432,
+      "step": 3058
+    },
+    {
+      "epoch": 0.8043664102058262,
+      "loss": 0.08576367795467377,
+      "loss_ce": 0.0013825736241415143,
+      "loss_iou": 0.578125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 525225432,
+      "step": 3058
+    },
+    {
+      "epoch": 0.8046294469652134,
+      "grad_norm": 3.4834198316912564,
+      "learning_rate": 5e-06,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 525397804,
+      "step": 3059
+    },
+    {
+      "epoch": 0.8046294469652134,
+      "loss": 0.07873048633337021,
+      "loss_ce": 0.0016125671099871397,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 525397804,
+      "step": 3059
+    },
+    {
+      "epoch": 0.8048924837246005,
+      "grad_norm": 2.9860188986605367,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 525570124,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8048924837246005,
+      "loss": 0.07787738740444183,
+      "loss_ce": 0.0008205035701394081,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 525570124,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8051555204839876,
+      "grad_norm": 14.35918831416577,
+      "learning_rate": 5e-06,
+      "loss": 0.1217,
+      "num_input_tokens_seen": 525742708,
+      "step": 3061
+    },
+    {
+      "epoch": 0.8051555204839876,
+      "loss": 0.10341215878725052,
+      "loss_ce": 0.00123931048437953,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 525742708,
+      "step": 3061
+    },
+    {
+      "epoch": 0.8054185572433747,
+      "grad_norm": 16.381501371718237,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 525914908,
+      "step": 3062
+    },
+    {
+      "epoch": 0.8054185572433747,
+      "loss": 0.10784236341714859,
+      "loss_ce": 0.0002678967430256307,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 525914908,
+      "step": 3062
+    },
+    {
+      "epoch": 0.8056815940027618,
+      "grad_norm": 4.73914065847523,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 526087040,
+      "step": 3063
+    },
+    {
+      "epoch": 0.8056815940027618,
+      "loss": 0.06849893927574158,
+      "loss_ce": 0.000368443870684132,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 526087040,
+      "step": 3063
+    },
+    {
+      "epoch": 0.8059446307621491,
+      "grad_norm": 8.562314005917766,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 526259428,
+      "step": 3064
+    },
+    {
+      "epoch": 0.8059446307621491,
+      "loss": 0.07737226039171219,
+      "loss_ce": 0.0030009234324097633,
+      "loss_iou": 0.59375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 526259428,
+      "step": 3064
+    },
+    {
+      "epoch": 0.8062076675215362,
+      "grad_norm": 12.302010843258703,
+      "learning_rate": 5e-06,
+      "loss": 0.1505,
+      "num_input_tokens_seen": 526431568,
+      "step": 3065
+    },
+    {
+      "epoch": 0.8062076675215362,
+      "loss": 0.20694774389266968,
+      "loss_ce": 0.0032123818527907133,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 526431568,
+      "step": 3065
+    },
+    {
+      "epoch": 0.8064707042809233,
+      "grad_norm": 8.1243600151207,
+      "learning_rate": 5e-06,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 526603728,
+      "step": 3066
+    },
+    {
+      "epoch": 0.8064707042809233,
+      "loss": 0.14467400312423706,
+      "loss_ce": 0.0004784482589457184,
+      "loss_iou": 0.59375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 526603728,
+      "step": 3066
+    },
+    {
+      "epoch": 0.8067337410403104,
+      "grad_norm": 92.56005912402708,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 526775572,
+      "step": 3067
+    },
+    {
+      "epoch": 0.8067337410403104,
+      "loss": 0.08829745650291443,
+      "loss_ce": 0.0011697689769789577,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 526775572,
+      "step": 3067
+    },
+    {
+      "epoch": 0.8069967777996975,
+      "grad_norm": 33.266184929246414,
+      "learning_rate": 5e-06,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 526947576,
+      "step": 3068
+    },
+    {
+      "epoch": 0.8069967777996975,
+      "loss": 0.060733191668987274,
+      "loss_ce": 0.0012391710188239813,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 526947576,
+      "step": 3068
+    },
+    {
+      "epoch": 0.8072598145590847,
+      "grad_norm": 5.97758373469706,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 527117908,
+      "step": 3069
+    },
+    {
+      "epoch": 0.8072598145590847,
+      "loss": 0.09366244077682495,
+      "loss_ce": 0.006046472117304802,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 527117908,
+      "step": 3069
+    },
+    {
+      "epoch": 0.8075228513184718,
+      "grad_norm": 5.290039917786788,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 527287100,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8075228513184718,
+      "loss": 0.10383239388465881,
+      "loss_ce": 0.00046935188584029675,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 527287100,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8077858880778589,
+      "grad_norm": 4.7891570756526125,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 527458888,
+      "step": 3071
+    },
+    {
+      "epoch": 0.8077858880778589,
+      "loss": 0.10491342842578888,
+      "loss_ce": 0.0008484934223815799,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 527458888,
+      "step": 3071
+    },
+    {
+      "epoch": 0.808048924837246,
+      "grad_norm": 11.230743830755795,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 527630920,
+      "step": 3072
+    },
+    {
+      "epoch": 0.808048924837246,
+      "loss": 0.06209864094853401,
+      "loss_ce": 0.0007583063561469316,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 527630920,
+      "step": 3072
+    },
+    {
+      "epoch": 0.8083119615966331,
+      "grad_norm": 5.040744603317089,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 527803168,
+      "step": 3073
+    },
+    {
+      "epoch": 0.8083119615966331,
+      "loss": 0.1508997231721878,
+      "loss_ce": 0.0011194492690265179,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 527803168,
+      "step": 3073
+    },
+    {
+      "epoch": 0.8085749983560202,
+      "grad_norm": 8.63208434905108,
+      "learning_rate": 5e-06,
+      "loss": 0.0993,
+      "num_input_tokens_seen": 527975596,
+      "step": 3074
+    },
+    {
+      "epoch": 0.8085749983560202,
+      "loss": 0.11231046169996262,
+      "loss_ce": 0.00017361767822876573,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 527975596,
+      "step": 3074
+    },
+    {
+      "epoch": 0.8088380351154074,
+      "grad_norm": 4.381253880078918,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 528147980,
+      "step": 3075
+    },
+    {
+      "epoch": 0.8088380351154074,
+      "loss": 0.11533799767494202,
+      "loss_ce": 0.0029875326436012983,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 528147980,
+      "step": 3075
+    },
+    {
+      "epoch": 0.8091010718747945,
+      "grad_norm": 8.161484057964593,
+      "learning_rate": 5e-06,
+      "loss": 0.1619,
+      "num_input_tokens_seen": 528320296,
+      "step": 3076
+    },
+    {
+      "epoch": 0.8091010718747945,
+      "loss": 0.16661548614501953,
+      "loss_ce": 0.0008439991506747901,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 528320296,
+      "step": 3076
+    },
+    {
+      "epoch": 0.8093641086341816,
+      "grad_norm": 4.683292048006889,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 528492424,
+      "step": 3077
+    },
+    {
+      "epoch": 0.8093641086341816,
+      "loss": 0.11115469038486481,
+      "loss_ce": 0.0029546155128628016,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 528492424,
+      "step": 3077
+    },
+    {
+      "epoch": 0.8096271453935687,
+      "grad_norm": 5.257960533659777,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 528664732,
+      "step": 3078
+    },
+    {
+      "epoch": 0.8096271453935687,
+      "loss": 0.21525058150291443,
+      "loss_ce": 0.001490199938416481,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 528664732,
+      "step": 3078
+    },
+    {
+      "epoch": 0.8098901821529558,
+      "grad_norm": 12.566112829085096,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 528837012,
+      "step": 3079
+    },
+    {
+      "epoch": 0.8098901821529558,
+      "loss": 0.08169254660606384,
+      "loss_ce": 0.00021060870494693518,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 528837012,
+      "step": 3079
+    },
+    {
+      "epoch": 0.810153218912343,
+      "grad_norm": 5.324934758940462,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 529009052,
+      "step": 3080
+    },
+    {
+      "epoch": 0.810153218912343,
+      "loss": 0.08878730237483978,
+      "loss_ce": 0.0001795147400116548,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 529009052,
+      "step": 3080
+    },
+    {
+      "epoch": 0.8104162556717301,
+      "grad_norm": 9.272723862546323,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 529181216,
+      "step": 3081
+    },
+    {
+      "epoch": 0.8104162556717301,
+      "loss": 0.0852864533662796,
+      "loss_ce": 0.0025990745052695274,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 529181216,
+      "step": 3081
+    },
+    {
+      "epoch": 0.8106792924311172,
+      "grad_norm": 28.988099538140567,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 529351788,
+      "step": 3082
+    },
+    {
+      "epoch": 0.8106792924311172,
+      "loss": 0.05319926142692566,
+      "loss_ce": 0.0007700645364820957,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 529351788,
+      "step": 3082
+    },
+    {
+      "epoch": 0.8109423291905044,
+      "grad_norm": 5.871269884706642,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 529524200,
+      "step": 3083
+    },
+    {
+      "epoch": 0.8109423291905044,
+      "loss": 0.180719256401062,
+      "loss_ce": 0.004815942607820034,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 529524200,
+      "step": 3083
+    },
+    {
+      "epoch": 0.8112053659498915,
+      "grad_norm": 9.034864384864637,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 529696672,
+      "step": 3084
+    },
+    {
+      "epoch": 0.8112053659498915,
+      "loss": 0.039623767137527466,
+      "loss_ce": 0.0003476430138107389,
+      "loss_iou": 0.546875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 529696672,
+      "step": 3084
+    },
+    {
+      "epoch": 0.8114684027092787,
+      "grad_norm": 7.571992947700351,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 529868932,
+      "step": 3085
+    },
+    {
+      "epoch": 0.8114684027092787,
+      "loss": 0.15128442645072937,
+      "loss_ce": 0.002373900031670928,
+      "loss_iou": 0.375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 529868932,
+      "step": 3085
+    },
+    {
+      "epoch": 0.8117314394686658,
+      "grad_norm": 29.21584689056005,
+      "learning_rate": 5e-06,
+      "loss": 0.0879,
+      "num_input_tokens_seen": 530041080,
+      "step": 3086
+    },
+    {
+      "epoch": 0.8117314394686658,
+      "loss": 0.11466438323259354,
+      "loss_ce": 0.001291584805585444,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 530041080,
+      "step": 3086
+    },
+    {
+      "epoch": 0.8119944762280529,
+      "grad_norm": 8.627125392444821,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 530210668,
+      "step": 3087
+    },
+    {
+      "epoch": 0.8119944762280529,
+      "loss": 0.1240130364894867,
+      "loss_ce": 0.0024767834693193436,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 530210668,
+      "step": 3087
+    },
+    {
+      "epoch": 0.81225751298744,
+      "grad_norm": 11.153523319241675,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 530382776,
+      "step": 3088
+    },
+    {
+      "epoch": 0.81225751298744,
+      "loss": 0.07181555032730103,
+      "loss_ce": 0.00026709536905400455,
+      "loss_iou": NaN,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 530382776,
+      "step": 3088
+    },
+    {
+      "epoch": 0.8125205497468271,
+      "grad_norm": 10.380517960562548,
+      "learning_rate": 5e-06,
+      "loss": 0.1332,
+      "num_input_tokens_seen": 530555004,
+      "step": 3089
+    },
+    {
+      "epoch": 0.8125205497468271,
+      "loss": 0.16954563558101654,
+      "loss_ce": 0.0021872336510568857,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 530555004,
+      "step": 3089
+    },
+    {
+      "epoch": 0.8127835865062143,
+      "grad_norm": 5.4446542029328775,
+      "learning_rate": 5e-06,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 530726976,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8127835865062143,
+      "loss": 0.08060289919376373,
+      "loss_ce": 0.00040270722820423543,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 530726976,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8130466232656014,
+      "grad_norm": 5.347357758568155,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 530898852,
+      "step": 3091
+    },
+    {
+      "epoch": 0.8130466232656014,
+      "loss": 0.08159644901752472,
+      "loss_ce": 0.0003891719679813832,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 530898852,
+      "step": 3091
+    },
+    {
+      "epoch": 0.8133096600249885,
+      "grad_norm": 4.89308048127463,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 531071100,
+      "step": 3092
+    },
+    {
+      "epoch": 0.8133096600249885,
+      "loss": 0.05772348493337631,
+      "loss_ce": 7.577867654617876e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 531071100,
+      "step": 3092
+    },
+    {
+      "epoch": 0.8135726967843756,
+      "grad_norm": 5.314071866923476,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 531243088,
+      "step": 3093
+    },
+    {
+      "epoch": 0.8135726967843756,
+      "loss": 0.16711823642253876,
+      "loss_ce": 0.002140207216143608,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 531243088,
+      "step": 3093
+    },
+    {
+      "epoch": 0.8138357335437627,
+      "grad_norm": 4.018224451020325,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 531415364,
+      "step": 3094
+    },
+    {
+      "epoch": 0.8138357335437627,
+      "loss": 0.07948748767375946,
+      "loss_ce": 0.0013472279533743858,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 531415364,
+      "step": 3094
+    },
+    {
+      "epoch": 0.8140987703031498,
+      "grad_norm": 5.643939653563094,
+      "learning_rate": 5e-06,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 531587860,
+      "step": 3095
+    },
+    {
+      "epoch": 0.8140987703031498,
+      "loss": 0.070250503718853,
+      "loss_ce": 0.00019740140123758465,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 531587860,
+      "step": 3095
+    },
+    {
+      "epoch": 0.814361807062537,
+      "grad_norm": 6.298167200858131,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 531760160,
+      "step": 3096
+    },
+    {
+      "epoch": 0.814361807062537,
+      "loss": 0.13414643704891205,
+      "loss_ce": 0.0014254867564886808,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 531760160,
+      "step": 3096
+    },
+    {
+      "epoch": 0.8146248438219241,
+      "grad_norm": 4.559829968544296,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 531932728,
+      "step": 3097
+    },
+    {
+      "epoch": 0.8146248438219241,
+      "loss": 0.11818031221628189,
+      "loss_ce": 0.0006265999400056899,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 531932728,
+      "step": 3097
+    },
+    {
+      "epoch": 0.8148878805813112,
+      "grad_norm": 6.798137203947019,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 532104748,
+      "step": 3098
+    },
+    {
+      "epoch": 0.8148878805813112,
+      "loss": 0.13361144065856934,
+      "loss_ce": 0.0008294496219605207,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 532104748,
+      "step": 3098
+    },
+    {
+      "epoch": 0.8151509173406983,
+      "grad_norm": 4.492953640528966,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 532276872,
+      "step": 3099
+    },
+    {
+      "epoch": 0.8151509173406983,
+      "loss": 0.23120509088039398,
+      "loss_ce": 0.004703632555902004,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 532276872,
+      "step": 3099
+    },
+    {
+      "epoch": 0.8154139541000854,
+      "grad_norm": 23.973104234846247,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 532449140,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8154139541000854,
+      "loss": 0.2094377875328064,
+      "loss_ce": 0.005488819442689419,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 532449140,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8156769908594727,
+      "grad_norm": 5.077265892021237,
+      "learning_rate": 5e-06,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 532621452,
+      "step": 3101
+    },
+    {
+      "epoch": 0.8156769908594727,
+      "loss": 0.07252339273691177,
+      "loss_ce": 0.001692092278972268,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 532621452,
+      "step": 3101
+    },
+    {
+      "epoch": 0.8159400276188598,
+      "grad_norm": 4.611749386992743,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 532793420,
+      "step": 3102
+    },
+    {
+      "epoch": 0.8159400276188598,
+      "loss": 0.10748874396085739,
+      "loss_ce": 0.0007687745383009315,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 532793420,
+      "step": 3102
+    },
+    {
+      "epoch": 0.8162030643782469,
+      "grad_norm": 8.46788739165104,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 532965672,
+      "step": 3103
+    },
+    {
+      "epoch": 0.8162030643782469,
+      "loss": 0.0986635610461235,
+      "loss_ce": 0.006317365914583206,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 532965672,
+      "step": 3103
+    },
+    {
+      "epoch": 0.816466101137634,
+      "grad_norm": 5.21012803335228,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 533137620,
+      "step": 3104
+    },
+    {
+      "epoch": 0.816466101137634,
+      "loss": 0.1294005811214447,
+      "loss_ce": 0.0007232190691865981,
+      "loss_iou": NaN,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 533137620,
+      "step": 3104
+    },
+    {
+      "epoch": 0.8167291378970211,
+      "grad_norm": 4.29918218259194,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 533308040,
+      "step": 3105
+    },
+    {
+      "epoch": 0.8167291378970211,
+      "loss": 0.07954747974872589,
+      "loss_ce": 0.0007816128781996667,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 533308040,
+      "step": 3105
+    },
+    {
+      "epoch": 0.8169921746564083,
+      "grad_norm": 7.9379046963597775,
+      "learning_rate": 5e-06,
+      "loss": 0.0867,
+      "num_input_tokens_seen": 533480088,
+      "step": 3106
+    },
+    {
+      "epoch": 0.8169921746564083,
+      "loss": 0.055092211812734604,
+      "loss_ce": 0.002663013059645891,
+      "loss_iou": 0.578125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 533480088,
+      "step": 3106
+    },
+    {
+      "epoch": 0.8172552114157954,
+      "grad_norm": 5.798174841670742,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 533652304,
+      "step": 3107
+    },
+    {
+      "epoch": 0.8172552114157954,
+      "loss": 0.10001754760742188,
+      "loss_ce": 0.0004997201031073928,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 533652304,
+      "step": 3107
+    },
+    {
+      "epoch": 0.8175182481751825,
+      "grad_norm": 4.397743796531851,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 533824784,
+      "step": 3108
+    },
+    {
+      "epoch": 0.8175182481751825,
+      "loss": 0.04015748202800751,
+      "loss_ce": 0.001079726149328053,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 533824784,
+      "step": 3108
+    },
+    {
+      "epoch": 0.8177812849345696,
+      "grad_norm": 3.0634489465977937,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 533997192,
+      "step": 3109
+    },
+    {
+      "epoch": 0.8177812849345696,
+      "loss": 0.11612477153539658,
+      "loss_ce": 0.0004936738405376673,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 533997192,
+      "step": 3109
+    },
+    {
+      "epoch": 0.8180443216939567,
+      "grad_norm": 11.903043667241318,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 534169324,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8180443216939567,
+      "loss": 0.12225233018398285,
+      "loss_ce": 0.002577648963779211,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 534169324,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8183073584533439,
+      "grad_norm": 11.232858169527498,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 534341512,
+      "step": 3111
+    },
+    {
+      "epoch": 0.8183073584533439,
+      "loss": 0.093436598777771,
+      "loss_ce": 0.00041902740485966206,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 534341512,
+      "step": 3111
+    },
+    {
+      "epoch": 0.818570395212731,
+      "grad_norm": 4.531836934905563,
+      "learning_rate": 5e-06,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 534511748,
+      "step": 3112
+    },
+    {
+      "epoch": 0.818570395212731,
+      "loss": 0.07575514912605286,
+      "loss_ce": 0.00020888875587843359,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 534511748,
+      "step": 3112
+    },
+    {
+      "epoch": 0.8188334319721181,
+      "grad_norm": 7.470057833509574,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 534683872,
+      "step": 3113
+    },
+    {
+      "epoch": 0.8188334319721181,
+      "loss": 0.10066039860248566,
+      "loss_ce": 0.00047118880320340395,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 534683872,
+      "step": 3113
+    },
+    {
+      "epoch": 0.8190964687315052,
+      "grad_norm": 6.233336264317783,
+      "learning_rate": 5e-06,
+      "loss": 0.139,
+      "num_input_tokens_seen": 534855832,
+      "step": 3114
+    },
+    {
+      "epoch": 0.8190964687315052,
+      "loss": 0.09569014608860016,
+      "loss_ce": 0.0034660203382372856,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 534855832,
+      "step": 3114
+    },
+    {
+      "epoch": 0.8193595054908923,
+      "grad_norm": 5.930104072036699,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 535027896,
+      "step": 3115
+    },
+    {
+      "epoch": 0.8193595054908923,
+      "loss": 0.13678929209709167,
+      "loss_ce": 0.001047111232765019,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 535027896,
+      "step": 3115
+    },
+    {
+      "epoch": 0.8196225422502795,
+      "grad_norm": 12.505587123966333,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 535200112,
+      "step": 3116
+    },
+    {
+      "epoch": 0.8196225422502795,
+      "loss": 0.0822412520647049,
+      "loss_ce": 0.0008356067701242864,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 535200112,
+      "step": 3116
+    },
+    {
+      "epoch": 0.8198855790096666,
+      "grad_norm": 4.231730639769997,
+      "learning_rate": 5e-06,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 535372284,
+      "step": 3117
+    },
+    {
+      "epoch": 0.8198855790096666,
+      "loss": 0.12019523978233337,
+      "loss_ce": 0.00018486013868823647,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 535372284,
+      "step": 3117
+    },
+    {
+      "epoch": 0.8201486157690537,
+      "grad_norm": 7.438830292525926,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 535544868,
+      "step": 3118
+    },
+    {
+      "epoch": 0.8201486157690537,
+      "loss": 0.16251003742218018,
+      "loss_ce": 0.0003091081453021616,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 535544868,
+      "step": 3118
+    },
+    {
+      "epoch": 0.8204116525284408,
+      "grad_norm": 4.777332996643992,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 535717156,
+      "step": 3119
+    },
+    {
+      "epoch": 0.8204116525284408,
+      "loss": 0.16800367832183838,
+      "loss_ce": 0.00139296252746135,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 535717156,
+      "step": 3119
+    },
+    {
+      "epoch": 0.8206746892878279,
+      "grad_norm": 9.684696947025552,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 535887380,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8206746892878279,
+      "loss": 0.0679091364145279,
+      "loss_ce": 0.0011366696562618017,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 535887380,
+      "step": 3120
+    },
+    {
+      "epoch": 0.820937726047215,
+      "grad_norm": 5.315834673220869,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 536059444,
+      "step": 3121
+    },
+    {
+      "epoch": 0.820937726047215,
+      "loss": 0.04389767348766327,
+      "loss_ce": 0.00013546722766477615,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 536059444,
+      "step": 3121
+    },
+    {
+      "epoch": 0.8212007628066023,
+      "grad_norm": 4.263271795831078,
+      "learning_rate": 5e-06,
+      "loss": 0.1419,
+      "num_input_tokens_seen": 536231404,
+      "step": 3122
+    },
+    {
+      "epoch": 0.8212007628066023,
+      "loss": 0.15159711241722107,
+      "loss_ce": 0.004105648957192898,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 536231404,
+      "step": 3122
+    },
+    {
+      "epoch": 0.8214637995659894,
+      "grad_norm": 7.367303112338102,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 536403668,
+      "step": 3123
+    },
+    {
+      "epoch": 0.8214637995659894,
+      "loss": 0.1084074005484581,
+      "loss_ce": 0.0006498372531495988,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 536403668,
+      "step": 3123
+    },
+    {
+      "epoch": 0.8217268363253765,
+      "grad_norm": 6.857717580081254,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 536575892,
+      "step": 3124
+    },
+    {
+      "epoch": 0.8217268363253765,
+      "loss": 0.0787474736571312,
+      "loss_ce": 0.00030203917413018644,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 536575892,
+      "step": 3124
+    },
+    {
+      "epoch": 0.8219898730847636,
+      "grad_norm": 4.826419943692509,
+      "learning_rate": 5e-06,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 536745964,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8219898730847636,
+      "loss": 0.06057834252715111,
+      "loss_ce": 7.724385795881972e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 536745964,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8222529098441507,
+      "grad_norm": 8.883739420539388,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 536918180,
+      "step": 3126
+    },
+    {
+      "epoch": 0.8222529098441507,
+      "loss": 0.11514750123023987,
+      "loss_ce": 0.0007676149252802134,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 536918180,
+      "step": 3126
+    },
+    {
+      "epoch": 0.8225159466035379,
+      "grad_norm": 4.391986789320263,
+      "learning_rate": 5e-06,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 537090196,
+      "step": 3127
+    },
+    {
+      "epoch": 0.8225159466035379,
+      "loss": 0.13754448294639587,
+      "loss_ce": 0.0006731519242748618,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 537090196,
+      "step": 3127
+    },
+    {
+      "epoch": 0.822778983362925,
+      "grad_norm": 7.64555343727262,
+      "learning_rate": 5e-06,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 537259572,
+      "step": 3128
+    },
+    {
+      "epoch": 0.822778983362925,
+      "loss": 0.05659133195877075,
+      "loss_ce": 5.7516066590324044e-05,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 537259572,
+      "step": 3128
+    },
+    {
+      "epoch": 0.8230420201223121,
+      "grad_norm": 4.3032702337998225,
+      "learning_rate": 5e-06,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 537431812,
+      "step": 3129
+    },
+    {
+      "epoch": 0.8230420201223121,
+      "loss": 0.07997481524944305,
+      "loss_ce": 0.0033604400232434273,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 537431812,
+      "step": 3129
+    },
+    {
+      "epoch": 0.8233050568816992,
+      "grad_norm": 11.32084254562734,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 537602468,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8233050568816992,
+      "loss": 0.12599240243434906,
+      "loss_ce": 0.0030370799358934164,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 537602468,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8235680936410863,
+      "grad_norm": 7.355242955773187,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 537774448,
+      "step": 3131
+    },
+    {
+      "epoch": 0.8235680936410863,
+      "loss": 0.18665780127048492,
+      "loss_ce": 0.0003784986911341548,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 537774448,
+      "step": 3131
+    },
+    {
+      "epoch": 0.8238311304004735,
+      "grad_norm": 11.757519786843565,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 537946740,
+      "step": 3132
+    },
+    {
+      "epoch": 0.8238311304004735,
+      "loss": 0.06623389571905136,
+      "loss_ce": 0.0015976695576682687,
+      "loss_iou": 0.359375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 537946740,
+      "step": 3132
+    },
+    {
+      "epoch": 0.8240941671598606,
+      "grad_norm": 17.264629874858777,
+      "learning_rate": 5e-06,
+      "loss": 0.1361,
+      "num_input_tokens_seen": 538118940,
+      "step": 3133
+    },
+    {
+      "epoch": 0.8240941671598606,
+      "loss": 0.09394903481006622,
+      "loss_ce": 0.0025488929823040962,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 538118940,
+      "step": 3133
+    },
+    {
+      "epoch": 0.8243572039192477,
+      "grad_norm": 12.736031512367697,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 538291488,
+      "step": 3134
+    },
+    {
+      "epoch": 0.8243572039192477,
+      "loss": 0.12837865948677063,
+      "loss_ce": 0.0019290748750790954,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 538291488,
+      "step": 3134
+    },
+    {
+      "epoch": 0.8246202406786348,
+      "grad_norm": 4.032167194049116,
+      "learning_rate": 5e-06,
+      "loss": 0.0797,
+      "num_input_tokens_seen": 538461628,
+      "step": 3135
+    },
+    {
+      "epoch": 0.8246202406786348,
+      "loss": 0.09231233596801758,
+      "loss_ce": 0.0006070120725780725,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 538461628,
+      "step": 3135
+    },
+    {
+      "epoch": 0.8248832774380219,
+      "grad_norm": 3.914309781520812,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 538633772,
+      "step": 3136
+    },
+    {
+      "epoch": 0.8248832774380219,
+      "loss": 0.07093091309070587,
+      "loss_ce": 0.00011487161100376397,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 538633772,
+      "step": 3136
+    },
+    {
+      "epoch": 0.8251463141974091,
+      "grad_norm": 3.7923535168173763,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 538805732,
+      "step": 3137
+    },
+    {
+      "epoch": 0.8251463141974091,
+      "loss": 0.1175466924905777,
+      "loss_ce": 0.0006643689121119678,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 538805732,
+      "step": 3137
+    },
+    {
+      "epoch": 0.8254093509567962,
+      "grad_norm": 10.05128274905235,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 538977916,
+      "step": 3138
+    },
+    {
+      "epoch": 0.8254093509567962,
+      "loss": 0.11163683980703354,
+      "loss_ce": 0.003146846778690815,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 538977916,
+      "step": 3138
+    },
+    {
+      "epoch": 0.8256723877161833,
+      "grad_norm": 12.00226096912509,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 539150104,
+      "step": 3139
+    },
+    {
+      "epoch": 0.8256723877161833,
+      "loss": 0.14204728603363037,
+      "loss_ce": 0.0005983082228340209,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 539150104,
+      "step": 3139
+    },
+    {
+      "epoch": 0.8259354244755704,
+      "grad_norm": 10.327954218233959,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 539322280,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8259354244755704,
+      "loss": 0.1829340159893036,
+      "loss_ce": 0.004543509799987078,
+      "loss_iou": 0.453125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 539322280,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8261984612349575,
+      "grad_norm": 3.552655492096545,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 539494448,
+      "step": 3141
+    },
+    {
+      "epoch": 0.8261984612349575,
+      "loss": 0.12512250244617462,
+      "loss_ce": 0.003601514268666506,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 539494448,
+      "step": 3141
+    },
+    {
+      "epoch": 0.8264614979943448,
+      "grad_norm": 4.0634516483762795,
+      "learning_rate": 5e-06,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 539667072,
+      "step": 3142
+    },
+    {
+      "epoch": 0.8264614979943448,
+      "loss": 0.1312233954668045,
+      "loss_ce": 0.0005166015471331775,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 539667072,
+      "step": 3142
+    },
+    {
+      "epoch": 0.8267245347537319,
+      "grad_norm": 12.104159402624534,
+      "learning_rate": 5e-06,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 539839460,
+      "step": 3143
+    },
+    {
+      "epoch": 0.8267245347537319,
+      "loss": 0.2201562076807022,
+      "loss_ce": 0.002535369014367461,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 539839460,
+      "step": 3143
+    },
+    {
+      "epoch": 0.826987571513119,
+      "grad_norm": 5.4839208249491245,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 540011628,
+      "step": 3144
+    },
+    {
+      "epoch": 0.826987571513119,
+      "loss": 0.1272164285182953,
+      "loss_ce": 0.0027352366596460342,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 540011628,
+      "step": 3144
+    },
+    {
+      "epoch": 0.8272506082725061,
+      "grad_norm": 4.535384525117057,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 540183608,
+      "step": 3145
+    },
+    {
+      "epoch": 0.8272506082725061,
+      "loss": 0.0952780544757843,
+      "loss_ce": 0.002458844566717744,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 540183608,
+      "step": 3145
+    },
+    {
+      "epoch": 0.8275136450318932,
+      "grad_norm": 4.264720708193596,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 540355724,
+      "step": 3146
+    },
+    {
+      "epoch": 0.8275136450318932,
+      "loss": 0.05311005562543869,
+      "loss_ce": 0.0009097411530092359,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 540355724,
+      "step": 3146
+    },
+    {
+      "epoch": 0.8277766817912803,
+      "grad_norm": 6.253031178644263,
+      "learning_rate": 5e-06,
+      "loss": 0.1647,
+      "num_input_tokens_seen": 540526424,
+      "step": 3147
+    },
+    {
+      "epoch": 0.8277766817912803,
+      "loss": 0.1789083182811737,
+      "loss_ce": 0.0011586775071918964,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 540526424,
+      "step": 3147
+    },
+    {
+      "epoch": 0.8280397185506675,
+      "grad_norm": 4.716126774756702,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 540698736,
+      "step": 3148
+    },
+    {
+      "epoch": 0.8280397185506675,
+      "loss": 0.053756728768348694,
+      "loss_ce": 0.0007171769393607974,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 540698736,
+      "step": 3148
+    },
+    {
+      "epoch": 0.8283027553100546,
+      "grad_norm": 15.946573913857675,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 540870876,
+      "step": 3149
+    },
+    {
+      "epoch": 0.8283027553100546,
+      "loss": 0.11821180582046509,
+      "loss_ce": 0.001833026995882392,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 540870876,
+      "step": 3149
+    },
+    {
+      "epoch": 0.8285657920694417,
+      "grad_norm": 3.1443507951345406,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 541041144,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8285657920694417,
+      "loss": 0.13489758968353271,
+      "loss_ce": 0.004984267987310886,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 541041144,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8288288288288288,
+      "grad_norm": 4.1136952366915605,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 541213632,
+      "step": 3151
+    },
+    {
+      "epoch": 0.8288288288288288,
+      "loss": 0.10275200009346008,
+      "loss_ce": 0.0014183830935508013,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 541213632,
+      "step": 3151
+    },
+    {
+      "epoch": 0.8290918655882159,
+      "grad_norm": 5.711362515317036,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 541384104,
+      "step": 3152
+    },
+    {
+      "epoch": 0.8290918655882159,
+      "loss": 0.08564618229866028,
+      "loss_ce": 0.0014176733093336225,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 541384104,
+      "step": 3152
+    },
+    {
+      "epoch": 0.8293549023476031,
+      "grad_norm": 7.096186366509263,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 541556240,
+      "step": 3153
+    },
+    {
+      "epoch": 0.8293549023476031,
+      "loss": 0.17091956734657288,
+      "loss_ce": 0.0015012390213087201,
+      "loss_iou": 0.421875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 541556240,
+      "step": 3153
+    },
+    {
+      "epoch": 0.8296179391069902,
+      "grad_norm": 4.29702608695741,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 541728356,
+      "step": 3154
+    },
+    {
+      "epoch": 0.8296179391069902,
+      "loss": 0.10837417840957642,
+      "loss_ce": 0.0006623809458687901,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 541728356,
+      "step": 3154
+    },
+    {
+      "epoch": 0.8298809758663773,
+      "grad_norm": 3.729285456687102,
+      "learning_rate": 5e-06,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 541898716,
+      "step": 3155
+    },
+    {
+      "epoch": 0.8298809758663773,
+      "loss": 0.062001317739486694,
+      "loss_ce": 0.0011492683552205563,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 541898716,
+      "step": 3155
+    },
+    {
+      "epoch": 0.8301440126257644,
+      "grad_norm": 6.543104796714217,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 542070820,
+      "step": 3156
+    },
+    {
+      "epoch": 0.8301440126257644,
+      "loss": 0.09034896641969681,
+      "loss_ce": 0.0022752326913177967,
+      "loss_iou": 0.390625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 542070820,
+      "step": 3156
+    },
+    {
+      "epoch": 0.8304070493851515,
+      "grad_norm": 3.814660695434685,
+      "learning_rate": 5e-06,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 542242936,
+      "step": 3157
+    },
+    {
+      "epoch": 0.8304070493851515,
+      "loss": 0.05618397891521454,
+      "loss_ce": 0.0004894005251117051,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 542242936,
+      "step": 3157
+    },
+    {
+      "epoch": 0.8306700861445387,
+      "grad_norm": 13.94223132380543,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 542415376,
+      "step": 3158
+    },
+    {
+      "epoch": 0.8306700861445387,
+      "loss": 0.09673337638378143,
+      "loss_ce": 0.0017016411293298006,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 542415376,
+      "step": 3158
+    },
+    {
+      "epoch": 0.8309331229039258,
+      "grad_norm": 46.47454058969395,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 542587700,
+      "step": 3159
+    },
+    {
+      "epoch": 0.8309331229039258,
+      "loss": 0.13622400164604187,
+      "loss_ce": 0.00045128766214475036,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 542587700,
+      "step": 3159
+    },
+    {
+      "epoch": 0.831196159663313,
+      "grad_norm": 7.76157996882612,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 542758148,
+      "step": 3160
+    },
+    {
+      "epoch": 0.831196159663313,
+      "loss": 0.11631490290164948,
+      "loss_ce": 0.004040728323161602,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 542758148,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8314591964227,
+      "grad_norm": 20.483608511309008,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 542930836,
+      "step": 3161
+    },
+    {
+      "epoch": 0.8314591964227,
+      "loss": 0.06239602342247963,
+      "loss_ce": 0.00014016299974173307,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 542930836,
+      "step": 3161
+    },
+    {
+      "epoch": 0.8317222331820872,
+      "grad_norm": 4.545805831684384,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 543102800,
+      "step": 3162
+    },
+    {
+      "epoch": 0.8317222331820872,
+      "loss": 0.12887202203273773,
+      "loss_ce": 0.0006066488567739725,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 543102800,
+      "step": 3162
+    },
+    {
+      "epoch": 0.8319852699414744,
+      "grad_norm": 6.107575063227538,
+      "learning_rate": 5e-06,
+      "loss": 0.1,
+      "num_input_tokens_seen": 543274984,
+      "step": 3163
+    },
+    {
+      "epoch": 0.8319852699414744,
+      "loss": 0.08694491535425186,
+      "loss_ce": 0.00036654339055530727,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 543274984,
+      "step": 3163
+    },
+    {
+      "epoch": 0.8322483067008615,
+      "grad_norm": 5.535589838242773,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 543447000,
+      "step": 3164
+    },
+    {
+      "epoch": 0.8322483067008615,
+      "loss": 0.12083543837070465,
+      "loss_ce": 0.006333490367978811,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 543447000,
+      "step": 3164
+    },
+    {
+      "epoch": 0.8325113434602486,
+      "grad_norm": 5.113582833614297,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 543619120,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8325113434602486,
+      "loss": 0.1699899584054947,
+      "loss_ce": 0.002250079531222582,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 543619120,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8327743802196357,
+      "grad_norm": 5.479089729021323,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 543791308,
+      "step": 3166
+    },
+    {
+      "epoch": 0.8327743802196357,
+      "loss": 0.08266595751047134,
+      "loss_ce": 0.00042108428897336125,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 543791308,
+      "step": 3166
+    },
+    {
+      "epoch": 0.8330374169790228,
+      "grad_norm": 4.786501058802981,
+      "learning_rate": 5e-06,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 543958632,
+      "step": 3167
+    },
+    {
+      "epoch": 0.8330374169790228,
+      "loss": 0.07515060901641846,
+      "loss_ce": 0.0012675554025918245,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 543958632,
+      "step": 3167
+    },
+    {
+      "epoch": 0.83330045373841,
+      "grad_norm": 8.594141920355682,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 544130848,
+      "step": 3168
+    },
+    {
+      "epoch": 0.83330045373841,
+      "loss": 0.09625629335641861,
+      "loss_ce": 0.0025978446938097477,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 544130848,
+      "step": 3168
+    },
+    {
+      "epoch": 0.8335634904977971,
+      "grad_norm": 3.7461805856215458,
+      "learning_rate": 5e-06,
+      "loss": 0.1468,
+      "num_input_tokens_seen": 544302820,
+      "step": 3169
+    },
+    {
+      "epoch": 0.8335634904977971,
+      "loss": 0.09222942590713501,
+      "loss_ce": 0.0022025699727237225,
+      "loss_iou": 0.375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 544302820,
+      "step": 3169
+    },
+    {
+      "epoch": 0.8338265272571842,
+      "grad_norm": 9.92373943246775,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 544474904,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8338265272571842,
+      "loss": 0.14077287912368774,
+      "loss_ce": 0.0030775703489780426,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 544474904,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8340895640165713,
+      "grad_norm": 14.235058865758765,
+      "learning_rate": 5e-06,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 544647036,
+      "step": 3171
+    },
+    {
+      "epoch": 0.8340895640165713,
+      "loss": 0.060213349759578705,
+      "loss_ce": 0.0022452091798186302,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 544647036,
+      "step": 3171
+    },
+    {
+      "epoch": 0.8343526007759584,
+      "grad_norm": 4.48891217014488,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 544817552,
+      "step": 3172
+    },
+    {
+      "epoch": 0.8343526007759584,
+      "loss": 0.1579454094171524,
+      "loss_ce": 0.0024888694752007723,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 544817552,
+      "step": 3172
+    },
+    {
+      "epoch": 0.8346156375353455,
+      "grad_norm": 3.708111692136732,
+      "learning_rate": 5e-06,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 544989748,
+      "step": 3173
+    },
+    {
+      "epoch": 0.8346156375353455,
+      "loss": 0.14915120601654053,
+      "loss_ce": 0.0004848288372159004,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 544989748,
+      "step": 3173
+    },
+    {
+      "epoch": 0.8348786742947327,
+      "grad_norm": 6.783609066408303,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 545161788,
+      "step": 3174
+    },
+    {
+      "epoch": 0.8348786742947327,
+      "loss": 0.13735045492649078,
+      "loss_ce": 0.003683460643514991,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 545161788,
+      "step": 3174
+    },
+    {
+      "epoch": 0.8351417110541198,
+      "grad_norm": 4.8955030087080775,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 545333884,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8351417110541198,
+      "loss": 0.08379638940095901,
+      "loss_ce": 0.0006665037362836301,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 545333884,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8354047478135069,
+      "grad_norm": 4.39565885197116,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 545506244,
+      "step": 3176
+    },
+    {
+      "epoch": 0.8354047478135069,
+      "loss": 0.12928782403469086,
+      "loss_ce": 0.000915632932446897,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 545506244,
+      "step": 3176
+    },
+    {
+      "epoch": 0.835667784572894,
+      "grad_norm": 6.2304214245063605,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 545678348,
+      "step": 3177
+    },
+    {
+      "epoch": 0.835667784572894,
+      "loss": 0.07726402580738068,
+      "loss_ce": 0.000832755584269762,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 545678348,
+      "step": 3177
+    },
+    {
+      "epoch": 0.8359308213322811,
+      "grad_norm": 4.950850483708038,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 545848136,
+      "step": 3178
+    },
+    {
+      "epoch": 0.8359308213322811,
+      "loss": 0.12010537087917328,
+      "loss_ce": 0.0009342365083284676,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 545848136,
+      "step": 3178
+    },
+    {
+      "epoch": 0.8361938580916684,
+      "grad_norm": 4.252976007842196,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 546020692,
+      "step": 3179
+    },
+    {
+      "epoch": 0.8361938580916684,
+      "loss": 0.12632903456687927,
+      "loss_ce": 0.00018462821026332676,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 546020692,
+      "step": 3179
+    },
+    {
+      "epoch": 0.8364568948510555,
+      "grad_norm": 16.684582759145172,
+      "learning_rate": 5e-06,
+      "loss": 0.073,
+      "num_input_tokens_seen": 546192760,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8364568948510555,
+      "loss": 0.09913001954555511,
+      "loss_ce": 0.0009549736278131604,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 546192760,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8367199316104426,
+      "grad_norm": 5.017020175241404,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 546363064,
+      "step": 3181
+    },
+    {
+      "epoch": 0.8367199316104426,
+      "loss": 0.1710553914308548,
+      "loss_ce": 0.00040109228575602174,
+      "loss_iou": NaN,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 546363064,
+      "step": 3181
+    },
+    {
+      "epoch": 0.8369829683698297,
+      "grad_norm": 17.167559763411923,
+      "learning_rate": 5e-06,
+      "loss": 0.1342,
+      "num_input_tokens_seen": 546535208,
+      "step": 3182
+    },
+    {
+      "epoch": 0.8369829683698297,
+      "loss": 0.15842683613300323,
+      "loss_ce": 0.0015664853854104877,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 546535208,
+      "step": 3182
+    },
+    {
+      "epoch": 0.8372460051292168,
+      "grad_norm": 3.9098429704759723,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 546707288,
+      "step": 3183
+    },
+    {
+      "epoch": 0.8372460051292168,
+      "loss": 0.08958999812602997,
+      "loss_ce": 0.00014297313464339823,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 546707288,
+      "step": 3183
+    },
+    {
+      "epoch": 0.837509041888604,
+      "grad_norm": 5.336892331569848,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 546879532,
+      "step": 3184
+    },
+    {
+      "epoch": 0.837509041888604,
+      "loss": 0.07368629425764084,
+      "loss_ce": 0.004197763279080391,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 546879532,
+      "step": 3184
+    },
+    {
+      "epoch": 0.8377720786479911,
+      "grad_norm": 3.9486138565288265,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 547051612,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8377720786479911,
+      "loss": 0.15984772145748138,
+      "loss_ce": 0.0009426883771084249,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 547051612,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8380351154073782,
+      "grad_norm": 6.643098269759205,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 547223768,
+      "step": 3186
+    },
+    {
+      "epoch": 0.8380351154073782,
+      "loss": 0.1492346227169037,
+      "loss_ce": 0.0017126407474279404,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 547223768,
+      "step": 3186
+    },
+    {
+      "epoch": 0.8382981521667653,
+      "grad_norm": 30.69973998174967,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 547395788,
+      "step": 3187
+    },
+    {
+      "epoch": 0.8382981521667653,
+      "loss": 0.13615994155406952,
+      "loss_ce": 0.00041775350109674037,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 547395788,
+      "step": 3187
+    },
+    {
+      "epoch": 0.8385611889261524,
+      "grad_norm": 35.72025172038926,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 547568000,
+      "step": 3188
+    },
+    {
+      "epoch": 0.8385611889261524,
+      "loss": 0.04795503616333008,
+      "loss_ce": 0.0002560606808401644,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 547568000,
+      "step": 3188
+    },
+    {
+      "epoch": 0.8388242256855396,
+      "grad_norm": 7.18874848401809,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 547740064,
+      "step": 3189
+    },
+    {
+      "epoch": 0.8388242256855396,
+      "loss": 0.12481731176376343,
+      "loss_ce": 0.0005802565719932318,
+      "loss_iou": 0.40625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 547740064,
+      "step": 3189
+    },
+    {
+      "epoch": 0.8390872624449267,
+      "grad_norm": 6.155592453443207,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 547912300,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8390872624449267,
+      "loss": 0.08178332448005676,
+      "loss_ce": 0.0008201911114156246,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 547912300,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8393502992043138,
+      "grad_norm": 12.72945010384761,
+      "learning_rate": 5e-06,
+      "loss": 0.139,
+      "num_input_tokens_seen": 548084548,
+      "step": 3191
+    },
+    {
+      "epoch": 0.8393502992043138,
+      "loss": 0.19623327255249023,
+      "loss_ce": 0.002812865423038602,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 548084548,
+      "step": 3191
+    },
+    {
+      "epoch": 0.8396133359637009,
+      "grad_norm": 5.353324449071839,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 548255176,
+      "step": 3192
+    },
+    {
+      "epoch": 0.8396133359637009,
+      "loss": 0.12437019497156143,
+      "loss_ce": 0.0003467575879767537,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 548255176,
+      "step": 3192
+    },
+    {
+      "epoch": 0.839876372723088,
+      "grad_norm": 4.347588948765136,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 548427488,
+      "step": 3193
+    },
+    {
+      "epoch": 0.839876372723088,
+      "loss": 0.08202692121267319,
+      "loss_ce": 0.0022234548814594746,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 548427488,
+      "step": 3193
+    },
+    {
+      "epoch": 0.8401394094824752,
+      "grad_norm": 5.210446302172759,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 548599828,
+      "step": 3194
+    },
+    {
+      "epoch": 0.8401394094824752,
+      "loss": 0.12387488037347794,
+      "loss_ce": 0.00018713258032221347,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 548599828,
+      "step": 3194
+    },
+    {
+      "epoch": 0.8404024462418623,
+      "grad_norm": 3.9642883700177514,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 548771932,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8404024462418623,
+      "loss": 0.12389804422855377,
+      "loss_ce": 0.00396396778523922,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 548771932,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8406654830012494,
+      "grad_norm": 4.444986430362012,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 548944332,
+      "step": 3196
+    },
+    {
+      "epoch": 0.8406654830012494,
+      "loss": 0.07641720026731491,
+      "loss_ce": 0.0017254289705306292,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 548944332,
+      "step": 3196
+    },
+    {
+      "epoch": 0.8409285197606365,
+      "grad_norm": 17.152304646652222,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 549116460,
+      "step": 3197
+    },
+    {
+      "epoch": 0.8409285197606365,
+      "loss": 0.07439431548118591,
+      "loss_ce": 0.0016404138877987862,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 549116460,
+      "step": 3197
+    },
+    {
+      "epoch": 0.8411915565200236,
+      "grad_norm": 6.835966777853977,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 549286920,
+      "step": 3198
+    },
+    {
+      "epoch": 0.8411915565200236,
+      "loss": 0.23670437932014465,
+      "loss_ce": 0.0033669895492494106,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 549286920,
+      "step": 3198
+    },
+    {
+      "epoch": 0.8414545932794107,
+      "grad_norm": 3.334370051059133,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 549458932,
+      "step": 3199
+    },
+    {
+      "epoch": 0.8414545932794107,
+      "loss": 0.10936430096626282,
+      "loss_ce": 0.001118454267270863,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 549458932,
+      "step": 3199
+    },
+    {
+      "epoch": 0.841717630038798,
+      "grad_norm": 3.3887650622610863,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 549630896,
+      "step": 3200
+    },
+    {
+      "epoch": 0.841717630038798,
+      "loss": 0.1543048769235611,
+      "loss_ce": 0.00280036055482924,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 549630896,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8419806667981851,
+      "grad_norm": 8.90667441822763,
+      "learning_rate": 5e-06,
+      "loss": 0.1431,
+      "num_input_tokens_seen": 549802980,
+      "step": 3201
+    },
+    {
+      "epoch": 0.8419806667981851,
+      "loss": 0.10668284446001053,
+      "loss_ce": 0.0037165414541959763,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 549802980,
+      "step": 3201
+    },
+    {
+      "epoch": 0.8422437035575722,
+      "grad_norm": 4.581013999909635,
+      "learning_rate": 5e-06,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 549973288,
+      "step": 3202
+    },
+    {
+      "epoch": 0.8422437035575722,
+      "loss": 0.05071475729346275,
+      "loss_ce": 0.0011084338184446096,
+      "loss_iou": 0.578125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 549973288,
+      "step": 3202
+    },
+    {
+      "epoch": 0.8425067403169593,
+      "grad_norm": 7.123665679875533,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 550143700,
+      "step": 3203
+    },
+    {
+      "epoch": 0.8425067403169593,
+      "loss": 0.1053222045302391,
+      "loss_ce": 0.0030272852163761854,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 550143700,
+      "step": 3203
+    },
+    {
+      "epoch": 0.8427697770763464,
+      "grad_norm": 14.86581378718403,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 550315752,
+      "step": 3204
+    },
+    {
+      "epoch": 0.8427697770763464,
+      "loss": 0.08674832433462143,
+      "loss_ce": 0.000688750937115401,
+      "loss_iou": NaN,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 550315752,
+      "step": 3204
+    },
+    {
+      "epoch": 0.8430328138357336,
+      "grad_norm": 15.15255979041917,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 550487764,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8430328138357336,
+      "loss": 0.11826883256435394,
+      "loss_ce": 0.001401771791279316,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 550487764,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8432958505951207,
+      "grad_norm": 5.341317771181548,
+      "learning_rate": 5e-06,
+      "loss": 0.1085,
+      "num_input_tokens_seen": 550659696,
+      "step": 3206
+    },
+    {
+      "epoch": 0.8432958505951207,
+      "loss": 0.11918849498033524,
+      "loss_ce": 0.0002920094411820173,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 550659696,
+      "step": 3206
+    },
+    {
+      "epoch": 0.8435588873545078,
+      "grad_norm": 5.395123389791678,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 550831904,
+      "step": 3207
+    },
+    {
+      "epoch": 0.8435588873545078,
+      "loss": 0.07498294115066528,
+      "loss_ce": 0.001038852147758007,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 550831904,
+      "step": 3207
+    },
+    {
+      "epoch": 0.8438219241138949,
+      "grad_norm": 3.643785530851578,
+      "learning_rate": 5e-06,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 551004156,
+      "step": 3208
+    },
+    {
+      "epoch": 0.8438219241138949,
+      "loss": 0.06928001344203949,
+      "loss_ce": 0.004369123373180628,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 551004156,
+      "step": 3208
+    },
+    {
+      "epoch": 0.844084960873282,
+      "grad_norm": 5.391792901404867,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 551176320,
+      "step": 3209
+    },
+    {
+      "epoch": 0.844084960873282,
+      "loss": 0.0887857973575592,
+      "loss_ce": 0.0004984364495612681,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 551176320,
+      "step": 3209
+    },
+    {
+      "epoch": 0.8443479976326692,
+      "grad_norm": 7.538916165469519,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 551348168,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8443479976326692,
+      "loss": 0.16849397122859955,
+      "loss_ce": 0.0030734348110854626,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 551348168,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8446110343920563,
+      "grad_norm": 7.685615482077233,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 551520728,
+      "step": 3211
+    },
+    {
+      "epoch": 0.8446110343920563,
+      "loss": 0.0783834308385849,
+      "loss_ce": 0.001479133265092969,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 551520728,
+      "step": 3211
+    },
+    {
+      "epoch": 0.8448740711514434,
+      "grad_norm": 8.632575689436024,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 551692804,
+      "step": 3212
+    },
+    {
+      "epoch": 0.8448740711514434,
+      "loss": 0.15737096965312958,
+      "loss_ce": 0.0006174240261316299,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 551692804,
+      "step": 3212
+    },
+    {
+      "epoch": 0.8451371079108305,
+      "grad_norm": 12.457163123139722,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 551864880,
+      "step": 3213
+    },
+    {
+      "epoch": 0.8451371079108305,
+      "loss": 0.07481381297111511,
+      "loss_ce": 0.003173800650984049,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 551864880,
+      "step": 3213
+    },
+    {
+      "epoch": 0.8454001446702176,
+      "grad_norm": 3.8683898896692166,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 552036812,
+      "step": 3214
+    },
+    {
+      "epoch": 0.8454001446702176,
+      "loss": 0.05259804055094719,
+      "loss_ce": 0.0005503093125298619,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 552036812,
+      "step": 3214
+    },
+    {
+      "epoch": 0.8456631814296048,
+      "grad_norm": 4.539951161935164,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 552208856,
+      "step": 3215
+    },
+    {
+      "epoch": 0.8456631814296048,
+      "loss": 0.07421931624412537,
+      "loss_ce": 0.0011754983570426702,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 552208856,
+      "step": 3215
+    },
+    {
+      "epoch": 0.8459262181889919,
+      "grad_norm": 10.573780213794782,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 552380904,
+      "step": 3216
+    },
+    {
+      "epoch": 0.8459262181889919,
+      "loss": 0.06666961312294006,
+      "loss_ce": 0.001804507803171873,
+      "loss_iou": 0.484375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 552380904,
+      "step": 3216
+    },
+    {
+      "epoch": 0.846189254948379,
+      "grad_norm": 4.275878567324078,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 552551132,
+      "step": 3217
+    },
+    {
+      "epoch": 0.846189254948379,
+      "loss": 0.09229104220867157,
+      "loss_ce": 0.0006162393838167191,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 552551132,
+      "step": 3217
+    },
+    {
+      "epoch": 0.8464522917077661,
+      "grad_norm": 4.658023344353002,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 552722824,
+      "step": 3218
+    },
+    {
+      "epoch": 0.8464522917077661,
+      "loss": 0.07020144164562225,
+      "loss_ce": 0.0020709503442049026,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 552722824,
+      "step": 3218
+    },
+    {
+      "epoch": 0.8467153284671532,
+      "grad_norm": 6.1002348516078095,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 552894948,
+      "step": 3219
+    },
+    {
+      "epoch": 0.8467153284671532,
+      "loss": 0.10183661431074142,
+      "loss_ce": 0.0007318751304410398,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 552894948,
+      "step": 3219
+    },
+    {
+      "epoch": 0.8469783652265405,
+      "grad_norm": 3.6957881460841646,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 553065492,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8469783652265405,
+      "loss": 0.20681847631931305,
+      "loss_ce": 0.005616086535155773,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 553065492,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8472414019859276,
+      "grad_norm": 31.1352481206466,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 553237384,
+      "step": 3221
+    },
+    {
+      "epoch": 0.8472414019859276,
+      "loss": 0.0710492879152298,
+      "loss_ce": 0.0004316139966249466,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 553237384,
+      "step": 3221
+    },
+    {
+      "epoch": 0.8475044387453147,
+      "grad_norm": 15.822969391935775,
+      "learning_rate": 5e-06,
+      "loss": 0.1572,
+      "num_input_tokens_seen": 553409788,
+      "step": 3222
+    },
+    {
+      "epoch": 0.8475044387453147,
+      "loss": 0.24513369798660278,
+      "loss_ce": 0.001847569365054369,
+      "loss_iou": NaN,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 553409788,
+      "step": 3222
+    },
+    {
+      "epoch": 0.8477674755047018,
+      "grad_norm": 4.903880792540357,
+      "learning_rate": 5e-06,
+      "loss": 0.082,
+      "num_input_tokens_seen": 553581532,
+      "step": 3223
+    },
+    {
+      "epoch": 0.8477674755047018,
+      "loss": 0.0573626384139061,
+      "loss_ce": 0.0015459894202649593,
+      "loss_iou": NaN,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 553581532,
+      "step": 3223
+    },
+    {
+      "epoch": 0.8480305122640889,
+      "grad_norm": 11.193676890263411,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 553753504,
+      "step": 3224
+    },
+    {
+      "epoch": 0.8480305122640889,
+      "loss": 0.12998200953006744,
+      "loss_ce": 0.0016861144686117768,
+      "loss_iou": 0.515625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 553753504,
+      "step": 3224
+    },
+    {
+      "epoch": 0.848293549023476,
+      "grad_norm": 6.892798676870787,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 553925232,
+      "step": 3225
+    },
+    {
+      "epoch": 0.848293549023476,
+      "loss": 0.06351655721664429,
+      "loss_ce": 0.0010470744455233216,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 553925232,
+      "step": 3225
+    },
+    {
+      "epoch": 0.8485565857828632,
+      "grad_norm": 3.94134933091549,
+      "learning_rate": 5e-06,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 554097380,
+      "step": 3226
+    },
+    {
+      "epoch": 0.8485565857828632,
+      "loss": 0.06772696226835251,
+      "loss_ce": 0.00040518559399060905,
+      "loss_iou": 0.5,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 554097380,
+      "step": 3226
+    },
+    {
+      "epoch": 0.8488196225422503,
+      "grad_norm": 4.553319445567292,
+      "learning_rate": 5e-06,
+      "loss": 0.1467,
+      "num_input_tokens_seen": 554269324,
+      "step": 3227
+    },
+    {
+      "epoch": 0.8488196225422503,
+      "loss": 0.06106797605752945,
+      "loss_ce": 0.00015489489305764437,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 554269324,
+      "step": 3227
+    },
+    {
+      "epoch": 0.8490826593016374,
+      "grad_norm": 8.122842269719726,
+      "learning_rate": 5e-06,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 554441420,
+      "step": 3228
+    },
+    {
+      "epoch": 0.8490826593016374,
+      "loss": 0.07994222640991211,
+      "loss_ce": 0.0005202332977205515,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 554441420,
+      "step": 3228
+    },
+    {
+      "epoch": 0.8493456960610245,
+      "grad_norm": 4.946244650717768,
+      "learning_rate": 5e-06,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 554613348,
+      "step": 3229
+    },
+    {
+      "epoch": 0.8493456960610245,
+      "loss": 0.07239595800638199,
+      "loss_ce": 0.0015646612737327814,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 554613348,
+      "step": 3229
+    },
+    {
+      "epoch": 0.8496087328204116,
+      "grad_norm": 10.327431498521296,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 554785544,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8496087328204116,
+      "loss": 0.1082451343536377,
+      "loss_ce": 0.001616714522242546,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 554785544,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8498717695797988,
+      "grad_norm": 6.2381116819087605,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 554957700,
+      "step": 3231
+    },
+    {
+      "epoch": 0.8498717695797988,
+      "loss": 0.05288837477564812,
+      "loss_ce": 0.0004286564071662724,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 554957700,
+      "step": 3231
+    },
+    {
+      "epoch": 0.8501348063391859,
+      "grad_norm": 5.665331437954715,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 555128400,
+      "step": 3232
+    },
+    {
+      "epoch": 0.8501348063391859,
+      "loss": 0.17892731726169586,
+      "loss_ce": 0.008456122130155563,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 555128400,
+      "step": 3232
+    },
+    {
+      "epoch": 0.850397843098573,
+      "grad_norm": 4.02543856443553,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 555300712,
+      "step": 3233
+    },
+    {
+      "epoch": 0.850397843098573,
+      "loss": 0.1149306371808052,
+      "loss_ce": 0.0023512912448495626,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 555300712,
+      "step": 3233
+    },
+    {
+      "epoch": 0.8506608798579601,
+      "grad_norm": 4.049090978454337,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 555472756,
+      "step": 3234
+    },
+    {
+      "epoch": 0.8506608798579601,
+      "loss": 0.14269746840000153,
+      "loss_ce": 0.0037509393878281116,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 555472756,
+      "step": 3234
+    },
+    {
+      "epoch": 0.8509239166173472,
+      "grad_norm": 4.24693988860905,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 555644856,
+      "step": 3235
+    },
+    {
+      "epoch": 0.8509239166173472,
+      "loss": 0.06741193681955338,
+      "loss_ce": 0.0012955997372046113,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 555644856,
+      "step": 3235
+    },
+    {
+      "epoch": 0.8511869533767344,
+      "grad_norm": 4.223553863328503,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 555817488,
+      "step": 3236
+    },
+    {
+      "epoch": 0.8511869533767344,
+      "loss": 0.09499558061361313,
+      "loss_ce": 0.0007572993636131287,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 555817488,
+      "step": 3236
+    },
+    {
+      "epoch": 0.8514499901361215,
+      "grad_norm": 8.070861336660833,
+      "learning_rate": 5e-06,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 555989844,
+      "step": 3237
+    },
+    {
+      "epoch": 0.8514499901361215,
+      "loss": 0.09148094058036804,
+      "loss_ce": 0.0017287411028519273,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 555989844,
+      "step": 3237
+    },
+    {
+      "epoch": 0.8517130268955087,
+      "grad_norm": 4.7628320746905235,
+      "learning_rate": 5e-06,
+      "loss": 0.13,
+      "num_input_tokens_seen": 556162144,
+      "step": 3238
+    },
+    {
+      "epoch": 0.8517130268955087,
+      "loss": 0.08755885809659958,
+      "loss_ce": 0.001667136326432228,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 556162144,
+      "step": 3238
+    },
+    {
+      "epoch": 0.8519760636548958,
+      "grad_norm": 8.201240707199219,
+      "learning_rate": 5e-06,
+      "loss": 0.068,
+      "num_input_tokens_seen": 556334020,
+      "step": 3239
+    },
+    {
+      "epoch": 0.8519760636548958,
+      "loss": 0.06216670200228691,
+      "loss_ce": 0.001772413495928049,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 556334020,
+      "step": 3239
+    },
+    {
+      "epoch": 0.8522391004142829,
+      "grad_norm": 5.1836633623539266,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 556502704,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8522391004142829,
+      "loss": 0.09547331184148788,
+      "loss_ce": 0.0012350315228104591,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 556502704,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8525021371736701,
+      "grad_norm": 4.1790455766130234,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 556674848,
+      "step": 3241
+    },
+    {
+      "epoch": 0.8525021371736701,
+      "loss": 0.15765714645385742,
+      "loss_ce": 0.00018642976647242904,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 556674848,
+      "step": 3241
+    },
+    {
+      "epoch": 0.8527651739330572,
+      "grad_norm": 32.42085523655999,
+      "learning_rate": 5e-06,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 556847120,
+      "step": 3242
+    },
+    {
+      "epoch": 0.8527651739330572,
+      "loss": 0.10435596853494644,
+      "loss_ce": 0.0030986424535512924,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 556847120,
+      "step": 3242
+    },
+    {
+      "epoch": 0.8530282106924443,
+      "grad_norm": 3.085990680432512,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 557015380,
+      "step": 3243
+    },
+    {
+      "epoch": 0.8530282106924443,
+      "loss": 0.12935911118984222,
+      "loss_ce": 0.0042217751033604145,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 557015380,
+      "step": 3243
+    },
+    {
+      "epoch": 0.8532912474518314,
+      "grad_norm": 5.554407683422717,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 557186020,
+      "step": 3244
+    },
+    {
+      "epoch": 0.8532912474518314,
+      "loss": 0.12462737411260605,
+      "loss_ce": 0.002557065337896347,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 557186020,
+      "step": 3244
+    },
+    {
+      "epoch": 0.8535542842112185,
+      "grad_norm": 77.39805630943022,
+      "learning_rate": 5e-06,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 557358056,
+      "step": 3245
+    },
+    {
+      "epoch": 0.8535542842112185,
+      "loss": 0.06247711926698685,
+      "loss_ce": 0.0029983618296682835,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 557358056,
+      "step": 3245
+    },
+    {
+      "epoch": 0.8538173209706057,
+      "grad_norm": 6.568132581838013,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 557530164,
+      "step": 3246
+    },
+    {
+      "epoch": 0.8538173209706057,
+      "loss": 0.10226649791002274,
+      "loss_ce": 0.0013143508695065975,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 557530164,
+      "step": 3246
+    },
+    {
+      "epoch": 0.8540803577299928,
+      "grad_norm": 8.945021217700889,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 557702352,
+      "step": 3247
+    },
+    {
+      "epoch": 0.8540803577299928,
+      "loss": 0.14076803624629974,
+      "loss_ce": 0.002218232722952962,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 557702352,
+      "step": 3247
+    },
+    {
+      "epoch": 0.8543433944893799,
+      "grad_norm": 5.970819800467794,
+      "learning_rate": 5e-06,
+      "loss": 0.113,
+      "num_input_tokens_seen": 557874756,
+      "step": 3248
+    },
+    {
+      "epoch": 0.8543433944893799,
+      "loss": 0.12330596148967743,
+      "loss_ce": 0.002425831276923418,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 557874756,
+      "step": 3248
+    },
+    {
+      "epoch": 0.854606431248767,
+      "grad_norm": 12.952896902366254,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 558045276,
+      "step": 3249
+    },
+    {
+      "epoch": 0.854606431248767,
+      "loss": 0.15818831324577332,
+      "loss_ce": 0.0018772899638861418,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 558045276,
+      "step": 3249
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "grad_norm": 10.077522334681213,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_websight_new_CIoU": 0.8994152843952179,
+      "eval_websight_new_GIoU": 0.9002452492713928,
+      "eval_websight_new_IoU": 0.9029170572757721,
+      "eval_websight_new_MAE_all": 0.014043471310287714,
+      "eval_websight_new_MAE_h": 0.007522843778133392,
+      "eval_websight_new_MAE_w": 0.020659465342760086,
+      "eval_websight_new_MAE_x": 0.021266265772283077,
+      "eval_websight_new_MAE_y": 0.006725311512127519,
+      "eval_websight_new_NUM_probability": 0.9999923408031464,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.07195720076560974,
+      "eval_websight_new_loss_ce": 4.601216005539754e-06,
+      "eval_websight_new_loss_iou": 0.31280517578125,
+      "eval_websight_new_loss_num": 0.012767791748046875,
+      "eval_websight_new_loss_xval": 0.06391143798828125,
+      "eval_websight_new_runtime": 56.9054,
+      "eval_websight_new_samples_per_second": 0.879,
+      "eval_websight_new_steps_per_second": 0.035,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_seeclick_CIoU": 0.6642794907093048,
+      "eval_seeclick_GIoU": 0.6613934338092804,
+      "eval_seeclick_IoU": 0.6828196048736572,
+      "eval_seeclick_MAE_all": 0.04043097607791424,
+      "eval_seeclick_MAE_h": 0.023812726140022278,
+      "eval_seeclick_MAE_w": 0.055972687900066376,
+      "eval_seeclick_MAE_x": 0.05744660459458828,
+      "eval_seeclick_MAE_y": 0.02449188195168972,
+      "eval_seeclick_NUM_probability": 0.999969094991684,
+      "eval_seeclick_inside_bbox": 0.9375,
+      "eval_seeclick_loss": 0.1842024326324463,
+      "eval_seeclick_loss_ce": 0.008998175617307425,
+      "eval_seeclick_loss_iou": 0.4769287109375,
+      "eval_seeclick_loss_num": 0.0336456298828125,
+      "eval_seeclick_loss_xval": 0.1681365966796875,
+      "eval_seeclick_runtime": 76.1272,
+      "eval_seeclick_samples_per_second": 0.565,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_icons_CIoU": 0.8727431297302246,
+      "eval_icons_GIoU": 0.8711867332458496,
+      "eval_icons_IoU": 0.8781991004943848,
+      "eval_icons_MAE_all": 0.017583131790161133,
+      "eval_icons_MAE_h": 0.019472193904221058,
+      "eval_icons_MAE_w": 0.015529958996921778,
+      "eval_icons_MAE_x": 0.015652839560061693,
+      "eval_icons_MAE_y": 0.019677532836794853,
+      "eval_icons_NUM_probability": 0.9999885261058807,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.06516695767641068,
+      "eval_icons_loss_ce": 9.223055712936912e-06,
+      "eval_icons_loss_iou": 0.504638671875,
+      "eval_icons_loss_num": 0.012132644653320312,
+      "eval_icons_loss_xval": 0.0607147216796875,
+      "eval_icons_runtime": 94.9537,
+      "eval_icons_samples_per_second": 0.527,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_screenspot_CIoU": 0.6023598512013754,
+      "eval_screenspot_GIoU": 0.5998436013857523,
+      "eval_screenspot_IoU": 0.6310188174247742,
+      "eval_screenspot_MAE_all": 0.06860506162047386,
+      "eval_screenspot_MAE_h": 0.04429138886431853,
+      "eval_screenspot_MAE_w": 0.11814649154742558,
+      "eval_screenspot_MAE_x": 0.06877896686395009,
+      "eval_screenspot_MAE_y": 0.043203407898545265,
+      "eval_screenspot_NUM_probability": 0.9998593727747599,
+      "eval_screenspot_inside_bbox": 0.8841666579246521,
+      "eval_screenspot_loss": 0.8844305276870728,
+      "eval_screenspot_loss_ce": 0.5899222294489542,
+      "eval_screenspot_loss_iou": 0.4657796223958333,
+      "eval_screenspot_loss_num": 0.057614644368489586,
+      "eval_screenspot_loss_xval": 0.2879842122395833,
+      "eval_screenspot_runtime": 153.6916,
+      "eval_screenspot_samples_per_second": 0.579,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "loss": 0.8671462535858154,
+      "loss_ce": 0.5839431285858154,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8551325047675412,
+      "grad_norm": 38.29206628356214,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 558389192,
+      "step": 3251
+    },
+    {
+      "epoch": 0.8551325047675412,
+      "loss": 0.21791958808898926,
+      "loss_ce": 0.0010311761870980263,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 558389192,
+      "step": 3251
+    },
+    {
+      "epoch": 0.8553955415269284,
+      "grad_norm": 7.6502620366688525,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 558561204,
+      "step": 3252
+    },
+    {
+      "epoch": 0.8553955415269284,
+      "loss": 0.16455963253974915,
+      "loss_ce": 0.0034573215525597334,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 558561204,
+      "step": 3252
+    },
+    {
+      "epoch": 0.8556585782863155,
+      "grad_norm": 7.815933727502156,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 558733140,
+      "step": 3253
+    },
+    {
+      "epoch": 0.8556585782863155,
+      "loss": 0.118372842669487,
+      "loss_ce": 0.0003461065352894366,
+      "loss_iou": NaN,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 558733140,
+      "step": 3253
+    },
+    {
+      "epoch": 0.8559216150457026,
+      "grad_norm": 26.03211500217736,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 558905528,
+      "step": 3254
+    },
+    {
+      "epoch": 0.8559216150457026,
+      "loss": 0.0649479404091835,
+      "loss_ce": 0.0013187924632802606,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 558905528,
+      "step": 3254
+    },
+    {
+      "epoch": 0.8561846518050897,
+      "grad_norm": 3.695118212351963,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 559077788,
+      "step": 3255
+    },
+    {
+      "epoch": 0.8561846518050897,
+      "loss": 0.05111432075500488,
+      "loss_ce": 0.00036358798388391733,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 559077788,
+      "step": 3255
+    },
+    {
+      "epoch": 0.8564476885644768,
+      "grad_norm": 4.828867302993965,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 559249740,
+      "step": 3256
+    },
+    {
+      "epoch": 0.8564476885644768,
+      "loss": 0.15594083070755005,
+      "loss_ce": 0.00011808436829596758,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 559249740,
+      "step": 3256
+    },
+    {
+      "epoch": 0.856710725323864,
+      "grad_norm": 4.4588084335736,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 559422008,
+      "step": 3257
+    },
+    {
+      "epoch": 0.856710725323864,
+      "loss": 0.1171468049287796,
+      "loss_ce": 0.005177808925509453,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 559422008,
+      "step": 3257
+    },
+    {
+      "epoch": 0.8569737620832512,
+      "grad_norm": 22.52084230262158,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 559594180,
+      "step": 3258
+    },
+    {
+      "epoch": 0.8569737620832512,
+      "loss": 0.12303026020526886,
+      "loss_ce": 0.00025803959579207003,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 559594180,
+      "step": 3258
+    },
+    {
+      "epoch": 0.8572367988426383,
+      "grad_norm": 5.00194835125284,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 559766068,
+      "step": 3259
+    },
+    {
+      "epoch": 0.8572367988426383,
+      "loss": 0.14650404453277588,
+      "loss_ce": 0.0012403683504089713,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 559766068,
+      "step": 3259
+    },
+    {
+      "epoch": 0.8574998356020254,
+      "grad_norm": 5.872594878613971,
+      "learning_rate": 5e-06,
+      "loss": 0.107,
+      "num_input_tokens_seen": 559937984,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8574998356020254,
+      "loss": 0.14809638261795044,
+      "loss_ce": 0.0009711501188576221,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 559937984,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8577628723614125,
+      "grad_norm": 7.89915891449399,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 560108224,
+      "step": 3261
+    },
+    {
+      "epoch": 0.8577628723614125,
+      "loss": 0.12339873611927032,
+      "loss_ce": 0.0029458554927259684,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 560108224,
+      "step": 3261
+    },
+    {
+      "epoch": 0.8580259091207997,
+      "grad_norm": 5.069283331721027,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 560280532,
+      "step": 3262
+    },
+    {
+      "epoch": 0.8580259091207997,
+      "loss": 0.12928983569145203,
+      "loss_ce": 0.004533977247774601,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 560280532,
+      "step": 3262
+    },
+    {
+      "epoch": 0.8582889458801868,
+      "grad_norm": 14.252906607697872,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 560452680,
+      "step": 3263
+    },
+    {
+      "epoch": 0.8582889458801868,
+      "loss": 0.10280074179172516,
+      "loss_ce": 0.0005515902303159237,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 560452680,
+      "step": 3263
+    },
+    {
+      "epoch": 0.8585519826395739,
+      "grad_norm": 4.981476516908149,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 560625236,
+      "step": 3264
+    },
+    {
+      "epoch": 0.8585519826395739,
+      "loss": 0.11402536928653717,
+      "loss_ce": 0.0004999822122044861,
+      "loss_iou": 0.453125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 560625236,
+      "step": 3264
+    },
+    {
+      "epoch": 0.858815019398961,
+      "grad_norm": 8.268610027664979,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 560797248,
+      "step": 3265
+    },
+    {
+      "epoch": 0.858815019398961,
+      "loss": 0.13858602941036224,
+      "loss_ce": 0.0017146880272775888,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 560797248,
+      "step": 3265
+    },
+    {
+      "epoch": 0.8590780561583481,
+      "grad_norm": 8.133733497165355,
+      "learning_rate": 5e-06,
+      "loss": 0.1637,
+      "num_input_tokens_seen": 560969220,
+      "step": 3266
+    },
+    {
+      "epoch": 0.8590780561583481,
+      "loss": 0.09112342447042465,
+      "loss_ce": 0.00012000648712273687,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 560969220,
+      "step": 3266
+    },
+    {
+      "epoch": 0.8593410929177353,
+      "grad_norm": 3.6754936276624406,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 561141552,
+      "step": 3267
+    },
+    {
+      "epoch": 0.8593410929177353,
+      "loss": 0.12336836755275726,
+      "loss_ce": 0.0018015915993601084,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 561141552,
+      "step": 3267
+    },
+    {
+      "epoch": 0.8596041296771224,
+      "grad_norm": 12.378662074106767,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 561313780,
+      "step": 3268
+    },
+    {
+      "epoch": 0.8596041296771224,
+      "loss": 0.14065909385681152,
+      "loss_ce": 0.0051305294036865234,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 561313780,
+      "step": 3268
+    },
+    {
+      "epoch": 0.8598671664365095,
+      "grad_norm": 3.0561071778648485,
+      "learning_rate": 5e-06,
+      "loss": 0.103,
+      "num_input_tokens_seen": 561486192,
+      "step": 3269
+    },
+    {
+      "epoch": 0.8598671664365095,
+      "loss": 0.0891718789935112,
+      "loss_ce": 0.003921023570001125,
+      "loss_iou": 0.453125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 561486192,
+      "step": 3269
+    },
+    {
+      "epoch": 0.8601302031958966,
+      "grad_norm": 4.439332180024113,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 561658324,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8601302031958966,
+      "loss": 0.0599624440073967,
+      "loss_ce": 0.0006515316781587899,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 561658324,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8603932399552837,
+      "grad_norm": 5.0489240884621305,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 561829100,
+      "step": 3271
+    },
+    {
+      "epoch": 0.8603932399552837,
+      "loss": 0.06052926927804947,
+      "loss_ce": 0.0006232602754607797,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 561829100,
+      "step": 3271
+    },
+    {
+      "epoch": 0.8606562767146708,
+      "grad_norm": 14.0110320202428,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 562001212,
+      "step": 3272
+    },
+    {
+      "epoch": 0.8606562767146708,
+      "loss": 0.05530541390180588,
+      "loss_ce": 0.0009230896248482168,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 562001212,
+      "step": 3272
+    },
+    {
+      "epoch": 0.860919313474058,
+      "grad_norm": 5.01456561485217,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 562173524,
+      "step": 3273
+    },
+    {
+      "epoch": 0.860919313474058,
+      "loss": 0.2010606825351715,
+      "loss_ce": 0.002330208197236061,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 562173524,
+      "step": 3273
+    },
+    {
+      "epoch": 0.8611823502334451,
+      "grad_norm": 9.881111822277372,
+      "learning_rate": 5e-06,
+      "loss": 0.1502,
+      "num_input_tokens_seen": 562345824,
+      "step": 3274
+    },
+    {
+      "epoch": 0.8611823502334451,
+      "loss": 0.106649249792099,
+      "loss_ce": 0.0010584269184619188,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 562345824,
+      "step": 3274
+    },
+    {
+      "epoch": 0.8614453869928322,
+      "grad_norm": 4.536681210731165,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 562517872,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8614453869928322,
+      "loss": 0.07371491193771362,
+      "loss_ce": 0.0001065141914295964,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 562517872,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8617084237522193,
+      "grad_norm": 4.65141834474055,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 562689796,
+      "step": 3276
+    },
+    {
+      "epoch": 0.8617084237522193,
+      "loss": 0.09085643291473389,
+      "loss_ce": 0.001287340302951634,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 562689796,
+      "step": 3276
+    },
+    {
+      "epoch": 0.8619714605116064,
+      "grad_norm": 8.721252840556735,
+      "learning_rate": 5e-06,
+      "loss": 0.1602,
+      "num_input_tokens_seen": 562860476,
+      "step": 3277
+    },
+    {
+      "epoch": 0.8619714605116064,
+      "loss": 0.1055789366364479,
+      "loss_ce": 0.00273469858802855,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 562860476,
+      "step": 3277
+    },
+    {
+      "epoch": 0.8622344972709937,
+      "grad_norm": 4.218638673445259,
+      "learning_rate": 5e-06,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 563032560,
+      "step": 3278
+    },
+    {
+      "epoch": 0.8622344972709937,
+      "loss": 0.10979656875133514,
+      "loss_ce": 0.00536542059853673,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 563032560,
+      "step": 3278
+    },
+    {
+      "epoch": 0.8624975340303808,
+      "grad_norm": 5.172843844259764,
+      "learning_rate": 5e-06,
+      "loss": 0.0986,
+      "num_input_tokens_seen": 563204720,
+      "step": 3279
+    },
+    {
+      "epoch": 0.8624975340303808,
+      "loss": 0.08518050611019135,
+      "loss_ce": 0.0019438066519796848,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 563204720,
+      "step": 3279
+    },
+    {
+      "epoch": 0.8627605707897679,
+      "grad_norm": 4.560338661852982,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 563377056,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8627605707897679,
+      "loss": 0.08894481509923935,
+      "loss_ce": 0.0001691804500296712,
+      "loss_iou": 0.578125,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 563377056,
+      "step": 3280
+    },
+    {
+      "epoch": 0.863023607549155,
+      "grad_norm": 10.082366990468211,
+      "learning_rate": 5e-06,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 563549604,
+      "step": 3281
+    },
+    {
+      "epoch": 0.863023607549155,
+      "loss": 0.1488599181175232,
+      "loss_ce": 0.0033978780265897512,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 563549604,
+      "step": 3281
+    },
+    {
+      "epoch": 0.8632866443085421,
+      "grad_norm": 4.832733016773012,
+      "learning_rate": 5e-06,
+      "loss": 0.081,
+      "num_input_tokens_seen": 563721852,
+      "step": 3282
+    },
+    {
+      "epoch": 0.8632866443085421,
+      "loss": 0.14535953104496002,
+      "loss_ce": 0.0013928530970588326,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 563721852,
+      "step": 3282
+    },
+    {
+      "epoch": 0.8635496810679293,
+      "grad_norm": 6.063993109800719,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 563893984,
+      "step": 3283
+    },
+    {
+      "epoch": 0.8635496810679293,
+      "loss": 0.06962516903877258,
+      "loss_ce": 0.0006554402643814683,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 563893984,
+      "step": 3283
+    },
+    {
+      "epoch": 0.8638127178273164,
+      "grad_norm": 4.985736636691599,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 564062972,
+      "step": 3284
+    },
+    {
+      "epoch": 0.8638127178273164,
+      "loss": 0.11227120459079742,
+      "loss_ce": 0.002926721004769206,
+      "loss_iou": 0.625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 564062972,
+      "step": 3284
+    },
+    {
+      "epoch": 0.8640757545867035,
+      "grad_norm": 10.90397210954732,
+      "learning_rate": 5e-06,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 564235088,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8640757545867035,
+      "loss": 0.1431923508644104,
+      "loss_ce": 0.0032692591194063425,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 564235088,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8643387913460906,
+      "grad_norm": 10.801058055973197,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 564407212,
+      "step": 3286
+    },
+    {
+      "epoch": 0.8643387913460906,
+      "loss": 0.17996424436569214,
+      "loss_ce": 0.005098515655845404,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 564407212,
+      "step": 3286
+    },
+    {
+      "epoch": 0.8646018281054777,
+      "grad_norm": 3.7908688756477646,
+      "learning_rate": 5e-06,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 564579544,
+      "step": 3287
+    },
+    {
+      "epoch": 0.8646018281054777,
+      "loss": 0.1904245764017105,
+      "loss_ce": 0.0021311198361217976,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 564579544,
+      "step": 3287
+    },
+    {
+      "epoch": 0.8648648648648649,
+      "grad_norm": 20.50727894197877,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 564751736,
+      "step": 3288
+    },
+    {
+      "epoch": 0.8648648648648649,
+      "loss": 0.13853441178798676,
+      "loss_ce": 0.006606926675885916,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 564751736,
+      "step": 3288
+    },
+    {
+      "epoch": 0.865127901624252,
+      "grad_norm": 7.515905813339519,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 564923836,
+      "step": 3289
+    },
+    {
+      "epoch": 0.865127901624252,
+      "loss": 0.09346568584442139,
+      "loss_ce": 0.0014246755745261908,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 564923836,
+      "step": 3289
+    },
+    {
+      "epoch": 0.8653909383836391,
+      "grad_norm": 4.659865141429364,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 565093036,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8653909383836391,
+      "loss": 0.05857858434319496,
+      "loss_ce": 0.0009613969596102834,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 565093036,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8656539751430262,
+      "grad_norm": 7.156818926737148,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 565264968,
+      "step": 3291
+    },
+    {
+      "epoch": 0.8656539751430262,
+      "loss": 0.05671301484107971,
+      "loss_ce": 0.0005454107304103673,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 565264968,
+      "step": 3291
+    },
+    {
+      "epoch": 0.8659170119024133,
+      "grad_norm": 4.159578348893538,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 565437008,
+      "step": 3292
+    },
+    {
+      "epoch": 0.8659170119024133,
+      "loss": 0.10393117368221283,
+      "loss_ce": 0.0047185225412249565,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 565437008,
+      "step": 3292
+    },
+    {
+      "epoch": 0.8661800486618005,
+      "grad_norm": 5.125947407861983,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 565609352,
+      "step": 3293
+    },
+    {
+      "epoch": 0.8661800486618005,
+      "loss": 0.07895916700363159,
+      "loss_ce": 0.004221613518893719,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 565609352,
+      "step": 3293
+    },
+    {
+      "epoch": 0.8664430854211876,
+      "grad_norm": 4.631562530302505,
+      "learning_rate": 5e-06,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 565781624,
+      "step": 3294
+    },
+    {
+      "epoch": 0.8664430854211876,
+      "loss": 0.10047349333763123,
+      "loss_ce": 0.0010014427825808525,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 565781624,
+      "step": 3294
+    },
+    {
+      "epoch": 0.8667061221805747,
+      "grad_norm": 12.748430441696817,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 565954124,
+      "step": 3295
+    },
+    {
+      "epoch": 0.8667061221805747,
+      "loss": 0.10027378797531128,
+      "loss_ce": 0.0005728579708375037,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 565954124,
+      "step": 3295
+    },
+    {
+      "epoch": 0.8669691589399618,
+      "grad_norm": 8.307592437145553,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 566124916,
+      "step": 3296
+    },
+    {
+      "epoch": 0.8669691589399618,
+      "loss": 0.10838001221418381,
+      "loss_ce": 0.006359752267599106,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 566124916,
+      "step": 3296
+    },
+    {
+      "epoch": 0.867232195699349,
+      "grad_norm": 4.778870142686005,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 566296912,
+      "step": 3297
+    },
+    {
+      "epoch": 0.867232195699349,
+      "loss": 0.04819861054420471,
+      "loss_ce": 0.00030127062927931547,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 566296912,
+      "step": 3297
+    },
+    {
+      "epoch": 0.867495232458736,
+      "grad_norm": 7.859655636877698,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 566467596,
+      "step": 3298
+    },
+    {
+      "epoch": 0.867495232458736,
+      "loss": 0.1346224993467331,
+      "loss_ce": 0.0007113683386705816,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 566467596,
+      "step": 3298
+    },
+    {
+      "epoch": 0.8677582692181233,
+      "grad_norm": 15.137416498429843,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 566639732,
+      "step": 3299
+    },
+    {
+      "epoch": 0.8677582692181233,
+      "loss": 0.10433374345302582,
+      "loss_ce": 0.00313745578750968,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 566639732,
+      "step": 3299
+    },
+    {
+      "epoch": 0.8680213059775104,
+      "grad_norm": 2.9506634806748657,
+      "learning_rate": 5e-06,
+      "loss": 0.118,
+      "num_input_tokens_seen": 566811792,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8680213059775104,
+      "loss": 0.07893712818622589,
+      "loss_ce": 0.0004764424229506403,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 566811792,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8682843427368975,
+      "grad_norm": 21.308475468927227,
+      "learning_rate": 5e-06,
+      "loss": 0.1217,
+      "num_input_tokens_seen": 566984152,
+      "step": 3301
+    },
+    {
+      "epoch": 0.8682843427368975,
+      "loss": 0.07327578961849213,
+      "loss_ce": 0.004519685637205839,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 566984152,
+      "step": 3301
+    },
+    {
+      "epoch": 0.8685473794962846,
+      "grad_norm": 4.854189489350901,
+      "learning_rate": 5e-06,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 567156660,
+      "step": 3302
+    },
+    {
+      "epoch": 0.8685473794962846,
+      "loss": 0.05097039043903351,
+      "loss_ce": 0.0012877746485173702,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 567156660,
+      "step": 3302
+    },
+    {
+      "epoch": 0.8688104162556717,
+      "grad_norm": 16.314064517801537,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 567328912,
+      "step": 3303
+    },
+    {
+      "epoch": 0.8688104162556717,
+      "loss": 0.11892714351415634,
+      "loss_ce": 0.0012513676192611456,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 567328912,
+      "step": 3303
+    },
+    {
+      "epoch": 0.8690734530150589,
+      "grad_norm": 16.431940991025456,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 567501148,
+      "step": 3304
+    },
+    {
+      "epoch": 0.8690734530150589,
+      "loss": 0.12847568094730377,
+      "loss_ce": 8.82247113622725e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 567501148,
+      "step": 3304
+    },
+    {
+      "epoch": 0.869336489774446,
+      "grad_norm": 3.641251656159218,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 567673160,
+      "step": 3305
+    },
+    {
+      "epoch": 0.869336489774446,
+      "loss": 0.2121918946504593,
+      "loss_ce": 0.0007966216653585434,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 567673160,
+      "step": 3305
+    },
+    {
+      "epoch": 0.8695995265338331,
+      "grad_norm": 27.660934774490972,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 567845644,
+      "step": 3306
+    },
+    {
+      "epoch": 0.8695995265338331,
+      "loss": 0.07302013039588928,
+      "loss_ce": 0.00043406913755461574,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 567845644,
+      "step": 3306
+    },
+    {
+      "epoch": 0.8698625632932202,
+      "grad_norm": 5.415514477165426,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 568017832,
+      "step": 3307
+    },
+    {
+      "epoch": 0.8698625632932202,
+      "loss": 0.14274545013904572,
+      "loss_ce": 0.0012964779743924737,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 568017832,
+      "step": 3307
+    },
+    {
+      "epoch": 0.8701256000526073,
+      "grad_norm": 6.457549289729974,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 568189908,
+      "step": 3308
+    },
+    {
+      "epoch": 0.8701256000526073,
+      "loss": 0.17567481100559235,
+      "loss_ce": 0.0026706610806286335,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 568189908,
+      "step": 3308
+    },
+    {
+      "epoch": 0.8703886368119945,
+      "grad_norm": 4.776984151346558,
+      "learning_rate": 5e-06,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 568358640,
+      "step": 3309
+    },
+    {
+      "epoch": 0.8703886368119945,
+      "loss": 0.050816282629966736,
+      "loss_ce": 0.0006148651009425521,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 568358640,
+      "step": 3309
+    },
+    {
+      "epoch": 0.8706516735713816,
+      "grad_norm": 8.715556903813102,
+      "learning_rate": 5e-06,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 568529372,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8706516735713816,
+      "loss": 0.055395372211933136,
+      "loss_ce": 0.002645737724378705,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 568529372,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8709147103307687,
+      "grad_norm": 4.571078001011392,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 568701592,
+      "step": 3311
+    },
+    {
+      "epoch": 0.8709147103307687,
+      "loss": 0.07437048852443695,
+      "loss_ce": 0.0011893401388078928,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 568701592,
+      "step": 3311
+    },
+    {
+      "epoch": 0.8711777470901558,
+      "grad_norm": 5.629346035154349,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 568871992,
+      "step": 3312
+    },
+    {
+      "epoch": 0.8711777470901558,
+      "loss": 0.08190266788005829,
+      "loss_ce": 0.001977135892957449,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 568871992,
+      "step": 3312
+    },
+    {
+      "epoch": 0.8714407838495429,
+      "grad_norm": 5.163413374700219,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 569044108,
+      "step": 3313
+    },
+    {
+      "epoch": 0.8714407838495429,
+      "loss": 0.08449655771255493,
+      "loss_ce": 0.0003901080635841936,
+      "loss_iou": 0.515625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 569044108,
+      "step": 3313
+    },
+    {
+      "epoch": 0.8717038206089301,
+      "grad_norm": 5.300317232023734,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 569216304,
+      "step": 3314
+    },
+    {
+      "epoch": 0.8717038206089301,
+      "loss": 0.18054433166980743,
+      "loss_ce": 9.388441685587168e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 569216304,
+      "step": 3314
+    },
+    {
+      "epoch": 0.8719668573683172,
+      "grad_norm": 5.650101854564751,
+      "learning_rate": 5e-06,
+      "loss": 0.101,
+      "num_input_tokens_seen": 569388684,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8719668573683172,
+      "loss": 0.08428364247083664,
+      "loss_ce": 0.0036256806924939156,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 569388684,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8722298941277044,
+      "grad_norm": 6.227695364300726,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 569558992,
+      "step": 3316
+    },
+    {
+      "epoch": 0.8722298941277044,
+      "loss": 0.18722307682037354,
+      "loss_ce": 0.005307785701006651,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 569558992,
+      "step": 3316
+    },
+    {
+      "epoch": 0.8724929308870915,
+      "grad_norm": 5.420469146952656,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 569731372,
+      "step": 3317
+    },
+    {
+      "epoch": 0.8724929308870915,
+      "loss": 0.09945103526115417,
+      "loss_ce": 0.002435657661408186,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 569731372,
+      "step": 3317
+    },
+    {
+      "epoch": 0.8727559676464786,
+      "grad_norm": 4.496256221139891,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 569902092,
+      "step": 3318
+    },
+    {
+      "epoch": 0.8727559676464786,
+      "loss": 0.09977222979068756,
+      "loss_ce": 8.65593392518349e-05,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 569902092,
+      "step": 3318
+    },
+    {
+      "epoch": 0.8730190044058658,
+      "grad_norm": 3.818210026049398,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 570074040,
+      "step": 3319
+    },
+    {
+      "epoch": 0.8730190044058658,
+      "loss": 0.11558879911899567,
+      "loss_ce": 0.0015446072211489081,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 570074040,
+      "step": 3319
+    },
+    {
+      "epoch": 0.8732820411652529,
+      "grad_norm": 48.02513571155541,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 570246368,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8732820411652529,
+      "loss": 0.1751258671283722,
+      "loss_ce": 0.00022961836657486856,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 570246368,
+      "step": 3320
+    },
+    {
+      "epoch": 0.87354507792464,
+      "grad_norm": 9.348108956967192,
+      "learning_rate": 5e-06,
+      "loss": 0.1217,
+      "num_input_tokens_seen": 570418348,
+      "step": 3321
+    },
+    {
+      "epoch": 0.87354507792464,
+      "loss": 0.1822582334280014,
+      "loss_ce": 0.001411056611686945,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 570418348,
+      "step": 3321
+    },
+    {
+      "epoch": 0.8738081146840271,
+      "grad_norm": 10.70912004570317,
+      "learning_rate": 5e-06,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 570590896,
+      "step": 3322
+    },
+    {
+      "epoch": 0.8738081146840271,
+      "loss": 0.1525057554244995,
+      "loss_ce": 0.001626854995265603,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 570590896,
+      "step": 3322
+    },
+    {
+      "epoch": 0.8740711514434142,
+      "grad_norm": 5.432623634439659,
+      "learning_rate": 5e-06,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 570763096,
+      "step": 3323
+    },
+    {
+      "epoch": 0.8740711514434142,
+      "loss": 0.10707151144742966,
+      "loss_ce": 0.00158750603441149,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 570763096,
+      "step": 3323
+    },
+    {
+      "epoch": 0.8743341882028013,
+      "grad_norm": 5.502873741651398,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 570935100,
+      "step": 3324
+    },
+    {
+      "epoch": 0.8743341882028013,
+      "loss": 0.0433497279882431,
+      "loss_ce": 0.0022425525821745396,
+      "loss_iou": 0.625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 570935100,
+      "step": 3324
+    },
+    {
+      "epoch": 0.8745972249621885,
+      "grad_norm": 9.035108225672646,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 571107528,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8745972249621885,
+      "loss": 0.11922352015972137,
+      "loss_ce": 0.008475230075418949,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 571107528,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8748602617215756,
+      "grad_norm": 4.9129107081845635,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 571279464,
+      "step": 3326
+    },
+    {
+      "epoch": 0.8748602617215756,
+      "loss": 0.09204280376434326,
+      "loss_ce": 0.0020769857801496983,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 571279464,
+      "step": 3326
+    },
+    {
+      "epoch": 0.8751232984809627,
+      "grad_norm": 4.808896887234646,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 571451676,
+      "step": 3327
+    },
+    {
+      "epoch": 0.8751232984809627,
+      "loss": 0.11273814737796783,
+      "loss_ce": 0.00046397349797189236,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 571451676,
+      "step": 3327
+    },
+    {
+      "epoch": 0.8753863352403498,
+      "grad_norm": 4.264250577253754,
+      "learning_rate": 5e-06,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 571620552,
+      "step": 3328
+    },
+    {
+      "epoch": 0.8753863352403498,
+      "loss": 0.06821378320455551,
+      "loss_ce": 0.0043710097670555115,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 571620552,
+      "step": 3328
+    },
+    {
+      "epoch": 0.8756493719997369,
+      "grad_norm": 19.986204474022973,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 571792724,
+      "step": 3329
+    },
+    {
+      "epoch": 0.8756493719997369,
+      "loss": 0.10326668620109558,
+      "loss_ce": 0.002741775708273053,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 571792724,
+      "step": 3329
+    },
+    {
+      "epoch": 0.8759124087591241,
+      "grad_norm": 4.794277486673457,
+      "learning_rate": 5e-06,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 571964880,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8759124087591241,
+      "loss": 0.11237187683582306,
+      "loss_ce": 0.00040288365562446415,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 571964880,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8761754455185112,
+      "grad_norm": 4.388563206165435,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 572137040,
+      "step": 3331
+    },
+    {
+      "epoch": 0.8761754455185112,
+      "loss": 0.1180691048502922,
+      "loss_ce": 0.003506115637719631,
+      "loss_iou": 0.5625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 572137040,
+      "step": 3331
+    },
+    {
+      "epoch": 0.8764384822778983,
+      "grad_norm": 4.318891449603682,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 572309112,
+      "step": 3332
+    },
+    {
+      "epoch": 0.8764384822778983,
+      "loss": 0.11371566355228424,
+      "loss_ce": 0.0035472088493406773,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 572309112,
+      "step": 3332
+    },
+    {
+      "epoch": 0.8767015190372854,
+      "grad_norm": 10.0048154448047,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 572481700,
+      "step": 3333
+    },
+    {
+      "epoch": 0.8767015190372854,
+      "loss": 0.21858729422092438,
+      "loss_ce": 0.0018819711403921247,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 572481700,
+      "step": 3333
+    },
+    {
+      "epoch": 0.8769645557966725,
+      "grad_norm": 20.86653328138894,
+      "learning_rate": 5e-06,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 572653588,
+      "step": 3334
+    },
+    {
+      "epoch": 0.8769645557966725,
+      "loss": 0.1189296618103981,
+      "loss_ce": 0.002413547597825527,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 572653588,
+      "step": 3334
+    },
+    {
+      "epoch": 0.8772275925560598,
+      "grad_norm": 4.936814293101831,
+      "learning_rate": 5e-06,
+      "loss": 0.1146,
+      "num_input_tokens_seen": 572825664,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8772275925560598,
+      "loss": 0.11313323676586151,
+      "loss_ce": 0.002995288698002696,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 572825664,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8774906293154469,
+      "grad_norm": 45.48466563185734,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 572997764,
+      "step": 3336
+    },
+    {
+      "epoch": 0.8774906293154469,
+      "loss": 0.12880939245224,
+      "loss_ce": 0.0020393752492964268,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 572997764,
+      "step": 3336
+    },
+    {
+      "epoch": 0.877753666074834,
+      "grad_norm": 5.329772367724437,
+      "learning_rate": 5e-06,
+      "loss": 0.145,
+      "num_input_tokens_seen": 573167692,
+      "step": 3337
+    },
+    {
+      "epoch": 0.877753666074834,
+      "loss": 0.06320846080780029,
+      "loss_ce": 0.0007694964297115803,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 573167692,
+      "step": 3337
+    },
+    {
+      "epoch": 0.8780167028342211,
+      "grad_norm": 9.592551272557182,
+      "learning_rate": 5e-06,
+      "loss": 0.0773,
+      "num_input_tokens_seen": 573340028,
+      "step": 3338
+    },
+    {
+      "epoch": 0.8780167028342211,
+      "loss": 0.08143293112516403,
+      "loss_ce": 0.0008054894860833883,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 573340028,
+      "step": 3338
+    },
+    {
+      "epoch": 0.8782797395936082,
+      "grad_norm": 3.2912653441237016,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 573510240,
+      "step": 3339
+    },
+    {
+      "epoch": 0.8782797395936082,
+      "loss": 0.11927518248558044,
+      "loss_ce": 0.0026980361435562372,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 573510240,
+      "step": 3339
+    },
+    {
+      "epoch": 0.8785427763529954,
+      "grad_norm": 6.074875569536377,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 573682136,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8785427763529954,
+      "loss": 0.14973485469818115,
+      "loss_ce": 0.0035251472145318985,
+      "loss_iou": 0.53125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 573682136,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8788058131123825,
+      "grad_norm": 4.101578026279459,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 573854492,
+      "step": 3341
+    },
+    {
+      "epoch": 0.8788058131123825,
+      "loss": 0.09308157861232758,
+      "loss_ce": 0.0016814331756904721,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 573854492,
+      "step": 3341
+    },
+    {
+      "epoch": 0.8790688498717696,
+      "grad_norm": 3.5139785635989704,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 574026312,
+      "step": 3342
+    },
+    {
+      "epoch": 0.8790688498717696,
+      "loss": 0.12709392607212067,
+      "loss_ce": 0.002124447375535965,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 574026312,
+      "step": 3342
+    },
+    {
+      "epoch": 0.8793318866311567,
+      "grad_norm": 8.712064685299605,
+      "learning_rate": 5e-06,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 574198272,
+      "step": 3343
+    },
+    {
+      "epoch": 0.8793318866311567,
+      "loss": 0.07467342913150787,
+      "loss_ce": 0.0009124410571530461,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 574198272,
+      "step": 3343
+    },
+    {
+      "epoch": 0.8795949233905438,
+      "grad_norm": 3.320527799830495,
+      "learning_rate": 5e-06,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 574370244,
+      "step": 3344
+    },
+    {
+      "epoch": 0.8795949233905438,
+      "loss": 0.04423138499259949,
+      "loss_ce": 0.00010297012340743095,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 574370244,
+      "step": 3344
+    },
+    {
+      "epoch": 0.879857960149931,
+      "grad_norm": 16.55083596075495,
+      "learning_rate": 5e-06,
+      "loss": 0.1285,
+      "num_input_tokens_seen": 574542464,
+      "step": 3345
+    },
+    {
+      "epoch": 0.879857960149931,
+      "loss": 0.05567498877644539,
+      "loss_ce": 0.0007433480932377279,
+      "loss_iou": 0.53125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 574542464,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8801209969093181,
+      "grad_norm": 4.233870416516864,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 574714876,
+      "step": 3346
+    },
+    {
+      "epoch": 0.8801209969093181,
+      "loss": 0.19878937304019928,
+      "loss_ce": 0.0015847685281187296,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 574714876,
+      "step": 3346
+    },
+    {
+      "epoch": 0.8803840336687052,
+      "grad_norm": 28.780242588246683,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 574886952,
+      "step": 3347
+    },
+    {
+      "epoch": 0.8803840336687052,
+      "loss": 0.11355656385421753,
+      "loss_ce": 0.0028998262714594603,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 574886952,
+      "step": 3347
+    },
+    {
+      "epoch": 0.8806470704280923,
+      "grad_norm": 3.1096156375201027,
+      "learning_rate": 5e-06,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 575059096,
+      "step": 3348
+    },
+    {
+      "epoch": 0.8806470704280923,
+      "loss": 0.049926742911338806,
+      "loss_ce": 0.0002136088878614828,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 575059096,
+      "step": 3348
+    },
+    {
+      "epoch": 0.8809101071874794,
+      "grad_norm": 11.92064516300679,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 575231232,
+      "step": 3349
+    },
+    {
+      "epoch": 0.8809101071874794,
+      "loss": 0.10825909674167633,
+      "loss_ce": 0.00019635571516118944,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 575231232,
+      "step": 3349
+    },
+    {
+      "epoch": 0.8811731439468665,
+      "grad_norm": 3.9827768470010754,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 575403592,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8811731439468665,
+      "loss": 0.11462774872779846,
+      "loss_ce": 0.005161193665117025,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 575403592,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8814361807062537,
+      "grad_norm": 6.779641757995467,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 575574024,
+      "step": 3351
+    },
+    {
+      "epoch": 0.8814361807062537,
+      "loss": 0.08754941821098328,
+      "loss_ce": 0.0006353590288199484,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 575574024,
+      "step": 3351
+    },
+    {
+      "epoch": 0.8816992174656408,
+      "grad_norm": 6.941492265981877,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 575746304,
+      "step": 3352
+    },
+    {
+      "epoch": 0.8816992174656408,
+      "loss": 0.08934507519006729,
+      "loss_ce": 0.0032244701869785786,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 575746304,
+      "step": 3352
+    },
+    {
+      "epoch": 0.8819622542250279,
+      "grad_norm": 5.409664650483643,
+      "learning_rate": 5e-06,
+      "loss": 0.1231,
+      "num_input_tokens_seen": 575918724,
+      "step": 3353
+    },
+    {
+      "epoch": 0.8819622542250279,
+      "loss": 0.08703694492578506,
+      "loss_ce": 0.0007637504604645073,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 575918724,
+      "step": 3353
+    },
+    {
+      "epoch": 0.882225290984415,
+      "grad_norm": 7.600652328701811,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 576090812,
+      "step": 3354
+    },
+    {
+      "epoch": 0.882225290984415,
+      "loss": 0.25224149227142334,
+      "loss_ce": 0.00353849190287292,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 576090812,
+      "step": 3354
+    },
+    {
+      "epoch": 0.8824883277438021,
+      "grad_norm": 8.07764479644441,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 576262592,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8824883277438021,
+      "loss": 0.08115795254707336,
+      "loss_ce": 0.0004999967059120536,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 576262592,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8827513645031894,
+      "grad_norm": 9.615405351973429,
+      "learning_rate": 5e-06,
+      "loss": 0.087,
+      "num_input_tokens_seen": 576434568,
+      "step": 3356
+    },
+    {
+      "epoch": 0.8827513645031894,
+      "loss": 0.08256545662879944,
+      "loss_ce": 0.0005952401552349329,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 576434568,
+      "step": 3356
+    },
+    {
+      "epoch": 0.8830144012625765,
+      "grad_norm": 4.944244594714697,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 576606792,
+      "step": 3357
+    },
+    {
+      "epoch": 0.8830144012625765,
+      "loss": 0.05560879409313202,
+      "loss_ce": 0.00032620219280943274,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 576606792,
+      "step": 3357
+    },
+    {
+      "epoch": 0.8832774380219636,
+      "grad_norm": 3.905791245202154,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 576778952,
+      "step": 3358
+    },
+    {
+      "epoch": 0.8832774380219636,
+      "loss": 0.05678252875804901,
+      "loss_ce": 0.0012634244048967957,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 576778952,
+      "step": 3358
+    },
+    {
+      "epoch": 0.8835404747813507,
+      "grad_norm": 3.3297332937077506,
+      "learning_rate": 5e-06,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 576951468,
+      "step": 3359
+    },
+    {
+      "epoch": 0.8835404747813507,
+      "loss": 0.15261146426200867,
+      "loss_ce": 0.0013968587154522538,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 576951468,
+      "step": 3359
+    },
+    {
+      "epoch": 0.8838035115407378,
+      "grad_norm": 7.82434473224169,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 577121952,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8838035115407378,
+      "loss": 0.10315507650375366,
+      "loss_ce": 0.0030421605333685875,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 577121952,
+      "step": 3360
+    },
+    {
+      "epoch": 0.884066548300125,
+      "grad_norm": 10.425535033707964,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 577293848,
+      "step": 3361
+    },
+    {
+      "epoch": 0.884066548300125,
+      "loss": 0.07322396337985992,
+      "loss_ce": 0.0005005778511986136,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 577293848,
+      "step": 3361
+    },
+    {
+      "epoch": 0.8843295850595121,
+      "grad_norm": 14.476290505686679,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 577466084,
+      "step": 3362
+    },
+    {
+      "epoch": 0.8843295850595121,
+      "loss": 0.1955878734588623,
+      "loss_ce": 0.0021064176689833403,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 577466084,
+      "step": 3362
+    },
+    {
+      "epoch": 0.8845926218188992,
+      "grad_norm": 7.100653555231675,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 577638416,
+      "step": 3363
+    },
+    {
+      "epoch": 0.8845926218188992,
+      "loss": 0.051397159695625305,
+      "loss_ce": 0.00037176761543378234,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 577638416,
+      "step": 3363
+    },
+    {
+      "epoch": 0.8848556585782863,
+      "grad_norm": 17.047045722794394,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 577810576,
+      "step": 3364
+    },
+    {
+      "epoch": 0.8848556585782863,
+      "loss": 0.11024264991283417,
+      "loss_ce": 0.0004404087667353451,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 577810576,
+      "step": 3364
+    },
+    {
+      "epoch": 0.8851186953376734,
+      "grad_norm": 6.6116170819185625,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 577982704,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8851186953376734,
+      "loss": 0.10162153840065002,
+      "loss_ce": 0.003171829041093588,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 577982704,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8853817320970606,
+      "grad_norm": 8.013299917102648,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 578154696,
+      "step": 3366
+    },
+    {
+      "epoch": 0.8853817320970606,
+      "loss": 0.09505804628133774,
+      "loss_ce": 0.002620304934680462,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 578154696,
+      "step": 3366
+    },
+    {
+      "epoch": 0.8856447688564477,
+      "grad_norm": 34.03888154985408,
+      "learning_rate": 5e-06,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 578326940,
+      "step": 3367
+    },
+    {
+      "epoch": 0.8856447688564477,
+      "loss": 0.10307721793651581,
+      "loss_ce": 0.0007899247575551271,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 578326940,
+      "step": 3367
+    },
+    {
+      "epoch": 0.8859078056158348,
+      "grad_norm": 5.847556882862909,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 578499332,
+      "step": 3368
+    },
+    {
+      "epoch": 0.8859078056158348,
+      "loss": 0.14421170949935913,
+      "loss_ce": 0.0017556664533913136,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 578499332,
+      "step": 3368
+    },
+    {
+      "epoch": 0.8861708423752219,
+      "grad_norm": 6.636470819413435,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 578671272,
+      "step": 3369
+    },
+    {
+      "epoch": 0.8861708423752219,
+      "loss": 0.128991961479187,
+      "loss_ce": 0.0024660732597112656,
+      "loss_iou": 0.421875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 578671272,
+      "step": 3369
+    },
+    {
+      "epoch": 0.886433879134609,
+      "grad_norm": 17.54919755928276,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 578842196,
+      "step": 3370
+    },
+    {
+      "epoch": 0.886433879134609,
+      "loss": 0.10547197610139847,
+      "loss_ce": 0.0005067643942311406,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 578842196,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8866969158939962,
+      "grad_norm": 7.184938374175427,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 579014236,
+      "step": 3371
+    },
+    {
+      "epoch": 0.8866969158939962,
+      "loss": 0.08108609914779663,
+      "loss_ce": 0.0009774556383490562,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 579014236,
+      "step": 3371
+    },
+    {
+      "epoch": 0.8869599526533833,
+      "grad_norm": 6.863755671185031,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 579186596,
+      "step": 3372
+    },
+    {
+      "epoch": 0.8869599526533833,
+      "loss": 0.1200333908200264,
+      "loss_ce": 0.00034345051972195506,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 579186596,
+      "step": 3372
+    },
+    {
+      "epoch": 0.8872229894127704,
+      "grad_norm": 7.255619753434164,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 579358828,
+      "step": 3373
+    },
+    {
+      "epoch": 0.8872229894127704,
+      "loss": 0.11177671700716019,
+      "loss_ce": 0.0016998156206682324,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 579358828,
+      "step": 3373
+    },
+    {
+      "epoch": 0.8874860261721575,
+      "grad_norm": 4.645601347618675,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 579531144,
+      "step": 3374
+    },
+    {
+      "epoch": 0.8874860261721575,
+      "loss": 0.13369636237621307,
+      "loss_ce": 0.0010364485206082463,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 579531144,
+      "step": 3374
+    },
+    {
+      "epoch": 0.8877490629315447,
+      "grad_norm": 19.381762342400947,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 579703288,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8877490629315447,
+      "loss": 0.15896877646446228,
+      "loss_ce": 0.0007961708470247686,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 579703288,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8880120996909318,
+      "grad_norm": 4.08558956843784,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 579875620,
+      "step": 3376
+    },
+    {
+      "epoch": 0.8880120996909318,
+      "loss": 0.06248597800731659,
+      "loss_ce": 0.0004437366151250899,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 579875620,
+      "step": 3376
+    },
+    {
+      "epoch": 0.888275136450319,
+      "grad_norm": 4.308623737664192,
+      "learning_rate": 5e-06,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 580047776,
+      "step": 3377
+    },
+    {
+      "epoch": 0.888275136450319,
+      "loss": 0.07517598569393158,
+      "loss_ce": 0.0003468855866231024,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 580047776,
+      "step": 3377
+    },
+    {
+      "epoch": 0.8885381732097061,
+      "grad_norm": 17.189878667773364,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 580219696,
+      "step": 3378
+    },
+    {
+      "epoch": 0.8885381732097061,
+      "loss": 0.1588488519191742,
+      "loss_ce": 0.00040158609044738114,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 580219696,
+      "step": 3378
+    },
+    {
+      "epoch": 0.8888012099690932,
+      "grad_norm": 17.94232408746683,
+      "learning_rate": 5e-06,
+      "loss": 0.1435,
+      "num_input_tokens_seen": 580389856,
+      "step": 3379
+    },
+    {
+      "epoch": 0.8888012099690932,
+      "loss": 0.16451287269592285,
+      "loss_ce": 0.002464542631059885,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 580389856,
+      "step": 3379
+    },
+    {
+      "epoch": 0.8890642467284803,
+      "grad_norm": 24.21781414429754,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 580562140,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8890642467284803,
+      "loss": 0.15492644906044006,
+      "loss_ce": 0.0021554557606577873,
+      "loss_iou": 0.515625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 580562140,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8893272834878674,
+      "grad_norm": 10.546114100067816,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 580734572,
+      "step": 3381
+    },
+    {
+      "epoch": 0.8893272834878674,
+      "loss": 0.10413020849227905,
+      "loss_ce": 0.0011639007134363055,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 580734572,
+      "step": 3381
+    },
+    {
+      "epoch": 0.8895903202472546,
+      "grad_norm": 6.233208405285995,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 580906660,
+      "step": 3382
+    },
+    {
+      "epoch": 0.8895903202472546,
+      "loss": 0.14456871151924133,
+      "loss_ce": 0.0018837791867554188,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 580906660,
+      "step": 3382
+    },
+    {
+      "epoch": 0.8898533570066417,
+      "grad_norm": 3.9495859544374516,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 581078540,
+      "step": 3383
+    },
+    {
+      "epoch": 0.8898533570066417,
+      "loss": 0.13160666823387146,
+      "loss_ce": 0.0018764439737424254,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 581078540,
+      "step": 3383
+    },
+    {
+      "epoch": 0.8901163937660288,
+      "grad_norm": 4.909648559933354,
+      "learning_rate": 5e-06,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 581247384,
+      "step": 3384
+    },
+    {
+      "epoch": 0.8901163937660288,
+      "loss": 0.06679339706897736,
+      "loss_ce": 0.00032610760536044836,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 581247384,
+      "step": 3384
+    },
+    {
+      "epoch": 0.8903794305254159,
+      "grad_norm": 4.667075414121075,
+      "learning_rate": 5e-06,
+      "loss": 0.1195,
+      "num_input_tokens_seen": 581419360,
+      "step": 3385
+    },
+    {
+      "epoch": 0.8903794305254159,
+      "loss": 0.1276930421590805,
+      "loss_ce": 0.0018690668512135744,
+      "loss_iou": 0.5,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 581419360,
+      "step": 3385
+    },
+    {
+      "epoch": 0.890642467284803,
+      "grad_norm": 4.483616610873386,
+      "learning_rate": 5e-06,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 581589952,
+      "step": 3386
+    },
+    {
+      "epoch": 0.890642467284803,
+      "loss": 0.047724828124046326,
+      "loss_ce": 0.00011740828631445765,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 581589952,
+      "step": 3386
+    },
+    {
+      "epoch": 0.8909055040441902,
+      "grad_norm": 21.859242645746537,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 581762156,
+      "step": 3387
+    },
+    {
+      "epoch": 0.8909055040441902,
+      "loss": 0.18419209122657776,
+      "loss_ce": 0.0005067941965535283,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 581762156,
+      "step": 3387
+    },
+    {
+      "epoch": 0.8911685408035773,
+      "grad_norm": 8.665372364747144,
+      "learning_rate": 5e-06,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 581934152,
+      "step": 3388
+    },
+    {
+      "epoch": 0.8911685408035773,
+      "loss": 0.08677740395069122,
+      "loss_ce": 0.00022955110762268305,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 581934152,
+      "step": 3388
+    },
+    {
+      "epoch": 0.8914315775629644,
+      "grad_norm": 4.1210884311067835,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 582106144,
+      "step": 3389
+    },
+    {
+      "epoch": 0.8914315775629644,
+      "loss": 0.13846494257450104,
+      "loss_ce": 0.0010137634817510843,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 582106144,
+      "step": 3389
+    },
+    {
+      "epoch": 0.8916946143223515,
+      "grad_norm": 11.500479931217354,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 582278400,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8916946143223515,
+      "loss": 0.12132581323385239,
+      "loss_ce": 0.002826055744662881,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 582278400,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8919576510817386,
+      "grad_norm": 4.055853466494655,
+      "learning_rate": 5e-06,
+      "loss": 0.094,
+      "num_input_tokens_seen": 582450564,
+      "step": 3391
+    },
+    {
+      "epoch": 0.8919576510817386,
+      "loss": 0.0762665793299675,
+      "loss_ce": 0.0018952443497255445,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 582450564,
+      "step": 3391
+    },
+    {
+      "epoch": 0.8922206878411258,
+      "grad_norm": 8.665290308457982,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 582622336,
+      "step": 3392
+    },
+    {
+      "epoch": 0.8922206878411258,
+      "loss": 0.16068054735660553,
+      "loss_ce": 0.0011956822127103806,
+      "loss_iou": 0.546875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 582622336,
+      "step": 3392
+    },
+    {
+      "epoch": 0.892483724600513,
+      "grad_norm": 14.955097188866974,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 582794384,
+      "step": 3393
+    },
+    {
+      "epoch": 0.892483724600513,
+      "loss": 0.08523661643266678,
+      "loss_ce": 0.0010691368952393532,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 582794384,
+      "step": 3393
+    },
+    {
+      "epoch": 0.8927467613599,
+      "grad_norm": 14.050392295123078,
+      "learning_rate": 5e-06,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 582966476,
+      "step": 3394
+    },
+    {
+      "epoch": 0.8927467613599,
+      "loss": 0.11920854449272156,
+      "loss_ce": 0.0003883557510562241,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 582966476,
+      "step": 3394
+    },
+    {
+      "epoch": 0.8930097981192872,
+      "grad_norm": 3.5189407633589225,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 583138656,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8930097981192872,
+      "loss": 0.23967677354812622,
+      "loss_ce": 0.0011361411307007074,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 583138656,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8932728348786743,
+      "grad_norm": 4.239345260315017,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 583310580,
+      "step": 3396
+    },
+    {
+      "epoch": 0.8932728348786743,
+      "loss": 0.045496270060539246,
+      "loss_ce": 0.00010137087519979104,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 583310580,
+      "step": 3396
+    },
+    {
+      "epoch": 0.8935358716380615,
+      "grad_norm": 4.318293293535835,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 583482984,
+      "step": 3397
+    },
+    {
+      "epoch": 0.8935358716380615,
+      "loss": 0.11194127053022385,
+      "loss_ce": 0.000826766830869019,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 583482984,
+      "step": 3397
+    },
+    {
+      "epoch": 0.8937989083974486,
+      "grad_norm": 4.660547370469876,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 583655292,
+      "step": 3398
+    },
+    {
+      "epoch": 0.8937989083974486,
+      "loss": 0.07680627703666687,
+      "loss_ce": 0.00017663151083979756,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 583655292,
+      "step": 3398
+    },
+    {
+      "epoch": 0.8940619451568357,
+      "grad_norm": 10.828455614089098,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 583827408,
+      "step": 3399
+    },
+    {
+      "epoch": 0.8940619451568357,
+      "loss": 0.1173757016658783,
+      "loss_ce": 3.5622346331365407e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 583827408,
+      "step": 3399
+    },
+    {
+      "epoch": 0.8943249819162228,
+      "grad_norm": 5.042018374187092,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 583999664,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8943249819162228,
+      "loss": 0.061997972428798676,
+      "loss_ce": 0.000367723434465006,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 583999664,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8945880186756099,
+      "grad_norm": 26.304936486213094,
+      "learning_rate": 5e-06,
+      "loss": 0.1601,
+      "num_input_tokens_seen": 584171616,
+      "step": 3401
+    },
+    {
+      "epoch": 0.8945880186756099,
+      "loss": 0.2535288333892822,
+      "loss_ce": 0.004078162834048271,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 584171616,
+      "step": 3401
+    },
+    {
+      "epoch": 0.894851055434997,
+      "grad_norm": 4.926581765464391,
+      "learning_rate": 5e-06,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 584343844,
+      "step": 3402
+    },
+    {
+      "epoch": 0.894851055434997,
+      "loss": 0.11322697252035141,
+      "loss_ce": 0.0008307351381517947,
+      "loss_iou": NaN,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 584343844,
+      "step": 3402
+    },
+    {
+      "epoch": 0.8951140921943842,
+      "grad_norm": 7.363528943726032,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 584516156,
+      "step": 3403
+    },
+    {
+      "epoch": 0.8951140921943842,
+      "loss": 0.08316925168037415,
+      "loss_ce": 0.002114569302648306,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 584516156,
+      "step": 3403
+    },
+    {
+      "epoch": 0.8953771289537713,
+      "grad_norm": 4.344737704222575,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 584688680,
+      "step": 3404
+    },
+    {
+      "epoch": 0.8953771289537713,
+      "loss": 0.061312295496463776,
+      "loss_ce": 0.0012231803266331553,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 584688680,
+      "step": 3404
+    },
+    {
+      "epoch": 0.8956401657131584,
+      "grad_norm": 9.026057514105759,
+      "learning_rate": 5e-06,
+      "loss": 0.1616,
+      "num_input_tokens_seen": 584860880,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8956401657131584,
+      "loss": 0.09903927892446518,
+      "loss_ce": 0.00197811983525753,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 584860880,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8959032024725455,
+      "grad_norm": 9.19169137185048,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 585032800,
+      "step": 3406
+    },
+    {
+      "epoch": 0.8959032024725455,
+      "loss": 0.06624776124954224,
+      "loss_ce": 0.00032979153911583126,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 585032800,
+      "step": 3406
+    },
+    {
+      "epoch": 0.8961662392319326,
+      "grad_norm": 9.011492701681583,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 585203264,
+      "step": 3407
+    },
+    {
+      "epoch": 0.8961662392319326,
+      "loss": 0.21042554080486298,
+      "loss_ce": 0.002722906181588769,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 585203264,
+      "step": 3407
+    },
+    {
+      "epoch": 0.8964292759913198,
+      "grad_norm": 17.979008103100615,
+      "learning_rate": 5e-06,
+      "loss": 0.093,
+      "num_input_tokens_seen": 585375592,
+      "step": 3408
+    },
+    {
+      "epoch": 0.8964292759913198,
+      "loss": 0.041413549333810806,
+      "loss_ce": 0.0023205317556858063,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 585375592,
+      "step": 3408
+    },
+    {
+      "epoch": 0.8966923127507069,
+      "grad_norm": 10.115194008381021,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 585546216,
+      "step": 3409
+    },
+    {
+      "epoch": 0.8966923127507069,
+      "loss": 0.15392959117889404,
+      "loss_ce": 0.0011891128960996866,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 585546216,
+      "step": 3409
+    },
+    {
+      "epoch": 0.896955349510094,
+      "grad_norm": 7.995711146610571,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 585718304,
+      "step": 3410
+    },
+    {
+      "epoch": 0.896955349510094,
+      "loss": 0.05361251160502434,
+      "loss_ce": 0.0008171012159436941,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 585718304,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8972183862694811,
+      "grad_norm": 4.960227305854337,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 585890728,
+      "step": 3411
+    },
+    {
+      "epoch": 0.8972183862694811,
+      "loss": 0.08083316683769226,
+      "loss_ce": 0.0013043570797890425,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 585890728,
+      "step": 3411
+    },
+    {
+      "epoch": 0.8974814230288682,
+      "grad_norm": 5.883674464139159,
+      "learning_rate": 5e-06,
+      "loss": 0.1462,
+      "num_input_tokens_seen": 586062896,
+      "step": 3412
+    },
+    {
+      "epoch": 0.8974814230288682,
+      "loss": 0.06225815415382385,
+      "loss_ce": 0.003023534081876278,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 586062896,
+      "step": 3412
+    },
+    {
+      "epoch": 0.8977444597882555,
+      "grad_norm": 5.4958687199216625,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 586235628,
+      "step": 3413
+    },
+    {
+      "epoch": 0.8977444597882555,
+      "loss": 0.08900677412748337,
+      "loss_ce": 0.0012687351554632187,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 586235628,
+      "step": 3413
+    },
+    {
+      "epoch": 0.8980074965476426,
+      "grad_norm": 6.158695953111894,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 586407736,
+      "step": 3414
+    },
+    {
+      "epoch": 0.8980074965476426,
+      "loss": 0.13948456943035126,
+      "loss_ce": 0.0007364002522081137,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 586407736,
+      "step": 3414
+    },
+    {
+      "epoch": 0.8982705333070297,
+      "grad_norm": 6.799680477506266,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 586579816,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8982705333070297,
+      "loss": 0.1316445767879486,
+      "loss_ce": 0.0035775681026279926,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 586579816,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8985335700664168,
+      "grad_norm": 4.757727392184302,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 586750092,
+      "step": 3416
+    },
+    {
+      "epoch": 0.8985335700664168,
+      "loss": 0.1803445667028427,
+      "loss_ce": 0.0017557005630806088,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 586750092,
+      "step": 3416
+    },
+    {
+      "epoch": 0.8987966068258039,
+      "grad_norm": 4.7130808920215825,
+      "learning_rate": 5e-06,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 586920724,
+      "step": 3417
+    },
+    {
+      "epoch": 0.8987966068258039,
+      "loss": 0.04298722371459007,
+      "loss_ce": 0.0010102951200678945,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 586920724,
+      "step": 3417
+    },
+    {
+      "epoch": 0.8990596435851911,
+      "grad_norm": 9.716073099163628,
+      "learning_rate": 5e-06,
+      "loss": 0.0909,
+      "num_input_tokens_seen": 587093308,
+      "step": 3418
+    },
+    {
+      "epoch": 0.8990596435851911,
+      "loss": 0.11212408542633057,
+      "loss_ce": 0.0025659759994596243,
+      "loss_iou": 0.78125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 587093308,
+      "step": 3418
+    },
+    {
+      "epoch": 0.8993226803445782,
+      "grad_norm": 9.72495754411886,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 587265328,
+      "step": 3419
+    },
+    {
+      "epoch": 0.8993226803445782,
+      "loss": 0.09303727746009827,
+      "loss_ce": 0.0017286788206547499,
+      "loss_iou": 0.53125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 587265328,
+      "step": 3419
+    },
+    {
+      "epoch": 0.8995857171039653,
+      "grad_norm": 27.58546147063481,
+      "learning_rate": 5e-06,
+      "loss": 0.1334,
+      "num_input_tokens_seen": 587435704,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8995857171039653,
+      "loss": 0.20503398776054382,
+      "loss_ce": 0.0012986233923584223,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 587435704,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8998487538633524,
+      "grad_norm": 3.780121082337908,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 587605756,
+      "step": 3421
+    },
+    {
+      "epoch": 0.8998487538633524,
+      "loss": 0.09629541635513306,
+      "loss_ce": 0.0002413361653452739,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 587605756,
+      "step": 3421
+    },
+    {
+      "epoch": 0.9001117906227395,
+      "grad_norm": 4.905276205162047,
+      "learning_rate": 5e-06,
+      "loss": 0.1407,
+      "num_input_tokens_seen": 587776128,
+      "step": 3422
+    },
+    {
+      "epoch": 0.9001117906227395,
+      "loss": 0.22563423216342926,
+      "loss_ce": 0.003191609401255846,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 587776128,
+      "step": 3422
+    },
+    {
+      "epoch": 0.9003748273821266,
+      "grad_norm": 4.24824425523058,
+      "learning_rate": 5e-06,
+      "loss": 0.0902,
+      "num_input_tokens_seen": 587948348,
+      "step": 3423
+    },
+    {
+      "epoch": 0.9003748273821266,
+      "loss": 0.04901735112071037,
+      "loss_ce": 0.0007690612110309303,
+      "loss_iou": 0.5,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 587948348,
+      "step": 3423
+    },
+    {
+      "epoch": 0.9006378641415138,
+      "grad_norm": 14.901804223286693,
+      "learning_rate": 5e-06,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 588115160,
+      "step": 3424
+    },
+    {
+      "epoch": 0.9006378641415138,
+      "loss": 0.0699070394039154,
+      "loss_ce": 0.001684993039816618,
+      "loss_iou": 0.515625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 588115160,
+      "step": 3424
+    },
+    {
+      "epoch": 0.9009009009009009,
+      "grad_norm": 4.750314737648371,
+      "learning_rate": 5e-06,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 588287148,
+      "step": 3425
+    },
+    {
+      "epoch": 0.9009009009009009,
+      "loss": 0.07479090988636017,
+      "loss_ce": 0.006248427089303732,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 588287148,
+      "step": 3425
+    },
+    {
+      "epoch": 0.901163937660288,
+      "grad_norm": 3.9857635354375183,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 588457468,
+      "step": 3426
+    },
+    {
+      "epoch": 0.901163937660288,
+      "loss": 0.04558904469013214,
+      "loss_ce": 0.0004535481857601553,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 588457468,
+      "step": 3426
+    },
+    {
+      "epoch": 0.9014269744196751,
+      "grad_norm": 7.929685479499518,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 588627832,
+      "step": 3427
+    },
+    {
+      "epoch": 0.9014269744196751,
+      "loss": 0.11351916939020157,
+      "loss_ce": 0.00022265504230745137,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 588627832,
+      "step": 3427
+    },
+    {
+      "epoch": 0.9016900111790622,
+      "grad_norm": 5.274686939603064,
+      "learning_rate": 5e-06,
+      "loss": 0.1085,
+      "num_input_tokens_seen": 588797844,
+      "step": 3428
+    },
+    {
+      "epoch": 0.9016900111790622,
+      "loss": 0.07151903212070465,
+      "loss_ce": 0.004655018448829651,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 588797844,
+      "step": 3428
+    },
+    {
+      "epoch": 0.9019530479384494,
+      "grad_norm": 5.241669176622665,
+      "learning_rate": 5e-06,
+      "loss": 0.1302,
+      "num_input_tokens_seen": 588966836,
+      "step": 3429
+    },
+    {
+      "epoch": 0.9019530479384494,
+      "loss": 0.12862293422222137,
+      "loss_ce": 0.0008153152884915471,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 588966836,
+      "step": 3429
+    },
+    {
+      "epoch": 0.9022160846978365,
+      "grad_norm": 7.584585654330743,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 589136916,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9022160846978365,
+      "loss": 0.1643849015235901,
+      "loss_ce": 0.0007801597821526229,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 589136916,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9024791214572236,
+      "grad_norm": 3.532701818396997,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 589309164,
+      "step": 3431
+    },
+    {
+      "epoch": 0.9024791214572236,
+      "loss": 0.077778160572052,
+      "loss_ce": 0.0013621454127132893,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 589309164,
+      "step": 3431
+    },
+    {
+      "epoch": 0.9027421582166107,
+      "grad_norm": 7.674994076422982,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 589481132,
+      "step": 3432
+    },
+    {
+      "epoch": 0.9027421582166107,
+      "loss": 0.09501226991415024,
+      "loss_ce": 0.0008197662536986172,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 589481132,
+      "step": 3432
+    },
+    {
+      "epoch": 0.9030051949759978,
+      "grad_norm": 6.592301292301914,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 589649864,
+      "step": 3433
+    },
+    {
+      "epoch": 0.9030051949759978,
+      "loss": 0.06211673840880394,
+      "loss_ce": 0.0013867560774087906,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 589649864,
+      "step": 3433
+    },
+    {
+      "epoch": 0.9032682317353851,
+      "grad_norm": 5.45319787734761,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 589822176,
+      "step": 3434
+    },
+    {
+      "epoch": 0.9032682317353851,
+      "loss": 0.06987213343381882,
+      "loss_ce": 0.0025808701757341623,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 589822176,
+      "step": 3434
+    },
+    {
+      "epoch": 0.9035312684947722,
+      "grad_norm": 35.99527436418574,
+      "learning_rate": 5e-06,
+      "loss": 0.1639,
+      "num_input_tokens_seen": 589992448,
+      "step": 3435
+    },
+    {
+      "epoch": 0.9035312684947722,
+      "loss": 0.2300114631652832,
+      "loss_ce": 0.0013890261761844158,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 589992448,
+      "step": 3435
+    },
+    {
+      "epoch": 0.9037943052541593,
+      "grad_norm": 8.340774703127748,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 590162576,
+      "step": 3436
+    },
+    {
+      "epoch": 0.9037943052541593,
+      "loss": 0.09980174899101257,
+      "loss_ce": 0.001687736832536757,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 590162576,
+      "step": 3436
+    },
+    {
+      "epoch": 0.9040573420135464,
+      "grad_norm": 3.893496744409823,
+      "learning_rate": 5e-06,
+      "loss": 0.1361,
+      "num_input_tokens_seen": 590334768,
+      "step": 3437
+    },
+    {
+      "epoch": 0.9040573420135464,
+      "loss": 0.14963126182556152,
+      "loss_ce": 0.002658609300851822,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 590334768,
+      "step": 3437
+    },
+    {
+      "epoch": 0.9043203787729335,
+      "grad_norm": 4.512105245241041,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 590506964,
+      "step": 3438
+    },
+    {
+      "epoch": 0.9043203787729335,
+      "loss": 0.2095954418182373,
+      "loss_ce": 0.0010688342154026031,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 590506964,
+      "step": 3438
+    },
+    {
+      "epoch": 0.9045834155323207,
+      "grad_norm": 10.79594707291913,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 590679132,
+      "step": 3439
+    },
+    {
+      "epoch": 0.9045834155323207,
+      "loss": 0.056604690849781036,
+      "loss_ce": 0.0010932188015431166,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 590679132,
+      "step": 3439
+    },
+    {
+      "epoch": 0.9048464522917078,
+      "grad_norm": 7.232598771359226,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 590851488,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9048464522917078,
+      "loss": 0.07088696211576462,
+      "loss_ce": 0.0023749994579702616,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 590851488,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9051094890510949,
+      "grad_norm": 5.761072356819141,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 591023672,
+      "step": 3441
+    },
+    {
+      "epoch": 0.9051094890510949,
+      "loss": 0.05779781565070152,
+      "loss_ce": 0.002149013802409172,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 591023672,
+      "step": 3441
+    },
+    {
+      "epoch": 0.905372525810482,
+      "grad_norm": 5.464589960258587,
+      "learning_rate": 5e-06,
+      "loss": 0.0756,
+      "num_input_tokens_seen": 591196064,
+      "step": 3442
+    },
+    {
+      "epoch": 0.905372525810482,
+      "loss": 0.11280819773674011,
+      "loss_ce": 0.004058802034705877,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 591196064,
+      "step": 3442
+    },
+    {
+      "epoch": 0.9056355625698691,
+      "grad_norm": 10.0338321043045,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 591368108,
+      "step": 3443
+    },
+    {
+      "epoch": 0.9056355625698691,
+      "loss": 0.11732570827007294,
+      "loss_ce": 0.0005044231074862182,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 591368108,
+      "step": 3443
+    },
+    {
+      "epoch": 0.9058985993292563,
+      "grad_norm": 4.9322678909564805,
+      "learning_rate": 5e-06,
+      "loss": 0.1497,
+      "num_input_tokens_seen": 591539968,
+      "step": 3444
+    },
+    {
+      "epoch": 0.9058985993292563,
+      "loss": 0.10426676273345947,
+      "loss_ce": 0.0025516818277537823,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 591539968,
+      "step": 3444
+    },
+    {
+      "epoch": 0.9061616360886434,
+      "grad_norm": 10.20363728877728,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 591709020,
+      "step": 3445
+    },
+    {
+      "epoch": 0.9061616360886434,
+      "loss": 0.049313947558403015,
+      "loss_ce": 0.0002569416828919202,
+      "loss_iou": 0.515625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 591709020,
+      "step": 3445
+    },
+    {
+      "epoch": 0.9064246728480305,
+      "grad_norm": 8.985092426433443,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 591881092,
+      "step": 3446
+    },
+    {
+      "epoch": 0.9064246728480305,
+      "loss": 0.0865587666630745,
+      "loss_ce": 0.0004686739994212985,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 591881092,
+      "step": 3446
+    },
+    {
+      "epoch": 0.9066877096074176,
+      "grad_norm": 27.73232811749845,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 592053524,
+      "step": 3447
+    },
+    {
+      "epoch": 0.9066877096074176,
+      "loss": 0.10780518501996994,
+      "loss_ce": 0.0006884862086735666,
+      "loss_iou": 0.5,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 592053524,
+      "step": 3447
+    },
+    {
+      "epoch": 0.9069507463668047,
+      "grad_norm": 11.824522545760333,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 592225468,
+      "step": 3448
+    },
+    {
+      "epoch": 0.9069507463668047,
+      "loss": 0.0678405836224556,
+      "loss_ce": 0.002670294838026166,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 592225468,
+      "step": 3448
+    },
+    {
+      "epoch": 0.9072137831261918,
+      "grad_norm": 5.549108113367569,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 592397784,
+      "step": 3449
+    },
+    {
+      "epoch": 0.9072137831261918,
+      "loss": 0.06828893721103668,
+      "loss_ce": 0.0015164725482463837,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 592397784,
+      "step": 3449
+    },
+    {
+      "epoch": 0.907476819885579,
+      "grad_norm": 34.02061333154032,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 592569716,
+      "step": 3450
+    },
+    {
+      "epoch": 0.907476819885579,
+      "loss": 0.18558475375175476,
+      "loss_ce": 0.002723418176174164,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 592569716,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9077398566449661,
+      "grad_norm": 8.419964301658604,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 592742332,
+      "step": 3451
+    },
+    {
+      "epoch": 0.9077398566449661,
+      "loss": 0.10704197734594345,
+      "loss_ce": 0.0033432499039918184,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 592742332,
+      "step": 3451
+    },
+    {
+      "epoch": 0.9080028934043533,
+      "grad_norm": 6.543315599280218,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 592914548,
+      "step": 3452
+    },
+    {
+      "epoch": 0.9080028934043533,
+      "loss": 0.1184120774269104,
+      "loss_ce": 0.00665670819580555,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 592914548,
+      "step": 3452
+    },
+    {
+      "epoch": 0.9082659301637404,
+      "grad_norm": 17.663325985728193,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 593086732,
+      "step": 3453
+    },
+    {
+      "epoch": 0.9082659301637404,
+      "loss": 0.10038851201534271,
+      "loss_ce": 0.0001993027253774926,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 593086732,
+      "step": 3453
+    },
+    {
+      "epoch": 0.9085289669231275,
+      "grad_norm": 4.446508836333715,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 593257004,
+      "step": 3454
+    },
+    {
+      "epoch": 0.9085289669231275,
+      "loss": 0.10031658411026001,
+      "loss_ce": 0.0005546216852962971,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 593257004,
+      "step": 3454
+    },
+    {
+      "epoch": 0.9087920036825147,
+      "grad_norm": 33.88108016172298,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 593429128,
+      "step": 3455
+    },
+    {
+      "epoch": 0.9087920036825147,
+      "loss": 0.08902574330568314,
+      "loss_ce": 0.0005247698863968253,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 593429128,
+      "step": 3455
+    },
+    {
+      "epoch": 0.9090550404419018,
+      "grad_norm": 5.647244857793579,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 593601516,
+      "step": 3456
+    },
+    {
+      "epoch": 0.9090550404419018,
+      "loss": 0.05231146514415741,
+      "loss_ce": 0.001621769741177559,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 593601516,
+      "step": 3456
+    },
+    {
+      "epoch": 0.9093180772012889,
+      "grad_norm": 9.778948998175075,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 593773792,
+      "step": 3457
+    },
+    {
+      "epoch": 0.9093180772012889,
+      "loss": 0.09074349701404572,
+      "loss_ce": 0.0006250919541344047,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 593773792,
+      "step": 3457
+    },
+    {
+      "epoch": 0.909581113960676,
+      "grad_norm": 8.900238633139468,
+      "learning_rate": 5e-06,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 593945832,
+      "step": 3458
+    },
+    {
+      "epoch": 0.909581113960676,
+      "loss": 0.10212016105651855,
+      "loss_ce": 0.0015037069097161293,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 593945832,
+      "step": 3458
+    },
+    {
+      "epoch": 0.9098441507200631,
+      "grad_norm": 5.072537904030542,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 594118188,
+      "step": 3459
+    },
+    {
+      "epoch": 0.9098441507200631,
+      "loss": 0.051810335367918015,
+      "loss_ce": 0.0029516899958252907,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 594118188,
+      "step": 3459
+    },
+    {
+      "epoch": 0.9101071874794503,
+      "grad_norm": 4.7625991934389305,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 594290712,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9101071874794503,
+      "loss": 0.07667580991983414,
+      "loss_ce": 0.00038186419988051057,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 594290712,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9103702242388374,
+      "grad_norm": 10.907454103811546,
+      "learning_rate": 5e-06,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 594461168,
+      "step": 3461
+    },
+    {
+      "epoch": 0.9103702242388374,
+      "loss": 0.07658274471759796,
+      "loss_ce": 0.00019724905723705888,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 594461168,
+      "step": 3461
+    },
+    {
+      "epoch": 0.9106332609982245,
+      "grad_norm": 6.097275668896845,
+      "learning_rate": 5e-06,
+      "loss": 0.1622,
+      "num_input_tokens_seen": 594633236,
+      "step": 3462
+    },
+    {
+      "epoch": 0.9106332609982245,
+      "loss": 0.16042682528495789,
+      "loss_ce": 0.0004689389606937766,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 594633236,
+      "step": 3462
+    },
+    {
+      "epoch": 0.9108962977576116,
+      "grad_norm": 4.28871295829322,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 594805240,
+      "step": 3463
+    },
+    {
+      "epoch": 0.9108962977576116,
+      "loss": 0.10126248002052307,
+      "loss_ce": 0.0024465657770633698,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 594805240,
+      "step": 3463
+    },
+    {
+      "epoch": 0.9111593345169987,
+      "grad_norm": 29.617769917448427,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 594977116,
+      "step": 3464
+    },
+    {
+      "epoch": 0.9111593345169987,
+      "loss": 0.08838079869747162,
+      "loss_ce": 0.0014514753129333258,
+      "loss_iou": 0.421875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 594977116,
+      "step": 3464
+    },
+    {
+      "epoch": 0.9114223712763859,
+      "grad_norm": 12.912626025390432,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 595148996,
+      "step": 3465
+    },
+    {
+      "epoch": 0.9114223712763859,
+      "loss": 0.05163312330842018,
+      "loss_ce": 0.0015690373256802559,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 595148996,
+      "step": 3465
+    },
+    {
+      "epoch": 0.911685408035773,
+      "grad_norm": 3.7162474056368,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 595321500,
+      "step": 3466
+    },
+    {
+      "epoch": 0.911685408035773,
+      "loss": 0.12145687639713287,
+      "loss_ce": 0.0029113469645380974,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 595321500,
+      "step": 3466
+    },
+    {
+      "epoch": 0.9119484447951601,
+      "grad_norm": 26.611038275341258,
+      "learning_rate": 5e-06,
+      "loss": 0.0815,
+      "num_input_tokens_seen": 595493996,
+      "step": 3467
+    },
+    {
+      "epoch": 0.9119484447951601,
+      "loss": 0.08623991906642914,
+      "loss_ce": 8.879496454028413e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 595493996,
+      "step": 3467
+    },
+    {
+      "epoch": 0.9122114815545472,
+      "grad_norm": 3.8019745447441835,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 595666232,
+      "step": 3468
+    },
+    {
+      "epoch": 0.9122114815545472,
+      "loss": 0.10731638222932816,
+      "loss_ce": 0.0009015857940539718,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 595666232,
+      "step": 3468
+    },
+    {
+      "epoch": 0.9124745183139343,
+      "grad_norm": 10.09490833472444,
+      "learning_rate": 5e-06,
+      "loss": 0.126,
+      "num_input_tokens_seen": 595838532,
+      "step": 3469
+    },
+    {
+      "epoch": 0.9124745183139343,
+      "loss": 0.16073833405971527,
+      "loss_ce": 0.0017417498165741563,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 595838532,
+      "step": 3469
+    },
+    {
+      "epoch": 0.9127375550733215,
+      "grad_norm": 2.9635391739801356,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 596010728,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9127375550733215,
+      "loss": 0.09987487643957138,
+      "loss_ce": 0.0005096413660794497,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 596010728,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9130005918327087,
+      "grad_norm": 5.819990486053208,
+      "learning_rate": 5e-06,
+      "loss": 0.077,
+      "num_input_tokens_seen": 596183036,
+      "step": 3471
+    },
+    {
+      "epoch": 0.9130005918327087,
+      "loss": 0.08611226826906204,
+      "loss_ce": 0.0005867574363946915,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 596183036,
+      "step": 3471
+    },
+    {
+      "epoch": 0.9132636285920958,
+      "grad_norm": 13.706964401152458,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 596353748,
+      "step": 3472
+    },
+    {
+      "epoch": 0.9132636285920958,
+      "loss": 0.11669529974460602,
+      "loss_ce": 0.007839101366698742,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 596353748,
+      "step": 3472
+    },
+    {
+      "epoch": 0.9135266653514829,
+      "grad_norm": 12.86793443272329,
+      "learning_rate": 5e-06,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 596525804,
+      "step": 3473
+    },
+    {
+      "epoch": 0.9135266653514829,
+      "loss": 0.14639021456241608,
+      "loss_ce": 0.002057329285889864,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 596525804,
+      "step": 3473
+    },
+    {
+      "epoch": 0.91378970211087,
+      "grad_norm": 3.549687957056835,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 596698176,
+      "step": 3474
+    },
+    {
+      "epoch": 0.91378970211087,
+      "loss": 0.07896921038627625,
+      "loss_ce": 0.0026447533164173365,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 596698176,
+      "step": 3474
+    },
+    {
+      "epoch": 0.9140527388702571,
+      "grad_norm": 7.626577083859758,
+      "learning_rate": 5e-06,
+      "loss": 0.0851,
+      "num_input_tokens_seen": 596870272,
+      "step": 3475
+    },
+    {
+      "epoch": 0.9140527388702571,
+      "loss": 0.110136017203331,
+      "loss_ce": 0.000928854919038713,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 596870272,
+      "step": 3475
+    },
+    {
+      "epoch": 0.9143157756296443,
+      "grad_norm": 4.749367923818828,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 597042624,
+      "step": 3476
+    },
+    {
+      "epoch": 0.9143157756296443,
+      "loss": 0.1048622876405716,
+      "loss_ce": 0.0006752688204869628,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 597042624,
+      "step": 3476
+    },
+    {
+      "epoch": 0.9145788123890314,
+      "grad_norm": 7.997778290032009,
+      "learning_rate": 5e-06,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 597214744,
+      "step": 3477
+    },
+    {
+      "epoch": 0.9145788123890314,
+      "loss": 0.0939270555973053,
+      "loss_ce": 0.00036016173544339836,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 597214744,
+      "step": 3477
+    },
+    {
+      "epoch": 0.9148418491484185,
+      "grad_norm": 9.635212836595876,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 597387080,
+      "step": 3478
+    },
+    {
+      "epoch": 0.9148418491484185,
+      "loss": 0.12316415458917618,
+      "loss_ce": 0.0010938385967165232,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 597387080,
+      "step": 3478
+    },
+    {
+      "epoch": 0.9151048859078056,
+      "grad_norm": 3.963240318321649,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 597559392,
+      "step": 3479
+    },
+    {
+      "epoch": 0.9151048859078056,
+      "loss": 0.05641660839319229,
+      "loss_ce": 0.0005084058502689004,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 597559392,
+      "step": 3479
+    },
+    {
+      "epoch": 0.9153679226671927,
+      "grad_norm": 3.4338009584504947,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 597731528,
+      "step": 3480
+    },
+    {
+      "epoch": 0.9153679226671927,
+      "loss": 0.09766215085983276,
+      "loss_ce": 0.001623332966119051,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 597731528,
+      "step": 3480
+    },
+    {
+      "epoch": 0.9156309594265799,
+      "grad_norm": 7.619842626421138,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 597903520,
+      "step": 3481
+    },
+    {
+      "epoch": 0.9156309594265799,
+      "loss": 0.04850924387574196,
+      "loss_ce": 0.0030533126555383205,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 597903520,
+      "step": 3481
+    },
+    {
+      "epoch": 0.915893996185967,
+      "grad_norm": 3.2410150384393765,
+      "learning_rate": 5e-06,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 598075560,
+      "step": 3482
+    },
+    {
+      "epoch": 0.915893996185967,
+      "loss": 0.06452830880880356,
+      "loss_ce": 0.00015147785597946495,
+      "loss_iou": NaN,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 598075560,
+      "step": 3482
+    },
+    {
+      "epoch": 0.9161570329453541,
+      "grad_norm": 3.618001442656621,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 598247796,
+      "step": 3483
+    },
+    {
+      "epoch": 0.9161570329453541,
+      "loss": 0.07420238852500916,
+      "loss_ce": 0.00047191951307468116,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 598247796,
+      "step": 3483
+    },
+    {
+      "epoch": 0.9164200697047412,
+      "grad_norm": 10.045861260608898,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 598419940,
+      "step": 3484
+    },
+    {
+      "epoch": 0.9164200697047412,
+      "loss": 0.1423761397600174,
+      "loss_ce": 0.0001794886775314808,
+      "loss_iou": NaN,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 598419940,
+      "step": 3484
+    },
+    {
+      "epoch": 0.9166831064641283,
+      "grad_norm": 13.006322984837828,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 598592140,
+      "step": 3485
+    },
+    {
+      "epoch": 0.9166831064641283,
+      "loss": 0.08574345707893372,
+      "loss_ce": 0.0002637250581756234,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 598592140,
+      "step": 3485
+    },
+    {
+      "epoch": 0.9169461432235155,
+      "grad_norm": 7.24999783542604,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 598764500,
+      "step": 3486
+    },
+    {
+      "epoch": 0.9169461432235155,
+      "loss": 0.0942230224609375,
+      "loss_ce": 0.002975467825308442,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 598764500,
+      "step": 3486
+    },
+    {
+      "epoch": 0.9172091799829026,
+      "grad_norm": 4.120990685906085,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 598936608,
+      "step": 3487
+    },
+    {
+      "epoch": 0.9172091799829026,
+      "loss": 0.08670195192098618,
+      "loss_ce": 0.0005203153123147786,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 598936608,
+      "step": 3487
+    },
+    {
+      "epoch": 0.9174722167422897,
+      "grad_norm": 7.691079480924726,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 599109116,
+      "step": 3488
+    },
+    {
+      "epoch": 0.9174722167422897,
+      "loss": 0.07726689428091049,
+      "loss_ce": 0.0006982971681281924,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 599109116,
+      "step": 3488
+    },
+    {
+      "epoch": 0.9177352535016768,
+      "grad_norm": 5.641902324014326,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 599281296,
+      "step": 3489
+    },
+    {
+      "epoch": 0.9177352535016768,
+      "loss": 0.07277127355337143,
+      "loss_ce": 0.0036947373300790787,
+      "loss_iou": 0.375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 599281296,
+      "step": 3489
+    },
+    {
+      "epoch": 0.9179982902610639,
+      "grad_norm": 5.439497936856968,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 599450632,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9179982902610639,
+      "loss": 0.08008137345314026,
+      "loss_ce": 0.0042757089249789715,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 599450632,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9182613270204512,
+      "grad_norm": 3.272614734962344,
+      "learning_rate": 5e-06,
+      "loss": 0.1474,
+      "num_input_tokens_seen": 599622648,
+      "step": 3491
+    },
+    {
+      "epoch": 0.9182613270204512,
+      "loss": 0.21320411562919617,
+      "loss_ce": 0.002327651483938098,
+      "loss_iou": 0.296875,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 599622648,
+      "step": 3491
+    },
+    {
+      "epoch": 0.9185243637798383,
+      "grad_norm": 4.968216314604122,
+      "learning_rate": 5e-06,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 599793016,
+      "step": 3492
+    },
+    {
+      "epoch": 0.9185243637798383,
+      "loss": 0.05503164976835251,
+      "loss_ce": 0.001076570013538003,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 599793016,
+      "step": 3492
+    },
+    {
+      "epoch": 0.9187874005392254,
+      "grad_norm": 13.600408653665847,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 599965196,
+      "step": 3493
+    },
+    {
+      "epoch": 0.9187874005392254,
+      "loss": 0.2306230664253235,
+      "loss_ce": 0.002626231173053384,
+      "loss_iou": 0.453125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 599965196,
+      "step": 3493
+    },
+    {
+      "epoch": 0.9190504372986125,
+      "grad_norm": 13.080198547497895,
+      "learning_rate": 5e-06,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 600137328,
+      "step": 3494
+    },
+    {
+      "epoch": 0.9190504372986125,
+      "loss": 0.0880010575056076,
+      "loss_ce": 0.0009954443667083979,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 600137328,
+      "step": 3494
+    },
+    {
+      "epoch": 0.9193134740579996,
+      "grad_norm": 6.526249469256959,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 600309400,
+      "step": 3495
+    },
+    {
+      "epoch": 0.9193134740579996,
+      "loss": 0.11348491907119751,
+      "loss_ce": 0.000966615742072463,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 600309400,
+      "step": 3495
+    },
+    {
+      "epoch": 0.9195765108173868,
+      "grad_norm": 8.360743966936655,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 600481616,
+      "step": 3496
+    },
+    {
+      "epoch": 0.9195765108173868,
+      "loss": 0.09168469160795212,
+      "loss_ce": 0.0012611029669642448,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 600481616,
+      "step": 3496
+    },
+    {
+      "epoch": 0.9198395475767739,
+      "grad_norm": 6.554644415896188,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 600653744,
+      "step": 3497
+    },
+    {
+      "epoch": 0.9198395475767739,
+      "loss": 0.11767074465751648,
+      "loss_ce": 0.0016429107636213303,
+      "loss_iou": 0.59375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 600653744,
+      "step": 3497
+    },
+    {
+      "epoch": 0.920102584336161,
+      "grad_norm": 16.679597431920058,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 600824128,
+      "step": 3498
+    },
+    {
+      "epoch": 0.920102584336161,
+      "loss": 0.14813083410263062,
+      "loss_ce": 0.0007919695926830173,
+      "loss_iou": NaN,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 600824128,
+      "step": 3498
+    },
+    {
+      "epoch": 0.9203656210955481,
+      "grad_norm": 4.858084855581349,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 600996452,
+      "step": 3499
+    },
+    {
+      "epoch": 0.9203656210955481,
+      "loss": 0.10890492051839828,
+      "loss_ce": 0.0007811367395333946,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 600996452,
+      "step": 3499
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "grad_norm": 5.0714335035388896,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_websight_new_CIoU": 0.8971402049064636,
+      "eval_websight_new_GIoU": 0.8995657861232758,
+      "eval_websight_new_IoU": 0.9006073176860809,
+      "eval_websight_new_MAE_all": 0.013850971590727568,
+      "eval_websight_new_MAE_h": 0.00682047987356782,
+      "eval_websight_new_MAE_w": 0.02169650699943304,
+      "eval_websight_new_MAE_x": 0.022679010406136513,
+      "eval_websight_new_MAE_y": 0.004207887570373714,
+      "eval_websight_new_NUM_probability": 0.9999927878379822,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.07072407752275467,
+      "eval_websight_new_loss_ce": 9.538403446640586e-06,
+      "eval_websight_new_loss_iou": 0.32720947265625,
+      "eval_websight_new_loss_num": 0.012699127197265625,
+      "eval_websight_new_loss_xval": 0.06354522705078125,
+      "eval_websight_new_runtime": 56.2114,
+      "eval_websight_new_samples_per_second": 0.889,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_seeclick_CIoU": 0.6764970123767853,
+      "eval_seeclick_GIoU": 0.6781544089317322,
+      "eval_seeclick_IoU": 0.6978051662445068,
+      "eval_seeclick_MAE_all": 0.040375180542469025,
+      "eval_seeclick_MAE_h": 0.021458005532622337,
+      "eval_seeclick_MAE_w": 0.05631308630108833,
+      "eval_seeclick_MAE_x": 0.06287308409810066,
+      "eval_seeclick_MAE_y": 0.02085655089467764,
+      "eval_seeclick_NUM_probability": 0.9999865889549255,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.17490381002426147,
+      "eval_seeclick_loss_ce": 0.008978934027254581,
+      "eval_seeclick_loss_iou": 0.46295166015625,
+      "eval_seeclick_loss_num": 0.031360626220703125,
+      "eval_seeclick_loss_xval": 0.1567230224609375,
+      "eval_seeclick_runtime": 74.0396,
+      "eval_seeclick_samples_per_second": 0.581,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_icons_CIoU": 0.8881092965602875,
+      "eval_icons_GIoU": 0.8860294818878174,
+      "eval_icons_IoU": 0.8926480114459991,
+      "eval_icons_MAE_all": 0.0159080708399415,
+      "eval_icons_MAE_h": 0.020047838799655437,
+      "eval_icons_MAE_w": 0.015083736274391413,
+      "eval_icons_MAE_x": 0.013403147924691439,
+      "eval_icons_MAE_y": 0.015097561292350292,
+      "eval_icons_NUM_probability": 0.9999927878379822,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.060060471296310425,
+      "eval_icons_loss_ce": 9.6267791604987e-06,
+      "eval_icons_loss_iou": 0.605712890625,
+      "eval_icons_loss_num": 0.01105499267578125,
+      "eval_icons_loss_xval": 0.05532073974609375,
+      "eval_icons_runtime": 80.8235,
+      "eval_icons_samples_per_second": 0.619,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_screenspot_CIoU": 0.5590948661168417,
+      "eval_screenspot_GIoU": 0.5591723521550497,
+      "eval_screenspot_IoU": 0.5979611476262411,
+      "eval_screenspot_MAE_all": 0.08105809738238652,
+      "eval_screenspot_MAE_h": 0.05730322003364563,
+      "eval_screenspot_MAE_w": 0.1365982194741567,
+      "eval_screenspot_MAE_x": 0.0790914719303449,
+      "eval_screenspot_MAE_y": 0.05123948057492574,
+      "eval_screenspot_NUM_probability": 0.9999733567237854,
+      "eval_screenspot_inside_bbox": 0.8841666579246521,
+      "eval_screenspot_loss": 0.9397080540657043,
+      "eval_screenspot_loss_ce": 0.5959697167078654,
+      "eval_screenspot_loss_iou": 0.539306640625,
+      "eval_screenspot_loss_num": 0.06738789876302083,
+      "eval_screenspot_loss_xval": 0.3368326822916667,
+      "eval_screenspot_runtime": 151.1273,
+      "eval_screenspot_samples_per_second": 0.589,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "loss": 0.9230542182922363,
+      "loss_ce": 0.5880932807922363,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9208916946143223,
+      "grad_norm": 3.0758731887113147,
+      "learning_rate": 5e-06,
+      "loss": 0.0803,
+      "num_input_tokens_seen": 601340840,
+      "step": 3501
+    },
+    {
+      "epoch": 0.9208916946143223,
+      "loss": 0.07921823859214783,
+      "loss_ce": 0.0005973259685561061,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 601340840,
+      "step": 3501
+    },
+    {
+      "epoch": 0.9211547313737095,
+      "grad_norm": 12.765305175609484,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 601513112,
+      "step": 3502
+    },
+    {
+      "epoch": 0.9211547313737095,
+      "loss": 0.13090460002422333,
+      "loss_ce": 0.002868090523406863,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 601513112,
+      "step": 3502
+    },
+    {
+      "epoch": 0.9214177681330966,
+      "grad_norm": 3.071857580938524,
+      "learning_rate": 5e-06,
+      "loss": 0.0751,
+      "num_input_tokens_seen": 601685348,
+      "step": 3503
+    },
+    {
+      "epoch": 0.9214177681330966,
+      "loss": 0.11101450026035309,
+      "loss_ce": 0.002280366839841008,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 601685348,
+      "step": 3503
+    },
+    {
+      "epoch": 0.9216808048924837,
+      "grad_norm": 3.8727649532640296,
+      "learning_rate": 5e-06,
+      "loss": 0.0905,
+      "num_input_tokens_seen": 601857460,
+      "step": 3504
+    },
+    {
+      "epoch": 0.9216808048924837,
+      "loss": 0.0590723380446434,
+      "loss_ce": 0.0022028274834156036,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 601857460,
+      "step": 3504
+    },
+    {
+      "epoch": 0.9219438416518708,
+      "grad_norm": 12.18479200262375,
+      "learning_rate": 5e-06,
+      "loss": 0.0989,
+      "num_input_tokens_seen": 602029992,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9219438416518708,
+      "loss": 0.12037432193756104,
+      "loss_ce": 0.0029426885303109884,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 602029992,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9222068784112579,
+      "grad_norm": 4.289786395943063,
+      "learning_rate": 5e-06,
+      "loss": 0.1091,
+      "num_input_tokens_seen": 602202184,
+      "step": 3506
+    },
+    {
+      "epoch": 0.9222068784112579,
+      "loss": 0.16438013315200806,
+      "loss_ce": 0.00022608340077567846,
+      "loss_iou": 0.453125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 602202184,
+      "step": 3506
+    },
+    {
+      "epoch": 0.9224699151706451,
+      "grad_norm": 27.091794274961455,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 602372792,
+      "step": 3507
+    },
+    {
+      "epoch": 0.9224699151706451,
+      "loss": 0.081387460231781,
+      "loss_ce": 0.0013398483861237764,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 602372792,
+      "step": 3507
+    },
+    {
+      "epoch": 0.9227329519300322,
+      "grad_norm": 8.942524792614973,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 602542048,
+      "step": 3508
+    },
+    {
+      "epoch": 0.9227329519300322,
+      "loss": 0.11642280220985413,
+      "loss_ce": 0.005247259978204966,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 602542048,
+      "step": 3508
+    },
+    {
+      "epoch": 0.9229959886894193,
+      "grad_norm": 5.673853173524807,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 602714148,
+      "step": 3509
+    },
+    {
+      "epoch": 0.9229959886894193,
+      "loss": 0.22935867309570312,
+      "loss_ce": 0.0008430513553321362,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 602714148,
+      "step": 3509
+    },
+    {
+      "epoch": 0.9232590254488064,
+      "grad_norm": 4.438849137936148,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 602886264,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9232590254488064,
+      "loss": 0.1615859568119049,
+      "loss_ce": 0.0012771158944815397,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 602886264,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9235220622081936,
+      "grad_norm": 4.872038444582204,
+      "learning_rate": 5e-06,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 603058732,
+      "step": 3511
+    },
+    {
+      "epoch": 0.9235220622081936,
+      "loss": 0.13943278789520264,
+      "loss_ce": 0.0005015181959606707,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 603058732,
+      "step": 3511
+    },
+    {
+      "epoch": 0.9237850989675808,
+      "grad_norm": 4.92552213324199,
+      "learning_rate": 5e-06,
+      "loss": 0.1144,
+      "num_input_tokens_seen": 603230928,
+      "step": 3512
+    },
+    {
+      "epoch": 0.9237850989675808,
+      "loss": 0.14707276225090027,
+      "loss_ce": 0.0023278831504285336,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 603230928,
+      "step": 3512
+    },
+    {
+      "epoch": 0.9240481357269679,
+      "grad_norm": 8.817456239528582,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 603403300,
+      "step": 3513
+    },
+    {
+      "epoch": 0.9240481357269679,
+      "loss": 0.12186002731323242,
+      "loss_ce": 0.0018343898700550199,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 603403300,
+      "step": 3513
+    },
+    {
+      "epoch": 0.924311172486355,
+      "grad_norm": 10.676170008596674,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 603575704,
+      "step": 3514
+    },
+    {
+      "epoch": 0.924311172486355,
+      "loss": 0.15692317485809326,
+      "loss_ce": 0.004670977126806974,
+      "loss_iou": NaN,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 603575704,
+      "step": 3514
+    },
+    {
+      "epoch": 0.9245742092457421,
+      "grad_norm": 11.658802940011123,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 603745852,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9245742092457421,
+      "loss": 0.08981953561306,
+      "loss_ce": 0.00015888996131252497,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 603745852,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9248372460051292,
+      "grad_norm": 11.699348951917594,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 603918360,
+      "step": 3516
+    },
+    {
+      "epoch": 0.9248372460051292,
+      "loss": 0.1186264157295227,
+      "loss_ce": 0.0009811592753976583,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 603918360,
+      "step": 3516
+    },
+    {
+      "epoch": 0.9251002827645164,
+      "grad_norm": 16.59746302136246,
+      "learning_rate": 5e-06,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 604090468,
+      "step": 3517
+    },
+    {
+      "epoch": 0.9251002827645164,
+      "loss": 0.15999768674373627,
+      "loss_ce": 0.0011536948150023818,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 604090468,
+      "step": 3517
+    },
+    {
+      "epoch": 0.9253633195239035,
+      "grad_norm": 7.69976984803875,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 604262492,
+      "step": 3518
+    },
+    {
+      "epoch": 0.9253633195239035,
+      "loss": 0.14189431071281433,
+      "loss_ce": 0.006487809121608734,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 604262492,
+      "step": 3518
+    },
+    {
+      "epoch": 0.9256263562832906,
+      "grad_norm": 12.75000933447967,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 604432856,
+      "step": 3519
+    },
+    {
+      "epoch": 0.9256263562832906,
+      "loss": 0.0861673578619957,
+      "loss_ce": 0.004319215193390846,
+      "loss_iou": 0.484375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 604432856,
+      "step": 3519
+    },
+    {
+      "epoch": 0.9258893930426777,
+      "grad_norm": 5.042427155046337,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 604605056,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9258893930426777,
+      "loss": 0.1412590742111206,
+      "loss_ce": 0.002571945311501622,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 604605056,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9261524298020648,
+      "grad_norm": 5.26588732086772,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 604775372,
+      "step": 3521
+    },
+    {
+      "epoch": 0.9261524298020648,
+      "loss": 0.05718105286359787,
+      "loss_ce": 0.0014254315756261349,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 604775372,
+      "step": 3521
+    },
+    {
+      "epoch": 0.926415466561452,
+      "grad_norm": 6.371747445648948,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 604947376,
+      "step": 3522
+    },
+    {
+      "epoch": 0.926415466561452,
+      "loss": 0.08819465339183807,
+      "loss_ce": 0.0007312724483199418,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 604947376,
+      "step": 3522
+    },
+    {
+      "epoch": 0.9266785033208391,
+      "grad_norm": 10.505886717736445,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 605119568,
+      "step": 3523
+    },
+    {
+      "epoch": 0.9266785033208391,
+      "loss": 0.17979584634304047,
+      "loss_ce": 0.004075629636645317,
+      "loss_iou": 0.609375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 605119568,
+      "step": 3523
+    },
+    {
+      "epoch": 0.9269415400802262,
+      "grad_norm": 6.455190531468335,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 605291524,
+      "step": 3524
+    },
+    {
+      "epoch": 0.9269415400802262,
+      "loss": 0.08140784502029419,
+      "loss_ce": 0.0016959276981651783,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 605291524,
+      "step": 3524
+    },
+    {
+      "epoch": 0.9272045768396133,
+      "grad_norm": 14.745208066661506,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 605463808,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9272045768396133,
+      "loss": 0.11558607965707779,
+      "loss_ce": 0.0017555101076141,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 605463808,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9274676135990004,
+      "grad_norm": 3.8942399340921825,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 605635872,
+      "step": 3526
+    },
+    {
+      "epoch": 0.9274676135990004,
+      "loss": 0.06215044856071472,
+      "loss_ce": 0.00019976735347881913,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 605635872,
+      "step": 3526
+    },
+    {
+      "epoch": 0.9277306503583875,
+      "grad_norm": 4.175922010381311,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 605807804,
+      "step": 3527
+    },
+    {
+      "epoch": 0.9277306503583875,
+      "loss": 0.16303026676177979,
+      "loss_ce": 0.0018974501872435212,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 605807804,
+      "step": 3527
+    },
+    {
+      "epoch": 0.9279936871177747,
+      "grad_norm": 3.844069149031415,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 605979952,
+      "step": 3528
+    },
+    {
+      "epoch": 0.9279936871177747,
+      "loss": 0.0688394084572792,
+      "loss_ce": 0.0023721237666904926,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 605979952,
+      "step": 3528
+    },
+    {
+      "epoch": 0.9282567238771618,
+      "grad_norm": 4.202436418234633,
+      "learning_rate": 5e-06,
+      "loss": 0.081,
+      "num_input_tokens_seen": 606150360,
+      "step": 3529
+    },
+    {
+      "epoch": 0.9282567238771618,
+      "loss": 0.04098789393901825,
+      "loss_ce": 0.00038425601087510586,
+      "loss_iou": 0.5,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 606150360,
+      "step": 3529
+    },
+    {
+      "epoch": 0.928519760636549,
+      "grad_norm": 3.743613714946928,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 606320544,
+      "step": 3530
+    },
+    {
+      "epoch": 0.928519760636549,
+      "loss": 0.12318438291549683,
+      "loss_ce": 0.003891176311299205,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 606320544,
+      "step": 3530
+    },
+    {
+      "epoch": 0.928782797395936,
+      "grad_norm": 7.541725516396188,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 606492452,
+      "step": 3531
+    },
+    {
+      "epoch": 0.928782797395936,
+      "loss": 0.05570812523365021,
+      "loss_ce": 0.0006238996866159141,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 606492452,
+      "step": 3531
+    },
+    {
+      "epoch": 0.9290458341553232,
+      "grad_norm": 7.283637997577956,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 606664904,
+      "step": 3532
+    },
+    {
+      "epoch": 0.9290458341553232,
+      "loss": 0.2060985267162323,
+      "loss_ce": 0.0019969542045146227,
+      "loss_iou": 0.40625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 606664904,
+      "step": 3532
+    },
+    {
+      "epoch": 0.9293088709147104,
+      "grad_norm": 3.96510750197539,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 606836960,
+      "step": 3533
+    },
+    {
+      "epoch": 0.9293088709147104,
+      "loss": 0.07297110557556152,
+      "loss_ce": 0.002353430027142167,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 606836960,
+      "step": 3533
+    },
+    {
+      "epoch": 0.9295719076740975,
