diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,27605 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.896551724137931,
+  "eval_steps": 250,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004597701149425287,
+      "grad_norm": 37.01336935001283,
+      "learning_rate": 5e-06,
+      "loss": 0.8637,
+      "num_input_tokens_seen": 90336,
+      "step": 1
+    },
+    {
+      "epoch": 0.004597701149425287,
+      "loss": 1.0519328117370605,
+      "loss_ce": 0.5021281838417053,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.10986328125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 90336,
+      "step": 1
+    },
+    {
+      "epoch": 0.009195402298850575,
+      "grad_norm": 40.354572183137385,
+      "learning_rate": 5e-06,
+      "loss": 0.8717,
+      "num_input_tokens_seen": 180800,
+      "step": 2
+    },
+    {
+      "epoch": 0.009195402298850575,
+      "loss": 0.7949165105819702,
+      "loss_ce": 0.3349556028842926,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.091796875,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 180800,
+      "step": 2
+    },
+    {
+      "epoch": 0.013793103448275862,
+      "grad_norm": 39.06584845163415,
+      "learning_rate": 5e-06,
+      "loss": 0.606,
+      "num_input_tokens_seen": 271252,
+      "step": 3
+    },
+    {
+      "epoch": 0.013793103448275862,
+      "loss": 0.5930109620094299,
+      "loss_ce": 0.20678049325942993,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 271252,
+      "step": 3
+    },
+    {
+      "epoch": 0.01839080459770115,
+      "grad_norm": 57.83141578006947,
+      "learning_rate": 5e-06,
+      "loss": 0.6455,
+      "num_input_tokens_seen": 361480,
+      "step": 4
+    },
+    {
+      "epoch": 0.01839080459770115,
+      "loss": 0.7589341402053833,
+      "loss_ce": 0.3553696274757385,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.08056640625,
+      "loss_xval": 0.404296875,
+      "num_input_tokens_seen": 361480,
+      "step": 4
+    },
+    {
+      "epoch": 0.022988505747126436,
+      "grad_norm": 42.90588851722674,
+      "learning_rate": 5e-06,
+      "loss": 1.0614,
+      "num_input_tokens_seen": 451876,
+      "step": 5
+    },
+    {
+      "epoch": 0.022988505747126436,
+      "loss": 1.2513668537139893,
+      "loss_ce": 0.703759491443634,
+      "loss_iou": 0.421875,
+      "loss_num": 0.109375,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 451876,
+      "step": 5
+    },
+    {
+      "epoch": 0.027586206896551724,
+      "grad_norm": 22.460688537328316,
+      "learning_rate": 5e-06,
+      "loss": 0.6396,
+      "num_input_tokens_seen": 542192,
+      "step": 6
+    },
+    {
+      "epoch": 0.027586206896551724,
+      "loss": 0.504480242729187,
+      "loss_ce": 0.05452907085418701,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.08984375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 542192,
+      "step": 6
+    },
+    {
+      "epoch": 0.03218390804597701,
+      "grad_norm": 23.415625594005842,
+      "learning_rate": 5e-06,
+      "loss": 0.874,
+      "num_input_tokens_seen": 632568,
+      "step": 7
+    },
+    {
+      "epoch": 0.03218390804597701,
+      "loss": 1.0449306964874268,
+      "loss_ce": 0.5966885685920715,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.08984375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 632568,
+      "step": 7
+    },
+    {
+      "epoch": 0.0367816091954023,
+      "grad_norm": 31.16740618162066,
+      "learning_rate": 5e-06,
+      "loss": 0.7589,
+      "num_input_tokens_seen": 723052,
+      "step": 8
+    },
+    {
+      "epoch": 0.0367816091954023,
+      "loss": 0.6076769828796387,
+      "loss_ce": 0.25794553756713867,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 723052,
+      "step": 8
+    },
+    {
+      "epoch": 0.041379310344827586,
+      "grad_norm": 79.07479774439705,
+      "learning_rate": 5e-06,
+      "loss": 0.687,
+      "num_input_tokens_seen": 813440,
+      "step": 9
+    },
+    {
+      "epoch": 0.041379310344827586,
+      "loss": 0.6652740240097046,
+      "loss_ce": 0.1359771490097046,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.10595703125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 813440,
+      "step": 9
+    },
+    {
+      "epoch": 0.04597701149425287,
+      "grad_norm": 44.77421252011234,
+      "learning_rate": 5e-06,
+      "loss": 0.6675,
+      "num_input_tokens_seen": 903664,
+      "step": 10
+    },
+    {
+      "epoch": 0.04597701149425287,
+      "loss": 0.6931918859481812,
+      "loss_ce": 0.2824862599372864,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 903664,
+      "step": 10
+    },
+    {
+      "epoch": 0.05057471264367816,
+      "grad_norm": 80.61281743018304,
+      "learning_rate": 5e-06,
+      "loss": 0.8424,
+      "num_input_tokens_seen": 994196,
+      "step": 11
+    },
+    {
+      "epoch": 0.05057471264367816,
+      "loss": 0.9065788984298706,
+      "loss_ce": 0.4577263593673706,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.08984375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 994196,
+      "step": 11
+    },
+    {
+      "epoch": 0.05517241379310345,
+      "grad_norm": 21.38207584365465,
+      "learning_rate": 5e-06,
+      "loss": 0.6596,
+      "num_input_tokens_seen": 1084712,
+      "step": 12
+    },
+    {
+      "epoch": 0.05517241379310345,
+      "loss": 0.5851885080337524,
+      "loss_ce": 0.21836715936660767,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 1084712,
+      "step": 12
+    },
+    {
+      "epoch": 0.059770114942528735,
+      "grad_norm": 27.964832638511165,
+      "learning_rate": 5e-06,
+      "loss": 0.5867,
+      "num_input_tokens_seen": 1175040,
+      "step": 13
+    },
+    {
+      "epoch": 0.059770114942528735,
+      "loss": 0.5128318071365356,
+      "loss_ce": 0.13587866723537445,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.376953125,
+      "num_input_tokens_seen": 1175040,
+      "step": 13
+    },
+    {
+      "epoch": 0.06436781609195402,
+      "grad_norm": 15.47526861039646,
+      "learning_rate": 5e-06,
+      "loss": 0.6604,
+      "num_input_tokens_seen": 1265560,
+      "step": 14
+    },
+    {
+      "epoch": 0.06436781609195402,
+      "loss": 0.6039849519729614,
+      "loss_ce": 0.20707331597805023,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.07958984375,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 1265560,
+      "step": 14
+    },
+    {
+      "epoch": 0.06896551724137931,
+      "grad_norm": 8.807051044677067,
+      "learning_rate": 5e-06,
+      "loss": 0.6509,
+      "num_input_tokens_seen": 1355984,
+      "step": 15
+    },
+    {
+      "epoch": 0.06896551724137931,
+      "loss": 0.6505950689315796,
+      "loss_ce": 0.1728118658065796,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.095703125,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 1355984,
+      "step": 15
+    },
+    {
+      "epoch": 0.0735632183908046,
+      "grad_norm": 16.39684002573081,
+      "learning_rate": 5e-06,
+      "loss": 0.5639,
+      "num_input_tokens_seen": 1446400,
+      "step": 16
+    },
+    {
+      "epoch": 0.0735632183908046,
+      "loss": 0.6162420511245728,
+      "loss_ce": 0.24746768176555634,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 1446400,
+      "step": 16
+    },
+    {
+      "epoch": 0.07816091954022988,
+      "grad_norm": 19.846643198159168,
+      "learning_rate": 5e-06,
+      "loss": 0.6321,
+      "num_input_tokens_seen": 1536776,
+      "step": 17
+    },
+    {
+      "epoch": 0.07816091954022988,
+      "loss": 0.6417566537857056,
+      "loss_ce": 0.18911993503570557,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 1536776,
+      "step": 17
+    },
+    {
+      "epoch": 0.08275862068965517,
+      "grad_norm": 28.522543896242876,
+      "learning_rate": 5e-06,
+      "loss": 0.6082,
+      "num_input_tokens_seen": 1627260,
+      "step": 18
+    },
+    {
+      "epoch": 0.08275862068965517,
+      "loss": 0.5867501497268677,
+      "loss_ce": 0.10921109467744827,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.095703125,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 1627260,
+      "step": 18
+    },
+    {
+      "epoch": 0.08735632183908046,
+      "grad_norm": 37.06692415264521,
+      "learning_rate": 5e-06,
+      "loss": 0.6748,
+      "num_input_tokens_seen": 1717724,
+      "step": 19
+    },
+    {
+      "epoch": 0.08735632183908046,
+      "loss": 0.691688060760498,
+      "loss_ce": 0.22208356857299805,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.09375,
+      "loss_xval": 0.46875,
+      "num_input_tokens_seen": 1717724,
+      "step": 19
+    },
+    {
+      "epoch": 0.09195402298850575,
+      "grad_norm": 113.58143371326585,
+      "learning_rate": 5e-06,
+      "loss": 0.5325,
+      "num_input_tokens_seen": 1808100,
+      "step": 20
+    },
+    {
+      "epoch": 0.09195402298850575,
+      "loss": 0.5246105194091797,
+      "loss_ce": 0.1267833560705185,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.07958984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 1808100,
+      "step": 20
+    },
+    {
+      "epoch": 0.09655172413793103,
+      "grad_norm": 45.45186371478247,
+      "learning_rate": 5e-06,
+      "loss": 0.672,
+      "num_input_tokens_seen": 1898532,
+      "step": 21
+    },
+    {
+      "epoch": 0.09655172413793103,
+      "loss": 0.6351579427719116,
+      "loss_ce": 0.06411299854516983,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.1142578125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 1898532,
+      "step": 21
+    },
+    {
+      "epoch": 0.10114942528735632,
+      "grad_norm": 23.89941149750651,
+      "learning_rate": 5e-06,
+      "loss": 0.5832,
+      "num_input_tokens_seen": 1988852,
+      "step": 22
+    },
+    {
+      "epoch": 0.10114942528735632,
+      "loss": 0.5167936682701111,
+      "loss_ce": 0.1544889509677887,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 1988852,
+      "step": 22
+    },
+    {
+      "epoch": 0.10574712643678161,
+      "grad_norm": 10.318774820543899,
+      "learning_rate": 5e-06,
+      "loss": 0.6361,
+      "num_input_tokens_seen": 2079304,
+      "step": 23
+    },
+    {
+      "epoch": 0.10574712643678161,
+      "loss": 0.5845872759819031,
+      "loss_ce": 0.07311265170574188,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 2079304,
+      "step": 23
+    },
+    {
+      "epoch": 0.1103448275862069,
+      "grad_norm": 25.73676580542981,
+      "learning_rate": 5e-06,
+      "loss": 0.568,
+      "num_input_tokens_seen": 2169628,
+      "step": 24
+    },
+    {
+      "epoch": 0.1103448275862069,
+      "loss": 0.5678774118423462,
+      "loss_ce": 0.06470361351966858,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.1005859375,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 2169628,
+      "step": 24
+    },
+    {
+      "epoch": 0.11494252873563218,
+      "grad_norm": 81.51890029791515,
+      "learning_rate": 5e-06,
+      "loss": 0.5453,
+      "num_input_tokens_seen": 2259984,
+      "step": 25
+    },
+    {
+      "epoch": 0.11494252873563218,
+      "loss": 0.5036299824714661,
+      "loss_ce": 0.05953817814588547,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0888671875,
+      "loss_xval": 0.443359375,
+      "num_input_tokens_seen": 2259984,
+      "step": 25
+    },
+    {
+      "epoch": 0.11954022988505747,
+      "grad_norm": 27.04495599462074,
+      "learning_rate": 5e-06,
+      "loss": 0.5524,
+      "num_input_tokens_seen": 2350292,
+      "step": 26
+    },
+    {
+      "epoch": 0.11954022988505747,
+      "loss": 0.6328690052032471,
+      "loss_ce": 0.08086705207824707,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.1103515625,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 2350292,
+      "step": 26
+    },
+    {
+      "epoch": 0.12413793103448276,
+      "grad_norm": 10.193343220359463,
+      "learning_rate": 5e-06,
+      "loss": 0.5916,
+      "num_input_tokens_seen": 2440640,
+      "step": 27
+    },
+    {
+      "epoch": 0.12413793103448276,
+      "loss": 0.5718714594841003,
+      "loss_ce": 0.14609020948410034,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 2440640,
+      "step": 27
+    },
+    {
+      "epoch": 0.12873563218390804,
+      "grad_norm": 19.9427592340637,
+      "learning_rate": 5e-06,
+      "loss": 0.581,
+      "num_input_tokens_seen": 2530228,
+      "step": 28
+    },
+    {
+      "epoch": 0.12873563218390804,
+      "loss": 0.6976370811462402,
+      "loss_ce": 0.15210485458374023,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.109375,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 2530228,
+      "step": 28
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 22.33962406561534,
+      "learning_rate": 5e-06,
+      "loss": 0.5247,
+      "num_input_tokens_seen": 2619848,
+      "step": 29
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "loss": 0.570274829864502,
+      "loss_ce": 0.09835098683834076,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.09423828125,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 2619848,
+      "step": 29
+    },
+    {
+      "epoch": 0.13793103448275862,
+      "grad_norm": 14.599136978515288,
+      "learning_rate": 5e-06,
+      "loss": 0.5767,
+      "num_input_tokens_seen": 2710140,
+      "step": 30
+    },
+    {
+      "epoch": 0.13793103448275862,
+      "loss": 0.603084146976471,
+      "loss_ce": 0.09063299000263214,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 2710140,
+      "step": 30
+    },
+    {
+      "epoch": 0.1425287356321839,
+      "grad_norm": 18.74413636198144,
+      "learning_rate": 5e-06,
+      "loss": 0.5167,
+      "num_input_tokens_seen": 2800536,
+      "step": 31
+    },
+    {
+      "epoch": 0.1425287356321839,
+      "loss": 0.547565758228302,
+      "loss_ce": 0.1607249677181244,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 2800536,
+      "step": 31
+    },
+    {
+      "epoch": 0.1471264367816092,
+      "grad_norm": 30.495448668876183,
+      "learning_rate": 5e-06,
+      "loss": 0.4526,
+      "num_input_tokens_seen": 2890960,
+      "step": 32
+    },
+    {
+      "epoch": 0.1471264367816092,
+      "loss": 0.42909184098243713,
+      "loss_ce": 0.10731448233127594,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 2890960,
+      "step": 32
+    },
+    {
+      "epoch": 0.15172413793103448,
+      "grad_norm": 82.31686845472512,
+      "learning_rate": 5e-06,
+      "loss": 0.5327,
+      "num_input_tokens_seen": 2979808,
+      "step": 33
+    },
+    {
+      "epoch": 0.15172413793103448,
+      "loss": 0.4499969482421875,
+      "loss_ce": 0.008224454708397388,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.08837890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 2979808,
+      "step": 33
+    },
+    {
+      "epoch": 0.15632183908045977,
+      "grad_norm": 10.971908242941907,
+      "learning_rate": 5e-06,
+      "loss": 0.4713,
+      "num_input_tokens_seen": 3070312,
+      "step": 34
+    },
+    {
+      "epoch": 0.15632183908045977,
+      "loss": 0.3486025929450989,
+      "loss_ce": 0.06405669450759888,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 3070312,
+      "step": 34
+    },
+    {
+      "epoch": 0.16091954022988506,
+      "grad_norm": 59.48550166875302,
+      "learning_rate": 5e-06,
+      "loss": 0.4692,
+      "num_input_tokens_seen": 3159280,
+      "step": 35
+    },
+    {
+      "epoch": 0.16091954022988506,
+      "loss": 0.5126940011978149,
+      "loss_ce": 0.08544795215129852,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 3159280,
+      "step": 35
+    },
+    {
+      "epoch": 0.16551724137931034,
+      "grad_norm": 60.760015794484936,
+      "learning_rate": 5e-06,
+      "loss": 0.4513,
+      "num_input_tokens_seen": 3249672,
+      "step": 36
+    },
+    {
+      "epoch": 0.16551724137931034,
+      "loss": 0.4749954044818878,
+      "loss_ce": 0.11220243573188782,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 3249672,
+      "step": 36
+    },
+    {
+      "epoch": 0.17011494252873563,
+      "grad_norm": 24.0106917483809,
+      "learning_rate": 5e-06,
+      "loss": 0.481,
+      "num_input_tokens_seen": 3339988,
+      "step": 37
+    },
+    {
+      "epoch": 0.17011494252873563,
+      "loss": 0.5111528635025024,
+      "loss_ce": 0.13212455809116364,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 3339988,
+      "step": 37
+    },
+    {
+      "epoch": 0.17471264367816092,
+      "grad_norm": 30.5271253034538,
+      "learning_rate": 5e-06,
+      "loss": 0.5096,
+      "num_input_tokens_seen": 3430444,
+      "step": 38
+    },
+    {
+      "epoch": 0.17471264367816092,
+      "loss": 0.4569355845451355,
+      "loss_ce": 0.03676954656839371,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 3430444,
+      "step": 38
+    },
+    {
+      "epoch": 0.1793103448275862,
+      "grad_norm": 8.31708311536423,
+      "learning_rate": 5e-06,
+      "loss": 0.417,
+      "num_input_tokens_seen": 3519396,
+      "step": 39
+    },
+    {
+      "epoch": 0.1793103448275862,
+      "loss": 0.4267890453338623,
+      "loss_ce": 0.0754707008600235,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 3519396,
+      "step": 39
+    },
+    {
+      "epoch": 0.1839080459770115,
+      "grad_norm": 20.02750098475025,
+      "learning_rate": 5e-06,
+      "loss": 0.3488,
+      "num_input_tokens_seen": 3609736,
+      "step": 40
+    },
+    {
+      "epoch": 0.1839080459770115,
+      "loss": 0.3892139196395874,
+      "loss_ce": 0.047050852328538895,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 3609736,
+      "step": 40
+    },
+    {
+      "epoch": 0.18850574712643678,
+      "grad_norm": 32.78061699301825,
+      "learning_rate": 5e-06,
+      "loss": 0.4256,
+      "num_input_tokens_seen": 3700148,
+      "step": 41
+    },
+    {
+      "epoch": 0.18850574712643678,
+      "loss": 0.3672151267528534,
+      "loss_ce": 0.013943652622401714,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 3700148,
+      "step": 41
+    },
+    {
+      "epoch": 0.19310344827586207,
+      "grad_norm": 9.890488159206065,
+      "learning_rate": 5e-06,
+      "loss": 0.4023,
+      "num_input_tokens_seen": 3790612,
+      "step": 42
+    },
+    {
+      "epoch": 0.19310344827586207,
+      "loss": 0.42205893993377686,
+      "loss_ce": 0.034973956644535065,
+      "loss_iou": 0.5625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 3790612,
+      "step": 42
+    },
+    {
+      "epoch": 0.19770114942528735,
+      "grad_norm": 24.69833771237315,
+      "learning_rate": 5e-06,
+      "loss": 0.5508,
+      "num_input_tokens_seen": 3880836,
+      "step": 43
+    },
+    {
+      "epoch": 0.19770114942528735,
+      "loss": 0.5896004438400269,
+      "loss_ce": 0.20044030249118805,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.078125,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 3880836,
+      "step": 43
+    },
+    {
+      "epoch": 0.20229885057471264,
+      "grad_norm": 16.669851454492647,
+      "learning_rate": 5e-06,
+      "loss": 0.4632,
+      "num_input_tokens_seen": 3971176,
+      "step": 44
+    },
+    {
+      "epoch": 0.20229885057471264,
+      "loss": 0.4887722134590149,
+      "loss_ce": 0.0513942651450634,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.08740234375,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 3971176,
+      "step": 44
+    },
+    {
+      "epoch": 0.20689655172413793,
+      "grad_norm": 18.6471842377231,
+      "learning_rate": 5e-06,
+      "loss": 0.4398,
+      "num_input_tokens_seen": 4061524,
+      "step": 45
+    },
+    {
+      "epoch": 0.20689655172413793,
+      "loss": 0.3761554956436157,
+      "loss_ce": 0.028621304780244827,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 4061524,
+      "step": 45
+    },
+    {
+      "epoch": 0.21149425287356322,
+      "grad_norm": 28.00486282610174,
+      "learning_rate": 5e-06,
+      "loss": 0.3932,
+      "num_input_tokens_seen": 4151828,
+      "step": 46
+    },
+    {
+      "epoch": 0.21149425287356322,
+      "loss": 0.3925134539604187,
+      "loss_ce": 0.04986211284995079,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 4151828,
+      "step": 46
+    },
+    {
+      "epoch": 0.2160919540229885,
+      "grad_norm": 24.306371497861868,
+      "learning_rate": 5e-06,
+      "loss": 0.4599,
+      "num_input_tokens_seen": 4241372,
+      "step": 47
+    },
+    {
+      "epoch": 0.2160919540229885,
+      "loss": 0.49177491664886475,
+      "loss_ce": 0.07002198696136475,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.08447265625,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 4241372,
+      "step": 47
+    },
+    {
+      "epoch": 0.2206896551724138,
+      "grad_norm": 6.774260279559146,
+      "learning_rate": 5e-06,
+      "loss": 0.414,
+      "num_input_tokens_seen": 4331848,
+      "step": 48
+    },
+    {
+      "epoch": 0.2206896551724138,
+      "loss": 0.3134571313858032,
+      "loss_ce": 0.009929286316037178,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 4331848,
+      "step": 48
+    },
+    {
+      "epoch": 0.22528735632183908,
+      "grad_norm": 7.185186363794483,
+      "learning_rate": 5e-06,
+      "loss": 0.3401,
+      "num_input_tokens_seen": 4422192,
+      "step": 49
+    },
+    {
+      "epoch": 0.22528735632183908,
+      "loss": 0.34919464588165283,
+      "loss_ce": 0.04316438362002373,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 4422192,
+      "step": 49
+    },
+    {
+      "epoch": 0.22988505747126436,
+      "grad_norm": 13.427495770074591,
+      "learning_rate": 5e-06,
+      "loss": 0.3785,
+      "num_input_tokens_seen": 4512568,
+      "step": 50
+    },
+    {
+      "epoch": 0.22988505747126436,
+      "loss": 0.3789505362510681,
+      "loss_ce": 0.001997418701648712,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.376953125,
+      "num_input_tokens_seen": 4512568,
+      "step": 50
+    },
+    {
+      "epoch": 0.23448275862068965,
+      "grad_norm": 13.227669876358155,
+      "learning_rate": 5e-06,
+      "loss": 0.4474,
+      "num_input_tokens_seen": 4602984,
+      "step": 51
+    },
+    {
+      "epoch": 0.23448275862068965,
+      "loss": 0.5125839710235596,
+      "loss_ce": 0.09559179842472076,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.08349609375,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 4602984,
+      "step": 51
+    },
+    {
+      "epoch": 0.23908045977011494,
+      "grad_norm": 23.649689887557532,
+      "learning_rate": 5e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 4693356,
+      "step": 52
+    },
+    {
+      "epoch": 0.23908045977011494,
+      "loss": 0.3423450291156769,
+      "loss_ce": 0.011290331371128559,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 4693356,
+      "step": 52
+    },
+    {
+      "epoch": 0.24367816091954023,
+      "grad_norm": 29.218941891807887,
+      "learning_rate": 5e-06,
+      "loss": 0.4166,
+      "num_input_tokens_seen": 4783796,
+      "step": 53
+    },
+    {
+      "epoch": 0.24367816091954023,
+      "loss": 0.4327765107154846,
+      "loss_ce": 0.06705383956432343,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 4783796,
+      "step": 53
+    },
+    {
+      "epoch": 0.2482758620689655,
+      "grad_norm": 12.83340723694157,
+      "learning_rate": 5e-06,
+      "loss": 0.4025,
+      "num_input_tokens_seen": 4874232,
+      "step": 54
+    },
+    {
+      "epoch": 0.2482758620689655,
+      "loss": 0.3716029226779938,
+      "loss_ce": 0.03688611835241318,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 4874232,
+      "step": 54
+    },
+    {
+      "epoch": 0.25287356321839083,
+      "grad_norm": 8.220941129491107,
+      "learning_rate": 5e-06,
+      "loss": 0.4638,
+      "num_input_tokens_seen": 4963396,
+      "step": 55
+    },
+    {
+      "epoch": 0.25287356321839083,
+      "loss": 0.5566645860671997,
+      "loss_ce": 0.040917523205280304,
+      "loss_iou": 0.421875,
+      "loss_num": 0.10302734375,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 4963396,
+      "step": 55
+    },
+    {
+      "epoch": 0.2574712643678161,
+      "grad_norm": 13.288705604735872,
+      "learning_rate": 5e-06,
+      "loss": 0.3137,
+      "num_input_tokens_seen": 5053628,
+      "step": 56
+    },
+    {
+      "epoch": 0.2574712643678161,
+      "loss": 0.3751257061958313,
+      "loss_ce": 0.020999712869524956,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 5053628,
+      "step": 56
+    },
+    {
+      "epoch": 0.2620689655172414,
+      "grad_norm": 39.13223161382933,
+      "learning_rate": 5e-06,
+      "loss": 0.4695,
+      "num_input_tokens_seen": 5144036,
+      "step": 57
+    },
+    {
+      "epoch": 0.2620689655172414,
+      "loss": 0.5425558686256409,
+      "loss_ce": 0.023757044225931168,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.103515625,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 5144036,
+      "step": 57
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 15.33841968913398,
+      "learning_rate": 5e-06,
+      "loss": 0.3458,
+      "num_input_tokens_seen": 5234456,
+      "step": 58
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "loss": 0.38235118985176086,
+      "loss_ce": 0.03310802951455116,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 5234456,
+      "step": 58
+    },
+    {
+      "epoch": 0.271264367816092,
+      "grad_norm": 26.180906675160497,
+      "learning_rate": 5e-06,
+      "loss": 0.3917,
+      "num_input_tokens_seen": 5324744,
+      "step": 59
+    },
+    {
+      "epoch": 0.271264367816092,
+      "loss": 0.43390488624572754,
+      "loss_ce": 0.039861951023340225,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 5324744,
+      "step": 59
+    },
+    {
+      "epoch": 0.27586206896551724,
+      "grad_norm": 9.816205473208605,
+      "learning_rate": 5e-06,
+      "loss": 0.3208,
+      "num_input_tokens_seen": 5415040,
+      "step": 60
+    },
+    {
+      "epoch": 0.27586206896551724,
+      "loss": 0.244655042886734,
+      "loss_ce": 0.01015797071158886,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 5415040,
+      "step": 60
+    },
+    {
+      "epoch": 0.28045977011494255,
+      "grad_norm": 15.328830837517142,
+      "learning_rate": 5e-06,
+      "loss": 0.4713,
+      "num_input_tokens_seen": 5505400,
+      "step": 61
+    },
+    {
+      "epoch": 0.28045977011494255,
+      "loss": 0.3520286977291107,
+      "loss_ce": 0.0024193418212234974,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 5505400,
+      "step": 61
+    },
+    {
+      "epoch": 0.2850574712643678,
+      "grad_norm": 25.32547400452274,
+      "learning_rate": 5e-06,
+      "loss": 0.367,
+      "num_input_tokens_seen": 5595868,
+      "step": 62
+    },
+    {
+      "epoch": 0.2850574712643678,
+      "loss": 0.39265066385269165,
+      "loss_ce": 0.026805955916643143,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 5595868,
+      "step": 62
+    },
+    {
+      "epoch": 0.2896551724137931,
+      "grad_norm": 8.379792746028329,
+      "learning_rate": 5e-06,
+      "loss": 0.3734,
+      "num_input_tokens_seen": 5686176,
+      "step": 63
+    },
+    {
+      "epoch": 0.2896551724137931,
+      "loss": 0.413876473903656,
+      "loss_ce": 0.0571870356798172,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 5686176,
+      "step": 63
+    },
+    {
+      "epoch": 0.2942528735632184,
+      "grad_norm": 9.587938754950969,
+      "learning_rate": 5e-06,
+      "loss": 0.431,
+      "num_input_tokens_seen": 5776472,
+      "step": 64
+    },
+    {
+      "epoch": 0.2942528735632184,
+      "loss": 0.45100903511047363,
+      "loss_ce": 0.017171159386634827,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0869140625,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 5776472,
+      "step": 64
+    },
+    {
+      "epoch": 0.2988505747126437,
+      "grad_norm": 6.157907021347152,
+      "learning_rate": 5e-06,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 5866900,
+      "step": 65
+    },
+    {
+      "epoch": 0.2988505747126437,
+      "loss": 0.3870934844017029,
+      "loss_ce": 0.006722383201122284,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.380859375,
+      "num_input_tokens_seen": 5866900,
+      "step": 65
+    },
+    {
+      "epoch": 0.30344827586206896,
+      "grad_norm": 9.170883238290122,
+      "learning_rate": 5e-06,
+      "loss": 0.3444,
+      "num_input_tokens_seen": 5957224,
+      "step": 66
+    },
+    {
+      "epoch": 0.30344827586206896,
+      "loss": 0.3455989360809326,
+      "loss_ce": 0.0012385793961584568,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 5957224,
+      "step": 66
+    },
+    {
+      "epoch": 0.3080459770114943,
+      "grad_norm": 27.234249102852328,
+      "learning_rate": 5e-06,
+      "loss": 0.3401,
+      "num_input_tokens_seen": 6047500,
+      "step": 67
+    },
+    {
+      "epoch": 0.3080459770114943,
+      "loss": 0.3533991575241089,
+      "loss_ce": 0.007085674442350864,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 6047500,
+      "step": 67
+    },
+    {
+      "epoch": 0.31264367816091954,
+      "grad_norm": 47.75785601894446,
+      "learning_rate": 5e-06,
+      "loss": 0.3646,
+      "num_input_tokens_seen": 6137912,
+      "step": 68
+    },
+    {
+      "epoch": 0.31264367816091954,
+      "loss": 0.3522525429725647,
+      "loss_ce": 0.019244728609919548,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 6137912,
+      "step": 68
+    },
+    {
+      "epoch": 0.31724137931034485,
+      "grad_norm": 6.987555924094217,
+      "learning_rate": 5e-06,
+      "loss": 0.3722,
+      "num_input_tokens_seen": 6228264,
+      "step": 69
+    },
+    {
+      "epoch": 0.31724137931034485,
+      "loss": 0.3856116533279419,
+      "loss_ce": 0.011588208377361298,
+      "loss_iou": 0.53125,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 6228264,
+      "step": 69
+    },
+    {
+      "epoch": 0.3218390804597701,
+      "grad_norm": 5.498036338797732,
+      "learning_rate": 5e-06,
+      "loss": 0.3389,
+      "num_input_tokens_seen": 6318588,
+      "step": 70
+    },
+    {
+      "epoch": 0.3218390804597701,
+      "loss": 0.31558507680892944,
+      "loss_ce": 0.0037564674858003855,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 6318588,
+      "step": 70
+    },
+    {
+      "epoch": 0.3264367816091954,
+      "grad_norm": 8.031546032564327,
+      "learning_rate": 5e-06,
+      "loss": 0.3435,
+      "num_input_tokens_seen": 6409052,
+      "step": 71
+    },
+    {
+      "epoch": 0.3264367816091954,
+      "loss": 0.25542575120925903,
+      "loss_ce": 0.016717255115509033,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 6409052,
+      "step": 71
+    },
+    {
+      "epoch": 0.3310344827586207,
+      "grad_norm": 6.547993935141339,
+      "learning_rate": 5e-06,
+      "loss": 0.4004,
+      "num_input_tokens_seen": 6499328,
+      "step": 72
+    },
+    {
+      "epoch": 0.3310344827586207,
+      "loss": 0.5037937164306641,
+      "loss_ce": 0.02991676703095436,
+      "loss_iou": 0.375,
+      "loss_num": 0.0947265625,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 6499328,
+      "step": 72
+    },
+    {
+      "epoch": 0.335632183908046,
+      "grad_norm": 68.22686102549073,
+      "learning_rate": 5e-06,
+      "loss": 0.4303,
+      "num_input_tokens_seen": 6589628,
+      "step": 73
+    },
+    {
+      "epoch": 0.335632183908046,
+      "loss": 0.39546552300453186,
+      "loss_ce": 0.03560223802924156,
+      "loss_iou": 0.40625,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 6589628,
+      "step": 73
+    },
+    {
+      "epoch": 0.34022988505747126,
+      "grad_norm": 13.503622709112387,
+      "learning_rate": 5e-06,
+      "loss": 0.345,
+      "num_input_tokens_seen": 6680112,
+      "step": 74
+    },
+    {
+      "epoch": 0.34022988505747126,
+      "loss": 0.3707137703895569,
+      "loss_ce": 0.03233487531542778,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 6680112,
+      "step": 74
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 13.855537520348836,
+      "learning_rate": 5e-06,
+      "loss": 0.3428,
+      "num_input_tokens_seen": 6770428,
+      "step": 75
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "loss": 0.34209129214286804,
+      "loss_ce": 0.017384245991706848,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 6770428,
+      "step": 75
+    },
+    {
+      "epoch": 0.34942528735632183,
+      "grad_norm": 23.91656100675138,
+      "learning_rate": 5e-06,
+      "loss": 0.395,
+      "num_input_tokens_seen": 6860876,
+      "step": 76
+    },
+    {
+      "epoch": 0.34942528735632183,
+      "loss": 0.3706984221935272,
+      "loss_ce": 0.04159684479236603,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 6860876,
+      "step": 76
+    },
+    {
+      "epoch": 0.35402298850574715,
+      "grad_norm": 12.200721545193472,
+      "learning_rate": 5e-06,
+      "loss": 0.3466,
+      "num_input_tokens_seen": 6950396,
+      "step": 77
+    },
+    {
+      "epoch": 0.35402298850574715,
+      "loss": 0.4722903072834015,
+      "loss_ce": 0.05505400151014328,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.08349609375,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 6950396,
+      "step": 77
+    },
+    {
+      "epoch": 0.3586206896551724,
+      "grad_norm": 8.19258423406812,
+      "learning_rate": 5e-06,
+      "loss": 0.3716,
+      "num_input_tokens_seen": 7040716,
+      "step": 78
+    },
+    {
+      "epoch": 0.3586206896551724,
+      "loss": 0.37182319164276123,
+      "loss_ce": 0.010861298069357872,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 7040716,
+      "step": 78
+    },
+    {
+      "epoch": 0.3632183908045977,
+      "grad_norm": 18.895525833487042,
+      "learning_rate": 5e-06,
+      "loss": 0.3798,
+      "num_input_tokens_seen": 7131028,
+      "step": 79
+    },
+    {
+      "epoch": 0.3632183908045977,
+      "loss": 0.2786239981651306,
+      "loss_ce": 0.0010361107997596264,
+      "loss_iou": 0.46875,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 7131028,
+      "step": 79
+    },
+    {
+      "epoch": 0.367816091954023,
+      "grad_norm": 14.693361209406406,
+      "learning_rate": 5e-06,
+      "loss": 0.33,
+      "num_input_tokens_seen": 7221608,
+      "step": 80
+    },
+    {
+      "epoch": 0.367816091954023,
+      "loss": 0.33968040347099304,
+      "loss_ce": 0.019367896020412445,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 7221608,
+      "step": 80
+    },
+    {
+      "epoch": 0.3724137931034483,
+      "grad_norm": 20.35099683349881,
+      "learning_rate": 5e-06,
+      "loss": 0.391,
+      "num_input_tokens_seen": 7311916,
+      "step": 81
+    },
+    {
+      "epoch": 0.3724137931034483,
+      "loss": 0.451404333114624,
+      "loss_ce": 0.04014945402741432,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.412109375,
+      "num_input_tokens_seen": 7311916,
+      "step": 81
+    },
+    {
+      "epoch": 0.37701149425287356,
+      "grad_norm": 9.233297779847412,
+      "learning_rate": 5e-06,
+      "loss": 0.3335,
+      "num_input_tokens_seen": 7402256,
+      "step": 82
+    },
+    {
+      "epoch": 0.37701149425287356,
+      "loss": 0.3448345959186554,
+      "loss_ce": 0.0010846068616956472,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 7402256,
+      "step": 82
+    },
+    {
+      "epoch": 0.3816091954022989,
+      "grad_norm": 2.170529749870518,
+      "learning_rate": 5e-06,
+      "loss": 0.206,
+      "num_input_tokens_seen": 7492744,
+      "step": 83
+    },
+    {
+      "epoch": 0.3816091954022989,
+      "loss": 0.19931316375732422,
+      "loss_ce": 0.002306930720806122,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 7492744,
+      "step": 83
+    },
+    {
+      "epoch": 0.38620689655172413,
+      "grad_norm": 5.8890959300403,
+      "learning_rate": 5e-06,
+      "loss": 0.3027,
+      "num_input_tokens_seen": 7583004,
+      "step": 84
+    },
+    {
+      "epoch": 0.38620689655172413,
+      "loss": 0.3211175799369812,
+      "loss_ce": 0.01050965953618288,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 7583004,
+      "step": 84
+    },
+    {
+      "epoch": 0.39080459770114945,
+      "grad_norm": 13.034484868845816,
+      "learning_rate": 5e-06,
+      "loss": 0.3487,
+      "num_input_tokens_seen": 7673256,
+      "step": 85
+    },
+    {
+      "epoch": 0.39080459770114945,
+      "loss": 0.36519187688827515,
+      "loss_ce": 0.003131319535896182,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 7673256,
+      "step": 85
+    },
+    {
+      "epoch": 0.3954022988505747,
+      "grad_norm": 13.756069652716036,
+      "learning_rate": 5e-06,
+      "loss": 0.335,
+      "num_input_tokens_seen": 7763604,
+      "step": 86
+    },
+    {
+      "epoch": 0.3954022988505747,
+      "loss": 0.37784039974212646,
+      "loss_ce": 0.018831633031368256,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 7763604,
+      "step": 86
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 15.575683221755641,
+      "learning_rate": 5e-06,
+      "loss": 0.3935,
+      "num_input_tokens_seen": 7853108,
+      "step": 87
+    },
+    {
+      "epoch": 0.4,
+      "loss": 0.36000126600265503,
+      "loss_ce": 0.04499881714582443,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 7853108,
+      "step": 87
+    },
+    {
+      "epoch": 0.4045977011494253,
+      "grad_norm": 25.9232530151757,
+      "learning_rate": 5e-06,
+      "loss": 0.3719,
+      "num_input_tokens_seen": 7943488,
+      "step": 88
+    },
+    {
+      "epoch": 0.4045977011494253,
+      "loss": 0.42258501052856445,
+      "loss_ce": 0.03049515187740326,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 7943488,
+      "step": 88
+    },
+    {
+      "epoch": 0.4091954022988506,
+      "grad_norm": 13.656470567398008,
+      "learning_rate": 5e-06,
+      "loss": 0.3132,
+      "num_input_tokens_seen": 8033868,
+      "step": 89
+    },
+    {
+      "epoch": 0.4091954022988506,
+      "loss": 0.2531309425830841,
+      "loss_ce": 0.0003233220777474344,
+      "loss_iou": 0.484375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 8033868,
+      "step": 89
+    },
+    {
+      "epoch": 0.41379310344827586,
+      "grad_norm": 16.547729090720072,
+      "learning_rate": 5e-06,
+      "loss": 0.318,
+      "num_input_tokens_seen": 8124172,
+      "step": 90
+    },
+    {
+      "epoch": 0.41379310344827586,
+      "loss": 0.30939018726348877,
+      "loss_ce": 0.001162638422101736,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 8124172,
+      "step": 90
+    },
+    {
+      "epoch": 0.41839080459770117,
+      "grad_norm": 15.99038307829354,
+      "learning_rate": 5e-06,
+      "loss": 0.3983,
+      "num_input_tokens_seen": 8213772,
+      "step": 91
+    },
+    {
+      "epoch": 0.41839080459770117,
+      "loss": 0.5176781415939331,
+      "loss_ce": 0.1472862958908081,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 8213772,
+      "step": 91
+    },
+    {
+      "epoch": 0.42298850574712643,
+      "grad_norm": 17.254415632547147,
+      "learning_rate": 5e-06,
+      "loss": 0.4036,
+      "num_input_tokens_seen": 8304136,
+      "step": 92
+    },
+    {
+      "epoch": 0.42298850574712643,
+      "loss": 0.44501999020576477,
+      "loss_ce": 0.025098130106925964,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 8304136,
+      "step": 92
+    },
+    {
+      "epoch": 0.42758620689655175,
+      "grad_norm": 22.369663390664186,
+      "learning_rate": 5e-06,
+      "loss": 0.3815,
+      "num_input_tokens_seen": 8394520,
+      "step": 93
+    },
+    {
+      "epoch": 0.42758620689655175,
+      "loss": 0.3709450364112854,
+      "loss_ce": 0.0024147892836481333,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 8394520,
+      "step": 93
+    },
+    {
+      "epoch": 0.432183908045977,
+      "grad_norm": 7.546128838220428,
+      "learning_rate": 5e-06,
+      "loss": 0.3913,
+      "num_input_tokens_seen": 8484876,
+      "step": 94
+    },
+    {
+      "epoch": 0.432183908045977,
+      "loss": 0.4254246652126312,
+      "loss_ce": 0.01648910902440548,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 8484876,
+      "step": 94
+    },
+    {
+      "epoch": 0.4367816091954023,
+      "grad_norm": 11.464537758049973,
+      "learning_rate": 5e-06,
+      "loss": 0.3445,
+      "num_input_tokens_seen": 8575288,
+      "step": 95
+    },
+    {
+      "epoch": 0.4367816091954023,
+      "loss": 0.4096546173095703,
+      "loss_ce": 0.011278166435658932,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.07958984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 8575288,
+      "step": 95
+    },
+    {
+      "epoch": 0.4413793103448276,
+      "grad_norm": 14.34926027066774,
+      "learning_rate": 5e-06,
+      "loss": 0.3164,
+      "num_input_tokens_seen": 8665812,
+      "step": 96
+    },
+    {
+      "epoch": 0.4413793103448276,
+      "loss": 0.3848586082458496,
+      "loss_ce": 0.016572486609220505,
+      "loss_iou": 0.359375,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 8665812,
+      "step": 96
+    },
+    {
+      "epoch": 0.4459770114942529,
+      "grad_norm": 9.735611757167003,
+      "learning_rate": 5e-06,
+      "loss": 0.3679,
+      "num_input_tokens_seen": 8756136,
+      "step": 97
+    },
+    {
+      "epoch": 0.4459770114942529,
+      "loss": 0.39260411262512207,
+      "loss_ce": 0.046046510338783264,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 8756136,
+      "step": 97
+    },
+    {
+      "epoch": 0.45057471264367815,
+      "grad_norm": 8.954384165667761,
+      "learning_rate": 5e-06,
+      "loss": 0.2693,
+      "num_input_tokens_seen": 8846560,
+      "step": 98
+    },
+    {
+      "epoch": 0.45057471264367815,
+      "loss": 0.2520179748535156,
+      "loss_ce": 0.0018959222361445427,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 8846560,
+      "step": 98
+    },
+    {
+      "epoch": 0.45517241379310347,
+      "grad_norm": 4.271408735001314,
+      "learning_rate": 5e-06,
+      "loss": 0.2796,
+      "num_input_tokens_seen": 8936096,
+      "step": 99
+    },
+    {
+      "epoch": 0.45517241379310347,
+      "loss": 0.2937536835670471,
+      "loss_ce": 0.01689821295440197,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 8936096,
+      "step": 99
+    },
+    {
+      "epoch": 0.45977011494252873,
+      "grad_norm": 45.09395035520587,
+      "learning_rate": 5e-06,
+      "loss": 0.341,
+      "num_input_tokens_seen": 9026468,
+      "step": 100
+    },
+    {
+      "epoch": 0.45977011494252873,
+      "loss": 0.3114800751209259,
+      "loss_ce": 0.009722266346216202,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 9026468,
+      "step": 100
+    },
+    {
+      "epoch": 0.46436781609195404,
+      "grad_norm": 19.359219772759506,
+      "learning_rate": 5e-06,
+      "loss": 0.341,
+      "num_input_tokens_seen": 9116736,
+      "step": 101
+    },
+    {
+      "epoch": 0.46436781609195404,
+      "loss": 0.3214173913002014,
+      "loss_ce": 0.003912519197911024,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 9116736,
+      "step": 101
+    },
+    {
+      "epoch": 0.4689655172413793,
+      "grad_norm": 6.534059918097954,
+      "learning_rate": 5e-06,
+      "loss": 0.3168,
+      "num_input_tokens_seen": 9207040,
+      "step": 102
+    },
+    {
+      "epoch": 0.4689655172413793,
+      "loss": 0.3264657258987427,
+      "loss_ce": 0.02580653503537178,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 9207040,
+      "step": 102
+    },
+    {
+      "epoch": 0.4735632183908046,
+      "grad_norm": 7.393393813109138,
+      "learning_rate": 5e-06,
+      "loss": 0.359,
+      "num_input_tokens_seen": 9297316,
+      "step": 103
+    },
+    {
+      "epoch": 0.4735632183908046,
+      "loss": 0.3772013187408447,
+      "loss_ce": 0.01831459254026413,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 9297316,
+      "step": 103
+    },
+    {
+      "epoch": 0.4781609195402299,
+      "grad_norm": 41.31290787104306,
+      "learning_rate": 5e-06,
+      "loss": 0.3541,
+      "num_input_tokens_seen": 9387640,
+      "step": 104
+    },
+    {
+      "epoch": 0.4781609195402299,
+      "loss": 0.32786232233047485,
+      "loss_ce": 0.0027890922501683235,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 9387640,
+      "step": 104
+    },
+    {
+      "epoch": 0.4827586206896552,
+      "grad_norm": 5.6085929794757,
+      "learning_rate": 5e-06,
+      "loss": 0.3398,
+      "num_input_tokens_seen": 9477976,
+      "step": 105
+    },
+    {
+      "epoch": 0.4827586206896552,
+      "loss": 0.3234696090221405,
+      "loss_ce": 0.02805946208536625,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 9477976,
+      "step": 105
+    },
+    {
+      "epoch": 0.48735632183908045,
+      "grad_norm": 20.069977336539267,
+      "learning_rate": 5e-06,
+      "loss": 0.4092,
+      "num_input_tokens_seen": 9568252,
+      "step": 106
+    },
+    {
+      "epoch": 0.48735632183908045,
+      "loss": 0.3579648733139038,
+      "loss_ce": 0.017388703301548958,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 9568252,
+      "step": 106
+    },
+    {
+      "epoch": 0.49195402298850577,
+      "grad_norm": 69.89835804812493,
+      "learning_rate": 5e-06,
+      "loss": 0.4106,
+      "num_input_tokens_seen": 9657112,
+      "step": 107
+    },
+    {
+      "epoch": 0.49195402298850577,
+      "loss": 0.38876211643218994,
+      "loss_ce": 0.019133206456899643,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 9657112,
+      "step": 107
+    },
+    {
+      "epoch": 0.496551724137931,
+      "grad_norm": 6.848175278978215,
+      "learning_rate": 5e-06,
+      "loss": 0.2565,
+      "num_input_tokens_seen": 9747548,
+      "step": 108
+    },
+    {
+      "epoch": 0.496551724137931,
+      "loss": 0.21308478713035583,
+      "loss_ce": 0.00849493220448494,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 9747548,
+      "step": 108
+    },
+    {
+      "epoch": 0.5011494252873563,
+      "grad_norm": 11.091512135924047,
+      "learning_rate": 5e-06,
+      "loss": 0.3075,
+      "num_input_tokens_seen": 9837824,
+      "step": 109
+    },
+    {
+      "epoch": 0.5011494252873563,
+      "loss": 0.3376855254173279,
+      "loss_ce": 0.01969236694276333,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 9837824,
+      "step": 109
+    },
+    {
+      "epoch": 0.5057471264367817,
+      "grad_norm": 8.760189132358677,
+      "learning_rate": 5e-06,
+      "loss": 0.2375,
+      "num_input_tokens_seen": 9928152,
+      "step": 110
+    },
+    {
+      "epoch": 0.5057471264367817,
+      "loss": 0.26674026250839233,
+      "loss_ce": 0.002091821748763323,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 9928152,
+      "step": 110
+    },
+    {
+      "epoch": 0.5103448275862069,
+      "grad_norm": 8.592645738220657,
+      "learning_rate": 5e-06,
+      "loss": 0.3073,
+      "num_input_tokens_seen": 10017708,
+      "step": 111
+    },
+    {
+      "epoch": 0.5103448275862069,
+      "loss": 0.29403021931648254,
+      "loss_ce": 0.005700131878256798,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 10017708,
+      "step": 111
+    },
+    {
+      "epoch": 0.5149425287356322,
+      "grad_norm": 13.883237322149405,
+      "learning_rate": 5e-06,
+      "loss": 0.2906,
+      "num_input_tokens_seen": 10108008,
+      "step": 112
+    },
+    {
+      "epoch": 0.5149425287356322,
+      "loss": 0.22113177180290222,
+      "loss_ce": 0.011414967477321625,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 10108008,
+      "step": 112
+    },
+    {
+      "epoch": 0.5195402298850574,
+      "grad_norm": 44.804966100254894,
+      "learning_rate": 5e-06,
+      "loss": 0.3082,
+      "num_input_tokens_seen": 10198460,
+      "step": 113
+    },
+    {
+      "epoch": 0.5195402298850574,
+      "loss": 0.26152580976486206,
+      "loss_ce": 0.0007836385047994554,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 10198460,
+      "step": 113
+    },
+    {
+      "epoch": 0.5241379310344828,
+      "grad_norm": 11.857004672937288,
+      "learning_rate": 5e-06,
+      "loss": 0.3335,
+      "num_input_tokens_seen": 10288856,
+      "step": 114
+    },
+    {
+      "epoch": 0.5241379310344828,
+      "loss": 0.3588264584541321,
+      "loss_ce": 0.039612580090761185,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 10288856,
+      "step": 114
+    },
+    {
+      "epoch": 0.5287356321839081,
+      "grad_norm": 12.190324307933235,
+      "learning_rate": 5e-06,
+      "loss": 0.2532,
+      "num_input_tokens_seen": 10379180,
+      "step": 115
+    },
+    {
+      "epoch": 0.5287356321839081,
+      "loss": 0.2677502930164337,
+      "loss_ce": 0.008900204673409462,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 10379180,
+      "step": 115
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 5.20934661237588,
+      "learning_rate": 5e-06,
+      "loss": 0.3062,
+      "num_input_tokens_seen": 10468788,
+      "step": 116
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "loss": 0.3348138928413391,
+      "loss_ce": 0.0026605918537825346,
+      "loss_iou": 0.40625,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 10468788,
+      "step": 116
+    },
+    {
+      "epoch": 0.5379310344827586,
+      "grad_norm": 102.83862228180482,
+      "learning_rate": 5e-06,
+      "loss": 0.2783,
+      "num_input_tokens_seen": 10559244,
+      "step": 117
+    },
+    {
+      "epoch": 0.5379310344827586,
+      "loss": 0.2607054114341736,
+      "loss_ce": 0.005578459706157446,
+      "loss_iou": 0.453125,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 10559244,
+      "step": 117
+    },
+    {
+      "epoch": 0.542528735632184,
+      "grad_norm": 11.560087419415964,
+      "learning_rate": 5e-06,
+      "loss": 0.2611,
+      "num_input_tokens_seen": 10649640,
+      "step": 118
+    },
+    {
+      "epoch": 0.542528735632184,
+      "loss": 0.2891330122947693,
+      "loss_ce": 0.017282411456108093,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 10649640,
+      "step": 118
+    },
+    {
+      "epoch": 0.5471264367816092,
+      "grad_norm": 6.8172804181644775,
+      "learning_rate": 5e-06,
+      "loss": 0.281,
+      "num_input_tokens_seen": 10740080,
+      "step": 119
+    },
+    {
+      "epoch": 0.5471264367816092,
+      "loss": 0.2734673321247101,
+      "loss_ce": 0.002349165268242359,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 10740080,
+      "step": 119
+    },
+    {
+      "epoch": 0.5517241379310345,
+      "grad_norm": 15.572102183682414,
+      "learning_rate": 5e-06,
+      "loss": 0.3282,
+      "num_input_tokens_seen": 10830516,
+      "step": 120
+    },
+    {
+      "epoch": 0.5517241379310345,
+      "loss": 0.37165123224258423,
+      "loss_ce": 0.0017782035283744335,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 10830516,
+      "step": 120
+    },
+    {
+      "epoch": 0.5563218390804597,
+      "grad_norm": 9.896381717450371,
+      "learning_rate": 5e-06,
+      "loss": 0.2514,
+      "num_input_tokens_seen": 10920860,
+      "step": 121
+    },
+    {
+      "epoch": 0.5563218390804597,
+      "loss": 0.24367234110832214,
+      "loss_ce": 0.0051469625905156136,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 10920860,
+      "step": 121
+    },
+    {
+      "epoch": 0.5609195402298851,
+      "grad_norm": 8.57733919240062,
+      "learning_rate": 5e-06,
+      "loss": 0.2907,
+      "num_input_tokens_seen": 11011300,
+      "step": 122
+    },
+    {
+      "epoch": 0.5609195402298851,
+      "loss": 0.30107301473617554,
+      "loss_ce": 0.005418718792498112,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 11011300,
+      "step": 122
+    },
+    {
+      "epoch": 0.5655172413793104,
+      "grad_norm": 19.20663757306324,
+      "learning_rate": 5e-06,
+      "loss": 0.208,
+      "num_input_tokens_seen": 11101728,
+      "step": 123
+    },
+    {
+      "epoch": 0.5655172413793104,
+      "loss": 0.1830858439207077,
+      "loss_ce": 0.006968907080590725,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 11101728,
+      "step": 123
+    },
+    {
+      "epoch": 0.5701149425287356,
+      "grad_norm": 15.412744082634788,
+      "learning_rate": 5e-06,
+      "loss": 0.3385,
+      "num_input_tokens_seen": 11192036,
+      "step": 124
+    },
+    {
+      "epoch": 0.5701149425287356,
+      "loss": 0.36701270937919617,
+      "loss_ce": 0.0038535220082849264,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 11192036,
+      "step": 124
+    },
+    {
+      "epoch": 0.5747126436781609,
+      "grad_norm": 56.494382183302136,
+      "learning_rate": 5e-06,
+      "loss": 0.3616,
+      "num_input_tokens_seen": 11282432,
+      "step": 125
+    },
+    {
+      "epoch": 0.5747126436781609,
+      "loss": 0.4529553949832916,
+      "loss_ce": 0.010572599247097969,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.08837890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 11282432,
+      "step": 125
+    },
+    {
+      "epoch": 0.5793103448275863,
+      "grad_norm": 10.883196608124821,
+      "learning_rate": 5e-06,
+      "loss": 0.2749,
+      "num_input_tokens_seen": 11372852,
+      "step": 126
+    },
+    {
+      "epoch": 0.5793103448275863,
+      "loss": 0.21984298527240753,
+      "loss_ce": 0.0012150609400123358,
+      "loss_iou": 0.421875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 11372852,
+      "step": 126
+    },
+    {
+      "epoch": 0.5839080459770115,
+      "grad_norm": 13.409455091165006,
+      "learning_rate": 5e-06,
+      "loss": 0.2527,
+      "num_input_tokens_seen": 11463312,
+      "step": 127
+    },
+    {
+      "epoch": 0.5839080459770115,
+      "loss": 0.21884769201278687,
+      "loss_ce": 0.008154332637786865,
+      "loss_iou": 0.453125,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 11463312,
+      "step": 127
+    },
+    {
+      "epoch": 0.5885057471264368,
+      "grad_norm": 66.84081956875737,
+      "learning_rate": 5e-06,
+      "loss": 0.3699,
+      "num_input_tokens_seen": 11553556,
+      "step": 128
+    },
+    {
+      "epoch": 0.5885057471264368,
+      "loss": 0.3832904100418091,
+      "loss_ce": 0.004384158179163933,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 11553556,
+      "step": 128
+    },
+    {
+      "epoch": 0.593103448275862,
+      "grad_norm": 12.161700433268052,
+      "learning_rate": 5e-06,
+      "loss": 0.3168,
+      "num_input_tokens_seen": 11643936,
+      "step": 129
+    },
+    {
+      "epoch": 0.593103448275862,
+      "loss": 0.29100292921066284,
+      "loss_ce": 0.004259749781340361,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 11643936,
+      "step": 129
+    },
+    {
+      "epoch": 0.5977011494252874,
+      "grad_norm": 29.82460546961223,
+      "learning_rate": 5e-06,
+      "loss": 0.4414,
+      "num_input_tokens_seen": 11734172,
+      "step": 130
+    },
+    {
+      "epoch": 0.5977011494252874,
+      "loss": 0.43849754333496094,
+      "loss_ce": 0.007955562323331833,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 11734172,
+      "step": 130
+    },
+    {
+      "epoch": 0.6022988505747127,
+      "grad_norm": 9.962930837362645,
+      "learning_rate": 5e-06,
+      "loss": 0.3169,
+      "num_input_tokens_seen": 11824480,
+      "step": 131
+    },
+    {
+      "epoch": 0.6022988505747127,
+      "loss": 0.28843504190444946,
+      "loss_ce": 0.01182370726019144,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 11824480,
+      "step": 131
+    },
+    {
+      "epoch": 0.6068965517241379,
+      "grad_norm": 19.583672279804937,
+      "learning_rate": 5e-06,
+      "loss": 0.33,
+      "num_input_tokens_seen": 11914836,
+      "step": 132
+    },
+    {
+      "epoch": 0.6068965517241379,
+      "loss": 0.43384528160095215,
+      "loss_ce": 0.0053784484043717384,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 11914836,
+      "step": 132
+    },
+    {
+      "epoch": 0.6114942528735632,
+      "grad_norm": 25.89247440808492,
+      "learning_rate": 5e-06,
+      "loss": 0.2403,
+      "num_input_tokens_seen": 12005236,
+      "step": 133
+    },
+    {
+      "epoch": 0.6114942528735632,
+      "loss": 0.23137599229812622,
+      "loss_ce": 0.003348652273416519,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 12005236,
+      "step": 133
+    },
+    {
+      "epoch": 0.6160919540229886,
+      "grad_norm": 11.009266151622725,
+      "learning_rate": 5e-06,
+      "loss": 0.3161,
+      "num_input_tokens_seen": 12095572,
+      "step": 134
+    },
+    {
+      "epoch": 0.6160919540229886,
+      "loss": 0.32933273911476135,
+      "loss_ce": 0.003038788214325905,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 12095572,
+      "step": 134
+    },
+    {
+      "epoch": 0.6206896551724138,
+      "grad_norm": 4.89493638128289,
+      "learning_rate": 5e-06,
+      "loss": 0.2481,
+      "num_input_tokens_seen": 12185852,
+      "step": 135
+    },
+    {
+      "epoch": 0.6206896551724138,
+      "loss": 0.17523905634880066,
+      "loss_ce": 0.0022654307540506124,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 12185852,
+      "step": 135
+    },
+    {
+      "epoch": 0.6252873563218391,
+      "grad_norm": 9.499104731899472,
+      "learning_rate": 5e-06,
+      "loss": 0.2763,
+      "num_input_tokens_seen": 12276216,
+      "step": 136
+    },
+    {
+      "epoch": 0.6252873563218391,
+      "loss": 0.24068671464920044,
+      "loss_ce": 0.010278991423547268,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 12276216,
+      "step": 136
+    },
+    {
+      "epoch": 0.6298850574712643,
+      "grad_norm": 16.393863320752942,
+      "learning_rate": 5e-06,
+      "loss": 0.3345,
+      "num_input_tokens_seen": 12366584,
+      "step": 137
+    },
+    {
+      "epoch": 0.6298850574712643,
+      "loss": 0.3390456438064575,
+      "loss_ce": 0.004023653920739889,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 12366584,
+      "step": 137
+    },
+    {
+      "epoch": 0.6344827586206897,
+      "grad_norm": 12.427779653501139,
+      "learning_rate": 5e-06,
+      "loss": 0.2663,
+      "num_input_tokens_seen": 12457056,
+      "step": 138
+    },
+    {
+      "epoch": 0.6344827586206897,
+      "loss": 0.2509799599647522,
+      "loss_ce": 0.004947238601744175,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 12457056,
+      "step": 138
+    },
+    {
+      "epoch": 0.639080459770115,
+      "grad_norm": 4.879940316834736,
+      "learning_rate": 5e-06,
+      "loss": 0.291,
+      "num_input_tokens_seen": 12547580,
+      "step": 139
+    },
+    {
+      "epoch": 0.639080459770115,
+      "loss": 0.28111234307289124,
+      "loss_ce": 0.0024258154444396496,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 12547580,
+      "step": 139
+    },
+    {
+      "epoch": 0.6436781609195402,
+      "grad_norm": 19.612477583005116,
+      "learning_rate": 5e-06,
+      "loss": 0.2142,
+      "num_input_tokens_seen": 12637988,
+      "step": 140
+    },
+    {
+      "epoch": 0.6436781609195402,
+      "loss": 0.21896348893642426,
+      "loss_ce": 0.0008238445734605193,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 12637988,
+      "step": 140
+    },
+    {
+      "epoch": 0.6482758620689655,
+      "grad_norm": 8.925670199142369,
+      "learning_rate": 5e-06,
+      "loss": 0.3375,
+      "num_input_tokens_seen": 12727624,
+      "step": 141
+    },
+    {
+      "epoch": 0.6482758620689655,
+      "loss": 0.3529576063156128,
+      "loss_ce": 0.010306272655725479,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 12727624,
+      "step": 141
+    },
+    {
+      "epoch": 0.6528735632183909,
+      "grad_norm": 12.46478278407722,
+      "learning_rate": 5e-06,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 12818080,
+      "step": 142
+    },
+    {
+      "epoch": 0.6528735632183909,
+      "loss": 0.2979605793952942,
+      "loss_ce": 0.011522573418915272,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 12818080,
+      "step": 142
+    },
+    {
+      "epoch": 0.6574712643678161,
+      "grad_norm": 6.157855338602633,
+      "learning_rate": 5e-06,
+      "loss": 0.2823,
+      "num_input_tokens_seen": 12908420,
+      "step": 143
+    },
+    {
+      "epoch": 0.6574712643678161,
+      "loss": 0.2000827193260193,
+      "loss_ce": 0.015390344895422459,
+      "loss_iou": 0.5,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 12908420,
+      "step": 143
+    },
+    {
+      "epoch": 0.6620689655172414,
+      "grad_norm": 7.342547102611114,
+      "learning_rate": 5e-06,
+      "loss": 0.3044,
+      "num_input_tokens_seen": 12998800,
+      "step": 144
+    },
+    {
+      "epoch": 0.6620689655172414,
+      "loss": 0.3009682297706604,
+      "loss_ce": 0.0017739273607730865,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 12998800,
+      "step": 144
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 4.580788750753373,
+      "learning_rate": 5e-06,
+      "loss": 0.251,
+      "num_input_tokens_seen": 13088988,
+      "step": 145
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "loss": 0.24675819277763367,
+      "loss_ce": 0.0006034071557223797,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 13088988,
+      "step": 145
+    },
+    {
+      "epoch": 0.671264367816092,
+      "grad_norm": 36.43993056681854,
+      "learning_rate": 5e-06,
+      "loss": 0.3002,
+      "num_input_tokens_seen": 13179296,
+      "step": 146
+    },
+    {
+      "epoch": 0.671264367816092,
+      "loss": 0.29325583577156067,
+      "loss_ce": 0.003033676417544484,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 13179296,
+      "step": 146
+    },
+    {
+      "epoch": 0.6758620689655173,
+      "grad_norm": 19.476664971027393,
+      "learning_rate": 5e-06,
+      "loss": 0.301,
+      "num_input_tokens_seen": 13269688,
+      "step": 147
+    },
+    {
+      "epoch": 0.6758620689655173,
+      "loss": 0.27462151646614075,
+      "loss_ce": 0.014977962709963322,
+      "loss_iou": 0.390625,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 13269688,
+      "step": 147
+    },
+    {
+      "epoch": 0.6804597701149425,
+      "grad_norm": 6.680497018215309,
+      "learning_rate": 5e-06,
+      "loss": 0.2906,
+      "num_input_tokens_seen": 13360096,
+      "step": 148
+    },
+    {
+      "epoch": 0.6804597701149425,
+      "loss": 0.31177955865859985,
+      "loss_ce": 0.007702387869358063,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 13360096,
+      "step": 148
+    },
+    {
+      "epoch": 0.6850574712643678,
+      "grad_norm": 42.907574919392374,
+      "learning_rate": 5e-06,
+      "loss": 0.2479,
+      "num_input_tokens_seen": 13450548,
+      "step": 149
+    },
+    {
+      "epoch": 0.6850574712643678,
+      "loss": 0.23805175721645355,
+      "loss_ce": 0.0016015599248930812,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 13450548,
+      "step": 149
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 26.91726277344863,
+      "learning_rate": 5e-06,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 13540784,
+      "step": 150
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "loss": 0.315399706363678,
+      "loss_ce": 0.0002141495351679623,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 13540784,
+      "step": 150
+    },
+    {
+      "epoch": 0.6942528735632184,
+      "grad_norm": 14.145343351845199,
+      "learning_rate": 5e-06,
+      "loss": 0.3234,
+      "num_input_tokens_seen": 13630388,
+      "step": 151
+    },
+    {
+      "epoch": 0.6942528735632184,
+      "loss": 0.29768356680870056,
+      "loss_ce": 0.0025175553746521473,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 13630388,
+      "step": 151
+    },
+    {
+      "epoch": 0.6988505747126437,
+      "grad_norm": 6.305485293216934,
+      "learning_rate": 5e-06,
+      "loss": 0.34,
+      "num_input_tokens_seen": 13720696,
+      "step": 152
+    },
+    {
+      "epoch": 0.6988505747126437,
+      "loss": 0.23166516423225403,
+      "loss_ce": 0.00021984206978231668,
+      "loss_iou": 0.375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 13720696,
+      "step": 152
+    },
+    {
+      "epoch": 0.7034482758620689,
+      "grad_norm": 7.601656082559478,
+      "learning_rate": 5e-06,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 13811092,
+      "step": 153
+    },
+    {
+      "epoch": 0.7034482758620689,
+      "loss": 0.32397621870040894,
+      "loss_ce": 0.010408090427517891,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0625,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 13811092,
+      "step": 153
+    },
+    {
+      "epoch": 0.7080459770114943,
+      "grad_norm": 9.200935279622126,
+      "learning_rate": 5e-06,
+      "loss": 0.2512,
+      "num_input_tokens_seen": 13900680,
+      "step": 154
+    },
+    {
+      "epoch": 0.7080459770114943,
+      "loss": 0.1441739946603775,
+      "loss_ce": 0.0004972339374944568,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 13900680,
+      "step": 154
+    },
+    {
+      "epoch": 0.7126436781609196,
+      "grad_norm": 15.640269655312782,
+      "learning_rate": 5e-06,
+      "loss": 0.2777,
+      "num_input_tokens_seen": 13991164,
+      "step": 155
+    },
+    {
+      "epoch": 0.7126436781609196,
+      "loss": 0.26036322116851807,
+      "loss_ce": 0.004870051983743906,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 13991164,
+      "step": 155
+    },
+    {
+      "epoch": 0.7172413793103448,
+      "grad_norm": 8.162614855844033,
+      "learning_rate": 5e-06,
+      "loss": 0.2297,
+      "num_input_tokens_seen": 14081500,
+      "step": 156
+    },
+    {
+      "epoch": 0.7172413793103448,
+      "loss": 0.2218717634677887,
+      "loss_ce": 0.000680366822052747,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 14081500,
+      "step": 156
+    },
+    {
+      "epoch": 0.7218390804597701,
+      "grad_norm": 3.320193610069822,
+      "learning_rate": 5e-06,
+      "loss": 0.2703,
+      "num_input_tokens_seen": 14171952,
+      "step": 157
+    },
+    {
+      "epoch": 0.7218390804597701,
+      "loss": 0.2780265808105469,
+      "loss_ce": 0.00043868436478078365,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 14171952,
+      "step": 157
+    },
+    {
+      "epoch": 0.7264367816091954,
+      "grad_norm": 35.54312110471689,
+      "learning_rate": 5e-06,
+      "loss": 0.2552,
+      "num_input_tokens_seen": 14262436,
+      "step": 158
+    },
+    {
+      "epoch": 0.7264367816091954,
+      "loss": 0.2401730865240097,
+      "loss_ce": 0.0017697698203846812,
+      "loss_iou": 0.421875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 14262436,
+      "step": 158
+    },
+    {
+      "epoch": 0.7310344827586207,
+      "grad_norm": 7.06971142407152,
+      "learning_rate": 5e-06,
+      "loss": 0.2596,
+      "num_input_tokens_seen": 14352756,
+      "step": 159
+    },
+    {
+      "epoch": 0.7310344827586207,
+      "loss": 0.20814064145088196,
+      "loss_ce": 0.00019387324573472142,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 14352756,
+      "step": 159
+    },
+    {
+      "epoch": 0.735632183908046,
+      "grad_norm": 3.0564053075210946,
+      "learning_rate": 5e-06,
+      "loss": 0.2989,
+      "num_input_tokens_seen": 14443136,
+      "step": 160
+    },
+    {
+      "epoch": 0.735632183908046,
+      "loss": 0.3017638325691223,
+      "loss_ce": 0.007879569195210934,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 14443136,
+      "step": 160
+    },
+    {
+      "epoch": 0.7402298850574712,
+      "grad_norm": 12.412697167878472,
+      "learning_rate": 5e-06,
+      "loss": 0.2787,
+      "num_input_tokens_seen": 14533508,
+      "step": 161
+    },
+    {
+      "epoch": 0.7402298850574712,
+      "loss": 0.2811751663684845,
+      "loss_ce": 0.006394892930984497,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 14533508,
+      "step": 161
+    },
+    {
+      "epoch": 0.7448275862068966,
+      "grad_norm": 7.776759221836541,
+      "learning_rate": 5e-06,
+      "loss": 0.2709,
+      "num_input_tokens_seen": 14623944,
+      "step": 162
+    },
+    {
+      "epoch": 0.7448275862068966,
+      "loss": 0.27529919147491455,
+      "loss_ce": 0.004486183635890484,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 14623944,
+      "step": 162
+    },
+    {
+      "epoch": 0.7494252873563219,
+      "grad_norm": 4.690047668781669,
+      "learning_rate": 5e-06,
+      "loss": 0.2994,
+      "num_input_tokens_seen": 14714176,
+      "step": 163
+    },
+    {
+      "epoch": 0.7494252873563219,
+      "loss": 0.3100579082965851,
+      "loss_ce": 0.01446463167667389,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 14714176,
+      "step": 163
+    },
+    {
+      "epoch": 0.7540229885057471,
+      "grad_norm": 14.854640525604244,
+      "learning_rate": 5e-06,
+      "loss": 0.3539,
+      "num_input_tokens_seen": 14804540,
+      "step": 164
+    },
+    {
+      "epoch": 0.7540229885057471,
+      "loss": 0.44931861758232117,
+      "loss_ce": 0.01682349294424057,
+      "loss_iou": 0.390625,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 14804540,
+      "step": 164
+    },
+    {
+      "epoch": 0.7586206896551724,
+      "grad_norm": 17.524154236618525,
+      "learning_rate": 5e-06,
+      "loss": 0.3198,
+      "num_input_tokens_seen": 14895024,
+      "step": 165
+    },
+    {
+      "epoch": 0.7586206896551724,
+      "loss": 0.27517974376678467,
+      "loss_ce": 0.0019253486534580588,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 14895024,
+      "step": 165
+    },
+    {
+      "epoch": 0.7632183908045977,
+      "grad_norm": 14.028282838728552,
+      "learning_rate": 5e-06,
+      "loss": 0.2588,
+      "num_input_tokens_seen": 14985468,
+      "step": 166
+    },
+    {
+      "epoch": 0.7632183908045977,
+      "loss": 0.26632797718048096,
+      "loss_ce": 0.021088741719722748,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 14985468,
+      "step": 166
+    },
+    {
+      "epoch": 0.767816091954023,
+      "grad_norm": 13.184310261988879,
+      "learning_rate": 5e-06,
+      "loss": 0.3099,
+      "num_input_tokens_seen": 15075900,
+      "step": 167
+    },
+    {
+      "epoch": 0.767816091954023,
+      "loss": 0.2652779519557953,
+      "loss_ce": 0.001728140632621944,
+      "loss_iou": 0.4375,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 15075900,
+      "step": 167
+    },
+    {
+      "epoch": 0.7724137931034483,
+      "grad_norm": 14.801504481488628,
+      "learning_rate": 5e-06,
+      "loss": 0.2864,
+      "num_input_tokens_seen": 15166300,
+      "step": 168
+    },
+    {
+      "epoch": 0.7724137931034483,
+      "loss": 0.2605416774749756,
+      "loss_ce": 0.012616850435733795,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 15166300,
+      "step": 168
+    },
+    {
+      "epoch": 0.7770114942528735,
+      "grad_norm": 7.804434978620636,
+      "learning_rate": 5e-06,
+      "loss": 0.2893,
+      "num_input_tokens_seen": 15256556,
+      "step": 169
+    },
+    {
+      "epoch": 0.7770114942528735,
+      "loss": 0.17921343445777893,
+      "loss_ce": 0.003676328808069229,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 15256556,
+      "step": 169
+    },
+    {
+      "epoch": 0.7816091954022989,
+      "grad_norm": 7.905968772421947,
+      "learning_rate": 5e-06,
+      "loss": 0.2628,
+      "num_input_tokens_seen": 15346920,
+      "step": 170
+    },
+    {
+      "epoch": 0.7816091954022989,
+      "loss": 0.26719164848327637,
+      "loss_ce": 0.0010783508187159896,
+      "loss_iou": 0.453125,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 15346920,
+      "step": 170
+    },
+    {
+      "epoch": 0.7862068965517242,
+      "grad_norm": 5.672105961071837,
+      "learning_rate": 5e-06,
+      "loss": 0.2997,
+      "num_input_tokens_seen": 15437236,
+      "step": 171
+    },
+    {
+      "epoch": 0.7862068965517242,
+      "loss": 0.30997809767723083,
+      "loss_ce": 0.0011402069358155131,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 15437236,
+      "step": 171
+    },
+    {
+      "epoch": 0.7908045977011494,
+      "grad_norm": 31.647759717380147,
+      "learning_rate": 5e-06,
+      "loss": 0.3455,
+      "num_input_tokens_seen": 15527680,
+      "step": 172
+    },
+    {
+      "epoch": 0.7908045977011494,
+      "loss": 0.4378126859664917,
+      "loss_ce": 0.006202578078955412,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 15527680,
+      "step": 172
+    },
+    {
+      "epoch": 0.7954022988505747,
+      "grad_norm": 9.751745462208614,
+      "learning_rate": 5e-06,
+      "loss": 0.3265,
+      "num_input_tokens_seen": 15618124,
+      "step": 173
+    },
+    {
+      "epoch": 0.7954022988505747,
+      "loss": 0.4681752622127533,
+      "loss_ce": 0.1417897641658783,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 15618124,
+      "step": 173
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 11.98242525805576,
+      "learning_rate": 5e-06,
+      "loss": 0.303,
+      "num_input_tokens_seen": 15708484,
+      "step": 174
+    },
+    {
+      "epoch": 0.8,
+      "loss": 0.30590367317199707,
+      "loss_ce": 0.0006058230064809322,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 15708484,
+      "step": 174
+    },
+    {
+      "epoch": 0.8045977011494253,
+      "grad_norm": 10.897026680920138,
+      "learning_rate": 5e-06,
+      "loss": 0.2461,
+      "num_input_tokens_seen": 15798884,
+      "step": 175
+    },
+    {
+      "epoch": 0.8045977011494253,
+      "loss": 0.20932422578334808,
+      "loss_ce": 0.0005229614907875657,
+      "loss_iou": 0.515625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 15798884,
+      "step": 175
+    },
+    {
+      "epoch": 0.8091954022988506,
+      "grad_norm": 8.597962137572907,
+      "learning_rate": 5e-06,
+      "loss": 0.2656,
+      "num_input_tokens_seen": 15889316,
+      "step": 176
+    },
+    {
+      "epoch": 0.8091954022988506,
+      "loss": 0.2625455856323242,
+      "loss_ce": 0.0009489042568020523,
+      "loss_iou": 0.453125,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 15889316,
+      "step": 176
+    },
+    {
+      "epoch": 0.8137931034482758,
+      "grad_norm": 11.519278966897533,
+      "learning_rate": 5e-06,
+      "loss": 0.2704,
+      "num_input_tokens_seen": 15979592,
+      "step": 177
+    },
+    {
+      "epoch": 0.8137931034482758,
+      "loss": 0.2799646854400635,
+      "loss_ce": 0.000789885874837637,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 15979592,
+      "step": 177
+    },
+    {
+      "epoch": 0.8183908045977012,
+      "grad_norm": 53.5897696229877,
+      "learning_rate": 5e-06,
+      "loss": 0.3061,
+      "num_input_tokens_seen": 16066948,
+      "step": 178
+    },
+    {
+      "epoch": 0.8183908045977012,
+      "loss": 0.3006019592285156,
+      "loss_ce": 0.00037002595490776,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 16066948,
+      "step": 178
+    },
+    {
+      "epoch": 0.8229885057471265,
+      "grad_norm": 8.142699994347275,
+      "learning_rate": 5e-06,
+      "loss": 0.2444,
+      "num_input_tokens_seen": 16157256,
+      "step": 179
+    },
+    {
+      "epoch": 0.8229885057471265,
+      "loss": 0.21453739702701569,
+      "loss_ce": 0.00024295142793562263,
+      "loss_iou": 0.484375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 16157256,
+      "step": 179
+    },
+    {
+      "epoch": 0.8275862068965517,
+      "grad_norm": 8.723664846573774,
+      "learning_rate": 5e-06,
+      "loss": 0.2936,
+      "num_input_tokens_seen": 16247552,
+      "step": 180
+    },
+    {
+      "epoch": 0.8275862068965517,
+      "loss": 0.30998989939689636,
+      "loss_ce": 0.008781395852565765,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 16247552,
+      "step": 180
+    },
+    {
+      "epoch": 0.832183908045977,
+      "grad_norm": 31.640092388654665,
+      "learning_rate": 5e-06,
+      "loss": 0.2839,
+      "num_input_tokens_seen": 16337132,
+      "step": 181
+    },
+    {
+      "epoch": 0.832183908045977,
+      "loss": 0.2720521092414856,
+      "loss_ce": 0.006427087355405092,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 16337132,
+      "step": 181
+    },
+    {
+      "epoch": 0.8367816091954023,
+      "grad_norm": 15.726684290210118,
+      "learning_rate": 5e-06,
+      "loss": 0.2782,
+      "num_input_tokens_seen": 16427532,
+      "step": 182
+    },
+    {
+      "epoch": 0.8367816091954023,
+      "loss": 0.22582070529460907,
+      "loss_ce": 0.0006009868229739368,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 16427532,
+      "step": 182
+    },
+    {
+      "epoch": 0.8413793103448276,
+      "grad_norm": 9.413293029313524,
+      "learning_rate": 5e-06,
+      "loss": 0.2648,
+      "num_input_tokens_seen": 16517996,
+      "step": 183
+    },
+    {
+      "epoch": 0.8413793103448276,
+      "loss": 0.2550521492958069,
+      "loss_ce": 0.0031600736547261477,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 16517996,
+      "step": 183
+    },
+    {
+      "epoch": 0.8459770114942529,
+      "grad_norm": 17.034560829091234,
+      "learning_rate": 5e-06,
+      "loss": 0.2503,
+      "num_input_tokens_seen": 16608392,
+      "step": 184
+    },
+    {
+      "epoch": 0.8459770114942529,
+      "loss": 0.24292418360710144,
+      "loss_ce": 0.006596065126359463,
+      "loss_iou": 0.46875,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 16608392,
+      "step": 184
+    },
+    {
+      "epoch": 0.8505747126436781,
+      "grad_norm": 24.671310704425558,
+      "learning_rate": 5e-06,
+      "loss": 0.2523,
+      "num_input_tokens_seen": 16698852,
+      "step": 185
+    },
+    {
+      "epoch": 0.8505747126436781,
+      "loss": 0.2351730614900589,
+      "loss_ce": 0.002690147841349244,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 16698852,
+      "step": 185
+    },
+    {
+      "epoch": 0.8551724137931035,
+      "grad_norm": 8.258454597087393,
+      "learning_rate": 5e-06,
+      "loss": 0.2756,
+      "num_input_tokens_seen": 16789340,
+      "step": 186
+    },
+    {
+      "epoch": 0.8551724137931035,
+      "loss": 0.3019639849662781,
+      "loss_ce": 0.01021592691540718,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 16789340,
+      "step": 186
+    },
+    {
+      "epoch": 0.8597701149425288,
+      "grad_norm": 4.472141198961523,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 16879736,
+      "step": 187
+    },
+    {
+      "epoch": 0.8597701149425288,
+      "loss": 0.19520393013954163,
+      "loss_ce": 0.003004225669428706,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 16879736,
+      "step": 187
+    },
+    {
+      "epoch": 0.864367816091954,
+      "grad_norm": 4.244846244353357,
+      "learning_rate": 5e-06,
+      "loss": 0.2684,
+      "num_input_tokens_seen": 16968620,
+      "step": 188
+    },
+    {
+      "epoch": 0.864367816091954,
+      "loss": 0.29818519949913025,
+      "loss_ce": 0.007596845738589764,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 16968620,
+      "step": 188
+    },
+    {
+      "epoch": 0.8689655172413793,
+      "grad_norm": 17.287226125768598,
+      "learning_rate": 5e-06,
+      "loss": 0.2103,
+      "num_input_tokens_seen": 17058968,
+      "step": 189
+    },
+    {
+      "epoch": 0.8689655172413793,
+      "loss": 0.23691387474536896,
+      "loss_ce": 0.00015851360512897372,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 17058968,
+      "step": 189
+    },
+    {
+      "epoch": 0.8735632183908046,
+      "grad_norm": 11.788400904878815,
+      "learning_rate": 5e-06,
+      "loss": 0.2889,
+      "num_input_tokens_seen": 17149248,
+      "step": 190
+    },
+    {
+      "epoch": 0.8735632183908046,
+      "loss": 0.24897444248199463,
+      "loss_ce": 0.00013410599785856903,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 17149248,
+      "step": 190
+    },
+    {
+      "epoch": 0.8781609195402299,
+      "grad_norm": 13.617723461222461,
+      "learning_rate": 5e-06,
+      "loss": 0.2228,
+      "num_input_tokens_seen": 17239628,
+      "step": 191
+    },
+    {
+      "epoch": 0.8781609195402299,
+      "loss": 0.23315443098545074,
+      "loss_ce": 0.005981584079563618,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 17239628,
+      "step": 191
+    },
+    {
+      "epoch": 0.8827586206896552,
+      "grad_norm": 22.65066129559751,
+      "learning_rate": 5e-06,
+      "loss": 0.218,
+      "num_input_tokens_seen": 17329208,
+      "step": 192
+    },
+    {
+      "epoch": 0.8827586206896552,
+      "loss": 0.23583698272705078,
+      "loss_ce": 0.004879951477050781,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 17329208,
+      "step": 192
+    },
+    {
+      "epoch": 0.8873563218390804,
+      "grad_norm": 21.965810935500244,
+      "learning_rate": 5e-06,
+      "loss": 0.2832,
+      "num_input_tokens_seen": 17419616,
+      "step": 193
+    },
+    {
+      "epoch": 0.8873563218390804,
+      "loss": 0.2855183780193329,
+      "loss_ce": 0.004756668582558632,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 17419616,
+      "step": 193
+    },
+    {
+      "epoch": 0.8919540229885058,
+      "grad_norm": 8.272351331856266,
+      "learning_rate": 5e-06,
+      "loss": 0.2343,
+      "num_input_tokens_seen": 17507752,
+      "step": 194
+    },
+    {
+      "epoch": 0.8919540229885058,
+      "loss": 0.2399004101753235,
+      "loss_ce": 0.0027177799493074417,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 17507752,
+      "step": 194
+    },
+    {
+      "epoch": 0.896551724137931,
+      "grad_norm": 5.04648132108404,
+      "learning_rate": 5e-06,
+      "loss": 0.247,
+      "num_input_tokens_seen": 17595748,
+      "step": 195
+    },
+    {
+      "epoch": 0.896551724137931,
+      "loss": 0.2049858421087265,
+      "loss_ce": 0.02108692191541195,
+      "loss_iou": 0.421875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 17595748,
+      "step": 195
+    },
+    {
+      "epoch": 0.9011494252873563,
+      "grad_norm": 11.7279486833154,
+      "learning_rate": 5e-06,
+      "loss": 0.3365,
+      "num_input_tokens_seen": 17686040,
+      "step": 196
+    },
+    {
+      "epoch": 0.9011494252873563,
+      "loss": 0.37410539388656616,
+      "loss_ce": 0.002889568218961358,
+      "loss_iou": 0.375,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 17686040,
+      "step": 196
+    },
+    {
+      "epoch": 0.9057471264367816,
+      "grad_norm": 9.565742677061259,
+      "learning_rate": 5e-06,
+      "loss": 0.293,
+      "num_input_tokens_seen": 17776348,
+      "step": 197
+    },
+    {
+      "epoch": 0.9057471264367816,
+      "loss": 0.3795655369758606,
+      "loss_ce": 0.0005372293526306748,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 17776348,
+      "step": 197
+    },
+    {
+      "epoch": 0.9103448275862069,
+      "grad_norm": 5.3718887593573275,
+      "learning_rate": 5e-06,
+      "loss": 0.245,
+      "num_input_tokens_seen": 17866780,
+      "step": 198
+    },
+    {
+      "epoch": 0.9103448275862069,
+      "loss": 0.23624387383460999,
+      "loss_ce": 0.004615448880940676,
+      "loss_iou": 0.515625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 17866780,
+      "step": 198
+    },
+    {
+      "epoch": 0.9149425287356322,
+      "grad_norm": 10.445265942368462,
+      "learning_rate": 5e-06,
+      "loss": 0.3073,
+      "num_input_tokens_seen": 17957092,
+      "step": 199
+    },
+    {
+      "epoch": 0.9149425287356322,
+      "loss": 0.4080125093460083,
+      "loss_ce": 0.0026475111953914165,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0810546875,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 17957092,
+      "step": 199
+    },
+    {
+      "epoch": 0.9195402298850575,
+      "grad_norm": 13.36367583066565,
+      "learning_rate": 5e-06,
+      "loss": 0.2221,
+      "num_input_tokens_seen": 18047500,
+      "step": 200
+    },
+    {
+      "epoch": 0.9195402298850575,
+      "loss": 0.18272346258163452,
+      "loss_ce": 0.0006555922445841134,
+      "loss_iou": 0.421875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 18047500,
+      "step": 200
+    },
+    {
+      "epoch": 0.9241379310344827,
+      "grad_norm": 7.033010833292816,
+      "learning_rate": 5e-06,
+      "loss": 0.3052,
+      "num_input_tokens_seen": 18137988,
+      "step": 201
+    },
+    {
+      "epoch": 0.9241379310344827,
+      "loss": 0.37599673867225647,
+      "loss_ce": 0.007222320418804884,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 18137988,
+      "step": 201
+    },
+    {
+      "epoch": 0.9287356321839081,
+      "grad_norm": 3.214939505030853,
+      "learning_rate": 5e-06,
+      "loss": 0.2099,
+      "num_input_tokens_seen": 18228316,
+      "step": 202
+    },
+    {
+      "epoch": 0.9287356321839081,
+      "loss": 0.2917129397392273,
+      "loss_ce": 0.0024673594161868095,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 18228316,
+      "step": 202
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 47.44674206476791,
+      "learning_rate": 5e-06,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 18318712,
+      "step": 203
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "loss": 0.27268558740615845,
+      "loss_ce": 0.0027881115674972534,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 18318712,
+      "step": 203
+    },
+    {
+      "epoch": 0.9379310344827586,
+      "grad_norm": 15.635747866161521,
+      "learning_rate": 5e-06,
+      "loss": 0.2178,
+      "num_input_tokens_seen": 18409128,
+      "step": 204
+    },
+    {
+      "epoch": 0.9379310344827586,
+      "loss": 0.21751438081264496,
+      "loss_ce": 0.009933818131685257,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 18409128,
+      "step": 204
+    },
+    {
+      "epoch": 0.9425287356321839,
+      "grad_norm": 14.842085686479773,
+      "learning_rate": 5e-06,
+      "loss": 0.2294,
+      "num_input_tokens_seen": 18499512,
+      "step": 205
+    },
+    {
+      "epoch": 0.9425287356321839,
+      "loss": 0.2684970498085022,
+      "loss_ce": 0.005435529164969921,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 18499512,
+      "step": 205
+    },
+    {
+      "epoch": 0.9471264367816092,
+      "grad_norm": 10.75647435799341,
+      "learning_rate": 5e-06,
+      "loss": 0.2631,
+      "num_input_tokens_seen": 18589756,
+      "step": 206
+    },
+    {
+      "epoch": 0.9471264367816092,
+      "loss": 0.3199033737182617,
+      "loss_ce": 0.0008725962834432721,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 18589756,
+      "step": 206
+    },
+    {
+      "epoch": 0.9517241379310345,
+      "grad_norm": 10.014451793263762,
+      "learning_rate": 5e-06,
+      "loss": 0.3506,
+      "num_input_tokens_seen": 18680168,
+      "step": 207
+    },
+    {
+      "epoch": 0.9517241379310345,
+      "loss": 0.41780880093574524,
+      "loss_ce": 0.004295613616704941,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 18680168,
+      "step": 207
+    },
+    {
+      "epoch": 0.9563218390804598,
+      "grad_norm": 81.22703034283288,
+      "learning_rate": 5e-06,
+      "loss": 0.2683,
+      "num_input_tokens_seen": 18768984,
+      "step": 208
+    },
+    {
+      "epoch": 0.9563218390804598,
+      "loss": 0.22123399376869202,
+      "loss_ce": 0.00040880130836740136,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 18768984,
+      "step": 208
+    },
+    {
+      "epoch": 0.960919540229885,
+      "grad_norm": 18.16487091628093,
+      "learning_rate": 5e-06,
+      "loss": 0.2017,
+      "num_input_tokens_seen": 18859384,
+      "step": 209
+    },
+    {
+      "epoch": 0.960919540229885,
+      "loss": 0.1849987506866455,
+      "loss_ce": 0.00650144275277853,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 18859384,
+      "step": 209
+    },
+    {
+      "epoch": 0.9655172413793104,
+      "grad_norm": 6.997216906550849,
+      "learning_rate": 5e-06,
+      "loss": 0.3129,
+      "num_input_tokens_seen": 18949688,
+      "step": 210
+    },
+    {
+      "epoch": 0.9655172413793104,
+      "loss": 0.38702672719955444,
+      "loss_ce": 0.0003079898888245225,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 18949688,
+      "step": 210
+    },
+    {
+      "epoch": 0.9701149425287356,
+      "grad_norm": 7.51470392722959,
+      "learning_rate": 5e-06,
+      "loss": 0.2452,
+      "num_input_tokens_seen": 19039948,
+      "step": 211
+    },
+    {
+      "epoch": 0.9701149425287356,
+      "loss": 0.30642956495285034,
+      "loss_ce": 0.0001551464811200276,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 19039948,
+      "step": 211
+    },
+    {
+      "epoch": 0.9747126436781609,
+      "grad_norm": 5.659981606083217,
+      "learning_rate": 5e-06,
+      "loss": 0.2912,
+      "num_input_tokens_seen": 19130216,
+      "step": 212
+    },
+    {
+      "epoch": 0.9747126436781609,
+      "loss": 0.28307363390922546,
+      "loss_ce": 0.0021898397244513035,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 19130216,
+      "step": 212
+    },
+    {
+      "epoch": 0.9793103448275862,
+      "grad_norm": 15.905977611839273,
+      "learning_rate": 5e-06,
+      "loss": 0.2863,
+      "num_input_tokens_seen": 19220568,
+      "step": 213
+    },
+    {
+      "epoch": 0.9793103448275862,
+      "loss": 0.39371466636657715,
+      "loss_ce": 0.0007703077862970531,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 19220568,
+      "step": 213
+    },
+    {
+      "epoch": 0.9839080459770115,
+      "grad_norm": 6.117701022668606,
+      "learning_rate": 5e-06,
+      "loss": 0.2468,
+      "num_input_tokens_seen": 19311004,
+      "step": 214
+    },
+    {
+      "epoch": 0.9839080459770115,
+      "loss": 0.2368617057800293,
+      "loss_ce": 0.001754299970343709,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.046875,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 19311004,
+      "step": 214
+    },
+    {
+      "epoch": 0.9885057471264368,
+      "grad_norm": 13.736398962240743,
+      "learning_rate": 5e-06,
+      "loss": 0.2513,
+      "num_input_tokens_seen": 19401384,
+      "step": 215
+    },
+    {
+      "epoch": 0.9885057471264368,
+      "loss": 0.24910268187522888,
+      "loss_ce": 0.004153335001319647,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 19401384,
+      "step": 215
+    },
+    {
+      "epoch": 0.993103448275862,
+      "grad_norm": 5.219663028608165,
+      "learning_rate": 5e-06,
+      "loss": 0.2163,
+      "num_input_tokens_seen": 19491860,
+      "step": 216
+    },
+    {
+      "epoch": 0.993103448275862,
+      "loss": 0.2344357669353485,
+      "loss_ce": 0.0007016360759735107,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 19491860,
+      "step": 216
+    },
+    {
+      "epoch": 0.9977011494252873,
+      "grad_norm": 9.37488702763018,
+      "learning_rate": 5e-06,
+      "loss": 0.1707,
+      "num_input_tokens_seen": 19582284,
+      "step": 217
+    },
+    {
+      "epoch": 0.9977011494252873,
+      "loss": 0.14838698506355286,
+      "loss_ce": 0.00046829067287035286,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 19582284,
+      "step": 217
+    },
+    {
+      "epoch": 0.9977011494252873,
+      "loss": 0.24481603503227234,
+      "loss_ce": 0.001377313630655408,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 19627492,
+      "step": 217
+    },
+    {
+      "epoch": 1.0022988505747126,
+      "grad_norm": 13.179380791300144,
+      "learning_rate": 5e-06,
+      "loss": 0.237,
+      "num_input_tokens_seen": 19672672,
+      "step": 218
+    },
+    {
+      "epoch": 1.0022988505747126,
+      "loss": 0.22918623685836792,
+      "loss_ce": 0.0003043994656763971,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 19672672,
+      "step": 218
+    },
+    {
+      "epoch": 1.006896551724138,
+      "grad_norm": 11.575549644295782,
+      "learning_rate": 5e-06,
+      "loss": 0.1865,
+      "num_input_tokens_seen": 19762928,
+      "step": 219
+    },
+    {
+      "epoch": 1.006896551724138,
+      "loss": 0.20134694874286652,
+      "loss_ce": 5.299979602568783e-05,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 19762928,
+      "step": 219
+    },
+    {
+      "epoch": 1.0114942528735633,
+      "grad_norm": 7.770217563711411,
+      "learning_rate": 5e-06,
+      "loss": 0.2716,
+      "num_input_tokens_seen": 19853264,
+      "step": 220
+    },
+    {
+      "epoch": 1.0114942528735633,
+      "loss": 0.23982341587543488,
+      "loss_ce": 0.0010538852075114846,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 19853264,
+      "step": 220
+    },
+    {
+      "epoch": 1.0160919540229885,
+      "grad_norm": 13.318468212320983,
+      "learning_rate": 5e-06,
+      "loss": 0.2188,
+      "num_input_tokens_seen": 19942828,
+      "step": 221
+    },
+    {
+      "epoch": 1.0160919540229885,
+      "loss": 0.19837090373039246,
+      "loss_ce": 0.000189754442544654,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 19942828,
+      "step": 221
+    },
+    {
+      "epoch": 1.0206896551724138,
+      "grad_norm": 4.7677670902469735,
+      "learning_rate": 5e-06,
+      "loss": 0.1753,
+      "num_input_tokens_seen": 20033268,
+      "step": 222
+    },
+    {
+      "epoch": 1.0206896551724138,
+      "loss": 0.23340031504631042,
+      "loss_ce": 0.01013370230793953,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 20033268,
+      "step": 222
+    },
+    {
+      "epoch": 1.025287356321839,
+      "grad_norm": 7.294150569078434,
+      "learning_rate": 5e-06,
+      "loss": 0.2063,
+      "num_input_tokens_seen": 20123744,
+      "step": 223
+    },
+    {
+      "epoch": 1.025287356321839,
+      "loss": 0.16170556843280792,
+      "loss_ce": 0.00222071073949337,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 20123744,
+      "step": 223
+    },
+    {
+      "epoch": 1.0298850574712644,
+      "grad_norm": 13.473703879708541,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 20214216,
+      "step": 224
+    },
+    {
+      "epoch": 1.0298850574712644,