+      "grad_norm": 2.973198469624849,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 607009360,
+      "step": 3534
+    },
+    {
+      "epoch": 0.9295719076740975,
+      "loss": 0.13494953513145447,
+      "loss_ce": 0.0017403117381036282,
+      "loss_iou": 0.4375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 607009360,
+      "step": 3534
+    },
+    {
+      "epoch": 0.9298349444334846,
+      "grad_norm": 3.119729004971168,
+      "learning_rate": 5e-06,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 607181304,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9298349444334846,
+      "loss": 0.06605780124664307,
+      "loss_ce": 0.002443910576403141,
+      "loss_iou": 0.421875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 607181304,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9300979811928717,
+      "grad_norm": 4.352773383034863,
+      "learning_rate": 5e-06,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 607353736,
+      "step": 3536
+    },
+    {
+      "epoch": 0.9300979811928717,
+      "loss": 0.06802303344011307,
+      "loss_ce": 0.001021688454784453,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 607353736,
+      "step": 3536
+    },
+    {
+      "epoch": 0.9303610179522588,
+      "grad_norm": 3.9578360400969794,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 607525920,
+      "step": 3537
+    },
+    {
+      "epoch": 0.9303610179522588,
+      "loss": 0.06541258096694946,
+      "loss_ce": 0.00288206129334867,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 607525920,
+      "step": 3537
+    },
+    {
+      "epoch": 0.930624054711646,
+      "grad_norm": 4.492300233458511,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 607697980,
+      "step": 3538
+    },
+    {
+      "epoch": 0.930624054711646,
+      "loss": 0.09697936475276947,
+      "loss_ce": 0.0013983015669509768,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 607697980,
+      "step": 3538
+    },
+    {
+      "epoch": 0.9308870914710331,
+      "grad_norm": 4.302295585236548,
+      "learning_rate": 5e-06,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 607870144,
+      "step": 3539
+    },
+    {
+      "epoch": 0.9308870914710331,
+      "loss": 0.10722782462835312,
+      "loss_ce": 0.000843545887619257,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 607870144,
+      "step": 3539
+    },
+    {
+      "epoch": 0.9311501282304202,
+      "grad_norm": 7.0499450082503365,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 608042240,
+      "step": 3540
+    },
+    {
+      "epoch": 0.9311501282304202,
+      "loss": 0.1015363559126854,
+      "loss_ce": 0.006443582940846682,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 608042240,
+      "step": 3540
+    },
+    {
+      "epoch": 0.9314131649898073,
+      "grad_norm": 6.586834009281099,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 608214284,
+      "step": 3541
+    },
+    {
+      "epoch": 0.9314131649898073,
+      "loss": 0.12908074259757996,
+      "loss_ce": 0.004782632924616337,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 608214284,
+      "step": 3541
+    },
+    {
+      "epoch": 0.9316762017491944,
+      "grad_norm": 4.53807614520661,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 608386616,
+      "step": 3542
+    },
+    {
+      "epoch": 0.9316762017491944,
+      "loss": 0.1289425790309906,
+      "loss_ce": 0.0027523916214704514,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 608386616,
+      "step": 3542
+    },
+    {
+      "epoch": 0.9319392385085816,
+      "grad_norm": 7.59401410995318,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 608558468,
+      "step": 3543
+    },
+    {
+      "epoch": 0.9319392385085816,
+      "loss": 0.16303950548171997,
+      "loss_ce": 0.00520259328186512,
+      "loss_iou": 0.4375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 608558468,
+      "step": 3543
+    },
+    {
+      "epoch": 0.9322022752679687,
+      "grad_norm": 8.796770729596103,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 608730916,
+      "step": 3544
+    },
+    {
+      "epoch": 0.9322022752679687,
+      "loss": 0.11866636574268341,
+      "loss_ce": 0.0005633389810100198,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 608730916,
+      "step": 3544
+    },
+    {
+      "epoch": 0.9324653120273558,
+      "grad_norm": 5.984167827436451,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 608902956,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9324653120273558,
+      "loss": 0.09633171558380127,
+      "loss_ce": 0.000903245760127902,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 608902956,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9327283487867429,
+      "grad_norm": 13.864562508217144,
+      "learning_rate": 5e-06,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 609072616,
+      "step": 3546
+    },
+    {
+      "epoch": 0.9327283487867429,
+      "loss": 0.11617599427700043,
+      "loss_ce": 0.0005448899464681745,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 609072616,
+      "step": 3546
+    },
+    {
+      "epoch": 0.93299138554613,
+      "grad_norm": 7.250119178294138,
+      "learning_rate": 5e-06,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 609244840,
+      "step": 3547
+    },
+    {
+      "epoch": 0.93299138554613,
+      "loss": 0.08952006697654724,
+      "loss_ce": 0.001415827078744769,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 609244840,
+      "step": 3547
+    },
+    {
+      "epoch": 0.9332544223055173,
+      "grad_norm": 8.965223152196344,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 609417116,
+      "step": 3548
+    },
+    {
+      "epoch": 0.9332544223055173,
+      "loss": 0.058124981820583344,
+      "loss_ce": 0.0005077911773696542,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 609417116,
+      "step": 3548
+    },
+    {
+      "epoch": 0.9335174590649044,
+      "grad_norm": 4.578416828961302,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 609589092,
+      "step": 3549
+    },
+    {
+      "epoch": 0.9335174590649044,
+      "loss": 0.06900876015424728,
+      "loss_ce": 0.0006951588438823819,
+      "loss_iou": 0.671875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 609589092,
+      "step": 3549
+    },
+    {
+      "epoch": 0.9337804958242915,
+      "grad_norm": 8.028931966527075,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 609761432,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9337804958242915,
+      "loss": 0.10969488322734833,
+      "loss_ce": 0.0008691949769854546,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 609761432,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9340435325836786,
+      "grad_norm": 3.0608586378113225,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 609933500,
+      "step": 3551
+    },
+    {
+      "epoch": 0.9340435325836786,
+      "loss": 0.1336784064769745,
+      "loss_ce": 0.003475167090073228,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 609933500,
+      "step": 3551
+    },
+    {
+      "epoch": 0.9343065693430657,
+      "grad_norm": 6.227248956666904,
+      "learning_rate": 5e-06,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 610105792,
+      "step": 3552
+    },
+    {
+      "epoch": 0.9343065693430657,
+      "loss": 0.08306320756673813,
+      "loss_ce": 0.007104953285306692,
+      "loss_iou": 0.453125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 610105792,
+      "step": 3552
+    },
+    {
+      "epoch": 0.9345696061024528,
+      "grad_norm": 5.469597791438911,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 610277880,
+      "step": 3553
+    },
+    {
+      "epoch": 0.9345696061024528,
+      "loss": 0.08409252762794495,
+      "loss_ce": 0.002595331287011504,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 610277880,
+      "step": 3553
+    },
+    {
+      "epoch": 0.93483264286184,
+      "grad_norm": 4.1047405914838455,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 610449924,
+      "step": 3554
+    },
+    {
+      "epoch": 0.93483264286184,
+      "loss": 0.10268253833055496,
+      "loss_ce": 0.0010895198211073875,
+      "loss_iou": 0.5,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 610449924,
+      "step": 3554
+    },
+    {
+      "epoch": 0.9350956796212271,
+      "grad_norm": 4.649820698561932,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 610622052,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9350956796212271,
+      "loss": 0.08575969934463501,
+      "loss_ce": 0.0018668812699615955,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 610622052,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9353587163806142,
+      "grad_norm": 9.738163582969452,
+      "learning_rate": 5e-06,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 610794168,
+      "step": 3556
+    },
+    {
+      "epoch": 0.9353587163806142,
+      "loss": 0.17649231851100922,
+      "loss_ce": 0.004281637258827686,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 610794168,
+      "step": 3556
+    },
+    {
+      "epoch": 0.9356217531400013,
+      "grad_norm": 60.60865892554417,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 610964304,
+      "step": 3557
+    },
+    {
+      "epoch": 0.9356217531400013,
+      "loss": 0.21576589345932007,
+      "loss_ce": 0.01731007918715477,
+      "loss_iou": 0.390625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 610964304,
+      "step": 3557
+    },
+    {
+      "epoch": 0.9358847898993884,
+      "grad_norm": 26.41744180605267,
+      "learning_rate": 5e-06,
+      "loss": 0.278,
+      "num_input_tokens_seen": 611136572,
+      "step": 3558
+    },
+    {
+      "epoch": 0.9358847898993884,
+      "loss": 0.2503662705421448,
+      "loss_ce": 0.15902717411518097,
+      "loss_iou": 0.515625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 611136572,
+      "step": 3558
+    },
+    {
+      "epoch": 0.9361478266587756,
+      "grad_norm": 31.460812663910158,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 611308712,
+      "step": 3559
+    },
+    {
+      "epoch": 0.9361478266587756,
+      "loss": 0.10697901993989944,
+      "loss_ce": 0.014724383130669594,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 611308712,
+      "step": 3559
+    },
+    {
+      "epoch": 0.9364108634181627,
+      "grad_norm": 125.0152995081179,
+      "learning_rate": 5e-06,
+      "loss": 0.2523,
+      "num_input_tokens_seen": 611481064,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9364108634181627,
+      "loss": 0.19752028584480286,
+      "loss_ce": 0.055674582719802856,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 611481064,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9366739001775498,
+      "grad_norm": 42.388335248969774,
+      "learning_rate": 5e-06,
+      "loss": 0.6327,
+      "num_input_tokens_seen": 611653432,
+      "step": 3561
+    },
+    {
+      "epoch": 0.9366739001775498,
+      "loss": 0.6363104581832886,
+      "loss_ce": 0.5759619474411011,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 611653432,
+      "step": 3561
+    },
+    {
+      "epoch": 0.9369369369369369,
+      "grad_norm": 34.483421318180156,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 611825552,
+      "step": 3562
+    },
+    {
+      "epoch": 0.9369369369369369,
+      "loss": 0.10395447909832001,
+      "loss_ce": 0.05541627109050751,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 611825552,
+      "step": 3562
+    },
+    {
+      "epoch": 0.937199973696324,
+      "grad_norm": 3.5466208788177513,
+      "learning_rate": 5e-06,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 611997960,
+      "step": 3563
+    },
+    {
+      "epoch": 0.937199973696324,
+      "loss": 0.1155683621764183,
+      "loss_ce": 0.0075666578486561775,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 611997960,
+      "step": 3563
+    },
+    {
+      "epoch": 0.9374630104557112,
+      "grad_norm": 6.556188606634138,
+      "learning_rate": 5e-06,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 612170180,
+      "step": 3564
+    },
+    {
+      "epoch": 0.9374630104557112,
+      "loss": 0.0452946312725544,
+      "loss_ce": 0.0018070839578285813,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 612170180,
+      "step": 3564
+    },
+    {
+      "epoch": 0.9377260472150983,
+      "grad_norm": 12.949003378413929,
+      "learning_rate": 5e-06,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 612342360,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9377260472150983,
+      "loss": 0.16198524832725525,
+      "loss_ce": 0.001249166438356042,
+      "loss_iou": 0.625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 612342360,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9379890839744854,
+      "grad_norm": 5.1711873858566335,
+      "learning_rate": 5e-06,
+      "loss": 0.074,
+      "num_input_tokens_seen": 612514340,
+      "step": 3566
+    },
+    {
+      "epoch": 0.9379890839744854,
+      "loss": 0.050079330801963806,
+      "loss_ce": 0.0008087016176432371,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 612514340,
+      "step": 3566
+    },
+    {
+      "epoch": 0.9382521207338725,
+      "grad_norm": 15.271109982643123,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 612686644,
+      "step": 3567
+    },
+    {
+      "epoch": 0.9382521207338725,
+      "loss": 0.051343828439712524,
+      "loss_ce": 0.0003336968075018376,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 612686644,
+      "step": 3567
+    },
+    {
+      "epoch": 0.9385151574932596,
+      "grad_norm": 6.894404249326668,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 612854000,
+      "step": 3568
+    },
+    {
+      "epoch": 0.9385151574932596,
+      "loss": 0.1327725201845169,
+      "loss_ce": 0.0005093337967991829,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 612854000,
+      "step": 3568
+    },
+    {
+      "epoch": 0.9387781942526469,
+      "grad_norm": 16.019697398064586,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 613026136,
+      "step": 3569
+    },
+    {
+      "epoch": 0.9387781942526469,
+      "loss": 0.14443224668502808,
+      "loss_ce": 0.0018541140016168356,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 613026136,
+      "step": 3569
+    },
+    {
+      "epoch": 0.939041231012034,
+      "grad_norm": 2.778217951943394,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 613198388,
+      "step": 3570
+    },
+    {
+      "epoch": 0.939041231012034,
+      "loss": 0.05536004900932312,
+      "loss_ce": 0.0027019698172807693,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 613198388,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9393042677714211,
+      "grad_norm": 4.873345772996428,
+      "learning_rate": 5e-06,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 613370288,
+      "step": 3571
+    },
+    {
+      "epoch": 0.9393042677714211,
+      "loss": 0.09475830942392349,
+      "loss_ce": 0.0007489121053367853,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 613370288,
+      "step": 3571
+    },
+    {
+      "epoch": 0.9395673045308082,
+      "grad_norm": 6.737346874423115,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 613542216,
+      "step": 3572
+    },
+    {
+      "epoch": 0.9395673045308082,
+      "loss": 0.2296835035085678,
+      "loss_ce": 0.005287751089781523,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 613542216,
+      "step": 3572
+    },
+    {
+      "epoch": 0.9398303412901953,
+      "grad_norm": 5.428921826665437,
+      "learning_rate": 5e-06,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 613714320,
+      "step": 3573
+    },
+    {
+      "epoch": 0.9398303412901953,
+      "loss": 0.04020649567246437,
+      "loss_ce": 0.002578320913016796,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 613714320,
+      "step": 3573
+    },
+    {
+      "epoch": 0.9400933780495825,
+      "grad_norm": 5.730418477266325,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 613884004,
+      "step": 3574
+    },
+    {
+      "epoch": 0.9400933780495825,
+      "loss": 0.12099497765302658,
+      "loss_ce": 0.000435287831351161,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 613884004,
+      "step": 3574
+    },
+    {
+      "epoch": 0.9403564148089696,
+      "grad_norm": 2.8371958686948315,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 614056360,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9403564148089696,
+      "loss": 0.06765338778495789,
+      "loss_ce": 0.00017902448598761111,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 614056360,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9406194515683567,
+      "grad_norm": 3.3483226181812173,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 614228248,
+      "step": 3576
+    },
+    {
+      "epoch": 0.9406194515683567,
+      "loss": 0.07497625052928925,
+      "loss_ce": 0.0009863873710855842,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 614228248,
+      "step": 3576
+    },
+    {
+      "epoch": 0.9408824883277438,
+      "grad_norm": 3.0470015747124837,
+      "learning_rate": 5e-06,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 614400460,
+      "step": 3577
+    },
+    {
+      "epoch": 0.9408824883277438,
+      "loss": 0.15504948794841766,
+      "loss_ce": 0.005391271784901619,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 614400460,
+      "step": 3577
+    },
+    {
+      "epoch": 0.9411455250871309,
+      "grad_norm": 14.590254214221755,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 614572668,
+      "step": 3578
+    },
+    {
+      "epoch": 0.9411455250871309,
+      "loss": 0.12589725852012634,
+      "loss_ce": 0.00352177070453763,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 614572668,
+      "step": 3578
+    },
+    {
+      "epoch": 0.941408561846518,
+      "grad_norm": 4.098895136914831,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 614744728,
+      "step": 3579
+    },
+    {
+      "epoch": 0.941408561846518,
+      "loss": 0.07378913462162018,
+      "loss_ce": 0.0005316926399245858,
+      "loss_iou": 0.5,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 614744728,
+      "step": 3579
+    },
+    {
+      "epoch": 0.9416715986059052,
+      "grad_norm": 6.551103187216315,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 614917064,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9416715986059052,
+      "loss": 0.11033067107200623,
+      "loss_ce": 0.0037785512395203114,
+      "loss_iou": 0.625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 614917064,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9419346353652923,
+      "grad_norm": 6.125579245234497,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 615089092,
+      "step": 3581
+    },
+    {
+      "epoch": 0.9419346353652923,
+      "loss": 0.09463340044021606,
+      "loss_ce": 0.002287208568304777,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 615089092,
+      "step": 3581
+    },
+    {
+      "epoch": 0.9421976721246794,
+      "grad_norm": 4.52462668676838,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 615259732,
+      "step": 3582
+    },
+    {
+      "epoch": 0.9421976721246794,
+      "loss": 0.09844870865345001,
+      "loss_ce": 0.0007771998061798513,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 615259732,
+      "step": 3582
+    },
+    {
+      "epoch": 0.9424607088840665,
+      "grad_norm": 4.261037832557463,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 615431984,
+      "step": 3583
+    },
+    {
+      "epoch": 0.9424607088840665,
+      "loss": 0.10644324868917465,
+      "loss_ce": 0.0039041785057634115,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 615431984,
+      "step": 3583
+    },
+    {
+      "epoch": 0.9427237456434536,
+      "grad_norm": 11.026204038590599,
+      "learning_rate": 5e-06,
+      "loss": 0.0974,
+      "num_input_tokens_seen": 615604052,
+      "step": 3584
+    },
+    {
+      "epoch": 0.9427237456434536,
+      "loss": 0.08766089379787445,
+      "loss_ce": 0.0031577199697494507,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 615604052,
+      "step": 3584
+    },
+    {
+      "epoch": 0.9429867824028408,
+      "grad_norm": 4.892521797802774,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 615776672,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9429867824028408,
+      "loss": 0.04693538695573807,
+      "loss_ce": 0.0009148788521997631,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 615776672,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9432498191622279,
+      "grad_norm": 8.102761106727687,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 615948804,
+      "step": 3586
+    },
+    {
+      "epoch": 0.9432498191622279,
+      "loss": 0.0826391950249672,
+      "loss_ce": 0.00152347341645509,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 615948804,
+      "step": 3586
+    },
+    {
+      "epoch": 0.943512855921615,
+      "grad_norm": 4.699200150667693,
+      "learning_rate": 5e-06,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 616120904,
+      "step": 3587
+    },
+    {
+      "epoch": 0.943512855921615,
+      "loss": 0.07182273268699646,
+      "loss_ce": 0.004500953480601311,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 616120904,
+      "step": 3587
+    },
+    {
+      "epoch": 0.9437758926810021,
+      "grad_norm": 5.979856684754754,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 616293232,
+      "step": 3588
+    },
+    {
+      "epoch": 0.9437758926810021,
+      "loss": 0.14250054955482483,
+      "loss_ce": 0.004179632291197777,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 616293232,
+      "step": 3588
+    },
+    {
+      "epoch": 0.9440389294403893,
+      "grad_norm": 4.477659380807632,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 616465544,
+      "step": 3589
+    },
+    {
+      "epoch": 0.9440389294403893,
+      "loss": 0.13571983575820923,
+      "loss_ce": 0.004356917925179005,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 616465544,
+      "step": 3589
+    },
+    {
+      "epoch": 0.9443019661997765,
+      "grad_norm": 5.354193803812341,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 616638112,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9443019661997765,
+      "loss": 0.08732321113348007,
+      "loss_ce": 0.0015383013524115086,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 616638112,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9445650029591636,
+      "grad_norm": 3.712673541690186,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 616808952,
+      "step": 3591
+    },
+    {
+      "epoch": 0.9445650029591636,
+      "loss": 0.0873071700334549,
+      "loss_ce": 0.0063440315425395966,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 616808952,
+      "step": 3591
+    },
+    {
+      "epoch": 0.9448280397185507,
+      "grad_norm": 5.018692274423124,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 616981056,
+      "step": 3592
+    },
+    {
+      "epoch": 0.9448280397185507,
+      "loss": 0.16758012771606445,
+      "loss_ce": 0.0011983029544353485,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 616981056,
+      "step": 3592
+    },
+    {
+      "epoch": 0.9450910764779378,
+      "grad_norm": 8.157994634855726,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 617152956,
+      "step": 3593
+    },
+    {
+      "epoch": 0.9450910764779378,
+      "loss": 0.09211128950119019,
+      "loss_ce": 0.0016877016751095653,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 617152956,
+      "step": 3593
+    },
+    {
+      "epoch": 0.9453541132373249,
+      "grad_norm": 6.815211067626931,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 617323192,
+      "step": 3594
+    },
+    {
+      "epoch": 0.9453541132373249,
+      "loss": 0.15205200016498566,
+      "loss_ce": 0.0006542917108163238,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 617323192,
+      "step": 3594
+    },
+    {
+      "epoch": 0.9456171499967121,
+      "grad_norm": 5.698573656915325,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 617495528,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9456171499967121,
+      "loss": 0.15452706813812256,
+      "loss_ce": 0.004380582831799984,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 617495528,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9458801867560992,
+      "grad_norm": 5.492995557036467,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 617667772,
+      "step": 3596
+    },
+    {
+      "epoch": 0.9458801867560992,
+      "loss": 0.11533143371343613,
+      "loss_ce": 0.0028283819556236267,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 617667772,
+      "step": 3596
+    },
+    {
+      "epoch": 0.9461432235154863,
+      "grad_norm": 3.750755987920789,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 617836324,
+      "step": 3597
+    },
+    {
+      "epoch": 0.9461432235154863,
+      "loss": 0.07357309758663177,
+      "loss_ce": 0.0006360823172144592,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 617836324,
+      "step": 3597
+    },
+    {
+      "epoch": 0.9464062602748734,
+      "grad_norm": 3.634136051792054,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 618006748,
+      "step": 3598
+    },
+    {
+      "epoch": 0.9464062602748734,
+      "loss": 0.05711160972714424,
+      "loss_ce": 0.0004099512880202383,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 618006748,
+      "step": 3598
+    },
+    {
+      "epoch": 0.9466692970342605,
+      "grad_norm": 17.367283166631303,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 618179204,
+      "step": 3599
+    },
+    {
+      "epoch": 0.9466692970342605,
+      "loss": 0.11188434064388275,
+      "loss_ce": 0.0004646638117264956,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 618179204,
+      "step": 3599
+    },
+    {
+      "epoch": 0.9469323337936476,
+      "grad_norm": 4.326428384534629,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 618351176,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9469323337936476,
+      "loss": 0.1435290426015854,
+      "loss_ce": 0.002629393944516778,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 618351176,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9471953705530348,
+      "grad_norm": 19.80317191157718,
+      "learning_rate": 5e-06,
+      "loss": 0.0993,
+      "num_input_tokens_seen": 618523308,
+      "step": 3601
+    },
+    {
+      "epoch": 0.9471953705530348,
+      "loss": 0.055401187390089035,
+      "loss_ce": 0.0005763589288108051,
+      "loss_iou": 0.5,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 618523308,
+      "step": 3601
+    },
+    {
+      "epoch": 0.9474584073124219,
+      "grad_norm": 7.650182889486073,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 618695336,
+      "step": 3602
+    },
+    {
+      "epoch": 0.9474584073124219,
+      "loss": 0.15820011496543884,
+      "loss_ce": 0.0013244987931102514,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 618695336,
+      "step": 3602
+    },
+    {
+      "epoch": 0.947721444071809,
+      "grad_norm": 15.018190748040105,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 618868120,
+      "step": 3603
+    },
+    {
+      "epoch": 0.947721444071809,
+      "loss": 0.08152879774570465,
+      "loss_ce": 0.0010081640211865306,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 618868120,
+      "step": 3603
+    },
+    {
+      "epoch": 0.9479844808311961,
+      "grad_norm": 3.724726783342823,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 619040216,
+      "step": 3604
+    },
+    {
+      "epoch": 0.9479844808311961,
+      "loss": 0.07213738560676575,
+      "loss_ce": 0.0031829241197556257,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 619040216,
+      "step": 3604
+    },
+    {
+      "epoch": 0.9482475175905832,
+      "grad_norm": 28.840812127557175,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 619212192,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9482475175905832,
+      "loss": 0.1626545786857605,
+      "loss_ce": 0.0024830668698996305,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 619212192,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9485105543499704,
+      "grad_norm": 5.069934241766405,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 619382656,
+      "step": 3606
+    },
+    {
+      "epoch": 0.9485105543499704,
+      "loss": 0.1947801113128662,
+      "loss_ce": 0.00343489833176136,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 619382656,
+      "step": 3606
+    },
+    {
+      "epoch": 0.9487735911093576,
+      "grad_norm": 13.189826643263768,
+      "learning_rate": 5e-06,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 619554772,
+      "step": 3607
+    },
+    {
+      "epoch": 0.9487735911093576,
+      "loss": 0.10128442198038101,
+      "loss_ce": 0.001995484111830592,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 619554772,
+      "step": 3607
+    },
+    {
+      "epoch": 0.9490366278687447,
+      "grad_norm": 3.7735949518372953,
+      "learning_rate": 5e-06,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 619727116,
+      "step": 3608
+    },
+    {
+      "epoch": 0.9490366278687447,
+      "loss": 0.11414018273353577,
+      "loss_ce": 0.005039841867983341,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 619727116,
+      "step": 3608
+    },
+    {
+      "epoch": 0.9492996646281318,
+      "grad_norm": 4.250123008721577,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 619899428,
+      "step": 3609
+    },
+    {
+      "epoch": 0.9492996646281318,
+      "loss": 0.1226678267121315,
+      "loss_ce": 0.007280868943780661,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 619899428,
+      "step": 3609
+    },
+    {
+      "epoch": 0.9495627013875189,
+      "grad_norm": 6.863676153540119,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 620071540,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9495627013875189,
+      "loss": 0.21938937902450562,
+      "loss_ce": 0.0038132029585540295,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 620071540,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9498257381469061,
+      "grad_norm": 3.813325984971825,
+      "learning_rate": 5e-06,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 620243836,
+      "step": 3611
+    },
+    {
+      "epoch": 0.9498257381469061,
+      "loss": 0.07250767946243286,
+      "loss_ce": 0.0018289745785295963,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 620243836,
+      "step": 3611
+    },
+    {
+      "epoch": 0.9500887749062932,
+      "grad_norm": 8.694293187792477,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 620416164,
+      "step": 3612
+    },
+    {
+      "epoch": 0.9500887749062932,
+      "loss": 0.08137984573841095,
+      "loss_ce": 0.0004167144070379436,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 620416164,
+      "step": 3612
+    },
+    {
+      "epoch": 0.9503518116656803,
+      "grad_norm": 4.011759905693181,
+      "learning_rate": 5e-06,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 620588248,
+      "step": 3613
+    },
+    {
+      "epoch": 0.9503518116656803,
+      "loss": 0.10577777028083801,
+      "loss_ce": 0.003009829204529524,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 620588248,
+      "step": 3613
+    },
+    {
+      "epoch": 0.9506148484250674,
+      "grad_norm": 4.8583016740360545,
+      "learning_rate": 5e-06,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 620760212,
+      "step": 3614
+    },
+    {
+      "epoch": 0.9506148484250674,
+      "loss": 0.07320687174797058,
+      "loss_ce": 0.00593086751177907,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 620760212,
+      "step": 3614
+    },
+    {
+      "epoch": 0.9508778851844545,
+      "grad_norm": 7.183426742063696,
+      "learning_rate": 5e-06,
+      "loss": 0.092,
+      "num_input_tokens_seen": 620932584,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9508778851844545,
+      "loss": 0.12085875123739243,
+      "loss_ce": 0.004022202454507351,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 620932584,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9511409219438417,
+      "grad_norm": 5.9763835890459855,
+      "learning_rate": 5e-06,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 621104624,
+      "step": 3616
+    },
+    {
+      "epoch": 0.9511409219438417,
+      "loss": 0.057411566376686096,
+      "loss_ce": 0.00405157683417201,
+      "loss_iou": 0.40625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 621104624,
+      "step": 3616
+    },
+    {
+      "epoch": 0.9514039587032288,
+      "grad_norm": 4.148575034087351,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 621276932,
+      "step": 3617
+    },
+    {
+      "epoch": 0.9514039587032288,
+      "loss": 0.17621394991874695,
+      "loss_ce": 0.0008599417633377016,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 621276932,
+      "step": 3617
+    },
+    {
+      "epoch": 0.9516669954626159,
+      "grad_norm": 9.831716470581155,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 621449128,
+      "step": 3618
+    },
+    {
+      "epoch": 0.9516669954626159,
+      "loss": 0.17716863751411438,
+      "loss_ce": 0.0019672252237796783,
+      "loss_iou": 0.5,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 621449128,
+      "step": 3618
+    },
+    {
+      "epoch": 0.951930032222003,
+      "grad_norm": 6.726193865130754,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 621621260,
+      "step": 3619
+    },
+    {
+      "epoch": 0.951930032222003,
+      "loss": 0.08456599712371826,
+      "loss_ce": 0.0014971550554037094,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 621621260,
+      "step": 3619
+    },
+    {
+      "epoch": 0.9521930689813901,
+      "grad_norm": 5.019143488565095,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 621792008,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9521930689813901,
+      "loss": 0.05885142832994461,
+      "loss_ce": 0.0001813878770917654,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 621792008,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9524561057407773,
+      "grad_norm": 55.831161754212744,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 621964056,
+      "step": 3621
+    },
+    {
+      "epoch": 0.9524561057407773,
+      "loss": 0.08644313365221024,
+      "loss_ce": 0.0015737485373392701,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 621964056,
+      "step": 3621
+    },
+    {
+      "epoch": 0.9527191425001644,
+      "grad_norm": 6.339787905898623,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 622136360,
+      "step": 3622
+    },
+    {
+      "epoch": 0.9527191425001644,
+      "loss": 0.07180093228816986,
+      "loss_ce": 0.0007102307863533497,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 622136360,
+      "step": 3622
+    },
+    {
+      "epoch": 0.9529821792595515,
+      "grad_norm": 5.256007258631962,
+      "learning_rate": 5e-06,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 622308160,
+      "step": 3623
+    },
+    {
+      "epoch": 0.9529821792595515,
+      "loss": 0.16570287942886353,
+      "loss_ce": 0.0014267577789723873,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 622308160,
+      "step": 3623
+    },
+    {
+      "epoch": 0.9532452160189386,
+      "grad_norm": 16.742735373121388,
+      "learning_rate": 5e-06,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 622480548,
+      "step": 3624
+    },
+    {
+      "epoch": 0.9532452160189386,
+      "loss": 0.052828967571258545,
+      "loss_ce": 0.0007049451814964414,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 622480548,
+      "step": 3624
+    },
+    {
+      "epoch": 0.9535082527783257,
+      "grad_norm": 17.572925497634134,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 622652520,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9535082527783257,
+      "loss": 0.0615709125995636,
+      "loss_ce": 0.00045946481986902654,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 622652520,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9537712895377128,
+      "grad_norm": 6.0047852962098975,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 622823148,
+      "step": 3626
+    },
+    {
+      "epoch": 0.9537712895377128,
+      "loss": 0.13855966925621033,
+      "loss_ce": 0.0012610815465450287,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 622823148,
+      "step": 3626
+    },
+    {
+      "epoch": 0.9540343262971,
+      "grad_norm": 7.792667970397866,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 622995452,
+      "step": 3627
+    },
+    {
+      "epoch": 0.9540343262971,
+      "loss": 0.13608847558498383,
+      "loss_ce": 0.001505955122411251,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 622995452,
+      "step": 3627
+    },
+    {
+      "epoch": 0.9542973630564872,
+      "grad_norm": 4.565665131118125,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 623167560,
+      "step": 3628
+    },
+    {
+      "epoch": 0.9542973630564872,
+      "loss": 0.11297139525413513,
+      "loss_ce": 0.0007735221879556775,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 623167560,
+      "step": 3628
+    },
+    {
+      "epoch": 0.9545603998158743,
+      "grad_norm": 5.57299165717365,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 623339872,
+      "step": 3629
+    },
+    {
+      "epoch": 0.9545603998158743,
+      "loss": 0.09495042264461517,
+      "loss_ce": 0.002497426699846983,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 623339872,
+      "step": 3629
+    },
+    {
+      "epoch": 0.9548234365752614,
+      "grad_norm": 21.302937351842843,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 623512180,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9548234365752614,
+      "loss": 0.11213900148868561,
+      "loss_ce": 0.0011923413258045912,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 623512180,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9550864733346485,
+      "grad_norm": 12.098614631817124,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 623680972,
+      "step": 3631
+    },
+    {
+      "epoch": 0.9550864733346485,
+      "loss": 0.21909213066101074,
+      "loss_ce": 0.0015017889672890306,
+      "loss_iou": 0.453125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 623680972,
+      "step": 3631
+    },
+    {
+      "epoch": 0.9553495100940357,
+      "grad_norm": 9.034625023288243,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 623851276,
+      "step": 3632
+    },
+    {
+      "epoch": 0.9553495100940357,
+      "loss": 0.0687929093837738,
+      "loss_ce": 0.0007081945077516139,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 623851276,
+      "step": 3632
+    },
+    {
+      "epoch": 0.9556125468534228,
+      "grad_norm": 4.18724157975137,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 624023500,
+      "step": 3633
+    },
+    {
+      "epoch": 0.9556125468534228,
+      "loss": 0.12112436443567276,
+      "loss_ce": 0.0027008973993360996,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 624023500,
+      "step": 3633
+    },
+    {
+      "epoch": 0.9558755836128099,
+      "grad_norm": 10.76826846477555,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 624195576,
+      "step": 3634
+    },
+    {
+      "epoch": 0.9558755836128099,
+      "loss": 0.11673011630773544,
+      "loss_ce": 0.004501717630773783,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 624195576,
+      "step": 3634
+    },
+    {
+      "epoch": 0.956138620372197,
+      "grad_norm": 3.420122712086701,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 624366376,
+      "step": 3635
+    },
+    {
+      "epoch": 0.956138620372197,
+      "loss": 0.13994066417217255,
+      "loss_ce": 0.0026115677319467068,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 624366376,
+      "step": 3635
+    },
+    {
+      "epoch": 0.9564016571315841,
+      "grad_norm": 3.188466600397663,
+      "learning_rate": 5e-06,
+      "loss": 0.1096,
+      "num_input_tokens_seen": 624538928,
+      "step": 3636
+    },
+    {
+      "epoch": 0.9564016571315841,
+      "loss": 0.19849437475204468,
+      "loss_ce": 0.0006184080266393721,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 624538928,
+      "step": 3636
+    },
+    {
+      "epoch": 0.9566646938909713,
+      "grad_norm": 4.145251385205127,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 624711036,
+      "step": 3637
+    },
+    {
+      "epoch": 0.9566646938909713,
+      "loss": 0.07516495883464813,
+      "loss_ce": 0.0005189694347791374,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 624711036,
+      "step": 3637
+    },
+    {
+      "epoch": 0.9569277306503584,
+      "grad_norm": 7.0181560986956395,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 624883520,
+      "step": 3638
+    },
+    {
+      "epoch": 0.9569277306503584,
+      "loss": 0.06267523765563965,
+      "loss_ce": 0.0029981140978634357,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 624883520,
+      "step": 3638
+    },
+    {
+      "epoch": 0.9571907674097455,
+      "grad_norm": 3.9660695052723813,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 625055640,
+      "step": 3639
+    },
+    {
+      "epoch": 0.9571907674097455,
+      "loss": 0.1359872817993164,
+      "loss_ce": 0.00042819694499485195,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 625055640,
+      "step": 3639
+    },
+    {
+      "epoch": 0.9574538041691326,
+      "grad_norm": 17.732753202882193,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 625227732,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9574538041691326,
+      "loss": 0.09507328271865845,
+      "loss_ce": 0.0008502600830979645,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 625227732,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9577168409285197,
+      "grad_norm": 3.4840402032914257,
+      "learning_rate": 5e-06,
+      "loss": 0.0918,
+      "num_input_tokens_seen": 625398028,
+      "step": 3641
+    },
+    {
+      "epoch": 0.9577168409285197,
+      "loss": 0.13700971007347107,
+      "loss_ce": 0.0006876978441141546,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 625398028,
+      "step": 3641
+    },
+    {
+      "epoch": 0.9579798776879069,
+      "grad_norm": 7.413398524469702,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 625570244,
+      "step": 3642
+    },
+    {
+      "epoch": 0.9579798776879069,
+      "loss": 0.12876945734024048,
+      "loss_ce": 0.00047354548587463796,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 625570244,
+      "step": 3642
+    },
+    {
+      "epoch": 0.958242914447294,
+      "grad_norm": 5.336702710138045,
+      "learning_rate": 5e-06,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 625742464,
+      "step": 3643
+    },
+    {
+      "epoch": 0.958242914447294,
+      "loss": 0.05379210785031319,
+      "loss_ce": 0.00020323891658335924,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 625742464,
+      "step": 3643
+    },
+    {
+      "epoch": 0.9585059512066811,
+      "grad_norm": 23.972493383254932,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 625914372,
+      "step": 3644
+    },
+    {
+      "epoch": 0.9585059512066811,
+      "loss": 0.08553168922662735,
+      "loss_ce": 0.0017609409987926483,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 625914372,
+      "step": 3644
+    },
+    {
+      "epoch": 0.9587689879660682,
+      "grad_norm": 10.340048159606546,
+      "learning_rate": 5e-06,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 626086592,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9587689879660682,
+      "loss": 0.234289288520813,
+      "loss_ce": 0.00026523511041887105,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 626086592,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9590320247254553,
+      "grad_norm": 9.300486446962857,
+      "learning_rate": 5e-06,
+      "loss": 0.094,
+      "num_input_tokens_seen": 626257416,
+      "step": 3646
+    },
+    {
+      "epoch": 0.9590320247254553,
+      "loss": 0.045688219368457794,
+      "loss_ce": 0.0002475440560374409,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 626257416,
+      "step": 3646
+    },
+    {
+      "epoch": 0.9592950614848426,
+      "grad_norm": 11.589258821015147,
+      "learning_rate": 5e-06,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 626427708,
+      "step": 3647
+    },
+    {
+      "epoch": 0.9592950614848426,
+      "loss": 0.10979291796684265,
+      "loss_ce": 0.0016691365744918585,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 626427708,
+      "step": 3647
+    },
+    {
+      "epoch": 0.9595580982442297,
+      "grad_norm": 3.2081047701574943,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 626599988,
+      "step": 3648
+    },
+    {
+      "epoch": 0.9595580982442297,
+      "loss": 0.09939119219779968,
+      "loss_ce": 0.0016281325370073318,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 626599988,
+      "step": 3648
+    },
+    {
+      "epoch": 0.9598211350036168,
+      "grad_norm": 4.783540700409292,
+      "learning_rate": 5e-06,
+      "loss": 0.1455,
+      "num_input_tokens_seen": 626772316,
+      "step": 3649
+    },
+    {
+      "epoch": 0.9598211350036168,
+      "loss": 0.09440785646438599,
+      "loss_ce": 0.0029161556158214808,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 626772316,
+      "step": 3649
+    },
+    {
+      "epoch": 0.9600841717630039,
+      "grad_norm": 5.125730804994831,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 626942620,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9600841717630039,
+      "loss": 0.06443943828344345,
+      "loss_ce": 0.0007797717116773129,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 626942620,
+      "step": 3650
+    },
+    {
+      "epoch": 0.960347208522391,
+      "grad_norm": 4.565795888601258,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 627114816,
+      "step": 3651
+    },
+    {
+      "epoch": 0.960347208522391,
+      "loss": 0.12303532660007477,
+      "loss_ce": 0.0003851845976896584,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 627114816,
+      "step": 3651
+    },
+    {
+      "epoch": 0.9606102452817781,
+      "grad_norm": 8.198006738555572,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 627286832,
+      "step": 3652
+    },
+    {
+      "epoch": 0.9606102452817781,
+      "loss": 0.12663108110427856,
+      "loss_ce": 0.0022109271958470345,
+      "loss_iou": 0.6875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 627286832,
+      "step": 3652
+    },
+    {
+      "epoch": 0.9608732820411653,
+      "grad_norm": 6.977304479694191,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 627458840,
+      "step": 3653
+    },
+    {
+      "epoch": 0.9608732820411653,
+      "loss": 0.08323599398136139,
+      "loss_ce": 0.003646154422312975,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 627458840,
+      "step": 3653
+    },
+    {
+      "epoch": 0.9611363188005524,
+      "grad_norm": 6.143584577276683,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 627631048,
+      "step": 3654
+    },
+    {
+      "epoch": 0.9611363188005524,
+      "loss": 0.08655130863189697,
+      "loss_ce": 0.0005985412281006575,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 627631048,
+      "step": 3654
+    },
+    {
+      "epoch": 0.9613993555599395,
+      "grad_norm": 4.962748401603593,
+      "learning_rate": 5e-06,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 627801572,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9613993555599395,
+      "loss": 0.06723140180110931,
+      "loss_ce": 0.0012218892807140946,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 627801572,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9616623923193266,
+      "grad_norm": 3.888233294845763,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 627973444,
+      "step": 3656
+    },
+    {
+      "epoch": 0.9616623923193266,
+      "loss": 0.19316411018371582,
+      "loss_ce": 0.0060761114582419395,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 627973444,
+      "step": 3656
+    },
+    {
+      "epoch": 0.9619254290787137,
+      "grad_norm": 12.231564786844833,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 628145692,
+      "step": 3657
+    },
+    {
+      "epoch": 0.9619254290787137,
+      "loss": 0.08886748552322388,
+      "loss_ce": 0.004608447663486004,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 628145692,
+      "step": 3657
+    },
+    {
+      "epoch": 0.9621884658381009,
+      "grad_norm": 6.681516105502576,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 628317968,
+      "step": 3658
+    },
+    {
+      "epoch": 0.9621884658381009,
+      "loss": 0.11374935507774353,
+      "loss_ce": 0.00020870784646831453,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 628317968,
+      "step": 3658
+    },
+    {
+      "epoch": 0.962451502597488,
+      "grad_norm": 9.221687537639268,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 628489248,
+      "step": 3659
+    },
+    {
+      "epoch": 0.962451502597488,
+      "loss": 0.07078136503696442,
+      "loss_ce": 0.000972404726780951,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 628489248,
+      "step": 3659
+    },
+    {
+      "epoch": 0.9627145393568751,
+      "grad_norm": 3.888238330257217,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 628661488,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9627145393568751,
+      "loss": 0.148333340883255,
+      "loss_ce": 0.001513272407464683,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 628661488,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9629775761162622,
+      "grad_norm": 3.548818147424885,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 628833384,
+      "step": 3661
+    },
+    {
+      "epoch": 0.9629775761162622,
+      "loss": 0.07117056846618652,
+      "loss_ce": 0.0004918586346320808,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 628833384,
+      "step": 3661
+    },
+    {
+      "epoch": 0.9632406128756493,
+      "grad_norm": 3.779136602687443,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 629005652,
+      "step": 3662
+    },
+    {
+      "epoch": 0.9632406128756493,
+      "loss": 0.06165676563978195,
+      "loss_ce": 0.0007131616584956646,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 629005652,
+      "step": 3662
+    },
+    {
+      "epoch": 0.9635036496350365,
+      "grad_norm": 5.7914523260663495,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 629177884,
+      "step": 3663
+    },
+    {
+      "epoch": 0.9635036496350365,
+      "loss": 0.09137916564941406,
+      "loss_ce": 0.00181007559876889,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 629177884,
+      "step": 3663
+    },
+    {
+      "epoch": 0.9637666863944236,
+      "grad_norm": 3.071560976269854,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 629350348,
+      "step": 3664
+    },
+    {
+      "epoch": 0.9637666863944236,
+      "loss": 0.08777488768100739,
+      "loss_ce": 0.0004946062108501792,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 629350348,
+      "step": 3664
+    },
+    {
+      "epoch": 0.9640297231538107,
+      "grad_norm": 4.779549366101704,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 629522792,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9640297231538107,
+      "loss": 0.1326950043439865,
+      "loss_ce": 0.0025680402759462595,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 629522792,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9642927599131978,
+      "grad_norm": 4.346102092926262,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 629694856,
+      "step": 3666
+    },
+    {
+      "epoch": 0.9642927599131978,
+      "loss": 0.11643362045288086,
+      "loss_ce": 0.0008330341661348939,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 629694856,
+      "step": 3666
+    },
+    {
+      "epoch": 0.964555796672585,
+      "grad_norm": 5.7406338456333845,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 629866908,
+      "step": 3667
+    },
+    {
+      "epoch": 0.964555796672585,
+      "loss": 0.13996919989585876,
+      "loss_ce": 0.0034030412789434195,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 629866908,
+      "step": 3667
+    },
+    {
+      "epoch": 0.9648188334319722,
+      "grad_norm": 7.17686270310257,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 630039516,
+      "step": 3668
+    },
+    {
+      "epoch": 0.9648188334319722,
+      "loss": 0.12104253470897675,
+      "loss_ce": 0.00020818831399083138,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 630039516,
+      "step": 3668
+    },
+    {
+      "epoch": 0.9650818701913593,
+      "grad_norm": 4.2613778122636035,
+      "learning_rate": 5e-06,
+      "loss": 0.0821,
+      "num_input_tokens_seen": 630211624,
+      "step": 3669
+    },
+    {
+      "epoch": 0.9650818701913593,
+      "loss": 0.062059760093688965,
+      "loss_ce": 0.0022910854313522577,
+      "loss_iou": 0.34375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 630211624,
+      "step": 3669
+    },
+    {
+      "epoch": 0.9653449069507464,
+      "grad_norm": 3.5238324652450235,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 630382332,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9653449069507464,
+      "loss": 0.15831097960472107,
+      "loss_ce": 0.00016888529353309423,
+      "loss_iou": 0.453125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 630382332,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9656079437101335,
+      "grad_norm": 3.564039297129955,
+      "learning_rate": 5e-06,
+      "loss": 0.0998,
+      "num_input_tokens_seen": 630552592,
+      "step": 3671
+    },
+    {
+      "epoch": 0.9656079437101335,
+      "loss": 0.08011811971664429,
+      "loss_ce": 0.0016269085463136435,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 630552592,
+      "step": 3671
+    },
+    {
+      "epoch": 0.9658709804695206,
+      "grad_norm": 9.76288401818362,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 630722244,
+      "step": 3672
+    },
+    {
+      "epoch": 0.9658709804695206,
+      "loss": 0.08490733802318573,
+      "loss_ce": 0.0020673726685345173,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 630722244,
+      "step": 3672
+    },
+    {
+      "epoch": 0.9661340172289078,
+      "grad_norm": 7.354300413013905,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 630894380,
+      "step": 3673
+    },
+    {
+      "epoch": 0.9661340172289078,
+      "loss": 0.13285204768180847,
+      "loss_ce": 0.0009245476103387773,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 630894380,
+      "step": 3673
+    },
+    {
+      "epoch": 0.9663970539882949,
+      "grad_norm": 5.188368697186112,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 631066652,
+      "step": 3674
+    },
+    {
+      "epoch": 0.9663970539882949,
+      "loss": 0.09987783432006836,
+      "loss_ce": 0.003976346459239721,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 631066652,
+      "step": 3674
+    },
+    {
+      "epoch": 0.966660090747682,
+      "grad_norm": 12.325083640976278,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 631237000,
+      "step": 3675
+    },
+    {
+      "epoch": 0.966660090747682,
+      "loss": 0.17316505312919617,
+      "loss_ce": 0.002648080699145794,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 631237000,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9669231275070691,
+      "grad_norm": 6.105729054233263,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 631409164,
+      "step": 3676
+    },
+    {
+      "epoch": 0.9669231275070691,
+      "loss": 0.09672191739082336,
+      "loss_ce": 0.0017207016935572028,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 631409164,
+      "step": 3676
+    },
+    {
+      "epoch": 0.9671861642664562,
+      "grad_norm": 4.682056592281961,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 631581116,
+      "step": 3677
+    },
+    {
+      "epoch": 0.9671861642664562,
+      "loss": 0.12396599352359772,
+      "loss_ce": 0.0040776850655674934,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 631581116,
+      "step": 3677
+    },
+    {
+      "epoch": 0.9674492010258433,
+      "grad_norm": 8.33340254914201,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 631750180,
+      "step": 3678
+    },
+    {
+      "epoch": 0.9674492010258433,
+      "loss": 0.09776239842176437,
+      "loss_ce": 0.0013878863537684083,
+      "loss_iou": 0.59375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 631750180,
+      "step": 3678
+    },
+    {
+      "epoch": 0.9677122377852305,
+      "grad_norm": 3.892518620403468,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 631922528,
+      "step": 3679
+    },
+    {
+      "epoch": 0.9677122377852305,
+      "loss": 0.10747776180505753,
+      "loss_ce": 0.0042520565912127495,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 631922528,
+      "step": 3679
+    },
+    {
+      "epoch": 0.9679752745446176,
+      "grad_norm": 7.586861538526353,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 632094768,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9679752745446176,
+      "loss": 0.08680423349142075,
+      "loss_ce": 0.001721223583444953,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 632094768,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9682383113040047,
+      "grad_norm": 4.2210475168735355,
+      "learning_rate": 5e-06,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 632266740,
+      "step": 3681
+    },
+    {
+      "epoch": 0.9682383113040047,
+      "loss": 0.05067894607782364,
+      "loss_ce": 0.0006911527598276734,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 632266740,
+      "step": 3681
+    },
+    {
+      "epoch": 0.9685013480633918,
+      "grad_norm": 3.8121173575745755,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 632439188,
+      "step": 3682
+    },
+    {
+      "epoch": 0.9685013480633918,
+      "loss": 0.11126217991113663,
+      "loss_ce": 0.00033078622072935104,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 632439188,
+      "step": 3682
+    },
+    {
+      "epoch": 0.9687643848227789,
+      "grad_norm": 5.838321531180536,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 632611332,
+      "step": 3683
+    },
+    {
+      "epoch": 0.9687643848227789,
+      "loss": 0.08182443678379059,
+      "loss_ce": 0.0006171565037220716,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 632611332,
+      "step": 3683
+    },
+    {
+      "epoch": 0.9690274215821661,
+      "grad_norm": 4.164644957819613,
+      "learning_rate": 5e-06,
+      "loss": 0.086,
+      "num_input_tokens_seen": 632783660,
+      "step": 3684
+    },
+    {
+      "epoch": 0.9690274215821661,
+      "loss": 0.12429676204919815,
+      "loss_ce": 0.0015245481627061963,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 632783660,
+      "step": 3684
+    },
+    {
+      "epoch": 0.9692904583415533,
+      "grad_norm": 26.321407570319618,
+      "learning_rate": 5e-06,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 632955748,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9692904583415533,
+      "loss": 0.05816134810447693,
+      "loss_ce": 0.0002847612486220896,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 632955748,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9695534951009404,
+      "grad_norm": 3.994252764794672,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 633126124,
+      "step": 3686
+    },
+    {
+      "epoch": 0.9695534951009404,
+      "loss": 0.10153771191835403,
+      "loss_ce": 0.00023461380624212325,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 633126124,
+      "step": 3686
+    },
+    {
+      "epoch": 0.9698165318603275,
+      "grad_norm": 7.846355332992627,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 633298344,
+      "step": 3687
+    },
+    {
+      "epoch": 0.9698165318603275,
+      "loss": 0.15658894181251526,
+      "loss_ce": 0.001040847273543477,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 633298344,
+      "step": 3687
+    },
+    {
+      "epoch": 0.9700795686197146,
+      "grad_norm": 6.085055991191323,
+      "learning_rate": 5e-06,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 633470408,
+      "step": 3688
+    },
+    {
+      "epoch": 0.9700795686197146,
+      "loss": 0.14665310084819794,
+      "loss_ce": 0.0056618861854076385,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 633470408,
+      "step": 3688
+    },
+    {
+      "epoch": 0.9703426053791018,
+      "grad_norm": 4.551457844454551,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 633642716,
+      "step": 3689
+    },
+    {
+      "epoch": 0.9703426053791018,
+      "loss": 0.0793912261724472,
+      "loss_ce": 0.0006558679160661995,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 633642716,
+      "step": 3689
+    },
+    {
+      "epoch": 0.9706056421384889,
+      "grad_norm": 5.945079605333328,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 633812440,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9706056421384889,
+      "loss": 0.08010639995336533,
+      "loss_ce": 0.0009438038687221706,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 633812440,
+      "step": 3690
+    },
+    {
+      "epoch": 0.970868678897876,
+      "grad_norm": 4.065153864479657,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 633985000,
+      "step": 3691
+    },
+    {
+      "epoch": 0.970868678897876,
+      "loss": 0.0722232460975647,
+      "loss_ce": 0.0003238367207814008,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 633985000,
+      "step": 3691
+    },
+    {
+      "epoch": 0.9711317156572631,
+      "grad_norm": 4.197621752026495,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 634157200,
+      "step": 3692
+    },
+    {
+      "epoch": 0.9711317156572631,
+      "loss": 0.09805089235305786,
+      "loss_ce": 0.003385363146662712,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 634157200,
+      "step": 3692
+    },
+    {
+      "epoch": 0.9713947524166502,
+      "grad_norm": 4.118683496322007,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 634329136,
+      "step": 3693
+    },
+    {
+      "epoch": 0.9713947524166502,
+      "loss": 0.06334918737411499,
+      "loss_ce": 0.0014290215913206339,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 634329136,
+      "step": 3693
+    },
+    {
+      "epoch": 0.9716577891760374,
+      "grad_norm": 5.569436401286201,
+      "learning_rate": 5e-06,
+      "loss": 0.0858,
+      "num_input_tokens_seen": 634501660,
+      "step": 3694
+    },
+    {
+      "epoch": 0.9716577891760374,
+      "loss": 0.056709855794906616,
+      "loss_ce": 0.0002981142024509609,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 634501660,
+      "step": 3694
+    },
+    {
+      "epoch": 0.9719208259354245,
+      "grad_norm": 15.633714461557659,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 634673676,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9719208259354245,
+      "loss": 0.16554518043994904,
+      "loss_ce": 0.0010554337641224265,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 634673676,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9721838626948116,
+      "grad_norm": 26.512182592600844,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 634845964,
+      "step": 3696
+    },
+    {
+      "epoch": 0.9721838626948116,
+      "loss": 0.08471856266260147,
+      "loss_ce": 0.0012224669335409999,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 634845964,
+      "step": 3696
+    },
+    {
+      "epoch": 0.9724468994541987,
+      "grad_norm": 6.8173523395664715,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 635018328,
+      "step": 3697
+    },
+    {
+      "epoch": 0.9724468994541987,
+      "loss": 0.08134950697422028,
+      "loss_ce": 0.0013324212050065398,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 635018328,
+      "step": 3697
+    },
+    {
+      "epoch": 0.9727099362135858,
+      "grad_norm": 4.383461318684907,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 635190608,
+      "step": 3698
+    },
+    {
+      "epoch": 0.9727099362135858,
+      "loss": 0.08322876691818237,
+      "loss_ce": 0.0011364765232428908,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 635190608,
+      "step": 3698
+    },
+    {
+      "epoch": 0.972972972972973,
+      "grad_norm": 8.267610666760818,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 635362588,
+      "step": 3699
+    },
+    {
+      "epoch": 0.972972972972973,
+      "loss": 0.11783481389284134,
+      "loss_ce": 0.0006015403778292239,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 635362588,
+      "step": 3699
+    },
+    {
+      "epoch": 0.9732360097323601,
+      "grad_norm": 4.332996360892621,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 635534740,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9732360097323601,
+      "loss": 0.1074044331908226,
+      "loss_ce": 0.00025721488054841757,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 635534740,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9734990464917472,
+      "grad_norm": 5.513199468514905,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 635707288,
+      "step": 3701
+    },
+    {
+      "epoch": 0.9734990464917472,
+      "loss": 0.09230555593967438,
+      "loss_ce": 0.0021413678769022226,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 635707288,
+      "step": 3701
+    },
+    {
+      "epoch": 0.9737620832511343,
+      "grad_norm": 3.768479254290149,
+      "learning_rate": 5e-06,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 635879648,
+      "step": 3702
+    },
+    {
+      "epoch": 0.9737620832511343,
+      "loss": 0.03509838879108429,
+      "loss_ce": 0.000766111770644784,
+      "loss_iou": 0.484375,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 635879648,
+      "step": 3702
+    },
+    {
+      "epoch": 0.9740251200105214,
+      "grad_norm": 3.566831404475063,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 636051432,
+      "step": 3703
+    },
+    {
+      "epoch": 0.9740251200105214,
+      "loss": 0.09813763201236725,
+      "loss_ce": 0.001030704821459949,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 636051432,
+      "step": 3703
+    },
+    {
+      "epoch": 0.9742881567699085,
+      "grad_norm": 3.4711728772501447,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 636223588,
+      "step": 3704
+    },
+    {
+      "epoch": 0.9742881567699085,
+      "loss": 0.09130216389894485,
+      "loss_ce": 0.0020840244833379984,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 636223588,
+      "step": 3704
+    },
+    {
+      "epoch": 0.9745511935292958,
+      "grad_norm": 10.080851555182159,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 636395540,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9745511935292958,
+      "loss": 0.07607042789459229,
+      "loss_ce": 0.0012871015351265669,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 636395540,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9748142302886829,
+      "grad_norm": 3.5401703294732814,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 636567872,
+      "step": 3706
+    },
+    {
+      "epoch": 0.9748142302886829,
+      "loss": 0.1415339708328247,
+      "loss_ce": 0.0009700124501250684,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 636567872,
+      "step": 3706
+    },
+    {
+      "epoch": 0.97507726704807,
+      "grad_norm": 4.65430606928134,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 636739796,
+      "step": 3707
+    },
+    {
+      "epoch": 0.97507726704807,
+      "loss": 0.06319437175989151,
+      "loss_ce": 0.0008469584863632917,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 636739796,
+      "step": 3707
+    },
+    {
+      "epoch": 0.9753403038074571,
+      "grad_norm": 12.08107208422745,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 636911900,
+      "step": 3708
+    },
+    {
+      "epoch": 0.9753403038074571,
+      "loss": 0.09893044084310532,
+      "loss_ce": 0.004631124436855316,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 636911900,
+      "step": 3708
+    },
+    {
+      "epoch": 0.9756033405668442,
+      "grad_norm": 4.1664851424909894,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 637083804,
+      "step": 3709
+    },
+    {
+      "epoch": 0.9756033405668442,
+      "loss": 0.14125725626945496,
+      "loss_ce": 0.0035619523841887712,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 637083804,
+      "step": 3709
+    },
+    {
+      "epoch": 0.9758663773262314,
+      "grad_norm": 4.483302931910663,
+      "learning_rate": 5e-06,
+      "loss": 0.0909,
+      "num_input_tokens_seen": 637253480,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9758663773262314,
+      "loss": 0.09771590679883957,
+      "loss_ce": 0.0010667359456419945,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 637253480,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9761294140856185,
+      "grad_norm": 4.570273140954518,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 637425832,
+      "step": 3711
+    },
+    {
+      "epoch": 0.9761294140856185,
+      "loss": 0.15927918255329132,
+      "loss_ce": 0.0016558904899284244,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 637425832,
+      "step": 3711
+    },
+    {
+      "epoch": 0.9763924508450056,
+      "grad_norm": 4.261644785229519,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 637597956,
+      "step": 3712
+    },
+    {
+      "epoch": 0.9763924508450056,
+      "loss": 0.09720858931541443,
+      "loss_ce": 0.005060770083218813,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 637597956,
+      "step": 3712
+    },
+    {
+      "epoch": 0.9766554876043927,
+      "grad_norm": 4.316121186379811,
+      "learning_rate": 5e-06,
+      "loss": 0.1502,
+      "num_input_tokens_seen": 637770516,
+      "step": 3713
+    },
+    {
+      "epoch": 0.9766554876043927,
+      "loss": 0.23752275109291077,
+      "loss_ce": 0.003071451559662819,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 637770516,
+      "step": 3713
+    },
+    {
+      "epoch": 0.9769185243637798,
+      "grad_norm": 3.9858618973194195,
+      "learning_rate": 5e-06,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 637940932,
+      "step": 3714
+    },
+    {
+      "epoch": 0.9769185243637798,
+      "loss": 0.09894842654466629,
+      "loss_ce": 0.0026502024848014116,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 637940932,
+      "step": 3714
+    },
+    {
+      "epoch": 0.977181561123167,
+      "grad_norm": 6.459702059191179,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 638112972,
+      "step": 3715
+    },
+    {
+      "epoch": 0.977181561123167,
+      "loss": 0.05996452271938324,
+      "loss_ce": 0.0013707715552300215,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 638112972,
+      "step": 3715
+    },
+    {
+      "epoch": 0.9774445978825541,
+      "grad_norm": 4.702404331881163,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 638283296,
+      "step": 3716
+    },
+    {
+      "epoch": 0.9774445978825541,
+      "loss": 0.09870034456253052,
+      "loss_ce": 0.0003421921283006668,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 638283296,
+      "step": 3716
+    },
+    {
+      "epoch": 0.9777076346419412,
+      "grad_norm": 6.311133724359332,
+      "learning_rate": 5e-06,
+      "loss": 0.1405,
+      "num_input_tokens_seen": 638455388,
+      "step": 3717
+    },
+    {
+      "epoch": 0.9777076346419412,
+      "loss": 0.10849727690219879,
+      "loss_ce": 0.001868859282694757,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 638455388,
+      "step": 3717
+    },
+    {
+      "epoch": 0.9779706714013283,
+      "grad_norm": 4.841088605766017,
+      "learning_rate": 5e-06,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 638627620,
+      "step": 3718
+    },
+    {
+      "epoch": 0.9779706714013283,
+      "loss": 0.0839821919798851,
+      "loss_ce": 0.0003640282666310668,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 638627620,
+      "step": 3718
+    },
+    {
+      "epoch": 0.9782337081607154,
+      "grad_norm": 4.2847622894404696,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 638799776,
+      "step": 3719
+    },
+    {
+      "epoch": 0.9782337081607154,
+      "loss": 0.18468543887138367,
+      "loss_ce": 0.0003745291323866695,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 638799776,
+      "step": 3719
+    },
+    {
+      "epoch": 0.9784967449201026,
+      "grad_norm": 4.11039448761818,
+      "learning_rate": 5e-06,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 638971952,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9784967449201026,
+      "loss": 0.1171189695596695,
+      "loss_ce": 0.002739093266427517,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 638971952,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9787597816794897,
+      "grad_norm": 6.78823750558351,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 639144380,
+      "step": 3721
+    },
+    {
+      "epoch": 0.9787597816794897,
+      "loss": 0.10253561288118362,
+      "loss_ce": 0.0012172528076916933,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 639144380,
+      "step": 3721
+    },
+    {
+      "epoch": 0.9790228184388768,
+      "grad_norm": 3.963795445916244,
+      "learning_rate": 5e-06,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 639316848,
+      "step": 3722
+    },
+    {
+      "epoch": 0.9790228184388768,
+      "loss": 0.13296961784362793,
+      "loss_ce": 0.0031783583108335733,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 639316848,
+      "step": 3722
+    },
+    {
+      "epoch": 0.9792858551982639,
+      "grad_norm": 11.532632559627872,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 639489024,
+      "step": 3723
+    },
+    {
+      "epoch": 0.9792858551982639,
+      "loss": 0.12053656578063965,
+      "loss_ce": 0.005637889727950096,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 639489024,
+      "step": 3723
+    },
+    {
+      "epoch": 0.979548891957651,
+      "grad_norm": 9.989053455854867,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 639661440,
+      "step": 3724
+    },
+    {
+      "epoch": 0.979548891957651,
+      "loss": 0.13625101745128632,
+      "loss_ce": 0.00029520769021473825,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 639661440,
+      "step": 3724
+    },
+    {
+      "epoch": 0.9798119287170383,
+      "grad_norm": 4.9459595738196285,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 639833676,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9798119287170383,
+      "loss": 0.1375160813331604,
+      "loss_ce": 0.0008125934982672334,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 639833676,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9800749654764254,
+      "grad_norm": 4.566071751318605,
+      "learning_rate": 5e-06,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 640006092,
+      "step": 3726
+    },
+    {
+      "epoch": 0.9800749654764254,
+      "loss": 0.15862302482128143,
+      "loss_ce": 0.006462385877966881,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 640006092,
+      "step": 3726
+    },
+    {
+      "epoch": 0.9803380022358125,
+      "grad_norm": 5.733708403269408,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 640178452,
+      "step": 3727
+    },
+    {
+      "epoch": 0.9803380022358125,
+      "loss": 0.045382432639598846,
+      "loss_ce": 0.00027745150146074593,
+      "loss_iou": 0.515625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 640178452,
+      "step": 3727
+    },
+    {
+      "epoch": 0.9806010389951996,
+      "grad_norm": 4.6254627269264095,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 640350624,
+      "step": 3728
+    },
+    {
+      "epoch": 0.9806010389951996,
+      "loss": 0.054897043853998184,
+      "loss_ce": 0.0018880083225667477,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 640350624,
+      "step": 3728
+    },
+    {
+      "epoch": 0.9808640757545867,
+      "grad_norm": 5.915011241669676,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 640522796,
+      "step": 3729
+    },
+    {
+      "epoch": 0.9808640757545867,
+      "loss": 0.12351959943771362,
+      "loss_ce": 0.002242741174995899,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 640522796,
+      "step": 3729
+    },
+    {
+      "epoch": 0.9811271125139738,
+      "grad_norm": 13.40274422198988,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 640694968,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9811271125139738,
+      "loss": 0.08854502439498901,
+      "loss_ce": 0.00027292766026221216,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 640694968,
+      "step": 3730
+    },
+    {
+      "epoch": 0.981390149273361,
+      "grad_norm": 6.656264975637515,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 640867112,
+      "step": 3731
+    },
+    {
+      "epoch": 0.981390149273361,
+      "loss": 0.12663085758686066,
+      "loss_ce": 0.0017834422178566456,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 640867112,
+      "step": 3731
+    },
+    {
+      "epoch": 0.9816531860327481,
+      "grad_norm": 11.247340076351778,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 641039384,
+      "step": 3732
+    },
+    {
+      "epoch": 0.9816531860327481,
+      "loss": 0.1410384476184845,
+      "loss_ce": 0.0051131523214280605,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 641039384,
+      "step": 3732
+    },
+    {
+      "epoch": 0.9819162227921352,
+      "grad_norm": 5.304544125644644,
+      "learning_rate": 5e-06,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 641211452,
+      "step": 3733
+    },
+    {
+      "epoch": 0.9819162227921352,
+      "loss": 0.12673211097717285,
+      "loss_ce": 0.004204033873975277,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 641211452,
+      "step": 3733
+    },
+    {
+      "epoch": 0.9821792595515223,
+      "grad_norm": 4.504449466935876,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 641383184,
+      "step": 3734
+    },
+    {
+      "epoch": 0.9821792595515223,
+      "loss": 0.09244327247142792,
+      "loss_ce": 0.0025995145551860332,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 641383184,
+      "step": 3734
+    },
+    {
+      "epoch": 0.9824422963109094,
+      "grad_norm": 4.706100049802098,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 641555380,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9824422963109094,
+      "loss": 0.08688107877969742,
+      "loss_ce": 0.0007604720303788781,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 641555380,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9827053330702966,
+      "grad_norm": 8.136156546275306,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 641727456,
+      "step": 3736
+    },
+    {
+      "epoch": 0.9827053330702966,
+      "loss": 0.11328569054603577,
+      "loss_ce": 0.0008589247590862215,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 641727456,
+      "step": 3736
+    },
+    {
+      "epoch": 0.9829683698296837,
+      "grad_norm": 4.005733802836262,
+      "learning_rate": 5e-06,
+      "loss": 0.074,
+      "num_input_tokens_seen": 641899684,
+      "step": 3737
+    },
+    {
+      "epoch": 0.9829683698296837,
+      "loss": 0.09893114119768143,
+      "loss_ce": 0.00038988247979432344,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 641899684,
+      "step": 3737
+    },
+    {
+      "epoch": 0.9832314065890708,
+      "grad_norm": 3.6593398060214795,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 642072000,
+      "step": 3738
+    },
+    {
+      "epoch": 0.9832314065890708,
+      "loss": 0.0708111971616745,
+      "loss_ce": 0.003092692233622074,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 642072000,
+      "step": 3738
+    },
+    {
+      "epoch": 0.9834944433484579,
+      "grad_norm": 112.80387932882391,
+      "learning_rate": 5e-06,
+      "loss": 0.075,
+      "num_input_tokens_seen": 642243916,
+      "step": 3739
+    },
+    {
+      "epoch": 0.9834944433484579,
+      "loss": 0.07938267290592194,
+      "loss_ce": 0.001837508985772729,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 642243916,
+      "step": 3739
+    },
+    {
+      "epoch": 0.983757480107845,
+      "grad_norm": 46.399369914383044,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 642415992,
+      "step": 3740
+    },
+    {
+      "epoch": 0.983757480107845,
+      "loss": 0.12807899713516235,
+      "loss_ce": 0.0024686530232429504,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 642415992,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9840205168672322,
+      "grad_norm": 4.972320382385805,
+      "learning_rate": 5e-06,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 642586472,
+      "step": 3741
+    },
+    {
+      "epoch": 0.9840205168672322,
+      "loss": 0.10946183651685715,
+      "loss_ce": 0.0011854701442644,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 642586472,
+      "step": 3741
+    },
+    {
+      "epoch": 0.9842835536266193,
+      "grad_norm": 3.8095943920850632,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 642758552,
+      "step": 3742
+    },
+    {
+      "epoch": 0.9842835536266193,
+      "loss": 0.1769047975540161,
+      "loss_ce": 0.002832533325999975,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 642758552,
+      "step": 3742
+    },
+    {
+      "epoch": 0.9845465903860064,
+      "grad_norm": 5.317599890556273,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 642930684,
+      "step": 3743
+    },
+    {
+      "epoch": 0.9845465903860064,
+      "loss": 0.05296643078327179,
+      "loss_ce": 0.0012238813797011971,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 642930684,
+      "step": 3743
+    },
+    {
+      "epoch": 0.9848096271453936,
+      "grad_norm": 4.846803333953094,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 643102812,
+      "step": 3744
+    },
+    {
+      "epoch": 0.9848096271453936,
+      "loss": 0.10566210746765137,
+      "loss_ce": 0.0007274242816492915,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 643102812,
+      "step": 3744
+    },
+    {
+      "epoch": 0.9850726639047807,
+      "grad_norm": 3.406688370945616,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 643275096,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9850726639047807,
+      "loss": 0.08728724718093872,
+      "loss_ce": 0.0003884438192471862,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 643275096,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9853357006641679,
+      "grad_norm": 5.344442258228798,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 643444760,
+      "step": 3746
+    },
+    {
+      "epoch": 0.9853357006641679,
+      "loss": 0.07199759781360626,
+      "loss_ce": 0.002844766713678837,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 643444760,
+      "step": 3746
+    },
+    {
+      "epoch": 0.985598737423555,
+      "grad_norm": 9.063261890916973,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 643617080,
+      "step": 3747
+    },
+    {
+      "epoch": 0.985598737423555,
+      "loss": 0.06594318896532059,
+      "loss_ce": 0.0005440223030745983,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 643617080,
+      "step": 3747
+    },
+    {
+      "epoch": 0.9858617741829421,
+      "grad_norm": 3.8879508865341124,
+      "learning_rate": 5e-06,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 643789448,
+      "step": 3748
+    },
+    {
+      "epoch": 0.9858617741829421,
+      "loss": 0.056218698620796204,
+      "loss_ce": 0.001195505610667169,
+      "loss_iou": 0.65625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 643789448,
+      "step": 3748
+    },
+    {
+      "epoch": 0.9861248109423292,
+      "grad_norm": 193.16068563077133,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 643958208,
+      "step": 3749
+    },
+    {
+      "epoch": 0.9861248109423292,
+      "loss": 0.057384688407182693,
+      "loss_ce": 0.003338058013468981,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 643958208,
+      "step": 3749
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "grad_norm": 11.212759382877062,
+      "learning_rate": 5e-06,
+      "loss": 0.0866,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_websight_new_CIoU": 0.8978322446346283,
+      "eval_websight_new_GIoU": 0.8991535305976868,
+      "eval_websight_new_IoU": 0.9012524485588074,
+      "eval_websight_new_MAE_all": 0.013891254551708698,
+      "eval_websight_new_MAE_h": 0.008930663112550974,
+      "eval_websight_new_MAE_w": 0.020785433240234852,
+      "eval_websight_new_MAE_x": 0.0205678790807724,
+      "eval_websight_new_MAE_y": 0.005281045567244291,
+      "eval_websight_new_NUM_probability": 0.9999656677246094,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.06887268275022507,
+      "eval_websight_new_loss_ce": 5.8660152717493474e-06,
+      "eval_websight_new_loss_iou": 0.30145263671875,
+      "eval_websight_new_loss_num": 0.012205123901367188,
+      "eval_websight_new_loss_xval": 0.06097412109375,
+      "eval_websight_new_runtime": 65.4719,
+      "eval_websight_new_samples_per_second": 0.764,
+      "eval_websight_new_steps_per_second": 0.031,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_seeclick_CIoU": 0.657441109418869,
+      "eval_seeclick_GIoU": 0.6616063714027405,
+      "eval_seeclick_IoU": 0.6793454885482788,
+      "eval_seeclick_MAE_all": 0.04169847071170807,
+      "eval_seeclick_MAE_h": 0.024878486059606075,
+      "eval_seeclick_MAE_w": 0.05510186776518822,
+      "eval_seeclick_MAE_x": 0.06434983387589455,
+      "eval_seeclick_MAE_y": 0.02246370818465948,
+      "eval_seeclick_NUM_probability": 0.9999178946018219,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.169124573469162,
+      "eval_seeclick_loss_ce": 0.009124522097408772,
+      "eval_seeclick_loss_iou": 0.455078125,
+      "eval_seeclick_loss_num": 0.030914306640625,
+      "eval_seeclick_loss_xval": 0.15460205078125,
+      "eval_seeclick_runtime": 75.5663,
+      "eval_seeclick_samples_per_second": 0.569,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_icons_CIoU": 0.8627608418464661,
+      "eval_icons_GIoU": 0.863149881362915,
+      "eval_icons_IoU": 0.8711875081062317,
+      "eval_icons_MAE_all": 0.022217202931642532,
+      "eval_icons_MAE_h": 0.01895691081881523,
+      "eval_icons_MAE_w": 0.0274124164134264,
+      "eval_icons_MAE_x": 0.0250897784717381,
+      "eval_icons_MAE_y": 0.017409704625606537,
+      "eval_icons_NUM_probability": 0.9999702572822571,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.07059822231531143,
+      "eval_icons_loss_ce": 1.0734858733485453e-05,
+      "eval_icons_loss_iou": 0.47100830078125,
+      "eval_icons_loss_num": 0.012834548950195312,
+      "eval_icons_loss_xval": 0.064178466796875,
+      "eval_icons_runtime": 81.7327,
+      "eval_icons_samples_per_second": 0.612,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_screenspot_CIoU": 0.6053812901178995,
+      "eval_screenspot_GIoU": 0.5979885856310526,
+      "eval_screenspot_IoU": 0.637059231599172,
+      "eval_screenspot_MAE_all": 0.07145863150556882,
+      "eval_screenspot_MAE_h": 0.04250209157665571,
+      "eval_screenspot_MAE_w": 0.12718970080216727,
+      "eval_screenspot_MAE_x": 0.07470711196462314,
+      "eval_screenspot_MAE_y": 0.04143562292059263,
+      "eval_screenspot_NUM_probability": 0.9999008377393087,
+      "eval_screenspot_inside_bbox": 0.8841666579246521,
+      "eval_screenspot_loss": 0.9072001576423645,
+      "eval_screenspot_loss_ce": 0.6087295611699423,
+      "eval_screenspot_loss_iou": 0.4524739583333333,
+      "eval_screenspot_loss_num": 0.058499654134114586,
+      "eval_screenspot_loss_xval": 0.29248046875,
+      "eval_screenspot_runtime": 146.6335,
+      "eval_screenspot_samples_per_second": 0.607,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 42.623229461756374,
+      "grad_norm": 24.37971363613738,
+      "learning_rate": 5e-06,
+      "loss": 1.249,
+      "num_input_tokens_seen": 644302160,
+      "step": 3751
+    },
+    {
+      "epoch": 42.623229461756374,
+      "loss": 1.3391211032867432,
+      "loss_ce": 0.9495947360992432,
+      "loss_iou": 0.609375,
+      "loss_num": 0.078125,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 644302160,
+      "step": 3751
+    },
+    {
+      "epoch": 42.63456090651558,
+      "grad_norm": 15.008470587136868,
+      "learning_rate": 5e-06,
+      "loss": 0.6148,
+      "num_input_tokens_seen": 644473668,
+      "step": 3752
+    },
+    {
+      "epoch": 42.63456090651558,
+      "loss": 0.5472804307937622,
+      "loss_ce": 0.2776271402835846,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 644473668,
+      "step": 3752
+    },
+    {
+      "epoch": 42.64589235127479,
+      "grad_norm": 10.153654090229104,
+      "learning_rate": 5e-06,
+      "loss": 0.5629,
+      "num_input_tokens_seen": 644644808,
+      "step": 3753
+    },
+    {
+      "epoch": 42.64589235127479,
+      "loss": 0.5728659629821777,
+      "loss_ce": 0.26402807235717773,
+      "loss_iou": 0.65625,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 644644808,
+      "step": 3753
+    },
+    {
+      "epoch": 42.657223796033996,
+      "grad_norm": 17.498808383644807,
+      "learning_rate": 5e-06,
+      "loss": 0.4414,
+      "num_input_tokens_seen": 644817044,
+      "step": 3754
+    },
+    {
+      "epoch": 42.657223796033996,
+      "loss": 0.4910026788711548,
+      "loss_ce": 0.17215505242347717,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 644817044,
+      "step": 3754
+    },
+    {
+      "epoch": 42.668555240793204,
+      "grad_norm": 23.60598186278669,
+      "learning_rate": 5e-06,
+      "loss": 0.5217,
+      "num_input_tokens_seen": 644988620,
+      "step": 3755
+    },
+    {
+      "epoch": 42.668555240793204,
+      "loss": 0.6383213996887207,
+      "loss_ce": 0.2723698318004608,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 644988620,
+      "step": 3755
+    },
+    {
+      "epoch": 42.67988668555241,
+      "grad_norm": 13.62518187033341,
+      "learning_rate": 5e-06,
+      "loss": 0.4168,
+      "num_input_tokens_seen": 645157052,
+      "step": 3756
+    },
+    {
+      "epoch": 42.67988668555241,
+      "loss": 0.4043606221675873,
+      "loss_ce": 0.03991969674825668,
+      "loss_iou": 0.0,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 645157052,
+      "step": 3756
+    },
+    {
+      "epoch": 42.69121813031161,
+      "grad_norm": 9.375746028718728,
+      "learning_rate": 5e-06,
+      "loss": 0.4246,
+      "num_input_tokens_seen": 645328860,
+      "step": 3757
+    },
+    {
+      "epoch": 42.69121813031161,
+      "loss": 0.450731098651886,
+      "loss_ce": 0.10713370144367218,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 645328860,
+      "step": 3757
+    },
+    {
+      "epoch": 42.70254957507082,
+      "grad_norm": 18.34093224440512,
+      "learning_rate": 5e-06,
+      "loss": 0.3826,
+      "num_input_tokens_seen": 645500924,
+      "step": 3758
+    },
+    {
+      "epoch": 42.70254957507082,
+      "loss": 0.5433092713356018,
+      "loss_ce": 0.164830282330513,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 645500924,
+      "step": 3758
+    },
+    {
+      "epoch": 42.713881019830026,
+      "grad_norm": 20.083649202913804,
+      "learning_rate": 5e-06,
+      "loss": 0.333,
+      "num_input_tokens_seen": 645672684,
+      "step": 3759
+    },
+    {
+      "epoch": 42.713881019830026,
+      "loss": 0.4810216724872589,
+      "loss_ce": 0.1741369217634201,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 645672684,
+      "step": 3759
+    },
+    {
+      "epoch": 42.725212464589234,
+      "grad_norm": 16.870768244182024,
+      "learning_rate": 5e-06,
+      "loss": 0.4587,
+      "num_input_tokens_seen": 645842788,
+      "step": 3760
+    },
+    {
+      "epoch": 42.725212464589234,
+      "loss": 0.3847481608390808,
+      "loss_ce": 0.2161385416984558,
+      "loss_iou": 0.0,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 645842788,
+      "step": 3760
+    },
+    {
+      "epoch": 42.73654390934844,
+      "grad_norm": 7.052248444847047,
+      "learning_rate": 5e-06,
+      "loss": 0.4233,
+      "num_input_tokens_seen": 646014564,
+      "step": 3761
+    },
+    {
+      "epoch": 42.73654390934844,
+      "loss": 0.37615275382995605,
+      "loss_ce": 0.07849954068660736,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 646014564,
+      "step": 3761
+    },
+    {
+      "epoch": 42.74787535410765,
+      "grad_norm": 59.35628679550276,
+      "learning_rate": 5e-06,
+      "loss": 0.3503,
+      "num_input_tokens_seen": 646185212,
+      "step": 3762
+    },
+    {
+      "epoch": 42.74787535410765,
+      "loss": 0.2602749466896057,
+      "loss_ce": 0.0961819440126419,
+      "loss_iou": 0.25,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 646185212,
+      "step": 3762
+    },
+    {
+      "epoch": 42.759206798866856,
+      "grad_norm": 11.196946849133175,
+      "learning_rate": 5e-06,
+      "loss": 0.3785,
+      "num_input_tokens_seen": 646356972,
+      "step": 3763
+    },
+    {
+      "epoch": 42.759206798866856,
+      "loss": 0.2895856499671936,
+      "loss_ce": 0.07242254912853241,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 646356972,
+      "step": 3763
+    },
+    {
+      "epoch": 42.77053824362606,
+      "grad_norm": 21.65002065464953,
+      "learning_rate": 5e-06,
+      "loss": 0.3237,
+      "num_input_tokens_seen": 646528444,
+      "step": 3764
+    },
+    {
+      "epoch": 42.77053824362606,
+      "loss": 0.271282434463501,
+      "loss_ce": 0.042797308415174484,
+      "loss_iou": 0.0,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 646528444,
+      "step": 3764
+    },
+    {
+      "epoch": 42.78186968838527,
+      "grad_norm": 8.94016361166771,
+      "learning_rate": 5e-06,
+      "loss": 0.2839,
+      "num_input_tokens_seen": 646699156,
+      "step": 3765
+    },
+    {
+      "epoch": 42.78186968838527,
+      "loss": 0.3400135338306427,
+      "loss_ce": 0.1177540272474289,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 646699156,
+      "step": 3765
+    },
+    {
+      "epoch": 42.79320113314448,
+      "grad_norm": 7.42291041299476,
+      "learning_rate": 5e-06,
+      "loss": 0.326,
+      "num_input_tokens_seen": 646869864,
+      "step": 3766
+    },
+    {
+      "epoch": 42.79320113314448,
+      "loss": 0.3942365050315857,
+      "loss_ce": 0.0757550597190857,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 646869864,
+      "step": 3766
+    },
+    {
+      "epoch": 42.804532577903686,
+      "grad_norm": 8.782066939888233,
+      "learning_rate": 5e-06,
+      "loss": 0.2947,
+      "num_input_tokens_seen": 647040996,
+      "step": 3767
+    },
+    {
+      "epoch": 42.804532577903686,
+      "loss": 0.35241279006004333,
+      "loss_ce": 0.08031807094812393,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 647040996,
+      "step": 3767
+    },
+    {
+      "epoch": 42.815864022662886,
+      "grad_norm": 16.45957031099631,
+      "learning_rate": 5e-06,
+      "loss": 0.3525,
+      "num_input_tokens_seen": 647212756,
+      "step": 3768
+    },
+    {
+      "epoch": 42.815864022662886,
+      "loss": 0.41619616746902466,
+      "loss_ce": 0.05816391855478287,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 647212756,
+      "step": 3768
+    },
+    {
+      "epoch": 42.827195467422094,
+      "grad_norm": 6.952175045712314,
+      "learning_rate": 5e-06,
+      "loss": 0.2681,
+      "num_input_tokens_seen": 647384520,
+      "step": 3769
+    },
+    {
+      "epoch": 42.827195467422094,
+      "loss": 0.2557736039161682,
+      "loss_ce": 0.06387907266616821,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 647384520,
+      "step": 3769
+    },
+    {
+      "epoch": 42.8385269121813,
+      "grad_norm": 17.73762005139639,
+      "learning_rate": 5e-06,
+      "loss": 0.347,
+      "num_input_tokens_seen": 647555936,
+      "step": 3770
+    },
+    {
+      "epoch": 42.8385269121813,
+      "loss": 0.4990995526313782,
+      "loss_ce": 0.04640180990099907,
+      "loss_iou": 0.0,
+      "loss_num": 0.09033203125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 647555936,
+      "step": 3770
+    },
+    {
+      "epoch": 42.84985835694051,
+      "grad_norm": 12.40804914890636,
+      "learning_rate": 5e-06,
+      "loss": 0.304,
+      "num_input_tokens_seen": 647727976,
+      "step": 3771
+    },
+    {
+      "epoch": 42.84985835694051,
+      "loss": 0.3800738453865051,
+      "loss_ce": 0.062141742557287216,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 647727976,
+      "step": 3771
+    },
+    {
+      "epoch": 42.861189801699716,
+      "grad_norm": 12.775334867173783,
+      "learning_rate": 5e-06,
+      "loss": 0.3116,
+      "num_input_tokens_seen": 647898980,
+      "step": 3772
+    },
+    {
+      "epoch": 42.861189801699716,
+      "loss": 0.43261992931365967,
+      "loss_ce": 0.05771148204803467,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 647898980,
+      "step": 3772
+    },
+    {
+      "epoch": 42.87252124645892,
+      "grad_norm": 58.24261672409072,
+      "learning_rate": 5e-06,
+      "loss": 0.3323,
+      "num_input_tokens_seen": 648070640,
+      "step": 3773
+    },
+    {
+      "epoch": 42.87252124645892,
+      "loss": 0.3606436252593994,
+      "loss_ce": 0.05900786817073822,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 648070640,
+      "step": 3773
+    },
+    {
+      "epoch": 42.88385269121813,
+      "grad_norm": 17.755111713808308,
+      "learning_rate": 5e-06,
+      "loss": 0.3556,
+      "num_input_tokens_seen": 648240800,
+      "step": 3774
+    },
+    {
+      "epoch": 42.88385269121813,
+      "loss": 0.3877432346343994,
+      "loss_ce": 0.0405142605304718,
+      "loss_iou": 0.076171875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 648240800,
+      "step": 3774
+    },
+    {
+      "epoch": 42.89518413597734,
+      "grad_norm": 21.60438828874626,
+      "learning_rate": 5e-06,
+      "loss": 0.2659,
+      "num_input_tokens_seen": 648410980,
+      "step": 3775
+    },
+    {
+      "epoch": 42.89518413597734,
+      "loss": 0.2268907129764557,
+      "loss_ce": 0.10134139657020569,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 648410980,
+      "step": 3775
+    },
+    {
+      "epoch": 42.906515580736546,
+      "grad_norm": 7.314877446779315,
+      "learning_rate": 5e-06,
+      "loss": 0.2199,
+      "num_input_tokens_seen": 648582436,
+      "step": 3776
+    },
+    {
+      "epoch": 42.906515580736546,
+      "loss": 0.19990015029907227,
+      "loss_ce": 0.06458520144224167,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 648582436,
+      "step": 3776
+    },
+    {
+      "epoch": 42.91784702549575,
+      "grad_norm": 89.40364631522068,
+      "learning_rate": 5e-06,
+      "loss": 0.2529,
+      "num_input_tokens_seen": 648753276,
+      "step": 3777
+    },
+    {
+      "epoch": 42.91784702549575,
+      "loss": 0.3249284625053406,
+      "loss_ce": 0.08383958786725998,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 648753276,
+      "step": 3777
+    },
+    {
+      "epoch": 42.92917847025496,
+      "grad_norm": 10.629610869016606,
+      "learning_rate": 5e-06,
+      "loss": 0.2321,
+      "num_input_tokens_seen": 648925076,
+      "step": 3778
+    },
+    {
+      "epoch": 42.92917847025496,
+      "loss": 0.24622215330600739,
+      "loss_ce": 0.03916039317846298,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 648925076,
+      "step": 3778
+    },
+    {
+      "epoch": 42.94050991501416,
+      "grad_norm": 8.143397457379272,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 649097140,
+      "step": 3779
+    },
+    {
+      "epoch": 42.94050991501416,
+      "loss": 0.22169527411460876,
+      "loss_ce": 0.04338107258081436,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 649097140,
+      "step": 3779
+    },
+    {
+      "epoch": 42.95184135977337,
+      "grad_norm": 12.910000035177736,
+      "learning_rate": 5e-06,
+      "loss": 0.2911,
+      "num_input_tokens_seen": 649267156,
+      "step": 3780
+    },
+    {
+      "epoch": 42.95184135977337,
+      "loss": 0.22911566495895386,
+      "loss_ce": 0.021168887615203857,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 649267156,
+      "step": 3780
+    },
+    {
+      "epoch": 42.963172804532576,
+      "grad_norm": 17.3966640002082,
+      "learning_rate": 5e-06,
+      "loss": 0.3775,
+      "num_input_tokens_seen": 649438392,
+      "step": 3781
+    },
+    {
+      "epoch": 42.963172804532576,
+      "loss": 0.47513455152511597,
+      "loss_ce": 0.01383085548877716,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.09228515625,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 649438392,
+      "step": 3781
+    },
+    {
+      "epoch": 42.97450424929178,
+      "grad_norm": 5.281441594031922,
+      "learning_rate": 5e-06,
+      "loss": 0.258,
+      "num_input_tokens_seen": 649608808,
+      "step": 3782
+    },
+    {
+      "epoch": 42.97450424929178,
+      "loss": 0.26870954036712646,
+      "loss_ce": 0.08236920088529587,
+      "loss_iou": 0.546875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 649608808,
+      "step": 3782
+    },
+    {
+      "epoch": 42.98583569405099,
+      "grad_norm": 9.712115026609329,
+      "learning_rate": 5e-06,
+      "loss": 0.3038,
+      "num_input_tokens_seen": 649779764,
+      "step": 3783
+    },
+    {
+      "epoch": 42.98583569405099,
+      "loss": 0.29391050338745117,
+      "loss_ce": 0.07393980026245117,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 649779764,
+      "step": 3783
+    },
+    {
+      "epoch": 42.9971671388102,
+      "grad_norm": 8.644881376613798,
+      "learning_rate": 5e-06,
+      "loss": 0.2818,
+      "num_input_tokens_seen": 649950780,
+      "step": 3784
+    },
+    {
+      "epoch": 42.9971671388102,
+      "loss": 0.33376365900039673,
+      "loss_ce": 0.06652123481035233,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 649950780,
+      "step": 3784
+    },
+    {
+      "epoch": 42.9971671388102,
+      "loss": 0.22381728887557983,
+      "loss_ce": 0.03976577892899513,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 649993692,
+      "step": 3784
+    },
+    {
+      "epoch": 43.008498583569406,
+      "grad_norm": 5.784889107414259,
+      "learning_rate": 5e-06,
+      "loss": 0.2592,
+      "num_input_tokens_seen": 650122464,
+      "step": 3785
+    },
+    {
+      "epoch": 43.008498583569406,
+      "loss": 0.2489549070596695,
+      "loss_ce": 0.0555039718747139,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 650122464,
+      "step": 3785
+    },
+    {
+      "epoch": 43.01983002832861,
+      "grad_norm": 14.651729063727553,
+      "learning_rate": 5e-06,
+      "loss": 0.2333,
+      "num_input_tokens_seen": 650293332,
+      "step": 3786
+    },
+    {
+      "epoch": 43.01983002832861,
+      "loss": 0.2854953408241272,
+      "loss_ce": 0.024783657863736153,
+      "loss_iou": 0.578125,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 650293332,
+      "step": 3786
+    },
+    {
+      "epoch": 43.03116147308782,
+      "grad_norm": 6.864405551935044,
+      "learning_rate": 5e-06,
+      "loss": 0.2131,
+      "num_input_tokens_seen": 650465228,
+      "step": 3787
+    },
+    {
+      "epoch": 43.03116147308782,
+      "loss": 0.21152222156524658,
+      "loss_ce": 0.05841553211212158,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 650465228,
+      "step": 3787
+    },
+    {
+      "epoch": 43.04249291784703,
+      "grad_norm": 30.52150954800119,
+      "learning_rate": 5e-06,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 650636904,
+      "step": 3788
+    },
+    {
+      "epoch": 43.04249291784703,
+      "loss": 0.20700110495090485,
+      "loss_ce": 0.03639259189367294,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 650636904,
+      "step": 3788
+    },
+    {
+      "epoch": 43.053824362606235,
+      "grad_norm": 6.722962112619755,
+      "learning_rate": 5e-06,
+      "loss": 0.232,
+      "num_input_tokens_seen": 650807920,
+      "step": 3789
+    },
+    {
+      "epoch": 43.053824362606235,
+      "loss": 0.15295374393463135,
+      "loss_ce": 0.05824244022369385,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 650807920,
+      "step": 3789
+    },
+    {
+      "epoch": 43.065155807365436,
+      "grad_norm": 13.160820060207227,
+      "learning_rate": 5e-06,
+      "loss": 0.2421,
+      "num_input_tokens_seen": 650980148,
+      "step": 3790
+    },
+    {
+      "epoch": 43.065155807365436,
+      "loss": 0.33120375871658325,
+      "loss_ce": 0.08941297233104706,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 650980148,
+      "step": 3790
+    },
+    {
+      "epoch": 43.07648725212464,
+      "grad_norm": 6.435419692026323,
+      "learning_rate": 5e-06,
+      "loss": 0.197,
+      "num_input_tokens_seen": 651151304,
+      "step": 3791
+    },
+    {
+      "epoch": 43.07648725212464,
+      "loss": 0.2795127034187317,
+      "loss_ce": 0.04800637066364288,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 651151304,
+      "step": 3791
+    },
+    {
+      "epoch": 43.08781869688385,
+      "grad_norm": 7.7632092093522544,
+      "learning_rate": 5e-06,
+      "loss": 0.2171,
+      "num_input_tokens_seen": 651323268,
+      "step": 3792
+    },
+    {
+      "epoch": 43.08781869688385,
+      "loss": 0.23102571070194244,
+      "loss_ce": 0.05686188489198685,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 651323268,
+      "step": 3792
+    },
+    {
+      "epoch": 43.09915014164306,
+      "grad_norm": 9.493350798082142,
+      "learning_rate": 5e-06,
+      "loss": 0.2472,
+      "num_input_tokens_seen": 651495456,
+      "step": 3793
+    },
+    {
+      "epoch": 43.09915014164306,
+      "loss": 0.22000005841255188,
+      "loss_ce": 0.034987252205610275,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 651495456,
+      "step": 3793
+    },
+    {
+      "epoch": 43.110481586402265,
+      "grad_norm": 31.530854864107187,
+      "learning_rate": 5e-06,
+      "loss": 0.2926,
+      "num_input_tokens_seen": 651665408,
+      "step": 3794
+    },
+    {
+      "epoch": 43.110481586402265,
+      "loss": 0.29464665055274963,
+      "loss_ce": 0.07449283450841904,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 651665408,
+      "step": 3794
+    },
+    {
+      "epoch": 43.12181303116147,
+      "grad_norm": 17.13192978487039,
+      "learning_rate": 5e-06,
+      "loss": 0.229,
+      "num_input_tokens_seen": 651836708,
+      "step": 3795
+    },
+    {
+      "epoch": 43.12181303116147,
+      "loss": 0.17648279666900635,
+      "loss_ce": 0.016387568786740303,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 651836708,
+      "step": 3795
+    },
+    {
+      "epoch": 43.13314447592068,
+      "grad_norm": 5.659905257802971,
+      "learning_rate": 5e-06,
+      "loss": 0.1888,
+      "num_input_tokens_seen": 652008812,
+      "step": 3796
+    },
+    {
+      "epoch": 43.13314447592068,
+      "loss": 0.24487656354904175,
+      "loss_ce": 0.04092760011553764,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 652008812,
+      "step": 3796
+    },
+    {
+      "epoch": 43.14447592067989,
+      "grad_norm": 11.602617114678928,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 652180336,
+      "step": 3797
+    },
+    {
+      "epoch": 43.14447592067989,
+      "loss": 0.25554385781288147,
+      "loss_ce": 0.02931704744696617,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 652180336,
+      "step": 3797
+    },
+    {
+      "epoch": 43.155807365439095,
+      "grad_norm": 29.420810232565373,
+      "learning_rate": 5e-06,
+      "loss": 0.239,
+      "num_input_tokens_seen": 652351044,
+      "step": 3798
+    },
+    {
+      "epoch": 43.155807365439095,
+      "loss": 0.3694992959499359,
+      "loss_ce": 0.031654439866542816,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 652351044,
+      "step": 3798
+    },
+    {
+      "epoch": 43.1671388101983,
+      "grad_norm": 7.506924438989472,
+      "learning_rate": 5e-06,
+      "loss": 0.2761,
+      "num_input_tokens_seen": 652522800,
+      "step": 3799
+    },
+    {
+      "epoch": 43.1671388101983,
+      "loss": 0.2833660840988159,
+      "loss_ce": 0.025339949876070023,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 652522800,
+      "step": 3799
+    },
+    {
+      "epoch": 43.17847025495751,
+      "grad_norm": 10.13119802584233,
+      "learning_rate": 5e-06,
+      "loss": 0.1714,
+      "num_input_tokens_seen": 652694216,
+      "step": 3800
+    },
+    {
+      "epoch": 43.17847025495751,
+      "loss": 0.22721588611602783,
+      "loss_ce": 0.017956862226128578,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 652694216,
+      "step": 3800
+    },
+    {
+      "epoch": 43.18980169971671,
+      "grad_norm": 6.188441992461635,
+      "learning_rate": 5e-06,
+      "loss": 0.3126,
+      "num_input_tokens_seen": 652866332,
+      "step": 3801
+    },
+    {
+      "epoch": 43.18980169971671,
+      "loss": 0.3227539658546448,
+      "loss_ce": 0.012817460112273693,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 652866332,
+      "step": 3801
+    },
+    {
+      "epoch": 43.20113314447592,
+      "grad_norm": 9.953767348454466,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 653037548,
+      "step": 3802
+    },
+    {
+      "epoch": 43.20113314447592,
+      "loss": 0.34050068259239197,
+      "loss_ce": 0.03166278079152107,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 653037548,
+      "step": 3802
+    },
+    {
+      "epoch": 43.212464589235125,
+      "grad_norm": 10.240115738305146,
+      "learning_rate": 5e-06,
+      "loss": 0.2311,
+      "num_input_tokens_seen": 653209240,
+      "step": 3803
+    },
+    {
+      "epoch": 43.212464589235125,
+      "loss": 0.2943469285964966,
+      "loss_ce": 0.02209961786866188,
+      "loss_iou": 0.59375,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 653209240,
+      "step": 3803
+    },
+    {
+      "epoch": 43.22379603399433,
+      "grad_norm": 9.75843735457791,
+      "learning_rate": 5e-06,
+      "loss": 0.2735,
+      "num_input_tokens_seen": 653379136,
+      "step": 3804
+    },
+    {
+      "epoch": 43.22379603399433,
+      "loss": 0.22456300258636475,
+      "loss_ce": 0.027694083750247955,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 653379136,
+      "step": 3804
+    },
+    {
+      "epoch": 43.23512747875354,
+      "grad_norm": 10.498601980364539,
+      "learning_rate": 5e-06,
+      "loss": 0.2699,
+      "num_input_tokens_seen": 653551180,
+      "step": 3805
+    },
+    {
+      "epoch": 43.23512747875354,
+      "loss": 0.2836833894252777,
+      "loss_ce": 0.02382621541619301,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 653551180,
+      "step": 3805
+    },
+    {
+      "epoch": 43.24645892351275,
+      "grad_norm": 18.66063402803507,
+      "learning_rate": 5e-06,
+      "loss": 0.2754,
+      "num_input_tokens_seen": 653722740,
+      "step": 3806
+    },
+    {
+      "epoch": 43.24645892351275,
+      "loss": 0.3721906542778015,
+      "loss_ce": 0.01601998507976532,
+      "loss_iou": 0.0,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 653722740,
+      "step": 3806
+    },
+    {
+      "epoch": 43.257790368271955,
+      "grad_norm": 11.386499638967939,
+      "learning_rate": 5e-06,
+      "loss": 0.206,
+      "num_input_tokens_seen": 653894636,
+      "step": 3807
+    },
+    {
+      "epoch": 43.257790368271955,
+      "loss": 0.24139291048049927,
+      "loss_ce": 0.08189278841018677,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 653894636,
+      "step": 3807
+    },
+    {
+      "epoch": 43.26912181303116,
+      "grad_norm": 13.489539747453755,
+      "learning_rate": 5e-06,
+      "loss": 0.1924,
+      "num_input_tokens_seen": 654065232,
+      "step": 3808
+    },
+    {
+      "epoch": 43.26912181303116,
+      "loss": 0.19118039309978485,
+      "loss_ce": 0.02858273684978485,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 654065232,
+      "step": 3808
+    },
+    {
+      "epoch": 43.28045325779037,
+      "grad_norm": 10.507455853530496,
+      "learning_rate": 5e-06,
+      "loss": 0.2809,
+      "num_input_tokens_seen": 654236760,
+      "step": 3809
+    },
+    {
+      "epoch": 43.28045325779037,
+      "loss": 0.27525657415390015,
+      "loss_ce": 0.09505028277635574,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 654236760,
+      "step": 3809
+    },
+    {
+      "epoch": 43.29178470254958,
+      "grad_norm": 8.614593715955264,
+      "learning_rate": 5e-06,
+      "loss": 0.2246,
+      "num_input_tokens_seen": 654408368,
+      "step": 3810
+    },
+    {
+      "epoch": 43.29178470254958,
+      "loss": 0.17899608612060547,
+      "loss_ce": 0.01819896697998047,
+      "loss_iou": 0.0,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 654408368,
+      "step": 3810
+    },
+    {
+      "epoch": 43.303116147308785,
+      "grad_norm": 4.433324089882691,
+      "learning_rate": 5e-06,
+      "loss": 0.1783,
+      "num_input_tokens_seen": 654580212,
+      "step": 3811
+    },
+    {
+      "epoch": 43.303116147308785,
+      "loss": 0.1583895981311798,
+      "loss_ce": 0.025638127699494362,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 654580212,
+      "step": 3811
+    },
+    {
+      "epoch": 43.314447592067985,
+      "grad_norm": 10.737819808808593,
+      "learning_rate": 5e-06,
+      "loss": 0.1974,
+      "num_input_tokens_seen": 654752004,
+      "step": 3812
+    },
+    {
+      "epoch": 43.314447592067985,
+      "loss": 0.25267672538757324,
+      "loss_ce": 0.03273654356598854,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 654752004,
+      "step": 3812
+    },
+    {
+      "epoch": 43.32577903682719,
+      "grad_norm": 9.169743103048214,
+      "learning_rate": 5e-06,
+      "loss": 0.2434,
+      "num_input_tokens_seen": 654921420,
+      "step": 3813
+    },
+    {
+      "epoch": 43.32577903682719,
+      "loss": 0.302659273147583,
+      "loss_ce": 0.07158016413450241,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 654921420,
+      "step": 3813
+    },
+    {
+      "epoch": 43.3371104815864,
+      "grad_norm": 6.7512630689625786,
+      "learning_rate": 5e-06,
+      "loss": 0.1827,
+      "num_input_tokens_seen": 655093140,
+      "step": 3814
+    },
+    {
+      "epoch": 43.3371104815864,
+      "loss": 0.17458195984363556,
+      "loss_ce": 0.02226872742176056,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 655093140,
+      "step": 3814
+    },
+    {
+      "epoch": 43.34844192634561,
+      "grad_norm": 14.476485058249372,
+      "learning_rate": 5e-06,
+      "loss": 0.2628,
+      "num_input_tokens_seen": 655265136,
+      "step": 3815
+    },
+    {
+      "epoch": 43.34844192634561,
+      "loss": 0.2875571846961975,
+      "loss_ce": 0.08492045104503632,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 655265136,
+      "step": 3815
+    },
+    {
+      "epoch": 43.359773371104815,
+      "grad_norm": 7.813083325716797,
+      "learning_rate": 5e-06,
+      "loss": 0.2828,
+      "num_input_tokens_seen": 655437064,
+      "step": 3816
+    },
+    {
+      "epoch": 43.359773371104815,
+      "loss": 0.34945645928382874,
+      "loss_ce": 0.04571498930454254,
+      "loss_iou": 0.390625,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 655437064,
+      "step": 3816
+    },
+    {
+      "epoch": 43.37110481586402,
+      "grad_norm": 7.336895629162358,
+      "learning_rate": 5e-06,
+      "loss": 0.222,
+      "num_input_tokens_seen": 655607676,
+      "step": 3817
+    },
+    {
+      "epoch": 43.37110481586402,
+      "loss": 0.16241437196731567,
+      "loss_ce": 0.04251081123948097,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 655607676,
+      "step": 3817
+    },
+    {
+      "epoch": 43.38243626062323,
+      "grad_norm": 5.472498623836622,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 655779600,
+      "step": 3818
+    },
+    {
+      "epoch": 43.38243626062323,
+      "loss": 0.13749703764915466,
+      "loss_ce": 0.05436716228723526,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 655779600,
+      "step": 3818
+    },
+    {
+      "epoch": 43.39376770538244,
+      "grad_norm": 11.062765407752119,
+      "learning_rate": 5e-06,
+      "loss": 0.2462,
+      "num_input_tokens_seen": 655950168,
+      "step": 3819
+    },
+    {
+      "epoch": 43.39376770538244,
+      "loss": 0.24120905995368958,
+      "loss_ce": 0.03503229841589928,
+      "loss_iou": 0.375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 655950168,
+      "step": 3819
+    },
+    {
+      "epoch": 43.405099150141645,
+      "grad_norm": 7.983856235610563,
+      "learning_rate": 5e-06,
+      "loss": 0.2226,
+      "num_input_tokens_seen": 656121288,
+      "step": 3820
+    },
+    {
+      "epoch": 43.405099150141645,
+      "loss": 0.2142389714717865,
+      "loss_ce": 0.053594425320625305,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 656121288,
+      "step": 3820
+    },
+    {
+      "epoch": 43.41643059490085,
+      "grad_norm": 8.372275065133358,
+      "learning_rate": 5e-06,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 656292508,
+      "step": 3821
+    },
+    {
+      "epoch": 43.41643059490085,
+      "loss": 0.27440422773361206,
+      "loss_ce": 0.02116938680410385,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 656292508,
+      "step": 3821
+    },
+    {
+      "epoch": 43.42776203966006,
+      "grad_norm": 15.602341705793288,
+      "learning_rate": 5e-06,
+      "loss": 0.2219,
+      "num_input_tokens_seen": 656461684,
+      "step": 3822
+    },
+    {
+      "epoch": 43.42776203966006,
+      "loss": 0.31255632638931274,
+      "loss_ce": 0.07775407284498215,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 656461684,
+      "step": 3822
+    },
+    {
+      "epoch": 43.43909348441926,
+      "grad_norm": 14.301675332298627,
+      "learning_rate": 5e-06,
+      "loss": 0.272,
+      "num_input_tokens_seen": 656633508,
+      "step": 3823
+    },
+    {
+      "epoch": 43.43909348441926,
+      "loss": 0.24964401125907898,
+      "loss_ce": 0.029185030609369278,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 656633508,
+      "step": 3823
+    },
+    {
+      "epoch": 43.45042492917847,
+      "grad_norm": 11.526420488961662,
+      "learning_rate": 5e-06,
+      "loss": 0.2866,
+      "num_input_tokens_seen": 656805284,
+      "step": 3824
+    },
+    {
+      "epoch": 43.45042492917847,
+      "loss": 0.3178904056549072,
+      "loss_ce": 0.019184362143278122,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 656805284,
+      "step": 3824
+    },
+    {
+      "epoch": 43.461756373937675,
+      "grad_norm": 12.819341592963644,
+      "learning_rate": 5e-06,
+      "loss": 0.1974,
+      "num_input_tokens_seen": 656977192,
+      "step": 3825
+    },
+    {
+      "epoch": 43.461756373937675,
+      "loss": 0.21698424220085144,
+      "loss_ce": 0.03781554847955704,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 656977192,
+      "step": 3825
+    },
+    {
+      "epoch": 43.47308781869688,
+      "grad_norm": 8.831140904488134,
+      "learning_rate": 5e-06,
+      "loss": 0.2008,
+      "num_input_tokens_seen": 657148044,
+      "step": 3826
+    },
+    {
+      "epoch": 43.47308781869688,
+      "loss": 0.2201002687215805,
+      "loss_ce": 0.011665214784443378,
+      "loss_iou": 0.59375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 657148044,
+      "step": 3826
+    },
+    {
+      "epoch": 43.48441926345609,
+      "grad_norm": 8.065696891811767,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 657320124,
+      "step": 3827
+    },
+    {
+      "epoch": 43.48441926345609,
+      "loss": 0.24065983295440674,
+      "loss_ce": 0.016554001718759537,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 657320124,
+      "step": 3827
+    },
+    {
+      "epoch": 43.4957507082153,
+      "grad_norm": 16.439324369048308,
+      "learning_rate": 5e-06,
+      "loss": 0.2246,
+      "num_input_tokens_seen": 657492284,
+      "step": 3828
+    },
+    {
+      "epoch": 43.4957507082153,
+      "loss": 0.21445706486701965,
+      "loss_ce": 0.04032377526164055,
+      "loss_iou": 0.328125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 657492284,
+      "step": 3828
+    },
+    {
+      "epoch": 43.507082152974505,
+      "grad_norm": 7.712628241498826,
+      "learning_rate": 5e-06,
+      "loss": 0.2453,
+      "num_input_tokens_seen": 657663228,
+      "step": 3829
+    },
+    {
+      "epoch": 43.507082152974505,
+      "loss": 0.20886492729187012,
+      "loss_ce": 0.03424334153532982,
+      "loss_iou": 0.46875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 657663228,
+      "step": 3829
+    },
+    {
+      "epoch": 43.51841359773371,
+      "grad_norm": 12.0529266773192,
+      "learning_rate": 5e-06,
+      "loss": 0.2182,
+      "num_input_tokens_seen": 657834276,
+      "step": 3830
+    },
+    {
+      "epoch": 43.51841359773371,
+      "loss": 0.2074132114648819,
+      "loss_ce": 0.037369273602962494,
+      "loss_iou": 0.46875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 657834276,
+      "step": 3830
+    },
+    {
+      "epoch": 43.52974504249292,
+      "grad_norm": 11.68380231415628,
+      "learning_rate": 5e-06,
+      "loss": 0.2204,
+      "num_input_tokens_seen": 658005484,
+      "step": 3831
+    },
+    {
+      "epoch": 43.52974504249292,
+      "loss": 0.2939346134662628,
+      "loss_ce": 0.04387357458472252,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 658005484,
+      "step": 3831
+    },
+    {
+      "epoch": 43.54107648725213,
+      "grad_norm": 21.100746918424953,
+      "learning_rate": 5e-06,
+      "loss": 0.2424,
+      "num_input_tokens_seen": 658177708,
+      "step": 3832
+    },
+    {
+      "epoch": 43.54107648725213,
+      "loss": 0.2990325391292572,
+      "loss_ce": 0.0315764844417572,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 658177708,
+      "step": 3832
+    },
+    {
+      "epoch": 43.552407932011334,
+      "grad_norm": 11.518224064790715,
+      "learning_rate": 5e-06,
+      "loss": 0.2318,
+      "num_input_tokens_seen": 658348500,
+      "step": 3833
+    },
+    {
+      "epoch": 43.552407932011334,
+      "loss": 0.20460152626037598,
+      "loss_ce": 0.06385447084903717,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 658348500,
+      "step": 3833
+    },
+    {
+      "epoch": 43.563739376770535,
+      "grad_norm": 6.272488097687055,
+      "learning_rate": 5e-06,
+      "loss": 0.2291,
+      "num_input_tokens_seen": 658519692,
+      "step": 3834
+    },
+    {
+      "epoch": 43.563739376770535,
+      "loss": 0.2233453094959259,
+      "loss_ce": 0.039995696395635605,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 658519692,
+      "step": 3834
+    },
+    {
+      "epoch": 43.57507082152974,
+      "grad_norm": 4.098578612905147,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 658690396,
+      "step": 3835
+    },
+    {
+      "epoch": 43.57507082152974,
+      "loss": 0.20045900344848633,
+      "loss_ce": 0.04628421366214752,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 658690396,
+      "step": 3835
+    },
+    {
+      "epoch": 43.58640226628895,
+      "grad_norm": 8.110726258019252,
+      "learning_rate": 5e-06,
+      "loss": 0.2675,
+      "num_input_tokens_seen": 658862380,
+      "step": 3836
+    },
+    {
+      "epoch": 43.58640226628895,
+      "loss": 0.36111244559288025,
+      "loss_ce": 0.04659830033779144,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 658862380,
+      "step": 3836
+    },
+    {
+      "epoch": 43.59773371104816,
+      "grad_norm": 7.458622707898409,
+      "learning_rate": 5e-06,
+      "loss": 0.2198,
+      "num_input_tokens_seen": 659034252,
+      "step": 3837
+    },
+    {
+      "epoch": 43.59773371104816,
+      "loss": 0.209273561835289,
+      "loss_ce": 0.0072471983730793,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 659034252,
+      "step": 3837
+    },
+    {
+      "epoch": 43.609065155807365,
+      "grad_norm": 8.432559943682879,
+      "learning_rate": 5e-06,
+      "loss": 0.2397,
+      "num_input_tokens_seen": 659204724,
+      "step": 3838
+    },
+    {
+      "epoch": 43.609065155807365,
+      "loss": 0.19880183041095734,
+      "loss_ce": 0.07355768978595734,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 659204724,
+      "step": 3838
+    },
+    {
+      "epoch": 43.62039660056657,
+      "grad_norm": 5.5823825751763,
+      "learning_rate": 5e-06,
+      "loss": 0.1696,
+      "num_input_tokens_seen": 659376692,
+      "step": 3839
+    },
+    {
+      "epoch": 43.62039660056657,
+      "loss": 0.09564369916915894,
+      "loss_ce": 0.013398820534348488,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 659376692,
+      "step": 3839
+    },
+    {
+      "epoch": 43.63172804532578,
+      "grad_norm": 10.559594019574956,
+      "learning_rate": 5e-06,
+      "loss": 0.2299,
+      "num_input_tokens_seen": 659547172,
+      "step": 3840
+    },
+    {
+      "epoch": 43.63172804532578,
+      "loss": 0.1778956651687622,
+      "loss_ce": 0.02066909149289131,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 659547172,
+      "step": 3840
+    },
+    {
+      "epoch": 43.64305949008499,
+      "grad_norm": 25.648096566696097,
+      "learning_rate": 5e-06,
+      "loss": 0.2321,
+      "num_input_tokens_seen": 659719140,
+      "step": 3841
+    },
+    {
+      "epoch": 43.64305949008499,
+      "loss": 0.26578018069267273,
+      "loss_ce": 0.052859045565128326,
+      "loss_iou": 0.40625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 659719140,
+      "step": 3841
+    },
+    {
+      "epoch": 43.654390934844194,
+      "grad_norm": 4.322128469344213,
+      "learning_rate": 5e-06,
+      "loss": 0.2071,
+      "num_input_tokens_seen": 659891276,
+      "step": 3842
+    },
+    {
+      "epoch": 43.654390934844194,
+      "loss": 0.1609874963760376,
+      "loss_ce": 0.058723084628582,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 659891276,
+      "step": 3842
+    },
+    {
+      "epoch": 43.6657223796034,
+      "grad_norm": 15.389118290922337,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 660063448,
+      "step": 3843
+    },
+    {
+      "epoch": 43.6657223796034,
+      "loss": 0.30868998169898987,
+      "loss_ce": 0.050419725477695465,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 660063448,
+      "step": 3843
+    },
+    {
+      "epoch": 43.67705382436261,
+      "grad_norm": 9.30800837634291,
+      "learning_rate": 5e-06,
+      "loss": 0.1926,
+      "num_input_tokens_seen": 660234844,
+      "step": 3844
+    },
+    {
+      "epoch": 43.67705382436261,
+      "loss": 0.24589240550994873,
+      "loss_ce": 0.052899233996868134,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 660234844,
+      "step": 3844
+    },
+    {
+      "epoch": 43.68838526912181,
+      "grad_norm": 7.544857056958143,
+      "learning_rate": 5e-06,
+      "loss": 0.2734,
+      "num_input_tokens_seen": 660407012,
+      "step": 3845
+    },
+    {
+      "epoch": 43.68838526912181,
+      "loss": 0.3219788372516632,
+      "loss_ce": 0.02855231799185276,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 660407012,
+      "step": 3845
+    },
+    {
+      "epoch": 43.69971671388102,
+      "grad_norm": 5.713854999747559,
+      "learning_rate": 5e-06,
+      "loss": 0.2058,
+      "num_input_tokens_seen": 660578960,
+      "step": 3846
+    },
+    {
+      "epoch": 43.69971671388102,
+      "loss": 0.2059517502784729,
+      "loss_ce": 0.0179939903318882,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 660578960,
+      "step": 3846
+    },
+    {
+      "epoch": 43.711048158640224,
+      "grad_norm": 7.675785504814342,
+      "learning_rate": 5e-06,
+      "loss": 0.2137,
+      "num_input_tokens_seen": 660750016,
+      "step": 3847
+    },
+    {
+      "epoch": 43.711048158640224,
+      "loss": 0.1930961310863495,
+      "loss_ce": 0.026409126818180084,
+      "loss_iou": 0.484375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 660750016,
+      "step": 3847
+    },
+    {
+      "epoch": 43.72237960339943,
+      "grad_norm": 8.297970210596503,
+      "learning_rate": 5e-06,
+      "loss": 0.2239,
+      "num_input_tokens_seen": 660919824,
+      "step": 3848
+    },
+    {
+      "epoch": 43.72237960339943,
+      "loss": 0.1422894448041916,
+      "loss_ce": 0.01771668717265129,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 660919824,
+      "step": 3848
+    },
+    {
+      "epoch": 43.73371104815864,
+      "grad_norm": 8.498874853495801,
+      "learning_rate": 5e-06,
+      "loss": 0.2264,
+      "num_input_tokens_seen": 661090136,
+      "step": 3849
+    },
+    {
+      "epoch": 43.73371104815864,
+      "loss": 0.2185337096452713,
+      "loss_ce": 0.038968276232481,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 661090136,
+      "step": 3849
+    },
+    {
+      "epoch": 43.74504249291785,
+      "grad_norm": 9.257081145105209,
+      "learning_rate": 5e-06,
+      "loss": 0.2249,
+      "num_input_tokens_seen": 661261156,
+      "step": 3850
+    },
+    {
+      "epoch": 43.74504249291785,
+      "loss": 0.20226441323757172,
+      "loss_ce": 0.028604142367839813,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 661261156,
+      "step": 3850
+    },
+    {
+      "epoch": 43.756373937677054,
+      "grad_norm": 5.628612071722162,
+      "learning_rate": 5e-06,
+      "loss": 0.2236,
+      "num_input_tokens_seen": 661432132,
+      "step": 3851
+    },
+    {
+      "epoch": 43.756373937677054,
+      "loss": 0.19719679653644562,
+      "loss_ce": 0.048637233674526215,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 661432132,
+      "step": 3851
+    },
+    {
+      "epoch": 43.76770538243626,
+      "grad_norm": 7.847435672271257,
+      "learning_rate": 5e-06,
+      "loss": 0.2018,
+      "num_input_tokens_seen": 661603952,
+      "step": 3852
+    },
+    {
+      "epoch": 43.76770538243626,
+      "loss": 0.20567631721496582,
+      "loss_ce": 0.023303259164094925,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 661603952,
+      "step": 3852
+    },
+    {
+      "epoch": 43.77903682719547,
+      "grad_norm": 5.306042708069077,
+      "learning_rate": 5e-06,
+      "loss": 0.2072,
+      "num_input_tokens_seen": 661775660,
+      "step": 3853
+    },
+    {
+      "epoch": 43.77903682719547,
+      "loss": 0.22403264045715332,
+      "loss_ce": 0.02073979191482067,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 661775660,
+      "step": 3853
+    },
+    {
+      "epoch": 43.79036827195468,
+      "grad_norm": 12.147321715502649,
+      "learning_rate": 5e-06,
+      "loss": 0.2544,
+      "num_input_tokens_seen": 661945204,
+      "step": 3854
+    },
+    {
+      "epoch": 43.79036827195468,
+      "loss": 0.1255451738834381,
+      "loss_ce": 0.03261914476752281,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 661945204,
+      "step": 3854
+    },
+    {
+      "epoch": 43.801699716713884,
+      "grad_norm": 16.31804533288108,
+      "learning_rate": 5e-06,
+      "loss": 0.2187,
+      "num_input_tokens_seen": 662115404,
+      "step": 3855
+    },
+    {
+      "epoch": 43.801699716713884,
+      "loss": 0.2279517650604248,
+      "loss_ce": 0.01835702732205391,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 662115404,
+      "step": 3855
+    },
+    {
+      "epoch": 43.81303116147309,
+      "grad_norm": 9.729461521704513,
+      "learning_rate": 5e-06,
+      "loss": 0.2394,
+      "num_input_tokens_seen": 662287296,
+      "step": 3856
+    },
+    {
+      "epoch": 43.81303116147309,
+      "loss": 0.15691012144088745,
+      "loss_ce": 0.003193075302988291,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 662287296,
+      "step": 3856
+    },
+    {
+      "epoch": 43.82436260623229,
+      "grad_norm": 8.11043735683055,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 662459236,
+      "step": 3857
+    },
+    {
+      "epoch": 43.82436260623229,
+      "loss": 0.22318127751350403,
+      "loss_ce": 0.026556532829999924,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 662459236,
+      "step": 3857
+    },
+    {
+      "epoch": 43.8356940509915,
+      "grad_norm": 9.073067304511525,
+      "learning_rate": 5e-06,
+      "loss": 0.2003,
+      "num_input_tokens_seen": 662631280,
+      "step": 3858
+    },
+    {
+      "epoch": 43.8356940509915,
+      "loss": 0.12552236020565033,
+      "loss_ce": 0.02737782895565033,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 662631280,
+      "step": 3858
+    },
+    {
+      "epoch": 43.84702549575071,
+      "grad_norm": 5.130539356409205,
+      "learning_rate": 5e-06,
+      "loss": 0.2124,
+      "num_input_tokens_seen": 662802136,
+      "step": 3859
+    },
+    {
+      "epoch": 43.84702549575071,
+      "loss": 0.16339465975761414,
+      "loss_ce": 0.03280993923544884,
+      "loss_iou": 0.46875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 662802136,
+      "step": 3859
+    },
+    {
+      "epoch": 43.858356940509914,
+      "grad_norm": 14.886448583020941,
+      "learning_rate": 5e-06,
+      "loss": 0.2103,
+      "num_input_tokens_seen": 662972652,
+      "step": 3860
+    },
+    {
+      "epoch": 43.858356940509914,
+      "loss": 0.1632852554321289,
+      "loss_ce": 0.00761509221047163,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 662972652,
+      "step": 3860
+    },
+    {
+      "epoch": 43.86968838526912,
+      "grad_norm": 18.515462307673364,
+      "learning_rate": 5e-06,
+      "loss": 0.2135,
+      "num_input_tokens_seen": 663144556,
+      "step": 3861
+    },
+    {
+      "epoch": 43.86968838526912,
+      "loss": 0.27687323093414307,
+      "loss_ce": 0.01797734946012497,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 663144556,
+      "step": 3861
+    },
+    {
+      "epoch": 43.88101983002833,
+      "grad_norm": 10.369040382279163,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 663315964,
+      "step": 3862
+    },
+    {
+      "epoch": 43.88101983002833,
+      "loss": 0.17530646920204163,
+      "loss_ce": 0.01565375365316868,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 663315964,
+      "step": 3862
+    },
+    {
+      "epoch": 43.892351274787536,
+      "grad_norm": 8.739823892770874,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 663487744,
+      "step": 3863
+    },
+    {
+      "epoch": 43.892351274787536,
+      "loss": 0.1954469382762909,
+      "loss_ce": 0.03522966429591179,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 663487744,
+      "step": 3863
+    },
+    {
+      "epoch": 43.903682719546744,
+      "grad_norm": 21.854777367265793,
+      "learning_rate": 5e-06,
+      "loss": 0.2322,
+      "num_input_tokens_seen": 663659864,
+      "step": 3864
+    },
+    {
+      "epoch": 43.903682719546744,
+      "loss": 0.308188796043396,
+      "loss_ce": 0.040427565574645996,
+      "loss_iou": 0.4375,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 663659864,
+      "step": 3864
+    },
+    {
+      "epoch": 43.91501416430595,
+      "grad_norm": 5.780298148519195,
+      "learning_rate": 5e-06,
+      "loss": 0.2676,
+      "num_input_tokens_seen": 663828604,
+      "step": 3865
+    },
+    {
+      "epoch": 43.91501416430595,
+      "loss": 0.17363348603248596,
+      "loss_ce": 0.05888739228248596,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 663828604,
+      "step": 3865
+    },
+    {
+      "epoch": 43.92634560906516,
+      "grad_norm": 11.531013536154632,
+      "learning_rate": 5e-06,
+      "loss": 0.2441,
+      "num_input_tokens_seen": 663999584,
+      "step": 3866
+    },
+    {
+      "epoch": 43.92634560906516,
+      "loss": 0.2707352638244629,
+      "loss_ce": 0.020613208413124084,
+      "loss_iou": 0.453125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 663999584,
+      "step": 3866
+    },
+    {
+      "epoch": 43.93767705382436,
+      "grad_norm": 4.158122993888739,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 664169836,
+      "step": 3867
+    },
+    {
+      "epoch": 43.93767705382436,
+      "loss": 0.191764235496521,
+      "loss_ce": 0.023490317165851593,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 664169836,
+      "step": 3867
+    },
+    {
+      "epoch": 43.94900849858357,
+      "grad_norm": 6.3836287404208285,
+      "learning_rate": 5e-06,
+      "loss": 0.2108,
+      "num_input_tokens_seen": 664341616,
+      "step": 3868
+    },
+    {
+      "epoch": 43.94900849858357,
+      "loss": 0.20721814036369324,
+      "loss_ce": 0.03509899228811264,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 664341616,
+      "step": 3868
+    },
+    {
+      "epoch": 43.960339943342774,
+      "grad_norm": 6.661686630134207,
+      "learning_rate": 5e-06,
+      "loss": 0.1539,
+      "num_input_tokens_seen": 664512632,
+      "step": 3869
+    },
+    {
+      "epoch": 43.960339943342774,
+      "loss": 0.12985064089298248,
+      "loss_ce": 0.00677324878051877,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 664512632,
+      "step": 3869
+    },
+    {
+      "epoch": 43.97167138810198,
+      "grad_norm": 11.260007791505094,
+      "learning_rate": 5e-06,
+      "loss": 0.2184,
+      "num_input_tokens_seen": 664684092,
+      "step": 3870
+    },
+    {
+      "epoch": 43.97167138810198,
+      "loss": 0.21081560850143433,
+      "loss_ce": 0.023742854595184326,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 664684092,
+      "step": 3870
+    },
+    {
+      "epoch": 43.98300283286119,
+      "grad_norm": 18.315713023686772,
+      "learning_rate": 5e-06,
+      "loss": 0.2193,
+      "num_input_tokens_seen": 664855708,
+      "step": 3871
+    },
+    {
+      "epoch": 43.98300283286119,
+      "loss": 0.19681121408939362,
+      "loss_ce": 0.03161955997347832,
+      "loss_iou": 0.515625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 664855708,
+      "step": 3871
+    },
+    {
+      "epoch": 43.994334277620396,
+      "grad_norm": 8.966144041465395,
+      "learning_rate": 5e-06,
+      "loss": 0.165,
+      "num_input_tokens_seen": 665027668,
+      "step": 3872
+    },
+    {
+      "epoch": 43.994334277620396,
+      "loss": 0.19843284785747528,
+      "loss_ce": 0.04087058827280998,
+      "loss_iou": 0.5,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 665027668,
+      "step": 3872
+    },
+    {
+      "epoch": 43.994334277620396,
+      "loss": 0.20960503816604614,
+      "loss_ce": 0.02045709453523159,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 665113460,
+      "step": 3872
+    },
+    {
+      "epoch": 44.005665722379604,
+      "grad_norm": 5.528582486793579,
+      "learning_rate": 5e-06,
+      "loss": 0.1555,
+      "num_input_tokens_seen": 665199152,
+      "step": 3873
+    },
+    {
+      "epoch": 44.005665722379604,
+      "loss": 0.16204151511192322,
+      "loss_ce": 0.02001270093023777,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 665199152,
+      "step": 3873
+    },
+    {
+      "epoch": 44.01699716713881,
+      "grad_norm": 4.395095908779968,
+      "learning_rate": 5e-06,
+      "loss": 0.1865,
+      "num_input_tokens_seen": 665370028,
+      "step": 3874
+    },
+    {
+      "epoch": 44.01699716713881,
+      "loss": 0.1789645254611969,
+      "loss_ce": 0.024179378524422646,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 665370028,
+      "step": 3874
+    },
+    {
+      "epoch": 44.02832861189802,
+      "grad_norm": 17.337216912045225,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 665541468,
+      "step": 3875
+    },
+    {
+      "epoch": 44.02832861189802,
+      "loss": 0.14606338739395142,
+      "loss_ce": 0.01892716810107231,
+      "loss_iou": 0.28125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 665541468,
+      "step": 3875
+    },
+    {
+      "epoch": 44.039660056657226,
+      "grad_norm": 4.722631917871475,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 665713312,
+      "step": 3876
+    },
+    {
+      "epoch": 44.039660056657226,
+      "loss": 0.1612386256456375,
+      "loss_ce": 0.028822844848036766,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 665713312,
+      "step": 3876
+    },
+    {
+      "epoch": 44.05099150141643,
+      "grad_norm": 4.3230895037327555,
+      "learning_rate": 5e-06,
+      "loss": 0.2116,
+      "num_input_tokens_seen": 665885268,
+      "step": 3877
+    },
+    {
+      "epoch": 44.05099150141643,
+      "loss": 0.20948238670825958,
+      "loss_ce": 0.044931601732969284,
+      "loss_iou": 0.078125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 665885268,
+      "step": 3877
+    },
+    {
+      "epoch": 44.06232294617564,
+      "grad_norm": 8.788249643225756,
+      "learning_rate": 5e-06,
+      "loss": 0.1393,
+      "num_input_tokens_seen": 666057036,
+      "step": 3878
+    },
+    {
+      "epoch": 44.06232294617564,
+      "loss": 0.12548784911632538,
+      "loss_ce": 0.01913408376276493,
+      "loss_iou": 0.09228515625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 666057036,
+      "step": 3878
+    },
+    {
+      "epoch": 44.07365439093484,
+      "grad_norm": 4.572224335134698,
+      "learning_rate": 5e-06,
+      "loss": 0.1863,
+      "num_input_tokens_seen": 666228044,
+      "step": 3879
+    },
+    {
+      "epoch": 44.07365439093484,
+      "loss": 0.17247647047042847,
+      "loss_ce": 0.014944737777113914,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 666228044,
+      "step": 3879
+    },
+    {
+      "epoch": 44.08498583569405,
+      "grad_norm": 5.784244766872415,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 666398588,
+      "step": 3880
+    },
+    {
+      "epoch": 44.08498583569405,
+      "loss": 0.22369036078453064,
+      "loss_ce": 0.03066667541861534,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 666398588,
+      "step": 3880
+    },
+    {
+      "epoch": 44.096317280453256,
+      "grad_norm": 7.298966303373416,
+      "learning_rate": 5e-06,
+      "loss": 0.146,
+      "num_input_tokens_seen": 666570360,
+      "step": 3881
+    },
+    {
+      "epoch": 44.096317280453256,
+      "loss": 0.17871423065662384,
+      "loss_ce": 0.016330188140273094,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 666570360,
+      "step": 3881
+    },
+    {
+      "epoch": 44.107648725212464,
+      "grad_norm": 5.166123627681899,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 666741996,
+      "step": 3882
+    },
+    {
+      "epoch": 44.107648725212464,
+      "loss": 0.11732202768325806,
+      "loss_ce": 0.04682641476392746,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 666741996,
+      "step": 3882
+    },
+    {
+      "epoch": 44.11898016997167,
+      "grad_norm": 20.384235799519953,
+      "learning_rate": 5e-06,
+      "loss": 0.2089,
+      "num_input_tokens_seen": 666913588,
+      "step": 3883
+    },
+    {
+      "epoch": 44.11898016997167,
+      "loss": 0.28551924228668213,
+      "loss_ce": 0.017361294478178024,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 666913588,
+      "step": 3883
+    },
+    {
+      "epoch": 44.13031161473088,
+      "grad_norm": 7.911628476498444,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 667085308,
+      "step": 3884
+    },
+    {
+      "epoch": 44.13031161473088,
+      "loss": 0.20761322975158691,
+      "loss_ce": 0.03772187978029251,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 667085308,
+      "step": 3884
+    },
+    {
+      "epoch": 44.141643059490086,
+      "grad_norm": 4.8956588594092185,
+      "learning_rate": 5e-06,
+      "loss": 0.1681,
+      "num_input_tokens_seen": 667257128,
+      "step": 3885
+    },
+    {
+      "epoch": 44.141643059490086,
+      "loss": 0.11757586896419525,
+      "loss_ce": 0.018393736332654953,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 667257128,
+      "step": 3885
+    },
+    {
+      "epoch": 44.15297450424929,
+      "grad_norm": 9.613726583708134,
+      "learning_rate": 5e-06,
+      "loss": 0.1702,
+      "num_input_tokens_seen": 667428216,
+      "step": 3886
+    },
+    {
+      "epoch": 44.15297450424929,
+      "loss": 0.13310682773590088,
+      "loss_ce": 0.027058254927396774,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 667428216,
+      "step": 3886
+    },
+    {
+      "epoch": 44.1643059490085,
+      "grad_norm": 4.528860521900117,
+      "learning_rate": 5e-06,
+      "loss": 0.1893,
+      "num_input_tokens_seen": 667600244,
+      "step": 3887
+    },
+    {
+      "epoch": 44.1643059490085,
+      "loss": 0.11852609366178513,
+      "loss_ce": 0.015498753637075424,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 667600244,
+      "step": 3887
+    },
+    {
+      "epoch": 44.17563739376771,
+      "grad_norm": 4.692495130189492,
+      "learning_rate": 5e-06,
+      "loss": 0.1662,
+      "num_input_tokens_seen": 667772056,
+      "step": 3888
+    },
+    {
+      "epoch": 44.17563739376771,
+      "loss": 0.16544538736343384,
+      "loss_ce": 0.009225909598171711,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 667772056,
+      "step": 3888
+    },
+    {
+      "epoch": 44.186968838526916,
+      "grad_norm": 4.571984747713864,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 667943060,
+      "step": 3889
+    },
+    {
+      "epoch": 44.186968838526916,
+      "loss": 0.14817777276039124,
+      "loss_ce": 0.030120527371764183,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 667943060,
+      "step": 3889
+    },
+    {
+      "epoch": 44.198300283286116,
+      "grad_norm": 21.334265923346287,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 668115148,
+      "step": 3890
+    },
+    {
+      "epoch": 44.198300283286116,
+      "loss": 0.17541903257369995,
+      "loss_ce": 0.039066486060619354,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 668115148,
+      "step": 3890
+    },
+    {
+      "epoch": 44.20963172804532,
+      "grad_norm": 7.800471377143766,
+      "learning_rate": 5e-06,
+      "loss": 0.2,
+      "num_input_tokens_seen": 668285564,
+      "step": 3891
+    },
+    {
+      "epoch": 44.20963172804532,
+      "loss": 0.12931157648563385,
+      "loss_ce": 0.03919317573308945,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 668285564,
+      "step": 3891
+    },
+    {
+      "epoch": 44.22096317280453,
+      "grad_norm": 6.6094242919048956,
+      "learning_rate": 5e-06,
+      "loss": 0.2124,
+      "num_input_tokens_seen": 668457312,
+      "step": 3892
+    },
+    {
+      "epoch": 44.22096317280453,
+      "loss": 0.19351902604103088,
+      "loss_ce": 0.01682223007082939,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 668457312,
+      "step": 3892
+    },
+    {
+      "epoch": 44.23229461756374,
+      "grad_norm": 3.763893800603745,
+      "learning_rate": 5e-06,
+      "loss": 0.1748,
+      "num_input_tokens_seen": 668628532,
+      "step": 3893
+    },
+    {
+      "epoch": 44.23229461756374,
+      "loss": 0.14614585041999817,
+      "loss_ce": 0.013119742274284363,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 668628532,
+      "step": 3893
+    },
+    {
+      "epoch": 44.243626062322946,
+      "grad_norm": 6.6147255674233385,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 668797504,
+      "step": 3894
+    },
+    {
+      "epoch": 44.243626062322946,
+      "loss": 0.1490989476442337,
+      "loss_ce": 0.0556846484541893,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 668797504,
+      "step": 3894
+    },
+    {
+      "epoch": 44.25495750708215,
+      "grad_norm": 3.739099364885454,
+      "learning_rate": 5e-06,
+      "loss": 0.1632,
+      "num_input_tokens_seen": 668968832,
+      "step": 3895
+    },
+    {
+      "epoch": 44.25495750708215,
+      "loss": 0.20089229941368103,
+      "loss_ce": 0.06338009238243103,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 668968832,
+      "step": 3895
+    },
+    {
+      "epoch": 44.26628895184136,
+      "grad_norm": 9.12464093646164,
+      "learning_rate": 5e-06,
+      "loss": 0.1581,
+      "num_input_tokens_seen": 669140524,
+      "step": 3896
+    },
+    {
+      "epoch": 44.26628895184136,
+      "loss": 0.1690121293067932,
+      "loss_ce": 0.017950113862752914,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 669140524,
+      "step": 3896
+    },
+    {
+      "epoch": 44.27762039660057,
+      "grad_norm": 6.8958968158456875,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 669312280,
+      "step": 3897
+    },
+    {
+      "epoch": 44.27762039660057,
+      "loss": 0.20437884330749512,
+      "loss_ce": 0.011233100667595863,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 669312280,
+      "step": 3897
+    },
+    {
+      "epoch": 44.288951841359776,
+      "grad_norm": 7.217350983333874,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 669484004,
+      "step": 3898
+    },
+    {
+      "epoch": 44.288951841359776,
+      "loss": 0.15820790827274323,
+      "loss_ce": 0.018986724317073822,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 669484004,
+      "step": 3898
+    },
+    {
+      "epoch": 44.30028328611898,
+      "grad_norm": 5.162043680061618,
+      "learning_rate": 5e-06,
+      "loss": 0.1978,
+      "num_input_tokens_seen": 669655768,
+      "step": 3899
+    },
+    {
+      "epoch": 44.30028328611898,
+      "loss": 0.2508530020713806,
+      "loss_ce": 0.013853507116436958,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 669655768,
+      "step": 3899
+    },
+    {
+      "epoch": 44.31161473087819,
+      "grad_norm": 5.31066240255363,
+      "learning_rate": 5e-06,
+      "loss": 0.189,
+      "num_input_tokens_seen": 669827588,
+      "step": 3900
+    },
+    {
+      "epoch": 44.31161473087819,
+      "loss": 0.18520700931549072,
+      "loss_ce": 0.008082992397248745,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 669827588,
+      "step": 3900
+    },
+    {
+      "epoch": 44.32294617563739,
+      "grad_norm": 4.28118504428304,
+      "learning_rate": 5e-06,
+      "loss": 0.1752,
+      "num_input_tokens_seen": 669997728,
+      "step": 3901
+    },
+    {
+      "epoch": 44.32294617563739,
+      "loss": 0.11126811802387238,
+      "loss_ce": 0.011841850355267525,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 669997728,
+      "step": 3901
+    },
+    {
+      "epoch": 44.3342776203966,
+      "grad_norm": 3.2329819093097476,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 670169376,
+      "step": 3902
+    },
+    {
+      "epoch": 44.3342776203966,
+      "loss": 0.16909259557724,
+      "loss_ce": 0.015558669343590736,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 670169376,
+      "step": 3902
+    },
+    {
+      "epoch": 44.345609065155806,
+      "grad_norm": 5.902571291429869,
+      "learning_rate": 5e-06,
+      "loss": 0.1749,
+      "num_input_tokens_seen": 670341112,
+      "step": 3903
+    },
+    {
+      "epoch": 44.345609065155806,
+      "loss": 0.1980319768190384,
+      "loss_ce": 0.04535253345966339,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 670341112,
+      "step": 3903
+    },
+    {
+      "epoch": 44.35694050991501,
+      "grad_norm": 6.6970159763365364,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 670510016,
+      "step": 3904
+    },
+    {
+      "epoch": 44.35694050991501,
+      "loss": 0.12334737181663513,
+      "loss_ce": 0.02100667729973793,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 670510016,
+      "step": 3904
+    },
+    {
+      "epoch": 44.36827195467422,
+      "grad_norm": 13.377695266846061,
+      "learning_rate": 5e-06,
+      "loss": 0.1944,
+      "num_input_tokens_seen": 670680452,
+      "step": 3905
+    },
+    {
+      "epoch": 44.36827195467422,
+      "loss": 0.148904949426651,
+      "loss_ce": 0.028024822473526,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 670680452,
+      "step": 3905
+    },
+    {
+      "epoch": 44.37960339943343,
+      "grad_norm": 7.838408355591025,
+      "learning_rate": 5e-06,
+      "loss": 0.1867,
+      "num_input_tokens_seen": 670852512,
+      "step": 3906
+    },
+    {
+      "epoch": 44.37960339943343,
+      "loss": 0.14574454724788666,
+      "loss_ce": 0.00911734439432621,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 670852512,
+      "step": 3906
+    },
+    {
+      "epoch": 44.390934844192635,
+      "grad_norm": 5.4060900424333616,
+      "learning_rate": 5e-06,
+      "loss": 0.1881,
+      "num_input_tokens_seen": 671024224,
+      "step": 3907
+    },
+    {
+      "epoch": 44.390934844192635,
+      "loss": 0.0828448235988617,
+      "loss_ce": 0.017628755420446396,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 671024224,
+      "step": 3907
+    },
+    {
+      "epoch": 44.40226628895184,
+      "grad_norm": 4.41292949501653,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 671196500,
+      "step": 3908
+    },
+    {
+      "epoch": 44.40226628895184,
+      "loss": 0.13619935512542725,
+      "loss_ce": 0.00952088087797165,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 671196500,
+      "step": 3908
+    },
+    {
+      "epoch": 44.41359773371105,
+      "grad_norm": 5.1494649175171014,
+      "learning_rate": 5e-06,
+      "loss": 0.21,
+      "num_input_tokens_seen": 671368548,
+      "step": 3909
+    },
+    {
+      "epoch": 44.41359773371105,
+      "loss": 0.1771545708179474,
+      "loss_ce": 0.00851443037390709,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 671368548,
+      "step": 3909
+    },
+    {
+      "epoch": 44.42492917847026,
+      "grad_norm": 10.00238723031077,
+      "learning_rate": 5e-06,
+      "loss": 0.2128,
+      "num_input_tokens_seen": 671540372,
+      "step": 3910
+    },
+    {
+      "epoch": 44.42492917847026,
+      "loss": 0.2260538935661316,
+      "loss_ce": 0.014048272743821144,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 671540372,
+      "step": 3910
+    },
+    {
+      "epoch": 44.436260623229465,
+      "grad_norm": 4.331142078752378,
+      "learning_rate": 5e-06,
+      "loss": 0.1383,
+      "num_input_tokens_seen": 671710528,
+      "step": 3911
+    },
+    {
+      "epoch": 44.436260623229465,
+      "loss": 0.1739613562822342,
+      "loss_ce": 0.023326583206653595,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 671710528,
+      "step": 3911
+    },
+    {
+      "epoch": 44.447592067988666,
+      "grad_norm": 8.844643532094778,
+      "learning_rate": 5e-06,
+      "loss": 0.1864,
+      "num_input_tokens_seen": 671881304,
+      "step": 3912
+    },
+    {
+      "epoch": 44.447592067988666,
+      "loss": 0.18756544589996338,
+      "loss_ce": 0.02997267246246338,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 671881304,
+      "step": 3912
+    },
+    {
+      "epoch": 44.45892351274787,
+      "grad_norm": 6.510764341126988,
+      "learning_rate": 5e-06,
+      "loss": 0.1857,
+      "num_input_tokens_seen": 672053348,
+      "step": 3913
+    },
+    {
+      "epoch": 44.45892351274787,
+      "loss": 0.2138848453760147,
+      "loss_ce": 0.013842117041349411,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 672053348,
+      "step": 3913
+    },
+    {
+      "epoch": 44.47025495750708,
+      "grad_norm": 9.006163268268532,
+      "learning_rate": 5e-06,
+      "loss": 0.1907,
+      "num_input_tokens_seen": 672224644,
+      "step": 3914
+    },
+    {
+      "epoch": 44.47025495750708,
+      "loss": 0.1696176528930664,
+      "loss_ce": 0.020905490964651108,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 672224644,
+      "step": 3914
+    },
+    {
+      "epoch": 44.48158640226629,
+      "grad_norm": 49.08920396255695,
+      "learning_rate": 5e-06,
+      "loss": 0.1507,
+      "num_input_tokens_seen": 672394948,
+      "step": 3915
+    },
+    {
+      "epoch": 44.48158640226629,
+      "loss": 0.09520304948091507,
+      "loss_ce": 0.02095378190279007,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 672394948,
+      "step": 3915
+    },
+    {
+      "epoch": 44.492917847025495,
+      "grad_norm": 22.289723145068706,
+      "learning_rate": 5e-06,
+      "loss": 0.2131,
+      "num_input_tokens_seen": 672567020,
+      "step": 3916
+    },
+    {
+      "epoch": 44.492917847025495,
+      "loss": 0.24901898205280304,
+      "loss_ce": 0.013942087069153786,
+      "loss_iou": 0.390625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 672567020,
+      "step": 3916
+    },
+    {
+      "epoch": 44.5042492917847,
+      "grad_norm": 9.872482670890951,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 672739200,
+      "step": 3917
+    },
+    {
+      "epoch": 44.5042492917847,
+      "loss": 0.20507213473320007,
+      "loss_ce": 0.009820652194321156,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 672739200,
+      "step": 3917
+    },
+    {
+      "epoch": 44.51558073654391,
+      "grad_norm": 4.701513319513321,
+      "learning_rate": 5e-06,
+      "loss": 0.1669,
+      "num_input_tokens_seen": 672910356,
+      "step": 3918
+    },
+    {
+      "epoch": 44.51558073654391,
+      "loss": 0.1403428018093109,
+      "loss_ce": 0.02788553573191166,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 672910356,
+      "step": 3918
+    },
+    {
+      "epoch": 44.52691218130312,
+      "grad_norm": 5.04437684335888,
+      "learning_rate": 5e-06,
+      "loss": 0.1928,
+      "num_input_tokens_seen": 673081496,
+      "step": 3919
+    },
+    {
+      "epoch": 44.52691218130312,
+      "loss": 0.18987755477428436,
+      "loss_ce": 0.017758414149284363,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 673081496,
+      "step": 3919
+    },
+    {
+      "epoch": 44.538243626062325,
+      "grad_norm": 11.325975447959623,
+      "learning_rate": 5e-06,
+      "loss": 0.2147,
+      "num_input_tokens_seen": 673253860,
+      "step": 3920
+    },
+    {
+      "epoch": 44.538243626062325,
+      "loss": 0.17633388936519623,
+      "loss_ce": 0.03412197530269623,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 673253860,
+      "step": 3920
+    },
+    {
+      "epoch": 44.54957507082153,
+      "grad_norm": 12.70179835878848,
+      "learning_rate": 5e-06,
+      "loss": 0.2076,
+      "num_input_tokens_seen": 673424932,
+      "step": 3921
+    },
+    {
+      "epoch": 44.54957507082153,
+      "loss": 0.1992478221654892,
+      "loss_ce": 0.0031723843421787024,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 673424932,
+      "step": 3921
+    },
+    {
+      "epoch": 44.56090651558074,
+      "grad_norm": 4.233158942637238,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 673596912,
+      "step": 3922
+    },
+    {
+      "epoch": 44.56090651558074,
+      "loss": 0.12535905838012695,
+      "loss_ce": 0.025475017726421356,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 673596912,
+      "step": 3922
+    },
+    {
+      "epoch": 44.57223796033994,
+      "grad_norm": 6.590398737913662,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 673767644,
+      "step": 3923
+    },
+    {
+      "epoch": 44.57223796033994,
+      "loss": 0.11935992538928986,
+      "loss_ce": 0.024755436927080154,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 673767644,
+      "step": 3923
+    },
+    {
+      "epoch": 44.58356940509915,
+      "grad_norm": 8.997132730978791,
+      "learning_rate": 5e-06,
+      "loss": 0.1728,
+      "num_input_tokens_seen": 673939180,
+      "step": 3924
+    },
+    {
+      "epoch": 44.58356940509915,
+      "loss": 0.25624096393585205,
+      "loss_ce": 0.020828373730182648,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 673939180,
+      "step": 3924
+    },
+    {
+      "epoch": 44.594900849858355,
+      "grad_norm": 4.469462741478929,
+      "learning_rate": 5e-06,
+      "loss": 0.1665,
+      "num_input_tokens_seen": 674110892,
+      "step": 3925
+    },
+    {
+      "epoch": 44.594900849858355,
+      "loss": 0.11365257948637009,
+      "loss_ce": 0.03238426148891449,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 674110892,
+      "step": 3925
+    },
+    {
+      "epoch": 44.60623229461756,
+      "grad_norm": 5.543728005249897,
+      "learning_rate": 5e-06,
+      "loss": 0.133,
+      "num_input_tokens_seen": 674282916,
+      "step": 3926
+    },
+    {
+      "epoch": 44.60623229461756,
+      "loss": 0.11739157140254974,
+      "loss_ce": 0.012136444449424744,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 674282916,
+      "step": 3926
+    },
+    {
+      "epoch": 44.61756373937677,
+      "grad_norm": 3.9444757987061574,
+      "learning_rate": 5e-06,
+      "loss": 0.1813,
+      "num_input_tokens_seen": 674454380,
+      "step": 3927
+    },
+    {
+      "epoch": 44.61756373937677,
+      "loss": 0.19512583315372467,
+      "loss_ce": 0.04400278627872467,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 674454380,
+      "step": 3927
+    },
+    {
+      "epoch": 44.62889518413598,
+      "grad_norm": 7.615596531975357,
+      "learning_rate": 5e-06,
+      "loss": 0.1729,
+      "num_input_tokens_seen": 674626708,
+      "step": 3928
+    },
+    {
+      "epoch": 44.62889518413598,
+      "loss": 0.21386294066905975,
+      "loss_ce": 0.012416411191225052,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 674626708,
+      "step": 3928
+    },
+    {
+      "epoch": 44.640226628895185,
+      "grad_norm": 13.07656539893205,
+      "learning_rate": 5e-06,
+      "loss": 0.145,
+      "num_input_tokens_seen": 674796932,
+      "step": 3929
+    },
+    {
+      "epoch": 44.640226628895185,
+      "loss": 0.13044890761375427,
+      "loss_ce": 0.022447196766734123,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 674796932,
+      "step": 3929
+    },
+    {
+      "epoch": 44.65155807365439,
+      "grad_norm": 5.034325857680527,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 674969036,
+      "step": 3930
+    },
+    {
+      "epoch": 44.65155807365439,
+      "loss": 0.08581889420747757,
+      "loss_ce": 0.014499313198029995,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 674969036,
+      "step": 3930
+    },
+    {
+      "epoch": 44.6628895184136,
+      "grad_norm": 8.033244599908855,
+      "learning_rate": 5e-06,
+      "loss": 0.1752,
+      "num_input_tokens_seen": 675139804,
+      "step": 3931
+    },
+    {
+      "epoch": 44.6628895184136,
+      "loss": 0.1502116322517395,
+      "loss_ce": 0.025913530960679054,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 675139804,
+      "step": 3931
+    },
+    {
+      "epoch": 44.67422096317281,
+      "grad_norm": 4.438394760476305,
+      "learning_rate": 5e-06,
+      "loss": 0.2002,
+      "num_input_tokens_seen": 675311184,
+      "step": 3932
+    },
+    {
+      "epoch": 44.67422096317281,
+      "loss": 0.18503566086292267,
+      "loss_ce": 0.03879541903734207,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 675311184,
+      "step": 3932
+    },
+    {
+      "epoch": 44.685552407932015,
+      "grad_norm": 15.97628154259988,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 675482768,
+      "step": 3933
+    },
+    {
+      "epoch": 44.685552407932015,
+      "loss": 0.3114229440689087,
+      "loss_ce": 0.017416590824723244,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 675482768,
+      "step": 3933
+    },
+    {
+      "epoch": 44.696883852691215,
+      "grad_norm": 6.345179098172619,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 675654556,
+      "step": 3934
+    },
+    {
+      "epoch": 44.696883852691215,
+      "loss": 0.13065677881240845,
+      "loss_ce": 0.008006628602743149,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 675654556,
+      "step": 3934
+    },
+    {
+      "epoch": 44.70821529745042,
+      "grad_norm": 6.999335904516425,
+      "learning_rate": 5e-06,
+      "loss": 0.1705,
+      "num_input_tokens_seen": 675826228,
+      "step": 3935
+    },
+    {
+      "epoch": 44.70821529745042,
+      "loss": 0.11878420412540436,
+      "loss_ce": 0.014200454577803612,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 675826228,
+      "step": 3935
+    },
+    {
+      "epoch": 44.71954674220963,
+      "grad_norm": 8.656115297011691,
+      "learning_rate": 5e-06,
+      "loss": 0.1705,
+      "num_input_tokens_seen": 675997212,
+      "step": 3936
+    },
+    {
+      "epoch": 44.71954674220963,
+      "loss": 0.17036622762680054,
+      "loss_ce": 0.07292358577251434,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 675997212,
+      "step": 3936
+    },
+    {
+      "epoch": 44.73087818696884,
+      "grad_norm": 5.791122415515461,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 676169256,
+      "step": 3937
+    },
+    {
+      "epoch": 44.73087818696884,
+      "loss": 0.14486059546470642,
+      "loss_ce": 0.010613763704895973,
+      "loss_iou": 0.578125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 676169256,
+      "step": 3937
+    },
+    {
+      "epoch": 44.742209631728045,
+      "grad_norm": 5.713523697325441,
+      "learning_rate": 5e-06,
+      "loss": 0.224,
+      "num_input_tokens_seen": 676340820,
+      "step": 3938
+    },
+    {
+      "epoch": 44.742209631728045,
+      "loss": 0.2946818768978119,
+      "loss_ce": 0.03241381421685219,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 676340820,
+      "step": 3938
+    },
+    {
+      "epoch": 44.75354107648725,
+      "grad_norm": 7.450730756549203,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 676511092,
+      "step": 3939
+    },
+    {
+      "epoch": 44.75354107648725,
+      "loss": 0.09902353584766388,
+      "loss_ce": 0.01612253114581108,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 676511092,
+      "step": 3939
+    },
+    {
+      "epoch": 44.76487252124646,
+      "grad_norm": 7.619284153921158,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 676682856,
+      "step": 3940
+    },
+    {
+      "epoch": 44.76487252124646,
+      "loss": 0.20546351373195648,
+      "loss_ce": 0.03425990790128708,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 676682856,
+      "step": 3940
+    },
+    {
+      "epoch": 44.77620396600567,
+      "grad_norm": 10.094560451988361,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 676854932,
+      "step": 3941
+    },
+    {
+      "epoch": 44.77620396600567,
+      "loss": 0.1597079336643219,
+      "loss_ce": 0.024301443248987198,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 676854932,
+      "step": 3941
+    },
+    {
+      "epoch": 44.787535410764875,
+      "grad_norm": 33.792708028340655,
+      "learning_rate": 5e-06,
+      "loss": 0.2305,
+      "num_input_tokens_seen": 677027040,
+      "step": 3942
+    },
+    {
+      "epoch": 44.787535410764875,
+      "loss": 0.17407982051372528,
+      "loss_ce": 0.017768792808055878,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 677027040,
+      "step": 3942
+    },
+    {
+      "epoch": 44.79886685552408,
+      "grad_norm": 71.07096406548798,
+      "learning_rate": 5e-06,
+      "loss": 0.2148,
+      "num_input_tokens_seen": 677198864,
+      "step": 3943
+    },
+    {
+      "epoch": 44.79886685552408,
+      "loss": 0.18417896330356598,
+      "loss_ce": 0.020955689251422882,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 677198864,
+      "step": 3943
+    },
+    {
+      "epoch": 44.81019830028329,
+      "grad_norm": 5.615255253349635,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 677370844,
+      "step": 3944
+    },
+    {
+      "epoch": 44.81019830028329,
+      "loss": 0.2352963089942932,
+      "loss_ce": 0.019262365996837616,
+      "loss_iou": 0.46875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 677370844,
+      "step": 3944
+    },
+    {
+      "epoch": 44.82152974504249,
+      "grad_norm": 5.8238037477455205,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 677542668,
+      "step": 3945
+    },
+    {
+      "epoch": 44.82152974504249,
+      "loss": 0.14084841310977936,
+      "loss_ce": 0.01236940547823906,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 677542668,
+      "step": 3945
+    },
+    {
+      "epoch": 44.8328611898017,
+      "grad_norm": 6.238451023563346,
+      "learning_rate": 5e-06,
+      "loss": 0.2078,
+      "num_input_tokens_seen": 677714584,
+      "step": 3946
+    },
+    {
+      "epoch": 44.8328611898017,
+      "loss": 0.15844382345676422,
+      "loss_ce": 0.015530005097389221,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 677714584,
+      "step": 3946
+    },
+    {
+      "epoch": 44.844192634560905,
+      "grad_norm": 6.717142131780718,
+      "learning_rate": 5e-06,
+      "loss": 0.1627,
+      "num_input_tokens_seen": 677886160,
+      "step": 3947
+    },
+    {
+      "epoch": 44.844192634560905,
+      "loss": 0.20618517696857452,
+      "loss_ce": 0.022469352930784225,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 677886160,
+      "step": 3947
+    },
+    {
+      "epoch": 44.85552407932011,
+      "grad_norm": 5.418955008781779,
+      "learning_rate": 5e-06,
+      "loss": 0.1681,
+      "num_input_tokens_seen": 678057140,
+      "step": 3948
+    },
+    {
+      "epoch": 44.85552407932011,
+      "loss": 0.16610217094421387,
+      "loss_ce": 0.01214098371565342,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 678057140,
+      "step": 3948
+    },
+    {
+      "epoch": 44.86685552407932,
+      "grad_norm": 6.785527307628056,
+      "learning_rate": 5e-06,
+      "loss": 0.1788,
+      "num_input_tokens_seen": 678228316,
+      "step": 3949
+    },
+    {
+      "epoch": 44.86685552407932,
+      "loss": 0.1471843719482422,
+      "loss_ce": 0.025724399834871292,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 678228316,
+      "step": 3949
+    },
+    {
+      "epoch": 44.87818696883853,
+      "grad_norm": 5.509305973758549,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 678398708,
+      "step": 3950
+    },
+    {
+      "epoch": 44.87818696883853,
+      "loss": 0.25689616799354553,
+      "loss_ce": 0.019652530550956726,
+      "loss_iou": 0.390625,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 678398708,
+      "step": 3950
+    },
+    {
+      "epoch": 44.889518413597735,
+      "grad_norm": 7.6875765451044105,
+      "learning_rate": 5e-06,
+      "loss": 0.2383,
+      "num_input_tokens_seen": 678570344,
+      "step": 3951
+    },
+    {
+      "epoch": 44.889518413597735,
+      "loss": 0.24902088940143585,
+      "loss_ce": 0.0131810512393713,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 678570344,
+      "step": 3951
+    },
+    {
+      "epoch": 44.90084985835694,
+      "grad_norm": 19.312373410910233,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 678740492,
+      "step": 3952
+    },
+    {
+      "epoch": 44.90084985835694,
+      "loss": 0.1345103681087494,
+      "loss_ce": 0.007191033102571964,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 678740492,
+      "step": 3952
+    },
+    {
+      "epoch": 44.91218130311615,
+      "grad_norm": 4.775684109064375,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 678911784,
+      "step": 3953
+    },
+    {
+      "epoch": 44.91218130311615,
+      "loss": 0.1061486005783081,
+      "loss_ce": 0.009407874196767807,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 678911784,
+      "step": 3953
+    },
+    {
+      "epoch": 44.92351274787536,
+      "grad_norm": 7.61891778490257,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 679083040,
+      "step": 3954
+    },
+    {
+      "epoch": 44.92351274787536,
+      "loss": 0.1314413845539093,
+      "loss_ce": 0.013857155106961727,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 679083040,
+      "step": 3954
+    },
+    {
+      "epoch": 44.934844192634564,
+      "grad_norm": 6.6918706288396175,
+      "learning_rate": 5e-06,
+      "loss": 0.1622,
+      "num_input_tokens_seen": 679253388,
+      "step": 3955
+    },
+    {
+      "epoch": 44.934844192634564,
+      "loss": 0.12126082181930542,
+      "loss_ce": 0.02278059720993042,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 679253388,
+      "step": 3955
+    },
+    {
+      "epoch": 44.946175637393765,
+      "grad_norm": 6.841514197216895,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 679423444,
+      "step": 3956
+    },
+    {
+      "epoch": 44.946175637393765,
+      "loss": 0.10983335971832275,
+      "loss_ce": 0.013764026574790478,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 679423444,
+      "step": 3956
+    },
+    {
+      "epoch": 44.95750708215297,
+      "grad_norm": 7.339101136394465,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 679595260,
+      "step": 3957
+    },
+    {
+      "epoch": 44.95750708215297,
+      "loss": 0.3390614092350006,
+      "loss_ce": 0.031169582158327103,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 679595260,
+      "step": 3957
+    },
+    {
+      "epoch": 44.96883852691218,
+      "grad_norm": 4.74519792464132,
+      "learning_rate": 5e-06,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 679766124,
+      "step": 3958
+    },
+    {
+      "epoch": 44.96883852691218,
+      "loss": 0.2327853888273239,
+      "loss_ce": 0.03436008840799332,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 679766124,
+      "step": 3958
+    },
+    {
+      "epoch": 44.98016997167139,
+      "grad_norm": 4.51509256827041,
+      "learning_rate": 5e-06,
+      "loss": 0.1314,
+      "num_input_tokens_seen": 679936368,
+      "step": 3959
+    },
+    {
+      "epoch": 44.98016997167139,
+      "loss": 0.1303640902042389,
+      "loss_ce": 0.05376496911048889,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 679936368,
+      "step": 3959
+    },
+    {
+      "epoch": 44.991501416430594,
+      "grad_norm": 5.104430236835701,
+      "learning_rate": 5e-06,
+      "loss": 0.1661,
+      "num_input_tokens_seen": 680107800,
+      "step": 3960
+    },
+    {
+      "epoch": 44.991501416430594,
+      "loss": 0.12421780079603195,
+      "loss_ce": 0.018718529492616653,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 680107800,
+      "step": 3960
+    },
+    {
+      "epoch": 44.991501416430594,
+      "loss": 0.20759907364845276,
+      "loss_ce": 0.02623310312628746,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 680235652,
+      "step": 3960
+    },
+    {
+      "epoch": 45.0028328611898,
+      "grad_norm": 9.245029315616877,
+      "learning_rate": 5e-06,
+      "loss": 0.1847,
+      "num_input_tokens_seen": 680278628,
+      "step": 3961
+    },
+    {
+      "epoch": 45.0028328611898,
+      "loss": 0.2199929654598236,
+      "loss_ce": 0.017203668132424355,
+      "loss_iou": 0.609375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 680278628,
+      "step": 3961
+    },
+    {
+      "epoch": 45.01416430594901,
+      "grad_norm": 13.978362735732325,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 680449940,
+      "step": 3962
+    },
+    {
+      "epoch": 45.01416430594901,
+      "loss": 0.1891382485628128,
+      "loss_ce": 0.019948795437812805,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 680449940,
+      "step": 3962
+    },
+    {
+      "epoch": 45.02549575070822,
+      "grad_norm": 15.361499531571992,
+      "learning_rate": 5e-06,
+      "loss": 0.1655,
+      "num_input_tokens_seen": 680621760,
+      "step": 3963
+    },
+    {
+      "epoch": 45.02549575070822,
+      "loss": 0.07439267635345459,
+      "loss_ce": 0.003317241556942463,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 680621760,
+      "step": 3963
+    },
+    {
+      "epoch": 45.036827195467424,
+      "grad_norm": 3.89474739029748,
+      "learning_rate": 5e-06,
+      "loss": 0.1453,
+      "num_input_tokens_seen": 680793512,
+      "step": 3964
+    },
+    {
+      "epoch": 45.036827195467424,
+      "loss": 0.10964441299438477,
+      "loss_ce": 0.020258424803614616,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 680793512,
+      "step": 3964
+    },
+    {
+      "epoch": 45.04815864022663,
+      "grad_norm": 10.564883519250367,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 680965480,
+      "step": 3965
+    },
+    {
+      "epoch": 45.04815864022663,
+      "loss": 0.12299640476703644,
+      "loss_ce": 0.023661689832806587,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 680965480,
+      "step": 3965
+    },
+    {
+      "epoch": 45.05949008498584,
+      "grad_norm": 5.677005087711661,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 681137032,
+      "step": 3966
+    },
+    {
+      "epoch": 45.05949008498584,
+      "loss": 0.18711864948272705,
+      "loss_ce": 0.008758679032325745,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 681137032,
+      "step": 3966
+    },
+    {
+      "epoch": 45.07082152974504,
+      "grad_norm": 3.1774431762700535,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 681308340,
+      "step": 3967
+    },
+    {
+      "epoch": 45.07082152974504,
+      "loss": 0.20559372007846832,
+      "loss_ce": 0.015316622331738472,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 681308340,
+      "step": 3967
+    },
+    {
+      "epoch": 45.08215297450425,
+      "grad_norm": 3.293653747544128,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 681479768,
+      "step": 3968
+    },
+    {
+      "epoch": 45.08215297450425,
+      "loss": 0.07989273220300674,
+      "loss_ce": 0.01559219416230917,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 681479768,
+      "step": 3968
+    },
+    {
+      "epoch": 45.093484419263454,
+      "grad_norm": 13.85451427557297,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 681649276,
+      "step": 3969
+    },
+    {
+      "epoch": 45.093484419263454,
+      "loss": 0.22615768015384674,
+      "loss_ce": 0.010337376967072487,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 681649276,
+      "step": 3969
+    },
+    {
+      "epoch": 45.10481586402266,
+      "grad_norm": 19.43299757201911,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 681819948,
+      "step": 3970
+    },
+    {
+      "epoch": 45.10481586402266,
+      "loss": 0.13635998964309692,
+      "loss_ce": 0.004218881484121084,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 681819948,
+      "step": 3970
+    },
+    {
+      "epoch": 45.11614730878187,
+      "grad_norm": 5.351304537964111,
+      "learning_rate": 5e-06,
+      "loss": 0.1722,
+      "num_input_tokens_seen": 681990612,
+      "step": 3971
+    },
+    {
+      "epoch": 45.11614730878187,
+      "loss": 0.29831600189208984,
+      "loss_ce": 0.024756435304880142,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 681990612,
+      "step": 3971
+    },
+    {
+      "epoch": 45.12747875354108,
+      "grad_norm": 24.48441549631094,
+      "learning_rate": 5e-06,
+      "loss": 0.1587,
+      "num_input_tokens_seen": 682162248,
+      "step": 3972
+    },
+    {
+      "epoch": 45.12747875354108,
+      "loss": 0.1765347421169281,
+      "loss_ce": 0.0029812734574079514,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 682162248,
+      "step": 3972
+    },
+    {
+      "epoch": 45.138810198300284,
+      "grad_norm": 4.795848369809877,
+      "learning_rate": 5e-06,
+      "loss": 0.1697,
+      "num_input_tokens_seen": 682331992,
+      "step": 3973
+    },
+    {
+      "epoch": 45.138810198300284,
+      "loss": 0.12189280241727829,
+      "loss_ce": 0.02449595369398594,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 682331992,
+      "step": 3973
+    },
+    {
+      "epoch": 45.15014164305949,
+      "grad_norm": 10.173440029675737,
+      "learning_rate": 5e-06,
+      "loss": 0.1363,
+      "num_input_tokens_seen": 682503792,
+      "step": 3974
+    },
+    {
+      "epoch": 45.15014164305949,
+      "loss": 0.22476442158222198,
+      "loss_ce": 0.015322279185056686,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 682503792,
+      "step": 3974
+    },
+    {
+      "epoch": 45.1614730878187,
+      "grad_norm": 6.554892812737551,
+      "learning_rate": 5e-06,
+      "loss": 0.2,
+      "num_input_tokens_seen": 682675432,
+      "step": 3975
+    },
+    {
+      "epoch": 45.1614730878187,
+      "loss": 0.2163558304309845,
+      "loss_ce": 0.0121016725897789,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 682675432,
+      "step": 3975
+    },
+    {
+      "epoch": 45.172804532577906,
+      "grad_norm": 9.94909079286914,
+      "learning_rate": 5e-06,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 682846604,
+      "step": 3976
+    },
+    {
+      "epoch": 45.172804532577906,
+      "loss": 0.1480620950460434,
+      "loss_ce": 0.013418532907962799,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 682846604,
+      "step": 3976
+    },
+    {
+      "epoch": 45.184135977337114,
+      "grad_norm": 4.859778556428821,
+      "learning_rate": 5e-06,
+      "loss": 0.1761,
+      "num_input_tokens_seen": 683018104,
+      "step": 3977
+    },
+    {
+      "epoch": 45.184135977337114,
+      "loss": 0.21864371001720428,
+      "loss_ce": 0.02110341563820839,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 683018104,
+      "step": 3977
+    },
+    {
+      "epoch": 45.195467422096314,
+      "grad_norm": 8.340452487209136,
+      "learning_rate": 5e-06,
+      "loss": 0.185,
+      "num_input_tokens_seen": 683188376,
+      "step": 3978
+    },
+    {
+      "epoch": 45.195467422096314,
+      "loss": 0.13148869574069977,
+      "loss_ce": 0.027393236756324768,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 683188376,
+      "step": 3978
+    },
+    {
+      "epoch": 45.20679886685552,
+      "grad_norm": 6.471837698938179,
+      "learning_rate": 5e-06,
+      "loss": 0.1516,
+      "num_input_tokens_seen": 683360000,
+      "step": 3979
+    },
+    {
+      "epoch": 45.20679886685552,
+      "loss": 0.2048691213130951,
+      "loss_ce": 0.00940403901040554,
+      "loss_iou": 0.052001953125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 683360000,
+      "step": 3979
+    },
+    {
+      "epoch": 45.21813031161473,
+      "grad_norm": 4.966360828511372,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 683531784,
+      "step": 3980
+    },
+    {
+      "epoch": 45.21813031161473,
+      "loss": 0.12496919929981232,
+      "loss_ce": 0.009032929316163063,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 683531784,
+      "step": 3980
+    },
+    {
+      "epoch": 45.22946175637394,
+      "grad_norm": 3.2656518214800245,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 683703388,
+      "step": 3981
+    },
+    {
+      "epoch": 45.22946175637394,
+      "loss": 0.17586511373519897,
+      "loss_ce": 0.0028914890717715025,
+      "loss_iou": 0.390625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 683703388,
+      "step": 3981
+    },
+    {
+      "epoch": 45.240793201133144,
+      "grad_norm": 7.119328119959129,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 683875044,
+      "step": 3982
+    },
+    {
+      "epoch": 45.240793201133144,
+      "loss": 0.14759066700935364,
+      "loss_ce": 0.042137183248996735,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 683875044,
+      "step": 3982
+    },
+    {
+      "epoch": 45.25212464589235,
+      "grad_norm": 5.439364781816015,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 684044320,
+      "step": 3983
+    },
+    {
+      "epoch": 45.25212464589235,
+      "loss": 0.11087010055780411,
+      "loss_ce": 0.008514144457876682,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 684044320,
+      "step": 3983
+    },
+    {
+      "epoch": 45.26345609065156,
+      "grad_norm": 5.3909633068184295,
+      "learning_rate": 5e-06,
+      "loss": 0.1855,
+      "num_input_tokens_seen": 684215692,
+      "step": 3984
+    },
+    {
+      "epoch": 45.26345609065156,
+      "loss": 0.12300942093133926,
+      "loss_ce": 0.007866598665714264,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 684215692,
+      "step": 3984
+    },
+    {
+      "epoch": 45.274787535410766,
+      "grad_norm": 7.406830467154554,
+      "learning_rate": 5e-06,
+      "loss": 0.1602,
+      "num_input_tokens_seen": 684387580,
+      "step": 3985
+    },
+    {
+      "epoch": 45.274787535410766,
+      "loss": 0.17128312587738037,
+      "loss_ce": 0.03551041707396507,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 684387580,
+      "step": 3985
+    },
+    {
+      "epoch": 45.286118980169974,
+      "grad_norm": 4.513940040204225,
+      "learning_rate": 5e-06,
+      "loss": 0.197,
+      "num_input_tokens_seen": 684558308,
+      "step": 3986
+    },
+    {
+      "epoch": 45.286118980169974,
+      "loss": 0.1830013245344162,
+      "loss_ce": 0.026324080303311348,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 684558308,
+      "step": 3986
+    },
+    {
+      "epoch": 45.29745042492918,
+      "grad_norm": 5.550275942700856,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 684730044,
+      "step": 3987
+    },
+    {
+      "epoch": 45.29745042492918,
+      "loss": 0.10014459490776062,
+      "loss_ce": 0.00367852789349854,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 684730044,
+      "step": 3987
+    },
+    {
+      "epoch": 45.30878186968839,
+      "grad_norm": 7.533654743288638,
+      "learning_rate": 5e-06,
+      "loss": 0.1355,
+      "num_input_tokens_seen": 684899472,
+      "step": 3988
+    },
+    {
+      "epoch": 45.30878186968839,
+      "loss": 0.12186650931835175,
+      "loss_ce": 0.012094780802726746,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 684899472,
+      "step": 3988
+    },
+    {
+      "epoch": 45.32011331444759,
+      "grad_norm": 4.292112600348843,
+      "learning_rate": 5e-06,
+      "loss": 0.1531,
+      "num_input_tokens_seen": 685070948,
+      "step": 3989
+    },
+    {
+      "epoch": 45.32011331444759,
+      "loss": 0.13786914944648743,
+      "loss_ce": 0.02712084725499153,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 685070948,
+      "step": 3989
+    },
+    {
+      "epoch": 45.331444759206796,
+      "grad_norm": 3.924196065163375,
+      "learning_rate": 5e-06,
+      "loss": 0.1543,
+      "num_input_tokens_seen": 685241192,
+      "step": 3990
+    },
+    {
+      "epoch": 45.331444759206796,
+      "loss": 0.13688303530216217,
+      "loss_ce": 0.023906951770186424,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 685241192,
+      "step": 3990
+    },
+    {
+      "epoch": 45.342776203966004,
+      "grad_norm": 3.5673849325899623,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 685412808,
+      "step": 3991
+    },
+    {
+      "epoch": 45.342776203966004,
+      "loss": 0.1533830612897873,
+      "loss_ce": 0.023591801524162292,
+      "loss_iou": 0.054443359375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 685412808,
+      "step": 3991
+    },
+    {
+      "epoch": 45.35410764872521,
+      "grad_norm": 32.92043305837096,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 685584656,
+      "step": 3992
+    },
+    {
+      "epoch": 45.35410764872521,
+      "loss": 0.13519130647182465,
+      "loss_ce": 0.006269804667681456,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 685584656,
+      "step": 3992
+    },
+    {
+      "epoch": 45.36543909348442,
+      "grad_norm": 4.149902785668814,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 685756264,
+      "step": 3993
+    },
+    {
+      "epoch": 45.36543909348442,
+      "loss": 0.12544085085391998,
+      "loss_ce": 0.012953061610460281,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 685756264,
+      "step": 3993
+    },
+    {
+      "epoch": 45.376770538243626,
+      "grad_norm": 9.920051263143561,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 685928208,
+      "step": 3994
+    },
+    {
+      "epoch": 45.376770538243626,
+      "loss": 0.11352415382862091,
+      "loss_ce": 0.01623411476612091,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 685928208,
+      "step": 3994
+    },
+    {
+      "epoch": 45.388101983002834,
+      "grad_norm": 7.464195854503767,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 686099572,
+      "step": 3995
+    },
+    {
+      "epoch": 45.388101983002834,
+      "loss": 0.1888955980539322,
+      "loss_ce": 0.02434481680393219,
+      "loss_iou": 0.0,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 686099572,
+      "step": 3995
+    },
+    {
+      "epoch": 45.39943342776204,
+      "grad_norm": 4.83817901538267,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 686269796,
+      "step": 3996
+    },
+    {
+      "epoch": 45.39943342776204,
+      "loss": 0.24119064211845398,
+      "loss_ce": 0.041269995272159576,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 686269796,
+      "step": 3996
+    },
+    {
+      "epoch": 45.41076487252125,
+      "grad_norm": 7.8935377101564494,
+      "learning_rate": 5e-06,
+      "loss": 0.1298,
+      "num_input_tokens_seen": 686438212,
+      "step": 3997
+    },
+    {
+      "epoch": 45.41076487252125,
+      "loss": 0.16225041449069977,
+      "loss_ce": 0.031085869297385216,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 686438212,
+      "step": 3997
+    },
+    {
+      "epoch": 45.422096317280456,
+      "grad_norm": 3.405448797930211,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 686610268,
+      "step": 3998
+    },
+    {
+      "epoch": 45.422096317280456,
+      "loss": 0.18029937148094177,
+      "loss_ce": 0.02063140645623207,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 686610268,
+      "step": 3998
+    },
+    {
+      "epoch": 45.43342776203966,
+      "grad_norm": 7.301927382845081,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 686781284,
+      "step": 3999
+    },
+    {
+      "epoch": 45.43342776203966,
+      "loss": 0.09374446421861649,
+      "loss_ce": 0.011224937625229359,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 686781284,
+      "step": 3999
+    },
+    {
+      "epoch": 45.444759206798864,
+      "grad_norm": 4.6130987466132085,
+      "learning_rate": 5e-06,
+      "loss": 0.1381,
+      "num_input_tokens_seen": 686953220,
+      "step": 4000
+    },
+    {
+      "epoch": 45.444759206798864,
+      "eval_seeclick_CIoU": 0.5753580033779144,
+      "eval_seeclick_GIoU": 0.5722742974758148,
+      "eval_seeclick_IoU": 0.6068568825721741,
+      "eval_seeclick_MAE_all": 0.057229211553931236,
+      "eval_seeclick_MAE_h": 0.03215831704437733,
+      "eval_seeclick_MAE_w": 0.07634955644607544,
+      "eval_seeclick_MAE_x": 0.08851699158549309,
+      "eval_seeclick_MAE_y": 0.03189198113977909,
+      "eval_seeclick_NUM_probability": 0.9997608363628387,
+      "eval_seeclick_inside_bbox": 0.9375,
+      "eval_seeclick_loss": 0.5520585775375366,
+      "eval_seeclick_loss_ce": 0.3357124775648117,
+      "eval_seeclick_loss_iou": 0.506103515625,
+      "eval_seeclick_loss_num": 0.042758941650390625,
+      "eval_seeclick_loss_xval": 0.2137908935546875,
+      "eval_seeclick_runtime": 70.0915,
+      "eval_seeclick_samples_per_second": 0.613,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 686953220,
+      "step": 4000
+    },
+    {
+      "epoch": 45.444759206798864,
+      "eval_icons_CIoU": 0.786271333694458,
+      "eval_icons_GIoU": 0.7868271470069885,
+      "eval_icons_IoU": 0.7971721887588501,
+      "eval_icons_MAE_all": 0.030403072014451027,
+      "eval_icons_MAE_h": 0.028422322124242783,
+      "eval_icons_MAE_w": 0.03643226437270641,
+      "eval_icons_MAE_x": 0.0266373991034925,
+      "eval_icons_MAE_y": 0.030120306648314,
+      "eval_icons_NUM_probability": 0.9990636110305786,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.1066146120429039,
+      "eval_icons_loss_ce": 0.00027707025583367795,
+      "eval_icons_loss_iou": 0.56201171875,
+      "eval_icons_loss_num": 0.019847869873046875,
+      "eval_icons_loss_xval": 0.0992279052734375,
+      "eval_icons_runtime": 79.7349,
+      "eval_icons_samples_per_second": 0.627,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 686953220,
+      "step": 4000
+    },
+    {
+      "epoch": 45.444759206798864,
+      "eval_screenspot_CIoU": 0.6420889695485433,
+      "eval_screenspot_GIoU": 0.6387250622113546,
+      "eval_screenspot_IoU": 0.6685307423273722,
+      "eval_screenspot_MAE_all": 0.0636286586523056,
+      "eval_screenspot_MAE_h": 0.03605052394171556,
+      "eval_screenspot_MAE_w": 0.11177605638901393,
+      "eval_screenspot_MAE_x": 0.0704745426774025,
+      "eval_screenspot_MAE_y": 0.03621351780990759,
+      "eval_screenspot_NUM_probability": 0.9997280637423197,
+      "eval_screenspot_inside_bbox": 0.8974999984105428,
+      "eval_screenspot_loss": 0.27599117159843445,
+      "eval_screenspot_loss_ce": 0.012172514184688529,
+      "eval_screenspot_loss_iou": 0.46875,
+      "eval_screenspot_loss_num": 0.05130767822265625,
+      "eval_screenspot_loss_xval": 0.2565511067708333,
+      "eval_screenspot_runtime": 137.9208,
+      "eval_screenspot_samples_per_second": 0.645,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 686953220,
+      "step": 4000
+    },
+    {
+      "epoch": 45.444759206798864,
+      "eval_compot_CIoU": 0.8052595555782318,
+      "eval_compot_GIoU": 0.7997151017189026,
+      "eval_compot_IoU": 0.8200662136077881,
+      "eval_compot_MAE_all": 0.034053971990942955,
+      "eval_compot_MAE_h": 0.029667741619050503,
+      "eval_compot_MAE_w": 0.04165242798626423,
+      "eval_compot_MAE_x": 0.03711581323295832,