+      "loss": 0.12107338011264801,
+      "loss_ce": 0.0005899769021198153,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 20214216,
+      "step": 224
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "grad_norm": 16.323294537818267,
+      "learning_rate": 5e-06,
+      "loss": 0.2242,
+      "num_input_tokens_seen": 20304620,
+      "step": 225
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "loss": 0.2116052806377411,
+      "loss_ce": 0.002529359422624111,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 20304620,
+      "step": 225
+    },
+    {
+      "epoch": 1.0390804597701149,
+      "grad_norm": 5.045052986814076,
+      "learning_rate": 5e-06,
+      "loss": 0.2254,
+      "num_input_tokens_seen": 20394932,
+      "step": 226
+    },
+    {
+      "epoch": 1.0390804597701149,
+      "loss": 0.19622498750686646,
+      "loss_ce": 0.0004242146678734571,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 20394932,
+      "step": 226
+    },
+    {
+      "epoch": 1.0436781609195402,
+      "grad_norm": 10.85002200148262,
+      "learning_rate": 5e-06,
+      "loss": 0.2326,
+      "num_input_tokens_seen": 20485364,
+      "step": 227
+    },
+    {
+      "epoch": 1.0436781609195402,
+      "loss": 0.16513219475746155,
+      "loss_ce": 0.0007034791633486748,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 20485364,
+      "step": 227
+    },
+    {
+      "epoch": 1.0482758620689656,
+      "grad_norm": 10.36785476657075,
+      "learning_rate": 5e-06,
+      "loss": 0.189,
+      "num_input_tokens_seen": 20575780,
+      "step": 228
+    },
+    {
+      "epoch": 1.0482758620689656,
+      "loss": 0.1746249496936798,
+      "loss_ce": 0.0005526923341676593,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 20575780,
+      "step": 228
+    },
+    {
+      "epoch": 1.0528735632183908,
+      "grad_norm": 11.993359880118176,
+      "learning_rate": 5e-06,
+      "loss": 0.2183,
+      "num_input_tokens_seen": 20666132,
+      "step": 229
+    },
+    {
+      "epoch": 1.0528735632183908,
+      "loss": 0.24332331120967865,
+      "loss_ce": 0.001074782107025385,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 20666132,
+      "step": 229
+    },
+    {
+      "epoch": 1.0574712643678161,
+      "grad_norm": 21.11605854531589,
+      "learning_rate": 5e-06,
+      "loss": 0.2293,
+      "num_input_tokens_seen": 20756592,
+      "step": 230
+    },
+    {
+      "epoch": 1.0574712643678161,
+      "loss": 0.21172873675823212,
+      "loss_ce": 0.005735091865062714,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 20756592,
+      "step": 230
+    },
+    {
+      "epoch": 1.0620689655172413,
+      "grad_norm": 13.639736851512211,
+      "learning_rate": 5e-06,
+      "loss": 0.1664,
+      "num_input_tokens_seen": 20846888,
+      "step": 231
+    },
+    {
+      "epoch": 1.0620689655172413,
+      "loss": 0.1343887746334076,
+      "loss_ce": 0.010151727125048637,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 20846888,
+      "step": 231
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 4.8416225246295195,
+      "learning_rate": 5e-06,
+      "loss": 0.2231,
+      "num_input_tokens_seen": 20937220,
+      "step": 232
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "loss": 0.19790634512901306,
+      "loss_ce": 0.00024397407833021134,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 20937220,
+      "step": 232
+    },
+    {
+      "epoch": 1.071264367816092,
+      "grad_norm": 8.735510518442098,
+      "learning_rate": 5e-06,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 21027540,
+      "step": 233
+    },
+    {
+      "epoch": 1.071264367816092,
+      "loss": 0.2129209041595459,
+      "loss_ce": 0.014556641690433025,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 21027540,
+      "step": 233
+    },
+    {
+      "epoch": 1.0758620689655172,
+      "grad_norm": 64.22232050337796,
+      "learning_rate": 5e-06,
+      "loss": 0.243,
+      "num_input_tokens_seen": 21117908,
+      "step": 234
+    },
+    {
+      "epoch": 1.0758620689655172,
+      "loss": 0.19178733229637146,
+      "loss_ce": 0.000259012304013595,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 21117908,
+      "step": 234
+    },
+    {
+      "epoch": 1.0804597701149425,
+      "grad_norm": 29.01468538309071,
+      "learning_rate": 5e-06,
+      "loss": 0.2216,
+      "num_input_tokens_seen": 21208340,
+      "step": 235
+    },
+    {
+      "epoch": 1.0804597701149425,
+      "loss": 0.22727806866168976,
+      "loss_ce": 0.0060866596177220345,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 21208340,
+      "step": 235
+    },
+    {
+      "epoch": 1.085057471264368,
+      "grad_norm": 11.877399551682112,
+      "learning_rate": 5e-06,
+      "loss": 0.193,
+      "num_input_tokens_seen": 21298700,
+      "step": 236
+    },
+    {
+      "epoch": 1.085057471264368,
+      "loss": 0.2526742219924927,
+      "loss_ce": 0.00011075517249992117,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 21298700,
+      "step": 236
+    },
+    {
+      "epoch": 1.089655172413793,
+      "grad_norm": 11.06433974753488,
+      "learning_rate": 5e-06,
+      "loss": 0.2513,
+      "num_input_tokens_seen": 21389112,
+      "step": 237
+    },
+    {
+      "epoch": 1.089655172413793,
+      "loss": 0.2866198718547821,
+      "loss_ce": 0.0012194736627861857,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 21389112,
+      "step": 237
+    },
+    {
+      "epoch": 1.0942528735632184,
+      "grad_norm": 5.522127581505815,
+      "learning_rate": 5e-06,
+      "loss": 0.2329,
+      "num_input_tokens_seen": 21479544,
+      "step": 238
+    },
+    {
+      "epoch": 1.0942528735632184,
+      "loss": 0.24488475918769836,
+      "loss_ce": 0.0002558681007940322,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 21479544,
+      "step": 238
+    },
+    {
+      "epoch": 1.0988505747126436,
+      "grad_norm": 20.5377108131709,
+      "learning_rate": 5e-06,
+      "loss": 0.183,
+      "num_input_tokens_seen": 21569880,
+      "step": 239
+    },
+    {
+      "epoch": 1.0988505747126436,
+      "loss": 0.17812752723693848,
+      "loss_ce": 2.6934067136608064e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 21569880,
+      "step": 239
+    },
+    {
+      "epoch": 1.103448275862069,
+      "grad_norm": 12.413266667265427,
+      "learning_rate": 5e-06,
+      "loss": 0.2067,
+      "num_input_tokens_seen": 21660268,
+      "step": 240
+    },
+    {
+      "epoch": 1.103448275862069,
+      "loss": 0.21825896203517914,
+      "loss_ce": 0.03881560266017914,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 21660268,
+      "step": 240
+    },
+    {
+      "epoch": 1.1080459770114943,
+      "grad_norm": 3.800385893974569,
+      "learning_rate": 5e-06,
+      "loss": 0.2301,
+      "num_input_tokens_seen": 21750640,
+      "step": 241
+    },
+    {
+      "epoch": 1.1080459770114943,
+      "loss": 0.2575361132621765,
+      "loss_ce": 0.000211891092476435,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 21750640,
+      "step": 241
+    },
+    {
+      "epoch": 1.1126436781609195,
+      "grad_norm": 8.560187437534255,
+      "learning_rate": 5e-06,
+      "loss": 0.2571,
+      "num_input_tokens_seen": 21841068,
+      "step": 242
+    },
+    {
+      "epoch": 1.1126436781609195,
+      "loss": 0.2956390976905823,
+      "loss_ce": 0.0056000337935984135,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 21841068,
+      "step": 242
+    },
+    {
+      "epoch": 1.1172413793103448,
+      "grad_norm": 20.63729856829291,
+      "learning_rate": 5e-06,
+      "loss": 0.2309,
+      "num_input_tokens_seen": 21931304,
+      "step": 243
+    },
+    {
+      "epoch": 1.1172413793103448,
+      "loss": 0.21533158421516418,
+      "loss_ce": 0.00045730240526609123,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 21931304,
+      "step": 243
+    },
+    {
+      "epoch": 1.1218390804597702,
+      "grad_norm": 25.579280798369926,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 22021568,
+      "step": 244
+    },
+    {
+      "epoch": 1.1218390804597702,
+      "loss": 0.25474822521209717,
+      "loss_ce": 0.0018490497022867203,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 22021568,
+      "step": 244
+    },
+    {
+      "epoch": 1.1264367816091954,
+      "grad_norm": 32.89831549609003,
+      "learning_rate": 5e-06,
+      "loss": 0.2793,
+      "num_input_tokens_seen": 22111816,
+      "step": 245
+    },
+    {
+      "epoch": 1.1264367816091954,
+      "loss": 0.2946988344192505,
+      "loss_ce": 0.0027066715992987156,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 22111816,
+      "step": 245
+    },
+    {
+      "epoch": 1.1310344827586207,
+      "grad_norm": 11.404530774167966,
+      "learning_rate": 5e-06,
+      "loss": 0.2252,
+      "num_input_tokens_seen": 22202224,
+      "step": 246
+    },
+    {
+      "epoch": 1.1310344827586207,
+      "loss": 0.17997342348098755,
+      "loss_ce": 0.0004079932114109397,
+      "loss_iou": 0.5,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 22202224,
+      "step": 246
+    },
+    {
+      "epoch": 1.1356321839080459,
+      "grad_norm": 24.132647296050727,
+      "learning_rate": 5e-06,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 22292492,
+      "step": 247
+    },
+    {
+      "epoch": 1.1356321839080459,
+      "loss": 0.17757512629032135,
+      "loss_ce": 0.000573178636841476,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 22292492,
+      "step": 247
+    },
+    {
+      "epoch": 1.1402298850574712,
+      "grad_norm": 7.434676789770062,
+      "learning_rate": 5e-06,
+      "loss": 0.2615,
+      "num_input_tokens_seen": 22382928,
+      "step": 248
+    },
+    {
+      "epoch": 1.1402298850574712,
+      "loss": 0.27751120924949646,
+      "loss_ce": 0.010482416488230228,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 22382928,
+      "step": 248
+    },
+    {
+      "epoch": 1.1448275862068966,
+      "grad_norm": 10.379288916413433,
+      "learning_rate": 5e-06,
+      "loss": 0.2081,
+      "num_input_tokens_seen": 22473304,
+      "step": 249
+    },
+    {
+      "epoch": 1.1448275862068966,
+      "loss": 0.14225530624389648,
+      "loss_ce": 0.0012641068315133452,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 22473304,
+      "step": 249
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "grad_norm": 7.839282118755005,
+      "learning_rate": 5e-06,
+      "loss": 0.2665,
+      "num_input_tokens_seen": 22563700,
+      "step": 250
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "eval_seeclick_CIoU": 0.4888755977153778,
+      "eval_seeclick_GIoU": 0.47952449321746826,
+      "eval_seeclick_IoU": 0.5278463363647461,
+      "eval_seeclick_MAE_all": 0.056948138400912285,
+      "eval_seeclick_MAE_h": 0.04946732707321644,
+      "eval_seeclick_MAE_w": 0.09706718474626541,
+      "eval_seeclick_MAE_x_boxes": 0.09927782043814659,
+      "eval_seeclick_MAE_y_boxes": 0.046883879229426384,
+      "eval_seeclick_NUM_probability": 0.9999977946281433,
+      "eval_seeclick_inside_bbox": 0.8451704680919647,
+      "eval_seeclick_loss": 0.3242720365524292,
+      "eval_seeclick_loss_ce": 0.040716785937547684,
+      "eval_seeclick_loss_iou": 0.467529296875,
+      "eval_seeclick_loss_num": 0.0596923828125,
+      "eval_seeclick_loss_xval": 0.29864501953125,
+      "eval_seeclick_runtime": 75.8262,
+      "eval_seeclick_samples_per_second": 0.567,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 22563700,
+      "step": 250
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "eval_icons_CIoU": 0.5218705832958221,
+      "eval_icons_GIoU": 0.5266680121421814,
+      "eval_icons_IoU": 0.5595913529396057,
+      "eval_icons_MAE_all": 0.043722933158278465,
+      "eval_icons_MAE_h": 0.08452420309185982,
+      "eval_icons_MAE_w": 0.06406798399984837,
+      "eval_icons_MAE_x_boxes": 0.05667761527001858,
+      "eval_icons_MAE_y_boxes": 0.08344150707125664,
+      "eval_icons_NUM_probability": 0.9999994039535522,
+      "eval_icons_inside_bbox": 0.7239583432674408,
+      "eval_icons_loss": 0.2156449854373932,
+      "eval_icons_loss_ce": 4.674579088259634e-07,
+      "eval_icons_loss_iou": 0.43145751953125,
+      "eval_icons_loss_num": 0.04747772216796875,
+      "eval_icons_loss_xval": 0.237518310546875,
+      "eval_icons_runtime": 84.5354,
+      "eval_icons_samples_per_second": 0.591,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 22563700,
+      "step": 250
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "eval_screenspot_CIoU": 0.30689453333616257,
+      "eval_screenspot_GIoU": 0.2762495552500089,
+      "eval_screenspot_IoU": 0.3886234064896901,
+      "eval_screenspot_MAE_all": 0.10496337960163753,
+      "eval_screenspot_MAE_h": 0.11666570603847504,
+      "eval_screenspot_MAE_w": 0.208540049691995,
+      "eval_screenspot_MAE_x_boxes": 0.18961978455384573,
+      "eval_screenspot_MAE_y_boxes": 0.1133890226483345,
+      "eval_screenspot_NUM_probability": 0.9999755620956421,
+      "eval_screenspot_inside_bbox": 0.6433333357175192,
+      "eval_screenspot_loss": 0.5237547755241394,
+      "eval_screenspot_loss_ce": 0.0012614075288486977,
+      "eval_screenspot_loss_iou": 0.3485921223958333,
+      "eval_screenspot_loss_num": 0.10777791341145833,
+      "eval_screenspot_loss_xval": 0.5385335286458334,
+      "eval_screenspot_runtime": 158.8594,
+      "eval_screenspot_samples_per_second": 0.56,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 22563700,
+      "step": 250
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "eval_compot_CIoU": 0.44808267056941986,
+      "eval_compot_GIoU": 0.4227418601512909,
+      "eval_compot_IoU": 0.5081988573074341,
+      "eval_compot_MAE_all": 0.05795424245297909,
+      "eval_compot_MAE_h": 0.09416088834404945,
+      "eval_compot_MAE_w": 0.09847152419388294,
+      "eval_compot_MAE_x_boxes": 0.08727088011801243,
+      "eval_compot_MAE_y_boxes": 0.09673519805073738,
+      "eval_compot_NUM_probability": 0.9999892115592957,
+      "eval_compot_inside_bbox": 0.7604166567325592,
+      "eval_compot_loss": 0.32618048787117004,
+      "eval_compot_loss_ce": 0.013961461605504155,
+      "eval_compot_loss_iou": 0.4866943359375,
+      "eval_compot_loss_num": 0.0546875,
+      "eval_compot_loss_xval": 0.273345947265625,
+      "eval_compot_runtime": 88.276,
+      "eval_compot_samples_per_second": 0.566,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 22563700,
+      "step": 250
+    },
+    {
+      "epoch": 1.1494252873563218,
+      "loss": 0.19636714458465576,
+      "loss_ce": 0.007402303162962198,
+      "loss_iou": 0.546875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 22563700,
+      "step": 250
+    },
+    {
+      "epoch": 1.1540229885057471,
+      "grad_norm": 7.43698227348526,
+      "learning_rate": 5e-06,
+      "loss": 0.1928,
+      "num_input_tokens_seen": 22654036,
+      "step": 251
+    },
+    {
+      "epoch": 1.1540229885057471,
+      "loss": 0.19022439420223236,
+      "loss_ce": 0.00022195614292286336,
+      "loss_iou": 0.5,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 22654036,
+      "step": 251
+    },
+    {
+      "epoch": 1.1586206896551725,
+      "grad_norm": 7.724863758957877,
+      "learning_rate": 5e-06,
+      "loss": 0.235,
+      "num_input_tokens_seen": 22744428,
+      "step": 252
+    },
+    {
+      "epoch": 1.1586206896551725,
+      "loss": 0.28669023513793945,
+      "loss_ce": 0.0006184765952639282,
+      "loss_iou": 0.375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 22744428,
+      "step": 252
+    },
+    {
+      "epoch": 1.1632183908045977,
+      "grad_norm": 52.54527766863086,
+      "learning_rate": 5e-06,
+      "loss": 0.2173,
+      "num_input_tokens_seen": 22834692,
+      "step": 253
+    },
+    {
+      "epoch": 1.1632183908045977,
+      "loss": 0.1744190752506256,
+      "loss_ce": 0.0014149189228191972,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 22834692,
+      "step": 253
+    },
+    {
+      "epoch": 1.167816091954023,
+      "grad_norm": 4.696366017744483,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 22924280,
+      "step": 254
+    },
+    {
+      "epoch": 1.167816091954023,
+      "loss": 0.12924961745738983,
+      "loss_ce": 0.0024185676593333483,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 22924280,
+      "step": 254
+    },
+    {
+      "epoch": 1.1724137931034484,
+      "grad_norm": 6.684225949401091,
+      "learning_rate": 5e-06,
+      "loss": 0.1772,
+      "num_input_tokens_seen": 23014664,
+      "step": 255
+    },
+    {
+      "epoch": 1.1724137931034484,
+      "loss": 0.1868850439786911,
+      "loss_ce": 0.00048367734416387975,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 23014664,
+      "step": 255
+    },
+    {
+      "epoch": 1.1770114942528735,
+      "grad_norm": 10.115688964897386,
+      "learning_rate": 5e-06,
+      "loss": 0.2414,
+      "num_input_tokens_seen": 23104224,
+      "step": 256
+    },
+    {
+      "epoch": 1.1770114942528735,
+      "loss": 0.2677534222602844,
+      "loss_ce": 0.00048048628377728164,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 23104224,
+      "step": 256
+    },
+    {
+      "epoch": 1.181609195402299,
+      "grad_norm": 5.045379741734736,
+      "learning_rate": 5e-06,
+      "loss": 0.2122,
+      "num_input_tokens_seen": 23194388,
+      "step": 257
+    },
+    {
+      "epoch": 1.181609195402299,
+      "loss": 0.24545620381832123,
+      "loss_ce": 9.487089118920267e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 23194388,
+      "step": 257
+    },
+    {
+      "epoch": 1.186206896551724,
+      "grad_norm": 26.131469875864838,
+      "learning_rate": 5e-06,
+      "loss": 0.2115,
+      "num_input_tokens_seen": 23284736,
+      "step": 258
+    },
+    {
+      "epoch": 1.186206896551724,
+      "loss": 0.21708114445209503,
+      "loss_ce": 0.000528403848875314,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 23284736,
+      "step": 258
+    },
+    {
+      "epoch": 1.1908045977011494,
+      "grad_norm": 19.645438145213465,
+      "learning_rate": 5e-06,
+      "loss": 0.2313,
+      "num_input_tokens_seen": 23375184,
+      "step": 259
+    },
+    {
+      "epoch": 1.1908045977011494,
+      "loss": 0.268694132566452,
+      "loss_ce": 0.0001394439022988081,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 23375184,
+      "step": 259
+    },
+    {
+      "epoch": 1.1954022988505748,
+      "grad_norm": 16.21307669612568,
+      "learning_rate": 5e-06,
+      "loss": 0.2792,
+      "num_input_tokens_seen": 23465472,
+      "step": 260
+    },
+    {
+      "epoch": 1.1954022988505748,
+      "loss": 0.29360634088516235,
+      "loss_ce": 0.0001798336743377149,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 23465472,
+      "step": 260
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 12.81808028387753,
+      "learning_rate": 5e-06,
+      "loss": 0.2153,
+      "num_input_tokens_seen": 23555724,
+      "step": 261
+    },
+    {
+      "epoch": 1.2,
+      "loss": 0.23135723173618317,
+      "loss_ce": 0.009524943307042122,
+      "loss_iou": 0.53125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 23555724,
+      "step": 261
+    },
+    {
+      "epoch": 1.2045977011494253,
+      "grad_norm": 4.1999709843832616,
+      "learning_rate": 5e-06,
+      "loss": 0.2374,
+      "num_input_tokens_seen": 23645988,
+      "step": 262
+    },
+    {
+      "epoch": 1.2045977011494253,
+      "loss": 0.22801579535007477,
+      "loss_ce": 0.0014532884815707803,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 23645988,
+      "step": 262
+    },
+    {
+      "epoch": 1.2091954022988505,
+      "grad_norm": 6.414285711751042,
+      "learning_rate": 5e-06,
+      "loss": 0.1998,
+      "num_input_tokens_seen": 23736268,
+      "step": 263
+    },
+    {
+      "epoch": 1.2091954022988505,
+      "loss": 0.15405645966529846,
+      "loss_ce": 0.0007971944869495928,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 23736268,
+      "step": 263
+    },
+    {
+      "epoch": 1.2137931034482758,
+      "grad_norm": 4.480437290662561,
+      "learning_rate": 5e-06,
+      "loss": 0.2126,
+      "num_input_tokens_seen": 23826792,
+      "step": 264
+    },
+    {
+      "epoch": 1.2137931034482758,
+      "loss": 0.20262828469276428,
+      "loss_ce": 0.000479854759760201,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 23826792,
+      "step": 264
+    },
+    {
+      "epoch": 1.2183908045977012,
+      "grad_norm": 15.731375012073059,
+      "learning_rate": 5e-06,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 23916428,
+      "step": 265
+    },
+    {
+      "epoch": 1.2183908045977012,
+      "loss": 0.15348592400550842,
+      "loss_ce": 0.00010457600728841498,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 23916428,
+      "step": 265
+    },
+    {
+      "epoch": 1.2229885057471264,
+      "grad_norm": 7.176014460557662,
+      "learning_rate": 5e-06,
+      "loss": 0.304,
+      "num_input_tokens_seen": 24006752,
+      "step": 266
+    },
+    {
+      "epoch": 1.2229885057471264,
+      "loss": 0.3059394955635071,
+      "loss_ce": 0.00015336027718149126,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 24006752,
+      "step": 266
+    },
+    {
+      "epoch": 1.2275862068965517,
+      "grad_norm": 6.940435918405699,
+      "learning_rate": 5e-06,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 24097012,
+      "step": 267
+    },
+    {
+      "epoch": 1.2275862068965517,
+      "loss": 0.18653042614459991,
+      "loss_ce": 0.0003121576155535877,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 24097012,
+      "step": 267
+    },
+    {
+      "epoch": 1.232183908045977,
+      "grad_norm": 3.7469836271055623,
+      "learning_rate": 5e-06,
+      "loss": 0.3164,
+      "num_input_tokens_seen": 24187256,
+      "step": 268
+    },
+    {
+      "epoch": 1.232183908045977,
+      "loss": 0.27794623374938965,
+      "loss_ce": 0.00569892255589366,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 24187256,
+      "step": 268
+    },
+    {
+      "epoch": 1.2367816091954023,
+      "grad_norm": 7.413553829750902,
+      "learning_rate": 5e-06,
+      "loss": 0.2141,
+      "num_input_tokens_seen": 24277688,
+      "step": 269
+    },
+    {
+      "epoch": 1.2367816091954023,
+      "loss": 0.19044940173625946,
+      "loss_ce": 8.075028017628938e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 24277688,
+      "step": 269
+    },
+    {
+      "epoch": 1.2413793103448276,
+      "grad_norm": 7.193196745882768,
+      "learning_rate": 5e-06,
+      "loss": 0.2068,
+      "num_input_tokens_seen": 24368024,
+      "step": 270
+    },
+    {
+      "epoch": 1.2413793103448276,
+      "loss": 0.24729135632514954,
+      "loss_ce": 3.7946036172797903e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 24368024,
+      "step": 270
+    },
+    {
+      "epoch": 1.245977011494253,
+      "grad_norm": 23.36412717520471,
+      "learning_rate": 5e-06,
+      "loss": 0.1889,
+      "num_input_tokens_seen": 24458400,
+      "step": 271
+    },
+    {
+      "epoch": 1.245977011494253,
+      "loss": 0.18987199664115906,
+      "loss_ce": 0.00023576414969284087,
+      "loss_iou": 0.421875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 24458400,
+      "step": 271
+    },
+    {
+      "epoch": 1.2505747126436781,
+      "grad_norm": 6.27939746197536,
+      "learning_rate": 5e-06,
+      "loss": 0.1642,
+      "num_input_tokens_seen": 24548712,
+      "step": 272
+    },
+    {
+      "epoch": 1.2505747126436781,
+      "loss": 0.193972647190094,
+      "loss_ce": 0.0008268996607512236,
+      "loss_iou": 0.484375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 24548712,
+      "step": 272
+    },
+    {
+      "epoch": 1.2551724137931035,
+      "grad_norm": 25.45105413808657,
+      "learning_rate": 5e-06,
+      "loss": 0.2115,
+      "num_input_tokens_seen": 24639064,
+      "step": 273
+    },
+    {
+      "epoch": 1.2551724137931035,
+      "loss": 0.2186853289604187,
+      "loss_ce": 0.0026818893384188414,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 24639064,
+      "step": 273
+    },
+    {
+      "epoch": 1.2597701149425287,
+      "grad_norm": 37.080183589133725,
+      "learning_rate": 5e-06,
+      "loss": 0.1826,
+      "num_input_tokens_seen": 24729388,
+      "step": 274
+    },
+    {
+      "epoch": 1.2597701149425287,
+      "loss": 0.19386309385299683,
+      "loss_ce": 0.0006868370110169053,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 24729388,
+      "step": 274
+    },
+    {
+      "epoch": 1.264367816091954,
+      "grad_norm": 16.018753076692136,
+      "learning_rate": 5e-06,
+      "loss": 0.2302,
+      "num_input_tokens_seen": 24819812,
+      "step": 275
+    },
+    {
+      "epoch": 1.264367816091954,
+      "loss": 0.2251913845539093,
+      "loss_ce": 0.0006735554779879749,
+      "loss_iou": 0.5,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 24819812,
+      "step": 275
+    },
+    {
+      "epoch": 1.2689655172413792,
+      "grad_norm": 6.147405946334659,
+      "learning_rate": 5e-06,
+      "loss": 0.1915,
+      "num_input_tokens_seen": 24910200,
+      "step": 276
+    },
+    {
+      "epoch": 1.2689655172413792,
+      "loss": 0.14948949217796326,
+      "loss_ce": 0.00031957216560840607,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 24910200,
+      "step": 276
+    },
+    {
+      "epoch": 1.2735632183908046,
+      "grad_norm": 6.051687547447403,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 25000476,
+      "step": 277
+    },
+    {
+      "epoch": 1.2735632183908046,
+      "loss": 0.2223796248435974,
+      "loss_ce": 0.008512439206242561,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 25000476,
+      "step": 277
+    },
+    {
+      "epoch": 1.27816091954023,
+      "grad_norm": 8.095544751091358,
+      "learning_rate": 5e-06,
+      "loss": 0.2494,
+      "num_input_tokens_seen": 25090868,
+      "step": 278
+    },
+    {
+      "epoch": 1.27816091954023,
+      "loss": 0.21112553775310516,
+      "loss_ce": 0.00047794863348826766,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 25090868,
+      "step": 278
+    },
+    {
+      "epoch": 1.282758620689655,
+      "grad_norm": 29.745999578899646,
+      "learning_rate": 5e-06,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 25181160,
+      "step": 279
+    },
+    {
+      "epoch": 1.282758620689655,
+      "loss": 0.16160300374031067,
+      "loss_ce": 0.00013449997641146183,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 25181160,
+      "step": 279
+    },
+    {
+      "epoch": 1.2873563218390804,
+      "grad_norm": 9.770076744230986,
+      "learning_rate": 5e-06,
+      "loss": 0.2388,
+      "num_input_tokens_seen": 25270696,
+      "step": 280
+    },
+    {
+      "epoch": 1.2873563218390804,
+      "loss": 0.2805434465408325,
+      "loss_ce": 0.00020895421039313078,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 25270696,
+      "step": 280
+    },
+    {
+      "epoch": 1.2919540229885058,
+      "grad_norm": 6.866286661032349,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 25361132,
+      "step": 281
+    },
+    {
+      "epoch": 1.2919540229885058,
+      "loss": 0.21573258936405182,
+      "loss_ce": 9.537945152260363e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 25361132,
+      "step": 281
+    },
+    {
+      "epoch": 1.296551724137931,
+      "grad_norm": 14.632378384320322,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 25451508,
+      "step": 282
+    },
+    {
+      "epoch": 1.296551724137931,
+      "loss": 0.14539626240730286,
+      "loss_ce": 0.002390899695456028,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 25451508,
+      "step": 282
+    },
+    {
+      "epoch": 1.3011494252873563,
+      "grad_norm": 10.586676404677785,
+      "learning_rate": 5e-06,
+      "loss": 0.2193,
+      "num_input_tokens_seen": 25541988,
+      "step": 283
+    },
+    {
+      "epoch": 1.3011494252873563,
+      "loss": 0.18449945747852325,
+      "loss_ce": 0.0038964273408055305,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 25541988,
+      "step": 283
+    },
+    {
+      "epoch": 1.3057471264367817,
+      "grad_norm": 20.999614027385967,
+      "learning_rate": 5e-06,
+      "loss": 0.2048,
+      "num_input_tokens_seen": 25632416,
+      "step": 284
+    },
+    {
+      "epoch": 1.3057471264367817,
+      "loss": 0.2276066094636917,
+      "loss_ce": 0.00018961683963425457,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 25632416,
+      "step": 284
+    },
+    {
+      "epoch": 1.3103448275862069,
+      "grad_norm": 5.543128202063141,
+      "learning_rate": 5e-06,
+      "loss": 0.1622,
+      "num_input_tokens_seen": 25722744,
+      "step": 285
+    },
+    {
+      "epoch": 1.3103448275862069,
+      "loss": 0.16011224687099457,
+      "loss_ce": 0.0001390949182678014,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 25722744,
+      "step": 285
+    },
+    {
+      "epoch": 1.3149425287356322,
+      "grad_norm": 11.112725830839327,
+      "learning_rate": 5e-06,
+      "loss": 0.2203,
+      "num_input_tokens_seen": 25813124,
+      "step": 286
+    },
+    {
+      "epoch": 1.3149425287356322,
+      "loss": 0.3086729943752289,
+      "loss_ce": 0.0008421677048318088,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 25813124,
+      "step": 286
+    },
+    {
+      "epoch": 1.3195402298850576,
+      "grad_norm": 9.783123070255048,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 25903612,
+      "step": 287
+    },
+    {
+      "epoch": 1.3195402298850576,
+      "loss": 0.17264115810394287,
+      "loss_ce": 0.0005983194569125772,
+      "loss_iou": 0.40625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 25903612,
+      "step": 287
+    },
+    {
+      "epoch": 1.3241379310344827,
+      "grad_norm": 4.636122741394509,
+      "learning_rate": 5e-06,
+      "loss": 0.2552,
+      "num_input_tokens_seen": 25994072,
+      "step": 288
+    },
+    {
+      "epoch": 1.3241379310344827,
+      "loss": 0.22548282146453857,
+      "loss_ce": 0.00018680733046494424,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 25994072,
+      "step": 288
+    },
+    {
+      "epoch": 1.328735632183908,
+      "grad_norm": 3.861529643068919,
+      "learning_rate": 5e-06,
+      "loss": 0.1707,
+      "num_input_tokens_seen": 26084532,
+      "step": 289
+    },
+    {
+      "epoch": 1.328735632183908,
+      "loss": 0.16129527986049652,
+      "loss_ce": 0.00016246250015683472,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 26084532,
+      "step": 289
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 57.38016500276973,
+      "learning_rate": 5e-06,
+      "loss": 0.2326,
+      "num_input_tokens_seen": 26174984,
+      "step": 290
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "loss": 0.15951338410377502,
+      "loss_ce": 0.00039472500793635845,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 26174984,
+      "step": 290
+    },
+    {
+      "epoch": 1.3379310344827586,
+      "grad_norm": 8.111819603516635,
+      "learning_rate": 5e-06,
+      "loss": 0.1983,
+      "num_input_tokens_seen": 26265348,
+      "step": 291
+    },
+    {
+      "epoch": 1.3379310344827586,
+      "loss": 0.1832209974527359,
+      "loss_ce": 0.0004817442095372826,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 26265348,
+      "step": 291
+    },
+    {
+      "epoch": 1.3425287356321838,
+      "grad_norm": 18.16139378563598,
+      "learning_rate": 5e-06,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 26355688,
+      "step": 292
+    },
+    {
+      "epoch": 1.3425287356321838,
+      "loss": 0.17206496000289917,
+      "loss_ce": 6.788483005948365e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 26355688,
+      "step": 292
+    },
+    {
+      "epoch": 1.3471264367816091,
+      "grad_norm": 20.460245063855925,
+      "learning_rate": 5e-06,
+      "loss": 0.237,
+      "num_input_tokens_seen": 26446172,
+      "step": 293
+    },
+    {
+      "epoch": 1.3471264367816091,
+      "loss": 0.23456808924674988,
+      "loss_ce": 0.005442116409540176,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 26446172,
+      "step": 293
+    },
+    {
+      "epoch": 1.3517241379310345,
+      "grad_norm": 12.205499115713149,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 26536628,
+      "step": 294
+    },
+    {
+      "epoch": 1.3517241379310345,
+      "loss": 0.2167300283908844,
+      "loss_ce": 0.00017729138198774308,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 26536628,
+      "step": 294
+    },
+    {
+      "epoch": 1.3563218390804597,
+      "grad_norm": 7.426284095641791,
+      "learning_rate": 5e-06,
+      "loss": 0.1613,
+      "num_input_tokens_seen": 26627064,
+      "step": 295
+    },
+    {
+      "epoch": 1.3563218390804597,
+      "loss": 0.19740846753120422,
+      "loss_ce": 0.0002038878737948835,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 26627064,
+      "step": 295
+    },
+    {
+      "epoch": 1.360919540229885,
+      "grad_norm": 39.074194628927266,
+      "learning_rate": 5e-06,
+      "loss": 0.2518,
+      "num_input_tokens_seen": 26717480,
+      "step": 296
+    },
+    {
+      "epoch": 1.360919540229885,
+      "loss": 0.3336067795753479,
+      "loss_ce": 0.0002937709796242416,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 26717480,
+      "step": 296
+    },
+    {
+      "epoch": 1.3655172413793104,
+      "grad_norm": 20.276024886618437,
+      "learning_rate": 5e-06,
+      "loss": 0.209,
+      "num_input_tokens_seen": 26807844,
+      "step": 297
+    },
+    {
+      "epoch": 1.3655172413793104,
+      "loss": 0.3174262046813965,
+      "loss_ce": 0.0001654599909670651,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 26807844,
+      "step": 297
+    },
+    {
+      "epoch": 1.3701149425287356,
+      "grad_norm": 4.705748915237839,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 26898224,
+      "step": 298
+    },
+    {
+      "epoch": 1.3701149425287356,
+      "loss": 0.24118977785110474,
+      "loss_ce": 0.00010090330033563077,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 26898224,
+      "step": 298
+    },
+    {
+      "epoch": 1.374712643678161,
+      "grad_norm": 25.22660122284678,
+      "learning_rate": 5e-06,
+      "loss": 0.2173,
+      "num_input_tokens_seen": 26988588,
+      "step": 299
+    },
+    {
+      "epoch": 1.374712643678161,
+      "loss": 0.18158341944217682,
+      "loss_ce": 6.486591883003712e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 26988588,
+      "step": 299
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "grad_norm": 8.272745007832095,
+      "learning_rate": 5e-06,
+      "loss": 0.229,
+      "num_input_tokens_seen": 27078844,
+      "step": 300
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "loss": 0.19429120421409607,
+      "loss_ce": 0.00010784986079670489,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 27078844,
+      "step": 300
+    },
+    {
+      "epoch": 1.3839080459770114,
+      "grad_norm": 11.158840002018794,
+      "learning_rate": 5e-06,
+      "loss": 0.1863,
+      "num_input_tokens_seen": 27169132,
+      "step": 301
+    },
+    {
+      "epoch": 1.3839080459770114,
+      "loss": 0.24286043643951416,
+      "loss_ce": 6.258698704186827e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 27169132,
+      "step": 301
+    },
+    {
+      "epoch": 1.3885057471264368,
+      "grad_norm": 11.626395028754814,
+      "learning_rate": 5e-06,
+      "loss": 0.1938,
+      "num_input_tokens_seen": 27259360,
+      "step": 302
+    },
+    {
+      "epoch": 1.3885057471264368,
+      "loss": 0.22401869297027588,
+      "loss_ce": 0.0002638171426951885,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 27259360,
+      "step": 302
+    },
+    {
+      "epoch": 1.3931034482758622,
+      "grad_norm": 8.347578373209931,
+      "learning_rate": 5e-06,
+      "loss": 0.2332,
+      "num_input_tokens_seen": 27349756,
+      "step": 303
+    },
+    {
+      "epoch": 1.3931034482758622,
+      "loss": 0.24985189735889435,
+      "loss_ce": 0.00015708088176324964,
+      "loss_iou": 0.421875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 27349756,
+      "step": 303
+    },
+    {
+      "epoch": 1.3977011494252873,
+      "grad_norm": 9.382990132866077,
+      "learning_rate": 5e-06,
+      "loss": 0.2427,
+      "num_input_tokens_seen": 27440160,
+      "step": 304
+    },
+    {
+      "epoch": 1.3977011494252873,
+      "loss": 0.23031748831272125,
+      "loss_ce": 0.002442737342789769,
+      "loss_iou": 0.390625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 27440160,
+      "step": 304
+    },
+    {
+      "epoch": 1.4022988505747127,
+      "grad_norm": 13.016846770254826,
+      "learning_rate": 5e-06,
+      "loss": 0.2354,
+      "num_input_tokens_seen": 27530484,
+      "step": 305
+    },
+    {
+      "epoch": 1.4022988505747127,
+      "loss": 0.2382659912109375,
+      "loss_ce": 0.0012664890382438898,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 27530484,
+      "step": 305
+    },
+    {
+      "epoch": 1.4068965517241379,
+      "grad_norm": 13.820143768808174,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 27620888,
+      "step": 306
+    },
+    {
+      "epoch": 1.4068965517241379,
+      "loss": 0.1968221664428711,
+      "loss_ce": 0.00025843450566753745,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 27620888,
+      "step": 306
+    },
+    {
+      "epoch": 1.4114942528735632,
+      "grad_norm": 3.5475816109452896,
+      "learning_rate": 5e-06,
+      "loss": 0.1937,
+      "num_input_tokens_seen": 27711272,
+      "step": 307
+    },
+    {
+      "epoch": 1.4114942528735632,
+      "loss": 0.1471329778432846,
+      "loss_ce": 0.00022135992185212672,
+      "loss_iou": 0.46875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 27711272,
+      "step": 307
+    },
+    {
+      "epoch": 1.4160919540229884,
+      "grad_norm": 10.22489016272008,
+      "learning_rate": 5e-06,
+      "loss": 0.2147,
+      "num_input_tokens_seen": 27801544,
+      "step": 308
+    },
+    {
+      "epoch": 1.4160919540229884,
+      "loss": 0.19165240228176117,
+      "loss_ce": 0.008699517697095871,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 27801544,
+      "step": 308
+    },
+    {
+      "epoch": 1.4206896551724137,
+      "grad_norm": 9.134110088295508,
+      "learning_rate": 5e-06,
+      "loss": 0.2337,
+      "num_input_tokens_seen": 27891884,
+      "step": 309
+    },
+    {
+      "epoch": 1.4206896551724137,
+      "loss": 0.2654152512550354,
+      "loss_ce": 0.005771718919277191,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 27891884,
+      "step": 309
+    },
+    {
+      "epoch": 1.4252873563218391,
+      "grad_norm": 10.161788829631169,
+      "learning_rate": 5e-06,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 27982204,
+      "step": 310
+    },
+    {
+      "epoch": 1.4252873563218391,
+      "loss": 0.2308727651834488,
+      "loss_ce": 0.003730440977960825,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 27982204,
+      "step": 310
+    },
+    {
+      "epoch": 1.4298850574712643,
+      "grad_norm": 8.186672197144189,
+      "learning_rate": 5e-06,
+      "loss": 0.1798,
+      "num_input_tokens_seen": 28072556,
+      "step": 311
+    },
+    {
+      "epoch": 1.4298850574712643,
+      "loss": 0.19206635653972626,
+      "loss_ce": 0.0002023405977524817,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 28072556,
+      "step": 311
+    },
+    {
+      "epoch": 1.4344827586206896,
+      "grad_norm": 20.204462885292834,
+      "learning_rate": 5e-06,
+      "loss": 0.1739,
+      "num_input_tokens_seen": 28162972,
+      "step": 312
+    },
+    {
+      "epoch": 1.4344827586206896,
+      "loss": 0.15177220106124878,
+      "loss_ce": 9.983143536373973e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 28162972,
+      "step": 312
+    },
+    {
+      "epoch": 1.439080459770115,
+      "grad_norm": 10.349632633496206,
+      "learning_rate": 5e-06,
+      "loss": 0.2239,
+      "num_input_tokens_seen": 28252592,
+      "step": 313
+    },
+    {
+      "epoch": 1.439080459770115,
+      "loss": 0.23639342188835144,
+      "loss_ce": 0.0029949769377708435,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 28252592,
+      "step": 313
+    },
+    {
+      "epoch": 1.4436781609195402,
+      "grad_norm": 4.037411009142002,
+      "learning_rate": 5e-06,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 28342884,
+      "step": 314
+    },
+    {
+      "epoch": 1.4436781609195402,
+      "loss": 0.14021849632263184,
+      "loss_ce": 0.00011228647781535983,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 28342884,
+      "step": 314
+    },
+    {
+      "epoch": 1.4482758620689655,
+      "grad_norm": 10.393150532292086,
+      "learning_rate": 5e-06,
+      "loss": 0.118,
+      "num_input_tokens_seen": 28433376,
+      "step": 315
+    },
+    {
+      "epoch": 1.4482758620689655,
+      "loss": 0.12029355764389038,
+      "loss_ce": 0.00014585554890800267,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 28433376,
+      "step": 315
+    },
+    {
+      "epoch": 1.452873563218391,
+      "grad_norm": 3.824936350612978,
+      "learning_rate": 5e-06,
+      "loss": 0.2174,
+      "num_input_tokens_seen": 28523648,
+      "step": 316
+    },
+    {
+      "epoch": 1.452873563218391,
+      "loss": 0.24938887357711792,
+      "loss_ce": 0.0014030258171260357,
+      "loss_iou": 0.484375,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 28523648,
+      "step": 316
+    },
+    {
+      "epoch": 1.457471264367816,
+      "grad_norm": 11.988045018065144,
+      "learning_rate": 5e-06,
+      "loss": 0.1848,
+      "num_input_tokens_seen": 28613988,
+      "step": 317
+    },
+    {
+      "epoch": 1.457471264367816,
+      "loss": 0.17838840186595917,
+      "loss_ce": 0.009259981103241444,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 28613988,
+      "step": 317
+    },
+    {
+      "epoch": 1.4620689655172414,
+      "grad_norm": 4.034511511486509,
+      "learning_rate": 5e-06,
+      "loss": 0.1713,
+      "num_input_tokens_seen": 28704372,
+      "step": 318
+    },
+    {
+      "epoch": 1.4620689655172414,
+      "loss": 0.1520136594772339,
+      "loss_ce": 0.0004633643548004329,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 28704372,
+      "step": 318
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 15.615735006728443,
+      "learning_rate": 5e-06,
+      "loss": 0.2133,
+      "num_input_tokens_seen": 28794788,
+      "step": 319
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "loss": 0.23856469988822937,
+      "loss_ce": 0.00043603702215477824,
+      "loss_iou": 0.34375,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 28794788,
+      "step": 319
+    },
+    {
+      "epoch": 1.471264367816092,
+      "grad_norm": 22.79226873920817,
+      "learning_rate": 5e-06,
+      "loss": 0.2221,
+      "num_input_tokens_seen": 28885240,
+      "step": 320
+    },
+    {
+      "epoch": 1.471264367816092,
+      "loss": 0.1695161610841751,
+      "loss_ce": 5.20510075148195e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 28885240,
+      "step": 320
+    },
+    {
+      "epoch": 1.4758620689655173,
+      "grad_norm": 12.555178366315296,
+      "learning_rate": 5e-06,
+      "loss": 0.2411,
+      "num_input_tokens_seen": 28975712,
+      "step": 321
+    },
+    {
+      "epoch": 1.4758620689655173,
+      "loss": 0.24558743834495544,
+      "loss_ce": 0.00040922165499068797,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 28975712,
+      "step": 321
+    },
+    {
+      "epoch": 1.4804597701149425,
+      "grad_norm": 6.643673356109838,
+      "learning_rate": 5e-06,
+      "loss": 0.2199,
+      "num_input_tokens_seen": 29066048,
+      "step": 322
+    },
+    {
+      "epoch": 1.4804597701149425,
+      "loss": 0.18256893754005432,
+      "loss_ce": 0.001660746755078435,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 29066048,
+      "step": 322
+    },
+    {
+      "epoch": 1.4850574712643678,
+      "grad_norm": 5.317104474065933,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 29156544,
+      "step": 323
+    },
+    {
+      "epoch": 1.4850574712643678,
+      "loss": 0.19987276196479797,
+      "loss_ce": 0.00025728094624355435,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 29156544,
+      "step": 323
+    },
+    {
+      "epoch": 1.489655172413793,
+      "grad_norm": 12.208298681282333,
+      "learning_rate": 5e-06,
+      "loss": 0.1693,
+      "num_input_tokens_seen": 29246832,
+      "step": 324
+    },
+    {
+      "epoch": 1.489655172413793,
+      "loss": 0.15619094669818878,
+      "loss_ce": 3.249588917242363e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 29246832,
+      "step": 324
+    },
+    {
+      "epoch": 1.4942528735632183,
+      "grad_norm": 14.551980500080187,
+      "learning_rate": 5e-06,
+      "loss": 0.2611,
+      "num_input_tokens_seen": 29337088,
+      "step": 325
+    },
+    {
+      "epoch": 1.4942528735632183,
+      "loss": 0.348785400390625,
+      "loss_ce": 0.0024414421059191227,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 29337088,
+      "step": 325
+    },
+    {
+      "epoch": 1.4988505747126437,
+      "grad_norm": 11.379206303993785,
+      "learning_rate": 5e-06,
+      "loss": 0.1849,
+      "num_input_tokens_seen": 29427352,
+      "step": 326
+    },
+    {
+      "epoch": 1.4988505747126437,
+      "loss": 0.22499850392341614,
+      "loss_ce": 0.0005417080246843398,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 29427352,
+      "step": 326
+    },
+    {
+      "epoch": 1.5034482758620689,
+      "grad_norm": 90.0439810100679,
+      "learning_rate": 5e-06,
+      "loss": 0.2161,
+      "num_input_tokens_seen": 29517724,
+      "step": 327
+    },
+    {
+      "epoch": 1.5034482758620689,
+      "loss": 0.18729592859745026,
+      "loss_ce": 0.0005893875495530665,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 29517724,
+      "step": 327
+    },
+    {
+      "epoch": 1.5080459770114942,
+      "grad_norm": 19.33964708457954,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 29608140,
+      "step": 328
+    },
+    {
+      "epoch": 1.5080459770114942,
+      "loss": 0.24188189208507538,
+      "loss_ce": 0.0005641538882628083,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 29608140,
+      "step": 328
+    },
+    {
+      "epoch": 1.5126436781609196,
+      "grad_norm": 6.833812883588884,
+      "learning_rate": 5e-06,
+      "loss": 0.2319,
+      "num_input_tokens_seen": 29697696,
+      "step": 329
+    },
+    {
+      "epoch": 1.5126436781609196,
+      "loss": 0.2876734733581543,
+      "loss_ce": 0.0008998108096420765,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 29697696,
+      "step": 329
+    },
+    {
+      "epoch": 1.5172413793103448,
+      "grad_norm": 10.08968474441476,
+      "learning_rate": 5e-06,
+      "loss": 0.1793,
+      "num_input_tokens_seen": 29787960,
+      "step": 330
+    },
+    {
+      "epoch": 1.5172413793103448,
+      "loss": 0.15798088908195496,
+      "loss_ce": 0.00012871227227151394,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 29787960,
+      "step": 330
+    },
+    {
+      "epoch": 1.5218390804597701,
+      "grad_norm": 10.48777960826385,
+      "learning_rate": 5e-06,
+      "loss": 0.2303,
+      "num_input_tokens_seen": 29878288,
+      "step": 331
+    },
+    {
+      "epoch": 1.5218390804597701,
+      "loss": 0.27822285890579224,
+      "loss_ce": 0.00014670401287730783,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 29878288,
+      "step": 331
+    },
+    {
+      "epoch": 1.5264367816091955,
+      "grad_norm": 15.669568573881612,
+      "learning_rate": 5e-06,
+      "loss": 0.1935,
+      "num_input_tokens_seen": 29968616,
+      "step": 332
+    },
+    {
+      "epoch": 1.5264367816091955,
+      "loss": 0.1327584981918335,
+      "loss_ce": 0.00019014105782844126,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 29968616,
+      "step": 332
+    },
+    {
+      "epoch": 1.5310344827586206,
+      "grad_norm": 10.895228295518322,
+      "learning_rate": 5e-06,
+      "loss": 0.1781,
+      "num_input_tokens_seen": 30058964,
+      "step": 333
+    },
+    {
+      "epoch": 1.5310344827586206,
+      "loss": 0.13389872014522552,
+      "loss_ce": 0.00044535251799970865,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 30058964,
+      "step": 333
+    },
+    {
+      "epoch": 1.535632183908046,
+      "grad_norm": 12.555340714374447,
+      "learning_rate": 5e-06,
+      "loss": 0.2026,
+      "num_input_tokens_seen": 30149276,
+      "step": 334
+    },
+    {
+      "epoch": 1.535632183908046,
+      "loss": 0.22426463663578033,
+      "loss_ce": 0.0006928644143044949,
+      "loss_iou": 0.390625,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 30149276,
+      "step": 334
+    },
+    {
+      "epoch": 1.5402298850574714,
+      "grad_norm": 6.653750036819891,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 30239660,
+      "step": 335
+    },
+    {
+      "epoch": 1.5402298850574714,
+      "loss": 0.2286517322063446,
+      "loss_ce": 0.009657589718699455,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 30239660,
+      "step": 335
+    },
+    {
+      "epoch": 1.5448275862068965,
+      "grad_norm": 7.196935162615249,
+      "learning_rate": 5e-06,
+      "loss": 0.2105,
+      "num_input_tokens_seen": 30328584,
+      "step": 336
+    },
+    {
+      "epoch": 1.5448275862068965,
+      "loss": 0.2146240770816803,
+      "loss_ce": 0.00045170937664806843,
+      "loss_iou": 0.34375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 30328584,
+      "step": 336
+    },
+    {
+      "epoch": 1.5494252873563217,
+      "grad_norm": 15.439783012231938,
+      "learning_rate": 5e-06,
+      "loss": 0.2042,
+      "num_input_tokens_seen": 30418984,
+      "step": 337
+    },
+    {
+      "epoch": 1.5494252873563217,
+      "loss": 0.2664060592651367,
+      "loss_ce": 0.000842072709929198,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 30418984,
+      "step": 337
+    },
+    {
+      "epoch": 1.5540229885057473,
+      "grad_norm": 9.972362426319057,
+      "learning_rate": 5e-06,
+      "loss": 0.1914,
+      "num_input_tokens_seen": 30509544,
+      "step": 338
+    },
+    {
+      "epoch": 1.5540229885057473,
+      "loss": 0.20261165499687195,
+      "loss_ce": 0.0002190732047893107,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 30509544,
+      "step": 338
+    },
+    {
+      "epoch": 1.5586206896551724,
+      "grad_norm": 15.96705640498333,
+      "learning_rate": 5e-06,
+      "loss": 0.1748,
+      "num_input_tokens_seen": 30599924,
+      "step": 339
+    },
+    {
+      "epoch": 1.5586206896551724,
+      "loss": 0.14146339893341064,
+      "loss_ce": 0.0009604630176909268,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 30599924,
+      "step": 339
+    },
+    {
+      "epoch": 1.5632183908045976,
+      "grad_norm": 24.02833413826726,
+      "learning_rate": 5e-06,
+      "loss": 0.1581,
+      "num_input_tokens_seen": 30690300,
+      "step": 340
+    },
+    {
+      "epoch": 1.5632183908045976,
+      "loss": 0.14789816737174988,
+      "loss_ce": 0.0007423943607136607,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 30690300,
+      "step": 340
+    },
+    {
+      "epoch": 1.567816091954023,
+      "grad_norm": 11.132936405519033,
+      "learning_rate": 5e-06,
+      "loss": 0.1858,
+      "num_input_tokens_seen": 30780564,
+      "step": 341
+    },
+    {
+      "epoch": 1.567816091954023,
+      "loss": 0.18376675248146057,
+      "loss_ce": 0.00038663047598674893,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 30780564,
+      "step": 341
+    },
+    {
+      "epoch": 1.5724137931034483,
+      "grad_norm": 11.301299582007728,
+      "learning_rate": 5e-06,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 30870936,
+      "step": 342
+    },
+    {
+      "epoch": 1.5724137931034483,
+      "loss": 0.15917402505874634,
+      "loss_ce": 0.00011640776210697368,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 30870936,
+      "step": 342
+    },
+    {
+      "epoch": 1.5770114942528735,
+      "grad_norm": 4.3022352764349225,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 30961368,
+      "step": 343
+    },
+    {
+      "epoch": 1.5770114942528735,
+      "loss": 0.13939973711967468,
+      "loss_ce": 0.0004837117448914796,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 30961368,
+      "step": 343
+    },
+    {
+      "epoch": 1.5816091954022988,
+      "grad_norm": 21.124622330389396,
+      "learning_rate": 5e-06,
+      "loss": 0.1911,
+      "num_input_tokens_seen": 31051716,
+      "step": 344
+    },
+    {
+      "epoch": 1.5816091954022988,
+      "loss": 0.2080269455909729,
+      "loss_ce": 0.00011068060121033341,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 31051716,
+      "step": 344
+    },
+    {
+      "epoch": 1.5862068965517242,
+      "grad_norm": 6.603147217700077,
+      "learning_rate": 5e-06,
+      "loss": 0.2052,
+      "num_input_tokens_seen": 31142164,
+      "step": 345
+    },
+    {
+      "epoch": 1.5862068965517242,
+      "loss": 0.19146263599395752,
+      "loss_ce": 5.637338836095296e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 31142164,
+      "step": 345
+    },
+    {
+      "epoch": 1.5908045977011493,
+      "grad_norm": 18.413342940258104,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 31232568,
+      "step": 346
+    },
+    {
+      "epoch": 1.5908045977011493,
+      "loss": 0.21431368589401245,
+      "loss_ce": 8.028043521335348e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 31232568,
+      "step": 346
+    },
+    {
+      "epoch": 1.5954022988505747,
+      "grad_norm": 3.750494996396292,
+      "learning_rate": 5e-06,
+      "loss": 0.2055,
+      "num_input_tokens_seen": 31322940,
+      "step": 347
+    },
+    {
+      "epoch": 1.5954022988505747,
+      "loss": 0.2030225545167923,
+      "loss_ce": 0.00014168729830998927,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 31322940,
+      "step": 347
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 19.608744946070097,
+      "learning_rate": 5e-06,
+      "loss": 0.2093,
+      "num_input_tokens_seen": 31412484,
+      "step": 348
+    },
+    {
+      "epoch": 1.6,
+      "loss": 0.21284618973731995,
+      "loss_ce": 0.00016918416076805443,
+      "loss_iou": 0.4375,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 31412484,
+      "step": 348
+    },
+    {
+      "epoch": 1.6045977011494252,
+      "grad_norm": 16.926454810756088,
+      "learning_rate": 5e-06,
+      "loss": 0.1911,
+      "num_input_tokens_seen": 31502796,
+      "step": 349
+    },
+    {
+      "epoch": 1.6045977011494252,
+      "loss": 0.236657053232193,
+      "loss_ce": 0.002068430185317993,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 31502796,
+      "step": 349
+    },
+    {
+      "epoch": 1.6091954022988506,
+      "grad_norm": 8.440280678490312,
+      "learning_rate": 5e-06,
+      "loss": 0.2148,
+      "num_input_tokens_seen": 31593140,
+      "step": 350
+    },
+    {
+      "epoch": 1.6091954022988506,
+      "loss": 0.1520470678806305,
+      "loss_ce": 3.901964009855874e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 31593140,
+      "step": 350
+    },
+    {
+      "epoch": 1.613793103448276,
+      "grad_norm": 12.531015324402237,
+      "learning_rate": 5e-06,
+      "loss": 0.2053,
+      "num_input_tokens_seen": 31683560,
+      "step": 351
+    },
+    {
+      "epoch": 1.613793103448276,
+      "loss": 0.28503406047821045,
+      "loss_ce": 0.00012196854368085042,
+      "loss_iou": 0.515625,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 31683560,
+      "step": 351
+    },
+    {
+      "epoch": 1.6183908045977011,
+      "grad_norm": 13.56199610824294,
+      "learning_rate": 5e-06,
+      "loss": 0.2573,
+      "num_input_tokens_seen": 31774020,
+      "step": 352
+    },
+    {
+      "epoch": 1.6183908045977011,
+      "loss": 0.2630327045917511,
+      "loss_ce": 0.0005510285845957696,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 31774020,
+      "step": 352
+    },
+    {
+      "epoch": 1.6229885057471263,
+      "grad_norm": 16.539672672228185,
+      "learning_rate": 5e-06,
+      "loss": 0.218,
+      "num_input_tokens_seen": 31864284,
+      "step": 353
+    },
+    {
+      "epoch": 1.6229885057471263,
+      "loss": 0.2166557013988495,
+      "loss_ce": 0.004497497342526913,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 31864284,
+      "step": 353
+    },
+    {
+      "epoch": 1.6275862068965519,
+      "grad_norm": 6.7039057056478555,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 31954736,
+      "step": 354
+    },
+    {
+      "epoch": 1.6275862068965519,
+      "loss": 0.12781822681427002,
+      "loss_ce": 0.00013267630129121244,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 31954736,
+      "step": 354
+    },
+    {
+      "epoch": 1.632183908045977,
+      "grad_norm": 24.22876257944078,
+      "learning_rate": 5e-06,
+      "loss": 0.1899,
+      "num_input_tokens_seen": 32045092,
+      "step": 355
+    },
+    {
+      "epoch": 1.632183908045977,
+      "loss": 0.21440809965133667,
+      "loss_ce": 5.264465289656073e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 32045092,
+      "step": 355
+    },
+    {
+      "epoch": 1.6367816091954022,
+      "grad_norm": 6.350667267782482,
+      "learning_rate": 5e-06,
+      "loss": 0.1784,
+      "num_input_tokens_seen": 32135436,
+      "step": 356
+    },
+    {
+      "epoch": 1.6367816091954022,
+      "loss": 0.20722705125808716,
+      "loss_ce": 0.005749986041337252,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 32135436,
+      "step": 356
+    },
+    {
+      "epoch": 1.6413793103448275,
+      "grad_norm": 24.037632779184456,
+      "learning_rate": 5e-06,
+      "loss": 0.2292,
+      "num_input_tokens_seen": 32225760,
+      "step": 357
+    },
+    {
+      "epoch": 1.6413793103448275,
+      "loss": 0.1975146234035492,
+      "loss_ce": 0.0004473668523132801,
+      "loss_iou": 0.546875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 32225760,
+      "step": 357
+    },
+    {
+      "epoch": 1.645977011494253,
+      "grad_norm": 10.97286648357755,
+      "learning_rate": 5e-06,
+      "loss": 0.2958,
+      "num_input_tokens_seen": 32314628,
+      "step": 358
+    },
+    {
+      "epoch": 1.645977011494253,
+      "loss": 0.30159226059913635,
+      "loss_ce": 7.8596654930152e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 32314628,
+      "step": 358
+    },
+    {
+      "epoch": 1.650574712643678,
+      "grad_norm": 14.188317923001321,
+      "learning_rate": 5e-06,
+      "loss": 0.1884,
+      "num_input_tokens_seen": 32405028,
+      "step": 359
+    },
+    {
+      "epoch": 1.650574712643678,
+      "loss": 0.1701948344707489,
+      "loss_ce": 0.0005018344381824136,
+      "loss_iou": 0.46875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 32405028,
+      "step": 359
+    },
+    {
+      "epoch": 1.6551724137931034,
+      "grad_norm": 15.866397059455883,
+      "learning_rate": 5e-06,
+      "loss": 0.2404,
+      "num_input_tokens_seen": 32495460,
+      "step": 360
+    },
+    {
+      "epoch": 1.6551724137931034,
+      "loss": 0.21389362215995789,
+      "loss_ce": 0.00789997074753046,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 32495460,
+      "step": 360
+    },
+    {
+      "epoch": 1.6597701149425288,
+      "grad_norm": 22.63781592648334,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 32585812,
+      "step": 361
+    },
+    {
+      "epoch": 1.6597701149425288,
+      "loss": 0.1410699486732483,
+      "loss_ce": 0.0007196052465587854,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 32585812,
+      "step": 361
+    },
+    {
+      "epoch": 1.664367816091954,
+      "grad_norm": 24.104025272898923,
+      "learning_rate": 5e-06,
+      "loss": 0.1888,
+      "num_input_tokens_seen": 32676080,
+      "step": 362
+    },
+    {
+      "epoch": 1.664367816091954,
+      "loss": 0.18728001415729523,
+      "loss_ce": 2.41522429860197e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 32676080,
+      "step": 362
+    },
+    {
+      "epoch": 1.6689655172413793,
+      "grad_norm": 4.251795484895364,
+      "learning_rate": 5e-06,
+      "loss": 0.2331,
+      "num_input_tokens_seen": 32766524,
+      "step": 363
+    },
+    {
+      "epoch": 1.6689655172413793,
+      "loss": 0.26753073930740356,
+      "loss_ce": 0.00013572408352047205,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 32766524,
+      "step": 363
+    },
+    {
+      "epoch": 1.6735632183908047,
+      "grad_norm": 16.06020227141594,
+      "learning_rate": 5e-06,
+      "loss": 0.2075,
+      "num_input_tokens_seen": 32856880,
+      "step": 364
+    },
+    {
+      "epoch": 1.6735632183908047,
+      "loss": 0.20815755426883698,
+      "loss_ce": 8.871030149748549e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 32856880,
+      "step": 364
+    },
+    {
+      "epoch": 1.6781609195402298,
+      "grad_norm": 4.038583152340203,
+      "learning_rate": 5e-06,
+      "loss": 0.199,
+      "num_input_tokens_seen": 32947228,
+      "step": 365
+    },
+    {
+      "epoch": 1.6781609195402298,
+      "loss": 0.1920887678861618,
+      "loss_ce": 0.00013320505968295038,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 32947228,
+      "step": 365
+    },
+    {
+      "epoch": 1.6827586206896552,
+      "grad_norm": 10.849275070130034,
+      "learning_rate": 5e-06,
+      "loss": 0.2501,
+      "num_input_tokens_seen": 33037680,
+      "step": 366
+    },
+    {
+      "epoch": 1.6827586206896552,
+      "loss": 0.28570684790611267,
+      "loss_ce": 0.00018438987899571657,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 33037680,
+      "step": 366
+    },
+    {
+      "epoch": 1.6873563218390806,
+      "grad_norm": 15.71736414697125,
+      "learning_rate": 5e-06,
+      "loss": 0.2526,
+      "num_input_tokens_seen": 33127980,
+      "step": 367
+    },
+    {
+      "epoch": 1.6873563218390806,
+      "loss": 0.19320005178451538,
+      "loss_ce": 0.00048154895193874836,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 33127980,
+      "step": 367
+    },
+    {
+      "epoch": 1.6919540229885057,
+      "grad_norm": 11.7136055813051,
+      "learning_rate": 5e-06,
+      "loss": 0.214,
+      "num_input_tokens_seen": 33218308,
+      "step": 368
+    },
+    {
+      "epoch": 1.6919540229885057,
+      "loss": 0.21528667211532593,
+      "loss_ce": 0.003677786560729146,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 33218308,
+      "step": 368
+    },
+    {
+      "epoch": 1.6965517241379309,
+      "grad_norm": 9.03276554803375,
+      "learning_rate": 5e-06,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 33308680,
+      "step": 369
+    },
+    {
+      "epoch": 1.6965517241379309,
+      "loss": 0.20362484455108643,
+      "loss_ce": 0.00010311185906175524,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 33308680,
+      "step": 369
+    },
+    {
+      "epoch": 1.7011494252873565,
+      "grad_norm": 4.661137477640627,
+      "learning_rate": 5e-06,
+      "loss": 0.2119,
+      "num_input_tokens_seen": 33399264,
+      "step": 370
+    },
+    {
+      "epoch": 1.7011494252873565,
+      "loss": 0.13792553544044495,
+      "loss_ce": 0.00016917982429731637,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 33399264,
+      "step": 370
+    },
+    {
+      "epoch": 1.7057471264367816,
+      "grad_norm": 3.2251449715057787,
+      "learning_rate": 5e-06,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 33489564,
+      "step": 371
+    },
+    {
+      "epoch": 1.7057471264367816,
+      "loss": 0.14613063633441925,
+      "loss_ce": 0.00024134977138601243,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 33489564,
+      "step": 371
+    },
+    {
+      "epoch": 1.7103448275862068,
+      "grad_norm": 8.964171049100557,
+      "learning_rate": 5e-06,
+      "loss": 0.2087,
+      "num_input_tokens_seen": 33579112,
+      "step": 372
+    },
+    {
+      "epoch": 1.7103448275862068,
+      "loss": 0.22618785500526428,
+      "loss_ce": 0.006644395180046558,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 33579112,
+      "step": 372
+    },
+    {
+      "epoch": 1.7149425287356321,
+      "grad_norm": 5.141600835893525,
+      "learning_rate": 5e-06,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 33669472,
+      "step": 373
+    },
+    {
+      "epoch": 1.7149425287356321,
+      "loss": 0.24420268833637238,
+      "loss_ce": 0.0002451670588925481,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 33669472,
+      "step": 373
+    },
+    {
+      "epoch": 1.7195402298850575,
+      "grad_norm": 8.029154993860924,
+      "learning_rate": 5e-06,
+      "loss": 0.2167,
+      "num_input_tokens_seen": 33759812,
+      "step": 374
+    },
+    {
+      "epoch": 1.7195402298850575,
+      "loss": 0.21689020097255707,
+      "loss_ce": 0.005037169903516769,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 33759812,
+      "step": 374
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 8.857319520624422,
+      "learning_rate": 5e-06,
+      "loss": 0.2094,
+      "num_input_tokens_seen": 33848696,
+      "step": 375
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "loss": 0.24398654699325562,
+      "loss_ce": 5.9540048823691905e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 33848696,
+      "step": 375
+    },
+    {
+      "epoch": 1.728735632183908,
+      "grad_norm": 29.84409998795184,
+      "learning_rate": 5e-06,
+      "loss": 0.1964,
+      "num_input_tokens_seen": 33939052,
+      "step": 376
+    },
+    {
+      "epoch": 1.728735632183908,
+      "loss": 0.2298891395330429,
+      "loss_ce": 3.074748383369297e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 33939052,
+      "step": 376
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 16.908297420855593,
+      "learning_rate": 5e-06,
+      "loss": 0.1799,
+      "num_input_tokens_seen": 34029428,
+      "step": 377
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "loss": 0.17225658893585205,
+      "loss_ce": 0.00032055945484898984,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 34029428,
+      "step": 377
+    },
+    {
+      "epoch": 1.7379310344827585,
+      "grad_norm": 5.924418877591673,
+      "learning_rate": 5e-06,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 34119072,
+      "step": 378
+    },
+    {
+      "epoch": 1.7379310344827585,
+      "loss": 0.1289907693862915,
+      "loss_ce": 0.00023710176174063236,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 34119072,
+      "step": 378
+    },
+    {
+      "epoch": 1.742528735632184,
+      "grad_norm": 4.126805999853441,
+      "learning_rate": 5e-06,
+      "loss": 0.216,
+      "num_input_tokens_seen": 34209448,
+      "step": 379
+    },
+    {
+      "epoch": 1.742528735632184,
+      "loss": 0.21730422973632812,
+      "loss_ce": 0.00011062939302064478,
+      "loss_iou": 0.5625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 34209448,
+      "step": 379
+    },
+    {
+      "epoch": 1.7471264367816093,
+      "grad_norm": 6.583943856738861,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 34299756,
+      "step": 380
+    },
+    {
+      "epoch": 1.7471264367816093,
+      "loss": 0.23842403292655945,
+      "loss_ce": 8.174288086593151e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 34299756,
+      "step": 380
+    },
+    {
+      "epoch": 1.7517241379310344,
+      "grad_norm": 6.121020487954506,
+      "learning_rate": 5e-06,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 34390068,
+      "step": 381
+    },
+    {
+      "epoch": 1.7517241379310344,
+      "loss": 0.2400810569524765,
+      "loss_ce": 0.00012134698044974357,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 34390068,
+      "step": 381
+    },
+    {
+      "epoch": 1.7563218390804598,
+      "grad_norm": 14.275086476453184,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 34480520,
+      "step": 382
+    },
+    {
+      "epoch": 1.7563218390804598,
+      "loss": 0.08874941617250443,
+      "loss_ce": 9.585011866874993e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 34480520,
+      "step": 382
+    },
+    {
+      "epoch": 1.7609195402298852,
+      "grad_norm": 10.514767544788414,
+      "learning_rate": 5e-06,
+      "loss": 0.1745,
+      "num_input_tokens_seen": 34570948,
+      "step": 383
+    },
+    {
+      "epoch": 1.7609195402298852,
+      "loss": 0.1786903589963913,
+      "loss_ce": 0.001032268744893372,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 34570948,
+      "step": 383
+    },
+    {
+      "epoch": 1.7655172413793103,
+      "grad_norm": 61.44054500522531,
+      "learning_rate": 5e-06,
+      "loss": 0.1866,
+      "num_input_tokens_seen": 34661360,
+      "step": 384
+    },
+    {
+      "epoch": 1.7655172413793103,
+      "loss": 0.2136577069759369,
+      "loss_ce": 0.0018352140905335546,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 34661360,
+      "step": 384
+    },
+    {
+      "epoch": 1.7701149425287355,
+      "grad_norm": 20.341892049512285,
+      "learning_rate": 5e-06,
+      "loss": 0.2173,
+      "num_input_tokens_seen": 34751736,
+      "step": 385
+    },
+    {
+      "epoch": 1.7701149425287355,
+      "loss": 0.29879891872406006,
+      "loss_ce": 0.00033700710628181696,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 34751736,
+      "step": 385
+    },
+    {
+      "epoch": 1.774712643678161,
+      "grad_norm": 10.713533548556878,
+      "learning_rate": 5e-06,
+      "loss": 0.2004,
+      "num_input_tokens_seen": 34841944,
+      "step": 386
+    },
+    {
+      "epoch": 1.774712643678161,
+      "loss": 0.2071501910686493,
+      "loss_ce": 0.00011893494956893846,
+      "loss_iou": 0.421875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 34841944,
+      "step": 386
+    },
+    {
+      "epoch": 1.7793103448275862,
+      "grad_norm": 6.168349303523133,
+      "learning_rate": 5e-06,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 34932320,
+      "step": 387
+    },
+    {
+      "epoch": 1.7793103448275862,
+      "loss": 0.18104185163974762,
+      "loss_ce": 7.260293932631612e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 34932320,
+      "step": 387
+    },
+    {
+      "epoch": 1.7839080459770114,
+      "grad_norm": 10.622510358830738,
+      "learning_rate": 5e-06,
+      "loss": 0.2128,
+      "num_input_tokens_seen": 35022772,
+      "step": 388
+    },
+    {
+      "epoch": 1.7839080459770114,
+      "loss": 0.18907198309898376,
+      "loss_ce": 0.00041230578790418804,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 35022772,
+      "step": 388
+    },
+    {
+      "epoch": 1.7885057471264367,
+      "grad_norm": 11.951897075599353,
+      "learning_rate": 5e-06,
+      "loss": 0.1647,
+      "num_input_tokens_seen": 35113260,
+      "step": 389
+    },
+    {
+      "epoch": 1.7885057471264367,
+      "loss": 0.16620348393917084,
+      "loss_ce": 0.0003404413000680506,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 35113260,
+      "step": 389
+    },
+    {
+      "epoch": 1.793103448275862,
+      "grad_norm": 4.828903526280168,
+      "learning_rate": 5e-06,
+      "loss": 0.2304,
+      "num_input_tokens_seen": 35203688,
+      "step": 390
+    },
+    {
+      "epoch": 1.793103448275862,
+      "loss": 0.29129573702812195,
+      "loss_ce": 0.0006463380996137857,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 35203688,
+      "step": 390
+    },
+    {
+      "epoch": 1.7977011494252872,
+      "grad_norm": 6.198908160148479,
+      "learning_rate": 5e-06,
+      "loss": 0.1866,
+      "num_input_tokens_seen": 35294020,
+      "step": 391
+    },
+    {
+      "epoch": 1.7977011494252872,
+      "loss": 0.18983140587806702,
+      "loss_ce": 0.00046984368236735463,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 35294020,
+      "step": 391
+    },
+    {
+      "epoch": 1.8022988505747126,
+      "grad_norm": 11.14648078385914,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 35384488,
+      "step": 392
+    },
+    {
+      "epoch": 1.8022988505747126,
+      "loss": 0.1650119572877884,
+      "loss_ce": 0.0010715241078287363,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 35384488,
+      "step": 392
+    },
+    {
+      "epoch": 1.806896551724138,
+      "grad_norm": 18.559559905574613,
+      "learning_rate": 5e-06,
+      "loss": 0.1779,
+      "num_input_tokens_seen": 35474708,
+      "step": 393
+    },
+    {
+      "epoch": 1.806896551724138,
+      "loss": 0.19601336121559143,
+      "loss_ce": 0.000517760228831321,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 35474708,
+      "step": 393
+    },
+    {
+      "epoch": 1.8114942528735631,
+      "grad_norm": 5.146339340993819,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 35564984,
+      "step": 394
+    },
+    {
+      "epoch": 1.8114942528735631,
+      "loss": 0.1250748336315155,
+      "loss_ce": 9.00992818060331e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 35564984,
+      "step": 394
+    },
+    {
+      "epoch": 1.8160919540229885,
+      "grad_norm": 13.805918822000008,
+      "learning_rate": 5e-06,
+      "loss": 0.2097,
+      "num_input_tokens_seen": 35655304,
+      "step": 395
+    },
+    {
+      "epoch": 1.8160919540229885,
+      "loss": 0.1994381844997406,
+      "loss_ce": 6.685496191494167e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 35655304,
+      "step": 395
+    },
+    {
+      "epoch": 1.8206896551724139,
+      "grad_norm": 17.32882624388584,
+      "learning_rate": 5e-06,
+      "loss": 0.2631,
+      "num_input_tokens_seen": 35744700,
+      "step": 396
+    },
+    {
+      "epoch": 1.8206896551724139,
+      "loss": 0.3162878155708313,
+      "loss_ce": 0.006351289339363575,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 35744700,
+      "step": 396
+    },
+    {
+      "epoch": 1.825287356321839,
+      "grad_norm": 26.732334449078184,
+      "learning_rate": 5e-06,
+      "loss": 0.2161,
+      "num_input_tokens_seen": 35835144,
+      "step": 397
+    },
+    {
+      "epoch": 1.825287356321839,
+      "loss": 0.2433795928955078,
+      "loss_ce": 0.0004596640937961638,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 35835144,
+      "step": 397
+    },
+    {
+      "epoch": 1.8298850574712644,
+      "grad_norm": 5.47457435472519,
+      "learning_rate": 5e-06,
+      "loss": 0.1831,
+      "num_input_tokens_seen": 35925388,
+      "step": 398
+    },
+    {
+      "epoch": 1.8298850574712644,
+      "loss": 0.1931520402431488,
+      "loss_ce": 0.00018938624998554587,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 35925388,
+      "step": 398
+    },
+    {
+      "epoch": 1.8344827586206898,
+      "grad_norm": 6.840227842822447,
+      "learning_rate": 5e-06,
+      "loss": 0.1699,
+      "num_input_tokens_seen": 36015004,
+      "step": 399
+    },
+    {
+      "epoch": 1.8344827586206898,
+      "loss": 0.142633855342865,
+      "loss_ce": 2.5207998987752944e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 36015004,
+      "step": 399
+    },
+    {
+      "epoch": 1.839080459770115,
+      "grad_norm": 8.746845766431045,
+      "learning_rate": 5e-06,
+      "loss": 0.1852,
+      "num_input_tokens_seen": 36105580,
+      "step": 400
+    },
+    {
+      "epoch": 1.839080459770115,
+      "loss": 0.15033434331417084,
+      "loss_ce": 0.000493036350235343,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 36105580,
+      "step": 400
+    },
+    {
+      "epoch": 1.84367816091954,
+      "grad_norm": 12.05362200433303,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 36196032,
+      "step": 401
+    },
+    {
+      "epoch": 1.84367816091954,
+      "loss": 0.10613197833299637,
+      "loss_ce": 0.00014442511019296944,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 36196032,
+      "step": 401
+    },
+    {
+      "epoch": 1.8482758620689657,
+      "grad_norm": 12.871135990489064,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 36286396,
+      "step": 402
+    },
+    {
+      "epoch": 1.8482758620689657,
+      "loss": 0.1379774510860443,
+      "loss_ce": 3.800613194471225e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 36286396,
+      "step": 402
+    },
+    {
+      "epoch": 1.8528735632183908,
+      "grad_norm": 14.529023016496748,
+      "learning_rate": 5e-06,
+      "loss": 0.193,
+      "num_input_tokens_seen": 36376740,
+      "step": 403
+    },
+    {
+      "epoch": 1.8528735632183908,
+      "loss": 0.23211072385311127,
+      "loss_ce": 0.003534059040248394,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 36376740,
+      "step": 403
+    },
+    {
+      "epoch": 1.857471264367816,
+      "grad_norm": 21.924185342651707,
+      "learning_rate": 5e-06,
+      "loss": 0.1735,
+      "num_input_tokens_seen": 36466984,
+      "step": 404
+    },
+    {
+      "epoch": 1.857471264367816,
+      "loss": 0.15595406293869019,
+      "loss_ce": 0.00013130568549968302,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 36466984,
+      "step": 404
+    },
+    {
+      "epoch": 1.8620689655172413,
+      "grad_norm": 13.574410791488862,
+      "learning_rate": 5e-06,
+      "loss": 0.1739,
+      "num_input_tokens_seen": 36557356,
+      "step": 405
+    },
+    {
+      "epoch": 1.8620689655172413,
+      "loss": 0.17355313897132874,
+      "loss_ce": 3.0175322535797022e-05,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 36557356,
+      "step": 405
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 7.0386023660360335,
+      "learning_rate": 5e-06,
+      "loss": 0.2389,
+      "num_input_tokens_seen": 36647736,
+      "step": 406
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "loss": 0.25896725058555603,
+      "loss_ce": 0.0009106049546971917,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 36647736,
+      "step": 406
+    },
+    {
+      "epoch": 1.8712643678160918,
+      "grad_norm": 22.160625349053383,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 36738104,
+      "step": 407
+    },
+    {
+      "epoch": 1.8712643678160918,
+      "loss": 0.1394781917333603,
+      "loss_ce": 7.390179962385446e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 36738104,
+      "step": 407
+    },
+    {
+      "epoch": 1.8758620689655172,
+      "grad_norm": 17.35116305455773,
+      "learning_rate": 5e-06,
+      "loss": 0.2099,
+      "num_input_tokens_seen": 36828500,
+      "step": 408
+    },
+    {
+      "epoch": 1.8758620689655172,
+      "loss": 0.1852053552865982,
+      "loss_ce": 0.00032986659789457917,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 36828500,
+      "step": 408
+    },
+    {
+      "epoch": 1.8804597701149426,
+      "grad_norm": 28.24772029390849,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "num_input_tokens_seen": 36918796,
+      "step": 409
+    },
+    {
+      "epoch": 1.8804597701149426,
+      "loss": 0.2876579761505127,
+      "loss_ce": 0.00016715031233616173,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 36918796,
+      "step": 409
+    },
+    {
+      "epoch": 1.8850574712643677,
+      "grad_norm": 11.025181667958632,
+      "learning_rate": 5e-06,
+      "loss": 0.1974,
+      "num_input_tokens_seen": 37008328,
+      "step": 410
+    },
+    {
+      "epoch": 1.8850574712643677,
+      "loss": 0.19846147298812866,
+      "loss_ce": 0.0004329003859311342,
+      "loss_iou": 0.40625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 37008328,
+      "step": 410
+    },
+    {
+      "epoch": 1.889655172413793,
+      "grad_norm": 9.446092449238963,
+      "learning_rate": 5e-06,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 37098728,
+      "step": 411
+    },
+    {
+      "epoch": 1.889655172413793,
+      "loss": 0.2067907452583313,
+      "loss_ce": 0.00021725523401983082,
+      "loss_iou": 0.453125,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 37098728,
+      "step": 411
+    },
+    {
+      "epoch": 1.8942528735632185,
+      "grad_norm": 5.183189456998249,
+      "learning_rate": 5e-06,
+      "loss": 0.159,
+      "num_input_tokens_seen": 37189052,
+      "step": 412
+    },
+    {
+      "epoch": 1.8942528735632185,
+      "loss": 0.13279348611831665,
+      "loss_ce": 4.201898264000192e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 37189052,
+      "step": 412
+    },
+    {
+      "epoch": 1.8988505747126436,
+      "grad_norm": 9.650000103120663,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 37279516,
+      "step": 413
+    },
+    {
+      "epoch": 1.8988505747126436,
+      "loss": 0.16115236282348633,
+      "loss_ce": 8.058699313551188e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 37279516,
+      "step": 413
+    },
+    {
+      "epoch": 1.903448275862069,
+      "grad_norm": 4.0534506430610495,
+      "learning_rate": 5e-06,
+      "loss": 0.2039,
+      "num_input_tokens_seen": 37369960,
+      "step": 414
+    },
+    {
+      "epoch": 1.903448275862069,
+      "loss": 0.19549965858459473,
+      "loss_ce": 0.0015299279475584626,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 37369960,
+      "step": 414
+    },
+    {
+      "epoch": 1.9080459770114944,
+      "grad_norm": 10.593561048071034,
+      "learning_rate": 5e-06,
+      "loss": 0.1738,
+      "num_input_tokens_seen": 37460424,
+      "step": 415
+    },
+    {
+      "epoch": 1.9080459770114944,
+      "loss": 0.16938260197639465,
+      "loss_ce": 0.00025418028235435486,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 37460424,
+      "step": 415
+    },
+    {
+      "epoch": 1.9126436781609195,
+      "grad_norm": 2.9328584393134003,
+      "learning_rate": 5e-06,
+      "loss": 0.1799,
+      "num_input_tokens_seen": 37550816,
+      "step": 416
+    },
+    {
+      "epoch": 1.9126436781609195,
+      "loss": 0.17176763713359833,
+      "loss_ce": 4.522378003457561e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 37550816,
+      "step": 416
+    },
+    {
+      "epoch": 1.9172413793103447,
+      "grad_norm": 6.966385105625737,
+      "learning_rate": 5e-06,
+      "loss": 0.1852,
+      "num_input_tokens_seen": 37640308,
+      "step": 417
+    },
+    {
+      "epoch": 1.9172413793103447,
+      "loss": 0.18929776549339294,
+      "loss_ce": 8.878001244738698e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 37640308,
+      "step": 417
+    },
+    {
+      "epoch": 1.9218390804597703,
+      "grad_norm": 28.585591957284574,
+      "learning_rate": 5e-06,
+      "loss": 0.1666,
+      "num_input_tokens_seen": 37730788,
+      "step": 418
+    },
+    {
+      "epoch": 1.9218390804597703,
+      "loss": 0.20673710107803345,
+      "loss_ce": 0.005260062403976917,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 37730788,
+      "step": 418
+    },
+    {
+      "epoch": 1.9264367816091954,
+      "grad_norm": 14.331276777730162,
+      "learning_rate": 5e-06,
+      "loss": 0.1901,
+      "num_input_tokens_seen": 37821240,
+      "step": 419
+    },
+    {
+      "epoch": 1.9264367816091954,
+      "loss": 0.17201045155525208,
+      "loss_ce": 0.0006237286143004894,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 37821240,
+      "step": 419
+    },
+    {
+      "epoch": 1.9310344827586206,
+      "grad_norm": 51.49279529680463,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 37911548,
+      "step": 420
+    },
+    {
+      "epoch": 1.9310344827586206,
+      "loss": 0.10174712538719177,
+      "loss_ce": 0.0005813572788611054,
+      "loss_iou": 0.5,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 37911548,
+      "step": 420
+    },
+    {
+      "epoch": 1.935632183908046,
+      "grad_norm": 20.70402147732612,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 38001920,
+      "step": 421
+    },
+    {
+      "epoch": 1.935632183908046,
+      "loss": 0.1458110809326172,
+      "loss_ce": 0.00012014327512588352,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 38001920,
+      "step": 421
+    },
+    {
+      "epoch": 1.9402298850574713,
+      "grad_norm": 15.948122093541322,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "num_input_tokens_seen": 38092224,
+      "step": 422
+    },
+    {
+      "epoch": 1.9402298850574713,
+      "loss": 0.23001374304294586,
+      "loss_ce": 0.00039948339690454304,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 38092224,
+      "step": 422
+    },
+    {
+      "epoch": 1.9448275862068964,
+      "grad_norm": 4.692813126881015,
+      "learning_rate": 5e-06,
+      "loss": 0.19,
+      "num_input_tokens_seen": 38181780,
+      "step": 423
+    },
+    {
+      "epoch": 1.9448275862068964,
+      "loss": 0.2313835322856903,
+      "loss_ce": 0.00016710199997760355,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 38181780,
+      "step": 423
+    },
+    {
+      "epoch": 1.9494252873563218,
+      "grad_norm": 10.416976166382188,
+      "learning_rate": 5e-06,
+      "loss": 0.1808,
+      "num_input_tokens_seen": 38272044,
+      "step": 424
+    },
+    {
+      "epoch": 1.9494252873563218,
+      "loss": 0.20144467055797577,
+      "loss_ce": 0.0002117574622388929,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 38272044,
+      "step": 424
+    },
+    {
+      "epoch": 1.9540229885057472,
+      "grad_norm": 19.032893916715878,
+      "learning_rate": 5e-06,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 38362384,
+      "step": 425
+    },
+    {
+      "epoch": 1.9540229885057472,
+      "loss": 0.14751462638378143,
+      "loss_ce": 0.00011472392361611128,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 38362384,
+      "step": 425
+    },
+    {
+      "epoch": 1.9586206896551723,
+      "grad_norm": 4.116116262537347,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 38452876,
+      "step": 426
+    },
+    {
+      "epoch": 1.9586206896551723,
+      "loss": 0.20857056975364685,
+      "loss_ce": 0.00031861409661360085,
+      "loss_iou": 0.375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 38452876,
+      "step": 426
+    },
+    {
+      "epoch": 1.9632183908045977,
+      "grad_norm": 12.769621257575608,
+      "learning_rate": 5e-06,
+      "loss": 0.2186,
+      "num_input_tokens_seen": 38543180,
+      "step": 427
+    },
+    {
+      "epoch": 1.9632183908045977,
+      "loss": 0.21470102667808533,
+      "loss_ce": 0.0008338369661942124,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 38543180,
+      "step": 427
+    },
+    {
+      "epoch": 1.967816091954023,
+      "grad_norm": 4.963161851563267,
+      "learning_rate": 5e-06,
+      "loss": 0.1961,
+      "num_input_tokens_seen": 38633460,
+      "step": 428
+    },
+    {
+      "epoch": 1.967816091954023,
+      "loss": 0.26360833644866943,
+      "loss_ce": 0.00018057512352243066,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 38633460,
+      "step": 428
+    },
+    {
+      "epoch": 1.9724137931034482,
+      "grad_norm": 9.041750000087218,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 38723828,
+      "step": 429
+    },
+    {
+      "epoch": 1.9724137931034482,
+      "loss": 0.12721416354179382,
+      "loss_ce": 7.794749399181455e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 38723828,
+      "step": 429
+    },
+    {
+      "epoch": 1.9770114942528736,
+      "grad_norm": 4.689071794501053,
+      "learning_rate": 5e-06,
+      "loss": 0.1893,
+      "num_input_tokens_seen": 38814248,
+      "step": 430
+    },
+    {
+      "epoch": 1.9770114942528736,
+      "loss": 0.1737414002418518,
+      "loss_ce": 0.0002489687467459589,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 38814248,
+      "step": 430
+    },
+    {
+      "epoch": 1.981609195402299,
+      "grad_norm": 27.154301658955358,
+      "learning_rate": 5e-06,
+      "loss": 0.1834,
+      "num_input_tokens_seen": 38904564,
+      "step": 431
+    },
+    {
+      "epoch": 1.981609195402299,
+      "loss": 0.1887126863002777,
+      "loss_ce": 5.3021052735857666e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 38904564,
+      "step": 431
+    },
+    {
+      "epoch": 1.986206896551724,
+      "grad_norm": 15.753748795396127,
+      "learning_rate": 5e-06,
+      "loss": 0.2578,
+      "num_input_tokens_seen": 38993440,
+      "step": 432
+    },
+    {
+      "epoch": 1.986206896551724,
+      "loss": 0.17184701561927795,
+      "loss_ce": 0.00027720603975467384,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 38993440,
+      "step": 432
+    },
+    {
+      "epoch": 1.9908045977011493,
+      "grad_norm": 16.878556629349514,
+      "learning_rate": 5e-06,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 39083816,
+      "step": 433
+    },
+    {
+      "epoch": 1.9908045977011493,
+      "loss": 0.18325263261795044,
+      "loss_ce": 2.5101442588493228e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 39083816,
+      "step": 433
+    },
+    {
+      "epoch": 1.9954022988505749,
+      "grad_norm": 12.650814063184473,
+      "learning_rate": 5e-06,
+      "loss": 0.1683,
+      "num_input_tokens_seen": 39174208,
+      "step": 434
+    },
+    {
+      "epoch": 1.9954022988505749,
+      "loss": 0.2204468846321106,
+      "loss_ce": 0.000476180954137817,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 39174208,
+      "step": 434
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 10.19888475335077,
+      "learning_rate": 5e-06,
+      "loss": 0.2245,
+      "num_input_tokens_seen": 39264536,
+      "step": 435
+    },
+    {
+      "epoch": 2.0,
+      "loss": 0.16308581829071045,
+      "loss_ce": 0.00015246294788084924,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 39264536,
+      "step": 435
+    },
+    {
+      "epoch": 2.004597701149425,
+      "grad_norm": 5.446086415432793,
+      "learning_rate": 5e-06,
+      "loss": 0.1492,
+      "num_input_tokens_seen": 39355008,
+      "step": 436
+    },
+    {
+      "epoch": 2.004597701149425,
+      "loss": 0.11079922318458557,
+      "loss_ce": 0.0006002452573738992,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 39355008,
+      "step": 436
+    },
+    {
+      "epoch": 2.0091954022988507,
+      "grad_norm": 11.525128525925364,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 39445364,
+      "step": 437
+    },
+    {
+      "epoch": 2.0091954022988507,
+      "loss": 0.1240125373005867,
+      "loss_ce": 0.0001722091546980664,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 39445364,
+      "step": 437
+    },
+    {
+      "epoch": 2.013793103448276,
+      "grad_norm": 5.953661546303923,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 39535692,
+      "step": 438
+    },
+    {
+      "epoch": 2.013793103448276,
+      "loss": 0.14717864990234375,
+      "loss_ce": 5.339417839422822e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 39535692,
+      "step": 438
+    },
+    {
+      "epoch": 2.018390804597701,
+      "grad_norm": 9.622272916635373,
+      "learning_rate": 5e-06,
+      "loss": 0.2175,
+      "num_input_tokens_seen": 39625276,
+      "step": 439
+    },
+    {
+      "epoch": 2.018390804597701,
+      "loss": 0.19843432307243347,
+      "loss_ce": 0.00040575824095867574,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 39625276,
+      "step": 439
+    },
+    {
+      "epoch": 2.0229885057471266,
+      "grad_norm": 11.722688812201046,
+      "learning_rate": 5e-06,
+      "loss": 0.1435,
+      "num_input_tokens_seen": 39715664,
+      "step": 440
+    },
+    {
+      "epoch": 2.0229885057471266,
+      "loss": 0.13606533408164978,
+      "loss_ce": 1.7956510419026017e-05,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 39715664,
+      "step": 440
+    },
+    {
+      "epoch": 2.027586206896552,
+      "grad_norm": 9.734047142024144,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 39806120,
+      "step": 441
+    },
+    {
+      "epoch": 2.027586206896552,
+      "loss": 0.15544164180755615,
+      "loss_ce": 0.00019872028497047722,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 39806120,
+      "step": 441
+    },
+    {
+      "epoch": 2.032183908045977,
+      "grad_norm": 4.317712136329391,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 39896512,
+      "step": 442
+    },
+    {
+      "epoch": 2.032183908045977,
+      "loss": 0.19860993325710297,
+      "loss_ce": 0.004579176660627127,
+      "loss_iou": 0.390625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 39896512,
+      "step": 442
+    },
+    {
+      "epoch": 2.036781609195402,
+      "grad_norm": 20.253389515904008,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 39987032,
+      "step": 443
+    },
+    {
+      "epoch": 2.036781609195402,
+      "loss": 0.19383695721626282,
+      "loss_ce": 0.0006912024109624326,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 39987032,
+      "step": 443
+    },
+    {
+      "epoch": 2.0413793103448277,
+      "grad_norm": 4.97212014387009,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 40077476,
+      "step": 444
+    },
+    {
+      "epoch": 2.0413793103448277,
+      "loss": 0.12523597478866577,
+      "loss_ce": 0.00011389970313757658,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 40077476,
+      "step": 444
+    },
+    {
+      "epoch": 2.045977011494253,
+      "grad_norm": 20.311552823682124,
+      "learning_rate": 5e-06,
+      "loss": 0.1806,
+      "num_input_tokens_seen": 40168036,
+      "step": 445
+    },
+    {
+      "epoch": 2.045977011494253,
+      "loss": 0.18241354823112488,
+      "loss_ce": 0.00022360245930030942,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 40168036,
+      "step": 445
+    },
+    {
+      "epoch": 2.050574712643678,
+      "grad_norm": 20.574323382128274,
+      "learning_rate": 5e-06,
+      "loss": 0.2583,
+      "num_input_tokens_seen": 40258332,
+      "step": 446
+    },
+    {
+      "epoch": 2.050574712643678,
+      "loss": 0.2653703987598419,