+      "eval_compot_MAE_y": 0.027779914438724518,
+      "eval_compot_NUM_probability": 0.9999364614486694,
+      "eval_compot_inside_bbox": 0.9131944477558136,
+      "eval_compot_loss": 0.11417640745639801,
+      "eval_compot_loss_ce": 0.00021567247313214466,
+      "eval_compot_loss_iou": 0.52294921875,
+      "eval_compot_loss_num": 0.020069122314453125,
+      "eval_compot_loss_xval": 0.10040283203125,
+      "eval_compot_runtime": 88.5259,
+      "eval_compot_samples_per_second": 0.565,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 686953220,
+      "step": 4000
+    },
+    {
+      "epoch": 45.444759206798864,
+      "eval_custom_ui_MAE_all": 0.028795880265533924,
+      "eval_custom_ui_MAE_x": 0.04842784069478512,
+      "eval_custom_ui_MAE_y": 0.009163921000435948,
+      "eval_custom_ui_NUM_probability": 0.9999248087406158,
+      "eval_custom_ui_loss": 0.22209425270557404,
+      "eval_custom_ui_loss_ce": 0.08117441087961197,
+      "eval_custom_ui_loss_num": 0.028009414672851562,
+      "eval_custom_ui_loss_xval": 0.1399688720703125,
+      "eval_custom_ui_runtime": 59.6647,
+      "eval_custom_ui_samples_per_second": 0.838,
+      "eval_custom_ui_steps_per_second": 0.034,
+      "num_input_tokens_seen": 686953220,
+      "step": 4000
+    },
+    {
+      "epoch": 45.444759206798864,
+      "loss": 0.270561158657074,
+      "loss_ce": 0.09734340012073517,
+      "loss_iou": 0.0,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 686953220,
+      "step": 4000
+    },
+    {
+      "epoch": 45.45609065155807,
+      "grad_norm": 9.31204032917625,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 687123868,
+      "step": 4001
+    },
+    {
+      "epoch": 45.45609065155807,
+      "loss": 0.09752388298511505,
+      "loss_ce": 0.011189653538167477,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 687123868,
+      "step": 4001
+    },
+    {
+      "epoch": 45.46742209631728,
+      "grad_norm": 12.269168177350808,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 687296036,
+      "step": 4002
+    },
+    {
+      "epoch": 45.46742209631728,
+      "loss": 0.11949057877063751,
+      "loss_ce": 0.009169531986117363,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 687296036,
+      "step": 4002
+    },
+    {
+      "epoch": 45.478753541076486,
+      "grad_norm": 4.147795239221182,
+      "learning_rate": 5e-06,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 687466952,
+      "step": 4003
+    },
+    {
+      "epoch": 45.478753541076486,
+      "loss": 0.1856471300125122,
+      "loss_ce": 0.008401032537221909,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 687466952,
+      "step": 4003
+    },
+    {
+      "epoch": 45.49008498583569,
+      "grad_norm": 14.007785582328102,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 687638980,
+      "step": 4004
+    },
+    {
+      "epoch": 45.49008498583569,
+      "loss": 0.09351881593465805,
+      "loss_ce": 0.013349141925573349,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 687638980,
+      "step": 4004
+    },
+    {
+      "epoch": 45.5014164305949,
+      "grad_norm": 6.070229664311195,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 687809768,
+      "step": 4005
+    },
+    {
+      "epoch": 45.5014164305949,
+      "loss": 0.09495474398136139,
+      "loss_ce": 0.0030357972718775272,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 687809768,
+      "step": 4005
+    },
+    {
+      "epoch": 45.51274787535411,
+      "grad_norm": 11.16057294591649,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 687981112,
+      "step": 4006
+    },
+    {
+      "epoch": 45.51274787535411,
+      "loss": 0.1584462821483612,
+      "loss_ce": 0.007689447142183781,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 687981112,
+      "step": 4006
+    },
+    {
+      "epoch": 45.524079320113316,
+      "grad_norm": 11.814343170315833,
+      "learning_rate": 5e-06,
+      "loss": 0.1808,
+      "num_input_tokens_seen": 688152364,
+      "step": 4007
+    },
+    {
+      "epoch": 45.524079320113316,
+      "loss": 0.2301245927810669,
+      "loss_ce": 0.008902676403522491,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 688152364,
+      "step": 4007
+    },
+    {
+      "epoch": 45.53541076487252,
+      "grad_norm": 4.871684150103701,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 688320480,
+      "step": 4008
+    },
+    {
+      "epoch": 45.53541076487252,
+      "loss": 0.17684242129325867,
+      "loss_ce": 0.012505274266004562,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 688320480,
+      "step": 4008
+    },
+    {
+      "epoch": 45.54674220963173,
+      "grad_norm": 8.719209365182106,
+      "learning_rate": 5e-06,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 688490572,
+      "step": 4009
+    },
+    {
+      "epoch": 45.54674220963173,
+      "loss": 0.09991415590047836,
+      "loss_ce": 0.011916719377040863,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 688490572,
+      "step": 4009
+    },
+    {
+      "epoch": 45.55807365439094,
+      "grad_norm": 9.053125357384218,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 688662528,
+      "step": 4010
+    },
+    {
+      "epoch": 45.55807365439094,
+      "loss": 0.09128035604953766,
+      "loss_ce": 0.008638758212327957,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 688662528,
+      "step": 4010
+    },
+    {
+      "epoch": 45.56940509915014,
+      "grad_norm": 8.352866762475179,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 688834664,
+      "step": 4011
+    },
+    {
+      "epoch": 45.56940509915014,
+      "loss": 0.13500797748565674,
+      "loss_ce": 0.027006270363926888,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 688834664,
+      "step": 4011
+    },
+    {
+      "epoch": 45.580736543909346,
+      "grad_norm": 5.040456032464371,
+      "learning_rate": 5e-06,
+      "loss": 0.1383,
+      "num_input_tokens_seen": 689006528,
+      "step": 4012
+    },
+    {
+      "epoch": 45.580736543909346,
+      "loss": 0.10428878664970398,
+      "loss_ce": 0.01514693908393383,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 689006528,
+      "step": 4012
+    },
+    {
+      "epoch": 45.59206798866855,
+      "grad_norm": 6.043612290767415,
+      "learning_rate": 5e-06,
+      "loss": 0.1865,
+      "num_input_tokens_seen": 689178116,
+      "step": 4013
+    },
+    {
+      "epoch": 45.59206798866855,
+      "loss": 0.3318237066268921,
+      "loss_ce": 0.018362397328019142,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 689178116,
+      "step": 4013
+    },
+    {
+      "epoch": 45.60339943342776,
+      "grad_norm": 6.376306942221769,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 689349720,
+      "step": 4014
+    },
+    {
+      "epoch": 45.60339943342776,
+      "loss": 0.16877371072769165,
+      "loss_ce": 0.018993424251675606,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 689349720,
+      "step": 4014
+    },
+    {
+      "epoch": 45.61473087818697,
+      "grad_norm": 5.246386012748272,
+      "learning_rate": 5e-06,
+      "loss": 0.1691,
+      "num_input_tokens_seen": 689520732,
+      "step": 4015
+    },
+    {
+      "epoch": 45.61473087818697,
+      "loss": 0.13252577185630798,
+      "loss_ce": 0.01536879874765873,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 689520732,
+      "step": 4015
+    },
+    {
+      "epoch": 45.626062322946176,
+      "grad_norm": 4.300499721567621,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 689691084,
+      "step": 4016
+    },
+    {
+      "epoch": 45.626062322946176,
+      "loss": 0.07665916532278061,
+      "loss_ce": 0.009581529535353184,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 689691084,
+      "step": 4016
+    },
+    {
+      "epoch": 45.63739376770538,
+      "grad_norm": 3.1191792672971483,
+      "learning_rate": 5e-06,
+      "loss": 0.1429,
+      "num_input_tokens_seen": 689862804,
+      "step": 4017
+    },
+    {
+      "epoch": 45.63739376770538,
+      "loss": 0.1220824122428894,
+      "loss_ce": 0.012005502358078957,
+      "loss_iou": 0.359375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 689862804,
+      "step": 4017
+    },
+    {
+      "epoch": 45.64872521246459,
+      "grad_norm": 4.719077406216629,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 690031992,
+      "step": 4018
+    },
+    {
+      "epoch": 45.64872521246459,
+      "loss": 0.10369390994310379,
+      "loss_ce": 0.02381414920091629,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 690031992,
+      "step": 4018
+    },
+    {
+      "epoch": 45.6600566572238,
+      "grad_norm": 8.181839908512726,
+      "learning_rate": 5e-06,
+      "loss": 0.1932,
+      "num_input_tokens_seen": 690203596,
+      "step": 4019
+    },
+    {
+      "epoch": 45.6600566572238,
+      "loss": 0.21765275299549103,
+      "loss_ce": 0.03378434106707573,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 690203596,
+      "step": 4019
+    },
+    {
+      "epoch": 45.671388101983005,
+      "grad_norm": 3.9158688906263914,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 690375272,
+      "step": 4020
+    },
+    {
+      "epoch": 45.671388101983005,
+      "loss": 0.12163268029689789,
+      "loss_ce": 0.019978631287813187,
+      "loss_iou": 0.625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 690375272,
+      "step": 4020
+    },
+    {
+      "epoch": 45.68271954674221,
+      "grad_norm": 7.975997913736339,
+      "learning_rate": 5e-06,
+      "loss": 0.1642,
+      "num_input_tokens_seen": 690546200,
+      "step": 4021
+    },
+    {
+      "epoch": 45.68271954674221,
+      "loss": 0.14920452237129211,
+      "loss_ce": 0.011081965640187263,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 690546200,
+      "step": 4021
+    },
+    {
+      "epoch": 45.69405099150141,
+      "grad_norm": 4.908049695166497,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 690716648,
+      "step": 4022
+    },
+    {
+      "epoch": 45.69405099150141,
+      "loss": 0.09378327429294586,
+      "loss_ce": 0.0029629673808813095,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 690716648,
+      "step": 4022
+    },
+    {
+      "epoch": 45.70538243626062,
+      "grad_norm": 10.419466484524843,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 690888552,
+      "step": 4023
+    },
+    {
+      "epoch": 45.70538243626062,
+      "loss": 0.07819560170173645,
+      "loss_ce": 0.015359912998974323,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 690888552,
+      "step": 4023
+    },
+    {
+      "epoch": 45.71671388101983,
+      "grad_norm": 8.4351898440096,
+      "learning_rate": 5e-06,
+      "loss": 0.1526,
+      "num_input_tokens_seen": 691060204,
+      "step": 4024
+    },
+    {
+      "epoch": 45.71671388101983,
+      "loss": 0.09885814040899277,
+      "loss_ce": 0.024441026151180267,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 691060204,
+      "step": 4024
+    },
+    {
+      "epoch": 45.728045325779036,
+      "grad_norm": 4.815286880513129,
+      "learning_rate": 5e-06,
+      "loss": 0.1802,
+      "num_input_tokens_seen": 691232156,
+      "step": 4025
+    },
+    {
+      "epoch": 45.728045325779036,
+      "loss": 0.10942128300666809,
+      "loss_ce": 0.02589466981589794,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 691232156,
+      "step": 4025
+    },
+    {
+      "epoch": 45.73937677053824,
+      "grad_norm": 7.476140142632498,
+      "learning_rate": 5e-06,
+      "loss": 0.17,
+      "num_input_tokens_seen": 691403528,
+      "step": 4026
+    },
+    {
+      "epoch": 45.73937677053824,
+      "loss": 0.25937730073928833,
+      "loss_ce": 0.03720935434103012,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 691403528,
+      "step": 4026
+    },
+    {
+      "epoch": 45.75070821529745,
+      "grad_norm": 7.256598943671983,
+      "learning_rate": 5e-06,
+      "loss": 0.1565,
+      "num_input_tokens_seen": 691575664,
+      "step": 4027
+    },
+    {
+      "epoch": 45.75070821529745,
+      "loss": 0.12446477264165878,
+      "loss_ce": 0.02204778790473938,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 691575664,
+      "step": 4027
+    },
+    {
+      "epoch": 45.76203966005666,
+      "grad_norm": 7.709819614324347,
+      "learning_rate": 5e-06,
+      "loss": 0.1923,
+      "num_input_tokens_seen": 691747364,
+      "step": 4028
+    },
+    {
+      "epoch": 45.76203966005666,
+      "loss": 0.15157550573349,
+      "loss_ce": 0.007502014748752117,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 691747364,
+      "step": 4028
+    },
+    {
+      "epoch": 45.773371104815865,
+      "grad_norm": 6.228833336730881,
+      "learning_rate": 5e-06,
+      "loss": 0.1534,
+      "num_input_tokens_seen": 691918140,
+      "step": 4029
+    },
+    {
+      "epoch": 45.773371104815865,
+      "loss": 0.16909784078598022,
+      "loss_ce": 0.01889032870531082,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 691918140,
+      "step": 4029
+    },
+    {
+      "epoch": 45.78470254957507,
+      "grad_norm": 4.262229610498088,
+      "learning_rate": 5e-06,
+      "loss": 0.1843,
+      "num_input_tokens_seen": 692088244,
+      "step": 4030
+    },
+    {
+      "epoch": 45.78470254957507,
+      "loss": 0.15146948397159576,
+      "loss_ce": 0.023326169699430466,
+      "loss_iou": 0.28125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 692088244,
+      "step": 4030
+    },
+    {
+      "epoch": 45.79603399433428,
+      "grad_norm": 5.439921049211111,
+      "learning_rate": 5e-06,
+      "loss": 0.1693,
+      "num_input_tokens_seen": 692260012,
+      "step": 4031
+    },
+    {
+      "epoch": 45.79603399433428,
+      "loss": 0.1990601271390915,
+      "loss_ce": 0.009759585373103619,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 692260012,
+      "step": 4031
+    },
+    {
+      "epoch": 45.80736543909349,
+      "grad_norm": 21.720364474416225,
+      "learning_rate": 5e-06,
+      "loss": 0.1778,
+      "num_input_tokens_seen": 692430088,
+      "step": 4032
+    },
+    {
+      "epoch": 45.80736543909349,
+      "loss": 0.17105695605278015,
+      "loss_ce": 0.017736639827489853,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 692430088,
+      "step": 4032
+    },
+    {
+      "epoch": 45.81869688385269,
+      "grad_norm": 4.291526005404919,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 692601696,
+      "step": 4033
+    },
+    {
+      "epoch": 45.81869688385269,
+      "loss": 0.10573385655879974,
+      "loss_ce": 0.019811607897281647,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 692601696,
+      "step": 4033
+    },
+    {
+      "epoch": 45.830028328611895,
+      "grad_norm": 14.738662069142217,
+      "learning_rate": 5e-06,
+      "loss": 0.1612,
+      "num_input_tokens_seen": 692771796,
+      "step": 4034
+    },
+    {
+      "epoch": 45.830028328611895,
+      "loss": 0.1369961053133011,
+      "loss_ce": 0.010378671810030937,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 692771796,
+      "step": 4034
+    },
+    {
+      "epoch": 45.8413597733711,
+      "grad_norm": 5.272358834271055,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 692942128,
+      "step": 4035
+    },
+    {
+      "epoch": 45.8413597733711,
+      "loss": 0.1546439379453659,
+      "loss_ce": 0.011699603870511055,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 692942128,
+      "step": 4035
+    },
+    {
+      "epoch": 45.85269121813031,
+      "grad_norm": 5.0499681071489775,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 693113752,
+      "step": 4036
+    },
+    {
+      "epoch": 45.85269121813031,
+      "loss": 0.10734754800796509,
+      "loss_ce": 0.003526749787852168,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 693113752,
+      "step": 4036
+    },
+    {
+      "epoch": 45.86402266288952,
+      "grad_norm": 12.33536913658161,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 693285384,
+      "step": 4037
+    },
+    {
+      "epoch": 45.86402266288952,
+      "loss": 0.15130847692489624,
+      "loss_ce": 0.01770252361893654,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 693285384,
+      "step": 4037
+    },
+    {
+      "epoch": 45.875354107648725,
+      "grad_norm": 5.575263259142814,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 693456296,
+      "step": 4038
+    },
+    {
+      "epoch": 45.875354107648725,
+      "loss": 0.13451847434043884,
+      "loss_ce": 0.019650310277938843,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 693456296,
+      "step": 4038
+    },
+    {
+      "epoch": 45.88668555240793,
+      "grad_norm": 4.394589401981415,
+      "learning_rate": 5e-06,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 693628360,
+      "step": 4039
+    },
+    {
+      "epoch": 45.88668555240793,
+      "loss": 0.16762179136276245,
+      "loss_ce": 0.016193576157093048,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 693628360,
+      "step": 4039
+    },
+    {
+      "epoch": 45.89801699716714,
+      "grad_norm": 14.174316130694905,
+      "learning_rate": 5e-06,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 693798768,
+      "step": 4040
+    },
+    {
+      "epoch": 45.89801699716714,
+      "loss": 0.0929897278547287,
+      "loss_ce": 0.0105922631919384,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 693798768,
+      "step": 4040
+    },
+    {
+      "epoch": 45.90934844192635,
+      "grad_norm": 4.384574378142213,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 693970572,
+      "step": 4041
+    },
+    {
+      "epoch": 45.90934844192635,
+      "loss": 0.16414324939250946,
+      "loss_ce": 0.00490253372117877,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 693970572,
+      "step": 4041
+    },
+    {
+      "epoch": 45.920679886685555,
+      "grad_norm": 7.035182863820019,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 694142172,
+      "step": 4042
+    },
+    {
+      "epoch": 45.920679886685555,
+      "loss": 0.1220024824142456,
+      "loss_ce": 0.014870522543787956,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 694142172,
+      "step": 4042
+    },
+    {
+      "epoch": 45.93201133144476,
+      "grad_norm": 4.525227088226603,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 694314212,
+      "step": 4043
+    },
+    {
+      "epoch": 45.93201133144476,
+      "loss": 0.0821773111820221,
+      "loss_ce": 0.0008479647221975029,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 694314212,
+      "step": 4043
+    },
+    {
+      "epoch": 45.94334277620396,
+      "grad_norm": 4.44986101318948,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 694483232,
+      "step": 4044
+    },
+    {
+      "epoch": 45.94334277620396,
+      "loss": 0.17371788620948792,
+      "loss_ce": 0.03776206821203232,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 694483232,
+      "step": 4044
+    },
+    {
+      "epoch": 45.95467422096317,
+      "grad_norm": 3.525747481949752,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 694655076,
+      "step": 4045
+    },
+    {
+      "epoch": 45.95467422096317,
+      "loss": 0.15462997555732727,
+      "loss_ce": 0.003934177104383707,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 694655076,
+      "step": 4045
+    },
+    {
+      "epoch": 45.96600566572238,
+      "grad_norm": 5.503909865333753,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 694826128,
+      "step": 4046
+    },
+    {
+      "epoch": 45.96600566572238,
+      "loss": 0.20056535303592682,
+      "loss_ce": 0.0060157934203743935,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 694826128,
+      "step": 4046
+    },
+    {
+      "epoch": 45.977337110481585,
+      "grad_norm": 6.673464510430702,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 694997956,
+      "step": 4047
+    },
+    {
+      "epoch": 45.977337110481585,
+      "loss": 0.08946847915649414,
+      "loss_ce": 0.019201762974262238,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 694997956,
+      "step": 4047
+    },
+    {
+      "epoch": 45.98866855524079,
+      "grad_norm": 4.230827994984814,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 695169776,
+      "step": 4048
+    },
+    {
+      "epoch": 45.98866855524079,
+      "loss": 0.11835239827632904,
+      "loss_ce": 0.01721714250743389,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 695169776,
+      "step": 4048
+    },
+    {
+      "epoch": 46.0,
+      "grad_norm": 37.442339599008115,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 695340616,
+      "step": 4049
+    },
+    {
+      "epoch": 46.0,
+      "loss": 0.20682911574840546,
+      "loss_ce": 0.010768943466246128,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 695340616,
+      "step": 4049
+    },
+    {
+      "epoch": 46.01133144475921,
+      "grad_norm": 8.029671612997193,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 695512220,
+      "step": 4050
+    },
+    {
+      "epoch": 46.01133144475921,
+      "loss": 0.19538497924804688,
+      "loss_ce": 0.016262037679553032,
+      "loss_iou": 0.0,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 695512220,
+      "step": 4050
+    },
+    {
+      "epoch": 46.022662889518415,
+      "grad_norm": 3.9366780952861826,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 695682644,
+      "step": 4051
+    },
+    {
+      "epoch": 46.022662889518415,
+      "loss": 0.10989221930503845,
+      "loss_ce": 0.015913337469100952,
+      "loss_iou": 0.421875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 695682644,
+      "step": 4051
+    },
+    {
+      "epoch": 46.03399433427762,
+      "grad_norm": 3.909104107158833,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 695853972,
+      "step": 4052
+    },
+    {
+      "epoch": 46.03399433427762,
+      "loss": 0.10027308762073517,
+      "loss_ce": 0.014610247686505318,
+      "loss_iou": 0.0,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 695853972,
+      "step": 4052
+    },
+    {
+      "epoch": 46.04532577903683,
+      "grad_norm": 4.78749313718534,
+      "learning_rate": 5e-06,
+      "loss": 0.1211,
+      "num_input_tokens_seen": 696023636,
+      "step": 4053
+    },
+    {
+      "epoch": 46.04532577903683,
+      "loss": 0.10963616520166397,
+      "loss_ce": 0.002824646420776844,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 696023636,
+      "step": 4053
+    },
+    {
+      "epoch": 46.05665722379604,
+      "grad_norm": 8.140274621598023,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 696195536,
+      "step": 4054
+    },
+    {
+      "epoch": 46.05665722379604,
+      "loss": 0.1319035291671753,
+      "loss_ce": 0.005499732214957476,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 696195536,
+      "step": 4054
+    },
+    {
+      "epoch": 46.06798866855524,
+      "grad_norm": 9.317605070566657,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 696365748,
+      "step": 4055
+    },
+    {
+      "epoch": 46.06798866855524,
+      "loss": 0.06587816029787064,
+      "loss_ce": 0.006170519627630711,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 696365748,
+      "step": 4055
+    },
+    {
+      "epoch": 46.079320113314445,
+      "grad_norm": 5.99944704025556,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 696536612,
+      "step": 4056
+    },
+    {
+      "epoch": 46.079320113314445,
+      "loss": 0.09519752860069275,
+      "loss_ce": 0.004468770697712898,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 696536612,
+      "step": 4056
+    },
+    {
+      "epoch": 46.09065155807365,
+      "grad_norm": 6.519834938236697,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 696708272,
+      "step": 4057
+    },
+    {
+      "epoch": 46.09065155807365,
+      "loss": 0.07906650751829147,
+      "loss_ce": 0.012721296399831772,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 696708272,
+      "step": 4057
+    },
+    {
+      "epoch": 46.10198300283286,
+      "grad_norm": 4.327929889530436,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 696879424,
+      "step": 4058
+    },
+    {
+      "epoch": 46.10198300283286,
+      "loss": 0.17151013016700745,
+      "loss_ce": 0.004304311703890562,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 696879424,
+      "step": 4058
+    },
+    {
+      "epoch": 46.11331444759207,
+      "grad_norm": 3.6334591029548866,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 697051788,
+      "step": 4059
+    },
+    {
+      "epoch": 46.11331444759207,
+      "loss": 0.1637030690908432,
+      "loss_ce": 0.010596380569040775,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 697051788,
+      "step": 4059
+    },
+    {
+      "epoch": 46.124645892351275,
+      "grad_norm": 4.770888159027178,
+      "learning_rate": 5e-06,
+      "loss": 0.1351,
+      "num_input_tokens_seen": 697221036,
+      "step": 4060
+    },
+    {
+      "epoch": 46.124645892351275,
+      "loss": 0.11457142978906631,
+      "loss_ce": 0.006813860032707453,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 697221036,
+      "step": 4060
+    },
+    {
+      "epoch": 46.13597733711048,
+      "grad_norm": 4.453661251315172,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 697392960,
+      "step": 4061
+    },
+    {
+      "epoch": 46.13597733711048,
+      "loss": 0.10582560300827026,
+      "loss_ce": 0.011312661692500114,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 697392960,
+      "step": 4061
+    },
+    {
+      "epoch": 46.14730878186969,
+      "grad_norm": 4.738120467915284,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 697564976,
+      "step": 4062
+    },
+    {
+      "epoch": 46.14730878186969,
+      "loss": 0.1082979291677475,
+      "loss_ce": 0.005758873652666807,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 697564976,
+      "step": 4062
+    },
+    {
+      "epoch": 46.1586402266289,
+      "grad_norm": 4.674716134505156,
+      "learning_rate": 5e-06,
+      "loss": 0.1854,
+      "num_input_tokens_seen": 697735944,
+      "step": 4063
+    },
+    {
+      "epoch": 46.1586402266289,
+      "loss": 0.19423949718475342,
+      "loss_ce": 0.005869742948561907,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 697735944,
+      "step": 4063
+    },
+    {
+      "epoch": 46.169971671388105,
+      "grad_norm": 10.410889751536967,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 697907684,
+      "step": 4064
+    },
+    {
+      "epoch": 46.169971671388105,
+      "loss": 0.08565324544906616,
+      "loss_ce": 0.01128191128373146,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 697907684,
+      "step": 4064
+    },
+    {
+      "epoch": 46.18130311614731,
+      "grad_norm": 4.3081013184249555,
+      "learning_rate": 5e-06,
+      "loss": 0.1726,
+      "num_input_tokens_seen": 698076996,
+      "step": 4065
+    },
+    {
+      "epoch": 46.18130311614731,
+      "loss": 0.1471741795539856,
+      "loss_ce": 0.046038925647735596,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 698076996,
+      "step": 4065
+    },
+    {
+      "epoch": 46.19263456090651,
+      "grad_norm": 4.47030046163472,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 698249656,
+      "step": 4066
+    },
+    {
+      "epoch": 46.19263456090651,
+      "loss": 0.12511605024337769,
+      "loss_ce": 0.004785236902534962,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 698249656,
+      "step": 4066
+    },
+    {
+      "epoch": 46.20396600566572,
+      "grad_norm": 6.0205544317267865,
+      "learning_rate": 5e-06,
+      "loss": 0.1596,
+      "num_input_tokens_seen": 698421400,
+      "step": 4067
+    },
+    {
+      "epoch": 46.20396600566572,
+      "loss": 0.10036885738372803,
+      "loss_ce": 0.004757285583764315,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 698421400,
+      "step": 4067
+    },
+    {
+      "epoch": 46.21529745042493,
+      "grad_norm": 3.8501652583748864,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 698593236,
+      "step": 4068
+    },
+    {
+      "epoch": 46.21529745042493,
+      "loss": 0.2151014804840088,
+      "loss_ce": 0.02143692597746849,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 698593236,
+      "step": 4068
+    },
+    {
+      "epoch": 46.226628895184135,
+      "grad_norm": 6.152689104657704,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 698764976,
+      "step": 4069
+    },
+    {
+      "epoch": 46.226628895184135,
+      "loss": 0.18791410326957703,
+      "loss_ce": 0.01158355362713337,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 698764976,
+      "step": 4069
+    },
+    {
+      "epoch": 46.23796033994334,
+      "grad_norm": 3.8575994737298793,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 698936224,
+      "step": 4070
+    },
+    {
+      "epoch": 46.23796033994334,
+      "loss": 0.14665040373802185,
+      "loss_ce": 0.007825931534171104,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 698936224,
+      "step": 4070
+    },
+    {
+      "epoch": 46.24929178470255,
+      "grad_norm": 5.036340117789579,
+      "learning_rate": 5e-06,
+      "loss": 0.159,
+      "num_input_tokens_seen": 699107260,
+      "step": 4071
+    },
+    {
+      "epoch": 46.24929178470255,
+      "loss": 0.2108308970928192,
+      "loss_ce": 0.006180016789585352,
+      "loss_iou": 0.59375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 699107260,
+      "step": 4071
+    },
+    {
+      "epoch": 46.26062322946176,
+      "grad_norm": 5.476714873348524,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 699279088,
+      "step": 4072
+    },
+    {
+      "epoch": 46.26062322946176,
+      "loss": 0.10727986693382263,
+      "loss_ce": 0.008814901113510132,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 699279088,
+      "step": 4072
+    },
+    {
+      "epoch": 46.271954674220964,
+      "grad_norm": 4.428868749580724,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 699449704,
+      "step": 4073
+    },
+    {
+      "epoch": 46.271954674220964,
+      "loss": 0.19007912278175354,
+      "loss_ce": 0.06129493936896324,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 699449704,
+      "step": 4073
+    },
+    {
+      "epoch": 46.28328611898017,
+      "grad_norm": 4.876000199126908,
+      "learning_rate": 5e-06,
+      "loss": 0.1278,
+      "num_input_tokens_seen": 699621788,
+      "step": 4074
+    },
+    {
+      "epoch": 46.28328611898017,
+      "loss": 0.08124564588069916,
+      "loss_ce": 0.014534223824739456,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 699621788,
+      "step": 4074
+    },
+    {
+      "epoch": 46.29461756373938,
+      "grad_norm": 7.153241046723841,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 699792328,
+      "step": 4075
+    },
+    {
+      "epoch": 46.29461756373938,
+      "loss": 0.14683113992214203,
+      "loss_ce": 0.01880989968776703,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 699792328,
+      "step": 4075
+    },
+    {
+      "epoch": 46.30594900849859,
+      "grad_norm": 5.106710145431983,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 699964152,
+      "step": 4076
+    },
+    {
+      "epoch": 46.30594900849859,
+      "loss": 0.08658863604068756,
+      "loss_ce": 0.008799327537417412,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 699964152,
+      "step": 4076
+    },
+    {
+      "epoch": 46.31728045325779,
+      "grad_norm": 3.9861398239406935,
+      "learning_rate": 5e-06,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 700136144,
+      "step": 4077
+    },
+    {
+      "epoch": 46.31728045325779,
+      "loss": 0.06945373117923737,
+      "loss_ce": 0.0023760886397212744,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 700136144,
+      "step": 4077
+    },
+    {
+      "epoch": 46.328611898016995,
+      "grad_norm": 5.222039549340686,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 700306940,
+      "step": 4078
+    },
+    {
+      "epoch": 46.328611898016995,
+      "loss": 0.10762211680412292,
+      "loss_ce": 0.009080860763788223,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 700306940,
+      "step": 4078
+    },
+    {
+      "epoch": 46.3399433427762,
+      "grad_norm": 10.256451617900025,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 700477872,
+      "step": 4079
+    },
+    {
+      "epoch": 46.3399433427762,
+      "loss": 0.10224519670009613,
+      "loss_ce": 0.0004232922801747918,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 700477872,
+      "step": 4079
+    },
+    {
+      "epoch": 46.35127478753541,
+      "grad_norm": 3.8981066547072576,
+      "learning_rate": 5e-06,
+      "loss": 0.113,
+      "num_input_tokens_seen": 700649372,
+      "step": 4080
+    },
+    {
+      "epoch": 46.35127478753541,
+      "loss": 0.1448293924331665,
+      "loss_ce": 0.022667525336146355,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 700649372,
+      "step": 4080
+    },
+    {
+      "epoch": 46.36260623229462,
+      "grad_norm": 5.488542117464678,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 700819904,
+      "step": 4081
+    },
+    {
+      "epoch": 46.36260623229462,
+      "loss": 0.08121670037508011,
+      "loss_ce": 0.007760888896882534,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 700819904,
+      "step": 4081
+    },
+    {
+      "epoch": 46.373937677053824,
+      "grad_norm": 5.459321748202955,
+      "learning_rate": 5e-06,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 700990904,
+      "step": 4082
+    },
+    {
+      "epoch": 46.373937677053824,
+      "loss": 0.1338321566581726,
+      "loss_ce": 0.018475698307156563,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 700990904,
+      "step": 4082
+    },
+    {
+      "epoch": 46.38526912181303,
+      "grad_norm": 4.375294198108324,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 701162804,
+      "step": 4083
+    },
+    {
+      "epoch": 46.38526912181303,
+      "loss": 0.08643153309822083,
+      "loss_ce": 0.003301647724583745,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 701162804,
+      "step": 4083
+    },
+    {
+      "epoch": 46.39660056657224,
+      "grad_norm": 3.7627644767980946,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 701334956,
+      "step": 4084
+    },
+    {
+      "epoch": 46.39660056657224,
+      "loss": 0.08527108281850815,
+      "loss_ce": 0.008702482096850872,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 701334956,
+      "step": 4084
+    },
+    {
+      "epoch": 46.40793201133145,
+      "grad_norm": 10.395005249273133,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 701505084,
+      "step": 4085
+    },
+    {
+      "epoch": 46.40793201133145,
+      "loss": 0.15912649035453796,
+      "loss_ce": 0.016670452430844307,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 701505084,
+      "step": 4085
+    },
+    {
+      "epoch": 46.419263456090654,
+      "grad_norm": 4.571563993193018,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 701677328,
+      "step": 4086
+    },
+    {
+      "epoch": 46.419263456090654,
+      "loss": 0.07297007739543915,
+      "loss_ce": 0.015505479648709297,
+      "loss_iou": 0.546875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 701677328,
+      "step": 4086
+    },
+    {
+      "epoch": 46.43059490084986,
+      "grad_norm": 6.295797853748418,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 701848872,
+      "step": 4087
+    },
+    {
+      "epoch": 46.43059490084986,
+      "loss": 0.14068666100502014,
+      "loss_ce": 0.026596687734127045,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 701848872,
+      "step": 4087
+    },
+    {
+      "epoch": 46.44192634560906,
+      "grad_norm": 3.410184659695416,
+      "learning_rate": 5e-06,
+      "loss": 0.1049,
+      "num_input_tokens_seen": 702021120,
+      "step": 4088
+    },
+    {
+      "epoch": 46.44192634560906,
+      "loss": 0.10812586545944214,
+      "loss_ce": 0.0036947091575711966,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 702021120,
+      "step": 4088
+    },
+    {
+      "epoch": 46.45325779036827,
+      "grad_norm": 5.255826511941344,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 702193012,
+      "step": 4089
+    },
+    {
+      "epoch": 46.45325779036827,
+      "loss": 0.11156702041625977,
+      "loss_ce": 0.010767453350126743,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 702193012,
+      "step": 4089
+    },
+    {
+      "epoch": 46.46458923512748,
+      "grad_norm": 5.6815189225046545,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 702365180,
+      "step": 4090
+    },
+    {
+      "epoch": 46.46458923512748,
+      "loss": 0.18132451176643372,
+      "loss_ce": 0.012318171560764313,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 702365180,
+      "step": 4090
+    },
+    {
+      "epoch": 46.475920679886684,
+      "grad_norm": 5.596616364565578,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 702536904,
+      "step": 4091
+    },
+    {
+      "epoch": 46.475920679886684,
+      "loss": 0.09863276034593582,
+      "loss_ce": 0.005737253464758396,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 702536904,
+      "step": 4091
+    },
+    {
+      "epoch": 46.48725212464589,
+      "grad_norm": 5.420601102713969,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 702709000,
+      "step": 4092
+    },
+    {
+      "epoch": 46.48725212464589,
+      "loss": 0.0952075719833374,
+      "loss_ce": 0.007103325333446264,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 702709000,
+      "step": 4092
+    },
+    {
+      "epoch": 46.4985835694051,
+      "grad_norm": 6.288228733485025,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 702880828,
+      "step": 4093
+    },
+    {
+      "epoch": 46.4985835694051,
+      "loss": 0.12521764636039734,
+      "loss_ce": 0.021213725209236145,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 702880828,
+      "step": 4093
+    },
+    {
+      "epoch": 46.50991501416431,
+      "grad_norm": 3.63778727264492,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 703052452,
+      "step": 4094
+    },
+    {
+      "epoch": 46.50991501416431,
+      "loss": 0.1453527957201004,
+      "loss_ce": 0.015439460054039955,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 703052452,
+      "step": 4094
+    },
+    {
+      "epoch": 46.521246458923514,
+      "grad_norm": 6.4617429890580595,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 703224348,
+      "step": 4095
+    },
+    {
+      "epoch": 46.521246458923514,
+      "loss": 0.10762103646993637,
+      "loss_ce": 0.01658710092306137,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 703224348,
+      "step": 4095
+    },
+    {
+      "epoch": 46.53257790368272,
+      "grad_norm": 11.668815896068267,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 703396032,
+      "step": 4096
+    },
+    {
+      "epoch": 46.53257790368272,
+      "loss": 0.11966902017593384,
+      "loss_ce": 0.0061436304822564125,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 703396032,
+      "step": 4096
+    },
+    {
+      "epoch": 46.54390934844193,
+      "grad_norm": 5.066378124433442,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 703567540,
+      "step": 4097
+    },
+    {
+      "epoch": 46.54390934844193,
+      "loss": 0.14190658926963806,
+      "loss_ce": 0.0061949086375534534,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 703567540,
+      "step": 4097
+    },
+    {
+      "epoch": 46.555240793201136,
+      "grad_norm": 7.279656555333125,
+      "learning_rate": 5e-06,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 703739108,
+      "step": 4098
+    },
+    {
+      "epoch": 46.555240793201136,
+      "loss": 0.23989951610565186,
+      "loss_ce": 0.015473244711756706,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 703739108,
+      "step": 4098
+    },
+    {
+      "epoch": 46.56657223796034,
+      "grad_norm": 3.967641706646963,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 703909940,
+      "step": 4099
+    },
+    {
+      "epoch": 46.56657223796034,
+      "loss": 0.12897495925426483,
+      "loss_ce": 0.007881210185587406,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 703909940,
+      "step": 4099
+    },
+    {
+      "epoch": 46.577903682719544,
+      "grad_norm": 6.193551355263466,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 704081276,
+      "step": 4100
+    },
+    {
+      "epoch": 46.577903682719544,
+      "loss": 0.09545089304447174,
+      "loss_ce": 0.00423385901376605,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 704081276,
+      "step": 4100
+    },
+    {
+      "epoch": 46.58923512747875,
+      "grad_norm": 5.394284180369022,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 704252848,
+      "step": 4101
+    },
+    {
+      "epoch": 46.58923512747875,
+      "loss": 0.11043091863393784,
+      "loss_ce": 0.005496219731867313,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 704252848,
+      "step": 4101
+    },
+    {
+      "epoch": 46.60056657223796,
+      "grad_norm": 4.0417234731558915,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 704422812,
+      "step": 4102
+    },
+    {
+      "epoch": 46.60056657223796,
+      "loss": 0.16895537078380585,
+      "loss_ce": 0.015024702996015549,
+      "loss_iou": 0.296875,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 704422812,
+      "step": 4102
+    },
+    {
+      "epoch": 46.611898016997166,
+      "grad_norm": 4.0477765416488145,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 704594832,
+      "step": 4103
+    },
+    {
+      "epoch": 46.611898016997166,
+      "loss": 0.07284734398126602,
+      "loss_ce": 0.011110285297036171,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 704594832,
+      "step": 4103
+    },
+    {
+      "epoch": 46.623229461756374,
+      "grad_norm": 4.507132621218288,
+      "learning_rate": 5e-06,
+      "loss": 0.1849,
+      "num_input_tokens_seen": 704765956,
+      "step": 4104
+    },
+    {
+      "epoch": 46.623229461756374,
+      "loss": 0.1683284044265747,
+      "loss_ce": 0.014763961546123028,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 704765956,
+      "step": 4104
+    },
+    {
+      "epoch": 46.63456090651558,
+      "grad_norm": 3.9266538127119053,
+      "learning_rate": 5e-06,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 704937096,
+      "step": 4105
+    },
+    {
+      "epoch": 46.63456090651558,
+      "loss": 0.16485007107257843,
+      "loss_ce": 0.02749045565724373,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 704937096,
+      "step": 4105
+    },
+    {
+      "epoch": 46.64589235127479,
+      "grad_norm": 4.362016053961835,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 705107796,
+      "step": 4106
+    },
+    {
+      "epoch": 46.64589235127479,
+      "loss": 0.09971150755882263,
+      "loss_ce": 0.01450643315911293,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 705107796,
+      "step": 4106
+    },
+    {
+      "epoch": 46.657223796033996,
+      "grad_norm": 6.681970585480897,
+      "learning_rate": 5e-06,
+      "loss": 0.145,
+      "num_input_tokens_seen": 705278624,
+      "step": 4107
+    },
+    {
+      "epoch": 46.657223796033996,
+      "loss": 0.1683436781167984,
+      "loss_ce": 0.004098069854080677,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 705278624,
+      "step": 4107
+    },
+    {
+      "epoch": 46.668555240793204,
+      "grad_norm": 8.049263162315144,
+      "learning_rate": 5e-06,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 705450604,
+      "step": 4108
+    },
+    {
+      "epoch": 46.668555240793204,
+      "loss": 0.14787597954273224,
+      "loss_ce": 0.0035125839058309793,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 705450604,
+      "step": 4108
+    },
+    {
+      "epoch": 46.67988668555241,
+      "grad_norm": 32.421335583148995,
+      "learning_rate": 5e-06,
+      "loss": 0.1359,
+      "num_input_tokens_seen": 705621788,
+      "step": 4109
+    },
+    {
+      "epoch": 46.67988668555241,
+      "loss": 0.1430978775024414,
+      "loss_ce": 0.01199435256421566,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 705621788,
+      "step": 4109
+    },
+    {
+      "epoch": 46.69121813031161,
+      "grad_norm": 5.701183868217729,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 705793576,
+      "step": 4110
+    },
+    {
+      "epoch": 46.69121813031161,
+      "loss": 0.1010262668132782,
+      "loss_ce": 0.005231581628322601,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 705793576,
+      "step": 4110
+    },
+    {
+      "epoch": 46.70254957507082,
+      "grad_norm": 6.564509490208478,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 705964484,
+      "step": 4111
+    },
+    {
+      "epoch": 46.70254957507082,
+      "loss": 0.13094480335712433,
+      "loss_ce": 0.007287577725946903,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 705964484,
+      "step": 4111
+    },
+    {
+      "epoch": 46.713881019830026,
+      "grad_norm": 9.280283493497476,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 706135580,
+      "step": 4112
+    },
+    {
+      "epoch": 46.713881019830026,
+      "loss": 0.12362485378980637,
+      "loss_ce": 0.006040620151907206,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 706135580,
+      "step": 4112
+    },
+    {
+      "epoch": 46.725212464589234,
+      "grad_norm": 10.852094869787907,
+      "learning_rate": 5e-06,
+      "loss": 0.1169,
+      "num_input_tokens_seen": 706306428,
+      "step": 4113
+    },
+    {
+      "epoch": 46.725212464589234,
+      "loss": 0.10285863280296326,
+      "loss_ce": 0.0033102878369390965,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 706306428,
+      "step": 4113
+    },
+    {
+      "epoch": 46.73654390934844,
+      "grad_norm": 3.4265361385303397,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 706477796,
+      "step": 4114
+    },
+    {
+      "epoch": 46.73654390934844,
+      "loss": 0.08991587162017822,
+      "loss_ce": 0.011989235877990723,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 706477796,
+      "step": 4114
+    },
+    {
+      "epoch": 46.74787535410765,
+      "grad_norm": 5.136046983358275,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 706649372,
+      "step": 4115
+    },
+    {
+      "epoch": 46.74787535410765,
+      "loss": 0.08165792375802994,
+      "loss_ce": 0.004875699989497662,
+      "loss_iou": 0.5,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 706649372,
+      "step": 4115
+    },
+    {
+      "epoch": 46.759206798866856,
+      "grad_norm": 3.890961109170352,
+      "learning_rate": 5e-06,
+      "loss": 0.0989,
+      "num_input_tokens_seen": 706820868,
+      "step": 4116
+    },
+    {
+      "epoch": 46.759206798866856,
+      "loss": 0.06502777338027954,
+      "loss_ce": 0.006983338855206966,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 706820868,
+      "step": 4116
+    },
+    {
+      "epoch": 46.77053824362606,
+      "grad_norm": 4.632384730094208,
+      "learning_rate": 5e-06,
+      "loss": 0.1497,
+      "num_input_tokens_seen": 706991388,
+      "step": 4117
+    },
+    {
+      "epoch": 46.77053824362606,
+      "loss": 0.10120939463376999,
+      "loss_ce": 0.01179288886487484,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 706991388,
+      "step": 4117
+    },
+    {
+      "epoch": 46.78186968838527,
+      "grad_norm": 4.449145455379127,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 707161544,
+      "step": 4118
+    },
+    {
+      "epoch": 46.78186968838527,
+      "loss": 0.13187825679779053,
+      "loss_ce": 0.027141934260725975,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 707161544,
+      "step": 4118
+    },
+    {
+      "epoch": 46.79320113314448,
+      "grad_norm": 3.93943171045403,
+      "learning_rate": 5e-06,
+      "loss": 0.13,
+      "num_input_tokens_seen": 707333404,
+      "step": 4119
+    },
+    {
+      "epoch": 46.79320113314448,
+      "loss": 0.13345091044902802,
+      "loss_ce": 0.0050329407677054405,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 707333404,
+      "step": 4119
+    },
+    {
+      "epoch": 46.804532577903686,
+      "grad_norm": 4.358186299723352,
+      "learning_rate": 5e-06,
+      "loss": 0.1668,
+      "num_input_tokens_seen": 707503460,
+      "step": 4120
+    },
+    {
+      "epoch": 46.804532577903686,
+      "loss": 0.20717239379882812,
+      "loss_ce": 0.010852810926735401,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 707503460,
+      "step": 4120
+    },
+    {
+      "epoch": 46.815864022662886,
+      "grad_norm": 4.131564781907263,
+      "learning_rate": 5e-06,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 707675436,
+      "step": 4121
+    },
+    {
+      "epoch": 46.815864022662886,
+      "loss": 0.2797163426876068,
+      "loss_ce": 0.01018509455025196,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 707675436,
+      "step": 4121
+    },
+    {
+      "epoch": 46.827195467422094,
+      "grad_norm": 7.321534278070325,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 707846220,
+      "step": 4122
+    },
+    {
+      "epoch": 46.827195467422094,
+      "loss": 0.10882514715194702,
+      "loss_ce": 0.002105175517499447,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 707846220,
+      "step": 4122
+    },
+    {
+      "epoch": 46.8385269121813,
+      "grad_norm": 6.76126905741619,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 708017504,
+      "step": 4123
+    },
+    {
+      "epoch": 46.8385269121813,
+      "loss": 0.11360522359609604,
+      "loss_ce": 0.008365360088646412,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 708017504,
+      "step": 4123
+    },
+    {
+      "epoch": 46.84985835694051,
+      "grad_norm": 5.4881715817973395,
+      "learning_rate": 5e-06,
+      "loss": 0.173,
+      "num_input_tokens_seen": 708187696,
+      "step": 4124
+    },
+    {
+      "epoch": 46.84985835694051,
+      "loss": 0.2519981563091278,
+      "loss_ce": 0.016066759824752808,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 708187696,
+      "step": 4124
+    },
+    {
+      "epoch": 46.861189801699716,
+      "grad_norm": 7.595222017014687,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 708359532,
+      "step": 4125
+    },
+    {
+      "epoch": 46.861189801699716,
+      "loss": 0.17828808724880219,
+      "loss_ce": 0.006352048367261887,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 708359532,
+      "step": 4125
+    },
+    {
+      "epoch": 46.87252124645892,
+      "grad_norm": 6.581851012640536,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 708531392,
+      "step": 4126
+    },
+    {
+      "epoch": 46.87252124645892,
+      "loss": 0.10337062180042267,
+      "loss_ce": 0.010292013175785542,
+      "loss_iou": 0.375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 708531392,
+      "step": 4126
+    },
+    {
+      "epoch": 46.88385269121813,
+      "grad_norm": 3.8732253010822775,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 708703060,
+      "step": 4127
+    },
+    {
+      "epoch": 46.88385269121813,
+      "loss": 0.10050459951162338,
+      "loss_ce": 0.023768151178956032,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 708703060,
+      "step": 4127
+    },
+    {
+      "epoch": 46.89518413597734,
+      "grad_norm": 4.6322022794767586,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 708874876,
+      "step": 4128
+    },
+    {
+      "epoch": 46.89518413597734,
+      "loss": 0.063858762383461,
+      "loss_ce": 0.00569226173684001,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 708874876,
+      "step": 4128
+    },
+    {
+      "epoch": 46.906515580736546,
+      "grad_norm": 3.297237030701353,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 709046612,
+      "step": 4129
+    },
+    {
+      "epoch": 46.906515580736546,
+      "loss": 0.19996875524520874,
+      "loss_ce": 0.011858412064611912,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 709046612,
+      "step": 4129
+    },
+    {
+      "epoch": 46.91784702549575,
+      "grad_norm": 5.620244892929863,
+      "learning_rate": 5e-06,
+      "loss": 0.1433,
+      "num_input_tokens_seen": 709218572,
+      "step": 4130
+    },
+    {
+      "epoch": 46.91784702549575,
+      "loss": 0.13293421268463135,
+      "loss_ce": 0.01260340679436922,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 709218572,
+      "step": 4130
+    },
+    {
+      "epoch": 46.92917847025496,
+      "grad_norm": 4.860258132306557,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 709389612,
+      "step": 4131
+    },
+    {
+      "epoch": 46.92917847025496,
+      "loss": 0.11076609790325165,
+      "loss_ce": 0.01744334027171135,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 709389612,
+      "step": 4131
+    },
+    {
+      "epoch": 46.94050991501416,
+      "grad_norm": 19.646166504744603,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 709559772,
+      "step": 4132
+    },
+    {
+      "epoch": 46.94050991501416,
+      "loss": 0.1330343782901764,
+      "loss_ce": 0.0077597107738256454,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 709559772,
+      "step": 4132
+    },
+    {
+      "epoch": 46.95184135977337,
+      "grad_norm": 6.2773890506999726,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 709730940,
+      "step": 4133
+    },
+    {
+      "epoch": 46.95184135977337,
+      "loss": 0.1601986587047577,
+      "loss_ce": 0.01643035188317299,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 709730940,
+      "step": 4133
+    },
+    {
+      "epoch": 46.963172804532576,
+      "grad_norm": 4.503412059648807,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 709901864,
+      "step": 4134
+    },
+    {
+      "epoch": 46.963172804532576,
+      "loss": 0.07518798857927322,
+      "loss_ce": 0.011528322473168373,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 709901864,
+      "step": 4134
+    },
+    {
+      "epoch": 46.97450424929178,
+      "grad_norm": 4.807239796706627,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 710073560,
+      "step": 4135
+    },
+    {
+      "epoch": 46.97450424929178,
+      "loss": 0.15487581491470337,
+      "loss_ce": 0.006224700249731541,
+      "loss_iou": 0.5,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 710073560,
+      "step": 4135
+    },
+    {
+      "epoch": 46.98583569405099,
+      "grad_norm": 3.6758352152746383,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 710245128,
+      "step": 4136
+    },
+    {
+      "epoch": 46.98583569405099,
+      "loss": 0.14955711364746094,
+      "loss_ce": 0.015829086303710938,
+      "loss_iou": 0.5,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 710245128,
+      "step": 4136
+    },
+    {
+      "epoch": 46.9971671388102,
+      "grad_norm": 5.416476818994581,
+      "learning_rate": 5e-06,
+      "loss": 0.1347,
+      "num_input_tokens_seen": 710417108,
+      "step": 4137
+    },
+    {
+      "epoch": 46.9971671388102,
+      "loss": 0.17205780744552612,
+      "loss_ce": 0.005218209698796272,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 710417108,
+      "step": 4137
+    },
+    {
+      "epoch": 46.9971671388102,
+      "loss": 0.08274796605110168,
+      "loss_ce": 0.0075984220020473,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 710460092,
+      "step": 4137
+    },
+    {
+      "epoch": 47.008498583569406,
+      "grad_norm": 3.7500122487300653,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 710588272,
+      "step": 4138
+    },
+    {
+      "epoch": 47.008498583569406,
+      "loss": 0.12801267206668854,
+      "loss_ce": 0.0033483668230473995,
+      "loss_iou": 0.6875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 710588272,
+      "step": 4138
+    },
+    {
+      "epoch": 47.01983002832861,
+      "grad_norm": 3.2823704561750566,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 710759344,
+      "step": 4139
+    },
+    {
+      "epoch": 47.01983002832861,
+      "loss": 0.15157178044319153,
+      "loss_ce": 0.00807812251150608,
+      "loss_iou": 0.453125,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 710759344,
+      "step": 4139
+    },
+    {
+      "epoch": 47.03116147308782,
+      "grad_norm": 3.2825986373188036,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 710930604,
+      "step": 4140
+    },
+    {
+      "epoch": 47.03116147308782,
+      "loss": 0.0933210551738739,
+      "loss_ce": 0.0437299944460392,
+      "loss_iou": 0.515625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 710930604,
+      "step": 4140
+    },
+    {
+      "epoch": 47.04249291784703,
+      "grad_norm": 3.274139828962442,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 711102008,
+      "step": 4141
+    },
+    {
+      "epoch": 47.04249291784703,
+      "loss": 0.15674403309822083,
+      "loss_ce": 0.00742152938619256,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 711102008,
+      "step": 4141
+    },
+    {
+      "epoch": 47.053824362606235,
+      "grad_norm": 3.3560075308653863,
+      "learning_rate": 5e-06,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 711273944,
+      "step": 4142
+    },
+    {
+      "epoch": 47.053824362606235,
+      "loss": 0.1561080813407898,
+      "loss_ce": 0.012370292097330093,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 711273944,
+      "step": 4142
+    },
+    {
+      "epoch": 47.065155807365436,
+      "grad_norm": 3.8465241877593734,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 711445736,
+      "step": 4143
+    },
+    {
+      "epoch": 47.065155807365436,
+      "loss": 0.1261669397354126,
+      "loss_ce": 0.012305853888392448,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 711445736,
+      "step": 4143
+    },
+    {
+      "epoch": 47.07648725212464,
+      "grad_norm": 3.4247367107228897,
+      "learning_rate": 5e-06,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 711617736,
+      "step": 4144
+    },
+    {
+      "epoch": 47.07648725212464,
+      "loss": 0.05091525614261627,
+      "loss_ce": 0.0004696962714660913,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 711617736,
+      "step": 4144
+    },
+    {
+      "epoch": 47.08781869688385,
+      "grad_norm": 4.703246043959707,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 711789700,
+      "step": 4145
+    },
+    {
+      "epoch": 47.08781869688385,
+      "loss": 0.09686382114887238,
+      "loss_ce": 0.0074168043211102486,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 711789700,
+      "step": 4145
+    },
+    {
+      "epoch": 47.09915014164306,
+      "grad_norm": 5.195075177504792,
+      "learning_rate": 5e-06,
+      "loss": 0.113,
+      "num_input_tokens_seen": 711961568,
+      "step": 4146
+    },
+    {
+      "epoch": 47.09915014164306,
+      "loss": 0.09385610371828079,
+      "loss_ce": 0.0018761223182082176,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 711961568,
+      "step": 4146
+    },
+    {
+      "epoch": 47.110481586402265,
+      "grad_norm": 6.566250934573623,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 712133288,
+      "step": 4147
+    },
+    {
+      "epoch": 47.110481586402265,
+      "loss": 0.13616234064102173,
+      "loss_ce": 0.00461631640791893,
+      "loss_iou": 0.11474609375,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 712133288,
+      "step": 4147
+    },
+    {
+      "epoch": 47.12181303116147,
+      "grad_norm": 8.871963562392512,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 712305604,
+      "step": 4148
+    },
+    {
+      "epoch": 47.12181303116147,
+      "loss": 0.09593099355697632,
+      "loss_ce": 0.007094325963407755,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 712305604,
+      "step": 4148
+    },
+    {
+      "epoch": 47.13314447592068,
+      "grad_norm": 3.2485259823090495,
+      "learning_rate": 5e-06,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 712474768,
+      "step": 4149
+    },
+    {
+      "epoch": 47.13314447592068,
+      "loss": 0.05707137659192085,
+      "loss_ce": 0.009921718388795853,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 712474768,
+      "step": 4149
+    },
+    {
+      "epoch": 47.14447592067989,
+      "grad_norm": 5.053475724918715,
+      "learning_rate": 5e-06,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 712644864,
+      "step": 4150
+    },
+    {
+      "epoch": 47.14447592067989,
+      "loss": 0.09276419132947922,
+      "loss_ce": 0.0073149725794792175,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 712644864,
+      "step": 4150
+    },
+    {
+      "epoch": 47.155807365439095,
+      "grad_norm": 6.185396896656014,
+      "learning_rate": 5e-06,
+      "loss": 0.1739,
+      "num_input_tokens_seen": 712815384,
+      "step": 4151
+    },
+    {
+      "epoch": 47.155807365439095,
+      "loss": 0.18949854373931885,
+      "loss_ce": 0.012435544282197952,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 712815384,
+      "step": 4151
+    },
+    {
+      "epoch": 47.1671388101983,
+      "grad_norm": 3.347347481858735,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 712986364,
+      "step": 4152
+    },
+    {
+      "epoch": 47.1671388101983,
+      "loss": 0.0958608090877533,
+      "loss_ce": 0.014439910650253296,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 712986364,
+      "step": 4152
+    },
+    {
+      "epoch": 47.17847025495751,
+      "grad_norm": 4.455102194275949,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 713157948,
+      "step": 4153
+    },
+    {
+      "epoch": 47.17847025495751,
+      "loss": 0.09848418086767197,
+      "loss_ce": 0.013126516714692116,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 713157948,
+      "step": 4153
+    },
+    {
+      "epoch": 47.18980169971671,
+      "grad_norm": 4.746513590040216,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 713328776,
+      "step": 4154
+    },
+    {
+      "epoch": 47.18980169971671,
+      "loss": 0.09090571105480194,
+      "loss_ce": 0.016290239989757538,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 713328776,
+      "step": 4154
+    },
+    {
+      "epoch": 47.20113314447592,
+      "grad_norm": 4.364856566897489,
+      "learning_rate": 5e-06,
+      "loss": 0.1657,
+      "num_input_tokens_seen": 713498652,
+      "step": 4155
+    },
+    {
+      "epoch": 47.20113314447592,
+      "loss": 0.15336136519908905,
+      "loss_ce": 0.002848668023943901,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 713498652,
+      "step": 4155
+    },
+    {
+      "epoch": 47.212464589235125,
+      "grad_norm": 5.365442246487729,
+      "learning_rate": 5e-06,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 713669348,
+      "step": 4156
+    },
+    {
+      "epoch": 47.212464589235125,
+      "loss": 0.10210458934307098,
+      "loss_ce": 0.009498994797468185,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 713669348,
+      "step": 4156
+    },
+    {
+      "epoch": 47.22379603399433,
+      "grad_norm": 5.2976462198843235,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 713841060,
+      "step": 4157
+    },
+    {
+      "epoch": 47.22379603399433,
+      "loss": 0.11391109973192215,
+      "loss_ce": 0.005604216363281012,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 713841060,
+      "step": 4157
+    },
+    {
+      "epoch": 47.23512747875354,
+      "grad_norm": 3.9077438210568856,
+      "learning_rate": 5e-06,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 714012740,
+      "step": 4158
+    },
+    {
+      "epoch": 47.23512747875354,
+      "loss": 0.058469705283641815,
+      "loss_ce": 0.004072119481861591,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 714012740,
+      "step": 4158
+    },
+    {
+      "epoch": 47.24645892351275,
+      "grad_norm": 4.182300246857357,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 714184500,
+      "step": 4159
+    },
+    {
+      "epoch": 47.24645892351275,
+      "loss": 0.13485759496688843,
+      "loss_ce": 0.002960630226880312,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 714184500,
+      "step": 4159
+    },
+    {
+      "epoch": 47.257790368271955,
+      "grad_norm": 3.878452961059349,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 714356332,
+      "step": 4160
+    },
+    {
+      "epoch": 47.257790368271955,
+      "loss": 0.0788886696100235,
+      "loss_ce": 0.007599605247378349,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 714356332,
+      "step": 4160
+    },
+    {
+      "epoch": 47.26912181303116,
+      "grad_norm": 3.308629479720532,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 714528512,
+      "step": 4161
+    },
+    {
+      "epoch": 47.26912181303116,
+      "loss": 0.08633209764957428,
+      "loss_ce": 0.008779304102063179,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 714528512,
+      "step": 4161
+    },
+    {
+      "epoch": 47.28045325779037,
+      "grad_norm": 4.59178044378594,
+      "learning_rate": 5e-06,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 714698852,
+      "step": 4162
+    },
+    {
+      "epoch": 47.28045325779037,
+      "loss": 0.192364901304245,
+      "loss_ce": 0.004376627504825592,
+      "loss_iou": 0.328125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 714698852,
+      "step": 4162
+    },
+    {
+      "epoch": 47.29178470254958,
+      "grad_norm": 3.6008658883683657,
+      "learning_rate": 5e-06,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 714869320,
+      "step": 4163
+    },
+    {
+      "epoch": 47.29178470254958,
+      "loss": 0.1373557448387146,
+      "loss_ce": 0.0013083870289847255,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 714869320,
+      "step": 4163
+    },
+    {
+      "epoch": 47.303116147308785,
+      "grad_norm": 4.302261990466185,
+      "learning_rate": 5e-06,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 715040692,
+      "step": 4164
+    },
+    {
+      "epoch": 47.303116147308785,
+      "loss": 0.06913058459758759,
+      "loss_ce": 0.0029989946633577347,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 715040692,
+      "step": 4164
+    },
+    {
+      "epoch": 47.314447592067985,
+      "grad_norm": 3.217772044208377,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 715212324,
+      "step": 4165
+    },
+    {
+      "epoch": 47.314447592067985,
+      "loss": 0.06825892627239227,
+      "loss_ce": 0.013235732913017273,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 715212324,
+      "step": 4165
+    },
+    {
+      "epoch": 47.32577903682719,
+      "grad_norm": 7.170243489217333,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 715384416,
+      "step": 4166
+    },
+    {
+      "epoch": 47.32577903682719,
+      "loss": 0.13868673145771027,
+      "loss_ce": 0.005263877101242542,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 715384416,
+      "step": 4166
+    },
+    {
+      "epoch": 47.3371104815864,
+      "grad_norm": 4.035680116667053,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 715554716,
+      "step": 4167
+    },
+    {
+      "epoch": 47.3371104815864,
+      "loss": 0.13755160570144653,
+      "loss_ce": 0.01664094813168049,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 715554716,
+      "step": 4167
+    },
+    {
+      "epoch": 47.34844192634561,
+      "grad_norm": 4.5163624861501255,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 715726904,
+      "step": 4168
+    },
+    {
+      "epoch": 47.34844192634561,
+      "loss": 0.08666570484638214,
+      "loss_ce": 0.004207202233374119,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 715726904,
+      "step": 4168
+    },
+    {
+      "epoch": 47.359773371104815,
+      "grad_norm": 8.89699127214451,
+      "learning_rate": 5e-06,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 715897916,
+      "step": 4169
+    },
+    {
+      "epoch": 47.359773371104815,
+      "loss": 0.12220755964517593,
+      "loss_ce": 0.00856009777635336,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 715897916,
+      "step": 4169
+    },
+    {
+      "epoch": 47.37110481586402,
+      "grad_norm": 7.708439530775382,
+      "learning_rate": 5e-06,
+      "loss": 0.1091,
+      "num_input_tokens_seen": 716070116,
+      "step": 4170
+    },
+    {
+      "epoch": 47.37110481586402,
+      "loss": 0.08520098030567169,
+      "loss_ce": 0.007686336524784565,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 716070116,
+      "step": 4170
+    },
+    {
+      "epoch": 47.38243626062323,
+      "grad_norm": 5.989922188487153,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 716241032,
+      "step": 4171
+    },
+    {
+      "epoch": 47.38243626062323,
+      "loss": 0.07092398405075073,
+      "loss_ce": 0.009919339790940285,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 716241032,
+      "step": 4171
+    },
+    {
+      "epoch": 47.39376770538244,
+      "grad_norm": 4.87162750552465,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 716412864,
+      "step": 4172
+    },
+    {
+      "epoch": 47.39376770538244,
+      "loss": 0.21069695055484772,
+      "loss_ce": 0.024722829461097717,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 716412864,
+      "step": 4172
+    },
+    {
+      "epoch": 47.405099150141645,
+      "grad_norm": 5.645400079701885,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 716583472,
+      "step": 4173
+    },
+    {
+      "epoch": 47.405099150141645,
+      "loss": 0.09405819326639175,
+      "loss_ce": 0.0018951030215248466,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 716583472,
+      "step": 4173
+    },
+    {
+      "epoch": 47.41643059490085,
+      "grad_norm": 4.006809787317644,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 716754880,
+      "step": 4174
+    },
+    {
+      "epoch": 47.41643059490085,
+      "loss": 0.08096863329410553,
+      "loss_ce": 0.005101929418742657,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 716754880,
+      "step": 4174
+    },
+    {
+      "epoch": 47.42776203966006,
+      "grad_norm": 4.708546500577695,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 716926804,
+      "step": 4175
+    },
+    {
+      "epoch": 47.42776203966006,
+      "loss": 0.1754830777645111,
+      "loss_ce": 0.01575406827032566,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 716926804,
+      "step": 4175
+    },
+    {
+      "epoch": 47.43909348441926,
+      "grad_norm": 3.9007423814955327,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 717097252,
+      "step": 4176
+    },
+    {
+      "epoch": 47.43909348441926,
+      "loss": 0.06068117544054985,
+      "loss_ce": 0.0019348366186022758,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 717097252,
+      "step": 4176
+    },
+    {
+      "epoch": 47.45042492917847,
+      "grad_norm": 3.777033326333161,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 717268800,
+      "step": 4177
+    },
+    {
+      "epoch": 47.45042492917847,
+      "loss": 0.11455395817756653,
+      "loss_ce": 0.0049958499148488045,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 717268800,
+      "step": 4177
+    },
+    {
+      "epoch": 47.461756373937675,
+      "grad_norm": 4.1211095172457615,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 717440720,
+      "step": 4178
+    },
+    {
+      "epoch": 47.461756373937675,
+      "loss": 0.09557903558015823,
+      "loss_ce": 0.005857359617948532,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 717440720,
+      "step": 4178
+    },
+    {
+      "epoch": 47.47308781869688,
+      "grad_norm": 16.199100882214815,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 717612424,
+      "step": 4179
+    },
+    {
+      "epoch": 47.47308781869688,
+      "loss": 0.1352303922176361,
+      "loss_ce": 0.010245651938021183,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 717612424,
+      "step": 4179
+    },
+    {
+      "epoch": 47.48441926345609,
+      "grad_norm": 2.84956766886783,
+      "learning_rate": 5e-06,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 717784756,
+      "step": 4180
+    },
+    {
+      "epoch": 47.48441926345609,
+      "loss": 0.06373220682144165,
+      "loss_ce": 0.009502465836703777,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 717784756,
+      "step": 4180
+    },
+    {
+      "epoch": 47.4957507082153,
+      "grad_norm": 3.491380691038172,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 717956336,
+      "step": 4181
+    },
+    {
+      "epoch": 47.4957507082153,
+      "loss": 0.17644673585891724,
+      "loss_ce": 0.009790237993001938,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 717956336,
+      "step": 4181
+    },
+    {
+      "epoch": 47.507082152974505,
+      "grad_norm": 155.16852225449458,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 718126756,
+      "step": 4182
+    },
+    {
+      "epoch": 47.507082152974505,
+      "loss": 0.1101723313331604,
+      "loss_ce": 0.009876305237412453,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 718126756,
+      "step": 4182
+    },
+    {
+      "epoch": 47.51841359773371,
+      "grad_norm": 3.8276867275783193,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 718298628,
+      "step": 4183
+    },
+    {
+      "epoch": 47.51841359773371,
+      "loss": 0.09529563039541245,
+      "loss_ce": 0.007557597476989031,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 718298628,
+      "step": 4183
+    },
+    {
+      "epoch": 47.52974504249292,
+      "grad_norm": 3.1537458110690726,
+      "learning_rate": 5e-06,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 718470576,
+      "step": 4184
+    },
+    {
+      "epoch": 47.52974504249292,
+      "loss": 0.05719384551048279,
+      "loss_ce": 0.009632197208702564,
+      "loss_iou": 0.5,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 718470576,
+      "step": 4184
+    },
+    {
+      "epoch": 47.54107648725213,
+      "grad_norm": 3.6755730472954578,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 718642428,
+      "step": 4185
+    },
+    {
+      "epoch": 47.54107648725213,
+      "loss": 0.08919580280780792,
+      "loss_ce": 0.005638672970235348,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 718642428,
+      "step": 4185
+    },
+    {
+      "epoch": 47.552407932011334,
+      "grad_norm": 4.051823991912136,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 718814548,
+      "step": 4186
+    },
+    {
+      "epoch": 47.552407932011334,
+      "loss": 0.14492866396903992,
+      "loss_ce": 0.00473091471940279,
+      "loss_iou": 0.5625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 718814548,
+      "step": 4186
+    },
+    {
+      "epoch": 47.563739376770535,
+      "grad_norm": 3.703186494278891,
+      "learning_rate": 5e-06,
+      "loss": 0.1534,
+      "num_input_tokens_seen": 718986472,
+      "step": 4187
+    },
+    {
+      "epoch": 47.563739376770535,
+      "loss": 0.100298210978508,
+      "loss_ce": 0.003938961774110794,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 718986472,
+      "step": 4187
+    },
+    {
+      "epoch": 47.57507082152974,
+      "grad_norm": 4.577818380793257,
+      "learning_rate": 5e-06,
+      "loss": 0.1332,
+      "num_input_tokens_seen": 719158252,
+      "step": 4188
+    },
+    {
+      "epoch": 47.57507082152974,
+      "loss": 0.06148518621921539,
+      "loss_ce": 0.007163895759731531,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 719158252,
+      "step": 4188
+    },
+    {
+      "epoch": 47.58640226628895,
+      "grad_norm": 8.840045535492111,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 719328264,
+      "step": 4189
+    },
+    {
+      "epoch": 47.58640226628895,
+      "loss": 0.10586181282997131,
+      "loss_ce": 0.011135252192616463,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 719328264,
+      "step": 4189
+    },
+    {
+      "epoch": 47.59773371104816,
+      "grad_norm": 4.076542713532686,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 719500244,
+      "step": 4190
+    },
+    {
+      "epoch": 47.59773371104816,
+      "loss": 0.11987441033124924,
+      "loss_ce": 0.0046705519780516624,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 719500244,
+      "step": 4190
+    },
+    {
+      "epoch": 47.609065155807365,
+      "grad_norm": 7.16829502025035,
+      "learning_rate": 5e-06,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 719670588,
+      "step": 4191
+    },
+    {
+      "epoch": 47.609065155807365,
+      "loss": 0.08642441034317017,
+      "loss_ce": 0.013853613287210464,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 719670588,
+      "step": 4191
+    },
+    {
+      "epoch": 47.62039660056657,
+      "grad_norm": 3.6589365251270713,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 719840676,
+      "step": 4192
+    },
+    {
+      "epoch": 47.62039660056657,
+      "loss": 0.13755139708518982,
+      "loss_ce": 0.006753057707101107,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 719840676,
+      "step": 4192
+    },
+    {
+      "epoch": 47.63172804532578,
+      "grad_norm": 3.407042032395625,
+      "learning_rate": 5e-06,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 720011548,
+      "step": 4193
+    },
+    {
+      "epoch": 47.63172804532578,
+      "loss": 0.07615301758050919,
+      "loss_ce": 0.016796328127384186,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 720011548,
+      "step": 4193
+    },
+    {
+      "epoch": 47.64305949008499,
+      "grad_norm": 4.3316498302145625,
+      "learning_rate": 5e-06,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 720183748,
+      "step": 4194
+    },
+    {
+      "epoch": 47.64305949008499,
+      "loss": 0.07157811522483826,
+      "loss_ce": 0.0006247470737434924,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 720183748,
+      "step": 4194
+    },
+    {
+      "epoch": 47.654390934844194,
+      "grad_norm": 6.905107492272185,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 720355300,
+      "step": 4195
+    },
+    {
+      "epoch": 47.654390934844194,
+      "loss": 0.10575076192617416,
+      "loss_ce": 0.019172394648194313,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 720355300,
+      "step": 4195
+    },
+    {
+      "epoch": 47.6657223796034,
+      "grad_norm": 4.192693541095733,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 720525276,
+      "step": 4196
+    },
+    {
+      "epoch": 47.6657223796034,
+      "loss": 0.11182039231061935,
+      "loss_ce": 0.009372884407639503,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 720525276,
+      "step": 4196
+    },
+    {
+      "epoch": 47.67705382436261,
+      "grad_norm": 4.891522799595915,
+      "learning_rate": 5e-06,
+      "loss": 0.1548,
+      "num_input_tokens_seen": 720696908,
+      "step": 4197
+    },
+    {
+      "epoch": 47.67705382436261,
+      "loss": 0.0823417603969574,
+      "loss_ce": 0.011205289512872696,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 720696908,
+      "step": 4197
+    },
+    {
+      "epoch": 47.68838526912181,
+      "grad_norm": 6.922247898464801,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 720868860,
+      "step": 4198
+    },
+    {
+      "epoch": 47.68838526912181,
+      "loss": 0.06526593118906021,
+      "loss_ce": 0.0011332416906952858,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 720868860,
+      "step": 4198
+    },
+    {
+      "epoch": 47.69971671388102,
+      "grad_norm": 4.363957584028134,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 721040436,
+      "step": 4199
+    },
+    {
+      "epoch": 47.69971671388102,
+      "loss": 0.07195132970809937,
+      "loss_ce": 0.0017151257488876581,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 721040436,
+      "step": 4199
+    },
+    {
+      "epoch": 47.711048158640224,
+      "grad_norm": 5.363117587429788,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 721208832,
+      "step": 4200
+    },
+    {
+      "epoch": 47.711048158640224,
+      "loss": 0.08068595826625824,
+      "loss_ce": 0.0013250020565465093,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 721208832,
+      "step": 4200
+    },
+    {
+      "epoch": 47.72237960339943,
+      "grad_norm": 6.431877585859448,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 721379924,
+      "step": 4201
+    },
+    {
+      "epoch": 47.72237960339943,
+      "loss": 0.1810862123966217,
+      "loss_ce": 0.003901155898347497,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 721379924,
+      "step": 4201
+    },
+    {
+      "epoch": 47.73371104815864,
+      "grad_norm": 4.165527307527694,
+      "learning_rate": 5e-06,
+      "loss": 0.138,
+      "num_input_tokens_seen": 721551444,
+      "step": 4202
+    },
+    {
+      "epoch": 47.73371104815864,
+      "loss": 0.17227022349834442,
+      "loss_ce": 0.011717259883880615,
+      "loss_iou": 0.095703125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 721551444,
+      "step": 4202
+    },
+    {
+      "epoch": 47.74504249291785,
+      "grad_norm": 4.002261686857825,
+      "learning_rate": 5e-06,
+      "loss": 0.1285,
+      "num_input_tokens_seen": 721723112,
+      "step": 4203
+    },
+    {
+      "epoch": 47.74504249291785,
+      "loss": 0.1651363968849182,
+      "loss_ce": 0.009649345651268959,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 721723112,
+      "step": 4203
+    },
+    {
+      "epoch": 47.756373937677054,
+      "grad_norm": 6.208699437208896,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 721894704,
+      "step": 4204
+    },
+    {
+      "epoch": 47.756373937677054,
+      "loss": 0.1591370552778244,
+      "loss_ce": 0.0043824203312397,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 721894704,
+      "step": 4204
+    },
+    {
+      "epoch": 47.76770538243626,
+      "grad_norm": 4.1244437800790275,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 722065572,
+      "step": 4205
+    },
+    {
+      "epoch": 47.76770538243626,
+      "loss": 0.09495656192302704,
+      "loss_ce": 0.006516617722809315,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 722065572,
+      "step": 4205
+    },
+    {
+      "epoch": 47.77903682719547,
+      "grad_norm": 4.4937411023770455,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 722237332,
+      "step": 4206
+    },
+    {
+      "epoch": 47.77903682719547,
+      "loss": 0.07483170926570892,
+      "loss_ce": 0.0135981859639287,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 722237332,
+      "step": 4206
+    },
+    {
+      "epoch": 47.79036827195468,
+      "grad_norm": 3.6510219836697395,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 722409240,
+      "step": 4207
+    },
+    {
+      "epoch": 47.79036827195468,
+      "loss": 0.09725276380777359,
+      "loss_ce": 0.006340898107737303,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 722409240,
+      "step": 4207
+    },
+    {
+      "epoch": 47.801699716713884,
+      "grad_norm": 4.261052119114076,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 722580796,
+      "step": 4208
+    },
+    {
+      "epoch": 47.801699716713884,
+      "loss": 0.10438238829374313,
+      "loss_ce": 0.007580631412565708,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 722580796,
+      "step": 4208
+    },
+    {
+      "epoch": 47.81303116147309,
+      "grad_norm": 7.865780838892762,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 722750844,
+      "step": 4209
+    },
+    {
+      "epoch": 47.81303116147309,
+      "loss": 0.1078704372048378,
+      "loss_ce": 0.002554281149059534,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 722750844,
+      "step": 4209
+    },
+    {
+      "epoch": 47.82436260623229,
+      "grad_norm": 4.28299915733703,
+      "learning_rate": 5e-06,
+      "loss": 0.1536,
+      "num_input_tokens_seen": 722922796,
+      "step": 4210
+    },
+    {
+      "epoch": 47.82436260623229,
+      "loss": 0.2250032275915146,
+      "loss_ce": 0.007596009410917759,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 722922796,
+      "step": 4210
+    },
+    {
+      "epoch": 47.8356940509915,
+      "grad_norm": 3.6998087357345852,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 723094576,
+      "step": 4211
+    },
+    {
+      "epoch": 47.8356940509915,
+      "loss": 0.14309553802013397,
+      "loss_ce": 0.004606769420206547,
+      "loss_iou": 0.421875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 723094576,
+      "step": 4211
+    },
+    {
+      "epoch": 47.84702549575071,
+      "grad_norm": 5.991102049383098,
+      "learning_rate": 5e-06,
+      "loss": 0.1603,
+      "num_input_tokens_seen": 723266368,
+      "step": 4212
+    },
+    {
+      "epoch": 47.84702549575071,
+      "loss": 0.18474189937114716,
+      "loss_ce": 0.026569288223981857,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 723266368,
+      "step": 4212
+    },
+    {
+      "epoch": 47.858356940509914,
+      "grad_norm": 3.8338759663685575,
+      "learning_rate": 5e-06,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 723437040,
+      "step": 4213
+    },
+    {
+      "epoch": 47.858356940509914,
+      "loss": 0.05602739006280899,
+      "loss_ce": 0.0008821301162242889,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 723437040,
+      "step": 4213
+    },
+    {
+      "epoch": 47.86968838526912,
+      "grad_norm": 4.322287049616213,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 723609056,
+      "step": 4214
+    },
+    {
+      "epoch": 47.86968838526912,
+      "loss": 0.07943825423717499,
+      "loss_ce": 0.0036936295218765736,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 723609056,
+      "step": 4214
+    },
+    {
+      "epoch": 47.88101983002833,
+      "grad_norm": 4.391115906578069,
+      "learning_rate": 5e-06,
+      "loss": 0.1255,
+      "num_input_tokens_seen": 723781028,
+      "step": 4215
+    },
+    {
+      "epoch": 47.88101983002833,
+      "loss": 0.14958599209785461,
+      "loss_ce": 0.006428041495382786,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 723781028,
+      "step": 4215
+    },
+    {
+      "epoch": 47.892351274787536,
+      "grad_norm": 5.005200616958996,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 723952968,
+      "step": 4216
+    },
+    {
+      "epoch": 47.892351274787536,
+      "loss": 0.1433696150779724,
+      "loss_ce": 0.00799363013356924,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 723952968,
+      "step": 4216
+    },
+    {
+      "epoch": 47.903682719546744,
+      "grad_norm": 4.095520063661307,
+      "learning_rate": 5e-06,
+      "loss": 0.0841,
+      "num_input_tokens_seen": 724124880,
+      "step": 4217
+    },
+    {
+      "epoch": 47.903682719546744,
+      "loss": 0.057315874844789505,
+      "loss_ce": 0.004840902052819729,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 724124880,
+      "step": 4217
+    },
+    {
+      "epoch": 47.91501416430595,
+      "grad_norm": 4.310502822772367,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 724296544,
+      "step": 4218
+    },
+    {
+      "epoch": 47.91501416430595,
+      "loss": 0.061356451362371445,
+      "loss_ce": 0.0025795954279601574,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 724296544,
+      "step": 4218
+    },
+    {
+      "epoch": 47.92634560906516,
+      "grad_norm": 4.333164150456548,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 724466208,
+      "step": 4219
+    },
+    {
+      "epoch": 47.92634560906516,
+      "loss": 0.0856182724237442,
+      "loss_ce": 0.006638777907937765,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 724466208,
+      "step": 4219
+    },
+    {
+      "epoch": 47.93767705382436,
+      "grad_norm": 5.764154800687036,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 724638180,
+      "step": 4220
+    },
+    {
+      "epoch": 47.93767705382436,
+      "loss": 0.11939413845539093,
+      "loss_ce": 0.009881798177957535,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 724638180,
+      "step": 4220
+    },
+    {
+      "epoch": 47.94900849858357,
+      "grad_norm": 3.3802125938357785,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 724808840,
+      "step": 4221
+    },
+    {
+      "epoch": 47.94900849858357,
+      "loss": 0.06741583347320557,
+      "loss_ce": 0.009401912800967693,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 724808840,
+      "step": 4221
+    },
+    {
+      "epoch": 47.960339943342774,
+      "grad_norm": 6.922033666320785,
+      "learning_rate": 5e-06,
+      "loss": 0.1674,
+      "num_input_tokens_seen": 724980784,
+      "step": 4222
+    },
+    {
+      "epoch": 47.960339943342774,
+      "loss": 0.09399766474962234,
+      "loss_ce": 0.007388778030872345,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 724980784,
+      "step": 4222
+    },
+    {
+      "epoch": 47.97167138810198,
+      "grad_norm": 4.452549019340513,
+      "learning_rate": 5e-06,
+      "loss": 0.1412,
+      "num_input_tokens_seen": 725152688,
+      "step": 4223
+    },
+    {
+      "epoch": 47.97167138810198,
+      "loss": 0.12679237127304077,
+      "loss_ce": 0.006232663057744503,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 725152688,
+      "step": 4223
+    },
+    {
+      "epoch": 47.98300283286119,
+      "grad_norm": 5.800637540562238,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 725324828,
+      "step": 4224
+    },
+    {
+      "epoch": 47.98300283286119,
+      "loss": 0.14737582206726074,
+      "loss_ce": 0.01047397032380104,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 725324828,
+      "step": 4224
+    },
+    {
+      "epoch": 47.994334277620396,
+      "grad_norm": 4.02003753009002,
+      "learning_rate": 5e-06,
+      "loss": 0.1497,
+      "num_input_tokens_seen": 725496612,
+      "step": 4225
+    },
+    {
+      "epoch": 47.994334277620396,
+      "loss": 0.18000835180282593,
+      "loss_ce": 0.012146404944360256,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 725496612,
+      "step": 4225
+    },
+    {
+      "epoch": 47.994334277620396,
+      "loss": 0.0791236162185669,
+      "loss_ce": 0.004401322919875383,
+      "loss_iou": 0.484375,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 725582460,
+      "step": 4225
+    },
+    {
+      "epoch": 48.005665722379604,
+      "grad_norm": 2.560986806828304,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 725666636,
+      "step": 4226
+    },
+    {
+      "epoch": 48.005665722379604,
+      "loss": 0.08964906632900238,
+      "loss_ce": 0.001926284865476191,
+      "loss_iou": 0.390625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 725666636,
+      "step": 4226
+    },
+    {
+      "epoch": 48.01699716713881,
+      "grad_norm": 2.2664165575224833,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 725838372,
+      "step": 4227
+    },
+    {
+      "epoch": 48.01699716713881,
+      "loss": 0.07389166951179504,
+      "loss_ce": 0.006768262945115566,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 725838372,
+      "step": 4227
+    },
+    {
+      "epoch": 48.02832861189802,
+      "grad_norm": 3.1319693365791594,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 726009644,
+      "step": 4228
+    },
+    {
+      "epoch": 48.02832861189802,
+      "loss": 0.0971192866563797,
+      "loss_ce": 0.0033235165756195784,
+      "loss_iou": 0.390625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 726009644,
+      "step": 4228
+    },
+    {
+      "epoch": 48.039660056657226,
+      "grad_norm": 5.2748469049709525,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 726181768,
+      "step": 4229
+    },
+    {
+      "epoch": 48.039660056657226,
+      "loss": 0.05843126028776169,
+      "loss_ce": 0.003163927933201194,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 726181768,
+      "step": 4229
+    },
+    {
+      "epoch": 48.05099150141643,
+      "grad_norm": 4.344460325867675,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 726353104,
+      "step": 4230
+    },
+    {
+      "epoch": 48.05099150141643,
+      "loss": 0.0806979387998581,
+      "loss_ce": 0.0011996516259387136,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 726353104,
+      "step": 4230
+    },
+    {
+      "epoch": 48.06232294617564,
+      "grad_norm": 5.095345151385122,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 726525064,
+      "step": 4231
+    },
+    {
+      "epoch": 48.06232294617564,
+      "loss": 0.08278858661651611,
+      "loss_ce": 0.004053235519677401,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 726525064,
+      "step": 4231
+    },
+    {
+      "epoch": 48.07365439093484,
+      "grad_norm": 4.667750883667177,
+      "learning_rate": 5e-06,
+      "loss": 0.1381,
+      "num_input_tokens_seen": 726696020,
+      "step": 4232
+    },
+    {
+      "epoch": 48.07365439093484,
+      "loss": 0.1957799196243286,
+      "loss_ce": 0.0008183744503185153,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 726696020,
+      "step": 4232
+    },
+    {
+      "epoch": 48.08498583569405,
+      "grad_norm": 5.079813233099534,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 726868240,
+      "step": 4233
+    },
+    {
+      "epoch": 48.08498583569405,
+      "loss": 0.07104089856147766,
+      "loss_ce": 0.004649905487895012,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 726868240,
+      "step": 4233
+    },
+    {
+      "epoch": 48.096317280453256,
+      "grad_norm": 5.060157686812156,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 727038508,
+      "step": 4234
+    },
+    {
+      "epoch": 48.096317280453256,
+      "loss": 0.10144554078578949,
+      "loss_ce": 0.004186021164059639,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 727038508,
+      "step": 4234
+    },
+    {
+      "epoch": 48.107648725212464,
+      "grad_norm": 3.615878582070035,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 727210044,
+      "step": 4235
+    },
+    {
+      "epoch": 48.107648725212464,
+      "loss": 0.16303175687789917,
+      "loss_ce": 0.004035164602100849,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 727210044,
+      "step": 4235
+    },
+    {
+      "epoch": 48.11898016997167,
+      "grad_norm": 4.016571185927729,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "num_input_tokens_seen": 727382172,
+      "step": 4236
+    },
+    {
+      "epoch": 48.11898016997167,
+      "loss": 0.15425026416778564,
+      "loss_ce": 0.0001975233171833679,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 727382172,
+      "step": 4236
+    },
+    {
+      "epoch": 48.13031161473088,
+      "grad_norm": 4.195520203080299,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 727553700,
+      "step": 4237
+    },
+    {
+      "epoch": 48.13031161473088,
+      "loss": 0.06917223334312439,
+      "loss_ce": 0.0008281111950054765,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 727553700,
+      "step": 4237
+    },
+    {
+      "epoch": 48.141643059490086,
+      "grad_norm": 8.27732720401473,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 727725472,
+      "step": 4238
+    },
+    {
+      "epoch": 48.141643059490086,
+      "loss": 0.13010528683662415,
+      "loss_ce": 0.017617493867874146,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 727725472,
+      "step": 4238
+    },
+    {
+      "epoch": 48.15297450424929,
+      "grad_norm": 3.307257998824753,
+      "learning_rate": 5e-06,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 727896288,
+      "step": 4239
+    },
+    {
+      "epoch": 48.15297450424929,
+      "loss": 0.10900919884443283,
+      "loss_ce": 0.0037235496565699577,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 727896288,
+      "step": 4239
+    },
+    {
+      "epoch": 48.1643059490085,
+      "grad_norm": 11.103666262859841,
+      "learning_rate": 5e-06,
+      "loss": 0.1706,
+      "num_input_tokens_seen": 728068536,
+      "step": 4240
+    },
+    {
+      "epoch": 48.1643059490085,
+      "loss": 0.1067027896642685,
+      "loss_ce": 0.0008067976450547576,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 728068536,
+      "step": 4240
+    },
+    {
+      "epoch": 48.17563739376771,
+      "grad_norm": 5.018892917937562,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 728240468,
+      "step": 4241
+    },
+    {
+      "epoch": 48.17563739376771,
+      "loss": 0.08491016924381256,
+      "loss_ce": 0.0035197832621634007,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 728240468,
+      "step": 4241
+    },
+    {
+      "epoch": 48.186968838526916,
+      "grad_norm": 4.185984092897652,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 728412140,
+      "step": 4242
+    },
+    {
+      "epoch": 48.186968838526916,
+      "loss": 0.1789214313030243,
+      "loss_ce": 0.005245887208729982,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 728412140,
+      "step": 4242
+    },
+    {
+      "epoch": 48.198300283286116,
+      "grad_norm": 5.1223715318213605,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 728583968,
+      "step": 4243
+    },
+    {
+      "epoch": 48.198300283286116,
+      "loss": 0.1572830080986023,
+      "loss_ce": 0.002650424838066101,
+      "loss_iou": 0.0,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 728583968,
+      "step": 4243
+    },
+    {
+      "epoch": 48.20963172804532,
+      "grad_norm": 3.993915516207532,
+      "learning_rate": 5e-06,
+      "loss": 0.0841,
+      "num_input_tokens_seen": 728755788,
+      "step": 4244
+    },
+    {
+      "epoch": 48.20963172804532,
+      "loss": 0.09096597135066986,
+      "loss_ce": 0.006157625466585159,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 728755788,
+      "step": 4244
+    },
+    {
+      "epoch": 48.22096317280453,
+      "grad_norm": 3.775856290174205,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 728926556,
+      "step": 4245
+    },
+    {
+      "epoch": 48.22096317280453,
+      "loss": 0.0689842626452446,
+      "loss_ce": 0.0012352368794381618,
+      "loss_iou": 0.064453125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 728926556,
+      "step": 4245
+    },
+    {
+      "epoch": 48.23229461756374,
+      "grad_norm": 4.133716410062589,
+      "learning_rate": 5e-06,
+      "loss": 0.1317,
+      "num_input_tokens_seen": 729098520,
+      "step": 4246
+    },
+    {
+      "epoch": 48.23229461756374,
+      "loss": 0.2406463772058487,
+      "loss_ce": 0.0025787455961108208,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 729098520,
+      "step": 4246
+    },
+    {
+      "epoch": 48.243626062322946,
+      "grad_norm": 2.7103875785594034,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 729270316,
+      "step": 4247
+    },
+    {
+      "epoch": 48.243626062322946,
+      "loss": 0.08812672644853592,
+      "loss_ce": 0.00307423691265285,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 729270316,
+      "step": 4247
+    },
+    {
+      "epoch": 48.25495750708215,
+      "grad_norm": 2.9468621257751746,
+      "learning_rate": 5e-06,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 729442276,
+      "step": 4248
+    },
+    {
+      "epoch": 48.25495750708215,
+      "loss": 0.06399865448474884,
+      "loss_ce": 0.008365103974938393,
+      "loss_iou": 0.05712890625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 729442276,
+      "step": 4248
+    },
+    {
+      "epoch": 48.26628895184136,
+      "grad_norm": 3.4277331328508582,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 729614272,
+      "step": 4249
+    },
+    {
+      "epoch": 48.26628895184136,
+      "loss": 0.08927109092473984,
+      "loss_ce": 0.001243135193362832,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 729614272,
+      "step": 4249
+    },
+    {
+      "epoch": 48.27762039660057,
+      "grad_norm": 4.05348556065702,
+      "learning_rate": 5e-06,
+      "loss": 0.1188,
+      "num_input_tokens_seen": 729786340,
+      "step": 4250
+    },
+    {
+      "epoch": 48.27762039660057,
+      "eval_seeclick_CIoU": 0.5490350723266602,
+      "eval_seeclick_GIoU": 0.5412544012069702,
+      "eval_seeclick_IoU": 0.582158088684082,
+      "eval_seeclick_MAE_all": 0.060669198632240295,
+      "eval_seeclick_MAE_h": 0.03776175528764725,
+      "eval_seeclick_MAE_w": 0.07905042171478271,
+      "eval_seeclick_MAE_x": 0.09220220521092415,
+      "eval_seeclick_MAE_y": 0.03366241417825222,
+      "eval_seeclick_NUM_probability": 0.999801903963089,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.67937833070755,
+      "eval_seeclick_loss_ce": 0.43770506978034973,
+      "eval_seeclick_loss_iou": 0.5579833984375,
+      "eval_seeclick_loss_num": 0.047847747802734375,
+      "eval_seeclick_loss_xval": 0.239410400390625,
+      "eval_seeclick_runtime": 69.0762,
+      "eval_seeclick_samples_per_second": 0.623,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 729786340,
+      "step": 4250
+    },
+    {
+      "epoch": 48.27762039660057,
+      "eval_icons_CIoU": 0.7743414342403412,
+      "eval_icons_GIoU": 0.7759648859500885,
+      "eval_icons_IoU": 0.7864299714565277,
+      "eval_icons_MAE_all": 0.03141877520829439,
+      "eval_icons_MAE_h": 0.02992552798241377,
+      "eval_icons_MAE_w": 0.03443619981408119,
+      "eval_icons_MAE_x": 0.029241597279906273,
+      "eval_icons_MAE_y": 0.032071780413389206,
+      "eval_icons_NUM_probability": 0.999067485332489,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.09602519869804382,
+      "eval_icons_loss_ce": 0.00017912417388288304,
+      "eval_icons_loss_iou": 0.5159912109375,
+      "eval_icons_loss_num": 0.017490386962890625,
+      "eval_icons_loss_xval": 0.0875244140625,
+      "eval_icons_runtime": 80.1421,
+      "eval_icons_samples_per_second": 0.624,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 729786340,
+      "step": 4250
+    },
+    {
+      "epoch": 48.27762039660057,
+      "eval_screenspot_CIoU": 0.6139720479647318,
+      "eval_screenspot_GIoU": 0.6172506213188171,
+      "eval_screenspot_IoU": 0.6374714374542236,
+      "eval_screenspot_MAE_all": 0.06436327720681827,
+      "eval_screenspot_MAE_h": 0.038560821364323296,
+      "eval_screenspot_MAE_w": 0.11435861885547638,
+      "eval_screenspot_MAE_x": 0.07099338124195735,
+      "eval_screenspot_MAE_y": 0.03354028550287088,
+      "eval_screenspot_NUM_probability": 0.999848206837972,
+      "eval_screenspot_inside_bbox": 0.9212500055631002,
+      "eval_screenspot_loss": 0.2953372299671173,
+      "eval_screenspot_loss_ce": 0.012427997930596272,
+      "eval_screenspot_loss_iou": 0.561279296875,
+      "eval_screenspot_loss_num": 0.056116739908854164,
+      "eval_screenspot_loss_xval": 0.2805582682291667,
+      "eval_screenspot_runtime": 144.2475,
+      "eval_screenspot_samples_per_second": 0.617,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 729786340,
+      "step": 4250
+    },
+    {
+      "epoch": 48.27762039660057,
+      "eval_compot_CIoU": 0.8351185619831085,
+      "eval_compot_GIoU": 0.8283049464225769,
+      "eval_compot_IoU": 0.8476720750331879,
+      "eval_compot_MAE_all": 0.028342776000499725,
+      "eval_compot_MAE_h": 0.02567121386528015,
+      "eval_compot_MAE_w": 0.031636047177016735,
+      "eval_compot_MAE_x": 0.030706957913935184,
+      "eval_compot_MAE_y": 0.02535688877105713,
+      "eval_compot_NUM_probability": 0.9999547600746155,
+      "eval_compot_inside_bbox": 0.9409722089767456,
+      "eval_compot_loss": 0.09002256393432617,
+      "eval_compot_loss_ce": 0.00013389794912654907,
+      "eval_compot_loss_iou": 0.5308837890625,
+      "eval_compot_loss_num": 0.0156402587890625,
+      "eval_compot_loss_xval": 0.0781707763671875,
+      "eval_compot_runtime": 89.1767,
+      "eval_compot_samples_per_second": 0.561,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 729786340,
+      "step": 4250
+    },
+    {
+      "epoch": 48.27762039660057,
+      "eval_custom_ui_MAE_all": 0.02408888377249241,
+      "eval_custom_ui_MAE_x": 0.03926929924637079,
+      "eval_custom_ui_MAE_y": 0.008908470161259174,
+      "eval_custom_ui_NUM_probability": 0.9998082518577576,
+      "eval_custom_ui_loss": 0.2017803192138672,
+      "eval_custom_ui_loss_ce": 0.07914393022656441,
+      "eval_custom_ui_loss_num": 0.024816513061523438,
+      "eval_custom_ui_loss_xval": 0.1241302490234375,
+      "eval_custom_ui_runtime": 57.9242,
+      "eval_custom_ui_samples_per_second": 0.863,
+      "eval_custom_ui_steps_per_second": 0.035,
+      "num_input_tokens_seen": 729786340,
+      "step": 4250
+    },
+    {
+      "epoch": 48.27762039660057,
+      "loss": 0.24452589452266693,
+      "loss_ce": 0.09440992772579193,
+      "loss_iou": 0.0,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 729786340,
+      "step": 4250
+    },
+    {
+      "epoch": 48.288951841359776,
+      "grad_norm": 5.008323932931067,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 729958032,
+      "step": 4251
+    },
+    {
+      "epoch": 48.288951841359776,
+      "loss": 0.1132562905550003,
+      "loss_ce": 0.002630067290738225,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 729958032,
+      "step": 4251
+    },
+    {
+      "epoch": 48.30028328611898,
+      "grad_norm": 4.965848227978041,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 730129792,
+      "step": 4252
+    },
+    {
+      "epoch": 48.30028328611898,
+      "loss": 0.06447215378284454,
+      "loss_ce": 0.010211902670562267,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 730129792,
+      "step": 4252
+    },
+    {
+      "epoch": 48.31161473087819,
+      "grad_norm": 4.285547477785948,
+      "learning_rate": 5e-06,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 730302020,
+      "step": 4253
+    },
+    {
+      "epoch": 48.31161473087819,
+      "loss": 0.09805352985858917,
+      "loss_ce": 0.0020757406018674374,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 730302020,
+      "step": 4253
+    },
+    {
+      "epoch": 48.32294617563739,
+      "grad_norm": 4.479127188763341,
+      "learning_rate": 5e-06,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 730473660,
+      "step": 4254
+    },
+    {
+      "epoch": 48.32294617563739,
+      "loss": 0.08305899053812027,
+      "loss_ce": 0.005117096938192844,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 730473660,
+      "step": 4254
+    },
+    {
+      "epoch": 48.3342776203966,
+      "grad_norm": 4.96026218858615,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 730644508,
+      "step": 4255
+    },
+    {
+      "epoch": 48.3342776203966,
+      "loss": 0.10590037703514099,
+      "loss_ce": 0.010822868905961514,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 730644508,
+      "step": 4255
+    },
+    {
+      "epoch": 48.345609065155806,
+      "grad_norm": 4.245440900984709,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 730816284,
+      "step": 4256
+    },
+    {
+      "epoch": 48.345609065155806,
+      "loss": 0.1167280375957489,
+      "loss_ce": 0.0030805806163698435,
+      "loss_iou": 0.390625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 730816284,
+      "step": 4256
+    },
+    {
+      "epoch": 48.35694050991501,
+      "grad_norm": 3.0107728653746815,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 730988000,
+      "step": 4257
+    },
+    {
+      "epoch": 48.35694050991501,
+      "loss": 0.048872120678424835,
+      "loss_ce": 0.001173144206404686,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 730988000,
+      "step": 4257
+    },
+    {
+      "epoch": 48.36827195467422,
+      "grad_norm": 2.979316255092292,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 731159896,
+      "step": 4258
+    },
+    {
+      "epoch": 48.36827195467422,
+      "loss": 0.1505395770072937,
+      "loss_ce": 0.0014612016966566443,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 731159896,
+      "step": 4258
+    },
+    {
+      "epoch": 48.37960339943343,
+      "grad_norm": 3.126901756899927,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 731331788,
+      "step": 4259
+    },
+    {
+      "epoch": 48.37960339943343,
+      "loss": 0.07996238768100739,
+      "loss_ce": 0.0031649041920900345,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 731331788,
+      "step": 4259
+    },
+    {
+      "epoch": 48.390934844192635,
+      "grad_norm": 5.319590957025481,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 731501628,
+      "step": 4260
+    },
+    {
+      "epoch": 48.390934844192635,
+      "loss": 0.16945302486419678,
+      "loss_ce": 0.006824860814958811,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 731501628,
+      "step": 4260
+    },
+    {
+      "epoch": 48.40226628895184,
+      "grad_norm": 4.84252851343208,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 731673104,
+      "step": 4261
+    },
+    {
+      "epoch": 48.40226628895184,
+      "loss": 0.1442323923110962,
+      "loss_ce": 0.0037599829956889153,
+      "loss_iou": 0.25,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 731673104,
+      "step": 4261
+    },
+    {
+      "epoch": 48.41359773371105,
+      "grad_norm": 3.4432359752470787,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 731844604,
+      "step": 4262
+    },
+    {
+      "epoch": 48.41359773371105,
+      "loss": 0.08415471762418747,
+      "loss_ce": 0.0027338173240423203,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 731844604,
+      "step": 4262
+    },
+    {
+      "epoch": 48.42492917847026,
+      "grad_norm": 4.286629721086909,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 732014724,
+      "step": 4263
+    },
+    {
+      "epoch": 48.42492917847026,
+      "loss": 0.11379265785217285,
+      "loss_ce": 0.004966974724084139,
+      "loss_iou": 0.53125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 732014724,
+      "step": 4263
+    },
+    {
+      "epoch": 48.436260623229465,
+      "grad_norm": 4.9569213446095555,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 732185356,
+      "step": 4264
+    },
+    {
+      "epoch": 48.436260623229465,
+      "loss": 0.07775212824344635,
+      "loss_ce": 0.0020227553322911263,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 732185356,
+      "step": 4264
+    },
+    {
+      "epoch": 48.447592067988666,
+      "grad_norm": 4.092167613692802,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 732356964,
+      "step": 4265
+    },
+    {
+      "epoch": 48.447592067988666,
+      "loss": 0.07262058556079865,
+      "loss_ce": 0.0035593120846897364,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 732356964,
+      "step": 4265
+    },
+    {
+      "epoch": 48.45892351274787,
+      "grad_norm": 5.196939899072797,
+      "learning_rate": 5e-06,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 732528184,
+      "step": 4266
+    },
+    {
+      "epoch": 48.45892351274787,
+      "loss": 0.10172464698553085,
+      "loss_ce": 0.0025730361230671406,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 732528184,
+      "step": 4266
+    },
+    {
+      "epoch": 48.47025495750708,
+      "grad_norm": 4.275484228140162,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 732700036,
+      "step": 4267
+    },
+    {
+      "epoch": 48.47025495750708,
+      "loss": 0.11507698893547058,
+      "loss_ce": 0.0015363376587629318,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 732700036,
+      "step": 4267
+    },
+    {
+      "epoch": 48.48158640226629,
+      "grad_norm": 3.356876264887818,
+      "learning_rate": 5e-06,
+      "loss": 0.1181,
+      "num_input_tokens_seen": 732870896,
+      "step": 4268
+    },
+    {
+      "epoch": 48.48158640226629,
+      "loss": 0.1799665242433548,
+      "loss_ce": 0.007267555221915245,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 732870896,
+      "step": 4268
+    },
+    {
+      "epoch": 48.492917847025495,
+      "grad_norm": 3.5321914434287955,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 733040180,
+      "step": 4269
+    },
+    {
+      "epoch": 48.492917847025495,
+      "loss": 0.12801069021224976,
+      "loss_ce": 0.007283156272023916,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 733040180,
+      "step": 4269
+    },
+    {
+      "epoch": 48.5042492917847,
+      "grad_norm": 4.115522581994122,
+      "learning_rate": 5e-06,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 733211112,
+      "step": 4270
+    },
+    {
+      "epoch": 48.5042492917847,
+      "loss": 0.0661100298166275,
+      "loss_ce": 0.007119554560631514,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 733211112,
+      "step": 4270
+    },
+    {
+      "epoch": 48.51558073654391,
+      "grad_norm": 3.4497487339955257,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 733382704,
+      "step": 4271
+    },
+    {
+      "epoch": 48.51558073654391,
+      "loss": 0.10695919394493103,
+      "loss_ce": 0.001643032068386674,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 733382704,
+      "step": 4271
+    },
+    {
+      "epoch": 48.52691218130312,
+      "grad_norm": 3.6149180820629065,
+      "learning_rate": 5e-06,
+      "loss": 0.099,
+      "num_input_tokens_seen": 733553896,
+      "step": 4272
+    },
+    {
+      "epoch": 48.52691218130312,
+      "loss": 0.13652561604976654,
+      "loss_ce": 0.003636822337284684,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 733553896,
+      "step": 4272
+    },
+    {
+      "epoch": 48.538243626062325,
+      "grad_norm": 8.280845339631705,
+      "learning_rate": 5e-06,
+      "loss": 0.1453,
+      "num_input_tokens_seen": 733725752,
+      "step": 4273
+    },
+    {
+      "epoch": 48.538243626062325,
+      "loss": 0.12543433904647827,
+      "loss_ce": 0.0014108981704339385,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 733725752,
+      "step": 4273
+    },
+    {
+      "epoch": 48.54957507082153,
+      "grad_norm": 4.665986699062646,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 733897512,
+      "step": 4274
+    },
+    {
+      "epoch": 48.54957507082153,
+      "loss": 0.14283835887908936,
+      "loss_ce": 0.003388291457667947,
+      "loss_iou": 0.390625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 733897512,
+      "step": 4274
+    },
+    {
+      "epoch": 48.56090651558074,
+      "grad_norm": 3.9022350004449065,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 734069692,
+      "step": 4275
+    },
+    {
+      "epoch": 48.56090651558074,
+      "loss": 0.06221405044198036,
+      "loss_ce": 0.007511289790272713,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 734069692,
+      "step": 4275
+    },
+    {
+      "epoch": 48.57223796033994,
+      "grad_norm": 3.680592793458854,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 734241692,
+      "step": 4276
+    },
+    {
+      "epoch": 48.57223796033994,
+      "loss": 0.0740753561258316,
+      "loss_ce": 0.0030151798855513334,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 734241692,
+      "step": 4276
+    },
+    {
+      "epoch": 48.58356940509915,
+      "grad_norm": 4.396541167149029,
+      "learning_rate": 5e-06,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 734413692,
+      "step": 4277
+    },
+    {
+      "epoch": 48.58356940509915,
+      "loss": 0.10261402279138565,
+      "loss_ce": 0.0065752062946558,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 734413692,
+      "step": 4277
+    },
+    {
+      "epoch": 48.594900849858355,
+      "grad_norm": 10.750248819532924,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 734585328,
+      "step": 4278
+    },
+    {
+      "epoch": 48.594900849858355,
+      "loss": 0.11873096972703934,
+      "loss_ce": 0.010576676577329636,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 734585328,
+      "step": 4278
+    },
+    {
+      "epoch": 48.60623229461756,
+      "grad_norm": 4.3787436246014275,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 734757128,
+      "step": 4279
+    },
+    {
+      "epoch": 48.60623229461756,
+      "loss": 0.08168354630470276,
+      "loss_ce": 0.007190130650997162,
+      "loss_iou": 0.46875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 734757128,
+      "step": 4279
+    },
+    {
+      "epoch": 48.61756373937677,
+      "grad_norm": 15.81769423746986,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 734928204,
+      "step": 4280
+    },
+    {
+      "epoch": 48.61756373937677,
+      "loss": 0.0950947254896164,
+      "loss_ce": 0.0025043943896889687,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 734928204,
+      "step": 4280
+    },
+    {
+      "epoch": 48.62889518413598,
+      "grad_norm": 4.1727358080254975,
+      "learning_rate": 5e-06,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 735099996,
+      "step": 4281
+    },
+    {
+      "epoch": 48.62889518413598,
+      "loss": 0.04747691750526428,
+      "loss_ce": 0.0012122668558731675,
+      "loss_iou": 0.53125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 735099996,
+      "step": 4281
+    },
+    {
+      "epoch": 48.640226628895185,
+      "grad_norm": 3.481485991237219,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 735270788,
+      "step": 4282
+    },
+    {
+      "epoch": 48.640226628895185,
+      "loss": 0.11616332828998566,
+      "loss_ce": 0.006788330618292093,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 735270788,
+      "step": 4282
+    },
+    {
+      "epoch": 48.65155807365439,
+      "grad_norm": 3.9990412778699294,
+      "learning_rate": 5e-06,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 735442308,
+      "step": 4283
+    },
+    {
+      "epoch": 48.65155807365439,
+      "loss": 0.10802154242992401,
+      "loss_ce": 0.004170226398855448,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 735442308,
+      "step": 4283
+    },
+    {
+      "epoch": 48.6628895184136,
+      "grad_norm": 3.6806368462051027,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 735614292,
+      "step": 4284
+    },
+    {
+      "epoch": 48.6628895184136,
+      "loss": 0.14905351400375366,
+      "loss_ce": 0.003728807670995593,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 735614292,
+      "step": 4284
+    },
+    {
+      "epoch": 48.67422096317281,
+      "grad_norm": 4.014967665069057,
+      "learning_rate": 5e-06,
+      "loss": 0.113,
+      "num_input_tokens_seen": 735786300,
+      "step": 4285
+    },
+    {
+      "epoch": 48.67422096317281,
+      "loss": 0.14360037446022034,
+      "loss_ce": 0.00444022286683321,
+      "loss_iou": 0.4375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 735786300,
+      "step": 4285
+    },
+    {
+      "epoch": 48.685552407932015,
+      "grad_norm": 3.4984328046370687,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 735958224,
+      "step": 4286
+    },
+    {
+      "epoch": 48.685552407932015,
+      "loss": 0.13294550776481628,
+      "loss_ce": 0.003978218417614698,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 735958224,
+      "step": 4286
+    },
+    {
+      "epoch": 48.696883852691215,
+      "grad_norm": 6.105849937970278,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 736129968,
+      "step": 4287
+    },
+    {
+      "epoch": 48.696883852691215,
+      "loss": 0.12372542917728424,
+      "loss_ce": 0.00708724232390523,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 736129968,
+      "step": 4287
+    },
+    {
+      "epoch": 48.70821529745042,
+      "grad_norm": 3.4128048536556217,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 736301272,
+      "step": 4288
+    },
+    {
+      "epoch": 48.70821529745042,
+      "loss": 0.16212065517902374,
+      "loss_ce": 0.0028189001604914665,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 736301272,
+      "step": 4288
+    },
+    {
+      "epoch": 48.71954674220963,
+      "grad_norm": 3.558755349484977,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 736472616,
+      "step": 4289
+    },
+    {
+      "epoch": 48.71954674220963,
+      "loss": 0.069485142827034,
+      "loss_ce": 0.004085969645529985,
+      "loss_iou": 0.01220703125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 736472616,
+      "step": 4289
+    },
+    {
+      "epoch": 48.73087818696884,
+      "grad_norm": 3.4913937377942807,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 736644160,
+      "step": 4290
+    },
+    {
+      "epoch": 48.73087818696884,
+      "loss": 0.05032903328537941,
+      "loss_ce": 0.002355400938540697,
+      "loss_iou": 0.453125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 736644160,
+      "step": 4290
+    },
+    {
+      "epoch": 48.742209631728045,
+      "grad_norm": 4.529124973874452,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 736815836,
+      "step": 4291
+    },
+    {
+      "epoch": 48.742209631728045,
+      "loss": 0.08149650692939758,
+      "loss_ce": 0.007888110354542732,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 736815836,
+      "step": 4291
+    },
+    {
+      "epoch": 48.75354107648725,
+      "grad_norm": 7.255218511608639,
+      "learning_rate": 5e-06,
+      "loss": 0.098,
+      "num_input_tokens_seen": 736987580,
+      "step": 4292
+    },
+    {
+      "epoch": 48.75354107648725,
+      "loss": 0.0904083102941513,
+      "loss_ce": 0.005233755335211754,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 736987580,
+      "step": 4292
+    },
+    {
+      "epoch": 48.76487252124646,
+      "grad_norm": 4.308651704191451,
+      "learning_rate": 5e-06,
+      "loss": 0.103,
+      "num_input_tokens_seen": 737158400,
+      "step": 4293
+    },
+    {
+      "epoch": 48.76487252124646,
+      "loss": 0.07376554608345032,
+      "loss_ce": 0.005879189353436232,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 737158400,
+      "step": 4293
+    },
+    {
+      "epoch": 48.77620396600567,
+      "grad_norm": 4.27711939400556,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 737329348,
+      "step": 4294
+    },
+    {
+      "epoch": 48.77620396600567,
+      "loss": 0.11947590112686157,
+      "loss_ce": 0.0006862251902930439,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 737329348,
+      "step": 4294
+    },
+    {
+      "epoch": 48.787535410764875,
+      "grad_norm": 5.172696697527694,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 737500868,
+      "step": 4295
+    },
+    {
+      "epoch": 48.787535410764875,
+      "loss": 0.07078903168439865,
+      "loss_ce": 0.007938079535961151,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 737500868,
+      "step": 4295
+    },
+    {
+      "epoch": 48.79886685552408,
+      "grad_norm": 3.61832815781928,
+      "learning_rate": 5e-06,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 737672476,
+      "step": 4296
+    },
+    {
+      "epoch": 48.79886685552408,
+      "loss": 0.11403404176235199,
+      "loss_ce": 0.0027364278212189674,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 737672476,
+      "step": 4296
+    },
+    {
+      "epoch": 48.81019830028329,
+      "grad_norm": 3.7446788983069093,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 737844208,
+      "step": 4297
+    },
+    {
+      "epoch": 48.81019830028329,
+      "loss": 0.07033340632915497,
+      "loss_ce": 0.00015060264558997005,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 737844208,
+      "step": 4297
+    },
+    {
+      "epoch": 48.82152974504249,
+      "grad_norm": 3.962563660047595,
+      "learning_rate": 5e-06,
+      "loss": 0.1593,
+      "num_input_tokens_seen": 738015844,
+      "step": 4298
+    },
+    {
+      "epoch": 48.82152974504249,
+      "loss": 0.19637420773506165,
+      "loss_ce": 0.0013058530166745186,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 738015844,
+      "step": 4298
+    },
+    {
+      "epoch": 48.8328611898017,
+      "grad_norm": 3.696844458100421,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 738186788,
+      "step": 4299
+    },
+    {
+      "epoch": 48.8328611898017,
+      "loss": 0.08086097240447998,
+      "loss_ce": 0.004719617776572704,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 738186788,
+      "step": 4299
+    },
+    {
+      "epoch": 48.844192634560905,
+      "grad_norm": 3.8948833797307976,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 738358256,
+      "step": 4300
+    },
+    {
+      "epoch": 48.844192634560905,
+      "loss": 0.17816084623336792,
+      "loss_ce": 0.0019828733056783676,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 738358256,
+      "step": 4300
+    },
+    {
+      "epoch": 48.85552407932011,
+      "grad_norm": 5.191495325960047,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 738528936,
+      "step": 4301
+    },
+    {
+      "epoch": 48.85552407932011,
+      "loss": 0.15114825963974,
+      "loss_ce": 0.0016121244989335537,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 738528936,
+      "step": 4301
+    },
+    {
+      "epoch": 48.86685552407932,
+      "grad_norm": 8.281709804459151,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 738699236,
+      "step": 4302
+    },
+    {
+      "epoch": 48.86685552407932,
+      "loss": 0.12180070579051971,
+      "loss_ce": 0.0003102283226326108,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 738699236,
+      "step": 4302
+    },
+    {
+      "epoch": 48.87818696883853,
+      "grad_norm": 4.655570414645217,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 738870800,
+      "step": 4303
+    },
+    {
+      "epoch": 48.87818696883853,
+      "loss": 0.10928734391927719,
+      "loss_ce": 0.003643119940534234,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 738870800,
+      "step": 4303
+    },
+    {
+      "epoch": 48.889518413597735,
+      "grad_norm": 3.1580637596584493,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 739042524,
+      "step": 4304
+    },
+    {
+      "epoch": 48.889518413597735,
+      "loss": 0.08541439473628998,
+      "loss_ce": 0.00014828561688773334,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 739042524,
+      "step": 4304
+    },
+    {
+      "epoch": 48.90084985835694,
+      "grad_norm": 3.7153373827408664,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 739214492,
+      "step": 4305
+    },
+    {
+      "epoch": 48.90084985835694,
+      "loss": 0.05097080022096634,
+      "loss_ce": 0.0017001696396619081,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 739214492,
+      "step": 4305
+    },
+    {
+      "epoch": 48.91218130311615,
+      "grad_norm": 3.910928125887995,
+      "learning_rate": 5e-06,
+      "loss": 0.1143,
+      "num_input_tokens_seen": 739386360,
+      "step": 4306
+    },
+    {
+      "epoch": 48.91218130311615,
+      "loss": 0.09835060685873032,
+      "loss_ce": 0.0038987009320408106,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 739386360,
+      "step": 4306
+    },
+    {
+      "epoch": 48.92351274787536,
+      "grad_norm": 3.83367633078769,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 739558112,
+      "step": 4307
+    },
+    {
+      "epoch": 48.92351274787536,
+      "loss": 0.06259407103061676,
+      "loss_ce": 0.004427570849657059,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 739558112,
+      "step": 4307
+    },
+    {
+      "epoch": 48.934844192634564,
+      "grad_norm": 3.5539803874064875,
+      "learning_rate": 5e-06,
+      "loss": 0.0791,
+      "num_input_tokens_seen": 739729436,
+      "step": 4308
+    },
+    {
+      "epoch": 48.934844192634564,
+      "loss": 0.05807384476065636,
+      "loss_ce": 0.012266959995031357,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 739729436,
+      "step": 4308
+    },
+    {
+      "epoch": 48.946175637393765,
+      "grad_norm": 3.675091285750069,
+      "learning_rate": 5e-06,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 739901028,
+      "step": 4309
+    },
+    {
+      "epoch": 48.946175637393765,
+      "loss": 0.04846800118684769,
+      "loss_ce": 0.0020965421572327614,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 739901028,
+      "step": 4309
+    },
+    {
+      "epoch": 48.95750708215297,
+      "grad_norm": 4.2305004566189455,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 740073252,
+      "step": 4310
+    },
+    {
+      "epoch": 48.95750708215297,
+      "loss": 0.04841400310397148,
+      "loss_ce": 0.002698671305552125,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 740073252,
+      "step": 4310
+    },
+    {
+      "epoch": 48.96883852691218,
+      "grad_norm": 3.9474841478106177,
+      "learning_rate": 5e-06,
+      "loss": 0.0879,
+      "num_input_tokens_seen": 740244172,
+      "step": 4311
+    },
+    {
+      "epoch": 48.96883852691218,
+      "loss": 0.07899753749370575,
+      "loss_ce": 0.005328100174665451,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 740244172,
+      "step": 4311
+    },
+    {
+      "epoch": 48.98016997167139,
+      "grad_norm": 4.575526650111768,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 740415300,
+      "step": 4312
+    },
+    {
+      "epoch": 48.98016997167139,
+      "loss": 0.0887562558054924,
+      "loss_ce": 0.001857450231909752,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 740415300,
+      "step": 4312
+    },
+    {
+      "epoch": 48.991501416430594,
+      "grad_norm": 3.652875275513445,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 740585464,
+      "step": 4313
+    },
+    {
+      "epoch": 48.991501416430594,
+      "loss": 0.10209809243679047,
+      "loss_ce": 0.0025497477035969496,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 740585464,
+      "step": 4313
+    },
+    {
+      "epoch": 48.991501416430594,
+      "loss": 0.0529276579618454,
+      "loss_ce": 0.0027567625511437654,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 740714468,
+      "step": 4313
+    },
+    {
+      "epoch": 49.0028328611898,
+      "grad_norm": 3.4731509067808997,
+      "learning_rate": 5e-06,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 740757324,
+      "step": 4314
+    },
+    {
+      "epoch": 49.0028328611898,
+      "loss": 0.07634000480175018,
+      "loss_ce": 0.0008089952752925456,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 740757324,
+      "step": 4314
+    },
+    {
+      "epoch": 49.01416430594901,
+      "grad_norm": 3.226841820356901,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 740925352,
+      "step": 4315
+    },
+    {
+      "epoch": 49.01416430594901,
+      "loss": 0.17698752880096436,
+      "loss_ce": 0.001114727696403861,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 740925352,
+      "step": 4315
+    },
+    {
+      "epoch": 49.02549575070822,
+      "grad_norm": 3.4968279206823474,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 741097600,
+      "step": 4316
+    },
+    {
+      "epoch": 49.02549575070822,
+      "loss": 0.074805349111557,
+      "loss_ce": 0.0007849633693695068,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 741097600,
+      "step": 4316
+    },
+    {
+      "epoch": 49.036827195467424,
+      "grad_norm": 3.030572576311112,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 741269636,
+      "step": 4317
+    },
+    {
+      "epoch": 49.036827195467424,
+      "loss": 0.05827585235238075,
+      "loss_ce": 0.0028254101052880287,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 741269636,
+      "step": 4317
+    },
+    {
+      "epoch": 49.04815864022663,
+      "grad_norm": 3.5442837172108175,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 741441176,
+      "step": 4318
+    },
+    {
+      "epoch": 49.04815864022663,
+      "loss": 0.05447140708565712,
+      "loss_ce": 0.0011266814544796944,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 741441176,
+      "step": 4318
+    },
+    {
+      "epoch": 49.05949008498584,
+      "grad_norm": 3.4420791807746145,
+      "learning_rate": 5e-06,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 741613384,
+      "step": 4319
+    },
+    {
+      "epoch": 49.05949008498584,
+      "loss": 0.05469021946191788,
+      "loss_ce": 0.00018582388292998075,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 741613384,
+      "step": 4319
+    },
+    {
+      "epoch": 49.07082152974504,
+      "grad_norm": 3.976612078863114,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 741784896,
+      "step": 4320
+    },
+    {
+      "epoch": 49.07082152974504,
+      "loss": 0.053845614194869995,
+      "loss_ce": 0.0018436636310070753,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 741784896,
+      "step": 4320
+    },
+    {
+      "epoch": 49.08215297450425,
+      "grad_norm": 3.3151647845175933,
+      "learning_rate": 5e-06,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 741956640,
+      "step": 4321
+    },
+    {
+      "epoch": 49.08215297450425,
+      "loss": 0.05855793505907059,
+      "loss_ce": 0.0009254844626411796,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 741956640,
+      "step": 4321
+    },
+    {
+      "epoch": 49.093484419263454,
+      "grad_norm": 6.875468956913932,
+      "learning_rate": 5e-06,
+      "loss": 0.0866,
+      "num_input_tokens_seen": 742127196,
+      "step": 4322
+    },
+    {
+      "epoch": 49.093484419263454,
+      "loss": 0.1033497005701065,
+      "loss_ce": 0.0015125458594411612,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 742127196,
+      "step": 4322
+    },
+    {
+      "epoch": 49.10481586402266,
+      "grad_norm": 4.227861006491061,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 742299164,
+      "step": 4323
+    },
+    {
+      "epoch": 49.10481586402266,
+      "loss": 0.055766358971595764,
+      "loss_ce": 0.0007584268460050225,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 742299164,
+      "step": 4323
+    },
+    {
+      "epoch": 49.11614730878187,
+      "grad_norm": 4.263018981776261,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 742469196,
+      "step": 4324
+    },
+    {
+      "epoch": 49.11614730878187,
+      "loss": 0.11103181540966034,
+      "loss_ce": 0.0019009571988135576,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 742469196,
+      "step": 4324
+    },
+    {
+      "epoch": 49.12747875354108,
+      "grad_norm": 4.245028883666309,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 742641656,
+      "step": 4325
+    },
+    {
+      "epoch": 49.12747875354108,
+      "loss": 0.1197550892829895,
+      "loss_ce": 0.0015299873193725944,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 742641656,
+      "step": 4325
+    },
+    {
+      "epoch": 49.138810198300284,
+      "grad_norm": 3.2887527616888583,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 742813156,
+      "step": 4326
+    },
+    {
+      "epoch": 49.138810198300284,
+      "loss": 0.07255055010318756,
+      "loss_ce": 0.0002544056042097509,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 742813156,
+      "step": 4326
+    },
+    {
+      "epoch": 49.15014164305949,
+      "grad_norm": 2.338941762486456,
+      "learning_rate": 5e-06,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 742984864,
+      "step": 4327
+    },
+    {
+      "epoch": 49.15014164305949,
+      "loss": 0.04754868894815445,
+      "loss_ce": 0.004213727544993162,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 742984864,
+      "step": 4327
+    },
+    {
+      "epoch": 49.1614730878187,
+      "grad_norm": 2.786666687353388,
+      "learning_rate": 5e-06,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 743156400,
+      "step": 4328
+    },
+    {
+      "epoch": 49.1614730878187,
+      "loss": 0.046610768884420395,
+      "loss_ce": 0.002360280603170395,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 743156400,
+      "step": 4328
+    },
+    {
+      "epoch": 49.172804532577906,
+      "grad_norm": 4.438183649509324,
+      "learning_rate": 5e-06,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 743327584,
+      "step": 4329
+    },
+    {
+      "epoch": 49.172804532577906,
+      "loss": 0.13865070044994354,
+      "loss_ce": 0.000741772644687444,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 743327584,
+      "step": 4329
+    },
+    {
+      "epoch": 49.184135977337114,
+      "grad_norm": 3.073314635685493,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 743498260,
+      "step": 4330
+    },
+    {
+      "epoch": 49.184135977337114,
+      "loss": 0.0806737095117569,
+      "loss_ce": 0.001450078096240759,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 743498260,
+      "step": 4330
+    },
+    {
+      "epoch": 49.195467422096314,
+      "grad_norm": 3.8487355782874886,
+      "learning_rate": 5e-06,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 743667916,
+      "step": 4331
+    },
+    {
+      "epoch": 49.195467422096314,
+      "loss": 0.07400716096162796,
+      "loss_ce": 0.0029317219741642475,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 743667916,
+      "step": 4331
+    },
+    {
+      "epoch": 49.20679886685552,
+      "grad_norm": 6.851903874822369,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 743838532,
+      "step": 4332
+    },
+    {
+      "epoch": 49.20679886685552,
+      "loss": 0.10344220697879791,
+      "loss_ce": 0.0023374767042696476,
+      "loss_iou": 0.46875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 743838532,
+      "step": 4332
+    },
+    {
+      "epoch": 49.21813031161473,
+      "grad_norm": 6.1283737427026415,
+      "learning_rate": 5e-06,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 744010164,
+      "step": 4333
+    },
+    {
+      "epoch": 49.21813031161473,
+      "loss": 0.06122565641999245,
+      "loss_ce": 0.0026319031603634357,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 744010164,
+      "step": 4333
+    },
+    {
+      "epoch": 49.22946175637394,
+      "grad_norm": 5.268752635010931,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 744181836,
+      "step": 4334
+    },
+    {
+      "epoch": 49.22946175637394,
+      "loss": 0.05788575857877731,
+      "loss_ce": 0.0014587538316845894,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 744181836,
+      "step": 4334
+    },
+    {
+      "epoch": 49.240793201133144,
+      "grad_norm": 4.210641814112757,
+      "learning_rate": 5e-06,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 744353392,
+      "step": 4335
+    },
+    {
+      "epoch": 49.240793201133144,
+      "loss": 0.09317013621330261,
+      "loss_ce": 0.0015716226771473885,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 744353392,
+      "step": 4335
+    },
+    {
+      "epoch": 49.25212464589235,
+      "grad_norm": 4.5924333386773055,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 744525052,
+      "step": 4336
+    },
+    {
+      "epoch": 49.25212464589235,
+      "loss": 0.06469232589006424,
+      "loss_ce": 0.0009411001810804009,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 744525052,
+      "step": 4336
+    },
+    {
+      "epoch": 49.26345609065156,
+      "grad_norm": 17.098168190024793,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 744694528,
+      "step": 4337
+    },
+    {
+      "epoch": 49.26345609065156,
+      "loss": 0.09466172009706497,
+      "loss_ce": 0.0008201682940125465,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 744694528,
+      "step": 4337
+    },
+    {
+      "epoch": 49.274787535410766,
+      "grad_norm": 4.185098376233496,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 744865708,
+      "step": 4338
+    },
+    {
+      "epoch": 49.274787535410766,
+      "loss": 0.06628775596618652,
+      "loss_ce": 0.002338170539587736,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 744865708,
+      "step": 4338
+    },
+    {
+      "epoch": 49.286118980169974,
+      "grad_norm": 3.888527606834987,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 745037548,
+      "step": 4339
+    },
+    {
+      "epoch": 49.286118980169974,
+      "loss": 0.08603458106517792,
+      "loss_ce": 0.0010889050317928195,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 745037548,
+      "step": 4339
+    },
+    {
+      "epoch": 49.29745042492918,
+      "grad_norm": 3.865975312395479,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 745209432,
+      "step": 4340
+    },
+    {
+      "epoch": 49.29745042492918,
+      "loss": 0.05419117584824562,
+      "loss_ce": 0.0012736956123262644,
+      "loss_iou": 0.5,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 745209432,
+      "step": 4340
+    },
+    {
+      "epoch": 49.30878186968839,
+      "grad_norm": 3.641625553205517,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 745381400,
+      "step": 4341
+    },
+    {
+      "epoch": 49.30878186968839,
+      "loss": 0.06112079694867134,
+      "loss_ce": 0.0028627419378608465,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 745381400,
+      "step": 4341
+    },
+    {
+      "epoch": 49.32011331444759,
+      "grad_norm": 5.268422727416257,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 745553100,
+      "step": 4342
+    },
+    {
+      "epoch": 49.32011331444759,
+      "loss": 0.13542711734771729,
+      "loss_ce": 0.004598258063197136,
+      "loss_iou": 0.265625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 745553100,
+      "step": 4342
+    },
+    {
+      "epoch": 49.331444759206796,
+      "grad_norm": 2.9015070743611715,
+      "learning_rate": 5e-06,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 745725268,
+      "step": 4343
+    },
+    {
+      "epoch": 49.331444759206796,
+      "loss": 0.06892704218626022,
+      "loss_ce": 0.000979652046225965,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 745725268,
+      "step": 4343
+    },
+    {
+      "epoch": 49.342776203966004,
+      "grad_norm": 4.031799282163996,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 745896452,
+      "step": 4344
+    },
+    {
+      "epoch": 49.342776203966004,
+      "loss": 0.09716005623340607,
+      "loss_ce": 0.0005261427140794694,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 745896452,
+      "step": 4344
+    },
+    {
+      "epoch": 49.35410764872521,
+      "grad_norm": 3.745710628841776,
+      "learning_rate": 5e-06,
+      "loss": 0.0815,
+      "num_input_tokens_seen": 746068412,
+      "step": 4345
+    },
+    {
+      "epoch": 49.35410764872521,
+      "loss": 0.08749920129776001,
+      "loss_ce": 0.0026603350415825844,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 746068412,
+      "step": 4345
+    },
+    {
+      "epoch": 49.36543909348442,
+      "grad_norm": 4.780133073611569,
+      "learning_rate": 5e-06,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 746239956,
+      "step": 4346
+    },
+    {
+      "epoch": 49.36543909348442,
+      "loss": 0.06789717078208923,
+      "loss_ce": 0.003260944038629532,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 746239956,
+      "step": 4346
+    },
+    {
+      "epoch": 49.376770538243626,
+      "grad_norm": 3.997057887700053,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 746411000,
+      "step": 4347
+    },
+    {
+      "epoch": 49.376770538243626,
+      "loss": 0.08519040793180466,
+      "loss_ce": 0.000809304416179657,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 746411000,
+      "step": 4347
+    },
+    {
+      "epoch": 49.388101983002834,
+      "grad_norm": 4.043479665739765,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 746582804,
+      "step": 4348
+    },
+    {
+      "epoch": 49.388101983002834,
+      "loss": 0.06669161468744278,
+      "loss_ce": 0.0005600225413218141,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 746582804,
+      "step": 4348
+    },
+    {
+      "epoch": 49.39943342776204,
+      "grad_norm": 4.625509585538065,
+      "learning_rate": 5e-06,
+      "loss": 0.099,
+      "num_input_tokens_seen": 746754476,
+      "step": 4349
+    },
+    {
+      "epoch": 49.39943342776204,
+      "loss": 0.19054612517356873,
+      "loss_ce": 0.0006352471536956728,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 746754476,
+      "step": 4349
+    },
+    {
+      "epoch": 49.41076487252125,
+      "grad_norm": 4.126685897996752,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 746926148,
+      "step": 4350
+    },
+    {
+      "epoch": 49.41076487252125,
+      "loss": 0.0921538919210434,
+      "loss_ce": 0.0005706397932954133,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 746926148,
+      "step": 4350
+    },
+    {
+      "epoch": 49.422096317280456,
+      "grad_norm": 4.8060166306817695,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 747095448,
+      "step": 4351
+    },
+    {
+      "epoch": 49.422096317280456,
+      "loss": 0.07864593714475632,
+      "loss_ce": 0.00299286306835711,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 747095448,
+      "step": 4351
+    },
+    {
+      "epoch": 49.43342776203966,
+      "grad_norm": 4.4387311828724005,
+      "learning_rate": 5e-06,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 747267176,
+      "step": 4352
+    },
+    {
+      "epoch": 49.43342776203966,
+      "loss": 0.05372954159975052,
+      "loss_ce": 0.0015139628667384386,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 747267176,
+      "step": 4352
+    },
+    {
+      "epoch": 49.444759206798864,
+      "grad_norm": 5.250254257051551,
+      "learning_rate": 5e-06,
+      "loss": 0.087,
+      "num_input_tokens_seen": 747438600,
+      "step": 4353
+    },
+    {
+      "epoch": 49.444759206798864,
+      "loss": 0.10443614423274994,
+      "loss_ce": 0.0013477625325322151,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 747438600,
+      "step": 4353
+    },
+    {
+      "epoch": 49.45609065155807,
+      "grad_norm": 4.095399647514437,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 747610484,
+      "step": 4354
+    },
+    {
+      "epoch": 49.45609065155807,
+      "loss": 0.08993314206600189,
+      "loss_ce": 0.0015237231273204088,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 747610484,
+      "step": 4354
+    },
+    {
+      "epoch": 49.46742209631728,
+      "grad_norm": 3.581291881022985,
+      "learning_rate": 5e-06,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 747781732,
+      "step": 4355
+    },
+    {
+      "epoch": 49.46742209631728,
+      "loss": 0.10985402762889862,
+      "loss_ce": 0.0005705712828785181,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 747781732,
+      "step": 4355
+    },
+    {
+      "epoch": 49.478753541076486,
+      "grad_norm": 9.460265397579896,
+      "learning_rate": 5e-06,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 747952732,
+      "step": 4356
+    },
+    {
+      "epoch": 49.478753541076486,
+      "loss": 0.03622867912054062,
+      "loss_ce": 0.0018964068731293082,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 747952732,
+      "step": 4356
+    },
+    {
+      "epoch": 49.49008498583569,
+      "grad_norm": 4.4833623848144875,
+      "learning_rate": 5e-06,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 748122908,
+      "step": 4357
+    },
+    {
+      "epoch": 49.49008498583569,
+      "loss": 0.06123455986380577,
+      "loss_ce": 0.003159608691930771,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 748122908,
+      "step": 4357
+    },
+    {
+      "epoch": 49.5014164305949,
+      "grad_norm": 5.778599854687138,
+      "learning_rate": 5e-06,
+      "loss": 0.0926,
+      "num_input_tokens_seen": 748294016,
+      "step": 4358
+    },
+    {
+      "epoch": 49.5014164305949,
+      "loss": 0.08962063491344452,
+      "loss_ce": 0.0027065714821219444,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 748294016,
+      "step": 4358
+    },
+    {
+      "epoch": 49.51274787535411,
+      "grad_norm": 4.351421035238239,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 748464184,
+      "step": 4359
+    },
+    {
+      "epoch": 49.51274787535411,
+      "loss": 0.05633925646543503,
+      "loss_ce": 0.002109519438818097,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 748464184,
+      "step": 4359
+    },
+    {
+      "epoch": 49.524079320113316,
+      "grad_norm": 5.106061808874131,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 748635376,
+      "step": 4360
+    },
+    {
+      "epoch": 49.524079320113316,
+      "loss": 0.06278373301029205,
+      "loss_ce": 0.001870643813163042,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 748635376,
+      "step": 4360
+    },
+    {
+      "epoch": 49.53541076487252,
+      "grad_norm": 3.813947211152109,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 748806940,
+      "step": 4361
+    },
+    {
+      "epoch": 49.53541076487252,
+      "loss": 0.10220496356487274,
+      "loss_ce": 0.0005509055918082595,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 748806940,
+      "step": 4361
+    },
+    {
+      "epoch": 49.54674220963173,
+      "grad_norm": 2.7406397459626497,
+      "learning_rate": 5e-06,
+      "loss": 0.068,
+      "num_input_tokens_seen": 748977128,
+      "step": 4362
+    },
+    {
+      "epoch": 49.54674220963173,
+      "loss": 0.05096760392189026,
+      "loss_ce": 0.001483348896726966,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 748977128,
+      "step": 4362
+    },
+    {
+      "epoch": 49.55807365439094,
+      "grad_norm": 3.572300868636267,
+      "learning_rate": 5e-06,
+      "loss": 0.0858,
+      "num_input_tokens_seen": 749148932,
+      "step": 4363
+    },
+    {
+      "epoch": 49.55807365439094,
+      "loss": 0.10987602174282074,
+      "loss_ce": 0.002179492497816682,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 749148932,
+      "step": 4363
+    },
+    {
+      "epoch": 49.56940509915014,
+      "grad_norm": 4.609996131348387,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 749320524,
+      "step": 4364
+    },
+    {
+      "epoch": 49.56940509915014,
+      "loss": 0.06587588042020798,
+      "loss_ce": 0.00015627789252903312,
+      "loss_iou": 0.0146484375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 749320524,
+      "step": 4364
+    },
+    {
+      "epoch": 49.580736543909346,
+      "grad_norm": 3.7454499124671576,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 749492104,
+      "step": 4365
+    },
+    {
+      "epoch": 49.580736543909346,
+      "loss": 0.10872483253479004,
+      "loss_ce": 0.001699693500995636,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 749492104,
+      "step": 4365
+    },
+    {
+      "epoch": 49.59206798866855,
+      "grad_norm": 4.217173046323239,
+      "learning_rate": 5e-06,
+      "loss": 0.0951,
+      "num_input_tokens_seen": 749664056,
+      "step": 4366
+    },
+    {
+      "epoch": 49.59206798866855,
+      "loss": 0.06376250088214874,
+      "loss_ce": 0.0031240719836205244,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 749664056,
+      "step": 4366
+    },
+    {
+      "epoch": 49.60339943342776,
+      "grad_norm": 7.324372451364367,
+      "learning_rate": 5e-06,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 749834768,
+      "step": 4367
+    },
+    {
+      "epoch": 49.60339943342776,
+      "loss": 0.07092300057411194,
+      "loss_ce": 0.0026704370975494385,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 749834768,
+      "step": 4367
+    },
+    {
+      "epoch": 49.61473087818697,
+      "grad_norm": 4.111071181194956,
+      "learning_rate": 5e-06,
+      "loss": 0.0918,
+      "num_input_tokens_seen": 750004936,
+      "step": 4368
+    },
+    {
+      "epoch": 49.61473087818697,
+      "loss": 0.04804462194442749,
+      "loss_ce": 0.0020088544115424156,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 750004936,
+      "step": 4368
+    },
+    {
+      "epoch": 49.626062322946176,
+      "grad_norm": 4.19101592959029,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 750176432,
+      "step": 4369
+    },
+    {
+      "epoch": 49.626062322946176,
+      "loss": 0.10773782432079315,
+      "loss_ce": 0.0014145805034786463,
+      "loss_iou": 0.46875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 750176432,
+      "step": 4369
+    },
+    {
+      "epoch": 49.63739376770538,
+      "grad_norm": 5.022033745972118,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 750348180,
+      "step": 4370
+    },
+    {
+      "epoch": 49.63739376770538,
+      "loss": 0.05284727364778519,
+      "loss_ce": 0.0014861896634101868,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 750348180,
+      "step": 4370
+    },
+    {
+      "epoch": 49.64872521246459,
+      "grad_norm": 3.209276110993788,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 750519856,
+      "step": 4371
+    },
+    {
+      "epoch": 49.64872521246459,
+      "loss": 0.039785709232091904,
+      "loss_ce": 0.0006621736101806164,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 750519856,
+      "step": 4371
+    },
+    {
+      "epoch": 49.6600566572238,
+      "grad_norm": 2.9342217290281902,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 750690804,
+      "step": 4372
+    },
+    {
+      "epoch": 49.6600566572238,
+      "loss": 0.04395214468240738,
+      "loss_ce": 0.0007087374688126147,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 750690804,
+      "step": 4372
+    },
+    {
+      "epoch": 49.671388101983005,
+      "grad_norm": 3.414709944889579,
+      "learning_rate": 5e-06,
+      "loss": 0.0717,
+      "num_input_tokens_seen": 750862236,
+      "step": 4373
+    },
+    {
+      "epoch": 49.671388101983005,
+      "loss": 0.06521345674991608,
+      "loss_ce": 0.003506913548335433,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 750862236,
+      "step": 4373
+    },
+    {
+      "epoch": 49.68271954674221,
+      "grad_norm": 3.791969349259422,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 751034176,
+      "step": 4374
+    },
+    {
+      "epoch": 49.68271954674221,
+      "loss": 0.13955862820148468,
+      "loss_ce": 0.0011614140821620822,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 751034176,
+      "step": 4374
+    },
+    {
+      "epoch": 49.69405099150141,
+      "grad_norm": 5.4588785586855675,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 751205332,
+      "step": 4375
+    },
+    {
+      "epoch": 49.69405099150141,
+      "loss": 0.11089522391557693,
+      "loss_ce": 0.0010929773561656475,
+      "loss_iou": 0.3125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 751205332,
+      "step": 4375
+    },
+    {
+      "epoch": 49.70538243626062,
+      "grad_norm": 7.065185276228404,
+      "learning_rate": 5e-06,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 751376916,
+      "step": 4376
+    },
+    {
+      "epoch": 49.70538243626062,
+      "loss": 0.11830483376979828,
+      "loss_ce": 0.001117331557907164,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 751376916,
+      "step": 4376
+    },
+    {
+      "epoch": 49.71671388101983,
+      "grad_norm": 7.133288924907278,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 751548420,
+      "step": 4377
+    },
+    {
+      "epoch": 49.71671388101983,
+      "loss": 0.2544563412666321,
+      "loss_ce": 0.00918656773865223,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 751548420,
+      "step": 4377
+    },
+    {
+      "epoch": 49.728045325779036,
+      "grad_norm": 5.21335164761946,
+      "learning_rate": 5e-06,
+      "loss": 0.0662,
+      "num_input_tokens_seen": 751720072,
+      "step": 4378
+    },
+    {
+      "epoch": 49.728045325779036,
+      "loss": 0.06574045121669769,
+      "loss_ce": 0.003942357841879129,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 751720072,
+      "step": 4378
+    },
+    {
+      "epoch": 49.73937677053824,
+      "grad_norm": 4.213015836699212,
+      "learning_rate": 5e-06,
+      "loss": 0.1556,
+      "num_input_tokens_seen": 751891104,
+      "step": 4379
+    },
+    {
+      "epoch": 49.73937677053824,
+      "loss": 0.17926564812660217,
+      "loss_ce": 0.010137236677110195,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 751891104,
+      "step": 4379
+    },
+    {
+      "epoch": 49.75070821529745,
+      "grad_norm": 5.515384881179111,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 752061492,
+      "step": 4380
+    },
+    {
+      "epoch": 49.75070821529745,
+      "loss": 0.05599920079112053,
+      "loss_ce": 0.000350397516740486,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 752061492,
+      "step": 4380
+    },
+    {
+      "epoch": 49.76203966005666,
+      "grad_norm": 5.509663136072773,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 752233440,
+      "step": 4381
+    },
+    {
+      "epoch": 49.76203966005666,
+      "loss": 0.10660643875598907,
+      "loss_ce": 0.004342036787420511,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 752233440,
+      "step": 4381
+    },
+    {
+      "epoch": 49.773371104815865,
+      "grad_norm": 4.026326589886566,
+      "learning_rate": 5e-06,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 752405320,
+      "step": 4382
+    },
+    {
+      "epoch": 49.773371104815865,
+      "loss": 0.15437199175357819,
+      "loss_ce": 0.0006854669190943241,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 752405320,
+      "step": 4382
+    },
+    {
+      "epoch": 49.78470254957507,
+      "grad_norm": 3.6386829417651048,
+      "learning_rate": 5e-06,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 752577152,
+      "step": 4383
+    },
+    {
+      "epoch": 49.78470254957507,
+      "loss": 0.05102023482322693,
+      "loss_ce": 0.0008188205538317561,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 752577152,
+      "step": 4383
+    },
+    {
+      "epoch": 49.79603399433428,
+      "grad_norm": 4.207026555063898,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 752745676,
+      "step": 4384
+    },
+    {
+      "epoch": 49.79603399433428,
+      "loss": 0.0880984216928482,
+      "loss_ce": 0.0020541041158139706,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 752745676,
+      "step": 4384
+    },
+    {
+      "epoch": 49.80736543909349,
+      "grad_norm": 3.2988619522199736,
+      "learning_rate": 5e-06,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 752916992,
+      "step": 4385
+    },
+    {
+      "epoch": 49.80736543909349,
+      "loss": 0.10638163983821869,
+      "loss_ce": 0.0005466811126098037,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 752916992,
+      "step": 4385
+    },
+    {
+      "epoch": 49.81869688385269,
+      "grad_norm": 4.523919041771532,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 753087748,
+      "step": 4386
+    },
+    {
+      "epoch": 49.81869688385269,
+      "loss": 0.07070009410381317,
+      "loss_ce": 0.001181046012789011,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 753087748,
+      "step": 4386
+    },
+    {
+      "epoch": 49.830028328611895,
+      "grad_norm": 3.5139723443398476,
+      "learning_rate": 5e-06,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 753259584,
+      "step": 4387
+    },
+    {
+      "epoch": 49.830028328611895,
+      "loss": 0.03835410624742508,
+      "loss_ce": 0.0027629798278212547,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 753259584,
+      "step": 4387
+    },
+    {
+      "epoch": 49.8413597733711,
+      "grad_norm": 3.5661981556230766,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 753431112,
+      "step": 4388
+    },
+    {
+      "epoch": 49.8413597733711,
+      "loss": 0.06541304290294647,
+      "loss_ce": 0.0003800879349000752,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 753431112,
+      "step": 4388
+    },
+    {
+      "epoch": 49.85269121813031,
+      "grad_norm": 2.855531912963118,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 753602968,
+      "step": 4389
+    },
+    {
+      "epoch": 49.85269121813031,
+      "loss": 0.28580358624458313,
+      "loss_ce": 0.0006778663373552263,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 753602968,
+      "step": 4389
+    },
+    {
+      "epoch": 49.86402266288952,
+      "grad_norm": 5.537335755281551,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 753775036,
+      "step": 4390
+    },
+    {
+      "epoch": 49.86402266288952,
+      "loss": 0.043919578194618225,
+      "loss_ce": 0.0005693574785254896,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 753775036,
+      "step": 4390
+    },
+    {
+      "epoch": 49.875354107648725,
+      "grad_norm": 2.817339374794205,
+      "learning_rate": 5e-06,
+      "loss": 0.0717,
+      "num_input_tokens_seen": 753945436,
+      "step": 4391
+    },
+    {
+      "epoch": 49.875354107648725,
+      "loss": 0.05330030620098114,
+      "loss_ce": 0.002702165860682726,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 753945436,
+      "step": 4391
+    },
+    {
+      "epoch": 49.88668555240793,
+      "grad_norm": 13.38166640926914,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 754116964,
+      "step": 4392
+    },
+    {
+      "epoch": 49.88668555240793,
+      "loss": 0.09472432732582092,
+      "loss_ce": 0.0010658869286999106,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 754116964,
+      "step": 4392
+    },
+    {
+      "epoch": 49.89801699716714,
+      "grad_norm": 3.241238386214431,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 754288068,
+      "step": 4393
+    },
+    {
+      "epoch": 49.89801699716714,
+      "loss": 0.1573636829853058,
+      "loss_ce": 0.0009305835701525211,
+      "loss_iou": 0.296875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 754288068,
+      "step": 4393
+    },
+    {
+      "epoch": 49.90934844192635,
+      "grad_norm": 8.678767483883833,
+      "learning_rate": 5e-06,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 754459996,
+      "step": 4394
+    },
+    {
+      "epoch": 49.90934844192635,
+      "loss": 0.1557530164718628,
+      "loss_ce": 0.0029972770716995,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 754459996,
+      "step": 4394
+    },
+    {
+      "epoch": 49.920679886685555,
+      "grad_norm": 3.7155969428994973,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 754631660,
+      "step": 4395
+    },
+    {
+      "epoch": 49.920679886685555,
+      "loss": 0.060144711285829544,
+      "loss_ce": 0.0011847498826682568,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 754631660,
+      "step": 4395
+    },
+    {
+      "epoch": 49.93201133144476,
+      "grad_norm": 3.622786070449283,
+      "learning_rate": 5e-06,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 754803276,
+      "step": 4396
+    },
+    {
+      "epoch": 49.93201133144476,
+      "loss": 0.06250564008951187,
+      "loss_ce": 0.0029200688004493713,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 754803276,
+      "step": 4396
+    },
+    {
+      "epoch": 49.94334277620396,
+      "grad_norm": 3.868323404788726,
+      "learning_rate": 5e-06,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 754974148,
+      "step": 4397
+    },
+    {
+      "epoch": 49.94334277620396,
+      "loss": 0.04745414853096008,
+      "loss_ce": 0.0037529736291617155,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 754974148,
+      "step": 4397
+    },
+    {
+      "epoch": 49.95467422096317,
+      "grad_norm": 3.9238463410260773,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 755145496,
+      "step": 4398
+    },
+    {
+      "epoch": 49.95467422096317,
+      "loss": 0.09703067690134048,
+      "loss_ce": 0.00090030551655218,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 755145496,
+      "step": 4398
+    },
+    {
+      "epoch": 49.96600566572238,
+      "grad_norm": 29.220471541875742,
+      "learning_rate": 5e-06,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 755317220,
+      "step": 4399
+    },
+    {
+      "epoch": 49.96600566572238,
+      "loss": 0.10200240463018417,
+      "loss_ce": 0.003247522981837392,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 755317220,
+      "step": 4399
+    },
+    {
+      "epoch": 49.977337110481585,
+      "grad_norm": 3.0335491516984656,
+      "learning_rate": 5e-06,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 755488908,
+      "step": 4400
+    },
+    {
+      "epoch": 49.977337110481585,
+      "loss": 0.10426551103591919,
+      "loss_ce": 0.004991832189261913,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 755488908,
+      "step": 4400
+    },
+    {
+      "epoch": 49.98866855524079,
+      "grad_norm": 5.726212766529932,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 755659036,
+      "step": 4401
+    },
+    {
+      "epoch": 49.98866855524079,
+      "loss": 0.13855154812335968,
+      "loss_ce": 0.004381020553410053,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 755659036,
+      "step": 4401
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 3.7630255373759285,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 755830676,
+      "step": 4402
+    },
+    {
+      "epoch": 50.0,
+      "loss": 0.10462699830532074,
+      "loss_ce": 0.0004552433965727687,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 755830676,
+      "step": 4402
+    },
+    {
+      "epoch": 50.01133144475921,
+      "grad_norm": 3.3401218698445185,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 756002576,
+      "step": 4403
+    },
+    {
+      "epoch": 50.01133144475921,
+      "loss": 0.07208964228630066,
+      "loss_ce": 0.0009150212281383574,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 756002576,
+      "step": 4403
+    },
+    {
+      "epoch": 50.022662889518415,
+      "grad_norm": 18.044934729647085,
+      "learning_rate": 5e-06,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 756174064,
+      "step": 4404
+    },
+    {
+      "epoch": 50.022662889518415,
+      "loss": 0.05712079256772995,
+      "loss_ce": 0.0008158613927662373,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 756174064,
+      "step": 4404
+    },
+    {
+      "epoch": 50.03399433427762,
+      "grad_norm": 3.7195738320771667,
+      "learning_rate": 5e-06,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 756346012,
+      "step": 4405
+    },
+    {
+      "epoch": 50.03399433427762,
+      "loss": 0.06001533195376396,
+      "loss_ce": 0.0006281236419454217,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 756346012,
+      "step": 4405
+    },
+    {
+      "epoch": 50.04532577903683,
+      "grad_norm": 4.137350334768581,
+      "learning_rate": 5e-06,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 756516772,
+      "step": 4406
+    },
+    {
+      "epoch": 50.04532577903683,
+      "loss": 0.05741478130221367,
+      "loss_ce": 0.0004689824709203094,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 756516772,
+      "step": 4406
+    },
+    {
+      "epoch": 50.05665722379604,
+      "grad_norm": 3.1581671903973527,
+      "learning_rate": 5e-06,
+      "loss": 0.0644,
+      "num_input_tokens_seen": 756687528,
+      "step": 4407
+    },
+    {
+      "epoch": 50.05665722379604,
+      "loss": 0.03790707886219025,
+      "loss_ce": 0.0001415750157320872,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 756687528,
+      "step": 4407
+    },
+    {
+      "epoch": 50.06798866855524,
+      "grad_norm": 3.435931554021553,
+      "learning_rate": 5e-06,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 756859136,
+      "step": 4408
+    },
+    {
+      "epoch": 50.06798866855524,
+      "loss": 0.11430996656417847,
+      "loss_ce": 0.000509918958414346,
+      "loss_iou": 0.12353515625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 756859136,
+      "step": 4408
+    },
+    {
+      "epoch": 50.079320113314445,
+      "grad_norm": 42.06504448100528,
+      "learning_rate": 5e-06,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 757031140,
+      "step": 4409
+    },
+    {
+      "epoch": 50.079320113314445,
+      "loss": 0.13308638334274292,
+      "loss_ce": 0.0005027586012147367,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 757031140,
+      "step": 4409
+    },
+    {
+      "epoch": 50.09065155807365,
+      "grad_norm": 4.650183993877952,
+      "learning_rate": 5e-06,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 757203080,
+      "step": 4410
+    },
+    {
+      "epoch": 50.09065155807365,
+      "loss": 0.12237624078989029,
+      "loss_ce": 0.004883565474301577,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 757203080,
+      "step": 4410
+    },
+    {
+      "epoch": 50.10198300283286,
+      "grad_norm": 4.093911039131734,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 757374752,
+      "step": 4411
+    },
+    {
+      "epoch": 50.10198300283286,
+      "loss": 0.09824614971876144,
+      "loss_ce": 0.000773002568166703,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 757374752,
+      "step": 4411
+    },
+    {
+      "epoch": 50.11331444759207,
+      "grad_norm": 4.103051129307374,
+      "learning_rate": 5e-06,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 757546344,
+      "step": 4412
+    },
+    {
+      "epoch": 50.11331444759207,
+      "loss": 0.09053158015012741,
+      "loss_ce": 0.00019954868184868246,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 757546344,
+      "step": 4412
+    },
+    {
+      "epoch": 50.124645892351275,
+      "grad_norm": 4.398128082821813,
+      "learning_rate": 5e-06,
+      "loss": 0.109,
+      "num_input_tokens_seen": 757717780,
+      "step": 4413
+    },
+    {
+      "epoch": 50.124645892351275,
+      "loss": 0.16719692945480347,
+      "loss_ce": 0.0008150916546583176,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 757717780,
+      "step": 4413
+    },
+    {
+      "epoch": 50.13597733711048,
+      "grad_norm": 3.9798703964823026,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 757887744,
+      "step": 4414
+    },
+    {
+      "epoch": 50.13597733711048,
+      "loss": 0.14836016297340393,
+      "loss_ce": 0.00016679619147907943,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 757887744,
+      "step": 4414
+    },
+    {
+      "epoch": 50.14730878186969,
+      "grad_norm": 3.721527603207931,
+      "learning_rate": 5e-06,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 758058608,
+      "step": 4415
+    },
+    {
+      "epoch": 50.14730878186969,
+      "loss": 0.11124084889888763,
+      "loss_ce": 0.0026898211799561977,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 758058608,
+      "step": 4415
+    },
+    {
+      "epoch": 50.1586402266289,
+      "grad_norm": 3.9387275366860792,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 758229648,
+      "step": 4416
+    },
+    {
+      "epoch": 50.1586402266289,
+      "loss": 0.07520405948162079,
+      "loss_ce": 0.0007259119884110987,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 758229648,
+      "step": 4416
+    },
+    {
+      "epoch": 50.169971671388105,
+      "grad_norm": 4.057770523118788,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 758401496,
+      "step": 4417
+    },
+    {
+      "epoch": 50.169971671388105,
+      "loss": 0.1181391105055809,
+      "loss_ce": 0.0007990228477865458,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 758401496,
+      "step": 4417
+    },
+    {
+      "epoch": 50.18130311614731,
+      "grad_norm": 2.160843232749934,
+      "learning_rate": 5e-06,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 758573248,
+      "step": 4418
+    },
+    {
+      "epoch": 50.18130311614731,
+      "loss": 0.0316898450255394,
+      "loss_ce": 0.00024148012744262815,
+      "loss_iou": 0.4375,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 758573248,
+      "step": 4418
+    },
+    {
+      "epoch": 50.19263456090651,
+      "grad_norm": 3.7319124431378903,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 758744244,
+      "step": 4419
+    },
+    {
+      "epoch": 50.19263456090651,
+      "loss": 0.048760753124952316,
+      "loss_ce": 0.0008634141413494945,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 758744244,
+      "step": 4419
+    },
+    {
+      "epoch": 50.20396600566572,
+      "grad_norm": 5.626406562416814,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 758916052,
+      "step": 4420
+    },
+    {
+      "epoch": 50.20396600566572,
+      "loss": 0.06947594881057739,
+      "loss_ce": 0.0016506274696439505,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 758916052,
+      "step": 4420
+    },
+    {
+      "epoch": 50.21529745042493,
+      "grad_norm": 5.012386939099503,
+      "learning_rate": 5e-06,
+      "loss": 0.0762,
+      "num_input_tokens_seen": 759086092,
+      "step": 4421
+    },
+    {
+      "epoch": 50.21529745042493,
+      "loss": 0.06383313238620758,
+      "loss_ce": 0.0006922596367076039,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 759086092,
+      "step": 4421
+    },
+    {
+      "epoch": 50.226628895184135,
+      "grad_norm": 4.184912909981786,
+      "learning_rate": 5e-06,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 759257144,
+      "step": 4422
+    },
+    {
+      "epoch": 50.226628895184135,
+      "loss": 0.061709772795438766,
+      "loss_ce": 0.0011628975626081228,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 759257144,
+      "step": 4422
+    },
+    {
+      "epoch": 50.23796033994334,
+      "grad_norm": 6.187060024118792,
+      "learning_rate": 5e-06,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 759428836,
+      "step": 4423
+    },
+    {
+      "epoch": 50.23796033994334,
+      "loss": 0.06169979274272919,
+      "loss_ce": 5.4287578677758574e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 759428836,
+      "step": 4423
+    },
+    {
+      "epoch": 50.24929178470255,
+      "grad_norm": 5.0446440281313585,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 759599648,
+      "step": 4424
+    },
+    {
+      "epoch": 50.24929178470255,
+      "loss": 0.05307450145483017,
+      "loss_ce": 0.0038954231422394514,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 759599648,
+      "step": 4424
+    },
+    {
+      "epoch": 50.26062322946176,
+      "grad_norm": 4.701344706992798,
+      "learning_rate": 5e-06,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 759771620,
+      "step": 4425
+    },
+    {
+      "epoch": 50.26062322946176,
+      "loss": 0.08179698139429092,
+      "loss_ce": 0.0017798923654481769,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 759771620,
+      "step": 4425
+    },
+    {
+      "epoch": 50.271954674220964,
+      "grad_norm": 4.460396666306289,
+      "learning_rate": 5e-06,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 759943252,
+      "step": 4426
+    },
+    {
+      "epoch": 50.271954674220964,
+      "loss": 0.058770958334207535,
+      "loss_ce": 0.001184286898933351,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 759943252,
+      "step": 4426
+    },
+    {
+      "epoch": 50.28328611898017,
+      "grad_norm": 4.170328484870514,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 760113352,
+      "step": 4427
+    },
+    {
+      "epoch": 50.28328611898017,
+      "loss": 0.08763612806797028,
+      "loss_ce": 0.00044740550220012665,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 760113352,
+      "step": 4427
+    },
+    {
+      "epoch": 50.29461756373938,
+      "grad_norm": 4.403793980424108,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 760284152,
+      "step": 4428
+    },
+    {
+      "epoch": 50.29461756373938,
+      "loss": 0.21055921912193298,
+      "loss_ce": 0.0005982845323160291,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 760284152,
+      "step": 4428
+    },
+    {
+      "epoch": 50.30594900849859,
+      "grad_norm": 4.6892535238443696,
+      "learning_rate": 5e-06,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 760456148,
+      "step": 4429
+    },
+    {
+      "epoch": 50.30594900849859,
+      "loss": 0.07114341109991074,
+      "loss_ce": 0.00046470039524137974,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 760456148,
+      "step": 4429
+    },
+    {
+      "epoch": 50.31728045325779,
+      "grad_norm": 4.225301058257953,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 760627588,
+      "step": 4430
+    },
+    {
+      "epoch": 50.31728045325779,
+      "loss": 0.10689513385295868,
+      "loss_ce": 0.0011517195962369442,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 760627588,
+      "step": 4430
+    },
+    {
+      "epoch": 50.328611898016995,
+      "grad_norm": 10.011086101252994,
+      "learning_rate": 5e-06,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 760798556,
+      "step": 4431
+    },
+    {
+      "epoch": 50.328611898016995,
+      "loss": 0.061522889882326126,
+      "loss_ce": 0.0006098012090660632,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 760798556,
+      "step": 4431
+    },
+    {
+      "epoch": 50.3399433427762,
+      "grad_norm": 7.002099777056922,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 760970140,
+      "step": 4432
+    },
+    {
+      "epoch": 50.3399433427762,
+      "loss": 0.07696007192134857,
+      "loss_ce": 0.0004677576362155378,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 760970140,
+      "step": 4432
+    },
+    {
+      "epoch": 50.35127478753541,
+      "grad_norm": 4.359764552948906,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 761141964,
+      "step": 4433
+    },
+    {
+      "epoch": 50.35127478753541,
+      "loss": 0.11063601076602936,
+      "loss_ce": 0.0008185015758499503,
+      "loss_iou": 0.15625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 761141964,
+      "step": 4433
+    },
+    {
+      "epoch": 50.36260623229462,
+      "grad_norm": 3.1975418134899174,
+      "learning_rate": 5e-06,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 761312964,
+      "step": 4434
+    },
+    {
+      "epoch": 50.36260623229462,
+      "loss": 0.10568847507238388,
+      "loss_ce": 0.0009521447354927659,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 761312964,
+      "step": 4434
+    },
+    {
+      "epoch": 50.373937677053824,
+      "grad_norm": 3.7055292500600423,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 761483964,
+      "step": 4435
+    },
+    {
+      "epoch": 50.373937677053824,
+      "loss": 0.1450173705816269,
+      "loss_ce": 0.00024198158644139767,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 761483964,
+      "step": 4435
+    },
+    {
+      "epoch": 50.38526912181303,
+      "grad_norm": 3.9033128277895988,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 761654848,
+      "step": 4436
+    },
+    {
+      "epoch": 50.38526912181303,
+      "loss": 0.05661904066801071,
+      "loss_ce": 0.0002530761994421482,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 761654848,
+      "step": 4436
+    },
+    {
+      "epoch": 50.39660056657224,
+      "grad_norm": 4.101962729604324,
+      "learning_rate": 5e-06,
+      "loss": 0.0651,
+      "num_input_tokens_seen": 761826036,
+      "step": 4437
+    },
+    {
+      "epoch": 50.39660056657224,
+      "loss": 0.07162822782993317,
+      "loss_ce": 0.0005222682375460863,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 761826036,
+      "step": 4437
+    },
+    {
+      "epoch": 50.40793201133145,
+      "grad_norm": 4.069166702192787,
+      "learning_rate": 5e-06,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 761997472,
+      "step": 4438
+    },
+    {
+      "epoch": 50.40793201133145,
+      "loss": 0.09274718910455704,
+      "loss_ce": 0.0003399592242203653,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 761997472,
+      "step": 4438
+    },
+    {
+      "epoch": 50.419263456090654,
+      "grad_norm": 4.884899028897731,
+      "learning_rate": 5e-06,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 762168420,
+      "step": 4439
+    },
+    {
+      "epoch": 50.419263456090654,
+      "loss": 0.12636074423789978,
+      "loss_ce": 0.0004146971623413265,
+      "loss_iou": 0.53125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 762168420,
+      "step": 4439
+    },
+    {
+      "epoch": 50.43059490084986,
+      "grad_norm": 4.508484589620329,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 762340384,
+      "step": 4440
+    },
+    {
+      "epoch": 50.43059490084986,
+      "loss": 0.12017138302326202,
+      "loss_ce": 0.0002983251179102808,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 762340384,
+      "step": 4440
+    },
+    {
+      "epoch": 50.44192634560906,
+      "grad_norm": 4.6213845222841385,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 762509820,
+      "step": 4441
+    },
+    {
+      "epoch": 50.44192634560906,
+      "loss": 0.1878926008939743,
+      "loss_ce": 0.0003926012432202697,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 762509820,
+      "step": 4441
+    },
+    {
+      "epoch": 50.45325779036827,
+      "grad_norm": 5.225287559911648,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 762681756,
+      "step": 4442
+    },
+    {
+      "epoch": 50.45325779036827,
+      "loss": 0.052958663552999496,
+      "loss_ce": 0.00040739180985838175,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 762681756,
+      "step": 4442
+    },
+    {
+      "epoch": 50.46458923512748,
+      "grad_norm": 3.6754381417200497,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 762853912,
+      "step": 4443
+    },
+    {
+      "epoch": 50.46458923512748,
+      "loss": 0.05573795735836029,
+      "loss_ce": 0.0003027716011274606,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 762853912,
+      "step": 4443
+    },
+    {
+      "epoch": 50.475920679886684,
+      "grad_norm": 3.8937946679832076,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 763024216,
+      "step": 4444
+    },
+    {
+      "epoch": 50.475920679886684,
+      "loss": 0.09364482760429382,
+      "loss_ce": 0.0005204459303058684,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 763024216,
+      "step": 4444
+    },
+    {
+      "epoch": 50.48725212464589,
+      "grad_norm": 4.669908733165509,
+      "learning_rate": 5e-06,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 763196380,
+      "step": 4445
+    },
+    {
+      "epoch": 50.48725212464589,
+      "loss": 0.04426318407058716,
+      "loss_ce": 0.000195800224901177,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 763196380,
+      "step": 4445
+    },
+    {
+      "epoch": 50.4985835694051,
+      "grad_norm": 4.324062435745885,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 763365028,
+      "step": 4446
+    },
+    {
+      "epoch": 50.4985835694051,
+      "loss": 0.14875555038452148,
+      "loss_ce": 0.0002570173528511077,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 763365028,
+      "step": 4446
+    },
+    {
+      "epoch": 50.50991501416431,
+      "grad_norm": 3.8630831844918543,
+      "learning_rate": 5e-06,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 763536276,
+      "step": 4447
+    },
+    {
+      "epoch": 50.50991501416431,
+      "loss": 0.09591014683246613,
+      "loss_ce": 0.001198843470774591,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 763536276,
+      "step": 4447
+    },
+    {
+      "epoch": 50.521246458923514,
+      "grad_norm": 17.249312436005795,
+      "learning_rate": 5e-06,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 763708156,
+      "step": 4448
+    },
+    {
+      "epoch": 50.521246458923514,
+      "loss": 0.07101692259311676,
+      "loss_ce": 0.000567091628909111,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 763708156,
+      "step": 4448
+    },
+    {
+      "epoch": 50.53257790368272,
+      "grad_norm": 3.6634734959179154,
+      "learning_rate": 5e-06,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 763879872,
+      "step": 4449
+    },
+    {
+      "epoch": 50.53257790368272,
+      "loss": 0.06236252188682556,
+      "loss_ce": 0.0010527074337005615,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 763879872,
+      "step": 4449
+    },
+    {
+      "epoch": 50.54390934844193,
+      "grad_norm": 3.1649803801412415,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 764051644,
+      "step": 4450
+    },
+    {
+      "epoch": 50.54390934844193,
+      "loss": 0.11230932921171188,
+      "loss_ce": 0.00046240483061410487,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 764051644,
+      "step": 4450
+    },
+    {
+      "epoch": 50.555240793201136,
+      "grad_norm": 3.004898679713995,
+      "learning_rate": 5e-06,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 764222032,
+      "step": 4451
+    },
+    {
+      "epoch": 50.555240793201136,
+      "loss": 0.052467942237854004,
+      "loss_ce": 0.0002523672883398831,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 764222032,
+      "step": 4451
+    },
+    {
+      "epoch": 50.56657223796034,
+      "grad_norm": 2.8458393998173763,
+      "learning_rate": 5e-06,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 764391780,
+      "step": 4452
+    },
+    {
+      "epoch": 50.56657223796034,
+      "loss": 0.06604805588722229,
+      "loss_ce": 0.0003589663829188794,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 764391780,
+      "step": 4452
+    },
+    {
+      "epoch": 50.577903682719544,
+      "grad_norm": 2.884652573213962,
+      "learning_rate": 5e-06,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 764561460,
+      "step": 4453
+    },
+    {
+      "epoch": 50.577903682719544,
+      "loss": 0.07057365775108337,
+      "loss_ce": 0.0003832242509815842,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 764561460,
+      "step": 4453
+    },
+    {
+      "epoch": 50.58923512747875,
+      "grad_norm": 2.9240991805472683,
+      "learning_rate": 5e-06,
+      "loss": 0.0966,
+      "num_input_tokens_seen": 764733044,
+      "step": 4454
+    },
+    {
+      "epoch": 50.58923512747875,
+      "loss": 0.21847650408744812,
+      "loss_ce": 0.0007335906266234815,
+      "loss_iou": 0.0966796875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 764733044,
+      "step": 4454
+    },
+    {
+      "epoch": 50.60056657223796,
+      "grad_norm": 3.781267890603106,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 764903636,
+      "step": 4455
+    },
+    {
+      "epoch": 50.60056657223796,
+      "loss": 0.15405893325805664,
+      "loss_ce": 0.0006623170920647681,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 764903636,
+      "step": 4455
+    },
+    {
+      "epoch": 50.611898016997166,
+      "grad_norm": 5.169681165561143,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 765075268,
+      "step": 4456
+    },
+    {
+      "epoch": 50.611898016997166,
+      "loss": 0.07309263944625854,
+      "loss_ce": 0.0002166681078961119,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 765075268,
+      "step": 4456
+    },
+    {
+      "epoch": 50.623229461756374,
+      "grad_norm": 4.091553070917247,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 765247104,
+      "step": 4457
+    },
+    {
+      "epoch": 50.623229461756374,
+      "loss": 0.07224041223526001,
+      "loss_ce": 0.001317555201239884,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 765247104,
+      "step": 4457
+    },
+    {
+      "epoch": 50.63456090651558,
+      "grad_norm": 4.126998992242619,
+      "learning_rate": 5e-06,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 765418476,
+      "step": 4458
+    },
+    {
+      "epoch": 50.63456090651558,
+      "loss": 0.08577851951122284,
+      "loss_ce": 0.000420852069510147,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 765418476,
+      "step": 4458
+    },
+    {
+      "epoch": 50.64589235127479,
+      "grad_norm": 3.720466276255888,
+      "learning_rate": 5e-06,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 765589536,
+      "step": 4459
+    },
+    {
+      "epoch": 50.64589235127479,
+      "loss": 0.07755723595619202,
+      "loss_ce": 0.0007444872171618044,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 765589536,
+      "step": 4459
+    },
+    {
+      "epoch": 50.657223796033996,
+      "grad_norm": 4.2967392019514286,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 765761248,
+      "step": 4460
+    },
+    {
+      "epoch": 50.657223796033996,
+      "loss": 0.10007596015930176,
+      "loss_ce": 0.00010038288019131869,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 765761248,
+      "step": 4460
+    },
+    {
+      "epoch": 50.668555240793204,
+      "grad_norm": 4.097532890372669,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 765931492,
+      "step": 4461
+    },
+    {
+      "epoch": 50.668555240793204,
+      "loss": 0.09184668213129044,
+      "loss_ce": 0.0006754158530384302,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 765931492,
+      "step": 4461
+    },
+    {
+      "epoch": 50.67988668555241,
+      "grad_norm": 3.7907628879818263,
+      "learning_rate": 5e-06,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 766103056,
+      "step": 4462
+    },
+    {
+      "epoch": 50.67988668555241,
+      "loss": 0.04951099306344986,
+      "loss_ce": 0.0002861374814528972,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 766103056,
+      "step": 4462
+    },
+    {
+      "epoch": 50.69121813031161,
+      "grad_norm": 3.7076227203087995,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 766274636,
+      "step": 4463
+    },
+    {
+      "epoch": 50.69121813031161,
+      "loss": 0.07821498066186905,
+      "loss_ce": 0.00042567052878439426,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 766274636,
+      "step": 4463
+    },
+    {
+      "epoch": 50.70254957507082,
+      "grad_norm": 3.4484857248147853,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 766446340,
+      "step": 4464
+    },
+    {
+      "epoch": 50.70254957507082,
+      "loss": 0.06428650766611099,
+      "loss_ce": 0.00032166283926926553,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 766446340,
+      "step": 4464
+    },
+    {
+      "epoch": 50.713881019830026,
+      "grad_norm": 4.003184014867428,
+      "learning_rate": 5e-06,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 766618132,
+      "step": 4465
+    },
+    {
+      "epoch": 50.713881019830026,
+      "loss": 0.05058712512254715,
+      "loss_ce": 0.0010494638700038195,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 766618132,
+      "step": 4465
+    },
+    {
+      "epoch": 50.725212464589234,
+      "grad_norm": 6.23183585450477,
+      "learning_rate": 5e-06,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 766787736,
+      "step": 4466
+    },
+    {
+      "epoch": 50.725212464589234,
+      "loss": 0.06207390874624252,
+      "loss_ce": 0.0004131448222324252,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 766787736,
+      "step": 4466
+    },
+    {
+      "epoch": 50.73654390934844,
+      "grad_norm": 5.782560540058673,
+      "learning_rate": 5e-06,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 766959388,
+      "step": 4467
+    },
+    {
+      "epoch": 50.73654390934844,
+      "loss": 0.07525616139173508,
+      "loss_ce": 0.00012188716209493577,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 766959388,
+      "step": 4467
+    },
+    {
+      "epoch": 50.74787535410765,
+      "grad_norm": 3.6699910332487518,
+      "learning_rate": 5e-06,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 767131616,
+      "step": 4468
+    },
+    {
+      "epoch": 50.74787535410765,
+      "loss": 0.0930500477552414,
+      "loss_ce": 0.002481503179296851,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 767131616,
+      "step": 4468
+    },
+    {
+      "epoch": 50.759206798866856,
+      "grad_norm": 5.082677311606937,
+      "learning_rate": 5e-06,
+      "loss": 0.0787,
+      "num_input_tokens_seen": 767302644,
+      "step": 4469
+    },
+    {
+      "epoch": 50.759206798866856,
+      "loss": 0.06062406674027443,
+      "loss_ce": 0.003785079112276435,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 767302644,
+      "step": 4469
+    },
+    {
+      "epoch": 50.77053824362606,
+      "grad_norm": 3.469927821657738,
+      "learning_rate": 5e-06,
+      "loss": 0.0819,
+      "num_input_tokens_seen": 767474644,
+      "step": 4470
+    },
+    {
+      "epoch": 50.77053824362606,
+      "loss": 0.10431724786758423,
+      "loss_ce": 0.00048118835547938943,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 767474644,
+      "step": 4470
+    },
+    {
+      "epoch": 50.78186968838527,
+      "grad_norm": 4.251362090542788,
+      "learning_rate": 5e-06,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 767646668,
+      "step": 4471
+    },
+    {
+      "epoch": 50.78186968838527,
+      "loss": 0.054449379444122314,
+      "loss_ce": 0.0012267231941223145,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 767646668,
+      "step": 4471
+    },
+    {
+      "epoch": 50.79320113314448,
+      "grad_norm": 3.1657645455821526,
+      "learning_rate": 5e-06,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 767818424,
+      "step": 4472
+    },
+    {
+      "epoch": 50.79320113314448,
+      "loss": 0.09054841101169586,
+      "loss_ce": 0.00015533754776697606,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 767818424,
+      "step": 4472
+    },
+    {
+      "epoch": 50.804532577903686,
+      "grad_norm": 18.121377094572424,
+      "learning_rate": 5e-06,
+      "loss": 0.078,
+      "num_input_tokens_seen": 767990356,
+      "step": 4473
+    },
+    {
+      "epoch": 50.804532577903686,
+      "loss": 0.08168047666549683,
+      "loss_ce": 0.00038164720172062516,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 767990356,
+      "step": 4473
+    },
+    {
+      "epoch": 50.815864022662886,
+      "grad_norm": 4.7741616377752685,
+      "learning_rate": 5e-06,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 768159880,
+      "step": 4474
+    },
+    {
+      "epoch": 50.815864022662886,
+      "loss": 0.0824337899684906,
+      "loss_ce": 0.00049409712664783,
+      "loss_iou": 0.453125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 768159880,
+      "step": 4474
+    },
+    {
+      "epoch": 50.827195467422094,