+      "loss_ce": 0.002186804311349988,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 40258332,
+      "step": 446
+    },
+    {
+      "epoch": 2.0551724137931036,
+      "grad_norm": 7.1556843698299,
+      "learning_rate": 5e-06,
+      "loss": 0.2328,
+      "num_input_tokens_seen": 40348728,
+      "step": 447
+    },
+    {
+      "epoch": 2.0551724137931036,
+      "loss": 0.2722781300544739,
+      "loss_ce": 0.00015285555855371058,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 40348728,
+      "step": 447
+    },
+    {
+      "epoch": 2.0597701149425287,
+      "grad_norm": 2.328436671702478,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 40439056,
+      "step": 448
+    },
+    {
+      "epoch": 2.0597701149425287,
+      "loss": 0.1083475798368454,
+      "loss_ce": 2.544172093621455e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 40439056,
+      "step": 448
+    },
+    {
+      "epoch": 2.064367816091954,
+      "grad_norm": 7.50649292907234,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 40529452,
+      "step": 449
+    },
+    {
+      "epoch": 2.064367816091954,
+      "loss": 0.08954399824142456,
+      "loss_ce": 6.646315159741789e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 40529452,
+      "step": 449
+    },
+    {
+      "epoch": 2.0689655172413794,
+      "grad_norm": 10.363487196000653,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 40619852,
+      "step": 450
+    },
+    {
+      "epoch": 2.0689655172413794,
+      "loss": 0.19379128515720367,
+      "loss_ce": 3.5169647162547335e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 40619852,
+      "step": 450
+    },
+    {
+      "epoch": 2.0735632183908046,
+      "grad_norm": 8.83788070087072,
+      "learning_rate": 5e-06,
+      "loss": 0.1502,
+      "num_input_tokens_seen": 40710184,
+      "step": 451
+    },
+    {
+      "epoch": 2.0735632183908046,
+      "loss": 0.1824960559606552,
+      "loss_ce": 3.146529707009904e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 40710184,
+      "step": 451
+    },
+    {
+      "epoch": 2.0781609195402297,
+      "grad_norm": 22.605606537183156,
+      "learning_rate": 5e-06,
+      "loss": 0.2029,
+      "num_input_tokens_seen": 40799700,
+      "step": 452
+    },
+    {
+      "epoch": 2.0781609195402297,
+      "loss": 0.24570101499557495,
+      "loss_ce": 0.00021761911921203136,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 40799700,
+      "step": 452
+    },
+    {
+      "epoch": 2.0827586206896553,
+      "grad_norm": 8.39747671515735,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 40890160,
+      "step": 453
+    },
+    {
+      "epoch": 2.0827586206896553,
+      "loss": 0.11327225714921951,
+      "loss_ce": 0.00023515010252594948,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 40890160,
+      "step": 453
+    },
+    {
+      "epoch": 2.0873563218390805,
+      "grad_norm": 12.581825524675565,
+      "learning_rate": 5e-06,
+      "loss": 0.161,
+      "num_input_tokens_seen": 40980460,
+      "step": 454
+    },
+    {
+      "epoch": 2.0873563218390805,
+      "loss": 0.19595982134342194,
+      "loss_ce": 0.0002200792368967086,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 40980460,
+      "step": 454
+    },
+    {
+      "epoch": 2.0919540229885056,
+      "grad_norm": 4.810948555506714,
+      "learning_rate": 5e-06,
+      "loss": 0.1872,
+      "num_input_tokens_seen": 41070800,
+      "step": 455
+    },
+    {
+      "epoch": 2.0919540229885056,
+      "loss": 0.195834219455719,
+      "loss_ce": 0.0006132656708359718,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 41070800,
+      "step": 455
+    },
+    {
+      "epoch": 2.0965517241379312,
+      "grad_norm": 16.611126353595154,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 41161136,
+      "step": 456
+    },
+    {
+      "epoch": 2.0965517241379312,
+      "loss": 0.18046724796295166,
+      "loss_ce": 4.7336572606582195e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 41161136,
+      "step": 456
+    },
+    {
+      "epoch": 2.1011494252873564,
+      "grad_norm": 4.0448675422540346,
+      "learning_rate": 5e-06,
+      "loss": 0.1711,
+      "num_input_tokens_seen": 41251520,
+      "step": 457
+    },
+    {
+      "epoch": 2.1011494252873564,
+      "loss": 0.17239469289779663,
+      "loss_ce": 0.0003060669405385852,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 41251520,
+      "step": 457
+    },
+    {
+      "epoch": 2.1057471264367815,
+      "grad_norm": 6.408813044691335,
+      "learning_rate": 5e-06,
+      "loss": 0.1917,
+      "num_input_tokens_seen": 41341836,
+      "step": 458
+    },
+    {
+      "epoch": 2.1057471264367815,
+      "loss": 0.22308608889579773,
+      "loss_ce": 0.034731604158878326,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 41341836,
+      "step": 458
+    },
+    {
+      "epoch": 2.110344827586207,
+      "grad_norm": 13.658825889797921,
+      "learning_rate": 5e-06,
+      "loss": 0.2127,
+      "num_input_tokens_seen": 41432320,
+      "step": 459
+    },
+    {
+      "epoch": 2.110344827586207,
+      "loss": 0.27129611372947693,
+      "loss_ce": 5.588199564954266e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 41432320,
+      "step": 459
+    },
+    {
+      "epoch": 2.1149425287356323,
+      "grad_norm": 48.461654169692885,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 41522668,
+      "step": 460
+    },
+    {
+      "epoch": 2.1149425287356323,
+      "loss": 0.15583747625350952,
+      "loss_ce": 0.0002588740608189255,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 41522668,
+      "step": 460
+    },
+    {
+      "epoch": 2.1195402298850574,
+      "grad_norm": 20.517893200082028,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 41613056,
+      "step": 461
+    },
+    {
+      "epoch": 2.1195402298850574,
+      "loss": 0.16513849794864655,
+      "loss_ce": 0.00016046430391725153,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 41613056,
+      "step": 461
+    },
+    {
+      "epoch": 2.1241379310344826,
+      "grad_norm": 5.0252606370642505,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 41703492,
+      "step": 462
+    },
+    {
+      "epoch": 2.1241379310344826,
+      "loss": 0.16197888553142548,
+      "loss_ce": 0.00014416445628739893,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 41703492,
+      "step": 462
+    },
+    {
+      "epoch": 2.128735632183908,
+      "grad_norm": 35.99843113408324,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 41792988,
+      "step": 463
+    },
+    {
+      "epoch": 2.128735632183908,
+      "loss": 0.11322343349456787,
+      "loss_ce": 3.226540684408974e-06,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 41792988,
+      "step": 463
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 19.56124215547498,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 41883320,
+      "step": 464
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "loss": 0.10433776676654816,
+      "loss_ce": 1.3423067684925627e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 41883320,
+      "step": 464
+    },
+    {
+      "epoch": 2.1379310344827585,
+      "grad_norm": 5.283946883842529,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 41972864,
+      "step": 465
+    },
+    {
+      "epoch": 2.1379310344827585,
+      "loss": 0.18815144896507263,
+      "loss_ce": 0.00022419335437007248,
+      "loss_iou": 0.5,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 41972864,
+      "step": 465
+    },
+    {
+      "epoch": 2.142528735632184,
+      "grad_norm": 12.855612844319221,
+      "learning_rate": 5e-06,
+      "loss": 0.1953,
+      "num_input_tokens_seen": 42063268,
+      "step": 466
+    },
+    {
+      "epoch": 2.142528735632184,
+      "loss": 0.1868751049041748,
+      "loss_ce": 4.649303446058184e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 42063268,
+      "step": 466
+    },
+    {
+      "epoch": 2.147126436781609,
+      "grad_norm": 9.976883452360386,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 42152812,
+      "step": 467
+    },
+    {
+      "epoch": 2.147126436781609,
+      "loss": 0.18522751331329346,
+      "loss_ce": 0.0004130652523599565,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 42152812,
+      "step": 467
+    },
+    {
+      "epoch": 2.1517241379310343,
+      "grad_norm": 16.988301795489004,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 42243232,
+      "step": 468
+    },
+    {
+      "epoch": 2.1517241379310343,
+      "loss": 0.16275331377983093,
+      "loss_ce": 0.0045196665450930595,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 42243232,
+      "step": 468
+    },
+    {
+      "epoch": 2.15632183908046,
+      "grad_norm": 18.341882617271445,
+      "learning_rate": 5e-06,
+      "loss": 0.1953,
+      "num_input_tokens_seen": 42333576,
+      "step": 469
+    },
+    {
+      "epoch": 2.15632183908046,
+      "loss": 0.2384186089038849,
+      "loss_ce": 0.0006256342749111354,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 42333576,
+      "step": 469
+    },
+    {
+      "epoch": 2.160919540229885,
+      "grad_norm": 23.406338359813535,
+      "learning_rate": 5e-06,
+      "loss": 0.1333,
+      "num_input_tokens_seen": 42423888,
+      "step": 470
+    },
+    {
+      "epoch": 2.160919540229885,
+      "loss": 0.09170129895210266,
+      "loss_ce": 1.1235326383030042e-05,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 42423888,
+      "step": 470
+    },
+    {
+      "epoch": 2.1655172413793102,
+      "grad_norm": 5.704299050854984,
+      "learning_rate": 5e-06,
+      "loss": 0.1674,
+      "num_input_tokens_seen": 42512692,
+      "step": 471
+    },
+    {
+      "epoch": 2.1655172413793102,
+      "loss": 0.1880715936422348,
+      "loss_ce": 5.279548349790275e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 42512692,
+      "step": 471
+    },
+    {
+      "epoch": 2.170114942528736,
+      "grad_norm": 7.3341829258742095,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 42602996,
+      "step": 472
+    },
+    {
+      "epoch": 2.170114942528736,
+      "loss": 0.13333070278167725,
+      "loss_ce": 0.00010622564150253311,
+      "loss_iou": 0.4375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 42602996,
+      "step": 472
+    },
+    {
+      "epoch": 2.174712643678161,
+      "grad_norm": 7.267255134108421,
+      "learning_rate": 5e-06,
+      "loss": 0.2403,
+      "num_input_tokens_seen": 42693276,
+      "step": 473
+    },
+    {
+      "epoch": 2.174712643678161,
+      "loss": 0.16120409965515137,
+      "loss_ce": 4.077212361153215e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 42693276,
+      "step": 473
+    },
+    {
+      "epoch": 2.179310344827586,
+      "grad_norm": 128.39041769732546,
+      "learning_rate": 5e-06,
+      "loss": 0.2092,
+      "num_input_tokens_seen": 42782804,
+      "step": 474
+    },
+    {
+      "epoch": 2.179310344827586,
+      "loss": 0.18164223432540894,
+      "loss_ce": 3.213401942048222e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 42782804,
+      "step": 474
+    },
+    {
+      "epoch": 2.1839080459770113,
+      "grad_norm": 21.574370414672206,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 42873080,
+      "step": 475
+    },
+    {
+      "epoch": 2.1839080459770113,
+      "loss": 0.14893415570259094,
+      "loss_ce": 8.381912266486324e-06,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 42873080,
+      "step": 475
+    },
+    {
+      "epoch": 2.188505747126437,
+      "grad_norm": 4.41738665455178,
+      "learning_rate": 5e-06,
+      "loss": 0.2296,
+      "num_input_tokens_seen": 42962612,
+      "step": 476
+    },
+    {
+      "epoch": 2.188505747126437,
+      "loss": 0.2648059129714966,
+      "loss_ce": 0.001317158224992454,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 42962612,
+      "step": 476
+    },
+    {
+      "epoch": 2.193103448275862,
+      "grad_norm": 6.382558470057758,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 43052192,
+      "step": 477
+    },
+    {
+      "epoch": 2.193103448275862,
+      "loss": 0.18961870670318604,
+      "loss_ce": 4.350763629190624e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 43052192,
+      "step": 477
+    },
+    {
+      "epoch": 2.197701149425287,
+      "grad_norm": 9.913316189631466,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 43142436,
+      "step": 478
+    },
+    {
+      "epoch": 2.197701149425287,
+      "loss": 0.20609432458877563,
+      "loss_ce": 0.0001006659513222985,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 43142436,
+      "step": 478
+    },
+    {
+      "epoch": 2.2022988505747128,
+      "grad_norm": 3.649045492362426,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 43232740,
+      "step": 479
+    },
+    {
+      "epoch": 2.2022988505747128,
+      "loss": 0.15660002827644348,
+      "loss_ce": 0.00016692971985321492,
+      "loss_iou": 0.34375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 43232740,
+      "step": 479
+    },
+    {
+      "epoch": 2.206896551724138,
+      "grad_norm": 16.113831960073153,
+      "learning_rate": 5e-06,
+      "loss": 0.1596,
+      "num_input_tokens_seen": 43323116,
+      "step": 480
+    },
+    {
+      "epoch": 2.206896551724138,
+      "loss": 0.14228999614715576,
+      "loss_ce": 0.00010859394387807697,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 43323116,
+      "step": 480
+    },
+    {
+      "epoch": 2.211494252873563,
+      "grad_norm": 13.400014157011977,
+      "learning_rate": 5e-06,
+      "loss": 0.2149,
+      "num_input_tokens_seen": 43413536,
+      "step": 481
+    },
+    {
+      "epoch": 2.211494252873563,
+      "loss": 0.181840181350708,
+      "loss_ce": 0.00013852809206582606,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 43413536,
+      "step": 481
+    },
+    {
+      "epoch": 2.2160919540229886,
+      "grad_norm": 4.019097019730536,
+      "learning_rate": 5e-06,
+      "loss": 0.1358,
+      "num_input_tokens_seen": 43503912,
+      "step": 482
+    },
+    {
+      "epoch": 2.2160919540229886,
+      "loss": 0.11794019490480423,
+      "loss_ce": 2.0275372662581503e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 43503912,
+      "step": 482
+    },
+    {
+      "epoch": 2.220689655172414,
+      "grad_norm": 10.618191683361344,
+      "learning_rate": 5e-06,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 43594288,
+      "step": 483
+    },
+    {
+      "epoch": 2.220689655172414,
+      "loss": 0.11793670058250427,
+      "loss_ce": 9.307506843470037e-05,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 43594288,
+      "step": 483
+    },
+    {
+      "epoch": 2.225287356321839,
+      "grad_norm": 3.8979874019779968,
+      "learning_rate": 5e-06,
+      "loss": 0.1525,
+      "num_input_tokens_seen": 43684704,
+      "step": 484
+    },
+    {
+      "epoch": 2.225287356321839,
+      "loss": 0.16006925702095032,
+      "loss_ce": 0.0007064603269100189,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 43684704,
+      "step": 484
+    },
+    {
+      "epoch": 2.2298850574712645,
+      "grad_norm": 6.611052035871228,
+      "learning_rate": 5e-06,
+      "loss": 0.19,
+      "num_input_tokens_seen": 43775036,
+      "step": 485
+    },
+    {
+      "epoch": 2.2298850574712645,
+      "loss": 0.1399611234664917,
+      "loss_ce": 6.85493359924294e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 43775036,
+      "step": 485
+    },
+    {
+      "epoch": 2.2344827586206897,
+      "grad_norm": 4.314197235501585,
+      "learning_rate": 5e-06,
+      "loss": 0.1728,
+      "num_input_tokens_seen": 43865412,
+      "step": 486
+    },
+    {
+      "epoch": 2.2344827586206897,
+      "loss": 0.2288285195827484,
+      "loss_ce": 0.00016030190454330295,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 43865412,
+      "step": 486
+    },
+    {
+      "epoch": 2.239080459770115,
+      "grad_norm": 9.468032363551144,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 43955720,
+      "step": 487
+    },
+    {
+      "epoch": 2.239080459770115,
+      "loss": 0.152041494846344,
+      "loss_ce": 9.446687181480229e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 43955720,
+      "step": 487
+    },
+    {
+      "epoch": 2.2436781609195404,
+      "grad_norm": 13.431440995922976,
+      "learning_rate": 5e-06,
+      "loss": 0.145,
+      "num_input_tokens_seen": 44046112,
+      "step": 488
+    },
+    {
+      "epoch": 2.2436781609195404,
+      "loss": 0.1250520944595337,
+      "loss_ce": 0.00011312306742183864,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 44046112,
+      "step": 488
+    },
+    {
+      "epoch": 2.2482758620689656,
+      "grad_norm": 3.5222867230041612,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "num_input_tokens_seen": 44136408,
+      "step": 489
+    },
+    {
+      "epoch": 2.2482758620689656,
+      "loss": 0.10490469634532928,
+      "loss_ce": 4.629993418348022e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 44136408,
+      "step": 489
+    },
+    {
+      "epoch": 2.2528735632183907,
+      "grad_norm": 11.11111499247831,
+      "learning_rate": 5e-06,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 44226780,
+      "step": 490
+    },
+    {
+      "epoch": 2.2528735632183907,
+      "loss": 0.196798175573349,
+      "loss_ce": 5.1357543270569295e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 44226780,
+      "step": 490
+    },
+    {
+      "epoch": 2.2574712643678163,
+      "grad_norm": 5.957507308754348,
+      "learning_rate": 5e-06,
+      "loss": 0.1419,
+      "num_input_tokens_seen": 44317112,
+      "step": 491
+    },
+    {
+      "epoch": 2.2574712643678163,
+      "loss": 0.18355637788772583,
+      "loss_ce": 0.004021458327770233,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 44317112,
+      "step": 491
+    },
+    {
+      "epoch": 2.2620689655172415,
+      "grad_norm": 6.742726892392994,
+      "learning_rate": 5e-06,
+      "loss": 0.1891,
+      "num_input_tokens_seen": 44407476,
+      "step": 492
+    },
+    {
+      "epoch": 2.2620689655172415,
+      "loss": 0.1890466809272766,
+      "loss_ce": 0.0002039193786913529,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 44407476,
+      "step": 492
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 11.363198722775557,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 44497844,
+      "step": 493
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "loss": 0.15371349453926086,
+      "loss_ce": 0.00021006805764045566,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 44497844,
+      "step": 493
+    },
+    {
+      "epoch": 2.2712643678160918,
+      "grad_norm": 20.995205197621296,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 44588252,
+      "step": 494
+    },
+    {
+      "epoch": 2.2712643678160918,
+      "loss": 0.14532320201396942,
+      "loss_ce": 5.9530953876674175e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 44588252,
+      "step": 494
+    },
+    {
+      "epoch": 2.2758620689655173,
+      "grad_norm": 13.929286977818037,
+      "learning_rate": 5e-06,
+      "loss": 0.2026,
+      "num_input_tokens_seen": 44678616,
+      "step": 495
+    },
+    {
+      "epoch": 2.2758620689655173,
+      "loss": 0.18880240619182587,
+      "loss_ce": 0.0003868812054861337,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 44678616,
+      "step": 495
+    },
+    {
+      "epoch": 2.2804597701149425,
+      "grad_norm": 5.471350156799274,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 44769036,
+      "step": 496
+    },
+    {
+      "epoch": 2.2804597701149425,
+      "loss": 0.09233726561069489,
+      "loss_ce": 9.788307215785608e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 44769036,
+      "step": 496
+    },
+    {
+      "epoch": 2.2850574712643676,
+      "grad_norm": 3.59565500880543,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 44859352,
+      "step": 497
+    },
+    {
+      "epoch": 2.2850574712643676,
+      "loss": 0.11752472817897797,
+      "loss_ce": 9.308746666647494e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 44859352,
+      "step": 497
+    },
+    {
+      "epoch": 2.2896551724137932,
+      "grad_norm": 9.014441044097715,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 44949672,
+      "step": 498
+    },
+    {
+      "epoch": 2.2896551724137932,
+      "loss": 0.17396917939186096,
+      "loss_ce": 0.00014104516594670713,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 44949672,
+      "step": 498
+    },
+    {
+      "epoch": 2.2942528735632184,
+      "grad_norm": 56.84361696439731,
+      "learning_rate": 5e-06,
+      "loss": 0.2118,
+      "num_input_tokens_seen": 45040108,
+      "step": 499
+    },
+    {
+      "epoch": 2.2942528735632184,
+      "loss": 0.16254353523254395,
+      "loss_ce": 0.0002510526101104915,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 45040108,
+      "step": 499
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "grad_norm": 4.499065914254272,
+      "learning_rate": 5e-06,
+      "loss": 0.2092,
+      "num_input_tokens_seen": 45130652,
+      "step": 500
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "eval_seeclick_CIoU": 0.4563731998205185,
+      "eval_seeclick_GIoU": 0.4370184391736984,
+      "eval_seeclick_IoU": 0.49749037623405457,
+      "eval_seeclick_MAE_all": 0.0592306274920702,
+      "eval_seeclick_MAE_h": 0.04618253372609615,
+      "eval_seeclick_MAE_w": 0.10425780341029167,
+      "eval_seeclick_MAE_x_boxes": 0.10805046185851097,
+      "eval_seeclick_MAE_y_boxes": 0.042274054139852524,
+      "eval_seeclick_NUM_probability": 0.9999992847442627,
+      "eval_seeclick_inside_bbox": 0.8607954680919647,
+      "eval_seeclick_loss": 0.3510092794895172,
+      "eval_seeclick_loss_ce": 0.05598363280296326,
+      "eval_seeclick_loss_iou": 0.519287109375,
+      "eval_seeclick_loss_num": 0.06208038330078125,
+      "eval_seeclick_loss_xval": 0.310455322265625,
+      "eval_seeclick_runtime": 73.6206,
+      "eval_seeclick_samples_per_second": 0.584,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 45130652,
+      "step": 500
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "eval_icons_CIoU": 0.5248142629861832,
+      "eval_icons_GIoU": 0.5294373333454132,
+      "eval_icons_IoU": 0.5619192123413086,
+      "eval_icons_MAE_all": 0.04526199400424957,
+      "eval_icons_MAE_h": 0.08578694611787796,
+      "eval_icons_MAE_w": 0.06393218040466309,
+      "eval_icons_MAE_x_boxes": 0.05692756548523903,
+      "eval_icons_MAE_y_boxes": 0.08431091904640198,
+      "eval_icons_NUM_probability": 0.9999994337558746,
+      "eval_icons_inside_bbox": 0.7118055522441864,
+      "eval_icons_loss": 0.22472724318504333,
+      "eval_icons_loss_ce": 1.0411458504222537e-06,
+      "eval_icons_loss_iou": 0.43853759765625,
+      "eval_icons_loss_num": 0.049304962158203125,
+      "eval_icons_loss_xval": 0.246337890625,
+      "eval_icons_runtime": 97.2319,
+      "eval_icons_samples_per_second": 0.514,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 45130652,
+      "step": 500
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "eval_screenspot_CIoU": 0.36121458808581036,
+      "eval_screenspot_GIoU": 0.3352409948905309,
+      "eval_screenspot_IoU": 0.4333700935045878,
+      "eval_screenspot_MAE_all": 0.09693960969646771,
+      "eval_screenspot_MAE_h": 0.09257866690556209,
+      "eval_screenspot_MAE_w": 0.19348373264074326,
+      "eval_screenspot_MAE_x_boxes": 0.18671841422716776,
+      "eval_screenspot_MAE_y_boxes": 0.08937582621971767,
+      "eval_screenspot_NUM_probability": 0.9999986489613851,
+      "eval_screenspot_inside_bbox": 0.7145833373069763,
+      "eval_screenspot_loss": 0.4828945994377136,
+      "eval_screenspot_loss_ce": 0.0005104693118482828,
+      "eval_screenspot_loss_iou": 0.3957926432291667,
+      "eval_screenspot_loss_num": 0.099456787109375,
+      "eval_screenspot_loss_xval": 0.4974365234375,
+      "eval_screenspot_runtime": 150.5805,
+      "eval_screenspot_samples_per_second": 0.591,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 45130652,
+      "step": 500
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "eval_compot_CIoU": 0.4746464341878891,
+      "eval_compot_GIoU": 0.444294735789299,
+      "eval_compot_IoU": 0.5337270200252533,
+      "eval_compot_MAE_all": 0.05769064649939537,
+      "eval_compot_MAE_h": 0.08894045650959015,
+      "eval_compot_MAE_w": 0.10224151611328125,
+      "eval_compot_MAE_x_boxes": 0.08807999640703201,
+      "eval_compot_MAE_y_boxes": 0.08862848207354546,
+      "eval_compot_NUM_probability": 0.9999935030937195,
+      "eval_compot_inside_bbox": 0.7638888955116272,
+      "eval_compot_loss": 0.31889885663986206,
+      "eval_compot_loss_ce": 0.01438705949112773,
+      "eval_compot_loss_iou": 0.5205078125,
+      "eval_compot_loss_num": 0.05315399169921875,
+      "eval_compot_loss_xval": 0.265472412109375,
+      "eval_compot_runtime": 86.6902,
+      "eval_compot_samples_per_second": 0.577,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 45130652,
+      "step": 500
+    },
+    {
+      "epoch": 2.2988505747126435,
+      "loss": 0.19853872060775757,
+      "loss_ce": 0.008414196781814098,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 45130652,
+      "step": 500
+    },
+    {
+      "epoch": 2.303448275862069,
+      "grad_norm": 25.629104321788386,
+      "learning_rate": 5e-06,
+      "loss": 0.1748,
+      "num_input_tokens_seen": 45221168,
+      "step": 501
+    },
+    {
+      "epoch": 2.303448275862069,
+      "loss": 0.22815854847431183,
+      "loss_ce": 0.00013120633957441896,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 45221168,
+      "step": 501
+    },
+    {
+      "epoch": 2.3080459770114943,
+      "grad_norm": 5.708666769215059,
+      "learning_rate": 5e-06,
+      "loss": 0.1472,
+      "num_input_tokens_seen": 45311552,
+      "step": 502
+    },
+    {
+      "epoch": 2.3080459770114943,
+      "loss": 0.14808592200279236,
+      "loss_ce": 0.00019773890380747616,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 45311552,
+      "step": 502
+    },
+    {
+      "epoch": 2.3126436781609194,
+      "grad_norm": 5.468912782546553,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 45401820,
+      "step": 503
+    },
+    {
+      "epoch": 2.3126436781609194,
+      "loss": 0.17252948880195618,
+      "loss_ce": 0.00022724125301465392,
+      "loss_iou": 0.4375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 45401820,
+      "step": 503
+    },
+    {
+      "epoch": 2.317241379310345,
+      "grad_norm": 3.4368379842335006,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 45491352,
+      "step": 504
+    },
+    {
+      "epoch": 2.317241379310345,
+      "loss": 0.17304499447345734,
+      "loss_ce": 0.004771072883158922,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 45491352,
+      "step": 504
+    },
+    {
+      "epoch": 2.32183908045977,
+      "grad_norm": 7.438752813468672,
+      "learning_rate": 5e-06,
+      "loss": 0.1578,
+      "num_input_tokens_seen": 45581828,
+      "step": 505
+    },
+    {
+      "epoch": 2.32183908045977,
+      "loss": 0.12591499090194702,
+      "loss_ce": 6.0488393501145765e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 45581828,
+      "step": 505
+    },
+    {
+      "epoch": 2.3264367816091953,
+      "grad_norm": 2.55941882010794,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 45672168,
+      "step": 506
+    },
+    {
+      "epoch": 2.3264367816091953,
+      "loss": 0.15659506618976593,
+      "loss_ce": 9.366869562654756e-06,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 45672168,
+      "step": 506
+    },
+    {
+      "epoch": 2.3310344827586205,
+      "grad_norm": 3.8697008798484838,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 45761816,
+      "step": 507
+    },
+    {
+      "epoch": 2.3310344827586205,
+      "loss": 0.1356024444103241,
+      "loss_ce": 0.00010439224570291117,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 45761816,
+      "step": 507
+    },
+    {
+      "epoch": 2.335632183908046,
+      "grad_norm": 9.154165977355914,
+      "learning_rate": 5e-06,
+      "loss": 0.165,
+      "num_input_tokens_seen": 45851328,
+      "step": 508
+    },
+    {
+      "epoch": 2.335632183908046,
+      "loss": 0.17103195190429688,
+      "loss_ce": 1.1456304491730407e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 45851328,
+      "step": 508
+    },
+    {
+      "epoch": 2.340229885057471,
+      "grad_norm": 6.692072817800835,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 45941700,
+      "step": 509
+    },
+    {
+      "epoch": 2.340229885057471,
+      "loss": 0.149922713637352,
+      "loss_ce": 0.00017295028374064714,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 45941700,
+      "step": 509
+    },
+    {
+      "epoch": 2.344827586206897,
+      "grad_norm": 6.611698710045344,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 46032036,
+      "step": 510
+    },
+    {
+      "epoch": 2.344827586206897,
+      "loss": 0.22002628445625305,
+      "loss_ce": 2.505266820662655e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 46032036,
+      "step": 510
+    },
+    {
+      "epoch": 2.349425287356322,
+      "grad_norm": 11.300552233375646,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 46122476,
+      "step": 511
+    },
+    {
+      "epoch": 2.349425287356322,
+      "loss": 0.1223573088645935,
+      "loss_ce": 4.286051989765838e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 46122476,
+      "step": 511
+    },
+    {
+      "epoch": 2.354022988505747,
+      "grad_norm": 4.419461671248112,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 46213036,
+      "step": 512
+    },
+    {
+      "epoch": 2.354022988505747,
+      "loss": 0.21517296135425568,
+      "loss_ce": 8.506829908583313e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 46213036,
+      "step": 512
+    },
+    {
+      "epoch": 2.3586206896551722,
+      "grad_norm": 4.8376770471091755,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 46303360,
+      "step": 513
+    },
+    {
+      "epoch": 2.3586206896551722,
+      "loss": 0.13146010041236877,
+      "loss_ce": 5.1406939746811986e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 46303360,
+      "step": 513
+    },
+    {
+      "epoch": 2.363218390804598,
+      "grad_norm": 14.293011444773292,
+      "learning_rate": 5e-06,
+      "loss": 0.1544,
+      "num_input_tokens_seen": 46393732,
+      "step": 514
+    },
+    {
+      "epoch": 2.363218390804598,
+      "loss": 0.1853524148464203,
+      "loss_ce": 0.000568474642932415,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 46393732,
+      "step": 514
+    },
+    {
+      "epoch": 2.367816091954023,
+      "grad_norm": 5.9219449761184135,
+      "learning_rate": 5e-06,
+      "loss": 0.1525,
+      "num_input_tokens_seen": 46484084,
+      "step": 515
+    },
+    {
+      "epoch": 2.367816091954023,
+      "loss": 0.13725048303604126,
+      "loss_ce": 0.005231434479355812,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 46484084,
+      "step": 515
+    },
+    {
+      "epoch": 2.372413793103448,
+      "grad_norm": 11.485628452257725,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 46574492,
+      "step": 516
+    },
+    {
+      "epoch": 2.372413793103448,
+      "loss": 0.10670308768749237,
+      "loss_ce": 4.415482544573024e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 46574492,
+      "step": 516
+    },
+    {
+      "epoch": 2.3770114942528737,
+      "grad_norm": 3.642033089699345,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 46664972,
+      "step": 517
+    },
+    {
+      "epoch": 2.3770114942528737,
+      "loss": 0.11098849028348923,
+      "loss_ce": 5.709293327527121e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 46664972,
+      "step": 517
+    },
+    {
+      "epoch": 2.381609195402299,
+      "grad_norm": 5.135666742103272,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 46755488,
+      "step": 518
+    },
+    {
+      "epoch": 2.381609195402299,
+      "loss": 0.10365065187215805,
+      "loss_ce": 1.295450420002453e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 46755488,
+      "step": 518
+    },
+    {
+      "epoch": 2.386206896551724,
+      "grad_norm": 14.48664391569232,
+      "learning_rate": 5e-06,
+      "loss": 0.1674,
+      "num_input_tokens_seen": 46845788,
+      "step": 519
+    },
+    {
+      "epoch": 2.386206896551724,
+      "loss": 0.1562279760837555,
+      "loss_ce": 3.902369280694984e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 46845788,
+      "step": 519
+    },
+    {
+      "epoch": 2.3908045977011496,
+      "grad_norm": 5.121624515541806,
+      "learning_rate": 5e-06,
+      "loss": 0.1334,
+      "num_input_tokens_seen": 46936132,
+      "step": 520
+    },
+    {
+      "epoch": 2.3908045977011496,
+      "loss": 0.15609031915664673,
+      "loss_ce": 0.002129129832610488,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 46936132,
+      "step": 520
+    },
+    {
+      "epoch": 2.3954022988505748,
+      "grad_norm": 12.984930412372549,
+      "learning_rate": 5e-06,
+      "loss": 0.1588,
+      "num_input_tokens_seen": 47026436,
+      "step": 521
+    },
+    {
+      "epoch": 2.3954022988505748,
+      "loss": 0.13674747943878174,
+      "loss_ce": 5.9245572629151866e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 47026436,
+      "step": 521
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 4.283346201311201,
+      "learning_rate": 5e-06,
+      "loss": 0.2249,
+      "num_input_tokens_seen": 47116864,
+      "step": 522
+    },
+    {
+      "epoch": 2.4,
+      "loss": 0.1752316951751709,
+      "loss_ce": 0.00030492368387058377,
+      "loss_iou": 0.421875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 47116864,
+      "step": 522
+    },
+    {
+      "epoch": 2.4045977011494255,
+      "grad_norm": 5.233495198593664,
+      "learning_rate": 5e-06,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 47207228,
+      "step": 523
+    },
+    {
+      "epoch": 2.4045977011494255,
+      "loss": 0.18661803007125854,
+      "loss_ce": 3.356490924488753e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 47207228,
+      "step": 523
+    },
+    {
+      "epoch": 2.4091954022988507,
+      "grad_norm": 5.2220413793779175,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 47297444,
+      "step": 524
+    },
+    {
+      "epoch": 2.4091954022988507,
+      "loss": 0.1606762409210205,
+      "loss_ce": 6.222333468031138e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 47297444,
+      "step": 524
+    },
+    {
+      "epoch": 2.413793103448276,
+      "grad_norm": 18.74625653768912,
+      "learning_rate": 5e-06,
+      "loss": 0.132,
+      "num_input_tokens_seen": 47387760,
+      "step": 525
+    },
+    {
+      "epoch": 2.413793103448276,
+      "loss": 0.15789783000946045,
+      "loss_ce": 0.00012196188617963344,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 47387760,
+      "step": 525
+    },
+    {
+      "epoch": 2.418390804597701,
+      "grad_norm": 5.191217680993693,
+      "learning_rate": 5e-06,
+      "loss": 0.1316,
+      "num_input_tokens_seen": 47478124,
+      "step": 526
+    },
+    {
+      "epoch": 2.418390804597701,
+      "loss": 0.13596881926059723,
+      "loss_ce": 4.352313044364564e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 47478124,
+      "step": 526
+    },
+    {
+      "epoch": 2.4229885057471265,
+      "grad_norm": 4.706845583881529,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 47568600,
+      "step": 527
+    },
+    {
+      "epoch": 2.4229885057471265,
+      "loss": 0.15037593245506287,
+      "loss_ce": 4.634057768271305e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 47568600,
+      "step": 527
+    },
+    {
+      "epoch": 2.4275862068965517,
+      "grad_norm": 7.095455232957096,
+      "learning_rate": 5e-06,
+      "loss": 0.1918,
+      "num_input_tokens_seen": 47658948,
+      "step": 528
+    },
+    {
+      "epoch": 2.4275862068965517,
+      "loss": 0.25855281949043274,
+      "loss_ce": 0.00025203393306583166,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 47658948,
+      "step": 528
+    },
+    {
+      "epoch": 2.432183908045977,
+      "grad_norm": 6.0875597343514265,
+      "learning_rate": 5e-06,
+      "loss": 0.1973,
+      "num_input_tokens_seen": 47749252,
+      "step": 529
+    },
+    {
+      "epoch": 2.432183908045977,
+      "loss": 0.21008528769016266,
+      "loss_ce": 0.001833330374211073,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 47749252,
+      "step": 529
+    },
+    {
+      "epoch": 2.4367816091954024,
+      "grad_norm": 43.1300948642533,
+      "learning_rate": 5e-06,
+      "loss": 0.1807,
+      "num_input_tokens_seen": 47838180,
+      "step": 530
+    },
+    {
+      "epoch": 2.4367816091954024,
+      "loss": 0.1620747148990631,
+      "loss_ce": 1.1108362741651945e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 47838180,
+      "step": 530
+    },
+    {
+      "epoch": 2.4413793103448276,
+      "grad_norm": 34.04511460331946,
+      "learning_rate": 5e-06,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 47928576,
+      "step": 531
+    },
+    {
+      "epoch": 2.4413793103448276,
+      "loss": 0.14182758331298828,
+      "loss_ce": 4.29103929491248e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 47928576,
+      "step": 531
+    },
+    {
+      "epoch": 2.4459770114942527,
+      "grad_norm": 14.495961279987304,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 48019012,
+      "step": 532
+    },
+    {
+      "epoch": 2.4459770114942527,
+      "loss": 0.205206036567688,
+      "loss_ce": 0.00012791430344805121,
+      "loss_iou": 0.421875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 48019012,
+      "step": 532
+    },
+    {
+      "epoch": 2.4505747126436783,
+      "grad_norm": 47.008419334105845,
+      "learning_rate": 5e-06,
+      "loss": 0.1181,
+      "num_input_tokens_seen": 48109540,
+      "step": 533
+    },
+    {
+      "epoch": 2.4505747126436783,
+      "loss": 0.09603258222341537,
+      "loss_ce": 8.53195961099118e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 48109540,
+      "step": 533
+    },
+    {
+      "epoch": 2.4551724137931035,
+      "grad_norm": 22.246765917362193,
+      "learning_rate": 5e-06,
+      "loss": 0.1418,
+      "num_input_tokens_seen": 48200004,
+      "step": 534
+    },
+    {
+      "epoch": 2.4551724137931035,
+      "loss": 0.15302670001983643,
+      "loss_ce": 7.259696576511487e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 48200004,
+      "step": 534
+    },
+    {
+      "epoch": 2.4597701149425286,
+      "grad_norm": 10.021203723893532,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 48290512,
+      "step": 535
+    },
+    {
+      "epoch": 2.4597701149425286,
+      "loss": 0.11590239405632019,
+      "loss_ce": 0.00024077783746179193,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 48290512,
+      "step": 535
+    },
+    {
+      "epoch": 2.464367816091954,
+      "grad_norm": 11.067971491143098,
+      "learning_rate": 5e-06,
+      "loss": 0.1704,
+      "num_input_tokens_seen": 48380860,
+      "step": 536
+    },
+    {
+      "epoch": 2.464367816091954,
+      "loss": 0.19783857464790344,
+      "loss_ce": 0.00023726305516902357,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 48380860,
+      "step": 536
+    },
+    {
+      "epoch": 2.4689655172413794,
+      "grad_norm": 5.653254921318319,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 48471104,
+      "step": 537
+    },
+    {
+      "epoch": 2.4689655172413794,
+      "loss": 0.1712353229522705,
+      "loss_ce": 6.221771764103323e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 48471104,
+      "step": 537
+    },
+    {
+      "epoch": 2.4735632183908045,
+      "grad_norm": 2.9377901589977085,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 48561516,
+      "step": 538
+    },
+    {
+      "epoch": 2.4735632183908045,
+      "loss": 0.12568363547325134,
+      "loss_ce": 1.2256616173544899e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 48561516,
+      "step": 538
+    },
+    {
+      "epoch": 2.4781609195402297,
+      "grad_norm": 90.38237762494742,
+      "learning_rate": 5e-06,
+      "loss": 0.1775,
+      "num_input_tokens_seen": 48651772,
+      "step": 539
+    },
+    {
+      "epoch": 2.4781609195402297,
+      "loss": 0.20320890843868256,
+      "loss_ce": 8.390971197513863e-05,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 48651772,
+      "step": 539
+    },
+    {
+      "epoch": 2.4827586206896552,
+      "grad_norm": 32.249804456709384,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 48742128,
+      "step": 540
+    },
+    {
+      "epoch": 2.4827586206896552,
+      "loss": 0.13325834274291992,
+      "loss_ce": 7.96284293755889e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 48742128,
+      "step": 540
+    },
+    {
+      "epoch": 2.4873563218390804,
+      "grad_norm": 4.3167432311394505,
+      "learning_rate": 5e-06,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 48832520,
+      "step": 541
+    },
+    {
+      "epoch": 2.4873563218390804,
+      "loss": 0.1864563375711441,
+      "loss_ce": 5.4961776186246425e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 48832520,
+      "step": 541
+    },
+    {
+      "epoch": 2.491954022988506,
+      "grad_norm": 9.206979571070748,
+      "learning_rate": 5e-06,
+      "loss": 0.1701,
+      "num_input_tokens_seen": 48922880,
+      "step": 542
+    },
+    {
+      "epoch": 2.491954022988506,
+      "loss": 0.14663755893707275,
+      "loss_ce": 0.00021421856945380569,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 48922880,
+      "step": 542
+    },
+    {
+      "epoch": 2.496551724137931,
+      "grad_norm": 9.250882650158083,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 49013144,
+      "step": 543
+    },
+    {
+      "epoch": 2.496551724137931,
+      "loss": 0.1472419798374176,
+      "loss_ce": 9.938361472450197e-06,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 49013144,
+      "step": 543
+    },
+    {
+      "epoch": 2.5011494252873563,
+      "grad_norm": 4.218071296940112,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 49103612,
+      "step": 544
+    },
+    {
+      "epoch": 2.5011494252873563,
+      "loss": 0.13012462854385376,
+      "loss_ce": 0.00018077107961289585,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 49103612,
+      "step": 544
+    },
+    {
+      "epoch": 2.5057471264367814,
+      "grad_norm": 5.7187751949461285,
+      "learning_rate": 5e-06,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 49194032,
+      "step": 545
+    },
+    {
+      "epoch": 2.5057471264367814,
+      "loss": 0.1249690055847168,
+      "loss_ce": 0.00015211128629744053,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 49194032,
+      "step": 545
+    },
+    {
+      "epoch": 2.510344827586207,
+      "grad_norm": 4.115869504252711,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 49284388,
+      "step": 546
+    },
+    {
+      "epoch": 2.510344827586207,
+      "loss": 0.13758164644241333,
+      "loss_ce": 6.943976768525317e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 49284388,
+      "step": 546
+    },
+    {
+      "epoch": 2.514942528735632,
+      "grad_norm": 9.743639046477583,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 49374780,
+      "step": 547
+    },
+    {
+      "epoch": 2.514942528735632,
+      "loss": 0.18289527297019958,
+      "loss_ce": 3.3943615562748164e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 49374780,
+      "step": 547
+    },
+    {
+      "epoch": 2.5195402298850573,
+      "grad_norm": 10.069950958132628,
+      "learning_rate": 5e-06,
+      "loss": 0.1805,
+      "num_input_tokens_seen": 49465116,
+      "step": 548
+    },
+    {
+      "epoch": 2.5195402298850573,
+      "loss": 0.1874997317790985,
+      "loss_ce": 3.024038960575126e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 49465116,
+      "step": 548
+    },
+    {
+      "epoch": 2.524137931034483,
+      "grad_norm": 19.870338599096762,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 49555324,
+      "step": 549
+    },
+    {
+      "epoch": 2.524137931034483,
+      "loss": 0.1507517397403717,
+      "loss_ce": 5.592878733295947e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 49555324,
+      "step": 549
+    },
+    {
+      "epoch": 2.528735632183908,
+      "grad_norm": 10.407984592997497,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 49645776,
+      "step": 550
+    },
+    {
+      "epoch": 2.528735632183908,
+      "loss": 0.18005026876926422,
+      "loss_ce": 2.7080646759714e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 49645776,
+      "step": 550
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 7.734767791315441,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 49736148,
+      "step": 551
+    },
+    {
+      "epoch": 2.533333333333333,
+      "loss": 0.16183626651763916,
+      "loss_ce": 0.00012362068810034543,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 49736148,
+      "step": 551
+    },
+    {
+      "epoch": 2.5379310344827584,
+      "grad_norm": 6.4853103024455505,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 49826532,
+      "step": 552
+    },
+    {
+      "epoch": 2.5379310344827584,
+      "loss": 0.10679163783788681,
+      "loss_ce": 7.166996510932222e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 49826532,
+      "step": 552
+    },
+    {
+      "epoch": 2.542528735632184,
+      "grad_norm": 37.47119278150993,
+      "learning_rate": 5e-06,
+      "loss": 0.1699,
+      "num_input_tokens_seen": 49917020,
+      "step": 553
+    },
+    {
+      "epoch": 2.542528735632184,
+      "loss": 0.1629483699798584,
+      "loss_ce": 0.0001065782635123469,
+      "loss_iou": 0.4375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 49917020,
+      "step": 553
+    },
+    {
+      "epoch": 2.547126436781609,
+      "grad_norm": 6.379696690058362,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 50007432,
+      "step": 554
+    },
+    {
+      "epoch": 2.547126436781609,
+      "loss": 0.15533311665058136,
+      "loss_ce": 4.4418171455617994e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 50007432,
+      "step": 554
+    },
+    {
+      "epoch": 2.5517241379310347,
+      "grad_norm": 4.14549282032518,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 50097820,
+      "step": 555
+    },
+    {
+      "epoch": 2.5517241379310347,
+      "loss": 0.1373203992843628,
+      "loss_ce": 5.233901902101934e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 50097820,
+      "step": 555
+    },
+    {
+      "epoch": 2.55632183908046,
+      "grad_norm": 3.4318598421901934,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 50188248,
+      "step": 556
+    },
+    {
+      "epoch": 2.55632183908046,