+      "grad_norm": 3.218696108190223,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 768331996,
+      "step": 4475
+    },
+    {
+      "epoch": 50.827195467422094,
+      "loss": 0.1260414570569992,
+      "loss_ce": 0.0010719744022935629,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 768331996,
+      "step": 4475
+    },
+    {
+      "epoch": 50.8385269121813,
+      "grad_norm": 3.881314956253814,
+      "learning_rate": 5e-06,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 768503540,
+      "step": 4476
+    },
+    {
+      "epoch": 50.8385269121813,
+      "loss": 0.06960465013980865,
+      "loss_ce": 0.000299228064250201,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 768503540,
+      "step": 4476
+    },
+    {
+      "epoch": 50.84985835694051,
+      "grad_norm": 7.3627371740088785,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 768673824,
+      "step": 4477
+    },
+    {
+      "epoch": 50.84985835694051,
+      "loss": 0.07985567301511765,
+      "loss_ce": 0.00043367242324166,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 768673824,
+      "step": 4477
+    },
+    {
+      "epoch": 50.861189801699716,
+      "grad_norm": 4.3201353556278015,
+      "learning_rate": 5e-06,
+      "loss": 0.0695,
+      "num_input_tokens_seen": 768844620,
+      "step": 4478
+    },
+    {
+      "epoch": 50.861189801699716,
+      "loss": 0.06958837807178497,
+      "loss_ce": 0.0011984816519543529,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 768844620,
+      "step": 4478
+    },
+    {
+      "epoch": 50.87252124645892,
+      "grad_norm": 4.718497359939087,
+      "learning_rate": 5e-06,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 769015160,
+      "step": 4479
+    },
+    {
+      "epoch": 50.87252124645892,
+      "loss": 0.06399042159318924,
+      "loss_ce": 0.0010784345213323832,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 769015160,
+      "step": 4479
+    },
+    {
+      "epoch": 50.88385269121813,
+      "grad_norm": 4.0812761461182205,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 769185676,
+      "step": 4480
+    },
+    {
+      "epoch": 50.88385269121813,
+      "loss": 0.11874992400407791,
+      "loss_ce": 0.0010131086455658078,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 769185676,
+      "step": 4480
+    },
+    {
+      "epoch": 50.89518413597734,
+      "grad_norm": 3.750528923218562,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 769356776,
+      "step": 4481
+    },
+    {
+      "epoch": 50.89518413597734,
+      "loss": 0.07799747586250305,
+      "loss_ce": 0.00039127099444158375,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 769356776,
+      "step": 4481
+    },
+    {
+      "epoch": 50.906515580736546,
+      "grad_norm": 7.689864593323736,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 769528424,
+      "step": 4482
+    },
+    {
+      "epoch": 50.906515580736546,
+      "loss": 0.06997948884963989,
+      "loss_ce": 6.371815106831491e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 769528424,
+      "step": 4482
+    },
+    {
+      "epoch": 50.91784702549575,
+      "grad_norm": 4.425615390648746,
+      "learning_rate": 5e-06,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 769700304,
+      "step": 4483
+    },
+    {
+      "epoch": 50.91784702549575,
+      "loss": 0.05891703814268112,
+      "loss_ce": 0.0003385467571206391,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 769700304,
+      "step": 4483
+    },
+    {
+      "epoch": 50.92917847025496,
+      "grad_norm": 3.4634851903456187,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 769870796,
+      "step": 4484
+    },
+    {
+      "epoch": 50.92917847025496,
+      "loss": 0.10744870454072952,
+      "loss_ce": 0.0008508105529472232,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 769870796,
+      "step": 4484
+    },
+    {
+      "epoch": 50.94050991501416,
+      "grad_norm": 3.316613958922671,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 770042420,
+      "step": 4485
+    },
+    {
+      "epoch": 50.94050991501416,
+      "loss": 0.06894853711128235,
+      "loss_ce": 0.001642022281885147,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 770042420,
+      "step": 4485
+    },
+    {
+      "epoch": 50.95184135977337,
+      "grad_norm": 2.991785197317611,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 770213472,
+      "step": 4486
+    },
+    {
+      "epoch": 50.95184135977337,
+      "loss": 0.14712882041931152,
+      "loss_ce": 0.00012564245844259858,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 770213472,
+      "step": 4486
+    },
+    {
+      "epoch": 50.963172804532576,
+      "grad_norm": 4.353569033815165,
+      "learning_rate": 5e-06,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 770384272,
+      "step": 4487
+    },
+    {
+      "epoch": 50.963172804532576,
+      "loss": 0.05455201119184494,
+      "loss_ce": 0.00015443185111507773,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 770384272,
+      "step": 4487
+    },
+    {
+      "epoch": 50.97450424929178,
+      "grad_norm": 5.072560782369086,
+      "learning_rate": 5e-06,
+      "loss": 0.066,
+      "num_input_tokens_seen": 770555772,
+      "step": 4488
+    },
+    {
+      "epoch": 50.97450424929178,
+      "loss": 0.05929575115442276,
+      "loss_ce": 0.0007325211772695184,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 770555772,
+      "step": 4488
+    },
+    {
+      "epoch": 50.98583569405099,
+      "grad_norm": 3.98101773270491,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 770725732,
+      "step": 4489
+    },
+    {
+      "epoch": 50.98583569405099,
+      "loss": 0.10329563915729523,
+      "loss_ce": 0.0006650229915976524,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 770725732,
+      "step": 4489
+    },
+    {
+      "epoch": 50.9971671388102,
+      "grad_norm": 3.543801989019078,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 770897584,
+      "step": 4490
+    },
+    {
+      "epoch": 50.9971671388102,
+      "loss": 0.07445745915174484,
+      "loss_ce": 0.0006354345241561532,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 770897584,
+      "step": 4490
+    },
+    {
+      "epoch": 50.9971671388102,
+      "loss": 0.09697633236646652,
+      "loss_ce": 0.0008612187812104821,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 770940412,
+      "step": 4490
+    },
+    {
+      "epoch": 51.008498583569406,
+      "grad_norm": 11.696029333854522,
+      "learning_rate": 5e-06,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 771068864,
+      "step": 4491
+    },
+    {
+      "epoch": 51.008498583569406,
+      "loss": 0.06903669238090515,
+      "loss_ce": 0.001249522203579545,
+      "loss_iou": 0.09765625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 771068864,
+      "step": 4491
+    },
+    {
+      "epoch": 51.01983002832861,
+      "grad_norm": 2.944264058756464,
+      "learning_rate": 5e-06,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 771240468,
+      "step": 4492
+    },
+    {
+      "epoch": 51.01983002832861,
+      "loss": 0.07568216323852539,
+      "loss_ce": 0.00032663423917256296,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 771240468,
+      "step": 4492
+    },
+    {
+      "epoch": 51.03116147308782,
+      "grad_norm": 4.038955340697767,
+      "learning_rate": 5e-06,
+      "loss": 0.0787,
+      "num_input_tokens_seen": 771412288,
+      "step": 4493
+    },
+    {
+      "epoch": 51.03116147308782,
+      "loss": 0.06470480561256409,
+      "loss_ce": 0.0003279761876910925,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 771412288,
+      "step": 4493
+    },
+    {
+      "epoch": 51.04249291784703,
+      "grad_norm": 4.118764173577385,
+      "learning_rate": 5e-06,
+      "loss": 0.079,
+      "num_input_tokens_seen": 771581744,
+      "step": 4494
+    },
+    {
+      "epoch": 51.04249291784703,
+      "loss": 0.06250278651714325,
+      "loss_ce": 6.38273122604005e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 771581744,
+      "step": 4494
+    },
+    {
+      "epoch": 51.053824362606235,
+      "grad_norm": 4.094204475591344,
+      "learning_rate": 5e-06,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 771753200,
+      "step": 4495
+    },
+    {
+      "epoch": 51.053824362606235,
+      "loss": 0.06505470722913742,
+      "loss_ce": 0.0005100309499539435,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 771753200,
+      "step": 4495
+    },
+    {
+      "epoch": 51.065155807365436,
+      "grad_norm": 3.955992170010095,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 771923520,
+      "step": 4496
+    },
+    {
+      "epoch": 51.065155807365436,
+      "loss": 0.047629572451114655,
+      "loss_ce": 0.00025103078223764896,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 771923520,
+      "step": 4496
+    },
+    {
+      "epoch": 51.07648725212464,
+      "grad_norm": 3.9878613105217866,
+      "learning_rate": 5e-06,
+      "loss": 0.0819,
+      "num_input_tokens_seen": 772095020,
+      "step": 4497
+    },
+    {
+      "epoch": 51.07648725212464,
+      "loss": 0.05086749047040939,
+      "loss_ce": 0.0003914149710908532,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 772095020,
+      "step": 4497
+    },
+    {
+      "epoch": 51.08781869688385,
+      "grad_norm": 4.2614251442128035,
+      "learning_rate": 5e-06,
+      "loss": 0.0695,
+      "num_input_tokens_seen": 772265236,
+      "step": 4498
+    },
+    {
+      "epoch": 51.08781869688385,
+      "loss": 0.10085298120975494,
+      "loss_ce": 0.0005417039501480758,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 772265236,
+      "step": 4498
+    },
+    {
+      "epoch": 51.09915014164306,
+      "grad_norm": 4.71628385813899,
+      "learning_rate": 5e-06,
+      "loss": 0.057,
+      "num_input_tokens_seen": 772436780,
+      "step": 4499
+    },
+    {
+      "epoch": 51.09915014164306,
+      "loss": 0.05079628527164459,
+      "loss_ce": 0.00028969591949135065,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 772436780,
+      "step": 4499
+    },
+    {
+      "epoch": 51.110481586402265,
+      "grad_norm": 4.168961813978523,
+      "learning_rate": 5e-06,
+      "loss": 0.0902,
+      "num_input_tokens_seen": 772608204,
+      "step": 4500
+    },
+    {
+      "epoch": 51.110481586402265,
+      "eval_seeclick_CIoU": 0.5465849936008453,
+      "eval_seeclick_GIoU": 0.5389738380908966,
+      "eval_seeclick_IoU": 0.5810232162475586,
+      "eval_seeclick_MAE_all": 0.06306532770395279,
+      "eval_seeclick_MAE_h": 0.03525807522237301,
+      "eval_seeclick_MAE_w": 0.08735432475805283,
+      "eval_seeclick_MAE_x": 0.0965481661260128,
+      "eval_seeclick_MAE_y": 0.03310075309127569,
+      "eval_seeclick_NUM_probability": 0.9998623430728912,
+      "eval_seeclick_inside_bbox": 0.8920454680919647,
+      "eval_seeclick_loss": 0.7668238878250122,
+      "eval_seeclick_loss_ce": 0.5261051058769226,
+      "eval_seeclick_loss_iou": 0.5706787109375,
+      "eval_seeclick_loss_num": 0.047580718994140625,
+      "eval_seeclick_loss_xval": 0.237640380859375,
+      "eval_seeclick_runtime": 68.9924,
+      "eval_seeclick_samples_per_second": 0.623,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 772608204,
+      "step": 4500
+    },
+    {
+      "epoch": 51.110481586402265,
+      "eval_icons_CIoU": 0.7577144801616669,
+      "eval_icons_GIoU": 0.7579652965068817,
+      "eval_icons_IoU": 0.7701258361339569,
+      "eval_icons_MAE_all": 0.032606909051537514,
+      "eval_icons_MAE_h": 0.031240517273545265,
+      "eval_icons_MAE_w": 0.03600175678730011,
+      "eval_icons_MAE_x": 0.028579371981322765,
+      "eval_icons_MAE_y": 0.03460598364472389,
+      "eval_icons_NUM_probability": 0.9976474046707153,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.11088309437036514,
+      "eval_icons_loss_ce": 0.001976564759388566,
+      "eval_icons_loss_iou": 0.5400390625,
+      "eval_icons_loss_num": 0.01996612548828125,
+      "eval_icons_loss_xval": 0.099884033203125,
+      "eval_icons_runtime": 84.5,
+      "eval_icons_samples_per_second": 0.592,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 772608204,
+      "step": 4500
+    },
+    {
+      "epoch": 51.110481586402265,
+      "eval_screenspot_CIoU": 0.6291436950365702,
+      "eval_screenspot_GIoU": 0.6315093835194906,
+      "eval_screenspot_IoU": 0.6529751420021057,
+      "eval_screenspot_MAE_all": 0.06452302634716034,
+      "eval_screenspot_MAE_h": 0.039918248231212296,
+      "eval_screenspot_MAE_w": 0.11111702273289363,
+      "eval_screenspot_MAE_x": 0.0706788773338,
+      "eval_screenspot_MAE_y": 0.0363779521236817,
+      "eval_screenspot_NUM_probability": 0.9998601078987122,
+      "eval_screenspot_inside_bbox": 0.8945833245913187,
+      "eval_screenspot_loss": 0.2845422625541687,
+      "eval_screenspot_loss_ce": 0.014367691474035382,
+      "eval_screenspot_loss_iou": 0.5061848958333334,
+      "eval_screenspot_loss_num": 0.053690592447916664,
+      "eval_screenspot_loss_xval": 0.2685139973958333,
+      "eval_screenspot_runtime": 137.4318,
+      "eval_screenspot_samples_per_second": 0.648,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 772608204,
+      "step": 4500
+    },
+    {
+      "epoch": 51.110481586402265,
+      "eval_compot_CIoU": 0.8065867125988007,
+      "eval_compot_GIoU": 0.8006526827812195,
+      "eval_compot_IoU": 0.8227367103099823,
+      "eval_compot_MAE_all": 0.029521239921450615,
+      "eval_compot_MAE_h": 0.022162611596286297,
+      "eval_compot_MAE_w": 0.03618418239057064,
+      "eval_compot_MAE_x": 0.03599496930837631,
+      "eval_compot_MAE_y": 0.02374320011585951,
+      "eval_compot_NUM_probability": 0.9999399483203888,
+      "eval_compot_inside_bbox": 0.9131944477558136,
+      "eval_compot_loss": 0.09009036421775818,
+      "eval_compot_loss_ce": 9.890348155749962e-05,
+      "eval_compot_loss_iou": 0.465087890625,
+      "eval_compot_loss_num": 0.015417098999023438,
+      "eval_compot_loss_xval": 0.07709503173828125,
+      "eval_compot_runtime": 79.8502,
+      "eval_compot_samples_per_second": 0.626,
+      "eval_compot_steps_per_second": 0.025,
+      "num_input_tokens_seen": 772608204,
+      "step": 4500
+    },
+    {
+      "epoch": 51.110481586402265,
+      "eval_custom_ui_MAE_all": 0.021320768631994724,
+      "eval_custom_ui_MAE_x": 0.0345577672123909,
+      "eval_custom_ui_MAE_y": 0.008083771681413054,
+      "eval_custom_ui_NUM_probability": 0.999765545129776,
+      "eval_custom_ui_loss": 0.20550864934921265,
+      "eval_custom_ui_loss_ce": 0.09546193853020668,
+      "eval_custom_ui_loss_num": 0.0206756591796875,
+      "eval_custom_ui_loss_xval": 0.1033477783203125,
+      "eval_custom_ui_runtime": 58.8978,
+      "eval_custom_ui_samples_per_second": 0.849,
+      "eval_custom_ui_steps_per_second": 0.034,
+      "num_input_tokens_seen": 772608204,
+      "step": 4500
+    },
+    {
+      "epoch": 51.110481586402265,
+      "loss": 0.24092400074005127,
+      "loss_ce": 0.11543572694063187,
+      "loss_iou": 0.0,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 772608204,
+      "step": 4500
+    },
+    {
+      "epoch": 51.12181303116147,
+      "grad_norm": 3.0836833964427015,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 772778084,
+      "step": 4501
+    },
+    {
+      "epoch": 51.12181303116147,
+      "loss": 0.044696711003780365,
+      "loss_ce": 0.0007666591554880142,
+      "loss_iou": 0.09765625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 772778084,
+      "step": 4501
+    },
+    {
+      "epoch": 51.13314447592068,
+      "grad_norm": 3.704876455578892,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 772949496,
+      "step": 4502
+    },
+    {
+      "epoch": 51.13314447592068,
+      "loss": 0.07046029716730118,
+      "loss_ce": 0.00042245464283041656,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 772949496,
+      "step": 4502
+    },
+    {
+      "epoch": 51.14447592067989,
+      "grad_norm": 3.444899198827256,
+      "learning_rate": 5e-06,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 773119972,
+      "step": 4503
+    },
+    {
+      "epoch": 51.14447592067989,
+      "loss": 0.040962524712085724,
+      "loss_ce": 0.00023681898892391473,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 773119972,
+      "step": 4503
+    },
+    {
+      "epoch": 51.155807365439095,
+      "grad_norm": 3.7466415073475305,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 773291744,
+      "step": 4504
+    },
+    {
+      "epoch": 51.155807365439095,
+      "loss": 0.1048392727971077,
+      "loss_ce": 0.00040812097722664475,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 773291744,
+      "step": 4504
+    },
+    {
+      "epoch": 51.1671388101983,
+      "grad_norm": 3.0811271862134704,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 773463464,
+      "step": 4505
+    },
+    {
+      "epoch": 51.1671388101983,
+      "loss": 0.04541376233100891,
+      "loss_ce": 0.00023248909565154463,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 773463464,
+      "step": 4505
+    },
+    {
+      "epoch": 51.17847025495751,
+      "grad_norm": 3.2550722332963824,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 773634660,
+      "step": 4506
+    },
+    {
+      "epoch": 51.17847025495751,
+      "loss": 0.08185526728630066,
+      "loss_ce": 0.0003122958296444267,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 773634660,
+      "step": 4506
+    },
+    {
+      "epoch": 51.18980169971671,
+      "grad_norm": 4.050945261020096,
+      "learning_rate": 5e-06,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 773804896,
+      "step": 4507
+    },
+    {
+      "epoch": 51.18980169971671,
+      "loss": 0.06780797243118286,
+      "loss_ce": 0.00022679413086734712,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 773804896,
+      "step": 4507
+    },
+    {
+      "epoch": 51.20113314447592,
+      "grad_norm": 4.021506772788571,
+      "learning_rate": 5e-06,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 773976180,
+      "step": 4508
+    },
+    {
+      "epoch": 51.20113314447592,
+      "loss": 0.059205252677202225,
+      "loss_ce": 0.00027580931782722473,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 773976180,
+      "step": 4508
+    },
+    {
+      "epoch": 51.212464589235125,
+      "grad_norm": 3.8404061210170335,
+      "learning_rate": 5e-06,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 774147332,
+      "step": 4509
+    },
+    {
+      "epoch": 51.212464589235125,
+      "loss": 0.10334266722202301,
+      "loss_ce": 0.0006510112434625626,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 774147332,
+      "step": 4509
+    },
+    {
+      "epoch": 51.22379603399433,
+      "grad_norm": 3.5631100162159806,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 774319096,
+      "step": 4510
+    },
+    {
+      "epoch": 51.22379603399433,
+      "loss": 0.2650887966156006,
+      "loss_ce": 0.0001657061802688986,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 774319096,
+      "step": 4510
+    },
+    {
+      "epoch": 51.23512747875354,
+      "grad_norm": 3.568186650477262,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 774489288,
+      "step": 4511
+    },
+    {
+      "epoch": 51.23512747875354,
+      "loss": 0.05329609289765358,
+      "loss_ce": 0.0003175779711455107,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 774489288,
+      "step": 4511
+    },
+    {
+      "epoch": 51.24645892351275,
+      "grad_norm": 4.063399714688638,
+      "learning_rate": 5e-06,
+      "loss": 0.1022,
+      "num_input_tokens_seen": 774660856,
+      "step": 4512
+    },
+    {
+      "epoch": 51.24645892351275,
+      "loss": 0.11217355728149414,
+      "loss_ce": 0.0005097426474094391,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 774660856,
+      "step": 4512
+    },
+    {
+      "epoch": 51.257790368271955,
+      "grad_norm": 5.245471536020726,
+      "learning_rate": 5e-06,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 774832468,
+      "step": 4513
+    },
+    {
+      "epoch": 51.257790368271955,
+      "loss": 0.07258614897727966,
+      "loss_ce": 0.00027474723174236715,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 774832468,
+      "step": 4513
+    },
+    {
+      "epoch": 51.26912181303116,
+      "grad_norm": 3.388952625301186,
+      "learning_rate": 5e-06,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 775004024,
+      "step": 4514
+    },
+    {
+      "epoch": 51.26912181303116,
+      "loss": 0.06552299857139587,
+      "loss_ce": 0.0005663403426297009,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 775004024,
+      "step": 4514
+    },
+    {
+      "epoch": 51.28045325779037,
+      "grad_norm": 3.541188551834626,
+      "learning_rate": 5e-06,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 775176000,
+      "step": 4515
+    },
+    {
+      "epoch": 51.28045325779037,
+      "loss": 0.039313673973083496,
+      "loss_ce": 0.00025117231416516006,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 775176000,
+      "step": 4515
+    },
+    {
+      "epoch": 51.29178470254958,
+      "grad_norm": 4.00918964159367,
+      "learning_rate": 5e-06,
+      "loss": 0.0726,
+      "num_input_tokens_seen": 775347296,
+      "step": 4516
+    },
+    {
+      "epoch": 51.29178470254958,
+      "loss": 0.05828949064016342,
+      "loss_ce": 0.00027557372231967747,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 775347296,
+      "step": 4516
+    },
+    {
+      "epoch": 51.303116147308785,
+      "grad_norm": 3.9411707848265465,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 775518092,
+      "step": 4517
+    },
+    {
+      "epoch": 51.303116147308785,
+      "loss": 0.10470222681760788,
+      "loss_ce": 0.008114089258015156,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 775518092,
+      "step": 4517
+    },
+    {
+      "epoch": 51.314447592067985,
+      "grad_norm": 3.2787565227625812,
+      "learning_rate": 5e-06,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 775689896,
+      "step": 4518
+    },
+    {
+      "epoch": 51.314447592067985,
+      "loss": 0.06945203244686127,
+      "loss_ce": 0.00022290655761025846,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 775689896,
+      "step": 4518
+    },
+    {
+      "epoch": 51.32577903682719,
+      "grad_norm": 3.592584442269907,
+      "learning_rate": 5e-06,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 775859404,
+      "step": 4519
+    },
+    {
+      "epoch": 51.32577903682719,
+      "loss": 0.08434545993804932,
+      "loss_ce": 0.000300046696793288,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 775859404,
+      "step": 4519
+    },
+    {
+      "epoch": 51.3371104815864,
+      "grad_norm": 3.0736497148178725,
+      "learning_rate": 5e-06,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 776031524,
+      "step": 4520
+    },
+    {
+      "epoch": 51.3371104815864,
+      "loss": 0.06985723972320557,
+      "loss_ce": 0.00020086189033463597,
+      "loss_iou": 0.4375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 776031524,
+      "step": 4520
+    },
+    {
+      "epoch": 51.34844192634561,
+      "grad_norm": 3.4876066097664657,
+      "learning_rate": 5e-06,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 776203388,
+      "step": 4521
+    },
+    {
+      "epoch": 51.34844192634561,
+      "loss": 0.09579696506261826,
+      "loss_ce": 0.001833343063481152,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 776203388,
+      "step": 4521
+    },
+    {
+      "epoch": 51.359773371104815,
+      "grad_norm": 6.06725668325794,
+      "learning_rate": 5e-06,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 776374720,
+      "step": 4522
+    },
+    {
+      "epoch": 51.359773371104815,
+      "loss": 0.10557658970355988,
+      "loss_ce": 0.0007639627438038588,
+      "loss_iou": 0.453125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 776374720,
+      "step": 4522
+    },
+    {
+      "epoch": 51.37110481586402,
+      "grad_norm": 4.304705218594438,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 776546292,
+      "step": 4523
+    },
+    {
+      "epoch": 51.37110481586402,
+      "loss": 0.04815925285220146,
+      "loss_ce": 0.0002619122969917953,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 776546292,
+      "step": 4523
+    },
+    {
+      "epoch": 51.38243626062323,
+      "grad_norm": 2.9298417830938344,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 776717500,
+      "step": 4524
+    },
+    {
+      "epoch": 51.38243626062323,
+      "loss": 0.05452612787485123,
+      "loss_ce": 0.00011328543041599914,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 776717500,
+      "step": 4524
+    },
+    {
+      "epoch": 51.39376770538244,
+      "grad_norm": 2.941561889973824,
+      "learning_rate": 5e-06,
+      "loss": 0.061,
+      "num_input_tokens_seen": 776889140,
+      "step": 4525
+    },
+    {
+      "epoch": 51.39376770538244,
+      "loss": 0.03288891538977623,
+      "loss_ce": 0.0002808834542520344,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 776889140,
+      "step": 4525
+    },
+    {
+      "epoch": 51.405099150141645,
+      "grad_norm": 4.11941007446635,
+      "learning_rate": 5e-06,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 777061364,
+      "step": 4526
+    },
+    {
+      "epoch": 51.405099150141645,
+      "loss": 0.08032729476690292,
+      "loss_ce": 0.00023391506692860276,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 777061364,
+      "step": 4526
+    },
+    {
+      "epoch": 51.41643059490085,
+      "grad_norm": 5.254428987401125,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 777232936,
+      "step": 4527
+    },
+    {
+      "epoch": 51.41643059490085,
+      "loss": 0.05055055767297745,
+      "loss_ce": 0.00045595254050567746,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 777232936,
+      "step": 4527
+    },
+    {
+      "epoch": 51.42776203966006,
+      "grad_norm": 3.7166516772707,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 777404436,
+      "step": 4528
+    },
+    {
+      "epoch": 51.42776203966006,
+      "loss": 0.07817244529724121,
+      "loss_ce": 0.00032210606150329113,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 777404436,
+      "step": 4528
+    },
+    {
+      "epoch": 51.43909348441926,
+      "grad_norm": 4.941066003490131,
+      "learning_rate": 5e-06,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 777576300,
+      "step": 4529
+    },
+    {
+      "epoch": 51.43909348441926,
+      "loss": 0.06884098052978516,
+      "loss_ce": 0.0003900496813002974,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 777576300,
+      "step": 4529
+    },
+    {
+      "epoch": 51.45042492917847,
+      "grad_norm": 4.477123869474375,
+      "learning_rate": 5e-06,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 777745720,
+      "step": 4530
+    },
+    {
+      "epoch": 51.45042492917847,
+      "loss": 0.05033246800303459,
+      "loss_ce": 0.00028363786987029016,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 777745720,
+      "step": 4530
+    },
+    {
+      "epoch": 51.461756373937675,
+      "grad_norm": 3.9073355502327325,
+      "learning_rate": 5e-06,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 777916876,
+      "step": 4531
+    },
+    {
+      "epoch": 51.461756373937675,
+      "loss": 0.07175743579864502,
+      "loss_ce": 0.0012160517508164048,
+      "loss_iou": 0.640625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 777916876,
+      "step": 4531
+    },
+    {
+      "epoch": 51.47308781869688,
+      "grad_norm": 4.938120073212937,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 778088448,
+      "step": 4532
+    },
+    {
+      "epoch": 51.47308781869688,
+      "loss": 0.04724579676985741,
+      "loss_ce": 6.56223128316924e-05,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 778088448,
+      "step": 4532
+    },
+    {
+      "epoch": 51.48441926345609,
+      "grad_norm": 4.323858802474057,
+      "learning_rate": 5e-06,
+      "loss": 0.0773,
+      "num_input_tokens_seen": 778258496,
+      "step": 4533
+    },
+    {
+      "epoch": 51.48441926345609,
+      "loss": 0.07330875098705292,
+      "loss_ce": 0.00021915065008215606,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 778258496,
+      "step": 4533
+    },
+    {
+      "epoch": 51.4957507082153,
+      "grad_norm": 3.755605542475779,
+      "learning_rate": 5e-06,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 778428840,
+      "step": 4534
+    },
+    {
+      "epoch": 51.4957507082153,
+      "loss": 0.12589004635810852,
+      "loss_ce": 0.0003102092305198312,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 778428840,
+      "step": 4534
+    },
+    {
+      "epoch": 51.507082152974505,
+      "grad_norm": 8.384106220042213,
+      "learning_rate": 5e-06,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 778600936,
+      "step": 4535
+    },
+    {
+      "epoch": 51.507082152974505,
+      "loss": 0.039587028324604034,
+      "loss_ce": 0.00020409331773407757,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 778600936,
+      "step": 4535
+    },
+    {
+      "epoch": 51.51841359773371,
+      "grad_norm": 3.2738725288189743,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 778769396,
+      "step": 4536
+    },
+    {
+      "epoch": 51.51841359773371,
+      "loss": 0.04192512109875679,
+      "loss_ce": 0.0002686251245904714,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 778769396,
+      "step": 4536
+    },
+    {
+      "epoch": 51.52974504249292,
+      "grad_norm": 4.698153922818547,
+      "learning_rate": 5e-06,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 778941060,
+      "step": 4537
+    },
+    {
+      "epoch": 51.52974504249292,
+      "loss": 0.12478635460138321,
+      "loss_ce": 0.0004272240912541747,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 778941060,
+      "step": 4537
+    },
+    {
+      "epoch": 51.54107648725213,
+      "grad_norm": 4.4347826307340155,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 779112156,
+      "step": 4538
+    },
+    {
+      "epoch": 51.54107648725213,
+      "loss": 0.06575482338666916,
+      "loss_ce": 0.00020305917132645845,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 779112156,
+      "step": 4538
+    },
+    {
+      "epoch": 51.552407932011334,
+      "grad_norm": 5.724012122037744,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 779284308,
+      "step": 4539
+    },
+    {
+      "epoch": 51.552407932011334,
+      "loss": 0.08919038623571396,
+      "loss_ce": 0.0003689788281917572,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 779284308,
+      "step": 4539
+    },
+    {
+      "epoch": 51.563739376770535,
+      "grad_norm": 3.523108598541958,
+      "learning_rate": 5e-06,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 779454308,
+      "step": 4540
+    },
+    {
+      "epoch": 51.563739376770535,
+      "loss": 0.05216085538268089,
+      "loss_ce": 0.00020467647118493915,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 779454308,
+      "step": 4540
+    },
+    {
+      "epoch": 51.57507082152974,
+      "grad_norm": 2.788066414864014,
+      "learning_rate": 5e-06,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 779624456,
+      "step": 4541
+    },
+    {
+      "epoch": 51.57507082152974,
+      "loss": 0.0518072135746479,
+      "loss_ce": 0.0002935428055934608,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 779624456,
+      "step": 4541
+    },
+    {
+      "epoch": 51.58640226628895,
+      "grad_norm": 5.1969517557770235,
+      "learning_rate": 5e-06,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 779794708,
+      "step": 4542
+    },
+    {
+      "epoch": 51.58640226628895,
+      "loss": 0.04793885722756386,
+      "loss_ce": 0.00030091824010014534,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 779794708,
+      "step": 4542
+    },
+    {
+      "epoch": 51.59773371104816,
+      "grad_norm": 3.967935626798619,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 779966024,
+      "step": 4543
+    },
+    {
+      "epoch": 51.59773371104816,
+      "loss": 0.17677834630012512,
+      "loss_ce": 0.00020362407667562366,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 779966024,
+      "step": 4543
+    },
+    {
+      "epoch": 51.609065155807365,
+      "grad_norm": 3.8957132359016216,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 780137820,
+      "step": 4544
+    },
+    {
+      "epoch": 51.609065155807365,
+      "loss": 0.12237225472927094,
+      "loss_ce": 0.00024091021623462439,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 780137820,
+      "step": 4544
+    },
+    {
+      "epoch": 51.62039660056657,
+      "grad_norm": 3.185600408235008,
+      "learning_rate": 5e-06,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 780309576,
+      "step": 4545
+    },
+    {
+      "epoch": 51.62039660056657,
+      "loss": 0.05723777785897255,
+      "loss_ce": 0.0002004253037739545,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 780309576,
+      "step": 4545
+    },
+    {
+      "epoch": 51.63172804532578,
+      "grad_norm": 4.995323500215864,
+      "learning_rate": 5e-06,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 780481248,
+      "step": 4546
+    },
+    {
+      "epoch": 51.63172804532578,
+      "loss": 0.058180827647447586,
+      "loss_ce": 0.000288982642814517,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 780481248,
+      "step": 4546
+    },
+    {
+      "epoch": 51.64305949008499,
+      "grad_norm": 4.228666600667304,
+      "learning_rate": 5e-06,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 780653384,
+      "step": 4547
+    },
+    {
+      "epoch": 51.64305949008499,
+      "loss": 0.057006604969501495,
+      "loss_ce": 0.00036597985308617353,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 780653384,
+      "step": 4547
+    },
+    {
+      "epoch": 51.654390934844194,
+      "grad_norm": 3.747763916070439,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 780824620,
+      "step": 4548
+    },
+    {
+      "epoch": 51.654390934844194,
+      "loss": 0.1148025393486023,
+      "loss_ce": 0.0010635234648361802,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 780824620,
+      "step": 4548
+    },
+    {
+      "epoch": 51.6657223796034,
+      "grad_norm": 3.6317036214257423,
+      "learning_rate": 5e-06,
+      "loss": 0.098,
+      "num_input_tokens_seen": 780995548,
+      "step": 4549
+    },
+    {
+      "epoch": 51.6657223796034,
+      "loss": 0.09171932935714722,
+      "loss_ce": 0.0002886661095544696,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 780995548,
+      "step": 4549
+    },
+    {
+      "epoch": 51.67705382436261,
+      "grad_norm": 3.916847526133361,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 781167400,
+      "step": 4550
+    },
+    {
+      "epoch": 51.67705382436261,
+      "loss": 0.06322573125362396,
+      "loss_ce": 0.0002984790480695665,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 781167400,
+      "step": 4550
+    },
+    {
+      "epoch": 51.68838526912181,
+      "grad_norm": 3.796101313892827,
+      "learning_rate": 5e-06,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 781338704,
+      "step": 4551
+    },
+    {
+      "epoch": 51.68838526912181,
+      "loss": 0.04833993688225746,
+      "loss_ce": 0.0007172580226324499,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 781338704,
+      "step": 4551
+    },
+    {
+      "epoch": 51.69971671388102,
+      "grad_norm": 5.437545814560452,
+      "learning_rate": 5e-06,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 781510388,
+      "step": 4552
+    },
+    {
+      "epoch": 51.69971671388102,
+      "loss": 0.044218339025974274,
+      "loss_ce": 0.000517164240591228,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 781510388,
+      "step": 4552
+    },
+    {
+      "epoch": 51.711048158640224,
+      "grad_norm": 5.024072145983963,
+      "learning_rate": 5e-06,
+      "loss": 0.0864,
+      "num_input_tokens_seen": 781682224,
+      "step": 4553
+    },
+    {
+      "epoch": 51.711048158640224,
+      "loss": 0.0491611622273922,
+      "loss_ce": 0.0002720020420383662,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 781682224,
+      "step": 4553
+    },
+    {
+      "epoch": 51.72237960339943,
+      "grad_norm": 3.147355411062974,
+      "learning_rate": 5e-06,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 781854112,
+      "step": 4554
+    },
+    {
+      "epoch": 51.72237960339943,
+      "loss": 0.04576756805181503,
+      "loss_ce": 0.00044896139297634363,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 781854112,
+      "step": 4554
+    },
+    {
+      "epoch": 51.73371104815864,
+      "grad_norm": 4.3830538468801095,
+      "learning_rate": 5e-06,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 782023732,
+      "step": 4555
+    },
+    {
+      "epoch": 51.73371104815864,
+      "loss": 0.11079318076372147,
+      "loss_ce": 0.0001822151243686676,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 782023732,
+      "step": 4555
+    },
+    {
+      "epoch": 51.74504249291785,
+      "grad_norm": 3.1154138772315147,
+      "learning_rate": 5e-06,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 782193224,
+      "step": 4556
+    },
+    {
+      "epoch": 51.74504249291785,
+      "loss": 0.05634358897805214,
+      "loss_ce": 0.00032857467886060476,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 782193224,
+      "step": 4556
+    },
+    {
+      "epoch": 51.756373937677054,
+      "grad_norm": 3.3295835669807796,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 782364844,
+      "step": 4557
+    },
+    {
+      "epoch": 51.756373937677054,
+      "loss": 0.16247408092021942,
+      "loss_ce": 0.00027314884937368333,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 782364844,
+      "step": 4557
+    },
+    {
+      "epoch": 51.76770538243626,
+      "grad_norm": 4.795776728751136,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 782536636,
+      "step": 4558
+    },
+    {
+      "epoch": 51.76770538243626,
+      "loss": 0.05933544412255287,
+      "loss_ce": 0.00034496455918997526,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 782536636,
+      "step": 4558
+    },
+    {
+      "epoch": 51.77903682719547,
+      "grad_norm": 3.456902876735184,
+      "learning_rate": 5e-06,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 782708580,
+      "step": 4559
+    },
+    {
+      "epoch": 51.77903682719547,
+      "loss": 0.07327303290367126,
+      "loss_ce": 0.00015291766612790525,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 782708580,
+      "step": 4559
+    },
+    {
+      "epoch": 51.79036827195468,
+      "grad_norm": 3.2982085404406645,
+      "learning_rate": 5e-06,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 782880456,
+      "step": 4560
+    },
+    {
+      "epoch": 51.79036827195468,
+      "loss": 0.09759436547756195,
+      "loss_ce": 0.00015173680731095374,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 782880456,
+      "step": 4560
+    },
+    {
+      "epoch": 51.801699716713884,
+      "grad_norm": 3.386624785449222,
+      "learning_rate": 5e-06,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 783052612,
+      "step": 4561
+    },
+    {
+      "epoch": 51.801699716713884,
+      "loss": 0.05019538849592209,
+      "loss_ce": 0.00031440702150575817,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 783052612,
+      "step": 4561
+    },
+    {
+      "epoch": 51.81303116147309,
+      "grad_norm": 3.6256909214450594,
+      "learning_rate": 5e-06,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 783223928,
+      "step": 4562
+    },
+    {
+      "epoch": 51.81303116147309,
+      "loss": 0.05543501302599907,
+      "loss_ce": 0.0008543243166059256,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 783223928,
+      "step": 4562
+    },
+    {
+      "epoch": 51.82436260623229,
+      "grad_norm": 3.43870692204241,
+      "learning_rate": 5e-06,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 783395124,
+      "step": 4563
+    },
+    {
+      "epoch": 51.82436260623229,
+      "loss": 0.11391442269086838,
+      "loss_ce": 0.00048058529500849545,
+      "loss_iou": 0.421875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 783395124,
+      "step": 4563
+    },
+    {
+      "epoch": 51.8356940509915,
+      "grad_norm": 3.8587525631730872,
+      "learning_rate": 5e-06,
+      "loss": 0.0915,
+      "num_input_tokens_seen": 783566948,
+      "step": 4564
+    },
+    {
+      "epoch": 51.8356940509915,
+      "loss": 0.047966137528419495,
+      "loss_ce": 0.0002824224065989256,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 783566948,
+      "step": 4564
+    },
+    {
+      "epoch": 51.84702549575071,
+      "grad_norm": 5.27205274234332,
+      "learning_rate": 5e-06,
+      "loss": 0.1211,
+      "num_input_tokens_seen": 783736480,
+      "step": 4565
+    },
+    {
+      "epoch": 51.84702549575071,
+      "loss": 0.17447087168693542,
+      "loss_ce": 0.0005511859781108797,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 783736480,
+      "step": 4565
+    },
+    {
+      "epoch": 51.858356940509914,
+      "grad_norm": 3.9523070232026734,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 783907644,
+      "step": 4566
+    },
+    {
+      "epoch": 51.858356940509914,
+      "loss": 0.06779085099697113,
+      "loss_ce": 0.000133380715851672,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 783907644,
+      "step": 4566
+    },
+    {
+      "epoch": 51.86968838526912,
+      "grad_norm": 3.332731983165318,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 784078368,
+      "step": 4567
+    },
+    {
+      "epoch": 51.86968838526912,
+      "loss": 0.07595255225896835,
+      "loss_ce": 0.00016214708739425987,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 784078368,
+      "step": 4567
+    },
+    {
+      "epoch": 51.88101983002833,
+      "grad_norm": 2.8011320176660166,
+      "learning_rate": 5e-06,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 784250188,
+      "step": 4568
+    },
+    {
+      "epoch": 51.88101983002833,
+      "loss": 0.05351635441184044,
+      "loss_ce": 0.0003547338128555566,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 784250188,
+      "step": 4568
+    },
+    {
+      "epoch": 51.892351274787536,
+      "grad_norm": 3.4640752341995116,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 784422232,
+      "step": 4569
+    },
+    {
+      "epoch": 51.892351274787536,
+      "loss": 0.08177356421947479,
+      "loss_ce": 0.0004899952909909189,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 784422232,
+      "step": 4569
+    },
+    {
+      "epoch": 51.903682719546744,
+      "grad_norm": 3.9566960634442965,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 784593788,
+      "step": 4570
+    },
+    {
+      "epoch": 51.903682719546744,
+      "loss": 0.03977704420685768,
+      "loss_ce": 0.0003788497415371239,
+      "loss_iou": 0.328125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 784593788,
+      "step": 4570
+    },
+    {
+      "epoch": 51.91501416430595,
+      "grad_norm": 4.176445520410173,
+      "learning_rate": 5e-06,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 784765404,
+      "step": 4571
+    },
+    {
+      "epoch": 51.91501416430595,
+      "loss": 0.09650145471096039,
+      "loss_ce": 0.0002490076585672796,
+      "loss_iou": 0.296875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 784765404,
+      "step": 4571
+    },
+    {
+      "epoch": 51.92634560906516,
+      "grad_norm": 3.6836944026912506,
+      "learning_rate": 5e-06,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 784937108,
+      "step": 4572
+    },
+    {
+      "epoch": 51.92634560906516,
+      "loss": 0.10871314257383347,
+      "loss_ce": 0.00025367323542013764,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 784937108,
+      "step": 4572
+    },
+    {
+      "epoch": 51.93767705382436,
+      "grad_norm": 3.322656807451097,
+      "learning_rate": 5e-06,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 785108972,
+      "step": 4573
+    },
+    {
+      "epoch": 51.93767705382436,
+      "loss": 0.08650954067707062,
+      "loss_ce": 0.00020583285368047655,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 785108972,
+      "step": 4573
+    },
+    {
+      "epoch": 51.94900849858357,
+      "grad_norm": 3.626391420245067,
+      "learning_rate": 5e-06,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 785280928,
+      "step": 4574
+    },
+    {
+      "epoch": 51.94900849858357,
+      "loss": 0.14615091681480408,
+      "loss_ce": 0.00021584474598057568,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 785280928,
+      "step": 4574
+    },
+    {
+      "epoch": 51.960339943342774,
+      "grad_norm": 3.74228068445649,
+      "learning_rate": 5e-06,
+      "loss": 0.062,
+      "num_input_tokens_seen": 785452408,
+      "step": 4575
+    },
+    {
+      "epoch": 51.960339943342774,
+      "loss": 0.040177762508392334,
+      "loss_ce": 0.00019973621238023043,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 785452408,
+      "step": 4575
+    },
+    {
+      "epoch": 51.97167138810198,
+      "grad_norm": 4.381000466982566,
+      "learning_rate": 5e-06,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 785623312,
+      "step": 4576
+    },
+    {
+      "epoch": 51.97167138810198,
+      "loss": 0.08159258216619492,
+      "loss_ce": 0.000278495776001364,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 785623312,
+      "step": 4576
+    },
+    {
+      "epoch": 51.98300283286119,
+      "grad_norm": 4.443961495423953,
+      "learning_rate": 5e-06,
+      "loss": 0.093,
+      "num_input_tokens_seen": 785795032,
+      "step": 4577
+    },
+    {
+      "epoch": 51.98300283286119,
+      "loss": 0.06167113408446312,
+      "loss_ce": 0.0026196218095719814,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 785795032,
+      "step": 4577
+    },
+    {
+      "epoch": 51.994334277620396,
+      "grad_norm": 5.640772146673305,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 785964900,
+      "step": 4578
+    },
+    {
+      "epoch": 51.994334277620396,
+      "loss": 0.1554902344942093,
+      "loss_ce": 9.472258534515277e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 785964900,
+      "step": 4578
+    },
+    {
+      "epoch": 51.994334277620396,
+      "loss": 0.08137950301170349,
+      "loss_ce": 0.0002027432492468506,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 786050860,
+      "step": 4578
+    },
+    {
+      "epoch": 52.005665722379604,
+      "grad_norm": 4.303144423608778,
+      "learning_rate": 5e-06,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 786136924,
+      "step": 4579
+    },
+    {
+      "epoch": 52.005665722379604,
+      "loss": 0.07516370713710785,
+      "loss_ce": 0.0003193450393155217,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 786136924,
+      "step": 4579
+    },
+    {
+      "epoch": 52.01699716713881,
+      "grad_norm": 3.921834133399401,
+      "learning_rate": 5e-06,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 786308744,
+      "step": 4580
+    },
+    {
+      "epoch": 52.01699716713881,
+      "loss": 0.06335844844579697,
+      "loss_ce": 0.0002175786648876965,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 786308744,
+      "step": 4580
+    },
+    {
+      "epoch": 52.02832861189802,
+      "grad_norm": 5.2965494335021,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 786480196,
+      "step": 4581
+    },
+    {
+      "epoch": 52.02832861189802,
+      "loss": 0.04880340397357941,
+      "loss_ce": 0.00014312355779111385,
+      "loss_iou": 0.40625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 786480196,
+      "step": 4581
+    },
+    {
+      "epoch": 52.039660056657226,
+      "grad_norm": 3.8913015508272104,
+      "learning_rate": 5e-06,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 786650312,
+      "step": 4582
+    },
+    {
+      "epoch": 52.039660056657226,
+      "loss": 0.06473094969987869,
+      "loss_ce": 0.0002167891652788967,
+      "loss_iou": 0.0,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 786650312,
+      "step": 4582
+    },
+    {
+      "epoch": 52.05099150141643,
+      "grad_norm": 3.239706300114128,
+      "learning_rate": 5e-06,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 786822288,
+      "step": 4583
+    },
+    {
+      "epoch": 52.05099150141643,
+      "loss": 0.06472610682249069,
+      "loss_ce": 0.0001203926804009825,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 786822288,
+      "step": 4583
+    },
+    {
+      "epoch": 52.06232294617564,
+      "grad_norm": 8.126112690511478,
+      "learning_rate": 5e-06,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 786992440,
+      "step": 4584
+    },
+    {
+      "epoch": 52.06232294617564,
+      "loss": 0.04443705454468727,
+      "loss_ce": 0.0002476015652064234,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 786992440,
+      "step": 4584
+    },
+    {
+      "epoch": 52.07365439093484,
+      "grad_norm": 2.8604273328543655,
+      "learning_rate": 5e-06,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 787164384,
+      "step": 4585
+    },
+    {
+      "epoch": 52.07365439093484,
+      "loss": 0.1946127712726593,
+      "loss_ce": 0.00018528368673287332,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 787164384,
+      "step": 4585
+    },
+    {
+      "epoch": 52.08498583569405,
+      "grad_norm": 3.781306563416115,
+      "learning_rate": 5e-06,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 787336264,
+      "step": 4586
+    },
+    {
+      "epoch": 52.08498583569405,
+      "loss": 0.05890447646379471,
+      "loss_ce": 0.00020391664293128997,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 787336264,
+      "step": 4586
+    },
+    {
+      "epoch": 52.096317280453256,
+      "grad_norm": 4.026259780403755,
+      "learning_rate": 5e-06,
+      "loss": 0.071,
+      "num_input_tokens_seen": 787507268,
+      "step": 4587
+    },
+    {
+      "epoch": 52.096317280453256,
+      "loss": 0.06553702801465988,
+      "loss_ce": 0.0017858067294582725,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 787507268,
+      "step": 4587
+    },
+    {
+      "epoch": 52.107648725212464,
+      "grad_norm": 4.201046247893455,
+      "learning_rate": 5e-06,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 787677964,
+      "step": 4588
+    },
+    {
+      "epoch": 52.107648725212464,
+      "loss": 0.12023103982210159,
+      "loss_ce": 0.00035798695171251893,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 787677964,
+      "step": 4588
+    },
+    {
+      "epoch": 52.11898016997167,
+      "grad_norm": 4.302483079826222,
+      "learning_rate": 5e-06,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 787848408,
+      "step": 4589
+    },
+    {
+      "epoch": 52.11898016997167,
+      "loss": 0.04813995957374573,
+      "loss_ce": 0.0005783134838566184,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 787848408,
+      "step": 4589
+    },
+    {
+      "epoch": 52.13031161473088,
+      "grad_norm": 3.9119184280850714,
+      "learning_rate": 5e-06,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 788018700,
+      "step": 4590
+    },
+    {
+      "epoch": 52.13031161473088,
+      "loss": 0.08638627827167511,
+      "loss_ce": 0.00031145173124969006,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 788018700,
+      "step": 4590
+    },
+    {
+      "epoch": 52.141643059490086,
+      "grad_norm": 4.143548816134912,
+      "learning_rate": 5e-06,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 788190580,
+      "step": 4591
+    },
+    {
+      "epoch": 52.141643059490086,
+      "loss": 0.06912362575531006,
+      "loss_ce": 0.00035226656473241746,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 788190580,
+      "step": 4591
+    },
+    {
+      "epoch": 52.15297450424929,
+      "grad_norm": 3.4365427981048535,
+      "learning_rate": 5e-06,
+      "loss": 0.0815,
+      "num_input_tokens_seen": 788362272,
+      "step": 4592
+    },
+    {
+      "epoch": 52.15297450424929,
+      "loss": 0.07606668770313263,
+      "loss_ce": 0.0003983561182394624,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 788362272,
+      "step": 4592
+    },
+    {
+      "epoch": 52.1643059490085,
+      "grad_norm": 2.283329325977374,
+      "learning_rate": 5e-06,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 788534284,
+      "step": 4593
+    },
+    {
+      "epoch": 52.1643059490085,
+      "loss": 0.11275344341993332,
+      "loss_ce": 0.0007691902574151754,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 788534284,
+      "step": 4593
+    },
+    {
+      "epoch": 52.17563739376771,
+      "grad_norm": 25.004455015184295,
+      "learning_rate": 5e-06,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 788704576,
+      "step": 4594
+    },
+    {
+      "epoch": 52.17563739376771,
+      "loss": 0.06830897182226181,
+      "loss_ce": 0.0002089940244331956,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 788704576,
+      "step": 4594
+    },
+    {
+      "epoch": 52.186968838526916,
+      "grad_norm": 3.933231076233221,
+      "learning_rate": 5e-06,
+      "loss": 0.1091,
+      "num_input_tokens_seen": 788876128,
+      "step": 4595
+    },
+    {
+      "epoch": 52.186968838526916,
+      "loss": 0.12454564869403839,
+      "loss_ce": 0.00024756102357059717,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 788876128,
+      "step": 4595
+    },
+    {
+      "epoch": 52.198300283286116,
+      "grad_norm": 3.882650550469579,
+      "learning_rate": 5e-06,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 789047916,
+      "step": 4596
+    },
+    {
+      "epoch": 52.198300283286116,
+      "loss": 0.07191193848848343,
+      "loss_ce": 0.0002566642942838371,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 789047916,
+      "step": 4596
+    },
+    {
+      "epoch": 52.20963172804532,
+      "grad_norm": 3.8972527625328928,
+      "learning_rate": 5e-06,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 789219720,
+      "step": 4597
+    },
+    {
+      "epoch": 52.20963172804532,
+      "loss": 0.060532379895448685,
+      "loss_ce": 0.00015335108037106693,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 789219720,
+      "step": 4597
+    },
+    {
+      "epoch": 52.22096317280453,
+      "grad_norm": 6.019367548412412,
+      "learning_rate": 5e-06,
+      "loss": 0.058,
+      "num_input_tokens_seen": 789390824,
+      "step": 4598
+    },
+    {
+      "epoch": 52.22096317280453,
+      "loss": 0.0595117025077343,
+      "loss_ce": 0.00018553137488197535,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 789390824,
+      "step": 4598
+    },
+    {
+      "epoch": 52.23229461756374,
+      "grad_norm": 5.295395724390834,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 789561784,
+      "step": 4599
+    },
+    {
+      "epoch": 52.23229461756374,
+      "loss": 0.07566888630390167,
+      "loss_ce": 0.00030573003459721804,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 789561784,
+      "step": 4599
+    },
+    {
+      "epoch": 52.243626062322946,
+      "grad_norm": 4.334072642231794,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 789732644,
+      "step": 4600
+    },
+    {
+      "epoch": 52.243626062322946,
+      "loss": 0.08430413901805878,
+      "loss_ce": 0.0003502837789710611,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 789732644,
+      "step": 4600
+    },
+    {
+      "epoch": 52.25495750708215,
+      "grad_norm": 5.6747098431544005,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 789902708,
+      "step": 4601
+    },
+    {
+      "epoch": 52.25495750708215,
+      "loss": 0.0486568883061409,
+      "loss_ce": 0.00019497414177749306,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 789902708,
+      "step": 4601
+    },
+    {
+      "epoch": 52.26628895184136,
+      "grad_norm": 3.0795139691449607,
+      "learning_rate": 5e-06,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 790074916,
+      "step": 4602
+    },
+    {
+      "epoch": 52.26628895184136,
+      "loss": 0.10147304832935333,
+      "loss_ce": 0.00018520957382861525,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 790074916,
+      "step": 4602
+    },
+    {
+      "epoch": 52.27762039660057,
+      "grad_norm": 2.5756460073002128,
+      "learning_rate": 5e-06,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 790244832,
+      "step": 4603
+    },
+    {
+      "epoch": 52.27762039660057,
+      "loss": 0.04581351578235626,
+      "loss_ce": 6.766716978745535e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 790244832,
+      "step": 4603
+    },
+    {
+      "epoch": 52.288951841359776,
+      "grad_norm": 3.059930235493123,
+      "learning_rate": 5e-06,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 790416556,
+      "step": 4604
+    },
+    {
+      "epoch": 52.288951841359776,
+      "loss": 0.04115888103842735,
+      "loss_ce": 0.00018903279851656407,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 790416556,
+      "step": 4604
+    },
+    {
+      "epoch": 52.30028328611898,
+      "grad_norm": 3.311641786260434,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 790585940,
+      "step": 4605
+    },
+    {
+      "epoch": 52.30028328611898,
+      "loss": 0.044871728867292404,
+      "loss_ce": 0.0004686516767833382,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 790585940,
+      "step": 4605
+    },
+    {
+      "epoch": 52.31161473087819,
+      "grad_norm": 4.606099171997949,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 790757840,
+      "step": 4606
+    },
+    {
+      "epoch": 52.31161473087819,
+      "loss": 0.07226390391588211,
+      "loss_ce": 0.0003950106038246304,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 790757840,
+      "step": 4606
+    },
+    {
+      "epoch": 52.32294617563739,
+      "grad_norm": 4.691173116312316,
+      "learning_rate": 5e-06,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 790929496,
+      "step": 4607
+    },
+    {
+      "epoch": 52.32294617563739,
+      "loss": 0.038962364196777344,
+      "loss_ce": 0.00041866640094667673,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 790929496,
+      "step": 4607
+    },
+    {
+      "epoch": 52.3342776203966,
+      "grad_norm": 5.360875976558649,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 791100772,
+      "step": 4608
+    },
+    {
+      "epoch": 52.3342776203966,
+      "loss": 0.18516650795936584,
+      "loss_ce": 0.00044359132880344987,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 791100772,
+      "step": 4608
+    },
+    {
+      "epoch": 52.345609065155806,
+      "grad_norm": 7.852135784214818,
+      "learning_rate": 5e-06,
+      "loss": 0.0797,
+      "num_input_tokens_seen": 791271060,
+      "step": 4609
+    },
+    {
+      "epoch": 52.345609065155806,
+      "loss": 0.14633364975452423,
+      "loss_ce": 0.00012392696226015687,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 791271060,
+      "step": 4609
+    },
+    {
+      "epoch": 52.35694050991501,
+      "grad_norm": 4.353403091836125,
+      "learning_rate": 5e-06,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 791442700,
+      "step": 4610
+    },
+    {
+      "epoch": 52.35694050991501,
+      "loss": 0.04589192941784859,
+      "loss_ce": 0.00013082032091915607,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 791442700,
+      "step": 4610
+    },
+    {
+      "epoch": 52.36827195467422,
+      "grad_norm": 5.7207745324118,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 791613768,
+      "step": 4611
+    },
+    {
+      "epoch": 52.36827195467422,
+      "loss": 0.08321557939052582,
+      "loss_ce": 0.00029931924655102193,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 791613768,
+      "step": 4611
+    },
+    {
+      "epoch": 52.37960339943343,
+      "grad_norm": 3.1196219284948743,
+      "learning_rate": 5e-06,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 791785176,
+      "step": 4612
+    },
+    {
+      "epoch": 52.37960339943343,
+      "loss": 0.04471583664417267,
+      "loss_ce": 0.00023646561021450907,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 791785176,
+      "step": 4612
+    },
+    {
+      "epoch": 52.390934844192635,
+      "grad_norm": 3.0644902369411224,
+      "learning_rate": 5e-06,
+      "loss": 0.0871,
+      "num_input_tokens_seen": 791957076,
+      "step": 4613
+    },
+    {
+      "epoch": 52.390934844192635,
+      "loss": 0.047982584685087204,
+      "loss_ce": 0.0002836083003785461,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 791957076,
+      "step": 4613
+    },
+    {
+      "epoch": 52.40226628895184,
+      "grad_norm": 3.562853539455721,
+      "learning_rate": 5e-06,
+      "loss": 0.078,
+      "num_input_tokens_seen": 792127860,
+      "step": 4614
+    },
+    {
+      "epoch": 52.40226628895184,
+      "loss": 0.1062203198671341,
+      "loss_ce": 0.0005074322107248008,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 792127860,
+      "step": 4614
+    },
+    {
+      "epoch": 52.41359773371105,
+      "grad_norm": 3.8692498889392914,
+      "learning_rate": 5e-06,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 792297432,
+      "step": 4615
+    },
+    {
+      "epoch": 52.41359773371105,
+      "loss": 0.07992237061262131,
+      "loss_ce": 0.00027149339439347386,
+      "loss_iou": 0.5625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 792297432,
+      "step": 4615
+    },
+    {
+      "epoch": 52.42492917847026,
+      "grad_norm": 3.234229716298798,
+      "learning_rate": 5e-06,
+      "loss": 0.061,
+      "num_input_tokens_seen": 792469232,
+      "step": 4616
+    },
+    {
+      "epoch": 52.42492917847026,
+      "loss": 0.04069376736879349,
+      "loss_ce": 0.00013590327580459416,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 792469232,
+      "step": 4616
+    },
+    {
+      "epoch": 52.436260623229465,
+      "grad_norm": 3.7335701325658466,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 792641164,
+      "step": 4617
+    },
+    {
+      "epoch": 52.436260623229465,
+      "loss": 0.04785492271184921,
+      "loss_ce": 0.00017120649863500148,
+      "loss_iou": 0.484375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 792641164,
+      "step": 4617
+    },
+    {
+      "epoch": 52.447592067988666,
+      "grad_norm": 4.038699857916988,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 792812784,
+      "step": 4618
+    },
+    {
+      "epoch": 52.447592067988666,
+      "loss": 0.19159315526485443,
+      "loss_ce": 0.0002784566313493997,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 792812784,
+      "step": 4618
+    },
+    {
+      "epoch": 52.45892351274787,
+      "grad_norm": 10.070736555907908,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 792983924,
+      "step": 4619
+    },
+    {
+      "epoch": 52.45892351274787,
+      "loss": 0.10200847685337067,
+      "loss_ce": 0.00023235470871441066,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 792983924,
+      "step": 4619
+    },
+    {
+      "epoch": 52.47025495750708,
+      "grad_norm": 4.21185622423892,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 793155532,
+      "step": 4620
+    },
+    {
+      "epoch": 52.47025495750708,
+      "loss": 0.08301263302564621,
+      "loss_ce": 0.0002184435143135488,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 793155532,
+      "step": 4620
+    },
+    {
+      "epoch": 52.48158640226629,
+      "grad_norm": 3.489045365033888,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 793327388,
+      "step": 4621
+    },
+    {
+      "epoch": 52.48158640226629,
+      "loss": 0.13099710643291473,
+      "loss_ce": 9.1952046204824e-05,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 793327388,
+      "step": 4621
+    },
+    {
+      "epoch": 52.492917847025495,
+      "grad_norm": 3.290680055827727,
+      "learning_rate": 5e-06,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 793499516,
+      "step": 4622
+    },
+    {
+      "epoch": 52.492917847025495,
+      "loss": 0.06874236464500427,
+      "loss_ce": 0.0002151418593712151,
+      "loss_iou": 0.40625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 793499516,
+      "step": 4622
+    },
+    {
+      "epoch": 52.5042492917847,
+      "grad_norm": 3.676025515682793,
+      "learning_rate": 5e-06,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 793669372,
+      "step": 4623
+    },
+    {
+      "epoch": 52.5042492917847,
+      "loss": 0.06475893408060074,
+      "loss_ce": 0.0003668441786430776,
+      "loss_iou": 0.625,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 793669372,
+      "step": 4623
+    },
+    {
+      "epoch": 52.51558073654391,
+      "grad_norm": 4.138553728613425,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 793841280,
+      "step": 4624
+    },
+    {
+      "epoch": 52.51558073654391,
+      "loss": 0.06110313534736633,
+      "loss_ce": 0.00012901413720101118,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 793841280,
+      "step": 4624
+    },
+    {
+      "epoch": 52.52691218130312,
+      "grad_norm": 3.7479659378994348,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 794012904,
+      "step": 4625
+    },
+    {
+      "epoch": 52.52691218130312,
+      "loss": 0.15677732229232788,
+      "loss_ce": 0.00010007420496549457,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 794012904,
+      "step": 4625
+    },
+    {
+      "epoch": 52.538243626062325,
+      "grad_norm": 3.9256908910914308,
+      "learning_rate": 5e-06,
+      "loss": 0.077,
+      "num_input_tokens_seen": 794182000,
+      "step": 4626
+    },
+    {
+      "epoch": 52.538243626062325,
+      "loss": 0.04677785187959671,
+      "loss_ce": 0.00033009928301908076,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 794182000,
+      "step": 4626
+    },
+    {
+      "epoch": 52.54957507082153,
+      "grad_norm": 2.7862732774895265,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 794353776,
+      "step": 4627
+    },
+    {
+      "epoch": 52.54957507082153,
+      "loss": 0.0538012720644474,
+      "loss_ce": 0.00022766315669286996,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 794353776,
+      "step": 4627
+    },
+    {
+      "epoch": 52.56090651558074,
+      "grad_norm": 3.1180152230786446,
+      "learning_rate": 5e-06,
+      "loss": 0.076,
+      "num_input_tokens_seen": 794525756,
+      "step": 4628
+    },
+    {
+      "epoch": 52.56090651558074,
+      "loss": 0.09337963163852692,
+      "loss_ce": 0.0001484311796957627,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 794525756,
+      "step": 4628
+    },
+    {
+      "epoch": 52.57223796033994,
+      "grad_norm": 4.366194404392181,
+      "learning_rate": 5e-06,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 794695428,
+      "step": 4629
+    },
+    {
+      "epoch": 52.57223796033994,
+      "loss": 0.07714550942182541,
+      "loss_ce": 0.003918584436178207,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 794695428,
+      "step": 4629
+    },
+    {
+      "epoch": 52.58356940509915,
+      "grad_norm": 3.69099901403253,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 794867264,
+      "step": 4630
+    },
+    {
+      "epoch": 52.58356940509915,
+      "loss": 0.05563777685165405,
+      "loss_ce": 9.578681056154892e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 794867264,
+      "step": 4630
+    },
+    {
+      "epoch": 52.594900849858355,
+      "grad_norm": 5.17556408270218,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 795039292,
+      "step": 4631
+    },
+    {
+      "epoch": 52.594900849858355,
+      "loss": 0.060076937079429626,
+      "loss_ce": 0.00044558741501532495,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 795039292,
+      "step": 4631
+    },
+    {
+      "epoch": 52.60623229461756,
+      "grad_norm": 4.477108568470924,
+      "learning_rate": 5e-06,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 795211032,
+      "step": 4632
+    },
+    {
+      "epoch": 52.60623229461756,
+      "loss": 0.06861227750778198,
+      "loss_ce": 0.0006954035488888621,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 795211032,
+      "step": 4632
+    },
+    {
+      "epoch": 52.61756373937677,
+      "grad_norm": 4.047173777819922,
+      "learning_rate": 5e-06,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 795382672,
+      "step": 4633
+    },
+    {
+      "epoch": 52.61756373937677,
+      "loss": 0.07145858556032181,
+      "loss_ce": 7.797317812219262e-05,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 795382672,
+      "step": 4633
+    },
+    {
+      "epoch": 52.62889518413598,
+      "grad_norm": 3.823208299584292,
+      "learning_rate": 5e-06,
+      "loss": 0.0857,
+      "num_input_tokens_seen": 795553456,
+      "step": 4634
+    },
+    {
+      "epoch": 52.62889518413598,
+      "loss": 0.06402266770601273,
+      "loss_ce": 0.0003019628638867289,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 795553456,
+      "step": 4634
+    },
+    {
+      "epoch": 52.640226628895185,
+      "grad_norm": 4.516449227754517,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 795722164,
+      "step": 4635
+    },
+    {
+      "epoch": 52.640226628895185,
+      "loss": 0.05346234142780304,
+      "loss_ce": 0.00034650013549253345,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 795722164,
+      "step": 4635
+    },
+    {
+      "epoch": 52.65155807365439,
+      "grad_norm": 3.675764736433771,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 795893256,
+      "step": 4636
+    },
+    {
+      "epoch": 52.65155807365439,
+      "loss": 0.07161383330821991,
+      "loss_ce": 5.011321263737045e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 795893256,
+      "step": 4636
+    },
+    {
+      "epoch": 52.6628895184136,
+      "grad_norm": 3.8623752727817084,
+      "learning_rate": 5e-06,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 796065108,
+      "step": 4637
+    },
+    {
+      "epoch": 52.6628895184136,
+      "loss": 0.050797734409570694,
+      "loss_ce": 9.277938806917518e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 796065108,
+      "step": 4637
+    },
+    {
+      "epoch": 52.67422096317281,
+      "grad_norm": 3.3647622613023693,
+      "learning_rate": 5e-06,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 796237156,
+      "step": 4638
+    },
+    {
+      "epoch": 52.67422096317281,
+      "loss": 0.051536764949560165,
+      "loss_ce": 0.0001146470385720022,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 796237156,
+      "step": 4638
+    },
+    {
+      "epoch": 52.685552407932015,
+      "grad_norm": 3.463757725684601,
+      "learning_rate": 5e-06,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 796408116,
+      "step": 4639
+    },
+    {
+      "epoch": 52.685552407932015,
+      "loss": 0.07609277963638306,
+      "loss_ce": 0.00016504891391377896,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 796408116,
+      "step": 4639
+    },
+    {
+      "epoch": 52.696883852691215,
+      "grad_norm": 3.6596971834167484,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 796578348,
+      "step": 4640
+    },
+    {
+      "epoch": 52.696883852691215,
+      "loss": 0.0635182112455368,
+      "loss_ce": 0.00037734012585133314,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 796578348,
+      "step": 4640
+    },
+    {
+      "epoch": 52.70821529745042,
+      "grad_norm": 3.6799844599966494,
+      "learning_rate": 5e-06,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 796749216,
+      "step": 4641
+    },
+    {
+      "epoch": 52.70821529745042,
+      "loss": 0.05242349952459335,
+      "loss_ce": 0.00013162585673853755,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 796749216,
+      "step": 4641
+    },
+    {
+      "epoch": 52.71954674220963,
+      "grad_norm": 3.100191065797761,
+      "learning_rate": 5e-06,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 796921008,
+      "step": 4642
+    },
+    {
+      "epoch": 52.71954674220963,
+      "loss": 0.10974981635808945,
+      "loss_ce": 5.437968502519652e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 796921008,
+      "step": 4642
+    },
+    {
+      "epoch": 52.73087818696884,
+      "grad_norm": 3.731808353896168,
+      "learning_rate": 5e-06,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 797092816,
+      "step": 4643
+    },
+    {
+      "epoch": 52.73087818696884,
+      "loss": 0.05236506834626198,
+      "loss_ce": 0.0005004440317861736,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 797092816,
+      "step": 4643
+    },
+    {
+      "epoch": 52.742209631728045,
+      "grad_norm": 4.16604243397578,
+      "learning_rate": 5e-06,
+      "loss": 0.0644,
+      "num_input_tokens_seen": 797264816,
+      "step": 4644
+    },
+    {
+      "epoch": 52.742209631728045,
+      "loss": 0.08338037133216858,
+      "loss_ce": 0.00026574498042464256,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 797264816,
+      "step": 4644
+    },
+    {
+      "epoch": 52.75354107648725,
+      "grad_norm": 3.3287556650560353,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 797436136,
+      "step": 4645
+    },
+    {
+      "epoch": 52.75354107648725,
+      "loss": 0.0746200680732727,
+      "loss_ce": 0.00046235433546826243,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 797436136,
+      "step": 4645
+    },
+    {
+      "epoch": 52.76487252124646,
+      "grad_norm": 3.1747378055032045,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 797607076,
+      "step": 4646
+    },
+    {
+      "epoch": 52.76487252124646,
+      "loss": 0.13902829587459564,
+      "loss_ce": 0.00012754701310768723,
+      "loss_iou": 0.375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 797607076,
+      "step": 4646
+    },
+    {
+      "epoch": 52.77620396600567,
+      "grad_norm": 3.1970781674124735,
+      "learning_rate": 5e-06,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 797778688,
+      "step": 4647
+    },
+    {
+      "epoch": 52.77620396600567,
+      "loss": 0.056480772793293,
+      "loss_ce": 0.0002216164139099419,
+      "loss_iou": 0.484375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 797778688,
+      "step": 4647
+    },
+    {
+      "epoch": 52.787535410764875,
+      "grad_norm": 3.6266024786709887,
+      "learning_rate": 5e-06,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 797950376,
+      "step": 4648
+    },
+    {
+      "epoch": 52.787535410764875,
+      "loss": 0.04782555252313614,
+      "loss_ce": 8.079847611952573e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 797950376,
+      "step": 4648
+    },
+    {
+      "epoch": 52.79886685552408,
+      "grad_norm": 3.8045115743989366,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 798121512,
+      "step": 4649
+    },
+    {
+      "epoch": 52.79886685552408,
+      "loss": 0.03150452673435211,
+      "loss_ce": 0.00019348980276845396,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 798121512,
+      "step": 4649
+    },
+    {
+      "epoch": 52.81019830028329,
+      "grad_norm": 9.042861900432891,
+      "learning_rate": 5e-06,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 798293176,
+      "step": 4650
+    },
+    {
+      "epoch": 52.81019830028329,
+      "loss": 0.0683264285326004,
+      "loss_ce": 0.00015015466487966478,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 798293176,
+      "step": 4650
+    },
+    {
+      "epoch": 52.82152974504249,
+      "grad_norm": 3.094945140900843,
+      "learning_rate": 5e-06,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 798464928,
+      "step": 4651
+    },
+    {
+      "epoch": 52.82152974504249,
+      "loss": 0.063038170337677,
+      "loss_ce": 0.00021773495245724916,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 798464928,
+      "step": 4651
+    },
+    {
+      "epoch": 52.8328611898017,
+      "grad_norm": 4.842813103805323,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 798636896,
+      "step": 4652
+    },
+    {
+      "epoch": 52.8328611898017,
+      "loss": 0.05820214003324509,
+      "loss_ce": 0.0001271892833756283,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 798636896,
+      "step": 4652
+    },
+    {
+      "epoch": 52.844192634560905,
+      "grad_norm": 3.99634349426944,
+      "learning_rate": 5e-06,
+      "loss": 0.0972,
+      "num_input_tokens_seen": 798808904,
+      "step": 4653
+    },
+    {
+      "epoch": 52.844192634560905,
+      "loss": 0.12208408862352371,
+      "loss_ce": 0.000563089910428971,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 798808904,
+      "step": 4653
+    },
+    {
+      "epoch": 52.85552407932011,
+      "grad_norm": 3.7155214222231163,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 798980428,
+      "step": 4654
+    },
+    {
+      "epoch": 52.85552407932011,
+      "loss": 0.1009514182806015,
+      "loss_ce": 0.0002434083289699629,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 798980428,
+      "step": 4654
+    },
+    {
+      "epoch": 52.86685552407932,
+      "grad_norm": 3.8699393792472097,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 799151540,
+      "step": 4655
+    },
+    {
+      "epoch": 52.86685552407932,
+      "loss": 0.03530982509255409,
+      "loss_ce": 0.00021460725110955536,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 799151540,
+      "step": 4655
+    },
+    {
+      "epoch": 52.87818696883853,
+      "grad_norm": 3.8444299838952127,
+      "learning_rate": 5e-06,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 799322788,
+      "step": 4656
+    },
+    {
+      "epoch": 52.87818696883853,
+      "loss": 0.06876468658447266,
+      "loss_ce": 0.00010013507562689483,
+      "loss_iou": 0.5,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 799322788,
+      "step": 4656
+    },
+    {
+      "epoch": 52.889518413597735,
+      "grad_norm": 3.245113541170879,
+      "learning_rate": 5e-06,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 799494620,
+      "step": 4657
+    },
+    {
+      "epoch": 52.889518413597735,
+      "loss": 0.06102427467703819,
+      "loss_ce": 0.00023325718939304352,
+      "loss_iou": 0.421875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 799494620,
+      "step": 4657
+    },
+    {
+      "epoch": 52.90084985835694,
+      "grad_norm": 3.3406453498419255,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 799666216,
+      "step": 4658
+    },
+    {
+      "epoch": 52.90084985835694,
+      "loss": 0.1429997831583023,
+      "loss_ce": 0.00010122729872819036,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 799666216,
+      "step": 4658
+    },
+    {
+      "epoch": 52.91218130311615,
+      "grad_norm": 3.5451157154569657,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 799837884,
+      "step": 4659
+    },
+    {
+      "epoch": 52.91218130311615,
+      "loss": 0.05081760138273239,
+      "loss_ce": 0.00031101072090677917,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 799837884,
+      "step": 4659
+    },
+    {
+      "epoch": 52.92351274787536,
+      "grad_norm": 2.9378663856248286,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 800008636,
+      "step": 4660
+    },
+    {
+      "epoch": 52.92351274787536,
+      "loss": 0.09843829274177551,
+      "loss_ce": 0.000141167503898032,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 800008636,
+      "step": 4660
+    },
+    {
+      "epoch": 52.934844192634564,
+      "grad_norm": 3.0602509644733886,
+      "learning_rate": 5e-06,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 800180580,
+      "step": 4661
+    },
+    {
+      "epoch": 52.934844192634564,
+      "loss": 0.043142370879650116,
+      "loss_ce": 0.00021939734870102257,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 800180580,
+      "step": 4661
+    },
+    {
+      "epoch": 52.946175637393765,
+      "grad_norm": 3.2067073291695167,
+      "learning_rate": 5e-06,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 800352152,
+      "step": 4662
+    },
+    {
+      "epoch": 52.946175637393765,
+      "loss": 0.05809144675731659,
+      "loss_ce": 7.753477257210761e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 800352152,
+      "step": 4662
+    },
+    {
+      "epoch": 52.95750708215297,
+      "grad_norm": 4.853015698655401,
+      "learning_rate": 5e-06,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 800523876,
+      "step": 4663
+    },
+    {
+      "epoch": 52.95750708215297,
+      "loss": 0.05939960479736328,
+      "loss_ce": 0.00021076375560369343,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 800523876,
+      "step": 4663
+    },
+    {
+      "epoch": 52.96883852691218,
+      "grad_norm": 3.8007122424660493,
+      "learning_rate": 5e-06,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 800695456,
+      "step": 4664
+    },
+    {
+      "epoch": 52.96883852691218,
+      "loss": 0.06708081811666489,
+      "loss_ce": 0.00036939175333827734,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 800695456,
+      "step": 4664
+    },
+    {
+      "epoch": 52.98016997167139,
+      "grad_norm": 3.5606164941183156,
+      "learning_rate": 5e-06,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 800865964,
+      "step": 4665
+    },
+    {
+      "epoch": 52.98016997167139,
+      "loss": 0.04869307577610016,
+      "loss_ce": 0.00023116446391213685,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 800865964,
+      "step": 4665
+    },
+    {
+      "epoch": 52.991501416430594,
+      "grad_norm": 5.552846294414013,
+      "learning_rate": 5e-06,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 801037732,
+      "step": 4666
+    },
+    {
+      "epoch": 52.991501416430594,
+      "loss": 0.04968515783548355,
+      "loss_ce": 0.00025431145331822336,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 801037732,
+      "step": 4666
+    },
+    {
+      "epoch": 52.991501416430594,
+      "loss": 0.08807050436735153,
+      "loss_ce": 0.0001188477617688477,
+      "loss_iou": 0.05029296875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 801166568,
+      "step": 4666
+    },
+    {
+      "epoch": 53.0028328611898,
+      "grad_norm": 3.6183639870460675,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 801209384,
+      "step": 4667
+    },
+    {
+      "epoch": 53.0028328611898,
+      "loss": 0.1584804654121399,
+      "loss_ce": 0.00015527110372204334,
+      "loss_iou": 0.7265625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 801209384,
+      "step": 4667
+    },
+    {
+      "epoch": 53.01416430594901,
+      "grad_norm": 4.737952616452396,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 801381352,
+      "step": 4668
+    },
+    {
+      "epoch": 53.01416430594901,
+      "loss": 0.0649605393409729,
+      "loss_ce": 0.00023275637067854404,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 801381352,
+      "step": 4668
+    },
+    {
+      "epoch": 53.02549575070822,
+      "grad_norm": 5.7467922276591175,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 801551968,
+      "step": 4669
+    },
+    {
+      "epoch": 53.02549575070822,
+      "loss": 0.11760212481021881,
+      "loss_ce": 0.00020100100664421916,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 801551968,
+      "step": 4669
+    },
+    {
+      "epoch": 53.036827195467424,
+      "grad_norm": 6.994592180657559,
+      "learning_rate": 5e-06,
+      "loss": 0.055,
+      "num_input_tokens_seen": 801720804,
+      "step": 4670
+    },
+    {
+      "epoch": 53.036827195467424,
+      "loss": 0.06153670325875282,
+      "loss_ce": 0.00019637009245343506,
+      "loss_iou": 0.484375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 801720804,
+      "step": 4670
+    },
+    {
+      "epoch": 53.04815864022663,
+      "grad_norm": 3.581475052913095,
+      "learning_rate": 5e-06,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 801892200,
+      "step": 4671
+    },
+    {
+      "epoch": 53.04815864022663,
+      "loss": 0.04765965789556503,
+      "loss_ce": 0.000250599579885602,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 801892200,
+      "step": 4671
+    },
+    {
+      "epoch": 53.05949008498584,
+      "grad_norm": 4.216413776684863,
+      "learning_rate": 5e-06,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 802064272,
+      "step": 4672
+    },
+    {
+      "epoch": 53.05949008498584,
+      "loss": 0.03942747414112091,
+      "loss_ce": 9.03129912330769e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 802064272,
+      "step": 4672
+    },
+    {
+      "epoch": 53.07082152974504,
+      "grad_norm": 3.534813613488904,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 802232532,
+      "step": 4673
+    },
+    {
+      "epoch": 53.07082152974504,
+      "loss": 0.07966461032629013,
+      "loss_ce": 0.00016632223560009152,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 802232532,
+      "step": 4673
+    },
+    {
+      "epoch": 53.08215297450425,
+      "grad_norm": 3.1714678883847984,
+      "learning_rate": 5e-06,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 802404136,
+      "step": 4674
+    },
+    {
+      "epoch": 53.08215297450425,
+      "loss": 0.05684254318475723,
+      "loss_ce": 0.00023243525356519967,
+      "loss_iou": 0.07421875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 802404136,
+      "step": 4674
+    },
+    {
+      "epoch": 53.093484419263454,
+      "grad_norm": 2.940926377124737,
+      "learning_rate": 5e-06,
+      "loss": 0.067,
+      "num_input_tokens_seen": 802574764,
+      "step": 4675
+    },
+    {
+      "epoch": 53.093484419263454,
+      "loss": 0.03710676729679108,
+      "loss_ce": 0.00036360337981022894,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 802574764,
+      "step": 4675
+    },
+    {
+      "epoch": 53.10481586402266,
+      "grad_norm": 2.8071621213057893,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 802746548,
+      "step": 4676
+    },
+    {
+      "epoch": 53.10481586402266,
+      "loss": 0.039691343903541565,
+      "loss_ce": 0.00017107937310356647,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 802746548,
+      "step": 4676
+    },
+    {
+      "epoch": 53.11614730878187,
+      "grad_norm": 4.215744155765197,
+      "learning_rate": 5e-06,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 802918028,
+      "step": 4677
+    },
+    {
+      "epoch": 53.11614730878187,
+      "loss": 0.08283141255378723,
+      "loss_ce": 0.00018980893946718425,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 802918028,
+      "step": 4677
+    },
+    {
+      "epoch": 53.12747875354108,
+      "grad_norm": 3.748771855538572,
+      "learning_rate": 5e-06,
+      "loss": 0.089,
+      "num_input_tokens_seen": 803089556,
+      "step": 4678
+    },
+    {
+      "epoch": 53.12747875354108,
+      "loss": 0.05010093003511429,
+      "loss_ce": 0.00023520737886428833,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 803089556,
+      "step": 4678
+    },
+    {
+      "epoch": 53.138810198300284,
+      "grad_norm": 3.3817925588098174,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 803261660,
+      "step": 4679
+    },
+    {
+      "epoch": 53.138810198300284,
+      "loss": 0.0462978333234787,
+      "loss_ce": 7.896107126725838e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 803261660,
+      "step": 4679
+    },
+    {
+      "epoch": 53.15014164305949,
+      "grad_norm": 3.210287937189868,
+      "learning_rate": 5e-06,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 803431452,
+      "step": 4680
+    },
+    {
+      "epoch": 53.15014164305949,
+      "loss": 0.03854640573263168,
+      "loss_ce": 0.00014003329852130264,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 803431452,
+      "step": 4680
+    },
+    {
+      "epoch": 53.1614730878187,
+      "grad_norm": 3.4720640179637225,
+      "learning_rate": 5e-06,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 803601552,
+      "step": 4681
+    },
+    {
+      "epoch": 53.1614730878187,
+      "loss": 0.04625174030661583,
+      "loss_ce": 0.0002159738796763122,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 803601552,
+      "step": 4681
+    },
+    {
+      "epoch": 53.172804532577906,
+      "grad_norm": 3.6961143800626117,
+      "learning_rate": 5e-06,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 803771352,
+      "step": 4682
+    },
+    {
+      "epoch": 53.172804532577906,
+      "loss": 0.07767904549837112,
+      "loss_ce": 0.00013387876970227808,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 803771352,
+      "step": 4682
+    },
+    {
+      "epoch": 53.184135977337114,
+      "grad_norm": 3.282811122322535,
+      "learning_rate": 5e-06,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 803943016,
+      "step": 4683
+    },
+    {
+      "epoch": 53.184135977337114,
+      "loss": 0.06307847797870636,
+      "loss_ce": 0.0001970052398974076,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 803943016,
+      "step": 4683
+    },
+    {
+      "epoch": 53.195467422096314,
+      "grad_norm": 5.781785936193978,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 804114688,
+      "step": 4684
+    },
+    {
+      "epoch": 53.195467422096314,
+      "loss": 0.052209023386240005,
+      "loss_ce": 0.000100260331237223,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 804114688,
+      "step": 4684
+    },
+    {
+      "epoch": 53.20679886685552,
+      "grad_norm": 4.025053206139338,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 804286568,
+      "step": 4685
+    },
+    {
+      "epoch": 53.20679886685552,
+      "loss": 0.03820769861340523,
+      "loss_ce": 0.00010650143667589873,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 804286568,
+      "step": 4685
+    },
+    {
+      "epoch": 53.21813031161473,
+      "grad_norm": 3.574836476854667,
+      "learning_rate": 5e-06,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 804458068,
+      "step": 4686
+    },
+    {
+      "epoch": 53.21813031161473,
+      "loss": 0.07929275184869766,
+      "loss_ce": 0.0001606728765182197,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 804458068,
+      "step": 4686
+    },
+    {
+      "epoch": 53.22946175637394,
+      "grad_norm": 2.9335493824511785,
+      "learning_rate": 5e-06,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 804629840,
+      "step": 4687
+    },
+    {
+      "epoch": 53.22946175637394,
+      "loss": 0.033859480172395706,
+      "loss_ce": 6.88940635882318e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 804629840,
+      "step": 4687
+    },
+    {
+      "epoch": 53.240793201133144,
+      "grad_norm": 3.4803271643586617,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 804801292,
+      "step": 4688
+    },
+    {
+      "epoch": 53.240793201133144,
+      "loss": 0.21241691708564758,
+      "loss_ce": 4.5087228500051424e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 804801292,
+      "step": 4688
+    },
+    {
+      "epoch": 53.25212464589235,
+      "grad_norm": 3.339061998052338,
+      "learning_rate": 5e-06,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 804972080,
+      "step": 4689
+    },
+    {
+      "epoch": 53.25212464589235,
+      "loss": 0.036198146641254425,
+      "loss_ce": 0.00033999019069597125,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 804972080,
+      "step": 4689
+    },
+    {
+      "epoch": 53.26345609065156,
+      "grad_norm": 3.4130839081378106,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 805144104,
+      "step": 4690
+    },
+    {
+      "epoch": 53.26345609065156,
+      "loss": 0.06861278414726257,
+      "loss_ce": 0.0001466000103391707,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 805144104,
+      "step": 4690
+    },
+    {
+      "epoch": 53.274787535410766,
+      "grad_norm": 4.101209750153936,
+      "learning_rate": 5e-06,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 805316084,
+      "step": 4691
+    },
+    {
+      "epoch": 53.274787535410766,
+      "loss": 0.09842968732118607,
+      "loss_ce": 0.00025463569909334183,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 805316084,
+      "step": 4691
+    },
+    {
+      "epoch": 53.286118980169974,
+      "grad_norm": 4.216783410876979,
+      "learning_rate": 5e-06,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 805488052,
+      "step": 4692
+    },
+    {
+      "epoch": 53.286118980169974,
+      "loss": 0.10868196189403534,
+      "loss_ce": 0.00013093504821881652,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 805488052,
+      "step": 4692
+    },
+    {
+      "epoch": 53.29745042492918,
+      "grad_norm": 3.709801538829773,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 805659744,
+      "step": 4693
+    },
+    {
+      "epoch": 53.29745042492918,
+      "loss": 0.08964075893163681,
+      "loss_ce": 7.166880823206156e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 805659744,
+      "step": 4693
+    },
+    {
+      "epoch": 53.30878186968839,
+      "grad_norm": 3.5890266602641114,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 805828452,
+      "step": 4694
+    },
+    {
+      "epoch": 53.30878186968839,
+      "loss": 0.05880039557814598,
+      "loss_ce": 5.405877891462296e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 805828452,
+      "step": 4694
+    },
+    {
+      "epoch": 53.32011331444759,
+      "grad_norm": 3.184967900399177,
+      "learning_rate": 5e-06,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 806000476,
+      "step": 4695
+    },
+    {
+      "epoch": 53.32011331444759,
+      "loss": 0.05769744887948036,
+      "loss_ce": 0.00011078018724219874,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 806000476,
+      "step": 4695
+    },
+    {
+      "epoch": 53.331444759206796,
+      "grad_norm": 3.7570691194562627,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 806172588,
+      "step": 4696
+    },
+    {
+      "epoch": 53.331444759206796,
+      "loss": 0.10235042870044708,
+      "loss_ce": 0.0007268953486345708,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 806172588,
+      "step": 4696
+    },
+    {
+      "epoch": 53.342776203966004,
+      "grad_norm": 3.362923314806938,
+      "learning_rate": 5e-06,
+      "loss": 0.054,
+      "num_input_tokens_seen": 806344556,
+      "step": 4697
+    },
+    {
+      "epoch": 53.342776203966004,
+      "loss": 0.049817316234111786,
+      "loss_ce": 0.0001652156061027199,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 806344556,
+      "step": 4697
+    },
+    {
+      "epoch": 53.35410764872521,
+      "grad_norm": 3.1716614065854,
+      "learning_rate": 5e-06,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 806515664,
+      "step": 4698
+    },
+    {
+      "epoch": 53.35410764872521,
+      "loss": 0.05348663777112961,
+      "loss_ce": 0.00017242977628484368,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 806515664,
+      "step": 4698
+    },
+    {
+      "epoch": 53.36543909348442,
+      "grad_norm": 3.115464620621151,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 806686296,
+      "step": 4699
+    },
+    {
+      "epoch": 53.36543909348442,
+      "loss": 0.08509091287851334,
+      "loss_ce": 0.0002673053531907499,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 806686296,
+      "step": 4699
+    },
+    {
+      "epoch": 53.376770538243626,
+      "grad_norm": 3.920512052086018,
+      "learning_rate": 5e-06,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 806858380,
+      "step": 4700
+    },
+    {
+      "epoch": 53.376770538243626,
+      "loss": 0.04476987570524216,
+      "loss_ce": 6.162213685456663e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 806858380,
+      "step": 4700
+    },
+    {
+      "epoch": 53.388101983002834,
+      "grad_norm": 3.8555820734061257,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 807030260,
+      "step": 4701
+    },
+    {
+      "epoch": 53.388101983002834,
+      "loss": 0.043867938220500946,
+      "loss_ce": 7.52117921365425e-05,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 807030260,
+      "step": 4701
+    },
+    {
+      "epoch": 53.39943342776204,
+      "grad_norm": 2.724184268662446,
+      "learning_rate": 5e-06,
+      "loss": 0.045,
+      "num_input_tokens_seen": 807201372,
+      "step": 4702
+    },
+    {
+      "epoch": 53.39943342776204,
+      "loss": 0.04772461578249931,
+      "loss_ce": 0.0002697823219932616,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 807201372,
+      "step": 4702
+    },
+    {
+      "epoch": 53.41076487252125,
+      "grad_norm": 2.8315897322615924,
+      "learning_rate": 5e-06,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 807372492,
+      "step": 4703
+    },
+    {
+      "epoch": 53.41076487252125,
+      "loss": 0.03447873890399933,
+      "loss_ce": 8.542610157746822e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 807372492,
+      "step": 4703
+    },
+    {
+      "epoch": 53.422096317280456,
+      "grad_norm": 5.854905697449762,
+      "learning_rate": 5e-06,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 807543508,
+      "step": 4704
+    },
+    {
+      "epoch": 53.422096317280456,
+      "loss": 0.11171075701713562,
+      "loss_ce": 7.745045877527446e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 807543508,
+      "step": 4704
+    },
+    {
+      "epoch": 53.43342776203966,
+      "grad_norm": 4.204693793117543,
+      "learning_rate": 5e-06,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 807715544,
+      "step": 4705
+    },
+    {
+      "epoch": 53.43342776203966,
+      "loss": 0.05382070690393448,
+      "loss_ce": 0.00024709681747481227,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 807715544,
+      "step": 4705
+    },
+    {
+      "epoch": 53.444759206798864,
+      "grad_norm": 3.9195419395884,
+      "learning_rate": 5e-06,
+      "loss": 0.103,
+      "num_input_tokens_seen": 807887324,
+      "step": 4706
+    },
+    {
+      "epoch": 53.444759206798864,
+      "loss": 0.061490319669246674,
+      "loss_ce": 0.00010421065235277638,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 807887324,
+      "step": 4706
+    },
+    {
+      "epoch": 53.45609065155807,
+      "grad_norm": 4.3423556259341884,
+      "learning_rate": 5e-06,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 808059196,
+      "step": 4707
+    },
+    {
+      "epoch": 53.45609065155807,
+      "loss": 0.04491560906171799,
+      "loss_ce": 0.00013106442929711193,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 808059196,
+      "step": 4707
+    },
+    {
+      "epoch": 53.46742209631728,
+      "grad_norm": 3.9545664113392722,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 808230340,
+      "step": 4708
+    },
+    {
+      "epoch": 53.46742209631728,
+      "loss": 0.07541818171739578,
+      "loss_ce": 0.00010080445645144209,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 808230340,
+      "step": 4708
+    },
+    {
+      "epoch": 53.478753541076486,
+      "grad_norm": 3.844563904266706,
+      "learning_rate": 5e-06,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 808402108,
+      "step": 4709
+    },
+    {
+      "epoch": 53.478753541076486,
+      "loss": 0.05846604332327843,
+      "loss_ce": 0.00014695180288981646,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 808402108,
+      "step": 4709
+    },
+    {
+      "epoch": 53.49008498583569,
+      "grad_norm": 3.8091888115848085,
+      "learning_rate": 5e-06,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 808573440,
+      "step": 4710
+    },
+    {
+      "epoch": 53.49008498583569,
+      "loss": 0.03224262222647667,
+      "loss_ce": 0.0001533899048808962,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 808573440,
+      "step": 4710
+    },
+    {
+      "epoch": 53.5014164305949,
+      "grad_norm": 3.288774773185425,
+      "learning_rate": 5e-06,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 808745364,
+      "step": 4711
+    },
+    {
+      "epoch": 53.5014164305949,
+      "loss": 0.0679798498749733,
+      "loss_ce": 6.297829531831667e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 808745364,
+      "step": 4711
+    },
+    {
+      "epoch": 53.51274787535411,
+      "grad_norm": 3.467909028755271,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 808916396,
+      "step": 4712
+    },
+    {
+      "epoch": 53.51274787535411,
+      "loss": 0.15797144174575806,
+      "loss_ce": 0.0001955716434167698,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 808916396,
+      "step": 4712
+    },
+    {
+      "epoch": 53.524079320113316,
+      "grad_norm": 3.083361360268773,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 809087332,
+      "step": 4713
+    },
+    {
+      "epoch": 53.524079320113316,
+      "loss": 0.04382530599832535,
+      "loss_ce": 3.258429205743596e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 809087332,
+      "step": 4713
+    },
+    {
+      "epoch": 53.53541076487252,
+      "grad_norm": 3.4923296446901952,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 809258808,
+      "step": 4714
+    },
+    {
+      "epoch": 53.53541076487252,
+      "loss": 0.05811478942632675,
+      "loss_ce": 8.561462163925171e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 809258808,
+      "step": 4714
+    },
+    {
+      "epoch": 53.54674220963173,
+      "grad_norm": 3.8627771173012437,
+      "learning_rate": 5e-06,
+      "loss": 0.083,
+      "num_input_tokens_seen": 809430676,
+      "step": 4715
+    },
+    {
+      "epoch": 53.54674220963173,
+      "loss": 0.05545062571763992,
+      "loss_ce": 9.174172737402841e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 809430676,
+      "step": 4715
+    },
+    {
+      "epoch": 53.55807365439094,
+      "grad_norm": 3.298441864008281,
+      "learning_rate": 5e-06,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 809601628,
+      "step": 4716
+    },
+    {
+      "epoch": 53.55807365439094,
+      "loss": 0.038653187453746796,
+      "loss_ce": 0.00030784669797867537,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 809601628,
+      "step": 4716
+    },
+    {
+      "epoch": 53.56940509915014,
+      "grad_norm": 3.812308088539133,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 809772656,
+      "step": 4717
+    },
+    {
+      "epoch": 53.56940509915014,
+      "loss": 0.08617973327636719,
+      "loss_ce": 5.913038330618292e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 809772656,
+      "step": 4717
+    },
+    {
+      "epoch": 53.580736543909346,
+      "grad_norm": 2.896232334976172,
+      "learning_rate": 5e-06,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 809944568,
+      "step": 4718
+    },
+    {
+      "epoch": 53.580736543909346,
+      "loss": 0.030894268304109573,
+      "loss_ce": 0.0001172898628283292,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 809944568,
+      "step": 4718
+    },
+    {
+      "epoch": 53.59206798866855,
+      "grad_norm": 3.0005585140462996,
+      "learning_rate": 5e-06,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 810116176,
+      "step": 4719
+    },
+    {
+      "epoch": 53.59206798866855,
+      "loss": 0.0554022379219532,
+      "loss_ce": 8.912795601645485e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 810116176,
+      "step": 4719
+    },
+    {
+      "epoch": 53.60339943342776,
+      "grad_norm": 3.6880261578395483,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 810285660,
+      "step": 4720
+    },
+    {
+      "epoch": 53.60339943342776,
+      "loss": 0.1042427197098732,
+      "loss_ce": 0.00016252268687821925,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 810285660,
+      "step": 4720
+    },
+    {
+      "epoch": 53.61473087818697,
+      "grad_norm": 3.639951122224193,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 810457344,
+      "step": 4721
+    },
+    {
+      "epoch": 53.61473087818697,
+      "loss": 0.0848560705780983,
+      "loss_ce": 0.00020030523592140526,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 810457344,
+      "step": 4721
+    },
+    {
+      "epoch": 53.626062322946176,
+      "grad_norm": 3.7371813028541077,
+      "learning_rate": 5e-06,
+      "loss": 0.0927,
+      "num_input_tokens_seen": 810628152,
+      "step": 4722
+    },
+    {
+      "epoch": 53.626062322946176,
+      "loss": 0.09531824290752411,
+      "loss_ce": 0.00016442826017737389,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 810628152,
+      "step": 4722
+    },
+    {
+      "epoch": 53.63739376770538,
+      "grad_norm": 3.6873669335555217,
+      "learning_rate": 5e-06,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 810799664,
+      "step": 4723
+    },
+    {
+      "epoch": 53.63739376770538,
+      "loss": 0.08019520342350006,
+      "loss_ce": 0.00014759181067347527,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 810799664,
+      "step": 4723
+    },
+    {
+      "epoch": 53.64872521246459,
+      "grad_norm": 5.009013222489445,
+      "learning_rate": 5e-06,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 810969668,
+      "step": 4724
+    },
+    {
+      "epoch": 53.64872521246459,
+      "loss": 0.05545923113822937,
+      "loss_ce": 5.456890357891098e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 810969668,
+      "step": 4724
+    },
+    {
+      "epoch": 53.6600566572238,
+      "grad_norm": 3.669644026865973,
+      "learning_rate": 5e-06,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 811140488,
+      "step": 4725
+    },
+    {
+      "epoch": 53.6600566572238,
+      "loss": 0.07152142375707626,
+      "loss_ce": 0.00035443459637463093,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 811140488,
+      "step": 4725
+    },
+    {
+      "epoch": 53.671388101983005,
+      "grad_norm": 5.499783017178718,
+      "learning_rate": 5e-06,
+      "loss": 0.066,
+      "num_input_tokens_seen": 811311344,
+      "step": 4726
+    },
+    {
+      "epoch": 53.671388101983005,
+      "loss": 0.06037700176239014,
+      "loss_ce": 0.0003184127854183316,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 811311344,
+      "step": 4726
+    },
+    {
+      "epoch": 53.68271954674221,
+      "grad_norm": 3.781040851411071,
+      "learning_rate": 5e-06,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 811483052,
+      "step": 4727
+    },
+    {
+      "epoch": 53.68271954674221,
+      "loss": 0.06411369144916534,
+      "loss_ce": 7.255301170516759e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 811483052,
+      "step": 4727
+    },
+    {
+      "epoch": 53.69405099150141,
+      "grad_norm": 3.9217344741463758,
+      "learning_rate": 5e-06,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 811655120,
+      "step": 4728
+    },
+    {
+      "epoch": 53.69405099150141,
+      "loss": 0.1315111666917801,
+      "loss_ce": 8.722052007215098e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 811655120,
+      "step": 4728
+    },
+    {
+      "epoch": 53.70538243626062,
+      "grad_norm": 3.7764690095459628,
+      "learning_rate": 5e-06,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 811826176,
+      "step": 4729
+    },
+    {
+      "epoch": 53.70538243626062,
+      "loss": 0.10992484539747238,
+      "loss_ce": 0.00016837533621583134,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 811826176,
+      "step": 4729
+    },
+    {
+      "epoch": 53.71671388101983,
+      "grad_norm": 5.493801959527736,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 811996400,
+      "step": 4730
+    },
+    {
+      "epoch": 53.71671388101983,
+      "loss": 0.07168786227703094,
+      "loss_ce": 0.00024620932526886463,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 811996400,
+      "step": 4730
+    },
+    {
+      "epoch": 53.728045325779036,
+      "grad_norm": 3.504280176538522,
+      "learning_rate": 5e-06,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 812168080,
+      "step": 4731
+    },
+    {
+      "epoch": 53.728045325779036,
+      "loss": 0.0738058090209961,
+      "loss_ce": 9.05952911125496e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 812168080,
+      "step": 4731
+    },
+    {
+      "epoch": 53.73937677053824,
+      "grad_norm": 3.3562273849600746,
+      "learning_rate": 5e-06,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 812339848,
+      "step": 4732
+    },
+    {
+      "epoch": 53.73937677053824,
+      "loss": 0.0745634213089943,
+      "loss_ce": 0.0009855427779257298,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 812339848,
+      "step": 4732
+    },
+    {
+      "epoch": 53.75070821529745,
+      "grad_norm": 2.992186732060685,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 812511904,
+      "step": 4733
+    },
+    {
+      "epoch": 53.75070821529745,
+      "loss": 0.04918941110372543,
+      "loss_ce": 0.00022395931591745466,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 812511904,
+      "step": 4733
+    },
+    {
+      "epoch": 53.76203966005666,
+      "grad_norm": 3.476255917572014,
+      "learning_rate": 5e-06,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 812683476,
+      "step": 4734
+    },
+    {
+      "epoch": 53.76203966005666,
+      "loss": 0.040006931871175766,
+      "loss_ce": 0.0001509758731117472,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 812683476,
+      "step": 4734
+    },
+    {
+      "epoch": 53.773371104815865,
+      "grad_norm": 4.74225071552281,
+      "learning_rate": 5e-06,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 812855144,
+      "step": 4735
+    },
+    {
+      "epoch": 53.773371104815865,
+      "loss": 0.058695729821920395,
+      "loss_ce": 8.671992691233754e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 812855144,
+      "step": 4735
+    },
+    {
+      "epoch": 53.78470254957507,
+      "grad_norm": 4.317753439416477,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 813026816,
+      "step": 4736
+    },
+    {
+      "epoch": 53.78470254957507,
+      "loss": 0.13734766840934753,
+      "loss_ce": 7.959892536746338e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 813026816,
+      "step": 4736
+    },
+    {
+      "epoch": 53.79603399433428,
+      "grad_norm": 4.06933961291216,
+      "learning_rate": 5e-06,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 813198868,
+      "step": 4737
+    },
+    {
+      "epoch": 53.79603399433428,
+      "loss": 0.06954001635313034,
+      "loss_ce": 9.726545977173373e-05,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 813198868,
+      "step": 4737
+    },
+    {
+      "epoch": 53.80736543909349,
+      "grad_norm": 4.732088376055132,
+      "learning_rate": 5e-06,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 813369232,
+      "step": 4738
+    },
+    {
+      "epoch": 53.80736543909349,
+      "loss": 0.09732776880264282,
+      "loss_ce": 6.825188029324636e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 813369232,
+      "step": 4738
+    },
+    {
+      "epoch": 53.81869688385269,
+      "grad_norm": 4.965105353213792,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 813540132,
+      "step": 4739
+    },
+    {
+      "epoch": 53.81869688385269,
+      "loss": 0.06733009219169617,
+      "loss_ce": 9.986882650991902e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 813540132,
+      "step": 4739
+    },
+    {
+      "epoch": 53.830028328611895,
+      "grad_norm": 3.440272746120115,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 813710804,
+      "step": 4740
+    },
+    {
+      "epoch": 53.830028328611895,
+      "loss": 0.13867655396461487,
+      "loss_ce": 0.00020304112695157528,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 813710804,
+      "step": 4740
+    },
+    {
+      "epoch": 53.8413597733711,
+      "grad_norm": 3.241899024244398,
+      "learning_rate": 5e-06,
+      "loss": 0.0793,
+      "num_input_tokens_seen": 813881780,
+      "step": 4741
+    },
+    {
+      "epoch": 53.8413597733711,
+      "loss": 0.04989277571439743,
+      "loss_ce": 2.7049361960962415e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 813881780,
+      "step": 4741
+    },
+    {
+      "epoch": 53.85269121813031,
+      "grad_norm": 3.3182526774776653,
+      "learning_rate": 5e-06,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 814053592,
+      "step": 4742
+    },
+    {
+      "epoch": 53.85269121813031,
+      "loss": 0.12495151907205582,
+      "loss_ce": 8.885106217348948e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 814053592,
+      "step": 4742
+    },
+    {
+      "epoch": 53.86402266288952,
+      "grad_norm": 3.300922566199744,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 814225508,
+      "step": 4743
+    },
+    {
+      "epoch": 53.86402266288952,
+      "loss": 0.1144113689661026,
+      "loss_ce": 4.674070078181103e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 814225508,
+      "step": 4743
+    },
+    {
+      "epoch": 53.875354107648725,
+      "grad_norm": 6.433502017316352,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 814395736,
+      "step": 4744
+    },
+    {
+      "epoch": 53.875354107648725,
+      "loss": 0.10933315008878708,
+      "loss_ce": 0.00015652200090698898,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 814395736,
+      "step": 4744
+    },
+    {
+      "epoch": 53.88668555240793,
+      "grad_norm": 3.1704300427332606,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 814567484,
+      "step": 4745
+    },
+    {
+      "epoch": 53.88668555240793,
+      "loss": 0.044446833431720734,
+      "loss_ce": 7.427643140545115e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 814567484,
+      "step": 4745
+    },
+    {
+      "epoch": 53.89801699716714,
+      "grad_norm": 3.0519088979500175,
+      "learning_rate": 5e-06,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 814739064,
+      "step": 4746
+    },
+    {
+      "epoch": 53.89801699716714,
+      "loss": 0.04025079309940338,
+      "loss_ce": 0.00012017632980132475,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 814739064,
+      "step": 4746
+    },
+    {
+      "epoch": 53.90934844192635,
+      "grad_norm": 2.7669753392707284,
+      "learning_rate": 5e-06,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 814909564,
+      "step": 4747
+    },
+    {
+      "epoch": 53.90934844192635,
+      "loss": 0.05583912879228592,
+      "loss_ce": 8.351239375770092e-05,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 814909564,
+      "step": 4747
+    },
+    {
+      "epoch": 53.920679886685555,
+      "grad_norm": 3.690798693609719,
+      "learning_rate": 5e-06,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 815080640,
+      "step": 4748
+    },
+    {
+      "epoch": 53.920679886685555,
+      "loss": 0.044106900691986084,
+      "loss_ce": 0.00011581404396565631,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 815080640,
+      "step": 4748
+    },
+    {
+      "epoch": 53.93201133144476,
+      "grad_norm": 3.709204693984996,
+      "learning_rate": 5e-06,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 815252568,
+      "step": 4749
+    },
+    {
+      "epoch": 53.93201133144476,
+      "loss": 0.08503993600606918,
+      "loss_ce": 0.00017055007629096508,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 815252568,
+      "step": 4749
+    },
+    {
+      "epoch": 53.94334277620396,
+      "grad_norm": 3.6783407282604093,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 815424420,
+      "step": 4750
+    },
+    {
+      "epoch": 53.94334277620396,
+      "eval_seeclick_CIoU": 0.5522235333919525,
+      "eval_seeclick_GIoU": 0.5461505651473999,
+      "eval_seeclick_IoU": 0.5867660045623779,
+      "eval_seeclick_MAE_all": 0.06227055564522743,
+      "eval_seeclick_MAE_h": 0.03112515900284052,
+      "eval_seeclick_MAE_w": 0.09046434238553047,
+      "eval_seeclick_MAE_x": 0.09699121490120888,
+      "eval_seeclick_MAE_y": 0.03050148393958807,
+      "eval_seeclick_NUM_probability": 0.9999385178089142,
+      "eval_seeclick_inside_bbox": 0.9076704680919647,
+      "eval_seeclick_loss": 0.8090967535972595,
+      "eval_seeclick_loss_ce": 0.5721153616905212,
+      "eval_seeclick_loss_iou": 0.54638671875,
+      "eval_seeclick_loss_num": 0.0470733642578125,
+      "eval_seeclick_loss_xval": 0.2353057861328125,
+      "eval_seeclick_runtime": 68.9236,
+      "eval_seeclick_samples_per_second": 0.624,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 815424420,
+      "step": 4750
+    },
+    {
+      "epoch": 53.94334277620396,
+      "eval_icons_CIoU": 0.7505115568637848,
+      "eval_icons_GIoU": 0.7503921985626221,
+      "eval_icons_IoU": 0.7632699310779572,
+      "eval_icons_MAE_all": 0.03318015206605196,
+      "eval_icons_MAE_h": 0.030470607802271843,
+      "eval_icons_MAE_w": 0.035882056690752506,
+      "eval_icons_MAE_x": 0.030328705441206694,
+      "eval_icons_MAE_y": 0.03603922761976719,
+      "eval_icons_NUM_probability": 0.998185783624649,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.11329381912946701,
+      "eval_icons_loss_ce": 0.002961780352052301,
+      "eval_icons_loss_iou": 0.53125,
+      "eval_icons_loss_num": 0.020236968994140625,
+      "eval_icons_loss_xval": 0.101165771484375,
+      "eval_icons_runtime": 78.9791,
+      "eval_icons_samples_per_second": 0.633,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 815424420,
+      "step": 4750
+    },
+    {
+      "epoch": 53.94334277620396,
+      "eval_screenspot_CIoU": 0.6188857158025106,
+      "eval_screenspot_GIoU": 0.6189959645271301,
+      "eval_screenspot_IoU": 0.6486753424008688,
+      "eval_screenspot_MAE_all": 0.06953854113817215,
+      "eval_screenspot_MAE_h": 0.04198170329133669,
+      "eval_screenspot_MAE_w": 0.12024158984422684,
+      "eval_screenspot_MAE_x": 0.08158182352781296,
+      "eval_screenspot_MAE_y": 0.03434905533989271,
+      "eval_screenspot_NUM_probability": 0.9999154011408488,
+      "eval_screenspot_inside_bbox": 0.8974999984105428,
+      "eval_screenspot_loss": 0.29695162177085876,
+      "eval_screenspot_loss_ce": 0.014681711016843716,
+      "eval_screenspot_loss_iou": 0.5032958984375,
+      "eval_screenspot_loss_num": 0.05596923828125,
+      "eval_screenspot_loss_xval": 0.2797444661458333,
+      "eval_screenspot_runtime": 147.3545,
+      "eval_screenspot_samples_per_second": 0.604,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 815424420,
+      "step": 4750
+    },
+    {
+      "epoch": 53.94334277620396,
+      "eval_compot_CIoU": 0.847744882106781,
+      "eval_compot_GIoU": 0.842746913433075,
+      "eval_compot_IoU": 0.8632494211196899,
+      "eval_compot_MAE_all": 0.028249988332390785,
+      "eval_compot_MAE_h": 0.022975671105086803,
+      "eval_compot_MAE_w": 0.034658296033740044,
+      "eval_compot_MAE_x": 0.03308155573904514,
+      "eval_compot_MAE_y": 0.022284426260739565,
+      "eval_compot_NUM_probability": 0.9999487102031708,
+      "eval_compot_inside_bbox": 0.9409722089767456,
+      "eval_compot_loss": 0.08661291003227234,
+      "eval_compot_loss_ce": 5.562421210925095e-05,
+      "eval_compot_loss_iou": 0.4775390625,
+      "eval_compot_loss_num": 0.014847755432128906,
+      "eval_compot_loss_xval": 0.07421875,
+      "eval_compot_runtime": 90.0244,
+      "eval_compot_samples_per_second": 0.555,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 815424420,
+      "step": 4750
+    },
+    {
+      "epoch": 53.94334277620396,
+      "eval_custom_ui_MAE_all": 0.020098551642149687,
+      "eval_custom_ui_MAE_x": 0.031903672963380814,
+      "eval_custom_ui_MAE_y": 0.00829342845827341,
+      "eval_custom_ui_NUM_probability": 0.999820202589035,
+      "eval_custom_ui_loss": 0.20859397947788239,
+      "eval_custom_ui_loss_ce": 0.10381979122757912,
+      "eval_custom_ui_loss_num": 0.01966094970703125,
+      "eval_custom_ui_loss_xval": 0.0983123779296875,
+      "eval_custom_ui_runtime": 58.7917,
+      "eval_custom_ui_samples_per_second": 0.85,
+      "eval_custom_ui_steps_per_second": 0.034,
+      "num_input_tokens_seen": 815424420,
+      "step": 4750
+    },
+    {
+      "epoch": 53.94334277620396,
+      "loss": 0.2475331723690033,
+      "loss_ce": 0.1268971860408783,
+      "loss_iou": 0.0,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 815424420,
+      "step": 4750
+    },
+    {
+      "epoch": 53.95467422096317,
+      "grad_norm": 3.0741142778363035,
+      "learning_rate": 5e-06,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 815596440,
+      "step": 4751
+    },
+    {
+      "epoch": 53.95467422096317,
+      "loss": 0.04185118153691292,
+      "loss_ce": 0.00011839439684990793,
+      "loss_iou": 0.328125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 815596440,
+      "step": 4751
+    },
+    {
+      "epoch": 53.96600566572238,
+      "grad_norm": 3.299156308551903,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 815767592,
+      "step": 4752
+    },
+    {
+      "epoch": 53.96600566572238,
+      "loss": 0.04948686435818672,
+      "loss_ce": 7.890626147855073e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 815767592,
+      "step": 4752
+    },
+    {
+      "epoch": 53.977337110481585,
+      "grad_norm": 3.5150675242539853,
+      "learning_rate": 5e-06,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 815938556,
+      "step": 4753
+    },
+    {
+      "epoch": 53.977337110481585,
+      "loss": 0.10413302481174469,
+      "loss_ce": 0.00014437444042414427,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 815938556,
+      "step": 4753
+    },
+    {
+      "epoch": 53.98866855524079,
+      "grad_norm": 3.897100742484984,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 816110692,
+      "step": 4754
+    },
+    {
+      "epoch": 53.98866855524079,
+      "loss": 0.168490469455719,
+      "loss_ce": 0.00014026153075974435,
+      "loss_iou": 0.34375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 816110692,
+      "step": 4754
+    },
+    {
+      "epoch": 54.0,
+      "grad_norm": 3.992371962658389,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 816282776,
+      "step": 4755
+    },
+    {
+      "epoch": 54.0,
+      "loss": 0.21594902873039246,
+      "loss_ce": 0.00014396820915862918,
+      "loss_iou": 0.546875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 816282776,
+      "step": 4755
+    },
+    {
+      "epoch": 54.01133144475921,
+      "grad_norm": 4.798275610049018,
+      "learning_rate": 5e-06,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 816454904,
+      "step": 4756
+    },
+    {
+      "epoch": 54.01133144475921,
+      "loss": 0.11086331307888031,
+      "loss_ce": 8.450080349575728e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 816454904,
+      "step": 4756
+    },
+    {
+      "epoch": 54.022662889518415,
+      "grad_norm": 4.0198215592283,
+      "learning_rate": 5e-06,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 816626276,
+      "step": 4757
+    },
+    {
+      "epoch": 54.022662889518415,
+      "loss": 0.07552963495254517,
+      "loss_ce": 0.00012069502554368228,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 816626276,
+      "step": 4757
+    },
+    {
+      "epoch": 54.03399433427762,
+      "grad_norm": 5.4977803106136145,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 816796748,
+      "step": 4758
+    },
+    {
+      "epoch": 54.03399433427762,
+      "loss": 0.062033962458372116,
+      "loss_ce": 0.00026638322742655873,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 816796748,
+      "step": 4758
+    },
+    {
+      "epoch": 54.04532577903683,
+      "grad_norm": 4.215787564761496,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 816967408,
+      "step": 4759
+    },
+    {
+      "epoch": 54.04532577903683,
+      "loss": 0.048968251794576645,
+      "loss_ce": 0.00010961154475808144,
+      "loss_iou": 0.0673828125,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 816967408,
+      "step": 4759
+    },
+    {
+      "epoch": 54.05665722379604,
+      "grad_norm": 2.854494039973909,
+      "learning_rate": 5e-06,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 817139676,
+      "step": 4760
+    },
+    {
+      "epoch": 54.05665722379604,
+      "loss": 0.07032491266727448,
+      "loss_ce": 0.00011922520934604108,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 817139676,
+      "step": 4760
+    },
+    {
+      "epoch": 54.06798866855524,
+      "grad_norm": 3.8089038966549555,
+      "learning_rate": 5e-06,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 817311660,
+      "step": 4761
+    },
+    {
+      "epoch": 54.06798866855524,
+      "loss": 0.060257233679294586,
+      "loss_ce": 7.656498928554356e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 817311660,
+      "step": 4761
+    },
+    {
+      "epoch": 54.079320113314445,
+      "grad_norm": 4.2336133692485145,
+      "learning_rate": 5e-06,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 817483628,
+      "step": 4762
+    },
+    {
+      "epoch": 54.079320113314445,
+      "loss": 0.054187577217817307,
+      "loss_ce": 0.00011043060658266768,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 817483628,
+      "step": 4762
+    },
+    {
+      "epoch": 54.09065155807365,
+      "grad_norm": 4.09759752186007,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 817654044,
+      "step": 4763
+    },
+    {
+      "epoch": 54.09065155807365,
+      "loss": 0.06972815841436386,
+      "loss_ce": 0.00020911383035127074,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 817654044,
+      "step": 4763
+    },
+    {
+      "epoch": 54.10198300283286,
+      "grad_norm": 4.304100044369193,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 817825252,
+      "step": 4764
+    },
+    {
+      "epoch": 54.10198300283286,
+      "loss": 0.07415296137332916,
+      "loss_ce": 0.0001173118653241545,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 817825252,
+      "step": 4764
+    },
+    {
+      "epoch": 54.11331444759207,
+      "grad_norm": 3.3664759933973154,
+      "learning_rate": 5e-06,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 817996288,
+      "step": 4765
+    },
+    {
+      "epoch": 54.11331444759207,
+      "loss": 0.043957240879535675,
+      "loss_ce": 7.296632247744128e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 817996288,
+      "step": 4765
+    },
+    {
+      "epoch": 54.124645892351275,
+      "grad_norm": 2.6768072844900317,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 818168404,
+      "step": 4766
+    },
+    {
+      "epoch": 54.124645892351275,
+      "loss": 0.08594170212745667,
+      "loss_ce": 4.9977148592006415e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 818168404,
+      "step": 4766
+    },
+    {
+      "epoch": 54.13597733711048,
+      "grad_norm": 3.0232286716919394,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 818338656,
+      "step": 4767
+    },
+    {
+      "epoch": 54.13597733711048,
+      "loss": 0.13524626195430756,
+      "loss_ce": 0.00016019925533328205,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 818338656,
+      "step": 4767
+    },
+    {
+      "epoch": 54.14730878186969,
+      "grad_norm": 3.923562074053229,
+      "learning_rate": 5e-06,
+      "loss": 0.1022,
+      "num_input_tokens_seen": 818510120,
+      "step": 4768
+    },
+    {
+      "epoch": 54.14730878186969,
+      "loss": 0.132354736328125,
+      "loss_ce": 0.00010681435378501192,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 818510120,
+      "step": 4768
+    },
+    {
+      "epoch": 54.1586402266289,
+      "grad_norm": 4.721469361154377,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 818681576,
+      "step": 4769
+    },
+    {
+      "epoch": 54.1586402266289,
+      "loss": 0.07277920097112656,
+      "loss_ce": 0.00013210689940024167,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 818681576,
+      "step": 4769
+    },
+    {
+      "epoch": 54.169971671388105,
+      "grad_norm": 3.110943331057615,
+      "learning_rate": 5e-06,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 818852528,
+      "step": 4770
+    },
+    {
+      "epoch": 54.169971671388105,
+      "loss": 0.03584219142794609,
+      "loss_ce": 2.9813183573423885e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 818852528,
+      "step": 4770
+    },
+    {
+      "epoch": 54.18130311614731,
+      "grad_norm": 4.983396811961203,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 819022796,
+      "step": 4771
+    },
+    {
+      "epoch": 54.18130311614731,
+      "loss": 0.05537346005439758,
+      "loss_ce": 6.03479020355735e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 819022796,
+      "step": 4771
+    },
+    {
+      "epoch": 54.19263456090651,
+      "grad_norm": 4.125428968528424,
+      "learning_rate": 5e-06,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 819192240,
+      "step": 4772
+    },
+    {
+      "epoch": 54.19263456090651,
+      "loss": 0.054470986127853394,
+      "loss_ce": 0.00018021788855548948,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 819192240,
+      "step": 4772
+    },
+    {
+      "epoch": 54.20396600566572,
+      "grad_norm": 3.196326019502915,
+      "learning_rate": 5e-06,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 819363680,
+      "step": 4773
+    },
+    {
+      "epoch": 54.20396600566572,
+      "loss": 0.06213907524943352,
+      "loss_ce": 6.63199316477403e-05,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 819363680,
+      "step": 4773
+    },
+    {
+      "epoch": 54.21529745042493,
+      "grad_norm": 3.220228013607888,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 819535452,
+      "step": 4774
+    },
+    {
+      "epoch": 54.21529745042493,
+      "loss": 0.059016890823841095,
+      "loss_ce": 0.00022477943275589496,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 819535452,
+      "step": 4774
+    },
+    {
+      "epoch": 54.226628895184135,
+      "grad_norm": 3.777664569826417,
+      "learning_rate": 5e-06,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 819706504,
+      "step": 4775
+    },
+    {
+      "epoch": 54.226628895184135,
+      "loss": 0.03237161785364151,
+      "loss_ce": 8.401776722166687e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 819706504,
+      "step": 4775
+    },
+    {
+      "epoch": 54.23796033994334,
+      "grad_norm": 3.3608117550637724,
+      "learning_rate": 5e-06,
+      "loss": 0.0672,
+      "num_input_tokens_seen": 819877244,
+      "step": 4776
+    },
+    {
+      "epoch": 54.23796033994334,
+      "loss": 0.04349207133054733,
+      "loss_ce": 0.00024866312742233276,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 819877244,
+      "step": 4776
+    },
+    {
+      "epoch": 54.24929178470255,
+      "grad_norm": 3.905544656208704,
+      "learning_rate": 5e-06,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 820048000,
+      "step": 4777
+    },
+    {
+      "epoch": 54.24929178470255,
+      "loss": 0.04804790019989014,
+      "loss_ce": 7.426677620969713e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 820048000,
+      "step": 4777
+    },
+    {
+      "epoch": 54.26062322946176,
+      "grad_norm": 4.476319027336858,
+      "learning_rate": 5e-06,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 820219732,
+      "step": 4778
+    },
+    {
+      "epoch": 54.26062322946176,
+      "loss": 0.05025298893451691,
+      "loss_ce": 8.208947838284075e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 820219732,
+      "step": 4778
+    },
+    {
+      "epoch": 54.271954674220964,
+      "grad_norm": 4.020817647951555,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 820391716,
+      "step": 4779
+    },
+    {
+      "epoch": 54.271954674220964,
+      "loss": 0.05510406196117401,
+      "loss_ce": 6.560588371939957e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 820391716,
+      "step": 4779
+    },
+    {
+      "epoch": 54.28328611898017,
+      "grad_norm": 3.3421576503910453,
+      "learning_rate": 5e-06,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 820563260,
+      "step": 4780
+    },
+    {
+      "epoch": 54.28328611898017,
+      "loss": 0.07333078980445862,
+      "loss_ce": 0.00027171504916623235,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 820563260,
+      "step": 4780
+    },
+    {
+      "epoch": 54.29461756373938,
+      "grad_norm": 3.928710372330599,
+      "learning_rate": 5e-06,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 820734788,
+      "step": 4781
+    },
+    {
+      "epoch": 54.29461756373938,
+      "loss": 0.04864327237010002,
+      "loss_ce": 5.9288166085025296e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 820734788,
+      "step": 4781
+    },
+    {
+      "epoch": 54.30594900849859,
+      "grad_norm": 4.136087283622599,
+      "learning_rate": 5e-06,
+      "loss": 0.047,
+      "num_input_tokens_seen": 820905200,
+      "step": 4782
+    },
+    {
+      "epoch": 54.30594900849859,
+      "loss": 0.04021323844790459,
+      "loss_ce": 9.788366151042283e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 820905200,
+      "step": 4782
+    },
+    {
+      "epoch": 54.31728045325779,
+      "grad_norm": 4.100370047611949,
+      "learning_rate": 5e-06,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 821076488,
+      "step": 4783
+    },
+    {
+      "epoch": 54.31728045325779,
+      "loss": 0.051346998661756516,
+      "loss_ce": 0.0001385059003951028,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 821076488,
+      "step": 4783
+    },
+    {
+      "epoch": 54.328611898016995,
+      "grad_norm": 4.355560878216246,
+      "learning_rate": 5e-06,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 821247328,
+      "step": 4784
+    },
+    {
+      "epoch": 54.328611898016995,
+      "loss": 0.03923734650015831,
+      "loss_ce": 8.329332194989547e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 821247328,
+      "step": 4784
+    },
+    {
+      "epoch": 54.3399433427762,
+      "grad_norm": 3.6971697671546213,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 821419012,
+      "step": 4785
+    },
+    {
+      "epoch": 54.3399433427762,
+      "loss": 0.08891374617815018,
+      "loss_ce": 7.707713666604832e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 821419012,
+      "step": 4785
+    },
+    {
+      "epoch": 54.35127478753541,
+      "grad_norm": 5.102105084891512,
+      "learning_rate": 5e-06,
+      "loss": 0.0793,
+      "num_input_tokens_seen": 821588980,
+      "step": 4786
+    },
+    {
+      "epoch": 54.35127478753541,
+      "loss": 0.12474104762077332,
+      "loss_ce": 3.0967781640356407e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 821588980,
+      "step": 4786
+    },
+    {
+      "epoch": 54.36260623229462,
+      "grad_norm": 2.9079707183029484,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 821761080,
+      "step": 4787
+    },
+    {
+      "epoch": 54.36260623229462,
+      "loss": 0.03385075181722641,
+      "loss_ce": 0.00013645895523950458,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 821761080,
+      "step": 4787
+    },
+    {
+      "epoch": 54.373937677053824,
+      "grad_norm": 4.197997791580651,
+      "learning_rate": 5e-06,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 821931384,
+      "step": 4788
+    },
+    {
+      "epoch": 54.373937677053824,
+      "loss": 0.07166539132595062,
+      "loss_ce": 0.0001474460877943784,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 821931384,
+      "step": 4788
+    },
+    {
+      "epoch": 54.38526912181303,
+      "grad_norm": 4.318176769670441,
+      "learning_rate": 5e-06,
+      "loss": 0.056,
+      "num_input_tokens_seen": 822102316,
+      "step": 4789
+    },
+    {
+      "epoch": 54.38526912181303,
+      "loss": 0.05376601964235306,
+      "loss_ce": 7.034085865598172e-05,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 822102316,
+      "step": 4789
+    },
+    {
+      "epoch": 54.39660056657224,
+      "grad_norm": 4.164191010859506,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 822273496,
+      "step": 4790
+    },
+    {
+      "epoch": 54.39660056657224,
+      "loss": 0.18763966858386993,
+      "loss_ce": 0.00012442198931239545,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 822273496,
+      "step": 4790
+    },
+    {
+      "epoch": 54.40793201133145,
+      "grad_norm": 4.5370911849606665,
+      "learning_rate": 5e-06,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 822445404,
+      "step": 4791
+    },
+    {
+      "epoch": 54.40793201133145,
+      "loss": 0.0644446611404419,
+      "loss_ce": 0.0001746386697050184,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 822445404,
+      "step": 4791
+    },
+    {
+      "epoch": 54.419263456090654,
+      "grad_norm": 4.603239099516089,
+      "learning_rate": 5e-06,
+      "loss": 0.1049,
+      "num_input_tokens_seen": 822615796,
+      "step": 4792
+    },
+    {
+      "epoch": 54.419263456090654,
+      "loss": 0.09498606622219086,
+      "loss_ce": 0.00010690891940612346,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 822615796,
+      "step": 4792
+    },
+    {
+      "epoch": 54.43059490084986,
+      "grad_norm": 4.175609154193794,
+      "learning_rate": 5e-06,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 822787900,
+      "step": 4793
+    },
+    {
+      "epoch": 54.43059490084986,
+      "loss": 0.038354288786649704,
+      "loss_ce": 0.00016153889009729028,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 822787900,
+      "step": 4793
+    },
+    {
+      "epoch": 54.44192634560906,
+      "grad_norm": 3.385730142833818,
+      "learning_rate": 5e-06,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 822959392,
+      "step": 4794
+    },
+    {
+      "epoch": 54.44192634560906,
+      "loss": 0.06145177036523819,
+      "loss_ce": 3.514348281896673e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 822959392,
+      "step": 4794
+    },
+    {
+      "epoch": 54.45325779036827,
+      "grad_norm": 4.109561322232012,
+      "learning_rate": 5e-06,
+      "loss": 0.0595,
+      "num_input_tokens_seen": 823131144,
+      "step": 4795
+    },
+    {
+      "epoch": 54.45325779036827,
+      "loss": 0.06852729618549347,
+      "loss_ce": 7.637016824446619e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 823131144,
+      "step": 4795
+    },
+    {
+      "epoch": 54.46458923512748,
+      "grad_norm": 3.3794133644466093,
+      "learning_rate": 5e-06,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 823301424,
+      "step": 4796
+    },
+    {
+      "epoch": 54.46458923512748,
+      "loss": 0.06543462723493576,
+      "loss_ce": 5.071881241747178e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 823301424,
+      "step": 4796
+    },
+    {
+      "epoch": 54.475920679886684,
+      "grad_norm": 3.5252525999746678,
+      "learning_rate": 5e-06,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 823473456,
+      "step": 4797
+    },
+    {
+      "epoch": 54.475920679886684,
+      "loss": 0.1188625916838646,
+      "loss_ce": 0.0002712833811528981,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 823473456,
+      "step": 4797
+    },
+    {
+      "epoch": 54.48725212464589,
+      "grad_norm": 2.5652412742463855,
+      "learning_rate": 5e-06,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 823645064,
+      "step": 4798
+    },
+    {
+      "epoch": 54.48725212464589,
+      "loss": 0.09908372908830643,
+      "loss_ce": 0.00010759633732959628,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 823645064,
+      "step": 4798
+    },
+    {
+      "epoch": 54.4985835694051,
+      "grad_norm": 2.946514585062964,
+      "learning_rate": 5e-06,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 823814164,
+      "step": 4799
+    },
+    {
+      "epoch": 54.4985835694051,
+      "loss": 0.07510152459144592,
+      "loss_ce": 0.0001656093809287995,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 823814164,
+      "step": 4799
+    },
+    {
+      "epoch": 54.50991501416431,
+      "grad_norm": 3.559158584586017,
+      "learning_rate": 5e-06,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 823984380,
+      "step": 4800
+    },
+    {
+      "epoch": 54.50991501416431,
+      "loss": 0.04265569522976875,
+      "loss_ce": 0.0001294516696361825,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 823984380,
+      "step": 4800
+    },
+    {
+      "epoch": 54.521246458923514,
+      "grad_norm": 2.2563742430502542,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 824154940,
+      "step": 4801
+    },
+    {
+      "epoch": 54.521246458923514,
+      "loss": 0.11463603377342224,
+      "loss_ce": 0.0001417134189978242,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 824154940,
+      "step": 4801
+    },
+    {
+      "epoch": 54.53257790368272,
+      "grad_norm": 3.9751600473707454,
+      "learning_rate": 5e-06,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 824326540,
+      "step": 4802
+    },
+    {
+      "epoch": 54.53257790368272,
+      "loss": 0.04263674467802048,
+      "loss_ce": 0.00011049696331610903,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 824326540,
+      "step": 4802
+    },
+    {
+      "epoch": 54.54390934844193,
+      "grad_norm": 3.293858694691053,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 824497492,
+      "step": 4803
+    },
+    {
+      "epoch": 54.54390934844193,
+      "loss": 0.10928931832313538,
+      "loss_ce": 2.112583206326235e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 824497492,
+      "step": 4803
+    },
+    {
+      "epoch": 54.555240793201136,
+      "grad_norm": 2.866017122208915,
+      "learning_rate": 5e-06,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 824668488,
+      "step": 4804
+    },
+    {
+      "epoch": 54.555240793201136,
+      "loss": 0.05335213243961334,
+      "loss_ce": 0.0001294738322030753,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 824668488,
+      "step": 4804
+    },
+    {
+      "epoch": 54.56657223796034,
+      "grad_norm": 3.6457935048700274,
+      "learning_rate": 5e-06,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 824840116,
+      "step": 4805
+    },
+    {
+      "epoch": 54.56657223796034,
+      "loss": 0.08859752118587494,
+      "loss_ce": 8.128270565066487e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 824840116,
+      "step": 4805
+    },
+    {
+      "epoch": 54.577903682719544,
+      "grad_norm": 3.7502007928505567,
+      "learning_rate": 5e-06,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 825012052,
+      "step": 4806
+    },
+    {
+      "epoch": 54.577903682719544,
+      "loss": 0.0466158390045166,
+      "loss_ce": 4.601352702593431e-05,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 825012052,
+      "step": 4806
+    },
+    {
+      "epoch": 54.58923512747875,
+      "grad_norm": 3.7225774111623346,
+      "learning_rate": 5e-06,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 825184044,
+      "step": 4807
+    },
+    {
+      "epoch": 54.58923512747875,
+      "loss": 0.04037688672542572,
+      "loss_ce": 0.00012420277926139534,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 825184044,
+      "step": 4807
+    },
+    {
+      "epoch": 54.60056657223796,
+      "grad_norm": 3.889638768464499,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 825352924,
+      "step": 4808
+    },
+    {
+      "epoch": 54.60056657223796,
+      "loss": 0.06766480952501297,
+      "loss_ce": 5.311829590937123e-05,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 825352924,
+      "step": 4808
+    },
+    {
+      "epoch": 54.611898016997166,
+      "grad_norm": 4.49605031857285,
+      "learning_rate": 5e-06,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 825525080,
+      "step": 4809
+    },
+    {
+      "epoch": 54.611898016997166,
+      "loss": 0.052674490958452225,
+      "loss_ce": 0.00015373813221231103,
+      "loss_iou": 0.390625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 825525080,
+      "step": 4809
+    },
+    {
+      "epoch": 54.623229461756374,
+      "grad_norm": 3.9350984044921895,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 825697280,
+      "step": 4810
+    },
+    {
+      "epoch": 54.623229461756374,
+      "loss": 0.09486401826143265,
+      "loss_ce": 0.00016797560965642333,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 825697280,
+      "step": 4810
+    },
+    {
+      "epoch": 54.63456090651558,
+      "grad_norm": 3.927121727963861,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 825866944,
+      "step": 4811
+    },
+    {
+      "epoch": 54.63456090651558,
+      "loss": 0.05210106074810028,
+      "loss_ce": 0.00010673302313080058,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 825866944,
+      "step": 4811
+    },
+    {
+      "epoch": 54.64589235127479,
+      "grad_norm": 3.6655260560736753,
+      "learning_rate": 5e-06,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 826037992,
+      "step": 4812
+    },
+    {
+      "epoch": 54.64589235127479,
+      "loss": 0.14142560958862305,
+      "loss_ce": 6.81814708514139e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 826037992,
+      "step": 4812
+    },
+    {
+      "epoch": 54.657223796033996,
+      "grad_norm": 3.7347683776950245,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 826209604,
+      "step": 4813
+    },
+    {
+      "epoch": 54.657223796033996,
+      "loss": 0.0608188733458519,
+      "loss_ce": 0.00016518283518962562,
+      "loss_iou": 0.5,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 826209604,
+      "step": 4813
+    },
+    {
+      "epoch": 54.668555240793204,
+      "grad_norm": 3.895897289203231,
+      "learning_rate": 5e-06,
+      "loss": 0.075,
+      "num_input_tokens_seen": 826381536,
+      "step": 4814
+    },
+    {
+      "epoch": 54.668555240793204,
+      "loss": 0.0637035071849823,
+      "loss_ce": 5.9096742916153744e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 826381536,
+      "step": 4814
+    },
+    {
+      "epoch": 54.67988668555241,
+      "grad_norm": 3.3592646523617717,
+      "learning_rate": 5e-06,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 826553720,
+      "step": 4815
+    },
+    {
+      "epoch": 54.67988668555241,
+      "loss": 0.034071944653987885,
+      "loss_ce": 0.0001211362105095759,
+      "loss_iou": 0.203125,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 826553720,
+      "step": 4815
+    },
+    {
+      "epoch": 54.69121813031161,
+      "grad_norm": 3.540193329895928,
+      "learning_rate": 5e-06,
+      "loss": 0.0803,
+      "num_input_tokens_seen": 826725536,
+      "step": 4816
+    },
+    {
+      "epoch": 54.69121813031161,
+      "loss": 0.04260273277759552,
+      "loss_ce": 3.071070386795327e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 826725536,
+      "step": 4816
+    },
+    {
+      "epoch": 54.70254957507082,
+      "grad_norm": 3.380776251926493,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 826897596,
+      "step": 4817
+    },
+    {
+      "epoch": 54.70254957507082,
+      "loss": 0.050521381199359894,
+      "loss_ce": 0.0001521162485005334,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 826897596,
+      "step": 4817
+    },
+    {
+      "epoch": 54.713881019830026,
+      "grad_norm": 2.8592284379065123,
+      "learning_rate": 5e-06,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 827069444,
+      "step": 4818
+    },
+    {
+      "epoch": 54.713881019830026,
+      "loss": 0.045377492904663086,
+      "loss_ce": 0.000303027278278023,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 827069444,
+      "step": 4818
+    },
+    {
+      "epoch": 54.725212464589234,
+      "grad_norm": 3.703633199262127,
+      "learning_rate": 5e-06,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 827241208,
+      "step": 4819
+    },
+    {
+      "epoch": 54.725212464589234,
+      "loss": 0.0570884607732296,
+      "loss_ce": 0.0001884365192381665,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 827241208,
+      "step": 4819
+    },
+    {
+      "epoch": 54.73654390934844,
+      "grad_norm": 12.420864520741524,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 827412980,
+      "step": 4820
+    },
+    {
+      "epoch": 54.73654390934844,
+      "loss": 0.11870403587818146,
+      "loss_ce": 5.168938514543697e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 827412980,
+      "step": 4820
+    },
+    {
+      "epoch": 54.74787535410765,
+      "grad_norm": 3.8378961402377056,
+      "learning_rate": 5e-06,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 827584712,
+      "step": 4821
+    },
+    {
+      "epoch": 54.74787535410765,
+      "loss": 0.04571519419550896,
+      "loss_ce": 4.5638960727956146e-05,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 827584712,
+      "step": 4821
+    },
+    {
+      "epoch": 54.759206798866856,
+      "grad_norm": 3.449814975254727,
+      "learning_rate": 5e-06,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 827756648,
+      "step": 4822
+    },
+    {
+      "epoch": 54.759206798866856,
+      "loss": 0.07373366504907608,
+      "loss_ce": 3.371223283465952e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 827756648,
+      "step": 4822
+    },
+    {
+      "epoch": 54.77053824362606,
+      "grad_norm": 3.4363965785193296,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 827928464,
+      "step": 4823
+    },
+    {
+      "epoch": 54.77053824362606,
+      "loss": 0.05223904922604561,
+      "loss_ce": 0.00032864813692867756,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 827928464,
+      "step": 4823
+    },
+    {
+      "epoch": 54.78186968838527,
+      "grad_norm": 4.089368975478395,
+      "learning_rate": 5e-06,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 828100456,
+      "step": 4824
+    },
+    {
+      "epoch": 54.78186968838527,
+      "loss": 0.1433524787425995,
+      "loss_ce": 0.00013347614731173962,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 828100456,
+      "step": 4824
+    },
+    {
+      "epoch": 54.79320113314448,
+      "grad_norm": 4.224494235415107,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 828271344,
+      "step": 4825
+    },
+    {
+      "epoch": 54.79320113314448,
+      "loss": 0.08384694159030914,
+      "loss_ce": 0.00013722653966397047,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 828271344,
+      "step": 4825
+    },
+    {
+      "epoch": 54.804532577903686,
+      "grad_norm": 4.132696797184806,
+      "learning_rate": 5e-06,
+      "loss": 0.068,
+      "num_input_tokens_seen": 828443304,
+      "step": 4826
+    },
+    {
+      "epoch": 54.804532577903686,
+      "loss": 0.060608796775341034,
+      "loss_ce": 0.00016873680579010397,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 828443304,
+      "step": 4826
+    },
+    {
+      "epoch": 54.815864022662886,
+      "grad_norm": 5.715146873866257,
+      "learning_rate": 5e-06,
+      "loss": 0.0745,
+      "num_input_tokens_seen": 828614272,
+      "step": 4827
+    },
+    {
+      "epoch": 54.815864022662886,
+      "loss": 0.15620847046375275,
+      "loss_ce": 0.00030941443401388824,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 828614272,
+      "step": 4827
+    },
+    {
+      "epoch": 54.827195467422094,
+      "grad_norm": 3.5342595585611107,
+      "learning_rate": 5e-06,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 828785468,
+      "step": 4828
+    },
+    {
+      "epoch": 54.827195467422094,
+      "loss": 0.07759758830070496,
+      "loss_ce": 0.0001744925684761256,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 828785468,
+      "step": 4828
+    },
+    {
+      "epoch": 54.8385269121813,
+      "grad_norm": 3.113392249446793,
+      "learning_rate": 5e-06,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 828957024,
+      "step": 4829
+    },
+    {
+      "epoch": 54.8385269121813,
+      "loss": 0.059903040528297424,
+      "loss_ce": 0.00011910706234630197,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 828957024,
+      "step": 4829
+    },
+    {
+      "epoch": 54.84985835694051,
+      "grad_norm": 3.376017094281157,
+      "learning_rate": 5e-06,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 829129308,
+      "step": 4830
+    },
+    {
+      "epoch": 54.84985835694051,
+      "loss": 0.08225090056657791,
+      "loss_ce": 0.00012809839972760528,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 829129308,
+      "step": 4830
+    },
+    {
+      "epoch": 54.861189801699716,
+      "grad_norm": 3.367628310907048,
+      "learning_rate": 5e-06,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 829299544,
+      "step": 4831
+    },
+    {
+      "epoch": 54.861189801699716,
+      "loss": 0.08068462461233139,
+      "loss_ce": 0.00016399394371546805,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 829299544,
+      "step": 4831
+    },
+    {
+      "epoch": 54.87252124645892,
+      "grad_norm": 4.490877263604367,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 829471672,
+      "step": 4832
+    },
+    {
+      "epoch": 54.87252124645892,
+      "loss": 0.0562872551381588,
+      "loss_ce": 0.0001043960073729977,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 829471672,
+      "step": 4832
+    },
+    {
+      "epoch": 54.88385269121813,
+      "grad_norm": 3.5382005892734947,
+      "learning_rate": 5e-06,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 829643252,
+      "step": 4833
+    },
+    {
+      "epoch": 54.88385269121813,
+      "loss": 0.085013747215271,
+      "loss_ce": 0.0001291025837417692,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 829643252,
+      "step": 4833
+    },
+    {
+      "epoch": 54.89518413597734,
+      "grad_norm": 3.604868679760141,
+      "learning_rate": 5e-06,
+      "loss": 0.059,
+      "num_input_tokens_seen": 829815276,
+      "step": 4834
+    },
+    {
+      "epoch": 54.89518413597734,
+      "loss": 0.0672743022441864,
+      "loss_ce": 0.00012036891712341458,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 829815276,
+      "step": 4834
+    },
+    {
+      "epoch": 54.906515580736546,
+      "grad_norm": 4.158991697365566,
+      "learning_rate": 5e-06,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 829986936,
+      "step": 4835
+    },
+    {
+      "epoch": 54.906515580736546,
+      "loss": 0.03940323740243912,
+      "loss_ce": 9.659463103162125e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 829986936,
+      "step": 4835
+    },
+    {
+      "epoch": 54.91784702549575,
+      "grad_norm": 4.790648861877688,
+      "learning_rate": 5e-06,
+      "loss": 0.069,
+      "num_input_tokens_seen": 830158848,
+      "step": 4836
+    },
+    {
+      "epoch": 54.91784702549575,
+      "loss": 0.03385428339242935,
+      "loss_ce": 0.00014761844067834318,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 830158848,
+      "step": 4836
+    },
+    {
+      "epoch": 54.92917847025496,
+      "grad_norm": 3.774530388782342,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 830330192,
+      "step": 4837
+    },
+    {
+      "epoch": 54.92917847025496,
+      "loss": 0.07731056958436966,
+      "loss_ce": 0.00013160944217815995,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 830330192,
+      "step": 4837
+    },
+    {
+      "epoch": 54.94050991501416,
+      "grad_norm": 3.100728104340838,
+      "learning_rate": 5e-06,
+      "loss": 0.058,
+      "num_input_tokens_seen": 830502468,
+      "step": 4838
+    },
+    {
+      "epoch": 54.94050991501416,
+      "loss": 0.06617963314056396,
+      "loss_ce": 0.00020063022384420037,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 830502468,
+      "step": 4838
+    },
+    {
+      "epoch": 54.95184135977337,
+      "grad_norm": 4.123130030874466,
+      "learning_rate": 5e-06,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 830674192,
+      "step": 4839
+    },
+    {
+      "epoch": 54.95184135977337,
+      "loss": 0.06947401165962219,
+      "loss_ce": 0.00016859086463227868,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 830674192,
+      "step": 4839
+    },
+    {
+      "epoch": 54.963172804532576,
+      "grad_norm": 4.231864916395142,
+      "learning_rate": 5e-06,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 830844016,
+      "step": 4840
+    },
+    {
+      "epoch": 54.963172804532576,
+      "loss": 0.13347838819026947,
+      "loss_ce": 0.00014708723756484687,
+      "loss_iou": 0.546875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 830844016,
+      "step": 4840
+    },
+    {
+      "epoch": 54.97450424929178,
+      "grad_norm": 3.291507432454373,
+      "learning_rate": 5e-06,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 831014900,
+      "step": 4841
+    },
+    {
+      "epoch": 54.97450424929178,
+      "loss": 0.0711250975728035,
+      "loss_ce": 0.00011069620086345822,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 831014900,
+      "step": 4841
+    },
+    {
+      "epoch": 54.98583569405099,
+      "grad_norm": 3.754441083026507,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 831186668,
+      "step": 4842
+    },
+    {
+      "epoch": 54.98583569405099,
+      "loss": 0.044196151196956635,
+      "loss_ce": 5.247210356174037e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 831186668,
+      "step": 4842
+    },
+    {
+      "epoch": 54.9971671388102,
+      "grad_norm": 4.812387360169728,
+      "learning_rate": 5e-06,
+      "loss": 0.088,
+      "num_input_tokens_seen": 831358496,
+      "step": 4843
+    },
+    {
+      "epoch": 54.9971671388102,
+      "loss": 0.12149156630039215,
+      "loss_ce": 0.0001841913181124255,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 831358496,
+      "step": 4843
+    },
+    {
+      "epoch": 54.9971671388102,
+      "loss": 0.05885420739650726,
+      "loss_ce": 0.000367268017726019,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 831401396,
+      "step": 4843
+    },
+    {
+      "epoch": 55.008498583569406,
+      "grad_norm": 3.08002992962033,
+      "learning_rate": 5e-06,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 831530268,
+      "step": 4844
+    },
+    {
+      "epoch": 55.008498583569406,
+      "loss": 0.049395687878131866,
+      "loss_ce": 7.928359264042228e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 831530268,
+      "step": 4844
+    },
+    {
+      "epoch": 55.01983002832861,
+      "grad_norm": 3.3385109741113417,
+      "learning_rate": 5e-06,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 831701532,
+      "step": 4845
+    },
+    {
+      "epoch": 55.01983002832861,
+      "loss": 0.08859927952289581,
+      "loss_ce": 0.00015934000839479268,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 831701532,
+      "step": 4845
+    },
+    {
+      "epoch": 55.03116147308782,
+      "grad_norm": 3.591264079351339,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 831873124,
+      "step": 4846
+    },
+    {
+      "epoch": 55.03116147308782,
+      "loss": 0.14700305461883545,
+      "loss_ce": 6.090020542615093e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 831873124,
+      "step": 4846
+    },
+    {
+      "epoch": 55.04249291784703,
+      "grad_norm": 3.751691612207764,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 832045180,
+      "step": 4847
+    },
+    {
+      "epoch": 55.04249291784703,
+      "loss": 0.040723513811826706,
+      "loss_ce": 7.409861427731812e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 832045180,
+      "step": 4847
+    },
+    {
+      "epoch": 55.053824362606235,
+      "grad_norm": 3.5488526355455243,
+      "learning_rate": 5e-06,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 832217084,
+      "step": 4848
+    },
+    {
+      "epoch": 55.053824362606235,
+      "loss": 0.06552620232105255,
+      "loss_ce": 2.0215240510879084e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 832217084,
+      "step": 4848
+    },
+    {
+      "epoch": 55.065155807365436,
+      "grad_norm": 3.451855976311762,
+      "learning_rate": 5e-06,
+      "loss": 0.052,
+      "num_input_tokens_seen": 832387800,
+      "step": 4849
+    },
+    {
+      "epoch": 55.065155807365436,
+      "loss": 0.050868839025497437,
+      "loss_ce": 0.00010284706513630226,
+      "loss_iou": 0.375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 832387800,
+      "step": 4849
+    },
+    {
+      "epoch": 55.07648725212464,
+      "grad_norm": 3.4960358945510177,
+      "learning_rate": 5e-06,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 832559164,
+      "step": 4850
+    },
+    {
+      "epoch": 55.07648725212464,
+      "loss": 0.04941345751285553,
+      "loss_ce": 5.127799522597343e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 832559164,
+      "step": 4850
+    },
+    {
+      "epoch": 55.08781869688385,
+      "grad_norm": 3.019830567129509,
+      "learning_rate": 5e-06,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 832729004,
+      "step": 4851
+    },
+    {
+      "epoch": 55.08781869688385,
+      "loss": 0.10447005927562714,
+      "loss_ce": 0.0003135710139758885,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 832729004,
+      "step": 4851
+    },
+    {
+      "epoch": 55.09915014164306,
+      "grad_norm": 5.171988480875635,
+      "learning_rate": 5e-06,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 832901252,
+      "step": 4852
+    },
+    {
+      "epoch": 55.09915014164306,
+      "loss": 0.04791152477264404,
+      "loss_ce": 0.00016677156963851303,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 832901252,
+      "step": 4852
+    },
+    {
+      "epoch": 55.110481586402265,
+      "grad_norm": 3.7103334345427466,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 833072592,
+      "step": 4853
+    },
+    {
+      "epoch": 55.110481586402265,
+      "loss": 0.05243786796927452,
+      "loss_ce": 5.44440554222092e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 833072592,
+      "step": 4853
+    },
+    {
+      "epoch": 55.12181303116147,
+      "grad_norm": 5.34505010837871,
+      "learning_rate": 5e-06,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 833244000,
+      "step": 4854
+    },
+    {
+      "epoch": 55.12181303116147,
+      "loss": 0.1246090978384018,
+      "loss_ce": 6.685879634460434e-05,
+      "loss_iou": 0.07421875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 833244000,
+      "step": 4854
+    },
+    {
+      "epoch": 55.13314447592068,
+      "grad_norm": 3.600249735058,
+      "learning_rate": 5e-06,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 833414296,
+      "step": 4855
+    },
+    {
+      "epoch": 55.13314447592068,
+      "loss": 0.08738004416227341,
+      "loss_ce": 0.00013028204557485878,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 833414296,
+      "step": 4855
+    },
+    {
+      "epoch": 55.14447592067989,
+      "grad_norm": 3.9518795251806416,
+      "learning_rate": 5e-06,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 833585820,
+      "step": 4856
+    },
+    {
+      "epoch": 55.14447592067989,
+      "loss": 0.03788881376385689,
+      "loss_ce": 4.70182130811736e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 833585820,
+      "step": 4856
+    },
+    {
+      "epoch": 55.155807365439095,
+      "grad_norm": 3.250529452619723,
+      "learning_rate": 5e-06,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 833757340,
+      "step": 4857
+    },
+    {
+      "epoch": 55.155807365439095,
+      "loss": 0.04628312587738037,
+      "loss_ce": 4.899408304481767e-05,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 833757340,
+      "step": 4857
+    },
+    {
+      "epoch": 55.1671388101983,
+      "grad_norm": 3.6477704130554094,
+      "learning_rate": 5e-06,
+      "loss": 0.099,
+      "num_input_tokens_seen": 833928936,
+      "step": 4858
+    },
+    {
+      "epoch": 55.1671388101983,
+      "loss": 0.11140298843383789,
+      "loss_ce": 7.486053073080257e-05,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 833928936,
+      "step": 4858
+    },
+    {
+      "epoch": 55.17847025495751,
+      "grad_norm": 4.049437142282047,
+      "learning_rate": 5e-06,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 834100300,
+      "step": 4859
+    },
+    {
+      "epoch": 55.17847025495751,
+      "loss": 0.0394081249833107,
+      "loss_ce": 0.00010148172441404313,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 834100300,
+      "step": 4859
+    },
+    {
+      "epoch": 55.18980169971671,
+      "grad_norm": 3.7360462053462715,
+      "learning_rate": 5e-06,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 834270868,
+      "step": 4860
+    },
+    {
+      "epoch": 55.18980169971671,
+      "loss": 0.05600510165095329,
+      "loss_ce": 0.00017319328617304564,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 834270868,
+      "step": 4860
+    },
+    {
+      "epoch": 55.20113314447592,
+      "grad_norm": 3.5912981043900216,
+      "learning_rate": 5e-06,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 834441052,
+      "step": 4861
+    },
+    {
+      "epoch": 55.20113314447592,
+      "loss": 0.04095540568232536,
+      "loss_ce": 0.0001686612085904926,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 834441052,
+      "step": 4861
+    },
+    {
+      "epoch": 55.212464589235125,
+      "grad_norm": 5.312096689896347,
+      "learning_rate": 5e-06,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 834612156,
+      "step": 4862
+    },
+    {
+      "epoch": 55.212464589235125,
+      "loss": 0.08147785067558289,
+      "loss_ce": 0.0001484984823036939,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 834612156,
+      "step": 4862
+    },
+    {
+      "epoch": 55.22379603399433,
+      "grad_norm": 5.652013947056333,
+      "learning_rate": 5e-06,
+      "loss": 0.0902,
+      "num_input_tokens_seen": 834783564,
+      "step": 4863
+    },
+    {
+      "epoch": 55.22379603399433,
+      "loss": 0.08377085626125336,
+      "loss_ce": 0.00018320902017876506,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 834783564,
+      "step": 4863
+    },
+    {
+      "epoch": 55.23512747875354,
+      "grad_norm": 2.9174302224508923,
+      "learning_rate": 5e-06,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 834955048,
+      "step": 4864
+    },
+    {
+      "epoch": 55.23512747875354,
+      "loss": 0.03336211293935776,
+      "loss_ce": 5.217511352384463e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 834955048,
+      "step": 4864
+    },
+    {
+      "epoch": 55.24645892351275,
+      "grad_norm": 3.183846767327532,
+      "learning_rate": 5e-06,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 835125456,
+      "step": 4865
+    },
+    {
+      "epoch": 55.24645892351275,
+      "loss": 0.12082712352275848,
+      "loss_ce": 6.90634551574476e-05,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 835125456,
+      "step": 4865
+    },
+    {
+      "epoch": 55.257790368271955,
+      "grad_norm": 3.6693916962039057,
+      "learning_rate": 5e-06,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 835297028,
+      "step": 4866
+    },
+    {
+      "epoch": 55.257790368271955,
+      "loss": 0.09813405573368073,
+      "loss_ce": 0.00011159271525684744,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 835297028,
+      "step": 4866
+    },
+    {
+      "epoch": 55.26912181303116,
+      "grad_norm": 3.6166673366715525,
+      "learning_rate": 5e-06,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 835468056,
+      "step": 4867
+    },
+    {
+      "epoch": 55.26912181303116,
+      "loss": 0.06933483481407166,
+      "loss_ce": 9.045004844665527e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 835468056,
+      "step": 4867
+    },
+    {
+      "epoch": 55.28045325779037,
+      "grad_norm": 3.859604724392222,
+      "learning_rate": 5e-06,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 835640024,
+      "step": 4868
+    },
+    {
+      "epoch": 55.28045325779037,
+      "loss": 0.06038837134838104,
+      "loss_ce": 3.986154479207471e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 835640024,
+      "step": 4868
+    },
+    {
+      "epoch": 55.29178470254958,
+      "grad_norm": 4.137460084964757,
+      "learning_rate": 5e-06,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 835812208,
+      "step": 4869
+    },
+    {
+      "epoch": 55.29178470254958,
+      "loss": 0.04981578141450882,
+      "loss_ce": 0.00013316590047907084,
+      "loss_iou": 0.5,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 835812208,
+      "step": 4869
+    },
+    {
+      "epoch": 55.303116147308785,
+      "grad_norm": 4.994204068695308,
+      "learning_rate": 5e-06,
+      "loss": 0.0607,
+      "num_input_tokens_seen": 835984376,
+      "step": 4870
+    },
+    {
+      "epoch": 55.303116147308785,
+      "loss": 0.07546278834342957,
+      "loss_ce": 0.0001759215083438903,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 835984376,
+      "step": 4870
+    },
+    {
+      "epoch": 55.314447592067985,
+      "grad_norm": 4.783889326640654,
+      "learning_rate": 5e-06,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 836153268,
+      "step": 4871
+    },
+    {
+      "epoch": 55.314447592067985,
+      "loss": 0.09268462657928467,
+      "loss_ce": 0.0008267149678431451,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 836153268,
+      "step": 4871
+    },
+    {
+      "epoch": 55.32577903682719,
+      "grad_norm": 3.2990718542812116,
+      "learning_rate": 5e-06,
+      "loss": 0.056,
+      "num_input_tokens_seen": 836325396,
+      "step": 4872
+    },
+    {
+      "epoch": 55.32577903682719,
+      "loss": 0.038956448435783386,
+      "loss_ce": 6.179563933983445e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 836325396,
+      "step": 4872
+    },
+    {
+      "epoch": 55.3371104815864,
+      "grad_norm": 3.2999850297171736,
+      "learning_rate": 5e-06,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 836497336,
+      "step": 4873
+    },
+    {
+      "epoch": 55.3371104815864,
+      "loss": 0.08170707523822784,
+      "loss_ce": 0.00019462211639620364,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 836497336,
+      "step": 4873
+    },
+    {
+      "epoch": 55.34844192634561,
+      "grad_norm": 4.309098550220041,
+      "learning_rate": 5e-06,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 836667720,
+      "step": 4874
+    },
+    {
+      "epoch": 55.34844192634561,
+      "loss": 0.09926150739192963,
+      "loss_ce": 7.937539339764044e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 836667720,
+      "step": 4874
+    },
+    {
+      "epoch": 55.359773371104815,
+      "grad_norm": 4.740192279373646,
+      "learning_rate": 5e-06,
+      "loss": 0.0819,
+      "num_input_tokens_seen": 836837976,
+      "step": 4875
+    },
+    {
+      "epoch": 55.359773371104815,
+      "loss": 0.055474989116191864,
+      "loss_ce": 7.032898429315537e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 836837976,
+      "step": 4875
+    },
+    {
+      "epoch": 55.37110481586402,
+      "grad_norm": 3.8193839405818504,
+      "learning_rate": 5e-06,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 837009992,
+      "step": 4876
+    },
+    {
+      "epoch": 55.37110481586402,
+      "loss": 0.0744498074054718,
+      "loss_ce": 0.00015476541011594236,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 837009992,
+      "step": 4876
+    },
+    {
+      "epoch": 55.38243626062323,
+      "grad_norm": 5.282860737524898,
+      "learning_rate": 5e-06,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 837181552,
+      "step": 4877
+    },
+    {
+      "epoch": 55.38243626062323,
+      "loss": 0.031676217913627625,
+      "loss_ce": 9.052344830706716e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 837181552,
+      "step": 4877
+    },
+    {
+      "epoch": 55.39376770538244,
+      "grad_norm": 68.58655013328075,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 837353896,
+      "step": 4878
+    },
+    {
+      "epoch": 55.39376770538244,
+      "loss": 0.05661048740148544,
+      "loss_ce": 6.141711492091417e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 837353896,
+      "step": 4878
+    },
+    {
+      "epoch": 55.405099150141645,
+      "grad_norm": 4.794209646326477,
+      "learning_rate": 5e-06,
+      "loss": 0.077,
+      "num_input_tokens_seen": 837526036,
+      "step": 4879
+    },
+    {
+      "epoch": 55.405099150141645,
+      "loss": 0.04730220139026642,
+      "loss_ce": 0.00010676590318325907,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 837526036,
+      "step": 4879
+    },
+    {
+      "epoch": 55.41643059490085,
+      "grad_norm": 3.983621763867161,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 837697824,
+      "step": 4880
+    },
+    {
+      "epoch": 55.41643059490085,
+      "loss": 0.03588370233774185,
+      "loss_ce": 0.00013235759979579598,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 837697824,
+      "step": 4880
+    },
+    {
+      "epoch": 55.42776203966006,
+      "grad_norm": 2.556705177928445,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 837868728,
+      "step": 4881
+    },
+    {
+      "epoch": 55.42776203966006,
+      "loss": 0.06567414104938507,
+      "loss_ce": 0.00016816367860883474,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 837868728,
+      "step": 4881
+    },
+    {
+      "epoch": 55.43909348441926,
+      "grad_norm": 2.214142232023204,
+      "learning_rate": 5e-06,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 838040748,
+      "step": 4882
+    },
+    {
+      "epoch": 55.43909348441926,
+      "loss": 0.05788538604974747,
+      "loss_ce": 5.457275256048888e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 838040748,
+      "step": 4882
+    },
+    {
+      "epoch": 55.45042492917847,
+      "grad_norm": 2.668098887875958,
+      "learning_rate": 5e-06,
+      "loss": 0.065,
+      "num_input_tokens_seen": 838210412,
+      "step": 4883
+    },
+    {
+      "epoch": 55.45042492917847,
+      "loss": 0.09227945655584335,
+      "loss_ce": 8.585143950767815e-05,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 838210412,
+      "step": 4883
+    },
+    {
+      "epoch": 55.461756373937675,
+      "grad_norm": 2.324162073048407,
+      "learning_rate": 5e-06,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 838381936,
+      "step": 4884
+    },
+    {
+      "epoch": 55.461756373937675,
+      "loss": 0.07587769627571106,
+      "loss_ce": 0.0003314313944429159,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 838381936,
+      "step": 4884
+    },
+    {
+      "epoch": 55.47308781869688,
+      "grad_norm": 3.295053985404213,
+      "learning_rate": 5e-06,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 838553800,
+      "step": 4885
+    },
+    {
+      "epoch": 55.47308781869688,
+      "loss": 0.11731519550085068,
+      "loss_ce": 0.00011243583867326379,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 838553800,
+      "step": 4885
+    },
+    {
+      "epoch": 55.48441926345609,
+      "grad_norm": 2.091044967669853,
+      "learning_rate": 5e-06,
+      "loss": 0.075,
+      "num_input_tokens_seen": 838725868,
+      "step": 4886
+    },
+    {
+      "epoch": 55.48441926345609,
+      "loss": 0.052465178072452545,
+      "loss_ce": 0.00018093621474690735,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 838725868,
+      "step": 4886
+    },
+    {
+      "epoch": 55.4957507082153,
+      "grad_norm": 5.63296129853245,
+      "learning_rate": 5e-06,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 838897860,
+      "step": 4887
+    },
+    {
+      "epoch": 55.4957507082153,
+      "loss": 0.09221188724040985,
+      "loss_ce": 0.00010983859829138964,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 838897860,
+      "step": 4887
+    },
+    {
+      "epoch": 55.507082152974505,
+      "grad_norm": 3.6102645727168037,
+      "learning_rate": 5e-06,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 839069112,
+      "step": 4888
+    },
+    {
+      "epoch": 55.507082152974505,
+      "loss": 0.044318750500679016,
+      "loss_ce": 5.300229895510711e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 839069112,
+      "step": 4888
+    },
+    {
+      "epoch": 55.51841359773371,
+      "grad_norm": 4.135972129451464,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 839239864,
+      "step": 4889
+    },
+    {
+      "epoch": 55.51841359773371,
+      "loss": 0.10798877477645874,
+      "loss_ce": 0.0002617257705423981,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 839239864,
+      "step": 4889
+    },
+    {
+      "epoch": 55.52974504249292,
+      "grad_norm": 4.650632633301889,
+      "learning_rate": 5e-06,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 839411936,
+      "step": 4890
+    },
+    {
+      "epoch": 55.52974504249292,
+      "loss": 0.10204128921031952,
+      "loss_ce": 0.00023463831166736782,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 839411936,
+      "step": 4890
+    },
+    {
+      "epoch": 55.54107648725213,
+      "grad_norm": 3.5102093662092635,
+      "learning_rate": 5e-06,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 839583056,
+      "step": 4891
+    },
+    {
+      "epoch": 55.54107648725213,
+      "loss": 0.038575440645217896,
+      "loss_ce": 0.00019958653138019145,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 839583056,
+      "step": 4891
+    },
+    {
+      "epoch": 55.552407932011334,
+      "grad_norm": 5.496362906048299,
+      "learning_rate": 5e-06,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 839753408,
+      "step": 4892
+    },
+    {
+      "epoch": 55.552407932011334,
+      "loss": 0.0450739860534668,
+      "loss_ce": 9.107418009079993e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 839753408,
+      "step": 4892
+    },
+    {
+      "epoch": 55.563739376770535,
+      "grad_norm": 4.258196540781646,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 839925240,
+      "step": 4893
+    },
+    {
+      "epoch": 55.563739376770535,
+      "loss": 0.10558530688285828,
+      "loss_ce": 8.604592585470527e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 839925240,
+      "step": 4893
+    },
+    {
+      "epoch": 55.57507082152974,
+      "grad_norm": 4.403422192478105,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 840096608,
+      "step": 4894
+    },
+    {
+      "epoch": 55.57507082152974,
+      "loss": 0.033832140266895294,
+      "loss_ce": 0.00011021541286027059,
+      "loss_iou": 0.5,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 840096608,
+      "step": 4894
+    },
+    {
+      "epoch": 55.58640226628895,
+      "grad_norm": 3.4135292072406562,
+      "learning_rate": 5e-06,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 840266596,
+      "step": 4895
+    },
+    {
+      "epoch": 55.58640226628895,
+      "loss": 0.17693571746349335,
+      "loss_ce": 0.00014738901518285275,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 840266596,
+      "step": 4895
+    },
+    {
+      "epoch": 55.59773371104816,
+      "grad_norm": 2.9171624453334646,
+      "learning_rate": 5e-06,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 840438620,
+      "step": 4896
+    },
+    {
+      "epoch": 55.59773371104816,
+      "loss": 0.06255581974983215,
+      "loss_ce": 7.10822205292061e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 840438620,
+      "step": 4896
+    },
+    {
+      "epoch": 55.609065155807365,
+      "grad_norm": 2.7276004161517036,
+      "learning_rate": 5e-06,
+      "loss": 0.043,
+      "num_input_tokens_seen": 840608808,
+      "step": 4897
+    },
+    {
+      "epoch": 55.609065155807365,
+      "loss": 0.028982864692807198,
+      "loss_ce": 6.74590264679864e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 840608808,
+      "step": 4897
+    },
+    {
+      "epoch": 55.62039660056657,
+      "grad_norm": 3.6868888325986107,
+      "learning_rate": 5e-06,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 840779572,
+      "step": 4898
+    },
+    {
+      "epoch": 55.62039660056657,
+      "loss": 0.092755526304245,
+      "loss_ce": 7.364529301412404e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 840779572,
+      "step": 4898
+    },
+    {
+      "epoch": 55.63172804532578,
+      "grad_norm": 6.472887507062862,
+      "learning_rate": 5e-06,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 840950552,
+      "step": 4899
+    },
+    {
+      "epoch": 55.63172804532578,
+      "loss": 0.05668313801288605,
+      "loss_ce": 8.82900640135631e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 840950552,
+      "step": 4899
+    },
+    {
+      "epoch": 55.64305949008499,
+      "grad_norm": 3.384872877294457,
+      "learning_rate": 5e-06,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 841122804,
+      "step": 4900
+    },
+    {
+      "epoch": 55.64305949008499,
+      "loss": 0.07760881632566452,
+      "loss_ce": 0.0001246868196176365,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 841122804,
+      "step": 4900
+    },
+    {
+      "epoch": 55.654390934844194,
+      "grad_norm": 4.040554348783327,
+      "learning_rate": 5e-06,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 841292968,
+      "step": 4901
+    },
+    {
+      "epoch": 55.654390934844194,
+      "loss": 0.04645824432373047,
+      "loss_ce": 8.678375161252916e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 841292968,
+      "step": 4901
+    },
+    {
+      "epoch": 55.6657223796034,
+      "grad_norm": 13.51059684780058,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 841463308,
+      "step": 4902
+    },
+    {
+      "epoch": 55.6657223796034,
+      "loss": 0.07271280884742737,
+      "loss_ce": 8.09723132988438e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 841463308,
+      "step": 4902
+    },
+    {
+      "epoch": 55.67705382436261,
+      "grad_norm": 3.770452936213153,
+      "learning_rate": 5e-06,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 841635424,
+      "step": 4903
+    },
+    {
+      "epoch": 55.67705382436261,
+      "loss": 0.04690537601709366,
+      "loss_ce": 0.0002134811511496082,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 841635424,
+      "step": 4903
+    },
+    {
+      "epoch": 55.68838526912181,
+      "grad_norm": 4.028212939468325,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 841807568,
+      "step": 4904
+    },
+    {
+      "epoch": 55.68838526912181,
+      "loss": 0.032003916800022125,
+      "loss_ce": 0.0002198598231188953,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 841807568,
+      "step": 4904
+    },
+    {
+      "epoch": 55.69971671388102,
+      "grad_norm": 5.4954994365558205,
+      "learning_rate": 5e-06,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 841979408,
+      "step": 4905
+    },
+    {
+      "epoch": 55.69971671388102,
+      "loss": 0.07235094904899597,
+      "loss_ce": 0.00016161345411092043,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 841979408,
+      "step": 4905
+    },
+    {
+      "epoch": 55.711048158640224,
+      "grad_norm": 3.7626026912339396,
+      "learning_rate": 5e-06,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 842151028,
+      "step": 4906
+    },
+    {
+      "epoch": 55.711048158640224,
+      "loss": 0.039993807673454285,
+      "loss_ce": 4.6300912799779326e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 842151028,
+      "step": 4906
+    },
+    {
+      "epoch": 55.72237960339943,
+      "grad_norm": 3.406112451312235,
+      "learning_rate": 5e-06,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 842323228,
+      "step": 4907
+    },
+    {
+      "epoch": 55.72237960339943,
+      "loss": 0.09761609137058258,
+      "loss_ce": 0.00011242822074564174,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 842323228,
+      "step": 4907
+    },
+    {
+      "epoch": 55.73371104815864,
+      "grad_norm": 3.9424127173175454,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 842495116,
+      "step": 4908
+    },
+    {
+      "epoch": 55.73371104815864,
+      "loss": 0.043002791702747345,
+      "loss_ce": 9.507859067525715e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 842495116,
+      "step": 4908
+    },
+    {
+      "epoch": 55.74504249291785,
+      "grad_norm": 4.796925740731655,
+      "learning_rate": 5e-06,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 842665972,
+      "step": 4909
+    },
+    {
+      "epoch": 55.74504249291785,
+      "loss": 0.10123119503259659,
+      "loss_ce": 0.00018749914306681603,
+      "loss_iou": 0.546875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 842665972,
+      "step": 4909
+    },
+    {
+      "epoch": 55.756373937677054,
+      "grad_norm": 3.8954137003505065,
+      "learning_rate": 5e-06,
+      "loss": 0.092,
+      "num_input_tokens_seen": 842836000,
+      "step": 4910
+    },
+    {
+      "epoch": 55.756373937677054,
+      "loss": 0.060989804565906525,
+      "loss_ce": 0.00018353178165853024,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 842836000,
+      "step": 4910
+    },
+    {
+      "epoch": 55.76770538243626,
+      "grad_norm": 3.424275686345672,
+      "learning_rate": 5e-06,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 843007948,
+      "step": 4911
+    },
+    {
+      "epoch": 55.76770538243626,
+      "loss": 0.1410728096961975,
+      "loss_ce": 6.634854071307927e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 843007948,
+      "step": 4911
+    },
+    {
+      "epoch": 55.77903682719547,
+      "grad_norm": 3.9696601319207447,
+      "learning_rate": 5e-06,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 843179012,
+      "step": 4912
+    },
+    {
+      "epoch": 55.77903682719547,
+      "loss": 0.05479143187403679,
+      "loss_ce": 0.0001497048360761255,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 843179012,
+      "step": 4912
+    },
+    {
+      "epoch": 55.79036827195468,
+      "grad_norm": 4.418036715516445,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 843349240,
+      "step": 4913
+    },
+    {
+      "epoch": 55.79036827195468,
+      "loss": 0.0803055465221405,
+      "loss_ce": 0.00012060943117830902,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 843349240,
+      "step": 4913
+    },
+    {
+      "epoch": 55.801699716713884,
+      "grad_norm": 3.85633454409002,
+      "learning_rate": 5e-06,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 843520008,
+      "step": 4914
+    },
+    {
+      "epoch": 55.801699716713884,
+      "loss": 0.04541558027267456,
+      "loss_ce": 5.1203820476075634e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 843520008,
+      "step": 4914
+    },
+    {
+      "epoch": 55.81303116147309,
+      "grad_norm": 4.506391250644138,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 843689300,
+      "step": 4915
+    },
+    {
+      "epoch": 55.81303116147309,
+      "loss": 0.06844595074653625,
+      "loss_ce": 0.0001323487376794219,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 843689300,
+      "step": 4915
+    },
+    {
+      "epoch": 55.82436260623229,
+      "grad_norm": 2.9060615384027937,
+      "learning_rate": 5e-06,
+      "loss": 0.054,
+      "num_input_tokens_seen": 843859248,
+      "step": 4916
+    },
+    {
+      "epoch": 55.82436260623229,
+      "loss": 0.03144977241754532,
+      "loss_ce": 0.00010821959585882723,
+      "loss_iou": 0.40625,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 843859248,
+      "step": 4916
+    },
+    {
+      "epoch": 55.8356940509915,
+      "grad_norm": 3.625623667593636,
+      "learning_rate": 5e-06,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 844029808,
+      "step": 4917
+    },
+    {
+      "epoch": 55.8356940509915,
+      "loss": 0.06820227205753326,
+      "loss_ce": 5.6525452237110585e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 844029808,
+      "step": 4917
+    },
+    {
+      "epoch": 55.84702549575071,
+      "grad_norm": 3.3309251540068745,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 844201496,
+      "step": 4918
+    },
+    {
+      "epoch": 55.84702549575071,
+      "loss": 0.05710037052631378,
+      "loss_ce": 0.00016982998931780457,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 844201496,
+      "step": 4918
+    },
+    {
+      "epoch": 55.858356940509914,
+      "grad_norm": 3.278909537406917,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 844373532,
+      "step": 4919
+    },
+    {
+      "epoch": 55.858356940509914,
+      "loss": 0.052809786051511765,
+      "loss_ce": 6.0152528021717444e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 844373532,
+      "step": 4919
+    },
+    {
+      "epoch": 55.86968838526912,
+      "grad_norm": 3.2660163344739854,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 844544464,
+      "step": 4920
+    },
+    {
+      "epoch": 55.86968838526912,
+      "loss": 0.059388354420661926,
+      "loss_ce": 7.74409927544184e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 844544464,
+      "step": 4920
+    },
+    {
+      "epoch": 55.88101983002833,
+      "grad_norm": 3.6625839864904806,
+      "learning_rate": 5e-06,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 844716164,
+      "step": 4921
+    },
+    {
+      "epoch": 55.88101983002833,
+      "loss": 0.03279288113117218,
+      "loss_ce": 0.0002153678797185421,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 844716164,
+      "step": 4921
+    },
+    {
+      "epoch": 55.892351274787536,
+      "grad_norm": 4.579621288372402,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 844887928,
+      "step": 4922
+    },
+    {
+      "epoch": 55.892351274787536,
+      "loss": 0.0599120631814003,
+      "loss_ce": 9.760991088114679e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 844887928,
+      "step": 4922
+    },
+    {
+      "epoch": 55.903682719546744,
+      "grad_norm": 3.7953442271450806,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 845059536,
+      "step": 4923
+    },
+    {
+      "epoch": 55.903682719546744,
+      "loss": 0.10065902769565582,
+      "loss_ce": 4.2575193219818175e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 845059536,
+      "step": 4923
+    },
+    {
+      "epoch": 55.91501416430595,
+      "grad_norm": 4.039896304998453,
+      "learning_rate": 5e-06,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 845230272,
+      "step": 4924
+    },
+    {
+      "epoch": 55.91501416430595,
+      "loss": 0.0445399172604084,
+      "loss_ce": 0.00010632298653945327,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 845230272,
+      "step": 4924
+    },
+    {
+      "epoch": 55.92634560906516,
+      "grad_norm": 3.7475278478992404,
+      "learning_rate": 5e-06,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 845400328,
+      "step": 4925
+    },
+    {
+      "epoch": 55.92634560906516,
+      "loss": 0.04840108007192612,
+      "loss_ce": 0.0006410708301700652,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 845400328,
+      "step": 4925
+    },
+    {
+      "epoch": 55.93767705382436,
+      "grad_norm": 3.631795058849131,
+      "learning_rate": 5e-06,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 845571388,
+      "step": 4926
+    },
+    {
+      "epoch": 55.93767705382436,
+      "loss": 0.05695248767733574,
+      "loss_ce": 6.772139749955386e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 845571388,
+      "step": 4926
+    },
+    {
+      "epoch": 55.94900849858357,
+      "grad_norm": 3.6403936228206923,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 845742144,
+      "step": 4927
+    },
+    {
+      "epoch": 55.94900849858357,
+      "loss": 0.0556352436542511,
+      "loss_ce": 0.00024584238417446613,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 845742144,
+      "step": 4927
+    },
+    {
+      "epoch": 55.960339943342774,
+      "grad_norm": 3.3421383799749287,
+      "learning_rate": 5e-06,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 845912708,
+      "step": 4928
+    },
+    {
+      "epoch": 55.960339943342774,
+      "loss": 0.03565491735935211,
+      "loss_ce": 5.615890404442325e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 845912708,
+      "step": 4928
+    },
+    {
+      "epoch": 55.97167138810198,
+      "grad_norm": 3.5515769376924964,
+      "learning_rate": 5e-06,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 846081776,
+      "step": 4929
+    },
+    {
+      "epoch": 55.97167138810198,
+      "loss": 0.03801915794610977,
+      "loss_ce": 5.529045301955193e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 846081776,
+      "step": 4929
+    },
+    {
+      "epoch": 55.98300283286119,
+      "grad_norm": 3.38914912074444,
+      "learning_rate": 5e-06,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 846253676,
+      "step": 4930
+    },
+    {
+      "epoch": 55.98300283286119,
+      "loss": 0.06063020974397659,
+      "loss_ce": 0.0002054050419246778,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 846253676,
+      "step": 4930
+    },
+    {
+      "epoch": 55.994334277620396,
+      "grad_norm": 2.964755871467462,
+      "learning_rate": 5e-06,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 846424632,
+      "step": 4931