+      "loss": 0.13454991579055786,
+      "loss_ce": 8.948062895797193e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 50188248,
+      "step": 556
+    },
+    {
+      "epoch": 2.560919540229885,
+      "grad_norm": 3.7413568064084655,
+      "learning_rate": 5e-06,
+      "loss": 0.1643,
+      "num_input_tokens_seen": 50278580,
+      "step": 557
+    },
+    {
+      "epoch": 2.560919540229885,
+      "loss": 0.10955231636762619,
+      "loss_ce": 8.576564869144931e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 50278580,
+      "step": 557
+    },
+    {
+      "epoch": 2.56551724137931,
+      "grad_norm": 16.960520342941066,
+      "learning_rate": 5e-06,
+      "loss": 0.1843,
+      "num_input_tokens_seen": 50368852,
+      "step": 558
+    },
+    {
+      "epoch": 2.56551724137931,
+      "loss": 0.19905152916908264,
+      "loss_ce": 1.5889523638179526e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 50368852,
+      "step": 558
+    },
+    {
+      "epoch": 2.5701149425287357,
+      "grad_norm": 2.4502760696851618,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 50459208,
+      "step": 559
+    },
+    {
+      "epoch": 2.5701149425287357,
+      "loss": 0.15352153778076172,
+      "loss_ce": 0.00010967279376927763,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 50459208,
+      "step": 559
+    },
+    {
+      "epoch": 2.574712643678161,
+      "grad_norm": 3.003683699720249,
+      "learning_rate": 5e-06,
+      "loss": 0.1435,
+      "num_input_tokens_seen": 50549548,
+      "step": 560
+    },
+    {
+      "epoch": 2.574712643678161,
+      "loss": 0.14211556315422058,
+      "loss_ce": 0.00011727242235792801,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 50549548,
+      "step": 560
+    },
+    {
+      "epoch": 2.5793103448275865,
+      "grad_norm": 8.360372471761355,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 50639992,
+      "step": 561
+    },
+    {
+      "epoch": 2.5793103448275865,
+      "loss": 0.12777529656887054,
+      "loss_ce": 4.39732575614471e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 50639992,
+      "step": 561
+    },
+    {
+      "epoch": 2.5839080459770116,
+      "grad_norm": 17.063605502973996,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 50730300,
+      "step": 562
+    },
+    {
+      "epoch": 2.5839080459770116,
+      "loss": 0.1004580706357956,
+      "loss_ce": 2.4719898647163063e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 50730300,
+      "step": 562
+    },
+    {
+      "epoch": 2.5885057471264368,
+      "grad_norm": 5.4210250306263585,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 50819996,
+      "step": 563
+    },
+    {
+      "epoch": 2.5885057471264368,
+      "loss": 0.18996095657348633,
+      "loss_ce": 1.954480103449896e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 50819996,
+      "step": 563
+    },
+    {
+      "epoch": 2.593103448275862,
+      "grad_norm": 7.275873200745777,
+      "learning_rate": 5e-06,
+      "loss": 0.1733,
+      "num_input_tokens_seen": 50910296,
+      "step": 564
+    },
+    {
+      "epoch": 2.593103448275862,
+      "loss": 0.1952449530363083,
+      "loss_ce": 8.50430442369543e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 50910296,
+      "step": 564
+    },
+    {
+      "epoch": 2.5977011494252875,
+      "grad_norm": 15.23898519773349,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 51000576,
+      "step": 565
+    },
+    {
+      "epoch": 2.5977011494252875,
+      "loss": 0.1579952836036682,
+      "loss_ce": 2.1030613424954936e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 51000576,
+      "step": 565
+    },
+    {
+      "epoch": 2.6022988505747127,
+      "grad_norm": 4.894484835323628,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 51090992,
+      "step": 566
+    },
+    {
+      "epoch": 2.6022988505747127,
+      "loss": 0.12843066453933716,
+      "loss_ce": 4.321142114349641e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 51090992,
+      "step": 566
+    },
+    {
+      "epoch": 2.606896551724138,
+      "grad_norm": 20.80440116305703,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 51181196,
+      "step": 567
+    },
+    {
+      "epoch": 2.606896551724138,
+      "loss": 0.14310401678085327,
+      "loss_ce": 3.761286643566564e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 51181196,
+      "step": 567
+    },
+    {
+      "epoch": 2.6114942528735634,
+      "grad_norm": 4.997998048110718,
+      "learning_rate": 5e-06,
+      "loss": 0.1867,
+      "num_input_tokens_seen": 51271612,
+      "step": 568
+    },
+    {
+      "epoch": 2.6114942528735634,
+      "loss": 0.24052223563194275,
+      "loss_ce": 4.3717802327591926e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 51271612,
+      "step": 568
+    },
+    {
+      "epoch": 2.6160919540229886,
+      "grad_norm": 14.548525343911681,
+      "learning_rate": 5e-06,
+      "loss": 0.1517,
+      "num_input_tokens_seen": 51362012,
+      "step": 569
+    },
+    {
+      "epoch": 2.6160919540229886,
+      "loss": 0.15177559852600098,
+      "loss_ce": 1.1683812772389501e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 51362012,
+      "step": 569
+    },
+    {
+      "epoch": 2.6206896551724137,
+      "grad_norm": 8.538017397932672,
+      "learning_rate": 5e-06,
+      "loss": 0.138,
+      "num_input_tokens_seen": 51452264,
+      "step": 570
+    },
+    {
+      "epoch": 2.6206896551724137,
+      "loss": 0.16324067115783691,
+      "loss_ce": 3.2669748179614544e-05,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 51452264,
+      "step": 570
+    },
+    {
+      "epoch": 2.625287356321839,
+      "grad_norm": 5.458750598266798,
+      "learning_rate": 5e-06,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 51542688,
+      "step": 571
+    },
+    {
+      "epoch": 2.625287356321839,
+      "loss": 0.1548556387424469,
+      "loss_ce": 3.9948965422809124e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 51542688,
+      "step": 571
+    },
+    {
+      "epoch": 2.6298850574712644,
+      "grad_norm": 6.9963748084758475,
+      "learning_rate": 5e-06,
+      "loss": 0.1511,
+      "num_input_tokens_seen": 51633028,
+      "step": 572
+    },
+    {
+      "epoch": 2.6298850574712644,
+      "loss": 0.10907353460788727,
+      "loss_ce": 0.00012577624875120819,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 51633028,
+      "step": 572
+    },
+    {
+      "epoch": 2.6344827586206896,
+      "grad_norm": 5.485383848252606,
+      "learning_rate": 5e-06,
+      "loss": 0.2015,
+      "num_input_tokens_seen": 51723304,
+      "step": 573
+    },
+    {
+      "epoch": 2.6344827586206896,
+      "loss": 0.20509332418441772,
+      "loss_ce": 0.009658743627369404,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 51723304,
+      "step": 573
+    },
+    {
+      "epoch": 2.639080459770115,
+      "grad_norm": 5.145735431943079,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 51813780,
+      "step": 574
+    },
+    {
+      "epoch": 2.639080459770115,
+      "loss": 0.14565084874629974,
+      "loss_ce": 5.1484184950822964e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 51813780,
+      "step": 574
+    },
+    {
+      "epoch": 2.6436781609195403,
+      "grad_norm": 11.268920979351746,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 51904132,
+      "step": 575
+    },
+    {
+      "epoch": 2.6436781609195403,
+      "loss": 0.11697958409786224,
+      "loss_ce": 6.674315227428451e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 51904132,
+      "step": 575
+    },
+    {
+      "epoch": 2.6482758620689655,
+      "grad_norm": 5.697746433151331,
+      "learning_rate": 5e-06,
+      "loss": 0.1411,
+      "num_input_tokens_seen": 51993764,
+      "step": 576
+    },
+    {
+      "epoch": 2.6482758620689655,
+      "loss": 0.12572216987609863,
+      "loss_ce": 5.002422767574899e-06,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 51993764,
+      "step": 576
+    },
+    {
+      "epoch": 2.6528735632183906,
+      "grad_norm": 23.221495764418574,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 52084092,
+      "step": 577
+    },
+    {
+      "epoch": 2.6528735632183906,
+      "loss": 0.10617563128471375,
+      "loss_ce": 3.5495380870997906e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 52084092,
+      "step": 577
+    },
+    {
+      "epoch": 2.657471264367816,
+      "grad_norm": 9.375076718437624,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 52174520,
+      "step": 578
+    },
+    {
+      "epoch": 2.657471264367816,
+      "loss": 0.15085318684577942,
+      "loss_ce": 3.5313376429257914e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 52174520,
+      "step": 578
+    },
+    {
+      "epoch": 2.6620689655172414,
+      "grad_norm": 14.09120273250363,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 52264944,
+      "step": 579
+    },
+    {
+      "epoch": 2.6620689655172414,
+      "loss": 0.13439422845840454,
+      "loss_ce": 2.5327437469968572e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 52264944,
+      "step": 579
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 7.150633563903932,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 52355440,
+      "step": 580
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "loss": 0.12812362611293793,
+      "loss_ce": 0.0001328981015831232,
+      "loss_iou": 0.421875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 52355440,
+      "step": 580
+    },
+    {
+      "epoch": 2.671264367816092,
+      "grad_norm": 17.68235051618041,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 52445736,
+      "step": 581
+    },
+    {
+      "epoch": 2.671264367816092,
+      "loss": 0.11226825416088104,
+      "loss_ce": 2.46073159360094e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 52445736,
+      "step": 581
+    },
+    {
+      "epoch": 2.6758620689655173,
+      "grad_norm": 9.382663357540876,
+      "learning_rate": 5e-06,
+      "loss": 0.1607,
+      "num_input_tokens_seen": 52536112,
+      "step": 582
+    },
+    {
+      "epoch": 2.6758620689655173,
+      "loss": 0.12408533692359924,
+      "loss_ce": 0.0001839685719460249,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 52536112,
+      "step": 582
+    },
+    {
+      "epoch": 2.6804597701149424,
+      "grad_norm": 4.0581740149048615,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 52626460,
+      "step": 583
+    },
+    {
+      "epoch": 2.6804597701149424,
+      "loss": 0.1088830754160881,
+      "loss_ce": 4.213307693134993e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 52626460,
+      "step": 583
+    },
+    {
+      "epoch": 2.6850574712643676,
+      "grad_norm": 21.41413357247865,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 52716824,
+      "step": 584
+    },
+    {
+      "epoch": 2.6850574712643676,
+      "loss": 0.1613665223121643,
+      "loss_ce": 8.111814531730488e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 52716824,
+      "step": 584
+    },
+    {
+      "epoch": 2.689655172413793,
+      "grad_norm": 4.051520688977973,
+      "learning_rate": 5e-06,
+      "loss": 0.1631,
+      "num_input_tokens_seen": 52807040,
+      "step": 585
+    },
+    {
+      "epoch": 2.689655172413793,
+      "loss": 0.14354471862316132,
+      "loss_ce": 5.1069248002022505e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 52807040,
+      "step": 585
+    },
+    {
+      "epoch": 2.6942528735632183,
+      "grad_norm": 8.143732797882166,
+      "learning_rate": 5e-06,
+      "loss": 0.2078,
+      "num_input_tokens_seen": 52897404,
+      "step": 586
+    },
+    {
+      "epoch": 2.6942528735632183,
+      "loss": 0.21973484754562378,
+      "loss_ce": 0.0002524274750612676,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 52897404,
+      "step": 586
+    },
+    {
+      "epoch": 2.698850574712644,
+      "grad_norm": 9.812036652439687,
+      "learning_rate": 5e-06,
+      "loss": 0.1997,
+      "num_input_tokens_seen": 52987688,
+      "step": 587
+    },
+    {
+      "epoch": 2.698850574712644,
+      "loss": 0.2230190932750702,
+      "loss_ce": 5.7667512010084465e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 52987688,
+      "step": 587
+    },
+    {
+      "epoch": 2.703448275862069,
+      "grad_norm": 11.85150221226106,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 53078184,
+      "step": 588
+    },
+    {
+      "epoch": 2.703448275862069,
+      "loss": 0.16252008080482483,
+      "loss_ce": 1.3969104657007847e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 53078184,
+      "step": 588
+    },
+    {
+      "epoch": 2.708045977011494,
+      "grad_norm": 4.7260277066471446,
+      "learning_rate": 5e-06,
+      "loss": 0.1668,
+      "num_input_tokens_seen": 53168516,
+      "step": 589
+    },
+    {
+      "epoch": 2.708045977011494,
+      "loss": 0.21362201869487762,
+      "loss_ce": 0.0016469230176880956,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 53168516,
+      "step": 589
+    },
+    {
+      "epoch": 2.7126436781609193,
+      "grad_norm": 18.524483984567052,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 53258748,
+      "step": 590
+    },
+    {
+      "epoch": 2.7126436781609193,
+      "loss": 0.1348852813243866,
+      "loss_ce": 2.8092332286178134e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 53258748,
+      "step": 590
+    },
+    {
+      "epoch": 2.717241379310345,
+      "grad_norm": 9.8553207385232,
+      "learning_rate": 5e-06,
+      "loss": 0.216,
+      "num_input_tokens_seen": 53349272,
+      "step": 591
+    },
+    {
+      "epoch": 2.717241379310345,
+      "loss": 0.26626330614089966,
+      "loss_ce": 8.898475061869249e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 53349272,
+      "step": 591
+    },
+    {
+      "epoch": 2.72183908045977,
+      "grad_norm": 10.070351523808545,
+      "learning_rate": 5e-06,
+      "loss": 0.1794,
+      "num_input_tokens_seen": 53438892,
+      "step": 592
+    },
+    {
+      "epoch": 2.72183908045977,
+      "loss": 0.17163211107254028,
+      "loss_ce": 6.230298458831385e-05,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 53438892,
+      "step": 592
+    },
+    {
+      "epoch": 2.7264367816091957,
+      "grad_norm": 4.7799609797973535,
+      "learning_rate": 5e-06,
+      "loss": 0.229,
+      "num_input_tokens_seen": 53529204,
+      "step": 593
+    },
+    {
+      "epoch": 2.7264367816091957,
+      "loss": 0.2294745147228241,
+      "loss_ce": 0.00010438873869134113,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 53529204,
+      "step": 593
+    },
+    {
+      "epoch": 2.731034482758621,
+      "grad_norm": 10.927919604630798,
+      "learning_rate": 5e-06,
+      "loss": 0.1659,
+      "num_input_tokens_seen": 53619564,
+      "step": 594
+    },
+    {
+      "epoch": 2.731034482758621,
+      "loss": 0.16697654128074646,
+      "loss_ce": 1.487006011302583e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 53619564,
+      "step": 594
+    },
+    {
+      "epoch": 2.735632183908046,
+      "grad_norm": 14.00357432861095,
+      "learning_rate": 5e-06,
+      "loss": 0.1539,
+      "num_input_tokens_seen": 53709964,
+      "step": 595
+    },
+    {
+      "epoch": 2.735632183908046,
+      "loss": 0.1489291936159134,
+      "loss_ce": 9.496298298472539e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 53709964,
+      "step": 595
+    },
+    {
+      "epoch": 2.740229885057471,
+      "grad_norm": 18.78198114746188,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 53799628,
+      "step": 596
+    },
+    {
+      "epoch": 2.740229885057471,
+      "loss": 0.17056533694267273,
+      "loss_ce": 0.006441800855100155,
+      "loss_iou": 0.375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 53799628,
+      "step": 596
+    },
+    {
+      "epoch": 2.7448275862068967,
+      "grad_norm": 10.48163904175623,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 53889936,
+      "step": 597
+    },
+    {
+      "epoch": 2.7448275862068967,
+      "loss": 0.13397589325904846,
+      "loss_ce": 1.898076334327925e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 53889936,
+      "step": 597
+    },
+    {
+      "epoch": 2.749425287356322,
+      "grad_norm": 7.682690937441725,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 53980312,
+      "step": 598
+    },
+    {
+      "epoch": 2.749425287356322,
+      "loss": 0.12832111120224,
+      "loss_ce": 7.098779315128922e-05,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 53980312,
+      "step": 598
+    },
+    {
+      "epoch": 2.754022988505747,
+      "grad_norm": 5.73861918938428,
+      "learning_rate": 5e-06,
+      "loss": 0.1738,
+      "num_input_tokens_seen": 54070552,
+      "step": 599
+    },
+    {
+      "epoch": 2.754022988505747,
+      "loss": 0.12556979060173035,
+      "loss_ce": 8.150518988259137e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 54070552,
+      "step": 599
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "grad_norm": 6.309494256468956,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 54161016,
+      "step": 600
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "loss": 0.08496278524398804,
+      "loss_ce": 3.236144038964994e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 54161016,
+      "step": 600
+    },
+    {
+      "epoch": 2.7632183908045977,
+      "grad_norm": 8.032143628331568,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 54251360,
+      "step": 601
+    },
+    {
+      "epoch": 2.7632183908045977,
+      "loss": 0.14277827739715576,
+      "loss_ce": 1.7043152183759958e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 54251360,
+      "step": 601
+    },
+    {
+      "epoch": 2.767816091954023,
+      "grad_norm": 4.475275047608208,
+      "learning_rate": 5e-06,
+      "loss": 0.1901,
+      "num_input_tokens_seen": 54340280,
+      "step": 602
+    },
+    {
+      "epoch": 2.767816091954023,
+      "loss": 0.16928395628929138,
+      "loss_ce": 3.346416269778274e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 54340280,
+      "step": 602
+    },
+    {
+      "epoch": 2.772413793103448,
+      "grad_norm": 20.093795622997725,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 54430644,
+      "step": 603
+    },
+    {
+      "epoch": 2.772413793103448,
+      "loss": 0.24863861501216888,
+      "loss_ce": 4.241707210894674e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 54430644,
+      "step": 603
+    },
+    {
+      "epoch": 2.7770114942528736,
+      "grad_norm": 7.529800946044408,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 54521100,
+      "step": 604
+    },
+    {
+      "epoch": 2.7770114942528736,
+      "loss": 0.060331616550683975,
+      "loss_ce": 0.002012524986639619,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 54521100,
+      "step": 604
+    },
+    {
+      "epoch": 2.781609195402299,
+      "grad_norm": 4.796037523923617,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 54611424,
+      "step": 605
+    },
+    {
+      "epoch": 2.781609195402299,
+      "loss": 0.1368238478899002,
+      "loss_ce": 1.3542239685193636e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 54611424,
+      "step": 605
+    },
+    {
+      "epoch": 2.7862068965517244,
+      "grad_norm": 6.220654742397298,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 54701824,
+      "step": 606
+    },
+    {
+      "epoch": 2.7862068965517244,
+      "loss": 0.12376809120178223,
+      "loss_ce": 4.983101462130435e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 54701824,
+      "step": 606
+    },
+    {
+      "epoch": 2.7908045977011495,
+      "grad_norm": 6.281261617719392,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 54792204,
+      "step": 607
+    },
+    {
+      "epoch": 2.7908045977011495,
+      "loss": 0.13932999968528748,
+      "loss_ce": 4.776205605594441e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 54792204,
+      "step": 607
+    },
+    {
+      "epoch": 2.7954022988505747,
+      "grad_norm": 5.897586685245029,
+      "learning_rate": 5e-06,
+      "loss": 0.1316,
+      "num_input_tokens_seen": 54882476,
+      "step": 608
+    },
+    {
+      "epoch": 2.7954022988505747,
+      "loss": 0.11914113909006119,
+      "loss_ce": 0.00033620704198256135,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 54882476,
+      "step": 608
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 18.865031038984814,
+      "learning_rate": 5e-06,
+      "loss": 0.1713,
+      "num_input_tokens_seen": 54972752,
+      "step": 609
+    },
+    {
+      "epoch": 2.8,
+      "loss": 0.1230660229921341,
+      "loss_ce": 1.914523818413727e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 54972752,
+      "step": 609
+    },
+    {
+      "epoch": 2.8045977011494254,
+      "grad_norm": 12.75408299711661,
+      "learning_rate": 5e-06,
+      "loss": 0.1575,
+      "num_input_tokens_seen": 55062340,
+      "step": 610
+    },
+    {
+      "epoch": 2.8045977011494254,
+      "loss": 0.16492819786071777,
+      "loss_ce": 1.1197053026990034e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 55062340,
+      "step": 610
+    },
+    {
+      "epoch": 2.8091954022988506,
+      "grad_norm": 12.5815565888982,
+      "learning_rate": 5e-06,
+      "loss": 0.1924,
+      "num_input_tokens_seen": 55151980,
+      "step": 611
+    },
+    {
+      "epoch": 2.8091954022988506,
+      "loss": 0.19428588449954987,
+      "loss_ce": 0.006968984380364418,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 55151980,
+      "step": 611
+    },
+    {
+      "epoch": 2.8137931034482757,
+      "grad_norm": 3.291240590141606,
+      "learning_rate": 5e-06,
+      "loss": 0.1739,
+      "num_input_tokens_seen": 55242284,
+      "step": 612
+    },
+    {
+      "epoch": 2.8137931034482757,
+      "loss": 0.15410566329956055,
+      "loss_ce": 0.00022076835739426315,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 55242284,
+      "step": 612
+    },
+    {
+      "epoch": 2.8183908045977013,
+      "grad_norm": 16.71758430849643,
+      "learning_rate": 5e-06,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 55332764,
+      "step": 613
+    },
+    {
+      "epoch": 2.8183908045977013,
+      "loss": 0.21246647834777832,
+      "loss_ce": 0.00018619894399307668,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 55332764,
+      "step": 613
+    },
+    {
+      "epoch": 2.8229885057471265,
+      "grad_norm": 9.529924266178037,
+      "learning_rate": 5e-06,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 55423028,
+      "step": 614
+    },
+    {
+      "epoch": 2.8229885057471265,
+      "loss": 0.10561984032392502,
+      "loss_ce": 2.9023341994616203e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 55423028,
+      "step": 614
+    },
+    {
+      "epoch": 2.8275862068965516,
+      "grad_norm": 4.245318747599123,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 55513508,
+      "step": 615
+    },
+    {
+      "epoch": 2.8275862068965516,
+      "loss": 0.1072404682636261,
+      "loss_ce": 3.2217059924732894e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 55513508,
+      "step": 615
+    },
+    {
+      "epoch": 2.8321839080459768,
+      "grad_norm": 18.179797794436265,
+      "learning_rate": 5e-06,
+      "loss": 0.1493,
+      "num_input_tokens_seen": 55603844,
+      "step": 616
+    },
+    {
+      "epoch": 2.8321839080459768,
+      "loss": 0.15227092802524567,
+      "loss_ce": 4.924844688503072e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 55603844,
+      "step": 616
+    },
+    {
+      "epoch": 2.8367816091954023,
+      "grad_norm": 5.57072118153668,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 55693488,
+      "step": 617
+    },
+    {
+      "epoch": 2.8367816091954023,
+      "loss": 0.13283629715442657,
+      "loss_ce": 0.00017638025747146457,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 55693488,
+      "step": 617
+    },
+    {
+      "epoch": 2.8413793103448275,
+      "grad_norm": 3.7001051290248053,
+      "learning_rate": 5e-06,
+      "loss": 0.23,
+      "num_input_tokens_seen": 55783768,
+      "step": 618
+    },
+    {
+      "epoch": 2.8413793103448275,
+      "loss": 0.18909384310245514,
+      "loss_ce": 3.744277637451887e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 55783768,
+      "step": 618
+    },
+    {
+      "epoch": 2.845977011494253,
+      "grad_norm": 8.633989870151986,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 55874088,
+      "step": 619
+    },
+    {
+      "epoch": 2.845977011494253,
+      "loss": 0.16746875643730164,
+      "loss_ce": 4.932547381031327e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 55874088,
+      "step": 619
+    },
+    {
+      "epoch": 2.8505747126436782,
+      "grad_norm": 7.508204824741638,
+      "learning_rate": 5e-06,
+      "loss": 0.1742,
+      "num_input_tokens_seen": 55964516,
+      "step": 620
+    },
+    {
+      "epoch": 2.8505747126436782,
+      "loss": 0.18894179165363312,
+      "loss_ce": 0.00011428301513660699,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 55964516,
+      "step": 620
+    },
+    {
+      "epoch": 2.8551724137931034,
+      "grad_norm": 11.720485380185297,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 56054816,
+      "step": 621
+    },
+    {
+      "epoch": 2.8551724137931034,
+      "loss": 0.12343352288007736,
+      "loss_ce": 2.0439181753317825e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 56054816,
+      "step": 621
+    },
+    {
+      "epoch": 2.8597701149425285,
+      "grad_norm": 6.132110793900122,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 56145008,
+      "step": 622
+    },
+    {
+      "epoch": 2.8597701149425285,
+      "loss": 0.11308827996253967,
+      "loss_ce": 2.0656258129747584e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 56145008,
+      "step": 622
+    },
+    {
+      "epoch": 2.864367816091954,
+      "grad_norm": 27.370691194991768,
+      "learning_rate": 5e-06,
+      "loss": 0.2575,
+      "num_input_tokens_seen": 56234656,
+      "step": 623
+    },
+    {
+      "epoch": 2.864367816091954,
+      "loss": 0.2561437487602234,
+      "loss_ce": 0.0002538765547797084,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 56234656,
+      "step": 623
+    },
+    {
+      "epoch": 2.8689655172413793,
+      "grad_norm": 4.250971216520788,
+      "learning_rate": 5e-06,
+      "loss": 0.1461,
+      "num_input_tokens_seen": 56325056,
+      "step": 624
+    },
+    {
+      "epoch": 2.8689655172413793,
+      "loss": 0.15083430707454681,
+      "loss_ce": 1.643798896111548e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 56325056,
+      "step": 624
+    },
+    {
+      "epoch": 2.873563218390805,
+      "grad_norm": 23.371099875991142,
+      "learning_rate": 5e-06,
+      "loss": 0.1696,
+      "num_input_tokens_seen": 56415548,
+      "step": 625
+    },
+    {
+      "epoch": 2.873563218390805,
+      "loss": 0.2244938164949417,
+      "loss_ce": 0.00018961615569423884,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 56415548,
+      "step": 625
+    },
+    {
+      "epoch": 2.87816091954023,
+      "grad_norm": 19.44155033716697,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 56505852,
+      "step": 626
+    },
+    {
+      "epoch": 2.87816091954023,
+      "loss": 0.12483800947666168,
+      "loss_ce": 0.008840695023536682,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 56505852,
+      "step": 626
+    },
+    {
+      "epoch": 2.882758620689655,
+      "grad_norm": 2.7000059987309135,
+      "learning_rate": 5e-06,
+      "loss": 0.132,
+      "num_input_tokens_seen": 56596200,
+      "step": 627
+    },
+    {
+      "epoch": 2.882758620689655,
+      "loss": 0.16553373634815216,
+      "loss_ce": 3.6913388612447307e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 56596200,
+      "step": 627
+    },
+    {
+      "epoch": 2.8873563218390803,
+      "grad_norm": 11.738060980857469,
+      "learning_rate": 5e-06,
+      "loss": 0.1943,
+      "num_input_tokens_seen": 56686468,
+      "step": 628
+    },
+    {
+      "epoch": 2.8873563218390803,
+      "loss": 0.13238102197647095,
+      "loss_ce": 0.00013309504720382392,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 56686468,
+      "step": 628
+    },
+    {
+      "epoch": 2.891954022988506,
+      "grad_norm": 5.374412860929261,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 56776724,
+      "step": 629
+    },
+    {
+      "epoch": 2.891954022988506,
+      "loss": 0.10836595296859741,
+      "loss_ce": 0.00018113740952685475,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 56776724,
+      "step": 629
+    },
+    {
+      "epoch": 2.896551724137931,
+      "grad_norm": 17.458934517040042,
+      "learning_rate": 5e-06,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 56867076,
+      "step": 630
+    },
+    {
+      "epoch": 2.896551724137931,
+      "loss": 0.0917268842458725,
+      "loss_ce": 8.260130562121049e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 56867076,
+      "step": 630
+    },
+    {
+      "epoch": 2.901149425287356,
+      "grad_norm": 7.144445161490093,
+      "learning_rate": 5e-06,
+      "loss": 0.1952,
+      "num_input_tokens_seen": 56957452,
+      "step": 631
+    },
+    {
+      "epoch": 2.901149425287356,
+      "loss": 0.17814010381698608,
+      "loss_ce": 3.951931284973398e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 56957452,
+      "step": 631
+    },
+    {
+      "epoch": 2.905747126436782,
+      "grad_norm": 4.881468932307776,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 57046320,
+      "step": 632
+    },
+    {
+      "epoch": 2.905747126436782,
+      "loss": 0.158255934715271,
+      "loss_ce": 8.333477308042347e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 57046320,
+      "step": 632
+    },
+    {
+      "epoch": 2.910344827586207,
+      "grad_norm": 6.020225715103822,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 57136760,
+      "step": 633
+    },
+    {
+      "epoch": 2.910344827586207,
+      "loss": 0.08915083855390549,
+      "loss_ce": 0.0001921012590173632,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 57136760,
+      "step": 633
+    },
+    {
+      "epoch": 2.914942528735632,
+      "grad_norm": 18.893533969376247,
+      "learning_rate": 5e-06,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 57227264,
+      "step": 634
+    },
+    {
+      "epoch": 2.914942528735632,
+      "loss": 0.0713462084531784,
+      "loss_ce": 8.766089740674943e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 57227264,
+      "step": 634
+    },
+    {
+      "epoch": 2.9195402298850572,
+      "grad_norm": 18.876600553359836,
+      "learning_rate": 5e-06,
+      "loss": 0.1505,
+      "num_input_tokens_seen": 57316096,
+      "step": 635
+    },
+    {
+      "epoch": 2.9195402298850572,
+      "loss": 0.14804166555404663,
+      "loss_ce": 0.0002297760802321136,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 57316096,
+      "step": 635
+    },
+    {
+      "epoch": 2.924137931034483,
+      "grad_norm": 14.779947666699496,
+      "learning_rate": 5e-06,
+      "loss": 0.1474,
+      "num_input_tokens_seen": 57406464,
+      "step": 636
+    },
+    {
+      "epoch": 2.924137931034483,
+      "loss": 0.1926122009754181,
+      "loss_ce": 1.5764764611958526e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 57406464,
+      "step": 636
+    },
+    {
+      "epoch": 2.928735632183908,
+      "grad_norm": 33.4524141554006,
+      "learning_rate": 5e-06,
+      "loss": 0.1697,
+      "num_input_tokens_seen": 57496896,
+      "step": 637
+    },
+    {
+      "epoch": 2.928735632183908,
+      "loss": 0.19599446654319763,
+      "loss_ce": 7.161433313740417e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 57496896,
+      "step": 637
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 8.837724303528487,
+      "learning_rate": 5e-06,
+      "loss": 0.1276,
+      "num_input_tokens_seen": 57587328,
+      "step": 638
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "loss": 0.11598608642816544,
+      "loss_ce": 8.032606274355203e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 57587328,
+      "step": 638
+    },
+    {
+      "epoch": 2.9379310344827587,
+      "grad_norm": 5.458845083322578,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 57676000,
+      "step": 639
+    },
+    {
+      "epoch": 2.9379310344827587,
+      "loss": 0.1924465298652649,
+      "loss_ce": 6.370765186147764e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 57676000,
+      "step": 639
+    },
+    {
+      "epoch": 2.942528735632184,
+      "grad_norm": 3.502931679203287,
+      "learning_rate": 5e-06,
+      "loss": 0.1517,
+      "num_input_tokens_seen": 57766436,
+      "step": 640
+    },
+    {
+      "epoch": 2.942528735632184,
+      "loss": 0.22913488745689392,
+      "loss_ce": 0.00022254293435253203,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 57766436,
+      "step": 640
+    },
+    {
+      "epoch": 2.947126436781609,
+      "grad_norm": 27.45097976056178,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 57856748,
+      "step": 641
+    },
+    {
+      "epoch": 2.947126436781609,
+      "loss": 0.1526871919631958,
+      "loss_ce": 9.930751548381522e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 57856748,
+      "step": 641
+    },
+    {
+      "epoch": 2.9517241379310346,
+      "grad_norm": 12.914503955582164,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 57946960,
+      "step": 642
+    },
+    {
+      "epoch": 2.9517241379310346,
+      "loss": 0.17462590336799622,
+      "loss_ce": 6.536281580338255e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 57946960,
+      "step": 642
+    },
+    {
+      "epoch": 2.9563218390804598,
+      "grad_norm": 11.39446414052775,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 58035688,
+      "step": 643
+    },
+    {
+      "epoch": 2.9563218390804598,
+      "loss": 0.11201904714107513,
+      "loss_ce": 8.057546801865101e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 58035688,
+      "step": 643
+    },
+    {
+      "epoch": 2.960919540229885,
+      "grad_norm": 3.1615458390497326,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 58126192,
+      "step": 644
+    },
+    {
+      "epoch": 2.960919540229885,
+      "loss": 0.11491385102272034,
+      "loss_ce": 0.00010671426571207121,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 58126192,
+      "step": 644
+    },
+    {
+      "epoch": 2.9655172413793105,
+      "grad_norm": 11.115986632833536,
+      "learning_rate": 5e-06,
+      "loss": 0.1975,
+      "num_input_tokens_seen": 58216656,
+      "step": 645
+    },
+    {
+      "epoch": 2.9655172413793105,
+      "loss": 0.18091078102588654,
+      "loss_ce": 0.00015516624262090772,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 58216656,
+      "step": 645
+    },
+    {
+      "epoch": 2.9701149425287356,
+      "grad_norm": 10.595330403567505,
+      "learning_rate": 5e-06,
+      "loss": 0.1393,
+      "num_input_tokens_seen": 58306940,
+      "step": 646
+    },
+    {
+      "epoch": 2.9701149425287356,
+      "loss": 0.18637780845165253,
+      "loss_ce": 3.746678703464568e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 58306940,
+      "step": 646
+    },
+    {
+      "epoch": 2.974712643678161,
+      "grad_norm": 12.224220904653137,
+      "learning_rate": 5e-06,
+      "loss": 0.1602,
+      "num_input_tokens_seen": 58397184,
+      "step": 647
+    },
+    {
+      "epoch": 2.974712643678161,
+      "loss": 0.1820899099111557,
+      "loss_ce": 2.2039628674974665e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 58397184,
+      "step": 647
+    },
+    {
+      "epoch": 2.979310344827586,
+      "grad_norm": 12.36377678211748,
+      "learning_rate": 5e-06,
+      "loss": 0.1605,
+      "num_input_tokens_seen": 58487644,
+      "step": 648
+    },
+    {
+      "epoch": 2.979310344827586,
+      "loss": 0.09505946934223175,
+      "loss_ce": 4.29864194302354e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 58487644,
+      "step": 648
+    },
+    {
+      "epoch": 2.9839080459770115,
+      "grad_norm": 44.77338962392459,
+      "learning_rate": 5e-06,
+      "loss": 0.1624,
+      "num_input_tokens_seen": 58578064,
+      "step": 649
+    },
+    {
+      "epoch": 2.9839080459770115,
+      "loss": 0.19933252036571503,
+      "loss_ce": 0.00029687874484807253,
+      "loss_iou": 0.4375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 58578064,
+      "step": 649
+    },
+    {
+      "epoch": 2.9885057471264367,
+      "grad_norm": 10.434155317304565,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 58668520,
+      "step": 650
+    },
+    {
+      "epoch": 2.9885057471264367,
+      "loss": 0.17232680320739746,
+      "loss_ce": 2.456527727190405e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 58668520,
+      "step": 650
+    },
+    {
+      "epoch": 2.9931034482758623,
+      "grad_norm": 11.343545620466347,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 58758840,
+      "step": 651
+    },
+    {
+      "epoch": 2.9931034482758623,
+      "loss": 0.17710255086421967,
+      "loss_ce": 2.4299104552483186e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 58758840,
+      "step": 651
+    },
+    {
+      "epoch": 2.9977011494252874,
+      "grad_norm": 4.426473466923714,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 58849212,
+      "step": 652
+    },
+    {
+      "epoch": 2.9977011494252874,
+      "loss": 0.1025293692946434,
+      "loss_ce": 0.00014289778482634574,
+      "loss_iou": 0.375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 58849212,
+      "step": 652
+    },
+    {
+      "epoch": 2.9977011494252874,
+      "loss": 0.18885542452335358,
+      "loss_ce": 1.2649360542127397e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 58894312,
+      "step": 652
+    },
+    {
+      "epoch": 3.0022988505747126,
+      "grad_norm": 5.109594045617307,
+      "learning_rate": 5e-06,
+      "loss": 0.1499,
+      "num_input_tokens_seen": 58939548,
+      "step": 653
+    },
+    {
+      "epoch": 3.0022988505747126,
+      "loss": 0.11101669818162918,
+      "loss_ce": 5.478527600644156e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 58939548,
+      "step": 653
+    },
+    {
+      "epoch": 3.0068965517241377,
+      "grad_norm": 10.518496135733074,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 59029104,
+      "step": 654
+    },
+    {
+      "epoch": 3.0068965517241377,
+      "loss": 0.15445493161678314,
+      "loss_ce": 0.00015805772272869945,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 59029104,
+      "step": 654
+    },
+    {
+      "epoch": 3.0114942528735633,
+      "grad_norm": 14.317668018009412,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 59119440,
+      "step": 655
+    },
+    {
+      "epoch": 3.0114942528735633,
+      "loss": 0.1157938614487648,
+      "loss_ce": 1.0168273547606077e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 59119440,
+      "step": 655
+    },
+    {
+      "epoch": 3.0160919540229885,
+      "grad_norm": 6.462788772652226,
+      "learning_rate": 5e-06,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 59209864,
+      "step": 656
+    },
+    {
+      "epoch": 3.0160919540229885,
+      "loss": 0.06505613029003143,
+      "loss_ce": 0.00011471970356069505,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 59209864,
+      "step": 656
+    },
+    {
+      "epoch": 3.0206896551724136,
+      "grad_norm": 5.315176931861487,
+      "learning_rate": 5e-06,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 59300316,
+      "step": 657
+    },
+    {
+      "epoch": 3.0206896551724136,
+      "loss": 0.064435213804245,
+      "loss_ce": 0.002515045227482915,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 59300316,
+      "step": 657
+    },
+    {
+      "epoch": 3.025287356321839,
+      "grad_norm": 4.182380926387102,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 59390752,
+      "step": 658
+    },
+    {
+      "epoch": 3.025287356321839,
+      "loss": 0.13093848526477814,
+      "loss_ce": 1.8076902051689103e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 59390752,
+      "step": 658
+    },
+    {
+      "epoch": 3.0298850574712644,
+      "grad_norm": 14.571610617843975,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 59481036,
+      "step": 659
+    },
+    {
+      "epoch": 3.0298850574712644,
+      "loss": 0.08019035309553146,
+      "loss_ce": 8.17138425190933e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 59481036,
+      "step": 659
+    },
+    {
+      "epoch": 3.0344827586206895,
+      "grad_norm": 16.65916183345602,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 59571308,
+      "step": 660
+    },
+    {
+      "epoch": 3.0344827586206895,
+      "loss": 0.15776613354682922,
+      "loss_ce": 5.127684198669158e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 59571308,
+      "step": 660
+    },
+    {
+      "epoch": 3.039080459770115,
+      "grad_norm": 3.372941981569247,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 59661712,
+      "step": 661
+    },
+    {
+      "epoch": 3.039080459770115,
+      "loss": 0.0939127653837204,
+      "loss_ce": 4.0693001210456714e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 59661712,
+      "step": 661
+    },
+    {
+      "epoch": 3.0436781609195402,
+      "grad_norm": 7.8989890189247,
+      "learning_rate": 5e-06,
+      "loss": 0.1298,
+      "num_input_tokens_seen": 59752088,
+      "step": 662
+    },
+    {
+      "epoch": 3.0436781609195402,
+      "loss": 0.11880500614643097,
+      "loss_ce": 3.058691436308436e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 59752088,
+      "step": 662
+    },
+    {
+      "epoch": 3.0482758620689654,
+      "grad_norm": 9.966522515655097,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 59842376,
+      "step": 663
+    },
+    {
+      "epoch": 3.0482758620689654,
+      "loss": 0.1415707916021347,
+      "loss_ce": 3.027227103302721e-05,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 59842376,
+      "step": 663
+    },
+    {
+      "epoch": 3.052873563218391,
+      "grad_norm": 14.738658620476503,
+      "learning_rate": 5e-06,
+      "loss": 0.1078,
+      "num_input_tokens_seen": 59932932,
+      "step": 664
+    },
+    {
+      "epoch": 3.052873563218391,
+      "loss": 0.15776753425598145,
+      "loss_ce": 2.216372377006337e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 59932932,
+      "step": 664
+    },
+    {
+      "epoch": 3.057471264367816,
+      "grad_norm": 11.861563289243291,
+      "learning_rate": 5e-06,
+      "loss": 0.161,
+      "num_input_tokens_seen": 60023360,
+      "step": 665
+    },
+    {
+      "epoch": 3.057471264367816,
+      "loss": 0.15695565938949585,
+      "loss_ce": 3.427857154747471e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 60023360,
+      "step": 665
+    },
+    {
+      "epoch": 3.0620689655172413,
+      "grad_norm": 8.820370554295812,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 60112940,
+      "step": 666
+    },
+    {
+      "epoch": 3.0620689655172413,
+      "loss": 0.13253626227378845,
+      "loss_ce": 8.997396798804402e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 60112940,
+      "step": 666
+    },
+    {
+      "epoch": 3.066666666666667,
+      "grad_norm": 3.1489432695249526,
+      "learning_rate": 5e-06,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 60203192,
+      "step": 667
+    },
+    {
+      "epoch": 3.066666666666667,
+      "loss": 0.14637088775634766,
+      "loss_ce": 8.589012395532336e-06,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 60203192,
+      "step": 667
+    },
+    {
+      "epoch": 3.071264367816092,
+      "grad_norm": 5.702383945793617,
+      "learning_rate": 5e-06,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 60293508,
+      "step": 668
+    },
+    {
+      "epoch": 3.071264367816092,
+      "loss": 0.09078279137611389,
+      "loss_ce": 2.3513095584348775e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 60293508,
+      "step": 668
+    },
+    {
+      "epoch": 3.075862068965517,
+      "grad_norm": 21.441294074851363,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 60383820,
+      "step": 669
+    },
+    {
+      "epoch": 3.075862068965517,
+      "loss": 0.10486976057291031,
+      "loss_ce": 1.1363279554643668e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 60383820,
+      "step": 669
+    },
+    {
+      "epoch": 3.0804597701149423,
+      "grad_norm": 3.914442501475495,
+      "learning_rate": 5e-06,
+      "loss": 0.193,
+      "num_input_tokens_seen": 60474148,
+      "step": 670
+    },
+    {
+      "epoch": 3.0804597701149423,
+      "loss": 0.16388007998466492,
+      "loss_ce": 1.594252353243064e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 60474148,
+      "step": 670
+    },
+    {
+      "epoch": 3.085057471264368,
+      "grad_norm": 4.9620660179246885,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 60564532,
+      "step": 671
+    },
+    {
+      "epoch": 3.085057471264368,
+      "loss": 0.18953979015350342,
+      "loss_ce": 8.66625050548464e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 60564532,
+      "step": 671
+    },
+    {
+      "epoch": 3.089655172413793,
+      "grad_norm": 16.0733034397525,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 60654828,
+      "step": 672
+    },
+    {
+      "epoch": 3.089655172413793,
+      "loss": 0.14448416233062744,
+      "loss_ce": 2.920163024100475e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 60654828,
+      "step": 672
+    },
+    {
+      "epoch": 3.094252873563218,
+      "grad_norm": 2.850437311995024,
+      "learning_rate": 5e-06,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 60745316,
+      "step": 673
+    },
+    {
+      "epoch": 3.094252873563218,
+      "loss": 0.08657945692539215,
+      "loss_ce": 0.00012315809726715088,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 60745316,
+      "step": 673
+    },
+    {
+      "epoch": 3.098850574712644,
+      "grad_norm": 6.711572483738216,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 60835616,
+      "step": 674
+    },
+    {
+      "epoch": 3.098850574712644,
+      "loss": 0.11418451368808746,
+      "loss_ce": 1.8256445400766097e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 60835616,
+      "step": 674
+    },
+    {
+      "epoch": 3.103448275862069,
+      "grad_norm": 2.9732106658271267,
+      "learning_rate": 5e-06,
+      "loss": 0.2022,
+      "num_input_tokens_seen": 60926020,
+      "step": 675
+    },
+    {
+      "epoch": 3.103448275862069,
+      "loss": 0.22187387943267822,
+      "loss_ce": 0.0012928310316056013,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 60926020,
+      "step": 675
+    },
+    {
+      "epoch": 3.108045977011494,
+      "grad_norm": 22.393897827054342,
+      "learning_rate": 5e-06,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 61016500,