+    },
+    {
+      "epoch": 55.994334277620396,
+      "loss": 0.11273901909589767,
+      "loss_ce": 8.337547478731722e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 846424632,
+      "step": 4931
+    },
+    {
+      "epoch": 55.994334277620396,
+      "loss": 0.0826747789978981,
+      "loss_ce": 4.843393253395334e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 846509616,
+      "step": 4931
+    },
+    {
+      "epoch": 56.005665722379604,
+      "grad_norm": 3.3536091692495864,
+      "learning_rate": 5e-06,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 846595100,
+      "step": 4932
+    },
+    {
+      "epoch": 56.005665722379604,
+      "loss": 0.05322595313191414,
+      "loss_ce": 0.00015588558744639158,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 846595100,
+      "step": 4932
+    },
+    {
+      "epoch": 56.01699716713881,
+      "grad_norm": 3.7673356202276587,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 846766744,
+      "step": 4933
+    },
+    {
+      "epoch": 56.01699716713881,
+      "loss": 0.05581562966108322,
+      "loss_ce": 0.0002888973103836179,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 846766744,
+      "step": 4933
+    },
+    {
+      "epoch": 56.02832861189802,
+      "grad_norm": 4.289372405513039,
+      "learning_rate": 5e-06,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 846936064,
+      "step": 4934
+    },
+    {
+      "epoch": 56.02832861189802,
+      "loss": 0.07163050025701523,
+      "loss_ce": 9.729446901474148e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 846936064,
+      "step": 4934
+    },
+    {
+      "epoch": 56.039660056657226,
+      "grad_norm": 4.189385506432874,
+      "learning_rate": 5e-06,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 847106620,
+      "step": 4935
+    },
+    {
+      "epoch": 56.039660056657226,
+      "loss": 0.10359985381364822,
+      "loss_ce": 8.422660903306678e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 847106620,
+      "step": 4935
+    },
+    {
+      "epoch": 56.05099150141643,
+      "grad_norm": 5.283206833997654,
+      "learning_rate": 5e-06,
+      "loss": 0.0695,
+      "num_input_tokens_seen": 847278548,
+      "step": 4936
+    },
+    {
+      "epoch": 56.05099150141643,
+      "loss": 0.08981213718652725,
+      "loss_ce": 0.00012097464059479535,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 847278548,
+      "step": 4936
+    },
+    {
+      "epoch": 56.06232294617564,
+      "grad_norm": 3.5717911775909035,
+      "learning_rate": 5e-06,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 847450156,
+      "step": 4937
+    },
+    {
+      "epoch": 56.06232294617564,
+      "loss": 0.050839874893426895,
+      "loss_ce": 0.00011966139572905377,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 847450156,
+      "step": 4937
+    },
+    {
+      "epoch": 56.07365439093484,
+      "grad_norm": 3.3854271070409374,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 847622068,
+      "step": 4938
+    },
+    {
+      "epoch": 56.07365439093484,
+      "loss": 0.09919365495443344,
+      "loss_ce": 7.255817763507366e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 847622068,
+      "step": 4938
+    },
+    {
+      "epoch": 56.08498583569405,
+      "grad_norm": 4.114400871868835,
+      "learning_rate": 5e-06,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 847793616,
+      "step": 4939
+    },
+    {
+      "epoch": 56.08498583569405,
+      "loss": 0.06629049777984619,
+      "loss_ce": 0.00011313016875647008,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 847793616,
+      "step": 4939
+    },
+    {
+      "epoch": 56.096317280453256,
+      "grad_norm": 3.5425123253729733,
+      "learning_rate": 5e-06,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 847962836,
+      "step": 4940
+    },
+    {
+      "epoch": 56.096317280453256,
+      "loss": 0.09083575010299683,
+      "loss_ce": 0.00025957508478313684,
+      "loss_iou": 0.375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 847962836,
+      "step": 4940
+    },
+    {
+      "epoch": 56.107648725212464,
+      "grad_norm": 2.996206786367124,
+      "learning_rate": 5e-06,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 848134244,
+      "step": 4941
+    },
+    {
+      "epoch": 56.107648725212464,
+      "loss": 0.031099818646907806,
+      "loss_ce": 7.869962428230792e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 848134244,
+      "step": 4941
+    },
+    {
+      "epoch": 56.11898016997167,
+      "grad_norm": 3.6048026164947617,
+      "learning_rate": 5e-06,
+      "loss": 0.0824,
+      "num_input_tokens_seen": 848305708,
+      "step": 4942
+    },
+    {
+      "epoch": 56.11898016997167,
+      "loss": 0.07235158234834671,
+      "loss_ce": 8.595604595029727e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 848305708,
+      "step": 4942
+    },
+    {
+      "epoch": 56.13031161473088,
+      "grad_norm": 3.328234090233558,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 848475716,
+      "step": 4943
+    },
+    {
+      "epoch": 56.13031161473088,
+      "loss": 0.04021462798118591,
+      "loss_ce": 5.349469211068936e-05,
+      "loss_iou": 0.07861328125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 848475716,
+      "step": 4943
+    },
+    {
+      "epoch": 56.141643059490086,
+      "grad_norm": 4.080033657337501,
+      "learning_rate": 5e-06,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 848646964,
+      "step": 4944
+    },
+    {
+      "epoch": 56.141643059490086,
+      "loss": 0.07201793789863586,
+      "loss_ce": 8.800456998869777e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 848646964,
+      "step": 4944
+    },
+    {
+      "epoch": 56.15297450424929,
+      "grad_norm": 3.3278663990506483,
+      "learning_rate": 5e-06,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 848818064,
+      "step": 4945
+    },
+    {
+      "epoch": 56.15297450424929,
+      "loss": 0.05643495172262192,
+      "loss_ce": 6.898325227666646e-05,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 848818064,
+      "step": 4945
+    },
+    {
+      "epoch": 56.1643059490085,
+      "grad_norm": 2.662876391110361,
+      "learning_rate": 5e-06,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 848990168,
+      "step": 4946
+    },
+    {
+      "epoch": 56.1643059490085,
+      "loss": 0.046533383429050446,
+      "loss_ce": 0.00011614571121754125,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 848990168,
+      "step": 4946
+    },
+    {
+      "epoch": 56.17563739376771,
+      "grad_norm": 3.098204345612474,
+      "learning_rate": 5e-06,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 849162136,
+      "step": 4947
+    },
+    {
+      "epoch": 56.17563739376771,
+      "loss": 0.03562714904546738,
+      "loss_ce": 7.41688854759559e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 849162136,
+      "step": 4947
+    },
+    {
+      "epoch": 56.186968838526916,
+      "grad_norm": 4.799208198369888,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 849332956,
+      "step": 4948
+    },
+    {
+      "epoch": 56.186968838526916,
+      "loss": 0.13538876175880432,
+      "loss_ce": 7.38299495424144e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 849332956,
+      "step": 4948
+    },
+    {
+      "epoch": 56.198300283286116,
+      "grad_norm": 3.9671172987701877,
+      "learning_rate": 5e-06,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 849505204,
+      "step": 4949
+    },
+    {
+      "epoch": 56.198300283286116,
+      "loss": 0.0647658109664917,
+      "loss_ce": 5.328988845576532e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 849505204,
+      "step": 4949
+    },
+    {
+      "epoch": 56.20963172804532,
+      "grad_norm": 3.900793327298166,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 849677564,
+      "step": 4950
+    },
+    {
+      "epoch": 56.20963172804532,
+      "loss": 0.042843788862228394,
+      "loss_ce": 8.866139978636056e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 849677564,
+      "step": 4950
+    },
+    {
+      "epoch": 56.22096317280453,
+      "grad_norm": 3.839341613292569,
+      "learning_rate": 5e-06,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 849849180,
+      "step": 4951
+    },
+    {
+      "epoch": 56.22096317280453,
+      "loss": 0.059122323989868164,
+      "loss_ce": 0.0001013303262880072,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 849849180,
+      "step": 4951
+    },
+    {
+      "epoch": 56.23229461756374,
+      "grad_norm": 3.8769282289676754,
+      "learning_rate": 5e-06,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 850019312,
+      "step": 4952
+    },
+    {
+      "epoch": 56.23229461756374,
+      "loss": 0.03659261763095856,
+      "loss_ce": 0.0001241089921677485,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 850019312,
+      "step": 4952
+    },
+    {
+      "epoch": 56.243626062322946,
+      "grad_norm": 3.4578204485483766,
+      "learning_rate": 5e-06,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 850190856,
+      "step": 4953
+    },
+    {
+      "epoch": 56.243626062322946,
+      "loss": 0.10320930182933807,
+      "loss_ce": 4.4626533053815365e-05,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 850190856,
+      "step": 4953
+    },
+    {
+      "epoch": 56.25495750708215,
+      "grad_norm": 4.7885427532956575,
+      "learning_rate": 5e-06,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 850363280,
+      "step": 4954
+    },
+    {
+      "epoch": 56.25495750708215,
+      "loss": 0.058364663273096085,
+      "loss_ce": 4.5572338422061875e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 850363280,
+      "step": 4954
+    },
+    {
+      "epoch": 56.26628895184136,
+      "grad_norm": 3.9535189843192984,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 850533712,
+      "step": 4955
+    },
+    {
+      "epoch": 56.26628895184136,
+      "loss": 0.13784638047218323,
+      "loss_ce": 2.901221341744531e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 850533712,
+      "step": 4955
+    },
+    {
+      "epoch": 56.27762039660057,
+      "grad_norm": 4.733945619596806,
+      "learning_rate": 5e-06,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 850705532,
+      "step": 4956
+    },
+    {
+      "epoch": 56.27762039660057,
+      "loss": 0.07580488920211792,
+      "loss_ce": 6.0261842008912936e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 850705532,
+      "step": 4956
+    },
+    {
+      "epoch": 56.288951841359776,
+      "grad_norm": 4.100497366778174,
+      "learning_rate": 5e-06,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 850877292,
+      "step": 4957
+    },
+    {
+      "epoch": 56.288951841359776,
+      "loss": 0.04398246854543686,
+      "loss_ce": 9.819120168685913e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 850877292,
+      "step": 4957
+    },
+    {
+      "epoch": 56.30028328611898,
+      "grad_norm": 3.62200045473798,
+      "learning_rate": 5e-06,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 851049008,
+      "step": 4958
+    },
+    {
+      "epoch": 56.30028328611898,
+      "loss": 0.05289608985185623,
+      "loss_ce": 7.016229210421443e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 851049008,
+      "step": 4958
+    },
+    {
+      "epoch": 56.31161473087819,
+      "grad_norm": 3.8524813387577854,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 851219144,
+      "step": 4959
+    },
+    {
+      "epoch": 56.31161473087819,
+      "loss": 0.07554268091917038,
+      "loss_ce": 0.00013374886475503445,
+      "loss_iou": 0.5,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 851219144,
+      "step": 4959
+    },
+    {
+      "epoch": 56.32294617563739,
+      "grad_norm": 3.4014691755529403,
+      "learning_rate": 5e-06,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 851390180,
+      "step": 4960
+    },
+    {
+      "epoch": 56.32294617563739,
+      "loss": 0.041571419686079025,
+      "loss_ce": 0.00014380962238647044,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 851390180,
+      "step": 4960
+    },
+    {
+      "epoch": 56.3342776203966,
+      "grad_norm": 3.3987210190568984,
+      "learning_rate": 5e-06,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 851561384,
+      "step": 4961
+    },
+    {
+      "epoch": 56.3342776203966,
+      "loss": 0.05235926806926727,
+      "loss_ce": 6.739788659615442e-05,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 851561384,
+      "step": 4961
+    },
+    {
+      "epoch": 56.345609065155806,
+      "grad_norm": 4.12563978245002,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 851732804,
+      "step": 4962
+    },
+    {
+      "epoch": 56.345609065155806,
+      "loss": 0.1039101779460907,
+      "loss_ce": 7.411520346067846e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 851732804,
+      "step": 4962
+    },
+    {
+      "epoch": 56.35694050991501,
+      "grad_norm": 6.658687136813664,
+      "learning_rate": 5e-06,
+      "loss": 0.0672,
+      "num_input_tokens_seen": 851904452,
+      "step": 4963
+    },
+    {
+      "epoch": 56.35694050991501,
+      "loss": 0.10025132447481155,
+      "loss_ce": 9.26262655411847e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 851904452,
+      "step": 4963
+    },
+    {
+      "epoch": 56.36827195467422,
+      "grad_norm": 3.8907108610126087,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 852074612,
+      "step": 4964
+    },
+    {
+      "epoch": 56.36827195467422,
+      "loss": 0.05475816875696182,
+      "loss_ce": 4.0155638998840004e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 852074612,
+      "step": 4964
+    },
+    {
+      "epoch": 56.37960339943343,
+      "grad_norm": 3.7545912140779767,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 852245048,
+      "step": 4965
+    },
+    {
+      "epoch": 56.37960339943343,
+      "loss": 0.1148814857006073,
+      "loss_ce": 0.00013538844359572977,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 852245048,
+      "step": 4965
+    },
+    {
+      "epoch": 56.390934844192635,
+      "grad_norm": 3.1588256694801804,
+      "learning_rate": 5e-06,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 852415672,
+      "step": 4966
+    },
+    {
+      "epoch": 56.390934844192635,
+      "loss": 0.042692095041275024,
+      "loss_ce": 8.955708472058177e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 852415672,
+      "step": 4966
+    },
+    {
+      "epoch": 56.40226628895184,
+      "grad_norm": 5.321244010620065,
+      "learning_rate": 5e-06,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 852586424,
+      "step": 4967
+    },
+    {
+      "epoch": 56.40226628895184,
+      "loss": 0.036578815430402756,
+      "loss_ce": 7.979491783771664e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 852586424,
+      "step": 4967
+    },
+    {
+      "epoch": 56.41359773371105,
+      "grad_norm": 5.584262332178128,
+      "learning_rate": 5e-06,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 852757076,
+      "step": 4968
+    },
+    {
+      "epoch": 56.41359773371105,
+      "loss": 0.03489917516708374,
+      "loss_ce": 4.8100184358190745e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 852757076,
+      "step": 4968
+    },
+    {
+      "epoch": 56.42492917847026,
+      "grad_norm": 3.3172669578355714,
+      "learning_rate": 5e-06,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 852928820,
+      "step": 4969
+    },
+    {
+      "epoch": 56.42492917847026,
+      "loss": 0.03755845129489899,
+      "loss_ce": 8.286676165880635e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 852928820,
+      "step": 4969
+    },
+    {
+      "epoch": 56.436260623229465,
+      "grad_norm": 3.3188128448066143,
+      "learning_rate": 5e-06,
+      "loss": 0.064,
+      "num_input_tokens_seen": 853098516,
+      "step": 4970
+    },
+    {
+      "epoch": 56.436260623229465,
+      "loss": 0.06092609092593193,
+      "loss_ce": 5.8779402024811134e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 853098516,
+      "step": 4970
+    },
+    {
+      "epoch": 56.447592067988666,
+      "grad_norm": 3.4500553346177925,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 853268508,
+      "step": 4971
+    },
+    {
+      "epoch": 56.447592067988666,
+      "loss": 0.0780632346868515,
+      "loss_ce": 2.9783921490889043e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 853268508,
+      "step": 4971
+    },
+    {
+      "epoch": 56.45892351274787,
+      "grad_norm": 2.636503485244163,
+      "learning_rate": 5e-06,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 853439300,
+      "step": 4972
+    },
+    {
+      "epoch": 56.45892351274787,
+      "loss": 0.050229206681251526,
+      "loss_ce": 4.305325273890048e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 853439300,
+      "step": 4972
+    },
+    {
+      "epoch": 56.47025495750708,
+      "grad_norm": 3.111984902578889,
+      "learning_rate": 5e-06,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 853611144,
+      "step": 4973
+    },
+    {
+      "epoch": 56.47025495750708,
+      "loss": 0.08155134320259094,
+      "loss_ce": 8.466400322504342e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 853611144,
+      "step": 4973
+    },
+    {
+      "epoch": 56.48158640226629,
+      "grad_norm": 3.624977827208369,
+      "learning_rate": 5e-06,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 853783024,
+      "step": 4974
+    },
+    {
+      "epoch": 56.48158640226629,
+      "loss": 0.03524838387966156,
+      "loss_ce": 9.213176235789433e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 853783024,
+      "step": 4974
+    },
+    {
+      "epoch": 56.492917847025495,
+      "grad_norm": 2.9029984017162103,
+      "learning_rate": 5e-06,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 853954680,
+      "step": 4975
+    },
+    {
+      "epoch": 56.492917847025495,
+      "loss": 0.035742372274398804,
+      "loss_ce": 6.732695328537375e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 853954680,
+      "step": 4975
+    },
+    {
+      "epoch": 56.5042492917847,
+      "grad_norm": 3.3800031713130623,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 854122288,
+      "step": 4976
+    },
+    {
+      "epoch": 56.5042492917847,
+      "loss": 0.03584957867860794,
+      "loss_ce": 9.82389465207234e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 854122288,
+      "step": 4976
+    },
+    {
+      "epoch": 56.51558073654391,
+      "grad_norm": 4.575082251332415,
+      "learning_rate": 5e-06,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 854294344,
+      "step": 4977
+    },
+    {
+      "epoch": 56.51558073654391,
+      "loss": 0.05696126073598862,
+      "loss_ce": 4.597748556989245e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 854294344,
+      "step": 4977
+    },
+    {
+      "epoch": 56.52691218130312,
+      "grad_norm": 3.288662489137845,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 854466328,
+      "step": 4978
+    },
+    {
+      "epoch": 56.52691218130312,
+      "loss": 0.047406382858753204,
+      "loss_ce": 0.0005161237204447389,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 854466328,
+      "step": 4978
+    },
+    {
+      "epoch": 56.538243626062325,
+      "grad_norm": 3.7791332314263353,
+      "learning_rate": 5e-06,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 854638396,
+      "step": 4979
+    },
+    {
+      "epoch": 56.538243626062325,
+      "loss": 0.05304059758782387,
+      "loss_ce": 4.682278085965663e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 854638396,
+      "step": 4979
+    },
+    {
+      "epoch": 56.54957507082153,
+      "grad_norm": 3.3139414531224483,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 854809012,
+      "step": 4980
+    },
+    {
+      "epoch": 56.54957507082153,
+      "loss": 0.08319170027971268,
+      "loss_ce": 0.0002906979061663151,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 854809012,
+      "step": 4980
+    },
+    {
+      "epoch": 56.56090651558074,
+      "grad_norm": 2.6589242281008043,
+      "learning_rate": 5e-06,
+      "loss": 0.1096,
+      "num_input_tokens_seen": 854979212,
+      "step": 4981
+    },
+    {
+      "epoch": 56.56090651558074,
+      "loss": 0.11985158175230026,
+      "loss_ce": 8.534132211934775e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 854979212,
+      "step": 4981
+    },
+    {
+      "epoch": 56.57223796033994,
+      "grad_norm": 2.946845232078595,
+      "learning_rate": 5e-06,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 855151068,
+      "step": 4982
+    },
+    {
+      "epoch": 56.57223796033994,
+      "loss": 0.07540497928857803,
+      "loss_ce": 5.7077835663221776e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 855151068,
+      "step": 4982
+    },
+    {
+      "epoch": 56.58356940509915,
+      "grad_norm": 3.252890628692685,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 855322920,
+      "step": 4983
+    },
+    {
+      "epoch": 56.58356940509915,
+      "loss": 0.03005708009004593,
+      "loss_ce": 7.356132118729874e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 855322920,
+      "step": 4983
+    },
+    {
+      "epoch": 56.594900849858355,
+      "grad_norm": 3.4605470952618327,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 855494476,
+      "step": 4984
+    },
+    {
+      "epoch": 56.594900849858355,
+      "loss": 0.040171485394239426,
+      "loss_ce": 7.138719956856221e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 855494476,
+      "step": 4984
+    },
+    {
+      "epoch": 56.60623229461756,
+      "grad_norm": 2.695984834480556,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 855665256,
+      "step": 4985
+    },
+    {
+      "epoch": 56.60623229461756,
+      "loss": 0.033694181591272354,
+      "loss_ce": 0.0002011396427405998,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 855665256,
+      "step": 4985
+    },
+    {
+      "epoch": 56.61756373937677,
+      "grad_norm": 3.1628748981828,
+      "learning_rate": 5e-06,
+      "loss": 0.0866,
+      "num_input_tokens_seen": 855836092,
+      "step": 4986
+    },
+    {
+      "epoch": 56.61756373937677,
+      "loss": 0.06583873182535172,
+      "loss_ce": 7.335042755585164e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 855836092,
+      "step": 4986
+    },
+    {
+      "epoch": 56.62889518413598,
+      "grad_norm": 5.016588449832622,
+      "learning_rate": 5e-06,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 856008108,
+      "step": 4987
+    },
+    {
+      "epoch": 56.62889518413598,
+      "loss": 0.05184105411171913,
+      "loss_ce": 6.798441609134898e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 856008108,
+      "step": 4987
+    },
+    {
+      "epoch": 56.640226628895185,
+      "grad_norm": 3.2625502096198398,
+      "learning_rate": 5e-06,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 856180012,
+      "step": 4988
+    },
+    {
+      "epoch": 56.640226628895185,
+      "loss": 0.043545953929424286,
+      "loss_ce": 7.366194040514529e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 856180012,
+      "step": 4988
+    },
+    {
+      "epoch": 56.65155807365439,
+      "grad_norm": 3.1762156173856084,
+      "learning_rate": 5e-06,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 856351032,
+      "step": 4989
+    },
+    {
+      "epoch": 56.65155807365439,
+      "loss": 0.10385333001613617,
+      "loss_ce": 9.355971997138113e-05,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 856351032,
+      "step": 4989
+    },
+    {
+      "epoch": 56.6628895184136,
+      "grad_norm": 3.367729780434394,
+      "learning_rate": 5e-06,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 856520400,
+      "step": 4990
+    },
+    {
+      "epoch": 56.6628895184136,
+      "loss": 0.04593687877058983,
+      "loss_ce": 0.0001299942086916417,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 856520400,
+      "step": 4990
+    },
+    {
+      "epoch": 56.67422096317281,
+      "grad_norm": 3.493860391116628,
+      "learning_rate": 5e-06,
+      "loss": 0.0725,
+      "num_input_tokens_seen": 856692220,
+      "step": 4991
+    },
+    {
+      "epoch": 56.67422096317281,
+      "loss": 0.06501106917858124,
+      "loss_ce": 5.4410804295912385e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 856692220,
+      "step": 4991
+    },
+    {
+      "epoch": 56.685552407932015,
+      "grad_norm": 3.489201211541105,
+      "learning_rate": 5e-06,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 856864100,
+      "step": 4992
+    },
+    {
+      "epoch": 56.685552407932015,
+      "loss": 0.0858091190457344,
+      "loss_ce": 6.998020398896188e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 856864100,
+      "step": 4992
+    },
+    {
+      "epoch": 56.696883852691215,
+      "grad_norm": 4.204124544672858,
+      "learning_rate": 5e-06,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 857035852,
+      "step": 4993
+    },
+    {
+      "epoch": 56.696883852691215,
+      "loss": 0.053366005420684814,
+      "loss_ce": 0.00023490253079216927,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 857035852,
+      "step": 4993
+    },
+    {
+      "epoch": 56.70821529745042,
+      "grad_norm": 3.886155176632275,
+      "learning_rate": 5e-06,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 857205796,
+      "step": 4994
+    },
+    {
+      "epoch": 56.70821529745042,
+      "loss": 0.05728646367788315,
+      "loss_ce": 0.00014230137458071113,
+      "loss_iou": 0.296875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 857205796,
+      "step": 4994
+    },
+    {
+      "epoch": 56.71954674220963,
+      "grad_norm": 4.2344686707047305,
+      "learning_rate": 5e-06,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 857376164,
+      "step": 4995
+    },
+    {
+      "epoch": 56.71954674220963,
+      "loss": 0.07924749702215195,
+      "loss_ce": 2.386427513556555e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 857376164,
+      "step": 4995
+    },
+    {
+      "epoch": 56.73087818696884,
+      "grad_norm": 10.997457627545906,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 857548108,
+      "step": 4996
+    },
+    {
+      "epoch": 56.73087818696884,
+      "loss": 0.06188258156180382,
+      "loss_ce": 0.00011500447726575658,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 857548108,
+      "step": 4996
+    },
+    {
+      "epoch": 56.742209631728045,
+      "grad_norm": 3.6828098030553496,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 857719680,
+      "step": 4997
+    },
+    {
+      "epoch": 56.742209631728045,
+      "loss": 0.12832649052143097,
+      "loss_ce": 0.0003662837261799723,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 857719680,
+      "step": 4997
+    },
+    {
+      "epoch": 56.75354107648725,
+      "grad_norm": 3.3609040150071428,
+      "learning_rate": 5e-06,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 857890152,
+      "step": 4998
+    },
+    {
+      "epoch": 56.75354107648725,
+      "loss": 0.06360742449760437,
+      "loss_ce": 3.9312657463597134e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 857890152,
+      "step": 4998
+    },
+    {
+      "epoch": 56.76487252124646,
+      "grad_norm": 3.610086961053706,
+      "learning_rate": 5e-06,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 858061716,
+      "step": 4999
+    },
+    {
+      "epoch": 56.76487252124646,
+      "loss": 0.05188732594251633,
+      "loss_ce": 8.373636228498071e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 858061716,
+      "step": 4999
+    },
+    {
+      "epoch": 56.77620396600567,
+      "grad_norm": 3.2722959488567205,
+      "learning_rate": 5e-06,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 858233480,
+      "step": 5000
+    },
+    {
+      "epoch": 56.77620396600567,
+      "eval_seeclick_CIoU": 0.5374038815498352,
+      "eval_seeclick_GIoU": 0.5343401730060577,
+      "eval_seeclick_IoU": 0.5729995667934418,
+      "eval_seeclick_MAE_all": 0.06275784783065319,
+      "eval_seeclick_MAE_h": 0.029513271525502205,
+      "eval_seeclick_MAE_w": 0.0937967300415039,
+      "eval_seeclick_MAE_x": 0.09877407923340797,
+      "eval_seeclick_MAE_y": 0.028947304002940655,
+      "eval_seeclick_NUM_probability": 0.999929815530777,
+      "eval_seeclick_inside_bbox": 0.8920454680919647,
+      "eval_seeclick_loss": 0.8652315139770508,
+      "eval_seeclick_loss_ce": 0.6316386461257935,
+      "eval_seeclick_loss_iou": 0.5211181640625,
+      "eval_seeclick_loss_num": 0.04671478271484375,
+      "eval_seeclick_loss_xval": 0.23345947265625,
+      "eval_seeclick_runtime": 68.9432,
+      "eval_seeclick_samples_per_second": 0.624,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 858233480,
+      "step": 5000
+    },
+    {
+      "epoch": 56.77620396600567,
+      "eval_icons_CIoU": 0.7602474093437195,
+      "eval_icons_GIoU": 0.7598395049571991,
+      "eval_icons_IoU": 0.772760659456253,
+      "eval_icons_MAE_all": 0.03311178460717201,
+      "eval_icons_MAE_h": 0.030518805608153343,
+      "eval_icons_MAE_w": 0.03753254935145378,
+      "eval_icons_MAE_x": 0.031239313073456287,
+      "eval_icons_MAE_y": 0.033156465739011765,
+      "eval_icons_NUM_probability": 0.9983026385307312,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.11245551705360413,
+      "eval_icons_loss_ce": 0.0024013028014451265,
+      "eval_icons_loss_iou": 0.5394287109375,
+      "eval_icons_loss_num": 0.020328521728515625,
+      "eval_icons_loss_xval": 0.1016693115234375,
+      "eval_icons_runtime": 78.7876,
+      "eval_icons_samples_per_second": 0.635,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 858233480,
+      "step": 5000
+    },
+    {
+      "epoch": 56.77620396600567,
+      "eval_screenspot_CIoU": 0.6211823423703512,
+      "eval_screenspot_GIoU": 0.6262733538945516,
+      "eval_screenspot_IoU": 0.6511431932449341,
+      "eval_screenspot_MAE_all": 0.06840701152880986,
+      "eval_screenspot_MAE_h": 0.033884188160300255,
+      "eval_screenspot_MAE_w": 0.12381209433078766,
+      "eval_screenspot_MAE_x": 0.08469116439421971,
+      "eval_screenspot_MAE_y": 0.031240610716243584,
+      "eval_screenspot_NUM_probability": 0.9999233881632487,
+      "eval_screenspot_inside_bbox": 0.8974999984105428,
+      "eval_screenspot_loss": 0.28809013962745667,
+      "eval_screenspot_loss_ce": 0.013613161475708088,
+      "eval_screenspot_loss_iou": 0.498046875,
+      "eval_screenspot_loss_num": 0.054570515950520836,
+      "eval_screenspot_loss_xval": 0.2729085286458333,
+      "eval_screenspot_runtime": 147.1347,
+      "eval_screenspot_samples_per_second": 0.605,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 858233480,
+      "step": 5000
+    },
+    {
+      "epoch": 56.77620396600567,
+      "eval_compot_CIoU": 0.8435192108154297,
+      "eval_compot_GIoU": 0.8384089767932892,
+      "eval_compot_IoU": 0.8579079210758209,
+      "eval_compot_MAE_all": 0.02599179372191429,
+      "eval_compot_MAE_h": 0.02002739906311035,
+      "eval_compot_MAE_w": 0.03220875933766365,
+      "eval_compot_MAE_x": 0.029916545376181602,
+      "eval_compot_MAE_y": 0.02181447669863701,
+      "eval_compot_NUM_probability": 0.9999608099460602,
+      "eval_compot_inside_bbox": 0.9565972089767456,
+      "eval_compot_loss": 0.08126524835824966,
+      "eval_compot_loss_ce": 5.167060953681357e-05,
+      "eval_compot_loss_iou": 0.50213623046875,
+      "eval_compot_loss_num": 0.014265060424804688,
+      "eval_compot_loss_xval": 0.07131195068359375,
+      "eval_compot_runtime": 82.3608,
+      "eval_compot_samples_per_second": 0.607,
+      "eval_compot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 858233480,
+      "step": 5000
+    },
+    {
+      "epoch": 56.77620396600567,
+      "eval_custom_ui_MAE_all": 0.019572661723941565,
+      "eval_custom_ui_MAE_x": 0.031017981469631195,
+      "eval_custom_ui_MAE_y": 0.008127343142405152,
+      "eval_custom_ui_NUM_probability": 0.9998641312122345,
+      "eval_custom_ui_loss": 0.2092946618795395,
+      "eval_custom_ui_loss_ce": 0.11008366197347641,
+      "eval_custom_ui_loss_num": 0.018358230590820312,
+      "eval_custom_ui_loss_xval": 0.091827392578125,
+      "eval_custom_ui_runtime": 64.8486,
+      "eval_custom_ui_samples_per_second": 0.771,
+      "eval_custom_ui_steps_per_second": 0.031,
+      "num_input_tokens_seen": 858233480,
+      "step": 5000
+    },
+    {
+      "epoch": 56.77620396600567,
+      "loss": 0.25011128187179565,
+      "loss_ce": 0.13557879626750946,
+      "loss_iou": 0.0,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 858233480,
+      "step": 5000
+    },
+    {
+      "epoch": 56.787535410764875,
+      "grad_norm": 3.531880139320847,
+      "learning_rate": 5e-06,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 858404320,
+      "step": 5001
+    },
+    {
+      "epoch": 56.787535410764875,
+      "loss": 0.06346552073955536,
+      "loss_ce": 6.525822391267866e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 858404320,
+      "step": 5001
+    },
+    {
+      "epoch": 56.79886685552408,
+      "grad_norm": 5.595290723932635,
+      "learning_rate": 5e-06,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 858575740,
+      "step": 5002
+    },
+    {
+      "epoch": 56.79886685552408,
+      "loss": 0.06861157715320587,
+      "loss_ce": 9.96118105831556e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 858575740,
+      "step": 5002
+    },
+    {
+      "epoch": 56.81019830028329,
+      "grad_norm": 3.731199623088894,
+      "learning_rate": 5e-06,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 858745836,
+      "step": 5003
+    },
+    {
+      "epoch": 56.81019830028329,
+      "loss": 0.04174460470676422,
+      "loss_ce": 0.00010337021376471967,
+      "loss_iou": 0.546875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 858745836,
+      "step": 5003
+    },
+    {
+      "epoch": 56.82152974504249,
+      "grad_norm": 4.050315692478664,
+      "learning_rate": 5e-06,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 858917428,
+      "step": 5004
+    },
+    {
+      "epoch": 56.82152974504249,
+      "loss": 0.04483303427696228,
+      "loss_ce": 0.00015530255041085184,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 858917428,
+      "step": 5004
+    },
+    {
+      "epoch": 56.8328611898017,
+      "grad_norm": 4.079478613238688,
+      "learning_rate": 5e-06,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 859086580,
+      "step": 5005
+    },
+    {
+      "epoch": 56.8328611898017,
+      "loss": 0.0678957849740982,
+      "loss_ce": 7.046588871162385e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 859086580,
+      "step": 5005
+    },
+    {
+      "epoch": 56.844192634560905,
+      "grad_norm": 3.8407059310069003,
+      "learning_rate": 5e-06,
+      "loss": 0.053,
+      "num_input_tokens_seen": 859258396,
+      "step": 5006
+    },
+    {
+      "epoch": 56.844192634560905,
+      "loss": 0.07439401745796204,
+      "loss_ce": 5.3192015911918133e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 859258396,
+      "step": 5006
+    },
+    {
+      "epoch": 56.85552407932011,
+      "grad_norm": 3.875403017964247,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 859429420,
+      "step": 5007
+    },
+    {
+      "epoch": 56.85552407932011,
+      "loss": 0.10895470529794693,
+      "loss_ce": 8.324792725034058e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 859429420,
+      "step": 5007
+    },
+    {
+      "epoch": 56.86685552407932,
+      "grad_norm": 2.989740407858414,
+      "learning_rate": 5e-06,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 859601012,
+      "step": 5008
+    },
+    {
+      "epoch": 56.86685552407932,
+      "loss": 0.038098469376564026,
+      "loss_ce": 2.7790672902483493e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 859601012,
+      "step": 5008
+    },
+    {
+      "epoch": 56.87818696883853,
+      "grad_norm": 2.907420240435011,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 859771720,
+      "step": 5009
+    },
+    {
+      "epoch": 56.87818696883853,
+      "loss": 0.060057006776332855,
+      "loss_ce": 0.0009597192984074354,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 859771720,
+      "step": 5009
+    },
+    {
+      "epoch": 56.889518413597735,
+      "grad_norm": 3.421985480487567,
+      "learning_rate": 5e-06,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 859943672,
+      "step": 5010
+    },
+    {
+      "epoch": 56.889518413597735,
+      "loss": 0.024871021509170532,
+      "loss_ce": 6.022902380209416e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 859943672,
+      "step": 5010
+    },
+    {
+      "epoch": 56.90084985835694,
+      "grad_norm": 6.224721474708366,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 860114784,
+      "step": 5011
+    },
+    {
+      "epoch": 56.90084985835694,
+      "loss": 0.09442857652902603,
+      "loss_ce": 9.873868839349598e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 860114784,
+      "step": 5011
+    },
+    {
+      "epoch": 56.91218130311615,
+      "grad_norm": 3.0724953723064012,
+      "learning_rate": 5e-06,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 860286620,
+      "step": 5012
+    },
+    {
+      "epoch": 56.91218130311615,
+      "loss": 0.07430398464202881,
+      "loss_ce": 5.4709242249373347e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 860286620,
+      "step": 5012
+    },
+    {
+      "epoch": 56.92351274787536,
+      "grad_norm": 3.575510818823419,
+      "learning_rate": 5e-06,
+      "loss": 0.049,
+      "num_input_tokens_seen": 860458152,
+      "step": 5013
+    },
+    {
+      "epoch": 56.92351274787536,
+      "loss": 0.03822827711701393,
+      "loss_ce": 3.5529570595826954e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 860458152,
+      "step": 5013
+    },
+    {
+      "epoch": 56.934844192634564,
+      "grad_norm": 3.9125080639132617,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 860630088,
+      "step": 5014
+    },
+    {
+      "epoch": 56.934844192634564,
+      "loss": 0.0815986916422844,
+      "loss_ce": 0.00016253037028945982,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 860630088,
+      "step": 5014
+    },
+    {
+      "epoch": 56.946175637393765,
+      "grad_norm": 3.9089289880742832,
+      "learning_rate": 5e-06,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 860800904,
+      "step": 5015
+    },
+    {
+      "epoch": 56.946175637393765,
+      "loss": 0.05884288251399994,
+      "loss_ce": 3.5510980524122715e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 860800904,
+      "step": 5015
+    },
+    {
+      "epoch": 56.95750708215297,
+      "grad_norm": 4.312881125873777,
+      "learning_rate": 5e-06,
+      "loss": 0.063,
+      "num_input_tokens_seen": 860972328,
+      "step": 5016
+    },
+    {
+      "epoch": 56.95750708215297,
+      "loss": 0.03629220277070999,
+      "loss_ce": 0.000983365927822888,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 860972328,
+      "step": 5016
+    },
+    {
+      "epoch": 56.96883852691218,
+      "grad_norm": 3.6306028847334093,
+      "learning_rate": 5e-06,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 861144208,
+      "step": 5017
+    },
+    {
+      "epoch": 56.96883852691218,
+      "loss": 0.03210487216711044,
+      "loss_ce": 0.00012245196558069438,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 861144208,
+      "step": 5017
+    },
+    {
+      "epoch": 56.98016997167139,
+      "grad_norm": 3.7858792747750267,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 861315960,
+      "step": 5018
+    },
+    {
+      "epoch": 56.98016997167139,
+      "loss": 0.06369799375534058,
+      "loss_ce": 0.0001451359858037904,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 861315960,
+      "step": 5018
+    },
+    {
+      "epoch": 56.991501416430594,
+      "grad_norm": 3.4509099325672494,
+      "learning_rate": 5e-06,
+      "loss": 0.073,
+      "num_input_tokens_seen": 861487568,
+      "step": 5019
+    },
+    {
+      "epoch": 56.991501416430594,
+      "loss": 0.09958165884017944,
+      "loss_ce": 9.43559207371436e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 861487568,
+      "step": 5019
+    },
+    {
+      "epoch": 56.991501416430594,
+      "loss": 0.09639353305101395,
+      "loss_ce": 0.00011057269148295745,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 861616320,
+      "step": 5019
+    },
+    {
+      "epoch": 57.0028328611898,
+      "grad_norm": 4.231669477564399,
+      "learning_rate": 5e-06,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 861658344,
+      "step": 5020
+    },
+    {
+      "epoch": 57.0028328611898,
+      "loss": 0.06899544596672058,
+      "loss_ce": 5.623399920295924e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 861658344,
+      "step": 5020
+    },
+    {
+      "epoch": 57.01416430594901,
+      "grad_norm": 3.537246798207537,
+      "learning_rate": 5e-06,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 861830260,
+      "step": 5021
+    },
+    {
+      "epoch": 57.01416430594901,
+      "loss": 0.0476437509059906,
+      "loss_ce": 0.0001583982666488737,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 861830260,
+      "step": 5021
+    },
+    {
+      "epoch": 57.02549575070822,
+      "grad_norm": 3.5774645402285126,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 862002040,
+      "step": 5022
+    },
+    {
+      "epoch": 57.02549575070822,
+      "loss": 0.05057121813297272,
+      "loss_ce": 5.6996053899638355e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 862002040,
+      "step": 5022
+    },
+    {
+      "epoch": 57.036827195467424,
+      "grad_norm": 3.257504467543487,
+      "learning_rate": 5e-06,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 862172640,
+      "step": 5023
+    },
+    {
+      "epoch": 57.036827195467424,
+      "loss": 0.07824724167585373,
+      "loss_ce": 0.0003663810493890196,
+      "loss_iou": 0.02001953125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 862172640,
+      "step": 5023
+    },
+    {
+      "epoch": 57.04815864022663,
+      "grad_norm": 3.65946973525709,
+      "learning_rate": 5e-06,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 862344820,
+      "step": 5024
+    },
+    {
+      "epoch": 57.04815864022663,
+      "loss": 0.1127917617559433,
+      "loss_ce": 5.981946742394939e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 862344820,
+      "step": 5024
+    },
+    {
+      "epoch": 57.05949008498584,
+      "grad_norm": 3.803995886183146,
+      "learning_rate": 5e-06,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 862516640,
+      "step": 5025
+    },
+    {
+      "epoch": 57.05949008498584,
+      "loss": 0.05670784413814545,
+      "loss_ce": 8.247532969107851e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 862516640,
+      "step": 5025
+    },
+    {
+      "epoch": 57.07082152974504,
+      "grad_norm": 3.931559119462654,
+      "learning_rate": 5e-06,
+      "loss": 0.0607,
+      "num_input_tokens_seen": 862688216,
+      "step": 5026
+    },
+    {
+      "epoch": 57.07082152974504,
+      "loss": 0.08011346310377121,
+      "loss_ce": 6.5854394051712e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 862688216,
+      "step": 5026
+    },
+    {
+      "epoch": 57.08215297450425,
+      "grad_norm": 3.3828929426001366,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 862860044,
+      "step": 5027
+    },
+    {
+      "epoch": 57.08215297450425,
+      "loss": 0.05361152067780495,
+      "loss_ce": 0.00015998186427168548,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 862860044,
+      "step": 5027
+    },
+    {
+      "epoch": 57.093484419263454,
+      "grad_norm": 3.6592721193873565,
+      "learning_rate": 5e-06,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 863031836,
+      "step": 5028
+    },
+    {
+      "epoch": 57.093484419263454,
+      "loss": 0.03794557601213455,
+      "loss_ce": 8.852001337800175e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 863031836,
+      "step": 5028
+    },
+    {
+      "epoch": 57.10481586402266,
+      "grad_norm": 3.238188279584168,
+      "learning_rate": 5e-06,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 863203528,
+      "step": 5029
+    },
+    {
+      "epoch": 57.10481586402266,
+      "loss": 0.05411809682846069,
+      "loss_ce": 7.146866119001061e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 863203528,
+      "step": 5029
+    },
+    {
+      "epoch": 57.11614730878187,
+      "grad_norm": 2.993312155786794,
+      "learning_rate": 5e-06,
+      "loss": 0.043,
+      "num_input_tokens_seen": 863374556,
+      "step": 5030
+    },
+    {
+      "epoch": 57.11614730878187,
+      "loss": 0.03923165798187256,
+      "loss_ce": 0.0001538988435640931,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 863374556,
+      "step": 5030
+    },
+    {
+      "epoch": 57.12747875354108,
+      "grad_norm": 7.261858447850296,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 863546088,
+      "step": 5031
+    },
+    {
+      "epoch": 57.12747875354108,
+      "loss": 0.08519823849201202,
+      "loss_ce": 5.41940753464587e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 863546088,
+      "step": 5031
+    },
+    {
+      "epoch": 57.138810198300284,
+      "grad_norm": 3.2578107247106916,
+      "learning_rate": 5e-06,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 863715504,
+      "step": 5032
+    },
+    {
+      "epoch": 57.138810198300284,
+      "loss": 0.038077641278505325,
+      "loss_ce": 0.0034401891753077507,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 863715504,
+      "step": 5032
+    },
+    {
+      "epoch": 57.15014164305949,
+      "grad_norm": 3.212333335213238,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 863886188,
+      "step": 5033
+    },
+    {
+      "epoch": 57.15014164305949,
+      "loss": 0.1206144168972969,
+      "loss_ce": 0.00011575574171729386,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 863886188,
+      "step": 5033
+    },
+    {
+      "epoch": 57.1614730878187,
+      "grad_norm": 2.941922916003128,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 864058052,
+      "step": 5034
+    },
+    {
+      "epoch": 57.1614730878187,
+      "loss": 0.03937336057424545,
+      "loss_ce": 0.00011249412636971101,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 864058052,
+      "step": 5034
+    },
+    {
+      "epoch": 57.172804532577906,
+      "grad_norm": 2.6937968117440767,
+      "learning_rate": 5e-06,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 864229764,
+      "step": 5035
+    },
+    {
+      "epoch": 57.172804532577906,
+      "loss": 0.05294235050678253,
+      "loss_ce": 4.012823774246499e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 864229764,
+      "step": 5035
+    },
+    {
+      "epoch": 57.184135977337114,
+      "grad_norm": 3.672482680881072,
+      "learning_rate": 5e-06,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 864401652,
+      "step": 5036
+    },
+    {
+      "epoch": 57.184135977337114,
+      "loss": 0.035760682076215744,
+      "loss_ce": 8.56327751534991e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 864401652,
+      "step": 5036
+    },
+    {
+      "epoch": 57.195467422096314,
+      "grad_norm": 3.7672993398283823,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 864573348,
+      "step": 5037
+    },
+    {
+      "epoch": 57.195467422096314,
+      "loss": 0.04081244766712189,
+      "loss_ce": 5.621928721666336e-05,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 864573348,
+      "step": 5037
+    },
+    {
+      "epoch": 57.20679886685552,
+      "grad_norm": 4.266719969087465,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 864745652,
+      "step": 5038
+    },
+    {
+      "epoch": 57.20679886685552,
+      "loss": 0.10175144672393799,
+      "loss_ce": 0.00012790832261089236,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 864745652,
+      "step": 5038
+    },
+    {
+      "epoch": 57.21813031161473,
+      "grad_norm": 4.058714396900332,
+      "learning_rate": 5e-06,
+      "loss": 0.081,
+      "num_input_tokens_seen": 864916432,
+      "step": 5039
+    },
+    {
+      "epoch": 57.21813031161473,
+      "loss": 0.041860103607177734,
+      "loss_ce": 0.0002188688813475892,
+      "loss_iou": 0.359375,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 864916432,
+      "step": 5039
+    },
+    {
+      "epoch": 57.22946175637394,
+      "grad_norm": 2.9693412464792606,
+      "learning_rate": 5e-06,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 865088232,
+      "step": 5040
+    },
+    {
+      "epoch": 57.22946175637394,
+      "loss": 0.047646526247262955,
+      "loss_ce": 2.3845044779591262e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 865088232,
+      "step": 5040
+    },
+    {
+      "epoch": 57.240793201133144,
+      "grad_norm": 3.1168266226033805,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 865260188,
+      "step": 5041
+    },
+    {
+      "epoch": 57.240793201133144,
+      "loss": 0.03999623283743858,
+      "loss_ce": 0.00012501695891842246,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 865260188,
+      "step": 5041
+    },
+    {
+      "epoch": 57.25212464589235,
+      "grad_norm": 3.4797757072260778,
+      "learning_rate": 5e-06,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 865430652,
+      "step": 5042
+    },
+    {
+      "epoch": 57.25212464589235,
+      "loss": 0.09753690659999847,
+      "loss_ce": 7.90227422839962e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 865430652,
+      "step": 5042
+    },
+    {
+      "epoch": 57.26345609065156,
+      "grad_norm": 4.862666830760718,
+      "learning_rate": 5e-06,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 865602488,
+      "step": 5043
+    },
+    {
+      "epoch": 57.26345609065156,
+      "loss": 0.05271115154027939,
+      "loss_ce": 0.00012936675921082497,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 865602488,
+      "step": 5043
+    },
+    {
+      "epoch": 57.274787535410766,
+      "grad_norm": 4.020972927791236,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 865773428,
+      "step": 5044
+    },
+    {
+      "epoch": 57.274787535410766,
+      "loss": 0.06453670561313629,
+      "loss_ce": 2.254248465760611e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 865773428,
+      "step": 5044
+    },
+    {
+      "epoch": 57.286118980169974,
+      "grad_norm": 6.731758623376648,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 865942904,
+      "step": 5045
+    },
+    {
+      "epoch": 57.286118980169974,
+      "loss": 0.12917166948318481,
+      "loss_ce": 0.00012808524479623884,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 865942904,
+      "step": 5045
+    },
+    {
+      "epoch": 57.29745042492918,
+      "grad_norm": 3.4722130684953743,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 866114776,
+      "step": 5046
+    },
+    {
+      "epoch": 57.29745042492918,
+      "loss": 0.044525355100631714,
+      "loss_ce": 9.175902232527733e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 866114776,
+      "step": 5046
+    },
+    {
+      "epoch": 57.30878186968839,
+      "grad_norm": 3.54394128499371,
+      "learning_rate": 5e-06,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 866286588,
+      "step": 5047
+    },
+    {
+      "epoch": 57.30878186968839,
+      "loss": 0.06828071177005768,
+      "loss_ce": 7.392392581095919e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 866286588,
+      "step": 5047
+    },
+    {
+      "epoch": 57.32011331444759,
+      "grad_norm": 4.8540153175909735,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 866457340,
+      "step": 5048
+    },
+    {
+      "epoch": 57.32011331444759,
+      "loss": 0.038827527314424515,
+      "loss_ce": 7.020324846962467e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 866457340,
+      "step": 5048
+    },
+    {
+      "epoch": 57.331444759206796,
+      "grad_norm": 3.5363898314808067,
+      "learning_rate": 5e-06,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 866627524,
+      "step": 5049
+    },
+    {
+      "epoch": 57.331444759206796,
+      "loss": 0.05021118372678757,
+      "loss_ce": 5.5545671784784645e-05,
+      "loss_iou": 0.5625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 866627524,
+      "step": 5049
+    },
+    {
+      "epoch": 57.342776203966004,
+      "grad_norm": 3.2470222610965322,
+      "learning_rate": 5e-06,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 866799256,
+      "step": 5050
+    },
+    {
+      "epoch": 57.342776203966004,
+      "loss": 0.03605237603187561,
+      "loss_ce": 8.740978228161111e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 866799256,
+      "step": 5050
+    },
+    {
+      "epoch": 57.35410764872521,
+      "grad_norm": 3.4114460330260985,
+      "learning_rate": 5e-06,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 866969368,
+      "step": 5051
+    },
+    {
+      "epoch": 57.35410764872521,
+      "loss": 0.05440719425678253,
+      "loss_ce": 4.01291654270608e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 866969368,
+      "step": 5051
+    },
+    {
+      "epoch": 57.36543909348442,
+      "grad_norm": 5.446001217826163,
+      "learning_rate": 5e-06,
+      "loss": 0.064,
+      "num_input_tokens_seen": 867140612,
+      "step": 5052
+    },
+    {
+      "epoch": 57.36543909348442,
+      "loss": 0.06678560376167297,
+      "loss_ce": 0.00016573088942095637,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 867140612,
+      "step": 5052
+    },
+    {
+      "epoch": 57.376770538243626,
+      "grad_norm": 4.087480069299749,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 867312368,
+      "step": 5053
+    },
+    {
+      "epoch": 57.376770538243626,
+      "loss": 0.08000503480434418,
+      "loss_ce": 6.423363811336458e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 867312368,
+      "step": 5053
+    },
+    {
+      "epoch": 57.388101983002834,
+      "grad_norm": 3.598545889710196,
+      "learning_rate": 5e-06,
+      "loss": 0.0918,
+      "num_input_tokens_seen": 867482772,
+      "step": 5054
+    },
+    {
+      "epoch": 57.388101983002834,
+      "loss": 0.13171756267547607,
+      "loss_ce": 0.00015627549146302044,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 867482772,
+      "step": 5054
+    },
+    {
+      "epoch": 57.39943342776204,
+      "grad_norm": 3.1050709183802456,
+      "learning_rate": 5e-06,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 867654648,
+      "step": 5055
+    },
+    {
+      "epoch": 57.39943342776204,
+      "loss": 0.046764541417360306,
+      "loss_ce": 0.00013367910287342966,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 867654648,
+      "step": 5055
+    },
+    {
+      "epoch": 57.41076487252125,
+      "grad_norm": 35.77623341430084,
+      "learning_rate": 5e-06,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 867823208,
+      "step": 5056
+    },
+    {
+      "epoch": 57.41076487252125,
+      "loss": 0.03876463323831558,
+      "loss_ce": 6.834713713033125e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 867823208,
+      "step": 5056
+    },
+    {
+      "epoch": 57.422096317280456,
+      "grad_norm": 3.996609357243811,
+      "learning_rate": 5e-06,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 867994520,
+      "step": 5057
+    },
+    {
+      "epoch": 57.422096317280456,
+      "loss": 0.10653401911258698,
+      "loss_ce": 2.766936995612923e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 867994520,
+      "step": 5057
+    },
+    {
+      "epoch": 57.43342776203966,
+      "grad_norm": 3.6658206174973396,
+      "learning_rate": 5e-06,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 868165596,
+      "step": 5058
+    },
+    {
+      "epoch": 57.43342776203966,
+      "loss": 0.1198868453502655,
+      "loss_ce": 9.008888446260244e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 868165596,
+      "step": 5058
+    },
+    {
+      "epoch": 57.444759206798864,
+      "grad_norm": 8.13883596178892,
+      "learning_rate": 5e-06,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 868337372,
+      "step": 5059
+    },
+    {
+      "epoch": 57.444759206798864,
+      "loss": 0.04465852305293083,
+      "loss_ce": 0.0001333790278295055,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 868337372,
+      "step": 5059
+    },
+    {
+      "epoch": 57.45609065155807,
+      "grad_norm": 3.621310775553928,
+      "learning_rate": 5e-06,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 868509060,
+      "step": 5060
+    },
+    {
+      "epoch": 57.45609065155807,
+      "loss": 0.05548835173249245,
+      "loss_ce": 0.0002973129157908261,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 868509060,
+      "step": 5060
+    },
+    {
+      "epoch": 57.46742209631728,
+      "grad_norm": 3.7753393288950763,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 868681164,
+      "step": 5061
+    },
+    {
+      "epoch": 57.46742209631728,
+      "loss": 0.03601599112153053,
+      "loss_ce": 3.576567905838601e-05,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 868681164,
+      "step": 5061
+    },
+    {
+      "epoch": 57.478753541076486,
+      "grad_norm": 4.391557164235902,
+      "learning_rate": 5e-06,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 868852404,
+      "step": 5062
+    },
+    {
+      "epoch": 57.478753541076486,
+      "loss": 0.050505250692367554,
+      "loss_ce": 0.0001359869638690725,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 868852404,
+      "step": 5062
+    },
+    {
+      "epoch": 57.49008498583569,
+      "grad_norm": 3.9887814992177333,
+      "learning_rate": 5e-06,
+      "loss": 0.0824,
+      "num_input_tokens_seen": 869024084,
+      "step": 5063
+    },
+    {
+      "epoch": 57.49008498583569,
+      "loss": 0.06514575332403183,
+      "loss_ce": 5.17590124218259e-05,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 869024084,
+      "step": 5063
+    },
+    {
+      "epoch": 57.5014164305949,
+      "grad_norm": 4.071979292489725,
+      "learning_rate": 5e-06,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 869194608,
+      "step": 5064
+    },
+    {
+      "epoch": 57.5014164305949,
+      "loss": 0.05516326054930687,
+      "loss_ce": 7.903205550974235e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 869194608,
+      "step": 5064
+    },
+    {
+      "epoch": 57.51274787535411,
+      "grad_norm": 4.116623607204772,
+      "learning_rate": 5e-06,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 869366760,
+      "step": 5065
+    },
+    {
+      "epoch": 57.51274787535411,
+      "loss": 0.10826753079891205,
+      "loss_ce": 5.219535159994848e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 869366760,
+      "step": 5065
+    },
+    {
+      "epoch": 57.524079320113316,
+      "grad_norm": 2.839576688607693,
+      "learning_rate": 5e-06,
+      "loss": 0.0651,
+      "num_input_tokens_seen": 869538644,
+      "step": 5066
+    },
+    {
+      "epoch": 57.524079320113316,
+      "loss": 0.06913868337869644,
+      "loss_ce": 6.214836321305484e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 869538644,
+      "step": 5066
+    },
+    {
+      "epoch": 57.53541076487252,
+      "grad_norm": 3.0964737999170335,
+      "learning_rate": 5e-06,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 869710768,
+      "step": 5067
+    },
+    {
+      "epoch": 57.53541076487252,
+      "loss": 0.0796612873673439,
+      "loss_ce": 0.00010196147923124954,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 869710768,
+      "step": 5067
+    },
+    {
+      "epoch": 57.54674220963173,
+      "grad_norm": 3.1266950009736094,
+      "learning_rate": 5e-06,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 869882580,
+      "step": 5068
+    },
+    {
+      "epoch": 57.54674220963173,
+      "loss": 0.04041702300310135,
+      "loss_ce": 0.0001033049775287509,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 869882580,
+      "step": 5068
+    },
+    {
+      "epoch": 57.55807365439094,
+      "grad_norm": 3.3143194949148764,
+      "learning_rate": 5e-06,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 870051768,
+      "step": 5069
+    },
+    {
+      "epoch": 57.55807365439094,
+      "loss": 0.035957276821136475,
+      "loss_ce": 3.808567998930812e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 870051768,
+      "step": 5069
+    },
+    {
+      "epoch": 57.56940509915014,
+      "grad_norm": 4.617062231752367,
+      "learning_rate": 5e-06,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 870223740,
+      "step": 5070
+    },
+    {
+      "epoch": 57.56940509915014,
+      "loss": 0.10611091554164886,
+      "loss_ce": 7.759387517580763e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 870223740,
+      "step": 5070
+    },
+    {
+      "epoch": 57.580736543909346,
+      "grad_norm": 3.9206413164352734,
+      "learning_rate": 5e-06,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 870395288,
+      "step": 5071
+    },
+    {
+      "epoch": 57.580736543909346,
+      "loss": 0.04212199151515961,
+      "loss_ce": 2.29953984671738e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 870395288,
+      "step": 5071
+    },
+    {
+      "epoch": 57.59206798866855,
+      "grad_norm": 3.72554947154378,
+      "learning_rate": 5e-06,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 870564132,
+      "step": 5072
+    },
+    {
+      "epoch": 57.59206798866855,
+      "loss": 0.05641186982393265,
+      "loss_ce": 9.168143151327968e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 870564132,
+      "step": 5072
+    },
+    {
+      "epoch": 57.60339943342776,
+      "grad_norm": 3.217229627382938,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 870735996,
+      "step": 5073
+    },
+    {
+      "epoch": 57.60339943342776,
+      "loss": 0.10920092463493347,
+      "loss_ce": 0.00010058139741886407,
+      "loss_iou": 0.5,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 870735996,
+      "step": 5073
+    },
+    {
+      "epoch": 57.61473087818697,
+      "grad_norm": 3.198346507225924,
+      "learning_rate": 5e-06,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 870906760,
+      "step": 5074
+    },
+    {
+      "epoch": 57.61473087818697,
+      "loss": 0.06260886788368225,
+      "loss_ce": 0.00015464352327398956,
+      "loss_iou": 0.0703125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 870906760,
+      "step": 5074
+    },
+    {
+      "epoch": 57.626062322946176,
+      "grad_norm": 3.597125684633523,
+      "learning_rate": 5e-06,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 871078044,
+      "step": 5075
+    },
+    {
+      "epoch": 57.626062322946176,
+      "loss": 0.08610302209854126,
+      "loss_ce": 7.396174623863772e-05,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 871078044,
+      "step": 5075
+    },
+    {
+      "epoch": 57.63739376770538,
+      "grad_norm": 3.3973737608801913,
+      "learning_rate": 5e-06,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 871249904,
+      "step": 5076
+    },
+    {
+      "epoch": 57.63739376770538,
+      "loss": 0.10708586871623993,
+      "loss_ce": 7.597856892971322e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 871249904,
+      "step": 5076
+    },
+    {
+      "epoch": 57.64872521246459,
+      "grad_norm": 14.15498967655936,
+      "learning_rate": 5e-06,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 871421708,
+      "step": 5077
+    },
+    {
+      "epoch": 57.64872521246459,
+      "loss": 0.12140925973653793,
+      "loss_ce": 0.0018719020299613476,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 871421708,
+      "step": 5077
+    },
+    {
+      "epoch": 57.6600566572238,
+      "grad_norm": 2.3009264601009862,
+      "learning_rate": 5e-06,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 871593536,
+      "step": 5078
+    },
+    {
+      "epoch": 57.6600566572238,
+      "loss": 0.03933015465736389,
+      "loss_ce": 6.166232196846977e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 871593536,
+      "step": 5078
+    },
+    {
+      "epoch": 57.671388101983005,
+      "grad_norm": 2.719883639560166,
+      "learning_rate": 5e-06,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 871765224,
+      "step": 5079
+    },
+    {
+      "epoch": 57.671388101983005,
+      "loss": 0.02782721072435379,
+      "loss_ce": 7.147227006498724e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 871765224,
+      "step": 5079
+    },
+    {
+      "epoch": 57.68271954674221,
+      "grad_norm": 5.392435032151135,
+      "learning_rate": 5e-06,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 871935504,
+      "step": 5080
+    },
+    {
+      "epoch": 57.68271954674221,
+      "loss": 0.12589111924171448,
+      "loss_ce": 0.0001281767472391948,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 871935504,
+      "step": 5080
+    },
+    {
+      "epoch": 57.69405099150141,
+      "grad_norm": 3.6050758135481114,
+      "learning_rate": 5e-06,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 872106924,
+      "step": 5081
+    },
+    {
+      "epoch": 57.69405099150141,
+      "loss": 0.06396076083183289,
+      "loss_ce": 8.746745879761875e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 872106924,
+      "step": 5081
+    },
+    {
+      "epoch": 57.70538243626062,
+      "grad_norm": 4.204151675840715,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 872278996,
+      "step": 5082
+    },
+    {
+      "epoch": 57.70538243626062,
+      "loss": 0.045700009912252426,
+      "loss_ce": 5.3342999308370054e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 872278996,
+      "step": 5082
+    },
+    {
+      "epoch": 57.71671388101983,
+      "grad_norm": 4.280723193083514,
+      "learning_rate": 5e-06,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 872449260,
+      "step": 5083
+    },
+    {
+      "epoch": 57.71671388101983,
+      "loss": 0.04482072964310646,
+      "loss_ce": 8.196244016289711e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 872449260,
+      "step": 5083
+    },
+    {
+      "epoch": 57.728045325779036,
+      "grad_norm": 3.682391061868857,
+      "learning_rate": 5e-06,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 872620680,
+      "step": 5084
+    },
+    {
+      "epoch": 57.728045325779036,
+      "loss": 0.04363110288977623,
+      "loss_ce": 5.963179864920676e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 872620680,
+      "step": 5084
+    },
+    {
+      "epoch": 57.73937677053824,
+      "grad_norm": 3.2672101962535125,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 872792372,
+      "step": 5085
+    },
+    {
+      "epoch": 57.73937677053824,
+      "loss": 0.04827268421649933,
+      "loss_ce": 0.00010068750270875171,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 872792372,
+      "step": 5085
+    },
+    {
+      "epoch": 57.75070821529745,
+      "grad_norm": 3.8539558843221764,
+      "learning_rate": 5e-06,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 872964324,
+      "step": 5086
+    },
+    {
+      "epoch": 57.75070821529745,
+      "loss": 0.07103857398033142,
+      "loss_ce": 3.9427824958693236e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 872964324,
+      "step": 5086
+    },
+    {
+      "epoch": 57.76203966005666,
+      "grad_norm": 3.6034569761092516,
+      "learning_rate": 5e-06,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 873136036,
+      "step": 5087
+    },
+    {
+      "epoch": 57.76203966005666,
+      "loss": 0.05179614573717117,
+      "loss_ce": 2.3074911950971e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 873136036,
+      "step": 5087
+    },
+    {
+      "epoch": 57.773371104815865,
+      "grad_norm": 6.347541657907292,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 873307896,
+      "step": 5088
+    },
+    {
+      "epoch": 57.773371104815865,
+      "loss": 0.06276465207338333,
+      "loss_ce": 0.0008902625413611531,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 873307896,
+      "step": 5088
+    },
+    {
+      "epoch": 57.78470254957507,
+      "grad_norm": 3.2255177959465247,
+      "learning_rate": 5e-06,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 873479704,
+      "step": 5089
+    },
+    {
+      "epoch": 57.78470254957507,
+      "loss": 0.039910003542900085,
+      "loss_ce": 5.4046304285293445e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 873479704,
+      "step": 5089
+    },
+    {
+      "epoch": 57.79603399433428,
+      "grad_norm": 5.388892145701267,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 873651572,
+      "step": 5090
+    },
+    {
+      "epoch": 57.79603399433428,
+      "loss": 0.07917419075965881,
+      "loss_ce": 2.685383878997527e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 873651572,
+      "step": 5090
+    },
+    {
+      "epoch": 57.80736543909349,
+      "grad_norm": 4.090374408229897,
+      "learning_rate": 5e-06,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 873822572,
+      "step": 5091
+    },
+    {
+      "epoch": 57.80736543909349,
+      "loss": 0.0569564551115036,
+      "loss_ce": 0.00010220690455753356,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 873822572,
+      "step": 5091
+    },
+    {
+      "epoch": 57.81869688385269,
+      "grad_norm": 3.07757802750749,
+      "learning_rate": 5e-06,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 873993632,
+      "step": 5092
+    },
+    {
+      "epoch": 57.81869688385269,
+      "loss": 0.08146277070045471,
+      "loss_ce": 0.004558474291115999,
+      "loss_iou": 0.53125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 873993632,
+      "step": 5092
+    },
+    {
+      "epoch": 57.830028328611895,
+      "grad_norm": 3.6462578497896283,
+      "learning_rate": 5e-06,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 874165400,
+      "step": 5093
+    },
+    {
+      "epoch": 57.830028328611895,
+      "loss": 0.04115299880504608,
+      "loss_ce": 0.00016789112123660743,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 874165400,
+      "step": 5093
+    },
+    {
+      "epoch": 57.8413597733711,
+      "grad_norm": 3.5431815528857205,
+      "learning_rate": 5e-06,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 874336344,
+      "step": 5094
+    },
+    {
+      "epoch": 57.8413597733711,
+      "loss": 0.06220860034227371,
+      "loss_ce": 0.00012058853462804109,
+      "loss_iou": 0.3125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 874336344,
+      "step": 5094
+    },
+    {
+      "epoch": 57.85269121813031,
+      "grad_norm": 3.541857387568089,
+      "learning_rate": 5e-06,
+      "loss": 0.0773,
+      "num_input_tokens_seen": 874508336,
+      "step": 5095
+    },
+    {
+      "epoch": 57.85269121813031,
+      "loss": 0.11195708811283112,
+      "loss_ce": 3.386967728147283e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 874508336,
+      "step": 5095
+    },
+    {
+      "epoch": 57.86402266288952,
+      "grad_norm": 4.784857267031445,
+      "learning_rate": 5e-06,
+      "loss": 0.063,
+      "num_input_tokens_seen": 874679984,
+      "step": 5096
+    },
+    {
+      "epoch": 57.86402266288952,
+      "loss": 0.03885423392057419,
+      "loss_ce": 6.639276398345828e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 874679984,
+      "step": 5096
+    },
+    {
+      "epoch": 57.875354107648725,
+      "grad_norm": 3.4170915279287724,
+      "learning_rate": 5e-06,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 874851824,
+      "step": 5097
+    },
+    {
+      "epoch": 57.875354107648725,
+      "loss": 0.037616003304719925,
+      "loss_ce": 0.00014041727990843356,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 874851824,
+      "step": 5097
+    },
+    {
+      "epoch": 57.88668555240793,
+      "grad_norm": 4.184896382903822,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 875020924,
+      "step": 5098
+    },
+    {
+      "epoch": 57.88668555240793,
+      "loss": 0.04111591726541519,
+      "loss_ce": 8.503426215611398e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 875020924,
+      "step": 5098
+    },
+    {
+      "epoch": 57.89801699716714,
+      "grad_norm": 3.352523644292825,
+      "learning_rate": 5e-06,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 875192168,
+      "step": 5099
+    },
+    {
+      "epoch": 57.89801699716714,
+      "loss": 0.05032181739807129,
+      "loss_ce": 8.988452464109287e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 875192168,
+      "step": 5099
+    },
+    {
+      "epoch": 57.90934844192635,
+      "grad_norm": 8.64567669140172,
+      "learning_rate": 5e-06,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 875362972,
+      "step": 5100
+    },
+    {
+      "epoch": 57.90934844192635,
+      "loss": 0.05239830166101456,
+      "loss_ce": 3.0138548027025536e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 875362972,
+      "step": 5100
+    },
+    {
+      "epoch": 57.920679886685555,
+      "grad_norm": 11.855779680858916,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 875532388,
+      "step": 5101
+    },
+    {
+      "epoch": 57.920679886685555,
+      "loss": 0.05897153541445732,
+      "loss_ce": 7.261127757374197e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 875532388,
+      "step": 5101
+    },
+    {
+      "epoch": 57.93201133144476,
+      "grad_norm": 4.5639913925062405,
+      "learning_rate": 5e-06,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 875704220,
+      "step": 5102
+    },
+    {
+      "epoch": 57.93201133144476,
+      "loss": 0.04093395918607712,
+      "loss_ce": 5.5663207604084164e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 875704220,
+      "step": 5102
+    },
+    {
+      "epoch": 57.94334277620396,
+      "grad_norm": 3.467964592377589,
+      "learning_rate": 5e-06,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 875876036,
+      "step": 5103
+    },
+    {
+      "epoch": 57.94334277620396,
+      "loss": 0.09246046841144562,
+      "loss_ce": 3.798675970756449e-05,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 875876036,
+      "step": 5103
+    },
+    {
+      "epoch": 57.95467422096317,
+      "grad_norm": 4.771503567332697,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 876047956,
+      "step": 5104
+    },
+    {
+      "epoch": 57.95467422096317,
+      "loss": 0.1134958416223526,
+      "loss_ce": 3.148883115500212e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 876047956,
+      "step": 5104
+    },
+    {
+      "epoch": 57.96600566572238,
+      "grad_norm": 4.364680975526661,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 876218100,
+      "step": 5105
+    },
+    {
+      "epoch": 57.96600566572238,
+      "loss": 0.043542273342609406,
+      "loss_ce": 6.998371100053191e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 876218100,
+      "step": 5105
+    },
+    {
+      "epoch": 57.977337110481585,
+      "grad_norm": 4.424021672230856,
+      "learning_rate": 5e-06,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 876389104,
+      "step": 5106
+    },
+    {
+      "epoch": 57.977337110481585,
+      "loss": 0.05820219963788986,
+      "loss_ce": 3.569648106349632e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 876389104,
+      "step": 5106
+    },
+    {
+      "epoch": 57.98866855524079,
+      "grad_norm": 4.141340864324125,
+      "learning_rate": 5e-06,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 876561396,
+      "step": 5107
+    },
+    {
+      "epoch": 57.98866855524079,
+      "loss": 0.03864654526114464,
+      "loss_ce": 4.1810115362750366e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 876561396,
+      "step": 5107
+    },
+    {
+      "epoch": 58.0,
+      "grad_norm": 4.19898040942282,
+      "learning_rate": 5e-06,
+      "loss": 0.064,
+      "num_input_tokens_seen": 876732192,
+      "step": 5108
+    },
+    {
+      "epoch": 58.0,
+      "loss": 0.03873483091592789,
+      "loss_ce": 6.905860936967656e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 876732192,
+      "step": 5108
+    },
+    {
+      "epoch": 58.01133144475921,
+      "grad_norm": 3.199857446432168,
+      "learning_rate": 5e-06,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 876903308,
+      "step": 5109
+    },
+    {
+      "epoch": 58.01133144475921,
+      "loss": 0.03641393035650253,
+      "loss_ce": 6.749634485458955e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 876903308,
+      "step": 5109
+    },
+    {
+      "epoch": 58.022662889518415,
+      "grad_norm": 3.3669725716022456,
+      "learning_rate": 5e-06,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 877075500,
+      "step": 5110
+    },
+    {
+      "epoch": 58.022662889518415,
+      "loss": 0.04136184975504875,
+      "loss_ce": 0.00034622280509211123,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 877075500,
+      "step": 5110
+    },
+    {
+      "epoch": 58.03399433427762,
+      "grad_norm": 2.8820647740383163,
+      "learning_rate": 5e-06,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 877247104,
+      "step": 5111
+    },
+    {
+      "epoch": 58.03399433427762,
+      "loss": 0.025820832699537277,
+      "loss_ce": 6.399747508112341e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 877247104,
+      "step": 5111
+    },
+    {
+      "epoch": 58.04532577903683,
+      "grad_norm": 2.3984366774928483,
+      "learning_rate": 5e-06,
+      "loss": 0.0742,
+      "num_input_tokens_seen": 877418828,
+      "step": 5112
+    },
+    {
+      "epoch": 58.04532577903683,
+      "loss": 0.04250626638531685,
+      "loss_ce": 0.00010208995081484318,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 877418828,
+      "step": 5112
+    },
+    {
+      "epoch": 58.05665722379604,
+      "grad_norm": 2.885038635840427,
+      "learning_rate": 5e-06,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 877591060,
+      "step": 5113
+    },
+    {
+      "epoch": 58.05665722379604,
+      "loss": 0.06191824749112129,
+      "loss_ce": 3.622656367952004e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 877591060,
+      "step": 5113
+    },
+    {
+      "epoch": 58.06798866855524,
+      "grad_norm": 11.275130982169811,
+      "learning_rate": 5e-06,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 877761884,
+      "step": 5114
+    },
+    {
+      "epoch": 58.06798866855524,
+      "loss": 0.07542091608047485,
+      "loss_ce": 2.7240759663982317e-05,
+      "loss_iou": 0.140625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 877761884,
+      "step": 5114
+    },
+    {
+      "epoch": 58.079320113314445,
+      "grad_norm": 3.8428706585620542,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 877934304,
+      "step": 5115
+    },
+    {
+      "epoch": 58.079320113314445,
+      "loss": 0.05714516341686249,
+      "loss_ce": 3.91458670492284e-05,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 877934304,
+      "step": 5115
+    },
+    {
+      "epoch": 58.09065155807365,
+      "grad_norm": 4.277177756148496,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 878105080,
+      "step": 5116
+    },
+    {
+      "epoch": 58.09065155807365,
+      "loss": 0.1516241729259491,
+      "loss_ce": 0.0002112043439410627,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 878105080,
+      "step": 5116
+    },
+    {
+      "epoch": 58.10198300283286,
+      "grad_norm": 3.4166333894226355,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 878276888,
+      "step": 5117
+    },
+    {
+      "epoch": 58.10198300283286,
+      "loss": 0.033880945295095444,
+      "loss_ce": 3.695099803735502e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 878276888,
+      "step": 5117
+    },
+    {
+      "epoch": 58.11331444759207,
+      "grad_norm": 3.9675535702788474,
+      "learning_rate": 5e-06,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 878448492,
+      "step": 5118
+    },
+    {
+      "epoch": 58.11331444759207,
+      "loss": 0.09718040376901627,
+      "loss_ce": 7.346836355281994e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 878448492,
+      "step": 5118
+    },
+    {
+      "epoch": 58.124645892351275,
+      "grad_norm": 3.089661459899956,
+      "learning_rate": 5e-06,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 878620404,
+      "step": 5119
+    },
+    {
+      "epoch": 58.124645892351275,
+      "loss": 0.043347954750061035,
+      "loss_ce": 4.3514530261745676e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 878620404,
+      "step": 5119
+    },
+    {
+      "epoch": 58.13597733711048,
+      "grad_norm": 2.9783339010224408,
+      "learning_rate": 5e-06,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 878792252,
+      "step": 5120
+    },
+    {
+      "epoch": 58.13597733711048,
+      "loss": 0.057560212910175323,
+      "loss_ce": 4.98374029120896e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 878792252,
+      "step": 5120
+    },
+    {
+      "epoch": 58.14730878186969,
+      "grad_norm": 3.6784884208077897,
+      "learning_rate": 5e-06,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 878962572,
+      "step": 5121
+    },
+    {
+      "epoch": 58.14730878186969,
+      "loss": 0.040791139006614685,
+      "loss_ce": 1.9656603399198502e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 878962572,
+      "step": 5121
+    },
+    {
+      "epoch": 58.1586402266289,
+      "grad_norm": 4.574891394109308,
+      "learning_rate": 5e-06,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 879134364,
+      "step": 5122
+    },
+    {
+      "epoch": 58.1586402266289,
+      "loss": 0.048217467963695526,
+      "loss_ce": 1.49508414324373e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 879134364,
+      "step": 5122
+    },
+    {
+      "epoch": 58.169971671388105,
+      "grad_norm": 3.5322630962007424,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 879305944,
+      "step": 5123
+    },
+    {
+      "epoch": 58.169971671388105,
+      "loss": 0.04073029011487961,
+      "loss_ce": 1.984263872145675e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 879305944,
+      "step": 5123
+    },
+    {
+      "epoch": 58.18130311614731,
+      "grad_norm": 3.9595663092352056,
+      "learning_rate": 5e-06,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 879478580,
+      "step": 5124
+    },
+    {
+      "epoch": 58.18130311614731,
+      "loss": 0.08592826128005981,
+      "loss_ce": 8.231262472691014e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 879478580,
+      "step": 5124
+    },
+    {
+      "epoch": 58.19263456090651,
+      "grad_norm": 3.6069704515292176,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 879647576,
+      "step": 5125
+    },
+    {
+      "epoch": 58.19263456090651,
+      "loss": 0.0516824796795845,
+      "loss_ce": 7.725724572082981e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 879647576,
+      "step": 5125
+    },
+    {
+      "epoch": 58.20396600566572,
+      "grad_norm": 3.045683902802767,
+      "learning_rate": 5e-06,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 879819492,
+      "step": 5126
+    },
+    {
+      "epoch": 58.20396600566572,
+      "loss": 0.05950441583991051,
+      "loss_ce": 0.00011721032205969095,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 879819492,
+      "step": 5126
+    },
+    {
+      "epoch": 58.21529745042493,
+      "grad_norm": 2.6772214076046112,
+      "learning_rate": 5e-06,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 879990924,
+      "step": 5127
+    },
+    {
+      "epoch": 58.21529745042493,
+      "loss": 0.03356531262397766,
+      "loss_ce": 4.93829429615289e-05,
+      "loss_iou": 0.05126953125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 879990924,
+      "step": 5127
+    },
+    {
+      "epoch": 58.226628895184135,
+      "grad_norm": 2.6516104702028374,
+      "learning_rate": 5e-06,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 880163040,
+      "step": 5128
+    },
+    {
+      "epoch": 58.226628895184135,
+      "loss": 0.03838310390710831,
+      "loss_ce": 6.828515324741602e-05,
+      "loss_iou": 0.033203125,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 880163040,
+      "step": 5128
+    },
+    {
+      "epoch": 58.23796033994334,
+      "grad_norm": 3.4412722254487997,
+      "learning_rate": 5e-06,
+      "loss": 0.076,
+      "num_input_tokens_seen": 880334264,
+      "step": 5129
+    },
+    {
+      "epoch": 58.23796033994334,
+      "loss": 0.03586754947900772,
+      "loss_ce": 0.00010094626486534253,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 880334264,
+      "step": 5129
+    },
+    {
+      "epoch": 58.24929178470255,
+      "grad_norm": 4.118062442913695,
+      "learning_rate": 5e-06,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 880505100,
+      "step": 5130
+    },
+    {
+      "epoch": 58.24929178470255,
+      "loss": 0.05262747406959534,
+      "loss_ce": 0.00010672125790733844,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 880505100,
+      "step": 5130
+    },
+    {
+      "epoch": 58.26062322946176,
+      "grad_norm": 3.615844659676223,
+      "learning_rate": 5e-06,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 880675108,
+      "step": 5131
+    },
+    {
+      "epoch": 58.26062322946176,
+      "loss": 0.09601166099309921,
+      "loss_ce": 3.38781246682629e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 880675108,
+      "step": 5131
+    },
+    {
+      "epoch": 58.271954674220964,
+      "grad_norm": 4.891314659232577,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 880846780,
+      "step": 5132
+    },
+    {
+      "epoch": 58.271954674220964,
+      "loss": 0.10980711132287979,
+      "loss_ce": 2.01282455236651e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 880846780,
+      "step": 5132
+    },
+    {
+      "epoch": 58.28328611898017,
+      "grad_norm": 2.8925024487179694,
+      "learning_rate": 5e-06,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 881017888,
+      "step": 5133
+    },
+    {
+      "epoch": 58.28328611898017,
+      "loss": 0.10723724216222763,
+      "loss_ce": 7.476101745851338e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 881017888,
+      "step": 5133
+    },
+    {
+      "epoch": 58.29461756373938,
+      "grad_norm": 3.2468648900083186,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 881188444,
+      "step": 5134
+    },
+    {
+      "epoch": 58.29461756373938,
+      "loss": 0.09116239845752716,
+      "loss_ce": 6.743149424437433e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 881188444,
+      "step": 5134
+    },
+    {
+      "epoch": 58.30594900849859,
+      "grad_norm": 3.7878268786147817,
+      "learning_rate": 5e-06,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 881360560,
+      "step": 5135
+    },
+    {
+      "epoch": 58.30594900849859,
+      "loss": 0.0402548685669899,
+      "loss_ce": 6.321677938103676e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 881360560,
+      "step": 5135
+    },
+    {
+      "epoch": 58.31728045325779,
+      "grad_norm": 3.3404631393697835,
+      "learning_rate": 5e-06,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 881531232,
+      "step": 5136
+    },
+    {
+      "epoch": 58.31728045325779,
+      "loss": 0.042157139629125595,
+      "loss_ce": 4.288295895094052e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 881531232,
+      "step": 5136
+    },
+    {
+      "epoch": 58.328611898016995,
+      "grad_norm": 3.090252804263699,
+      "learning_rate": 5e-06,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 881702996,
+      "step": 5137
+    },
+    {
+      "epoch": 58.328611898016995,
+      "loss": 0.05525463446974754,
+      "loss_ce": 3.307961014797911e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 881702996,
+      "step": 5137
+    },
+    {
+      "epoch": 58.3399433427762,
+      "grad_norm": 3.604927215276543,
+      "learning_rate": 5e-06,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 881874200,
+      "step": 5138
+    },
+    {
+      "epoch": 58.3399433427762,
+      "loss": 0.0328078530728817,
+      "loss_ce": 6.249104626476765e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 881874200,
+      "step": 5138
+    },
+    {
+      "epoch": 58.35127478753541,
+      "grad_norm": 4.0498817920183345,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 882046348,
+      "step": 5139
+    },
+    {
+      "epoch": 58.35127478753541,
+      "loss": 0.05541135370731354,
+      "loss_ce": 9.824558947002515e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 882046348,
+      "step": 5139
+    },
+    {
+      "epoch": 58.36260623229462,
+      "grad_norm": 3.263545120224719,
+      "learning_rate": 5e-06,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 882216184,
+      "step": 5140
+    },
+    {
+      "epoch": 58.36260623229462,
+      "loss": 0.06498118489980698,
+      "loss_ce": 5.5038894060999155e-05,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 882216184,
+      "step": 5140
+    },
+    {
+      "epoch": 58.373937677053824,
+      "grad_norm": 3.594097870683096,
+      "learning_rate": 5e-06,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 882387708,
+      "step": 5141
+    },
+    {
+      "epoch": 58.373937677053824,
+      "loss": 0.05924582853913307,
+      "loss_ce": 7.224692672025412e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 882387708,
+      "step": 5141
+    },
+    {
+      "epoch": 58.38526912181303,
+      "grad_norm": 3.0042299427516297,
+      "learning_rate": 5e-06,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 882559580,
+      "step": 5142
+    },
+    {
+      "epoch": 58.38526912181303,
+      "loss": 0.03157265484333038,
+      "loss_ce": 4.799876478500664e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 882559580,
+      "step": 5142
+    },
+    {
+      "epoch": 58.39660056657224,
+      "grad_norm": 3.41089944767485,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 882731532,
+      "step": 5143
+    },
+    {
+      "epoch": 58.39660056657224,
+      "loss": 0.11940383911132812,
+      "loss_ce": 3.434465907048434e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 882731532,
+      "step": 5143
+    },
+    {
+      "epoch": 58.40793201133145,
+      "grad_norm": 2.842345429791182,
+      "learning_rate": 5e-06,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 882902512,
+      "step": 5144
+    },
+    {
+      "epoch": 58.40793201133145,
+      "loss": 0.10355561971664429,
+      "loss_ce": 8.57689228723757e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 882902512,
+      "step": 5144
+    },
+    {
+      "epoch": 58.419263456090654,
+      "grad_norm": 2.5457588820237618,
+      "learning_rate": 5e-06,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 883072764,
+      "step": 5145
+    },
+    {
+      "epoch": 58.419263456090654,
+      "loss": 0.10356974601745605,
+      "loss_ce": 3.8865608075866476e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 883072764,
+      "step": 5145
+    },
+    {
+      "epoch": 58.43059490084986,
+      "grad_norm": 2.2274371942951032,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 883244632,
+      "step": 5146
+    },
+    {
+      "epoch": 58.43059490084986,
+      "loss": 0.02465543895959854,
+      "loss_ce": 5.82701395615004e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 883244632,
+      "step": 5146
+    },
+    {
+      "epoch": 58.44192634560906,
+      "grad_norm": 2.425816204474589,
+      "learning_rate": 5e-06,
+      "loss": 0.064,
+      "num_input_tokens_seen": 883416764,
+      "step": 5147
+    },
+    {
+      "epoch": 58.44192634560906,
+      "loss": 0.06607544422149658,
+      "loss_ce": 7.355008710874245e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 883416764,
+      "step": 5147
+    },
+    {
+      "epoch": 58.45325779036827,
+      "grad_norm": 3.0425603181572747,
+      "learning_rate": 5e-06,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 883588312,
+      "step": 5148
+    },
+    {
+      "epoch": 58.45325779036827,
+      "loss": 0.05393122881650925,
+      "loss_ce": 6.770497566321865e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 883588312,
+      "step": 5148
+    },
+    {
+      "epoch": 58.46458923512748,
+      "grad_norm": 3.0562728917512296,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 883759892,
+      "step": 5149
+    },
+    {
+      "epoch": 58.46458923512748,
+      "loss": 0.06445635855197906,
+      "loss_ce": 1.8499664292903617e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 883759892,
+      "step": 5149
+    },
+    {
+      "epoch": 58.475920679886684,
+      "grad_norm": 3.227847404238047,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 883931692,
+      "step": 5150
+    },
+    {
+      "epoch": 58.475920679886684,
+      "loss": 0.07732360810041428,
+      "loss_ce": 0.00015991358668543398,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 883931692,
+      "step": 5150
+    },
+    {
+      "epoch": 58.48725212464589,
+      "grad_norm": 3.377039550405065,
+      "learning_rate": 5e-06,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 884102024,
+      "step": 5151
+    },
+    {
+      "epoch": 58.48725212464589,
+      "loss": 0.06637821346521378,
+      "loss_ce": 7.877300959080458e-05,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 884102024,
+      "step": 5151
+    },
+    {
+      "epoch": 58.4985835694051,
+      "grad_norm": 4.0823108696331865,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 884272964,
+      "step": 5152
+    },
+    {
+      "epoch": 58.4985835694051,
+      "loss": 0.03338304162025452,
+      "loss_ce": 4.2590116208884865e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 884272964,
+      "step": 5152
+    },
+    {
+      "epoch": 58.50991501416431,
+      "grad_norm": 3.5536594124255125,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 884444676,
+      "step": 5153
+    },
+    {
+      "epoch": 58.50991501416431,
+      "loss": 0.04385187476873398,
+      "loss_ce": 7.440572517225519e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 884444676,
+      "step": 5153
+    },
+    {
+      "epoch": 58.521246458923514,
+      "grad_norm": 3.993798626008827,
+      "learning_rate": 5e-06,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 884614212,
+      "step": 5154
+    },
+    {
+      "epoch": 58.521246458923514,
+      "loss": 0.13052082061767578,
+      "loss_ce": 5.818325371365063e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 884614212,
+      "step": 5154
+    },
+    {
+      "epoch": 58.53257790368272,
+      "grad_norm": 3.857913453011595,
+      "learning_rate": 5e-06,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 884786108,
+      "step": 5155
+    },
+    {
+      "epoch": 58.53257790368272,
+      "loss": 0.05422041565179825,
+      "loss_ce": 0.00025008045486174524,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 884786108,
+      "step": 5155
+    },
+    {
+      "epoch": 58.54390934844193,
+      "grad_norm": 4.665716681310218,
+      "learning_rate": 5e-06,
+      "loss": 0.052,
+      "num_input_tokens_seen": 884956924,
+      "step": 5156
+    },
+    {
+      "epoch": 58.54390934844193,
+      "loss": 0.07115911692380905,
+      "loss_ce": 6.841756112407893e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 884956924,
+      "step": 5156
+    },
+    {
+      "epoch": 58.555240793201136,
+      "grad_norm": 3.0631782360908972,
+      "learning_rate": 5e-06,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 885128788,
+      "step": 5157
+    },
+    {
+      "epoch": 58.555240793201136,
+      "loss": 0.06514810770750046,
+      "loss_ce": 4.648451431421563e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 885128788,
+      "step": 5157
+    },
+    {
+      "epoch": 58.56657223796034,
+      "grad_norm": 3.6429955683677684,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 885300476,
+      "step": 5158
+    },
+    {
+      "epoch": 58.56657223796034,
+      "loss": 0.038397423923015594,
+      "loss_ce": 0.0002657122095115483,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 885300476,
+      "step": 5158
+    },
+    {
+      "epoch": 58.577903682719544,
+      "grad_norm": 3.832394927390381,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 885472168,
+      "step": 5159
+    },
+    {
+      "epoch": 58.577903682719544,
+      "loss": 0.04183327034115791,
+      "loss_ce": 3.9447215385735035e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 885472168,
+      "step": 5159
+    },
+    {
+      "epoch": 58.58923512747875,
+      "grad_norm": 3.7582317401369183,
+      "learning_rate": 5e-06,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 885644112,
+      "step": 5160
+    },
+    {
+      "epoch": 58.58923512747875,
+      "loss": 0.07095743715763092,
+      "loss_ce": 6.51002919767052e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 885644112,
+      "step": 5160
+    },
+    {
+      "epoch": 58.60056657223796,
+      "grad_norm": 3.948321749939692,
+      "learning_rate": 5e-06,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 885815256,
+      "step": 5161
+    },
+    {
+      "epoch": 58.60056657223796,
+      "loss": 0.05903956666588783,
+      "loss_ce": 0.00014063986600376666,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 885815256,
+      "step": 5161
+    },
+    {
+      "epoch": 58.611898016997166,
+      "grad_norm": 4.667675028409224,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 885986980,
+      "step": 5162
+    },
+    {
+      "epoch": 58.611898016997166,
+      "loss": 0.041775137186050415,
+      "loss_ce": 0.00010338243009755388,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 885986980,
+      "step": 5162
+    },
+    {
+      "epoch": 58.623229461756374,
+      "grad_norm": 3.509228240218512,
+      "learning_rate": 5e-06,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 886158724,
+      "step": 5163
+    },
+    {
+      "epoch": 58.623229461756374,
+      "loss": 0.06796180456876755,
+      "loss_ce": 7.545328844571486e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 886158724,
+      "step": 5163
+    },
+    {
+      "epoch": 58.63456090651558,
+      "grad_norm": 4.0130861192513345,
+      "learning_rate": 5e-06,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 886330920,
+      "step": 5164
+    },
+    {
+      "epoch": 58.63456090651558,
+      "loss": 0.0440477654337883,
+      "loss_ce": 4.141558747505769e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 886330920,
+      "step": 5164
+    },
+    {
+      "epoch": 58.64589235127479,
+      "grad_norm": 3.934199158970189,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 886502768,
+      "step": 5165
+    },
+    {
+      "epoch": 58.64589235127479,
+      "loss": 0.03873293846845627,
+      "loss_ce": 6.716544885421172e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 886502768,
+      "step": 5165
+    },
+    {
+      "epoch": 58.657223796033996,
+      "grad_norm": 3.856637103531575,
+      "learning_rate": 5e-06,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 886673020,
+      "step": 5166
+    },
+    {
+      "epoch": 58.657223796033996,
+      "loss": 0.03267926722764969,
+      "loss_ce": 2.545849565649405e-05,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 886673020,
+      "step": 5166
+    },
+    {
+      "epoch": 58.668555240793204,
+      "grad_norm": 4.324162407777541,
+      "learning_rate": 5e-06,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 886844684,
+      "step": 5167
+    },
+    {
+      "epoch": 58.668555240793204,
+      "loss": 0.05115539953112602,
+      "loss_ce": 3.845775790978223e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 886844684,
+      "step": 5167
+    },
+    {
+      "epoch": 58.67988668555241,
+      "grad_norm": 2.951253382292852,
+      "learning_rate": 5e-06,
+      "loss": 0.073,
+      "num_input_tokens_seen": 887016588,
+      "step": 5168
+    },
+    {
+      "epoch": 58.67988668555241,
+      "loss": 0.06818795204162598,
+      "loss_ce": 5.745126327383332e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 887016588,
+      "step": 5168
+    },
+    {
+      "epoch": 58.69121813031161,
+      "grad_norm": 3.5326689245700233,
+      "learning_rate": 5e-06,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 887186476,
+      "step": 5169
+    },
+    {
+      "epoch": 58.69121813031161,
+      "loss": 0.04181846231222153,
+      "loss_ce": 3.989890683442354e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 887186476,
+      "step": 5169
+    },
+    {
+      "epoch": 58.70254957507082,
+      "grad_norm": 3.5115756826015962,
+      "learning_rate": 5e-06,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 887356768,
+      "step": 5170
+    },
+    {
+      "epoch": 58.70254957507082,
+      "loss": 0.056678298860788345,
+      "loss_ce": 0.00011397062917239964,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 887356768,
+      "step": 5170
+    },
+    {
+      "epoch": 58.713881019830026,
+      "grad_norm": 3.445329232365934,
+      "learning_rate": 5e-06,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 887526332,
+      "step": 5171
+    },
+    {
+      "epoch": 58.713881019830026,
+      "loss": 0.07594342529773712,
+      "loss_ce": 4.620824256562628e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 887526332,
+      "step": 5171
+    },
+    {
+      "epoch": 58.725212464589234,
+      "grad_norm": 3.8084227247381373,
+      "learning_rate": 5e-06,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 887698288,
+      "step": 5172
+    },
+    {
+      "epoch": 58.725212464589234,
+      "loss": 0.059559352695941925,
+      "loss_ce": 3.481722160358913e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 887698288,
+      "step": 5172
+    },
+    {
+      "epoch": 58.73654390934844,
+      "grad_norm": 6.1021537043841825,
+      "learning_rate": 5e-06,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 887870352,
+      "step": 5173
+    },
+    {
+      "epoch": 58.73654390934844,
+      "loss": 0.03974830359220505,
+      "loss_ce": 6.019495049258694e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 887870352,
+      "step": 5173
+    },
+    {
+      "epoch": 58.74787535410765,
+      "grad_norm": 24.118149560993597,
+      "learning_rate": 5e-06,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 888042324,
+      "step": 5174
+    },
+    {
+      "epoch": 58.74787535410765,
+      "loss": 0.05540543794631958,
+      "loss_ce": 0.00012284755939617753,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 888042324,
+      "step": 5174
+    },
+    {
+      "epoch": 58.759206798866856,
+      "grad_norm": 3.5913669121689704,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 888213448,
+      "step": 5175
+    },
+    {
+      "epoch": 58.759206798866856,
+      "loss": 0.05376358702778816,
+      "loss_ce": 6.79100994602777e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 888213448,
+      "step": 5175
+    },
+    {
+      "epoch": 58.77053824362606,
+      "grad_norm": 3.125276980338516,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 888384316,
+      "step": 5176
+    },
+    {
+      "epoch": 58.77053824362606,
+      "loss": 0.07048118859529495,
+      "loss_ce": 0.00015342731785494834,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 888384316,
+      "step": 5176
+    },
+    {
+      "epoch": 58.78186968838527,
+      "grad_norm": 3.2676573211304083,
+      "learning_rate": 5e-06,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 888555352,
+      "step": 5177
+    },
+    {
+      "epoch": 58.78186968838527,
+      "loss": 0.027033204212784767,
+      "loss_ce": 8.618322317488492e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 888555352,
+      "step": 5177
+    },
+    {
+      "epoch": 58.79320113314448,
+      "grad_norm": 3.8466464162662635,
+      "learning_rate": 5e-06,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 888727452,
+      "step": 5178
+    },
+    {
+      "epoch": 58.79320113314448,
+      "loss": 0.06564351916313171,
+      "loss_ce": 6.124786159489304e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 888727452,
+      "step": 5178
+    },
+    {
+      "epoch": 58.804532577903686,
+      "grad_norm": 3.4246407041435205,
+      "learning_rate": 5e-06,
+      "loss": 0.0672,
+      "num_input_tokens_seen": 888898764,
+      "step": 5179
+    },
+    {
+      "epoch": 58.804532577903686,
+      "loss": 0.048203542828559875,
+      "loss_ce": 4.68058860860765e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 888898764,
+      "step": 5179
+    },
+    {
+      "epoch": 58.815864022662886,
+      "grad_norm": 3.1318405152470103,
+      "learning_rate": 5e-06,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 889069668,
+      "step": 5180
+    },
+    {
+      "epoch": 58.815864022662886,
+      "loss": 0.12884651124477386,
+      "loss_ce": 0.000382764614187181,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 889069668,
+      "step": 5180
+    },
+    {
+      "epoch": 58.827195467422094,
+      "grad_norm": 3.7127060891935946,
+      "learning_rate": 5e-06,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 889241380,
+      "step": 5181
+    },
+    {
+      "epoch": 58.827195467422094,
+      "loss": 0.03743062913417816,
+      "loss_ce": 7.71126797189936e-05,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 889241380,
+      "step": 5181
+    },
+    {
+      "epoch": 58.8385269121813,
+      "grad_norm": 3.32036666150757,
+      "learning_rate": 5e-06,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 889412572,
+      "step": 5182
+    },
+    {
+      "epoch": 58.8385269121813,
+      "loss": 0.029161836951971054,
+      "loss_ce": 6.332558405119926e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 889412572,
+      "step": 5182
+    },
+    {
+      "epoch": 58.84985835694051,
+      "grad_norm": 3.1431802741623733,
+      "learning_rate": 5e-06,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 889584372,
+      "step": 5183
+    },
+    {
+      "epoch": 58.84985835694051,
+      "loss": 0.04124344140291214,
+      "loss_ce": 2.1819667381350882e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 889584372,
+      "step": 5183
+    },
+    {
+      "epoch": 58.861189801699716,
+      "grad_norm": 3.1697984764414184,
+      "learning_rate": 5e-06,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 889755696,
+      "step": 5184
+    },
+    {
+      "epoch": 58.861189801699716,
+      "loss": 0.09251611679792404,
+      "loss_ce": 3.259401273680851e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 889755696,
+      "step": 5184
+    },
+    {
+      "epoch": 58.87252124645892,
+      "grad_norm": 3.2758392657934152,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 889927732,
+      "step": 5185
+    },
+    {
+      "epoch": 58.87252124645892,
+      "loss": 0.08032210171222687,
+      "loss_ce": 0.00012190362031105906,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 889927732,
+      "step": 5185
+    },
+    {
+      "epoch": 58.88385269121813,
+      "grad_norm": 3.5441915976507445,
+      "learning_rate": 5e-06,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 890099496,
+      "step": 5186
+    },
+    {
+      "epoch": 58.88385269121813,
+      "loss": 0.036501601338386536,
+      "loss_ce": 0.0002467167214490473,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 890099496,
+      "step": 5186
+    },
+    {
+      "epoch": 58.89518413597734,
+      "grad_norm": 8.553759434547434,
+      "learning_rate": 5e-06,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 890271624,
+      "step": 5187
+    },
+    {
+      "epoch": 58.89518413597734,
+      "loss": 0.05966384336352348,
+      "loss_ce": 9.353039786219597e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 890271624,
+      "step": 5187
+    },
+    {
+      "epoch": 58.906515580736546,
+      "grad_norm": 4.349920012260608,
+      "learning_rate": 5e-06,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 890442888,
+      "step": 5188
+    },
+    {
+      "epoch": 58.906515580736546,
+      "loss": 0.04077528044581413,
+      "loss_ce": 4.957479177392088e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 890442888,
+      "step": 5188
+    },
+    {
+      "epoch": 58.91784702549575,
+      "grad_norm": 2.59131701536466,
+      "learning_rate": 5e-06,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 890614716,
+      "step": 5189
+    },
+    {
+      "epoch": 58.91784702549575,
+      "loss": 0.04095759987831116,
+      "loss_ce": 9.456434054300189e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 890614716,
+      "step": 5189
+    },
+    {
+      "epoch": 58.92917847025496,
+      "grad_norm": 4.765012906974223,
+      "learning_rate": 5e-06,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 890786476,
+      "step": 5190
+    },
+    {
+      "epoch": 58.92917847025496,
+      "loss": 0.027431871742010117,
+      "loss_ce": 4.234358493704349e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 890786476,
+      "step": 5190
+    },
+    {
+      "epoch": 58.94050991501416,
+      "grad_norm": 3.8250754101430284,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 890957028,
+      "step": 5191
+    },
+    {
+      "epoch": 58.94050991501416,
+      "loss": 0.030107181519269943,
+      "loss_ce": 7.788444054313004e-05,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 890957028,
+      "step": 5191
+    },
+    {
+      "epoch": 58.95184135977337,
+      "grad_norm": 3.492034853285189,
+      "learning_rate": 5e-06,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 891128872,
+      "step": 5192
+    },
+    {
+      "epoch": 58.95184135977337,
+      "loss": 0.0569198802113533,
+      "loss_ce": 5.037500523030758e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 891128872,
+      "step": 5192
+    },
+    {
+      "epoch": 58.963172804532576,
+      "grad_norm": 4.031746483819515,
+      "learning_rate": 5e-06,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 891299784,
+      "step": 5193
+    },
+    {
+      "epoch": 58.963172804532576,
+      "loss": 0.052659302949905396,
+      "loss_ce": 4.700115096056834e-05,
+      "loss_iou": 0.625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 891299784,
+      "step": 5193
+    },
+    {
+      "epoch": 58.97450424929178,
+      "grad_norm": 3.5584400734985246,
+      "learning_rate": 5e-06,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 891471652,
+      "step": 5194
+    },
+    {
+      "epoch": 58.97450424929178,
+      "loss": 0.05989007651805878,
+      "loss_ce": 7.562487007817253e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 891471652,
+      "step": 5194
+    },
+    {
+      "epoch": 58.98583569405099,
+      "grad_norm": 3.56007655017724,
+      "learning_rate": 5e-06,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 891642620,
+      "step": 5195
+    },
+    {
+      "epoch": 58.98583569405099,
+      "loss": 0.034561313688755035,
+      "loss_ce": 7.644950528629124e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 891642620,
+      "step": 5195
+    },
+    {
+      "epoch": 58.9971671388102,
+      "grad_norm": 3.3759239239938763,
+      "learning_rate": 5e-06,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 891814196,
+      "step": 5196
+    },
+    {
+      "epoch": 58.9971671388102,
+      "loss": 0.046778351068496704,
+      "loss_ce": 5.59394211450126e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 891814196,
+      "step": 5196
+    },
+    {
+      "epoch": 58.9971671388102,
+      "loss": 0.03991572558879852,
+      "loss_ce": 5.9768499340862036e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 891857332,
+      "step": 5196
+    },
+    {
+      "epoch": 59.008498583569406,
+      "grad_norm": 2.9512422980350177,
+      "learning_rate": 5e-06,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 891986148,
+      "step": 5197
+    },
+    {
+      "epoch": 59.008498583569406,
+      "loss": 0.07488942891359329,
+      "loss_ce": 7.558612560387701e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 891986148,
+      "step": 5197
+    },
+    {
+      "epoch": 59.01983002832861,
+      "grad_norm": 3.000726492663069,
+      "learning_rate": 5e-06,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 892157096,
+      "step": 5198
+    },
+    {
+      "epoch": 59.01983002832861,
+      "loss": 0.03526885062456131,
+      "loss_ce": 5.156617407919839e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 892157096,
+      "step": 5198
+    },
+    {
+      "epoch": 59.03116147308782,
+      "grad_norm": 2.8948697206372023,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 892327480,
+      "step": 5199
+    },
+    {
+      "epoch": 59.03116147308782,
+      "loss": 0.036264244467020035,
+      "loss_ce": 5.5136340961325914e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 892327480,
+      "step": 5199
+    },
+    {
+      "epoch": 59.04249291784703,
+      "grad_norm": 3.0108412687307133,
+      "learning_rate": 5e-06,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 892495460,
+      "step": 5200
+    },
+    {
+      "epoch": 59.04249291784703,
+      "loss": 0.038533806800842285,
+      "loss_ce": 5.113949009682983e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 892495460,
+      "step": 5200
+    },
+    {
+      "epoch": 59.053824362606235,
+      "grad_norm": 3.4335686742362515,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 892667304,
+      "step": 5201
+    },
+    {
+      "epoch": 59.053824362606235,
+      "loss": 0.057238124310970306,
+      "loss_ce": 4.818146044271998e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 892667304,
+      "step": 5201
+    },
+    {
+      "epoch": 59.065155807365436,
+      "grad_norm": 3.737309829311134,
+      "learning_rate": 5e-06,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 892839164,
+      "step": 5202
+    },
+    {
+      "epoch": 59.065155807365436,
+      "loss": 0.03735791891813278,
+      "loss_ce": 0.00041638925904408097,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 892839164,
+      "step": 5202
+    },
+    {
+      "epoch": 59.07648725212464,
+      "grad_norm": 4.897844191340123,
+      "learning_rate": 5e-06,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 893009204,
+      "step": 5203
+    },
+    {
+      "epoch": 59.07648725212464,
+      "loss": 0.08405373990535736,
+      "loss_ce": 6.936582212802023e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 893009204,
+      "step": 5203
+    },
+    {
+      "epoch": 59.08781869688385,
+      "grad_norm": 3.048425046986099,
+      "learning_rate": 5e-06,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 893180392,
+      "step": 5204
+    },
+    {
+      "epoch": 59.08781869688385,
+      "loss": 0.033896736800670624,
+      "loss_ce": 3.748424569494091e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 893180392,
+      "step": 5204
+    },
+    {
+      "epoch": 59.09915014164306,
+      "grad_norm": 3.8873056091322047,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 893352384,
+      "step": 5205
+    },
+    {
+      "epoch": 59.09915014164306,
+      "loss": 0.036965228617191315,
+      "loss_ce": 8.473488560412079e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 893352384,
+      "step": 5205
+    },
+    {
+      "epoch": 59.110481586402265,
+      "grad_norm": 3.578444002014332,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 893523704,
+      "step": 5206
+    },
+    {
+      "epoch": 59.110481586402265,
+      "loss": 0.09712836146354675,
+      "loss_ce": 9.772434714250267e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 893523704,
+      "step": 5206
+    },
+    {
+      "epoch": 59.12181303116147,
+      "grad_norm": 3.0985569110304,
+      "learning_rate": 5e-06,
+      "loss": 0.065,
+      "num_input_tokens_seen": 893695352,
+      "step": 5207
+    },
+    {
+      "epoch": 59.12181303116147,
+      "loss": 0.03881976008415222,
+      "loss_ce": 3.1915416911942884e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 893695352,
+      "step": 5207
+    },
+    {
+      "epoch": 59.13314447592068,
+      "grad_norm": 8.824055311547783,
+      "learning_rate": 5e-06,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 893867392,
+      "step": 5208
+    },
+    {
+      "epoch": 59.13314447592068,
+      "loss": 0.07464145123958588,
+      "loss_ce": 2.5969651687773876e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 893867392,
+      "step": 5208
+    },
+    {
+      "epoch": 59.14447592067989,
+      "grad_norm": 3.500292296272771,
+      "learning_rate": 5e-06,
+      "loss": 0.045,
+      "num_input_tokens_seen": 894038532,
+      "step": 5209
+    },
+    {
+      "epoch": 59.14447592067989,
+      "loss": 0.031523656100034714,
+      "loss_ce": 4.4773987610824406e-05,
+      "loss_iou": 0.640625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 894038532,
+      "step": 5209
+    },
+    {
+      "epoch": 59.155807365439095,
+      "grad_norm": 2.9003618968807303,
+      "learning_rate": 5e-06,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 894210628,
+      "step": 5210
+    },
+    {
+      "epoch": 59.155807365439095,
+      "loss": 0.04914901405572891,
+      "loss_ce": 6.148905958980322e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 894210628,
+      "step": 5210
+    },
+    {
+      "epoch": 59.1671388101983,
+      "grad_norm": 3.661745755725081,
+      "learning_rate": 5e-06,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 894381088,
+      "step": 5211
+    },
+    {
+      "epoch": 59.1671388101983,
+      "loss": 0.031463753432035446,
+      "loss_ce": 4.590559910866432e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 894381088,
+      "step": 5211
+    },
+    {
+      "epoch": 59.17847025495751,
+      "grad_norm": 3.6140270793788654,
+      "learning_rate": 5e-06,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 894552936,
+      "step": 5212
+    },
+    {
+      "epoch": 59.17847025495751,
+      "loss": 0.06581361591815948,
+      "loss_ce": 7.875532901380211e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 894552936,
+      "step": 5212
+    },
+    {
+      "epoch": 59.18980169971671,
+      "grad_norm": 3.0262770442006057,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 894724828,
+      "step": 5213
+    },
+    {
+      "epoch": 59.18980169971671,
+      "loss": 0.040536098182201385,
+      "loss_ce": 3.927417128579691e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 894724828,
+      "step": 5213
+    },
+    {
+      "epoch": 59.20113314447592,
+      "grad_norm": 3.8619316168824627,
+      "learning_rate": 5e-06,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 894896716,
+      "step": 5214
+    },
+    {
+      "epoch": 59.20113314447592,
+      "loss": 0.061334170401096344,
+      "loss_ce": 2.436098839098122e-05,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 894896716,
+      "step": 5214
+    },
+    {
+      "epoch": 59.212464589235125,
+      "grad_norm": 3.5741880313937777,
+      "learning_rate": 5e-06,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 895068848,
+      "step": 5215
+    },
+    {
+      "epoch": 59.212464589235125,
+      "loss": 0.031410180032253265,
+      "loss_ce": 2.2852716938359663e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 895068848,
+      "step": 5215
+    },
+    {
+      "epoch": 59.22379603399433,
+      "grad_norm": 3.3876071852006344,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 895240408,
+      "step": 5216
+    },
+    {
+      "epoch": 59.22379603399433,
+      "loss": 0.0444064661860466,
+      "loss_ce": 5.6797223805915564e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 895240408,
+      "step": 5216
+    },
+    {
+      "epoch": 59.23512747875354,
+      "grad_norm": 3.6376875567123848,
+      "learning_rate": 5e-06,
+      "loss": 0.0644,
+      "num_input_tokens_seen": 895412100,
+      "step": 5217
+    },
+    {
+      "epoch": 59.23512747875354,
+      "loss": 0.058473944664001465,
+      "loss_ce": 3.278174699516967e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 895412100,
+      "step": 5217
+    },
+    {
+      "epoch": 59.24645892351275,
+      "grad_norm": 3.5705671608171876,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 895583924,
+      "step": 5218
+    },
+    {
+      "epoch": 59.24645892351275,
+      "loss": 0.03600358963012695,
+      "loss_ce": 6.91401437507011e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 895583924,
+      "step": 5218
+    },
+    {
+      "epoch": 59.257790368271955,
+      "grad_norm": 3.2717295880101074,
+      "learning_rate": 5e-06,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 895755580,
+      "step": 5219
+    },
+    {
+      "epoch": 59.257790368271955,
+      "loss": 0.16085468232631683,
+      "loss_ce": 2.703846439544577e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 895755580,
+      "step": 5219
+    },
+    {
+      "epoch": 59.26912181303116,
+      "grad_norm": 3.8188275027294445,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 895927500,
+      "step": 5220
+    },
+    {
+      "epoch": 59.26912181303116,
+      "loss": 0.05126499384641647,
+      "loss_ce": 7.175881182774901e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 895927500,
+      "step": 5220
+    },
+    {
+      "epoch": 59.28045325779037,
+      "grad_norm": 3.348527184903433,
+      "learning_rate": 5e-06,
+      "loss": 0.0736,
+      "num_input_tokens_seen": 896098612,
+      "step": 5221
+    },
+    {
+      "epoch": 59.28045325779037,
+      "loss": 0.08834994584321976,
+      "loss_ce": 6.259631481952965e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 896098612,
+      "step": 5221
+    },
+    {
+      "epoch": 59.29178470254958,
+      "grad_norm": 3.6352127779974093,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 896269664,
+      "step": 5222
+    },
+    {
+      "epoch": 59.29178470254958,
+      "loss": 0.042771488428115845,
+      "loss_ce": 7.739839202258736e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 896269664,
+      "step": 5222
+    },
+    {
+      "epoch": 59.303116147308785,
+      "grad_norm": 5.631886452199878,
+      "learning_rate": 5e-06,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 896441164,
+      "step": 5223
+    },
+    {
+      "epoch": 59.303116147308785,
+      "loss": 0.14781562983989716,
+      "loss_ce": 0.0005072755157016218,
+      "loss_iou": 0.484375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 896441164,
+      "step": 5223
+    },
+    {
+      "epoch": 59.314447592067985,
+      "grad_norm": 19.52152245455981,
+      "learning_rate": 5e-06,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 896612956,
+      "step": 5224
+    },
+    {
+      "epoch": 59.314447592067985,
+      "loss": 0.026554375886917114,
+      "loss_ce": 8.037681982386857e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 896612956,
+      "step": 5224
+    },
+    {
+      "epoch": 59.32577903682719,
+      "grad_norm": 2.491045027064484,
+      "learning_rate": 5e-06,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 896784588,
+      "step": 5225
+    },
+    {
+      "epoch": 59.32577903682719,
+      "loss": 0.045461613684892654,
+      "loss_ce": 2.0940171452821232e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 896784588,
+      "step": 5225
+    },
+    {
+      "epoch": 59.3371104815864,
+      "grad_norm": 2.6585516503788598,
+      "learning_rate": 5e-06,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 896956444,
+      "step": 5226
+    },
+    {
+      "epoch": 59.3371104815864,
+      "loss": 0.034350357949733734,
+      "loss_ce": 4.8599016736261547e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 896956444,
+      "step": 5226
+    },
+    {
+      "epoch": 59.34844192634561,
+      "grad_norm": 3.215248566430737,
+      "learning_rate": 5e-06,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 897125964,
+      "step": 5227
+    },
+    {
+      "epoch": 59.34844192634561,
+      "loss": 0.09650050848722458,
+      "loss_ce": 9.547673835186288e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 897125964,
+      "step": 5227
+    },
+    {
+      "epoch": 59.359773371104815,
+      "grad_norm": 3.5119163950093584,
+      "learning_rate": 5e-06,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 897297824,
+      "step": 5228
+    },
+    {
+      "epoch": 59.359773371104815,
+      "loss": 0.07572523504495621,
+      "loss_ce": 4.164027632214129e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 897297824,
+      "step": 5228
+    },
+    {
+      "epoch": 59.37110481586402,
+      "grad_norm": 5.764402596600396,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 897467512,
+      "step": 5229
+    },
+    {
+      "epoch": 59.37110481586402,
+      "loss": 0.14661215245723724,
+      "loss_ce": 6.673127063550055e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 897467512,
+      "step": 5229
+    },
+    {
+      "epoch": 59.38243626062323,
+      "grad_norm": 4.043773771106399,
+      "learning_rate": 5e-06,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 897639088,
+      "step": 5230
+    },
+    {
+      "epoch": 59.38243626062323,
+      "loss": 0.12071627378463745,
+      "loss_ce": 4.977144271833822e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 897639088,
+      "step": 5230
+    },
+    {
+      "epoch": 59.39376770538244,
+      "grad_norm": 3.3223077903414184,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 897810984,
+      "step": 5231
+    },
+    {
+      "epoch": 59.39376770538244,
+      "loss": 0.044003166258335114,
+      "loss_ce": 2.7336178391124122e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 897810984,
+      "step": 5231
+    },
+    {
+      "epoch": 59.405099150141645,
+      "grad_norm": 3.795589652525898,
+      "learning_rate": 5e-06,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 897981576,
+      "step": 5232
+    },
+    {
+      "epoch": 59.405099150141645,
+      "loss": 0.04296860471367836,
+      "loss_ce": 6.0891936300322413e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 897981576,
+      "step": 5232
+    },
+    {
+      "epoch": 59.41643059490085,
+      "grad_norm": 3.152132433951108,
+      "learning_rate": 5e-06,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 898153336,
+      "step": 5233
+    },
+    {
+      "epoch": 59.41643059490085,
+      "loss": 0.044174984097480774,
+      "loss_ce": 6.182699144119397e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 898153336,
+      "step": 5233
+    },
+    {
+      "epoch": 59.42776203966006,
+      "grad_norm": 3.530469751948677,
+      "learning_rate": 5e-06,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 898324248,
+      "step": 5234
+    },
+    {
+      "epoch": 59.42776203966006,
+      "loss": 0.08770953118801117,
+      "loss_ce": 4.7785593778826296e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 898324248,
+      "step": 5234
+    },
+    {
+      "epoch": 59.43909348441926,
+      "grad_norm": 3.1972254841473653,
+      "learning_rate": 5e-06,
+      "loss": 0.038,
+      "num_input_tokens_seen": 898495984,
+      "step": 5235
+    },
+    {
+      "epoch": 59.43909348441926,
+      "loss": 0.03576774150133133,
+      "loss_ce": 0.000817483349237591,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 898495984,
+      "step": 5235
+    },
+    {
+      "epoch": 59.45042492917847,
+      "grad_norm": 3.6598191103957824,
+      "learning_rate": 5e-06,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 898667248,
+      "step": 5236
+    },
+    {
+      "epoch": 59.45042492917847,
+      "loss": 0.046595871448516846,
+      "loss_ce": 0.00014811850269325078,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 898667248,
+      "step": 5236
+    },
+    {
+      "epoch": 59.461756373937675,
+      "grad_norm": 3.9673486446916817,
+      "learning_rate": 5e-06,
+      "loss": 0.053,
+      "num_input_tokens_seen": 898838312,
+      "step": 5237
+    },
+    {
+      "epoch": 59.461756373937675,
+      "loss": 0.05976229906082153,
+      "loss_ce": 6.991454574745148e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 898838312,
+      "step": 5237
+    },
+    {
+      "epoch": 59.47308781869688,
+      "grad_norm": 4.6056600842978845,
+      "learning_rate": 5e-06,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 899009204,
+      "step": 5238
+    },
+    {
+      "epoch": 59.47308781869688,
+      "loss": 0.08491309732198715,
+      "loss_ce": 8.948677714215592e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 899009204,
+      "step": 5238
+    },
+    {
+      "epoch": 59.48441926345609,
+      "grad_norm": 8.445198532074523,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 899181288,
+      "step": 5239
+    },
+    {
+      "epoch": 59.48441926345609,
+      "loss": 0.07698958367109299,
+      "loss_ce": 5.476978185470216e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 899181288,
+      "step": 5239
+    },
+    {
+      "epoch": 59.4957507082153,
+      "grad_norm": 4.764378479825633,
+      "learning_rate": 5e-06,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 899352900,
+      "step": 5240
+    },
+    {
+      "epoch": 59.4957507082153,
+      "loss": 0.05390908196568489,
+      "loss_ce": 4.555407213047147e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 899352900,
+      "step": 5240
+    },
+    {
+      "epoch": 59.507082152974505,
+      "grad_norm": 3.1921177712850275,
+      "learning_rate": 5e-06,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 899524552,
+      "step": 5241
+    },
+    {
+      "epoch": 59.507082152974505,
+      "loss": 0.04019215330481529,
+      "loss_ce": 7.679581176489592e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 899524552,
+      "step": 5241
+    },
+    {
+      "epoch": 59.51841359773371,
+      "grad_norm": 3.0275404441557745,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 899695716,
+      "step": 5242
+    },
+    {
+      "epoch": 59.51841359773371,
+      "loss": 0.036375295370817184,
+      "loss_ce": 0.002149831270799041,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 899695716,
+      "step": 5242
+    },
+    {
+      "epoch": 59.52974504249292,
+      "grad_norm": 4.720293188026689,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 899868096,
+      "step": 5243
+    },
+    {
+      "epoch": 59.52974504249292,
+      "loss": 0.04349971190094948,
+      "loss_ce": 0.00014949182514101267,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 899868096,
+      "step": 5243
+    },
+    {
+      "epoch": 59.54107648725213,
+      "grad_norm": 3.7776323495494184,
+      "learning_rate": 5e-06,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 900040268,
+      "step": 5244
+    },
+    {
+      "epoch": 59.54107648725213,
+      "loss": 0.04592818766832352,
+      "loss_ce": 0.00010604485578369349,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 900040268,
+      "step": 5244
+    },
+    {
+      "epoch": 59.552407932011334,
+      "grad_norm": 3.279990577561121,
+      "learning_rate": 5e-06,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 900212156,
+      "step": 5245
+    },
+    {
+      "epoch": 59.552407932011334,
+      "loss": 0.03187783807516098,
+      "loss_ce": 3.27452908095438e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 900212156,
+      "step": 5245
+    },
+    {
+      "epoch": 59.563739376770535,
+      "grad_norm": 3.411208085270571,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 900384188,
+      "step": 5246
+    },
+    {
+      "epoch": 59.563739376770535,
+      "loss": 0.05028306692838669,
+      "loss_ce": 5.1135484682163224e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 900384188,
+      "step": 5246
+    },
+    {
+      "epoch": 59.57507082152974,
+      "grad_norm": 3.999552577502403,
+      "learning_rate": 5e-06,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 900556476,
+      "step": 5247
+    },
+    {
+      "epoch": 59.57507082152974,
+      "loss": 0.03434481471776962,
+      "loss_ce": 4.305379115976393e-05,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 900556476,
+      "step": 5247
+    },
+    {
+      "epoch": 59.58640226628895,
+      "grad_norm": 3.6926286319582684,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 900728344,
+      "step": 5248
+    },
+    {
+      "epoch": 59.58640226628895,
+      "loss": 0.06600113213062286,
+      "loss_ce": 5.264939318294637e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 900728344,
+      "step": 5248
+    },
+    {
+      "epoch": 59.59773371104816,
+      "grad_norm": 3.3382681546609305,
+      "learning_rate": 5e-06,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 900900272,
+      "step": 5249
+    },
+    {
+      "epoch": 59.59773371104816,
+      "loss": 0.06957980245351791,
+      "loss_ce": 3.0236533348215744e-05,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 900900272,
+      "step": 5249
+    },
+    {
+      "epoch": 59.609065155807365,
+      "grad_norm": 3.1123895424490073,
+      "learning_rate": 5e-06,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 901072064,
+      "step": 5250
+    },
+    {
+      "epoch": 59.609065155807365,
+      "eval_seeclick_CIoU": 0.5399854183197021,
+      "eval_seeclick_GIoU": 0.5371536016464233,
+      "eval_seeclick_IoU": 0.5758621692657471,
+      "eval_seeclick_MAE_all": 0.0638058539479971,
+      "eval_seeclick_MAE_h": 0.03312877286225557,
+      "eval_seeclick_MAE_w": 0.0970098078250885,
+      "eval_seeclick_MAE_x": 0.09429314360022545,
+      "eval_seeclick_MAE_y": 0.03079168125987053,
+      "eval_seeclick_NUM_probability": 0.9999721050262451,
+      "eval_seeclick_inside_bbox": 0.9232954680919647,
+      "eval_seeclick_loss": 0.9146884679794312,
+      "eval_seeclick_loss_ce": 0.663954883813858,
+      "eval_seeclick_loss_iou": 0.5419921875,
+      "eval_seeclick_loss_num": 0.050067901611328125,
+      "eval_seeclick_loss_xval": 0.250457763671875,
+      "eval_seeclick_runtime": 69.1178,
+      "eval_seeclick_samples_per_second": 0.622,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 901072064,
+      "step": 5250
+    },
+    {
+      "epoch": 59.609065155807365,
+      "eval_icons_CIoU": 0.7491226494312286,
+      "eval_icons_GIoU": 0.7482729554176331,
+      "eval_icons_IoU": 0.7629626095294952,
+      "eval_icons_MAE_all": 0.03358728997409344,
+      "eval_icons_MAE_h": 0.030434874817728996,
+      "eval_icons_MAE_w": 0.03648512065410614,
+      "eval_icons_MAE_x": 0.033146674279123545,
+      "eval_icons_MAE_y": 0.03428248316049576,
+      "eval_icons_NUM_probability": 0.9996849894523621,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.12134695798158646,
+      "eval_icons_loss_ce": 0.005554302595555782,
+      "eval_icons_loss_iou": 0.50933837890625,
+      "eval_icons_loss_num": 0.020902633666992188,
+      "eval_icons_loss_xval": 0.1045379638671875,
+      "eval_icons_runtime": 86.7469,
+      "eval_icons_samples_per_second": 0.576,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 901072064,
+      "step": 5250
+    },
+    {
+      "epoch": 59.609065155807365,
+      "eval_screenspot_CIoU": 0.6173333923021952,
+      "eval_screenspot_GIoU": 0.6209428906440735,
+      "eval_screenspot_IoU": 0.6434433460235596,
+      "eval_screenspot_MAE_all": 0.06658324226737022,
+      "eval_screenspot_MAE_h": 0.03863982359568278,
+      "eval_screenspot_MAE_w": 0.11977884421745937,
+      "eval_screenspot_MAE_x": 0.07868607218066852,
+      "eval_screenspot_MAE_y": 0.02922821355362733,
+      "eval_screenspot_NUM_probability": 0.9999502499898275,
+      "eval_screenspot_inside_bbox": 0.8974999984105428,
+      "eval_screenspot_loss": 0.28868868947029114,
+      "eval_screenspot_loss_ce": 0.013192149965713421,
+      "eval_screenspot_loss_iou": 0.5154622395833334,
+      "eval_screenspot_loss_num": 0.054911295572916664,
+      "eval_screenspot_loss_xval": 0.27447509765625,
+      "eval_screenspot_runtime": 137.1174,
+      "eval_screenspot_samples_per_second": 0.649,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 901072064,
+      "step": 5250
+    },
+    {
+      "epoch": 59.609065155807365,
+      "eval_compot_CIoU": 0.8520741164684296,
+      "eval_compot_GIoU": 0.8481800258159637,
+      "eval_compot_IoU": 0.8633108735084534,
+      "eval_compot_MAE_all": 0.02417416125535965,
+      "eval_compot_MAE_h": 0.018058547750115395,
+      "eval_compot_MAE_w": 0.03456522338092327,
+      "eval_compot_MAE_x": 0.02924160286784172,
+      "eval_compot_MAE_y": 0.01483126450330019,
+      "eval_compot_NUM_probability": 0.9999614357948303,
+      "eval_compot_inside_bbox": 0.9409722089767456,
+      "eval_compot_loss": 0.0808219239115715,
+      "eval_compot_loss_ce": 3.6658964745583944e-05,
+      "eval_compot_loss_iou": 0.520263671875,
+      "eval_compot_loss_num": 0.01422882080078125,
+      "eval_compot_loss_xval": 0.07110595703125,
+      "eval_compot_runtime": 89.3881,
+      "eval_compot_samples_per_second": 0.559,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 901072064,
+      "step": 5250
+    },
+    {
+      "epoch": 59.609065155807365,
+      "eval_custom_ui_MAE_all": 0.020768867805600166,
+      "eval_custom_ui_MAE_x": 0.03414425626397133,
+      "eval_custom_ui_MAE_y": 0.0073934795800596476,
+      "eval_custom_ui_NUM_probability": 0.9998302757740021,
+      "eval_custom_ui_loss": 0.2251778095960617,
+      "eval_custom_ui_loss_ce": 0.11784247308969498,
+      "eval_custom_ui_loss_num": 0.020610809326171875,
+      "eval_custom_ui_loss_xval": 0.102996826171875,
+      "eval_custom_ui_runtime": 59.5926,
+      "eval_custom_ui_samples_per_second": 0.839,
+      "eval_custom_ui_steps_per_second": 0.034,
+      "num_input_tokens_seen": 901072064,
+      "step": 5250
+    },
+    {
+      "epoch": 59.609065155807365,
+      "loss": 0.2747749090194702,
+      "loss_ce": 0.14831003546714783,
+      "loss_iou": 0.0,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 901072064,
+      "step": 5250
+    },
+    {
+      "epoch": 59.62039660056657,
+      "grad_norm": 3.7087665501915996,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 901244140,
+      "step": 5251
+    },
+    {
+      "epoch": 59.62039660056657,
+      "loss": 0.029397062957286835,
+      "loss_ce": 0.000893645454198122,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 901244140,
+      "step": 5251
+    },
+    {
+      "epoch": 59.63172804532578,
+      "grad_norm": 3.283835748950128,
+      "learning_rate": 5e-06,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 901415972,
+      "step": 5252
+    },
+    {
+      "epoch": 59.63172804532578,
+      "loss": 0.05694088712334633,
+      "loss_ce": 7.138182263588533e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 901415972,
+      "step": 5252
+    },
+    {
+      "epoch": 59.64305949008499,
+      "grad_norm": 3.06353840442789,
+      "learning_rate": 5e-06,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 901587804,
+      "step": 5253
+    },
+    {
+      "epoch": 59.64305949008499,
+      "loss": 0.06405947357416153,
+      "loss_ce": 6.411396316252649e-05,
+      "loss_iou": 0.1220703125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 901587804,
+      "step": 5253
+    },
+    {
+      "epoch": 59.654390934844194,
+      "grad_norm": 2.9528373098068994,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 901758880,
+      "step": 5254
+    },
+    {
+      "epoch": 59.654390934844194,
+      "loss": 0.05358899384737015,
+      "loss_ce": 6.116217991802841e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 901758880,
+      "step": 5254
+    },
+    {
+      "epoch": 59.6657223796034,
+      "grad_norm": 3.2228902111648075,
+      "learning_rate": 5e-06,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 901930576,
+      "step": 5255
+    },
+    {
+      "epoch": 59.6657223796034,
+      "loss": 0.11216942965984344,
+      "loss_ce": 4.784986231243238e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 901930576,
+      "step": 5255
+    },
+    {
+      "epoch": 59.67705382436261,
+      "grad_norm": 3.584780983850795,
+      "learning_rate": 5e-06,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 902101944,
+      "step": 5256
+    },
+    {
+      "epoch": 59.67705382436261,
+      "loss": 0.10403447598218918,
+      "loss_ce": 4.58312060800381e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 902101944,
+      "step": 5256
+    },
+    {
+      "epoch": 59.68838526912181,
+      "grad_norm": 3.7482166240001744,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 902273520,
+      "step": 5257
+    },
+    {
+      "epoch": 59.68838526912181,
+      "loss": 0.03523241728544235,
+      "loss_ce": 4.565247581922449e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 902273520,
+      "step": 5257
+    },
+    {
+      "epoch": 59.69971671388102,
+      "grad_norm": 3.209212969824453,
+      "learning_rate": 5e-06,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 902445200,
+      "step": 5258
+    },
+    {
+      "epoch": 59.69971671388102,
+      "loss": 0.050422340631484985,
+      "loss_ce": 8.359376806765795e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 902445200,
+      "step": 5258
+    },
+    {
+      "epoch": 59.711048158640224,
+      "grad_norm": 2.9227306083120066,
+      "learning_rate": 5e-06,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 902616932,
+      "step": 5259
+    },
+    {
+      "epoch": 59.711048158640224,
+      "loss": 0.046733736991882324,
+      "loss_ce": 0.00011813243327196687,
+      "loss_iou": 0.40625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 902616932,
+      "step": 5259
+    },
+    {
+      "epoch": 59.72237960339943,
+      "grad_norm": 2.5986721588811226,
+      "learning_rate": 5e-06,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 902788236,
+      "step": 5260
+    },
+    {
+      "epoch": 59.72237960339943,
+      "loss": 0.0629258081316948,
+      "loss_ce": 0.00021981516329105943,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 902788236,
+      "step": 5260
+    },
+    {
+      "epoch": 59.73371104815864,
+      "grad_norm": 2.7199419352021823,
+      "learning_rate": 5e-06,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 902958860,
+      "step": 5261
+    },
+    {
+      "epoch": 59.73371104815864,
+      "loss": 0.03853485360741615,
+      "loss_ce": 2.167032471334096e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 902958860,
+      "step": 5261
+    },
+    {
+      "epoch": 59.74504249291785,
+      "grad_norm": 2.445189939451211,
+      "learning_rate": 5e-06,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 903127396,
+      "step": 5262
+    },
+    {
+      "epoch": 59.74504249291785,
+      "loss": 0.029985571280121803,
+      "loss_ce": 7.834483403712511e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 903127396,
+      "step": 5262
+    },
+    {
+      "epoch": 59.756373937677054,
+      "grad_norm": 2.2026778973020398,
+      "learning_rate": 5e-06,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 903299152,
+      "step": 5263
+    },
+    {
+      "epoch": 59.756373937677054,
+      "loss": 0.04307733476161957,
+      "loss_ce": 9.332680201623589e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 903299152,
+      "step": 5263
+    },
+    {
+      "epoch": 59.76770538243626,
+      "grad_norm": 1.933058244066133,
+      "learning_rate": 5e-06,
+      "loss": 0.048,
+      "num_input_tokens_seen": 903470880,
+      "step": 5264
+    },
+    {
+      "epoch": 59.76770538243626,
+      "loss": 0.0282057486474514,
+      "loss_ce": 0.00011431855091359466,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 903470880,
+      "step": 5264
+    },
+    {
+      "epoch": 59.77903682719547,
+      "grad_norm": 2.370442082220845,
+      "learning_rate": 5e-06,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 903642872,
+      "step": 5265
+    },
+    {
+      "epoch": 59.77903682719547,
+      "loss": 0.0547727569937706,
+      "loss_ce": 5.473729470395483e-05,
+      "loss_iou": 0.0810546875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 903642872,
+      "step": 5265
+    },
+    {
+      "epoch": 59.79036827195468,
+      "grad_norm": 4.494736490360524,
+      "learning_rate": 5e-06,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 903814820,
+      "step": 5266
+    },
+    {
+      "epoch": 59.79036827195468,
+      "loss": 0.11013898253440857,
+      "loss_ce": 0.00017652772658038884,
+      "loss_iou": 0.265625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 903814820,
+      "step": 5266
+    },
+    {
+      "epoch": 59.801699716713884,
+      "grad_norm": 3.984608772692165,
+      "learning_rate": 5e-06,
+      "loss": 0.039,
+      "num_input_tokens_seen": 903986920,
+      "step": 5267
+    },
+    {
+      "epoch": 59.801699716713884,
+      "loss": 0.050803352147340775,
+      "loss_ce": 0.004447150509804487,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 903986920,
+      "step": 5267
+    },
+    {
+      "epoch": 59.81303116147309,
+      "grad_norm": 4.291915277877288,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 904155872,
+      "step": 5268
+    },
+    {
+      "epoch": 59.81303116147309,
+      "loss": 0.03763923421502113,
+      "loss_ce": 5.683595372829586e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 904155872,
+      "step": 5268
+    },
+    {
+      "epoch": 59.82436260623229,
+      "grad_norm": 3.291402256615581,
+      "learning_rate": 5e-06,
+      "loss": 0.0787,
+      "num_input_tokens_seen": 904325228,
+      "step": 5269
+    },
+    {
+      "epoch": 59.82436260623229,
+      "loss": 0.04456984996795654,
+      "loss_ce": 7.522162195527926e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 904325228,
+      "step": 5269
+    },
+    {
+      "epoch": 59.8356940509915,
+      "grad_norm": 2.919817446506232,
+      "learning_rate": 5e-06,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 904496380,
+      "step": 5270
+    },
+    {
+      "epoch": 59.8356940509915,
+      "loss": 0.0895550325512886,
+      "loss_ce": 9.274869807995856e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 904496380,
+      "step": 5270
+    },
+    {
+      "epoch": 59.84702549575071,
+      "grad_norm": 3.499973683797222,
+      "learning_rate": 5e-06,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 904665796,
+      "step": 5271
+    },
+    {
+      "epoch": 59.84702549575071,
+      "loss": 0.03309710696339607,
+      "loss_ce": 3.894151814165525e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 904665796,
+      "step": 5271
+    },
+    {
+      "epoch": 59.858356940509914,
+      "grad_norm": 4.9206719201400455,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 904836296,
+      "step": 5272
+    },
+    {
+      "epoch": 59.858356940509914,
+      "loss": 0.052924949675798416,
+      "loss_ce": 6.850512727396563e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 904836296,
+      "step": 5272
+    },
+    {
+      "epoch": 59.86968838526912,
+      "grad_norm": 4.0376322760594325,
+      "learning_rate": 5e-06,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 905008360,
+      "step": 5273
+    },
+    {
+      "epoch": 59.86968838526912,
+      "loss": 0.03543814271688461,
+      "loss_ce": 0.00019034132128581405,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 905008360,
+      "step": 5273
+    },
+    {
+      "epoch": 59.88101983002833,
+      "grad_norm": 4.674511053048756,
+      "learning_rate": 5e-06,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 905178776,
+      "step": 5274
+    },
+    {
+      "epoch": 59.88101983002833,
+      "loss": 0.09513377398252487,
+      "loss_ce": 0.00011729764810297638,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 905178776,
+      "step": 5274
+    },
+    {
+      "epoch": 59.892351274787536,
+      "grad_norm": 5.714946555958703,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 905350724,
+      "step": 5275
+    },
+    {
+      "epoch": 59.892351274787536,
+      "loss": 0.10863351821899414,
+      "loss_ce": 0.00015879135753493756,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 905350724,
+      "step": 5275
+    },
+    {
+      "epoch": 59.903682719546744,
+      "grad_norm": 3.1390227953572136,
+      "learning_rate": 5e-06,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 905521936,
+      "step": 5276
+    },
+    {
+      "epoch": 59.903682719546744,
+      "loss": 0.10067965090274811,
+      "loss_ce": 1.741788764775265e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 905521936,
+      "step": 5276
+    },
+    {
+      "epoch": 59.91501416430595,
+      "grad_norm": 2.9155259803237623,
+      "learning_rate": 5e-06,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 905693844,
+      "step": 5277
+    },
+    {
+      "epoch": 59.91501416430595,
+      "loss": 0.047902293503284454,
+      "loss_ce": 3.546860898495652e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 905693844,
+      "step": 5277
+    },
+    {
+      "epoch": 59.92634560906516,
+      "grad_norm": 3.7189901168004087,
+      "learning_rate": 5e-06,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 905864384,
+      "step": 5278
+    },
+    {
+      "epoch": 59.92634560906516,
+      "loss": 0.06744060665369034,
+      "loss_ce": 7.305352482944727e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 905864384,
+      "step": 5278
+    },
+    {
+      "epoch": 59.93767705382436,
+      "grad_norm": 3.336170011931883,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 906036168,
+      "step": 5279
+    },
+    {
+      "epoch": 59.93767705382436,
+      "loss": 0.03950412571430206,
+      "loss_ce": 2.9637725674547255e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 906036168,
+      "step": 5279
+    },
+    {
+      "epoch": 59.94900849858357,
+      "grad_norm": 3.600718726467341,
+      "learning_rate": 5e-06,
+      "loss": 0.059,
+      "num_input_tokens_seen": 906207388,
+      "step": 5280
+    },
+    {
+      "epoch": 59.94900849858357,
+      "loss": 0.03594730421900749,
+      "loss_ce": 7.389260281343013e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 906207388,
+      "step": 5280
+    },
+    {
+      "epoch": 59.960339943342774,
+      "grad_norm": 3.683771661353578,
+      "learning_rate": 5e-06,
+      "loss": 0.0683,
+      "num_input_tokens_seen": 906378016,
+      "step": 5281
+    },
+    {
+      "epoch": 59.960339943342774,
+      "loss": 0.10890936851501465,
+      "loss_ce": 2.265237708343193e-05,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 906378016,
+      "step": 5281
+    },
+    {
+      "epoch": 59.97167138810198,
+      "grad_norm": 3.2901430966928698,
+      "learning_rate": 5e-06,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 906549748,
+      "step": 5282
+    },
+    {
+      "epoch": 59.97167138810198,
+      "loss": 0.031769223511219025,
+      "loss_ce": 6.146270607132465e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 906549748,
+      "step": 5282
+    },
+    {
+      "epoch": 59.98300283286119,
+      "grad_norm": 2.7343532032088333,
+      "learning_rate": 5e-06,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 906719144,
+      "step": 5283
+    },
+    {
+      "epoch": 59.98300283286119,
+      "loss": 0.06259878724813461,
+      "loss_ce": 8.352575241588056e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 906719144,
+      "step": 5283
+    },
+    {
+      "epoch": 59.994334277620396,
+      "grad_norm": 3.524429963076134,
+      "learning_rate": 5e-06,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 906891176,
+      "step": 5284
+    },
+    {
+      "epoch": 59.994334277620396,
+      "loss": 0.12320826202630997,
+      "loss_ce": 2.404916813247837e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 906891176,
+      "step": 5284
+    },
+    {
+      "epoch": 59.994334277620396,
+      "loss": 0.06530926376581192,
+      "loss_ce": 4.7420820919796824e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 906976940,
+      "step": 5284
+    },
+    {
+      "epoch": 60.005665722379604,
+      "grad_norm": 9.944597566678567,
+      "learning_rate": 5e-06,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 907063212,
+      "step": 5285
+    },
+    {
+      "epoch": 60.005665722379604,
+      "loss": 0.07469999045133591,
+      "loss_ce": 6.16243269178085e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 907063212,
+      "step": 5285
+    },
+    {
+      "epoch": 60.01699716713881,
+      "grad_norm": 3.233336838936552,
+      "learning_rate": 5e-06,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 907235352,
+      "step": 5286
+    },
+    {
+      "epoch": 60.01699716713881,
+      "loss": 0.11833537369966507,
+      "loss_ce": 0.0033298793714493513,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 907235352,
+      "step": 5286
+    },
+    {
+      "epoch": 60.02832861189802,
+      "grad_norm": 3.155411714739412,
+      "learning_rate": 5e-06,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 907405296,
+      "step": 5287
+    },
+    {
+      "epoch": 60.02832861189802,
+      "loss": 0.08510617166757584,
+      "loss_ce": 5.368136044126004e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 907405296,
+      "step": 5287
+    },
+    {
+      "epoch": 60.039660056657226,
+      "grad_norm": 3.7161595690155895,
+      "learning_rate": 5e-06,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 907576600,
+      "step": 5288
+    },
+    {
+      "epoch": 60.039660056657226,
+      "loss": 0.03138658404350281,
+      "loss_ce": 9.080891322810203e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 907576600,
+      "step": 5288
+    },
+    {
+      "epoch": 60.05099150141643,
+      "grad_norm": 4.321721882023258,
+      "learning_rate": 5e-06,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 907744644,
+      "step": 5289
+    },
+    {
+      "epoch": 60.05099150141643,
+      "loss": 0.044658586382865906,
+      "loss_ce": 0.00010292250226484612,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 907744644,
+      "step": 5289
+    },
+    {
+      "epoch": 60.06232294617564,
+      "grad_norm": 5.325421575238035,
+      "learning_rate": 5e-06,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 907916728,
+      "step": 5290
+    },
+    {
+      "epoch": 60.06232294617564,
+      "loss": 0.028624074533581734,
+      "loss_ce": 4.436278686625883e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 907916728,
+      "step": 5290
+    },
+    {
+      "epoch": 60.07365439093484,
+      "grad_norm": 3.3122030502063566,
+      "learning_rate": 5e-06,
+      "loss": 0.044,
+      "num_input_tokens_seen": 908088780,
+      "step": 5291
+    },
+    {
+      "epoch": 60.07365439093484,
+      "loss": 0.034543078392744064,
+      "loss_ce": 7.347343489527702e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 908088780,
+      "step": 5291
+    },
+    {
+      "epoch": 60.08498583569405,
+      "grad_norm": 4.7144702983211975,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 908260860,
+      "step": 5292
+    },
+    {
+      "epoch": 60.08498583569405,
+      "loss": 0.06500456482172012,
+      "loss_ce": 0.0008108344045467675,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 908260860,
+      "step": 5292
+    },
+    {
+      "epoch": 60.096317280453256,
+      "grad_norm": 3.0513645968633516,
+      "learning_rate": 5e-06,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 908431884,
+      "step": 5293
+    },
+    {
+      "epoch": 60.096317280453256,
+      "loss": 0.05299338698387146,
+      "loss_ce": 9.11634269868955e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 908431884,
+      "step": 5293
+    },
+    {
+      "epoch": 60.107648725212464,
+      "grad_norm": 3.4499735471378012,
+      "learning_rate": 5e-06,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 908602524,
+      "step": 5294
+    },
+    {
+      "epoch": 60.107648725212464,
+      "loss": 0.04691363126039505,
+      "loss_ce": 9.966388461180031e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 908602524,
+      "step": 5294
+    },
+    {
+      "epoch": 60.11898016997167,
+      "grad_norm": 3.4328010263247335,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 908773032,
+      "step": 5295
+    },
+    {
+      "epoch": 60.11898016997167,
+      "loss": 0.04888949170708656,
+      "loss_ce": 0.0011142236180603504,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 908773032,
+      "step": 5295
+    },
+    {
+      "epoch": 60.13031161473088,
+      "grad_norm": 4.543315143245784,
+      "learning_rate": 5e-06,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 908940228,
+      "step": 5296
+    },
+    {
+      "epoch": 60.13031161473088,
+      "loss": 0.037237007170915604,
+      "loss_ce": 2.082024366245605e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 908940228,
+      "step": 5296
+    },
+    {
+      "epoch": 60.141643059490086,
+      "grad_norm": 4.005370055505921,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 909110940,
+      "step": 5297
+    },
+    {
+      "epoch": 60.141643059490086,
+      "loss": 0.02901434898376465,
+      "loss_ce": 0.0002515295927878469,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 909110940,
+      "step": 5297
+    },
+    {
+      "epoch": 60.15297450424929,
+      "grad_norm": 2.996543023093513,
+      "learning_rate": 5e-06,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 909281900,
+      "step": 5298
+    },
+    {
+      "epoch": 60.15297450424929,
+      "loss": 0.043112967163324356,
+      "loss_ce": 5.266504012979567e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 909281900,
+      "step": 5298
+    },
+    {
+      "epoch": 60.1643059490085,
+      "grad_norm": 3.2434793767078345,
+      "learning_rate": 5e-06,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 909453544,
+      "step": 5299
+    },
+    {
+      "epoch": 60.1643059490085,
+      "loss": 0.03535942733287811,
+      "loss_ce": 5.0591705075930804e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 909453544,
+      "step": 5299
+    },
+    {
+      "epoch": 60.17563739376771,
+      "grad_norm": 4.44120916068292,
+      "learning_rate": 5e-06,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 909624588,
+      "step": 5300
+    },
+    {
+      "epoch": 60.17563739376771,
+      "loss": 0.042779356241226196,
+      "loss_ce": 7.000465848250315e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 909624588,
+      "step": 5300
+    },
+    {
+      "epoch": 60.186968838526916,
+      "grad_norm": 3.5546218501379356,
+      "learning_rate": 5e-06,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 909795992,
+      "step": 5301
+    },
+    {
+      "epoch": 60.186968838526916,
+      "loss": 0.02905501052737236,
+      "loss_ce": 4.8052814236143604e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 909795992,
+      "step": 5301
+    },
+    {
+      "epoch": 60.198300283286116,
+      "grad_norm": 3.6992703774051408,
+      "learning_rate": 5e-06,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 909967688,
+      "step": 5302
+    },
+    {
+      "epoch": 60.198300283286116,
+      "loss": 0.03620871156454086,
+      "loss_ce": 4.538108260021545e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 909967688,
+      "step": 5302
+    },
+    {
+      "epoch": 60.20963172804532,
+      "grad_norm": 3.7704361175605765,
+      "learning_rate": 5e-06,
+      "loss": 0.1169,
+      "num_input_tokens_seen": 910138364,
+      "step": 5303
+    },
+    {
+      "epoch": 60.20963172804532,
+      "loss": 0.2508850693702698,
+      "loss_ce": 9.162837523035705e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 910138364,
+      "step": 5303
+    },
+    {
+      "epoch": 60.22096317280453,
+      "grad_norm": 4.082134267957282,
+      "learning_rate": 5e-06,
+      "loss": 0.075,
+      "num_input_tokens_seen": 910310248,
+      "step": 5304
+    },
+    {
+      "epoch": 60.22096317280453,
+      "loss": 0.04449412226676941,
+      "loss_ce": 7.578609802294523e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 910310248,
+      "step": 5304
+    },
+    {
+      "epoch": 60.23229461756374,
+      "grad_norm": 3.2077888659000227,
+      "learning_rate": 5e-06,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 910481940,
+      "step": 5305
+    },
+    {
+      "epoch": 60.23229461756374,
+      "loss": 0.041572362184524536,
+      "loss_ce": 0.00011423211981309578,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 910481940,
+      "step": 5305
+    },
+    {
+      "epoch": 60.243626062322946,
+      "grad_norm": 52.62992156400096,
+      "learning_rate": 5e-06,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 910650624,
+      "step": 5306
+    },
+    {
+      "epoch": 60.243626062322946,
+      "loss": 0.03883575648069382,
+      "loss_ce": 7.843222556402907e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 910650624,
+      "step": 5306
+    },
+    {
+      "epoch": 60.25495750708215,
+      "grad_norm": 60.626483052214766,
+      "learning_rate": 5e-06,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 910822336,
+      "step": 5307
+    },
+    {
+      "epoch": 60.25495750708215,
+      "loss": 0.04873540252447128,
+      "loss_ce": 0.00021245228708721697,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 910822336,
+      "step": 5307
+    },
+    {
+      "epoch": 60.26628895184136,
+      "grad_norm": 3.7409132307538333,
+      "learning_rate": 5e-06,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 910994308,
+      "step": 5308
+    },
+    {
+      "epoch": 60.26628895184136,
+      "loss": 0.08972421288490295,
+      "loss_ce": 4.830794205190614e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 910994308,
+      "step": 5308
+    },
+    {
+      "epoch": 60.27762039660057,
+      "grad_norm": 4.172696941141727,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 911166440,
+      "step": 5309
+    },
+    {
+      "epoch": 60.27762039660057,
+      "loss": 0.0516594834625721,
+      "loss_ce": 5.4256794101092964e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 911166440,
+      "step": 5309
+    },
+    {
+      "epoch": 60.288951841359776,
+      "grad_norm": 4.892729380505663,
+      "learning_rate": 5e-06,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 911338324,
+      "step": 5310
+    },
+    {
+      "epoch": 60.288951841359776,
+      "loss": 0.041725482791662216,
+      "loss_ce": 0.00019106024410575628,
+      "loss_iou": 0.515625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 911338324,
+      "step": 5310
+    },
+    {
+      "epoch": 60.30028328611898,
+      "grad_norm": 4.534346362503417,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 911509600,
+      "step": 5311
+    },
+    {
+      "epoch": 60.30028328611898,
+      "loss": 0.059051379561424255,
+      "loss_ce": 0.00013719552953261882,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 911509600,
+      "step": 5311
+    },
+    {
+      "epoch": 60.31161473087819,
+      "grad_norm": 3.552033149940994,
+      "learning_rate": 5e-06,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 911681232,
+      "step": 5312
+    },
+    {
+      "epoch": 60.31161473087819,
+      "loss": 0.10575592517852783,
+      "loss_ce": 5.8291636378271505e-05,
+      "loss_iou": 0.08251953125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 911681232,
+      "step": 5312
+    },
+    {
+      "epoch": 60.32294617563739,
+      "grad_norm": 2.938119773227793,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 911853084,
+      "step": 5313
+    },
+    {
+      "epoch": 60.32294617563739,
+      "loss": 0.07814521342515945,
+      "loss_ce": 0.00014227983774617314,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 911853084,
+      "step": 5313
+    },
+    {
+      "epoch": 60.3342776203966,
+      "grad_norm": 2.9562902498742267,
+      "learning_rate": 5e-06,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 912023336,
+      "step": 5314
+    },
+    {
+      "epoch": 60.3342776203966,
+      "loss": 0.04107864946126938,
+      "loss_ce": 7.828190427972004e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 912023336,
+      "step": 5314
+    },
+    {
+      "epoch": 60.345609065155806,
+      "grad_norm": 7.32597120194911,
+      "learning_rate": 5e-06,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 912195080,
+      "step": 5315
+    },
+    {
+      "epoch": 60.345609065155806,
+      "loss": 0.03294748812913895,
+      "loss_ce": 4.953733514412306e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 912195080,
+      "step": 5315
+    },
+    {
+      "epoch": 60.35694050991501,
+      "grad_norm": 3.6730861898071594,
+      "learning_rate": 5e-06,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 912366752,
+      "step": 5316
+    },
+    {
+      "epoch": 60.35694050991501,
+      "loss": 0.05385345220565796,
+      "loss_ce": 6.622144428547472e-05,
+      "loss_iou": 0.11865234375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 912366752,
+      "step": 5316
+    },
+    {
+      "epoch": 60.36827195467422,
+      "grad_norm": 3.6236791811335443,
+      "learning_rate": 5e-06,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 912538200,
+      "step": 5317
+    },
+    {
+      "epoch": 60.36827195467422,
+      "loss": 0.04094957932829857,
+      "loss_ce": 0.00013231889170128852,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 912538200,
+      "step": 5317
+    },
+    {
+      "epoch": 60.37960339943343,
+      "grad_norm": 6.647115198108427,
+      "learning_rate": 5e-06,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 912709472,
+      "step": 5318
+    },
+    {
+      "epoch": 60.37960339943343,
+      "loss": 0.04388439282774925,
+      "loss_ce": 7.6408134191297e-05,
+      "loss_iou": 0.1220703125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 912709472,
+      "step": 5318
+    },
+    {
+      "epoch": 60.390934844192635,
+      "grad_norm": 3.0380153621279327,
+      "learning_rate": 5e-06,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 912881332,
+      "step": 5319
+    },
+    {
+      "epoch": 60.390934844192635,
+      "loss": 0.03983393311500549,
+      "loss_ce": 0.00010004534851759672,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 912881332,
+      "step": 5319
+    },
+    {
+      "epoch": 60.40226628895184,
+      "grad_norm": 3.9709569958688324,
+      "learning_rate": 5e-06,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 913052384,
+      "step": 5320
+    },
+    {
+      "epoch": 60.40226628895184,
+      "loss": 0.06085922196507454,
+      "loss_ce": 5.294629227137193e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 913052384,
+      "step": 5320
+    },
+    {
+      "epoch": 60.41359773371105,
+      "grad_norm": 3.9302525097261847,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 913223456,
+      "step": 5321
+    },
+    {
+      "epoch": 60.41359773371105,
+      "loss": 0.04768989607691765,
+      "loss_ce": 3.669855868793093e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 913223456,
+      "step": 5321
+    },
+    {
+      "epoch": 60.42492917847026,
+      "grad_norm": 3.5528166998725377,
+      "learning_rate": 5e-06,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 913394428,
+      "step": 5322
+    },
+    {
+      "epoch": 60.42492917847026,
+      "loss": 0.0415625125169754,
+      "loss_ce": 0.00010437953460495919,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 913394428,
+      "step": 5322
+    },
+    {
+      "epoch": 60.436260623229465,
+      "grad_norm": 3.7763049516416567,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 913566468,
+      "step": 5323
+    },
+    {
+      "epoch": 60.436260623229465,
+      "loss": 0.06610424071550369,
+      "loss_ce": 0.001162838307209313,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 913566468,
+      "step": 5323
+    },
+    {
+      "epoch": 60.447592067988666,
+      "grad_norm": 4.832197538193646,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 913738448,
+      "step": 5324
+    },
+    {
+      "epoch": 60.447592067988666,
+      "loss": 0.03582945838570595,
+      "loss_ce": 0.00013915127783548087,
+      "loss_iou": 0.46875,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 913738448,
+      "step": 5324
+    },
+    {
+      "epoch": 60.45892351274787,
+      "grad_norm": 3.9103831486990317,
+      "learning_rate": 5e-06,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 913910044,
+      "step": 5325
+    },
+    {
+      "epoch": 60.45892351274787,
+      "loss": 0.056947849690914154,
+      "loss_ce": 0.0001698917185422033,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 913910044,
+      "step": 5325
+    },
+    {
+      "epoch": 60.47025495750708,
+      "grad_norm": 4.331391604625984,
+      "learning_rate": 5e-06,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 914080700,
+      "step": 5326
+    },
+    {
+      "epoch": 60.47025495750708,
+      "loss": 0.050633084028959274,
+      "loss_ce": 5.019925447413698e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 914080700,
+      "step": 5326
+    },
+    {
+      "epoch": 60.48158640226629,
+      "grad_norm": 3.0465676268319557,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 914252472,
+      "step": 5327
+    },
+    {
+      "epoch": 60.48158640226629,
+      "loss": 0.039646558463573456,
+      "loss_ce": 6.526013748953119e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 914252472,
+      "step": 5327
+    },
+    {
+      "epoch": 60.492917847025495,
+      "grad_norm": 4.63397528210791,
+      "learning_rate": 5e-06,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 914422816,
+      "step": 5328
+    },
+    {
+      "epoch": 60.492917847025495,
+      "loss": 0.07953919470310211,
+      "loss_ce": 5.616456473944709e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 914422816,
+      "step": 5328
+    },
+    {
+      "epoch": 60.5042492917847,
+      "grad_norm": 3.417403666928378,
+      "learning_rate": 5e-06,
+      "loss": 0.0857,
+      "num_input_tokens_seen": 914594468,
+      "step": 5329
+    },
+    {
+      "epoch": 60.5042492917847,
+      "loss": 0.12426836043596268,
+      "loss_ce": 0.001282525365240872,
+      "loss_iou": 0.3125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 914594468,
+      "step": 5329
+    },
+    {
+      "epoch": 60.51558073654391,
+      "grad_norm": 3.6345513073395743,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 914765384,
+      "step": 5330
+    },
+    {
+      "epoch": 60.51558073654391,
+      "loss": 0.09104451537132263,
+      "loss_ce": 7.161663961596787e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 914765384,
+      "step": 5330
+    },
+    {
+      "epoch": 60.52691218130312,
+      "grad_norm": 3.428790681170709,
+      "learning_rate": 5e-06,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 914936652,
+      "step": 5331
+    },
+    {
+      "epoch": 60.52691218130312,
+      "loss": 0.033823419362306595,
+      "loss_ce": 4.0460443415213376e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 914936652,
+      "step": 5331
+    },
+    {
+      "epoch": 60.538243626062325,
+      "grad_norm": 3.3356038372650394,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 915107168,
+      "step": 5332
+    },
+    {
+      "epoch": 60.538243626062325,
+      "loss": 0.045993655920028687,
+      "loss_ce": 0.00023254757979884744,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 915107168,
+      "step": 5332
+    },
+    {
+      "epoch": 60.54957507082153,
+      "grad_norm": 3.844161256637789,
+      "learning_rate": 5e-06,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 915279200,
+      "step": 5333
+    },
+    {
+      "epoch": 60.54957507082153,
+      "loss": 0.06420257687568665,
+      "loss_ce": 0.00013092387234792113,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 915279200,
+      "step": 5333
+    },
+    {
+      "epoch": 60.56090651558074,
+      "grad_norm": 3.5094774430612423,
+      "learning_rate": 5e-06,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 915450712,
+      "step": 5334
+    },
+    {
+      "epoch": 60.56090651558074,
+      "loss": 0.039232831448316574,
+      "loss_ce": 0.00018559047020971775,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 915450712,
+      "step": 5334
+    },
+    {
+      "epoch": 60.57223796033994,
+      "grad_norm": 3.6725382736535677,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 915622488,
+      "step": 5335
+    },
+    {
+      "epoch": 60.57223796033994,
+      "loss": 0.05054699629545212,
+      "loss_ce": 2.5146975531242788e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 915622488,
+      "step": 5335
+    },
+    {
+      "epoch": 60.58356940509915,
+      "grad_norm": 3.1941661988372863,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 915794368,
+      "step": 5336
+    },
+    {
+      "epoch": 60.58356940509915,
+      "loss": 0.05514456331729889,
+      "loss_ce": 9.085034253075719e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 915794368,
+      "step": 5336
+    },
+    {
+      "epoch": 60.594900849858355,
+      "grad_norm": 2.985224376866991,
+      "learning_rate": 5e-06,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 915965932,
+      "step": 5337
+    },
+    {
+      "epoch": 60.594900849858355,
+      "loss": 0.06457200646400452,
+      "loss_ce": 8.836544293444604e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 915965932,
+      "step": 5337
+    },
+    {
+      "epoch": 60.60623229461756,
+      "grad_norm": 3.1195290302707748,
+      "learning_rate": 5e-06,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 916137716,
+      "step": 5338
+    },
+    {
+      "epoch": 60.60623229461756,
+      "loss": 0.03704243153333664,
+      "loss_ce": 7.038851617835462e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 916137716,
+      "step": 5338
+    },
+    {
+      "epoch": 60.61756373937677,
+      "grad_norm": 3.144355509678711,
+      "learning_rate": 5e-06,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 916308984,
+      "step": 5339
+    },
+    {
+      "epoch": 60.61756373937677,
+      "loss": 0.041377291083335876,
+      "loss_ce": 1.0714086783991661e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 916308984,
+      "step": 5339
+    },
+    {
+      "epoch": 60.62889518413598,
+      "grad_norm": 3.9728630362030253,
+      "learning_rate": 5e-06,
+      "loss": 0.0644,
+      "num_input_tokens_seen": 916480872,
+      "step": 5340
+    },
+    {
+      "epoch": 60.62889518413598,
+      "loss": 0.06372785568237305,
+      "loss_ce": 3.766483496292494e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 916480872,
+      "step": 5340
+    },
+    {
+      "epoch": 60.640226628895185,
+      "grad_norm": 3.434787622281573,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 916651212,
+      "step": 5341
+    },
+    {
+      "epoch": 60.640226628895185,
+      "loss": 0.07078741490840912,
+      "loss_ce": 7.818300946382806e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 916651212,
+      "step": 5341
+    },
+    {
+      "epoch": 60.65155807365439,
+      "grad_norm": 2.99494251309212,
+      "learning_rate": 5e-06,
+      "loss": 0.049,
+      "num_input_tokens_seen": 916822764,
+      "step": 5342
+    },
+    {
+      "epoch": 60.65155807365439,
+      "loss": 0.05739685520529747,
+      "loss_ce": 5.432831676444039e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 916822764,
+      "step": 5342
+    },
+    {
+      "epoch": 60.6628895184136,
+      "grad_norm": 2.866935456467115,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 916994532,
+      "step": 5343
+    },
+    {
+      "epoch": 60.6628895184136,
+      "loss": 0.05307865887880325,
+      "loss_ce": 0.0002527314645703882,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 916994532,
+      "step": 5343
+    },
+    {
+      "epoch": 60.67422096317281,
+      "grad_norm": 2.8775239754470103,
+      "learning_rate": 5e-06,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 917164824,
+      "step": 5344
+    },
+    {
+      "epoch": 60.67422096317281,
+      "loss": 0.11749410629272461,
+      "loss_ce": 3.1954506994225085e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 917164824,
+      "step": 5344
+    },
+    {
+      "epoch": 60.685552407932015,
+      "grad_norm": 4.672608535448603,
+      "learning_rate": 5e-06,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 917336516,
+      "step": 5345
+    },
+    {
+      "epoch": 60.685552407932015,
+      "loss": 0.044775884598493576,
+      "loss_ce": 6.763250712538138e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 917336516,
+      "step": 5345
+    },
+    {
+      "epoch": 60.696883852691215,
+      "grad_norm": 3.436934965950403,
+      "learning_rate": 5e-06,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 917507284,
+      "step": 5346
+    },
+    {
+      "epoch": 60.696883852691215,
+      "loss": 0.0323917530477047,
+      "loss_ce": 7.36360889277421e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 917507284,
+      "step": 5346
+    },
+    {
+      "epoch": 60.70821529745042,
+      "grad_norm": 2.7588964267339633,
+      "learning_rate": 5e-06,
+      "loss": 0.039,
+      "num_input_tokens_seen": 917679244,
+      "step": 5347
+    },
+    {
+      "epoch": 60.70821529745042,
+      "loss": 0.02689727395772934,
+      "loss_ce": 6.46931875962764e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 917679244,
+      "step": 5347
+    },
+    {
+      "epoch": 60.71954674220963,
+      "grad_norm": 3.896795808320474,
+      "learning_rate": 5e-06,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 917850096,
+      "step": 5348
+    },
+    {
+      "epoch": 60.71954674220963,
+      "loss": 0.06668774783611298,
+      "loss_ce": 6.78746000630781e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 917850096,
+      "step": 5348
+    },
+    {
+      "epoch": 60.73087818696884,
+      "grad_norm": 3.5464775819922627,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 918021984,
+      "step": 5349
+    },
+    {
+      "epoch": 60.73087818696884,
+      "loss": 0.04377872869372368,
+      "loss_ce": 4.703795275418088e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 918021984,
+      "step": 5349
+    },
+    {
+      "epoch": 60.742209631728045,
+      "grad_norm": 3.5358960923237963,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 918193020,
+      "step": 5350
+    },
+    {
+      "epoch": 60.742209631728045,
+      "loss": 0.04534516483545303,
+      "loss_ce": 4.1820010665105656e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 918193020,
+      "step": 5350
+    },
+    {
+      "epoch": 60.75354107648725,
+      "grad_norm": 3.459368774951292,
+      "learning_rate": 5e-06,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 918363096,
+      "step": 5351
+    },
+    {
+      "epoch": 60.75354107648725,
+      "loss": 0.08535854518413544,
+      "loss_ce": 6.191064312588423e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 918363096,
+      "step": 5351
+    },
+    {
+      "epoch": 60.76487252124646,
+      "grad_norm": 3.6421239954789297,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 918534720,
+      "step": 5352
+    },
+    {
+      "epoch": 60.76487252124646,
+      "loss": 0.04928767308592796,
+      "loss_ce": 1.704102396615781e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 918534720,
+      "step": 5352
+    },
+    {
+      "epoch": 60.77620396600567,
+      "grad_norm": 8.577733510934467,
+      "learning_rate": 5e-06,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 918704856,
+      "step": 5353
+    },
+    {
+      "epoch": 60.77620396600567,
+      "loss": 0.09034101665019989,
+      "loss_ce": 0.00010054362792288885,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 918704856,
+      "step": 5353
+    },
+    {
+      "epoch": 60.787535410764875,
+      "grad_norm": 4.279460044950278,
+      "learning_rate": 5e-06,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 918875896,
+      "step": 5354
+    },
+    {
+      "epoch": 60.787535410764875,
+      "loss": 0.11280453205108643,
+      "loss_ce": 2.682180274860002e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 918875896,
+      "step": 5354
+    },
+    {
+      "epoch": 60.79886685552408,
+      "grad_norm": 3.1644655289872343,
+      "learning_rate": 5e-06,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 919046916,
+      "step": 5355
+    },
+    {
+      "epoch": 60.79886685552408,
+      "loss": 0.02776140347123146,
+      "loss_ce": 0.0001124780610552989,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 919046916,
+      "step": 5355
+    },
+    {
+      "epoch": 60.81019830028329,
+      "grad_norm": 3.4591929620229958,
+      "learning_rate": 5e-06,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 919218324,
+      "step": 5356
+    },
+    {
+      "epoch": 60.81019830028329,
+      "loss": 0.027796726673841476,
+      "loss_ce": 2.5732038920978084e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 919218324,
+      "step": 5356
+    },
+    {
+      "epoch": 60.82152974504249,
+      "grad_norm": 3.8068347489283454,
+      "learning_rate": 5e-06,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 919389136,
+      "step": 5357
+    },
+    {
+      "epoch": 60.82152974504249,
+      "loss": 0.08701546490192413,
+      "loss_ce": 4.0369253838434815e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 919389136,
+      "step": 5357
+    },
+    {
+      "epoch": 60.8328611898017,
+      "grad_norm": 3.838767399604408,
+      "learning_rate": 5e-06,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 919561256,
+      "step": 5358
+    },
+    {
+      "epoch": 60.8328611898017,
+      "loss": 0.04236666113138199,
+      "loss_ce": 3.8779722672188655e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 919561256,
+      "step": 5358
+    },
+    {
+      "epoch": 60.844192634560905,
+      "grad_norm": 3.5630442421661446,
+      "learning_rate": 5e-06,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 919733196,
+      "step": 5359
+    },
+    {
+      "epoch": 60.844192634560905,
+      "loss": 0.043702252209186554,
+      "loss_ce": 1.634146974538453e-05,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 919733196,
+      "step": 5359
+    },
+    {
+      "epoch": 60.85552407932011,
+      "grad_norm": 3.5274798459111123,
+      "learning_rate": 5e-06,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 919903784,
+      "step": 5360
+    },
+    {
+      "epoch": 60.85552407932011,
+      "loss": 0.029463166370987892,
+      "loss_ce": 8.999784768093377e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 919903784,
+      "step": 5360
+    },
+    {
+      "epoch": 60.86685552407932,
+      "grad_norm": 3.065690085216179,
+      "learning_rate": 5e-06,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 920075336,
+      "step": 5361
+    },
+    {
+      "epoch": 60.86685552407932,
+      "loss": 0.05938621237874031,
+      "loss_ce": 2.9526741855079308e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 920075336,
+      "step": 5361
+    },
+    {
+      "epoch": 60.87818696883853,
+      "grad_norm": 3.6584399703373247,
+      "learning_rate": 5e-06,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 920247412,
+      "step": 5362
+    },
+    {
+      "epoch": 60.87818696883853,
+      "loss": 0.04484322667121887,
+      "loss_ce": 2.81646462099161e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 920247412,
+      "step": 5362
+    },
+    {
+      "epoch": 60.889518413597735,
+      "grad_norm": 3.4592669651519206,
+      "learning_rate": 5e-06,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 920418924,
+      "step": 5363
+    },
+    {
+      "epoch": 60.889518413597735,
+      "loss": 0.038255900144577026,
+      "loss_ce": 9.366725134896114e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 920418924,
+      "step": 5363
+    },
+    {
+      "epoch": 60.90084985835694,
+      "grad_norm": 3.2966777464399364,
+      "learning_rate": 5e-06,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 920589732,
+      "step": 5364
+    },
+    {
+      "epoch": 60.90084985835694,
+      "loss": 0.03835352510213852,
+      "loss_ce": 2.3445025362889282e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 920589732,
+      "step": 5364
+    },
+    {
+      "epoch": 60.91218130311615,
+      "grad_norm": 3.3396297880577026,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 920760848,
+      "step": 5365
+    },
+    {
+      "epoch": 60.91218130311615,
+      "loss": 0.04530102759599686,
+      "loss_ce": 8.923888526624069e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 920760848,
+      "step": 5365
+    },
+    {
+      "epoch": 60.92351274787536,
+      "grad_norm": 3.6937501855135495,
+      "learning_rate": 5e-06,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 920932928,
+      "step": 5366
+    },
+    {
+      "epoch": 60.92351274787536,
+      "loss": 0.051009755581617355,
+      "loss_ce": 2.25132389459759e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 920932928,
+      "step": 5366
+    },
+    {
+      "epoch": 60.934844192634564,
+      "grad_norm": 3.669876031423358,
+      "learning_rate": 5e-06,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 921104200,
+      "step": 5367
+    },
+    {
+      "epoch": 60.934844192634564,
+      "loss": 0.033807940781116486,
+      "loss_ce": 4.0241549868369475e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 921104200,
+      "step": 5367
+    },
+    {
+      "epoch": 60.946175637393765,
+      "grad_norm": 3.578982783201232,
+      "learning_rate": 5e-06,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 921276148,
+      "step": 5368
+    },
+    {
+      "epoch": 60.946175637393765,
+      "loss": 0.06229458749294281,
+      "loss_ce": 0.0010000327602028847,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 921276148,
+      "step": 5368
+    },
+    {
+      "epoch": 60.95750708215297,
+      "grad_norm": 3.676618433402416,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 921448072,
+      "step": 5369
+    },
+    {
+      "epoch": 60.95750708215297,
+      "loss": 0.04133474826812744,
+      "loss_ce": 5.9723417507484555e-05,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 921448072,
+      "step": 5369
+    },
+    {
+      "epoch": 60.96883852691218,
+      "grad_norm": 3.895485292129811,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 921619728,
+      "step": 5370
+    },
+    {
+      "epoch": 60.96883852691218,
+      "loss": 0.14817720651626587,
+      "loss_ce": 7.540077058365569e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 921619728,
+      "step": 5370
+    },
+    {
+      "epoch": 60.98016997167139,
+      "grad_norm": 3.331073725054409,
+      "learning_rate": 5e-06,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 921790584,
+      "step": 5371
+    },
+    {
+      "epoch": 60.98016997167139,
+      "loss": 0.03188817575573921,
+      "loss_ce": 0.00011937714589294046,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 921790584,
+      "step": 5371
+    },
+    {
+      "epoch": 60.991501416430594,
+      "grad_norm": 2.958003639440145,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 921962312,
+      "step": 5372
+    },
+    {
+      "epoch": 60.991501416430594,
+      "loss": 0.11879308521747589,
+      "loss_ce": 9.496501297689974e-05,
+      "loss_iou": 0.04638671875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 921962312,
+      "step": 5372
+    },
+    {
+      "epoch": 60.991501416430594,
+      "loss": 0.06609611213207245,
+      "loss_ce": 7.133104372769594e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 922091072,
+      "step": 5372
+    },
+    {
+      "epoch": 61.0028328611898,
+      "grad_norm": 2.618421301790613,
+      "learning_rate": 5e-06,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 922132364,
+      "step": 5373
+    },
+    {
+      "epoch": 61.0028328611898,
+      "loss": 0.05547528713941574,
+      "loss_ce": 0.00025372812524437904,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 922132364,
+      "step": 5373
+    },
+    {
+      "epoch": 61.01416430594901,
+      "grad_norm": 2.7079544177949773,
+      "learning_rate": 5e-06,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 922303744,
+      "step": 5374
+    },
+    {
+      "epoch": 61.01416430594901,
+      "loss": 0.08013202995061874,
+      "loss_ce": 2.3384365704259835e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 922303744,
+      "step": 5374
+    },
+    {
+      "epoch": 61.02549575070822,
+      "grad_norm": 3.094933194899561,
+      "learning_rate": 5e-06,
+      "loss": 0.045,
+      "num_input_tokens_seen": 922474460,
+      "step": 5375
+    },
+    {
+      "epoch": 61.02549575070822,
+      "loss": 0.04491021856665611,
+      "loss_ce": 6.463790487032384e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 922474460,
+      "step": 5375
+    },
+    {
+      "epoch": 61.036827195467424,
+      "grad_norm": 3.970055090364111,
+      "learning_rate": 5e-06,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 922646220,
+      "step": 5376
+    },
+    {
+      "epoch": 61.036827195467424,
+      "loss": 0.03952575474977493,
+      "loss_ce": 8.178333519026637e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 922646220,
+      "step": 5376
+    },
+    {
+      "epoch": 61.04815864022663,
+      "grad_norm": 3.8113416756828795,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 922816960,
+      "step": 5377
+    },
+    {
+      "epoch": 61.04815864022663,
+      "loss": 0.05297220125794411,
+      "loss_ce": 3.946200740756467e-05,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 922816960,
+      "step": 5377
+    },
+    {
+      "epoch": 61.05949008498584,
+      "grad_norm": 4.287461186986569,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 922988924,
+      "step": 5378
+    },
+    {
+      "epoch": 61.05949008498584,
+      "loss": 0.08201511204242706,
+      "loss_ce": 4.490065111895092e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 922988924,
+      "step": 5378
+    },
+    {
+      "epoch": 61.07082152974504,
+      "grad_norm": 2.99957455436037,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 923160468,
+      "step": 5379
+    },
+    {
+      "epoch": 61.07082152974504,
+      "loss": 0.0688992589712143,
+      "loss_ce": 8.212377724703401e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 923160468,
+      "step": 5379
+    },
+    {
+      "epoch": 61.08215297450425,
+      "grad_norm": 4.947147071926928,
+      "learning_rate": 5e-06,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 923332612,
+      "step": 5380
+    },
+    {
+      "epoch": 61.08215297450425,
+      "loss": 0.0749097466468811,
+      "loss_ce": 7.301415462279692e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 923332612,
+      "step": 5380
+    },
+    {
+      "epoch": 61.093484419263454,
+      "grad_norm": 2.9007247427808136,
+      "learning_rate": 5e-06,
+      "loss": 0.063,
+      "num_input_tokens_seen": 923504372,
+      "step": 5381
+    },
+    {
+      "epoch": 61.093484419263454,
+      "loss": 0.043455302715301514,
+      "loss_ce": 2.8789188945665956e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 923504372,
+      "step": 5381
+    },
+    {
+      "epoch": 61.10481586402266,
+      "grad_norm": 8.553309062017176,
+      "learning_rate": 5e-06,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 923675068,
+      "step": 5382
+    },
+    {
+      "epoch": 61.10481586402266,
+      "loss": 0.08925221860408783,
+      "loss_ce": 3.407484109629877e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 923675068,
+      "step": 5382
+    },
+    {
+      "epoch": 61.11614730878187,
+      "grad_norm": 2.7025777183912227,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 923845204,
+      "step": 5383
+    },
+    {
+      "epoch": 61.11614730878187,
+      "loss": 0.03930659964680672,
+      "loss_ce": 0.00012202812649775296,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 923845204,
+      "step": 5383
+    },
+    {
+      "epoch": 61.12747875354108,
+      "grad_norm": 2.4315243336572014,
+      "learning_rate": 5e-06,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 924017264,
+      "step": 5384
+    },
+    {
+      "epoch": 61.12747875354108,
+      "loss": 0.037681084126234055,
+      "loss_ce": 2.2392483515432104e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 924017264,
+      "step": 5384
+    },
+    {
+      "epoch": 61.138810198300284,
+      "grad_norm": 2.9323200141645396,
+      "learning_rate": 5e-06,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 924187328,
+      "step": 5385
+    },
+    {
+      "epoch": 61.138810198300284,
+      "loss": 0.029375692829489708,
+      "loss_ce": 4.8299727495759726e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 924187328,
+      "step": 5385
+    },
+    {
+      "epoch": 61.15014164305949,
+      "grad_norm": 3.8294529155714025,
+      "learning_rate": 5e-06,
+      "loss": 0.051,
+      "num_input_tokens_seen": 924358196,
+      "step": 5386
+    },
+    {
+      "epoch": 61.15014164305949,
+      "loss": 0.0677463710308075,
+      "loss_ce": 2.7861919079441577e-05,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 924358196,
+      "step": 5386
+    },
+    {
+      "epoch": 61.1614730878187,
+      "grad_norm": 3.535512680210106,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 924530108,