+      "step": 676
+    },
+    {
+      "epoch": 3.108045977011494,
+      "loss": 0.17077524960041046,
+      "loss_ce": 0.0004261271096765995,
+      "loss_iou": 0.34375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 61016500,
+      "step": 676
+    },
+    {
+      "epoch": 3.1126436781609197,
+      "grad_norm": 4.958182554412022,
+      "learning_rate": 5e-06,
+      "loss": 0.1796,
+      "num_input_tokens_seen": 61106876,
+      "step": 677
+    },
+    {
+      "epoch": 3.1126436781609197,
+      "loss": 0.17057394981384277,
+      "loss_ce": 8.749846165301278e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 61106876,
+      "step": 677
+    },
+    {
+      "epoch": 3.117241379310345,
+      "grad_norm": 5.763483977522711,
+      "learning_rate": 5e-06,
+      "loss": 0.1333,
+      "num_input_tokens_seen": 61197312,
+      "step": 678
+    },
+    {
+      "epoch": 3.117241379310345,
+      "loss": 0.16354021430015564,
+      "loss_ce": 0.00014910995378158987,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 61197312,
+      "step": 678
+    },
+    {
+      "epoch": 3.12183908045977,
+      "grad_norm": 22.61759363539356,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 61287748,
+      "step": 679
+    },
+    {
+      "epoch": 3.12183908045977,
+      "loss": 0.16320019960403442,
+      "loss_ce": 8.373754099011421e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 61287748,
+      "step": 679
+    },
+    {
+      "epoch": 3.1264367816091956,
+      "grad_norm": 21.37560854861523,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 61378228,
+      "step": 680
+    },
+    {
+      "epoch": 3.1264367816091956,
+      "loss": 0.1652674674987793,
+      "loss_ce": 4.528817953541875e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 61378228,
+      "step": 680
+    },
+    {
+      "epoch": 3.1310344827586207,
+      "grad_norm": 9.912277304010486,
+      "learning_rate": 5e-06,
+      "loss": 0.168,
+      "num_input_tokens_seen": 61468564,
+      "step": 681
+    },
+    {
+      "epoch": 3.1310344827586207,
+      "loss": 0.14338386058807373,
+      "loss_ce": 7.332136738114059e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 61468564,
+      "step": 681
+    },
+    {
+      "epoch": 3.135632183908046,
+      "grad_norm": 14.24831883027225,
+      "learning_rate": 5e-06,
+      "loss": 0.1808,
+      "num_input_tokens_seen": 61559020,
+      "step": 682
+    },
+    {
+      "epoch": 3.135632183908046,
+      "loss": 0.15874823927879333,
+      "loss_ce": 0.000117871692054905,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 61559020,
+      "step": 682
+    },
+    {
+      "epoch": 3.1402298850574715,
+      "grad_norm": 10.805661680356696,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 61649428,
+      "step": 683
+    },
+    {
+      "epoch": 3.1402298850574715,
+      "loss": 0.11267475038766861,
+      "loss_ce": 8.014314516913146e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 61649428,
+      "step": 683
+    },
+    {
+      "epoch": 3.1448275862068966,
+      "grad_norm": 10.580707465651063,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 61739904,
+      "step": 684
+    },
+    {
+      "epoch": 3.1448275862068966,
+      "loss": 0.11008670926094055,
+      "loss_ce": 2.5071400159504265e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 61739904,
+      "step": 684
+    },
+    {
+      "epoch": 3.1494252873563218,
+      "grad_norm": 9.787018998948723,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 61830168,
+      "step": 685
+    },
+    {
+      "epoch": 3.1494252873563218,
+      "loss": 0.11129388213157654,
+      "loss_ce": 7.25619393051602e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 61830168,
+      "step": 685
+    },
+    {
+      "epoch": 3.154022988505747,
+      "grad_norm": 25.016543912767162,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 61920580,
+      "step": 686
+    },
+    {
+      "epoch": 3.154022988505747,
+      "loss": 0.11655561625957489,
+      "loss_ce": 8.985316526377574e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 61920580,
+      "step": 686
+    },
+    {
+      "epoch": 3.1586206896551725,
+      "grad_norm": 27.789033510951985,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 62010984,
+      "step": 687
+    },
+    {
+      "epoch": 3.1586206896551725,
+      "loss": 0.10948251187801361,
+      "loss_ce": 1.596645415702369e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 62010984,
+      "step": 687
+    },
+    {
+      "epoch": 3.1632183908045977,
+      "grad_norm": 3.436728324433976,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 62101364,
+      "step": 688
+    },
+    {
+      "epoch": 3.1632183908045977,
+      "loss": 0.12181434035301208,
+      "loss_ce": 0.0001102387614082545,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 62101364,
+      "step": 688
+    },
+    {
+      "epoch": 3.167816091954023,
+      "grad_norm": 9.993289616742318,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 62190176,
+      "step": 689
+    },
+    {
+      "epoch": 3.167816091954023,
+      "loss": 0.14744696021080017,
+      "loss_ce": 0.003327697515487671,
+      "loss_iou": 0.375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 62190176,
+      "step": 689
+    },
+    {
+      "epoch": 3.1724137931034484,
+      "grad_norm": 12.827182492237142,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 62280480,
+      "step": 690
+    },
+    {
+      "epoch": 3.1724137931034484,
+      "loss": 0.15000127255916595,
+      "loss_ce": 3.789434413192794e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 62280480,
+      "step": 690
+    },
+    {
+      "epoch": 3.1770114942528735,
+      "grad_norm": 2.7933249990009013,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 62370884,
+      "step": 691
+    },
+    {
+      "epoch": 3.1770114942528735,
+      "loss": 0.13926616311073303,
+      "loss_ce": 0.00022808580251876265,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 62370884,
+      "step": 691
+    },
+    {
+      "epoch": 3.1816091954022987,
+      "grad_norm": 3.400237110893925,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 62461280,
+      "step": 692
+    },
+    {
+      "epoch": 3.1816091954022987,
+      "loss": 0.07589490711688995,
+      "loss_ce": 5.872425026609562e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 62461280,
+      "step": 692
+    },
+    {
+      "epoch": 3.1862068965517243,
+      "grad_norm": 10.582817694578983,
+      "learning_rate": 5e-06,
+      "loss": 0.137,
+      "num_input_tokens_seen": 62551580,
+      "step": 693
+    },
+    {
+      "epoch": 3.1862068965517243,
+      "loss": 0.14650990068912506,
+      "loss_ce": 2.5522436772007495e-05,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 62551580,
+      "step": 693
+    },
+    {
+      "epoch": 3.1908045977011494,
+      "grad_norm": 13.951673209631243,
+      "learning_rate": 5e-06,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 62641896,
+      "step": 694
+    },
+    {
+      "epoch": 3.1908045977011494,
+      "loss": 0.10015565156936646,
+      "loss_ce": 0.00011902584810741246,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 62641896,
+      "step": 694
+    },
+    {
+      "epoch": 3.1954022988505746,
+      "grad_norm": 8.105786030944435,
+      "learning_rate": 5e-06,
+      "loss": 0.171,
+      "num_input_tokens_seen": 62731508,
+      "step": 695
+    },
+    {
+      "epoch": 3.1954022988505746,
+      "loss": 0.2043384611606598,
+      "loss_ce": 0.0001148240189650096,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 62731508,
+      "step": 695
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 29.137424088774004,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 62821956,
+      "step": 696
+    },
+    {
+      "epoch": 3.2,
+      "loss": 0.13569773733615875,
+      "loss_ce": 7.761328015476465e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 62821956,
+      "step": 696
+    },
+    {
+      "epoch": 3.2045977011494253,
+      "grad_norm": 15.53880122139471,
+      "learning_rate": 5e-06,
+      "loss": 0.1688,
+      "num_input_tokens_seen": 62912308,
+      "step": 697
+    },
+    {
+      "epoch": 3.2045977011494253,
+      "loss": 0.1253717541694641,
+      "loss_ce": 8.183569298125803e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 62912308,
+      "step": 697
+    },
+    {
+      "epoch": 3.2091954022988505,
+      "grad_norm": 10.527367140209382,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 63002620,
+      "step": 698
+    },
+    {
+      "epoch": 3.2091954022988505,
+      "loss": 0.1609758883714676,
+      "loss_ce": 5.669322854373604e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 63002620,
+      "step": 698
+    },
+    {
+      "epoch": 3.213793103448276,
+      "grad_norm": 18.955512801085465,
+      "learning_rate": 5e-06,
+      "loss": 0.0993,
+      "num_input_tokens_seen": 63093068,
+      "step": 699
+    },
+    {
+      "epoch": 3.213793103448276,
+      "loss": 0.09753237664699554,
+      "loss_ce": 2.871520700864494e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 63093068,
+      "step": 699
+    },
+    {
+      "epoch": 3.218390804597701,
+      "grad_norm": 19.835239072673776,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 63183464,
+      "step": 700
+    },
+    {
+      "epoch": 3.218390804597701,
+      "loss": 0.14680443704128265,
+      "loss_ce": 0.00010644025314832106,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 63183464,
+      "step": 700
+    },
+    {
+      "epoch": 3.2229885057471264,
+      "grad_norm": 8.057607073965688,
+      "learning_rate": 5e-06,
+      "loss": 0.1078,
+      "num_input_tokens_seen": 63273704,
+      "step": 701
+    },
+    {
+      "epoch": 3.2229885057471264,
+      "loss": 0.10005685687065125,
+      "loss_ce": 1.2611541023943573e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 63273704,
+      "step": 701
+    },
+    {
+      "epoch": 3.227586206896552,
+      "grad_norm": 8.719404497052812,
+      "learning_rate": 5e-06,
+      "loss": 0.1474,
+      "num_input_tokens_seen": 63363992,
+      "step": 702
+    },
+    {
+      "epoch": 3.227586206896552,
+      "loss": 0.12116096913814545,
+      "loss_ce": 3.67055217793677e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 63363992,
+      "step": 702
+    },
+    {
+      "epoch": 3.232183908045977,
+      "grad_norm": 11.659344031825642,
+      "learning_rate": 5e-06,
+      "loss": 0.1215,
+      "num_input_tokens_seen": 63454440,
+      "step": 703
+    },
+    {
+      "epoch": 3.232183908045977,
+      "loss": 0.0965924859046936,
+      "loss_ce": 1.9609713490353897e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 63454440,
+      "step": 703
+    },
+    {
+      "epoch": 3.2367816091954023,
+      "grad_norm": 8.032198591188585,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 63544912,
+      "step": 704
+    },
+    {
+      "epoch": 3.2367816091954023,
+      "loss": 0.1306506246328354,
+      "loss_ce": 6.59109718981199e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 63544912,
+      "step": 704
+    },
+    {
+      "epoch": 3.2413793103448274,
+      "grad_norm": 6.297596133480101,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 63634348,
+      "step": 705
+    },
+    {
+      "epoch": 3.2413793103448274,
+      "loss": 0.19925595819950104,
+      "loss_ce": 9.824423614190891e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 63634348,
+      "step": 705
+    },
+    {
+      "epoch": 3.245977011494253,
+      "grad_norm": 34.35621373849989,
+      "learning_rate": 5e-06,
+      "loss": 0.2007,
+      "num_input_tokens_seen": 63724764,
+      "step": 706
+    },
+    {
+      "epoch": 3.245977011494253,
+      "loss": 0.22871080040931702,
+      "loss_ce": 0.0001493933523306623,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 63724764,
+      "step": 706
+    },
+    {
+      "epoch": 3.250574712643678,
+      "grad_norm": 11.328882741656939,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 63815068,
+      "step": 707
+    },
+    {
+      "epoch": 3.250574712643678,
+      "loss": 0.12251278012990952,
+      "loss_ce": 4.574016566039063e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 63815068,
+      "step": 707
+    },
+    {
+      "epoch": 3.2551724137931033,
+      "grad_norm": 2.795600172154129,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 63905392,
+      "step": 708
+    },
+    {
+      "epoch": 3.2551724137931033,
+      "loss": 0.1475527286529541,
+      "loss_ce": 0.00013756597763858736,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 63905392,
+      "step": 708
+    },
+    {
+      "epoch": 3.259770114942529,
+      "grad_norm": 6.617540471617317,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 63995760,
+      "step": 709
+    },
+    {
+      "epoch": 3.259770114942529,
+      "loss": 0.1187780499458313,
+      "loss_ce": 0.00021725612168665975,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 63995760,
+      "step": 709
+    },
+    {
+      "epoch": 3.264367816091954,
+      "grad_norm": 3.758683889886017,
+      "learning_rate": 5e-06,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 64086168,
+      "step": 710
+    },
+    {
+      "epoch": 3.264367816091954,
+      "loss": 0.12687164545059204,
+      "loss_ce": 0.0001016306850942783,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 64086168,
+      "step": 710
+    },
+    {
+      "epoch": 3.268965517241379,
+      "grad_norm": 7.276014322969384,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 64174996,
+      "step": 711
+    },
+    {
+      "epoch": 3.268965517241379,
+      "loss": 0.10789632797241211,
+      "loss_ce": 1.669170342211146e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 64174996,
+      "step": 711
+    },
+    {
+      "epoch": 3.2735632183908048,
+      "grad_norm": 9.531874918868848,
+      "learning_rate": 5e-06,
+      "loss": 0.1484,
+      "num_input_tokens_seen": 64265440,
+      "step": 712
+    },
+    {
+      "epoch": 3.2735632183908048,
+      "loss": 0.14500431716442108,
+      "loss_ce": 4.581706161843613e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 64265440,
+      "step": 712
+    },
+    {
+      "epoch": 3.27816091954023,
+      "grad_norm": 4.769697015733171,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 64355800,
+      "step": 713
+    },
+    {
+      "epoch": 3.27816091954023,
+      "loss": 0.1640397310256958,
+      "loss_ce": 6.877434498164803e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 64355800,
+      "step": 713
+    },
+    {
+      "epoch": 3.282758620689655,
+      "grad_norm": 10.346574150082782,
+      "learning_rate": 5e-06,
+      "loss": 0.147,
+      "num_input_tokens_seen": 64446020,
+      "step": 714
+    },
+    {
+      "epoch": 3.282758620689655,
+      "loss": 0.13765498995780945,
+      "loss_ce": 2.0713385310955346e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 64446020,
+      "step": 714
+    },
+    {
+      "epoch": 3.2873563218390807,
+      "grad_norm": 3.797921190886921,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 64536336,
+      "step": 715
+    },
+    {
+      "epoch": 3.2873563218390807,
+      "loss": 0.11108782887458801,
+      "loss_ce": 6.487606151495129e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 64536336,
+      "step": 715
+    },
+    {
+      "epoch": 3.291954022988506,
+      "grad_norm": 7.680464395220625,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 64626672,
+      "step": 716
+    },
+    {
+      "epoch": 3.291954022988506,
+      "loss": 0.0874263346195221,
+      "loss_ce": 8.502809214405715e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 64626672,
+      "step": 716
+    },
+    {
+      "epoch": 3.296551724137931,
+      "grad_norm": 29.00223362689999,
+      "learning_rate": 5e-06,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 64717188,
+      "step": 717
+    },
+    {
+      "epoch": 3.296551724137931,
+      "loss": 0.09892787039279938,
+      "loss_ce": 2.0401252186275087e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 64717188,
+      "step": 717
+    },
+    {
+      "epoch": 3.301149425287356,
+      "grad_norm": 2.963145490619517,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 64806948,
+      "step": 718
+    },
+    {
+      "epoch": 3.301149425287356,
+      "loss": 0.13925355672836304,
+      "loss_ce": 1.7103782738558948e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 64806948,
+      "step": 718
+    },
+    {
+      "epoch": 3.3057471264367817,
+      "grad_norm": 3.512627488510696,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 64897416,
+      "step": 719
+    },
+    {
+      "epoch": 3.3057471264367817,
+      "loss": 0.09657322615385056,
+      "loss_ce": 1.5609241017955355e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 64897416,
+      "step": 719
+    },
+    {
+      "epoch": 3.310344827586207,
+      "grad_norm": 9.222756441936497,
+      "learning_rate": 5e-06,
+      "loss": 0.1217,
+      "num_input_tokens_seen": 64986932,
+      "step": 720
+    },
+    {
+      "epoch": 3.310344827586207,
+      "loss": 0.11704221367835999,
+      "loss_ce": 3.782210478675552e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 64986932,
+      "step": 720
+    },
+    {
+      "epoch": 3.314942528735632,
+      "grad_norm": 9.715496271101662,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 65077284,
+      "step": 721
+    },
+    {
+      "epoch": 3.314942528735632,
+      "loss": 0.09352642297744751,
+      "loss_ce": 9.685942495707422e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 65077284,
+      "step": 721
+    },
+    {
+      "epoch": 3.3195402298850576,
+      "grad_norm": 5.296050089954301,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 65167528,
+      "step": 722
+    },
+    {
+      "epoch": 3.3195402298850576,
+      "loss": 0.19446220993995667,
+      "loss_ce": 4.202499439998064e-06,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 65167528,
+      "step": 722
+    },
+    {
+      "epoch": 3.3241379310344827,
+      "grad_norm": 4.298113327898983,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 65257968,
+      "step": 723
+    },
+    {
+      "epoch": 3.3241379310344827,
+      "loss": 0.12031020224094391,
+      "loss_ce": 9.91147771856049e-06,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 65257968,
+      "step": 723
+    },
+    {
+      "epoch": 3.328735632183908,
+      "grad_norm": 5.779893006341727,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 65348448,
+      "step": 724
+    },
+    {
+      "epoch": 3.328735632183908,
+      "loss": 0.19137105345726013,
+      "loss_ce": 5.6363111070822924e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 65348448,
+      "step": 724
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 22.187024011658394,
+      "learning_rate": 5e-06,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 65438876,
+      "step": 725
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "loss": 0.06496837735176086,
+      "loss_ce": 2.6973326384904794e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 65438876,
+      "step": 725
+    },
+    {
+      "epoch": 3.3379310344827586,
+      "grad_norm": 12.062384100708867,
+      "learning_rate": 5e-06,
+      "loss": 0.1188,
+      "num_input_tokens_seen": 65529280,
+      "step": 726
+    },
+    {
+      "epoch": 3.3379310344827586,
+      "loss": 0.1274305284023285,
+      "loss_ce": 1.9637809600681067e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 65529280,
+      "step": 726
+    },
+    {
+      "epoch": 3.342528735632184,
+      "grad_norm": 3.5029991469109394,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 65619760,
+      "step": 727
+    },
+    {
+      "epoch": 3.342528735632184,
+      "loss": 0.11032354086637497,
+      "loss_ce": 1.7757560272002593e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 65619760,
+      "step": 727
+    },
+    {
+      "epoch": 3.3471264367816094,
+      "grad_norm": 4.52005838961759,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 65710148,
+      "step": 728
+    },
+    {
+      "epoch": 3.3471264367816094,
+      "loss": 0.2110876739025116,
+      "loss_ce": 0.00016542623052373528,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 65710148,
+      "step": 728
+    },
+    {
+      "epoch": 3.3517241379310345,
+      "grad_norm": 6.440524482881178,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 65800676,
+      "step": 729
+    },
+    {
+      "epoch": 3.3517241379310345,
+      "loss": 0.10761566460132599,
+      "loss_ce": 1.0684560038498603e-05,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 65800676,
+      "step": 729
+    },
+    {
+      "epoch": 3.3563218390804597,
+      "grad_norm": 4.6993643427058425,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 65890984,
+      "step": 730
+    },
+    {
+      "epoch": 3.3563218390804597,
+      "loss": 0.15184611082077026,
+      "loss_ce": 0.003957913722842932,
+      "loss_iou": 0.421875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 65890984,
+      "step": 730
+    },
+    {
+      "epoch": 3.360919540229885,
+      "grad_norm": 5.235041599676305,
+      "learning_rate": 5e-06,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 65981292,
+      "step": 731
+    },
+    {
+      "epoch": 3.360919540229885,
+      "loss": 0.08815214037895203,
+      "loss_ce": 6.314841448329389e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 65981292,
+      "step": 731
+    },
+    {
+      "epoch": 3.3655172413793104,
+      "grad_norm": 2.2738505705233805,
+      "learning_rate": 5e-06,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 66071596,
+      "step": 732
+    },
+    {
+      "epoch": 3.3655172413793104,
+      "loss": 0.10933637619018555,
+      "loss_ce": 2.240698267996777e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 66071596,
+      "step": 732
+    },
+    {
+      "epoch": 3.3701149425287356,
+      "grad_norm": 3.85647849045901,
+      "learning_rate": 5e-06,
+      "loss": 0.1753,
+      "num_input_tokens_seen": 66162068,
+      "step": 733
+    },
+    {
+      "epoch": 3.3701149425287356,
+      "loss": 0.20265939831733704,
+      "loss_ce": 8.371137664653361e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 66162068,
+      "step": 733
+    },
+    {
+      "epoch": 3.374712643678161,
+      "grad_norm": 8.57033500179484,
+      "learning_rate": 5e-06,
+      "loss": 0.1791,
+      "num_input_tokens_seen": 66252320,
+      "step": 734
+    },
+    {
+      "epoch": 3.374712643678161,
+      "loss": 0.14232100546360016,
+      "loss_ce": 1.754688855726272e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 66252320,
+      "step": 734
+    },
+    {
+      "epoch": 3.3793103448275863,
+      "grad_norm": 9.318942386130336,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 66342832,
+      "step": 735
+    },
+    {
+      "epoch": 3.3793103448275863,
+      "loss": 0.11867457628250122,
+      "loss_ce": 2.2227392037166283e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 66342832,
+      "step": 735
+    },
+    {
+      "epoch": 3.3839080459770114,
+      "grad_norm": 19.001081916459317,
+      "learning_rate": 5e-06,
+      "loss": 0.1317,
+      "num_input_tokens_seen": 66433196,
+      "step": 736
+    },
+    {
+      "epoch": 3.3839080459770114,
+      "loss": 0.12269207835197449,
+      "loss_ce": 1.1412482308514882e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 66433196,
+      "step": 736
+    },
+    {
+      "epoch": 3.3885057471264366,
+      "grad_norm": 24.510560741814796,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 66523456,
+      "step": 737
+    },
+    {
+      "epoch": 3.3885057471264366,
+      "loss": 0.12029193341732025,
+      "loss_ce": 6.896796548971906e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 66523456,
+      "step": 737
+    },
+    {
+      "epoch": 3.393103448275862,
+      "grad_norm": 5.6235077208306095,
+      "learning_rate": 5e-06,
+      "loss": 0.1803,
+      "num_input_tokens_seen": 66613868,
+      "step": 738
+    },
+    {
+      "epoch": 3.393103448275862,
+      "loss": 0.16011008620262146,
+      "loss_ce": 1.4873921827529557e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 66613868,
+      "step": 738
+    },
+    {
+      "epoch": 3.3977011494252873,
+      "grad_norm": 5.027102839955047,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 66704308,
+      "step": 739
+    },
+    {
+      "epoch": 3.3977011494252873,
+      "loss": 0.17483371496200562,
+      "loss_ce": 0.00015872399671934545,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 66704308,
+      "step": 739
+    },
+    {
+      "epoch": 3.4022988505747125,
+      "grad_norm": 7.661403465008269,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 66794608,
+      "step": 740
+    },
+    {
+      "epoch": 3.4022988505747125,
+      "loss": 0.1427173763513565,
+      "loss_ce": 1.7180107533931732e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 66794608,
+      "step": 740
+    },
+    {
+      "epoch": 3.406896551724138,
+      "grad_norm": 10.312505826375784,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 66884952,
+      "step": 741
+    },
+    {
+      "epoch": 3.406896551724138,
+      "loss": 0.19520951807498932,
+      "loss_ce": 1.9079348930972628e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 66884952,
+      "step": 741
+    },
+    {
+      "epoch": 3.4114942528735632,
+      "grad_norm": 13.89068989533051,
+      "learning_rate": 5e-06,
+      "loss": 0.1618,
+      "num_input_tokens_seen": 66975348,
+      "step": 742
+    },
+    {
+      "epoch": 3.4114942528735632,
+      "loss": 0.15027475357055664,
+      "loss_ce": 0.00012826563033740968,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 66975348,
+      "step": 742
+    },
+    {
+      "epoch": 3.4160919540229884,
+      "grad_norm": 4.196181753619327,
+      "learning_rate": 5e-06,
+      "loss": 0.161,
+      "num_input_tokens_seen": 67065664,
+      "step": 743
+    },
+    {
+      "epoch": 3.4160919540229884,
+      "loss": 0.13950249552726746,
+      "loss_ce": 0.00028130554710514843,
+      "loss_iou": 0.390625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 67065664,
+      "step": 743
+    },
+    {
+      "epoch": 3.420689655172414,
+      "grad_norm": 211.3346828186652,
+      "learning_rate": 5e-06,
+      "loss": 0.137,
+      "num_input_tokens_seen": 67155980,
+      "step": 744
+    },
+    {
+      "epoch": 3.420689655172414,
+      "loss": 0.17403674125671387,
+      "loss_ce": 2.5510042178211734e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 67155980,
+      "step": 744
+    },
+    {
+      "epoch": 3.425287356321839,
+      "grad_norm": 13.754629395643422,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 67246368,
+      "step": 745
+    },
+    {
+      "epoch": 3.425287356321839,
+      "loss": 0.13543261587619781,
+      "loss_ce": 0.0008653545519337058,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 67246368,
+      "step": 745
+    },
+    {
+      "epoch": 3.4298850574712643,
+      "grad_norm": 3.5887963625227863,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 67336752,
+      "step": 746
+    },
+    {
+      "epoch": 3.4298850574712643,
+      "loss": 0.08411843329668045,
+      "loss_ce": 1.1989444828941487e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 67336752,
+      "step": 746
+    },
+    {
+      "epoch": 3.43448275862069,
+      "grad_norm": 7.58822953581897,
+      "learning_rate": 5e-06,
+      "loss": 0.1642,
+      "num_input_tokens_seen": 67427208,
+      "step": 747
+    },
+    {
+      "epoch": 3.43448275862069,
+      "loss": 0.15234288573265076,
+      "loss_ce": 6.0174254031153396e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 67427208,
+      "step": 747
+    },
+    {
+      "epoch": 3.439080459770115,
+      "grad_norm": 11.338073894134098,
+      "learning_rate": 5e-06,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 67517560,
+      "step": 748
+    },
+    {
+      "epoch": 3.439080459770115,
+      "loss": 0.13652345538139343,
+      "loss_ce": 1.832299676607363e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 67517560,
+      "step": 748
+    },
+    {
+      "epoch": 3.44367816091954,
+      "grad_norm": 6.967548952942209,
+      "learning_rate": 5e-06,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 67607920,
+      "step": 749
+    },
+    {
+      "epoch": 3.44367816091954,
+      "loss": 0.08051057159900665,
+      "loss_ce": 2.0457886421354488e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 67607920,
+      "step": 749
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "grad_norm": 6.965059583863805,
+      "learning_rate": 5e-06,
+      "loss": 0.1495,
+      "num_input_tokens_seen": 67698360,
+      "step": 750
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "eval_seeclick_CIoU": 0.484468474984169,
+      "eval_seeclick_GIoU": 0.467425137758255,
+      "eval_seeclick_IoU": 0.5228271484375,
+      "eval_seeclick_MAE_all": 0.05892105959355831,
+      "eval_seeclick_MAE_h": 0.04424383118748665,
+      "eval_seeclick_MAE_w": 0.1074240393936634,
+      "eval_seeclick_MAE_x_boxes": 0.10459480062127113,
+      "eval_seeclick_MAE_y_boxes": 0.04638782888650894,
+      "eval_seeclick_NUM_probability": 0.9999992251396179,
+      "eval_seeclick_inside_bbox": 0.8764204680919647,
+      "eval_seeclick_loss": 0.3601701557636261,
+      "eval_seeclick_loss_ce": 0.06872005760669708,
+      "eval_seeclick_loss_iou": 0.46417236328125,
+      "eval_seeclick_loss_num": 0.061737060546875,
+      "eval_seeclick_loss_xval": 0.30859375,
+      "eval_seeclick_runtime": 76.4319,
+      "eval_seeclick_samples_per_second": 0.563,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 67698360,
+      "step": 750
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "eval_icons_CIoU": 0.5398552417755127,
+      "eval_icons_GIoU": 0.5399449467658997,
+      "eval_icons_IoU": 0.5800551474094391,
+      "eval_icons_MAE_all": 0.044397372752428055,
+      "eval_icons_MAE_h": 0.0814935453236103,
+      "eval_icons_MAE_w": 0.0699278824031353,
+      "eval_icons_MAE_x_boxes": 0.06737468019127846,
+      "eval_icons_MAE_y_boxes": 0.08113038912415504,
+      "eval_icons_NUM_probability": 0.9999993741512299,
+      "eval_icons_inside_bbox": 0.7795138955116272,
+      "eval_icons_loss": 0.22252009809017181,
+      "eval_icons_loss_ce": 7.370136074769107e-07,
+      "eval_icons_loss_iou": 0.4132080078125,
+      "eval_icons_loss_num": 0.047031402587890625,
+      "eval_icons_loss_xval": 0.234954833984375,
+      "eval_icons_runtime": 96.2118,
+      "eval_icons_samples_per_second": 0.52,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 67698360,
+      "step": 750
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "eval_screenspot_CIoU": 0.41030613084634143,
+      "eval_screenspot_GIoU": 0.3897818972667058,
+      "eval_screenspot_IoU": 0.4746982256571452,
+      "eval_screenspot_MAE_all": 0.08831173926591873,
+      "eval_screenspot_MAE_h": 0.0866956611474355,
+      "eval_screenspot_MAE_w": 0.17451148480176926,
+      "eval_screenspot_MAE_x_boxes": 0.16739005843798319,
+      "eval_screenspot_MAE_y_boxes": 0.08148317784070969,
+      "eval_screenspot_NUM_probability": 0.9999986886978149,
+      "eval_screenspot_inside_bbox": 0.753333330154419,
+      "eval_screenspot_loss": 0.44167619943618774,
+      "eval_screenspot_loss_ce": 0.00013617607813406116,
+      "eval_screenspot_loss_iou": 0.4222005208333333,
+      "eval_screenspot_loss_num": 0.09005228678385417,
+      "eval_screenspot_loss_xval": 0.4504191080729167,
+      "eval_screenspot_runtime": 157.7137,
+      "eval_screenspot_samples_per_second": 0.564,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 67698360,
+      "step": 750
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "eval_compot_CIoU": 0.4846196174621582,
+      "eval_compot_GIoU": 0.46163466572761536,
+      "eval_compot_IoU": 0.5454895496368408,
+      "eval_compot_MAE_all": 0.05586290545761585,
+      "eval_compot_MAE_h": 0.07578141614794731,
+      "eval_compot_MAE_w": 0.11417003348469734,
+      "eval_compot_MAE_x_boxes": 0.09730804339051247,
+      "eval_compot_MAE_y_boxes": 0.07610557973384857,
+      "eval_compot_NUM_probability": 0.9999949038028717,
+      "eval_compot_inside_bbox": 0.7638888955116272,
+      "eval_compot_loss": 0.30647769570350647,
+      "eval_compot_loss_ce": 0.012538184644654393,
+      "eval_compot_loss_iou": 0.5220947265625,
+      "eval_compot_loss_num": 0.050449371337890625,
+      "eval_compot_loss_xval": 0.2524261474609375,
+      "eval_compot_runtime": 90.9711,
+      "eval_compot_samples_per_second": 0.55,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 67698360,
+      "step": 750
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "loss": 0.18523982167243958,
+      "loss_ce": 0.007658033166080713,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 67698360,
+      "step": 750
+    },
+    {
+      "epoch": 3.452873563218391,
+      "grad_norm": 5.387146730993583,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 67788820,
+      "step": 751
+    },
+    {
+      "epoch": 3.452873563218391,
+      "loss": 0.1267109513282776,
+      "loss_ce": 3.247974018449895e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 67788820,
+      "step": 751
+    },
+    {
+      "epoch": 3.457471264367816,
+      "grad_norm": 13.911926672220762,
+      "learning_rate": 5e-06,
+      "loss": 0.1078,
+      "num_input_tokens_seen": 67879216,
+      "step": 752
+    },
+    {
+      "epoch": 3.457471264367816,
+      "loss": 0.10455399006605148,
+      "loss_ce": 6.180434866109863e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 67879216,
+      "step": 752
+    },
+    {
+      "epoch": 3.462068965517241,
+      "grad_norm": 5.6856250871143486,
+      "learning_rate": 5e-06,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 67969500,
+      "step": 753
+    },
+    {
+      "epoch": 3.462068965517241,
+      "loss": 0.19283702969551086,
+      "loss_ce": 8.800413343124092e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 67969500,
+      "step": 753
+    },
+    {
+      "epoch": 3.466666666666667,
+      "grad_norm": 7.868167270167063,
+      "learning_rate": 5e-06,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 68059952,
+      "step": 754
+    },
+    {
+      "epoch": 3.466666666666667,
+      "loss": 0.05673067271709442,
+      "loss_ce": 1.3753228813584428e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 68059952,
+      "step": 754
+    },
+    {
+      "epoch": 3.471264367816092,
+      "grad_norm": 12.024869069661348,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 68150204,
+      "step": 755
+    },
+    {
+      "epoch": 3.471264367816092,
+      "loss": 0.16925933957099915,
+      "loss_ce": 8.86403904587496e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 68150204,
+      "step": 755
+    },
+    {
+      "epoch": 3.475862068965517,
+      "grad_norm": 10.067626482591868,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 68240748,
+      "step": 756
+    },
+    {
+      "epoch": 3.475862068965517,
+      "loss": 0.12415396422147751,
+      "loss_ce": 8.463855920126662e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 68240748,
+      "step": 756
+    },
+    {
+      "epoch": 3.4804597701149427,
+      "grad_norm": 13.763593458927147,
+      "learning_rate": 5e-06,
+      "loss": 0.2071,
+      "num_input_tokens_seen": 68331096,
+      "step": 757
+    },
+    {
+      "epoch": 3.4804597701149427,
+      "loss": 0.22060084342956543,
+      "loss_ce": 1.9778890418820083e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 68331096,
+      "step": 757
+    },
+    {
+      "epoch": 3.485057471264368,
+      "grad_norm": 37.000929048972885,
+      "learning_rate": 5e-06,
+      "loss": 0.1891,
+      "num_input_tokens_seen": 68421584,
+      "step": 758
+    },
+    {
+      "epoch": 3.485057471264368,
+      "loss": 0.1451530158519745,
+      "loss_ce": 7.243985601235181e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 68421584,
+      "step": 758
+    },
+    {
+      "epoch": 3.489655172413793,
+      "grad_norm": 12.987634056120026,
+      "learning_rate": 5e-06,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 68511940,
+      "step": 759
+    },
+    {
+      "epoch": 3.489655172413793,
+      "loss": 0.051099397242069244,
+      "loss_ce": 1.2974151104572229e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 68511940,
+      "step": 759
+    },
+    {
+      "epoch": 3.4942528735632186,
+      "grad_norm": 4.05554262967813,
+      "learning_rate": 5e-06,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 68602372,
+      "step": 760
+    },
+    {
+      "epoch": 3.4942528735632186,
+      "loss": 0.10796726495027542,
+      "loss_ce": 1.1330963388900273e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 68602372,
+      "step": 760
+    },
+    {
+      "epoch": 3.4988505747126437,
+      "grad_norm": 2.872197937123995,
+      "learning_rate": 5e-06,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 68692692,
+      "step": 761
+    },
+    {
+      "epoch": 3.4988505747126437,
+      "loss": 0.10546976327896118,
+      "loss_ce": 1.009384050121298e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 68692692,
+      "step": 761
+    },
+    {
+      "epoch": 3.503448275862069,
+      "grad_norm": 2.7436969971531573,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 68783112,
+      "step": 762
+    },
+    {
+      "epoch": 3.503448275862069,
+      "loss": 0.0926920622587204,
+      "loss_ce": 1.017252270685276e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 68783112,
+      "step": 762
+    },
+    {
+      "epoch": 3.508045977011494,
+      "grad_norm": 12.740068722129191,
+      "learning_rate": 5e-06,
+      "loss": 0.1955,
+      "num_input_tokens_seen": 68872560,
+      "step": 763
+    },
+    {
+      "epoch": 3.508045977011494,
+      "loss": 0.20338374376296997,
+      "loss_ce": 0.0019982485100626945,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 68872560,
+      "step": 763
+    },
+    {
+      "epoch": 3.5126436781609196,
+      "grad_norm": 7.842380690749216,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 68962972,
+      "step": 764
+    },
+    {
+      "epoch": 3.5126436781609196,
+      "loss": 0.14544086158275604,
+      "loss_ce": 2.4600009055575356e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 68962972,
+      "step": 764
+    },
+    {
+      "epoch": 3.5172413793103448,
+      "grad_norm": 5.499498755138326,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 69053244,
+      "step": 765
+    },
+    {
+      "epoch": 3.5172413793103448,
+      "loss": 0.09770219027996063,
+      "loss_ce": 4.593187622958794e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 69053244,
+      "step": 765
+    },
+    {
+      "epoch": 3.5218390804597703,
+      "grad_norm": 4.548325999877122,
+      "learning_rate": 5e-06,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 69143592,
+      "step": 766
+    },
+    {
+      "epoch": 3.5218390804597703,
+      "loss": 0.06963898241519928,
+      "loss_ce": 2.8384643883327954e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 69143592,
+      "step": 766
+    },
+    {
+      "epoch": 3.5264367816091955,
+      "grad_norm": 13.276586752236186,
+      "learning_rate": 5e-06,
+      "loss": 0.1276,
+      "num_input_tokens_seen": 69233884,
+      "step": 767
+    },
+    {
+      "epoch": 3.5264367816091955,
+      "loss": 0.14585262537002563,
+      "loss_ce": 0.0010162012185901403,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 69233884,
+      "step": 767
+    },
+    {
+      "epoch": 3.5310344827586206,
+      "grad_norm": 3.288042195364263,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 69324272,
+      "step": 768
+    },
+    {
+      "epoch": 3.5310344827586206,
+      "loss": 0.10515020787715912,
+      "loss_ce": 0.00013922779180575162,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 69324272,
+      "step": 768
+    },
+    {
+      "epoch": 3.535632183908046,
+      "grad_norm": 20.003220474729524,
+      "learning_rate": 5e-06,
+      "loss": 0.1411,
+      "num_input_tokens_seen": 69413868,
+      "step": 769
+    },
+    {
+      "epoch": 3.535632183908046,
+      "loss": 0.1137920469045639,
+      "loss_ce": 2.2520007405546494e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 69413868,
+      "step": 769
+    },
+    {
+      "epoch": 3.5402298850574714,
+      "grad_norm": 14.09650460702761,
+      "learning_rate": 5e-06,
+      "loss": 0.1811,
+      "num_input_tokens_seen": 69504204,
+      "step": 770
+    },
+    {
+      "epoch": 3.5402298850574714,
+      "loss": 0.16392265260219574,
+      "loss_ce": 1.2739261364913546e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 69504204,
+      "step": 770
+    },
+    {
+      "epoch": 3.5448275862068965,
+      "grad_norm": 5.589999305942861,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 69594556,
+      "step": 771
+    },
+    {
+      "epoch": 3.5448275862068965,
+      "loss": 0.1477949023246765,
+      "loss_ce": 2.8777922125300393e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 69594556,
+      "step": 771
+    },
+    {
+      "epoch": 3.5494252873563217,
+      "grad_norm": 6.623212624230002,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 69684820,
+      "step": 772
+    },
+    {
+      "epoch": 3.5494252873563217,
+      "loss": 0.06927981972694397,
+      "loss_ce": 4.912180884275585e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 69684820,
+      "step": 772
+    },
+    {
+      "epoch": 3.5540229885057473,
+      "grad_norm": 6.206244364237178,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 69773688,
+      "step": 773
+    },
+    {
+      "epoch": 3.5540229885057473,
+      "loss": 0.12721136212348938,
+      "loss_ce": 1.4094775906414725e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 69773688,
+      "step": 773
+    },
+    {
+      "epoch": 3.5586206896551724,
+      "grad_norm": 6.055994010471258,
+      "learning_rate": 5e-06,
+      "loss": 0.1533,
+      "num_input_tokens_seen": 69864012,
+      "step": 774
+    },
+    {
+      "epoch": 3.5586206896551724,
+      "loss": 0.1802106499671936,
+      "loss_ce": 9.590051922714338e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 69864012,
+      "step": 774
+    },
+    {
+      "epoch": 3.5632183908045976,
+      "grad_norm": 4.739416821882011,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 69954324,
+      "step": 775
+    },
+    {
+      "epoch": 3.5632183908045976,
+      "loss": 0.16380281746387482,
+      "loss_ce": 0.0002896270889323205,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 69954324,
+      "step": 775
+    },
+    {
+      "epoch": 3.5678160919540227,
+      "grad_norm": 8.653175282625192,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 70044580,
+      "step": 776
+    },
+    {
+      "epoch": 3.5678160919540227,
+      "loss": 0.09520787745714188,
+      "loss_ce": 8.29543569125235e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 70044580,
+      "step": 776
+    },
+    {
+      "epoch": 3.5724137931034483,
+      "grad_norm": 21.81495303272478,
+      "learning_rate": 5e-06,
+      "loss": 0.1423,
+      "num_input_tokens_seen": 70134884,
+      "step": 777
+    },
+    {
+      "epoch": 3.5724137931034483,
+      "loss": 0.11333857476711273,
+      "loss_ce": 0.000103099322586786,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 70134884,
+      "step": 777
+    },
+    {
+      "epoch": 3.5770114942528735,
+      "grad_norm": 15.12571431595152,
+      "learning_rate": 5e-06,
+      "loss": 0.1656,
+      "num_input_tokens_seen": 70225164,
+      "step": 778
+    },
+    {
+      "epoch": 3.5770114942528735,
+      "loss": 0.2089380919933319,
+      "loss_ce": 1.4762490536668338e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 70225164,
+      "step": 778
+    },
+    {
+      "epoch": 3.581609195402299,
+      "grad_norm": 10.383950864076555,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 70315464,
+      "step": 779
+    },
+    {
+      "epoch": 3.581609195402299,
+      "loss": 0.11370711028575897,
+      "loss_ce": 2.912801573984325e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 70315464,
+      "step": 779
+    },
+    {
+      "epoch": 3.586206896551724,
+      "grad_norm": 2.1421328839083684,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 70405788,
+      "step": 780
+    },
+    {
+      "epoch": 3.586206896551724,
+      "loss": 0.1896582543849945,
+      "loss_ce": 2.202134237450082e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 70405788,
+      "step": 780
+    },
+    {
+      "epoch": 3.5908045977011493,
+      "grad_norm": 9.380525583505861,
+      "learning_rate": 5e-06,
+      "loss": 0.2045,
+      "num_input_tokens_seen": 70496136,
+      "step": 781
+    },
+    {
+      "epoch": 3.5908045977011493,
+      "loss": 0.21958574652671814,
+      "loss_ce": 1.1782582078012638e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 70496136,
+      "step": 781
+    },
+    {
+      "epoch": 3.5954022988505745,
+      "grad_norm": 3.9371848580056517,
+      "learning_rate": 5e-06,
+      "loss": 0.0866,
+      "num_input_tokens_seen": 70586496,
+      "step": 782
+    },
+    {
+      "epoch": 3.5954022988505745,
+      "loss": 0.05908702313899994,
+      "loss_ce": 2.025284084083978e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 70586496,
+      "step": 782
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 10.508397334914141,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 70676784,
+      "step": 783
+    },
+    {
+      "epoch": 3.6,
+      "loss": 0.11284907907247543,
+      "loss_ce": 1.0333604222978465e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 70676784,
+      "step": 783
+    },
+    {
+      "epoch": 3.6045977011494252,
+      "grad_norm": 12.149659037494654,
+      "learning_rate": 5e-06,
+      "loss": 0.1,
+      "num_input_tokens_seen": 70767128,
+      "step": 784
+    },
+    {
+      "epoch": 3.6045977011494252,
+      "loss": 0.10474172979593277,
+      "loss_ce": 5.399918336479459e-06,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 70767128,
+      "step": 784
+    },
+    {
+      "epoch": 3.609195402298851,
+      "grad_norm": 32.31492723615795,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 70857476,
+      "step": 785
+    },
+    {
+      "epoch": 3.609195402298851,
+      "loss": 0.11612387001514435,
+      "loss_ce": 4.4922308006789535e-06,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 70857476,
+      "step": 785
+    },
+    {
+      "epoch": 3.613793103448276,
+      "grad_norm": 13.784691745076751,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 70947968,
+      "step": 786
+    },
+    {
+      "epoch": 3.613793103448276,
+      "loss": 0.09685492515563965,
+      "loss_ce": 2.2650790924672037e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 70947968,
+      "step": 786
+    },
+    {