+      "step": 5387
+    },
+    {
+      "epoch": 61.1614730878187,
+      "loss": 0.0360972099006176,
+      "loss_ce": 2.5431319954805076e-05,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 924530108,
+      "step": 5387
+    },
+    {
+      "epoch": 61.172804532577906,
+      "grad_norm": 3.659544842438588,
+      "learning_rate": 5e-06,
+      "loss": 0.068,
+      "num_input_tokens_seen": 924700868,
+      "step": 5388
+    },
+    {
+      "epoch": 61.172804532577906,
+      "loss": 0.04517899453639984,
+      "loss_ce": 4.3497555452631786e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 924700868,
+      "step": 5388
+    },
+    {
+      "epoch": 61.184135977337114,
+      "grad_norm": 3.5042503761228523,
+      "learning_rate": 5e-06,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 924871528,
+      "step": 5389
+    },
+    {
+      "epoch": 61.184135977337114,
+      "loss": 0.07018998265266418,
+      "loss_ce": 1.4817609553574584e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 924871528,
+      "step": 5389
+    },
+    {
+      "epoch": 61.195467422096314,
+      "grad_norm": 3.7407213081920694,
+      "learning_rate": 5e-06,
+      "loss": 0.04,
+      "num_input_tokens_seen": 925043544,
+      "step": 5390
+    },
+    {
+      "epoch": 61.195467422096314,
+      "loss": 0.03325444832444191,
+      "loss_ce": 8.184273610822856e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 925043544,
+      "step": 5390
+    },
+    {
+      "epoch": 61.20679886685552,
+      "grad_norm": 3.372763482189903,
+      "learning_rate": 5e-06,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 925213444,
+      "step": 5391
+    },
+    {
+      "epoch": 61.20679886685552,
+      "loss": 0.04542173072695732,
+      "loss_ce": 0.0001641612616367638,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 925213444,
+      "step": 5391
+    },
+    {
+      "epoch": 61.21813031161473,
+      "grad_norm": 3.4511652855575896,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 925383564,
+      "step": 5392
+    },
+    {
+      "epoch": 61.21813031161473,
+      "loss": 0.03346575051546097,
+      "loss_ce": 6.425984611269087e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 925383564,
+      "step": 5392
+    },
+    {
+      "epoch": 61.22946175637394,
+      "grad_norm": 4.378422567240052,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 925555040,
+      "step": 5393
+    },
+    {
+      "epoch": 61.22946175637394,
+      "loss": 0.04979715496301651,
+      "loss_ce": 3.8242134905885905e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 925555040,
+      "step": 5393
+    },
+    {
+      "epoch": 61.240793201133144,
+      "grad_norm": 9.225634993019883,
+      "learning_rate": 5e-06,
+      "loss": 0.065,
+      "num_input_tokens_seen": 925725252,
+      "step": 5394
+    },
+    {
+      "epoch": 61.240793201133144,
+      "loss": 0.04012008756399155,
+      "loss_ce": 6.576583837158978e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 925725252,
+      "step": 5394
+    },
+    {
+      "epoch": 61.25212464589235,
+      "grad_norm": 4.097715928365581,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 925897652,
+      "step": 5395
+    },
+    {
+      "epoch": 61.25212464589235,
+      "loss": 0.08319907635450363,
+      "loss_ce": 3.867210398311727e-05,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 925897652,
+      "step": 5395
+    },
+    {
+      "epoch": 61.26345609065156,
+      "grad_norm": 7.551747478651342,
+      "learning_rate": 5e-06,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 926069616,
+      "step": 5396
+    },
+    {
+      "epoch": 61.26345609065156,
+      "loss": 0.07369707524776459,
+      "loss_ce": 0.00025651822215877473,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 926069616,
+      "step": 5396
+    },
+    {
+      "epoch": 61.274787535410766,
+      "grad_norm": 3.6511756459188347,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 926241012,
+      "step": 5397
+    },
+    {
+      "epoch": 61.274787535410766,
+      "loss": 0.04547043517231941,
+      "loss_ce": 6.027755807735957e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 926241012,
+      "step": 5397
+    },
+    {
+      "epoch": 61.286118980169974,
+      "grad_norm": 4.382312423281042,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 926412948,
+      "step": 5398
+    },
+    {
+      "epoch": 61.286118980169974,
+      "loss": 0.030807040631771088,
+      "loss_ce": 4.5323533413466066e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 926412948,
+      "step": 5398
+    },
+    {
+      "epoch": 61.29745042492918,
+      "grad_norm": 3.5702317902278486,
+      "learning_rate": 5e-06,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 926584352,
+      "step": 5399
+    },
+    {
+      "epoch": 61.29745042492918,
+      "loss": 0.07534250617027283,
+      "loss_ce": 7.09006781107746e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 926584352,
+      "step": 5399
+    },
+    {
+      "epoch": 61.30878186968839,
+      "grad_norm": 4.353963779393881,
+      "learning_rate": 5e-06,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 926756256,
+      "step": 5400
+    },
+    {
+      "epoch": 61.30878186968839,
+      "loss": 0.03362511098384857,
+      "loss_ce": 4.0518236346542835e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 926756256,
+      "step": 5400
+    },
+    {
+      "epoch": 61.32011331444759,
+      "grad_norm": 4.2806917431010145,
+      "learning_rate": 5e-06,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 926925480,
+      "step": 5401
+    },
+    {
+      "epoch": 61.32011331444759,
+      "loss": 0.06202850490808487,
+      "loss_ce": 0.00019989105931017548,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 926925480,
+      "step": 5401
+    },
+    {
+      "epoch": 61.331444759206796,
+      "grad_norm": 4.184087435412052,
+      "learning_rate": 5e-06,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 927096704,
+      "step": 5402
+    },
+    {
+      "epoch": 61.331444759206796,
+      "loss": 0.03770472854375839,
+      "loss_ce": 4.6038785512791947e-05,
+      "loss_iou": 0.5625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 927096704,
+      "step": 5402
+    },
+    {
+      "epoch": 61.342776203966004,
+      "grad_norm": 4.6483500108472295,
+      "learning_rate": 5e-06,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 927268792,
+      "step": 5403
+    },
+    {
+      "epoch": 61.342776203966004,
+      "loss": 0.041017964482307434,
+      "loss_ce": 1.7597076293895952e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 927268792,
+      "step": 5403
+    },
+    {
+      "epoch": 61.35410764872521,
+      "grad_norm": 3.541180097637101,
+      "learning_rate": 5e-06,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 927439980,
+      "step": 5404
+    },
+    {
+      "epoch": 61.35410764872521,
+      "loss": 0.03730998933315277,
+      "loss_ce": 6.328700692392886e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 927439980,
+      "step": 5404
+    },
+    {
+      "epoch": 61.36543909348442,
+      "grad_norm": 4.06912400471719,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 927611752,
+      "step": 5405
+    },
+    {
+      "epoch": 61.36543909348442,
+      "loss": 0.09848879277706146,
+      "loss_ce": 0.00017641550221014768,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 927611752,
+      "step": 5405
+    },
+    {
+      "epoch": 61.376770538243626,
+      "grad_norm": 6.7557225539952155,
+      "learning_rate": 5e-06,
+      "loss": 0.103,
+      "num_input_tokens_seen": 927783696,
+      "step": 5406
+    },
+    {
+      "epoch": 61.376770538243626,
+      "loss": 0.08808835595846176,
+      "loss_ce": 6.039925938239321e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 927783696,
+      "step": 5406
+    },
+    {
+      "epoch": 61.388101983002834,
+      "grad_norm": 2.5775862143684343,
+      "learning_rate": 5e-06,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 927955968,
+      "step": 5407
+    },
+    {
+      "epoch": 61.388101983002834,
+      "loss": 0.03690294921398163,
+      "loss_ce": 3.771438787225634e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 927955968,
+      "step": 5407
+    },
+    {
+      "epoch": 61.39943342776204,
+      "grad_norm": 3.1407236906164826,
+      "learning_rate": 5e-06,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 928127184,
+      "step": 5408
+    },
+    {
+      "epoch": 61.39943342776204,
+      "loss": 0.05802961811423302,
+      "loss_ce": 3.096037107752636e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 928127184,
+      "step": 5408
+    },
+    {
+      "epoch": 61.41076487252125,
+      "grad_norm": 2.3814743568995707,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 928298580,
+      "step": 5409
+    },
+    {
+      "epoch": 61.41076487252125,
+      "loss": 0.02485290728509426,
+      "loss_ce": 2.6857247576117516e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 928298580,
+      "step": 5409
+    },
+    {
+      "epoch": 61.422096317280456,
+      "grad_norm": 5.518554237820242,
+      "learning_rate": 5e-06,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 928470272,
+      "step": 5410
+    },
+    {
+      "epoch": 61.422096317280456,
+      "loss": 0.034210801124572754,
+      "loss_ce": 1.5853276636335067e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 928470272,
+      "step": 5410
+    },
+    {
+      "epoch": 61.43342776203966,
+      "grad_norm": 8.528453867087968,
+      "learning_rate": 5e-06,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 928641288,
+      "step": 5411
+    },
+    {
+      "epoch": 61.43342776203966,
+      "loss": 0.03900233283638954,
+      "loss_ce": 1.6126345144584775e-05,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 928641288,
+      "step": 5411
+    },
+    {
+      "epoch": 61.444759206798864,
+      "grad_norm": 3.084682555867012,
+      "learning_rate": 5e-06,
+      "loss": 0.0656,
+      "num_input_tokens_seen": 928812436,
+      "step": 5412
+    },
+    {
+      "epoch": 61.444759206798864,
+      "loss": 0.0273964311927557,
+      "loss_ce": 5.267925007501617e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 928812436,
+      "step": 5412
+    },
+    {
+      "epoch": 61.45609065155807,
+      "grad_norm": 3.9639327219346874,
+      "learning_rate": 5e-06,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 928983856,
+      "step": 5413
+    },
+    {
+      "epoch": 61.45609065155807,
+      "loss": 0.029340803623199463,
+      "loss_ce": 0.00010496339382370934,
+      "loss_iou": 0.46875,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 928983856,
+      "step": 5413
+    },
+    {
+      "epoch": 61.46742209631728,
+      "grad_norm": 3.6539222112964658,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 929155308,
+      "step": 5414
+    },
+    {
+      "epoch": 61.46742209631728,
+      "loss": 0.04240777716040611,
+      "loss_ce": 0.00024011486675590277,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 929155308,
+      "step": 5414
+    },
+    {
+      "epoch": 61.478753541076486,
+      "grad_norm": 3.5068252122486903,
+      "learning_rate": 5e-06,
+      "loss": 0.077,
+      "num_input_tokens_seen": 929325240,
+      "step": 5415
+    },
+    {
+      "epoch": 61.478753541076486,
+      "loss": 0.126019686460495,
+      "loss_ce": 4.312150122132152e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 929325240,
+      "step": 5415
+    },
+    {
+      "epoch": 61.49008498583569,
+      "grad_norm": 3.2674185502066058,
+      "learning_rate": 5e-06,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 929497020,
+      "step": 5416
+    },
+    {
+      "epoch": 61.49008498583569,
+      "loss": 0.027205945923924446,
+      "loss_ce": 0.0001368547382298857,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 929497020,
+      "step": 5416
+    },
+    {
+      "epoch": 61.5014164305949,
+      "grad_norm": 5.000935789033505,
+      "learning_rate": 5e-06,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 929667324,
+      "step": 5417
+    },
+    {
+      "epoch": 61.5014164305949,
+      "loss": 0.034466326236724854,
+      "loss_ce": 1.1979979717580136e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 929667324,
+      "step": 5417
+    },
+    {
+      "epoch": 61.51274787535411,
+      "grad_norm": 4.028097059557899,
+      "learning_rate": 5e-06,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 929838736,
+      "step": 5418
+    },
+    {
+      "epoch": 61.51274787535411,
+      "loss": 0.05556664243340492,
+      "loss_ce": 5.516869350685738e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 929838736,
+      "step": 5418
+    },
+    {
+      "epoch": 61.524079320113316,
+      "grad_norm": 3.6396761912297717,
+      "learning_rate": 5e-06,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 930010732,
+      "step": 5419
+    },
+    {
+      "epoch": 61.524079320113316,
+      "loss": 0.0308145210146904,
+      "loss_ce": 7.568874571006745e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 930010732,
+      "step": 5419
+    },
+    {
+      "epoch": 61.53541076487252,
+      "grad_norm": 3.989282732550078,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 930182116,
+      "step": 5420
+    },
+    {
+      "epoch": 61.53541076487252,
+      "loss": 0.0341818705201149,
+      "loss_ce": 4.796050052391365e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 930182116,
+      "step": 5420
+    },
+    {
+      "epoch": 61.54674220963173,
+      "grad_norm": 3.4325899624936826,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 930353272,
+      "step": 5421
+    },
+    {
+      "epoch": 61.54674220963173,
+      "loss": 0.08812987804412842,
+      "loss_ce": 4.088549758307636e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 930353272,
+      "step": 5421
+    },
+    {
+      "epoch": 61.55807365439094,
+      "grad_norm": 3.190805977217918,
+      "learning_rate": 5e-06,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 930523944,
+      "step": 5422
+    },
+    {
+      "epoch": 61.55807365439094,
+      "loss": 0.05038332939147949,
+      "loss_ce": 6.747568841092288e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 930523944,
+      "step": 5422
+    },
+    {
+      "epoch": 61.56940509915014,
+      "grad_norm": 3.28246754590637,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 930695872,
+      "step": 5423
+    },
+    {
+      "epoch": 61.56940509915014,
+      "loss": 0.03591139242053032,
+      "loss_ce": 3.797807221417315e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 930695872,
+      "step": 5423
+    },
+    {
+      "epoch": 61.580736543909346,
+      "grad_norm": 2.9441316974884915,
+      "learning_rate": 5e-06,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 930867124,
+      "step": 5424
+    },
+    {
+      "epoch": 61.580736543909346,
+      "loss": 0.07019766420125961,
+      "loss_ce": 5.3008283430244774e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 930867124,
+      "step": 5424
+    },
+    {
+      "epoch": 61.59206798866855,
+      "grad_norm": 18.563942399602713,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 931039048,
+      "step": 5425
+    },
+    {
+      "epoch": 61.59206798866855,
+      "loss": 0.02848649024963379,
+      "loss_ce": 2.8848489819210954e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 931039048,
+      "step": 5425
+    },
+    {
+      "epoch": 61.60339943342776,
+      "grad_norm": 3.589626787427431,
+      "learning_rate": 5e-06,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 931210696,
+      "step": 5426
+    },
+    {
+      "epoch": 61.60339943342776,
+      "loss": 0.050541266798973083,
+      "loss_ce": 4.993191396351904e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 931210696,
+      "step": 5426
+    },
+    {
+      "epoch": 61.61473087818697,
+      "grad_norm": 3.473028753065151,
+      "learning_rate": 5e-06,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 931381840,
+      "step": 5427
+    },
+    {
+      "epoch": 61.61473087818697,
+      "loss": 0.07088002562522888,
+      "loss_ce": 3.3468684705439955e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 931381840,
+      "step": 5427
+    },
+    {
+      "epoch": 61.626062322946176,
+      "grad_norm": 4.658931952452781,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 931553444,
+      "step": 5428
+    },
+    {
+      "epoch": 61.626062322946176,
+      "loss": 0.0438983254134655,
+      "loss_ce": 4.456478200154379e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 931553444,
+      "step": 5428
+    },
+    {
+      "epoch": 61.63739376770538,
+      "grad_norm": 4.018127900473281,
+      "learning_rate": 5e-06,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 931725036,
+      "step": 5429
+    },
+    {
+      "epoch": 61.63739376770538,
+      "loss": 0.0372612327337265,
+      "loss_ce": 4.5044809667160735e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 931725036,
+      "step": 5429
+    },
+    {
+      "epoch": 61.64872521246459,
+      "grad_norm": 4.453898987378134,
+      "learning_rate": 5e-06,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 931897016,
+      "step": 5430
+    },
+    {
+      "epoch": 61.64872521246459,
+      "loss": 0.04600450396537781,
+      "loss_ce": 0.0004265017923898995,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 931897016,
+      "step": 5430
+    },
+    {
+      "epoch": 61.6600566572238,
+      "grad_norm": 3.216479772630932,
+      "learning_rate": 5e-06,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 932068764,
+      "step": 5431
+    },
+    {
+      "epoch": 61.6600566572238,
+      "loss": 0.04468665271997452,
+      "loss_ce": 2.4175857106456533e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 932068764,
+      "step": 5431
+    },
+    {
+      "epoch": 61.671388101983005,
+      "grad_norm": 3.9802162349397903,
+      "learning_rate": 5e-06,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 932238280,
+      "step": 5432
+    },
+    {
+      "epoch": 61.671388101983005,
+      "loss": 0.030636565759778023,
+      "loss_ce": 4.269458440830931e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 932238280,
+      "step": 5432
+    },
+    {
+      "epoch": 61.68271954674221,
+      "grad_norm": 2.986396616669738,
+      "learning_rate": 5e-06,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 932409984,
+      "step": 5433
+    },
+    {
+      "epoch": 61.68271954674221,
+      "loss": 0.0468275249004364,
+      "loss_ce": 2.8818783903261647e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 932409984,
+      "step": 5433
+    },
+    {
+      "epoch": 61.69405099150141,
+      "grad_norm": 3.0210660949127752,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 932581864,
+      "step": 5434
+    },
+    {
+      "epoch": 61.69405099150141,
+      "loss": 0.031948283314704895,
+      "loss_ce": 4.2154802940785885e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 932581864,
+      "step": 5434
+    },
+    {
+      "epoch": 61.70538243626062,
+      "grad_norm": 3.501685231665008,
+      "learning_rate": 5e-06,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 932753596,
+      "step": 5435
+    },
+    {
+      "epoch": 61.70538243626062,
+      "loss": 0.057972513139247894,
+      "loss_ce": 3.4894284908659756e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 932753596,
+      "step": 5435
+    },
+    {
+      "epoch": 61.71671388101983,
+      "grad_norm": 4.665618159130267,
+      "learning_rate": 5e-06,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 932925300,
+      "step": 5436
+    },
+    {
+      "epoch": 61.71671388101983,
+      "loss": 0.05014738440513611,
+      "loss_ce": 2.226119249826297e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 932925300,
+      "step": 5436
+    },
+    {
+      "epoch": 61.728045325779036,
+      "grad_norm": 3.6334451061987476,
+      "learning_rate": 5e-06,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 933097384,
+      "step": 5437
+    },
+    {
+      "epoch": 61.728045325779036,
+      "loss": 0.03699394688010216,
+      "loss_ce": 3.716022183652967e-05,
+      "loss_iou": 0.640625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 933097384,
+      "step": 5437
+    },
+    {
+      "epoch": 61.73937677053824,
+      "grad_norm": 3.460290664743299,
+      "learning_rate": 5e-06,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 933269376,
+      "step": 5438
+    },
+    {
+      "epoch": 61.73937677053824,
+      "loss": 0.06431925296783447,
+      "loss_ce": 0.00011027281288988888,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 933269376,
+      "step": 5438
+    },
+    {
+      "epoch": 61.75070821529745,
+      "grad_norm": 3.217820309660579,
+      "learning_rate": 5e-06,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 933441000,
+      "step": 5439
+    },
+    {
+      "epoch": 61.75070821529745,
+      "loss": 0.06744620203971863,
+      "loss_ce": 3.286795617896132e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 933441000,
+      "step": 5439
+    },
+    {
+      "epoch": 61.76203966005666,
+      "grad_norm": 11.72594330252447,
+      "learning_rate": 5e-06,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 933611336,
+      "step": 5440
+    },
+    {
+      "epoch": 61.76203966005666,
+      "loss": 0.06040842831134796,
+      "loss_ce": 2.939722799055744e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 933611336,
+      "step": 5440
+    },
+    {
+      "epoch": 61.773371104815865,
+      "grad_norm": 3.0682254511155818,
+      "learning_rate": 5e-06,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 933783524,
+      "step": 5441
+    },
+    {
+      "epoch": 61.773371104815865,
+      "loss": 0.040234826505184174,
+      "loss_ce": 7.369305239990354e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 933783524,
+      "step": 5441
+    },
+    {
+      "epoch": 61.78470254957507,
+      "grad_norm": 3.0250770804577334,
+      "learning_rate": 5e-06,
+      "loss": 0.049,
+      "num_input_tokens_seen": 933955356,
+      "step": 5442
+    },
+    {
+      "epoch": 61.78470254957507,
+      "loss": 0.04656115174293518,
+      "loss_ce": 2.184302320529241e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 933955356,
+      "step": 5442
+    },
+    {
+      "epoch": 61.79603399433428,
+      "grad_norm": 2.544432106404559,
+      "learning_rate": 5e-06,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 934125340,
+      "step": 5443
+    },
+    {
+      "epoch": 61.79603399433428,
+      "loss": 0.14097346365451813,
+      "loss_ce": 7.380479655694216e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 934125340,
+      "step": 5443
+    },
+    {
+      "epoch": 61.80736543909349,
+      "grad_norm": 2.51052929600517,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 934294652,
+      "step": 5444
+    },
+    {
+      "epoch": 61.80736543909349,
+      "loss": 0.037977270781993866,
+      "loss_ce": 5.918265014770441e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 934294652,
+      "step": 5444
+    },
+    {
+      "epoch": 61.81869688385269,
+      "grad_norm": 2.930109162078187,
+      "learning_rate": 5e-06,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 934465460,
+      "step": 5445
+    },
+    {
+      "epoch": 61.81869688385269,
+      "loss": 0.03072969615459442,
+      "loss_ce": 5.952788342256099e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 934465460,
+      "step": 5445
+    },
+    {
+      "epoch": 61.830028328611895,
+      "grad_norm": 3.5571746383023672,
+      "learning_rate": 5e-06,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 934637528,
+      "step": 5446
+    },
+    {
+      "epoch": 61.830028328611895,
+      "loss": 0.03334224969148636,
+      "loss_ce": 4.757177157443948e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 934637528,
+      "step": 5446
+    },
+    {
+      "epoch": 61.8413597733711,
+      "grad_norm": 3.990280030304733,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 934809404,
+      "step": 5447
+    },
+    {
+      "epoch": 61.8413597733711,
+      "loss": 0.05740930512547493,
+      "loss_ce": 3.626099351095036e-05,
+      "loss_iou": 0.59375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 934809404,
+      "step": 5447
+    },
+    {
+      "epoch": 61.85269121813031,
+      "grad_norm": 3.718655360816984,
+      "learning_rate": 5e-06,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 934980428,
+      "step": 5448
+    },
+    {
+      "epoch": 61.85269121813031,
+      "loss": 0.03991135209798813,
+      "loss_ce": 2.4881439458113164e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 934980428,
+      "step": 5448
+    },
+    {
+      "epoch": 61.86402266288952,
+      "grad_norm": 3.374143159873171,
+      "learning_rate": 5e-06,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 935152444,
+      "step": 5449
+    },
+    {
+      "epoch": 61.86402266288952,
+      "loss": 0.038515567779541016,
+      "loss_ce": 0.00013971282169222832,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 935152444,
+      "step": 5449
+    },
+    {
+      "epoch": 61.875354107648725,
+      "grad_norm": 3.887096494678402,
+      "learning_rate": 5e-06,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 935322656,
+      "step": 5450
+    },
+    {
+      "epoch": 61.875354107648725,
+      "loss": 0.05174320936203003,
+      "loss_ce": 9.220830543199554e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 935322656,
+      "step": 5450
+    },
+    {
+      "epoch": 61.88668555240793,
+      "grad_norm": 3.2053296501397277,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 935491536,
+      "step": 5451
+    },
+    {
+      "epoch": 61.88668555240793,
+      "loss": 0.03982760012149811,
+      "loss_ce": 4.79383124911692e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 935491536,
+      "step": 5451
+    },
+    {
+      "epoch": 61.89801699716714,
+      "grad_norm": 3.5288835585458127,
+      "learning_rate": 5e-06,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 935663660,
+      "step": 5452
+    },
+    {
+      "epoch": 61.89801699716714,
+      "loss": 0.03726642578840256,
+      "loss_ce": 1.9720599084394053e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 935663660,
+      "step": 5452
+    },
+    {
+      "epoch": 61.90934844192635,
+      "grad_norm": 4.201719011974456,
+      "learning_rate": 5e-06,
+      "loss": 0.0497,
+      "num_input_tokens_seen": 935834640,
+      "step": 5453
+    },
+    {
+      "epoch": 61.90934844192635,
+      "loss": 0.048681341111660004,
+      "loss_ce": 3.632364678196609e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 935834640,
+      "step": 5453
+    },
+    {
+      "epoch": 61.920679886685555,
+      "grad_norm": 3.7062454895882597,
+      "learning_rate": 5e-06,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 936004300,
+      "step": 5454
+    },
+    {
+      "epoch": 61.920679886685555,
+      "loss": 0.05486631765961647,
+      "loss_ce": 2.6231682568322867e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 936004300,
+      "step": 5454
+    },
+    {
+      "epoch": 61.93201133144476,
+      "grad_norm": 3.482678929249557,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 936175564,
+      "step": 5455
+    },
+    {
+      "epoch": 61.93201133144476,
+      "loss": 0.051267560571432114,
+      "loss_ce": 7.432389247696847e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 936175564,
+      "step": 5455
+    },
+    {
+      "epoch": 61.94334277620396,
+      "grad_norm": 3.435541656780739,
+      "learning_rate": 5e-06,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 936343884,
+      "step": 5456
+    },
+    {
+      "epoch": 61.94334277620396,
+      "loss": 0.029880985617637634,
+      "loss_ce": 3.479361112113111e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 936343884,
+      "step": 5456
+    },
+    {
+      "epoch": 61.95467422096317,
+      "grad_norm": 2.5019860673659955,
+      "learning_rate": 5e-06,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 936515464,
+      "step": 5457
+    },
+    {
+      "epoch": 61.95467422096317,
+      "loss": 0.04240552335977554,
+      "loss_ce": 4.712356167146936e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 936515464,
+      "step": 5457
+    },
+    {
+      "epoch": 61.96600566572238,
+      "grad_norm": 2.820645150690986,
+      "learning_rate": 5e-06,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 936686680,
+      "step": 5458
+    },
+    {
+      "epoch": 61.96600566572238,
+      "loss": 0.08181725442409515,
+      "loss_ce": 4.540765075944364e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 936686680,
+      "step": 5458
+    },
+    {
+      "epoch": 61.977337110481585,
+      "grad_norm": 3.1310054560317986,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 936858448,
+      "step": 5459
+    },
+    {
+      "epoch": 61.977337110481585,
+      "loss": 0.03853434696793556,
+      "loss_ce": 2.1163112251088023e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 936858448,
+      "step": 5459
+    },
+    {
+      "epoch": 61.98866855524079,
+      "grad_norm": 3.7846895611219433,
+      "learning_rate": 5e-06,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 937030340,
+      "step": 5460
+    },
+    {
+      "epoch": 61.98866855524079,
+      "loss": 0.031084464862942696,
+      "loss_ce": 3.2829411793500185e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 937030340,
+      "step": 5460
+    },
+    {
+      "epoch": 62.0,
+      "grad_norm": 3.417153388904375,
+      "learning_rate": 5e-06,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 937201416,
+      "step": 5461
+    },
+    {
+      "epoch": 62.0,
+      "loss": 0.062279924750328064,
+      "loss_ce": 0.00011561553401406854,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 937201416,
+      "step": 5461
+    },
+    {
+      "epoch": 62.01133144475921,
+      "grad_norm": 3.2218375417976497,
+      "learning_rate": 5e-06,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 937372452,
+      "step": 5462
+    },
+    {
+      "epoch": 62.01133144475921,
+      "loss": 0.06710394471883774,
+      "loss_ce": 4.156710201641545e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 937372452,
+      "step": 5462
+    },
+    {
+      "epoch": 62.022662889518415,
+      "grad_norm": 3.061302474910644,
+      "learning_rate": 5e-06,
+      "loss": 0.048,
+      "num_input_tokens_seen": 937542480,
+      "step": 5463
+    },
+    {
+      "epoch": 62.022662889518415,
+      "loss": 0.04750329256057739,
+      "loss_ce": 3.319942334201187e-05,
+      "loss_iou": 0.0830078125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 937542480,
+      "step": 5463
+    },
+    {
+      "epoch": 62.03399433427762,
+      "grad_norm": 4.791495385985432,
+      "learning_rate": 5e-06,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 937713684,
+      "step": 5464
+    },
+    {
+      "epoch": 62.03399433427762,
+      "loss": 0.05480201169848442,
+      "loss_ce": 3.8218680856516585e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 937713684,
+      "step": 5464
+    },
+    {
+      "epoch": 62.04532577903683,
+      "grad_norm": 4.773380508106855,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 937885840,
+      "step": 5465
+    },
+    {
+      "epoch": 62.04532577903683,
+      "loss": 0.11668197810649872,
+      "loss_ce": 4.379051097203046e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 937885840,
+      "step": 5465
+    },
+    {
+      "epoch": 62.05665722379604,
+      "grad_norm": 3.748227812173109,
+      "learning_rate": 5e-06,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 938057976,
+      "step": 5466
+    },
+    {
+      "epoch": 62.05665722379604,
+      "loss": 0.048764150589704514,
+      "loss_ce": 2.7580414098338224e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 938057976,
+      "step": 5466
+    },
+    {
+      "epoch": 62.06798866855524,
+      "grad_norm": 5.729863651225758,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 938229656,
+      "step": 5467
+    },
+    {
+      "epoch": 62.06798866855524,
+      "loss": 0.06794430315494537,
+      "loss_ce": 4.269663622835651e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 938229656,
+      "step": 5467
+    },
+    {
+      "epoch": 62.079320113314445,
+      "grad_norm": 3.465089578231126,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 938401440,
+      "step": 5468
+    },
+    {
+      "epoch": 62.079320113314445,
+      "loss": 0.0363333523273468,
+      "loss_ce": 7.847221422707662e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 938401440,
+      "step": 5468
+    },
+    {
+      "epoch": 62.09065155807365,
+      "grad_norm": 3.636549317837481,
+      "learning_rate": 5e-06,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 938572824,
+      "step": 5469
+    },
+    {
+      "epoch": 62.09065155807365,
+      "loss": 0.04511266201734543,
+      "loss_ce": 0.00014501233818009496,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 938572824,
+      "step": 5469
+    },
+    {
+      "epoch": 62.10198300283286,
+      "grad_norm": 3.3519042758105075,
+      "learning_rate": 5e-06,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 938744636,
+      "step": 5470
+    },
+    {
+      "epoch": 62.10198300283286,
+      "loss": 0.031431086361408234,
+      "loss_ce": 5.9018824686063454e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 938744636,
+      "step": 5470
+    },
+    {
+      "epoch": 62.11331444759207,
+      "grad_norm": 2.659883634851869,
+      "learning_rate": 5e-06,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 938916256,
+      "step": 5471
+    },
+    {
+      "epoch": 62.11331444759207,
+      "loss": 0.03059856966137886,
+      "loss_ce": 5.810384755022824e-05,
+      "loss_iou": 0.0283203125,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 938916256,
+      "step": 5471
+    },
+    {
+      "epoch": 62.124645892351275,
+      "grad_norm": 2.855979792314044,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 939087900,
+      "step": 5472
+    },
+    {
+      "epoch": 62.124645892351275,
+      "loss": 0.11063708364963531,
+      "loss_ce": 4.138428266742267e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 939087900,
+      "step": 5472
+    },
+    {
+      "epoch": 62.13597733711048,
+      "grad_norm": 2.1534098796645025,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 939257536,
+      "step": 5473
+    },
+    {
+      "epoch": 62.13597733711048,
+      "loss": 0.028540601953864098,
+      "loss_ce": 8.296081068692729e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 939257536,
+      "step": 5473
+    },
+    {
+      "epoch": 62.14730878186969,
+      "grad_norm": 3.2350856560883776,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 939427492,
+      "step": 5474
+    },
+    {
+      "epoch": 62.14730878186969,
+      "loss": 0.03937472403049469,
+      "loss_ce": 4.5195130951469764e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 939427492,
+      "step": 5474
+    },
+    {
+      "epoch": 62.1586402266289,
+      "grad_norm": 4.100332152486143,
+      "learning_rate": 5e-06,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 939599192,
+      "step": 5475
+    },
+    {
+      "epoch": 62.1586402266289,
+      "loss": 0.027330735698342323,
+      "loss_ce": 1.7503924027550966e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 939599192,
+      "step": 5475
+    },
+    {
+      "epoch": 62.169971671388105,
+      "grad_norm": 3.507712436120644,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 939770736,
+      "step": 5476
+    },
+    {
+      "epoch": 62.169971671388105,
+      "loss": 0.040435679256916046,
+      "loss_ce": 4.566271309158765e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 939770736,
+      "step": 5476
+    },
+    {
+      "epoch": 62.18130311614731,
+      "grad_norm": 3.345223746675795,
+      "learning_rate": 5e-06,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 939940736,
+      "step": 5477
+    },
+    {
+      "epoch": 62.18130311614731,
+      "loss": 0.03670191019773483,
+      "loss_ce": 0.0003554762515705079,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 939940736,
+      "step": 5477
+    },
+    {
+      "epoch": 62.19263456090651,
+      "grad_norm": 3.634020319367164,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 940112220,
+      "step": 5478
+    },
+    {
+      "epoch": 62.19263456090651,
+      "loss": 0.03957882523536682,
+      "loss_ce": 2.804466384986881e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 940112220,
+      "step": 5478
+    },
+    {
+      "epoch": 62.20396600566572,
+      "grad_norm": 3.734574033117681,
+      "learning_rate": 5e-06,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 940281772,
+      "step": 5479
+    },
+    {
+      "epoch": 62.20396600566572,
+      "loss": 0.04302915185689926,
+      "loss_ce": 4.514193278737366e-05,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 940281772,
+      "step": 5479
+    },
+    {
+      "epoch": 62.21529745042493,
+      "grad_norm": 3.3620158978122068,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 940453712,
+      "step": 5480
+    },
+    {
+      "epoch": 62.21529745042493,
+      "loss": 0.0472903847694397,
+      "loss_ce": 6.44295578240417e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 940453712,
+      "step": 5480
+    },
+    {
+      "epoch": 62.226628895184135,
+      "grad_norm": 3.4598195091750634,
+      "learning_rate": 5e-06,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 940625912,
+      "step": 5481
+    },
+    {
+      "epoch": 62.226628895184135,
+      "loss": 0.050176411867141724,
+      "loss_ce": 5.128864722792059e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 940625912,
+      "step": 5481
+    },
+    {
+      "epoch": 62.23796033994334,
+      "grad_norm": 3.638312621458632,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 940797020,
+      "step": 5482
+    },
+    {
+      "epoch": 62.23796033994334,
+      "loss": 0.06003417819738388,
+      "loss_ce": 3.662324888864532e-05,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 940797020,
+      "step": 5482
+    },
+    {
+      "epoch": 62.24929178470255,
+      "grad_norm": 3.6491319345394113,
+      "learning_rate": 5e-06,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 940968596,
+      "step": 5483
+    },
+    {
+      "epoch": 62.24929178470255,
+      "loss": 0.03468136861920357,
+      "loss_ce": 5.917662929277867e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 940968596,
+      "step": 5483
+    },
+    {
+      "epoch": 62.26062322946176,
+      "grad_norm": 3.4337978968828584,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 941140248,
+      "step": 5484
+    },
+    {
+      "epoch": 62.26062322946176,
+      "loss": 0.052647706121206284,
+      "loss_ce": 2.014332130784169e-05,
+      "loss_iou": 0.1044921875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 941140248,
+      "step": 5484
+    },
+    {
+      "epoch": 62.271954674220964,
+      "grad_norm": 3.0039884288887255,
+      "learning_rate": 5e-06,
+      "loss": 0.048,
+      "num_input_tokens_seen": 941312176,
+      "step": 5485
+    },
+    {
+      "epoch": 62.271954674220964,
+      "loss": 0.0806206688284874,
+      "loss_ce": 0.00013055690214969218,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 941312176,
+      "step": 5485
+    },
+    {
+      "epoch": 62.28328611898017,
+      "grad_norm": 3.2068092276286273,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 941483908,
+      "step": 5486
+    },
+    {
+      "epoch": 62.28328611898017,
+      "loss": 0.03270648419857025,
+      "loss_ce": 5.267691449262202e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 941483908,
+      "step": 5486
+    },
+    {
+      "epoch": 62.29461756373938,
+      "grad_norm": 3.054246704007408,
+      "learning_rate": 5e-06,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 941654524,
+      "step": 5487
+    },
+    {
+      "epoch": 62.29461756373938,
+      "loss": 0.0500912144780159,
+      "loss_ce": 2.7128768124384806e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 941654524,
+      "step": 5487
+    },
+    {
+      "epoch": 62.30594900849859,
+      "grad_norm": 2.8691608604965206,
+      "learning_rate": 5e-06,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 941826104,
+      "step": 5488
+    },
+    {
+      "epoch": 62.30594900849859,
+      "loss": 0.029837291687726974,
+      "loss_ce": 2.1616673620883375e-05,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 941826104,
+      "step": 5488
+    },
+    {
+      "epoch": 62.31728045325779,
+      "grad_norm": 4.056745157682888,
+      "learning_rate": 5e-06,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 941996984,
+      "step": 5489
+    },
+    {
+      "epoch": 62.31728045325779,
+      "loss": 0.027645211666822433,
+      "loss_ce": 5.7322162319906056e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 941996984,
+      "step": 5489
+    },
+    {
+      "epoch": 62.328611898016995,
+      "grad_norm": 3.71155986718287,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 942167972,
+      "step": 5490
+    },
+    {
+      "epoch": 62.328611898016995,
+      "loss": 0.034503042697906494,
+      "loss_ce": 0.00010973081225529313,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 942167972,
+      "step": 5490
+    },
+    {
+      "epoch": 62.3399433427762,
+      "grad_norm": 3.2239677368004456,
+      "learning_rate": 5e-06,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 942339772,
+      "step": 5491
+    },
+    {
+      "epoch": 62.3399433427762,
+      "loss": 0.036976248025894165,
+      "loss_ce": 1.9460418116068467e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 942339772,
+      "step": 5491
+    },
+    {
+      "epoch": 62.35127478753541,
+      "grad_norm": 3.4934915350009472,
+      "learning_rate": 5e-06,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 942511676,
+      "step": 5492
+    },
+    {
+      "epoch": 62.35127478753541,
+      "loss": 0.07291790843009949,
+      "loss_ce": 4.193507265881635e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 942511676,
+      "step": 5492
+    },
+    {
+      "epoch": 62.36260623229462,
+      "grad_norm": 2.95771765077045,
+      "learning_rate": 5e-06,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 942683112,
+      "step": 5493
+    },
+    {
+      "epoch": 62.36260623229462,
+      "loss": 0.05229192227125168,
+      "loss_ce": 0.0002747105318121612,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 942683112,
+      "step": 5493
+    },
+    {
+      "epoch": 62.373937677053824,
+      "grad_norm": 71.51418829355738,
+      "learning_rate": 5e-06,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 942855304,
+      "step": 5494
+    },
+    {
+      "epoch": 62.373937677053824,
+      "loss": 0.03032710961997509,
+      "loss_ce": 2.3153936126618646e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 942855304,
+      "step": 5494
+    },
+    {
+      "epoch": 62.38526912181303,
+      "grad_norm": 3.4614384727158227,
+      "learning_rate": 5e-06,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 943027188,
+      "step": 5495
+    },
+    {
+      "epoch": 62.38526912181303,
+      "loss": 0.03571552410721779,
+      "loss_ce": 2.5214503693860024e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 943027188,
+      "step": 5495
+    },
+    {
+      "epoch": 62.39660056657224,
+      "grad_norm": 3.5745985231228183,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 943199220,
+      "step": 5496
+    },
+    {
+      "epoch": 62.39660056657224,
+      "loss": 0.036427609622478485,
+      "loss_ce": 5.065386358182877e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 943199220,
+      "step": 5496
+    },
+    {
+      "epoch": 62.40793201133145,
+      "grad_norm": 3.731658545464063,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 943370800,
+      "step": 5497
+    },
+    {
+      "epoch": 62.40793201133145,
+      "loss": 0.06790690124034882,
+      "loss_ce": 2.054810647678096e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 943370800,
+      "step": 5497
+    },
+    {
+      "epoch": 62.419263456090654,
+      "grad_norm": 3.4609450095495955,
+      "learning_rate": 5e-06,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 943542468,
+      "step": 5498
+    },
+    {
+      "epoch": 62.419263456090654,
+      "loss": 0.043365925550460815,
+      "loss_ce": 4.622228880180046e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 943542468,
+      "step": 5498
+    },
+    {
+      "epoch": 62.43059490084986,
+      "grad_norm": 2.917480575959165,
+      "learning_rate": 5e-06,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 943712368,
+      "step": 5499
+    },
+    {
+      "epoch": 62.43059490084986,
+      "loss": 0.02478640154004097,
+      "loss_ce": 6.716401549056172e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 943712368,
+      "step": 5499
+    },
+    {
+      "epoch": 62.44192634560906,
+      "grad_norm": 3.639346077728529,
+      "learning_rate": 5e-06,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 943884380,
+      "step": 5500
+    },
+    {
+      "epoch": 62.44192634560906,
+      "eval_seeclick_CIoU": 0.5294385850429535,
+      "eval_seeclick_GIoU": 0.5267061293125153,
+      "eval_seeclick_IoU": 0.5655473172664642,
+      "eval_seeclick_MAE_all": 0.0660373866558075,
+      "eval_seeclick_MAE_h": 0.03382791392505169,
+      "eval_seeclick_MAE_w": 0.0966530367732048,
+      "eval_seeclick_MAE_x": 0.10127667337656021,
+      "eval_seeclick_MAE_y": 0.03239190950989723,
+      "eval_seeclick_NUM_probability": 0.9999770522117615,
+      "eval_seeclick_inside_bbox": 0.9232954680919647,
+      "eval_seeclick_loss": 0.9397633671760559,
+      "eval_seeclick_loss_ce": 0.6926635801792145,
+      "eval_seeclick_loss_iou": 0.5479736328125,
+      "eval_seeclick_loss_num": 0.049381256103515625,
+      "eval_seeclick_loss_xval": 0.2469482421875,
+      "eval_seeclick_runtime": 68.7854,
+      "eval_seeclick_samples_per_second": 0.625,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 943884380,
+      "step": 5500
+    },
+    {
+      "epoch": 62.44192634560906,
+      "eval_icons_CIoU": 0.7322515547275543,
+      "eval_icons_GIoU": 0.7308847606182098,
+      "eval_icons_IoU": 0.7460240721702576,
+      "eval_icons_MAE_all": 0.03638903237879276,
+      "eval_icons_MAE_h": 0.03358772397041321,
+      "eval_icons_MAE_w": 0.042488014325499535,
+      "eval_icons_MAE_x": 0.03324949275702238,
+      "eval_icons_MAE_y": 0.03623090125620365,
+      "eval_icons_NUM_probability": 0.9991548657417297,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.13571447134017944,
+      "eval_icons_loss_ce": 0.00491101061925292,
+      "eval_icons_loss_iou": 0.5789794921875,
+      "eval_icons_loss_num": 0.02399444580078125,
+      "eval_icons_loss_xval": 0.1199951171875,
+      "eval_icons_runtime": 78.55,
+      "eval_icons_samples_per_second": 0.637,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 943884380,
+      "step": 5500
+    },
+    {
+      "epoch": 62.44192634560906,
+      "eval_screenspot_CIoU": 0.6412812868754069,
+      "eval_screenspot_GIoU": 0.6449100772539774,
+      "eval_screenspot_IoU": 0.6669768492380778,
+      "eval_screenspot_MAE_all": 0.06233343109488487,
+      "eval_screenspot_MAE_h": 0.030469733600815136,
+      "eval_screenspot_MAE_w": 0.11368518508970737,
+      "eval_screenspot_MAE_x": 0.07766263497372468,
+      "eval_screenspot_MAE_y": 0.027516182512044907,
+      "eval_screenspot_NUM_probability": 0.9999338785807291,
+      "eval_screenspot_inside_bbox": 0.9079166650772095,
+      "eval_screenspot_loss": 0.2680523097515106,
+      "eval_screenspot_loss_ce": 0.01657482422888279,
+      "eval_screenspot_loss_iou": 0.4791666666666667,
+      "eval_screenspot_loss_num": 0.0505218505859375,
+      "eval_screenspot_loss_xval": 0.2526448567708333,
+      "eval_screenspot_runtime": 149.5419,
+      "eval_screenspot_samples_per_second": 0.595,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 943884380,
+      "step": 5500
+    },
+    {
+      "epoch": 62.44192634560906,
+      "eval_compot_CIoU": 0.8782232403755188,
+      "eval_compot_GIoU": 0.8780380487442017,
+      "eval_compot_IoU": 0.8868246972560883,
+      "eval_compot_MAE_all": 0.018250909633934498,
+      "eval_compot_MAE_h": 0.008052447810769081,
+      "eval_compot_MAE_w": 0.028046160470694304,
+      "eval_compot_MAE_x": 0.02766910195350647,
+      "eval_compot_MAE_y": 0.009235929697751999,
+      "eval_compot_NUM_probability": 0.9999649822711945,
+      "eval_compot_inside_bbox": 0.96875,
+      "eval_compot_loss": 0.06666712462902069,
+      "eval_compot_loss_ce": 3.294740054116119e-05,
+      "eval_compot_loss_iou": 0.4637451171875,
+      "eval_compot_loss_num": 0.011798858642578125,
+      "eval_compot_loss_xval": 0.0589752197265625,
+      "eval_compot_runtime": 83.2066,
+      "eval_compot_samples_per_second": 0.601,
+      "eval_compot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 943884380,
+      "step": 5500
+    },
+    {
+      "epoch": 62.44192634560906,
+      "eval_custom_ui_MAE_all": 0.019400109071284533,
+      "eval_custom_ui_MAE_x": 0.03052842989563942,
+      "eval_custom_ui_MAE_y": 0.008271789643913507,
+      "eval_custom_ui_NUM_probability": 0.9998880922794342,
+      "eval_custom_ui_loss": 0.21498550474643707,
+      "eval_custom_ui_loss_ce": 0.11672582477331161,
+      "eval_custom_ui_loss_num": 0.018308639526367188,
+      "eval_custom_ui_loss_xval": 0.0915069580078125,
+      "eval_custom_ui_runtime": 58.2084,
+      "eval_custom_ui_samples_per_second": 0.859,
+      "eval_custom_ui_steps_per_second": 0.034,
+      "num_input_tokens_seen": 943884380,
+      "step": 5500
+    },
+    {
+      "epoch": 62.44192634560906,
+      "loss": 0.259590744972229,
+      "loss_ce": 0.1462026834487915,
+      "loss_iou": 0.0,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 943884380,
+      "step": 5500
+    },
+    {
+      "epoch": 62.45325779036827,
+      "grad_norm": 2.8989652265084795,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 944056040,
+      "step": 5501
+    },
+    {
+      "epoch": 62.45325779036827,
+      "loss": 0.05497933179140091,
+      "loss_ce": 1.7175698303617537e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 944056040,
+      "step": 5501
+    },
+    {
+      "epoch": 62.46458923512748,
+      "grad_norm": 3.886888631254645,
+      "learning_rate": 5e-06,
+      "loss": 0.037,
+      "num_input_tokens_seen": 944227688,
+      "step": 5502
+    },
+    {
+      "epoch": 62.46458923512748,
+      "loss": 0.031036769971251488,
+      "loss_ce": 3.0910578061593696e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 944227688,
+      "step": 5502
+    },
+    {
+      "epoch": 62.475920679886684,
+      "grad_norm": 4.631258754104667,
+      "learning_rate": 5e-06,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 944399352,
+      "step": 5503
+    },
+    {
+      "epoch": 62.475920679886684,
+      "loss": 0.09276417642831802,
+      "loss_ce": 2.1252646547509357e-05,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 944399352,
+      "step": 5503
+    },
+    {
+      "epoch": 62.48725212464589,
+      "grad_norm": 3.233491652720797,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 944569148,
+      "step": 5504
+    },
+    {
+      "epoch": 62.48725212464589,
+      "loss": 0.05277272313833237,
+      "loss_ce": 2.3091110051609576e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 944569148,
+      "step": 5504
+    },
+    {
+      "epoch": 62.4985835694051,
+      "grad_norm": 3.012208039440752,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 944740044,
+      "step": 5505
+    },
+    {
+      "epoch": 62.4985835694051,
+      "loss": 0.04457823932170868,
+      "loss_ce": 5.309501648298465e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 944740044,
+      "step": 5505
+    },
+    {
+      "epoch": 62.50991501416431,
+      "grad_norm": 3.183093027172892,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 944911944,
+      "step": 5506
+    },
+    {
+      "epoch": 62.50991501416431,
+      "loss": 0.038533765822649,
+      "loss_ce": 3.5839791962644085e-05,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 944911944,
+      "step": 5506
+    },
+    {
+      "epoch": 62.521246458923514,
+      "grad_norm": 3.2110582643211774,
+      "learning_rate": 5e-06,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 945083412,
+      "step": 5507
+    },
+    {
+      "epoch": 62.521246458923514,
+      "loss": 0.036099180579185486,
+      "loss_ce": 4.266292671673e-05,
+      "loss_iou": 0.025390625,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 945083412,
+      "step": 5507
+    },
+    {
+      "epoch": 62.53257790368272,
+      "grad_norm": 3.4422862511589054,
+      "learning_rate": 5e-06,
+      "loss": 0.044,
+      "num_input_tokens_seen": 945255504,
+      "step": 5508
+    },
+    {
+      "epoch": 62.53257790368272,
+      "loss": 0.04410179704427719,
+      "loss_ce": 3.441364970058203e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 945255504,
+      "step": 5508
+    },
+    {
+      "epoch": 62.54390934844193,
+      "grad_norm": 3.0136358965578594,
+      "learning_rate": 5e-06,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 945427568,
+      "step": 5509
+    },
+    {
+      "epoch": 62.54390934844193,
+      "loss": 0.026692409068346024,
+      "loss_ce": 3.530406320351176e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 945427568,
+      "step": 5509
+    },
+    {
+      "epoch": 62.555240793201136,
+      "grad_norm": 2.9336353248864806,
+      "learning_rate": 5e-06,
+      "loss": 0.0374,
+      "num_input_tokens_seen": 945598324,
+      "step": 5510
+    },
+    {
+      "epoch": 62.555240793201136,
+      "loss": 0.041790224611759186,
+      "loss_ce": 5.743668953073211e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 945598324,
+      "step": 5510
+    },
+    {
+      "epoch": 62.56657223796034,
+      "grad_norm": 4.196889928738389,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 945769936,
+      "step": 5511
+    },
+    {
+      "epoch": 62.56657223796034,
+      "loss": 0.031056372448801994,
+      "loss_ce": 5.051364132668823e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 945769936,
+      "step": 5511
+    },
+    {
+      "epoch": 62.577903682719544,
+      "grad_norm": 3.2909186270744875,
+      "learning_rate": 5e-06,
+      "loss": 0.055,
+      "num_input_tokens_seen": 945939532,
+      "step": 5512
+    },
+    {
+      "epoch": 62.577903682719544,
+      "loss": 0.035352952778339386,
+      "loss_ce": 4.411242844071239e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 945939532,
+      "step": 5512
+    },
+    {
+      "epoch": 62.58923512747875,
+      "grad_norm": 3.7816456933003115,
+      "learning_rate": 5e-06,
+      "loss": 0.042,
+      "num_input_tokens_seen": 946111400,
+      "step": 5513
+    },
+    {
+      "epoch": 62.58923512747875,
+      "loss": 0.04019385948777199,
+      "loss_ce": 4.798402369488031e-05,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 946111400,
+      "step": 5513
+    },
+    {
+      "epoch": 62.60056657223796,
+      "grad_norm": 4.076746931196385,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 946280812,
+      "step": 5514
+    },
+    {
+      "epoch": 62.60056657223796,
+      "loss": 0.035268209874629974,
+      "loss_ce": 2.041032348643057e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 946280812,
+      "step": 5514
+    },
+    {
+      "epoch": 62.611898016997166,
+      "grad_norm": 10.867864889594042,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 946452328,
+      "step": 5515
+    },
+    {
+      "epoch": 62.611898016997166,
+      "loss": 0.05627760291099548,
+      "loss_ce": 0.0004151762695983052,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 946452328,
+      "step": 5515
+    },
+    {
+      "epoch": 62.623229461756374,
+      "grad_norm": 2.649761157902133,
+      "learning_rate": 5e-06,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 946624120,
+      "step": 5516
+    },
+    {
+      "epoch": 62.623229461756374,
+      "loss": 0.03688281029462814,
+      "loss_ce": 4.80938506370876e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 946624120,
+      "step": 5516
+    },
+    {
+      "epoch": 62.63456090651558,
+      "grad_norm": 3.051612995672834,
+      "learning_rate": 5e-06,
+      "loss": 0.04,
+      "num_input_tokens_seen": 946796408,
+      "step": 5517
+    },
+    {
+      "epoch": 62.63456090651558,
+      "loss": 0.05551644042134285,
+      "loss_ce": 3.5485689295455813e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 946796408,
+      "step": 5517
+    },
+    {
+      "epoch": 62.64589235127479,
+      "grad_norm": 3.5373073118656984,
+      "learning_rate": 5e-06,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 946966452,
+      "step": 5518
+    },
+    {
+      "epoch": 62.64589235127479,
+      "loss": 0.04456561431288719,
+      "loss_ce": 4.046925823786296e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 946966452,
+      "step": 5518
+    },
+    {
+      "epoch": 62.657223796033996,
+      "grad_norm": 3.907036335038993,
+      "learning_rate": 5e-06,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 947137888,
+      "step": 5519
+    },
+    {
+      "epoch": 62.657223796033996,
+      "loss": 0.05694613233208656,
+      "loss_ce": 3.84789273084607e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 947137888,
+      "step": 5519
+    },
+    {
+      "epoch": 62.668555240793204,
+      "grad_norm": 3.2000291273963035,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 947309924,
+      "step": 5520
+    },
+    {
+      "epoch": 62.668555240793204,
+      "loss": 0.03819970414042473,
+      "loss_ce": 3.747270966414362e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 947309924,
+      "step": 5520
+    },
+    {
+      "epoch": 62.67988668555241,
+      "grad_norm": 3.5812497904922735,
+      "learning_rate": 5e-06,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 947482232,
+      "step": 5521
+    },
+    {
+      "epoch": 62.67988668555241,
+      "loss": 0.032056666910648346,
+      "loss_ce": 4.3724794522859156e-05,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 947482232,
+      "step": 5521
+    },
+    {
+      "epoch": 62.69121813031161,
+      "grad_norm": 3.546502527072104,
+      "learning_rate": 5e-06,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 947653988,
+      "step": 5522
+    },
+    {
+      "epoch": 62.69121813031161,
+      "loss": 0.10170534998178482,
+      "loss_ce": 0.0003412201476749033,
+      "loss_iou": 0.3125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 947653988,
+      "step": 5522
+    },
+    {
+      "epoch": 62.70254957507082,
+      "grad_norm": 2.9130508687657453,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 947825392,
+      "step": 5523
+    },
+    {
+      "epoch": 62.70254957507082,
+      "loss": 0.03968769684433937,
+      "loss_ce": 4.5361222873907536e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 947825392,
+      "step": 5523
+    },
+    {
+      "epoch": 62.713881019830026,
+      "grad_norm": 2.7849107269109816,
+      "learning_rate": 5e-06,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 947997548,
+      "step": 5524
+    },
+    {
+      "epoch": 62.713881019830026,
+      "loss": 0.038623061031103134,
+      "loss_ce": 7.936110341688618e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 947997548,
+      "step": 5524
+    },
+    {
+      "epoch": 62.725212464589234,
+      "grad_norm": 2.7275279847315534,
+      "learning_rate": 5e-06,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 948169724,
+      "step": 5525
+    },
+    {
+      "epoch": 62.725212464589234,
+      "loss": 0.08988450467586517,
+      "loss_ce": 2.5498316972516477e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 948169724,
+      "step": 5525
+    },
+    {
+      "epoch": 62.73654390934844,
+      "grad_norm": 3.580844911917427,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 948340084,
+      "step": 5526
+    },
+    {
+      "epoch": 62.73654390934844,
+      "loss": 0.0536431148648262,
+      "loss_ce": 0.00025260995607823133,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 948340084,
+      "step": 5526
+    },
+    {
+      "epoch": 62.74787535410765,
+      "grad_norm": 4.0624235856668065,
+      "learning_rate": 5e-06,
+      "loss": 0.043,
+      "num_input_tokens_seen": 948510764,
+      "step": 5527
+    },
+    {
+      "epoch": 62.74787535410765,
+      "loss": 0.05362735688686371,
+      "loss_ce": 6.900899461470544e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 948510764,
+      "step": 5527
+    },
+    {
+      "epoch": 62.759206798866856,
+      "grad_norm": 4.225018529027337,
+      "learning_rate": 5e-06,
+      "loss": 0.0742,
+      "num_input_tokens_seen": 948681964,
+      "step": 5528
+    },
+    {
+      "epoch": 62.759206798866856,
+      "loss": 0.07007186114788055,
+      "loss_ce": 7.979686051839963e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 948681964,
+      "step": 5528
+    },
+    {
+      "epoch": 62.77053824362606,
+      "grad_norm": 2.6253563894603347,
+      "learning_rate": 5e-06,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 948853932,
+      "step": 5529
+    },
+    {
+      "epoch": 62.77053824362606,
+      "loss": 0.043707218021154404,
+      "loss_ce": 5.1821691158693284e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 948853932,
+      "step": 5529
+    },
+    {
+      "epoch": 62.78186968838527,
+      "grad_norm": 2.7099694501533813,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 949025976,
+      "step": 5530
+    },
+    {
+      "epoch": 62.78186968838527,
+      "loss": 0.09038908779621124,
+      "loss_ce": 4.942140367347747e-05,
+      "loss_iou": 0.05078125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 949025976,
+      "step": 5530
+    },
+    {
+      "epoch": 62.79320113314448,
+      "grad_norm": 3.5593713543248726,
+      "learning_rate": 5e-06,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 949195152,
+      "step": 5531
+    },
+    {
+      "epoch": 62.79320113314448,
+      "loss": 0.047094546258449554,
+      "loss_ce": 6.695705815218389e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 949195152,
+      "step": 5531
+    },
+    {
+      "epoch": 62.804532577903686,
+      "grad_norm": 3.626068397511478,
+      "learning_rate": 5e-06,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 949366200,
+      "step": 5532
+    },
+    {
+      "epoch": 62.804532577903686,
+      "loss": 0.03735312819480896,
+      "loss_ce": 3.012992601725273e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 949366200,
+      "step": 5532
+    },
+    {
+      "epoch": 62.815864022662886,
+      "grad_norm": 3.4964932596175315,
+      "learning_rate": 5e-06,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 949538252,
+      "step": 5533
+    },
+    {
+      "epoch": 62.815864022662886,
+      "loss": 0.0271298848092556,
+      "loss_ce": 1.5017081750556827e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 949538252,
+      "step": 5533
+    },
+    {
+      "epoch": 62.827195467422094,
+      "grad_norm": 3.563074364395963,
+      "learning_rate": 5e-06,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 949708288,
+      "step": 5534
+    },
+    {
+      "epoch": 62.827195467422094,
+      "loss": 0.09923435002565384,
+      "loss_ce": 6.74857001286e-05,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 949708288,
+      "step": 5534
+    },
+    {
+      "epoch": 62.8385269121813,
+      "grad_norm": 7.6493647738468,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 949879900,
+      "step": 5535
+    },
+    {
+      "epoch": 62.8385269121813,
+      "loss": 0.08445686101913452,
+      "loss_ce": 4.5232605771161616e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 949879900,
+      "step": 5535
+    },
+    {
+      "epoch": 62.84985835694051,
+      "grad_norm": 3.6836378734952415,
+      "learning_rate": 5e-06,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 950051596,
+      "step": 5536
+    },
+    {
+      "epoch": 62.84985835694051,
+      "loss": 0.04199357330799103,
+      "loss_ce": 3.190220013493672e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 950051596,
+      "step": 5536
+    },
+    {
+      "epoch": 62.861189801699716,
+      "grad_norm": 4.1354539553392105,
+      "learning_rate": 5e-06,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 950222536,
+      "step": 5537
+    },
+    {
+      "epoch": 62.861189801699716,
+      "loss": 0.08515620976686478,
+      "loss_ce": 0.00013423789641819894,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 950222536,
+      "step": 5537
+    },
+    {
+      "epoch": 62.87252124645892,
+      "grad_norm": 3.330818098883854,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 950392056,
+      "step": 5538
+    },
+    {
+      "epoch": 62.87252124645892,
+      "loss": 0.0704437643289566,
+      "loss_ce": 2.4455370294163004e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 950392056,
+      "step": 5538
+    },
+    {
+      "epoch": 62.88385269121813,
+      "grad_norm": 46.86637353936134,
+      "learning_rate": 5e-06,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 950562524,
+      "step": 5539
+    },
+    {
+      "epoch": 62.88385269121813,
+      "loss": 0.034051500260829926,
+      "loss_ce": 2.440443859086372e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 950562524,
+      "step": 5539
+    },
+    {
+      "epoch": 62.89518413597734,
+      "grad_norm": 4.284276290861703,
+      "learning_rate": 5e-06,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 950734480,
+      "step": 5540
+    },
+    {
+      "epoch": 62.89518413597734,
+      "loss": 0.04192068427801132,
+      "loss_ce": 5.056730515207164e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 950734480,
+      "step": 5540
+    },
+    {
+      "epoch": 62.906515580736546,
+      "grad_norm": 3.747367498548348,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 950906020,
+      "step": 5541
+    },
+    {
+      "epoch": 62.906515580736546,
+      "loss": 0.02904198318719864,
+      "loss_ce": 5.028243322158232e-05,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 950906020,
+      "step": 5541
+    },
+    {
+      "epoch": 62.91784702549575,
+      "grad_norm": 4.109480185749345,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 951077940,
+      "step": 5542
+    },
+    {
+      "epoch": 62.91784702549575,
+      "loss": 0.02756473794579506,
+      "loss_ce": 2.2624190023634583e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 951077940,
+      "step": 5542
+    },
+    {
+      "epoch": 62.92917847025496,
+      "grad_norm": 2.967862491880175,
+      "learning_rate": 5e-06,
+      "loss": 0.062,
+      "num_input_tokens_seen": 951249684,
+      "step": 5543
+    },
+    {
+      "epoch": 62.92917847025496,
+      "loss": 0.048396818339824677,
+      "loss_ce": 5.697394226444885e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 951249684,
+      "step": 5543
+    },
+    {
+      "epoch": 62.94050991501416,
+      "grad_norm": 2.757708236252594,
+      "learning_rate": 5e-06,
+      "loss": 0.0497,
+      "num_input_tokens_seen": 951421544,
+      "step": 5544
+    },
+    {
+      "epoch": 62.94050991501416,
+      "loss": 0.08457933366298676,
+      "loss_ce": 6.090298847993836e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 951421544,
+      "step": 5544
+    },
+    {
+      "epoch": 62.95184135977337,
+      "grad_norm": 3.1680387240008447,
+      "learning_rate": 5e-06,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 951591828,
+      "step": 5545
+    },
+    {
+      "epoch": 62.95184135977337,
+      "loss": 0.047847069799900055,
+      "loss_ce": 2.602555468911305e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 951591828,
+      "step": 5545
+    },
+    {
+      "epoch": 62.963172804532576,
+      "grad_norm": 3.49766816083566,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 951762252,
+      "step": 5546
+    },
+    {
+      "epoch": 62.963172804532576,
+      "loss": 0.04163876175880432,
+      "loss_ce": 4.3301130062900484e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 951762252,
+      "step": 5546
+    },
+    {
+      "epoch": 62.97450424929178,
+      "grad_norm": 3.5518985165046186,
+      "learning_rate": 5e-06,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 951933848,
+      "step": 5547
+    },
+    {
+      "epoch": 62.97450424929178,
+      "loss": 0.0535704642534256,
+      "loss_ce": 2.736984970397316e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 951933848,
+      "step": 5547
+    },
+    {
+      "epoch": 62.98583569405099,
+      "grad_norm": 5.483441912411955,
+      "learning_rate": 5e-06,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 952105424,
+      "step": 5548
+    },
+    {
+      "epoch": 62.98583569405099,
+      "loss": 0.03674546256661415,
+      "loss_ce": 4.807391087524593e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 952105424,
+      "step": 5548
+    },
+    {
+      "epoch": 62.9971671388102,
+      "grad_norm": 3.222507785108215,
+      "learning_rate": 5e-06,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 952275744,
+      "step": 5549
+    },
+    {
+      "epoch": 62.9971671388102,
+      "loss": 0.03499796986579895,
+      "loss_ce": 4.008049290860072e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 952275744,
+      "step": 5549
+    },
+    {
+      "epoch": 62.9971671388102,
+      "loss": 0.022512448951601982,
+      "loss_ce": 4.388198794913478e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 952318660,
+      "step": 5549
+    },
+    {
+      "epoch": 63.008498583569406,
+      "grad_norm": 3.0726285121250587,
+      "learning_rate": 5e-06,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 952446708,
+      "step": 5550
+    },
+    {
+      "epoch": 63.008498583569406,
+      "loss": 0.0453701987862587,
+      "loss_ce": 3.633755841292441e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 952446708,
+      "step": 5550
+    },
+    {
+      "epoch": 63.01983002832861,
+      "grad_norm": 8.066609775491115,
+      "learning_rate": 5e-06,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 952618676,
+      "step": 5551
+    },
+    {
+      "epoch": 63.01983002832861,
+      "loss": 0.07753059267997742,
+      "loss_ce": 1.5940637240419164e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 952618676,
+      "step": 5551
+    },
+    {
+      "epoch": 63.03116147308782,
+      "grad_norm": 3.4605414647529296,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 952789532,
+      "step": 5552
+    },
+    {
+      "epoch": 63.03116147308782,
+      "loss": 0.03073236159980297,
+      "loss_ce": 3.1678409868618473e-05,
+      "loss_iou": 0.072265625,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 952789532,
+      "step": 5552
+    },
+    {
+      "epoch": 63.04249291784703,
+      "grad_norm": 3.516790101144911,
+      "learning_rate": 5e-06,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 952960124,
+      "step": 5553
+    },
+    {
+      "epoch": 63.04249291784703,
+      "loss": 0.04897896945476532,
+      "loss_ce": 8.980993879958987e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 952960124,
+      "step": 5553
+    },
+    {
+      "epoch": 63.053824362606235,
+      "grad_norm": 3.1040856488604938,
+      "learning_rate": 5e-06,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 953131736,
+      "step": 5554
+    },
+    {
+      "epoch": 63.053824362606235,
+      "loss": 0.026958251371979713,
+      "loss_ce": 2.648955342010595e-05,
+      "loss_iou": 0.09130859375,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 953131736,
+      "step": 5554
+    },
+    {
+      "epoch": 63.065155807365436,
+      "grad_norm": 2.2572768312065623,
+      "learning_rate": 5e-06,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 953302820,
+      "step": 5555
+    },
+    {
+      "epoch": 63.065155807365436,
+      "loss": 0.11343389749526978,
+      "loss_ce": 1.53264518303331e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 953302820,
+      "step": 5555
+    },
+    {
+      "epoch": 63.07648725212464,
+      "grad_norm": 1.6245283851401011,
+      "learning_rate": 5e-06,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 953472924,
+      "step": 5556
+    },
+    {
+      "epoch": 63.07648725212464,
+      "loss": 0.09962118417024612,
+      "loss_ce": 2.7067460905527696e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 953472924,
+      "step": 5556
+    },
+    {
+      "epoch": 63.08781869688385,
+      "grad_norm": 1.845728616668121,
+      "learning_rate": 5e-06,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 953642624,
+      "step": 5557
+    },
+    {
+      "epoch": 63.08781869688385,
+      "loss": 0.05160154402256012,
+      "loss_ce": 4.2095598473679274e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 953642624,
+      "step": 5557
+    },
+    {
+      "epoch": 63.09915014164306,
+      "grad_norm": 2.983284150019853,
+      "learning_rate": 5e-06,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 953813464,
+      "step": 5558
+    },
+    {
+      "epoch": 63.09915014164306,
+      "loss": 0.04296734929084778,
+      "loss_ce": 2.9116959922248498e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 953813464,
+      "step": 5558
+    },
+    {
+      "epoch": 63.110481586402265,
+      "grad_norm": 3.4295234938795858,
+      "learning_rate": 5e-06,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 953985092,
+      "step": 5559
+    },
+    {
+      "epoch": 63.110481586402265,
+      "loss": 0.04968930780887604,
+      "loss_ce": 2.195028901041951e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 953985092,
+      "step": 5559
+    },
+    {
+      "epoch": 63.12181303116147,
+      "grad_norm": 3.1002552986878786,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 954156588,
+      "step": 5560
+    },
+    {
+      "epoch": 63.12181303116147,
+      "loss": 0.05151050165295601,
+      "loss_ce": 4.260712012182921e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 954156588,
+      "step": 5560
+    },
+    {
+      "epoch": 63.13314447592068,
+      "grad_norm": 3.0713185503779683,
+      "learning_rate": 5e-06,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 954328908,
+      "step": 5561
+    },
+    {
+      "epoch": 63.13314447592068,
+      "loss": 0.05751762539148331,
+      "loss_ce": 5.3026731620775536e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 954328908,
+      "step": 5561
+    },
+    {
+      "epoch": 63.14447592067989,
+      "grad_norm": 3.7568378889500935,
+      "learning_rate": 5e-06,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 954501452,
+      "step": 5562
+    },
+    {
+      "epoch": 63.14447592067989,
+      "loss": 0.05231466516852379,
+      "loss_ce": 3.80574056180194e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 954501452,
+      "step": 5562
+    },
+    {
+      "epoch": 63.155807365439095,
+      "grad_norm": 3.879245840564028,
+      "learning_rate": 5e-06,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 954672428,
+      "step": 5563
+    },
+    {
+      "epoch": 63.155807365439095,
+      "loss": 0.03197059407830238,
+      "loss_ce": 4.920885112369433e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 954672428,
+      "step": 5563
+    },
+    {
+      "epoch": 63.1671388101983,
+      "grad_norm": 4.029669843159996,
+      "learning_rate": 5e-06,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 954842956,
+      "step": 5564
+    },
+    {
+      "epoch": 63.1671388101983,
+      "loss": 0.0496329739689827,
+      "loss_ce": 4.1908337152563035e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 954842956,
+      "step": 5564
+    },
+    {
+      "epoch": 63.17847025495751,
+      "grad_norm": 3.405893534186781,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 955014900,
+      "step": 5565
+    },
+    {
+      "epoch": 63.17847025495751,
+      "loss": 0.06293866783380508,
+      "loss_ce": 4.193765198579058e-05,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 955014900,
+      "step": 5565
+    },
+    {
+      "epoch": 63.18980169971671,
+      "grad_norm": 3.6858397617786784,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 955185804,
+      "step": 5566
+    },
+    {
+      "epoch": 63.18980169971671,
+      "loss": 0.03252874314785004,
+      "loss_ce": 2.7520247385837138e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 955185804,
+      "step": 5566
+    },
+    {
+      "epoch": 63.20113314447592,
+      "grad_norm": 3.4877211384962377,
+      "learning_rate": 5e-06,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 955357280,
+      "step": 5567
+    },
+    {
+      "epoch": 63.20113314447592,
+      "loss": 0.0366751104593277,
+      "loss_ce": 2.3499407689087093e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 955357280,
+      "step": 5567
+    },
+    {
+      "epoch": 63.212464589235125,
+      "grad_norm": 2.789096633978577,
+      "learning_rate": 5e-06,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 955528236,
+      "step": 5568
+    },
+    {
+      "epoch": 63.212464589235125,
+      "loss": 0.03407943993806839,
+      "loss_ce": 2.1822495909873396e-05,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 955528236,
+      "step": 5568
+    },
+    {
+      "epoch": 63.22379603399433,
+      "grad_norm": 3.1484394802352003,
+      "learning_rate": 5e-06,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 955699884,
+      "step": 5569
+    },
+    {
+      "epoch": 63.22379603399433,
+      "loss": 0.09011122584342957,
+      "loss_ce": 0.001205886946991086,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 955699884,
+      "step": 5569
+    },
+    {
+      "epoch": 63.23512747875354,
+      "grad_norm": 3.5739299590467675,
+      "learning_rate": 5e-06,
+      "loss": 0.033,
+      "num_input_tokens_seen": 955871648,
+      "step": 5570
+    },
+    {
+      "epoch": 63.23512747875354,
+      "loss": 0.036277756094932556,
+      "loss_ce": 3.8133817724883556e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 955871648,
+      "step": 5570
+    },
+    {
+      "epoch": 63.24645892351275,
+      "grad_norm": 3.2392655078018016,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 956042996,
+      "step": 5571
+    },
+    {
+      "epoch": 63.24645892351275,
+      "loss": 0.029277481138706207,
+      "loss_ce": 2.6382684154668823e-05,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 956042996,
+      "step": 5571
+    },
+    {
+      "epoch": 63.257790368271955,
+      "grad_norm": 2.5369535907591034,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 956215116,
+      "step": 5572
+    },
+    {
+      "epoch": 63.257790368271955,
+      "loss": 0.030111264437437057,
+      "loss_ce": 2.093338844133541e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 956215116,
+      "step": 5572
+    },
+    {
+      "epoch": 63.26912181303116,
+      "grad_norm": 17.48850390691409,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 956385932,
+      "step": 5573
+    },
+    {
+      "epoch": 63.26912181303116,
+      "loss": 0.04315242916345596,
+      "loss_ce": 4.6352201025001705e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 956385932,
+      "step": 5573
+    },
+    {
+      "epoch": 63.28045325779037,
+      "grad_norm": 3.780119713259165,
+      "learning_rate": 5e-06,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 956557840,
+      "step": 5574
+    },
+    {
+      "epoch": 63.28045325779037,
+      "loss": 0.06662124395370483,
+      "loss_ce": 1.663565490162e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 956557840,
+      "step": 5574
+    },
+    {
+      "epoch": 63.29178470254958,
+      "grad_norm": 3.856453084559645,
+      "learning_rate": 5e-06,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 956728512,
+      "step": 5575
+    },
+    {
+      "epoch": 63.29178470254958,
+      "loss": 0.11761099100112915,
+      "loss_ce": 7.253909279825166e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 956728512,
+      "step": 5575
+    },
+    {
+      "epoch": 63.303116147308785,
+      "grad_norm": 4.096762063490481,
+      "learning_rate": 5e-06,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 956900412,
+      "step": 5576
+    },
+    {
+      "epoch": 63.303116147308785,
+      "loss": 0.03702767938375473,
+      "loss_ce": 4.037451435578987e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 956900412,
+      "step": 5576
+    },
+    {
+      "epoch": 63.314447592067985,
+      "grad_norm": 3.464112655615027,
+      "learning_rate": 5e-06,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 957071180,
+      "step": 5577
+    },
+    {
+      "epoch": 63.314447592067985,
+      "loss": 0.03874988108873367,
+      "loss_ce": 5.3590476454701275e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 957071180,
+      "step": 5577
+    },
+    {
+      "epoch": 63.32577903682719,
+      "grad_norm": 3.7135803523094784,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 957239820,
+      "step": 5578
+    },
+    {
+      "epoch": 63.32577903682719,
+      "loss": 0.0559682622551918,
+      "loss_ce": 7.531804294558242e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 957239820,
+      "step": 5578
+    },
+    {
+      "epoch": 63.3371104815864,
+      "grad_norm": 3.674811826499546,
+      "learning_rate": 5e-06,
+      "loss": 0.0851,
+      "num_input_tokens_seen": 957409972,
+      "step": 5579
+    },
+    {
+      "epoch": 63.3371104815864,
+      "loss": 0.10021825134754181,
+      "loss_ce": 2.904361826949753e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 957409972,
+      "step": 5579
+    },
+    {
+      "epoch": 63.34844192634561,
+      "grad_norm": 3.0014446453364187,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 957581456,
+      "step": 5580
+    },
+    {
+      "epoch": 63.34844192634561,
+      "loss": 0.13484829664230347,
+      "loss_ce": 5.214768316363916e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 957581456,
+      "step": 5580
+    },
+    {
+      "epoch": 63.359773371104815,
+      "grad_norm": 4.104451018042735,
+      "learning_rate": 5e-06,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 957752524,
+      "step": 5581
+    },
+    {
+      "epoch": 63.359773371104815,
+      "loss": 0.03547465801239014,
+      "loss_ce": 2.8488504540291615e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 957752524,
+      "step": 5581
+    },
+    {
+      "epoch": 63.37110481586402,
+      "grad_norm": 3.572085229707181,
+      "learning_rate": 5e-06,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 957923436,
+      "step": 5582
+    },
+    {
+      "epoch": 63.37110481586402,
+      "loss": 0.04619760066270828,
+      "loss_ce": 2.450751344440505e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 957923436,
+      "step": 5582
+    },
+    {
+      "epoch": 63.38243626062323,
+      "grad_norm": 4.316242948746455,
+      "learning_rate": 5e-06,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 958095076,
+      "step": 5583
+    },
+    {
+      "epoch": 63.38243626062323,
+      "loss": 0.077049620449543,
+      "loss_ce": 3.088476660195738e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 958095076,
+      "step": 5583
+    },
+    {
+      "epoch": 63.39376770538244,
+      "grad_norm": 3.1183898338672473,
+      "learning_rate": 5e-06,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 958265652,
+      "step": 5584
+    },
+    {
+      "epoch": 63.39376770538244,
+      "loss": 0.026530340313911438,
+      "loss_ce": 5.6342134485021234e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 958265652,
+      "step": 5584
+    },
+    {
+      "epoch": 63.405099150141645,
+      "grad_norm": 3.245622560958497,
+      "learning_rate": 5e-06,
+      "loss": 0.045,
+      "num_input_tokens_seen": 958436604,
+      "step": 5585
+    },
+    {
+      "epoch": 63.405099150141645,
+      "loss": 0.03134666383266449,
+      "loss_ce": 2.0372652215883136e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 958436604,
+      "step": 5585
+    },
+    {
+      "epoch": 63.41643059490085,
+      "grad_norm": 3.511396698237713,
+      "learning_rate": 5e-06,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 958608500,
+      "step": 5586
+    },
+    {
+      "epoch": 63.41643059490085,
+      "loss": 0.09213487803936005,
+      "loss_ce": 3.281996032455936e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 958608500,
+      "step": 5586
+    },
+    {
+      "epoch": 63.42776203966006,
+      "grad_norm": 2.7535135754958016,
+      "learning_rate": 5e-06,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 958780004,
+      "step": 5587
+    },
+    {
+      "epoch": 63.42776203966006,
+      "loss": 0.024539779871702194,
+      "loss_ce": 3.416433901293203e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 958780004,
+      "step": 5587
+    },
+    {
+      "epoch": 63.43909348441926,
+      "grad_norm": 2.2488181785794845,
+      "learning_rate": 5e-06,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 958951728,
+      "step": 5588
+    },
+    {
+      "epoch": 63.43909348441926,
+      "loss": 0.03219490498304367,
+      "loss_ce": 1.411808352713706e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 958951728,
+      "step": 5588
+    },
+    {
+      "epoch": 63.45042492917847,
+      "grad_norm": 4.044006306081819,
+      "learning_rate": 5e-06,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 959123764,
+      "step": 5589
+    },
+    {
+      "epoch": 63.45042492917847,
+      "loss": 0.027514085173606873,
+      "loss_ce": 0.0017343595391139388,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 959123764,
+      "step": 5589
+    },
+    {
+      "epoch": 63.461756373937675,
+      "grad_norm": 2.0575514676075994,
+      "learning_rate": 5e-06,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 959294788,
+      "step": 5590
+    },
+    {
+      "epoch": 63.461756373937675,
+      "loss": 0.04627084732055664,
+      "loss_ce": 2.1456908143591136e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 959294788,
+      "step": 5590
+    },
+    {
+      "epoch": 63.47308781869688,
+      "grad_norm": 2.3870953048577026,
+      "learning_rate": 5e-06,
+      "loss": 0.067,
+      "num_input_tokens_seen": 959466308,
+      "step": 5591
+    },
+    {
+      "epoch": 63.47308781869688,
+      "loss": 0.1199498176574707,
+      "loss_ce": 7.677252870053053e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 959466308,
+      "step": 5591
+    },
+    {
+      "epoch": 63.48441926345609,
+      "grad_norm": 4.139935714053537,
+      "learning_rate": 5e-06,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 959637912,
+      "step": 5592
+    },
+    {
+      "epoch": 63.48441926345609,
+      "loss": 0.05224212259054184,
+      "loss_ce": 2.6541882107267156e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 959637912,
+      "step": 5592
+    },
+    {
+      "epoch": 63.4957507082153,
+      "grad_norm": 4.056118180792055,
+      "learning_rate": 5e-06,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 959809196,
+      "step": 5593
+    },
+    {
+      "epoch": 63.4957507082153,
+      "loss": 0.03815827891230583,
+      "loss_ce": 2.6566141968942247e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 959809196,
+      "step": 5593
+    },
+    {
+      "epoch": 63.507082152974505,
+      "grad_norm": 3.423991928067274,
+      "learning_rate": 5e-06,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 959978684,
+      "step": 5594
+    },
+    {
+      "epoch": 63.507082152974505,
+      "loss": 0.04080747812986374,
+      "loss_ce": 0.00018858144176192582,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 959978684,
+      "step": 5594
+    },
+    {
+      "epoch": 63.51841359773371,
+      "grad_norm": 3.04059163856433,
+      "learning_rate": 5e-06,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 960150596,
+      "step": 5595
+    },
+    {
+      "epoch": 63.51841359773371,
+      "loss": 0.06325870752334595,
+      "loss_ce": 7.206252485048026e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 960150596,
+      "step": 5595
+    },
+    {
+      "epoch": 63.52974504249292,
+      "grad_norm": 3.2057911955594505,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 960322312,
+      "step": 5596
+    },
+    {
+      "epoch": 63.52974504249292,
+      "loss": 0.03003763034939766,
+      "loss_ce": 3.885219484800473e-05,
+      "loss_iou": 0.051513671875,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 960322312,
+      "step": 5596
+    },
+    {
+      "epoch": 63.54107648725213,
+      "grad_norm": 3.0136179664321525,
+      "learning_rate": 5e-06,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 960494360,
+      "step": 5597
+    },
+    {
+      "epoch": 63.54107648725213,
+      "loss": 0.04219797998666763,
+      "loss_ce": 2.268829121021554e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 960494360,
+      "step": 5597
+    },
+    {
+      "epoch": 63.552407932011334,
+      "grad_norm": 3.304524073190295,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 960665812,
+      "step": 5598
+    },
+    {
+      "epoch": 63.552407932011334,
+      "loss": 0.05926670506596565,
+      "loss_ce": 0.00016941546346060932,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 960665812,
+      "step": 5598
+    },
+    {
+      "epoch": 63.563739376770535,
+      "grad_norm": 3.08255476666748,
+      "learning_rate": 5e-06,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 960836720,
+      "step": 5599
+    },
+    {
+      "epoch": 63.563739376770535,
+      "loss": 0.09824319183826447,
+      "loss_ce": 3.7625391996698454e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 960836720,
+      "step": 5599
+    },
+    {
+      "epoch": 63.57507082152974,
+      "grad_norm": 3.0751660740568574,
+      "learning_rate": 5e-06,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 961007856,
+      "step": 5600
+    },
+    {
+      "epoch": 63.57507082152974,
+      "loss": 0.03729415684938431,
+      "loss_ce": 3.2191150239668787e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 961007856,
+      "step": 5600
+    },
+    {
+      "epoch": 63.58640226628895,
+      "grad_norm": 3.238049387634643,
+      "learning_rate": 5e-06,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 961179468,
+      "step": 5601
+    },
+    {
+      "epoch": 63.58640226628895,
+      "loss": 0.1232166439294815,
+      "loss_ce": 6.296264473348856e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 961179468,
+      "step": 5601
+    },
+    {
+      "epoch": 63.59773371104816,
+      "grad_norm": 3.559327683167962,
+      "learning_rate": 5e-06,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 961349880,
+      "step": 5602
+    },
+    {
+      "epoch": 63.59773371104816,
+      "loss": 0.0860898345708847,
+      "loss_ce": 6.0780505009461194e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 961349880,
+      "step": 5602
+    },
+    {
+      "epoch": 63.609065155807365,
+      "grad_norm": 3.319702796243184,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 961521876,
+      "step": 5603
+    },
+    {
+      "epoch": 63.609065155807365,
+      "loss": 0.035224370658397675,
+      "loss_ce": 6.81184756103903e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 961521876,
+      "step": 5603
+    },
+    {
+      "epoch": 63.62039660056657,
+      "grad_norm": 4.349612899861339,
+      "learning_rate": 5e-06,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 961693552,
+      "step": 5604
+    },
+    {
+      "epoch": 63.62039660056657,
+      "loss": 0.0656881183385849,
+      "loss_ce": 0.00013636643416248262,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 961693552,
+      "step": 5604
+    },
+    {
+      "epoch": 63.63172804532578,
+      "grad_norm": 3.4759568008340995,
+      "learning_rate": 5e-06,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 961864420,
+      "step": 5605
+    },
+    {
+      "epoch": 63.63172804532578,
+      "loss": 0.091413713991642,
+      "loss_ce": 1.356533721263986e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 961864420,
+      "step": 5605
+    },
+    {
+      "epoch": 63.64305949008499,
+      "grad_norm": 3.291281051661945,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 962036372,
+      "step": 5606
+    },
+    {
+      "epoch": 63.64305949008499,
+      "loss": 0.04150788113474846,
+      "loss_ce": 1.9234055798733607e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 962036372,
+      "step": 5606
+    },
+    {
+      "epoch": 63.654390934844194,
+      "grad_norm": 3.307902196414728,
+      "learning_rate": 5e-06,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 962208460,
+      "step": 5607
+    },
+    {
+      "epoch": 63.654390934844194,
+      "loss": 0.05206012725830078,
+      "loss_ce": 5.817593773826957e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 962208460,
+      "step": 5607
+    },
+    {
+      "epoch": 63.6657223796034,
+      "grad_norm": 3.84998803033237,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 962379968,
+      "step": 5608
+    },
+    {
+      "epoch": 63.6657223796034,
+      "loss": 0.0449778214097023,
+      "loss_ce": 4.06864273827523e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 962379968,
+      "step": 5608
+    },
+    {
+      "epoch": 63.67705382436261,
+      "grad_norm": 3.390590423753857,
+      "learning_rate": 5e-06,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 962550852,
+      "step": 5609
+    },
+    {
+      "epoch": 63.67705382436261,
+      "loss": 0.044905029237270355,
+      "loss_ce": 5.944907752564177e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 962550852,
+      "step": 5609
+    },
+    {
+      "epoch": 63.68838526912181,
+      "grad_norm": 3.242737635489924,
+      "learning_rate": 5e-06,
+      "loss": 0.054,
+      "num_input_tokens_seen": 962721656,
+      "step": 5610
+    },
+    {
+      "epoch": 63.68838526912181,
+      "loss": 0.03350004553794861,
+      "loss_ce": 5.2782394050154835e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 962721656,
+      "step": 5610
+    },
+    {
+      "epoch": 63.69971671388102,
+      "grad_norm": 3.5341141303856554,
+      "learning_rate": 5e-06,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 962893592,
+      "step": 5611
+    },
+    {
+      "epoch": 63.69971671388102,
+      "loss": 0.02900099754333496,
+      "loss_ce": 7.033486326690763e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 962893592,
+      "step": 5611
+    },
+    {
+      "epoch": 63.711048158640224,
+      "grad_norm": 3.648832531736882,
+      "learning_rate": 5e-06,
+      "loss": 0.067,
+      "num_input_tokens_seen": 963065260,
+      "step": 5612
+    },
+    {
+      "epoch": 63.711048158640224,
+      "loss": 0.03863013535737991,
+      "loss_ce": 2.5400549930054694e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 963065260,
+      "step": 5612
+    },
+    {
+      "epoch": 63.72237960339943,
+      "grad_norm": 3.963323910051783,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 963236236,
+      "step": 5613
+    },
+    {
+      "epoch": 63.72237960339943,
+      "loss": 0.08686865866184235,
+      "loss_ce": 1.56275782501325e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 963236236,
+      "step": 5613
+    },
+    {
+      "epoch": 63.73371104815864,
+      "grad_norm": 3.0671449103716277,
+      "learning_rate": 5e-06,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 963406420,
+      "step": 5614
+    },
+    {
+      "epoch": 63.73371104815864,
+      "loss": 0.04987305402755737,
+      "loss_ce": 3.7846475606784225e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 963406420,
+      "step": 5614
+    },
+    {
+      "epoch": 63.74504249291785,
+      "grad_norm": 2.853354158370971,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 963578108,
+      "step": 5615
+    },
+    {
+      "epoch": 63.74504249291785,
+      "loss": 0.03520198538899422,
+      "loss_ce": 3.047668724320829e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 963578108,
+      "step": 5615
+    },
+    {
+      "epoch": 63.756373937677054,
+      "grad_norm": 7.058838399571512,
+      "learning_rate": 5e-06,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 963749856,
+      "step": 5616
+    },
+    {
+      "epoch": 63.756373937677054,
+      "loss": 0.13913656771183014,
+      "loss_ce": 3.744324931176379e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 963749856,
+      "step": 5616
+    },
+    {
+      "epoch": 63.76770538243626,
+      "grad_norm": 3.105532113430941,
+      "learning_rate": 5e-06,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 963921884,
+      "step": 5617
+    },
+    {
+      "epoch": 63.76770538243626,
+      "loss": 0.04394109547138214,
+      "loss_ce": 1.1040867320843972e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 963921884,
+      "step": 5617
+    },
+    {
+      "epoch": 63.77903682719547,
+      "grad_norm": 4.326486056073228,
+      "learning_rate": 5e-06,
+      "loss": 0.047,
+      "num_input_tokens_seen": 964093564,
+      "step": 5618
+    },
+    {
+      "epoch": 63.77903682719547,
+      "loss": 0.03333599492907524,
+      "loss_ce": 2.605864210636355e-05,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 964093564,
+      "step": 5618
+    },
+    {
+      "epoch": 63.79036827195468,
+      "grad_norm": 3.1073288459533672,
+      "learning_rate": 5e-06,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 964265312,
+      "step": 5619
+    },
+    {
+      "epoch": 63.79036827195468,
+      "loss": 0.08995333313941956,
+      "loss_ce": 0.00010958635539282113,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 964265312,
+      "step": 5619
+    },
+    {
+      "epoch": 63.801699716713884,
+      "grad_norm": 2.665485201457284,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 964433932,
+      "step": 5620
+    },
+    {
+      "epoch": 63.801699716713884,
+      "loss": 0.03465600311756134,
+      "loss_ce": 4.90689417347312e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 964433932,
+      "step": 5620
+    },
+    {
+      "epoch": 63.81303116147309,
+      "grad_norm": 3.001507436378676,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 964605876,
+      "step": 5621
+    },
+    {
+      "epoch": 63.81303116147309,
+      "loss": 0.028206557035446167,
+      "loss_ce": 3.883427052642219e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 964605876,
+      "step": 5621
+    },
+    {
+      "epoch": 63.82436260623229,
+      "grad_norm": 3.592585108157437,
+      "learning_rate": 5e-06,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 964776820,
+      "step": 5622
+    },
+    {
+      "epoch": 63.82436260623229,
+      "loss": 0.06478553265333176,
+      "loss_ce": 5.774579040007666e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 964776820,
+      "step": 5622
+    },
+    {
+      "epoch": 63.8356940509915,
+      "grad_norm": 3.322501747688564,
+      "learning_rate": 5e-06,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 964947572,
+      "step": 5623
+    },
+    {
+      "epoch": 63.8356940509915,
+      "loss": 0.10199524462223053,
+      "loss_ce": 3.6022509448230267e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 964947572,
+      "step": 5623
+    },
+    {
+      "epoch": 63.84702549575071,
+      "grad_norm": 2.790117742190208,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 965119232,
+      "step": 5624
+    },
+    {
+      "epoch": 63.84702549575071,
+      "loss": 0.03153211623430252,
+      "loss_ce": 5.323695586412214e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 965119232,
+      "step": 5624
+    },
+    {
+      "epoch": 63.858356940509914,
+      "grad_norm": 3.495267003240254,
+      "learning_rate": 5e-06,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 965291060,
+      "step": 5625
+    },
+    {
+      "epoch": 63.858356940509914,
+      "loss": 0.0597134605050087,
+      "loss_ce": 4.3967294914182276e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 965291060,
+      "step": 5625
+    },
+    {
+      "epoch": 63.86968838526912,
+      "grad_norm": 3.5538715811434285,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 965461892,
+      "step": 5626
+    },
+    {
+      "epoch": 63.86968838526912,
+      "loss": 0.04024675861001015,
+      "loss_ce": 2.4590355678810738e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 965461892,
+      "step": 5626
+    },
+    {
+      "epoch": 63.88101983002833,
+      "grad_norm": 3.342196276751497,
+      "learning_rate": 5e-06,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 965633768,
+      "step": 5627
+    },
+    {
+      "epoch": 63.88101983002833,
+      "loss": 0.04424550384283066,
+      "loss_ce": 5.605052865575999e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 965633768,
+      "step": 5627
+    },
+    {
+      "epoch": 63.892351274787536,
+      "grad_norm": 3.9672739991304624,
+      "learning_rate": 5e-06,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 965804404,
+      "step": 5628
+    },
+    {
+      "epoch": 63.892351274787536,
+      "loss": 0.055883850902318954,
+      "loss_ce": 2.1421208657557145e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 965804404,
+      "step": 5628
+    },
+    {
+      "epoch": 63.903682719546744,
+      "grad_norm": 3.8018861919290634,
+      "learning_rate": 5e-06,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 965974528,
+      "step": 5629
+    },
+    {
+      "epoch": 63.903682719546744,
+      "loss": 0.07362597435712814,
+      "loss_ce": 4.8089077608892694e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 965974528,
+      "step": 5629
+    },
+    {
+      "epoch": 63.91501416430595,
+      "grad_norm": 3.8577425551825084,
+      "learning_rate": 5e-06,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 966146368,
+      "step": 5630
+    },
+    {
+      "epoch": 63.91501416430595,
+      "loss": 0.057474687695503235,
+      "loss_ce": 4.0604638343211263e-05,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 966146368,
+      "step": 5630
+    },
+    {
+      "epoch": 63.92634560906516,
+      "grad_norm": 3.0837934175541104,
+      "learning_rate": 5e-06,
+      "loss": 0.061,
+      "num_input_tokens_seen": 966318108,
+      "step": 5631
+    },
+    {
+      "epoch": 63.92634560906516,
+      "loss": 0.08532983064651489,
+      "loss_ce": 4.8456768126925454e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 966318108,
+      "step": 5631
+    },
+    {
+      "epoch": 63.93767705382436,
+      "grad_norm": 2.5832236290080757,
+      "learning_rate": 5e-06,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 966488820,
+      "step": 5632
+    },
+    {
+      "epoch": 63.93767705382436,
+      "loss": 0.03581484034657478,
+      "loss_ce": 3.2980107789626345e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 966488820,
+      "step": 5632
+    },
+    {
+      "epoch": 63.94900849858357,
+      "grad_norm": 3.659230430175558,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 966660548,
+      "step": 5633
+    },
+    {
+      "epoch": 63.94900849858357,
+      "loss": 0.029485203325748444,
+      "loss_ce": 3.574071160983294e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 966660548,
+      "step": 5633
+    },
+    {
+      "epoch": 63.960339943342774,
+      "grad_norm": 3.420272409614578,
+      "learning_rate": 5e-06,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 966832440,
+      "step": 5634
+    },
+    {
+      "epoch": 63.960339943342774,
+      "loss": 0.05185551568865776,
+      "loss_ce": 2.1407164240372367e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 966832440,
+      "step": 5634
+    },
+    {
+      "epoch": 63.97167138810198,
+      "grad_norm": 3.0803276120719647,
+      "learning_rate": 5e-06,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 967004168,
+      "step": 5635
+    },
+    {
+      "epoch": 63.97167138810198,
+      "loss": 0.040978141129016876,
+      "loss_ce": 3.881048542098142e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 967004168,
+      "step": 5635
+    },
+    {
+      "epoch": 63.98300283286119,
+      "grad_norm": 3.1702988538022328,
+      "learning_rate": 5e-06,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 967175868,
+      "step": 5636
+    },
+    {
+      "epoch": 63.98300283286119,
+      "loss": 0.029574017971754074,
+      "loss_ce": 0.0001092967067961581,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 967175868,
+      "step": 5636
+    },
+    {
+      "epoch": 63.994334277620396,
+      "grad_norm": 3.194526194787991,
+      "learning_rate": 5e-06,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 967345240,
+      "step": 5637
+    },
+    {
+      "epoch": 63.994334277620396,
+      "loss": 0.03083980828523636,
+      "loss_ce": 4.757327405968681e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 967345240,
+      "step": 5637
+    },
+    {
+      "epoch": 63.994334277620396,
+      "loss": 0.06418736279010773,
+      "loss_ce": 0.00016148097347468138,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 967431080,
+      "step": 5637
+    },
+    {
+      "epoch": 64.0056657223796,
+      "grad_norm": 3.4043921666610912,
+      "learning_rate": 5e-06,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 967517088,
+      "step": 5638
+    },
+    {
+      "epoch": 64.0056657223796,
+      "loss": 0.05650923401117325,
+      "loss_ce": 0.00011274752614554018,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 967517088,
+      "step": 5638
+    },
+    {
+      "epoch": 64.01699716713881,
+      "grad_norm": 3.3673918855447624,
+      "learning_rate": 5e-06,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 967688416,
+      "step": 5639
+    },
+    {
+      "epoch": 64.01699716713881,
+      "loss": 0.04392354190349579,
+      "loss_ce": 3.926271165255457e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 967688416,
+      "step": 5639
+    },
+    {
+      "epoch": 64.02832861189802,
+      "grad_norm": 2.7462258795519543,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 967857656,
+      "step": 5640
+    },
+    {
+      "epoch": 64.02832861189802,
+      "loss": 0.053232863545417786,
+      "loss_ce": 4.072372757946141e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 967857656,
+      "step": 5640
+    },
+    {
+      "epoch": 64.03966005665723,
+      "grad_norm": 3.321336390282397,
+      "learning_rate": 5e-06,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 968029768,
+      "step": 5641
+    },
+    {
+      "epoch": 64.03966005665723,
+      "loss": 0.0696585476398468,
+      "loss_ce": 1.74314554897137e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 968029768,
+      "step": 5641
+    },
+    {
+      "epoch": 64.05099150141643,
+      "grad_norm": 3.45511423659337,
+      "learning_rate": 5e-06,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 968201456,
+      "step": 5642
+    },
+    {
+      "epoch": 64.05099150141643,
+      "loss": 0.04316634684801102,
+      "loss_ce": 1.4490167814074084e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 968201456,
+      "step": 5642
+    },
+    {
+      "epoch": 64.06232294617564,
+      "grad_norm": 3.660624579416374,
+      "learning_rate": 5e-06,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 968373120,
+      "step": 5643
+    },
+    {
+      "epoch": 64.06232294617564,
+      "loss": 0.05655057728290558,
+      "loss_ce": 1.6763298845035024e-05,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 968373120,
+      "step": 5643
+    },
+    {
+      "epoch": 64.07365439093485,
+      "grad_norm": 3.1247914860741974,
+      "learning_rate": 5e-06,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 968542272,
+      "step": 5644
+    },
+    {
+      "epoch": 64.07365439093485,
+      "loss": 0.03748254477977753,
+      "loss_ce": 2.22196031245403e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 968542272,
+      "step": 5644
+    },
+    {
+      "epoch": 64.08498583569406,
+      "grad_norm": 2.7439084129009568,
+      "learning_rate": 5e-06,
+      "loss": 0.059,
+      "num_input_tokens_seen": 968712796,
+      "step": 5645
+    },
+    {
+      "epoch": 64.08498583569406,
+      "loss": 0.0821637213230133,
+      "loss_ce": 2.5654127966845408e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 968712796,
+      "step": 5645
+    },
+    {
+      "epoch": 64.09631728045326,
+      "grad_norm": 2.269957941464663,
+      "learning_rate": 5e-06,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 968884624,
+      "step": 5646
+    },
+    {
+      "epoch": 64.09631728045326,
+      "loss": 0.023347310721874237,
+      "loss_ce": 2.4251186914625578e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.004669189453125,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 968884624,
+      "step": 5646
+    },
+    {
+      "epoch": 64.10764872521247,
+      "grad_norm": 2.556019841203874,
+      "learning_rate": 5e-06,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 969056416,
+      "step": 5647
+    },
+    {
+      "epoch": 64.10764872521247,
+      "loss": 0.03283655643463135,
+      "loss_ce": 1.4902190741850063e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 969056416,
+      "step": 5647
+    },
+    {
+      "epoch": 64.11898016997168,
+      "grad_norm": 2.652508352726603,
+      "learning_rate": 5e-06,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 969227468,
+      "step": 5648
+    },
+    {
+      "epoch": 64.11898016997168,
+      "loss": 0.0322101004421711,
+      "loss_ce": 2.9315209758351557e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 969227468,
+      "step": 5648
+    },
+    {
+      "epoch": 64.13031161473087,
+      "grad_norm": 3.314436439791771,
+      "learning_rate": 5e-06,
+      "loss": 0.0783,
+      "num_input_tokens_seen": 969397608,
+      "step": 5649
+    },
+    {
+      "epoch": 64.13031161473087,
+      "loss": 0.09146112203598022,
+      "loss_ce": 4.571592580759898e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 969397608,
+      "step": 5649
+    },
+    {
+      "epoch": 64.14164305949008,
+      "grad_norm": 3.292731792346954,
+      "learning_rate": 5e-06,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 969569360,
+      "step": 5650
+    },
+    {
+      "epoch": 64.14164305949008,
+      "loss": 0.04446880519390106,
+      "loss_ce": 5.0470534915803e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 969569360,
+      "step": 5650
+    },
+    {
+      "epoch": 64.15297450424929,
+      "grad_norm": 3.1997910092185937,
+      "learning_rate": 5e-06,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 969740256,
+      "step": 5651
+    },
+    {
+      "epoch": 64.15297450424929,
+      "loss": 0.040051646530628204,
+      "loss_ce": 1.2582495401147753e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 969740256,
+      "step": 5651
+    },
+    {
+      "epoch": 64.1643059490085,
+      "grad_norm": 2.833640324178782,
+      "learning_rate": 5e-06,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 969911480,
+      "step": 5652
+    },
+    {
+      "epoch": 64.1643059490085,
+      "loss": 0.05397683382034302,
+      "loss_ce": 3.701582318171859e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 969911480,
+      "step": 5652
+    },
+    {
+      "epoch": 64.1756373937677,
+      "grad_norm": 3.052946854182395,
+      "learning_rate": 5e-06,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 970083080,
+      "step": 5653
+    },
+    {
+      "epoch": 64.1756373937677,
+      "loss": 0.07143557071685791,
+      "loss_ce": 5.496014273376204e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 970083080,
+      "step": 5653
+    },
+    {
+      "epoch": 64.18696883852691,
+      "grad_norm": 3.574714947392389,
+      "learning_rate": 5e-06,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 970255200,
+      "step": 5654
+    },
+    {
+      "epoch": 64.18696883852691,
+      "loss": 0.055493105202913284,
+      "loss_ce": 2.740791933319997e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 970255200,
+      "step": 5654
+    },
+    {
+      "epoch": 64.19830028328612,
+      "grad_norm": 8.874588734579559,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 970426736,
+      "step": 5655
+    },
+    {
+      "epoch": 64.19830028328612,
+      "loss": 0.0505923330783844,
+      "loss_ce": 3.996703162556514e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 970426736,
+      "step": 5655
+    },
+    {
+      "epoch": 64.20963172804532,
+      "grad_norm": 4.4667537098046886,
+      "learning_rate": 5e-06,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 970598584,
+      "step": 5656
+    },
+    {
+      "epoch": 64.20963172804532,
+      "loss": 0.0637560784816742,
+      "loss_ce": 4.300387081457302e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 970598584,
+      "step": 5656
+    },
+    {
+      "epoch": 64.22096317280453,
+      "grad_norm": 3.9314811588230287,
+      "learning_rate": 5e-06,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 970770300,
+      "step": 5657
+    },
+    {
+      "epoch": 64.22096317280453,
+      "loss": 0.029440617188811302,
+      "loss_ce": 3.6931080103386194e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 970770300,
+      "step": 5657
+    },
+    {
+      "epoch": 64.23229461756374,
+      "grad_norm": 4.7000634688927345,
+      "learning_rate": 5e-06,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 970941796,
+      "step": 5658
+    },
+    {
+      "epoch": 64.23229461756374,
+      "loss": 0.04870712012052536,
+      "loss_ce": 3.158202889608219e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 970941796,
+      "step": 5658
+    },
+    {
+      "epoch": 64.24362606232295,
+      "grad_norm": 3.3144629295899684,
+      "learning_rate": 5e-06,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 971112680,
+      "step": 5659
+    },
+    {
+      "epoch": 64.24362606232295,
+      "loss": 0.03153107315301895,
+      "loss_ce": 2.1673440642189234e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 971112680,
+      "step": 5659
+    },
+    {
+      "epoch": 64.25495750708215,
+      "grad_norm": 3.7107457610403656,
+      "learning_rate": 5e-06,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 971283676,
+      "step": 5660
+    },
+    {
+      "epoch": 64.25495750708215,
+      "loss": 0.03189505636692047,
+      "loss_ce": 1.9446108126430772e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 971283676,
+      "step": 5660
+    },
+    {
+      "epoch": 64.26628895184136,
+      "grad_norm": 3.3812816437690736,
+      "learning_rate": 5e-06,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 971455240,
+      "step": 5661
+    },
+    {
+      "epoch": 64.26628895184136,
+      "loss": 0.03896826505661011,
+      "loss_ce": 5.835268530063331e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 971455240,
+      "step": 5661
+    },
+    {
+      "epoch": 64.27762039660057,
+      "grad_norm": 3.5732360849154614,
+      "learning_rate": 5e-06,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 971626224,
+      "step": 5662
+    },
+    {
+      "epoch": 64.27762039660057,
+      "loss": 0.05453784763813019,
+      "loss_ce": 1.8197564713773318e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 971626224,
+      "step": 5662
+    },
+    {
+      "epoch": 64.28895184135978,
+      "grad_norm": 3.20346346620919,
+      "learning_rate": 5e-06,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 971796696,
+      "step": 5663
+    },
+    {
+      "epoch": 64.28895184135978,
+      "loss": 0.027630165219306946,
+      "loss_ce": 4.2273662984371185e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 971796696,
+      "step": 5663
+    },
+    {
+      "epoch": 64.30028328611898,
+      "grad_norm": 3.9955071759580933,
+      "learning_rate": 5e-06,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 971966764,
+      "step": 5664
+    },
+    {
+      "epoch": 64.30028328611898,
+      "loss": 0.04953522980213165,
+      "loss_ce": 5.097504617879167e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 971966764,
+      "step": 5664
+    },
+    {
+      "epoch": 64.31161473087819,
+      "grad_norm": 3.1604054124686374,
+      "learning_rate": 5e-06,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 972138996,
+      "step": 5665
+    },
+    {
+      "epoch": 64.31161473087819,
+      "loss": 0.030261635780334473,
+      "loss_ce": 3.397395630599931e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 972138996,
+      "step": 5665
+    },
+    {
+      "epoch": 64.3229461756374,
+      "grad_norm": 3.0790661118265463,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 972310740,
+      "step": 5666
+    },
+    {
+      "epoch": 64.3229461756374,
+      "loss": 0.036920733749866486,
+      "loss_ce": 2.4980761736514978e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 972310740,
+      "step": 5666
+    },
+    {
+      "epoch": 64.3342776203966,
+      "grad_norm": 3.63271527234115,
+      "learning_rate": 5e-06,
+      "loss": 0.068,
+      "num_input_tokens_seen": 972482284,
+      "step": 5667
+    },
+    {
+      "epoch": 64.3342776203966,
+      "loss": 0.065069779753685,
+      "loss_ce": 3.6813093174714595e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 972482284,
+      "step": 5667
+    },
+    {
+      "epoch": 64.34560906515581,
+      "grad_norm": 5.398248023016888,
+      "learning_rate": 5e-06,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 972653332,
+      "step": 5668
+    },
+    {
+      "epoch": 64.34560906515581,
+      "loss": 0.0391923226416111,
+      "loss_ce": 5.3527455747826025e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 972653332,
+      "step": 5668
+    },
+    {
+      "epoch": 64.35694050991502,
+      "grad_norm": 3.6619567478599806,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 972825500,
+      "step": 5669
+    },
+    {
+      "epoch": 64.35694050991502,
+      "loss": 0.03753100335597992,
+      "loss_ce": 4.015903323306702e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 972825500,
+      "step": 5669
+    },
+    {
+      "epoch": 64.36827195467423,
+      "grad_norm": 3.72495940807384,
+      "learning_rate": 5e-06,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 972997324,
+      "step": 5670
+    },
+    {
+      "epoch": 64.36827195467423,
+      "loss": 0.033766455948352814,
+      "loss_ce": 2.9272661777213216e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 972997324,
+      "step": 5670
+    },
+    {
+      "epoch": 64.37960339943342,
+      "grad_norm": 3.608396455096228,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 973167204,
+      "step": 5671
+    },
+    {
+      "epoch": 64.37960339943342,
+      "loss": 0.03851725906133652,
+      "loss_ce": 1.9336674085934646e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 973167204,
+      "step": 5671
+    },
+    {
+      "epoch": 64.39093484419263,
+      "grad_norm": 3.486497386678485,
+      "learning_rate": 5e-06,
+      "loss": 0.045,
+      "num_input_tokens_seen": 973338916,
+      "step": 5672
+    },
+    {
+      "epoch": 64.39093484419263,
+      "loss": 0.03376801311969757,
+      "loss_ce": 4.608709423337132e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 973338916,
+      "step": 5672
+    },
+    {
+      "epoch": 64.40226628895184,
+      "grad_norm": 3.539497225739517,
+      "learning_rate": 5e-06,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 973508200,
+      "step": 5673
+    },
+    {
+      "epoch": 64.40226628895184,
+      "loss": 0.03594028577208519,
+      "loss_ce": 2.109771958203055e-05,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 973508200,
+      "step": 5673
+    },
+    {
+      "epoch": 64.41359773371104,
+      "grad_norm": 3.514136880019449,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 973680328,
+      "step": 5674
+    },
+    {
+      "epoch": 64.41359773371104,
+      "loss": 0.030760731548070908,
+      "loss_ce": 0.0013112699380144477,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 973680328,
+      "step": 5674
+    },
+    {
+      "epoch": 64.42492917847025,
+      "grad_norm": 3.686446307197577,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 973850304,
+      "step": 5675
+    },
+    {
+      "epoch": 64.42492917847025,
+      "loss": 0.03783392906188965,
+      "loss_ce": 2.2651593098999e-05,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 973850304,
+      "step": 5675
+    },
+    {
+      "epoch": 64.43626062322946,
+      "grad_norm": 3.458810633124514,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 974022192,
+      "step": 5676
+    },
+    {
+      "epoch": 64.43626062322946,
+      "loss": 0.047905053943395615,
+      "loss_ce": 2.297363789693918e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 974022192,
+      "step": 5676
+    },
+    {
+      "epoch": 64.44759206798867,
+      "grad_norm": 3.559519439300565,
+      "learning_rate": 5e-06,
+      "loss": 0.089,
+      "num_input_tokens_seen": 974193944,
+      "step": 5677
+    },
+    {
+      "epoch": 64.44759206798867,
+      "loss": 0.1032581552863121,
+      "loss_ce": 1.71900937857572e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 974193944,
+      "step": 5677
+    },
+    {
+      "epoch": 64.45892351274787,
+      "grad_norm": 3.520029139844324,
+      "learning_rate": 5e-06,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 974365960,
+      "step": 5678
+    },
+    {
+      "epoch": 64.45892351274787,
+      "loss": 0.049592651426792145,
+      "loss_ce": 4.736337359645404e-05,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 974365960,
+      "step": 5678
+    },
+    {
+      "epoch": 64.47025495750708,
+      "grad_norm": 3.7315271964070154,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 974537520,
+      "step": 5679
+    },
+    {
+      "epoch": 64.47025495750708,
+      "loss": 0.03711051493883133,
+      "loss_ce": 3.1655843486078084e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 974537520,
+      "step": 5679
+    },
+    {
+      "epoch": 64.48158640226629,
+      "grad_norm": 4.648282599937621,
+      "learning_rate": 5e-06,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 974707524,
+      "step": 5680
+    },
+    {
+      "epoch": 64.48158640226629,
+      "loss": 0.03814130276441574,
+      "loss_ce": 2.4845245206961408e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 974707524,
+      "step": 5680
+    },
+    {
+      "epoch": 64.4929178470255,
+      "grad_norm": 3.2678062672198256,
+      "learning_rate": 5e-06,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 974879300,
+      "step": 5681
+    },
+    {
+      "epoch": 64.4929178470255,
+      "loss": 0.07701968401670456,
+      "loss_ce": 0.00013827778457198292,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 974879300,
+      "step": 5681
+    },
+    {
+      "epoch": 64.5042492917847,
+      "grad_norm": 3.2989228814855935,
+      "learning_rate": 5e-06,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 975051216,
+      "step": 5682
+    },
+    {
+      "epoch": 64.5042492917847,
+      "loss": 0.029823914170265198,
+      "loss_ce": 3.875717084156349e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 975051216,
+      "step": 5682
+    },
+    {
+      "epoch": 64.51558073654391,
+      "grad_norm": 3.2641555983620685,
+      "learning_rate": 5e-06,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 975221584,
+      "step": 5683
+    },
+    {
+      "epoch": 64.51558073654391,
+      "loss": 0.03420019894838333,
+      "loss_ce": 3.5772707633441314e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 975221584,
+      "step": 5683
+    },
+    {
+      "epoch": 64.52691218130312,
+      "grad_norm": 3.1877772493264906,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 975393476,
+      "step": 5684
+    },
+    {
+      "epoch": 64.52691218130312,
+      "loss": 0.058207154273986816,
+      "loss_ce": 7.117014320101589e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 975393476,
+      "step": 5684
+    },
+    {
+      "epoch": 64.53824362606233,
+      "grad_norm": 3.915987478606102,
+      "learning_rate": 5e-06,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 975565764,
+      "step": 5685
+    },
+    {
+      "epoch": 64.53824362606233,
+      "loss": 0.04126343876123428,
+      "loss_ce": 3.4191343729617074e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 975565764,
+      "step": 5685
+    },
+    {
+      "epoch": 64.54957507082153,
+      "grad_norm": 3.5377990606580147,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 975737520,
+      "step": 5686
+    },
+    {
+      "epoch": 64.54957507082153,
+      "loss": 0.035324983298778534,
+      "loss_ce": 7.718152482993901e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 975737520,
+      "step": 5686
+    },
+    {
+      "epoch": 64.56090651558074,
+      "grad_norm": 3.490277098444185,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 975908960,
+      "step": 5687
+    },
+    {
+      "epoch": 64.56090651558074,
+      "loss": 0.04415760934352875,
+      "loss_ce": 1.3927259715273976e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 975908960,
+      "step": 5687
+    },
+    {
+      "epoch": 64.57223796033995,
+      "grad_norm": 5.138649997461146,
+      "learning_rate": 5e-06,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 976080856,
+      "step": 5688
+    },
+    {
+      "epoch": 64.57223796033995,
+      "loss": 0.03360157459974289,
+      "loss_ce": 1.6977472114376724e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 976080856,
+      "step": 5688
+    },
+    {
+      "epoch": 64.58356940509915,
+      "grad_norm": 2.928120706613792,
+      "learning_rate": 5e-06,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 976251688,
+      "step": 5689
+    },
+    {
+      "epoch": 64.58356940509915,
+      "loss": 0.026877131313085556,
+      "loss_ce": 6.4029891291284e-06,
+      "loss_iou": 0.1015625,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 976251688,
+      "step": 5689
+    },
+    {
+      "epoch": 64.59490084985836,
+      "grad_norm": 2.6874185357359237,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 976422544,
+      "step": 5690
+    },
+    {
+      "epoch": 64.59490084985836,
+      "loss": 0.03900842368602753,
+      "loss_ce": 6.799708353355527e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 976422544,
+      "step": 5690
+    },
+    {
+      "epoch": 64.60623229461757,
+      "grad_norm": 3.1095773453465476,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 976594244,
+      "step": 5691
+    },
+    {
+      "epoch": 64.60623229461757,
+      "loss": 0.043078623712062836,
+      "loss_ce": 1.832183261285536e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 976594244,
+      "step": 5691
+    },
+    {
+      "epoch": 64.61756373937678,
+      "grad_norm": 2.9965862760586988,
+      "learning_rate": 5e-06,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 976766444,
+      "step": 5692
+    },
+    {
+      "epoch": 64.61756373937678,
+      "loss": 0.07904849946498871,
+      "loss_ce": 0.00011478344822535291,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 976766444,
+      "step": 5692
+    },
+    {
+      "epoch": 64.62889518413597,
+      "grad_norm": 6.873534064891267,
+      "learning_rate": 5e-06,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 976937280,
+      "step": 5693
+    },
+    {
+      "epoch": 64.62889518413597,
+      "loss": 0.05382588505744934,
+      "loss_ce": 5.391139711719006e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 976937280,
+      "step": 5693
+    },
+    {
+      "epoch": 64.64022662889518,
+      "grad_norm": 3.5559464002296726,
+      "learning_rate": 5e-06,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 977105888,
+      "step": 5694
+    },
+    {
+      "epoch": 64.64022662889518,
+      "loss": 0.04109952971339226,
+      "loss_ce": 2.2868898668093607e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 977105888,
+      "step": 5694
+    },
+    {
+      "epoch": 64.65155807365439,
+      "grad_norm": 4.000632100500444,
+      "learning_rate": 5e-06,
+      "loss": 0.0595,
+      "num_input_tokens_seen": 977277820,
+      "step": 5695
+    },
+    {
+      "epoch": 64.65155807365439,
+      "loss": 0.03450049087405205,
+      "loss_ce": 7.666256715310737e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 977277820,
+      "step": 5695
+    },
+    {
+      "epoch": 64.66288951841359,
+      "grad_norm": 4.133296833399541,
+      "learning_rate": 5e-06,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 977449952,
+      "step": 5696
+    },
+    {
+      "epoch": 64.66288951841359,
+      "loss": 0.08129125833511353,
+      "loss_ce": 3.820850542979315e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 977449952,
+      "step": 5696
+    },
+    {
+      "epoch": 64.6742209631728,
+      "grad_norm": 4.627647126351046,
+      "learning_rate": 5e-06,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 977621140,
+      "step": 5697
+    },
+    {
+      "epoch": 64.6742209631728,
+      "loss": 0.04074341058731079,
+      "loss_ce": 1.770470043993555e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 977621140,
+      "step": 5697
+    },
+    {
+      "epoch": 64.68555240793201,
+      "grad_norm": 3.50543315635961,
+      "learning_rate": 5e-06,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 977792744,
+      "step": 5698
+    },
+    {
+      "epoch": 64.68555240793201,
+      "loss": 0.0386078767478466,
+      "loss_ce": 6.417420809157193e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 977792744,
+      "step": 5698
+    },
+    {
+      "epoch": 64.69688385269122,
+      "grad_norm": 2.686637750687325,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 977963340,
+      "step": 5699
+    },
+    {
+      "epoch": 64.69688385269122,
+      "loss": 0.04813689738512039,
+      "loss_ce": 2.5937277314369567e-05,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 977963340,
+      "step": 5699
+    },
+    {
+      "epoch": 64.70821529745042,
+      "grad_norm": 2.720560968972478,
+      "learning_rate": 5e-06,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 978134064,
+      "step": 5700
+    },
+    {
+      "epoch": 64.70821529745042,
+      "loss": 0.044515762478113174,
+      "loss_ce": 3.6394052585819736e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 978134064,
+      "step": 5700
+    },
+    {
+      "epoch": 64.71954674220963,
+      "grad_norm": 51.96241600684382,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 978306100,
+      "step": 5701
+    },
+    {
+      "epoch": 64.71954674220963,
+      "loss": 0.028195591643452644,
+      "loss_ce": 2.786701952572912e-05,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 978306100,
+      "step": 5701
+    },
+    {
+      "epoch": 64.73087818696884,
+      "grad_norm": 3.0046149853037902,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 978477960,
+      "step": 5702
+    },
+    {
+      "epoch": 64.73087818696884,
+      "loss": 0.04333553463220596,
+      "loss_ce": 4.6350498450919986e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 978477960,
+      "step": 5702
+    },
+    {
+      "epoch": 64.74220963172804,
+      "grad_norm": 6.296504067866645,
+      "learning_rate": 5e-06,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 978649964,
+      "step": 5703
+    },
+    {
+      "epoch": 64.74220963172804,
+      "loss": 0.05528460443019867,
+      "loss_ce": 0.000993833295069635,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 978649964,
+      "step": 5703
+    },
+    {
+      "epoch": 64.75354107648725,
+      "grad_norm": 3.576862711492624,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 978821816,
+      "step": 5704
+    },
+    {
+      "epoch": 64.75354107648725,
+      "loss": 0.06209405139088631,
+      "loss_ce": 3.6554796679411083e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 978821816,
+      "step": 5704
+    },
+    {
+      "epoch": 64.76487252124646,
+      "grad_norm": 2.8915771054261503,
+      "learning_rate": 5e-06,
+      "loss": 0.043,
+      "num_input_tokens_seen": 978993128,
+      "step": 5705
+    },
+    {
+      "epoch": 64.76487252124646,
+      "loss": 0.05067801475524902,
+      "loss_ce": 3.409415512578562e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 978993128,
+      "step": 5705
+    },
+    {
+      "epoch": 64.77620396600567,
+      "grad_norm": 3.3214370040191206,
+      "learning_rate": 5e-06,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 979164708,
+      "step": 5706
+    },
+    {
+      "epoch": 64.77620396600567,
+      "loss": 0.05327049642801285,
+      "loss_ce": 3.257872594986111e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 979164708,
+      "step": 5706
+    },
+    {
+      "epoch": 64.78753541076487,
+      "grad_norm": 3.8183336050826213,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 979336676,
+      "step": 5707
+    },
+    {
+      "epoch": 64.78753541076487,
+      "loss": 0.03260376676917076,
+      "loss_ce": 2.6249990696669556e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 979336676,
+      "step": 5707
+    },
+    {
+      "epoch": 64.79886685552408,
+      "grad_norm": 3.939627741630551,
+      "learning_rate": 5e-06,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 979506688,
+      "step": 5708
+    },
+    {
+      "epoch": 64.79886685552408,
+      "loss": 0.040414802730083466,
+      "loss_ce": 2.4786560970824212e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 979506688,
+      "step": 5708
+    },
+    {
+      "epoch": 64.81019830028329,
+      "grad_norm": 5.572343923519142,
+      "learning_rate": 5e-06,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 979678420,
+      "step": 5709
+    },
+    {
+      "epoch": 64.81019830028329,
+      "loss": 0.08625856041908264,
+      "loss_ce": 4.6400498831644654e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 979678420,
+      "step": 5709
+    },
+    {
+      "epoch": 64.8215297450425,
+      "grad_norm": 2.9141808544941203,
+      "learning_rate": 5e-06,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 979849404,
+      "step": 5710
+    },
+    {
+      "epoch": 64.8215297450425,
+      "loss": 0.05351022630929947,
+      "loss_ce": 4.3428146454971284e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 979849404,
+      "step": 5710
+    },
+    {
+      "epoch": 64.8328611898017,
+      "grad_norm": 3.3921465956807886,
+      "learning_rate": 5e-06,
+      "loss": 0.034,
+      "num_input_tokens_seen": 980021316,
+      "step": 5711
+    },
+    {
+      "epoch": 64.8328611898017,
+      "loss": 0.04531057924032211,
+      "loss_ce": 5.300892371451482e-05,
+      "loss_iou": 0.130859375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 980021316,
+      "step": 5711
+    },
+    {
+      "epoch": 64.84419263456091,
+      "grad_norm": 3.6504093743421926,
+      "learning_rate": 5e-06,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 980193032,
+      "step": 5712
+    },
+    {
+      "epoch": 64.84419263456091,
+      "loss": 0.13299715518951416,
+      "loss_ce": 4.7332356189144775e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 980193032,
+      "step": 5712
+    },
+    {
+      "epoch": 64.85552407932012,
+      "grad_norm": 3.666979601813156,
+      "learning_rate": 5e-06,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 980363416,
+      "step": 5713
+    },
+    {
+      "epoch": 64.85552407932012,
+      "loss": 0.03740091994404793,
+      "loss_ce": 1.6887715901248157e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 980363416,
+      "step": 5713
+    },
+    {
+      "epoch": 64.86685552407933,
+      "grad_norm": 3.5495682185182016,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 980534876,
+      "step": 5714
+    },
+    {
+      "epoch": 64.86685552407933,
+      "loss": 0.037795357406139374,
+      "loss_ce": 2.9854807507945225e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 980534876,
+      "step": 5714
+    },
+    {
+      "epoch": 64.87818696883852,
+      "grad_norm": 2.8497783315438867,
+      "learning_rate": 5e-06,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 980706424,
+      "step": 5715
+    },
+    {
+      "epoch": 64.87818696883852,
+      "loss": 0.031192386522889137,
+      "loss_ce": 4.9198362830793485e-05,
+      "loss_iou": 0.12353515625,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 980706424,
+      "step": 5715
+    },
+    {
+      "epoch": 64.88951841359773,
+      "grad_norm": 2.287255791980991,
+      "learning_rate": 5e-06,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 980878004,
+      "step": 5716
+    },
+    {
+      "epoch": 64.88951841359773,
+      "loss": 0.04274478554725647,
+      "loss_ce": 2.0179269995423965e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 980878004,
+      "step": 5716
+    },
+    {
+      "epoch": 64.90084985835693,
+      "grad_norm": 4.263638228683565,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 981049860,
+      "step": 5717
+    },
+    {
+      "epoch": 64.90084985835693,
+      "loss": 0.1389244645833969,
+      "loss_ce": 5.422658796305768e-05,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 981049860,
+      "step": 5717
+    },
+    {
+      "epoch": 64.91218130311614,
+      "grad_norm": 3.803421017811128,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 981221956,
+      "step": 5718
+    },
+    {
+      "epoch": 64.91218130311614,
+      "loss": 0.04325833544135094,
+      "loss_ce": 4.544557668850757e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 981221956,
+      "step": 5718
+    },
+    {
+      "epoch": 64.92351274787535,
+      "grad_norm": 4.0302904918822575,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 981388288,
+      "step": 5719
+    },
+    {
+      "epoch": 64.92351274787535,
+      "loss": 0.0562330037355423,
+      "loss_ce": 9.591891284799203e-05,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 981388288,
+      "step": 5719
+    },
+    {
+      "epoch": 64.93484419263456,
+      "grad_norm": 3.566501978859216,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 981560188,
+      "step": 5720
+    },
+    {
+      "epoch": 64.93484419263456,
+      "loss": 0.08308477699756622,
+      "loss_ce": 3.118738459306769e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 981560188,
+      "step": 5720
+    },
+    {
+      "epoch": 64.94617563739376,
+      "grad_norm": 3.584624047164368,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 981731952,
+      "step": 5721
+    },
+    {
+      "epoch": 64.94617563739376,
+      "loss": 0.06078848987817764,
+      "loss_ce": 2.7986985514871776e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 981731952,
+      "step": 5721
+    },
+    {
+      "epoch": 64.95750708215297,
+      "grad_norm": 3.354499966578575,
+      "learning_rate": 5e-06,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 981904072,
+      "step": 5722
+    },
+    {
+      "epoch": 64.95750708215297,
+      "loss": 0.07813683152198792,
+      "loss_ce": 4.2348918213974684e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 981904072,
+      "step": 5722
+    },
+    {
+      "epoch": 64.96883852691218,
+      "grad_norm": 3.627788052273909,
+      "learning_rate": 5e-06,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 982074112,
+      "step": 5723
+    },
+    {
+      "epoch": 64.96883852691218,
+      "loss": 0.04120307043194771,
+      "loss_ce": 4.24871759605594e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 982074112,
+      "step": 5723
+    },
+    {
+      "epoch": 64.98016997167139,
+      "grad_norm": 3.8320040003143294,
+      "learning_rate": 5e-06,
+      "loss": 0.038,
+      "num_input_tokens_seen": 982242872,
+      "step": 5724
+    },
+    {
+      "epoch": 64.98016997167139,
+      "loss": 0.04043585807085037,
+      "loss_ce": 3.058449146919884e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 982242872,
+      "step": 5724
+    },
+    {
+      "epoch": 64.9915014164306,
+      "grad_norm": 3.908828546511955,
+      "learning_rate": 5e-06,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 982414804,
+      "step": 5725
+    },
+    {
+      "epoch": 64.9915014164306,
+      "loss": 0.09888176620006561,
+      "loss_ce": 2.0067374862264842e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 982414804,
+      "step": 5725
+    },
+    {
+      "epoch": 64.9915014164306,
+      "loss": 0.03593401610851288,
+      "loss_ce": 3.008566091011744e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 982543972,
+      "step": 5725
+    },
+    {
+      "epoch": 65.0028328611898,
+      "grad_norm": 4.278739973485845,
+      "learning_rate": 5e-06,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 982587144,
+      "step": 5726
+    },
+    {
+      "epoch": 65.0028328611898,
+      "loss": 0.04389403015375137,
+      "loss_ce": 4.027242539450526e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 982587144,
+      "step": 5726
+    },
+    {
+      "epoch": 65.01416430594901,
+      "grad_norm": 3.344390384944367,
+      "learning_rate": 5e-06,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 982758360,
+      "step": 5727
+    },
+    {
+      "epoch": 65.01416430594901,
+      "loss": 0.03912629932165146,
+      "loss_ce": 1.80219085450517e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 982758360,
+      "step": 5727
+    },
+    {
+      "epoch": 65.02549575070822,
+      "grad_norm": 5.0895530611280995,
+      "learning_rate": 5e-06,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 982930100,
+      "step": 5728
+    },
+    {
+      "epoch": 65.02549575070822,
+      "loss": 0.02324196696281433,
+      "loss_ce": 7.149645534809679e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 982930100,
+      "step": 5728
+    },
+    {
+      "epoch": 65.03682719546742,
+      "grad_norm": 3.244496383794755,
+      "learning_rate": 5e-06,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 983101820,
+      "step": 5729
+    },
+    {
+      "epoch": 65.03682719546742,
+      "loss": 0.043377332389354706,
+      "loss_ce": 4.237061148160137e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 983101820,
+      "step": 5729
+    },
+    {
+      "epoch": 65.04815864022663,
+      "grad_norm": 3.1005707560814937,
+      "learning_rate": 5e-06,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 983273340,
+      "step": 5730
+    },
+    {
+      "epoch": 65.04815864022663,
+      "loss": 0.0363239049911499,
+      "loss_ce": 3.0871364288032055e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 983273340,
+      "step": 5730
+    },
+    {
+      "epoch": 65.05949008498584,
+      "grad_norm": 3.1630157186545462,
+      "learning_rate": 5e-06,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 983444220,
+      "step": 5731
+    },
+    {
+      "epoch": 65.05949008498584,
+      "loss": 0.05001368746161461,
+      "loss_ce": 4.115112096769735e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 983444220,
+      "step": 5731
+    },
+    {
+      "epoch": 65.07082152974505,
+      "grad_norm": 2.601181032677839,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 983616740,
+      "step": 5732
+    },
+    {
+      "epoch": 65.07082152974505,
+      "loss": 0.029378995299339294,
+      "loss_ce": 3.634267341112718e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 983616740,
+      "step": 5732
+    },
+    {
+      "epoch": 65.08215297450425,
+      "grad_norm": 3.265062334190223,
+      "learning_rate": 5e-06,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 983788724,
+      "step": 5733
+    },
+    {
+      "epoch": 65.08215297450425,
+      "loss": 0.03413556516170502,
+      "loss_ce": 3.217053017579019e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 983788724,
+      "step": 5733
+    },
+    {
+      "epoch": 65.09348441926346,
+      "grad_norm": 3.9142167297086194,
+      "learning_rate": 5e-06,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 983959888,
+      "step": 5734
+    },
+    {
+      "epoch": 65.09348441926346,
+      "loss": 0.0876886248588562,
+      "loss_ce": 4.2142957681789994e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 983959888,
+      "step": 5734
+    },
+    {
+      "epoch": 65.10481586402267,
+      "grad_norm": 3.731325356539566,
+      "learning_rate": 5e-06,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 984131464,
+      "step": 5735
+    },
+    {
+      "epoch": 65.10481586402267,
+      "loss": 0.13541018962860107,
+      "loss_ce": 6.472597306128591e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 984131464,
+      "step": 5735
+    },
+    {
+      "epoch": 65.11614730878188,
+      "grad_norm": 2.9642035457082985,
+      "learning_rate": 5e-06,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 984303212,
+      "step": 5736
+    },
+    {
+      "epoch": 65.11614730878188,
+      "loss": 0.058792926371097565,
+      "loss_ce": 0.00016866040823515505,
+      "loss_iou": 0.078125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 984303212,
+      "step": 5736
+    },
+    {
+      "epoch": 65.12747875354107,
+      "grad_norm": 3.14199247012568,
+      "learning_rate": 5e-06,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 984473684,
+      "step": 5737
+    },
+    {
+      "epoch": 65.12747875354107,
+      "loss": 0.12547801434993744,
+      "loss_ce": 2.0248595319571905e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 984473684,
+      "step": 5737
+    },
+    {
+      "epoch": 65.13881019830028,
+      "grad_norm": 3.810742643843869,
+      "learning_rate": 5e-06,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 984645292,
+      "step": 5738
+    },
+    {
+      "epoch": 65.13881019830028,
+      "loss": 0.03057882748544216,
+      "loss_ce": 4.599188105203211e-05,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 984645292,
+      "step": 5738
+    },
+    {
+      "epoch": 65.15014164305948,
+      "grad_norm": 3.6153608489384146,
+      "learning_rate": 5e-06,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 984817140,
+      "step": 5739
+    },
+    {
+      "epoch": 65.15014164305948,
+      "loss": 0.05022459849715233,
+      "loss_ce": 2.3182514269137755e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 984817140,
+      "step": 5739
+    },
+    {
+      "epoch": 65.16147308781869,
+      "grad_norm": 3.2427250268432157,
+      "learning_rate": 5e-06,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 984989060,
+      "step": 5740
+    },
+    {
+      "epoch": 65.16147308781869,
+      "loss": 0.04532986879348755,
+      "loss_ce": 2.652549846970942e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 984989060,
+      "step": 5740
+    },
+    {
+      "epoch": 65.1728045325779,
+      "grad_norm": 2.8996045963031767,
+      "learning_rate": 5e-06,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 985160680,
+      "step": 5741
+    },
+    {
+      "epoch": 65.1728045325779,
+      "loss": 0.04038078337907791,
+      "loss_ce": 3.654730608104728e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 985160680,
+      "step": 5741
+    },
+    {
+      "epoch": 65.1841359773371,
+      "grad_norm": 2.7577053711303443,
+      "learning_rate": 5e-06,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 985332472,
+      "step": 5742
+    },
+    {
+      "epoch": 65.1841359773371,
+      "loss": 0.03965802863240242,
+      "loss_ce": 6.146983650978655e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 985332472,
+      "step": 5742
+    },
+    {
+      "epoch": 65.19546742209631,
+      "grad_norm": 3.132942549717698,
+      "learning_rate": 5e-06,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 985502956,
+      "step": 5743
+    },
+    {
+      "epoch": 65.19546742209631,
+      "loss": 0.0280813779681921,
+      "loss_ce": 1.283594156120671e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 985502956,
+      "step": 5743
+    },
+    {
+      "epoch": 65.20679886685552,
+      "grad_norm": 2.745008962198351,
+      "learning_rate": 5e-06,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 985674840,
+      "step": 5744
+    },
+    {
+      "epoch": 65.20679886685552,
+      "loss": 0.025214754045009613,
+      "loss_ce": 2.2494146833196282e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 985674840,
+      "step": 5744
+    },
+    {
+      "epoch": 65.21813031161473,
+      "grad_norm": 2.9213145978826693,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 985846348,
+      "step": 5745
+    },
+    {
+      "epoch": 65.21813031161473,
+      "loss": 0.03277846798300743,
+      "loss_ce": 4.8366287956014276e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 985846348,
+      "step": 5745
+    },
+    {
+      "epoch": 65.22946175637394,
+      "grad_norm": 3.0008774580000157,
+      "learning_rate": 5e-06,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 986016048,
+      "step": 5746
+    },
+    {
+      "epoch": 65.22946175637394,
+      "loss": 0.030263621360063553,
+      "loss_ce": 2.0701359972008504e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 986016048,
+      "step": 5746
+    },
+    {
+      "epoch": 65.24079320113314,
+      "grad_norm": 3.593550432236033,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 986187112,
+      "step": 5747
+    },
+    {
+      "epoch": 65.24079320113314,
+      "loss": 0.03626236319541931,
+      "loss_ce": 3.799736441578716e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 986187112,
+      "step": 5747
+    },
+    {
+      "epoch": 65.25212464589235,
+      "grad_norm": 4.1807311587745595,
+      "learning_rate": 5e-06,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 986359172,
+      "step": 5748
+    },
+    {
+      "epoch": 65.25212464589235,
+      "loss": 0.03177310526371002,
+      "loss_ce": 3.482650572550483e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 986359172,
+      "step": 5748
+    },
+    {
+      "epoch": 65.26345609065156,
+      "grad_norm": 3.1281716946654208,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 986531024,
+      "step": 5749
+    },
+    {
+      "epoch": 65.26345609065156,
+      "loss": 0.035262130200862885,
+      "loss_ce": 3.7216508644632995e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 986531024,
+      "step": 5749
+    },
+    {
+      "epoch": 65.27478753541077,
+      "grad_norm": 3.853304110181165,
+      "learning_rate": 5e-06,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 986702852,
+      "step": 5750
+    },
+    {
+      "epoch": 65.27478753541077,
+      "eval_seeclick_CIoU": 0.5279392600059509,
+      "eval_seeclick_GIoU": 0.5279070436954498,
+      "eval_seeclick_IoU": 0.5653687417507172,
+      "eval_seeclick_MAE_all": 0.0654074177145958,
+      "eval_seeclick_MAE_h": 0.032020159997045994,
+      "eval_seeclick_MAE_w": 0.09991196915507317,
+      "eval_seeclick_MAE_x": 0.1016213670372963,
+      "eval_seeclick_MAE_y": 0.028076169081032276,
+      "eval_seeclick_NUM_probability": 0.9999834299087524,
+      "eval_seeclick_inside_bbox": 0.9076704680919647,
+      "eval_seeclick_loss": 0.9440920352935791,
+      "eval_seeclick_loss_ce": 0.6966625452041626,
+      "eval_seeclick_loss_iou": 0.5435791015625,
+      "eval_seeclick_loss_num": 0.049468994140625,
+      "eval_seeclick_loss_xval": 0.247283935546875,
+      "eval_seeclick_runtime": 69.1799,
+      "eval_seeclick_samples_per_second": 0.622,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 986702852,
+      "step": 5750
+    },
+    {
+      "epoch": 65.27478753541077,
+      "eval_icons_CIoU": 0.7469396889209747,
+      "eval_icons_GIoU": 0.7457391917705536,
+      "eval_icons_IoU": 0.7601137757301331,
+      "eval_icons_MAE_all": 0.03410924784839153,
+      "eval_icons_MAE_h": 0.031143520027399063,
+      "eval_icons_MAE_w": 0.038319991901516914,
+      "eval_icons_MAE_x": 0.033546026796102524,
+      "eval_icons_MAE_y": 0.033427443355321884,
+      "eval_icons_NUM_probability": 0.9996732473373413,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.11560887098312378,
+      "eval_icons_loss_ce": 0.003657496767118573,
+      "eval_icons_loss_iou": 0.54168701171875,
+      "eval_icons_loss_num": 0.020341873168945312,
+      "eval_icons_loss_xval": 0.101776123046875,
+      "eval_icons_runtime": 79.6161,
+      "eval_icons_samples_per_second": 0.628,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 986702852,
+      "step": 5750
+    },
+    {
+      "epoch": 65.27478753541077,
+      "eval_screenspot_CIoU": 0.6210390130678812,
+      "eval_screenspot_GIoU": 0.617551843325297,
+      "eval_screenspot_IoU": 0.6457635164260864,
+      "eval_screenspot_MAE_all": 0.0647940809528033,
+      "eval_screenspot_MAE_h": 0.04260993810991446,
+      "eval_screenspot_MAE_w": 0.11057011783123016,
+      "eval_screenspot_MAE_x": 0.07184754063685735,
+      "eval_screenspot_MAE_y": 0.0341487160573403,
+      "eval_screenspot_NUM_probability": 0.9999564091364542,
+      "eval_screenspot_inside_bbox": 0.9183333317438761,
+      "eval_screenspot_loss": 0.29010704159736633,
+      "eval_screenspot_loss_ce": 0.0171851241029799,
+      "eval_screenspot_loss_iou": 0.5324300130208334,
+      "eval_screenspot_loss_num": 0.054255167643229164,
+      "eval_screenspot_loss_xval": 0.2713623046875,
+      "eval_screenspot_runtime": 136.7024,
+      "eval_screenspot_samples_per_second": 0.651,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 986702852,
+      "step": 5750
+    },
+    {
+      "epoch": 65.27478753541077,
+      "eval_compot_CIoU": 0.8379127681255341,
+      "eval_compot_GIoU": 0.8338889479637146,
+      "eval_compot_IoU": 0.8521289229393005,
+      "eval_compot_MAE_all": 0.026224092580378056,
+      "eval_compot_MAE_h": 0.020950539968907833,
+      "eval_compot_MAE_w": 0.032208927907049656,
+      "eval_compot_MAE_x": 0.030544021166861057,
+      "eval_compot_MAE_y": 0.02119287569075823,
+      "eval_compot_NUM_probability": 0.9999770820140839,
+      "eval_compot_inside_bbox": 0.9409722089767456,
+      "eval_compot_loss": 0.08464618772268295,
+      "eval_compot_loss_ce": 2.4417404347332194e-05,
+      "eval_compot_loss_iou": 0.5771484375,
+      "eval_compot_loss_num": 0.014888763427734375,
+      "eval_compot_loss_xval": 0.0743865966796875,
+      "eval_compot_runtime": 87.2073,
+      "eval_compot_samples_per_second": 0.573,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 986702852,
+      "step": 5750
+    },
+    {
+      "epoch": 65.27478753541077,
+      "eval_custom_ui_MAE_all": 0.0198229500092566,
+      "eval_custom_ui_MAE_x": 0.03207499347627163,
+      "eval_custom_ui_MAE_y": 0.007570905378088355,
+      "eval_custom_ui_NUM_probability": 0.9998688995838165,
+      "eval_custom_ui_loss": 0.22339992225170135,
+      "eval_custom_ui_loss_ce": 0.12163322418928146,
+      "eval_custom_ui_loss_num": 0.019514083862304688,
+      "eval_custom_ui_loss_xval": 0.09759521484375,
+      "eval_custom_ui_runtime": 65.9456,
+      "eval_custom_ui_samples_per_second": 0.758,
+      "eval_custom_ui_steps_per_second": 0.03,
+      "num_input_tokens_seen": 986702852,
+      "step": 5750
+    },
+    {
+      "epoch": 65.27478753541077,
+      "loss": 0.2739031910896301,
+      "loss_ce": 0.15262635052204132,
+      "loss_iou": 0.0,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 986702852,
+      "step": 5750
+    },
+    {
+      "epoch": 65.28611898016997,
+      "grad_norm": 3.5189714360163653,
+      "learning_rate": 5e-06,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 986875044,
+      "step": 5751
+    },
+    {
+      "epoch": 65.28611898016997,
+      "loss": 0.040284451097249985,
+      "loss_ce": 3.176531754434109e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 986875044,
+      "step": 5751
+    },
+    {
+      "epoch": 65.29745042492918,
+      "grad_norm": 2.918074328603341,
+      "learning_rate": 5e-06,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 987045164,
+      "step": 5752
+    },
+    {
+      "epoch": 65.29745042492918,
+      "loss": 0.03508681431412697,
+      "loss_ce": 2.2118176275398582e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 987045164,
+      "step": 5752
+    },
+    {
+      "epoch": 65.30878186968839,
+      "grad_norm": 2.7106561879634805,
+      "learning_rate": 5e-06,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 987216812,
+      "step": 5753
+    },
+    {
+      "epoch": 65.30878186968839,
+      "loss": 0.05032968521118164,
+      "loss_ce": 2.1456542526721023e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 987216812,
+      "step": 5753
+    },
+    {
+      "epoch": 65.3201133144476,
+      "grad_norm": 2.8396150977942467,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 987388264,
+      "step": 5754
+    },
+    {
+      "epoch": 65.3201133144476,
+      "loss": 0.029339594766497612,
+      "loss_ce": 4.2719671910163015e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 987388264,
+      "step": 5754
+    },
+    {
+      "epoch": 65.3314447592068,
+      "grad_norm": 3.563886772336644,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 987560316,
+      "step": 5755
+    },
+    {
+      "epoch": 65.3314447592068,
+      "loss": 0.03773288428783417,
+      "loss_ce": 1.3158956789993681e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 987560316,
+      "step": 5755
+    },
+    {
+      "epoch": 65.34277620396601,
+      "grad_norm": 3.188420084164555,
+      "learning_rate": 5e-06,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 987732324,
+      "step": 5756
+    },
+    {
+      "epoch": 65.34277620396601,
+      "loss": 0.07110798358917236,
+      "loss_ce": 1.7283775378018618e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 987732324,
+      "step": 5756
+    },
+    {
+      "epoch": 65.35410764872522,
+      "grad_norm": 3.3629373273694183,
+      "learning_rate": 5e-06,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 987902472,
+      "step": 5757
+    },
+    {
+      "epoch": 65.35410764872522,
+      "loss": 0.03837970271706581,
+      "loss_ce": 1.9107495973003097e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 987902472,
+      "step": 5757
+    },
+    {
+      "epoch": 65.36543909348443,
+      "grad_norm": 3.7747000969198683,
+      "learning_rate": 5e-06,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 988073228,
+      "step": 5758
+    },
+    {
+      "epoch": 65.36543909348443,
+      "loss": 0.033994633704423904,
+      "loss_ce": 1.3310223948792554e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 988073228,
+      "step": 5758
+    },
+    {
+      "epoch": 65.37677053824362,
+      "grad_norm": 2.8496583040792607,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 988241896,
+      "step": 5759
+    },
+    {
+      "epoch": 65.37677053824362,
+      "loss": 0.026833347976207733,
+      "loss_ce": 3.8915659388294443e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 988241896,
+      "step": 5759
+    },
+    {
+      "epoch": 65.38810198300283,
+      "grad_norm": 2.848140053213339,
+      "learning_rate": 5e-06,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 988413896,
+      "step": 5760
+    },
+    {
+      "epoch": 65.38810198300283,
+      "loss": 0.040435243397951126,
+      "loss_ce": 1.470960251026554e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 988413896,
+      "step": 5760
+    },
+    {
+      "epoch": 65.39943342776203,
+      "grad_norm": 3.2017273348398727,
+      "learning_rate": 5e-06,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 988585760,
+      "step": 5761
+    },
+    {
+      "epoch": 65.39943342776203,
+      "loss": 0.08214667439460754,
+      "loss_ce": 2.3867934942245483e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 988585760,
+      "step": 5761
+    },
+    {
+      "epoch": 65.41076487252124,
+      "grad_norm": 2.515146429526058,
+      "learning_rate": 5e-06,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 988757560,
+      "step": 5762
+    },
+    {
+      "epoch": 65.41076487252124,
+      "loss": 0.09435395151376724,
+      "loss_ce": 2.4117889552144334e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 988757560,
+      "step": 5762
+    },
+    {
+      "epoch": 65.42209631728045,
+      "grad_norm": 3.169178426645544,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 988929536,
+      "step": 5763
+    },
+    {
+      "epoch": 65.42209631728045,
+      "loss": 0.04145709052681923,
+      "loss_ce": 1.4219082004274242e-05,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 988929536,
+      "step": 5763
+    },
+    {
+      "epoch": 65.43342776203966,
+      "grad_norm": 3.4570929116556885,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 989100740,
+      "step": 5764
+    },
+    {
+      "epoch": 65.43342776203966,
+      "loss": 0.06846779584884644,
+      "loss_ce": 3.212683805031702e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 989100740,
+      "step": 5764
+    },
+    {
+      "epoch": 65.44475920679886,
+      "grad_norm": 3.65635693285648,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 989272332,
+      "step": 5765
+    },
+    {
+      "epoch": 65.44475920679886,
+      "loss": 0.03286335617303848,
+      "loss_ce": 1.1184663890162483e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 989272332,
+      "step": 5765
+    },
+    {
+      "epoch": 65.45609065155807,
+      "grad_norm": 14.478673610780442,
+      "learning_rate": 5e-06,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 989441612,
+      "step": 5766
+    },
+    {
+      "epoch": 65.45609065155807,
+      "loss": 0.02933388203382492,
+      "loss_ce": 3.70066991308704e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 989441612,
+      "step": 5766
+    },
+    {
+      "epoch": 65.46742209631728,
+      "grad_norm": 3.3031396609222394,
+      "learning_rate": 5e-06,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 989613176,
+      "step": 5767
+    },
+    {
+      "epoch": 65.46742209631728,
+      "loss": 0.02758553810417652,
+      "loss_ce": 2.8164911782369018e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 989613176,
+      "step": 5767
+    },
+    {
+      "epoch": 65.47875354107649,
+      "grad_norm": 3.2221096835054857,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 989785412,
+      "step": 5768
+    },
+    {
+      "epoch": 65.47875354107649,
+      "loss": 0.08748716115951538,
+      "loss_ce": 2.377748205617536e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 989785412,
+      "step": 5768
+    },
+    {
+      "epoch": 65.4900849858357,
+      "grad_norm": 3.1833275123196416,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 989957212,
+      "step": 5769
+    },
+    {
+      "epoch": 65.4900849858357,
+      "loss": 0.030664196237921715,
+      "loss_ce": 3.980598921771161e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 989957212,
+      "step": 5769
+    },
+    {
+      "epoch": 65.5014164305949,
+      "grad_norm": 2.8106543301552502,
+      "learning_rate": 5e-06,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 990127676,
+      "step": 5770
+    },
+    {
+      "epoch": 65.5014164305949,
+      "loss": 0.07636883854866028,
+      "loss_ce": 2.9118898964952677e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 990127676,
+      "step": 5770
+    },
+    {
+      "epoch": 65.51274787535411,
+      "grad_norm": 3.0117490827801823,
+      "learning_rate": 5e-06,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 990299612,
+      "step": 5771
+    },
+    {
+      "epoch": 65.51274787535411,
+      "loss": 0.03499307483434677,
+      "loss_ce": 0.00016488679102621973,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 990299612,
+      "step": 5771
+    },
+    {
+      "epoch": 65.52407932011332,
+      "grad_norm": 3.481769144821567,
+      "learning_rate": 5e-06,
+      "loss": 0.0321,
+      "num_input_tokens_seen": 990471752,
+      "step": 5772
+    },
+    {
+      "epoch": 65.52407932011332,
+      "loss": 0.0201481394469738,
+      "loss_ce": 3.705561539391056e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0040283203125,
+      "loss_xval": 0.0201416015625,
+      "num_input_tokens_seen": 990471752,
+      "step": 5772
+    },
+    {
+      "epoch": 65.53541076487252,
+      "grad_norm": 3.808287545096577,
+      "learning_rate": 5e-06,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 990643820,
+      "step": 5773
+    },
+    {
+      "epoch": 65.53541076487252,
+      "loss": 0.036998577415943146,
+      "loss_ce": 1.127464565797709e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 990643820,
+      "step": 5773
+    },
+    {
+      "epoch": 65.54674220963173,
+      "grad_norm": 3.557916191116322,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 990814096,
+      "step": 5774
+    },
+    {
+      "epoch": 65.54674220963173,
+      "loss": 0.04732809215784073,
+      "loss_ce": 2.5847934011835605e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 990814096,
+      "step": 5774
+    },
+    {
+      "epoch": 65.55807365439094,
+      "grad_norm": 3.0852354078520934,
+      "learning_rate": 5e-06,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 990986084,
+      "step": 5775
+    },
+    {
+      "epoch": 65.55807365439094,
+      "loss": 0.05569403991103172,
+      "loss_ce": 4.5235283323563635e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 990986084,
+      "step": 5775
+    },
+    {
+      "epoch": 65.56940509915015,
+      "grad_norm": 4.03868127531456,
+      "learning_rate": 5e-06,
+      "loss": 0.062,
+      "num_input_tokens_seen": 991158164,
+      "step": 5776
+    },
+    {
+      "epoch": 65.56940509915015,
+      "loss": 0.042034778743982315,
+      "loss_ce": 4.259090928826481e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 991158164,
+      "step": 5776
+    },
+    {
+      "epoch": 65.58073654390935,
+      "grad_norm": 4.921513598889633,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 991329256,
+      "step": 5777
+    },
+    {
+      "epoch": 65.58073654390935,
+      "loss": 0.04119446873664856,
+      "loss_ce": 2.6259327569277957e-05,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 991329256,
+      "step": 5777
+    },
+    {
+      "epoch": 65.59206798866856,
+      "grad_norm": 3.858369560899777,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 991499740,
+      "step": 5778
+    },
+    {
+      "epoch": 65.59206798866856,
+      "loss": 0.0308595709502697,
+      "loss_ce": 3.681503949337639e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 991499740,
+      "step": 5778
+    },
+    {
+      "epoch": 65.60339943342777,
+      "grad_norm": 3.219289885126241,
+      "learning_rate": 5e-06,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 991671760,
+      "step": 5779
+    },
+    {
+      "epoch": 65.60339943342777,
+      "loss": 0.04247557371854782,
+      "loss_ce": 2.5621367967687547e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 991671760,
+      "step": 5779
+    },
+    {
+      "epoch": 65.61473087818698,
+      "grad_norm": 3.3675850115156463,
+      "learning_rate": 5e-06,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 991843572,
+      "step": 5780
+    },
+    {
+      "epoch": 65.61473087818698,
+      "loss": 0.04054023325443268,
+      "loss_ce": 0.00010444331564940512,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 991843572,
+      "step": 5780
+    },
+    {
+      "epoch": 65.62606232294617,
+      "grad_norm": 3.2036837375601754,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 992014720,
+      "step": 5781
+    },
+    {
+      "epoch": 65.62606232294617,
+      "loss": 0.037506699562072754,
+      "loss_ce": 0.00030577252618968487,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 992014720,
+      "step": 5781
+    },
+    {
+      "epoch": 65.63739376770538,
+      "grad_norm": 5.414410987094799,
+      "learning_rate": 5e-06,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 992185968,
+      "step": 5782
+    },
+    {
+      "epoch": 65.63739376770538,
+      "loss": 0.032044366002082825,
+      "loss_ce": 1.616914596525021e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 992185968,
+      "step": 5782
+    },
+    {
+      "epoch": 65.64872521246458,
+      "grad_norm": 4.543922422501434,
+      "learning_rate": 5e-06,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 992355448,
+      "step": 5783
+    },
+    {
+      "epoch": 65.64872521246458,
+      "loss": 0.08454081416130066,
+      "loss_ce": 2.238055458292365e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 992355448,
+      "step": 5783
+    },
+    {
+      "epoch": 65.66005665722379,
+      "grad_norm": 2.7286435562118303,
+      "learning_rate": 5e-06,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 992527064,
+      "step": 5784
+    },
+    {
+      "epoch": 65.66005665722379,
+      "loss": 0.028632130473852158,
+      "loss_ce": 2.1901407308178023e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 992527064,
+      "step": 5784
+    },
+    {
+      "epoch": 65.671388101983,
+      "grad_norm": 3.118595815003266,
+      "learning_rate": 5e-06,
+      "loss": 0.061,
+      "num_input_tokens_seen": 992698188,
+      "step": 5785
+    },
+    {
+      "epoch": 65.671388101983,
+      "loss": 0.030240904539823532,
+      "loss_ce": 0.00018871987413149327,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 992698188,
+      "step": 5785
+    },
+    {
+      "epoch": 65.6827195467422,
+      "grad_norm": 3.233464806094293,
+      "learning_rate": 5e-06,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 992869924,
+      "step": 5786
+    },
+    {
+      "epoch": 65.6827195467422,
+      "loss": 0.03187212720513344,
+      "loss_ce": 4.229322075843811e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 992869924,
+      "step": 5786
+    },
+    {
+      "epoch": 65.69405099150141,
+      "grad_norm": 3.5123250941426436,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 993040912,
+      "step": 5787
+    },
+    {
+      "epoch": 65.69405099150141,
+      "loss": 0.03946268558502197,
+      "loss_ce": 3.397458203835413e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 993040912,
+      "step": 5787
+    },
+    {
+      "epoch": 65.70538243626062,
+      "grad_norm": 3.3410576426367093,
+      "learning_rate": 5e-06,
+      "loss": 0.039,
+      "num_input_tokens_seen": 993212884,
+      "step": 5788
+    },
+    {
+      "epoch": 65.70538243626062,
+      "loss": 0.026257412508130074,
+      "loss_ce": 5.807226261822507e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 993212884,
+      "step": 5788
+    },
+    {
+      "epoch": 65.71671388101983,
+      "grad_norm": 4.329024998225538,
+      "learning_rate": 5e-06,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 993383984,
+      "step": 5789
+    },
+    {
+      "epoch": 65.71671388101983,
+      "loss": 0.10554572194814682,
+      "loss_ce": 1.59339870151598e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 993383984,
+      "step": 5789
+    },
+    {
+      "epoch": 65.72804532577904,
+      "grad_norm": 3.3322995804244333,
+      "learning_rate": 5e-06,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 993555656,
+      "step": 5790
+    },
+    {
+      "epoch": 65.72804532577904,
+      "loss": 0.041612058877944946,
+      "loss_ce": 1.659831832512282e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 993555656,
+      "step": 5790
+    },
+    {
+      "epoch": 65.73937677053824,
+      "grad_norm": 3.1685012912060966,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 993727760,
+      "step": 5791
+    },
+    {
+      "epoch": 65.73937677053824,
+      "loss": 0.04165208339691162,
+      "loss_ce": 2.610619048937224e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 993727760,
+      "step": 5791
+    },
+    {
+      "epoch": 65.75070821529745,
+      "grad_norm": 3.3555801751964274,
+      "learning_rate": 5e-06,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 993898768,
+      "step": 5792
+    },
+    {
+      "epoch": 65.75070821529745,
+      "loss": 0.03631053492426872,
+      "loss_ce": 3.276379356975667e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 993898768,
+      "step": 5792
+    },
+    {
+      "epoch": 65.76203966005666,
+      "grad_norm": 3.4643511357462073,
+      "learning_rate": 5e-06,
+      "loss": 0.059,
+      "num_input_tokens_seen": 994070592,
+      "step": 5793
+    },
+    {
+      "epoch": 65.76203966005666,
+      "loss": 0.052137747406959534,
+      "loss_ce": 1.372660062770592e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 994070592,
+      "step": 5793
+    },
+    {
+      "epoch": 65.77337110481587,
+      "grad_norm": 4.011102195483074,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 994240024,
+      "step": 5794
+    },
+    {
+      "epoch": 65.77337110481587,
+      "loss": 0.0688139796257019,
+      "loss_ce": 2.7361627871869132e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 994240024,
+      "step": 5794
+    },
+    {
+      "epoch": 65.78470254957507,
+      "grad_norm": 3.676975177737194,
+      "learning_rate": 5e-06,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 994410604,
+      "step": 5795
+    },
+    {
+      "epoch": 65.78470254957507,
+      "loss": 0.03442037105560303,
+      "loss_ce": 7.283472223207355e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 994410604,
+      "step": 5795
+    },
+    {
+      "epoch": 65.79603399433428,
+      "grad_norm": 3.958102057118438,
+      "learning_rate": 5e-06,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 994580852,
+      "step": 5796
+    },
+    {
+      "epoch": 65.79603399433428,
+      "loss": 0.05996246263384819,
+      "loss_ce": 5.6456676247762516e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 994580852,
+      "step": 5796
+    },
+    {
+      "epoch": 65.80736543909349,
+      "grad_norm": 3.6256979084322625,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 994752808,
+      "step": 5797
+    },
+    {
+      "epoch": 65.80736543909349,
+      "loss": 0.04044720157980919,
+      "loss_ce": 7.244764128699899e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 994752808,
+      "step": 5797
+    },
+    {
+      "epoch": 65.8186968838527,
+      "grad_norm": 3.6506547708510024,
+      "learning_rate": 5e-06,
+      "loss": 0.051,
+      "num_input_tokens_seen": 994921444,
+      "step": 5798
+    },
+    {
+      "epoch": 65.8186968838527,
+      "loss": 0.05915772542357445,
+      "loss_ce": 7.569455192424357e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 994921444,
+      "step": 5798
+    },
+    {
+      "epoch": 65.8300283286119,
+      "grad_norm": 4.0529672243563475,
+      "learning_rate": 5e-06,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 995093628,
+      "step": 5799
+    },
+    {
+      "epoch": 65.8300283286119,
+      "loss": 0.09639350324869156,
+      "loss_ce": 1.8988373994943686e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 995093628,
+      "step": 5799
+    },
+    {
+      "epoch": 65.84135977337111,
+      "grad_norm": 2.760159503163887,
+      "learning_rate": 5e-06,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 995265672,
+      "step": 5800
+    },
+    {
+      "epoch": 65.84135977337111,
+      "loss": 0.027931788936257362,
+      "loss_ce": 2.346429755561985e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 995265672,
+      "step": 5800
+    },
+    {
+      "epoch": 65.85269121813032,
+      "grad_norm": 2.0083589946602425,
+      "learning_rate": 5e-06,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 995436604,
+      "step": 5801
+    },
+    {
+      "epoch": 65.85269121813032,
+      "loss": 0.038099292665719986,
+      "loss_ce": 1.3354522707231808e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 995436604,
+      "step": 5801
+    },
+    {
+      "epoch": 65.86402266288952,
+      "grad_norm": 3.1245451031496283,
+      "learning_rate": 5e-06,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 995608516,
+      "step": 5802
+    },
+    {
+      "epoch": 65.86402266288952,
+      "loss": 0.043090157210826874,
+      "loss_ce": 0.00021295633632689714,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 995608516,
+      "step": 5802
+    },
+    {
+      "epoch": 65.87535410764872,
+      "grad_norm": 3.583962028136142,
+      "learning_rate": 5e-06,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 995779672,
+      "step": 5803
+    },
+    {
+      "epoch": 65.87535410764872,
+      "loss": 0.04351666197180748,
+      "loss_ce": 2.91143442154862e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 995779672,
+      "step": 5803
+    },
+    {
+      "epoch": 65.88668555240793,
+      "grad_norm": 3.7307450487507774,
+      "learning_rate": 5e-06,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 995950140,
+      "step": 5804
+    },
+    {
+      "epoch": 65.88668555240793,
+      "loss": 0.04242198169231415,
+      "loss_ce": 1.780654929461889e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 995950140,
+      "step": 5804
+    },
+    {
+      "epoch": 65.89801699716713,
+      "grad_norm": 3.333835225797573,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 996122144,
+      "step": 5805
+    },
+    {
+      "epoch": 65.89801699716713,
+      "loss": 0.03899308294057846,
+      "loss_ce": 2.213602783740498e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 996122144,
+      "step": 5805
+    },
+    {
+      "epoch": 65.90934844192634,
+      "grad_norm": 3.088832301858489,
+      "learning_rate": 5e-06,
+      "loss": 0.049,
+      "num_input_tokens_seen": 996293124,
+      "step": 5806
+    },
+    {
+      "epoch": 65.90934844192634,
+      "loss": 0.04271329194307327,
+      "loss_ce": 1.919774877023883e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 996293124,
+      "step": 5806
+    },
+    {
+      "epoch": 65.92067988668555,
+      "grad_norm": 3.1918272762212734,
+      "learning_rate": 5e-06,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 996465304,
+      "step": 5807
+    },
+    {
+      "epoch": 65.92067988668555,
+      "loss": 0.06512096524238586,
+      "loss_ce": 7.274914969457313e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 996465304,
+      "step": 5807
+    },
+    {
+      "epoch": 65.93201133144476,
+      "grad_norm": 3.3837897582855914,
+      "learning_rate": 5e-06,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 996637152,
+      "step": 5808
+    },
+    {
+      "epoch": 65.93201133144476,
+      "loss": 0.03420586884021759,
+      "loss_ce": 5.6695262173889205e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 996637152,
+      "step": 5808
+    },
+    {
+      "epoch": 65.94334277620396,
+      "grad_norm": 3.5474891351220488,
+      "learning_rate": 5e-06,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 996809196,
+      "step": 5809
+    },
+    {
+      "epoch": 65.94334277620396,
+      "loss": 0.06417332589626312,
+      "loss_ce": 2.5378167265444063e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 996809196,
+      "step": 5809
+    },
+    {
+      "epoch": 65.95467422096317,
+      "grad_norm": 3.358706827509532,
+      "learning_rate": 5e-06,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 996980928,
+      "step": 5810
+    },
+    {
+      "epoch": 65.95467422096317,
+      "loss": 0.028481513261795044,
+      "loss_ce": 2.3870634322520345e-05,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 996980928,
+      "step": 5810
+    },
+    {
+      "epoch": 65.96600566572238,
+      "grad_norm": 3.570636376145596,
+      "learning_rate": 5e-06,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 997152636,
+      "step": 5811
+    },
+    {
+      "epoch": 65.96600566572238,
+      "loss": 0.09663854539394379,
+      "loss_ce": 0.00020299361494835466,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 997152636,
+      "step": 5811
+    },
+    {
+      "epoch": 65.97733711048159,
+      "grad_norm": 3.5896584375412366,
+      "learning_rate": 5e-06,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 997322024,
+      "step": 5812
+    },
+    {
+      "epoch": 65.97733711048159,
+      "loss": 0.03467239439487457,
+      "loss_ce": 3.49411930073984e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 997322024,
+      "step": 5812
+    },
+    {
+      "epoch": 65.98866855524079,
+      "grad_norm": 3.19433135796106,
+      "learning_rate": 5e-06,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 997493992,
+      "step": 5813
+    },
+    {
+      "epoch": 65.98866855524079,
+      "loss": 0.061878662556409836,
+      "loss_ce": 3.479280348983593e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 997493992,
+      "step": 5813
+    },
+    {
+      "epoch": 66.0,
+      "grad_norm": 3.2508703732801822,
+      "learning_rate": 5e-06,
+      "loss": 0.036,
+      "num_input_tokens_seen": 997665704,
+      "step": 5814
+    },
+    {
+      "epoch": 66.0,
+      "loss": 0.029385238885879517,
+      "loss_ce": 1.2070284356013872e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 997665704,
+      "step": 5814
+    },
+    {
+      "epoch": 66.01133144475921,
+      "grad_norm": 2.982087293680888,
+      "learning_rate": 5e-06,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 997836092,
+      "step": 5815
+    },
+    {
+      "epoch": 66.01133144475921,
+      "loss": 0.04796624928712845,
+      "loss_ce": 2.3131056877900846e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 997836092,
+      "step": 5815
+    },
+    {
+      "epoch": 66.02266288951841,
+      "grad_norm": 2.9031186038490406,
+      "learning_rate": 5e-06,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 998007956,
+      "step": 5816
+    },
+    {
+      "epoch": 66.02266288951841,
+      "loss": 0.044926710426807404,
+      "loss_ce": 3.5350502002984285e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 998007956,
+      "step": 5816
+    },
+    {
+      "epoch": 66.03399433427762,
+      "grad_norm": 2.879831266219969,
+      "learning_rate": 5e-06,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 998177812,
+      "step": 5817
+    },
+    {
+      "epoch": 66.03399433427762,
+      "loss": 0.040174782276153564,
+      "loss_ce": 1.3646911611431278e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 998177812,
+      "step": 5817
+    },
+    {
+      "epoch": 66.04532577903683,
+      "grad_norm": 3.080295728619409,
+      "learning_rate": 5e-06,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 998349064,
+      "step": 5818
+    },
+    {
+      "epoch": 66.04532577903683,
+      "loss": 0.022834401577711105,
+      "loss_ce": 2.2511003407998942e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.0228271484375,
+      "num_input_tokens_seen": 998349064,
+      "step": 5818
+    },
+    {
+      "epoch": 66.05665722379604,
+      "grad_norm": 3.173499252796179,
+      "learning_rate": 5e-06,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 998520968,
+      "step": 5819
+    },
+    {
+      "epoch": 66.05665722379604,
+      "loss": 0.038249991834163666,
+      "loss_ce": 1.1462408110674005e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 998520968,
+      "step": 5819
+    },
+    {
+      "epoch": 66.06798866855524,
+      "grad_norm": 2.6150134399018246,
+      "learning_rate": 5e-06,
+      "loss": 0.027,
+      "num_input_tokens_seen": 998692948,
+      "step": 5820
+    },
+    {
+      "epoch": 66.06798866855524,
+      "loss": 0.029565272852778435,
+      "loss_ce": 5.4774878663010895e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 998692948,
+      "step": 5820
+    },
+    {
+      "epoch": 66.07932011331445,
+      "grad_norm": 3.025580036903358,
+      "learning_rate": 5e-06,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 998863436,
+      "step": 5821
+    },
+    {
+      "epoch": 66.07932011331445,
+      "loss": 0.026316750794649124,
+      "loss_ce": 1.8229262423119508e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 998863436,
+      "step": 5821
+    },
+    {
+      "epoch": 66.09065155807366,
+      "grad_norm": 4.718583219379033,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 999034668,
+      "step": 5822
+    },
+    {
+      "epoch": 66.09065155807366,
+      "loss": 0.09526698291301727,
+      "loss_ce": 2.161873271688819e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 999034668,
+      "step": 5822
+    },
+    {
+      "epoch": 66.10198300283287,
+      "grad_norm": 3.260550952286194,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 999206140,
+      "step": 5823
+    },
+    {
+      "epoch": 66.10198300283287,
+      "loss": 0.066210076212883,
+      "loss_ce": 1.7453430700697936e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 999206140,
+      "step": 5823
+    },
+    {
+      "epoch": 66.11331444759207,
+      "grad_norm": 3.7022550385730697,
+      "learning_rate": 5e-06,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 999378324,
+      "step": 5824
+    },
+    {
+      "epoch": 66.11331444759207,
+      "loss": 0.03564482182264328,
+      "loss_ce": 0.001808458473533392,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 999378324,
+      "step": 5824
+    },
+    {
+      "epoch": 66.12464589235128,
+      "grad_norm": 4.090211646471778,
+      "learning_rate": 5e-06,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 999549100,
+      "step": 5825
+    },
+    {
+      "epoch": 66.12464589235128,
+      "loss": 0.04278557002544403,
+      "loss_ce": 3.044380537176039e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 999549100,
+      "step": 5825
+    },
+    {
+      "epoch": 66.13597733711048,
+      "grad_norm": 4.372960864223768,
+      "learning_rate": 5e-06,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 999721040,
+      "step": 5826
+    },
+    {
+      "epoch": 66.13597733711048,
+      "loss": 0.04125019162893295,
+      "loss_ce": 2.0940937247360125e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 999721040,
+      "step": 5826
+    },
+    {
+      "epoch": 66.14730878186968,
+      "grad_norm": 3.346314970089101,
+      "learning_rate": 5e-06,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 999893296,
+      "step": 5827
+    },
+    {
+      "epoch": 66.14730878186968,
+      "loss": 0.033320240676403046,
+      "loss_ce": 2.556304752943106e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 999893296,
+      "step": 5827
+    },
+    {
+      "epoch": 66.15864022662889,
+      "grad_norm": 3.5593086325218426,
+      "learning_rate": 5e-06,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 1000065396,
+      "step": 5828
+    },
+    {
+      "epoch": 66.15864022662889,
+      "loss": 0.06338623911142349,
+      "loss_ce": 4.7003570216475055e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 1000065396,
+      "step": 5828
+    },
+    {
+      "epoch": 66.1699716713881,
+      "grad_norm": 3.659081421388046,
+      "learning_rate": 5e-06,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 1000235576,
+      "step": 5829
+    },
+    {
+      "epoch": 66.1699716713881,
+      "loss": 0.09226740151643753,
+      "loss_ce": 4.327727583586238e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 1000235576,
+      "step": 5829
+    },
+    {
+      "epoch": 66.1813031161473,
+      "grad_norm": 3.053870960199708,
+      "learning_rate": 5e-06,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 1000407280,
+      "step": 5830
+    },
+    {
+      "epoch": 66.1813031161473,
+      "loss": 0.03429713100194931,
+      "loss_ce": 2.5887809897540137e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1000407280,
+      "step": 5830
+    },
+    {
+      "epoch": 66.19263456090651,
+      "grad_norm": 3.547001296451918,
+      "learning_rate": 5e-06,
+      "loss": 0.033,
+      "num_input_tokens_seen": 1000579428,
+      "step": 5831
+    },
+    {
+      "epoch": 66.19263456090651,
+      "loss": 0.028814319521188736,
+      "loss_ce": 2.0983618014724925e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 1000579428,
+      "step": 5831
+    },
+    {
+      "epoch": 66.20396600566572,
+      "grad_norm": 3.5174642503992857,
+      "learning_rate": 5e-06,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 1000751072,
+      "step": 5832
+    },
+    {
+      "epoch": 66.20396600566572,
+      "loss": 0.030492212623357773,
+      "loss_ce": 2.0411091099958867e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1000751072,
+      "step": 5832
+    },
+    {
+      "epoch": 66.21529745042493,
+      "grad_norm": 3.403598196342013,
+      "learning_rate": 5e-06,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 1000922844,
+      "step": 5833
+    },
+    {
+      "epoch": 66.21529745042493,
+      "loss": 0.03076525591313839,
+      "loss_ce": 8.74602046678774e-05,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 1000922844,
+      "step": 5833
+    },
+    {
+      "epoch": 66.22662889518413,
+      "grad_norm": 3.667321037271246,
+      "learning_rate": 5e-06,
+      "loss": 0.038,
+      "num_input_tokens_seen": 1001094860,
+      "step": 5834
+    },
+    {
+      "epoch": 66.22662889518413,
+      "loss": 0.04786504805088043,
+      "loss_ce": 2.874444362532813e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 1001094860,
+      "step": 5834
+    },
+    {
+      "epoch": 66.23796033994334,
+      "grad_norm": 3.0871553739144177,
+      "learning_rate": 5e-06,
+      "loss": 0.049,
+      "num_input_tokens_seen": 1001266512,
+      "step": 5835
+    },
+    {
+      "epoch": 66.23796033994334,
+      "loss": 0.04659239575266838,
+      "loss_ce": 2.2572769012185745e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 1001266512,
+      "step": 5835
+    },
+    {
+      "epoch": 66.24929178470255,
+      "grad_norm": 2.9355038930060147,
+      "learning_rate": 5e-06,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 1001438652,
+      "step": 5836
+    },
+    {
+      "epoch": 66.24929178470255,
+      "loss": 0.0538221038877964,
+      "loss_ce": 1.9614308257587254e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 1001438652,
+      "step": 5836
+    },
+    {
+      "epoch": 66.26062322946176,
+      "grad_norm": 3.0633631698576447,
+      "learning_rate": 5e-06,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 1001609572,
+      "step": 5837
+    },
+    {
+      "epoch": 66.26062322946176,
+      "loss": 0.03917674720287323,
+      "loss_ce": 4.5582084567286074e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 1001609572,
+      "step": 5837
+    },
+    {
+      "epoch": 66.27195467422096,
+      "grad_norm": 3.196188132065316,
+      "learning_rate": 5e-06,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 1001780432,
+      "step": 5838
+    },
+    {
+      "epoch": 66.27195467422096,
+      "loss": 0.08353699743747711,
+      "loss_ce": 2.5648923838161863e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 1001780432,
+      "step": 5838
+    },
+    {
+      "epoch": 66.28328611898017,
+      "grad_norm": 3.1527980281563592,
+      "learning_rate": 5e-06,
+      "loss": 0.036,
+      "num_input_tokens_seen": 1001952348,
+      "step": 5839
+    },
+    {
+      "epoch": 66.28328611898017,
+      "loss": 0.04768617823719978,
+      "loss_ce": 4.8238645831588656e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 1001952348,
+      "step": 5839
+    },
+    {
+      "epoch": 66.29461756373938,
+      "grad_norm": 3.027742268942204,
+      "learning_rate": 5e-06,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 1002124200,
+      "step": 5840
+    },
+    {
+      "epoch": 66.29461756373938,
+      "loss": 0.028493424877524376,
+      "loss_ce": 2.0525785657810047e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 1002124200,
+      "step": 5840
+    },
+    {
+      "epoch": 66.30594900849859,
+      "grad_norm": 2.396562150462242,
+      "learning_rate": 5e-06,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 1002295348,
+      "step": 5841
+    },
+    {
+      "epoch": 66.30594900849859,
+      "loss": 0.030366579070687294,
+      "loss_ce": 4.736658956971951e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1002295348,
+      "step": 5841
+    },
+    {
+      "epoch": 66.3172804532578,
+      "grad_norm": 2.464358527557364,
+      "learning_rate": 5e-06,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 1002465672,
+      "step": 5842
+    },
+    {
+      "epoch": 66.3172804532578,
+      "loss": 0.03490632772445679,
+      "loss_ce": 3.999529144493863e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1002465672,
+      "step": 5842
+    },
+    {
+      "epoch": 66.328611898017,
+      "grad_norm": 3.0308350853691026,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 1002636504,
+      "step": 5843
+    },
+    {
+      "epoch": 66.328611898017,
+      "loss": 0.047676317393779755,
+      "loss_ce": 5.3635478252545e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 1002636504,
+      "step": 5843
+    },
+    {
+      "epoch": 66.33994334277621,
+      "grad_norm": 3.519476984782888,
+      "learning_rate": 5e-06,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 1002807508,
+      "step": 5844
+    },
+    {
+      "epoch": 66.33994334277621,
+      "loss": 0.028714781627058983,
+      "loss_ce": 4.351743700681254e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 1002807508,
+      "step": 5844
+    },
+    {
+      "epoch": 66.35127478753542,
+      "grad_norm": 3.6832435047484644,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 1002979232,
+      "step": 5845
+    },
+    {
+      "epoch": 66.35127478753542,
+      "loss": 0.058221690356731415,
+      "loss_ce": 2.4667113393661566e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 1002979232,
+      "step": 5845
+    },
+    {
+      "epoch": 66.36260623229462,
+      "grad_norm": 3.2213159575847463,
+      "learning_rate": 5e-06,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 1003150108,
+      "step": 5846
+    },
+    {
+      "epoch": 66.36260623229462,
+      "loss": 0.03850887715816498,
+      "loss_ce": 2.6208439521724358e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1003150108,
+      "step": 5846
+    },
+    {
+      "epoch": 66.37393767705383,
+      "grad_norm": 3.0780758587924186,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 1003322116,
+      "step": 5847
+    },
+    {
+      "epoch": 66.37393767705383,
+      "loss": 0.1336781233549118,
+      "loss_ce": 2.6381814677733928e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 1003322116,
+      "step": 5847
+    },
+    {
+      "epoch": 66.38526912181302,
+      "grad_norm": 3.0114202711595164,
+      "learning_rate": 5e-06,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 1003493668,
+      "step": 5848
+    },
+    {
+      "epoch": 66.38526912181302,
+      "loss": 0.03140972927212715,
+      "loss_ce": 6.817728717578575e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1003493668,
+      "step": 5848
+    },
+    {
+      "epoch": 66.39660056657223,
+      "grad_norm": 3.0600816238815436,
+      "learning_rate": 5e-06,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 1003664036,
+      "step": 5849
+    },
+    {
+      "epoch": 66.39660056657223,
+      "loss": 0.05011823773384094,
+      "loss_ce": 9.992759441956878e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 1003664036,
+      "step": 5849
+    },
+    {
+      "epoch": 66.40793201133144,
+      "grad_norm": 3.1787078847168337,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 1003835124,
+      "step": 5850
+    },
+    {
+      "epoch": 66.40793201133144,
+      "loss": 0.045853085815906525,
+      "loss_ce": 1.568091283843387e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 1003835124,
+      "step": 5850
+    },
+    {
+      "epoch": 66.41926345609065,
+      "grad_norm": 2.8499363990472206,
+      "learning_rate": 5e-06,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 1004006716,
+      "step": 5851
+    },
+    {
+      "epoch": 66.41926345609065,
+      "loss": 0.02343943528831005,
+      "loss_ce": 3.245188054279424e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.004669189453125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 1004006716,
+      "step": 5851
+    },
+    {
+      "epoch": 66.43059490084985,
+      "grad_norm": 2.90798252892652,
+      "learning_rate": 5e-06,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 1004178412,
+      "step": 5852
+    },
+    {
+      "epoch": 66.43059490084985,
+      "loss": 0.03418688476085663,
+      "loss_ce": 3.771146293729544e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1004178412,
+      "step": 5852
+    },
+    {
+      "epoch": 66.44192634560906,
+      "grad_norm": 3.7013428476250754,
+      "learning_rate": 5e-06,
+      "loss": 0.0695,
+      "num_input_tokens_seen": 1004350224,
+      "step": 5853
+    },
+    {
+      "epoch": 66.44192634560906,
+      "loss": 0.06857158988714218,
+      "loss_ce": 1.3844116438122e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 1004350224,
+      "step": 5853
+    },
+    {
+      "epoch": 66.45325779036827,
+      "grad_norm": 3.6578450552378907,
+      "learning_rate": 5e-06,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 1004521616,
+      "step": 5854
+    },
+    {
+      "epoch": 66.45325779036827,
+      "loss": 0.03935757279396057,
+      "loss_ce": 2.0414892787812278e-05,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 1004521616,
+      "step": 5854
+    },
+    {
+      "epoch": 66.46458923512748,
+      "grad_norm": 3.5320625754576453,
+      "learning_rate": 5e-06,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 1004693676,
+      "step": 5855
+    },
+    {
+      "epoch": 66.46458923512748,
+      "loss": 0.03869687393307686,
+      "loss_ce": 1.5844016161281615e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1004693676,
+      "step": 5855
+    },
+    {
+      "epoch": 66.47592067988668,
+      "grad_norm": 3.3561761166729176,
+      "learning_rate": 5e-06,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 1004863876,
+      "step": 5856
+    },
+    {
+      "epoch": 66.47592067988668,
+      "loss": 0.034067802131175995,
+      "loss_ce": 2.544432572904043e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1004863876,
+      "step": 5856
+    },
+    {
+      "epoch": 66.48725212464589,
+      "grad_norm": 3.0898957578953783,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 1005035548,
+      "step": 5857
+    },
+    {
+      "epoch": 66.48725212464589,
+      "loss": 0.0312093123793602,
+      "loss_ce": 2.0346453311503865e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1005035548,
+      "step": 5857
+    },
+    {
+      "epoch": 66.4985835694051,
+      "grad_norm": 5.9490567310327345,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 1005206636,
+      "step": 5858
+    },
+    {
+      "epoch": 66.4985835694051,
+      "loss": 0.07717415690422058,
+      "loss_ce": 0.0008725836523808539,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 1005206636,
+      "step": 5858
+    },
+    {
+      "epoch": 66.5099150141643,
+      "grad_norm": 3.587159520297805,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1005377448,
+      "step": 5859
+    },
+    {
+      "epoch": 66.5099150141643,
+      "loss": 0.0374593585729599,
+      "loss_ce": 4.480699135456234e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 1005377448,
+      "step": 5859
+    },
+    {
+      "epoch": 66.52124645892351,
+      "grad_norm": 3.221606922633493,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 1005548260,
+      "step": 5860
+    },
+    {
+      "epoch": 66.52124645892351,
+      "loss": 0.07446883618831635,
+      "loss_ce": 3.6456749512581155e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 1005548260,
+      "step": 5860
+    },
+    {
+      "epoch": 66.53257790368272,
+      "grad_norm": 2.9724617301496763,
+      "learning_rate": 5e-06,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 1005718152,
+      "step": 5861
+    },
+    {
+      "epoch": 66.53257790368272,
+      "loss": 0.08820773661136627,
+      "loss_ce": 7.29785897419788e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 1005718152,
+      "step": 5861
+    },
+    {
+      "epoch": 66.54390934844193,
+      "grad_norm": 2.2221241336411763,
+      "learning_rate": 5e-06,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 1005889012,
+      "step": 5862
+    },
+    {
+      "epoch": 66.54390934844193,
+      "loss": 0.02373218536376953,
+      "loss_ce": 2.002680230361875e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 1005889012,
+      "step": 5862
+    },
+    {
+      "epoch": 66.55524079320114,
+      "grad_norm": 2.2178180780421948,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1006059780,
+      "step": 5863
+    },
+    {
+      "epoch": 66.55524079320114,
+      "loss": 0.029174644500017166,
+      "loss_ce": 0.001701194210909307,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 1006059780,
+      "step": 5863
+    },
+    {
+      "epoch": 66.56657223796034,
+      "grad_norm": 2.4287566193503576,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 1006231408,
+      "step": 5864
+    },
+    {
+      "epoch": 66.56657223796034,
+      "loss": 0.05618676915764809,
+      "loss_ce": 1.916783003252931e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 1006231408,
+      "step": 5864
+    },
+    {
+      "epoch": 66.57790368271955,
+      "grad_norm": 2.8344261082103213,
+      "learning_rate": 5e-06,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 1006402524,
+      "step": 5865
+    },
+    {
+      "epoch": 66.57790368271955,
+      "loss": 0.026119187474250793,
+      "loss_ce": 2.665856300154701e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 1006402524,
+      "step": 5865
+    },
+    {
+      "epoch": 66.58923512747876,
+      "grad_norm": 3.8200780796833893,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 1006573884,
+      "step": 5866
+    },
+    {
+      "epoch": 66.58923512747876,
+      "loss": 0.03293679282069206,
+      "loss_ce": 2.3586859242641367e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 1006573884,
+      "step": 5866
+    },
+    {
+      "epoch": 66.60056657223797,
+      "grad_norm": 4.397566459606763,
+      "learning_rate": 5e-06,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 1006745588,
+      "step": 5867
+    },
+    {
+      "epoch": 66.60056657223797,
+      "loss": 0.03528425097465515,
+      "loss_ce": 3.645121250883676e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1006745588,
+      "step": 5867
+    },
+    {
+      "epoch": 66.61189801699717,
+      "grad_norm": 3.5239528076772406,
+      "learning_rate": 5e-06,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 1006917580,
+      "step": 5868
+    },
+    {
+      "epoch": 66.61189801699717,
+      "loss": 0.03473358973860741,
+      "loss_ce": 3.510230453684926e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1006917580,
+      "step": 5868
+    },
+    {
+      "epoch": 66.62322946175638,
+      "grad_norm": 3.080813349138261,
+      "learning_rate": 5e-06,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 1007087732,
+      "step": 5869
+    },
+    {
+      "epoch": 66.62322946175638,
+      "loss": 0.034828707575798035,
+      "loss_ce": 2.3409364075632766e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1007087732,
+      "step": 5869
+    },
+    {
+      "epoch": 66.63456090651557,
+      "grad_norm": 3.083926074968894,
+      "learning_rate": 5e-06,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 1007259412,
+      "step": 5870
+    },
+    {
+      "epoch": 66.63456090651557,
+      "loss": 0.09153534471988678,
+      "loss_ce": 1.3125488294463139e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 1007259412,
+      "step": 5870
+    },
+    {
+      "epoch": 66.64589235127478,
+      "grad_norm": 3.3350508360299043,
+      "learning_rate": 5e-06,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 1007430116,
+      "step": 5871
+    },
+    {
+      "epoch": 66.64589235127478,
+      "loss": 0.029761623591184616,
+      "loss_ce": 3.750433825189248e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 1007430116,
+      "step": 5871
+    },
+    {
+      "epoch": 66.65722379603399,
+      "grad_norm": 3.482152951698697,
+      "learning_rate": 5e-06,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 1007602320,
+      "step": 5872
+    },
+    {
+      "epoch": 66.65722379603399,
+      "loss": 0.04584764689207077,
+      "loss_ce": 4.0761849959380925e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 1007602320,
+      "step": 5872
+    },
+    {
+      "epoch": 66.6685552407932,
+      "grad_norm": 3.8432140256900147,
+      "learning_rate": 5e-06,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 1007774248,
+      "step": 5873
+    },
+    {
+      "epoch": 66.6685552407932,
+      "loss": 0.06982676684856415,
+      "loss_ce": 1.7805064999265596e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 1007774248,
+      "step": 5873
+    },
+    {
+      "epoch": 66.6798866855524,
+      "grad_norm": 3.0378203476707664,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 1007945096,
+      "step": 5874
+    },
+    {
+      "epoch": 66.6798866855524,
+      "loss": 0.028539368882775307,
+      "loss_ce": 9.698656504042447e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 1007945096,
+      "step": 5874
+    },
+    {
+      "epoch": 66.69121813031161,
+      "grad_norm": 3.340986507941867,
+      "learning_rate": 5e-06,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 1008116756,
+      "step": 5875
+    },
+    {
+      "epoch": 66.69121813031161,
+      "loss": 0.03618955239653587,
+      "loss_ce": 2.6222069209325127e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 1008116756,
+      "step": 5875
+    },
+    {
+      "epoch": 66.70254957507082,
+      "grad_norm": 3.7052307835458067,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 1008288816,
+      "step": 5876
+    },
+    {
+      "epoch": 66.70254957507082,
+      "loss": 0.03358769416809082,
+      "loss_ce": 1.8356109649175778e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1008288816,
+      "step": 5876
+    },
+    {
+      "epoch": 66.71388101983003,
+      "grad_norm": 3.5810112715152584,
+      "learning_rate": 5e-06,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 1008459372,
+      "step": 5877
+    },
+    {
+      "epoch": 66.71388101983003,
+      "loss": 0.0730247050523758,
+      "loss_ce": 1.1404594260966405e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 1008459372,
+      "step": 5877
+    },
+    {
+      "epoch": 66.72521246458923,
+      "grad_norm": 3.7009198375560497,
+      "learning_rate": 5e-06,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 1008631252,
+      "step": 5878
+    },
+    {
+      "epoch": 66.72521246458923,
+      "loss": 0.054692186415195465,
+      "loss_ce": 1.9942915969295427e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 1008631252,
+      "step": 5878
+    },
+    {
+      "epoch": 66.73654390934844,
+      "grad_norm": 3.5753158759883266,
+      "learning_rate": 5e-06,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 1008803048,
+      "step": 5879
+    },
+    {
+      "epoch": 66.73654390934844,
+      "loss": 0.06965494900941849,
+      "loss_ce": 2.9097438527969643e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 1008803048,
+      "step": 5879
+    },
+    {
+      "epoch": 66.74787535410765,
+      "grad_norm": 4.073487495975577,
+      "learning_rate": 5e-06,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 1008973744,
+      "step": 5880
+    },
+    {
+      "epoch": 66.74787535410765,
+      "loss": 0.04496660828590393,
+      "loss_ce": 7.524955435656011e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 1008973744,
+      "step": 5880
+    },
+    {
+      "epoch": 66.75920679886686,
+      "grad_norm": 4.021521978359537,
+      "learning_rate": 5e-06,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 1009144380,
+      "step": 5881
+    },
+    {
+      "epoch": 66.75920679886686,
+      "loss": 0.037859391421079636,
+      "loss_ce": 3.285380080342293e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 1009144380,
+      "step": 5881
+    },
+    {
+      "epoch": 66.77053824362606,
+      "grad_norm": 3.495021723455332,
+      "learning_rate": 5e-06,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 1009315944,
+      "step": 5882
+    },
+    {
+      "epoch": 66.77053824362606,
+      "loss": 0.03854941576719284,
+      "loss_ce": 5.149029311724007e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1009315944,
+      "step": 5882
+    },
+    {
+      "epoch": 66.78186968838527,
+      "grad_norm": 3.2580233992851557,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 1009485472,
+      "step": 5883
+    },
+    {
+      "epoch": 66.78186968838527,
+      "loss": 0.04334977641701698,
+      "loss_ce": 1.4816567272646353e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 1009485472,
+      "step": 5883
+    },
+    {
+      "epoch": 66.79320113314448,
+      "grad_norm": 3.5475812380333536,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 1009657520,
+      "step": 5884
+    },
+    {
+      "epoch": 66.79320113314448,
+      "loss": 0.07149433344602585,
+      "loss_ce": 6.911292075528763e-06,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 1009657520,
+      "step": 5884
+    },
+    {
+      "epoch": 66.80453257790369,
+      "grad_norm": 3.9318048365168767,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 1009829400,
+      "step": 5885
+    },
+    {
+      "epoch": 66.80453257790369,
+      "loss": 0.04538831114768982,
+      "loss_ce": 2.3933404008857906e-05,
+      "loss_iou": 0.578125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 1009829400,
+      "step": 5885
+    },
+    {
+      "epoch": 66.8158640226629,
+      "grad_norm": 3.9569411520948616,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 1010001112,
+      "step": 5886
+    },
+    {
+      "epoch": 66.8158640226629,
+      "loss": 0.04939759895205498,
+      "loss_ce": 9.645023965276778e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 1010001112,
+      "step": 5886
+    },
+    {
+      "epoch": 66.8271954674221,
+      "grad_norm": 3.267906343917386,
+      "learning_rate": 5e-06,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 1010173532,
+      "step": 5887
+    },
+    {
+      "epoch": 66.8271954674221,
+      "loss": 0.06323733925819397,
+      "loss_ce": 2.017248516494874e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 1010173532,
+      "step": 5887
+    },
+    {
+      "epoch": 66.83852691218131,
+      "grad_norm": 3.4780046126140705,
+      "learning_rate": 5e-06,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 1010345404,
+      "step": 5888
+    },
+    {
+      "epoch": 66.83852691218131,
+      "loss": 0.07771638035774231,
+      "loss_ce": 3.38833051500842e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 1010345404,
+      "step": 5888
+    },
+    {
+      "epoch": 66.84985835694052,
+      "grad_norm": 3.846023019233165,
+      "learning_rate": 5e-06,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 1010516620,
+      "step": 5889
+    },
+    {
+      "epoch": 66.84985835694052,
+      "loss": 0.03444063663482666,
+      "loss_ce": 3.206765541108325e-05,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 1010516620,
+      "step": 5889
+    },
+    {
+      "epoch": 66.86118980169972,
+      "grad_norm": 3.638996902667991,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 1010688608,
+      "step": 5890
+    },
+    {
+      "epoch": 66.86118980169972,
+      "loss": 0.04404052719473839,
+      "loss_ce": 1.8919668946182355e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 1010688608,
+      "step": 5890
+    },
+    {
+      "epoch": 66.87252124645893,
+      "grad_norm": 3.6262065352479524,
+      "learning_rate": 5e-06,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 1010860780,
+      "step": 5891
+    },
+    {
+      "epoch": 66.87252124645893,
+      "loss": 0.03483825922012329,
+      "loss_ce": 1.7704283891362138e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1010860780,
+      "step": 5891
+    },
+    {
+      "epoch": 66.88385269121812,
+      "grad_norm": 3.7192873429292064,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 1011032796,
+      "step": 5892
+    },
+    {
+      "epoch": 66.88385269121812,
+      "loss": 0.040547676384449005,
+      "loss_ce": 2.033288728853222e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1011032796,
+      "step": 5892
+    },
+    {
+      "epoch": 66.89518413597733,
+      "grad_norm": 3.230528540400533,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 1011203144,
+      "step": 5893
+    },
+    {
+      "epoch": 66.89518413597733,
+      "loss": 0.04866734892129898,
+      "loss_ce": 0.00826207548379898,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1011203144,
+      "step": 5893
+    },
+    {
+      "epoch": 66.90651558073654,
+      "grad_norm": 3.128661556011401,
+      "learning_rate": 5e-06,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 1011375216,
+      "step": 5894
+    },
+    {
+      "epoch": 66.90651558073654,
+      "loss": 0.03125831484794617,
+      "loss_ce": 5.40924520464614e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1011375216,
+      "step": 5894
+    },
+    {
+      "epoch": 66.91784702549575,
+      "grad_norm": 3.4362861421303217,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 1011547284,
+      "step": 5895
+    },
+    {
+      "epoch": 66.91784702549575,
+      "loss": 0.037881843745708466,
+      "loss_ce": 2.478659553162288e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 1011547284,
+      "step": 5895
+    },
+    {
+      "epoch": 66.92917847025495,
+      "grad_norm": 4.216842845805076,
+      "learning_rate": 5e-06,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 1011719284,
+      "step": 5896
+    },
+    {
+      "epoch": 66.92917847025495,
+      "loss": 0.029956012964248657,
+      "loss_ce": 1.8267914128955454e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 1011719284,
+      "step": 5896
+    },
+    {
+      "epoch": 66.94050991501416,
+      "grad_norm": 3.3632649210293906,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 1011890616,
+      "step": 5897
+    },
+    {
+      "epoch": 66.94050991501416,
+      "loss": 0.06394051015377045,
+      "loss_ce": 2.1438916519400664e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 1011890616,
+      "step": 5897
+    },
+    {
+      "epoch": 66.95184135977337,
+      "grad_norm": 3.5603245536371477,
+      "learning_rate": 5e-06,
+      "loss": 0.067,
+      "num_input_tokens_seen": 1012062656,
+      "step": 5898
+    },
+    {
+      "epoch": 66.95184135977337,
+      "loss": 0.0314258374273777,
+      "loss_ce": 2.325181230844464e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 1012062656,
+      "step": 5898
+    },
+    {
+      "epoch": 66.96317280453258,
+      "grad_norm": 3.295416702105483,
+      "learning_rate": 5e-06,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 1012233696,
+      "step": 5899
+    },
+    {
+      "epoch": 66.96317280453258,
+      "loss": 0.06347407400608063,
+      "loss_ce": 0.0002569097268860787,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 1012233696,
+      "step": 5899
+    },
+    {
+      "epoch": 66.97450424929178,
+      "grad_norm": 2.776204452955552,
+      "learning_rate": 5e-06,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 1012405572,
+      "step": 5900
+    },
+    {
+      "epoch": 66.97450424929178,
+      "loss": 0.032131895422935486,
+      "loss_ce": 4.266363612259738e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1012405572,
+      "step": 5900
+    },
+    {
+      "epoch": 66.98583569405099,
+      "grad_norm": 2.4209620667450964,
+      "learning_rate": 5e-06,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 1012577560,
+      "step": 5901
+    },
+    {
+      "epoch": 66.98583569405099,
+      "loss": 0.03457998111844063,
+      "loss_ce": 2.6453566533746198e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1012577560,
+      "step": 5901
+    },
+    {
+      "epoch": 66.9971671388102,
+      "grad_norm": 2.9366039666022847,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 1012748648,
+      "step": 5902
+    },
+    {
+      "epoch": 66.9971671388102,
+      "loss": 0.03843795508146286,
+      "loss_ce": 3.158029721817002e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1012748648,
+      "step": 5902
+    },
+    {
+      "epoch": 66.9971671388102,
+      "loss": 0.032693568617105484,
+      "loss_ce": 9.243114618584514e-06,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 1012791668,
+      "step": 5902
+    },
+    {
+      "epoch": 67.0084985835694,
+      "grad_norm": 3.410598001843645,
+      "learning_rate": 5e-06,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 1012921052,
+      "step": 5903
+    },
+    {
+      "epoch": 67.0084985835694,
+      "loss": 0.0994865819811821,
+      "loss_ce": 1.453044023946859e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 1012921052,
+      "step": 5903
+    },
+    {
+      "epoch": 67.01983002832861,
+      "grad_norm": 3.139367601624398,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1013091632,
+      "step": 5904
+    },
+    {
+      "epoch": 67.01983002832861,
+      "loss": 0.07225705683231354,
+      "loss_ce": 2.194325134041719e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 1013091632,
+      "step": 5904
+    },
+    {
+      "epoch": 67.03116147308782,
+      "grad_norm": 3.170914475670413,
+      "learning_rate": 5e-06,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 1013263640,
+      "step": 5905
+    },
+    {
+      "epoch": 67.03116147308782,
+      "loss": 0.028303295373916626,
+      "loss_ce": 4.401723708724603e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 1013263640,
+      "step": 5905
+    },
+    {
+      "epoch": 67.04249291784703,
+      "grad_norm": 3.4233038333469907,
+      "learning_rate": 5e-06,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 1013435128,
+      "step": 5906
+    },
+    {
+      "epoch": 67.04249291784703,
+      "loss": 0.06967861205339432,
+      "loss_ce": 6.983099865465192e-06,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 1013435128,
+      "step": 5906
+    },
+    {
+      "epoch": 67.05382436260624,
+      "grad_norm": 2.9230778941112296,
+      "learning_rate": 5e-06,
+      "loss": 0.0651,
+      "num_input_tokens_seen": 1013606892,
+      "step": 5907
+    },
+    {
+      "epoch": 67.05382436260624,
+      "loss": 0.03308812528848648,
+      "loss_ce": 3.7587400584015995e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 1013606892,
+      "step": 5907
+    },
+    {
+      "epoch": 67.06515580736544,
+      "grad_norm": 3.149083378186109,
+      "learning_rate": 5e-06,
+      "loss": 0.059,
+      "num_input_tokens_seen": 1013777608,
+      "step": 5908
+    },
+    {
+      "epoch": 67.06515580736544,
+      "loss": 0.03846941143274307,
+      "loss_ce": 1.7260059394175187e-05,
+      "loss_iou": 0.115234375,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1013777608,
+      "step": 5908
+    },
+    {
+      "epoch": 67.07648725212465,
+      "grad_norm": 3.822322036963989,
+      "learning_rate": 5e-06,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 1013949232,
+      "step": 5909
+    },
+    {
+      "epoch": 67.07648725212465,
+      "loss": 0.0286090187728405,
+      "loss_ce": 5.98249607719481e-05,
+      "loss_iou": 0.1201171875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 1013949232,
+      "step": 5909
+    },
+    {
+      "epoch": 67.08781869688386,
+      "grad_norm": 3.059393793797116,
+      "learning_rate": 5e-06,
+      "loss": 0.054,
+      "num_input_tokens_seen": 1014120344,
+      "step": 5910
+    },
+    {
+      "epoch": 67.08781869688386,
+      "loss": 0.07710748165845871,
+      "loss_ce": 0.00017267190560232848,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 1014120344,
+      "step": 5910
+    },
+    {
+      "epoch": 67.09915014164307,
+      "grad_norm": 3.357287027752127,
+      "learning_rate": 5e-06,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 1014292544,
+      "step": 5911
+    },
+    {
+      "epoch": 67.09915014164307,
+      "loss": 0.026154689490795135,
+      "loss_ce": 3.164316149195656e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 1014292544,
+      "step": 5911
+    },
+    {
+      "epoch": 67.11048158640227,
+      "grad_norm": 3.8670329388988627,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 1014464684,
+      "step": 5912
+    },
+    {
+      "epoch": 67.11048158640227,
+      "loss": 0.026670530438423157,
+      "loss_ce": 2.8684116841759533e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 1014464684,
+      "step": 5912
+    },
+    {
+      "epoch": 67.12181303116148,
+      "grad_norm": 4.266413390132541,
+      "learning_rate": 5e-06,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 1014635884,
+      "step": 5913
+    },
+    {
+      "epoch": 67.12181303116148,
+      "loss": 0.04703632742166519,
+      "loss_ce": 2.399746881565079e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 1014635884,
+      "step": 5913
+    },
+    {
+      "epoch": 67.13314447592067,
+      "grad_norm": 3.56594202447463,
+      "learning_rate": 5e-06,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 1014806688,
+      "step": 5914
+    },
+    {
+      "epoch": 67.13314447592067,
+      "loss": 0.034070972353219986,
+      "loss_ce": 4.387211083667353e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1014806688,
+      "step": 5914
+    },
+    {
+      "epoch": 67.14447592067988,
+      "grad_norm": 3.8889687679784086,
+      "learning_rate": 5e-06,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 1014977536,
+      "step": 5915
+    },
+    {
+      "epoch": 67.14447592067988,
+      "loss": 0.0505722351372242,
+      "loss_ce": 1.9869721654686145e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 1014977536,
+      "step": 5915
+    },
+    {
+      "epoch": 67.15580736543909,
+      "grad_norm": 4.063078445855036,
+      "learning_rate": 5e-06,
+      "loss": 0.06,
+      "num_input_tokens_seen": 1015148744,
+      "step": 5916
+    },
+    {
+      "epoch": 67.15580736543909,
+      "loss": 0.040352530777454376,
+      "loss_ce": 2.3553355276817456e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 1015148744,
+      "step": 5916
+    },
+    {
+      "epoch": 67.1671388101983,
+      "grad_norm": 3.6187104370536463,
+      "learning_rate": 5e-06,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 1015321004,
+      "step": 5917
+    },
+    {
+      "epoch": 67.1671388101983,
+      "loss": 0.03494476526975632,
+      "loss_ce": 1.7398597265128046e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1015321004,
+      "step": 5917
+    },
+    {
+      "epoch": 67.1784702549575,
+      "grad_norm": 3.874309356516632,
+      "learning_rate": 5e-06,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 1015492936,
+      "step": 5918
+    },
+    {
+      "epoch": 67.1784702549575,
+      "loss": 0.05753310024738312,
+      "loss_ce": 2.2725695089320652e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 1015492936,
+      "step": 5918
+    },
+    {
+      "epoch": 67.18980169971671,
+      "grad_norm": 3.495275834336859,
+      "learning_rate": 5e-06,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 1015664848,
+      "step": 5919
+    },
+    {
+      "epoch": 67.18980169971671,
+      "loss": 0.03310427442193031,
+      "loss_ce": 2.322067302884534e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1015664848,
+      "step": 5919
+    },
+    {
+      "epoch": 67.20113314447592,
+      "grad_norm": 5.61487747988574,
+      "learning_rate": 5e-06,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 1015835460,
+      "step": 5920
+    },
+    {
+      "epoch": 67.20113314447592,
+      "loss": 0.057879336178302765,
+      "loss_ce": 1.8009741324931383e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 1015835460,
+      "step": 5920
+    },
+    {
+      "epoch": 67.21246458923513,
+      "grad_norm": 3.28088184153324,
+      "learning_rate": 5e-06,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 1016006280,
+      "step": 5921
+    },
+    {
+      "epoch": 67.21246458923513,
+      "loss": 0.03148248791694641,
+      "loss_ce": 1.8867329345084727e-05,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 1016006280,
+      "step": 5921
+    },
+    {
+      "epoch": 67.22379603399433,
+      "grad_norm": 3.1613892748712806,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 1016176752,
+      "step": 5922
+    },
+    {
+      "epoch": 67.22379603399433,
+      "loss": 0.033312857151031494,
+      "loss_ce": 1.818059172364883e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1016176752,
+      "step": 5922
+    },
+    {
+      "epoch": 67.23512747875354,
+      "grad_norm": 3.2289602236315087,
+      "learning_rate": 5e-06,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 1016348216,
+      "step": 5923
+    },
+    {
+      "epoch": 67.23512747875354,
+      "loss": 0.0476088672876358,
+      "loss_ce": 1.6705838788766414e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 1016348216,
+      "step": 5923
+    },
+    {
+      "epoch": 67.24645892351275,
+      "grad_norm": 4.005952449048984,
+      "learning_rate": 5e-06,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 1016520284,
+      "step": 5924
+    },
+    {
+      "epoch": 67.24645892351275,
+      "loss": 0.03556230291724205,
+      "loss_ce": 9.32509919948643e-06,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 1016520284,
+      "step": 5924
+    },
+    {
+      "epoch": 67.25779036827196,
+      "grad_norm": 3.807551435154973,
+      "learning_rate": 5e-06,
+      "loss": 0.047,
+      "num_input_tokens_seen": 1016691804,
+      "step": 5925
+    },
+    {
+      "epoch": 67.25779036827196,
+      "loss": 0.027643997222185135,
+      "loss_ce": 2.5588775315554813e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 1016691804,
+      "step": 5925
+    },
+    {
+      "epoch": 67.26912181303116,
+      "grad_norm": 2.1898374351273358,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 1016861324,
+      "step": 5926
+    },
+    {
+      "epoch": 67.26912181303116,
+      "loss": 0.027906589210033417,
+      "loss_ce": 4.4037507905159146e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 1016861324,
+      "step": 5926
+    },
+    {
+      "epoch": 67.28045325779037,
+      "grad_norm": 2.6943783282324074,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 1017033324,
+      "step": 5927
+    },
+    {
+      "epoch": 67.28045325779037,
+      "loss": 0.03989099711179733,
+      "loss_ce": 5.030314059695229e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 1017033324,
+      "step": 5927
+    },
+    {
+      "epoch": 67.29178470254958,
+      "grad_norm": 3.8670128059336544,
+      "learning_rate": 5e-06,
+      "loss": 0.0683,
+      "num_input_tokens_seen": 1017204996,
+      "step": 5928
+    },
+    {
+      "epoch": 67.29178470254958,
+      "loss": 0.0622432604432106,
+      "loss_ce": 1.791580689314287e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 1017204996,
+      "step": 5928
+    },
+    {
+      "epoch": 67.30311614730878,
+      "grad_norm": 3.4525496258784427,
+      "learning_rate": 5e-06,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 1017375868,
+      "step": 5929
+    },
+    {
+      "epoch": 67.30311614730878,
+      "loss": 0.028353175148367882,
+      "loss_ce": 3.286193896201439e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 1017375868,
+      "step": 5929
+    },
+    {
+      "epoch": 67.31444759206799,
+      "grad_norm": 3.3125732237198084,
+      "learning_rate": 5e-06,
+      "loss": 0.052,
+      "num_input_tokens_seen": 1017547372,
+      "step": 5930
+    },
+    {
+      "epoch": 67.31444759206799,
+      "loss": 0.03600083291530609,
+      "loss_ce": 2.0605415556929074e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1017547372,
+      "step": 5930
+    },
+    {
+      "epoch": 67.3257790368272,
+      "grad_norm": 2.6714252256092736,
+      "learning_rate": 5e-06,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 1017719404,
+      "step": 5931
+    },
+    {
+      "epoch": 67.3257790368272,
+      "loss": 0.04466455802321434,
+      "loss_ce": 4.785820055985823e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 1017719404,
+      "step": 5931
+    },
+    {
+      "epoch": 67.33711048158641,
+      "grad_norm": 2.62488654195433,
+      "learning_rate": 5e-06,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 1017889532,
+      "step": 5932
+    },
+    {
+      "epoch": 67.33711048158641,
+      "loss": 0.07380382716655731,
+      "loss_ce": 1.2323328519414645e-05,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 1017889532,
+      "step": 5932
+    },
+    {
+      "epoch": 67.34844192634561,
+      "grad_norm": 2.9472113088538143,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 1018061060,
+      "step": 5933
+    },
+    {
+      "epoch": 67.34844192634561,
+      "loss": 0.02997974306344986,
+      "loss_ce": 2.6739504392025992e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 1018061060,
+      "step": 5933
+    },
+    {
+      "epoch": 67.35977337110482,
+      "grad_norm": 3.2866108667328433,
+      "learning_rate": 5e-06,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 1018233152,
+      "step": 5934
+    },
+    {
+      "epoch": 67.35977337110482,
+      "loss": 0.04820363596081734,
+      "loss_ce": 1.6378886357415468e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 1018233152,
+      "step": 5934
+    },
+    {
+      "epoch": 67.37110481586403,
+      "grad_norm": 3.776423617043664,
+      "learning_rate": 5e-06,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 1018404816,
+      "step": 5935
+    },
+    {
+      "epoch": 67.37110481586403,
+      "loss": 0.032211679965257645,
+      "loss_ce": 1.5637106116628274e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1018404816,
+      "step": 5935
+    },
+    {
+      "epoch": 67.38243626062322,
+      "grad_norm": 3.1408827241364103,
+      "learning_rate": 5e-06,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 1018575060,
+      "step": 5936
+    },
+    {
+      "epoch": 67.38243626062322,
+      "loss": 0.029682811349630356,
+      "loss_ce": 3.498440491966903e-05,
+      "loss_iou": 0.05419921875,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 1018575060,
+      "step": 5936
+    },
+    {
+      "epoch": 67.39376770538243,
+      "grad_norm": 3.1382655961215646,
+      "learning_rate": 5e-06,
+      "loss": 0.0736,
+      "num_input_tokens_seen": 1018746740,
+      "step": 5937
+    },
+    {
+      "epoch": 67.39376770538243,
+      "loss": 0.09658536314964294,
+      "loss_ce": 1.248701119038742e-05,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 1018746740,
+      "step": 5937
+    },
+    {
+      "epoch": 67.40509915014164,
+      "grad_norm": 3.2910981273672024,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 1018918468,
+      "step": 5938
+    },
+    {
+      "epoch": 67.40509915014164,
+      "loss": 0.023494010791182518,
+      "loss_ce": 5.651146057061851e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 1018918468,
+      "step": 5938
+    },
+    {
+      "epoch": 67.41643059490085,
+      "grad_norm": 3.689264062316093,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 1019089508,
+      "step": 5939
+    },
+    {
+      "epoch": 67.41643059490085,
+      "loss": 0.03473558649420738,
+      "loss_ce": 5.235806020209566e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1019089508,
+      "step": 5939
+    },
+    {
+      "epoch": 67.42776203966005,
+      "grad_norm": 3.3120753030258636,
+      "learning_rate": 5e-06,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 1019261244,
+      "step": 5940
+    },
+    {
+      "epoch": 67.42776203966005,
+      "loss": 0.034588318318128586,
+      "loss_ce": 2.7160000172443688e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1019261244,
+      "step": 5940
+    },
+    {
+      "epoch": 67.43909348441926,
+      "grad_norm": 2.014337567173122,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 1019431908,
+      "step": 5941
+    },
+    {
+      "epoch": 67.43909348441926,
+      "loss": 0.08788033574819565,
+      "loss_ce": 2.0221694285282865e-05,
+      "loss_iou": 0.115234375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 1019431908,
+      "step": 5941
+    },
+    {
+      "epoch": 67.45042492917847,
+      "grad_norm": 1.8326181119829843,
+      "learning_rate": 5e-06,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 1019602880,
+      "step": 5942
+    },
+    {
+      "epoch": 67.45042492917847,
+      "loss": 0.05754459649324417,
+      "loss_ce": 3.421810470172204e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 1019602880,
+      "step": 5942
+    },
+    {
+      "epoch": 67.46175637393767,
+      "grad_norm": 2.4943084113623866,
+      "learning_rate": 5e-06,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 1019773856,
+      "step": 5943
+    },
+    {
+      "epoch": 67.46175637393767,
+      "loss": 0.03813718259334564,
+      "loss_ce": 2.835897248587571e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 1019773856,
+      "step": 5943
+    },
+    {
+      "epoch": 67.47308781869688,
+      "grad_norm": 3.3729050981254973,
+      "learning_rate": 5e-06,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 1019945988,
+      "step": 5944
+    },
+    {
+      "epoch": 67.47308781869688,
+      "loss": 0.031304746866226196,
+      "loss_ce": 8.967769645096269e-06,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1019945988,
+      "step": 5944
+    },
+    {
+      "epoch": 67.48441926345609,
+      "grad_norm": 3.6145136190826728,
+      "learning_rate": 5e-06,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 1020117512,
+      "step": 5945
+    },
+    {
+      "epoch": 67.48441926345609,
+      "loss": 0.06070343777537346,
+      "loss_ce": 1.923059426189866e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 1020117512,
+      "step": 5945
+    },
+    {
+      "epoch": 67.4957507082153,
+      "grad_norm": 3.7382029337158165,
+      "learning_rate": 5e-06,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 1020289056,
+      "step": 5946
+    },
+    {
+      "epoch": 67.4957507082153,
+      "loss": 0.08486691117286682,
+      "loss_ce": 1.278760282730218e-05,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 1020289056,
+      "step": 5946
+    },
+    {
+      "epoch": 67.5070821529745,
+      "grad_norm": 3.9162870187893954,
+      "learning_rate": 5e-06,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 1020460748,
+      "step": 5947
+    },
+    {
+      "epoch": 67.5070821529745,
+      "loss": 0.03843928873538971,
+      "loss_ce": 3.2913085306063294e-05,
+      "loss_iou": 0.0546875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1020460748,
+      "step": 5947
+    },
+    {
+      "epoch": 67.51841359773371,
+      "grad_norm": 3.178981981993921,
+      "learning_rate": 5e-06,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 1020629756,
+      "step": 5948
+    },
+    {
+      "epoch": 67.51841359773371,
+      "loss": 0.03202046826481819,
+      "loss_ce": 2.2787029593018815e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1020629756,
+      "step": 5948
+    },
+    {
+      "epoch": 67.52974504249292,
+      "grad_norm": 4.49468163895135,
+      "learning_rate": 5e-06,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 1020801676,
+      "step": 5949
+    },
+    {
+      "epoch": 67.52974504249292,
+      "loss": 0.03027253784239292,
+      "loss_ce": 1.435909689462278e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1020801676,
+      "step": 5949
+    },
+    {
+      "epoch": 67.54107648725213,
+      "grad_norm": 2.870764415016361,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 1020973340,
+      "step": 5950
+    },
+    {
+      "epoch": 67.54107648725213,
+      "loss": 0.04656238853931427,
+      "loss_ce": 2.308426337549463e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 1020973340,
+      "step": 5950
+    },
+    {
+      "epoch": 67.55240793201133,
+      "grad_norm": 2.152731324318681,
+      "learning_rate": 5e-06,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 1021144032,
+      "step": 5951
+    },
+    {
+      "epoch": 67.55240793201133,
+      "loss": 0.09358450025320053,
+      "loss_ce": 1.760596023814287e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 1021144032,
+      "step": 5951
+    },
+    {
+      "epoch": 67.56373937677054,
+      "grad_norm": 2.614658031229655,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 1021315680,
+      "step": 5952
+    },
+    {
+      "epoch": 67.56373937677054,
+      "loss": 0.029144007712602615,
+      "loss_ce": 1.4978529179643374e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 1021315680,
+      "step": 5952
+    },
+    {
+      "epoch": 67.57507082152975,
+      "grad_norm": 2.455817125054252,
+      "learning_rate": 5e-06,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 1021487188,
+      "step": 5953
+    },
+    {
+      "epoch": 67.57507082152975,
+      "loss": 0.04054902493953705,
+      "loss_ce": 6.425943865906447e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1021487188,
+      "step": 5953
+    },
+    {
+      "epoch": 67.58640226628896,
+      "grad_norm": 2.7790801756106007,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 1021657144,
+      "step": 5954
+    },
+    {
+      "epoch": 67.58640226628896,
+      "loss": 0.05249221622943878,
+      "loss_ce": 2.4869255867088214e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 1021657144,
+      "step": 5954
+    },
+    {
+      "epoch": 67.59773371104816,
+      "grad_norm": 3.990036489675684,
+      "learning_rate": 5e-06,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 1021828840,
+      "step": 5955
+    },
+    {
+      "epoch": 67.59773371104816,
+      "loss": 0.11182501912117004,
+      "loss_ce": 0.0006342322449199855,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 1021828840,
+      "step": 5955
+    },
+    {
+      "epoch": 67.60906515580737,
+      "grad_norm": 3.3139858742748847,
+      "learning_rate": 5e-06,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 1022000628,
+      "step": 5956
+    },
+    {
+      "epoch": 67.60906515580737,
+      "loss": 0.024956900626420975,
+      "loss_ce": 2.4039345589699224e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 1022000628,
+      "step": 5956
+    },
+    {
+      "epoch": 67.62039660056658,
+      "grad_norm": 3.325913767312133,
+      "learning_rate": 5e-06,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 1022171756,
+      "step": 5957
+    },
+    {
+      "epoch": 67.62039660056658,
+      "loss": 0.03688191995024681,
+      "loss_ce": 1.6683925423421897e-05,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1022171756,
+      "step": 5957
+    },
+    {
+      "epoch": 67.63172804532577,
+      "grad_norm": 3.9717894907036566,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 1022341876,
+      "step": 5958
+    },
+    {
+      "epoch": 67.63172804532577,
+      "loss": 0.0380752868950367,
+      "loss_ce": 1.9868448362103663e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 1022341876,
+      "step": 5958
+    },
+    {
+      "epoch": 67.64305949008498,
+      "grad_norm": 3.0809299599576145,
+      "learning_rate": 5e-06,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 1022511024,
+      "step": 5959
+    },
+    {
+      "epoch": 67.64305949008498,
+      "loss": 0.03251413628458977,
+      "loss_ce": 1.2914847502543125e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 1022511024,
+      "step": 5959
+    },
+    {
+      "epoch": 67.65439093484419,
+      "grad_norm": 2.9675342226452934,
+      "learning_rate": 5e-06,
+      "loss": 0.037,
+      "num_input_tokens_seen": 1022683032,
+      "step": 5960
+    },
+    {
+      "epoch": 67.65439093484419,
+      "loss": 0.04448346048593521,
+      "loss_ce": 0.011158263310790062,
+      "loss_iou": 0.34375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1022683032,
+      "step": 5960
+    },
+    {
+      "epoch": 67.6657223796034,
+      "grad_norm": 3.3878513421076475,
+      "learning_rate": 5e-06,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 1022854104,
+      "step": 5961
+    },
+    {
+      "epoch": 67.6657223796034,
+      "loss": 0.04200320318341255,
+      "loss_ce": 4.1534469346515834e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1022854104,
+      "step": 5961
+    },
+    {
+      "epoch": 67.6770538243626,
+      "grad_norm": 2.931597839566919,
+      "learning_rate": 5e-06,
+      "loss": 0.062,
+      "num_input_tokens_seen": 1023026068,
+      "step": 5962
+    },
+    {
+      "epoch": 67.6770538243626,
+      "loss": 0.09601771831512451,
+      "loss_ce": 9.412933650310151e-06,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 1023026068,
+      "step": 5962
+    },
+    {
+      "epoch": 67.68838526912181,
+      "grad_norm": 2.9075194371976374,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 1023197708,
+      "step": 5963
+    },
+    {
+      "epoch": 67.68838526912181,
+      "loss": 0.027083203196525574,
+      "loss_ce": 2.173969551222399e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1023197708,
+      "step": 5963
+    },
+    {
+      "epoch": 67.69971671388102,
+      "grad_norm": 3.06052980489186,
+      "learning_rate": 5e-06,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 1023369932,
+      "step": 5964
+    },
+    {
+      "epoch": 67.69971671388102,
+      "loss": 0.047607045620679855,
+      "loss_ce": 3.0140927265165374e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 1023369932,
+      "step": 5964
+    },
+    {
+      "epoch": 67.71104815864022,
+      "grad_norm": 2.233890716038348,
+      "learning_rate": 5e-06,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 1023541852,
+      "step": 5965
+    },
+    {
+      "epoch": 67.71104815864022,
+      "loss": 0.020587986335158348,
+      "loss_ce": 2.676712756510824e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.004119873046875,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 1023541852,
+      "step": 5965
+    },
+    {
+      "epoch": 67.72237960339943,
+      "grad_norm": 2.5856713842759476,
+      "learning_rate": 5e-06,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 1023712040,
+      "step": 5966
+    },
+    {
+      "epoch": 67.72237960339943,
+      "loss": 0.04573599249124527,
+      "loss_ce": 2.066115484922193e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 1023712040,
+      "step": 5966
+    },
+    {
+      "epoch": 67.73371104815864,
+      "grad_norm": 2.9594529508452845,
+      "learning_rate": 5e-06,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 1023883848,
+      "step": 5967
+    },
+    {
+      "epoch": 67.73371104815864,
+      "loss": 0.038973450660705566,
+      "loss_ce": 2.5389812435605563e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 1023883848,
+      "step": 5967
+    },
+    {
+      "epoch": 67.74504249291785,
+      "grad_norm": 4.5825447113779205,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 1024055500,
+      "step": 5968
+    },
+    {
+      "epoch": 67.74504249291785,
+      "loss": 0.028818221762776375,
+      "loss_ce": 3.251616726629436e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 1024055500,
+      "step": 5968
+    },
+    {
+      "epoch": 67.75637393767705,
+      "grad_norm": 3.553929893449383,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 1024226744,
+      "step": 5969
+    },
+    {
+      "epoch": 67.75637393767705,
+      "loss": 0.08989392220973969,
+      "loss_ce": 3.491756797302514e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 1024226744,
+      "step": 5969
+    },
+    {
+      "epoch": 67.76770538243626,
+      "grad_norm": 3.081348734401079,
+      "learning_rate": 5e-06,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 1024398684,
+      "step": 5970
+    },
+    {
+      "epoch": 67.76770538243626,
+      "loss": 0.02868819236755371,
+      "loss_ce": 3.218708297936246e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 1024398684,
+      "step": 5970
+    },
+    {
+      "epoch": 67.77903682719547,
+      "grad_norm": 3.6556474210840806,
+      "learning_rate": 5e-06,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 1024570544,
+      "step": 5971
+    },
+    {
+      "epoch": 67.77903682719547,
+      "loss": 0.03137829899787903,
+      "loss_ce": 2.148860221495852e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1024570544,
+      "step": 5971
+    },
+    {
+      "epoch": 67.79036827195468,
+      "grad_norm": 4.246381103833915,
+      "learning_rate": 5e-06,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 1024742620,
+      "step": 5972
+    },
+    {
+      "epoch": 67.79036827195468,
+      "loss": 0.04452531412243843,
+      "loss_ce": 3.068404839723371e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 1024742620,
+      "step": 5972
+    },
+    {
+      "epoch": 67.80169971671388,
+      "grad_norm": 3.622694740462554,
+      "learning_rate": 5e-06,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 1024914520,
+      "step": 5973
+    },
+    {
+      "epoch": 67.80169971671388,
+      "loss": 0.03686343505978584,
+      "loss_ce": 2.8719499823637307e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1024914520,
+      "step": 5973
+    },
+    {
+      "epoch": 67.81303116147309,
+      "grad_norm": 4.308399468186588,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 1025082984,
+      "step": 5974
+    },
+    {
+      "epoch": 67.81303116147309,
+      "loss": 0.09234784543514252,
+      "loss_ce": 0.0019852956756949425,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 1025082984,
+      "step": 5974
+    },
+    {
+      "epoch": 67.8243626062323,
+      "grad_norm": 3.6874468633881823,
+      "learning_rate": 5e-06,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 1025254168,
+      "step": 5975
+    },
+    {
+      "epoch": 67.8243626062323,
+      "loss": 0.04916127026081085,
+      "loss_ce": 0.00010426143853692338,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 1025254168,
+      "step": 5975
+    },
+    {
+      "epoch": 67.8356940509915,
+      "grad_norm": 3.1187827386990836,
+      "learning_rate": 5e-06,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 1025425440,
+      "step": 5976
+    },
+    {
+      "epoch": 67.8356940509915,
+      "loss": 0.0417618528008461,
+      "loss_ce": 2.9063037800369784e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 1025425440,
+      "step": 5976
+    },
+    {
+      "epoch": 67.84702549575071,
+      "grad_norm": 3.2357237380576946,
+      "learning_rate": 5e-06,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 1025597704,
+      "step": 5977
+    },
+    {
+      "epoch": 67.84702549575071,
+      "loss": 0.02934318408370018,
+      "loss_ce": 3.1050603865878657e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 1025597704,
+      "step": 5977
+    },
+    {
+      "epoch": 67.85835694050992,
+      "grad_norm": 3.711116833290631,
+      "learning_rate": 5e-06,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 1025767168,
+      "step": 5978
+    },
+    {
+      "epoch": 67.85835694050992,
+      "loss": 0.0322192944586277,
+      "loss_ce": 3.85077582905069e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1025767168,
+      "step": 5978
+    },
+    {
+      "epoch": 67.86968838526913,
+      "grad_norm": 3.4783556474564588,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 1025939408,
+      "step": 5979
+    },
+    {
+      "epoch": 67.86968838526913,
+      "loss": 0.13107489049434662,
+      "loss_ce": 1.7151229258161038e-05,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 1025939408,
+      "step": 5979
+    },
+    {
+      "epoch": 67.88101983002832,
+      "grad_norm": 3.548318530409953,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 1026111376,
+      "step": 5980
+    },
+    {
+      "epoch": 67.88101983002832,
+      "loss": 0.03401817008852959,
+      "loss_ce": 2.1587828086921945e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1026111376,
+      "step": 5980
+    },
+    {
+      "epoch": 67.89235127478753,
+      "grad_norm": 3.207230953688128,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 1026282652,
+      "step": 5981
+    },
+    {
+      "epoch": 67.89235127478753,
+      "loss": 0.03093486651778221,
+      "loss_ce": 2.0560566554195248e-05,
+      "loss_iou": 0.671875,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 1026282652,
+      "step": 5981
+    },
+    {
+      "epoch": 67.90368271954674,
+      "grad_norm": 3.64447854110498,
+      "learning_rate": 5e-06,
+      "loss": 0.045,
+      "num_input_tokens_seen": 1026454576,
+      "step": 5982
+    },
+    {
+      "epoch": 67.90368271954674,
+      "loss": 0.02514209970831871,
+      "loss_ce": 2.613234391901642e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 1026454576,
+      "step": 5982
+    },
+    {
+      "epoch": 67.91501416430594,
+      "grad_norm": 3.8475526626672796,
+      "learning_rate": 5e-06,
+      "loss": 0.062,
+      "num_input_tokens_seen": 1026625752,
+      "step": 5983
+    },
+    {
+      "epoch": 67.91501416430594,
+      "loss": 0.09264905005693436,
+      "loss_ce": 0.0014014916960150003,
+      "loss_iou": 0.453125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 1026625752,
+      "step": 5983
+    },
+    {
+      "epoch": 67.92634560906515,
+      "grad_norm": 3.9230716331848154,
+      "learning_rate": 5e-06,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 1026797564,
+      "step": 5984
+    },
+    {
+      "epoch": 67.92634560906515,
+      "loss": 0.04877150431275368,
+      "loss_ce": 0.0004011422861367464,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 1026797564,
+      "step": 5984
+    },
+    {
+      "epoch": 67.93767705382436,
+      "grad_norm": 3.0421690646658455,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 1026969224,
+      "step": 5985
+    },
+    {
+      "epoch": 67.93767705382436,
+      "loss": 0.04318626597523689,
+      "loss_ce": 3.441072112764232e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1026969224,
+      "step": 5985
+    },
+    {
+      "epoch": 67.94900849858357,
+      "grad_norm": 2.6075163387190443,
+      "learning_rate": 5e-06,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 1027141024,
+      "step": 5986
+    },
+    {
+      "epoch": 67.94900849858357,
+      "loss": 0.024318058043718338,
+      "loss_ce": 0.00027020624838769436,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 1027141024,
+      "step": 5986
+    },
+    {
+      "epoch": 67.96033994334277,
+      "grad_norm": 2.949921975660462,
+      "learning_rate": 5e-06,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 1027311160,
+      "step": 5987
+    },
+    {
+      "epoch": 67.96033994334277,
+      "loss": 0.04512227326631546,
+      "loss_ce": 0.0001241056015715003,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 1027311160,
+      "step": 5987
+    },
+    {
+      "epoch": 67.97167138810198,
+      "grad_norm": 3.4021908133242165,
+      "learning_rate": 5e-06,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 1027481832,
+      "step": 5988
+    },
+    {
+      "epoch": 67.97167138810198,
+      "loss": 0.024713270366191864,
+      "loss_ce": 3.980904148193076e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 1027481832,
+      "step": 5988
+    },
+    {
+      "epoch": 67.98300283286119,
+      "grad_norm": 3.6481015501735263,
+      "learning_rate": 5e-06,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 1027653396,
+      "step": 5989
+    },
+    {
+      "epoch": 67.98300283286119,
+      "loss": 0.036867640912532806,
+      "loss_ce": 0.00012447308108676225,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1027653396,
+      "step": 5989
+    },
+    {
+      "epoch": 67.9943342776204,
+      "grad_norm": 3.1972775382174485,
+      "learning_rate": 5e-06,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 1027824828,
+      "step": 5990
+    },
+    {
+      "epoch": 67.9943342776204,
+      "loss": 0.031084056943655014,
+      "loss_ce": 9.345810394734144e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1027824828,
+      "step": 5990
+    },
+    {
+      "epoch": 67.9943342776204,
+      "loss": 0.044801052659749985,
+      "loss_ce": 0.00045901184785179794,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 1027910640,
+      "step": 5990
+    },
+    {
+      "epoch": 68.0056657223796,
+      "grad_norm": 3.3392750182851,
+      "learning_rate": 5e-06,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 1027996508,
+      "step": 5991
+    },
+    {
+      "epoch": 68.0056657223796,
+      "loss": 0.02234051190316677,
+      "loss_ce": 4.742142846225761e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.0223388671875,
+      "num_input_tokens_seen": 1027996508,
+      "step": 5991
+    },
+    {
+      "epoch": 68.01699716713881,
+      "grad_norm": 3.717837503755839,
+      "learning_rate": 5e-06,
+      "loss": 0.059,
+      "num_input_tokens_seen": 1028167196,
+      "step": 5992
+    },
+    {
+      "epoch": 68.01699716713881,
+      "loss": 0.040140435099601746,
+      "loss_ce": 0.005914973560720682,
+      "loss_iou": 0.08642578125,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1028167196,
+      "step": 5992
+    },
+    {
+      "epoch": 68.02832861189802,
+      "grad_norm": 3.76317119270108,
+      "learning_rate": 5e-06,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 1028339244,
+      "step": 5993
+    },
+    {
+      "epoch": 68.02832861189802,
+      "loss": 0.03492102399468422,
+      "loss_ce": 0.0001615022192709148,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1028339244,
+      "step": 5993
+    },
+    {
+      "epoch": 68.03966005665723,
+      "grad_norm": 3.6665597602083944,
+      "learning_rate": 5e-06,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 1028511300,
+      "step": 5994
+    },
+    {
+      "epoch": 68.03966005665723,
+      "loss": 0.03729729726910591,
+      "loss_ce": 0.003819514298811555,
+      "loss_iou": 0.59375,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1028511300,
+      "step": 5994
+    },
+    {
+      "epoch": 68.05099150141643,
+      "grad_norm": 3.3119987573277734,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 1028683484,
+      "step": 5995
+    },
+    {
+      "epoch": 68.05099150141643,
+      "loss": 0.03806358575820923,
+      "loss_ce": 2.3425462131854147e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 1028683484,
+      "step": 5995
+    },
+    {
+      "epoch": 68.06232294617564,
+      "grad_norm": 3.5869846460505164,
+      "learning_rate": 5e-06,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 1028855680,
+      "step": 5996
+    },
+    {
+      "epoch": 68.06232294617564,
+      "loss": 0.058862023055553436,
+      "loss_ce": 6.990783731453121e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 1028855680,
+      "step": 5996
+    },
+    {
+      "epoch": 68.07365439093485,
+      "grad_norm": 4.217078139634235,
+      "learning_rate": 5e-06,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 1029027400,
+      "step": 5997
+    },
+    {
+      "epoch": 68.07365439093485,
+      "loss": 0.06039019301533699,
+      "loss_ce": 2.642503022798337e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 1029027400,
+      "step": 5997
+    },
+    {
+      "epoch": 68.08498583569406,
+      "grad_norm": 3.9664111072443577,
+      "learning_rate": 5e-06,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 1029199416,
+      "step": 5998
+    },
+    {
+      "epoch": 68.08498583569406,
+      "loss": 0.0984577089548111,
+      "loss_ce": 2.325842433492653e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 1029199416,
+      "step": 5998
+    },
+    {
+      "epoch": 68.09631728045326,
+      "grad_norm": 3.3733826022513007,
+      "learning_rate": 5e-06,
+      "loss": 0.031,
+      "num_input_tokens_seen": 1029371352,
+      "step": 5999
+    },
+    {
+      "epoch": 68.09631728045326,
+      "loss": 0.035213347524404526,
+      "loss_ce": 0.0036123963072896004,
+      "loss_iou": 0.46875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 1029371352,
+      "step": 5999
+    },
+    {
+      "epoch": 68.10764872521247,
+      "grad_norm": 3.252815059622613,
+      "learning_rate": 5e-06,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 1029543136,
+      "step": 6000
+    },
+    {
+      "epoch": 68.10764872521247,
+      "eval_seeclick_CIoU": 0.5355914235115051,
+      "eval_seeclick_GIoU": 0.5389807820320129,
+      "eval_seeclick_IoU": 0.5723366141319275,
+      "eval_seeclick_MAE_all": 0.06617512553930283,
+      "eval_seeclick_MAE_h": 0.030604134313762188,
+      "eval_seeclick_MAE_w": 0.10251282155513763,
+      "eval_seeclick_MAE_x": 0.10159862786531448,
+      "eval_seeclick_MAE_y": 0.02998490259051323,
+      "eval_seeclick_NUM_probability": 0.9999902844429016,
+      "eval_seeclick_inside_bbox": 0.9076704680919647,
+      "eval_seeclick_loss": 0.9886494874954224,
+      "eval_seeclick_loss_ce": 0.7358945608139038,
+      "eval_seeclick_loss_iou": 0.53857421875,
+      "eval_seeclick_loss_num": 0.05078125,
+      "eval_seeclick_loss_xval": 0.25390625,
+      "eval_seeclick_runtime": 68.2118,
+      "eval_seeclick_samples_per_second": 0.63,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 1029543136,
+      "step": 6000
+    },
+    {
+      "epoch": 68.10764872521247,
+      "eval_icons_CIoU": 0.7497312128543854,
+      "eval_icons_GIoU": 0.7520346641540527,
+      "eval_icons_IoU": 0.7618784606456757,
+      "eval_icons_MAE_all": 0.03181108087301254,
+      "eval_icons_MAE_h": 0.02700522541999817,
+      "eval_icons_MAE_w": 0.03983544930815697,
+      "eval_icons_MAE_x": 0.031000439543277025,
+      "eval_icons_MAE_y": 0.02940320409834385,
+      "eval_icons_NUM_probability": 0.9996767938137054,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.12723614275455475,
+      "eval_icons_loss_ce": 0.008824684657156467,
+      "eval_icons_loss_iou": 0.5606689453125,
+      "eval_icons_loss_num": 0.02147674560546875,
+      "eval_icons_loss_xval": 0.107391357421875,
+      "eval_icons_runtime": 90.7617,
+      "eval_icons_samples_per_second": 0.551,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 1029543136,
+      "step": 6000
+    },
+    {
+      "epoch": 68.10764872521247,
+      "eval_screenspot_CIoU": 0.650463879108429,
+      "eval_screenspot_GIoU": 0.6525246898333231,
+      "eval_screenspot_IoU": 0.6760209004084269,
+      "eval_screenspot_MAE_all": 0.061041269451379776,
+      "eval_screenspot_MAE_h": 0.033623721450567245,
+      "eval_screenspot_MAE_w": 0.10974500266214211,
+      "eval_screenspot_MAE_x": 0.07187512516975403,
+      "eval_screenspot_MAE_y": 0.02892123783628146,
+      "eval_screenspot_NUM_probability": 0.9999533494313558,
+      "eval_screenspot_inside_bbox": 0.9287499984105428,
+      "eval_screenspot_loss": 0.2678432762622833,
+      "eval_screenspot_loss_ce": 0.015532740236570438,
+      "eval_screenspot_loss_iou": 0.503173828125,
+      "eval_screenspot_loss_num": 0.050409952799479164,
+      "eval_screenspot_loss_xval": 0.2519938151041667,
+      "eval_screenspot_runtime": 143.366,
+      "eval_screenspot_samples_per_second": 0.621,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 1029543136,
+      "step": 6000
+    },
+    {
+      "epoch": 68.10764872521247,
+      "eval_compot_CIoU": 0.8502991795539856,
+      "eval_compot_GIoU": 0.844149112701416,
+      "eval_compot_IoU": 0.8653054535388947,
+      "eval_compot_MAE_all": 0.027687517926096916,
+      "eval_compot_MAE_h": 0.0234012259170413,
+      "eval_compot_MAE_w": 0.03197004646062851,
+      "eval_compot_MAE_x": 0.03141937404870987,
+      "eval_compot_MAE_y": 0.02395942062139511,
+      "eval_compot_NUM_probability": 0.9999739825725555,
+      "eval_compot_inside_bbox": 0.9409722089767456,
+      "eval_compot_loss": 0.08844759315252304,
+      "eval_compot_loss_ce": 1.9227887605666183e-05,
+      "eval_compot_loss_iou": 0.5478515625,
+      "eval_compot_loss_num": 0.0153656005859375,
+      "eval_compot_loss_xval": 0.0768280029296875,
+      "eval_compot_runtime": 81.9061,
+      "eval_compot_samples_per_second": 0.61,
+      "eval_compot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 1029543136,
+      "step": 6000
+    },
+    {
+      "epoch": 68.10764872521247,
+      "eval_custom_ui_MAE_all": 0.019856389611959457,
+      "eval_custom_ui_MAE_x": 0.03132188878953457,
+      "eval_custom_ui_MAE_y": 0.00839089136570692,
+      "eval_custom_ui_NUM_probability": 0.9998889267444611,
+      "eval_custom_ui_loss": 0.21336433291435242,
+      "eval_custom_ui_loss_ce": 0.11450638249516487,
+      "eval_custom_ui_loss_num": 0.019094467163085938,
+      "eval_custom_ui_loss_xval": 0.0953521728515625,
+      "eval_custom_ui_runtime": 59.4084,
+      "eval_custom_ui_samples_per_second": 0.842,
+      "eval_custom_ui_steps_per_second": 0.034,
+      "num_input_tokens_seen": 1029543136,
+      "step": 6000
+    },
+    {
+      "epoch": 68.10764872521247,
+      "loss": 0.2626144289970398,
+      "loss_ce": 0.1441451907157898,
+      "loss_iou": 0.0,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 1029543136,
+      "step": 6000
+    },
+    {
+      "epoch": 68.11898016997168,
+      "grad_norm": 3.5271743295168365,
+      "learning_rate": 5e-06,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 1029713776,
+      "step": 6001
+    },
+    {
+      "epoch": 68.11898016997168,
+      "loss": 0.07258082181215286,
+      "loss_ce": 5.579826029133983e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 1029713776,
+      "step": 6001
+    },
+    {
+      "epoch": 68.13031161473087,
+      "grad_norm": 3.7402497158029813,
+      "learning_rate": 5e-06,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 1029884776,
+      "step": 6002
+    },
+    {
+      "epoch": 68.13031161473087,
+      "loss": 0.05135255306959152,
+      "loss_ce": 3.724159978446551e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 1029884776,
+      "step": 6002
+    },
+    {
+      "epoch": 68.14164305949008,
+      "grad_norm": 4.126340138413504,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 1030055976,
+      "step": 6003
+    },
+    {
+      "epoch": 68.14164305949008,
+      "loss": 0.05936101824045181,
+      "loss_ce": 0.006351982709020376,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 1030055976,
+      "step": 6003
+    },
+    {
+      "epoch": 68.15297450424929,
+      "grad_norm": 3.428165531831982,
+      "learning_rate": 5e-06,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 1030226216,
+      "step": 6004
+    },
+    {
+      "epoch": 68.15297450424929,
+      "loss": 0.05081775784492493,
+      "loss_ce": 0.0006468593492172658,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 1030226216,
+      "step": 6004
+    },
+    {
+      "epoch": 68.1643059490085,
+      "grad_norm": 3.043971150622534,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 1030397028,
+      "step": 6005
+    },
+    {
+      "epoch": 68.1643059490085,
+      "loss": 0.03183009475469589,
+      "loss_ce": 0.0040896194986999035,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 1030397028,
+      "step": 6005
+    },
+    {
+      "epoch": 68.1756373937677,
+      "grad_norm": 3.081223206313506,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 1030568936,
+      "step": 6006
+    },
+    {
+      "epoch": 68.1756373937677,
+      "loss": 0.03006034716963768,
+      "loss_ce": 0.00068717694375664,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 1030568936,
+      "step": 6006
+    },
+    {
+      "epoch": 68.18696883852691,
+      "grad_norm": 3.4644385077962907,
+      "learning_rate": 5e-06,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 1030740296,
+      "step": 6007
+    },
+    {
+      "epoch": 68.18696883852691,
+      "loss": 0.030875112861394882,
+      "loss_ce": 0.001700307708233595,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 1030740296,
+      "step": 6007
+    },
+    {
+      "epoch": 68.19830028328612,
+      "grad_norm": 3.693014341918968,
+      "learning_rate": 5e-06,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 1030911176,
+      "step": 6008
+    },
+    {
+      "epoch": 68.19830028328612,
+      "loss": 0.06895988434553146,
+      "loss_ce": 3.5931345337303355e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 1030911176,
+      "step": 6008
+    },
+    {
+      "epoch": 68.20963172804532,
+      "grad_norm": 3.5069177145608217,
+      "learning_rate": 5e-06,
+      "loss": 0.049,
+      "num_input_tokens_seen": 1031082608,
+      "step": 6009
+    },
+    {
+      "epoch": 68.20963172804532,
+      "loss": 0.05237337946891785,
+      "loss_ce": 0.000707118830177933,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 1031082608,
+      "step": 6009
+    },
+    {
+      "epoch": 68.22096317280453,
+      "grad_norm": 3.771492340230315,
+      "learning_rate": 5e-06,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 1031253728,
+      "step": 6010
+    },
+    {
+      "epoch": 68.22096317280453,
+      "loss": 0.05987983196973801,
+      "loss_ce": 0.0001264110323973,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 1031253728,
+      "step": 6010
+    },
+    {
+      "epoch": 68.23229461756374,
+      "grad_norm": 3.518149160080209,
+      "learning_rate": 5e-06,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 1031425328,
+      "step": 6011
+    },
+    {
+      "epoch": 68.23229461756374,
+      "loss": 0.04251479730010033,
+      "loss_ce": 0.0012397740501910448,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 1031425328,
+      "step": 6011
+    },
+    {
+      "epoch": 68.24362606232295,
+      "grad_norm": 3.318930889226568,
+      "learning_rate": 5e-06,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 1031595404,
+      "step": 6012
+    },
+    {
+      "epoch": 68.24362606232295,
+      "loss": 0.022058993577957153,
+      "loss_ce": 8.633681864012033e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00439453125,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 1031595404,
+      "step": 6012
+    },
+    {
+      "epoch": 68.25495750708215,
+      "grad_norm": 3.3077254795503803,
+      "learning_rate": 5e-06,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 1031767040,
+      "step": 6013
+    },
+    {
+      "epoch": 68.25495750708215,
+      "loss": 0.03557591885328293,
+      "loss_ce": 0.00011449439625721425,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 1031767040,
+      "step": 6013
+    },
+    {
+      "epoch": 68.26628895184136,
+      "grad_norm": 3.3666912902008925,
+      "learning_rate": 5e-06,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 1031938692,
+      "step": 6014
+    },
+    {
+      "epoch": 68.26628895184136,
+      "loss": 0.030247028917074203,
+      "loss_ce": 0.0002329906856175512,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1031938692,
+      "step": 6014
+    },
+    {
+      "epoch": 68.27762039660057,
+      "grad_norm": 3.4014988637377868,
+      "learning_rate": 5e-06,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 1032110584,
+      "step": 6015
+    },
+    {
+      "epoch": 68.27762039660057,
+      "loss": 0.05519992485642433,
+      "loss_ce": 4.7031175199663267e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 1032110584,
+      "step": 6015
+    },
+    {
+      "epoch": 68.28895184135978,
+      "grad_norm": 3.3531783133478994,
+      "learning_rate": 5e-06,
+      "loss": 0.057,
+      "num_input_tokens_seen": 1032281612,
+      "step": 6016
+    },
+    {
+      "epoch": 68.28895184135978,
+      "loss": 0.06379108130931854,
+      "loss_ce": 5.511831113835797e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 1032281612,
+      "step": 6016
+    },
+    {
+      "epoch": 68.30028328611898,
+      "grad_norm": 3.305365183510959,
+      "learning_rate": 5e-06,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 1032453304,
+      "step": 6017
+    },
+    {
+      "epoch": 68.30028328611898,
+      "loss": 0.02550710365176201,
+      "loss_ce": 2.492560633982066e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 1032453304,
+      "step": 6017
+    },
+    {
+      "epoch": 68.31161473087819,
+      "grad_norm": 2.913452668035862,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 1032624204,
+      "step": 6018
+    },
+    {
+      "epoch": 68.31161473087819,
+      "loss": 0.044340990483760834,
+      "loss_ce": 0.00015153891581576318,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 1032624204,
+      "step": 6018
+    },
+    {
+      "epoch": 68.3229461756374,
+      "grad_norm": 3.2352085898190484,
+      "learning_rate": 5e-06,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 1032794584,
+      "step": 6019
+    },
+    {
+      "epoch": 68.3229461756374,
+      "loss": 0.0685746893286705,
+      "loss_ce": 9.324547863798216e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 1032794584,
+      "step": 6019
+    },
+    {
+      "epoch": 68.3342776203966,
+      "grad_norm": 3.6128382610473855,
+      "learning_rate": 5e-06,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 1032966740,
+      "step": 6020
+    },
+    {
+      "epoch": 68.3342776203966,
+      "loss": 0.03370404243469238,
+      "loss_ce": 1.2636179235414602e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1032966740,
+      "step": 6020
+    },
+    {
+      "epoch": 68.34560906515581,
+      "grad_norm": 3.4112636185759064,
+      "learning_rate": 5e-06,
+      "loss": 0.038,
+      "num_input_tokens_seen": 1033138656,
+      "step": 6021
+    },
+    {
+      "epoch": 68.34560906515581,
+      "loss": 0.042105741798877716,
+      "loss_ce": 9.829607733990997e-05,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1033138656,
+      "step": 6021
+    },
+    {
+      "epoch": 68.35694050991502,
+      "grad_norm": 3.4430701279976796,
+      "learning_rate": 5e-06,
+      "loss": 0.045,
+      "num_input_tokens_seen": 1033310520,
+      "step": 6022
+    },
+    {
+      "epoch": 68.35694050991502,
+      "loss": 0.033855095505714417,
+      "loss_ce": 5.687963493983261e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1033310520,
+      "step": 6022
+    },
+    {
+      "epoch": 68.36827195467423,
+      "grad_norm": 3.694661922592041,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 1033481756,
+      "step": 6023
+    },
+    {
+      "epoch": 68.36827195467423,
+      "loss": 0.04631166160106659,
+      "loss_ce": 1.6495927411597222e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 1033481756,
+      "step": 6023
+    },
+    {
+      "epoch": 68.37960339943342,
+      "grad_norm": 3.6295488126088777,
+      "learning_rate": 5e-06,
+      "loss": 0.072,
+      "num_input_tokens_seen": 1033653796,
+      "step": 6024
+    },
+    {
+      "epoch": 68.37960339943342,
+      "loss": 0.09208717197179794,
+      "loss_ce": 1.563353362143971e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 1033653796,
+      "step": 6024
+    },
+    {
+      "epoch": 68.39093484419263,
+      "grad_norm": 4.276016949571233,
+      "learning_rate": 5e-06,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 1033825684,
+      "step": 6025
+    },
+    {
+      "epoch": 68.39093484419263,
+      "loss": 0.028254598379135132,
+      "loss_ce": 4.1098268411587924e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 1033825684,
+      "step": 6025
+    },
+    {
+      "epoch": 68.40226628895184,
+      "grad_norm": 3.3259976906337756,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 1033996880,
+      "step": 6026
+    },
+    {
+      "epoch": 68.40226628895184,
+      "loss": 0.061343878507614136,
+      "loss_ce": 0.0004918276681564748,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 1033996880,
+      "step": 6026
+    },
+    {
+      "epoch": 68.41359773371104,
+      "grad_norm": 3.474480196527888,
+      "learning_rate": 5e-06,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 1034168664,
+      "step": 6027
+    },
+    {
+      "epoch": 68.41359773371104,
+      "loss": 0.041671402752399445,
+      "loss_ce": 7.594195631099865e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 1034168664,
+      "step": 6027
+    },
+    {
+      "epoch": 68.42492917847025,
+      "grad_norm": 3.5426219987918084,
+      "learning_rate": 5e-06,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 1034340060,
+      "step": 6028
+    },
+    {
+      "epoch": 68.42492917847025,
+      "loss": 0.02955089882016182,
+      "loss_ce": 2.514255902497098e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 1034340060,
+      "step": 6028
+    },
+    {
+      "epoch": 68.43626062322946,
+      "grad_norm": 3.1110351160121583,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 1034510928,
+      "step": 6029
+    },
+    {
+      "epoch": 68.43626062322946,
+      "loss": 0.026297051459550858,
+      "loss_ce": 2.9046279450994916e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 1034510928,
+      "step": 6029
+    },
+    {
+      "epoch": 68.44759206798867,
+      "grad_norm": 3.886738075438796,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 1034681744,
+      "step": 6030
+    },
+    {
+      "epoch": 68.44759206798867,
+      "loss": 0.0462028905749321,
+      "loss_ce": 5.268330642138608e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 1034681744,
+      "step": 6030
+    },
+    {
+      "epoch": 68.45892351274787,
+      "grad_norm": 3.5529192312233326,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 1034852356,
+      "step": 6031
+    },
+    {
+      "epoch": 68.45892351274787,
+      "loss": 0.039018433541059494,
+      "loss_ce": 0.0005968009936623275,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1034852356,
+      "step": 6031
+    },
+    {
+      "epoch": 68.47025495750708,
+      "grad_norm": 2.371103549395048,
+      "learning_rate": 5e-06,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 1035024596,
+      "step": 6032
+    },
+    {
+      "epoch": 68.47025495750708,
+      "loss": 0.033117782324552536,
+      "loss_ce": 2.146774932043627e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1035024596,
+      "step": 6032
+    },
+    {
+      "epoch": 68.48158640226629,
+      "grad_norm": 2.7509865723152487,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 1035196280,
+      "step": 6033
+    },
+    {
+      "epoch": 68.48158640226629,
+      "loss": 0.04057038575410843,
+      "loss_ce": 1.2522734323283657e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1035196280,
+      "step": 6033
+    },
+    {
+      "epoch": 68.4929178470255,
+      "grad_norm": 2.7816715261384037,
+      "learning_rate": 5e-06,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 1035365140,
+      "step": 6034
+    },
+    {
+      "epoch": 68.4929178470255,
+      "loss": 0.028911568224430084,
+      "loss_ce": 0.0002479329996276647,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 1035365140,
+      "step": 6034
+    },
+    {
+      "epoch": 68.5042492917847,
+      "grad_norm": 2.8451632858893956,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 1035536688,
+      "step": 6035
+    },
+    {
+      "epoch": 68.5042492917847,
+      "loss": 0.04116325080394745,
+      "loss_ce": 5.607565981335938e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1035536688,
+      "step": 6035
+    },
+    {
+      "epoch": 68.51558073654391,
+      "grad_norm": 4.862706201586702,
+      "learning_rate": 5e-06,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 1035708404,
+      "step": 6036
+    },
+    {
+      "epoch": 68.51558073654391,
+      "loss": 0.031239386647939682,
+      "loss_ce": 0.004773015156388283,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 1035708404,
+      "step": 6036
+    },
+    {
+      "epoch": 68.52691218130312,
+      "grad_norm": 3.2819575596366137,
+      "learning_rate": 5e-06,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 1035877140,
+      "step": 6037
+    },
+    {
+      "epoch": 68.52691218130312,
+      "loss": 0.08378823101520538,
+      "loss_ce": 0.00030739742214791477,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 1035877140,
+      "step": 6037
+    },
+    {
+      "epoch": 68.53824362606233,
+      "grad_norm": 2.968974868863053,
+      "learning_rate": 5e-06,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 1036047680,
+      "step": 6038
+    },
+    {
+      "epoch": 68.53824362606233,
+      "loss": 0.03451596945524216,
+      "loss_ce": 9.214033343596384e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 1036047680,
+      "step": 6038
+    },
+    {
+      "epoch": 68.54957507082153,
+      "grad_norm": 3.528375599277384,
+      "learning_rate": 5e-06,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 1036219264,
+      "step": 6039
+    },
+    {
+      "epoch": 68.54957507082153,
+      "loss": 0.026953136548399925,
+      "loss_ce": 3.663220559246838e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 1036219264,
+      "step": 6039
+    },
+    {
+      "epoch": 68.56090651558074,
+      "grad_norm": 3.956795338038998,
+      "learning_rate": 5e-06,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 1036391036,
+      "step": 6040
+    },
+    {
+      "epoch": 68.56090651558074,
+      "loss": 0.03081545978784561,
+      "loss_ce": 9.95180889731273e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 1036391036,
+      "step": 6040
+    },
+    {
+      "epoch": 68.57223796033995,
+      "grad_norm": 4.597195948699677,
+      "learning_rate": 5e-06,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 1036562056,
+      "step": 6041
+    },
+    {
+      "epoch": 68.57223796033995,
+      "loss": 0.042367078363895416,
+      "loss_ce": 8.497034286847338e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 1036562056,
+      "step": 6041
+    },
+    {
+      "epoch": 68.58356940509915,
+      "grad_norm": 4.148584366220655,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 1036732424,
+      "step": 6042
+    },
+    {
+      "epoch": 68.58356940509915,
+      "loss": 0.036121152341365814,
+      "loss_ce": 1.885701931314543e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 1036732424,
+      "step": 6042
+    },
+    {
+      "epoch": 68.59490084985836,
+      "grad_norm": 3.700407683064666,
+      "learning_rate": 5e-06,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 1036900404,
+      "step": 6043
+    },
+    {
+      "epoch": 68.59490084985836,
+      "loss": 0.04376261681318283,
+      "loss_ce": 1.5666475519537926e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 1036900404,
+      "step": 6043
+    },
+    {
+      "epoch": 68.60623229461757,
+      "grad_norm": 3.6403288671672596,
+      "learning_rate": 5e-06,
+      "loss": 0.05,
+      "num_input_tokens_seen": 1037070436,
+      "step": 6044
+    },
+    {
+      "epoch": 68.60623229461757,
+      "loss": 0.04341422766447067,
+      "loss_ce": 1.8233891751151532e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 1037070436,
+      "step": 6044
+    },
+    {
+      "epoch": 68.61756373937678,
+      "grad_norm": 3.518003676297302,
+      "learning_rate": 5e-06,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 1037240848,
+      "step": 6045
+    },
+    {
+      "epoch": 68.61756373937678,
+      "loss": 0.0429302416741848,
+      "loss_ce": 2.2526497559738345e-05,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 1037240848,
+      "step": 6045
+    },
+    {
+      "epoch": 68.62889518413597,
+      "grad_norm": 3.0585796312948714,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 1037412604,
+      "step": 6046
+    },
+    {
+      "epoch": 68.62889518413597,
+      "loss": 0.03132706135511398,
+      "loss_ce": 3.128579191979952e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1037412604,
+      "step": 6046
+    },
+    {
+      "epoch": 68.64022662889518,
+      "grad_norm": 3.275455915419227,
+      "learning_rate": 5e-06,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 1037584524,
+      "step": 6047
+    },
+    {
+      "epoch": 68.64022662889518,
+      "loss": 0.034866221249103546,
+      "loss_ce": 0.00024402780400123447,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1037584524,
+      "step": 6047
+    },
+    {
+      "epoch": 68.65155807365439,
+      "grad_norm": 3.4777212867909446,
+      "learning_rate": 5e-06,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 1037756352,
+      "step": 6048
+    },
+    {
+      "epoch": 68.65155807365439,
+      "loss": 0.04108915105462074,
+      "loss_ce": 2.7749953005695716e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1037756352,
+      "step": 6048
+    },
+    {
+      "epoch": 68.66288951841359,
+      "grad_norm": 3.7213501227500942,
+      "learning_rate": 5e-06,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 1037928264,
+      "step": 6049
+    },
+    {
+      "epoch": 68.66288951841359,
+      "loss": 0.030276209115982056,
+      "loss_ce": 3.3289456041529775e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1037928264,
+      "step": 6049
+    },
+    {
+      "epoch": 68.6742209631728,
+      "grad_norm": 3.6707126448892713,
+      "learning_rate": 5e-06,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 1038099788,
+      "step": 6050
+    },
+    {
+      "epoch": 68.6742209631728,
+      "loss": 0.04867839068174362,
+      "loss_ce": 7.914793968666345e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 1038099788,
+      "step": 6050
+    },
+    {
+      "epoch": 68.68555240793201,
+      "grad_norm": 2.854682568962777,
+      "learning_rate": 5e-06,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 1038271472,
+      "step": 6051
+    },
+    {
+      "epoch": 68.68555240793201,
+      "loss": 0.04150331765413284,
+      "loss_ce": 0.0005029530730098486,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1038271472,
+      "step": 6051
+    },
+    {
+      "epoch": 68.69688385269122,
+      "grad_norm": 2.9646780319420847,
+      "learning_rate": 5e-06,
+      "loss": 0.0373,
+      "num_input_tokens_seen": 1038443476,
+      "step": 6052
+    },
+    {
+      "epoch": 68.69688385269122,
+      "loss": 0.043342575430870056,
+      "loss_ce": 2.2872329282108694e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1038443476,
+      "step": 6052
+    },
+    {
+      "epoch": 68.70821529745042,
+      "grad_norm": 3.3077502423371548,
+      "learning_rate": 5e-06,
+      "loss": 0.043,
+      "num_input_tokens_seen": 1038613584,
+      "step": 6053
+    },
+    {
+      "epoch": 68.70821529745042,
+      "loss": 0.038875751197338104,
+      "loss_ce": 4.2135252442676574e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 1038613584,
+      "step": 6053
+    },
+    {
+      "epoch": 68.71954674220963,
+      "grad_norm": 3.131680385838785,
+      "learning_rate": 5e-06,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 1038785480,
+      "step": 6054
+    },
+    {
+      "epoch": 68.71954674220963,
+      "loss": 0.030011728405952454,
+      "loss_ce": 2.0577959730871953e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1038785480,
+      "step": 6054
+    },
+    {
+      "epoch": 68.73087818696884,
+      "grad_norm": 3.1863758737240397,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 1038956232,
+      "step": 6055
+    },
+    {
+      "epoch": 68.73087818696884,
+      "loss": 0.03920143470168114,
+      "loss_ce": 3.2123578421305865e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 1038956232,
+      "step": 6055
+    },
+    {
+      "epoch": 68.74220963172804,
+      "grad_norm": 3.3607580232360257,
+      "learning_rate": 5e-06,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 1039128224,
+      "step": 6056
+    },
+    {
+      "epoch": 68.74220963172804,
+      "loss": 0.06432376801967621,
+      "loss_ce": 4.612480552168563e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 1039128224,
+      "step": 6056
+    },
+    {
+      "epoch": 68.75354107648725,
+      "grad_norm": 3.7702577525963137,
+      "learning_rate": 5e-06,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 1039299376,
+      "step": 6057
+    },
+    {
+      "epoch": 68.75354107648725,
+      "loss": 0.03325468301773071,
+      "loss_ce": 2.1041156287537888e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1039299376,
+      "step": 6057
+    },
+    {
+      "epoch": 68.76487252124646,
+      "grad_norm": 3.1813173480590247,
+      "learning_rate": 5e-06,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 1039470276,
+      "step": 6058
+    },
+    {
+      "epoch": 68.76487252124646,
+      "loss": 0.0703522115945816,
+      "loss_ce": 0.0001770367380231619,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 1039470276,
+      "step": 6058
+    },
+    {
+      "epoch": 68.77620396600567,
+      "grad_norm": 2.77467348214069,
+      "learning_rate": 5e-06,
+      "loss": 0.0301,
+      "num_input_tokens_seen": 1039641980,
+      "step": 6059
+    },
+    {
+      "epoch": 68.77620396600567,
+      "loss": 0.030268246307969093,
+      "loss_ce": 2.5325391106889583e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1039641980,
+      "step": 6059
+    },
+    {
+      "epoch": 68.78753541076487,
+      "grad_norm": 2.8478234165742937,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 1039813768,
+      "step": 6060
+    },
+    {
+      "epoch": 68.78753541076487,
+      "loss": 0.06594222784042358,
+      "loss_ce": 7.003456266829744e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 1039813768,
+      "step": 6060
+    },
+    {
+      "epoch": 68.79886685552408,
+      "grad_norm": 3.4791135533331805,
+      "learning_rate": 5e-06,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 1039985460,
+      "step": 6061
+    },
+    {
+      "epoch": 68.79886685552408,
+      "loss": 0.05560876056551933,
+      "loss_ce": 8.202873868867755e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 1039985460,
+      "step": 6061
+    },
+    {
+      "epoch": 68.81019830028329,
+      "grad_norm": 3.430999117936259,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 1040156356,
+      "step": 6062
+    },
+    {
+      "epoch": 68.81019830028329,
+      "loss": 0.070872962474823,
+      "loss_ce": 0.0006672763847745955,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 1040156356,
+      "step": 6062
+    },
+    {
+      "epoch": 68.8215297450425,
+      "grad_norm": 3.3756290846917683,
+      "learning_rate": 5e-06,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 1040328108,
+      "step": 6063
+    },
+    {
+      "epoch": 68.8215297450425,
+      "loss": 0.03579780459403992,
+      "loss_ce": 1.594540663063526e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1040328108,
+      "step": 6063
+    },
+    {
+      "epoch": 68.8328611898017,
+      "grad_norm": 3.1679235989591086,
+      "learning_rate": 5e-06,
+      "loss": 0.071,
+      "num_input_tokens_seen": 1040499692,
+      "step": 6064
+    },
+    {
+      "epoch": 68.8328611898017,
+      "loss": 0.0366927795112133,
+      "loss_ce": 2.5910276235663332e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1040499692,
+      "step": 6064
+    },
+    {
+      "epoch": 68.84419263456091,
+      "grad_norm": 4.155807624272627,
+      "learning_rate": 5e-06,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 1040670536,
+      "step": 6065
+    },
+    {
+      "epoch": 68.84419263456091,
+      "loss": 0.03323636204004288,
+      "loss_ce": 4.849334800383076e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1040670536,
+      "step": 6065
+    },
+    {
+      "epoch": 68.85552407932012,
+      "grad_norm": 3.208425084220272,
+      "learning_rate": 5e-06,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 1040841752,
+      "step": 6066
+    },
+    {
+      "epoch": 68.85552407932012,
+      "loss": 0.1210227832198143,
+      "loss_ce": 3.584111254895106e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 1040841752,
+      "step": 6066
+    },
+    {
+      "epoch": 68.86685552407933,
+      "grad_norm": 4.031176393493782,
+      "learning_rate": 5e-06,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 1041013608,
+      "step": 6067
+    },
+    {
+      "epoch": 68.86685552407933,
+      "loss": 0.03931416571140289,
+      "loss_ce": 0.0010451258858665824,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1041013608,
+      "step": 6067
+    },
+    {
+      "epoch": 68.87818696883852,
+      "grad_norm": 3.8254088541235234,
+      "learning_rate": 5e-06,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 1041183664,
+      "step": 6068
+    },
+    {
+      "epoch": 68.87818696883852,
+      "loss": 0.042976222932338715,
+      "loss_ce": 0.0006788604077883065,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 1041183664,
+      "step": 6068
+    },
+    {
+      "epoch": 68.88951841359773,
+      "grad_norm": 3.5545591987673903,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 1041355312,
+      "step": 6069
+    },
+    {
+      "epoch": 68.88951841359773,
+      "loss": 0.07258428633213043,
+      "loss_ce": 2.874880374292843e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 1041355312,
+      "step": 6069
+    },
+    {
+      "epoch": 68.90084985835693,
+      "grad_norm": 3.7588090899489215,
+      "learning_rate": 5e-06,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 1041526628,
+      "step": 6070
+    },
+    {
+      "epoch": 68.90084985835693,
+      "loss": 0.08877602219581604,
+      "loss_ce": 0.0004734083195216954,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 1041526628,
+      "step": 6070
+    },
+    {
+      "epoch": 68.91218130311614,
+      "grad_norm": 2.922613969510062,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 1041696912,
+      "step": 6071
+    },
+    {
+      "epoch": 68.91218130311614,
+      "loss": 0.030344445258378983,
+      "loss_ce": 0.0001167829759651795,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1041696912,
+      "step": 6071
+    },
+    {
+      "epoch": 68.92351274787535,
+      "grad_norm": 2.7330679061523746,
+      "learning_rate": 5e-06,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 1041868040,
+      "step": 6072
+    },
+    {
+      "epoch": 68.92351274787535,
+      "loss": 0.02817266434431076,
+      "loss_ce": 0.00018041461589746177,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 1041868040,
+      "step": 6072
+    },
+    {
+      "epoch": 68.93484419263456,
+      "grad_norm": 3.009534152062361,
+      "learning_rate": 5e-06,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 1042039836,
+      "step": 6073
+    },
+    {
+      "epoch": 68.93484419263456,
+      "loss": 0.023865772411227226,
+      "loss_ce": 7.732024823781103e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 1042039836,
+      "step": 6073
+    },
+    {
+      "epoch": 68.94617563739376,
+      "grad_norm": 3.2542066267616097,
+      "learning_rate": 5e-06,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 1042211748,
+      "step": 6074
+    },
+    {
+      "epoch": 68.94617563739376,
+      "loss": 0.09823054820299149,
+      "loss_ce": 7.075207395246252e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 1042211748,
+      "step": 6074
+    },
+    {
+      "epoch": 68.95750708215297,
+      "grad_norm": 2.7210391719931013,
+      "learning_rate": 5e-06,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 1042382716,
+      "step": 6075
+    },
+    {
+      "epoch": 68.95750708215297,
+      "loss": 0.02861340343952179,
+      "loss_ce": 4.8950609198072925e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 1042382716,
+      "step": 6075
+    },
+    {
+      "epoch": 68.96883852691218,
+      "grad_norm": 3.784531080526908,
+      "learning_rate": 5e-06,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 1042553516,
+      "step": 6076
+    },
+    {
+      "epoch": 68.96883852691218,
+      "loss": 0.03051602467894554,
+      "loss_ce": 9.000018326332793e-05,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 1042553516,
+      "step": 6076
+    },
+    {
+      "epoch": 68.98016997167139,
+      "grad_norm": 3.4738884577869746,
+      "learning_rate": 5e-06,
+      "loss": 0.031,
+      "num_input_tokens_seen": 1042724344,
+      "step": 6077
+    },
+    {
+      "epoch": 68.98016997167139,
+      "loss": 0.03250674903392792,
+      "loss_ce": 6.656155892414972e-05,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 1042724344,
+      "step": 6077
+    },
+    {
+      "epoch": 68.9915014164306,
+      "grad_norm": 3.6709092426138783,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 1042896016,
+      "step": 6078
+    },
+    {
+      "epoch": 68.9915014164306,
+      "loss": 0.032034341245889664,
+      "loss_ce": 2.1402305719675496e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1042896016,
+      "step": 6078
+    },
+    {
+      "epoch": 68.9915014164306,
+      "loss": 0.04184401407837868,
+      "loss_ce": 8.070863987086341e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 1043023892,
+      "step": 6078
+    },
+    {
+      "epoch": 69.0028328611898,
+      "grad_norm": 3.566578336891107,
+      "learning_rate": 5e-06,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 1043067080,
+      "step": 6079
+    },
+    {
+      "epoch": 69.0028328611898,
+      "loss": 0.04661869630217552,
+      "loss_ce": 1.8353712221141905e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 1043067080,
+      "step": 6079
+    },
+    {
+      "epoch": 69.01416430594901,
+      "grad_norm": 2.9009560791276146,
+      "learning_rate": 5e-06,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 1043238072,
+      "step": 6080
+    },
+    {
+      "epoch": 69.01416430594901,
+      "loss": 0.027608666568994522,
+      "loss_ce": 2.8403321266523562e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 1043238072,
+      "step": 6080
+    },
+    {
+      "epoch": 69.02549575070822,
+      "grad_norm": 2.8979708147926844,
+      "learning_rate": 5e-06,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 1043408260,
+      "step": 6081
+    },
+    {
+      "epoch": 69.02549575070822,
+      "loss": 0.023163681849837303,
+      "loss_ce": 1.6099183994811028e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 1043408260,
+      "step": 6081
+    },
+    {
+      "epoch": 69.03682719546742,
+      "grad_norm": 3.3045306706829276,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 1043576916,
+      "step": 6082
+    },
+    {
+      "epoch": 69.03682719546742,
+      "loss": 0.036885086447000504,
+      "loss_ce": 3.5111792385578156e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1043576916,
+      "step": 6082
+    },
+    {
+      "epoch": 69.04815864022663,
+      "grad_norm": 4.533060458499613,
+      "learning_rate": 5e-06,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 1043747496,
+      "step": 6083
+    },
+    {
+      "epoch": 69.04815864022663,
+      "loss": 0.07012467086315155,
+      "loss_ce": 2.579538158897776e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 1043747496,
+      "step": 6083
+    },
+    {
+      "epoch": 69.05949008498584,
+      "grad_norm": 2.925154068098337,
+      "learning_rate": 5e-06,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 1043918968,
+      "step": 6084
+    },
+    {
+      "epoch": 69.05949008498584,
+      "loss": 0.032461017370224,
+      "loss_ce": 0.0001123846450354904,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1043918968,
+      "step": 6084
+    },
+    {
+      "epoch": 69.07082152974505,
+      "grad_norm": 2.4533193899115484,
+      "learning_rate": 5e-06,
+      "loss": 0.032,
+      "num_input_tokens_seen": 1044090704,
+      "step": 6085
+    },
+    {
+      "epoch": 69.07082152974505,
+      "loss": 0.03167867660522461,
+      "loss_ce": 0.00013875971490051597,
+      "loss_iou": 0.328125,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 1044090704,
+      "step": 6085
+    },
+    {
+      "epoch": 69.08215297450425,
+      "grad_norm": 2.782693400322336,
+      "learning_rate": 5e-06,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 1044261496,
+      "step": 6086
+    },
+    {
+      "epoch": 69.08215297450425,
+      "loss": 0.02466420829296112,
+      "loss_ce": 3.652416489785537e-05,
+      "loss_iou": 0.125,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 1044261496,
+      "step": 6086
+    },
+    {
+      "epoch": 69.09348441926346,
+      "grad_norm": 2.4640254531508496,
+      "learning_rate": 5e-06,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 1044433476,
+      "step": 6087
+    },
+    {
+      "epoch": 69.09348441926346,
+      "loss": 0.04142580181360245,
+      "loss_ce": 1.3446344382828102e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 1044433476,
+      "step": 6087
+    },
+    {
+      "epoch": 69.10481586402267,
+      "grad_norm": 3.8157631205734632,
+      "learning_rate": 5e-06,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 1044605324,
+      "step": 6088
+    },
+    {
+      "epoch": 69.10481586402267,
+      "loss": 0.04222281277179718,
+      "loss_ce": 1.7004045730573125e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 1044605324,
+      "step": 6088
+    },
+    {
+      "epoch": 69.11614730878188,
+      "grad_norm": 3.564928900768224,
+      "learning_rate": 5e-06,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 1044776076,
+      "step": 6089
+    },
+    {
+      "epoch": 69.11614730878188,
+      "loss": 0.031930938363075256,
+      "loss_ce": 1.7182930605486035e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1044776076,
+      "step": 6089
+    },
+    {
+      "epoch": 69.12747875354107,
+      "grad_norm": 3.1182661273176424,
+      "learning_rate": 5e-06,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 1044947544,
+      "step": 6090
+    },
+    {
+      "epoch": 69.12747875354107,
+      "loss": 0.03167203441262245,
+      "loss_ce": 4.8195892304647714e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 1044947544,
+      "step": 6090
+    },
+    {
+      "epoch": 69.13881019830028,
+      "grad_norm": 3.1791290479697523,
+      "learning_rate": 5e-06,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 1045119312,
+      "step": 6091
+    },
+    {
+      "epoch": 69.13881019830028,
+      "loss": 0.02324497140944004,
+      "loss_ce": 2.109505658154376e-05,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 1045119312,
+      "step": 6091
+    },
+    {
+      "epoch": 69.15014164305948,
+      "grad_norm": 3.860080170310789,
+      "learning_rate": 5e-06,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 1045288128,
+      "step": 6092
+    },
+    {
+      "epoch": 69.15014164305948,
+      "loss": 0.10384989529848099,
+      "loss_ce": 1.3841778127243742e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 1045288128,
+      "step": 6092
+    },
+    {
+      "epoch": 69.16147308781869,
+      "grad_norm": 4.252370975724673,
+      "learning_rate": 5e-06,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 1045458920,
+      "step": 6093
+    },
+    {
+      "epoch": 69.16147308781869,
+      "loss": 0.059459540992975235,
+      "loss_ce": 2.6554151190794073e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 1045458920,
+      "step": 6093
+    },
+    {
+      "epoch": 69.1728045325779,
+      "grad_norm": 10.401991581393569,
+      "learning_rate": 5e-06,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 1045630384,
+      "step": 6094
+    },
+    {
+      "epoch": 69.1728045325779,
+      "loss": 0.041744448244571686,
+      "loss_ce": 2.691516056074761e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 1045630384,
+      "step": 6094
+    },
+    {
+      "epoch": 69.1841359773371,
+      "grad_norm": 3.379250977751393,
+      "learning_rate": 5e-06,
+      "loss": 0.039,
+      "num_input_tokens_seen": 1045802072,
+      "step": 6095
+    },
+    {
+      "epoch": 69.1841359773371,
+      "loss": 0.03261017054319382,
+      "loss_ce": 6.317358202068135e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 1045802072,
+      "step": 6095
+    },
+    {
+      "epoch": 69.19546742209631,
+      "grad_norm": 2.907778734033868,
+      "learning_rate": 5e-06,
+      "loss": 0.039,
+      "num_input_tokens_seen": 1045974228,
+      "step": 6096
+    },
+    {
+      "epoch": 69.19546742209631,
+      "loss": 0.03515955060720444,
+      "loss_ce": 2.6189010895905085e-05,
+      "loss_iou": 0.09228515625,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1045974228,
+      "step": 6096
+    },
+    {
+      "epoch": 69.20679886685552,
+      "grad_norm": 2.401889926366466,
+      "learning_rate": 5e-06,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 1046145732,
+      "step": 6097
+    },
+    {
+      "epoch": 69.20679886685552,
+      "loss": 0.03377799689769745,
+      "loss_ce": 1.0294672392774373e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1046145732,
+      "step": 6097
+    },
+    {
+      "epoch": 69.21813031161473,
+      "grad_norm": 2.516323955612636,
+      "learning_rate": 5e-06,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 1046315712,
+      "step": 6098
+    },
+    {
+      "epoch": 69.21813031161473,
+      "loss": 0.1046159565448761,
+      "loss_ce": 6.273268081713468e-05,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 1046315712,
+      "step": 6098
+    },
+    {
+      "epoch": 69.22946175637394,
+      "grad_norm": 2.8023313680286677,
+      "learning_rate": 5e-06,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 1046487624,
+      "step": 6099
+    },
+    {
+      "epoch": 69.22946175637394,
+      "loss": 0.06589950621128082,
+      "loss_ce": 4.2571951780701056e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 1046487624,
+      "step": 6099
+    },
+    {
+      "epoch": 69.24079320113314,
+      "grad_norm": 3.1140863974854156,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 1046658000,
+      "step": 6100
+    },
+    {
+      "epoch": 69.24079320113314,
+      "loss": 0.026163093745708466,
+      "loss_ce": 2.4787481379462406e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 1046658000,
+      "step": 6100
+    },
+    {
+      "epoch": 69.25212464589235,
+      "grad_norm": 3.690686521105264,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 1046829060,
+      "step": 6101
+    },
+    {
+      "epoch": 69.25212464589235,
+      "loss": 0.09821905195713043,
+      "loss_ce": 2.874925485230051e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 1046829060,
+      "step": 6101
+    },
+    {
+      "epoch": 69.26345609065156,
+      "grad_norm": 3.663949545387579,
+      "learning_rate": 5e-06,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 1047000952,
+      "step": 6102
+    },
+    {
+      "epoch": 69.26345609065156,
+      "loss": 0.03768531233072281,
+      "loss_ce": 2.6622426958056167e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 1047000952,
+      "step": 6102
+    },
+    {
+      "epoch": 69.27478753541077,
+      "grad_norm": 3.480404452511672,
+      "learning_rate": 5e-06,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 1047172624,
+      "step": 6103
+    },
+    {
+      "epoch": 69.27478753541077,
+      "loss": 0.031147504225373268,
+      "loss_ce": 0.00017216356354765594,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1047172624,
+      "step": 6103
+    },
+    {
+      "epoch": 69.28611898016997,
+      "grad_norm": 3.8079999909703814,
+      "learning_rate": 5e-06,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 1047344452,
+      "step": 6104
+    },
+    {
+      "epoch": 69.28611898016997,
+      "loss": 0.052981358021497726,
+      "loss_ce": 4.861808702116832e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 1047344452,
+      "step": 6104
+    },
+    {
+      "epoch": 69.29745042492918,
+      "grad_norm": 3.725031553337833,
+      "learning_rate": 5e-06,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 1047516296,
+      "step": 6105
+    },
+    {
+      "epoch": 69.29745042492918,
+      "loss": 0.06311427056789398,
+      "loss_ce": 0.00015650886052753776,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 1047516296,
+      "step": 6105
+    },
+    {
+      "epoch": 69.30878186968839,
+      "grad_norm": 3.6927872218784086,
+      "learning_rate": 5e-06,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 1047687796,
+      "step": 6106
+    },
+    {
+      "epoch": 69.30878186968839,
+      "loss": 0.054484352469444275,
+      "loss_ce": 4.0995531890075654e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 1047687796,
+      "step": 6106
+    },
+    {
+      "epoch": 69.3201133144476,
+      "grad_norm": 5.54216898432565,
+      "learning_rate": 5e-06,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 1047857392,
+      "step": 6107
+    },
+    {
+      "epoch": 69.3201133144476,
+      "loss": 0.07165655493736267,
+      "loss_ce": 0.0010846630902960896,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 1047857392,
+      "step": 6107
+    },
+    {
+      "epoch": 69.3314447592068,
+      "grad_norm": 2.7671178100971905,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 1048028528,
+      "step": 6108
+    },
+    {
+      "epoch": 69.3314447592068,
+      "loss": 0.03650432825088501,
+      "loss_ce": 1.2933372090628836e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 1048028528,
+      "step": 6108
+    },
+    {
+      "epoch": 69.34277620396601,
+      "grad_norm": 3.26740394776147,
+      "learning_rate": 5e-06,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 1048199580,
+      "step": 6109
+    },
+    {
+      "epoch": 69.34277620396601,
+      "loss": 0.040207166224718094,
+      "loss_ce": 3.0773429898545146e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 1048199580,
+      "step": 6109
+    },
+    {
+      "epoch": 69.35410764872522,
+      "grad_norm": 3.3191409920318256,
+      "learning_rate": 5e-06,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 1048370948,
+      "step": 6110
+    },
+    {
+      "epoch": 69.35410764872522,
+      "loss": 0.03141346946358681,
+      "loss_ce": 4.139923112234101e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1048370948,
+      "step": 6110
+    },
+    {
+      "epoch": 69.36543909348443,
+      "grad_norm": 3.4791550491549947,
+      "learning_rate": 5e-06,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 1048542868,
+      "step": 6111
+    },
+    {
+      "epoch": 69.36543909348443,
+      "loss": 0.02632526308298111,
+      "loss_ce": 2.674159622984007e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 1048542868,
+      "step": 6111
+    },
+    {
+      "epoch": 69.37677053824362,
+      "grad_norm": 4.522326384274253,
+      "learning_rate": 5e-06,
+      "loss": 0.055,
+      "num_input_tokens_seen": 1048714876,
+      "step": 6112
+    },
+    {
+      "epoch": 69.37677053824362,
+      "loss": 0.09521997720003128,
+      "loss_ce": 0.009602908976376057,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 1048714876,
+      "step": 6112
+    },
+    {
+      "epoch": 69.38810198300283,
+      "grad_norm": 2.906494423595446,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 1048886344,
+      "step": 6113
+    },
+    {
+      "epoch": 69.38810198300283,
+      "loss": 0.03917405381798744,
+      "loss_ce": 8.103667641989887e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 1048886344,
+      "step": 6113
+    },
+    {
+      "epoch": 69.39943342776203,
+      "grad_norm": 2.1402710245133654,
+      "learning_rate": 5e-06,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 1049058144,
+      "step": 6114
+    },
+    {
+      "epoch": 69.39943342776203,
+      "loss": 0.028720121830701828,
+      "loss_ce": 1.834042814152781e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 1049058144,
+      "step": 6114
+    },
+    {
+      "epoch": 69.41076487252124,
+      "grad_norm": 2.815526873758044,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1049229860,
+      "step": 6115
+    },
+    {
+      "epoch": 69.41076487252124,
+      "loss": 0.045954931527376175,
+      "loss_ce": 0.0001480501814512536,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 1049229860,
+      "step": 6115
+    },
+    {
+      "epoch": 69.42209631728045,
+      "grad_norm": 3.9225679743618254,
+      "learning_rate": 5e-06,
+      "loss": 0.065,
+      "num_input_tokens_seen": 1049400552,
+      "step": 6116
+    },
+    {
+      "epoch": 69.42209631728045,
+      "loss": 0.03415348753333092,
+      "loss_ce": 9.586734813638031e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1049400552,
+      "step": 6116
+    },
+    {
+      "epoch": 69.43342776203966,
+      "grad_norm": 3.918168041140486,
+      "learning_rate": 5e-06,
+      "loss": 0.049,
+      "num_input_tokens_seen": 1049572756,
+      "step": 6117
+    },
+    {
+      "epoch": 69.43342776203966,
+      "loss": 0.08863445371389389,
+      "loss_ce": 1.140386302722618e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 1049572756,
+      "step": 6117
+    },
+    {
+      "epoch": 69.44475920679886,
+      "grad_norm": 4.933180234363589,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 1049744256,
+      "step": 6118
+    },
+    {
+      "epoch": 69.44475920679886,
+      "loss": 0.03854266554117203,
+      "loss_ce": 2.9484068363672122e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1049744256,
+      "step": 6118
+    },
+    {
+      "epoch": 69.45609065155807,
+      "grad_norm": 2.5139675123367624,
+      "learning_rate": 5e-06,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 1049915984,
+      "step": 6119
+    },
+    {
+      "epoch": 69.45609065155807,
+      "loss": 0.08057437092065811,
+      "loss_ce": 3.848466803901829e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 1049915984,
+      "step": 6119
+    },
+    {
+      "epoch": 69.46742209631728,
+      "grad_norm": 2.682071752666328,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 1050086952,
+      "step": 6120
+    },
+    {
+      "epoch": 69.46742209631728,
+      "loss": 0.047244325280189514,
+      "loss_ce": 3.362995630595833e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 1050086952,
+      "step": 6120
+    },
+    {
+      "epoch": 69.47875354107649,
+      "grad_norm": 3.2324562057725745,
+      "learning_rate": 5e-06,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 1050259208,
+      "step": 6121
+    },
+    {
+      "epoch": 69.47875354107649,
+      "loss": 0.033484216779470444,
+      "loss_ce": 2.169225990655832e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1050259208,
+      "step": 6121
+    },
+    {
+      "epoch": 69.4900849858357,
+      "grad_norm": 3.343413611918484,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1050431172,
+      "step": 6122
+    },
+    {
+      "epoch": 69.4900849858357,
+      "loss": 0.02710139751434326,
+      "loss_ce": 1.704711394268088e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1050431172,
+      "step": 6122
+    },
+    {
+      "epoch": 69.5014164305949,
+      "grad_norm": 2.8679378961770907,
+      "learning_rate": 5e-06,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 1050601156,
+      "step": 6123
+    },
+    {
+      "epoch": 69.5014164305949,
+      "loss": 0.023616516962647438,
+      "loss_ce": 4.1689767385832965e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.0235595703125,
+      "num_input_tokens_seen": 1050601156,
+      "step": 6123
+    },
+    {
+      "epoch": 69.51274787535411,
+      "grad_norm": 2.8652092682585577,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1050772208,
+      "step": 6124
+    },
+    {
+      "epoch": 69.51274787535411,
+      "loss": 0.030777039006352425,
+      "loss_ce": 3.057961293961853e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 1050772208,
+      "step": 6124
+    },
+    {
+      "epoch": 69.52407932011332,
+      "grad_norm": 3.6032131609225653,
+      "learning_rate": 5e-06,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 1050943468,
+      "step": 6125
+    },
+    {
+      "epoch": 69.52407932011332,
+      "loss": 0.031950563192367554,
+      "loss_ce": 1.3917691831011325e-05,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1050943468,
+      "step": 6125
+    },
+    {
+      "epoch": 69.53541076487252,
+      "grad_norm": 3.7051864058376247,
+      "learning_rate": 5e-06,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 1051111868,
+      "step": 6126
+    },
+    {
+      "epoch": 69.53541076487252,
+      "loss": 0.045482244342565536,
+      "loss_ce": 2.631138158903923e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 1051111868,
+      "step": 6126
+    },
+    {
+      "epoch": 69.54674220963173,
+      "grad_norm": 3.2277313792402915,
+      "learning_rate": 5e-06,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 1051282032,
+      "step": 6127
+    },
+    {
+      "epoch": 69.54674220963173,
+      "loss": 0.031909506767988205,
+      "loss_ce": 3.389604898984544e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1051282032,
+      "step": 6127
+    },
+    {
+      "epoch": 69.55807365439094,
+      "grad_norm": 3.516043378714864,
+      "learning_rate": 5e-06,
+      "loss": 0.036,
+      "num_input_tokens_seen": 1051453444,
+      "step": 6128
+    },
+    {
+      "epoch": 69.55807365439094,
+      "loss": 0.03425643965601921,
+      "loss_ce": 0.00019882083870470524,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1051453444,
+      "step": 6128
+    },
+    {
+      "epoch": 69.56940509915015,
+      "grad_norm": 3.814911899148198,
+      "learning_rate": 5e-06,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 1051625236,
+      "step": 6129
+    },
+    {
+      "epoch": 69.56940509915015,
+      "loss": 0.03843668848276138,
+      "loss_ce": 0.0009458417189307511,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 1051625236,
+      "step": 6129
+    },
+    {
+      "epoch": 69.58073654390935,
+      "grad_norm": 3.506933612733227,
+      "learning_rate": 5e-06,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 1051797336,
+      "step": 6130
+    },
+    {
+      "epoch": 69.58073654390935,
+      "loss": 0.07583128660917282,
+      "loss_ce": 2.562140070949681e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 1051797336,
+      "step": 6130
+    },
+    {
+      "epoch": 69.59206798866856,
+      "grad_norm": 3.050175570988111,
+      "learning_rate": 5e-06,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 1051969108,
+      "step": 6131
+    },
+    {
+      "epoch": 69.59206798866856,
+      "loss": 0.07978345453739166,
+      "loss_ce": 5.627945211017504e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 1051969108,
+      "step": 6131
+    },
+    {
+      "epoch": 69.60339943342777,
+      "grad_norm": 3.292593363396408,
+      "learning_rate": 5e-06,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 1052140812,
+      "step": 6132
+    },
+    {
+      "epoch": 69.60339943342777,
+      "loss": 0.027203774079680443,
+      "loss_ce": 2.7870884878211655e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 1052140812,
+      "step": 6132
+    },
+    {
+      "epoch": 69.61473087818698,
+      "grad_norm": 3.75680394740765,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 1052311980,
+      "step": 6133
+    },
+    {
+      "epoch": 69.61473087818698,
+      "loss": 0.05178377404808998,
+      "loss_ce": 4.1219303966499865e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 1052311980,
+      "step": 6133
+    },
+    {
+      "epoch": 69.62606232294617,
+      "grad_norm": 3.413427238896476,
+      "learning_rate": 5e-06,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 1052483728,
+      "step": 6134
+    },
+    {
+      "epoch": 69.62606232294617,
+      "loss": 0.08037745952606201,
+      "loss_ce": 0.0006350207841023803,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 1052483728,
+      "step": 6134
+    },
+    {
+      "epoch": 69.63739376770538,
+      "grad_norm": 2.861953296818038,
+      "learning_rate": 5e-06,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 1052654588,
+      "step": 6135
+    },
+    {
+      "epoch": 69.63739376770538,
+      "loss": 0.03404511511325836,
+      "loss_ce": 0.00044526217970997095,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1052654588,
+      "step": 6135
+    },
+    {
+      "epoch": 69.64872521246458,
+      "grad_norm": 3.1769476984958165,
+      "learning_rate": 5e-06,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 1052826496,
+      "step": 6136
+    },
+    {
+      "epoch": 69.64872521246458,
+      "loss": 0.03207669034600258,
+      "loss_ce": 1.797280492610298e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1052826496,
+      "step": 6136
+    },
+    {
+      "epoch": 69.66005665722379,
+      "grad_norm": 2.9367353324832517,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 1052998300,
+      "step": 6137
+    },
+    {
+      "epoch": 69.66005665722379,
+      "loss": 0.022776728495955467,
+      "loss_ce": 1.824560240493156e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 1052998300,
+      "step": 6137
+    },
+    {
+      "epoch": 69.671388101983,
+      "grad_norm": 3.1937218268913146,
+      "learning_rate": 5e-06,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 1053170284,
+      "step": 6138
+    },
+    {
+      "epoch": 69.671388101983,
+      "loss": 0.08735427260398865,
+      "loss_ce": 2.822013630066067e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 1053170284,
+      "step": 6138
+    },
+    {
+      "epoch": 69.6827195467422,
+      "grad_norm": 3.0279243998692658,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 1053341504,
+      "step": 6139
+    },
+    {
+      "epoch": 69.6827195467422,
+      "loss": 0.05386228859424591,
+      "loss_ce": 2.9281665774760768e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 1053341504,
+      "step": 6139
+    },
+    {
+      "epoch": 69.69405099150141,
+      "grad_norm": 3.3349301100090076,
+      "learning_rate": 5e-06,
+      "loss": 0.079,
+      "num_input_tokens_seen": 1053512476,
+      "step": 6140
+    },
+    {
+      "epoch": 69.69405099150141,
+      "loss": 0.114628367125988,
+      "loss_ce": 1.960219378815964e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 1053512476,
+      "step": 6140
+    },
+    {
+      "epoch": 69.70538243626062,
+      "grad_norm": 3.1141144975063,
+      "learning_rate": 5e-06,
+      "loss": 0.0257,
+      "num_input_tokens_seen": 1053684084,
+      "step": 6141
+    },
+    {
+      "epoch": 69.70538243626062,
+      "loss": 0.02537723071873188,
+      "loss_ce": 0.00013919416232965887,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 1053684084,
+      "step": 6141
+    },
+    {
+      "epoch": 69.71671388101983,
+      "grad_norm": 3.0929758989747924,
+      "learning_rate": 5e-06,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 1053856048,
+      "step": 6142
+    },
+    {
+      "epoch": 69.71671388101983,
+      "loss": 0.026500869542360306,
+      "loss_ce": 1.9242288544774055e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 1053856048,
+      "step": 6142
+    },
+    {
+      "epoch": 69.72804532577904,
+      "grad_norm": 3.5319484659013742,
+      "learning_rate": 5e-06,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 1054026288,
+      "step": 6143
+    },
+    {
+      "epoch": 69.72804532577904,
+      "loss": 0.030305106192827225,
+      "loss_ce": 6.218641647137702e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1054026288,
+      "step": 6143
+    },
+    {
+      "epoch": 69.73937677053824,
+      "grad_norm": 3.3771881888549613,
+      "learning_rate": 5e-06,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 1054193228,
+      "step": 6144
+    },
+    {
+      "epoch": 69.73937677053824,
+      "loss": 0.03407318890094757,
+      "loss_ce": 0.00015289938892237842,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1054193228,
+      "step": 6144
+    },
+    {
+      "epoch": 69.75070821529745,
+      "grad_norm": 3.049368731658451,
+      "learning_rate": 5e-06,
+      "loss": 0.0255,
+      "num_input_tokens_seen": 1054365340,
+      "step": 6145
+    },
+    {
+      "epoch": 69.75070821529745,
+      "loss": 0.026761401444673538,
+      "loss_ce": 2.800256697810255e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 1054365340,
+      "step": 6145
+    },
+    {
+      "epoch": 69.76203966005666,
+      "grad_norm": 3.1954434838993593,
+      "learning_rate": 5e-06,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 1054537048,
+      "step": 6146
+    },
+    {
+      "epoch": 69.76203966005666,
+      "loss": 0.040646135807037354,
+      "loss_ce": 4.2498337279539555e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1054537048,
+      "step": 6146
+    },
+    {
+      "epoch": 69.77337110481587,
+      "grad_norm": 3.467828717186582,
+      "learning_rate": 5e-06,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 1054708616,
+      "step": 6147
+    },
+    {
+      "epoch": 69.77337110481587,
+      "loss": 0.02807961031794548,
+      "loss_ce": 0.0005832723691128194,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 1054708616,
+      "step": 6147
+    },
+    {
+      "epoch": 69.78470254957507,
+      "grad_norm": 3.2862520091862293,
+      "learning_rate": 5e-06,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 1054880828,
+      "step": 6148
+    },
+    {
+      "epoch": 69.78470254957507,
+      "loss": 0.06495119631290436,
+      "loss_ce": 4.030777563457377e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 1054880828,
+      "step": 6148
+    },
+    {
+      "epoch": 69.79603399433428,
+      "grad_norm": 3.3807713252664224,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 1055052056,
+      "step": 6149
+    },
+    {
+      "epoch": 69.79603399433428,
+      "loss": 0.035884223878383636,
+      "loss_ce": 1.0808475053636357e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1055052056,
+      "step": 6149
+    },
+    {
+      "epoch": 69.80736543909349,
+      "grad_norm": 3.416896191566224,
+      "learning_rate": 5e-06,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 1055223968,
+      "step": 6150
+    },
+    {
+      "epoch": 69.80736543909349,
+      "loss": 0.03593236207962036,
+      "loss_ce": 0.002881825203076005,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 1055223968,
+      "step": 6150
+    },
+    {
+      "epoch": 69.8186968838527,
+      "grad_norm": 3.273938069814868,
+      "learning_rate": 5e-06,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 1055394816,
+      "step": 6151
+    },
+    {
+      "epoch": 69.8186968838527,
+      "loss": 0.03881031274795532,
+      "loss_ce": 2.2470598196377978e-05,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 1055394816,
+      "step": 6151
+    },
+    {
+      "epoch": 69.8300283286119,
+      "grad_norm": 3.1745237028102844,
+      "learning_rate": 5e-06,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 1055566528,
+      "step": 6152
+    },
+    {
+      "epoch": 69.8300283286119,
+      "loss": 0.051968708634376526,
+      "loss_ce": 2.778843554551713e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 1055566528,
+      "step": 6152
+    },
+    {
+      "epoch": 69.84135977337111,
+      "grad_norm": 3.076305780753445,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 1055736848,
+      "step": 6153
+    },
+    {
+      "epoch": 69.84135977337111,
+      "loss": 0.037253763526678085,
+      "loss_ce": 0.00020542369748000056,
+      "loss_iou": 0.421875,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 1055736848,
+      "step": 6153
+    },
+    {
+      "epoch": 69.85269121813032,
+      "grad_norm": 3.7851088014842476,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1055908736,
+      "step": 6154
+    },
+    {
+      "epoch": 69.85269121813032,
+      "loss": 0.044917941093444824,
+      "loss_ce": 5.710304685635492e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 1055908736,
+      "step": 6154
+    },
+    {
+      "epoch": 69.86402266288952,
+      "grad_norm": 3.490260886009078,
+      "learning_rate": 5e-06,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 1056080536,
+      "step": 6155
+    },
+    {
+      "epoch": 69.86402266288952,
+      "loss": 0.08909451216459274,
+      "loss_ce": 0.0021194161381572485,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 1056080536,
+      "step": 6155
+    },
+    {
+      "epoch": 69.87535410764872,
+      "grad_norm": 3.109214370172217,
+      "learning_rate": 5e-06,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 1056251592,
+      "step": 6156
+    },
+    {
+      "epoch": 69.87535410764872,
+      "loss": 0.05837704613804817,
+      "loss_ce": 7.32170301489532e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 1056251592,
+      "step": 6156
+    },
+    {
+      "epoch": 69.88668555240793,
+      "grad_norm": 3.2973669778192445,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 1056422588,
+      "step": 6157
+    },
+    {
+      "epoch": 69.88668555240793,
+      "loss": 0.03679753839969635,
+      "loss_ce": 5.437579602585174e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1056422588,
+      "step": 6157
+    },
+    {
+      "epoch": 69.89801699716713,
+      "grad_norm": 2.832491080343421,
+      "learning_rate": 5e-06,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 1056592052,
+      "step": 6158
+    },
+    {
+      "epoch": 69.89801699716713,
+      "loss": 0.02813900262117386,
+      "loss_ce": 3.2312989787897095e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 1056592052,
+      "step": 6158
+    },
+    {
+      "epoch": 69.90934844192634,
+      "grad_norm": 3.1540004610022834,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 1056763104,
+      "step": 6159
+    },
+    {
+      "epoch": 69.90934844192634,
+      "loss": 0.03876408934593201,
+      "loss_ce": 6.767540980945341e-06,
+      "loss_iou": 0.515625,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 1056763104,
+      "step": 6159
+    },
+    {
+      "epoch": 69.92067988668555,
+      "grad_norm": 3.540570572833432,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 1056934912,
+      "step": 6160
+    },
+    {
+      "epoch": 69.92067988668555,
+      "loss": 0.03748191148042679,
+      "loss_ce": 3.684240800794214e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 1056934912,
+      "step": 6160
+    },
+    {
+      "epoch": 69.93201133144476,
+      "grad_norm": 3.555012099284406,
+      "learning_rate": 5e-06,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 1057106784,
+      "step": 6161
+    },
+    {
+      "epoch": 69.93201133144476,
+      "loss": 0.035901911556720734,
+      "loss_ce": 2.8501468477770686e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1057106784,
+      "step": 6161
+    },
+    {
+      "epoch": 69.94334277620396,
+      "grad_norm": 2.911021570874157,
+      "learning_rate": 5e-06,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 1057278580,
+      "step": 6162
+    },
+    {
+      "epoch": 69.94334277620396,
+      "loss": 0.047409527003765106,
+      "loss_ce": 0.00010728441702667624,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 1057278580,
+      "step": 6162
+    },
+    {
+      "epoch": 69.95467422096317,
+      "grad_norm": 4.711453198508495,
+      "learning_rate": 5e-06,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 1057449572,
+      "step": 6163
+    },
+    {
+      "epoch": 69.95467422096317,
+      "loss": 0.03106820210814476,
+      "loss_ce": 0.0005124761955812573,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1057449572,
+      "step": 6163
+    },
+    {
+      "epoch": 69.96600566572238,
+      "grad_norm": 3.3138851567082996,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 1057621108,
+      "step": 6164
+    },
+    {
+      "epoch": 69.96600566572238,
+      "loss": 0.07658272236585617,
+      "loss_ce": 3.319292954984121e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 1057621108,
+      "step": 6164
+    },
+    {
+      "epoch": 69.97733711048159,
+      "grad_norm": 3.5273757360648847,
+      "learning_rate": 5e-06,
+      "loss": 0.04,
+      "num_input_tokens_seen": 1057791844,
+      "step": 6165
+    },
+    {
+      "epoch": 69.97733711048159,
+      "loss": 0.03083958476781845,
+      "loss_ce": 4.73485270049423e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 1057791844,
+      "step": 6165
+    },
+    {
+      "epoch": 69.98866855524079,
+      "grad_norm": 3.2129878766061672,
+      "learning_rate": 5e-06,
+      "loss": 0.037,
+      "num_input_tokens_seen": 1057962708,
+      "step": 6166
+    },
+    {
+      "epoch": 69.98866855524079,
+      "loss": 0.029260292649269104,
+      "loss_ce": 2.4452669094898738e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 1057962708,
+      "step": 6166
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 3.4529212212668274,
+      "learning_rate": 5e-06,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 1058134280,
+      "step": 6167
+    },
+    {
+      "epoch": 70.0,
+      "loss": 0.03967504948377609,
+      "loss_ce": 1.745303234201856e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 1058134280,
+      "step": 6167
+    },
+    {
+      "epoch": 70.01133144475921,
+      "grad_norm": 4.509035215887961,
+      "learning_rate": 5e-06,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 1058306384,
+      "step": 6168
+    },
+    {
+      "epoch": 70.01133144475921,
+      "loss": 0.08040237426757812,
+      "loss_ce": 6.484663754235953e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 1058306384,
+      "step": 6168
+    },
+    {
+      "epoch": 70.02266288951841,
+      "grad_norm": 3.088021551782971,
+      "learning_rate": 5e-06,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 1058478524,
+      "step": 6169
+    },
+    {
+      "epoch": 70.02266288951841,
+      "loss": 0.03102484717965126,
+      "loss_ce": 3.424622991587967e-05,
+      "loss_iou": 0.59375,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1058478524,
+      "step": 6169
+    },
+    {
+      "epoch": 70.03399433427762,
+      "grad_norm": 2.911710590305498,
+      "learning_rate": 5e-06,
+      "loss": 0.0272,
+      "num_input_tokens_seen": 1058648464,
+      "step": 6170
+    },
+    {
+      "epoch": 70.03399433427762,
+      "loss": 0.024744171649217606,
+      "loss_ce": 2.4934066459536552e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 1058648464,
+      "step": 6170
+    },
+    {
+      "epoch": 70.04532577903683,
+      "grad_norm": 3.307225420282024,
+      "learning_rate": 5e-06,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 1058820208,
+      "step": 6171
+    },
+    {
+      "epoch": 70.04532577903683,
+      "loss": 0.035976119339466095,
+      "loss_ce": 1.1152664228575304e-05,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1058820208,
+      "step": 6171
+    },
+    {
+      "epoch": 70.05665722379604,
+      "grad_norm": 3.253539414051452,
+      "learning_rate": 5e-06,
+      "loss": 0.048,
+      "num_input_tokens_seen": 1058991988,
+      "step": 6172
+    },
+    {
+      "epoch": 70.05665722379604,
+      "loss": 0.04432141035795212,
+      "loss_ce": 1.7517057131044567e-05,
+      "loss_iou": 0.06103515625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 1058991988,
+      "step": 6172
+    },
+    {
+      "epoch": 70.06798866855524,
+      "grad_norm": 3.7577538459775837,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 1059163612,
+      "step": 6173
+    },
+    {
+      "epoch": 70.06798866855524,
+      "loss": 0.041993699967861176,
+      "loss_ce": 4.728862768388353e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1059163612,
+      "step": 6173
+    },
+    {
+      "epoch": 70.07932011331445,
+      "grad_norm": 3.2451374050671786,
+      "learning_rate": 5e-06,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 1059335328,
+      "step": 6174
+    },
+    {
+      "epoch": 70.07932011331445,
+      "loss": 0.03083517961204052,
+      "loss_ce": 2.7682533982442692e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 1059335328,
+      "step": 6174
+    },
+    {
+      "epoch": 70.09065155807366,
+      "grad_norm": 2.7313674203319174,
+      "learning_rate": 5e-06,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 1059507120,
+      "step": 6175
+    },
+    {
+      "epoch": 70.09065155807366,
+      "loss": 0.06758061051368713,
+      "loss_ce": 6.0467391449492425e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 1059507120,
+      "step": 6175
+    },
+    {
+      "epoch": 70.10198300283287,
+      "grad_norm": 2.959345007315852,
+      "learning_rate": 5e-06,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 1059678680,
+      "step": 6176
+    },
+    {
+      "epoch": 70.10198300283287,
+      "loss": 0.03468804061412811,
+      "loss_ce": 2.0072940969839692e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1059678680,
+      "step": 6176
+    },
+    {
+      "epoch": 70.11331444759207,
+      "grad_norm": 3.4026638160561733,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 1059850668,
+      "step": 6177
+    },
+    {
+      "epoch": 70.11331444759207,
+      "loss": 0.06843946129083633,
+      "loss_ce": 4.957188502885401e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 1059850668,
+      "step": 6177
+    },
+    {
+      "epoch": 70.12464589235128,
+      "grad_norm": 3.6583013106266984,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 1060022348,
+      "step": 6178
+    },
+    {
+      "epoch": 70.12464589235128,
+      "loss": 0.10209149122238159,
+      "loss_ce": 5.5965760111575946e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 1060022348,
+      "step": 6178
+    },
+    {
+      "epoch": 70.13597733711048,
+      "grad_norm": 3.714873490747718,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 1060191120,
+      "step": 6179
+    },
+    {
+      "epoch": 70.13597733711048,
+      "loss": 0.03477904945611954,
+      "loss_ce": 6.530592509079725e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1060191120,
+      "step": 6179
+    },
+    {
+      "epoch": 70.14730878186968,
+      "grad_norm": 4.0095924521720345,
+      "learning_rate": 5e-06,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 1060362696,
+      "step": 6180
+    },
+    {
+      "epoch": 70.14730878186968,
+      "loss": 0.03978504613041878,
+      "loss_ce": 2.0641280570998788e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 1060362696,
+      "step": 6180
+    },
+    {
+      "epoch": 70.15864022662889,
+      "grad_norm": 3.6641240301162243,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 1060532088,
+      "step": 6181
+    },
+    {
+      "epoch": 70.15864022662889,
+      "loss": 0.025972293689846992,
+      "loss_ce": 3.235208714613691e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 1060532088,
+      "step": 6181
+    },
+    {
+      "epoch": 70.1699716713881,
+      "grad_norm": 3.7563183727561134,
+      "learning_rate": 5e-06,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 1060704268,
+      "step": 6182
+    },
+    {
+      "epoch": 70.1699716713881,
+      "loss": 0.03175430744886398,
+      "loss_ce": 4.654276926885359e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 1060704268,
+      "step": 6182
+    },
+    {
+      "epoch": 70.1813031161473,
+      "grad_norm": 3.885898202831319,
+      "learning_rate": 5e-06,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 1060876196,
+      "step": 6183
+    },
+    {
+      "epoch": 70.1813031161473,
+      "loss": 0.027663579210639,
+      "loss_ce": 2.991233122884296e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 1060876196,
+      "step": 6183
+    },
+    {
+      "epoch": 70.19263456090651,
+      "grad_norm": 4.256689402027802,
+      "learning_rate": 5e-06,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 1061047692,
+      "step": 6184
+    },
+    {
+      "epoch": 70.19263456090651,
+      "loss": 0.06845005601644516,
+      "loss_ce": 4.4908549170941114e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 1061047692,
+      "step": 6184
+    },
+    {
+      "epoch": 70.20396600566572,
+      "grad_norm": 3.109819554699469,
+      "learning_rate": 5e-06,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 1061217068,
+      "step": 6185
+    },
+    {
+      "epoch": 70.20396600566572,
+      "loss": 0.02164578251540661,
+      "loss_ce": 3.9335609471891075e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 1061217068,
+      "step": 6185
+    },
+    {
+      "epoch": 70.21529745042493,
+      "grad_norm": 3.5351829257635243,
+      "learning_rate": 5e-06,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 1061388988,
+      "step": 6186
+    },
+    {
+      "epoch": 70.21529745042493,
+      "loss": 0.028955310583114624,
+      "loss_ce": 2.4647413738421164e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 1061388988,
+      "step": 6186
+    },
+    {
+      "epoch": 70.22662889518413,
+      "grad_norm": 3.6217002103751734,
+      "learning_rate": 5e-06,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 1061560576,
+      "step": 6187
+    },
+    {
+      "epoch": 70.22662889518413,
+      "loss": 0.03518831729888916,
+      "loss_ce": 1.6808910004328936e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1061560576,
+      "step": 6187
+    },
+    {
+      "epoch": 70.23796033994334,
+      "grad_norm": 3.2643411458164104,
+      "learning_rate": 5e-06,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 1061732068,
+      "step": 6188
+    },
+    {
+      "epoch": 70.23796033994334,
+      "loss": 0.0447344109416008,
+      "loss_ce": 5.6677912652958184e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 1061732068,
+      "step": 6188
+    },
+    {
+      "epoch": 70.24929178470255,
+      "grad_norm": 3.0986623092849768,
+      "learning_rate": 5e-06,
+      "loss": 0.0373,
+      "num_input_tokens_seen": 1061903412,
+      "step": 6189
+    },
+    {
+      "epoch": 70.24929178470255,
+      "loss": 0.033926405012607574,
+      "loss_ce": 9.766806761035696e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1061903412,
+      "step": 6189
+    },
+    {
+      "epoch": 70.26062322946176,
+      "grad_norm": 2.7389124235081392,
+      "learning_rate": 5e-06,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 1062075140,
+      "step": 6190
+    },
+    {
+      "epoch": 70.26062322946176,
+      "loss": 0.03864557296037674,
+      "loss_ce": 4.084153624717146e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1062075140,
+      "step": 6190
+    },
+    {
+      "epoch": 70.27195467422096,
+      "grad_norm": 2.3240647748971157,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 1062244632,
+      "step": 6191
+    },
+    {
+      "epoch": 70.27195467422096,
+      "loss": 0.030351605266332626,
+      "loss_ce": 0.00021549731900449842,
+      "loss_iou": 0.34375,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 1062244632,
+      "step": 6191
+    },
+    {
+      "epoch": 70.28328611898017,
+      "grad_norm": 3.2711494664173673,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 1062416580,
+      "step": 6192
+    },
+    {
+      "epoch": 70.28328611898017,
+      "loss": 0.04531951993703842,
+      "loss_ce": 3.143498906865716e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 1062416580,
+      "step": 6192
+    },
+    {
+      "epoch": 70.29461756373938,
+      "grad_norm": 3.0356899201502197,
+      "learning_rate": 5e-06,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 1062587472,
+      "step": 6193
+    },
+    {
+      "epoch": 70.29461756373938,
+      "loss": 0.03430301323533058,
+      "loss_ce": 7.755000842735171e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1062587472,
+      "step": 6193
+    },
+    {
+      "epoch": 70.30594900849859,
+      "grad_norm": 2.687015101448378,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 1062757956,
+      "step": 6194
+    },
+    {
+      "epoch": 70.30594900849859,
+      "loss": 0.028414569795131683,
+      "loss_ce": 1.796483229554724e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 1062757956,
+      "step": 6194
+    },
+    {
+      "epoch": 70.3172804532578,
+      "grad_norm": 3.270507006458377,
+      "learning_rate": 5e-06,
+      "loss": 0.033,
+      "num_input_tokens_seen": 1062929548,
+      "step": 6195
+    },
+    {
+      "epoch": 70.3172804532578,
+      "loss": 0.03262282907962799,
+      "loss_ce": 0.0016780063742771745,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1062929548,
+      "step": 6195
+    },
+    {
+      "epoch": 70.328611898017,
+      "grad_norm": 3.420441285138797,
+      "learning_rate": 5e-06,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 1063101376,
+      "step": 6196
+    },
+    {
+      "epoch": 70.328611898017,
+      "loss": 0.050790995359420776,
+      "loss_ce": 4.026099486509338e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 1063101376,
+      "step": 6196
+    },
+    {
+      "epoch": 70.33994334277621,
+      "grad_norm": 3.021693814612,
+      "learning_rate": 5e-06,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 1063272232,
+      "step": 6197
+    },
+    {
+      "epoch": 70.33994334277621,
+      "loss": 0.025040145963430405,
+      "loss_ce": 0.00013780235894955695,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 1063272232,
+      "step": 6197
+    },
+    {
+      "epoch": 70.35127478753542,
+      "grad_norm": 2.777533714470306,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 1063442568,
+      "step": 6198
+    },
+    {
+      "epoch": 70.35127478753542,
+      "loss": 0.07694695889949799,
+      "loss_ce": 0.00021050882060080767,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 1063442568,
+      "step": 6198
+    },
+    {
+      "epoch": 70.36260623229462,
+      "grad_norm": 3.0098235443338974,
+      "learning_rate": 5e-06,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 1063614480,
+      "step": 6199
+    },
+    {
+      "epoch": 70.36260623229462,
+      "loss": 0.0431799441576004,
+      "loss_ce": 1.282903667743085e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1063614480,
+      "step": 6199
+    },
+    {
+      "epoch": 70.37393767705383,
+      "grad_norm": 3.5299715853278415,
+      "learning_rate": 5e-06,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 1063786596,
+      "step": 6200
+    },
+    {
+      "epoch": 70.37393767705383,
+      "loss": 0.0365288183093071,
+      "loss_ce": 1.4537613424181473e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1063786596,
+      "step": 6200
+    },
+    {
+      "epoch": 70.38526912181302,
+      "grad_norm": 3.2553877384854326,
+      "learning_rate": 5e-06,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 1063958432,
+      "step": 6201
+    },
+    {
+      "epoch": 70.38526912181302,
+      "loss": 0.030552154406905174,
+      "loss_ce": 4.983360122423619e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1063958432,
+      "step": 6201
+    },
+    {
+      "epoch": 70.39660056657223,
+      "grad_norm": 3.0266139359761626,
+      "learning_rate": 5e-06,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 1064129860,
+      "step": 6202
+    },
+    {
+      "epoch": 70.39660056657223,
+      "loss": 0.09253968298435211,
+      "loss_ce": 4.090438596904278e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 1064129860,
+      "step": 6202
+    },
+    {
+      "epoch": 70.40793201133144,
+      "grad_norm": 3.556413640662341,
+      "learning_rate": 5e-06,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 1064301736,
+      "step": 6203
+    },
+    {
+      "epoch": 70.40793201133144,
+      "loss": 0.09668193757534027,
+      "loss_ce": 3.2764088246040046e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 1064301736,
+      "step": 6203
+    },
+    {
+      "epoch": 70.41926345609065,
+      "grad_norm": 3.690919878043717,
+      "learning_rate": 5e-06,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 1064473712,
+      "step": 6204
+    },
+    {
+      "epoch": 70.41926345609065,
+      "loss": 0.03813214600086212,
+      "loss_ce": 1.569224514241796e-05,
+      "loss_iou": 0.0625,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 1064473712,
+      "step": 6204
+    },
+    {
+      "epoch": 70.43059490084985,
+      "grad_norm": 3.4682080094381607,
+      "learning_rate": 5e-06,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 1064643888,
+      "step": 6205
+    },
+    {
+      "epoch": 70.43059490084985,
+      "loss": 0.0844365656375885,
+      "loss_ce": 4.020178676000796e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 1064643888,
+      "step": 6205
+    },
+    {
+      "epoch": 70.44192634560906,
+      "grad_norm": 2.621481060362728,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 1064814664,
+      "step": 6206
+    },
+    {
+      "epoch": 70.44192634560906,
+      "loss": 0.07125361263751984,
+      "loss_ce": 4.084207466803491e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 1064814664,
+      "step": 6206
+    },
+    {
+      "epoch": 70.45325779036827,
+      "grad_norm": 2.4895973992645364,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 1064986300,
+      "step": 6207
+    },
+    {
+      "epoch": 70.45325779036827,
+      "loss": 0.07286182045936584,
+      "loss_ce": 1.6363295799237676e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 1064986300,
+      "step": 6207
+    },
+    {
+      "epoch": 70.46458923512748,
+      "grad_norm": 3.4498795981142663,
+      "learning_rate": 5e-06,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 1065158052,
+      "step": 6208
+    },
+    {
+      "epoch": 70.46458923512748,
+      "loss": 0.09581515192985535,
+      "loss_ce": 2.0473917174967937e-05,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 1065158052,
+      "step": 6208
+    },
+    {
+      "epoch": 70.47592067988668,
+      "grad_norm": 3.687708636754502,
+      "learning_rate": 5e-06,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 1065328704,
+      "step": 6209
+    },
+    {
+      "epoch": 70.47592067988668,
+      "loss": 0.04114636778831482,
+      "loss_ce": 6.970532558625564e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1065328704,
+      "step": 6209
+    },
+    {
+      "epoch": 70.48725212464589,
+      "grad_norm": 3.4492124862625517,
+      "learning_rate": 5e-06,
+      "loss": 0.051,
+      "num_input_tokens_seen": 1065500172,
+      "step": 6210
+    },
+    {
+      "epoch": 70.48725212464589,
+      "loss": 0.02557147666811943,
+      "loss_ce": 2.8261265470064245e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 1065500172,
+      "step": 6210
+    },
+    {
+      "epoch": 70.4985835694051,
+      "grad_norm": 3.253813081175315,
+      "learning_rate": 5e-06,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 1065671412,
+      "step": 6211
+    },
+    {
+      "epoch": 70.4985835694051,
+      "loss": 0.028385695070028305,
+      "loss_ce": 3.486586865619756e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 1065671412,
+      "step": 6211
+    },
+    {
+      "epoch": 70.5099150141643,
+      "grad_norm": 2.851093667591735,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 1065842528,
+      "step": 6212
+    },
+    {
+      "epoch": 70.5099150141643,
+      "loss": 0.03827948495745659,
+      "loss_ce": 4.096064003533684e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1065842528,
+      "step": 6212
+    },
+    {
+      "epoch": 70.52124645892351,
+      "grad_norm": 2.738830959647769,
+      "learning_rate": 5e-06,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 1066013720,
+      "step": 6213
+    },
+    {
+      "epoch": 70.52124645892351,
+      "loss": 0.027378059923648834,
+      "loss_ce": 4.9569291149964556e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 1066013720,
+      "step": 6213
+    },
+    {
+      "epoch": 70.53257790368272,
+      "grad_norm": 2.63645023513762,
+      "learning_rate": 5e-06,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 1066184636,
+      "step": 6214
+    },
+    {
+      "epoch": 70.53257790368272,
+      "loss": 0.033459797501564026,
+      "loss_ce": 2.7791174943558872e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1066184636,
+      "step": 6214
+    },
+    {
+      "epoch": 70.54390934844193,
+      "grad_norm": 2.893925016911087,
+      "learning_rate": 5e-06,
+      "loss": 0.052,
+      "num_input_tokens_seen": 1066357252,
+      "step": 6215
+    },
+    {
+      "epoch": 70.54390934844193,
+      "loss": 0.09516491740942001,
+      "loss_ce": 2.6371020794613287e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 1066357252,
+      "step": 6215
+    },
+    {
+      "epoch": 70.55524079320114,
+      "grad_norm": 2.5061209317037147,
+      "learning_rate": 5e-06,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 1066528296,
+      "step": 6216
+    },
+    {
+      "epoch": 70.55524079320114,
+      "loss": 0.03342249616980553,
+      "loss_ce": 2.1009296688134782e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1066528296,
+      "step": 6216
+    },
+    {
+      "epoch": 70.56657223796034,
+      "grad_norm": 2.7913668419555995,
+      "learning_rate": 5e-06,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 1066699960,
+      "step": 6217
+    },
+    {
+      "epoch": 70.56657223796034,
+      "loss": 0.023264724761247635,
+      "loss_ce": 4.084629108547233e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 1066699960,
+      "step": 6217
+    },
+    {
+      "epoch": 70.57790368271955,
+      "grad_norm": 3.243231601320792,
+      "learning_rate": 5e-06,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 1066871824,
+      "step": 6218
+    },
+    {
+      "epoch": 70.57790368271955,
+      "loss": 0.04991789162158966,
+      "loss_ce": 5.217043508309871e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 1066871824,
+      "step": 6218
+    },
+    {
+      "epoch": 70.58923512747876,
+      "grad_norm": 3.1959772068900536,
+      "learning_rate": 5e-06,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 1067042872,
+      "step": 6219
+    },
+    {
+      "epoch": 70.58923512747876,
+      "loss": 0.06666183471679688,
+      "loss_ce": 4.19584393966943e-05,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 1067042872,
+      "step": 6219
+    },
+    {
+      "epoch": 70.60056657223797,
+      "grad_norm": 3.7139893822689927,
+      "learning_rate": 5e-06,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 1067213744,
+      "step": 6220
+    },
+    {
+      "epoch": 70.60056657223797,
+      "loss": 0.04193349927663803,
+      "loss_ce": 3.2866824767552316e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1067213744,
+      "step": 6220
+    },
+    {
+      "epoch": 70.61189801699717,
+      "grad_norm": 2.879566284772091,
+      "learning_rate": 5e-06,
+      "loss": 0.038,
+      "num_input_tokens_seen": 1067385320,
+      "step": 6221
+    },
+    {
+      "epoch": 70.61189801699717,
+      "loss": 0.030481917783617973,
+      "loss_ce": 5.589351712842472e-05,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 1067385320,
+      "step": 6221
+    },
+    {
+      "epoch": 70.62322946175638,
+      "grad_norm": 2.9003876610040393,
+      "learning_rate": 5e-06,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 1067557464,
+      "step": 6222
+    },
+    {
+      "epoch": 70.62322946175638,
+      "loss": 0.02531801536679268,
+      "loss_ce": 1.8942057067761198e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 1067557464,
+      "step": 6222
+    },
+    {
+      "epoch": 70.63456090651557,
+      "grad_norm": 3.1934085265570475,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 1067728976,
+      "step": 6223
+    },
+    {
+      "epoch": 70.63456090651557,
+      "loss": 0.03895484656095505,
+      "loss_ce": 1.4415631085284986e-05,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 1067728976,
+      "step": 6223
+    },
+    {
+      "epoch": 70.64589235127478,
+      "grad_norm": 3.215257014153987,
+      "learning_rate": 5e-06,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 1067900632,
+      "step": 6224
+    },
+    {
+      "epoch": 70.64589235127478,
+      "loss": 0.024823594838380814,
+      "loss_ce": 5.857882206328213e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 1067900632,
+      "step": 6224
+    },
+    {
+      "epoch": 70.65722379603399,
+      "grad_norm": 3.3511728802599636,
+      "learning_rate": 5e-06,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 1068072332,
+      "step": 6225
+    },
+    {
+      "epoch": 70.65722379603399,
+      "loss": 0.0377066433429718,
+      "loss_ce": 4.795142012881115e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 1068072332,
+      "step": 6225
+    },
+    {
+      "epoch": 70.6685552407932,
+      "grad_norm": 3.134878021635956,
+      "learning_rate": 5e-06,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 1068243588,
+      "step": 6226
+    },
+    {
+      "epoch": 70.6685552407932,
+      "loss": 0.03613574802875519,
+      "loss_ce": 1.8194015865447e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 1068243588,
+      "step": 6226
+    },
+    {
+      "epoch": 70.6798866855524,
+      "grad_norm": 2.243108998814011,
+      "learning_rate": 5e-06,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 1068415116,
+      "step": 6227
+    },
+    {
+      "epoch": 70.6798866855524,
+      "loss": 0.02113475650548935,
+      "loss_ce": 1.659293775446713e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.00421142578125,
+      "loss_xval": 0.0211181640625,
+      "num_input_tokens_seen": 1068415116,
+      "step": 6227
+    },
+    {
+      "epoch": 70.69121813031161,
+      "grad_norm": 2.949245700762008,
+      "learning_rate": 5e-06,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 1068586076,
+      "step": 6228
+    },
+    {
+      "epoch": 70.69121813031161,
+      "loss": 0.03588490933179855,
+      "loss_ce": 2.675283030839637e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1068586076,
+      "step": 6228
+    },
+    {
+      "epoch": 70.70254957507082,
+      "grad_norm": 3.5533962202450087,
+      "learning_rate": 5e-06,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 1068757740,
+      "step": 6229
+    },
+    {
+      "epoch": 70.70254957507082,
+      "loss": 0.03083064593374729,
+      "loss_ce": 7.891836503404193e-06,
+      "loss_iou": 0.578125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 1068757740,
+      "step": 6229
+    },
+    {
+      "epoch": 70.71388101983003,
+      "grad_norm": 3.391818726883472,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 1068929096,
+      "step": 6230
+    },
+    {
+      "epoch": 70.71388101983003,
+      "loss": 0.04288551211357117,
+      "loss_ce": 5.409139703260735e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 1068929096,
+      "step": 6230
+    },
+    {
+      "epoch": 70.72521246458923,
+      "grad_norm": 3.249833817117871,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 1069100744,
+      "step": 6231
+    },
+    {
+      "epoch": 70.72521246458923,
+      "loss": 0.03534594550728798,
+      "loss_ce": 8.28854099381715e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1069100744,
+      "step": 6231
+    },
+    {
+      "epoch": 70.73654390934844,
+      "grad_norm": 3.3667022486172273,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 1069272520,
+      "step": 6232
+    },
+    {
+      "epoch": 70.73654390934844,
+      "loss": 0.027780842036008835,
+      "loss_ce": 2.5106468456215225e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 1069272520,
+      "step": 6232
+    },
+    {
+      "epoch": 70.74787535410765,
+      "grad_norm": 3.240958815165247,
+      "learning_rate": 5e-06,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 1069443876,
+      "step": 6233
+    },
+    {
+      "epoch": 70.74787535410765,
+      "loss": 0.043085820972919464,
+      "loss_ce": 7.12929613655433e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 1069443876,
+      "step": 6233
+    },
+    {
+      "epoch": 70.75920679886686,
+      "grad_norm": 3.2056680044434955,
+      "learning_rate": 5e-06,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 1069614980,
+      "step": 6234
+    },
+    {
+      "epoch": 70.75920679886686,
+      "loss": 0.08273632824420929,
+      "loss_ce": 9.47221415117383e-05,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 1069614980,
+      "step": 6234
+    },
+    {
+      "epoch": 70.77053824362606,
+      "grad_norm": 3.260752791093577,
+      "learning_rate": 5e-06,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 1069785840,
+      "step": 6235
+    },
+    {
+      "epoch": 70.77053824362606,
+      "loss": 0.04071637988090515,
+      "loss_ce": 5.928438895352883e-06,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1069785840,
+      "step": 6235
+    },
+    {
+      "epoch": 70.78186968838527,
+      "grad_norm": 4.037914205742755,
+      "learning_rate": 5e-06,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 1069957884,
+      "step": 6236
+    },
+    {
+      "epoch": 70.78186968838527,
+      "loss": 0.03360344097018242,
+      "loss_ce": 0.00011039771197829396,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1069957884,
+      "step": 6236
+    },
+    {
+      "epoch": 70.79320113314448,
+      "grad_norm": 2.606737444567429,
+      "learning_rate": 5e-06,
+      "loss": 0.067,
+      "num_input_tokens_seen": 1070128984,
+      "step": 6237
+    },
+    {
+      "epoch": 70.79320113314448,
+      "loss": 0.03347121924161911,
+      "loss_ce": 5.446997965918854e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1070128984,
+      "step": 6237
+    },
+    {
+      "epoch": 70.80453257790369,
+      "grad_norm": 2.4293053177601283,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 1070300848,
+      "step": 6238
+    },
+    {
+      "epoch": 70.80453257790369,
+      "loss": 0.0309780091047287,
+      "loss_ce": 4.844508657697588e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 1070300848,
+      "step": 6238
+    },
+    {
+      "epoch": 70.8158640226629,
+      "grad_norm": 2.8708135379197763,
+      "learning_rate": 5e-06,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 1070471432,
+      "step": 6239
+    },
+    {
+      "epoch": 70.8158640226629,
+      "loss": 0.04072948172688484,
+      "loss_ce": 3.429186108405702e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1070471432,
+      "step": 6239
+    },
+    {
+      "epoch": 70.8271954674221,
+      "grad_norm": 3.569295927151639,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 1070643176,
+      "step": 6240
+    },
+    {
+      "epoch": 70.8271954674221,
+      "loss": 0.07012691348791122,
+      "loss_ce": 4.329534567659721e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 1070643176,
+      "step": 6240
+    },
+    {
+      "epoch": 70.83852691218131,
+      "grad_norm": 3.2310263945865003,
+      "learning_rate": 5e-06,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 1070814120,
+      "step": 6241
+    },
+    {
+      "epoch": 70.83852691218131,
+      "loss": 0.029400072991847992,
+      "loss_ce": 2.6904177502729e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 1070814120,
+      "step": 6241
+    },
+    {
+      "epoch": 70.84985835694052,
+      "grad_norm": 3.6011500650156387,
+      "learning_rate": 5e-06,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 1070985284,
+      "step": 6242
+    },
+    {
+      "epoch": 70.84985835694052,
+      "loss": 0.03525716811418533,
+      "loss_ce": 0.00020772943389602005,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1070985284,
+      "step": 6242
+    },
+    {
+      "epoch": 70.86118980169972,
+      "grad_norm": 3.28721032710094,
+      "learning_rate": 5e-06,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 1071157168,
+      "step": 6243
+    },
+    {
+      "epoch": 70.86118980169972,
+      "loss": 0.05845300480723381,
+      "loss_ce": 1.1842974345199764e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 1071157168,
+      "step": 6243
+    },
+    {
+      "epoch": 70.87252124645893,
+      "grad_norm": 3.649765057288374,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 1071329260,
+      "step": 6244
+    },
+    {
+      "epoch": 70.87252124645893,
+      "loss": 0.05465655401349068,
+      "loss_ce": 1.4832143278908916e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 1071329260,
+      "step": 6244
+    },
+    {
+      "epoch": 70.88385269121812,
+      "grad_norm": 3.344622770963368,
+      "learning_rate": 5e-06,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 1071500976,
+      "step": 6245
+    },
+    {
+      "epoch": 70.88385269121812,
+      "loss": 0.03332934528589249,
+      "loss_ce": 1.9410406821407378e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1071500976,
+      "step": 6245
+    },
+    {
+      "epoch": 70.89518413597733,
+      "grad_norm": 3.415172816286984,
+      "learning_rate": 5e-06,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 1071672856,
+      "step": 6246
+    },
+    {
+      "epoch": 70.89518413597733,
+      "loss": 0.04589034616947174,
+      "loss_ce": 2.2425159841077402e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 1071672856,
+      "step": 6246
+    },
+    {
+      "epoch": 70.90651558073654,
+      "grad_norm": 3.3976247946749623,
+      "learning_rate": 5e-06,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 1071844692,
+      "step": 6247
+    },
+    {
+      "epoch": 70.90651558073654,
+      "loss": 0.10620458424091339,
+      "loss_ce": 1.8668979464564472e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 1071844692,
+      "step": 6247
+    },
+    {
+      "epoch": 70.91784702549575,
+      "grad_norm": 3.409508605869081,
+      "learning_rate": 5e-06,
+      "loss": 0.0298,
+      "num_input_tokens_seen": 1072016720,
+      "step": 6248
+    },
+    {
+      "epoch": 70.91784702549575,
+      "loss": 0.03481808304786682,
+      "loss_ce": 2.804333416861482e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1072016720,
+      "step": 6248
+    },
+    {
+      "epoch": 70.92917847025495,
+      "grad_norm": 3.3234102062615842,
+      "learning_rate": 5e-06,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 1072187036,
+      "step": 6249
+    },
+    {
+      "epoch": 70.92917847025495,
+      "loss": 0.04293372482061386,
+      "loss_ce": 7.178887608461082e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 1072187036,
+      "step": 6249
+    },
+    {
+      "epoch": 70.94050991501416,
+      "grad_norm": 3.2652082636012816,
+      "learning_rate": 5e-06,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 1072358776,
+      "step": 6250
+    },
+    {
+      "epoch": 70.94050991501416,
+      "eval_seeclick_CIoU": 0.5332036912441254,
+      "eval_seeclick_GIoU": 0.537353515625,
+      "eval_seeclick_IoU": 0.5715103447437286,
+      "eval_seeclick_MAE_all": 0.06658566743135452,
+      "eval_seeclick_MAE_h": 0.03053866233676672,
+      "eval_seeclick_MAE_w": 0.10175251960754395,
+      "eval_seeclick_MAE_x": 0.10434956476092339,
+      "eval_seeclick_MAE_y": 0.029701927676796913,
+      "eval_seeclick_NUM_probability": 0.999990314245224,
+      "eval_seeclick_inside_bbox": 0.8778409063816071,
+      "eval_seeclick_loss": 1.0003641843795776,
+      "eval_seeclick_loss_ce": 0.7595674097537994,
+      "eval_seeclick_loss_iou": 0.527099609375,
+      "eval_seeclick_loss_num": 0.0487518310546875,
+      "eval_seeclick_loss_xval": 0.24383544921875,
+      "eval_seeclick_runtime": 69.491,
+      "eval_seeclick_samples_per_second": 0.619,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 1072358776,
+      "step": 6250
+    },
+    {
+      "epoch": 70.94050991501416,
+      "eval_icons_CIoU": 0.7184293568134308,
+      "eval_icons_GIoU": 0.7168136835098267,
+      "eval_icons_IoU": 0.7327116429805756,
+      "eval_icons_MAE_all": 0.0377997811883688,
+      "eval_icons_MAE_h": 0.033207567408680916,
+      "eval_icons_MAE_w": 0.04584744665771723,
+      "eval_icons_MAE_x": 0.036356305703520775,
+      "eval_icons_MAE_y": 0.0357878003269434,
+      "eval_icons_NUM_probability": 0.9997987747192383,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.14739839732646942,
+      "eval_icons_loss_ce": 0.005564313847571611,
+      "eval_icons_loss_iou": 0.6092529296875,
+      "eval_icons_loss_num": 0.0262298583984375,
+      "eval_icons_loss_xval": 0.131103515625,
+      "eval_icons_runtime": 79.2524,
+      "eval_icons_samples_per_second": 0.631,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 1072358776,
+      "step": 6250
+    },
+    {
+      "epoch": 70.94050991501416,
+      "eval_screenspot_CIoU": 0.6451834638913473,
+      "eval_screenspot_GIoU": 0.6446922222773234,
+      "eval_screenspot_IoU": 0.6738998492558798,
+      "eval_screenspot_MAE_all": 0.0641996053357919,
+      "eval_screenspot_MAE_h": 0.03544317185878754,
+      "eval_screenspot_MAE_w": 0.11415816657245159,
+      "eval_screenspot_MAE_x": 0.0754590214540561,
+      "eval_screenspot_MAE_y": 0.031738064251840115,
+      "eval_screenspot_NUM_probability": 0.9999789396921793,
+      "eval_screenspot_inside_bbox": 0.9183333317438761,
+      "eval_screenspot_loss": 0.2720736265182495,
+      "eval_screenspot_loss_ce": 0.015694058500230312,
+      "eval_screenspot_loss_iou": 0.458251953125,
+      "eval_screenspot_loss_num": 0.051142374674479164,
+      "eval_screenspot_loss_xval": 0.2555033365885417,
+      "eval_screenspot_runtime": 154.6758,
+      "eval_screenspot_samples_per_second": 0.575,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 1072358776,
+      "step": 6250
+    },
+    {
+      "epoch": 70.94050991501416,
+      "eval_compot_CIoU": 0.8608261346817017,
+      "eval_compot_GIoU": 0.8591076731681824,
+      "eval_compot_IoU": 0.8748774826526642,
+      "eval_compot_MAE_all": 0.026053864508867264,
+      "eval_compot_MAE_h": 0.02242235280573368,
+      "eval_compot_MAE_w": 0.030955156311392784,
+      "eval_compot_MAE_x": 0.028336072340607643,
+      "eval_compot_MAE_y": 0.02250187285244465,
+      "eval_compot_NUM_probability": 0.99997878074646,
+      "eval_compot_inside_bbox": 0.9409722089767456,
+      "eval_compot_loss": 0.08311405032873154,
+      "eval_compot_loss_ce": 1.6617660548945423e-05,
+      "eval_compot_loss_iou": 0.4324951171875,
+      "eval_compot_loss_num": 0.014416694641113281,
+      "eval_compot_loss_xval": 0.07208251953125,
+      "eval_compot_runtime": 82.8768,
+      "eval_compot_samples_per_second": 0.603,
+      "eval_compot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 1072358776,
+      "step": 6250
+    },
+    {
+      "epoch": 70.94050991501416,
+      "eval_custom_ui_MAE_all": 0.018017619382590055,
+      "eval_custom_ui_MAE_x": 0.029217696748673916,
+      "eval_custom_ui_MAE_y": 0.00681754294782877,
+      "eval_custom_ui_NUM_probability": 0.9999310374259949,
+      "eval_custom_ui_loss": 0.20993298292160034,
+      "eval_custom_ui_loss_ce": 0.1193423680961132,
+      "eval_custom_ui_loss_num": 0.017267227172851562,
+      "eval_custom_ui_loss_xval": 0.08635711669921875,
+      "eval_custom_ui_runtime": 60.0096,
+      "eval_custom_ui_samples_per_second": 0.833,
+      "eval_custom_ui_steps_per_second": 0.033,
+      "num_input_tokens_seen": 1072358776,
+      "step": 6250
+    },
+    {
+      "epoch": 70.94050991501416,
+      "loss": 0.25332245230674744,
+      "loss_ce": 0.14626678824424744,
+      "loss_iou": 0.0,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 1072358776,
+      "step": 6250
+    },
+    {
+      "epoch": 70.95184135977337,
+      "grad_norm": 2.6022644252433853,
+      "learning_rate": 5e-06,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 1072531044,
+      "step": 6251
+    },
+    {
+      "epoch": 70.95184135977337,
+      "loss": 0.028813624754548073,
+      "loss_ce": 2.792044688249007e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 1072531044,
+      "step": 6251
+    },
+    {
+      "epoch": 70.96317280453258,
+      "grad_norm": 3.6093790969367756,
+      "learning_rate": 5e-06,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 1072702736,
+      "step": 6252
+    },
+    {
+      "epoch": 70.96317280453258,
+      "loss": 0.03342428058385849,
+      "loss_ce": 3.805007872870192e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1072702736,
+      "step": 6252
+    },
+    {
+      "epoch": 70.97450424929178,
+      "grad_norm": 3.51340177743106,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 1072874292,
+      "step": 6253
+    },
+    {
+      "epoch": 70.97450424929178,
+      "loss": 0.05357645824551582,
+      "loss_ce": 3.336972076795064e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 1072874292,
+      "step": 6253
+    },
+    {
+      "epoch": 70.98583569405099,
+      "grad_norm": 3.7240316323212315,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 1073045900,
+      "step": 6254
+    },
+    {
+      "epoch": 70.98583569405099,
+      "loss": 0.03787202015519142,
+      "loss_ce": 4.548008291749284e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 1073045900,
+      "step": 6254
+    },
+    {
+      "epoch": 70.9971671388102,
+      "grad_norm": 3.090858616541715,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 1073216328,
+      "step": 6255
+    },
+    {
+      "epoch": 70.9971671388102,
+      "loss": 0.0370880588889122,
+      "loss_ce": 9.20179627428297e-06,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 1073216328,
+      "step": 6255
+    },
+    {
+      "epoch": 70.9971671388102,
+      "loss": 0.07131224125623703,
+      "loss_ce": 2.318295810255222e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 1073259208,
+      "step": 6255
+    },
+    {
+      "epoch": 71.0084985835694,
+      "grad_norm": 3.8275486719057987,
+      "learning_rate": 5e-06,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 1073387672,
+      "step": 6256
+    },
+    {
+      "epoch": 71.0084985835694,
+      "loss": 0.040035270154476166,
+      "loss_ce": 1.9094208255410194e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 1073387672,
+      "step": 6256
+    },
+    {
+      "epoch": 71.01983002832861,
+      "grad_norm": 3.981111185185952,
+      "learning_rate": 5e-06,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 1073559764,
+      "step": 6257
+    },
+    {
+      "epoch": 71.01983002832861,
+      "loss": 0.02991669252514839,
+      "loss_ce": 2.4726070478209294e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 1073559764,
+      "step": 6257
+    },
+    {
+      "epoch": 71.03116147308782,
+      "grad_norm": 5.3028674476185165,
+      "learning_rate": 5e-06,
+      "loss": 0.047,
+      "num_input_tokens_seen": 1073731528,
+      "step": 6258
+    },
+    {
+      "epoch": 71.03116147308782,
+      "loss": 0.03628016263246536,
+      "loss_ce": 5.57987586944364e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 1073731528,
+      "step": 6258
+    },
+    {
+      "epoch": 71.04249291784703,
+      "grad_norm": 3.6641537516413427,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 1073902840,
+      "step": 6259
+    },
+    {
+      "epoch": 71.04249291784703,
+      "loss": 0.041697751730680466,
+      "loss_ce": 1.073991552402731e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 1073902840,
+      "step": 6259
+    },
+    {
+      "epoch": 71.05382436260624,
+      "grad_norm": 3.650230366892165,
+      "learning_rate": 5e-06,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 1074074256,
+      "step": 6260
+    },
+    {
+      "epoch": 71.05382436260624,
+      "loss": 0.05637975037097931,
+      "loss_ce": 2.9039892979199067e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 1074074256,
+      "step": 6260
+    },
+    {
+      "epoch": 71.06515580736544,
+      "grad_norm": 3.377017852906457,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 1074244400,
+      "step": 6261
+    },
+    {
+      "epoch": 71.06515580736544,
+      "loss": 0.05887112393975258,
+      "loss_ce": 1.7974434740608558e-05,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 1074244400,
+      "step": 6261
+    },
+    {
+      "epoch": 71.07648725212465,
+      "grad_norm": 3.519817402950062,
+      "learning_rate": 5e-06,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 1074416888,
+      "step": 6262
+    },
+    {
+      "epoch": 71.07648725212465,
+      "loss": 0.04325985908508301,
+      "loss_ce": 3.171001299051568e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1074416888,
+      "step": 6262
+    },
+    {
+      "epoch": 71.08781869688386,
+      "grad_norm": 3.730597520861243,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 1074589212,
+      "step": 6263
+    },
+    {
+      "epoch": 71.08781869688386,
+      "loss": 0.030119143426418304,
+      "loss_ce": 2.881279942812398e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1074589212,
+      "step": 6263
+    },
+    {
+      "epoch": 71.09915014164307,
+      "grad_norm": 3.427052563276431,
+      "learning_rate": 5e-06,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 1074759160,
+      "step": 6264
+    },
+    {
+      "epoch": 71.09915014164307,
+      "loss": 0.029640046879649162,
+      "loss_ce": 3.7994970625732094e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 1074759160,
+      "step": 6264
+    },
+    {
+      "epoch": 71.11048158640227,
+      "grad_norm": 2.8740436245657053,
+      "learning_rate": 5e-06,
+      "loss": 0.0497,
+      "num_input_tokens_seen": 1074929984,
+      "step": 6265
+    },
+    {
+      "epoch": 71.11048158640227,
+      "loss": 0.043136321008205414,
+      "loss_ce": 3.024235411430709e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1074929984,
+      "step": 6265
+    },
+    {
+      "epoch": 71.12181303116148,
+      "grad_norm": 2.695746382285947,
+      "learning_rate": 5e-06,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 1075102372,
+      "step": 6266
+    },
+    {
+      "epoch": 71.12181303116148,
+      "loss": 0.026429925113916397,
+      "loss_ce": 3.221843144274317e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 1075102372,
+      "step": 6266
+    },
+    {
+      "epoch": 71.13314447592067,
+      "grad_norm": 2.9772407460159775,
+      "learning_rate": 5e-06,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 1075274072,
+      "step": 6267
+    },
+    {
+      "epoch": 71.13314447592067,
+      "loss": 0.025593260303139687,
+      "loss_ce": 1.9531062207533978e-05,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 1075274072,
+      "step": 6267
+    },
+    {
+      "epoch": 71.14447592067988,
+      "grad_norm": 2.8037580673440234,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 1075445940,
+      "step": 6268
+    },
+    {
+      "epoch": 71.14447592067988,
+      "loss": 0.026813849806785583,
+      "loss_ce": 1.941429582075216e-05,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 1075445940,
+      "step": 6268
+    },
+    {
+      "epoch": 71.15580736543909,
+      "grad_norm": 2.5285215725299155,
+      "learning_rate": 5e-06,
+      "loss": 0.037,
+      "num_input_tokens_seen": 1075617656,
+      "step": 6269
+    },
+    {
+      "epoch": 71.15580736543909,
+      "loss": 0.031994231045246124,
+      "loss_ce": 2.7069992938777432e-05,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1075617656,
+      "step": 6269
+    },
+    {
+      "epoch": 71.1671388101983,
+      "grad_norm": 3.0378333829680857,
+      "learning_rate": 5e-06,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 1075789716,
+      "step": 6270
+    },
+    {
+      "epoch": 71.1671388101983,
+      "loss": 0.022927870973944664,
+      "loss_ce": 1.679891829553526e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 1075789716,
+      "step": 6270
+    },
+    {
+      "epoch": 71.1784702549575,
+      "grad_norm": 2.9116909751112456,
+      "learning_rate": 5e-06,
+      "loss": 0.064,
+      "num_input_tokens_seen": 1075961440,
+      "step": 6271
+    },
+    {
+      "epoch": 71.1784702549575,
+      "loss": 0.02443828620016575,
+      "loss_ce": 1.6594827684457414e-05,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 1075961440,
+      "step": 6271
+    },
+    {
+      "epoch": 71.18980169971671,
+      "grad_norm": 3.0447241125512035,
+      "learning_rate": 5e-06,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 1076131520,
+      "step": 6272
+    },
+    {
+      "epoch": 71.18980169971671,
+      "loss": 0.12112133949995041,
+      "loss_ce": 2.7589607270783745e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 1076131520,
+      "step": 6272
+    },
+    {
+      "epoch": 71.20113314447592,
+      "grad_norm": 3.0092335967304904,
+      "learning_rate": 5e-06,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 1076300164,
+      "step": 6273
+    },
+    {
+      "epoch": 71.20113314447592,
+      "loss": 0.039823926985263824,
+      "loss_ce": 5.952330684522167e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 1076300164,
+      "step": 6273
+    },
+    {
+      "epoch": 71.21246458923513,
+      "grad_norm": 3.7401573267971195,
+      "learning_rate": 5e-06,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 1076472136,
+      "step": 6274
+    },
+    {
+      "epoch": 71.21246458923513,
+      "loss": 0.050376009196043015,
+      "loss_ce": 2.200378366978839e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 1076472136,
+      "step": 6274
+    },
+    {
+      "epoch": 71.22379603399433,
+      "grad_norm": 4.310655285148014,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 1076642556,
+      "step": 6275
+    },
+    {
+      "epoch": 71.22379603399433,
+      "loss": 0.032685473561286926,
+      "loss_ce": 3.1665284041082487e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 1076642556,
+      "step": 6275
+    },
+    {
+      "epoch": 71.23512747875354,
+      "grad_norm": 2.9037329739437503,
+      "learning_rate": 5e-06,
+      "loss": 0.044,
+      "num_input_tokens_seen": 1076814060,
+      "step": 6276
+    },
+    {
+      "epoch": 71.23512747875354,
+      "loss": 0.0578373521566391,
+      "loss_ce": 2.1802028641104698e-05,
+      "loss_iou": 0.08544921875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 1076814060,
+      "step": 6276
+    },
+    {
+      "epoch": 71.24645892351275,
+      "grad_norm": 2.839188700892403,
+      "learning_rate": 5e-06,
+      "loss": 0.0371,
+      "num_input_tokens_seen": 1076985036,
+      "step": 6277
+    },
+    {
+      "epoch": 71.24645892351275,
+      "loss": 0.04363429546356201,
+      "loss_ce": 2.4674371161381714e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 1076985036,
+      "step": 6277
+    },
+    {
+      "epoch": 71.25779036827196,
+      "grad_norm": 3.8292807486423244,
+      "learning_rate": 5e-06,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 1077156980,
+      "step": 6278
+    },
+    {
+      "epoch": 71.25779036827196,
+      "loss": 0.0277617909014225,
+      "loss_ce": 2.1310916054062545e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 1077156980,
+      "step": 6278
+    },
+    {
+      "epoch": 71.26912181303116,
+      "grad_norm": 4.019725194955129,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 1077328852,
+      "step": 6279
+    },
+    {
+      "epoch": 71.26912181303116,
+      "loss": 0.05417455732822418,
+      "loss_ce": 2.1115471099619754e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 1077328852,
+      "step": 6279
+    },
+    {
+      "epoch": 71.28045325779037,
+      "grad_norm": 3.2862892866898465,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 1077498664,
+      "step": 6280
+    },
+    {
+      "epoch": 71.28045325779037,
+      "loss": 0.11885473132133484,
+      "loss_ce": 4.980745870852843e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 1077498664,
+      "step": 6280
+    },
+    {
+      "epoch": 71.29178470254958,
+      "grad_norm": 2.9884669215106188,
+      "learning_rate": 5e-06,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 1077669776,
+      "step": 6281
+    },
+    {
+      "epoch": 71.29178470254958,
+      "loss": 0.03479103744029999,
+      "loss_ce": 2.388862230873201e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1077669776,
+      "step": 6281
+    },
+    {
+      "epoch": 71.30311614730878,
+      "grad_norm": 2.472156314154236,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 1077841404,
+      "step": 6282
+    },
+    {
+      "epoch": 71.30311614730878,
+      "loss": 0.02248840592801571,
+      "loss_ce": 5.79866609768942e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 1077841404,
+      "step": 6282
+    },
+    {
+      "epoch": 71.31444759206799,
+      "grad_norm": 2.4002984493697572,
+      "learning_rate": 5e-06,
+      "loss": 0.0284,
+      "num_input_tokens_seen": 1078013028,
+      "step": 6283
+    },
+    {
+      "epoch": 71.31444759206799,
+      "loss": 0.02115441858768463,
+      "loss_ce": 2.099564517266117e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.00421142578125,
+      "loss_xval": 0.0211181640625,
+      "num_input_tokens_seen": 1078013028,
+      "step": 6283
+    },
+    {
+      "epoch": 71.3257790368272,
+      "grad_norm": 2.7621721933387104,
+      "learning_rate": 5e-06,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 1078184604,
+      "step": 6284
+    },
+    {
+      "epoch": 71.3257790368272,
+      "loss": 0.043080884963274,
+      "loss_ce": 5.110111305839382e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 1078184604,
+      "step": 6284
+    },
+    {
+      "epoch": 71.33711048158641,
+      "grad_norm": 2.806500798875873,
+      "learning_rate": 5e-06,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 1078356264,
+      "step": 6285
+    },
+    {
+      "epoch": 71.33711048158641,
+      "loss": 0.03789878264069557,
+      "loss_ce": 2.6467809220775962e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 1078356264,
+      "step": 6285
+    },
+    {
+      "epoch": 71.34844192634561,
+      "grad_norm": 2.671688308401394,
+      "learning_rate": 5e-06,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 1078528144,
+      "step": 6286
+    },
+    {
+      "epoch": 71.34844192634561,
+      "loss": 0.024138683453202248,
+      "loss_ce": 2.9797256502206437e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 1078528144,
+      "step": 6286
+    },
+    {
+      "epoch": 71.35977337110482,
+      "grad_norm": 3.112345052585686,
+      "learning_rate": 5e-06,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 1078698068,
+      "step": 6287
+    },
+    {
+      "epoch": 71.35977337110482,
+      "loss": 0.08573763817548752,
+      "loss_ce": 1.3763445167569444e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 1078698068,
+      "step": 6287
+    },
+    {
+      "epoch": 71.37110481586403,
+      "grad_norm": 3.1752111271016727,
+      "learning_rate": 5e-06,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 1078869732,
+      "step": 6288
+    },
+    {
+      "epoch": 71.37110481586403,
+      "loss": 0.051510654389858246,
+      "loss_ce": 1.2237802366144024e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 1078869732,
+      "step": 6288
+    },
+    {
+      "epoch": 71.38243626062322,
+      "grad_norm": 3.397525029542879,
+      "learning_rate": 5e-06,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 1079040728,
+      "step": 6289
+    },
+    {
+      "epoch": 71.38243626062322,
+      "loss": 0.0289616659283638,
+      "loss_ce": 0.0004887655377388,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 1079040728,
+      "step": 6289
+    },
+    {
+      "epoch": 71.39376770538243,
+      "grad_norm": 3.0440196126369647,
+      "learning_rate": 5e-06,
+      "loss": 0.063,
+      "num_input_tokens_seen": 1079212388,
+      "step": 6290
+    },
+    {
+      "epoch": 71.39376770538243,
+      "loss": 0.029235374182462692,
+      "loss_ce": 3.0051163776079193e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 1079212388,
+      "step": 6290
+    },
+    {
+      "epoch": 71.40509915014164,
+      "grad_norm": 3.155561962055097,
+      "learning_rate": 5e-06,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 1079384088,
+      "step": 6291
+    },
+    {
+      "epoch": 71.40509915014164,
+      "loss": 0.0399649553000927,
+      "loss_ce": 1.7446262063458562e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 1079384088,
+      "step": 6291
+    },
+    {
+      "epoch": 71.41643059490085,
+      "grad_norm": 3.6339630246857952,
+      "learning_rate": 5e-06,
+      "loss": 0.061,
+      "num_input_tokens_seen": 1079556112,
+      "step": 6292
+    },
+    {
+      "epoch": 71.41643059490085,
+      "loss": 0.08425532281398773,
+      "loss_ce": 1.1549636838026345e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 1079556112,
+      "step": 6292
+    },
+    {
+      "epoch": 71.42776203966005,
+      "grad_norm": 3.1195031790617396,
+      "learning_rate": 5e-06,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 1079728104,
+      "step": 6293
+    },
+    {
+      "epoch": 71.42776203966005,
+      "loss": 0.03666820749640465,
+      "loss_ce": 4.711440487881191e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1079728104,
+      "step": 6293
+    },
+    {
+      "epoch": 71.43909348441926,
+      "grad_norm": 2.52210980621293,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 1079898968,
+      "step": 6294
+    },
+    {
+      "epoch": 71.43909348441926,
+      "loss": 0.04158343747258186,
+      "loss_ce": 1.849461295932997e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 1079898968,
+      "step": 6294
+    },
+    {
+      "epoch": 71.45042492917847,
+      "grad_norm": 2.592978632615565,
+      "learning_rate": 5e-06,
+      "loss": 0.043,
+      "num_input_tokens_seen": 1080067824,
+      "step": 6295
+    },
+    {
+      "epoch": 71.45042492917847,
+      "loss": 0.05258346349000931,
+      "loss_ce": 3.219364589313045e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 1080067824,
+      "step": 6295
+    },
+    {
+      "epoch": 71.46175637393767,
+      "grad_norm": 3.4400110356131206,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1080239864,
+      "step": 6296
+    },
+    {
+      "epoch": 71.46175637393767,
+      "loss": 0.04466085135936737,
+      "loss_ce": 1.3634474271384533e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 1080239864,
+      "step": 6296
+    },
+    {
+      "epoch": 71.47308781869688,
+      "grad_norm": 3.4743187480599094,
+      "learning_rate": 5e-06,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 1080411512,
+      "step": 6297
+    },
+    {
+      "epoch": 71.47308781869688,
+      "loss": 0.06579624116420746,
+      "loss_ce": 1.560635973874014e-05,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 1080411512,
+      "step": 6297
+    },
+    {
+      "epoch": 71.48441926345609,
+      "grad_norm": 3.3283175783501693,
+      "learning_rate": 5e-06,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 1080583192,
+      "step": 6298
+    },
+    {
+      "epoch": 71.48441926345609,
+      "loss": 0.03355266526341438,
+      "loss_ce": 1.3849156857759226e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1080583192,
+      "step": 6298
+    },
+    {
+      "epoch": 71.4957507082153,
+      "grad_norm": 3.848911347362622,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1080753388,
+      "step": 6299
+    },
+    {
+      "epoch": 71.4957507082153,
+      "loss": 0.029399413615465164,
+      "loss_ce": 2.6242038074997254e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 1080753388,
+      "step": 6299
+    },
+    {
+      "epoch": 71.5070821529745,
+      "grad_norm": 7.918903226565688,
+      "learning_rate": 5e-06,
+      "loss": 0.043,
+      "num_input_tokens_seen": 1080925680,
+      "step": 6300
+    },
+    {
+      "epoch": 71.5070821529745,
+      "loss": 0.044917866587638855,
+      "loss_ce": 2.6508409064263105e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 1080925680,
+      "step": 6300
+    },
+    {
+      "epoch": 71.51841359773371,
+      "grad_norm": 3.2990587227078425,
+      "learning_rate": 5e-06,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 1081097496,
+      "step": 6301
+    },
+    {
+      "epoch": 71.51841359773371,
+      "loss": 0.12951254844665527,
+      "loss_ce": 1.1201225788681768e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 1081097496,
+      "step": 6301
+    },
+    {
+      "epoch": 71.52974504249292,
+      "grad_norm": 3.2439060139701823,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1081269304,
+      "step": 6302
+    },
+    {
+      "epoch": 71.52974504249292,
+      "loss": 0.04331529513001442,
+      "loss_ce": 4.137136784265749e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1081269304,
+      "step": 6302
+    },
+    {
+      "epoch": 71.54107648725213,
+      "grad_norm": 3.060696665030482,
+      "learning_rate": 5e-06,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 1081440292,
+      "step": 6303
+    },
+    {
+      "epoch": 71.54107648725213,
+      "loss": 0.06857012212276459,
+      "loss_ce": 8.868220902513713e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 1081440292,
+      "step": 6303
+    },
+    {
+      "epoch": 71.55240793201133,
+      "grad_norm": 3.340584040354602,
+      "learning_rate": 5e-06,
+      "loss": 0.067,
+      "num_input_tokens_seen": 1081610408,
+      "step": 6304
+    },
+    {
+      "epoch": 71.55240793201133,
+      "loss": 0.1014086976647377,
+      "loss_ce": 9.033540118252859e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 1081610408,
+      "step": 6304
+    },
+    {
+      "epoch": 71.56373937677054,
+      "grad_norm": 3.4659722910630024,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 1081782428,
+      "step": 6305
+    },
+    {
+      "epoch": 71.56373937677054,
+      "loss": 0.04283825308084488,
+      "loss_ce": 2.208867226727307e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 1081782428,
+      "step": 6305
+    },
+    {
+      "epoch": 71.57507082152975,
+      "grad_norm": 3.580784678437838,
+      "learning_rate": 5e-06,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 1081954660,
+      "step": 6306
+    },
+    {
+      "epoch": 71.57507082152975,
+      "loss": 0.07088121771812439,
+      "loss_ce": 3.465483314357698e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 1081954660,
+      "step": 6306
+    },
+    {
+      "epoch": 71.58640226628896,
+      "grad_norm": 3.5295961317650923,
+      "learning_rate": 5e-06,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 1082124756,
+      "step": 6307
+    },
+    {
+      "epoch": 71.58640226628896,
+      "loss": 0.036429621279239655,
+      "loss_ce": 2.2147392883198336e-05,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 1082124756,
+      "step": 6307
+    },
+    {
+      "epoch": 71.59773371104816,
+      "grad_norm": 3.700651667499769,
+      "learning_rate": 5e-06,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 1082295712,
+      "step": 6308
+    },
+    {
+      "epoch": 71.59773371104816,
+      "loss": 0.04259459674358368,
+      "loss_ce": 2.257281812489964e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 1082295712,
+      "step": 6308
+    },
+    {
+      "epoch": 71.60906515580737,
+      "grad_norm": 3.2547996891083355,
+      "learning_rate": 5e-06,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 1082467384,
+      "step": 6309
+    },
+    {
+      "epoch": 71.60906515580737,
+      "loss": 0.03559117391705513,
+      "loss_ce": 2.2936623281566426e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 1082467384,
+      "step": 6309
+    },
+    {
+      "epoch": 71.62039660056658,
+      "grad_norm": 3.0060754687615217,
+      "learning_rate": 5e-06,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 1082639748,
+      "step": 6310
+    },
+    {
+      "epoch": 71.62039660056658,
+      "loss": 0.04740023985505104,
+      "loss_ce": 2.1699372155126184e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 1082639748,
+      "step": 6310
+    },
+    {
+      "epoch": 71.63172804532577,
+      "grad_norm": 3.1697521022581125,
+      "learning_rate": 5e-06,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 1082811936,
+      "step": 6311
+    },
+    {
+      "epoch": 71.63172804532577,
+      "loss": 0.07100522518157959,
+      "loss_ce": 2.1335294150048867e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 1082811936,
+      "step": 6311
+    },
+    {
+      "epoch": 71.64305949008498,
+      "grad_norm": 3.2591715848672007,
+      "learning_rate": 5e-06,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 1082982144,
+      "step": 6312
+    },
+    {
+      "epoch": 71.64305949008498,
+      "loss": 0.02785702608525753,
+      "loss_ce": 1.7365549865644425e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 1082982144,
+      "step": 6312
+    },
+    {
+      "epoch": 71.65439093484419,
+      "grad_norm": 3.3285965882845083,
+      "learning_rate": 5e-06,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 1083152036,
+      "step": 6313
+    },
+    {
+      "epoch": 71.65439093484419,
+      "loss": 0.029422510415315628,
+      "loss_ce": 3.4084452636307105e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 1083152036,
+      "step": 6313
+    },
+    {
+      "epoch": 71.6657223796034,
+      "grad_norm": 3.2564390359104487,
+      "learning_rate": 5e-06,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 1083323460,
+      "step": 6314
+    },
+    {
+      "epoch": 71.6657223796034,
+      "loss": 0.038522012531757355,
+      "loss_ce": 8.828123100101948e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1083323460,
+      "step": 6314
+    },
+    {
+      "epoch": 71.6770538243626,
+      "grad_norm": 2.7396137946480557,
+      "learning_rate": 5e-06,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 1083495424,
+      "step": 6315
+    },
+    {
+      "epoch": 71.6770538243626,
+      "loss": 0.05260499194264412,
+      "loss_ce": 3.84659506380558e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 1083495424,
+      "step": 6315
+    },
+    {
+      "epoch": 71.68838526912181,
+      "grad_norm": 2.266224632014876,
+      "learning_rate": 5e-06,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 1083667124,
+      "step": 6316
+    },
+    {
+      "epoch": 71.68838526912181,
+      "loss": 0.030978474766016006,
+      "loss_ce": 2.6021647499874234e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1083667124,
+      "step": 6316
+    },
+    {
+      "epoch": 71.69971671388102,
+      "grad_norm": 2.4898532380264133,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 1083838976,
+      "step": 6317
+    },
+    {
+      "epoch": 71.69971671388102,
+      "loss": 0.024016916751861572,
+      "loss_ce": 2.2471380361821502e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 1083838976,
+      "step": 6317
+    },
+    {
+      "epoch": 71.71104815864022,
+      "grad_norm": 3.105282815254955,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 1084010628,
+      "step": 6318
+    },
+    {
+      "epoch": 71.71104815864022,
+      "loss": 0.029156381264328957,
+      "loss_ce": 1.2093717487005051e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 1084010628,
+      "step": 6318
+    },
+    {
+      "epoch": 71.72237960339943,
+      "grad_norm": 3.18531159092337,
+      "learning_rate": 5e-06,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 1084182672,
+      "step": 6319
+    },
+    {
+      "epoch": 71.72237960339943,
+      "loss": 0.023192856460809708,
+      "loss_ce": 3.0014671210665256e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 1084182672,
+      "step": 6319
+    },
+    {
+      "epoch": 71.73371104815864,
+      "grad_norm": 3.3128075410602986,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 1084354528,
+      "step": 6320
+    },
+    {
+      "epoch": 71.73371104815864,
+      "loss": 0.08003723621368408,
+      "loss_ce": 1.2516074093582574e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 1084354528,
+      "step": 6320
+    },
+    {
+      "epoch": 71.74504249291785,
+      "grad_norm": 3.3971306092465654,
+      "learning_rate": 5e-06,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 1084526304,
+      "step": 6321
+    },
+    {
+      "epoch": 71.74504249291785,
+      "loss": 0.04324670508503914,
+      "loss_ce": 1.855498157965485e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1084526304,
+      "step": 6321
+    },
+    {
+      "epoch": 71.75637393767705,
+      "grad_norm": 3.389335234027946,
+      "learning_rate": 5e-06,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 1084696772,
+      "step": 6322
+    },
+    {
+      "epoch": 71.75637393767705,
+      "loss": 0.03128332644701004,
+      "loss_ce": 1.8069695215672255e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1084696772,
+      "step": 6322
+    },
+    {
+      "epoch": 71.76770538243626,
+      "grad_norm": 4.15401820229542,
+      "learning_rate": 5e-06,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 1084868796,
+      "step": 6323
+    },
+    {
+      "epoch": 71.76770538243626,
+      "loss": 0.04818715155124664,
+      "loss_ce": 1.5153706044657156e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 1084868796,
+      "step": 6323
+    },
+    {
+      "epoch": 71.77903682719547,
+      "grad_norm": 3.825831923996868,
+      "learning_rate": 5e-06,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 1085040796,
+      "step": 6324
+    },
+    {
+      "epoch": 71.77903682719547,
+      "loss": 0.08951297402381897,
+      "loss_ce": 3.5430595744401217e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 1085040796,
+      "step": 6324
+    },
+    {
+      "epoch": 71.79036827195468,
+      "grad_norm": 3.210139623039176,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 1085212396,
+      "step": 6325
+    },
+    {
+      "epoch": 71.79036827195468,
+      "loss": 0.0444817915558815,
+      "loss_ce": 6.345735164359212e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 1085212396,
+      "step": 6325
+    },
+    {
+      "epoch": 71.80169971671388,
+      "grad_norm": 2.9837470222283824,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 1085380304,
+      "step": 6326
+    },
+    {
+      "epoch": 71.80169971671388,
+      "loss": 0.058926261961460114,
+      "loss_ce": 4.259422712493688e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 1085380304,
+      "step": 6326
+    },
+    {
+      "epoch": 71.81303116147309,
+      "grad_norm": 3.0765421742915606,
+      "learning_rate": 5e-06,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 1085552212,
+      "step": 6327
+    },
+    {
+      "epoch": 71.81303116147309,
+      "loss": 0.03482535108923912,
+      "loss_ce": 2.0052239051437937e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1085552212,
+      "step": 6327
+    },
+    {
+      "epoch": 71.8243626062323,
+      "grad_norm": 3.1333489212861947,
+      "learning_rate": 5e-06,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 1085723968,
+      "step": 6328
+    },
+    {
+      "epoch": 71.8243626062323,
+      "loss": 0.031688421964645386,
+      "loss_ce": 1.1178676686540712e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 1085723968,
+      "step": 6328
+    },
+    {
+      "epoch": 71.8356940509915,
+      "grad_norm": 3.5824008093267814,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1085894568,
+      "step": 6329
+    },
+    {
+      "epoch": 71.8356940509915,
+      "loss": 0.03835016116499901,
+      "loss_ce": 2.008280534937512e-05,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1085894568,
+      "step": 6329
+    },
+    {
+      "epoch": 71.84702549575071,
+      "grad_norm": 3.9530717543199807,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 1086066124,
+      "step": 6330
+    },
+    {
+      "epoch": 71.84702549575071,
+      "loss": 0.034444116055965424,
+      "loss_ce": 2.02841947611887e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 1086066124,
+      "step": 6330
+    },
+    {
+      "epoch": 71.85835694050992,
+      "grad_norm": 3.0279630139462412,
+      "learning_rate": 5e-06,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 1086238232,
+      "step": 6331
+    },
+    {
+      "epoch": 71.85835694050992,
+      "loss": 0.02691538631916046,
+      "loss_ce": 1.4142431609798223e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 1086238232,
+      "step": 6331
+    },
+    {
+      "epoch": 71.86968838526913,
+      "grad_norm": 2.759156970980906,
+      "learning_rate": 5e-06,
+      "loss": 0.035,
+      "num_input_tokens_seen": 1086409028,
+      "step": 6332
+    },
+    {
+      "epoch": 71.86968838526913,
+      "loss": 0.040620580315589905,
+      "loss_ce": 1.6944686649367213e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1086409028,
+      "step": 6332
+    },
+    {
+      "epoch": 71.88101983002832,
+      "grad_norm": 3.123038518789632,
+      "learning_rate": 5e-06,
+      "loss": 0.0242,
+      "num_input_tokens_seen": 1086579040,
+      "step": 6333
+    },
+    {
+      "epoch": 71.88101983002832,
+      "loss": 0.027104049921035767,
+      "loss_ce": 1.9699866243172437e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1086579040,
+      "step": 6333
+    },
+    {
+      "epoch": 71.89235127478753,
+      "grad_norm": 3.0824204519206835,
+      "learning_rate": 5e-06,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 1086750344,
+      "step": 6334
+    },
+    {
+      "epoch": 71.89235127478753,
+      "loss": 0.04207644611597061,
+      "loss_ce": 7.965642907947768e-06,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1086750344,
+      "step": 6334
+    },
+    {
+      "epoch": 71.90368271954674,
+      "grad_norm": 3.155146531260981,
+      "learning_rate": 5e-06,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 1086921492,
+      "step": 6335
+    },
+    {
+      "epoch": 71.90368271954674,
+      "loss": 0.029619436711072922,
+      "loss_ce": 5.5533313570776954e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 1086921492,
+      "step": 6335
+    },
+    {
+      "epoch": 71.91501416430594,
+      "grad_norm": 3.419832680079542,
+      "learning_rate": 5e-06,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 1087092360,
+      "step": 6336
+    },
+    {
+      "epoch": 71.91501416430594,
+      "loss": 0.026729252189397812,
+      "loss_ce": 1.1113129403383937e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 1087092360,
+      "step": 6336
+    },
+    {
+      "epoch": 71.92634560906515,
+      "grad_norm": 2.609690859679022,
+      "learning_rate": 5e-06,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 1087263292,
+      "step": 6337
+    },
+    {
+      "epoch": 71.92634560906515,
+      "loss": 0.024927429854869843,
+      "loss_ce": 4.7973782784538344e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 1087263292,
+      "step": 6337
+    },
+    {
+      "epoch": 71.93767705382436,
+      "grad_norm": 6.157433000399018,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 1087434068,
+      "step": 6338
+    },
+    {
+      "epoch": 71.93767705382436,
+      "loss": 0.03840063512325287,
+      "loss_ce": 3.2410272979177535e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1087434068,
+      "step": 6338
+    },
+    {
+      "epoch": 71.94900849858357,
+      "grad_norm": 3.7948587291065867,
+      "learning_rate": 5e-06,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 1087605904,
+      "step": 6339
+    },
+    {
+      "epoch": 71.94900849858357,
+      "loss": 0.12592989206314087,
+      "loss_ce": 1.4367837138706818e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 1087605904,
+      "step": 6339
+    },
+    {
+      "epoch": 71.96033994334277,
+      "grad_norm": 2.9600195851625672,
+      "learning_rate": 5e-06,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 1087777780,
+      "step": 6340
+    },
+    {
+      "epoch": 71.96033994334277,
+      "loss": 0.08376624435186386,
+      "loss_ce": 4.1264436731580645e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 1087777780,
+      "step": 6340
+    },
+    {
+      "epoch": 71.97167138810198,
+      "grad_norm": 3.0581412759152875,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 1087949588,
+      "step": 6341
+    },
+    {
+      "epoch": 71.97167138810198,
+      "loss": 0.03042149543762207,
+      "loss_ce": 2.598770151962526e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 1087949588,
+      "step": 6341
+    },
+    {
+      "epoch": 71.98300283286119,
+      "grad_norm": 3.0108472863819835,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 1088121404,
+      "step": 6342
+    },
+    {
+      "epoch": 71.98300283286119,
+      "loss": 0.039963483810424805,
+      "loss_ce": 3.123074566246942e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 1088121404,
+      "step": 6342
+    },
+    {
+      "epoch": 71.9943342776204,
+      "grad_norm": 2.9866481980623267,
+      "learning_rate": 5e-06,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 1088290780,
+      "step": 6343
+    },
+    {
+      "epoch": 71.9943342776204,
+      "loss": 0.07404462993144989,
+      "loss_ce": 8.989324669528287e-06,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 1088290780,
+      "step": 6343
+    },
+    {
+      "epoch": 71.9943342776204,
+      "loss": 0.03748292475938797,
+      "loss_ce": 0.00014466721040662378,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 1088376584,
+      "step": 6343
+    },
+    {
+      "epoch": 72.0056657223796,
+      "grad_norm": 2.6230742453159093,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 1088461620,
+      "step": 6344
+    },
+    {
+      "epoch": 72.0056657223796,
+      "loss": 0.021838389337062836,
+      "loss_ce": 2.5949284463422373e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.0218505859375,
+      "num_input_tokens_seen": 1088461620,
+      "step": 6344
+    },
+    {
+      "epoch": 72.01699716713881,
+      "grad_norm": 2.5743993105138983,
+      "learning_rate": 5e-06,
+      "loss": 0.034,
+      "num_input_tokens_seen": 1088632324,
+      "step": 6345
+    },
+    {
+      "epoch": 72.01699716713881,
+      "loss": 0.02753208950161934,
+      "loss_ce": 1.286371480091475e-05,
+      "loss_iou": 0.06201171875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 1088632324,
+      "step": 6345
+    },
+    {
+      "epoch": 72.02832861189802,
+      "grad_norm": 4.0657011342861855,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 1088804568,
+      "step": 6346
+    },
+    {
+      "epoch": 72.02832861189802,
+      "loss": 0.061970219016075134,
+      "loss_ce": 1.95354332390707e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 1088804568,
+      "step": 6346
+    },
+    {
+      "epoch": 72.03966005665723,
+      "grad_norm": 3.697535219229695,
+      "learning_rate": 5e-06,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 1088975984,
+      "step": 6347
+    },
+    {
+      "epoch": 72.03966005665723,
+      "loss": 0.05731949955224991,
+      "loss_ce": 2.274712096550502e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 1088975984,
+      "step": 6347
+    },
+    {
+      "epoch": 72.05099150141643,
+      "grad_norm": 3.057934850127661,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 1089148072,
+      "step": 6348
+    },
+    {
+      "epoch": 72.05099150141643,
+      "loss": 0.035268183797597885,
+      "loss_ce": 1.2753221199091058e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1089148072,
+      "step": 6348
+    },
+    {
+      "epoch": 72.06232294617564,
+      "grad_norm": 3.1599802504278465,
+      "learning_rate": 5e-06,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 1089319728,
+      "step": 6349
+    },
+    {
+      "epoch": 72.06232294617564,
+      "loss": 0.058867134153842926,
+      "loss_ce": 2.924150430771988e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 1089319728,
+      "step": 6349
+    },
+    {
+      "epoch": 72.07365439093485,
+      "grad_norm": 3.0273473143408074,
+      "learning_rate": 5e-06,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 1089491660,
+      "step": 6350
+    },
+    {
+      "epoch": 72.07365439093485,
+      "loss": 0.027279410511255264,
+      "loss_ce": 1.1955982699873857e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 1089491660,
+      "step": 6350
+    },
+    {
+      "epoch": 72.08498583569406,
+      "grad_norm": 3.073425291741844,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 1089663228,
+      "step": 6351
+    },
+    {
+      "epoch": 72.08498583569406,
+      "loss": 0.053060173988342285,
+      "loss_ce": 6.640057836193591e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 1089663228,
+      "step": 6351
+    },
+    {
+      "epoch": 72.09631728045326,
+      "grad_norm": 3.7254015656290926,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 1089834216,
+      "step": 6352
+    },
+    {
+      "epoch": 72.09631728045326,
+      "loss": 0.03060598112642765,
+      "loss_ce": 2.7369242161512375e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1089834216,
+      "step": 6352
+    },
+    {
+      "epoch": 72.10764872521247,
+      "grad_norm": 3.1308690975300855,
+      "learning_rate": 5e-06,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 1090005696,
+      "step": 6353
+    },
+    {
+      "epoch": 72.10764872521247,
+      "loss": 0.025884132832288742,
+      "loss_ce": 2.0485880668275058e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 1090005696,
+      "step": 6353
+    },
+    {
+      "epoch": 72.11898016997168,
+      "grad_norm": 3.1659734315883097,
+      "learning_rate": 5e-06,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 1090177840,
+      "step": 6354
+    },
+    {
+      "epoch": 72.11898016997168,
+      "loss": 0.049676500260829926,
+      "loss_ce": 3.965653741033748e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 1090177840,
+      "step": 6354
+    },
+    {
+      "epoch": 72.13031161473087,
+      "grad_norm": 3.0206836599688476,
+      "learning_rate": 5e-06,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 1090348972,
+      "step": 6355
+    },
+    {
+      "epoch": 72.13031161473087,
+      "loss": 0.022258980199694633,
+      "loss_ce": 1.1666428690659814e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 1090348972,
+      "step": 6355
+    },
+    {
+      "epoch": 72.14164305949008,
+      "grad_norm": 2.8412842413214427,
+      "learning_rate": 5e-06,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 1090520284,
+      "step": 6356
+    },
+    {
+      "epoch": 72.14164305949008,
+      "loss": 0.08763731271028519,
+      "loss_ce": 0.00028074689907953143,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 1090520284,
+      "step": 6356
+    },
+    {
+      "epoch": 72.15297450424929,
+      "grad_norm": 3.9974403631731033,
+      "learning_rate": 5e-06,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 1090692048,
+      "step": 6357
+    },
+    {
+      "epoch": 72.15297450424929,
+      "loss": 0.100965216755867,
+      "loss_ce": 1.307215279666707e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 1090692048,
+      "step": 6357
+    },
+    {
+      "epoch": 72.1643059490085,
+      "grad_norm": 3.5454203929180657,
+      "learning_rate": 5e-06,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 1090862808,
+      "step": 6358
+    },
+    {
+      "epoch": 72.1643059490085,
+      "loss": 0.03559213876724243,
+      "loss_ce": 2.390237023064401e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 1090862808,
+      "step": 6358
+    },
+    {
+      "epoch": 72.1756373937677,
+      "grad_norm": 3.8905664284263115,
+      "learning_rate": 5e-06,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 1091034328,
+      "step": 6359
+    },
+    {
+      "epoch": 72.1756373937677,
+      "loss": 0.08154311776161194,
+      "loss_ce": 1.540574521641247e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 1091034328,
+      "step": 6359
+    },
+    {
+      "epoch": 72.18696883852691,
+      "grad_norm": 3.601051258444796,
+      "learning_rate": 5e-06,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 1091206096,
+      "step": 6360
+    },
+    {
+      "epoch": 72.18696883852691,
+      "loss": 0.029772158712148666,
+      "loss_ce": 1.7518334061605856e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 1091206096,
+      "step": 6360
+    },
+    {
+      "epoch": 72.19830028328612,
+      "grad_norm": 4.007055157856261,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 1091378120,
+      "step": 6361
+    },
+    {
+      "epoch": 72.19830028328612,
+      "loss": 0.043407365679740906,
+      "loss_ce": 5.714629151043482e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 1091378120,
+      "step": 6361
+    },
+    {
+      "epoch": 72.20963172804532,
+      "grad_norm": 2.8065246633678336,
+      "learning_rate": 5e-06,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 1091549844,
+      "step": 6362
+    },
+    {
+      "epoch": 72.20963172804532,
+      "loss": 0.027665428817272186,
+      "loss_ce": 4.702120349975303e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 1091549844,
+      "step": 6362
+    },
+    {
+      "epoch": 72.22096317280453,
+      "grad_norm": 2.691046088896478,
+      "learning_rate": 5e-06,
+      "loss": 0.0267,
+      "num_input_tokens_seen": 1091721780,
+      "step": 6363
+    },
+    {
+      "epoch": 72.22096317280453,
+      "loss": 0.029430441558361053,
+      "loss_ce": 2.6752926714834757e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 1091721780,
+      "step": 6363
+    },
+    {
+      "epoch": 72.23229461756374,
+      "grad_norm": 3.1806246620934266,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 1091893040,
+      "step": 6364
+    },
+    {
+      "epoch": 72.23229461756374,
+      "loss": 0.06900182366371155,
+      "loss_ce": 1.683681330177933e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 1091893040,
+      "step": 6364
+    },
+    {
+      "epoch": 72.24362606232295,
+      "grad_norm": 2.8295989546131324,
+      "learning_rate": 5e-06,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 1092064772,
+      "step": 6365
+    },
+    {
+      "epoch": 72.24362606232295,
+      "loss": 0.027438584715127945,
+      "loss_ce": 2.617051541164983e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 1092064772,
+      "step": 6365
+    },
+    {
+      "epoch": 72.25495750708215,
+      "grad_norm": 2.788002037871266,
+      "learning_rate": 5e-06,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 1092236716,
+      "step": 6366
+    },
+    {
+      "epoch": 72.25495750708215,
+      "loss": 0.04069644585251808,
+      "loss_ce": 1.6512391084688716e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1092236716,
+      "step": 6366
+    },
+    {
+      "epoch": 72.26628895184136,
+      "grad_norm": 2.593735249886273,
+      "learning_rate": 5e-06,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 1092408476,
+      "step": 6367
+    },
+    {
+      "epoch": 72.26628895184136,
+      "loss": 0.02480863593518734,
+      "loss_ce": 1.3103700439387467e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 1092408476,
+      "step": 6367
+    },
+    {
+      "epoch": 72.27762039660057,
+      "grad_norm": 2.994600128926469,
+      "learning_rate": 5e-06,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 1092579344,
+      "step": 6368
+    },
+    {
+      "epoch": 72.27762039660057,
+      "loss": 0.13209834694862366,
+      "loss_ce": 1.8278251445735805e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 1092579344,
+      "step": 6368
+    },
+    {
+      "epoch": 72.28895184135978,
+      "grad_norm": 3.356586619114031,
+      "learning_rate": 5e-06,
+      "loss": 0.061,
+      "num_input_tokens_seen": 1092751416,
+      "step": 6369
+    },
+    {
+      "epoch": 72.28895184135978,
+      "loss": 0.05163809284567833,
+      "loss_ce": 1.7609203496249393e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 1092751416,
+      "step": 6369
+    },
+    {
+      "epoch": 72.30028328611898,
+      "grad_norm": 3.3015117766675948,
+      "learning_rate": 5e-06,
+      "loss": 0.0295,
+      "num_input_tokens_seen": 1092922696,
+      "step": 6370
+    },
+    {
+      "epoch": 72.30028328611898,
+      "loss": 0.03412819281220436,
+      "loss_ce": 2.4800161554594524e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1092922696,
+      "step": 6370
+    },
+    {
+      "epoch": 72.31161473087819,
+      "grad_norm": 3.163256542833529,
+      "learning_rate": 5e-06,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 1093094364,
+      "step": 6371
+    },
+    {
+      "epoch": 72.31161473087819,
+      "loss": 0.040805984288454056,
+      "loss_ce": 1.924399839481339e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1093094364,
+      "step": 6371
+    },
+    {
+      "epoch": 72.3229461756374,
+      "grad_norm": 2.966859222890234,
+      "learning_rate": 5e-06,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 1093265244,
+      "step": 6372
+    },
+    {
+      "epoch": 72.3229461756374,
+      "loss": 0.030220340937376022,
+      "loss_ce": 7.93865092418855e-06,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1093265244,
+      "step": 6372
+    },
+    {
+      "epoch": 72.3342776203966,
+      "grad_norm": 2.881963139230349,
+      "learning_rate": 5e-06,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 1093436892,
+      "step": 6373
+    },
+    {
+      "epoch": 72.3342776203966,
+      "loss": 0.02472321316599846,
+      "loss_ce": 3.449180439929478e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 1093436892,
+      "step": 6373
+    },
+    {
+      "epoch": 72.34560906515581,
+      "grad_norm": 4.050290191683477,
+      "learning_rate": 5e-06,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 1093606436,
+      "step": 6374
+    },
+    {
+      "epoch": 72.34560906515581,
+      "loss": 0.09272155910730362,
+      "loss_ce": 9.157141903415322e-06,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 1093606436,
+      "step": 6374
+    },
+    {
+      "epoch": 72.35694050991502,
+      "grad_norm": 3.673134995852733,
+      "learning_rate": 5e-06,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 1093777372,
+      "step": 6375
+    },
+    {
+      "epoch": 72.35694050991502,
+      "loss": 0.03644733875989914,
+      "loss_ce": 3.986688534496352e-05,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 1093777372,
+      "step": 6375
+    },
+    {
+      "epoch": 72.36827195467423,
+      "grad_norm": 3.026457567967279,
+      "learning_rate": 5e-06,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 1093948476,
+      "step": 6376
+    },
+    {
+      "epoch": 72.36827195467423,
+      "loss": 0.036039236932992935,
+      "loss_ce": 1.3235137885203585e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 1093948476,
+      "step": 6376
+    },
+    {
+      "epoch": 72.37960339943342,
+      "grad_norm": 3.2663509329947313,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 1094120208,
+      "step": 6377
+    },
+    {
+      "epoch": 72.37960339943342,
+      "loss": 0.040823910385370255,
+      "loss_ce": 6.649356237176107e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1094120208,
+      "step": 6377
+    },
+    {
+      "epoch": 72.39093484419263,
+      "grad_norm": 3.7452499230689535,
+      "learning_rate": 5e-06,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 1094290888,
+      "step": 6378
+    },
+    {
+      "epoch": 72.39093484419263,
+      "loss": 0.03225569427013397,
+      "loss_ce": 2.9134240321582183e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1094290888,
+      "step": 6378
+    },
+    {
+      "epoch": 72.40226628895184,
+      "grad_norm": 3.6645911498181,
+      "learning_rate": 5e-06,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 1094462640,
+      "step": 6379
+    },
+    {
+      "epoch": 72.40226628895184,
+      "loss": 0.02852441370487213,
+      "loss_ce": 2.099523408105597e-05,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 1094462640,
+      "step": 6379
+    },
+    {
+      "epoch": 72.41359773371104,
+      "grad_norm": 3.2392320312589855,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 1094634252,
+      "step": 6380
+    },
+    {
+      "epoch": 72.41359773371104,
+      "loss": 0.05362439528107643,
+      "loss_ce": 2.0267836589482613e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 1094634252,
+      "step": 6380
+    },
+    {
+      "epoch": 72.42492917847025,
+      "grad_norm": 2.637737527351458,
+      "learning_rate": 5e-06,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 1094803536,
+      "step": 6381
+    },
+    {
+      "epoch": 72.42492917847025,
+      "loss": 0.028267571702599525,
+      "loss_ce": 5.406998752732761e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 1094803536,
+      "step": 6381
+    },
+    {
+      "epoch": 72.43626062322946,
+      "grad_norm": 3.0048239657208,
+      "learning_rate": 5e-06,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 1094975200,
+      "step": 6382
+    },
+    {
+      "epoch": 72.43626062322946,
+      "loss": 0.03899305313825607,
+      "loss_ce": 6.84635597281158e-06,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 1094975200,
+      "step": 6382
+    },
+    {
+      "epoch": 72.44759206798867,
+      "grad_norm": 2.7316506971804104,
+      "learning_rate": 5e-06,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 1095147272,
+      "step": 6383
+    },
+    {
+      "epoch": 72.44759206798867,
+      "loss": 0.07629137486219406,
+      "loss_ce": 2.795118598442059e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 1095147272,
+      "step": 6383
+    },
+    {
+      "epoch": 72.45892351274787,
+      "grad_norm": 3.363695395337474,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 1095318468,
+      "step": 6384
+    },
+    {
+      "epoch": 72.45892351274787,
+      "loss": 0.036173634231090546,
+      "loss_ce": 1.030434032145422e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 1095318468,
+      "step": 6384
+    },
+    {
+      "epoch": 72.47025495750708,
+      "grad_norm": 3.096272130763295,
+      "learning_rate": 5e-06,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 1095489368,
+      "step": 6385
+    },
+    {
+      "epoch": 72.47025495750708,
+      "loss": 0.025361312553286552,
+      "loss_ce": 0.00011564575834199786,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 1095489368,
+      "step": 6385
+    },
+    {
+      "epoch": 72.48158640226629,
+      "grad_norm": 2.8884396188120927,
+      "learning_rate": 5e-06,
+      "loss": 0.077,
+      "num_input_tokens_seen": 1095660904,
+      "step": 6386
+    },
+    {
+      "epoch": 72.48158640226629,
+      "loss": 0.040947325527668,
+      "loss_ce": 0.00013006487279199064,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1095660904,
+      "step": 6386
+    },
+    {
+      "epoch": 72.4929178470255,
+      "grad_norm": 3.23385595715523,
+      "learning_rate": 5e-06,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 1095832840,
+      "step": 6387
+    },
+    {
+      "epoch": 72.4929178470255,
+      "loss": 0.02521863952279091,
+      "loss_ce": 1.1119105693069287e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 1095832840,
+      "step": 6387
+    },
+    {
+      "epoch": 72.5042492917847,
+      "grad_norm": 3.8777402460745356,
+      "learning_rate": 5e-06,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 1096004648,
+      "step": 6388
+    },
+    {
+      "epoch": 72.5042492917847,
+      "loss": 0.03499479219317436,
+      "loss_ce": 3.6908062611473724e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1096004648,
+      "step": 6388
+    },
+    {
+      "epoch": 72.51558073654391,
+      "grad_norm": 3.3672274204661914,
+      "learning_rate": 5e-06,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 1096176320,
+      "step": 6389
+    },
+    {
+      "epoch": 72.51558073654391,
+      "loss": 0.07335026562213898,
+      "loss_ce": 1.6526042600162327e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 1096176320,
+      "step": 6389
+    },
+    {
+      "epoch": 72.52691218130312,
+      "grad_norm": 3.0684152286121025,
+      "learning_rate": 5e-06,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 1096346712,
+      "step": 6390
+    },
+    {
+      "epoch": 72.52691218130312,
+      "loss": 0.05328311771154404,
+      "loss_ce": 2.994591704919003e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 1096346712,
+      "step": 6390
+    },
+    {
+      "epoch": 72.53824362606233,
+      "grad_norm": 2.809210817791173,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 1096518388,
+      "step": 6391
+    },
+    {
+      "epoch": 72.53824362606233,
+      "loss": 0.038925543427467346,
+      "loss_ce": 3.0889827030478045e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 1096518388,
+      "step": 6391
+    },
+    {
+      "epoch": 72.54957507082153,
+      "grad_norm": 2.70614106490957,
+      "learning_rate": 5e-06,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 1096690044,
+      "step": 6392
+    },
+    {
+      "epoch": 72.54957507082153,
+      "loss": 0.03132588788866997,
+      "loss_ce": 1.4854080291115679e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1096690044,
+      "step": 6392
+    },
+    {
+      "epoch": 72.56090651558074,
+      "grad_norm": 3.662964555163521,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 1096860176,
+      "step": 6393
+    },
+    {
+      "epoch": 72.56090651558074,
+      "loss": 0.03108474239706993,
+      "loss_ce": 1.7848155039246194e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1096860176,
+      "step": 6393
+    },
+    {
+      "epoch": 72.57223796033995,
+      "grad_norm": 2.9861319328524516,
+      "learning_rate": 5e-06,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 1097030844,
+      "step": 6394
+    },
+    {
+      "epoch": 72.57223796033995,
+      "loss": 0.059183087199926376,
+      "loss_ce": 2.4766082788119093e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 1097030844,
+      "step": 6394
+    },
+    {
+      "epoch": 72.58356940509915,
+      "grad_norm": 3.126706457541378,
+      "learning_rate": 5e-06,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 1097202252,
+      "step": 6395
+    },
+    {
+      "epoch": 72.58356940509915,
+      "loss": 0.02511187270283699,
+      "loss_ce": 3.405200186534785e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 1097202252,
+      "step": 6395
+    },
+    {
+      "epoch": 72.59490084985836,
+      "grad_norm": 3.3284161292466363,
+      "learning_rate": 5e-06,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 1097374284,
+      "step": 6396
+    },
+    {
+      "epoch": 72.59490084985836,
+      "loss": 0.12613177299499512,
+      "loss_ce": 3.313073830213398e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 1097374284,
+      "step": 6396
+    },
+    {
+      "epoch": 72.60623229461757,
+      "grad_norm": 3.3783026259225273,
+      "learning_rate": 5e-06,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 1097544512,
+      "step": 6397
+    },
+    {
+      "epoch": 72.60623229461757,
+      "loss": 0.03002443164587021,
+      "loss_ce": 1.0393116099294275e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1097544512,
+      "step": 6397
+    },
+    {
+      "epoch": 72.61756373937678,
+      "grad_norm": 3.3631935856148014,
+      "learning_rate": 5e-06,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 1097715356,
+      "step": 6398
+    },
+    {
+      "epoch": 72.61756373937678,
+      "loss": 0.0634152889251709,
+      "loss_ce": 4.5536198740592226e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 1097715356,
+      "step": 6398
+    },
+    {
+      "epoch": 72.62889518413597,
+      "grad_norm": 2.880927828933064,
+      "learning_rate": 5e-06,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 1097885468,
+      "step": 6399
+    },
+    {
+      "epoch": 72.62889518413597,
+      "loss": 0.025341372936964035,
+      "loss_ce": 1.1783835361711681e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 1097885468,
+      "step": 6399
+    },
+    {
+      "epoch": 72.64022662889518,
+      "grad_norm": 2.705033329201722,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 1098057020,
+      "step": 6400
+    },
+    {
+      "epoch": 72.64022662889518,
+      "loss": 0.05277795344591141,
+      "loss_ce": 1.3056976058578584e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 1098057020,
+      "step": 6400
+    },
+    {
+      "epoch": 72.65155807365439,
+      "grad_norm": 3.162785202256377,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 1098227424,
+      "step": 6401
+    },
+    {
+      "epoch": 72.65155807365439,
+      "loss": 0.027142629027366638,
+      "loss_ce": 2.77605122391833e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1098227424,
+      "step": 6401
+    },
+    {
+      "epoch": 72.66288951841359,
+      "grad_norm": 3.156325189735493,
+      "learning_rate": 5e-06,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 1098398900,
+      "step": 6402
+    },
+    {
+      "epoch": 72.66288951841359,
+      "loss": 0.03227756917476654,
+      "loss_ce": 1.2857719411840662e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1098398900,
+      "step": 6402
+    },
+    {
+      "epoch": 72.6742209631728,
+      "grad_norm": 2.5847939039886376,
+      "learning_rate": 5e-06,
+      "loss": 0.0271,
+      "num_input_tokens_seen": 1098570032,
+      "step": 6403
+    },
+    {
+      "epoch": 72.6742209631728,
+      "loss": 0.020491670817136765,
+      "loss_ce": 1.4374811144080013e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.00408935546875,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 1098570032,
+      "step": 6403
+    },
+    {
+      "epoch": 72.68555240793201,
+      "grad_norm": 2.8781350671666925,
+      "learning_rate": 5e-06,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 1098741204,
+      "step": 6404
+    },
+    {
+      "epoch": 72.68555240793201,
+      "loss": 0.0322367399930954,
+      "loss_ce": 2.5438566808588803e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1098741204,
+      "step": 6404
+    },
+    {
+      "epoch": 72.69688385269122,
+      "grad_norm": 3.0884883214756598,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 1098913076,
+      "step": 6405
+    },
+    {
+      "epoch": 72.69688385269122,
+      "loss": 0.07983721047639847,
+      "loss_ce": 3.374271545908414e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 1098913076,
+      "step": 6405
+    },
+    {
+      "epoch": 72.70821529745042,
+      "grad_norm": 3.5041441826181363,
+      "learning_rate": 5e-06,
+      "loss": 0.057,
+      "num_input_tokens_seen": 1099084620,
+      "step": 6406
+    },
+    {
+      "epoch": 72.70821529745042,
+      "loss": 0.0833563581109047,
+      "loss_ce": 1.2848513506469317e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 1099084620,
+      "step": 6406
+    },
+    {
+      "epoch": 72.71954674220963,
+      "grad_norm": 3.381943111102298,
+      "learning_rate": 5e-06,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 1099254128,
+      "step": 6407
+    },
+    {
+      "epoch": 72.71954674220963,
+      "loss": 0.04095417633652687,
+      "loss_ce": 1.4843945791653823e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1099254128,
+      "step": 6407
+    },
+    {
+      "epoch": 72.73087818696884,
+      "grad_norm": 3.0092830599290936,
+      "learning_rate": 5e-06,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 1099424932,
+      "step": 6408
+    },
+    {
+      "epoch": 72.73087818696884,
+      "loss": 0.03320252522826195,
+      "loss_ce": 4.5176191633800045e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1099424932,
+      "step": 6408
+    },
+    {
+      "epoch": 72.74220963172804,
+      "grad_norm": 3.048968242766737,
+      "learning_rate": 5e-06,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 1099597220,
+      "step": 6409
+    },
+    {
+      "epoch": 72.74220963172804,
+      "loss": 0.03239120915532112,
+      "loss_ce": 1.9688148313434795e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 1099597220,
+      "step": 6409
+    },
+    {
+      "epoch": 72.75354107648725,
+      "grad_norm": 3.1656607363930127,
+      "learning_rate": 5e-06,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 1099768968,
+      "step": 6410
+    },
+    {
+      "epoch": 72.75354107648725,
+      "loss": 0.040279995650053024,
+      "loss_ce": 1.20499844342703e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 1099768968,
+      "step": 6410
+    },
+    {
+      "epoch": 72.76487252124646,
+      "grad_norm": 3.3913853820335262,
+      "learning_rate": 5e-06,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 1099938672,
+      "step": 6411
+    },
+    {
+      "epoch": 72.76487252124646,
+      "loss": 0.028848998248577118,
+      "loss_ce": 5.5662807426415384e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 1099938672,
+      "step": 6411
+    },
+    {
+      "epoch": 72.77620396600567,
+      "grad_norm": 2.9403937630499297,
+      "learning_rate": 5e-06,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 1100110368,
+      "step": 6412
+    },
+    {
+      "epoch": 72.77620396600567,
+      "loss": 0.048872239887714386,
+      "loss_ce": 1.3594909432868008e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 1100110368,
+      "step": 6412
+    },
+    {
+      "epoch": 72.78753541076487,
+      "grad_norm": 2.385304148353764,
+      "learning_rate": 5e-06,
+      "loss": 0.042,
+      "num_input_tokens_seen": 1100282236,
+      "step": 6413
+    },
+    {
+      "epoch": 72.78753541076487,
+      "loss": 0.03129076957702637,
+      "loss_ce": 1.0252973879687488e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1100282236,
+      "step": 6413
+    },
+    {
+      "epoch": 72.79886685552408,
+      "grad_norm": 3.1936189494910256,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 1100454052,
+      "step": 6414
+    },
+    {
+      "epoch": 72.79886685552408,
+      "loss": 0.04210112988948822,
+      "loss_ce": 1.7386093531968072e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1100454052,
+      "step": 6414
+    },
+    {
+      "epoch": 72.81019830028329,
+      "grad_norm": 3.9225931916393053,
+      "learning_rate": 5e-06,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 1100625908,
+      "step": 6415
+    },
+    {
+      "epoch": 72.81019830028329,
+      "loss": 0.03038671612739563,
+      "loss_ce": 3.698662112583406e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 1100625908,
+      "step": 6415
+    },
+    {
+      "epoch": 72.8215297450425,
+      "grad_norm": 3.6949332877671903,
+      "learning_rate": 5e-06,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 1100797744,
+      "step": 6416
+    },
+    {
+      "epoch": 72.8215297450425,
+      "loss": 0.06131046265363693,
+      "loss_ce": 1.5906749467831105e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 1100797744,
+      "step": 6416
+    },
+    {
+      "epoch": 72.8328611898017,
+      "grad_norm": 3.1794617188706633,
+      "learning_rate": 5e-06,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 1100969404,
+      "step": 6417
+    },
+    {
+      "epoch": 72.8328611898017,
+      "loss": 0.02156449295580387,
+      "loss_ce": 1.9082250219071284e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.004302978515625,
+      "loss_xval": 0.021484375,
+      "num_input_tokens_seen": 1100969404,
+      "step": 6417
+    },
+    {
+      "epoch": 72.84419263456091,
+      "grad_norm": 2.7363584770403735,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1101141208,
+      "step": 6418
+    },
+    {
+      "epoch": 72.84419263456091,
+      "loss": 0.03769973665475845,
+      "loss_ce": 8.682251063873991e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 1101141208,
+      "step": 6418
+    },
+    {
+      "epoch": 72.85552407932012,
+      "grad_norm": 3.0160878846217303,
+      "learning_rate": 5e-06,
+      "loss": 0.055,
+      "num_input_tokens_seen": 1101313280,
+      "step": 6419
+    },
+    {
+      "epoch": 72.85552407932012,
+      "loss": 0.033302415162324905,
+      "loss_ce": 2.2996295228949748e-05,
+      "loss_iou": 0.578125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1101313280,
+      "step": 6419
+    },
+    {
+      "epoch": 72.86685552407933,
+      "grad_norm": 2.5464081025820207,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 1101484928,
+      "step": 6420
+    },
+    {
+      "epoch": 72.86685552407933,
+      "loss": 0.0377458892762661,
+      "loss_ce": 7.193784404080361e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 1101484928,
+      "step": 6420
+    },
+    {
+      "epoch": 72.87818696883852,
+      "grad_norm": 2.1841767032141584,
+      "learning_rate": 5e-06,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 1101655796,
+      "step": 6421
+    },
+    {
+      "epoch": 72.87818696883852,
+      "loss": 0.04816635698080063,
+      "loss_ce": 0.00011643294419627637,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 1101655796,
+      "step": 6421
+    },
+    {
+      "epoch": 72.88951841359773,
+      "grad_norm": 2.3475046303012523,
+      "learning_rate": 5e-06,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 1101826844,
+      "step": 6422
+    },
+    {
+      "epoch": 72.88951841359773,
+      "loss": 0.0436820313334465,
+      "loss_ce": 1.137554590968648e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 1101826844,
+      "step": 6422
+    },
+    {
+      "epoch": 72.90084985835693,
+      "grad_norm": 3.3492603447476252,
+      "learning_rate": 5e-06,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 1101997848,
+      "step": 6423
+    },
+    {
+      "epoch": 72.90084985835693,
+      "loss": 0.06725621223449707,
+      "loss_ce": 4.124757106183097e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 1101997848,
+      "step": 6423
+    },
+    {
+      "epoch": 72.91218130311614,
+      "grad_norm": 3.4347234155492328,
+      "learning_rate": 5e-06,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 1102169952,
+      "step": 6424
+    },
+    {
+      "epoch": 72.91218130311614,
+      "loss": 0.04631930589675903,
+      "loss_ce": 2.4137785658240318e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 1102169952,
+      "step": 6424
+    },
+    {
+      "epoch": 72.92351274787535,
+      "grad_norm": 3.022699629832468,
+      "learning_rate": 5e-06,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 1102340788,
+      "step": 6425
+    },
+    {
+      "epoch": 72.92351274787535,
+      "loss": 0.041486747562885284,
+      "loss_ce": 4.3875479605048895e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 1102340788,
+      "step": 6425
+    },
+    {
+      "epoch": 72.93484419263456,
+      "grad_norm": 3.1689008663761293,
+      "learning_rate": 5e-06,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 1102511728,
+      "step": 6426
+    },
+    {
+      "epoch": 72.93484419263456,
+      "loss": 0.0775311291217804,
+      "loss_ce": 1.6485340893268585e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 1102511728,
+      "step": 6426
+    },
+    {
+      "epoch": 72.94617563739376,
+      "grad_norm": 3.467805795385136,
+      "learning_rate": 5e-06,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 1102682212,
+      "step": 6427
+    },
+    {
+      "epoch": 72.94617563739376,
+      "loss": 0.030984679237008095,
+      "loss_ce": 9.336401490145363e-06,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1102682212,
+      "step": 6427
+    },
+    {
+      "epoch": 72.95750708215297,
+      "grad_norm": 3.4446084932003944,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 1102854436,
+      "step": 6428
+    },
+    {
+      "epoch": 72.95750708215297,
+      "loss": 0.046675343066453934,
+      "loss_ce": 4.448468826012686e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 1102854436,
+      "step": 6428
+    },
+    {
+      "epoch": 72.96883852691218,
+      "grad_norm": 3.106152647942425,
+      "learning_rate": 5e-06,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 1103026072,
+      "step": 6429
+    },
+    {
+      "epoch": 72.96883852691218,
+      "loss": 0.08778481185436249,
+      "loss_ce": 4.677658944274299e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 1103026072,
+      "step": 6429
+    },
+    {
+      "epoch": 72.98016997167139,
+      "grad_norm": 2.910520697482817,
+      "learning_rate": 5e-06,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 1103195616,
+      "step": 6430
+    },
+    {
+      "epoch": 72.98016997167139,
+      "loss": 0.05527765303850174,
+      "loss_ce": 4.083516250830144e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 1103195616,
+      "step": 6430
+    },
+    {
+      "epoch": 72.9915014164306,
+      "grad_norm": 2.981832397353389,
+      "learning_rate": 5e-06,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 1103366048,
+      "step": 6431
+    },
+    {
+      "epoch": 72.9915014164306,
+      "loss": 0.04713235795497894,
+      "loss_ce": 1.3217547348176595e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 1103366048,
+      "step": 6431
+    },
+    {
+      "epoch": 72.9915014164306,
+      "loss": 0.10227605700492859,
+      "loss_ce": 2.691092959139496e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 1103495140,
+      "step": 6431
+    },
+    {
+      "epoch": 73.0028328611898,
+      "grad_norm": 3.6994591782662516,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 1103537952,
+      "step": 6432
+    },
+    {
+      "epoch": 73.0028328611898,
+      "loss": 0.027685847133398056,
+      "loss_ce": 6.402898634405574e-06,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 1103537952,
+      "step": 6432
+    },
+    {
+      "epoch": 73.01416430594901,
+      "grad_norm": 3.4648379887529073,
+      "learning_rate": 5e-06,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 1103708580,
+      "step": 6433
+    },
+    {
+      "epoch": 73.01416430594901,
+      "loss": 0.03935055807232857,
+      "loss_ce": 1.3397737347986549e-05,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 1103708580,
+      "step": 6433
+    },
+    {
+      "epoch": 73.02549575070822,
+      "grad_norm": 2.9207199662965877,
+      "learning_rate": 5e-06,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 1103880156,
+      "step": 6434
+    },
+    {
+      "epoch": 73.02549575070822,
+      "loss": 0.03269508481025696,
+      "loss_ce": 4.1278435674030334e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 1103880156,
+      "step": 6434
+    },
+    {
+      "epoch": 73.03682719546742,
+      "grad_norm": 2.905982254364295,
+      "learning_rate": 5e-06,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 1104052064,
+      "step": 6435
+    },
+    {
+      "epoch": 73.03682719546742,
+      "loss": 0.036076027899980545,
+      "loss_ce": 0.000141577998874709,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1104052064,
+      "step": 6435
+    },
+    {
+      "epoch": 73.04815864022663,
+      "grad_norm": 2.640863974585965,
+      "learning_rate": 5e-06,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 1104223684,
+      "step": 6436
+    },
+    {
+      "epoch": 73.04815864022663,
+      "loss": 0.048409804701805115,
+      "loss_ce": 2.418357689748518e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 1104223684,
+      "step": 6436
+    },
+    {
+      "epoch": 73.05949008498584,
+      "grad_norm": 3.313332227660754,
+      "learning_rate": 5e-06,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 1104393612,
+      "step": 6437
+    },
+    {
+      "epoch": 73.05949008498584,
+      "loss": 0.05031883716583252,
+      "loss_ce": 6.401594146154821e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 1104393612,
+      "step": 6437
+    },
+    {
+      "epoch": 73.07082152974505,
+      "grad_norm": 3.450378233183144,
+      "learning_rate": 5e-06,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 1104563424,
+      "step": 6438
+    },
+    {
+      "epoch": 73.07082152974505,
+      "loss": 0.03137843310832977,
+      "loss_ce": 2.1618816390400752e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1104563424,
+      "step": 6438
+    },
+    {
+      "epoch": 73.08215297450425,
+      "grad_norm": 2.8020120825118693,
+      "learning_rate": 5e-06,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 1104735500,
+      "step": 6439
+    },
+    {
+      "epoch": 73.08215297450425,
+      "loss": 0.026850799098610878,
+      "loss_ce": 1.0589408702799119e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 1104735500,
+      "step": 6439
+    },
+    {
+      "epoch": 73.09348441926346,
+      "grad_norm": 2.6340272115492276,
+      "learning_rate": 5e-06,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 1104906872,
+      "step": 6440
+    },
+    {
+      "epoch": 73.09348441926346,
+      "loss": 0.07926710695028305,
+      "loss_ce": 2.0583902369253337e-05,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 1104906872,
+      "step": 6440
+    },
+    {
+      "epoch": 73.10481586402267,
+      "grad_norm": 2.584400808954087,
+      "learning_rate": 5e-06,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 1105077888,
+      "step": 6441
+    },
+    {
+      "epoch": 73.10481586402267,
+      "loss": 0.020703855901956558,
+      "loss_ce": 2.8195639970363118e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.004150390625,
+      "loss_xval": 0.0206298828125,
+      "num_input_tokens_seen": 1105077888,
+      "step": 6441
+    },
+    {
+      "epoch": 73.11614730878188,
+      "grad_norm": 2.0846174342460033,
+      "learning_rate": 5e-06,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 1105249828,
+      "step": 6442
+    },
+    {
+      "epoch": 73.11614730878188,
+      "loss": 0.030573632568120956,
+      "loss_ce": 7.894216105341911e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1105249828,
+      "step": 6442
+    },
+    {
+      "epoch": 73.12747875354107,
+      "grad_norm": 2.696051878153036,
+      "learning_rate": 5e-06,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 1105418924,
+      "step": 6443
+    },
+    {
+      "epoch": 73.12747875354107,
+      "loss": 0.04201575368642807,
+      "loss_ce": 2.356819823035039e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1105418924,
+      "step": 6443
+    },
+    {
+      "epoch": 73.13881019830028,
+      "grad_norm": 3.58811776792575,
+      "learning_rate": 5e-06,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 1105589376,
+      "step": 6444
+    },
+    {
+      "epoch": 73.13881019830028,
+      "loss": 0.03687825798988342,
+      "loss_ce": 1.30264616018394e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1105589376,
+      "step": 6444
+    },
+    {
+      "epoch": 73.15014164305948,
+      "grad_norm": 3.749253289314688,
+      "learning_rate": 5e-06,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 1105760624,
+      "step": 6445
+    },
+    {
+      "epoch": 73.15014164305948,
+      "loss": 0.042613156139850616,
+      "loss_ce": 1.0616137842589524e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 1105760624,
+      "step": 6445
+    },
+    {
+      "epoch": 73.16147308781869,
+      "grad_norm": 3.877064264441872,
+      "learning_rate": 5e-06,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 1105932508,
+      "step": 6446
+    },
+    {
+      "epoch": 73.16147308781869,
+      "loss": 0.04025866091251373,
+      "loss_ce": 2.12349186767824e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 1105932508,
+      "step": 6446
+    },
+    {
+      "epoch": 73.1728045325779,
+      "grad_norm": 3.8991843133533117,
+      "learning_rate": 5e-06,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 1106103540,
+      "step": 6447
+    },
+    {
+      "epoch": 73.1728045325779,
+      "loss": 0.07381882518529892,
+      "loss_ce": 1.2061761481163558e-05,
+      "loss_iou": 0.1298828125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 1106103540,
+      "step": 6447
+    },
+    {
+      "epoch": 73.1841359773371,
+      "grad_norm": 3.500376497813227,
+      "learning_rate": 5e-06,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 1106275008,
+      "step": 6448
+    },
+    {
+      "epoch": 73.1841359773371,
+      "loss": 0.03309686481952667,
+      "loss_ce": 1.58109196490841e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 1106275008,
+      "step": 6448
+    },
+    {
+      "epoch": 73.19546742209631,
+      "grad_norm": 2.6903158899230797,
+      "learning_rate": 5e-06,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 1106446248,
+      "step": 6449
+    },
+    {
+      "epoch": 73.19546742209631,
+      "loss": 0.08586322516202927,
+      "loss_ce": 1.7282774933846667e-05,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 1106446248,
+      "step": 6449
+    },
+    {
+      "epoch": 73.20679886685552,
+      "grad_norm": 2.436789809227241,
+      "learning_rate": 5e-06,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 1106618140,
+      "step": 6450
+    },
+    {
+      "epoch": 73.20679886685552,
+      "loss": 0.03357922285795212,
+      "loss_ce": 2.5147364794975147e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1106618140,
+      "step": 6450
+    },
+    {
+      "epoch": 73.21813031161473,
+      "grad_norm": 2.848381549558379,
+      "learning_rate": 5e-06,
+      "loss": 0.032,
+      "num_input_tokens_seen": 1106789616,
+      "step": 6451
+    },
+    {
+      "epoch": 73.21813031161473,
+      "loss": 0.04093830659985542,
+      "loss_ce": 1.4233784895623103e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1106789616,
+      "step": 6451
+    },
+    {
+      "epoch": 73.22946175637394,
+      "grad_norm": 3.1241714664854245,
+      "learning_rate": 5e-06,
+      "loss": 0.041,
+      "num_input_tokens_seen": 1106959604,
+      "step": 6452
+    },
+    {
+      "epoch": 73.22946175637394,
+      "loss": 0.03366551548242569,
+      "loss_ce": 3.514260606607422e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1106959604,
+      "step": 6452
+    },
+    {
+      "epoch": 73.24079320113314,
+      "grad_norm": 3.0840479155024445,
+      "learning_rate": 5e-06,
+      "loss": 0.037,
+      "num_input_tokens_seen": 1107131352,
+      "step": 6453
+    },
+    {
+      "epoch": 73.24079320113314,
+      "loss": 0.030013732612133026,
+      "loss_ce": 1.49521238199668e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1107131352,
+      "step": 6453
+    },
+    {
+      "epoch": 73.25212464589235,
+      "grad_norm": 3.5992573840646767,
+      "learning_rate": 5e-06,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 1107303088,
+      "step": 6454
+    },
+    {
+      "epoch": 73.25212464589235,
+      "loss": 0.04085526615381241,
+      "loss_ce": 2.2747990442439914e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1107303088,
+      "step": 6454
+    },
+    {
+      "epoch": 73.26345609065156,
+      "grad_norm": 3.8252708912968534,
+      "learning_rate": 5e-06,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 1107474824,
+      "step": 6455
+    },
+    {
+      "epoch": 73.26345609065156,
+      "loss": 0.0406356155872345,
+      "loss_ce": 1.6718397091608495e-05,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1107474824,
+      "step": 6455
+    },
+    {
+      "epoch": 73.27478753541077,
+      "grad_norm": 4.017300771050497,
+      "learning_rate": 5e-06,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 1107646632,
+      "step": 6456
+    },
+    {
+      "epoch": 73.27478753541077,
+      "loss": 0.034910544753074646,
+      "loss_ce": 1.3692781067220494e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1107646632,
+      "step": 6456
+    },
+    {
+      "epoch": 73.28611898016997,
+      "grad_norm": 2.989494704346733,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1107818212,
+      "step": 6457
+    },
+    {
+      "epoch": 73.28611898016997,
+      "loss": 0.03674791380763054,
+      "loss_ce": 2.0007406419608742e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1107818212,
+      "step": 6457
+    },
+    {
+      "epoch": 73.29745042492918,
+      "grad_norm": 3.1306413996893503,
+      "learning_rate": 5e-06,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 1107986928,
+      "step": 6458
+    },
+    {
+      "epoch": 73.29745042492918,
+      "loss": 0.031121093779802322,
+      "loss_ce": 2.368139939790126e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 1107986928,
+      "step": 6458
+    },
+    {
+      "epoch": 73.30878186968839,
+      "grad_norm": 3.0920400310808733,
+      "learning_rate": 5e-06,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 1108158608,
+      "step": 6459
+    },
+    {
+      "epoch": 73.30878186968839,
+      "loss": 0.04605382680892944,
+      "loss_ce": 1.8060127331409603e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 1108158608,
+      "step": 6459
+    },
+    {
+      "epoch": 73.3201133144476,
+      "grad_norm": 5.88684457277583,
+      "learning_rate": 5e-06,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 1108330740,
+      "step": 6460
+    },
+    {
+      "epoch": 73.3201133144476,
+      "loss": 0.032196275889873505,
+      "loss_ce": 1.5490251826122403e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1108330740,
+      "step": 6460
+    },
+    {
+      "epoch": 73.3314447592068,
+      "grad_norm": 3.0814263094685934,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1108502448,
+      "step": 6461
+    },
+    {
+      "epoch": 73.3314447592068,
+      "loss": 0.0262615904211998,
+      "loss_ce": 1.6473162759211846e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 1108502448,
+      "step": 6461
+    },
+    {
+      "epoch": 73.34277620396601,
+      "grad_norm": 2.751703744798392,
+      "learning_rate": 5e-06,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 1108673192,
+      "step": 6462
+    },
+    {
+      "epoch": 73.34277620396601,
+      "loss": 0.023171426728367805,
+      "loss_ce": 1.6213718481594697e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 1108673192,
+      "step": 6462
+    },
+    {
+      "epoch": 73.35410764872522,
+      "grad_norm": 3.094501974301736,
+      "learning_rate": 5e-06,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 1108842956,
+      "step": 6463
+    },
+    {
+      "epoch": 73.35410764872522,
+      "loss": 0.028249815106391907,
+      "loss_ce": 2.105399835272692e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 1108842956,
+      "step": 6463
+    },
+    {
+      "epoch": 73.36543909348443,
+      "grad_norm": 3.5133096138128637,
+      "learning_rate": 5e-06,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 1109014732,
+      "step": 6464
+    },
+    {
+      "epoch": 73.36543909348443,
+      "loss": 0.03406064584851265,
+      "loss_ce": 0.0035278077702969313,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1109014732,
+      "step": 6464
+    },
+    {
+      "epoch": 73.37677053824362,
+      "grad_norm": 3.1754817078113504,
+      "learning_rate": 5e-06,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 1109184920,
+      "step": 6465
+    },
+    {
+      "epoch": 73.37677053824362,
+      "loss": 0.05430317670106888,
+      "loss_ce": 2.7667218091664836e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 1109184920,
+      "step": 6465
+    },
+    {
+      "epoch": 73.38810198300283,
+      "grad_norm": 2.9395621782941648,
+      "learning_rate": 5e-06,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 1109356608,
+      "step": 6466
+    },
+    {
+      "epoch": 73.38810198300283,
+      "loss": 0.031933143734931946,
+      "loss_ce": 1.1758004802686628e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1109356608,
+      "step": 6466
+    },
+    {
+      "epoch": 73.39943342776203,
+      "grad_norm": 3.5523172752121637,
+      "learning_rate": 5e-06,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 1109526300,
+      "step": 6467
+    },
+    {
+      "epoch": 73.39943342776203,
+      "loss": 0.05787474662065506,
+      "loss_ce": 1.341482038696995e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 1109526300,
+      "step": 6467
+    },
+    {
+      "epoch": 73.41076487252124,
+      "grad_norm": 2.9660119784878916,
+      "learning_rate": 5e-06,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 1109698084,
+      "step": 6468
+    },
+    {
+      "epoch": 73.41076487252124,
+      "loss": 0.079731285572052,
+      "loss_ce": 1.9377464923309162e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 1109698084,
+      "step": 6468
+    },
+    {
+      "epoch": 73.42209631728045,
+      "grad_norm": 2.586709458627176,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 1109868764,
+      "step": 6469
+    },
+    {
+      "epoch": 73.42209631728045,
+      "loss": 0.05429849773645401,
+      "loss_ce": 2.2983678718446754e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 1109868764,
+      "step": 6469
+    },
+    {
+      "epoch": 73.43342776203966,
+      "grad_norm": 3.306600132488779,
+      "learning_rate": 5e-06,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 1110040360,
+      "step": 6470
+    },
+    {
+      "epoch": 73.43342776203966,
+      "loss": 0.06186573952436447,
+      "loss_ce": 3.712794205057435e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 1110040360,
+      "step": 6470
+    },
+    {
+      "epoch": 73.44475920679886,
+      "grad_norm": 3.312855548906545,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 1110212048,
+      "step": 6471
+    },
+    {
+      "epoch": 73.44475920679886,
+      "loss": 0.035165928304195404,
+      "loss_ce": 9.676637091615703e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1110212048,
+      "step": 6471
+    },
+    {
+      "epoch": 73.45609065155807,
+      "grad_norm": 3.28920872797524,
+      "learning_rate": 5e-06,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 1110383036,
+      "step": 6472
+    },
+    {
+      "epoch": 73.45609065155807,
+      "loss": 0.03040408156812191,
+      "loss_ce": 8.572325896238908e-06,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 1110383036,
+      "step": 6472
+    },
+    {
+      "epoch": 73.46742209631728,
+      "grad_norm": 3.2592469296490307,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 1110553644,
+      "step": 6473
+    },
+    {
+      "epoch": 73.46742209631728,
+      "loss": 0.043555401265621185,
+      "loss_ce": 3.733702396857552e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 1110553644,
+      "step": 6473
+    },
+    {
+      "epoch": 73.47875354107649,
+      "grad_norm": 2.6549269921249867,
+      "learning_rate": 5e-06,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 1110725256,
+      "step": 6474
+    },
+    {
+      "epoch": 73.47875354107649,
+      "loss": 0.026280013844370842,
+      "loss_ce": 1.963696922757663e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 1110725256,
+      "step": 6474
+    },
+    {
+      "epoch": 73.4900849858357,
+      "grad_norm": 2.8951843138895166,
+      "learning_rate": 5e-06,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 1110896804,
+      "step": 6475
+    },
+    {
+      "epoch": 73.4900849858357,
+      "loss": 0.11417710036039352,
+      "loss_ce": 5.661951581714675e-05,
+      "loss_iou": 0.03662109375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 1110896804,
+      "step": 6475
+    },
+    {
+      "epoch": 73.5014164305949,
+      "grad_norm": 3.7301456964179827,
+      "learning_rate": 5e-06,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 1111068760,
+      "step": 6476
+    },
+    {
+      "epoch": 73.5014164305949,
+      "loss": 0.06310199201107025,
+      "loss_ce": 8.319588960148394e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 1111068760,
+      "step": 6476
+    },
+    {
+      "epoch": 73.51274787535411,
+      "grad_norm": 3.498005795239106,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 1111240540,
+      "step": 6477
+    },
+    {
+      "epoch": 73.51274787535411,
+      "loss": 0.04817181080579758,
+      "loss_ce": 1.5074914699653164e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 1111240540,
+      "step": 6477
+    },
+    {
+      "epoch": 73.52407932011332,
+      "grad_norm": 3.055170153768895,
+      "learning_rate": 5e-06,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 1111412424,
+      "step": 6478
+    },
+    {
+      "epoch": 73.52407932011332,
+      "loss": 0.04850431904196739,
+      "loss_ce": 2.714867878239602e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 1111412424,
+      "step": 6478
+    },
+    {
+      "epoch": 73.53541076487252,
+      "grad_norm": 3.352723256610269,
+      "learning_rate": 5e-06,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 1111584168,
+      "step": 6479
+    },
+    {
+      "epoch": 73.53541076487252,
+      "loss": 0.04059762507677078,
+      "loss_ce": 9.247715752280783e-06,
+      "loss_iou": 0.515625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1111584168,
+      "step": 6479
+    },
+    {
+      "epoch": 73.54674220963173,
+      "grad_norm": 4.009041258935215,
+      "learning_rate": 5e-06,
+      "loss": 0.0695,
+      "num_input_tokens_seen": 1111754700,
+      "step": 6480
+    },
+    {
+      "epoch": 73.54674220963173,
+      "loss": 0.04311896860599518,
+      "loss_ce": 1.2887299817521125e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1111754700,
+      "step": 6480
+    },
+    {
+      "epoch": 73.55807365439094,
+      "grad_norm": 3.1537946213291375,
+      "learning_rate": 5e-06,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 1111925588,
+      "step": 6481
+    },
+    {
+      "epoch": 73.55807365439094,
+      "loss": 0.031039375811815262,
+      "loss_ce": 1.825577419367619e-05,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 1111925588,
+      "step": 6481
+    },
+    {
+      "epoch": 73.56940509915015,
+      "grad_norm": 2.2943597392871053,
+      "learning_rate": 5e-06,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 1112097356,
+      "step": 6482
+    },
+    {
+      "epoch": 73.56940509915015,
+      "loss": 0.03549940139055252,
+      "loss_ce": 7.459177140844986e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 1112097356,
+      "step": 6482
+    },
+    {
+      "epoch": 73.58073654390935,
+      "grad_norm": 2.090884541047774,
+      "learning_rate": 5e-06,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 1112269016,
+      "step": 6483
+    },
+    {
+      "epoch": 73.58073654390935,
+      "loss": 0.07406432181596756,
+      "loss_ce": 4.3938751332461834e-05,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 1112269016,
+      "step": 6483
+    },
+    {
+      "epoch": 73.59206798866856,
+      "grad_norm": 3.1708341647785434,
+      "learning_rate": 5e-06,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 1112440168,
+      "step": 6484
+    },
+    {
+      "epoch": 73.59206798866856,
+      "loss": 0.03127949312329292,
+      "loss_ce": 1.4234448826755397e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1112440168,
+      "step": 6484
+    },
+    {
+      "epoch": 73.60339943342777,
+      "grad_norm": 3.8607870766871435,
+      "learning_rate": 5e-06,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 1112609292,
+      "step": 6485
+    },
+    {
+      "epoch": 73.60339943342777,
+      "loss": 0.03704557567834854,
+      "loss_ce": 2.7751655579777434e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 1112609292,
+      "step": 6485
+    },
+    {
+      "epoch": 73.61473087818698,
+      "grad_norm": 2.2399730241428277,
+      "learning_rate": 5e-06,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 1112779556,
+      "step": 6486
+    },
+    {
+      "epoch": 73.61473087818698,
+      "loss": 0.02919117361307144,
+      "loss_ce": 2.3999054974410683e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 1112779556,
+      "step": 6486
+    },
+    {
+      "epoch": 73.62606232294617,
+      "grad_norm": 3.8101749347443814,
+      "learning_rate": 5e-06,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 1112950660,
+      "step": 6487
+    },
+    {
+      "epoch": 73.62606232294617,
+      "loss": 0.022443873807787895,
+      "loss_ce": 8.97480858839117e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.0223388671875,
+      "num_input_tokens_seen": 1112950660,
+      "step": 6487
+    },
+    {
+      "epoch": 73.63739376770538,
+      "grad_norm": 3.894889633043946,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 1113120940,
+      "step": 6488
+    },
+    {
+      "epoch": 73.63739376770538,
+      "loss": 0.03205784782767296,
+      "loss_ce": 1.4391994227480609e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1113120940,
+      "step": 6488
+    },
+    {
+      "epoch": 73.64872521246458,
+      "grad_norm": 3.120787083301744,
+      "learning_rate": 5e-06,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 1113292748,
+      "step": 6489
+    },
+    {
+      "epoch": 73.64872521246458,
+      "loss": 0.02735951915383339,
+      "loss_ce": 3.1029641831992194e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 1113292748,
+      "step": 6489
+    },
+    {
+      "epoch": 73.66005665722379,
+      "grad_norm": 3.221535969834847,
+      "learning_rate": 5e-06,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 1113464820,
+      "step": 6490
+    },
+    {
+      "epoch": 73.66005665722379,
+      "loss": 0.03254421055316925,
+      "loss_ce": 4.2991916416212916e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 1113464820,
+      "step": 6490
+    },
+    {
+      "epoch": 73.671388101983,
+      "grad_norm": 4.151994683365662,
+      "learning_rate": 5e-06,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 1113636328,
+      "step": 6491
+    },
+    {
+      "epoch": 73.671388101983,
+      "loss": 0.03663531690835953,
+      "loss_ce": 2.9481716410373338e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1113636328,
+      "step": 6491
+    },
+    {
+      "epoch": 73.6827195467422,
+      "grad_norm": 3.2698866394695214,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 1113807864,
+      "step": 6492
+    },
+    {
+      "epoch": 73.6827195467422,
+      "loss": 0.02959362044930458,
+      "loss_ce": 2.2087617253419012e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 1113807864,
+      "step": 6492
+    },
+    {
+      "epoch": 73.69405099150141,
+      "grad_norm": 3.244195012469785,
+      "learning_rate": 5e-06,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 1113978272,
+      "step": 6493
+    },
+    {
+      "epoch": 73.69405099150141,
+      "loss": 0.035467371344566345,
+      "loss_ce": 2.120135832228698e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 1113978272,
+      "step": 6493
+    },
+    {
+      "epoch": 73.70538243626062,
+      "grad_norm": 3.655037573224194,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 1114149392,
+      "step": 6494
+    },
+    {
+      "epoch": 73.70538243626062,
+      "loss": 0.044869210571050644,
+      "loss_ce": 8.370522664336022e-06,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 1114149392,
+      "step": 6494
+    },
+    {
+      "epoch": 73.71671388101983,
+      "grad_norm": 3.7782426891480454,
+      "learning_rate": 5e-06,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 1114318948,
+      "step": 6495
+    },
+    {
+      "epoch": 73.71671388101983,
+      "loss": 0.03719808906316757,
+      "loss_ce": 4.293832171242684e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 1114318948,
+      "step": 6495
+    },
+    {
+      "epoch": 73.72804532577904,
+      "grad_norm": 3.499634486668161,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 1114489916,
+      "step": 6496
+    },
+    {
+      "epoch": 73.72804532577904,
+      "loss": 0.030175620689988136,
+      "loss_ce": 2.425476850476116e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 1114489916,
+      "step": 6496
+    },
+    {
+      "epoch": 73.73937677053824,
+      "grad_norm": 2.8426321920918576,
+      "learning_rate": 5e-06,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 1114661624,
+      "step": 6497
+    },
+    {
+      "epoch": 73.73937677053824,
+      "loss": 0.026565423235297203,
+      "loss_ce": 2.2759606508770958e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 1114661624,
+      "step": 6497
+    },
+    {
+      "epoch": 73.75070821529745,
+      "grad_norm": 3.121434993905259,
+      "learning_rate": 5e-06,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 1114833784,
+      "step": 6498
+    },
+    {
+      "epoch": 73.75070821529745,
+      "loss": 0.07860241830348969,
+      "loss_ce": 1.9653536583064124e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 1114833784,
+      "step": 6498
+    },
+    {
+      "epoch": 73.76203966005666,
+      "grad_norm": 3.0212671657824584,
+      "learning_rate": 5e-06,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 1115005612,
+      "step": 6499
+    },
+    {
+      "epoch": 73.76203966005666,
+      "loss": 0.026860613375902176,
+      "loss_ce": 3.566223676898517e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 1115005612,
+      "step": 6499
+    },
+    {
+      "epoch": 73.77337110481587,
+      "grad_norm": 3.305158194069835,
+      "learning_rate": 5e-06,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 1115176432,
+      "step": 6500
+    },
+    {
+      "epoch": 73.77337110481587,
+      "eval_seeclick_CIoU": 0.5174305289983749,
+      "eval_seeclick_GIoU": 0.5203062295913696,
+      "eval_seeclick_IoU": 0.5562175512313843,
+      "eval_seeclick_MAE_all": 0.06904058903455734,
+      "eval_seeclick_MAE_h": 0.034702371805906296,
+      "eval_seeclick_MAE_w": 0.1056600883603096,
+      "eval_seeclick_MAE_x": 0.10088222846388817,
+      "eval_seeclick_MAE_y": 0.03491765260696411,
+      "eval_seeclick_NUM_probability": 0.9999951124191284,
+      "eval_seeclick_inside_bbox": 0.9076704680919647,
+      "eval_seeclick_loss": 1.0270713567733765,
+      "eval_seeclick_loss_ce": 0.7699527740478516,
+      "eval_seeclick_loss_iou": 0.560791015625,
+      "eval_seeclick_loss_num": 0.052215576171875,
+      "eval_seeclick_loss_xval": 0.2611083984375,
+      "eval_seeclick_runtime": 70.9732,
+      "eval_seeclick_samples_per_second": 0.606,
+      "eval_seeclick_steps_per_second": 0.028,
+      "num_input_tokens_seen": 1115176432,
+      "step": 6500
+    },
+    {
+      "epoch": 73.77337110481587,
+      "eval_icons_CIoU": 0.7356639802455902,
+      "eval_icons_GIoU": 0.7380691468715668,
+      "eval_icons_IoU": 0.7486778497695923,
+      "eval_icons_MAE_all": 0.03320657089352608,
+      "eval_icons_MAE_h": 0.02907582465559244,
+      "eval_icons_MAE_w": 0.039673191495239735,
+      "eval_icons_MAE_x": 0.03304527560248971,
+      "eval_icons_MAE_y": 0.031031984835863113,
+      "eval_icons_NUM_probability": 0.9998223185539246,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.13039979338645935,
+      "eval_icons_loss_ce": 0.004626711364835501,
+      "eval_icons_loss_iou": 0.56982421875,
+      "eval_icons_loss_num": 0.0230712890625,
+      "eval_icons_loss_xval": 0.115386962890625,
+      "eval_icons_runtime": 80.9779,
+      "eval_icons_samples_per_second": 0.617,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 1115176432,
+      "step": 6500
+    },
+    {
+      "epoch": 73.77337110481587,
+      "eval_screenspot_CIoU": 0.6431978742281595,
+      "eval_screenspot_GIoU": 0.6427284081776937,
+      "eval_screenspot_IoU": 0.6707419554392496,
+      "eval_screenspot_MAE_all": 0.06348163510362308,
+      "eval_screenspot_MAE_h": 0.03672616680463155,
+      "eval_screenspot_MAE_w": 0.1126464003076156,
+      "eval_screenspot_MAE_x": 0.07129109464585781,
+      "eval_screenspot_MAE_y": 0.033262877725064754,
+      "eval_screenspot_NUM_probability": 0.9999578396479288,
+      "eval_screenspot_inside_bbox": 0.9183333317438761,
+      "eval_screenspot_loss": 0.2698996067047119,
+      "eval_screenspot_loss_ce": 0.014380348846316338,
+      "eval_screenspot_loss_iou": 0.4943033854166667,
+      "eval_screenspot_loss_num": 0.050984700520833336,
+      "eval_screenspot_loss_xval": 0.2549641927083333,
+      "eval_screenspot_runtime": 152.8165,
+      "eval_screenspot_samples_per_second": 0.582,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 1115176432,
+      "step": 6500
+    },
+    {
+      "epoch": 73.77337110481587,
+      "eval_compot_CIoU": 0.8599472343921661,
+      "eval_compot_GIoU": 0.8595451414585114,
+      "eval_compot_IoU": 0.8748769760131836,
+      "eval_compot_MAE_all": 0.024935631081461906,
+      "eval_compot_MAE_h": 0.022948280908167362,
+      "eval_compot_MAE_w": 0.02818493079394102,
+      "eval_compot_MAE_x": 0.02519217785447836,
+      "eval_compot_MAE_y": 0.023417134769260883,
+      "eval_compot_NUM_probability": 0.9999797642230988,
+      "eval_compot_inside_bbox": 0.9409722089767456,
+      "eval_compot_loss": 0.08053144812583923,
+      "eval_compot_loss_ce": 1.2657139905059012e-05,
+      "eval_compot_loss_iou": 0.522216796875,
+      "eval_compot_loss_num": 0.013930320739746094,
+      "eval_compot_loss_xval": 0.069610595703125,
+      "eval_compot_runtime": 86.145,
+      "eval_compot_samples_per_second": 0.58,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 1115176432,
+      "step": 6500
+    },
+    {
+      "epoch": 73.77337110481587,
+      "eval_custom_ui_MAE_all": 0.020075481850653887,
+      "eval_custom_ui_MAE_x": 0.03277571313083172,
+      "eval_custom_ui_MAE_y": 0.007375251268967986,
+      "eval_custom_ui_NUM_probability": 0.999919593334198,
+      "eval_custom_ui_loss": 0.22876690328121185,
+      "eval_custom_ui_loss_ce": 0.1282486617565155,
+      "eval_custom_ui_loss_num": 0.019617080688476562,
+      "eval_custom_ui_loss_xval": 0.0980987548828125,
+      "eval_custom_ui_runtime": 64.9955,
+      "eval_custom_ui_samples_per_second": 0.769,
+      "eval_custom_ui_steps_per_second": 0.031,
+      "num_input_tokens_seen": 1115176432,
+      "step": 6500
+    },
+    {
+      "epoch": 73.77337110481587,
+      "loss": 0.27542316913604736,
+      "loss_ce": 0.15579429268836975,
+      "loss_iou": 0.0,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 1115176432,
+      "step": 6500
+    },
+    {
+      "epoch": 73.78470254957507,
+      "grad_norm": 3.074138302570329,
+      "learning_rate": 5e-06,
+      "loss": 0.036,
+      "num_input_tokens_seen": 1115346552,
+      "step": 6501
+    },
+    {
+      "epoch": 73.78470254957507,
+      "loss": 0.03765908628702164,
+      "loss_ce": 1.5651125067961402e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 1115346552,
+      "step": 6501
+    },
+    {
+      "epoch": 73.79603399433428,
+      "grad_norm": 4.332527683362927,
+      "learning_rate": 5e-06,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 1115517988,
+      "step": 6502
+    },
+    {
+      "epoch": 73.79603399433428,
+      "loss": 0.1721544861793518,
+      "loss_ce": 2.7715950636775233e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 1115517988,
+      "step": 6502
+    },
+    {
+      "epoch": 73.80736543909349,
+      "grad_norm": 4.103666190926778,
+      "learning_rate": 5e-06,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 1115689260,
+      "step": 6503
+    },
+    {
+      "epoch": 73.80736543909349,
+      "loss": 0.03792804479598999,
+      "loss_ce": 9.954062988981605e-06,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 1115689260,
+      "step": 6503
+    },
+    {
+      "epoch": 73.8186968838527,
+      "grad_norm": 3.183386145933696,
+      "learning_rate": 5e-06,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 1115860240,
+      "step": 6504
+    },
+    {
+      "epoch": 73.8186968838527,
+      "loss": 0.06231614574790001,
+      "loss_ce": 1.4506962543237023e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 1115860240,
+      "step": 6504
+    },
+    {
+      "epoch": 73.8300283286119,
+      "grad_norm": 2.6778214579147934,
+      "learning_rate": 5e-06,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 1116029880,
+      "step": 6505
+    },
+    {
+      "epoch": 73.8300283286119,
+      "loss": 0.03555383160710335,
+      "loss_ce": 6.951671093702316e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 1116029880,
+      "step": 6505
+    },
+    {
+      "epoch": 73.84135977337111,
+      "grad_norm": 3.2073385362650293,
+      "learning_rate": 5e-06,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 1116201856,
+      "step": 6506
+    },
+    {
+      "epoch": 73.84135977337111,
+      "loss": 0.032043490558862686,
+      "loss_ce": 3.0551869713235646e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1116201856,
+      "step": 6506
+    },
+    {
+      "epoch": 73.85269121813032,
+      "grad_norm": 3.370755909312059,
+      "learning_rate": 5e-06,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 1116373880,
+      "step": 6507
+    },
+    {
+      "epoch": 73.85269121813032,
+      "loss": 0.035537585616111755,
+      "loss_ce": 1.5125366189749911e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 1116373880,
+      "step": 6507
+    },
+    {
+      "epoch": 73.86402266288952,
+      "grad_norm": 3.708180200731744,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 1116545764,
+      "step": 6508
+    },
+    {
+      "epoch": 73.86402266288952,
+      "loss": 0.035140153020620346,
+      "loss_ce": 1.4421272680920083e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1116545764,
+      "step": 6508
+    },
+    {
+      "epoch": 73.87535410764872,
+      "grad_norm": 3.2301353830037516,
+      "learning_rate": 5e-06,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 1116717468,
+      "step": 6509
+    },
+    {
+      "epoch": 73.87535410764872,
+      "loss": 0.043307363986968994,
+      "loss_ce": 1.8178719983552583e-05,
+      "loss_iou": 0.625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 1116717468,
+      "step": 6509
+    },
+    {
+      "epoch": 73.88668555240793,
+      "grad_norm": 4.004722300627511,
+      "learning_rate": 5e-06,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 1116888892,
+      "step": 6510
+    },
+    {
+      "epoch": 73.88668555240793,
+      "loss": 0.06704437732696533,
+      "loss_ce": 2.777582085400354e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 1116888892,
+      "step": 6510
+    },
+    {
+      "epoch": 73.89801699716713,
+      "grad_norm": 4.163893794625848,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 1117060236,
+      "step": 6511
+    },
+    {
+      "epoch": 73.89801699716713,
+      "loss": 0.043467551469802856,
+      "loss_ce": 2.5777859264053404e-05,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 1117060236,
+      "step": 6511
+    },
+    {
+      "epoch": 73.90934844192634,
+      "grad_norm": 3.7593655545216818,
+      "learning_rate": 5e-06,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 1117231864,
+      "step": 6512
+    },
+    {
+      "epoch": 73.90934844192634,
+      "loss": 0.03475530445575714,
+      "loss_ce": 7.207525050034747e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1117231864,
+      "step": 6512
+    },
+    {
+      "epoch": 73.92067988668555,
+      "grad_norm": 3.0722646414303862,
+      "learning_rate": 5e-06,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 1117402148,
+      "step": 6513
+    },
+    {
+      "epoch": 73.92067988668555,
+      "loss": 0.02711726725101471,
+      "loss_ce": 1.765920387697406e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1117402148,
+      "step": 6513
+    },
+    {
+      "epoch": 73.93201133144476,
+      "grad_norm": 3.0359591532116337,
+      "learning_rate": 5e-06,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 1117573852,
+      "step": 6514
+    },
+    {
+      "epoch": 73.93201133144476,
+      "loss": 0.059854790568351746,
+      "loss_ce": 2.5079119950532913e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 1117573852,
+      "step": 6514
+    },
+    {
+      "epoch": 73.94334277620396,
+      "grad_norm": 2.897343816417393,
+      "learning_rate": 5e-06,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 1117745864,
+      "step": 6515
+    },
+    {
+      "epoch": 73.94334277620396,
+      "loss": 0.03228772059082985,
+      "loss_ce": 3.826897227554582e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1117745864,
+      "step": 6515
+    },
+    {
+      "epoch": 73.95467422096317,
+      "grad_norm": 4.028246615971039,
+      "learning_rate": 5e-06,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 1117917648,
+      "step": 6516
+    },
+    {
+      "epoch": 73.95467422096317,
+      "loss": 0.06714378297328949,
+      "loss_ce": 2.0364366719149984e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 1117917648,
+      "step": 6516
+    },
+    {
+      "epoch": 73.96600566572238,
+      "grad_norm": 3.9567889367108693,
+      "learning_rate": 5e-06,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 1118089424,
+      "step": 6517
+    },
+    {
+      "epoch": 73.96600566572238,
+      "loss": 0.03859155997633934,
+      "loss_ce": 1.7340553313260898e-05,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 1118089424,
+      "step": 6517
+    },
+    {
+      "epoch": 73.97733711048159,
+      "grad_norm": 2.9648063966099025,
+      "learning_rate": 5e-06,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 1118261220,
+      "step": 6518
+    },
+    {
+      "epoch": 73.97733711048159,
+      "loss": 0.055747825652360916,
+      "loss_ce": 3.798906254814938e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 1118261220,
+      "step": 6518
+    },
+    {
+      "epoch": 73.98866855524079,
+      "grad_norm": 2.9961977730138494,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1118433176,
+      "step": 6519
+    },
+    {
+      "epoch": 73.98866855524079,
+      "loss": 0.04088219255208969,
+      "loss_ce": 1.9152939785271883e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1118433176,
+      "step": 6519
+    },
+    {
+      "epoch": 74.0,
+      "grad_norm": 3.3755395370940406,
+      "learning_rate": 5e-06,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 1118602448,
+      "step": 6520
+    },
+    {
+      "epoch": 74.0,
+      "loss": 0.032127220183610916,
+      "loss_ce": 7.469081538147293e-06,
+      "loss_iou": 0.421875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1118602448,
+      "step": 6520
+    },
+    {
+      "epoch": 74.01133144475921,
+      "grad_norm": 3.092078476524234,
+      "learning_rate": 5e-06,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 1118773860,
+      "step": 6521
+    },
+    {
+      "epoch": 74.01133144475921,
+      "loss": 0.036174528300762177,
+      "loss_ce": 1.120026263379259e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 1118773860,
+      "step": 6521
+    },
+    {
+      "epoch": 74.02266288951841,
+      "grad_norm": 2.971499851196099,
+      "learning_rate": 5e-06,
+      "loss": 0.053,
+      "num_input_tokens_seen": 1118945720,
+      "step": 6522
+    },
+    {
+      "epoch": 74.02266288951841,
+      "loss": 0.07449185103178024,
+      "loss_ce": 1.3699782357434742e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 1118945720,
+      "step": 6522
+    },
+    {
+      "epoch": 74.03399433427762,
+      "grad_norm": 5.598040763860579,
+      "learning_rate": 5e-06,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 1119117352,
+      "step": 6523
+    },
+    {
+      "epoch": 74.03399433427762,
+      "loss": 0.025059839710593224,
+      "loss_ce": 2.0166649846942164e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 1119117352,
+      "step": 6523
+    },
+    {
+      "epoch": 74.04532577903683,
+      "grad_norm": 20.947211042536978,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 1119287792,
+      "step": 6524
+    },
+    {
+      "epoch": 74.04532577903683,
+      "loss": 0.03474951535463333,
+      "loss_ce": 3.5771379771176726e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1119287792,
+      "step": 6524
+    },
+    {
+      "epoch": 74.05665722379604,
+      "grad_norm": 3.3105554612824744,
+      "learning_rate": 5e-06,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 1119459256,
+      "step": 6525
+    },
+    {
+      "epoch": 74.05665722379604,
+      "loss": 0.06944504380226135,
+      "loss_ce": 1.755080302245915e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 1119459256,
+      "step": 6525
+    },
+    {
+      "epoch": 74.06798866855524,
+      "grad_norm": 2.67025463899964,
+      "learning_rate": 5e-06,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 1119630180,
+      "step": 6526
+    },
+    {
+      "epoch": 74.06798866855524,
+      "loss": 0.040065426379442215,
+      "loss_ce": 1.1105581506853923e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 1119630180,
+      "step": 6526
+    },
+    {
+      "epoch": 74.07932011331445,
+      "grad_norm": 2.9205934349374885,
+      "learning_rate": 5e-06,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 1119802516,
+      "step": 6527
+    },
+    {
+      "epoch": 74.07932011331445,
+      "loss": 0.04559585079550743,
+      "loss_ce": 1.0219937394140288e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 1119802516,
+      "step": 6527
+    },
+    {
+      "epoch": 74.09065155807366,
+      "grad_norm": 2.9843199474765485,
+      "learning_rate": 5e-06,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 1119974392,
+      "step": 6528
+    },
+    {
+      "epoch": 74.09065155807366,
+      "loss": 0.026788223534822464,
+      "loss_ce": 9.048206266015768e-06,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 1119974392,
+      "step": 6528
+    },
+    {
+      "epoch": 74.10198300283287,
+      "grad_norm": 3.04783184834557,
+      "learning_rate": 5e-06,
+      "loss": 0.042,
+      "num_input_tokens_seen": 1120146056,
+      "step": 6529
+    },
+    {
+      "epoch": 74.10198300283287,
+      "loss": 0.030398353934288025,
+      "loss_ce": 1.8105823983205482e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 1120146056,
+      "step": 6529
+    },
+    {
+      "epoch": 74.11331444759207,
+      "grad_norm": 3.2375599818467187,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1120317720,
+      "step": 6530
+    },
+    {
+      "epoch": 74.11331444759207,
+      "loss": 0.06411470472812653,
+      "loss_ce": 1.2532809705589898e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 1120317720,
+      "step": 6530
+    },
+    {
+      "epoch": 74.12464589235128,
+      "grad_norm": 3.345333962402265,
+      "learning_rate": 5e-06,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 1120489804,
+      "step": 6531
+    },
+    {
+      "epoch": 74.12464589235128,
+      "loss": 0.026301445439457893,
+      "loss_ce": 1.0550304978096392e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 1120489804,
+      "step": 6531
+    },
+    {
+      "epoch": 74.13597733711048,
+      "grad_norm": 3.3609352826898613,
+      "learning_rate": 5e-06,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 1120661268,
+      "step": 6532
+    },
+    {
+      "epoch": 74.13597733711048,
+      "loss": 0.10038816183805466,
+      "loss_ce": 1.5849176634219475e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 1120661268,
+      "step": 6532
+    },
+    {
+      "epoch": 74.14730878186968,
+      "grad_norm": 3.426619580685186,
+      "learning_rate": 5e-06,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 1120833440,
+      "step": 6533
+    },
+    {
+      "epoch": 74.14730878186968,
+      "loss": 0.027686811983585358,
+      "loss_ce": 2.262873385916464e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 1120833440,
+      "step": 6533
+    },
+    {
+      "epoch": 74.15864022662889,
+      "grad_norm": 3.5166757680499914,
+      "learning_rate": 5e-06,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 1121005280,
+      "step": 6534
+    },
+    {
+      "epoch": 74.15864022662889,
+      "loss": 0.03233487904071808,
+      "loss_ce": 1.6764171959948726e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1121005280,
+      "step": 6534
+    },
+    {
+      "epoch": 74.1699716713881,
+      "grad_norm": 4.6375297336868355,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 1121176436,
+      "step": 6535
+    },
+    {
+      "epoch": 74.1699716713881,
+      "loss": 0.06014683470129967,
+      "loss_ce": 2.7205438527744263e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 1121176436,
+      "step": 6535
+    },
+    {
+      "epoch": 74.1813031161473,
+      "grad_norm": 3.0243636504944513,
+      "learning_rate": 5e-06,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 1121348124,
+      "step": 6536
+    },
+    {
+      "epoch": 74.1813031161473,
+      "loss": 0.026662033051252365,
+      "loss_ce": 1.255888128071092e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 1121348124,
+      "step": 6536
+    },
+    {
+      "epoch": 74.19263456090651,
+      "grad_norm": 2.600093201754161,
+      "learning_rate": 5e-06,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 1121519948,
+      "step": 6537
+    },
+    {
+      "epoch": 74.19263456090651,
+      "loss": 0.033793188631534576,
+      "loss_ce": 2.549035161791835e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1121519948,
+      "step": 6537
+    },
+    {
+      "epoch": 74.20396600566572,
+      "grad_norm": 2.6841947473673113,
+      "learning_rate": 5e-06,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 1121691796,
+      "step": 6538
+    },
+    {
+      "epoch": 74.20396600566572,
+      "loss": 0.022242574021220207,
+      "loss_ce": 1.0517702321521938e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 1121691796,
+      "step": 6538
+    },
+    {
+      "epoch": 74.21529745042493,
+      "grad_norm": 2.9045826869659304,
+      "learning_rate": 5e-06,
+      "loss": 0.059,
+      "num_input_tokens_seen": 1121863312,
+      "step": 6539
+    },
+    {
+      "epoch": 74.21529745042493,
+      "loss": 0.053002290427684784,
+      "loss_ce": 2.377333294134587e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 1121863312,
+      "step": 6539
+    },
+    {
+      "epoch": 74.22662889518413,
+      "grad_norm": 2.644822881678392,
+      "learning_rate": 5e-06,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 1122035228,
+      "step": 6540
+    },
+    {
+      "epoch": 74.22662889518413,
+      "loss": 0.06761403381824493,
+      "loss_ce": 1.7599053535377607e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 1122035228,
+      "step": 6540
+    },
+    {
+      "epoch": 74.23796033994334,
+      "grad_norm": 2.866182283587977,
+      "learning_rate": 5e-06,
+      "loss": 0.0301,
+      "num_input_tokens_seen": 1122206848,
+      "step": 6541
+    },
+    {
+      "epoch": 74.23796033994334,
+      "loss": 0.034088488668203354,
+      "loss_ce": 1.56136757141212e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1122206848,
+      "step": 6541
+    },
+    {
+      "epoch": 74.24929178470255,
+      "grad_norm": 3.092252378469409,
+      "learning_rate": 5e-06,
+      "loss": 0.0304,
+      "num_input_tokens_seen": 1122378008,
+      "step": 6542
+    },
+    {
+      "epoch": 74.24929178470255,
+      "loss": 0.0221841000020504,
+      "loss_ce": 2.070872506010346e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.004425048828125,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 1122378008,
+      "step": 6542
+    },
+    {
+      "epoch": 74.26062322946176,
+      "grad_norm": 3.3128737241389437,
+      "learning_rate": 5e-06,
+      "loss": 0.034,
+      "num_input_tokens_seen": 1122548344,
+      "step": 6543
+    },
+    {
+      "epoch": 74.26062322946176,
+      "loss": 0.036753054708242416,
+      "loss_ce": 9.889366992865689e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1122548344,
+      "step": 6543
+    },
+    {
+      "epoch": 74.27195467422096,
+      "grad_norm": 3.0350620237385955,
+      "learning_rate": 5e-06,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 1122720252,
+      "step": 6544
+    },
+    {
+      "epoch": 74.27195467422096,
+      "loss": 0.03843335062265396,
+      "loss_ce": 1.9351271475898102e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1122720252,
+      "step": 6544
+    },
+    {
+      "epoch": 74.28328611898017,
+      "grad_norm": 2.880081483896375,
+      "learning_rate": 5e-06,
+      "loss": 0.048,
+      "num_input_tokens_seen": 1122890520,
+      "step": 6545
+    },
+    {
+      "epoch": 74.28328611898017,
+      "loss": 0.09767642617225647,
+      "loss_ce": 2.0175570170977153e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 1122890520,
+      "step": 6545
+    },
+    {
+      "epoch": 74.29461756373938,
+      "grad_norm": 5.350790177112331,
+      "learning_rate": 5e-06,
+      "loss": 0.047,
+      "num_input_tokens_seen": 1123061460,
+      "step": 6546
+    },
+    {
+      "epoch": 74.29461756373938,
+      "loss": 0.024730706587433815,
+      "loss_ce": 1.1468640877865255e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 1123061460,
+      "step": 6546
+    },
+    {
+      "epoch": 74.30594900849859,
+      "grad_norm": 2.9657035232887274,
+      "learning_rate": 5e-06,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 1123233092,
+      "step": 6547
+    },
+    {
+      "epoch": 74.30594900849859,
+      "loss": 0.039753399789333344,
+      "loss_ce": 1.951216472662054e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 1123233092,
+      "step": 6547
+    },
+    {
+      "epoch": 74.3172804532578,
+      "grad_norm": 3.127766017975965,
+      "learning_rate": 5e-06,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 1123401852,
+      "step": 6548
+    },
+    {
+      "epoch": 74.3172804532578,
+      "loss": 0.03144387528300285,
+      "loss_ce": 1.076921216736082e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 1123401852,
+      "step": 6548
+    },
+    {
+      "epoch": 74.328611898017,
+      "grad_norm": 3.0473684714921467,
+      "learning_rate": 5e-06,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 1123573480,
+      "step": 6549
+    },
+    {
+      "epoch": 74.328611898017,
+      "loss": 0.02534647099673748,
+      "loss_ce": 1.6881414921954274e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 1123573480,
+      "step": 6549
+    },
+    {
+      "epoch": 74.33994334277621,
+      "grad_norm": 2.8736710243541985,
+      "learning_rate": 5e-06,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 1123745096,
+      "step": 6550
+    },
+    {
+      "epoch": 74.33994334277621,
+      "loss": 0.06466744095087051,
+      "loss_ce": 1.5952380636008456e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 1123745096,
+      "step": 6550
+    },
+    {
+      "epoch": 74.35127478753542,
+      "grad_norm": 2.1977943765302212,
+      "learning_rate": 5e-06,
+      "loss": 0.027,
+      "num_input_tokens_seen": 1123916736,
+      "step": 6551
+    },
+    {
+      "epoch": 74.35127478753542,
+      "loss": 0.01735638827085495,
+      "loss_ce": 3.0034945666557178e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0034637451171875,
+      "loss_xval": 0.017333984375,
+      "num_input_tokens_seen": 1123916736,
+      "step": 6551
+    },
+    {
+      "epoch": 74.36260623229462,
+      "grad_norm": 2.1753931528727954,
+      "learning_rate": 5e-06,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 1124086696,
+      "step": 6552
+    },
+    {
+      "epoch": 74.36260623229462,
+      "loss": 0.03582029789686203,
+      "loss_ce": 7.922475560917519e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1124086696,
+      "step": 6552
+    },
+    {
+      "epoch": 74.37393767705383,
+      "grad_norm": 6.772045594611746,
+      "learning_rate": 5e-06,
+      "loss": 0.04,
+      "num_input_tokens_seen": 1124258948,
+      "step": 6553
+    },
+    {
+      "epoch": 74.37393767705383,
+      "loss": 0.030590107664465904,
+      "loss_ce": 1.912384141178336e-05,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1124258948,
+      "step": 6553
+    },
+    {
+      "epoch": 74.38526912181302,
+      "grad_norm": 3.1339157702441343,
+      "learning_rate": 5e-06,
+      "loss": 0.042,
+      "num_input_tokens_seen": 1124430820,
+      "step": 6554
+    },
+    {
+      "epoch": 74.38526912181302,
+      "loss": 0.07429294288158417,
+      "loss_ce": 1.3157720786693972e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 1124430820,
+      "step": 6554
+    },
+    {
+      "epoch": 74.39660056657223,
+      "grad_norm": 3.1952496038681786,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1124602920,
+      "step": 6555
+    },
+    {
+      "epoch": 74.39660056657223,
+      "loss": 0.03208663687109947,
+      "loss_ce": 1.2664571841014549e-05,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1124602920,
+      "step": 6555
+    },
+    {
+      "epoch": 74.40793201133144,
+      "grad_norm": 4.043248135548104,
+      "learning_rate": 5e-06,
+      "loss": 0.053,
+      "num_input_tokens_seen": 1124773948,
+      "step": 6556
+    },
+    {
+      "epoch": 74.40793201133144,
+      "loss": 0.03340116888284683,
+      "loss_ce": 7.309532975341426e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1124773948,
+      "step": 6556
+    },
+    {
+      "epoch": 74.41926345609065,
+      "grad_norm": 2.9120655542665035,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 1124946244,
+      "step": 6557
+    },
+    {
+      "epoch": 74.41926345609065,
+      "loss": 0.03089434839785099,
+      "loss_ce": 1.0557387213339098e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 1124946244,
+      "step": 6557
+    },
+    {
+      "epoch": 74.43059490084985,
+      "grad_norm": 3.376758922847797,
+      "learning_rate": 5e-06,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 1125118024,
+      "step": 6558
+    },
+    {
+      "epoch": 74.43059490084985,
+      "loss": 0.027084695175290108,
+      "loss_ce": 1.5603432984789833e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1125118024,
+      "step": 6558
+    },
+    {
+      "epoch": 74.44192634560906,
+      "grad_norm": 3.3811898333494916,
+      "learning_rate": 5e-06,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 1125287312,
+      "step": 6559
+    },
+    {
+      "epoch": 74.44192634560906,
+      "loss": 0.04278423264622688,
+      "loss_ce": 1.3848143680661451e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 1125287312,
+      "step": 6559
+    },
+    {
+      "epoch": 74.45325779036827,
+      "grad_norm": 3.295039162291036,
+      "learning_rate": 5e-06,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 1125456656,
+      "step": 6560
+    },
+    {
+      "epoch": 74.45325779036827,
+      "loss": 0.048660967499017715,
+      "loss_ce": 1.5948513464536518e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 1125456656,
+      "step": 6560
+    },
+    {
+      "epoch": 74.46458923512748,
+      "grad_norm": 4.141489323193328,
+      "learning_rate": 5e-06,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 1125627848,
+      "step": 6561
+    },
+    {
+      "epoch": 74.46458923512748,
+      "loss": 0.028617870062589645,
+      "loss_ce": 7.641958291060291e-06,
+      "loss_iou": 0.515625,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 1125627848,
+      "step": 6561
+    },
+    {
+      "epoch": 74.47592067988668,
+      "grad_norm": 2.987924891331927,
+      "learning_rate": 5e-06,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 1125797816,
+      "step": 6562
+    },
+    {
+      "epoch": 74.47592067988668,
+      "loss": 0.028473205864429474,
+      "loss_ce": 1.5564743080176413e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 1125797816,
+      "step": 6562
+    },
+    {
+      "epoch": 74.48725212464589,
+      "grad_norm": 2.881362799091236,
+      "learning_rate": 5e-06,
+      "loss": 0.03,
+      "num_input_tokens_seen": 1125968112,
+      "step": 6563
+    },
+    {
+      "epoch": 74.48725212464589,
+      "loss": 0.04204528033733368,
+      "loss_ce": 2.2578688003704883e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1125968112,
+      "step": 6563
+    },
+    {
+      "epoch": 74.4985835694051,
+      "grad_norm": 2.5457582393322116,
+      "learning_rate": 5e-06,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 1126139936,
+      "step": 6564
+    },
+    {
+      "epoch": 74.4985835694051,
+      "loss": 0.03218161314725876,
+      "loss_ce": 8.456287105218507e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1126139936,
+      "step": 6564
+    },
+    {
+      "epoch": 74.5099150141643,
+      "grad_norm": 3.242619201626592,
+      "learning_rate": 5e-06,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 1126311664,
+      "step": 6565
+    },
+    {
+      "epoch": 74.5099150141643,
+      "loss": 0.03974199295043945,
+      "loss_ce": 2.336677789571695e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 1126311664,
+      "step": 6565
+    },
+    {
+      "epoch": 74.52124645892351,
+      "grad_norm": 3.8942997766225345,
+      "learning_rate": 5e-06,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 1126483416,
+      "step": 6566
+    },
+    {
+      "epoch": 74.52124645892351,
+      "loss": 0.10694268345832825,
+      "loss_ce": 9.09206755750347e-06,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 1126483416,
+      "step": 6566
+    },
+    {
+      "epoch": 74.53257790368272,
+      "grad_norm": 2.8869488214410244,
+      "learning_rate": 5e-06,
+      "loss": 0.026,
+      "num_input_tokens_seen": 1126654220,
+      "step": 6567
+    },
+    {
+      "epoch": 74.53257790368272,
+      "loss": 0.021920349448919296,
+      "loss_ce": 8.728484317543916e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.00439453125,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 1126654220,
+      "step": 6567
+    },
+    {
+      "epoch": 74.54390934844193,
+      "grad_norm": 2.5256706389271173,
+      "learning_rate": 5e-06,
+      "loss": 0.06,
+      "num_input_tokens_seen": 1126825992,
+      "step": 6568
+    },
+    {
+      "epoch": 74.54390934844193,
+      "loss": 0.10881047695875168,
+      "loss_ce": 4.583359259413555e-05,
+      "loss_iou": 0.12158203125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 1126825992,
+      "step": 6568
+    },
+    {
+      "epoch": 74.55524079320114,
+      "grad_norm": 2.2031247186967446,
+      "learning_rate": 5e-06,
+      "loss": 0.0268,
+      "num_input_tokens_seen": 1126997916,
+      "step": 6569
+    },
+    {
+      "epoch": 74.55524079320114,
+      "loss": 0.03448734059929848,
+      "loss_ce": 5.5885753681650385e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 1126997916,
+      "step": 6569
+    },
+    {
+      "epoch": 74.56657223796034,
+      "grad_norm": 2.623623901419336,
+      "learning_rate": 5e-06,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 1127169720,
+      "step": 6570
+    },
+    {
+      "epoch": 74.56657223796034,
+      "loss": 0.09014096111059189,
+      "loss_ce": 2.2551414076588117e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 1127169720,
+      "step": 6570
+    },
+    {
+      "epoch": 74.57790368271955,
+      "grad_norm": 7.897089597829697,
+      "learning_rate": 5e-06,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 1127341224,
+      "step": 6571
+    },
+    {
+      "epoch": 74.57790368271955,
+      "loss": 0.03404213488101959,
+      "loss_ce": 1.5037266166473273e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1127341224,
+      "step": 6571
+    },
+    {
+      "epoch": 74.58923512747876,
+      "grad_norm": 3.261585871586935,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1127512132,
+      "step": 6572
+    },
+    {
+      "epoch": 74.58923512747876,
+      "loss": 0.03402257338166237,
+      "loss_ce": 1.0732976079452783e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1127512132,
+      "step": 6572
+    },
+    {
+      "epoch": 74.60056657223797,
+      "grad_norm": 2.919091564399831,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 1127683784,
+      "step": 6573
+    },
+    {
+      "epoch": 74.60056657223797,
+      "loss": 0.028045453131198883,
+      "loss_ce": 3.031630149052944e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 1127683784,
+      "step": 6573
+    },
+    {
+      "epoch": 74.61189801699717,
+      "grad_norm": 3.067986514972103,
+      "learning_rate": 5e-06,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 1127854752,
+      "step": 6574
+    },
+    {
+      "epoch": 74.61189801699717,
+      "loss": 0.08889612555503845,
+      "loss_ce": 2.8932770874234848e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 1127854752,
+      "step": 6574
+    },
+    {
+      "epoch": 74.62322946175638,
+      "grad_norm": 3.4098896002097594,
+      "learning_rate": 5e-06,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 1128026192,
+      "step": 6575
+    },
+    {
+      "epoch": 74.62322946175638,
+      "loss": 0.11072533577680588,
+      "loss_ce": 7.562517566839233e-06,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 1128026192,
+      "step": 6575
+    },
+    {
+      "epoch": 74.63456090651557,
+      "grad_norm": 3.129265554680959,
+      "learning_rate": 5e-06,
+      "loss": 0.048,
+      "num_input_tokens_seen": 1128198076,
+      "step": 6576
+    },
+    {
+      "epoch": 74.63456090651557,
+      "loss": 0.07285108417272568,
+      "loss_ce": 5.140165376360528e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 1128198076,
+      "step": 6576
+    },
+    {
+      "epoch": 74.64589235127478,
+      "grad_norm": 6.5740103890432335,
+      "learning_rate": 5e-06,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 1128368572,
+      "step": 6577
+    },
+    {
+      "epoch": 74.64589235127478,
+      "loss": 0.025021247565746307,
+      "loss_ce": 2.7350888558430597e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 1128368572,
+      "step": 6577
+    },
+    {
+      "epoch": 74.65722379603399,
+      "grad_norm": 3.0016955243098815,
+      "learning_rate": 5e-06,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 1128540320,
+      "step": 6578
+    },
+    {
+      "epoch": 74.65722379603399,
+      "loss": 0.06106710806488991,
+      "loss_ce": 6.246630800887942e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 1128540320,
+      "step": 6578
+    },
+    {
+      "epoch": 74.6685552407932,
+      "grad_norm": 2.9110242886428654,
+      "learning_rate": 5e-06,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 1128712092,
+      "step": 6579
+    },
+    {
+      "epoch": 74.6685552407932,
+      "loss": 0.03408673405647278,
+      "loss_ce": 1.3858853890269529e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 1128712092,
+      "step": 6579
+    },
+    {
+      "epoch": 74.6798866855524,
+      "grad_norm": 3.5661725253959573,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 1128882144,
+      "step": 6580
+    },
+    {
+      "epoch": 74.6798866855524,
+      "loss": 0.04359672963619232,
+      "loss_ce": 0.0007195360958576202,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 1128882144,
+      "step": 6580
+    },
+    {
+      "epoch": 74.69121813031161,
+      "grad_norm": 2.731680176846638,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 1129053908,
+      "step": 6581
+    },
+    {
+      "epoch": 74.69121813031161,
+      "loss": 0.06613113731145859,
+      "loss_ce": 1.4803251360717695e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 1129053908,
+      "step": 6581
+    },
+    {
+      "epoch": 74.70254957507082,
+      "grad_norm": 2.4624581703642607,
+      "learning_rate": 5e-06,
+      "loss": 0.052,
+      "num_input_tokens_seen": 1129224084,
+      "step": 6582
+    },
+    {
+      "epoch": 74.70254957507082,
+      "loss": 0.04896687716245651,
+      "loss_ce": 0.002076619304716587,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 1129224084,
+      "step": 6582
+    },
+    {
+      "epoch": 74.71388101983003,
+      "grad_norm": 2.2279051896986664,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1129395740,
+      "step": 6583
+    },
+    {
+      "epoch": 74.71388101983003,
+      "loss": 0.02393813617527485,
+      "loss_ce": 1.2354178579698782e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 1129395740,
+      "step": 6583
+    },
+    {
+      "epoch": 74.72521246458923,
+      "grad_norm": 2.214563796199562,
+      "learning_rate": 5e-06,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 1129567948,
+      "step": 6584
+    },
+    {
+      "epoch": 74.72521246458923,
+      "loss": 0.03841949999332428,
+      "loss_ce": 1.3128978025633842e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 1129567948,
+      "step": 6584
+    },
+    {
+      "epoch": 74.73654390934844,
+      "grad_norm": 2.4723858598221367,
+      "learning_rate": 5e-06,
+      "loss": 0.0248,
+      "num_input_tokens_seen": 1129738824,
+      "step": 6585
+    },
+    {
+      "epoch": 74.73654390934844,
+      "loss": 0.02932172268629074,
+      "loss_ce": 2.484569813532289e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 1129738824,
+      "step": 6585
+    },
+    {
+      "epoch": 74.74787535410765,
+      "grad_norm": 2.852778627022531,
+      "learning_rate": 5e-06,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 1129910312,
+      "step": 6586
+    },
+    {
+      "epoch": 74.74787535410765,
+      "loss": 0.03491400182247162,
+      "loss_ce": 3.240866135456599e-05,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 1129910312,
+      "step": 6586
+    },
+    {
+      "epoch": 74.75920679886686,
+      "grad_norm": 2.9994155111322427,
+      "learning_rate": 5e-06,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 1130080956,
+      "step": 6587
+    },
+    {
+      "epoch": 74.75920679886686,
+      "loss": 0.07636642456054688,
+      "loss_ce": 1.1439266017987393e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 1130080956,
+      "step": 6587
+    },
+    {
+      "epoch": 74.77053824362606,
+      "grad_norm": 3.705925377055238,
+      "learning_rate": 5e-06,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 1130251972,
+      "step": 6588
+    },
+    {
+      "epoch": 74.77053824362606,
+      "loss": 0.07284902036190033,
+      "loss_ce": 1.8822984202415682e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 1130251972,
+      "step": 6588
+    },
+    {
+      "epoch": 74.78186968838527,
+      "grad_norm": 3.104827330766222,
+      "learning_rate": 5e-06,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 1130422384,
+      "step": 6589
+    },
+    {
+      "epoch": 74.78186968838527,
+      "loss": 0.033627867698669434,
+      "loss_ce": 1.2753389455610886e-05,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1130422384,
+      "step": 6589
+    },
+    {
+      "epoch": 74.79320113314448,
+      "grad_norm": 3.38921178910297,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 1130594036,
+      "step": 6590
+    },
+    {
+      "epoch": 74.79320113314448,
+      "loss": 0.05058871954679489,
+      "loss_ce": 2.1094274416100234e-05,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 1130594036,
+      "step": 6590
+    },
+    {
+      "epoch": 74.80453257790369,
+      "grad_norm": 3.2946389524645587,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 1130764112,
+      "step": 6591
+    },
+    {
+      "epoch": 74.80453257790369,
+      "loss": 0.026711052283644676,
+      "loss_ce": 8.171422450686805e-06,
+      "loss_iou": 0.5,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 1130764112,
+      "step": 6591
+    },
+    {
+      "epoch": 74.8158640226629,
+      "grad_norm": 3.5085739993113663,
+      "learning_rate": 5e-06,
+      "loss": 0.035,
+      "num_input_tokens_seen": 1130934424,
+      "step": 6592
+    },
+    {
+      "epoch": 74.8158640226629,
+      "loss": 0.03191527724266052,
+      "loss_ce": 9.152002348855603e-06,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1130934424,
+      "step": 6592
+    },
+    {
+      "epoch": 74.8271954674221,
+      "grad_norm": 2.9625154039273123,
+      "learning_rate": 5e-06,
+      "loss": 0.032,
+      "num_input_tokens_seen": 1131106348,
+      "step": 6593
+    },
+    {
+      "epoch": 74.8271954674221,
+      "loss": 0.03468397259712219,
+      "loss_ce": 3.126372757833451e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1131106348,
+      "step": 6593
+    },
+    {
+      "epoch": 74.83852691218131,
+      "grad_norm": 2.238460167599019,
+      "learning_rate": 5e-06,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 1131277076,
+      "step": 6594
+    },
+    {
+      "epoch": 74.83852691218131,
+      "loss": 0.03013141080737114,
+      "loss_ce": 2.581985609140247e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 1131277076,
+      "step": 6594
+    },
+    {
+      "epoch": 74.84985835694052,
+      "grad_norm": 2.745086066254167,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 1131448800,
+      "step": 6595
+    },
+    {
+      "epoch": 74.84985835694052,
+      "loss": 0.0662766546010971,
+      "loss_ce": 0.0001145425922004506,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 1131448800,
+      "step": 6595
+    },
+    {
+      "epoch": 74.86118980169972,
+      "grad_norm": 4.161388633340837,
+      "learning_rate": 5e-06,
+      "loss": 0.039,
+      "num_input_tokens_seen": 1131620540,
+      "step": 6596
+    },
+    {
+      "epoch": 74.86118980169972,
+      "loss": 0.03462936729192734,
+      "loss_ce": 7.173584890551865e-06,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 1131620540,
+      "step": 6596
+    },
+    {
+      "epoch": 74.87252124645893,
+      "grad_norm": 3.942770467044283,
+      "learning_rate": 5e-06,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 1131792352,
+      "step": 6597
+    },
+    {
+      "epoch": 74.87252124645893,
+      "loss": 0.030213255435228348,
+      "loss_ce": 1.6111145669128746e-05,
+      "loss_iou": 0.59375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 1131792352,
+      "step": 6597
+    },
+    {
+      "epoch": 74.88385269121812,
+      "grad_norm": 6.28398597284707,
+      "learning_rate": 5e-06,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 1131963328,
+      "step": 6598
+    },
+    {
+      "epoch": 74.88385269121812,
+      "loss": 0.04276979714632034,
+      "loss_ce": 1.4671622921014205e-05,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 1131963328,
+      "step": 6598
+    },
+    {
+      "epoch": 74.89518413597733,
+      "grad_norm": 3.117787571982213,
+      "learning_rate": 5e-06,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 1132134912,
+      "step": 6599
+    },
+    {
+      "epoch": 74.89518413597733,
+      "loss": 0.02718474715948105,
+      "loss_ce": 8.842358511174098e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 1132134912,
+      "step": 6599
+    },
+    {
+      "epoch": 74.90651558073654,
+      "grad_norm": 3.377153752117538,
+      "learning_rate": 5e-06,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 1132306592,
+      "step": 6600
+    },
+    {
+      "epoch": 74.90651558073654,
+      "loss": 0.04378780722618103,
+      "loss_ce": 1.0341590495954733e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 1132306592,
+      "step": 6600
+    },
+    {
+      "epoch": 74.91784702549575,
+      "grad_norm": 3.2312934907367965,
+      "learning_rate": 5e-06,
+      "loss": 0.0304,
+      "num_input_tokens_seen": 1132477732,
+      "step": 6601
+    },
+    {
+      "epoch": 74.91784702549575,
+      "loss": 0.02760060504078865,
+      "loss_ce": 1.2714146578218788e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 1132477732,
+      "step": 6601
+    },
+    {
+      "epoch": 74.92917847025495,
+      "grad_norm": 2.9390795376256085,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 1132648924,
+      "step": 6602
+    },
+    {
+      "epoch": 74.92917847025495,
+      "loss": 0.11069358885288239,
+      "loss_ce": 2.158638380933553e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 1132648924,
+      "step": 6602
+    },
+    {
+      "epoch": 74.94050991501416,
+      "grad_norm": 2.675223345899106,
+      "learning_rate": 5e-06,
+      "loss": 0.046,
+      "num_input_tokens_seen": 1132820768,
+      "step": 6603
+    },
+    {
+      "epoch": 74.94050991501416,
+      "loss": 0.06753931939601898,
+      "loss_ce": 1.1552236173884012e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 1132820768,
+      "step": 6603
+    },
+    {
+      "epoch": 74.95184135977337,
+      "grad_norm": 3.6814864893805255,
+      "learning_rate": 5e-06,
+      "loss": 0.035,
+      "num_input_tokens_seen": 1132992448,
+      "step": 6604
+    },
+    {
+      "epoch": 74.95184135977337,
+      "loss": 0.05241741985082626,
+      "loss_ce": 3.399672277737409e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 1132992448,
+      "step": 6604
+    },
+    {
+      "epoch": 74.96317280453258,
+      "grad_norm": 3.4656831250717595,
+      "learning_rate": 5e-06,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 1133163448,
+      "step": 6605
+    },
+    {
+      "epoch": 74.96317280453258,
+      "loss": 0.044289443641901016,
+      "loss_ce": 8.437264114036225e-06,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 1133163448,
+      "step": 6605
+    },
+    {
+      "epoch": 74.97450424929178,
+      "grad_norm": 3.4958870464672467,
+      "learning_rate": 5e-06,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 1133334400,
+      "step": 6606
+    },
+    {
+      "epoch": 74.97450424929178,
+      "loss": 0.04712475836277008,
+      "loss_ce": 2.087727261823602e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 1133334400,
+      "step": 6606
+    },
+    {
+      "epoch": 74.98583569405099,
+      "grad_norm": 3.188127732940439,
+      "learning_rate": 5e-06,
+      "loss": 0.051,
+      "num_input_tokens_seen": 1133506188,
+      "step": 6607
+    },
+    {
+      "epoch": 74.98583569405099,
+      "loss": 0.043926019221544266,
+      "loss_ce": 1.1226005881326273e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 1133506188,
+      "step": 6607
+    },
+    {
+      "epoch": 74.9971671388102,
+      "grad_norm": 3.032565899470877,
+      "learning_rate": 5e-06,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 1133674920,
+      "step": 6608
+    },
+    {
+      "epoch": 74.9971671388102,
+      "loss": 0.058987658470869064,
+      "loss_ce": 4.2952389776473865e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 1133674920,
+      "step": 6608
+    },
+    {
+      "epoch": 74.9971671388102,
+      "loss": 0.040857791900634766,
+      "loss_ce": 1.0014868166763335e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1133717820,
+      "step": 6608
+    },
+    {
+      "epoch": 75.0084985835694,
+      "grad_norm": 2.758072465777096,
+      "learning_rate": 5e-06,
+      "loss": 0.0294,
+      "num_input_tokens_seen": 1133846548,
+      "step": 6609
+    },
+    {
+      "epoch": 75.0084985835694,
+      "loss": 0.02699139714241028,
+      "loss_ce": 2.9117847589077428e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 1133846548,
+      "step": 6609
+    },
+    {
+      "epoch": 75.01983002832861,
+      "grad_norm": 2.7561372054264255,
+      "learning_rate": 5e-06,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 1134017688,
+      "step": 6610
+    },
+    {
+      "epoch": 75.01983002832861,
+      "loss": 0.029750000685453415,
+      "loss_ce": 1.0622742593113799e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 1134017688,
+      "step": 6610
+    },
+    {
+      "epoch": 75.03116147308782,
+      "grad_norm": 2.7277961566773232,
+      "learning_rate": 5e-06,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 1134189520,
+      "step": 6611
+    },
+    {
+      "epoch": 75.03116147308782,
+      "loss": 0.03339288383722305,
+      "loss_ce": 5.243175837676972e-05,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 1134189520,
+      "step": 6611
+    },
+    {
+      "epoch": 75.04249291784703,
+      "grad_norm": 3.3672931179972703,
+      "learning_rate": 5e-06,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 1134360152,
+      "step": 6612
+    },
+    {
+      "epoch": 75.04249291784703,
+      "loss": 0.024752387776970863,
+      "loss_ce": 1.026110567181604e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 1134360152,
+      "step": 6612
+    },
+    {
+      "epoch": 75.05382436260624,
+      "grad_norm": 2.8869852001999785,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 1134531532,
+      "step": 6613
+    },
+    {
+      "epoch": 75.05382436260624,
+      "loss": 0.027328990399837494,
+      "loss_ce": 3.1015144486445934e-05,
+      "loss_iou": 0.6875,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 1134531532,
+      "step": 6613
+    },
+    {
+      "epoch": 75.06515580736544,
+      "grad_norm": 2.999026999517069,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 1134703164,
+      "step": 6614
+    },
+    {
+      "epoch": 75.06515580736544,
+      "loss": 0.14103229343891144,
+      "loss_ce": 1.0569621736067347e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 1134703164,
+      "step": 6614
+    },
+    {
+      "epoch": 75.07648725212465,
+      "grad_norm": 6.40104962066838,
+      "learning_rate": 5e-06,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 1134875064,
+      "step": 6615
+    },
+    {
+      "epoch": 75.07648725212465,
+      "loss": 0.030785927549004555,
+      "loss_ce": 8.950310984801035e-06,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 1134875064,
+      "step": 6615
+    },
+    {
+      "epoch": 75.08781869688386,
+      "grad_norm": 3.9704458714465303,
+      "learning_rate": 5e-06,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 1135046852,
+      "step": 6616
+    },
+    {
+      "epoch": 75.08781869688386,
+      "loss": 0.041131727397441864,
+      "loss_ce": 9.290861271438189e-06,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1135046852,
+      "step": 6616
+    },
+    {
+      "epoch": 75.09915014164307,
+      "grad_norm": 3.6034962388222778,
+      "learning_rate": 5e-06,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 1135219272,
+      "step": 6617
+    },
+    {
+      "epoch": 75.09915014164307,
+      "loss": 0.03451300039887428,
+      "loss_ce": 1.2876675100415014e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 1135219272,
+      "step": 6617
+    },
+    {
+      "epoch": 75.11048158640227,
+      "grad_norm": 3.054477993280501,
+      "learning_rate": 5e-06,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 1135389292,
+      "step": 6618
+    },
+    {
+      "epoch": 75.11048158640227,
+      "loss": 0.11907505989074707,
+      "loss_ce": 1.072060149454046e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 1135389292,
+      "step": 6618
+    },
+    {
+      "epoch": 75.12181303116148,
+      "grad_norm": 3.0063760272545856,
+      "learning_rate": 5e-06,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 1135560732,
+      "step": 6619
+    },
+    {
+      "epoch": 75.12181303116148,
+      "loss": 0.0369563102722168,
+      "loss_ce": 2.2411819372791797e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1135560732,
+      "step": 6619
+    },
+    {
+      "epoch": 75.13314447592067,
+      "grad_norm": 3.349644319625914,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 1135731944,
+      "step": 6620
+    },
+    {
+      "epoch": 75.13314447592067,
+      "loss": 0.026946287602186203,
+      "loss_ce": 1.4525796359521337e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 1135731944,
+      "step": 6620
+    },
+    {
+      "epoch": 75.14447592067988,
+      "grad_norm": 3.4479739877604243,
+      "learning_rate": 5e-06,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 1135903928,
+      "step": 6621
+    },
+    {
+      "epoch": 75.14447592067988,
+      "loss": 0.06711386889219284,
+      "loss_ce": 2.097502874676138e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 1135903928,
+      "step": 6621
+    },
+    {
+      "epoch": 75.15580736543909,
+      "grad_norm": 3.7447745089891695,
+      "learning_rate": 5e-06,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 1136075736,
+      "step": 6622
+    },
+    {
+      "epoch": 75.15580736543909,
+      "loss": 0.02858702465891838,
+      "loss_ce": 1.4943962924007792e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 1136075736,
+      "step": 6622
+    },
+    {
+      "epoch": 75.1671388101983,
+      "grad_norm": 4.059687802559946,
+      "learning_rate": 5e-06,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 1136244004,
+      "step": 6623
+    },
+    {
+      "epoch": 75.1671388101983,
+      "loss": 0.027831776067614555,
+      "loss_ce": 7.375477707682876e-06,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 1136244004,
+      "step": 6623
+    },
+    {
+      "epoch": 75.1784702549575,
+      "grad_norm": 3.3001793878709997,
+      "learning_rate": 5e-06,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 1136415752,
+      "step": 6624
+    },
+    {
+      "epoch": 75.1784702549575,
+      "loss": 0.028650036081671715,
+      "loss_ce": 9.288190994993784e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 1136415752,
+      "step": 6624
+    },
+    {
+      "epoch": 75.18980169971671,
+      "grad_norm": 2.9926771387093587,
+      "learning_rate": 5e-06,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 1136587840,
+      "step": 6625
+    },
+    {
+      "epoch": 75.18980169971671,
+      "loss": 0.02985565736889839,
+      "loss_ce": 1.7094267604989e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 1136587840,
+      "step": 6625
+    },
+    {
+      "epoch": 75.20113314447592,
+      "grad_norm": 3.011683975089285,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 1136759296,
+      "step": 6626
+    },
+    {
+      "epoch": 75.20113314447592,
+      "loss": 0.036683566868305206,
+      "loss_ce": 1.6696771126589738e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1136759296,
+      "step": 6626
+    },
+    {
+      "epoch": 75.21246458923513,
+      "grad_norm": 2.589423643540145,
+      "learning_rate": 5e-06,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 1136930908,
+      "step": 6627
+    },
+    {
+      "epoch": 75.21246458923513,
+      "loss": 0.028652256354689598,
+      "loss_ce": 2.6767382223624736e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 1136930908,
+      "step": 6627
+    },
+    {
+      "epoch": 75.22379603399433,
+      "grad_norm": 3.0961173867377507,
+      "learning_rate": 5e-06,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 1137102796,
+      "step": 6628
+    },
+    {
+      "epoch": 75.22379603399433,
+      "loss": 0.05981450155377388,
+      "loss_ce": 1.5307865396607667e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 1137102796,
+      "step": 6628
+    },
+    {
+      "epoch": 75.23512747875354,
+      "grad_norm": 3.2605992830967643,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 1137274736,
+      "step": 6629
+    },
+    {
+      "epoch": 75.23512747875354,
+      "loss": 0.04676087200641632,
+      "loss_ce": 7.94199058873346e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 1137274736,
+      "step": 6629
+    },
+    {
+      "epoch": 75.24645892351275,
+      "grad_norm": 3.5075056754683884,
+      "learning_rate": 5e-06,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 1137446040,
+      "step": 6630
+    },
+    {
+      "epoch": 75.24645892351275,
+      "loss": 0.0396280474960804,
+      "loss_ce": 1.623117714188993e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 1137446040,
+      "step": 6630
+    },
+    {
+      "epoch": 75.25779036827196,
+      "grad_norm": 3.203424984960221,
+      "learning_rate": 5e-06,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 1137617756,
+      "step": 6631
+    },
+    {
+      "epoch": 75.25779036827196,
+      "loss": 0.03802847862243652,
+      "loss_ce": 1.883343793451786e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 1137617756,
+      "step": 6631
+    },
+    {
+      "epoch": 75.26912181303116,
+      "grad_norm": 3.59352516117109,
+      "learning_rate": 5e-06,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 1137788248,
+      "step": 6632
+    },
+    {
+      "epoch": 75.26912181303116,
+      "loss": 0.061858586966991425,
+      "loss_ce": 1.4712988559040241e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 1137788248,
+      "step": 6632
+    },
+    {
+      "epoch": 75.28045325779037,
+      "grad_norm": 3.7772396063241342,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 1137959916,
+      "step": 6633
+    },
+    {
+      "epoch": 75.28045325779037,
+      "loss": 0.03259987384080887,
+      "loss_ce": 2.2356794943334535e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 1137959916,
+      "step": 6633
+    },
+    {
+      "epoch": 75.29178470254958,
+      "grad_norm": 2.8745239306164696,
+      "learning_rate": 5e-06,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 1138130020,
+      "step": 6634
+    },
+    {
+      "epoch": 75.29178470254958,
+      "loss": 0.034460075199604034,
+      "loss_ce": 5.150468132342212e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 1138130020,
+      "step": 6634
+    },
+    {
+      "epoch": 75.30311614730878,
+      "grad_norm": 3.3563284691669337,
+      "learning_rate": 5e-06,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 1138302012,
+      "step": 6635
+    },
+    {
+      "epoch": 75.30311614730878,
+      "loss": 0.04271052032709122,
+      "loss_ce": 1.6429548850283027e-05,
+      "loss_iou": 0.53125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 1138302012,
+      "step": 6635
+    },
+    {
+      "epoch": 75.31444759206799,
+      "grad_norm": 3.803427072091682,
+      "learning_rate": 5e-06,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 1138474204,
+      "step": 6636
+    },
+    {
+      "epoch": 75.31444759206799,
+      "loss": 0.040617987513542175,
+      "loss_ce": 1.4352742255141493e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1138474204,
+      "step": 6636
+    },
+    {
+      "epoch": 75.3257790368272,
+      "grad_norm": 3.6134514059485423,
+      "learning_rate": 5e-06,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 1138646060,
+      "step": 6637
+    },
+    {
+      "epoch": 75.3257790368272,
+      "loss": 0.02564416080713272,
+      "loss_ce": 9.39374240260804e-06,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 1138646060,
+      "step": 6637
+    },
+    {
+      "epoch": 75.33711048158641,
+      "grad_norm": 3.6706366778505313,
+      "learning_rate": 5e-06,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 1138816984,
+      "step": 6638
+    },
+    {
+      "epoch": 75.33711048158641,
+      "loss": 0.044599637389183044,
+      "loss_ce": 1.3456691704050172e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 1138816984,
+      "step": 6638
+    },
+    {
+      "epoch": 75.34844192634561,
+      "grad_norm": 2.9328785578990813,
+      "learning_rate": 5e-06,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 1138989100,
+      "step": 6639
+    },
+    {
+      "epoch": 75.34844192634561,
+      "loss": 0.026982787996530533,
+      "loss_ce": 3.5766301152762026e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 1138989100,
+      "step": 6639
+    },
+    {
+      "epoch": 75.35977337110482,
+      "grad_norm": 2.7000132793099856,
+      "learning_rate": 5e-06,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 1139160796,
+      "step": 6640
+    },
+    {
+      "epoch": 75.35977337110482,
+      "loss": 0.03656289726495743,
+      "loss_ce": 1.8098895452567376e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 1139160796,
+      "step": 6640
+    },
+    {
+      "epoch": 75.37110481586403,
+      "grad_norm": 2.7058270581428956,
+      "learning_rate": 5e-06,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 1139332176,
+      "step": 6641
+    },
+    {
+      "epoch": 75.37110481586403,
+      "loss": 0.025388851761817932,
+      "loss_ce": 1.3486058378475718e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 1139332176,
+      "step": 6641
+    },
+    {
+      "epoch": 75.38243626062322,
+      "grad_norm": 3.201883920930135,
+      "learning_rate": 5e-06,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 1139503828,
+      "step": 6642
+    },
+    {
+      "epoch": 75.38243626062322,
+      "loss": 0.02916206791996956,
+      "loss_ce": 3.304060373920947e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 1139503828,
+      "step": 6642
+    },
+    {
+      "epoch": 75.39376770538243,
+      "grad_norm": 2.5807513155674044,
+      "learning_rate": 5e-06,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 1139674364,
+      "step": 6643
+    },
+    {
+      "epoch": 75.39376770538243,
+      "loss": 0.07299471646547318,
+      "loss_ce": 3.4817956475308165e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 1139674364,
+      "step": 6643
+    },
+    {
+      "epoch": 75.40509915014164,
+      "grad_norm": 2.3856992311643794,
+      "learning_rate": 5e-06,
+      "loss": 0.063,
+      "num_input_tokens_seen": 1139845776,
+      "step": 6644
+    },
+    {
+      "epoch": 75.40509915014164,
+      "loss": 0.10186062008142471,
+      "loss_ce": 5.397787026595324e-05,
+      "loss_iou": 0.12109375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 1139845776,
+      "step": 6644
+    },
+    {
+      "epoch": 75.41643059490085,
+      "grad_norm": 1.9521959466451755,
+      "learning_rate": 5e-06,
+      "loss": 0.0302,
+      "num_input_tokens_seen": 1140017744,
+      "step": 6645
+    },
+    {
+      "epoch": 75.41643059490085,
+      "loss": 0.02789829671382904,
+      "loss_ce": 2.048969327006489e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 1140017744,
+      "step": 6645
+    },
+    {
+      "epoch": 75.42776203966005,
+      "grad_norm": 2.3466580034481934,
+      "learning_rate": 5e-06,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 1140189228,
+      "step": 6646
+    },
+    {
+      "epoch": 75.42776203966005,
+      "loss": 0.030522137880325317,
+      "loss_ce": 1.9818293367279693e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 1140189228,
+      "step": 6646
+    },
+    {
+      "epoch": 75.43909348441926,
+      "grad_norm": 2.8872076815482277,
+      "learning_rate": 5e-06,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 1140359360,
+      "step": 6647
+    },
+    {
+      "epoch": 75.43909348441926,
+      "loss": 0.024180948734283447,
+      "loss_ce": 1.1026722859242e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 1140359360,
+      "step": 6647
+    },
+    {
+      "epoch": 75.45042492917847,
+      "grad_norm": 3.031967190159842,
+      "learning_rate": 5e-06,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 1140531128,
+      "step": 6648
+    },
+    {
+      "epoch": 75.45042492917847,
+      "loss": 0.03687503561377525,
+      "loss_ce": 2.5059895051526837e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1140531128,
+      "step": 6648
+    },
+    {
+      "epoch": 75.46175637393767,
+      "grad_norm": 2.9603860164549545,
+      "learning_rate": 5e-06,
+      "loss": 0.033,
+      "num_input_tokens_seen": 1140702992,
+      "step": 6649
+    },
+    {
+      "epoch": 75.46175637393767,
+      "loss": 0.029325708746910095,
+      "loss_ce": 2.12053710129112e-05,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 1140702992,
+      "step": 6649
+    },
+    {
+      "epoch": 75.47308781869688,
+      "grad_norm": 3.2910846177482047,
+      "learning_rate": 5e-06,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 1140874604,
+      "step": 6650
+    },
+    {
+      "epoch": 75.47308781869688,
+      "loss": 0.028882550075650215,
+      "loss_ce": 2.8181069865240715e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 1140874604,
+      "step": 6650
+    },
+    {
+      "epoch": 75.48441926345609,
+      "grad_norm": 3.5063045319901307,
+      "learning_rate": 5e-06,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 1141044580,
+      "step": 6651
+    },
+    {
+      "epoch": 75.48441926345609,
+      "loss": 0.03537952899932861,
+      "loss_ce": 2.49135719059268e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 1141044580,
+      "step": 6651
+    },
+    {
+      "epoch": 75.4957507082153,
+      "grad_norm": 4.529549886685166,
+      "learning_rate": 5e-06,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 1141216568,
+      "step": 6652
+    },
+    {
+      "epoch": 75.4957507082153,
+      "loss": 0.0756077691912651,
+      "loss_ce": 4.624552093446255e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 1141216568,
+      "step": 6652
+    },
+    {
+      "epoch": 75.5070821529745,
+      "grad_norm": 5.702531039173888,
+      "learning_rate": 5e-06,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 1141388436,
+      "step": 6653
+    },
+    {
+      "epoch": 75.5070821529745,
+      "loss": 0.041098494082689285,
+      "loss_ce": 2.1833089704159647e-05,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1141388436,
+      "step": 6653
+    },
+    {
+      "epoch": 75.51841359773371,
+      "grad_norm": 4.173235326074557,
+      "learning_rate": 5e-06,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 1141559964,
+      "step": 6654
+    },
+    {
+      "epoch": 75.51841359773371,
+      "loss": 0.06591375172138214,
+      "loss_ce": 1.1045767678297125e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 1141559964,
+      "step": 6654
+    },
+    {
+      "epoch": 75.52974504249292,
+      "grad_norm": 3.272257811938323,
+      "learning_rate": 5e-06,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 1141730152,
+      "step": 6655
+    },
+    {
+      "epoch": 75.52974504249292,
+      "loss": 0.03391244262456894,
+      "loss_ce": 7.413479579554405e-06,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1141730152,
+      "step": 6655
+    },
+    {
+      "epoch": 75.54107648725213,
+      "grad_norm": 3.8810884567390467,
+      "learning_rate": 5e-06,
+      "loss": 0.051,
+      "num_input_tokens_seen": 1141902064,
+      "step": 6656
+    },
+    {
+      "epoch": 75.54107648725213,
+      "loss": 0.032190896570682526,
+      "loss_ce": 1.0111092706210911e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 1141902064,
+      "step": 6656
+    },
+    {
+      "epoch": 75.55240793201133,
+      "grad_norm": 3.7569852476853343,
+      "learning_rate": 5e-06,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 1142073736,
+      "step": 6657
+    },
+    {
+      "epoch": 75.55240793201133,
+      "loss": 0.03570391237735748,
+      "loss_ce": 1.3606133506982587e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 1142073736,
+      "step": 6657
+    },
+    {
+      "epoch": 75.56373937677054,
+      "grad_norm": 3.2002252008567087,
+      "learning_rate": 5e-06,
+      "loss": 0.039,
+      "num_input_tokens_seen": 1142244908,
+      "step": 6658
+    },
+    {
+      "epoch": 75.56373937677054,
+      "loss": 0.053096137940883636,
+      "loss_ce": 1.0810408639372326e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 1142244908,
+      "step": 6658
+    },
+    {
+      "epoch": 75.57507082152975,
+      "grad_norm": 3.2672547875452085,
+      "learning_rate": 5e-06,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 1142416592,
+      "step": 6659
+    },
+    {
+      "epoch": 75.57507082152975,
+      "loss": 0.06007232889533043,
+      "loss_ce": 2.8991431463509798e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 1142416592,
+      "step": 6659
+    },
+    {
+      "epoch": 75.58640226628896,
+      "grad_norm": 2.2120304026232414,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1142588436,
+      "step": 6660
+    },
+    {
+      "epoch": 75.58640226628896,
+      "loss": 0.03027614951133728,
+      "loss_ce": 7.900712080299854e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 1142588436,
+      "step": 6660
+    },
+    {
+      "epoch": 75.59773371104816,
+      "grad_norm": 2.119201606880293,
+      "learning_rate": 5e-06,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 1142760220,
+      "step": 6661
+    },
+    {
+      "epoch": 75.59773371104816,
+      "loss": 0.027178173884749413,
+      "loss_ce": 3.278976146248169e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1142760220,
+      "step": 6661
+    },
+    {
+      "epoch": 75.60906515580737,
+      "grad_norm": 2.403888445968798,
+      "learning_rate": 5e-06,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 1142932264,
+      "step": 6662
+    },
+    {
+      "epoch": 75.60906515580737,
+      "loss": 0.03135679289698601,
+      "loss_ce": 3.049948463740293e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1142932264,
+      "step": 6662
+    },
+    {
+      "epoch": 75.62039660056658,
+      "grad_norm": 3.39996957938274,
+      "learning_rate": 5e-06,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 1143104196,
+      "step": 6663
+    },
+    {
+      "epoch": 75.62039660056658,
+      "loss": 0.027434950694441795,
+      "loss_ce": 1.4907400327501819e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 1143104196,
+      "step": 6663
+    },
+    {
+      "epoch": 75.63172804532577,
+      "grad_norm": 3.437044465632727,
+      "learning_rate": 5e-06,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 1143276156,
+      "step": 6664
+    },
+    {
+      "epoch": 75.63172804532577,
+      "loss": 0.0396089032292366,
+      "loss_ce": 4.2866828152909875e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 1143276156,
+      "step": 6664
+    },
+    {
+      "epoch": 75.64305949008498,
+      "grad_norm": 2.812134809677353,
+      "learning_rate": 5e-06,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 1143446468,
+      "step": 6665
+    },
+    {
+      "epoch": 75.64305949008498,
+      "loss": 0.03706753998994827,
+      "loss_ce": 3.4461074392311275e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 1143446468,
+      "step": 6665
+    },
+    {
+      "epoch": 75.65439093484419,
+      "grad_norm": 2.9006621397520758,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1143618708,
+      "step": 6666
+    },
+    {
+      "epoch": 75.65439093484419,
+      "loss": 0.04202726483345032,
+      "loss_ce": 1.9816994608845562e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1143618708,
+      "step": 6666
+    },
+    {
+      "epoch": 75.6657223796034,
+      "grad_norm": 3.3625161917795054,
+      "learning_rate": 5e-06,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 1143789444,
+      "step": 6667
+    },
+    {
+      "epoch": 75.6657223796034,
+      "loss": 0.05386757850646973,
+      "loss_ce": 0.00023293087724596262,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 1143789444,
+      "step": 6667
+    },
+    {
+      "epoch": 75.6770538243626,
+      "grad_norm": 3.7325123012343937,
+      "learning_rate": 5e-06,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 1143960568,
+      "step": 6668
+    },
+    {
+      "epoch": 75.6770538243626,
+      "loss": 0.04736229404807091,
+      "loss_ce": 1.427180086466251e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 1143960568,
+      "step": 6668
+    },
+    {
+      "epoch": 75.68838526912181,
+      "grad_norm": 5.078434440900139,
+      "learning_rate": 5e-06,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 1144132432,
+      "step": 6669
+    },
+    {
+      "epoch": 75.68838526912181,
+      "loss": 0.0710034891963005,
+      "loss_ce": 1.9605802663136274e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 1144132432,
+      "step": 6669
+    },
+    {
+      "epoch": 75.69971671388102,
+      "grad_norm": 4.1809444971230505,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1144302040,
+      "step": 6670
+    },
+    {
+      "epoch": 75.69971671388102,
+      "loss": 0.030401837080717087,
+      "loss_ce": 2.1587624360108748e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 1144302040,
+      "step": 6670
+    },
+    {
+      "epoch": 75.71104815864022,
+      "grad_norm": 3.175617142640843,
+      "learning_rate": 5e-06,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 1144473412,
+      "step": 6671
+    },
+    {
+      "epoch": 75.71104815864022,
+      "loss": 0.02501879446208477,
+      "loss_ce": 9.640024472901132e-06,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 1144473412,
+      "step": 6671
+    },
+    {
+      "epoch": 75.72237960339943,
+      "grad_norm": 3.2636468707945774,
+      "learning_rate": 5e-06,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 1144645004,
+      "step": 6672
+    },
+    {
+      "epoch": 75.72237960339943,
+      "loss": 0.07613909244537354,
+      "loss_ce": 5.8768029703060165e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 1144645004,
+      "step": 6672
+    },
+    {
+      "epoch": 75.73371104815864,
+      "grad_norm": 2.173765635881127,
+      "learning_rate": 5e-06,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 1144816820,
+      "step": 6673
+    },
+    {
+      "epoch": 75.73371104815864,
+      "loss": 0.027086056768894196,
+      "loss_ce": 2.459430106682703e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 1144816820,
+      "step": 6673
+    },
+    {
+      "epoch": 75.74504249291785,
+      "grad_norm": 2.5653880769482424,
+      "learning_rate": 5e-06,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 1144988400,
+      "step": 6674
+    },
+    {
+      "epoch": 75.74504249291785,
+      "loss": 0.04108980670571327,
+      "loss_ce": 1.3146077435521875e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 1144988400,
+      "step": 6674
+    },
+    {
+      "epoch": 75.75637393767705,
+      "grad_norm": 3.112950775261105,
+      "learning_rate": 5e-06,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 1145160336,
+      "step": 6675
+    },
+    {
+      "epoch": 75.75637393767705,
+      "loss": 0.03883012756705284,
+      "loss_ce": 1.1768001058953814e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 1145160336,
+      "step": 6675
+    },
+    {
+      "epoch": 75.76770538243626,
+      "grad_norm": 4.058973674720043,
+      "learning_rate": 5e-06,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 1145332772,
+      "step": 6676
+    },
+    {
+      "epoch": 75.76770538243626,
+      "loss": 0.15601879358291626,
+      "loss_ce": 1.2941011846123729e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 1145332772,
+      "step": 6676
+    },
+    {
+      "epoch": 75.77903682719547,
+      "grad_norm": 3.4316266732291565,
+      "learning_rate": 5e-06,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 1145503140,
+      "step": 6677
+    },
+    {
+      "epoch": 75.77903682719547,
+      "loss": 0.028674378991127014,
+      "loss_ce": 1.0742652193584945e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 1145503140,
+      "step": 6677
+    },
+    {
+      "epoch": 75.79036827195468,
+      "grad_norm": 2.6864191523547776,
+      "learning_rate": 5e-06,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 1145674124,
+      "step": 6678
+    },
+    {
+      "epoch": 75.79036827195468,
+      "loss": 0.03138051554560661,
+      "loss_ce": 8.442595571978018e-06,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1145674124,
+      "step": 6678
+    },
+    {
+      "epoch": 75.80169971671388,
+      "grad_norm": 2.585479455315279,
+      "learning_rate": 5e-06,
+      "loss": 0.0254,
+      "num_input_tokens_seen": 1145845716,
+      "step": 6679
+    },
+    {
+      "epoch": 75.80169971671388,
+      "loss": 0.024859212338924408,
+      "loss_ce": 7.893960719229653e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 1145845716,
+      "step": 6679
+    },
+    {
+      "epoch": 75.81303116147309,
+      "grad_norm": 2.9421066973364973,
+      "learning_rate": 5e-06,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 1146016148,
+      "step": 6680
+    },
+    {
+      "epoch": 75.81303116147309,
+      "loss": 0.06626441329717636,
+      "loss_ce": 1.0752217349363491e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 1146016148,
+      "step": 6680
+    },
+    {
+      "epoch": 75.8243626062323,
+      "grad_norm": 3.2144953504060023,
+      "learning_rate": 5e-06,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 1146188196,
+      "step": 6681
+    },
+    {
+      "epoch": 75.8243626062323,
+      "loss": 0.024401476606726646,
+      "loss_ce": 7.896759052528068e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 1146188196,
+      "step": 6681
+    },
+    {
+      "epoch": 75.8356940509915,
+      "grad_norm": 3.259132412895314,
+      "learning_rate": 5e-06,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 1146359360,
+      "step": 6682
+    },
+    {
+      "epoch": 75.8356940509915,
+      "loss": 0.02609214186668396,
+      "loss_ce": 1.4871469829813577e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 1146359360,
+      "step": 6682
+    },
+    {
+      "epoch": 75.84702549575071,
+      "grad_norm": 3.5376500694855406,
+      "learning_rate": 5e-06,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 1146531176,
+      "step": 6683
+    },
+    {
+      "epoch": 75.84702549575071,
+      "loss": 0.04434838145971298,
+      "loss_ce": 2.1600109903374687e-05,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 1146531176,
+      "step": 6683
+    },
+    {
+      "epoch": 75.85835694050992,
+      "grad_norm": 3.174695679723189,
+      "learning_rate": 5e-06,
+      "loss": 0.03,
+      "num_input_tokens_seen": 1146701748,
+      "step": 6684
+    },
+    {
+      "epoch": 75.85835694050992,
+      "loss": 0.035643965005874634,
+      "loss_ce": 1.4690532225358766e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 1146701748,
+      "step": 6684
+    },
+    {
+      "epoch": 75.86968838526913,
+      "grad_norm": 3.469891437466592,
+      "learning_rate": 5e-06,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 1146873696,
+      "step": 6685
+    },
+    {
+      "epoch": 75.86968838526913,
+      "loss": 0.03279561549425125,
+      "loss_ce": 4.4745111154043116e-06,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 1146873696,
+      "step": 6685
+    },
+    {
+      "epoch": 75.88101983002832,
+      "grad_norm": 3.64185882665407,
+      "learning_rate": 5e-06,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 1147044404,
+      "step": 6686
+    },
+    {
+      "epoch": 75.88101983002832,
+      "loss": 0.06192648783326149,
+      "loss_ce": 6.3236630012397654e-06,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 1147044404,
+      "step": 6686
+    },
+    {
+      "epoch": 75.89235127478753,
+      "grad_norm": 3.2899063958324746,
+      "learning_rate": 5e-06,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 1147216440,
+      "step": 6687
+    },
+    {
+      "epoch": 75.89235127478753,
+      "loss": 0.030222224071621895,
+      "loss_ce": 2.508055513317231e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 1147216440,
+      "step": 6687
+    },
+    {
+      "epoch": 75.90368271954674,
+      "grad_norm": 3.514346245156699,
+      "learning_rate": 5e-06,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 1147387340,
+      "step": 6688
+    },
+    {
+      "epoch": 75.90368271954674,
+      "loss": 0.03265424072742462,
+      "loss_ce": 1.5691683074692264e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 1147387340,
+      "step": 6688
+    },
+    {
+      "epoch": 75.91501416430594,
+      "grad_norm": 3.3733432012211915,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 1147558264,
+      "step": 6689
+    },
+    {
+      "epoch": 75.91501416430594,
+      "loss": 0.04083312302827835,
+      "loss_ce": 1.586188045621384e-05,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 1147558264,
+      "step": 6689
+    },
+    {
+      "epoch": 75.92634560906515,
+      "grad_norm": 3.6304490685557522,
+      "learning_rate": 5e-06,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 1147729988,
+      "step": 6690
+    },
+    {
+      "epoch": 75.92634560906515,
+      "loss": 0.09520262479782104,
+      "loss_ce": 1.8297483620699495e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 1147729988,
+      "step": 6690
+    },
+    {
+      "epoch": 75.93767705382436,
+      "grad_norm": 3.504647030603883,
+      "learning_rate": 5e-06,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 1147902024,
+      "step": 6691
+    },
+    {
+      "epoch": 75.93767705382436,
+      "loss": 0.03515082225203514,
+      "loss_ce": 3.271982495789416e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 1147902024,
+      "step": 6691
+    },
+    {
+      "epoch": 75.94900849858357,
+      "grad_norm": 3.6373306013619557,
+      "learning_rate": 5e-06,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 1148073580,
+      "step": 6692
+    },
+    {
+      "epoch": 75.94900849858357,
+      "loss": 0.033994030207395554,
+      "loss_ce": 1.2707678251899779e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1148073580,
+      "step": 6692
+    },
+    {
+      "epoch": 75.96033994334277,
+      "grad_norm": 3.623874562672049,
+      "learning_rate": 5e-06,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 1148245624,
+      "step": 6693
+    },
+    {
+      "epoch": 75.96033994334277,
+      "loss": 0.029669811949133873,
+      "loss_ce": 2.1983621991239488e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 1148245624,
+      "step": 6693
+    },
+    {
+      "epoch": 75.97167138810198,
+      "grad_norm": 3.423271052882423,
+      "learning_rate": 5e-06,
+      "loss": 0.042,
+      "num_input_tokens_seen": 1148416304,
+      "step": 6694
+    },
+    {
+      "epoch": 75.97167138810198,
+      "loss": 0.033641472458839417,
+      "loss_ce": 1.1105205885542091e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1148416304,
+      "step": 6694
+    },
+    {
+      "epoch": 75.98300283286119,
+      "grad_norm": 3.4081389156869983,
+      "learning_rate": 5e-06,
+      "loss": 0.04,
+      "num_input_tokens_seen": 1148588124,
+      "step": 6695
+    },
+    {
+      "epoch": 75.98300283286119,
+      "loss": 0.07447037100791931,
+      "loss_ce": 3.800016929744743e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 1148588124,
+      "step": 6695
+    },
+    {
+      "epoch": 75.9943342776204,
+      "grad_norm": 2.8942014844608597,
+      "learning_rate": 5e-06,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 1148758284,
+      "step": 6696
+    },
+    {
+      "epoch": 75.9943342776204,
+      "loss": 0.03029724769294262,
+      "loss_ce": 8.551374776288867e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 1148758284,
+      "step": 6696
+    },
+    {
+      "epoch": 75.9943342776204,
+      "loss": 0.020227624103426933,
+      "loss_ce": 9.727918040880468e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0040283203125,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 1148844128,
+      "step": 6696
+    },
+    {
+      "epoch": 76.0056657223796,
+      "grad_norm": 2.142813520793836,
+      "learning_rate": 5e-06,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 1148930216,
+      "step": 6697
+    },
+    {
+      "epoch": 76.0056657223796,
+      "loss": 0.023491788655519485,
+      "loss_ce": 8.513385182595812e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 1148930216,
+      "step": 6697
+    },
+    {
+      "epoch": 76.01699716713881,
+      "grad_norm": 2.364388857413835,
+      "learning_rate": 5e-06,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 1149100288,
+      "step": 6698
+    },
+    {
+      "epoch": 76.01699716713881,
+      "loss": 0.04649578034877777,
+      "loss_ce": 1.7512007616460323e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 1149100288,
+      "step": 6698
+    },
+    {
+      "epoch": 76.02832861189802,
+      "grad_norm": 2.1706255465948185,
+      "learning_rate": 5e-06,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 1149272144,
+      "step": 6699
+    },
+    {
+      "epoch": 76.02832861189802,
+      "loss": 0.03590993583202362,
+      "loss_ce": 2.12640061363345e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 1149272144,
+      "step": 6699
+    },
+    {
+      "epoch": 76.03966005665723,
+      "grad_norm": 2.378750537462879,
+      "learning_rate": 5e-06,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 1149443848,
+      "step": 6700
+    },
+    {
+      "epoch": 76.03966005665723,
+      "loss": 0.038892585784196854,
+      "loss_ce": 1.3190378012950532e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 1149443848,
+      "step": 6700
+    },
+    {
+      "epoch": 76.05099150141643,
+      "grad_norm": 3.51565247921851,
+      "learning_rate": 5e-06,
+      "loss": 0.04,
+      "num_input_tokens_seen": 1149615740,
+      "step": 6701
+    },
+    {
+      "epoch": 76.05099150141643,
+      "loss": 0.018701083958148956,
+      "loss_ce": 1.6695030353730544e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0037384033203125,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 1149615740,
+      "step": 6701
+    },
+    {
+      "epoch": 76.06232294617564,
+      "grad_norm": 2.74667367170532,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 1149787284,
+      "step": 6702
+    },
+    {
+      "epoch": 76.06232294617564,
+      "loss": 0.027399137616157532,
+      "loss_ce": 1.7241462046513334e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 1149787284,
+      "step": 6702
+    },
+    {
+      "epoch": 76.07365439093485,
+      "grad_norm": 6.632180810782623,
+      "learning_rate": 5e-06,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 1149959316,
+      "step": 6703
+    },
+    {
+      "epoch": 76.07365439093485,
+      "loss": 0.024576185271143913,
+      "loss_ce": 9.535842764307745e-06,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 1149959316,
+      "step": 6703
+    },
+    {
+      "epoch": 76.08498583569406,
+      "grad_norm": 3.796454105620394,
+      "learning_rate": 5e-06,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 1150130608,
+      "step": 6704
+    },
+    {
+      "epoch": 76.08498583569406,
+      "loss": 0.09224532544612885,
+      "loss_ce": 2.1197325622779317e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 1150130608,
+      "step": 6704
+    },
+    {
+      "epoch": 76.09631728045326,
+      "grad_norm": 3.3551287482854293,
+      "learning_rate": 5e-06,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 1150302596,
+      "step": 6705
+    },
+    {
+      "epoch": 76.09631728045326,
+      "loss": 0.03742213919758797,
+      "loss_ce": 7.587786967633292e-06,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 1150302596,
+      "step": 6705
+    },
+    {
+      "epoch": 76.10764872521247,
+      "grad_norm": 8.126262764623439,
+      "learning_rate": 5e-06,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 1150473404,
+      "step": 6706
+    },
+    {
+      "epoch": 76.10764872521247,
+      "loss": 0.034440137445926666,
+      "loss_ce": 3.156549428240396e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 1150473404,
+      "step": 6706
+    },
+    {
+      "epoch": 76.11898016997168,
+      "grad_norm": 4.338004033444712,
+      "learning_rate": 5e-06,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 1150645052,
+      "step": 6707
+    },
+    {
+      "epoch": 76.11898016997168,
+      "loss": 0.03648258000612259,
+      "loss_ce": 2.933320138254203e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 1150645052,
+      "step": 6707
+    },
+    {
+      "epoch": 76.13031161473087,
+      "grad_norm": 3.158468614834595,
+      "learning_rate": 5e-06,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 1150816712,
+      "step": 6708
+    },
+    {
+      "epoch": 76.13031161473087,
+      "loss": 0.03373480588197708,
+      "loss_ce": 1.2880462236353196e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 1150816712,
+      "step": 6708
+    },
+    {
+      "epoch": 76.14164305949008,
+      "grad_norm": 4.015923163663293,
+      "learning_rate": 5e-06,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 1150987940,
+      "step": 6709
+    },
+    {
+      "epoch": 76.14164305949008,
+      "loss": 0.03205908462405205,
+      "loss_ce": 1.562702891533263e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 1150987940,
+      "step": 6709
+    },
+    {
+      "epoch": 76.15297450424929,
+      "grad_norm": 4.982810922612806,
+      "learning_rate": 5e-06,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 1151156912,
+      "step": 6710
+    },
+    {
+      "epoch": 76.15297450424929,
+      "loss": 0.026900295168161392,
+      "loss_ce": 6.678551471850369e-06,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 1151156912,
+      "step": 6710
+    },
+    {
+      "epoch": 76.1643059490085,
+      "grad_norm": 3.186404904076927,
+      "learning_rate": 5e-06,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 1151328616,
+      "step": 6711
+    },
+    {
+      "epoch": 76.1643059490085,
+      "loss": 0.03951674699783325,
+      "loss_ce": 4.226102464599535e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 1151328616,
+      "step": 6711
+    },
+    {
+      "epoch": 76.1756373937677,
+      "grad_norm": 2.986152407553737,
+      "learning_rate": 5e-06,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 1151500404,
+      "step": 6712
+    },
+    {
+      "epoch": 76.1756373937677,
+      "loss": 0.09450555592775345,
+      "loss_ce": 7.873508366174065e-06,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 1151500404,
+      "step": 6712
+    },
+    {
+      "epoch": 76.18696883852691,
+      "grad_norm": 2.7997377199803486,
+      "learning_rate": 5e-06,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 1151671476,
+      "step": 6713
+    },
+    {
+      "epoch": 76.18696883852691,
+      "loss": 0.024938397109508514,
+      "loss_ce": 1.3164338270144071e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 1151671476,
+      "step": 6713
+    },
+    {
+      "epoch": 76.19830028328612,
+      "grad_norm": 2.7542865452134406,
+      "learning_rate": 5e-06,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 1151841756,
+      "step": 6714
+    },
+    {
+      "epoch": 76.19830028328612,
+      "loss": 0.0285926666110754,
+      "loss_ce": 1.295272886636667e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 1151841756,
+      "step": 6714
+    },
+    {
+      "epoch": 76.20963172804532,
+      "grad_norm": 2.850649834079787,
+      "learning_rate": 5e-06,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 1152012624,
+      "step": 6715
+    },
+    {
+      "epoch": 76.20963172804532,
+      "loss": 0.030105112120509148,
+      "loss_ce": 1.4781064237467945e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1152012624,
+      "step": 6715
+    },
+    {
+      "epoch": 76.22096317280453,
+      "grad_norm": 2.7709607289535336,
+      "learning_rate": 5e-06,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 1152182652,
+      "step": 6716
+    },
+    {
+      "epoch": 76.22096317280453,
+      "loss": 0.03405708819627762,
+      "loss_ce": 1.4731094779563136e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1152182652,
+      "step": 6716
+    },
+    {
+      "epoch": 76.23229461756374,
+      "grad_norm": 2.7755466240660986,
+      "learning_rate": 5e-06,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 1152352324,
+      "step": 6717
+    },
+    {
+      "epoch": 76.23229461756374,
+      "loss": 0.053192902356386185,
+      "loss_ce": 1.602433985681273e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 1152352324,
+      "step": 6717
+    },
+    {
+      "epoch": 76.24362606232295,
+      "grad_norm": 2.6765404393720997,
+      "learning_rate": 5e-06,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 1152524428,
+      "step": 6718
+    },
+    {
+      "epoch": 76.24362606232295,
+      "loss": 0.025339476764202118,
+      "loss_ce": 4.040456406073645e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 1152524428,
+      "step": 6718
+    },
+    {
+      "epoch": 76.25495750708215,
+      "grad_norm": 2.778012442088198,
+      "learning_rate": 5e-06,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 1152695868,
+      "step": 6719
+    },
+    {
+      "epoch": 76.25495750708215,
+      "loss": 0.04451717436313629,
+      "loss_ce": 2.254739956697449e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 1152695868,
+      "step": 6719
+    },
+    {
+      "epoch": 76.26628895184136,
+      "grad_norm": 2.581844634505008,
+      "learning_rate": 5e-06,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 1152867256,
+      "step": 6720
+    },
+    {
+      "epoch": 76.26628895184136,
+      "loss": 0.03184906765818596,
+      "loss_ce": 1.9235512809245847e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 1152867256,
+      "step": 6720
+    },
+    {
+      "epoch": 76.27762039660057,
+      "grad_norm": 2.959147014006936,
+      "learning_rate": 5e-06,
+      "loss": 0.0285,
+      "num_input_tokens_seen": 1153039436,
+      "step": 6721
+    },
+    {
+      "epoch": 76.27762039660057,
+      "loss": 0.023907314985990524,
+      "loss_ce": 1.205211810884066e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 1153039436,
+      "step": 6721
+    },
+    {
+      "epoch": 76.28895184135978,
+      "grad_norm": 3.259455427741046,
+      "learning_rate": 5e-06,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 1153210448,
+      "step": 6722
+    },
+    {
+      "epoch": 76.28895184135978,
+      "loss": 0.02499602735042572,
+      "loss_ce": 1.7390973880537786e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 1153210448,
+      "step": 6722
+    },
+    {
+      "epoch": 76.30028328611898,
+      "grad_norm": 3.7055552445953435,
+      "learning_rate": 5e-06,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 1153381396,
+      "step": 6723
+    },
+    {
+      "epoch": 76.30028328611898,
+      "loss": 0.039533935487270355,
+      "loss_ce": 2.8930055123055354e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 1153381396,
+      "step": 6723
+    },
+    {
+      "epoch": 76.31161473087819,
+      "grad_norm": 2.980235199301115,
+      "learning_rate": 5e-06,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 1153552208,
+      "step": 6724
+    },
+    {
+      "epoch": 76.31161473087819,
+      "loss": 0.03134167566895485,
+      "loss_ce": 1.5381949197035283e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 1153552208,
+      "step": 6724
+    },
+    {
+      "epoch": 76.3229461756374,
+      "grad_norm": 3.451790865187317,
+      "learning_rate": 5e-06,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 1153723456,
+      "step": 6725
+    },
+    {
+      "epoch": 76.3229461756374,
+      "loss": 0.03680173307657242,
+      "loss_ce": 2.8052258130628616e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1153723456,
+      "step": 6725
+    },
+    {
+      "epoch": 76.3342776203966,
+      "grad_norm": 3.882913298644648,
+      "learning_rate": 5e-06,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 1153895360,
+      "step": 6726
+    },
+    {
+      "epoch": 76.3342776203966,
+      "loss": 0.06896818429231644,
+      "loss_ce": 1.3716375178773887e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 1153895360,
+      "step": 6726
+    },
+    {
+      "epoch": 76.34560906515581,
+      "grad_norm": 3.5824310438613276,
+      "learning_rate": 5e-06,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 1154065480,
+      "step": 6727
+    },
+    {
+      "epoch": 76.34560906515581,
+      "loss": 0.031630292534828186,
+      "loss_ce": 1.4083432688494213e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 1154065480,
+      "step": 6727
+    },
+    {
+      "epoch": 76.35694050991502,
+      "grad_norm": 3.5855519577328705,
+      "learning_rate": 5e-06,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 1154237336,
+      "step": 6728
+    },
+    {
+      "epoch": 76.35694050991502,
+      "loss": 0.03963451087474823,
+      "loss_ce": 2.2692787752021104e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 1154237336,
+      "step": 6728
+    },
+    {
+      "epoch": 76.36827195467423,
+      "grad_norm": 3.107590296644265,
+      "learning_rate": 5e-06,
+      "loss": 0.036,
+      "num_input_tokens_seen": 1154408496,
+      "step": 6729
+    },
+    {
+      "epoch": 76.36827195467423,
+      "loss": 0.037957750260829926,
+      "loss_ce": 2.4401826522080228e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 1154408496,
+      "step": 6729
+    },
+    {
+      "epoch": 76.37960339943342,
+      "grad_norm": 4.057694296249576,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 1154580376,
+      "step": 6730
+    },
+    {
+      "epoch": 76.37960339943342,
+      "loss": 0.050042614340782166,
+      "loss_ce": 2.4300499717355706e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 1154580376,
+      "step": 6730
+    },
+    {
+      "epoch": 76.39093484419263,
+      "grad_norm": 3.8431018046852685,
+      "learning_rate": 5e-06,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 1154749188,
+      "step": 6731
+    },
+    {
+      "epoch": 76.39093484419263,
+      "loss": 0.04194425791501999,
+      "loss_ce": 0.005170574877411127,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1154749188,
+      "step": 6731
+    },
+    {
+      "epoch": 76.40226628895184,
+      "grad_norm": 3.222343304965176,
+      "learning_rate": 5e-06,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 1154920744,
+      "step": 6732
+    },
+    {
+      "epoch": 76.40226628895184,
+      "loss": 0.0338994562625885,
+      "loss_ce": 2.494702857802622e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 1154920744,
+      "step": 6732
+    },
+    {
+      "epoch": 76.41359773371104,
+      "grad_norm": 3.254521229914388,
+      "learning_rate": 5e-06,
+      "loss": 0.036,
+      "num_input_tokens_seen": 1155092700,
+      "step": 6733
+    },
+    {
+      "epoch": 76.41359773371104,
+      "loss": 0.04597647860646248,
+      "loss_ce": 4.7525096306344494e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 1155092700,
+      "step": 6733
+    },
+    {
+      "epoch": 76.42492917847025,
+      "grad_norm": 3.5812796985408597,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 1155264796,
+      "step": 6734
+    },
+    {
+      "epoch": 76.42492917847025,
+      "loss": 0.04866945743560791,
+      "loss_ce": 2.444002529955469e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 1155264796,
+      "step": 6734
+    },
+    {
+      "epoch": 76.43626062322946,
+      "grad_norm": 3.1875034200454926,
+      "learning_rate": 5e-06,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 1155436384,
+      "step": 6735
+    },
+    {
+      "epoch": 76.43626062322946,
+      "loss": 0.06429353356361389,
+      "loss_ce": 2.3516251530963928e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 1155436384,
+      "step": 6735
+    },
+    {
+      "epoch": 76.44759206798867,
+      "grad_norm": 3.1861560871577215,
+      "learning_rate": 5e-06,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 1155607288,
+      "step": 6736
+    },
+    {
+      "epoch": 76.44759206798867,
+      "loss": 0.030070578679442406,
+      "loss_ce": 4.1281000449089333e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1155607288,
+      "step": 6736
+    },
+    {
+      "epoch": 76.45892351274787,
+      "grad_norm": 3.3144382378426442,
+      "learning_rate": 5e-06,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 1155779228,
+      "step": 6737
+    },
+    {
+      "epoch": 76.45892351274787,
+      "loss": 0.08017963916063309,
+      "loss_ce": 9.964962373487651e-06,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 1155779228,
+      "step": 6737
+    },
+    {
+      "epoch": 76.47025495750708,
+      "grad_norm": 3.383780044199627,
+      "learning_rate": 5e-06,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 1155950196,
+      "step": 6738
+    },
+    {
+      "epoch": 76.47025495750708,
+      "loss": 0.04567316919565201,
+      "loss_ce": 0.00014094336074776947,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 1155950196,
+      "step": 6738
+    },
+    {
+      "epoch": 76.48158640226629,
+      "grad_norm": 3.622071716453547,
+      "learning_rate": 5e-06,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 1156121144,
+      "step": 6739
+    },
+    {
+      "epoch": 76.48158640226629,
+      "loss": 0.05220004916191101,
+      "loss_ce": 0.0009152608108706772,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 1156121144,
+      "step": 6739
+    },
+    {
+      "epoch": 76.4929178470255,
+      "grad_norm": 3.4324059357214742,
+      "learning_rate": 5e-06,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 1156292092,
+      "step": 6740
+    },
+    {
+      "epoch": 76.4929178470255,
+      "loss": 0.05812738835811615,
+      "loss_ce": 8.295338193420321e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 1156292092,
+      "step": 6740
+    },
+    {
+      "epoch": 76.5042492917847,
+      "grad_norm": 3.112818253550665,
+      "learning_rate": 5e-06,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 1156463808,
+      "step": 6741
+    },
+    {
+      "epoch": 76.5042492917847,
+      "loss": 0.03158782422542572,
+      "loss_ce": 1.7389553249813616e-05,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 1156463808,
+      "step": 6741
+    },
+    {
+      "epoch": 76.51558073654391,
+      "grad_norm": 2.4148675084999134,
+      "learning_rate": 5e-06,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 1156635544,
+      "step": 6742
+    },
+    {
+      "epoch": 76.51558073654391,
+      "loss": 0.036333635449409485,
+      "loss_ce": 5.586558472714387e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 1156635544,
+      "step": 6742
+    },
+    {
+      "epoch": 76.52691218130312,
+      "grad_norm": 2.1242609993958737,
+      "learning_rate": 5e-06,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 1156807580,
+      "step": 6743
+    },
+    {
+      "epoch": 76.52691218130312,
+      "loss": 0.03686174005270004,
+      "loss_ce": 7.279974670382217e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 1156807580,
+      "step": 6743
+    },
+    {
+      "epoch": 76.53824362606233,
+      "grad_norm": 3.1722154420360895,
+      "learning_rate": 5e-06,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 1156978424,
+      "step": 6744
+    },
+    {
+      "epoch": 76.53824362606233,
+      "loss": 0.040512923151254654,
+      "loss_ce": 7.713256491115317e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 1156978424,
+      "step": 6744
+    },
+    {
+      "epoch": 76.54957507082153,
+      "grad_norm": 2.7252823904909125,
+      "learning_rate": 5e-06,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 1157150032,
+      "step": 6745
+    },
+    {
+      "epoch": 76.54957507082153,
+      "loss": 0.047656770795583725,
+      "loss_ce": 1.1202017958567012e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 1157150032,
+      "step": 6745
+    },
+    {
+      "epoch": 76.56090651558074,
+      "grad_norm": 3.6473200451005217,
+      "learning_rate": 5e-06,
+      "loss": 0.032,
+      "num_input_tokens_seen": 1157322088,
+      "step": 6746
+    },
+    {
+      "epoch": 76.56090651558074,
+      "loss": 0.024936577305197716,
+      "loss_ce": 3.423343878239393e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 1157322088,
+      "step": 6746
+    },
+    {
+      "epoch": 76.57223796033995,
+      "grad_norm": 3.2244182499202094,
+      "learning_rate": 5e-06,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 1157493908,
+      "step": 6747
+    },
+    {
+      "epoch": 76.57223796033995,
+      "loss": 0.04200422018766403,
+      "loss_ce": 1.2032525773975067e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 1157493908,
+      "step": 6747
+    },
+    {
+      "epoch": 76.58356940509915,
+      "grad_norm": 2.812770553290911,
+      "learning_rate": 5e-06,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 1157664756,
+      "step": 6748
+    },
+    {
+      "epoch": 76.58356940509915,
+      "loss": 0.026366740465164185,
+      "loss_ce": 1.4810613720328547e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 1157664756,
+      "step": 6748
+    },
+    {
+      "epoch": 76.59490084985836,
+      "grad_norm": 4.0153110644259264,
+      "learning_rate": 5e-06,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 1157836356,
+      "step": 6749
+    },
+    {
+      "epoch": 76.59490084985836,
+      "loss": 0.03164807707071304,
+      "loss_ce": 0.001649297191761434,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 1157836356,
+      "step": 6749
+    },
+    {
+      "epoch": 76.60623229461757,
+      "grad_norm": 3.387376251531648,
+      "learning_rate": 5e-06,
+      "loss": 0.0286,
+      "num_input_tokens_seen": 1158008468,
+      "step": 6750
+    },
+    {
+      "epoch": 76.60623229461757,
+      "eval_seeclick_CIoU": 0.5175180286169052,
+      "eval_seeclick_GIoU": 0.5201136469841003,
+      "eval_seeclick_IoU": 0.5568788349628448,
+      "eval_seeclick_MAE_all": 0.06802032887935638,
+      "eval_seeclick_MAE_h": 0.03438388183712959,
+      "eval_seeclick_MAE_w": 0.10334005951881409,
+      "eval_seeclick_MAE_x": 0.10081024095416069,
+      "eval_seeclick_MAE_y": 0.033547136932611465,
+      "eval_seeclick_NUM_probability": 0.9999966025352478,
+      "eval_seeclick_inside_bbox": 0.9076704680919647,
+      "eval_seeclick_loss": 1.0442501306533813,
+      "eval_seeclick_loss_ce": 0.7959894239902496,
+      "eval_seeclick_loss_iou": 0.5408935546875,
+      "eval_seeclick_loss_num": 0.050487518310546875,
+      "eval_seeclick_loss_xval": 0.252197265625,
+      "eval_seeclick_runtime": 73.7402,
+      "eval_seeclick_samples_per_second": 0.583,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 1158008468,
+      "step": 6750
+    },
+    {
+      "epoch": 76.60623229461757,
+      "eval_icons_CIoU": 0.7266184091567993,
+      "eval_icons_GIoU": 0.7282552719116211,
+      "eval_icons_IoU": 0.7413917481899261,
+      "eval_icons_MAE_all": 0.035040028393268585,
+      "eval_icons_MAE_h": 0.02840566076338291,
+      "eval_icons_MAE_w": 0.04360402841120958,
+      "eval_icons_MAE_x": 0.03670475818216801,
+      "eval_icons_MAE_y": 0.03144566807895899,
+      "eval_icons_NUM_probability": 0.9998024106025696,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.13592630624771118,
+      "eval_icons_loss_ce": 0.005310217617079616,
+      "eval_icons_loss_iou": 0.569580078125,
+      "eval_icons_loss_num": 0.02378082275390625,
+      "eval_icons_loss_xval": 0.1189117431640625,
+      "eval_icons_runtime": 84.2962,
+      "eval_icons_samples_per_second": 0.593,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 1158008468,
+      "step": 6750
+    },
+    {
+      "epoch": 76.60623229461757,
+      "eval_screenspot_CIoU": 0.6396422783533732,
+      "eval_screenspot_GIoU": 0.6394848624865214,
+      "eval_screenspot_IoU": 0.6678770184516907,
+      "eval_screenspot_MAE_all": 0.06549978007872899,
+      "eval_screenspot_MAE_h": 0.0410217580695947,
+      "eval_screenspot_MAE_w": 0.11402194077769916,
+      "eval_screenspot_MAE_x": 0.073229289924105,
+      "eval_screenspot_MAE_y": 0.03372615793099006,
+      "eval_screenspot_NUM_probability": 0.9999787012736002,
+      "eval_screenspot_inside_bbox": 0.9287499984105428,
+      "eval_screenspot_loss": 0.277182012796402,
+      "eval_screenspot_loss_ce": 0.015680634416639805,
+      "eval_screenspot_loss_iou": 0.49853515625,
+      "eval_screenspot_loss_num": 0.051724751790364586,
+      "eval_screenspot_loss_xval": 0.258544921875,
+      "eval_screenspot_runtime": 148.4232,
+      "eval_screenspot_samples_per_second": 0.6,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 1158008468,
+      "step": 6750
+    },
+    {
+      "epoch": 76.60623229461757,
+      "eval_compot_CIoU": 0.8883132338523865,
+      "eval_compot_GIoU": 0.8871381878852844,
+      "eval_compot_IoU": 0.8967113792896271,
+      "eval_compot_MAE_all": 0.017100587021559477,
+      "eval_compot_MAE_h": 0.00995214655995369,
+      "eval_compot_MAE_w": 0.02563900128006935,
+      "eval_compot_MAE_x": 0.023243038915097713,
+      "eval_compot_MAE_y": 0.009568159701302648,
+      "eval_compot_NUM_probability": 0.9999822676181793,
+      "eval_compot_inside_bbox": 0.96875,
+      "eval_compot_loss": 0.06645932793617249,
+      "eval_compot_loss_ce": 1.2750736914313165e-05,
+      "eval_compot_loss_iou": 0.4945068359375,
+      "eval_compot_loss_num": 0.011624336242675781,
+      "eval_compot_loss_xval": 0.05809783935546875,
+      "eval_compot_runtime": 84.7839,
+      "eval_compot_samples_per_second": 0.59,
+      "eval_compot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 1158008468,
+      "step": 6750
+    },
+    {
+      "epoch": 76.60623229461757,
+      "eval_custom_ui_MAE_all": 0.019065924920141697,
+      "eval_custom_ui_MAE_x": 0.03074595332145691,
+      "eval_custom_ui_MAE_y": 0.007385896518826485,
+      "eval_custom_ui_NUM_probability": 0.9998641610145569,
+      "eval_custom_ui_loss": 0.21941247582435608,
+      "eval_custom_ui_loss_ce": 0.12301512807607651,
+      "eval_custom_ui_loss_num": 0.0182952880859375,
+      "eval_custom_ui_loss_xval": 0.0915069580078125,
+      "eval_custom_ui_runtime": 65.79,
+      "eval_custom_ui_samples_per_second": 0.76,
+      "eval_custom_ui_steps_per_second": 0.03,
+      "num_input_tokens_seen": 1158008468,
+      "step": 6750
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8096,
+  "num_input_tokens_seen": 1158008468,
+  "num_train_epochs": 92,
+  "save_steps": 250,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8354828616859648.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}