+      "epoch": 3.618390804597701,
+      "grad_norm": 5.938316209673259,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 71038404,
+      "step": 787
+    },
+    {
+      "epoch": 3.618390804597701,
+      "loss": 0.13221478462219238,
+      "loss_ce": 1.2635980965569615e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 71038404,
+      "step": 787
+    },
+    {
+      "epoch": 3.6229885057471263,
+      "grad_norm": 13.011958974161733,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 71128788,
+      "step": 788
+    },
+    {
+      "epoch": 3.6229885057471263,
+      "loss": 0.14329373836517334,
+      "loss_ce": 1.3719867638428695e-05,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 71128788,
+      "step": 788
+    },
+    {
+      "epoch": 3.627586206896552,
+      "grad_norm": 11.936048890187932,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 71219196,
+      "step": 789
+    },
+    {
+      "epoch": 3.627586206896552,
+      "loss": 0.11237190663814545,
+      "loss_ce": 8.247944788308814e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 71219196,
+      "step": 789
+    },
+    {
+      "epoch": 3.632183908045977,
+      "grad_norm": 6.308330166762513,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 71309520,
+      "step": 790
+    },
+    {
+      "epoch": 3.632183908045977,
+      "loss": 0.1292654573917389,
+      "loss_ce": 5.403875547926873e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 71309520,
+      "step": 790
+    },
+    {
+      "epoch": 3.636781609195402,
+      "grad_norm": 5.137471316885716,
+      "learning_rate": 5e-06,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 71399780,
+      "step": 791
+    },
+    {
+      "epoch": 3.636781609195402,
+      "loss": 0.13990908861160278,
+      "loss_ce": 1.6500023775734007e-05,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 71399780,
+      "step": 791
+    },
+    {
+      "epoch": 3.6413793103448278,
+      "grad_norm": 9.275105119831409,
+      "learning_rate": 5e-06,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 71490120,
+      "step": 792
+    },
+    {
+      "epoch": 3.6413793103448278,
+      "loss": 0.06399580091238022,
+      "loss_ce": 3.09558781736996e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 71490120,
+      "step": 792
+    },
+    {
+      "epoch": 3.645977011494253,
+      "grad_norm": 11.0637012290359,
+      "learning_rate": 5e-06,
+      "loss": 0.1358,
+      "num_input_tokens_seen": 71580584,
+      "step": 793
+    },
+    {
+      "epoch": 3.645977011494253,
+      "loss": 0.15851935744285583,
+      "loss_ce": 4.156233626417816e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 71580584,
+      "step": 793
+    },
+    {
+      "epoch": 3.650574712643678,
+      "grad_norm": 16.625197592400877,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 71670196,
+      "step": 794
+    },
+    {
+      "epoch": 3.650574712643678,
+      "loss": 0.15688209235668182,
+      "loss_ce": 2.1741370801464655e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 71670196,
+      "step": 794
+    },
+    {
+      "epoch": 3.655172413793103,
+      "grad_norm": 5.3921763962194005,
+      "learning_rate": 5e-06,
+      "loss": 0.1317,
+      "num_input_tokens_seen": 71760612,
+      "step": 795
+    },
+    {
+      "epoch": 3.655172413793103,
+      "loss": 0.10780765116214752,
+      "loss_ce": 4.304832600610098e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 71760612,
+      "step": 795
+    },
+    {
+      "epoch": 3.659770114942529,
+      "grad_norm": 11.704393512510705,
+      "learning_rate": 5e-06,
+      "loss": 0.1639,
+      "num_input_tokens_seen": 71850916,
+      "step": 796
+    },
+    {
+      "epoch": 3.659770114942529,
+      "loss": 0.1306782066822052,
+      "loss_ce": 1.929806785483379e-06,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 71850916,
+      "step": 796
+    },
+    {
+      "epoch": 3.664367816091954,
+      "grad_norm": 16.992640998013997,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 71941204,
+      "step": 797
+    },
+    {
+      "epoch": 3.664367816091954,
+      "loss": 0.11268985271453857,
+      "loss_ce": 1.8956783605972305e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 71941204,
+      "step": 797
+    },
+    {
+      "epoch": 3.6689655172413795,
+      "grad_norm": 11.094649026288351,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 72031516,
+      "step": 798
+    },
+    {
+      "epoch": 3.6689655172413795,
+      "loss": 0.15380123257637024,
+      "loss_ce": 2.316595782758668e-05,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 72031516,
+      "step": 798
+    },
+    {
+      "epoch": 3.6735632183908047,
+      "grad_norm": 9.218816872173734,
+      "learning_rate": 5e-06,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 72121944,
+      "step": 799
+    },
+    {
+      "epoch": 3.6735632183908047,
+      "loss": 0.10079614073038101,
+      "loss_ce": 2.710282933549024e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 72121944,
+      "step": 799
+    },
+    {
+      "epoch": 3.67816091954023,
+      "grad_norm": 15.692187876785969,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 72212332,
+      "step": 800
+    },
+    {
+      "epoch": 3.67816091954023,
+      "loss": 0.08001083880662918,
+      "loss_ce": 9.010436770040542e-06,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 72212332,
+      "step": 800
+    },
+    {
+      "epoch": 3.682758620689655,
+      "grad_norm": 3.364192050864013,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 72302688,
+      "step": 801
+    },
+    {
+      "epoch": 3.682758620689655,
+      "loss": 0.060924116522073746,
+      "loss_ce": 0.00010258361726300791,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 72302688,
+      "step": 801
+    },
+    {
+      "epoch": 3.6873563218390806,
+      "grad_norm": 13.859096616510685,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 72393104,
+      "step": 802
+    },
+    {
+      "epoch": 3.6873563218390806,
+      "loss": 0.1117105633020401,
+      "loss_ce": 1.622791023692116e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 72393104,
+      "step": 802
+    },
+    {
+      "epoch": 3.6919540229885057,
+      "grad_norm": 13.956089000343894,
+      "learning_rate": 5e-06,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 72483416,
+      "step": 803
+    },
+    {
+      "epoch": 3.6919540229885057,
+      "loss": 0.0864303782582283,
+      "loss_ce": 4.60219871456502e-06,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 72483416,
+      "step": 803
+    },
+    {
+      "epoch": 3.696551724137931,
+      "grad_norm": 10.769508987312188,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 72573912,
+      "step": 804
+    },
+    {
+      "epoch": 3.696551724137931,
+      "loss": 0.12728144228458405,
+      "loss_ce": 7.876707968534902e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 72573912,
+      "step": 804
+    },
+    {
+      "epoch": 3.7011494252873565,
+      "grad_norm": 3.244021220688608,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 72664244,
+      "step": 805
+    },
+    {
+      "epoch": 3.7011494252873565,
+      "loss": 0.14200298488140106,
+      "loss_ce": 5.0472979637561366e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 72664244,
+      "step": 805
+    },
+    {
+      "epoch": 3.7057471264367816,
+      "grad_norm": 4.828124192128039,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 72753736,
+      "step": 806
+    },
+    {
+      "epoch": 3.7057471264367816,
+      "loss": 0.0866774469614029,
+      "loss_ce": 7.529814411100233e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 72753736,
+      "step": 806
+    },
+    {
+      "epoch": 3.7103448275862068,
+      "grad_norm": 10.090685737843089,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 72844080,
+      "step": 807
+    },
+    {
+      "epoch": 3.7103448275862068,
+      "loss": 0.14415577054023743,
+      "loss_ce": 3.650196231319569e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 72844080,
+      "step": 807
+    },
+    {
+      "epoch": 3.714942528735632,
+      "grad_norm": 11.901790776803901,
+      "learning_rate": 5e-06,
+      "loss": 0.196,
+      "num_input_tokens_seen": 72934468,
+      "step": 808
+    },
+    {
+      "epoch": 3.714942528735632,
+      "loss": 0.21217146515846252,
+      "loss_ce": 2.8517068130895495e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 72934468,
+      "step": 808
+    },
+    {
+      "epoch": 3.7195402298850575,
+      "grad_norm": 24.55270500907039,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 73024692,
+      "step": 809
+    },
+    {
+      "epoch": 3.7195402298850575,
+      "loss": 0.10771137475967407,
+      "loss_ce": 4.536272172117606e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 73024692,
+      "step": 809
+    },
+    {
+      "epoch": 3.7241379310344827,
+      "grad_norm": 15.553335565571802,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 73115056,
+      "step": 810
+    },
+    {
+      "epoch": 3.7241379310344827,
+      "loss": 0.14749327301979065,
+      "loss_ce": 3.232985909562558e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 73115056,
+      "step": 810
+    },
+    {
+      "epoch": 3.7287356321839082,
+      "grad_norm": 21.65190443149484,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 73205556,
+      "step": 811
+    },
+    {
+      "epoch": 3.7287356321839082,
+      "loss": 0.10954000800848007,
+      "loss_ce": 0.0008363968227058649,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 73205556,
+      "step": 811
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 3.28808735511797,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 73296004,
+      "step": 812
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "loss": 0.09380966424942017,
+      "loss_ce": 1.3889082765672356e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 73296004,
+      "step": 812
+    },
+    {
+      "epoch": 3.7379310344827585,
+      "grad_norm": 7.828759411600456,
+      "learning_rate": 5e-06,
+      "loss": 0.1538,
+      "num_input_tokens_seen": 73386480,
+      "step": 813
+    },
+    {
+      "epoch": 3.7379310344827585,
+      "loss": 0.167840838432312,
+      "loss_ce": 0.0001162344342446886,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 73386480,
+      "step": 813
+    },
+    {
+      "epoch": 3.7425287356321837,
+      "grad_norm": 27.164938076027006,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 73476720,
+      "step": 814
+    },
+    {
+      "epoch": 3.7425287356321837,
+      "loss": 0.13790945708751678,
+      "loss_ce": 6.155119626782835e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 73476720,
+      "step": 814
+    },
+    {
+      "epoch": 3.7471264367816093,
+      "grad_norm": 7.036276194280488,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 73566368,
+      "step": 815
+    },
+    {
+      "epoch": 3.7471264367816093,
+      "loss": 0.11970978230237961,
+      "loss_ce": 3.510632086545229e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 73566368,
+      "step": 815
+    },
+    {
+      "epoch": 3.7517241379310344,
+      "grad_norm": 6.291798562723388,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 73656748,
+      "step": 816
+    },
+    {
+      "epoch": 3.7517241379310344,
+      "loss": 0.15857738256454468,
+      "loss_ce": 3.092413680860773e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 73656748,
+      "step": 816
+    },
+    {
+      "epoch": 3.75632183908046,
+      "grad_norm": 4.395310325624689,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 73747096,
+      "step": 817
+    },
+    {
+      "epoch": 3.75632183908046,
+      "loss": 0.1381591111421585,
+      "loss_ce": 6.042053428245708e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 73747096,
+      "step": 817
+    },
+    {
+      "epoch": 3.760919540229885,
+      "grad_norm": 4.574856648294534,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 73837368,
+      "step": 818
+    },
+    {
+      "epoch": 3.760919540229885,
+      "loss": 0.1333150714635849,
+      "loss_ce": 0.00028894448769278824,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 73837368,
+      "step": 818
+    },
+    {
+      "epoch": 3.7655172413793103,
+      "grad_norm": 20.895031812009428,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 73927844,
+      "step": 819
+    },
+    {
+      "epoch": 3.7655172413793103,
+      "loss": 0.10864730179309845,
+      "loss_ce": 3.5239758290117607e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 73927844,
+      "step": 819
+    },
+    {
+      "epoch": 3.7701149425287355,
+      "grad_norm": 15.134371186789252,
+      "learning_rate": 5e-06,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 74018240,
+      "step": 820
+    },
+    {
+      "epoch": 3.7701149425287355,
+      "loss": 0.09329765290021896,
+      "loss_ce": 0.00020377803593873978,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 74018240,
+      "step": 820
+    },
+    {
+      "epoch": 3.774712643678161,
+      "grad_norm": 15.175525511944207,
+      "learning_rate": 5e-06,
+      "loss": 0.1241,
+      "num_input_tokens_seen": 74108676,
+      "step": 821
+    },
+    {
+      "epoch": 3.774712643678161,
+      "loss": 0.13372212648391724,
+      "loss_ce": 7.039559568511322e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 74108676,
+      "step": 821
+    },
+    {
+      "epoch": 3.779310344827586,
+      "grad_norm": 27.19521596346353,
+      "learning_rate": 5e-06,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 74199076,
+      "step": 822
+    },
+    {
+      "epoch": 3.779310344827586,
+      "loss": 0.14935675263404846,
+      "loss_ce": 0.00011052708578063175,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 74199076,
+      "step": 822
+    },
+    {
+      "epoch": 3.7839080459770114,
+      "grad_norm": 10.150253946725481,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 74288004,
+      "step": 823
+    },
+    {
+      "epoch": 3.7839080459770114,
+      "loss": 0.1666683554649353,
+      "loss_ce": 0.0001491915318183601,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 74288004,
+      "step": 823
+    },
+    {
+      "epoch": 3.788505747126437,
+      "grad_norm": 3.7134569300100755,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 74378480,
+      "step": 824
+    },
+    {
+      "epoch": 3.788505747126437,
+      "loss": 0.118888720870018,
+      "loss_ce": 5.327704275259748e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 74378480,
+      "step": 824
+    },
+    {
+      "epoch": 3.793103448275862,
+      "grad_norm": 5.136391279757606,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 74468904,
+      "step": 825
+    },
+    {
+      "epoch": 3.793103448275862,
+      "loss": 0.10247914493083954,
+      "loss_ce": 0.00015370690380223095,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 74468904,
+      "step": 825
+    },
+    {
+      "epoch": 3.7977011494252872,
+      "grad_norm": 3.4776597012299777,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 74559308,
+      "step": 826
+    },
+    {
+      "epoch": 3.7977011494252872,
+      "loss": 0.08007881045341492,
+      "loss_ce": 0.00010749474313342944,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 74559308,
+      "step": 826
+    },
+    {
+      "epoch": 3.8022988505747124,
+      "grad_norm": 13.304246207016769,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 74649636,
+      "step": 827
+    },
+    {
+      "epoch": 3.8022988505747124,
+      "loss": 0.144487202167511,
+      "loss_ce": 1.7336749351670733e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 74649636,
+      "step": 827
+    },
+    {
+      "epoch": 3.806896551724138,
+      "grad_norm": 8.020543075324909,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 74740060,
+      "step": 828
+    },
+    {
+      "epoch": 3.806896551724138,
+      "loss": 0.09062141180038452,
+      "loss_ce": 1.4721297702635638e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 74740060,
+      "step": 828
+    },
+    {
+      "epoch": 3.811494252873563,
+      "grad_norm": 14.35279728790665,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 74830480,
+      "step": 829
+    },
+    {
+      "epoch": 3.811494252873563,
+      "loss": 0.06638382375240326,
+      "loss_ce": 8.095349585346412e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 74830480,
+      "step": 829
+    },
+    {
+      "epoch": 3.8160919540229887,
+      "grad_norm": 14.243365874652318,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 74920808,
+      "step": 830
+    },
+    {
+      "epoch": 3.8160919540229887,
+      "loss": 0.1177852526307106,
+      "loss_ce": 2.654241143318359e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 74920808,
+      "step": 830
+    },
+    {
+      "epoch": 3.820689655172414,
+      "grad_norm": 10.427621446470681,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 75011112,
+      "step": 831
+    },
+    {
+      "epoch": 3.820689655172414,
+      "loss": 0.09321253001689911,
+      "loss_ce": 0.00024072162341326475,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 75011112,
+      "step": 831
+    },
+    {
+      "epoch": 3.825287356321839,
+      "grad_norm": 3.5623265997795768,
+      "learning_rate": 5e-06,
+      "loss": 0.1278,
+      "num_input_tokens_seen": 75101356,
+      "step": 832
+    },
+    {
+      "epoch": 3.825287356321839,
+      "loss": 0.1735125035047531,
+      "loss_ce": 5.05839052493684e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 75101356,
+      "step": 832
+    },
+    {
+      "epoch": 3.829885057471264,
+      "grad_norm": 4.146150429296354,
+      "learning_rate": 5e-06,
+      "loss": 0.1608,
+      "num_input_tokens_seen": 75190144,
+      "step": 833
+    },
+    {
+      "epoch": 3.829885057471264,
+      "loss": 0.1830693781375885,
+      "loss_ce": 0.0004521902301348746,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 75190144,
+      "step": 833
+    },
+    {
+      "epoch": 3.8344827586206898,
+      "grad_norm": 6.028978250099342,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 75280528,
+      "step": 834
+    },
+    {
+      "epoch": 3.8344827586206898,
+      "loss": 0.12013056129217148,
+      "loss_ce": 4.389049718156457e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 75280528,
+      "step": 834
+    },
+    {
+      "epoch": 3.839080459770115,
+      "grad_norm": 5.005804908205812,
+      "learning_rate": 5e-06,
+      "loss": 0.1431,
+      "num_input_tokens_seen": 75370936,
+      "step": 835
+    },
+    {
+      "epoch": 3.839080459770115,
+      "loss": 0.17678186297416687,
+      "loss_ce": 2.404209772066679e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 75370936,
+      "step": 835
+    },
+    {
+      "epoch": 3.84367816091954,
+      "grad_norm": 11.780827894531459,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 75461384,
+      "step": 836
+    },
+    {
+      "epoch": 3.84367816091954,
+      "loss": 0.0846051275730133,
+      "loss_ce": 8.669264207128435e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 75461384,
+      "step": 836
+    },
+    {
+      "epoch": 3.8482758620689657,
+      "grad_norm": 9.132134187365157,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 75551680,
+      "step": 837
+    },
+    {
+      "epoch": 3.8482758620689657,
+      "loss": 0.12721320986747742,
+      "loss_ce": 6.894965736137237e-07,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 75551680,
+      "step": 837
+    },
+    {
+      "epoch": 3.852873563218391,
+      "grad_norm": 26.114035834080024,
+      "learning_rate": 5e-06,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 75642112,
+      "step": 838
+    },
+    {
+      "epoch": 3.852873563218391,
+      "loss": 0.08660604059696198,
+      "loss_ce": 5.8186371461488307e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 75642112,
+      "step": 838
+    },
+    {
+      "epoch": 3.857471264367816,
+      "grad_norm": 11.804335104556394,
+      "learning_rate": 5e-06,
+      "loss": 0.0909,
+      "num_input_tokens_seen": 75732404,
+      "step": 839
+    },
+    {
+      "epoch": 3.857471264367816,
+      "loss": 0.06891857087612152,
+      "loss_ce": 9.877001502900384e-06,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 75732404,
+      "step": 839
+    },
+    {
+      "epoch": 3.862068965517241,
+      "grad_norm": 8.148540211603276,
+      "learning_rate": 5e-06,
+      "loss": 0.1853,
+      "num_input_tokens_seen": 75822708,
+      "step": 840
+    },
+    {
+      "epoch": 3.862068965517241,
+      "loss": 0.1470450758934021,
+      "loss_ce": 1.1382823686290067e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 75822708,
+      "step": 840
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "grad_norm": 2.959579079163402,
+      "learning_rate": 5e-06,
+      "loss": 0.1515,
+      "num_input_tokens_seen": 75912948,
+      "step": 841
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "loss": 0.14687013626098633,
+      "loss_ce": 4.280491793906549e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 75912948,
+      "step": 841
+    },
+    {
+      "epoch": 3.871264367816092,
+      "grad_norm": 3.2788587846844477,
+      "learning_rate": 5e-06,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 76003272,
+      "step": 842
+    },
+    {
+      "epoch": 3.871264367816092,
+      "loss": 0.1434180587530136,
+      "loss_ce": 1.5959709344315343e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 76003272,
+      "step": 842
+    },
+    {
+      "epoch": 3.8758620689655174,
+      "grad_norm": 6.603617127888935,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 76093704,
+      "step": 843
+    },
+    {
+      "epoch": 3.8758620689655174,
+      "loss": 0.18056106567382812,
+      "loss_ce": 4.960661317454651e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 76093704,
+      "step": 843
+    },
+    {
+      "epoch": 3.8804597701149426,
+      "grad_norm": 2.2614424128970096,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 76183952,
+      "step": 844
+    },
+    {
+      "epoch": 3.8804597701149426,
+      "loss": 0.11200746148824692,
+      "loss_ce": 7.948267921165098e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 76183952,
+      "step": 844
+    },
+    {
+      "epoch": 3.8850574712643677,
+      "grad_norm": 9.237162950267727,
+      "learning_rate": 5e-06,
+      "loss": 0.1355,
+      "num_input_tokens_seen": 76274384,
+      "step": 845
+    },
+    {
+      "epoch": 3.8850574712643677,
+      "loss": 0.13711686432361603,
+      "loss_ce": 3.19016762659885e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 76274384,
+      "step": 845
+    },
+    {
+      "epoch": 3.889655172413793,
+      "grad_norm": 3.4174539263582626,
+      "learning_rate": 5e-06,
+      "loss": 0.1565,
+      "num_input_tokens_seen": 76364656,
+      "step": 846
+    },
+    {
+      "epoch": 3.889655172413793,
+      "loss": 0.15261296927928925,
+      "loss_ce": 5.560090721701272e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 76364656,
+      "step": 846
+    },
+    {
+      "epoch": 3.8942528735632185,
+      "grad_norm": 5.269387019291706,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 76455084,
+      "step": 847
+    },
+    {
+      "epoch": 3.8942528735632185,
+      "loss": 0.12015949189662933,
+      "loss_ce": 4.230613194522448e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 76455084,
+      "step": 847
+    },
+    {
+      "epoch": 3.8988505747126436,
+      "grad_norm": 33.611586771121246,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 76545436,
+      "step": 848
+    },
+    {
+      "epoch": 3.8988505747126436,
+      "loss": 0.11951969563961029,
+      "loss_ce": 1.2851842257077806e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 76545436,
+      "step": 848
+    },
+    {
+      "epoch": 3.903448275862069,
+      "grad_norm": 3.5055722837258783,
+      "learning_rate": 5e-06,
+      "loss": 0.1269,
+      "num_input_tokens_seen": 76634196,
+      "step": 849
+    },
+    {
+      "epoch": 3.903448275862069,
+      "loss": 0.12913568317890167,
+      "loss_ce": 1.5821904526092112e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 76634196,
+      "step": 849
+    },
+    {
+      "epoch": 3.9080459770114944,
+      "grad_norm": 8.97172584427251,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 76724524,
+      "step": 850
+    },
+    {
+      "epoch": 3.9080459770114944,
+      "loss": 0.10759300738573074,
+      "loss_ce": 1.8539070879342034e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 76724524,
+      "step": 850
+    },
+    {
+      "epoch": 3.9126436781609195,
+      "grad_norm": 2.3536210529744888,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 76815068,
+      "step": 851
+    },
+    {
+      "epoch": 3.9126436781609195,
+      "loss": 0.0927945151925087,
+      "loss_ce": 0.00011263469059485942,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 76815068,
+      "step": 851
+    },
+    {
+      "epoch": 3.9172413793103447,
+      "grad_norm": 6.886385253300276,
+      "learning_rate": 5e-06,
+      "loss": 0.0821,
+      "num_input_tokens_seen": 76905484,
+      "step": 852
+    },
+    {
+      "epoch": 3.9172413793103447,
+      "loss": 0.09015928208827972,
+      "loss_ce": 7.139760418795049e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 76905484,
+      "step": 852
+    },
+    {
+      "epoch": 3.9218390804597703,
+      "grad_norm": 8.891761743105253,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 76995800,
+      "step": 853
+    },
+    {
+      "epoch": 3.9218390804597703,
+      "loss": 0.15627999603748322,
+      "loss_ce": 3.000300239364151e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 76995800,
+      "step": 853
+    },
+    {
+      "epoch": 3.9264367816091954,
+      "grad_norm": 5.132513236049143,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 77085280,
+      "step": 854
+    },
+    {
+      "epoch": 3.9264367816091954,
+      "loss": 0.10330615192651749,
+      "loss_ce": 4.150414042669581e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 77085280,
+      "step": 854
+    },
+    {
+      "epoch": 3.9310344827586206,
+      "grad_norm": 11.013320716371243,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 77174948,
+      "step": 855
+    },
+    {
+      "epoch": 3.9310344827586206,
+      "loss": 0.09394749999046326,
+      "loss_ce": 1.439205152564682e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 77174948,
+      "step": 855
+    },
+    {
+      "epoch": 3.935632183908046,
+      "grad_norm": 6.10534980851738,
+      "learning_rate": 5e-06,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 77265376,
+      "step": 856
+    },
+    {
+      "epoch": 3.935632183908046,
+      "loss": 0.06424582004547119,
+      "loss_ce": 6.319404747046065e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 77265376,
+      "step": 856
+    },
+    {
+      "epoch": 3.9402298850574713,
+      "grad_norm": 15.651592669210814,
+      "learning_rate": 5e-06,
+      "loss": 0.0864,
+      "num_input_tokens_seen": 77355760,
+      "step": 857
+    },
+    {
+      "epoch": 3.9402298850574713,
+      "loss": 0.0786062479019165,
+      "loss_ce": 2.3480024538002908e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 77355760,
+      "step": 857
+    },
+    {
+      "epoch": 3.9448275862068964,
+      "grad_norm": 6.800574472479352,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 77446208,
+      "step": 858
+    },
+    {
+      "epoch": 3.9448275862068964,
+      "loss": 0.09560946375131607,
+      "loss_ce": 0.0001809898967621848,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 77446208,
+      "step": 858
+    },
+    {
+      "epoch": 3.9494252873563216,
+      "grad_norm": 14.49037795832371,
+      "learning_rate": 5e-06,
+      "loss": 0.1543,
+      "num_input_tokens_seen": 77536448,
+      "step": 859
+    },
+    {
+      "epoch": 3.9494252873563216,
+      "loss": 0.14268243312835693,
+      "loss_ce": 1.2764152415911667e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 77536448,
+      "step": 859
+    },
+    {
+      "epoch": 3.954022988505747,
+      "grad_norm": 3.136445093906524,
+      "learning_rate": 5e-06,
+      "loss": 0.099,
+      "num_input_tokens_seen": 77626764,
+      "step": 860
+    },
+    {
+      "epoch": 3.954022988505747,
+      "loss": 0.09950034320354462,
+      "loss_ce": 1.3036394193477463e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 77626764,
+      "step": 860
+    },
+    {
+      "epoch": 3.9586206896551723,
+      "grad_norm": 5.399769066083071,
+      "learning_rate": 5e-06,
+      "loss": 0.2064,
+      "num_input_tokens_seen": 77717140,
+      "step": 861
+    },
+    {
+      "epoch": 3.9586206896551723,
+      "loss": 0.19695694744586945,
+      "loss_ce": 1.1762541362259071e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 77717140,
+      "step": 861
+    },
+    {
+      "epoch": 3.963218390804598,
+      "grad_norm": 18.891060577417118,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 77807488,
+      "step": 862
+    },
+    {
+      "epoch": 3.963218390804598,
+      "loss": 0.09799706935882568,
+      "loss_ce": 5.1299293772899546e-06,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 77807488,
+      "step": 862
+    },
+    {
+      "epoch": 3.967816091954023,
+      "grad_norm": 5.482893632604601,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 77897924,
+      "step": 863
+    },
+    {
+      "epoch": 3.967816091954023,
+      "loss": 0.08787819743156433,
+      "loss_ce": 1.808757588150911e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 77897924,
+      "step": 863
+    },
+    {
+      "epoch": 3.972413793103448,
+      "grad_norm": 22.690601947302575,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 77988320,
+      "step": 864
+    },
+    {
+      "epoch": 3.972413793103448,
+      "loss": 0.14658576250076294,
+      "loss_ce": 9.828477232076693e-06,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 77988320,
+      "step": 864
+    },
+    {
+      "epoch": 3.9770114942528734,
+      "grad_norm": 6.60885795449724,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 78078664,
+      "step": 865
+    },
+    {
+      "epoch": 3.9770114942528734,
+      "loss": 0.14846225082874298,
+      "loss_ce": 4.0008031646721065e-05,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 78078664,
+      "step": 865
+    },
+    {
+      "epoch": 3.981609195402299,
+      "grad_norm": 8.750514149287007,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 78168928,
+      "step": 866
+    },
+    {
+      "epoch": 3.981609195402299,
+      "loss": 0.11572016775608063,
+      "loss_ce": 2.8033886337652802e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 78168928,
+      "step": 866
+    },
+    {
+      "epoch": 3.986206896551724,
+      "grad_norm": 7.203279541438859,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 78259188,
+      "step": 867
+    },
+    {
+      "epoch": 3.986206896551724,
+      "loss": 0.10868757218122482,
+      "loss_ce": 1.4477924196398817e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 78259188,
+      "step": 867
+    },
+    {
+      "epoch": 3.9908045977011493,
+      "grad_norm": 3.5527677092629077,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 78349456,
+      "step": 868
+    },
+    {
+      "epoch": 3.9908045977011493,
+      "loss": 0.061215296387672424,
+      "loss_ce": 2.7547748686629348e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 78349456,
+      "step": 868
+    },
+    {
+      "epoch": 3.995402298850575,
+      "grad_norm": 13.601719729469064,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 78439776,
+      "step": 869
+    },
+    {
+      "epoch": 3.995402298850575,
+      "loss": 0.06812618672847748,
+      "loss_ce": 2.621453495521564e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 78439776,
+      "step": 869
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 12.850021260799155,
+      "learning_rate": 5e-06,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 78530044,
+      "step": 870
+    },
+    {
+      "epoch": 4.0,
+      "loss": 0.07683277875185013,
+      "loss_ce": 4.778397851623595e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 78530044,
+      "step": 870
+    },
+    {
+      "epoch": 4.004597701149425,
+      "grad_norm": 12.896033435436056,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 78620484,
+      "step": 871
+    },
+    {
+      "epoch": 4.004597701149425,
+      "loss": 0.07385321706533432,
+      "loss_ce": 1.593406886968296e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 78620484,
+      "step": 871
+    },
+    {
+      "epoch": 4.00919540229885,
+      "grad_norm": 10.038095509850997,
+      "learning_rate": 5e-06,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 78710856,
+      "step": 872
+    },
+    {
+      "epoch": 4.00919540229885,
+      "loss": 0.12039355933666229,
+      "loss_ce": 0.00012378332030493766,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 78710856,
+      "step": 872
+    },
+    {
+      "epoch": 4.0137931034482754,
+      "grad_norm": 7.577320908621721,
+      "learning_rate": 5e-06,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 78801316,
+      "step": 873
+    },
+    {
+      "epoch": 4.0137931034482754,
+      "loss": 0.08987805247306824,
+      "loss_ce": 3.429901698837057e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 78801316,
+      "step": 873
+    },
+    {
+      "epoch": 4.0183908045977015,
+      "grad_norm": 8.350163864558647,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 78891520,
+      "step": 874
+    },
+    {
+      "epoch": 4.0183908045977015,
+      "loss": 0.0876561775803566,
+      "loss_ce": 9.694716936792247e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 78891520,
+      "step": 874
+    },
+    {
+      "epoch": 4.022988505747127,
+      "grad_norm": 8.769669621263414,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 78981876,
+      "step": 875
+    },
+    {
+      "epoch": 4.022988505747127,
+      "loss": 0.15618135035037994,
+      "loss_ce": 2.289763870066963e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 78981876,
+      "step": 875
+    },
+    {
+      "epoch": 4.027586206896552,
+      "grad_norm": 16.167283950116676,
+      "learning_rate": 5e-06,
+      "loss": 0.1022,
+      "num_input_tokens_seen": 79072252,
+      "step": 876
+    },
+    {
+      "epoch": 4.027586206896552,
+      "loss": 0.09752378612756729,
+      "loss_ce": 2.0128632968408056e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 79072252,
+      "step": 876
+    },
+    {
+      "epoch": 4.032183908045977,
+      "grad_norm": 8.509588312264029,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 79162600,
+      "step": 877
+    },
+    {
+      "epoch": 4.032183908045977,
+      "loss": 0.1270662546157837,
+      "loss_ce": 8.26121904538013e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 79162600,
+      "step": 877
+    },
+    {
+      "epoch": 4.036781609195402,
+      "grad_norm": 10.729375299726154,
+      "learning_rate": 5e-06,
+      "loss": 0.0858,
+      "num_input_tokens_seen": 79253000,
+      "step": 878
+    },
+    {
+      "epoch": 4.036781609195402,
+      "loss": 0.07850313931703568,
+      "loss_ce": 1.1927713785553351e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 79253000,
+      "step": 878
+    },
+    {
+      "epoch": 4.041379310344827,
+      "grad_norm": 9.480711461561928,
+      "learning_rate": 5e-06,
+      "loss": 0.092,
+      "num_input_tokens_seen": 79343448,
+      "step": 879
+    },
+    {
+      "epoch": 4.041379310344827,
+      "loss": 0.11301976442337036,
+      "loss_ce": 4.369192538433708e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 79343448,
+      "step": 879
+    },
+    {
+      "epoch": 4.045977011494253,
+      "grad_norm": 20.962984913528285,
+      "learning_rate": 5e-06,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 79433948,
+      "step": 880
+    },
+    {
+      "epoch": 4.045977011494253,
+      "loss": 0.060304559767246246,
+      "loss_ce": 1.8244732018501963e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 79433948,
+      "step": 880
+    },
+    {
+      "epoch": 4.050574712643678,
+      "grad_norm": 27.220075486414753,
+      "learning_rate": 5e-06,
+      "loss": 0.082,
+      "num_input_tokens_seen": 79524228,
+      "step": 881
+    },
+    {
+      "epoch": 4.050574712643678,
+      "loss": 0.07480692863464355,
+      "loss_ce": 8.341698048752733e-06,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 79524228,
+      "step": 881
+    },
+    {
+      "epoch": 4.055172413793104,
+      "grad_norm": 9.60955652874984,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 79614560,
+      "step": 882
+    },
+    {
+      "epoch": 4.055172413793104,
+      "loss": 0.2088443636894226,
+      "loss_ce": 0.00010411619587102905,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 79614560,
+      "step": 882
+    },
+    {
+      "epoch": 4.059770114942529,
+      "grad_norm": 13.229041315882611,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 79705048,
+      "step": 883
+    },
+    {
+      "epoch": 4.059770114942529,
+      "loss": 0.11971049755811691,
+      "loss_ce": 0.00014262790500652045,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 79705048,
+      "step": 883
+    },
+    {
+      "epoch": 4.064367816091954,
+      "grad_norm": 11.345287474225346,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 79795468,
+      "step": 884
+    },
+    {
+      "epoch": 4.064367816091954,
+      "loss": 0.15525750815868378,
+      "loss_ce": 7.562241080449894e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 79795468,
+      "step": 884
+    },
+    {
+      "epoch": 4.068965517241379,
+      "grad_norm": 5.046401225257881,
+      "learning_rate": 5e-06,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 79885904,
+      "step": 885
+    },
+    {
+      "epoch": 4.068965517241379,
+      "loss": 0.09430328011512756,
+      "loss_ce": 3.965928954130504e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 79885904,
+      "step": 885
+    },
+    {
+      "epoch": 4.073563218390804,
+      "grad_norm": 7.149591467541043,
+      "learning_rate": 5e-06,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 79976308,
+      "step": 886
+    },
+    {
+      "epoch": 4.073563218390804,
+      "loss": 0.06827942281961441,
+      "loss_ce": 2.6857467673835345e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 79976308,
+      "step": 886
+    },
+    {
+      "epoch": 4.07816091954023,
+      "grad_norm": 14.125533156236072,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 80066680,
+      "step": 887
+    },
+    {
+      "epoch": 4.07816091954023,
+      "loss": 0.11485080420970917,
+      "loss_ce": 1.3159661648387555e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 80066680,
+      "step": 887
+    },
+    {
+      "epoch": 4.082758620689655,
+      "grad_norm": 7.34272733165029,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 80157060,
+      "step": 888
+    },
+    {
+      "epoch": 4.082758620689655,
+      "loss": 0.13511279225349426,
+      "loss_ce": 7.250368071254343e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 80157060,
+      "step": 888
+    },
+    {
+      "epoch": 4.0873563218390805,
+      "grad_norm": 7.121447843023311,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 80247628,
+      "step": 889
+    },
+    {
+      "epoch": 4.0873563218390805,
+      "loss": 0.1320275515317917,
+      "loss_ce": 8.516525667801034e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 80247628,
+      "step": 889
+    },
+    {
+      "epoch": 4.091954022988506,
+      "grad_norm": 9.329754076401972,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 80337164,
+      "step": 890
+    },
+    {
+      "epoch": 4.091954022988506,
+      "loss": 0.10165956616401672,
+      "loss_ce": 5.516031251318054e-06,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 80337164,
+      "step": 890
+    },
+    {
+      "epoch": 4.096551724137931,
+      "grad_norm": 4.7416331340636875,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 80426632,
+      "step": 891
+    },
+    {
+      "epoch": 4.096551724137931,
+      "loss": 0.14343947172164917,
+      "loss_ce": 3.7382742448244244e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 80426632,
+      "step": 891
+    },
+    {
+      "epoch": 4.101149425287356,
+      "grad_norm": 23.466418685698162,
+      "learning_rate": 5e-06,
+      "loss": 0.1543,
+      "num_input_tokens_seen": 80516208,
+      "step": 892
+    },
+    {
+      "epoch": 4.101149425287356,
+      "loss": 0.18008266389369965,
+      "loss_ce": 2.8955022571608424e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 80516208,
+      "step": 892
+    },
+    {
+      "epoch": 4.105747126436782,
+      "grad_norm": 20.195736171276238,
+      "learning_rate": 5e-06,
+      "loss": 0.103,
+      "num_input_tokens_seen": 80606536,
+      "step": 893
+    },
+    {
+      "epoch": 4.105747126436782,
+      "loss": 0.09322860836982727,
+      "loss_ce": 5.844266706844792e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 80606536,
+      "step": 893
+    },
+    {
+      "epoch": 4.110344827586207,
+      "grad_norm": 4.925478130148734,
+      "learning_rate": 5e-06,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 80696904,
+      "step": 894
+    },
+    {
+      "epoch": 4.110344827586207,
+      "loss": 0.1146705150604248,
+      "loss_ce": 1.5975076166796498e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 80696904,
+      "step": 894
+    },
+    {
+      "epoch": 4.114942528735632,
+      "grad_norm": 18.783331502764156,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 80787324,
+      "step": 895
+    },
+    {
+      "epoch": 4.114942528735632,
+      "loss": 0.10798490047454834,
+      "loss_ce": 0.0011123453732579947,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 80787324,
+      "step": 895
+    },
+    {
+      "epoch": 4.119540229885057,
+      "grad_norm": 6.031295114157052,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 80877804,
+      "step": 896
+    },
+    {
+      "epoch": 4.119540229885057,
+      "loss": 0.12141025811433792,
+      "loss_ce": 8.762812649365515e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 80877804,
+      "step": 896
+    },
+    {
+      "epoch": 4.124137931034483,
+      "grad_norm": 4.270748379873812,
+      "learning_rate": 5e-06,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 80967380,
+      "step": 897
+    },
+    {
+      "epoch": 4.124137931034483,
+      "loss": 0.08645440638065338,
+      "loss_ce": 2.8627207939280197e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 80967380,
+      "step": 897
+    },
+    {
+      "epoch": 4.128735632183908,
+      "grad_norm": 4.246728442816423,
+      "learning_rate": 5e-06,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 81057784,
+      "step": 898
+    },
+    {
+      "epoch": 4.128735632183908,
+      "loss": 0.09482555836439133,
+      "loss_ce": 6.847563054179773e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 81057784,
+      "step": 898
+    },
+    {
+      "epoch": 4.133333333333334,
+      "grad_norm": 4.522025434777465,
+      "learning_rate": 5e-06,
+      "loss": 0.1549,
+      "num_input_tokens_seen": 81148164,
+      "step": 899
+    },
+    {
+      "epoch": 4.133333333333334,
+      "loss": 0.20047786831855774,
+      "loss_ce": 3.8414596929214895e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 81148164,
+      "step": 899
+    },
+    {
+      "epoch": 4.137931034482759,
+      "grad_norm": 12.60515016043542,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 81237064,
+      "step": 900
+    },
+    {
+      "epoch": 4.137931034482759,
+      "loss": 0.09033560007810593,
+      "loss_ce": 3.5681719054991845e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 81237064,
+      "step": 900
+    },
+    {
+      "epoch": 4.142528735632184,
+      "grad_norm": 12.203513946052261,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 81327320,
+      "step": 901
+    },
+    {
+      "epoch": 4.142528735632184,
+      "loss": 0.09698610007762909,
+      "loss_ce": 1.236280354532937e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 81327320,
+      "step": 901
+    },
+    {
+      "epoch": 4.147126436781609,
+      "grad_norm": 5.265820346765914,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 81417700,
+      "step": 902
+    },
+    {
+      "epoch": 4.147126436781609,
+      "loss": 0.13237115740776062,
+      "loss_ce": 1.6427336959168315e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 81417700,
+      "step": 902
+    },
+    {
+      "epoch": 4.151724137931034,
+      "grad_norm": 5.402431273469456,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 81508048,
+      "step": 903
+    },
+    {
+      "epoch": 4.151724137931034,
+      "loss": 0.10200782120227814,
+      "loss_ce": 1.8078684661304578e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 81508048,
+      "step": 903
+    },
+    {
+      "epoch": 4.1563218390804595,
+      "grad_norm": 7.802227527691613,
+      "learning_rate": 5e-06,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 81598384,
+      "step": 904
+    },
+    {
+      "epoch": 4.1563218390804595,
+      "loss": 0.0613405816257,
+      "loss_ce": 4.60283481515944e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 81598384,
+      "step": 904
+    },
+    {
+      "epoch": 4.160919540229885,
+      "grad_norm": 16.9420578658568,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 81688732,
+      "step": 905
+    },
+    {
+      "epoch": 4.160919540229885,
+      "loss": 0.16473174095153809,
+      "loss_ce": 1.3099732314003631e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 81688732,
+      "step": 905
+    },
+    {
+      "epoch": 4.165517241379311,
+      "grad_norm": 3.5038961317329633,
+      "learning_rate": 5e-06,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 81779124,
+      "step": 906
+    },
+    {
+      "epoch": 4.165517241379311,
+      "loss": 0.09057803452014923,
+      "loss_ce": 1.8635427068147692e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 81779124,
+      "step": 906
+    },
+    {
+      "epoch": 4.170114942528736,
+      "grad_norm": 6.320699544904618,
+      "learning_rate": 5e-06,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 81869528,
+      "step": 907
+    },
+    {
+      "epoch": 4.170114942528736,
+      "loss": 0.0921754315495491,
+      "loss_ce": 4.286535840947181e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 81869528,
+      "step": 907
+    },
+    {
+      "epoch": 4.174712643678161,
+      "grad_norm": 20.78428241303839,
+      "learning_rate": 5e-06,
+      "loss": 0.1552,
+      "num_input_tokens_seen": 81959744,
+      "step": 908
+    },
+    {
+      "epoch": 4.174712643678161,
+      "loss": 0.17669041454792023,
+      "loss_ce": 8.907296432880685e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 81959744,
+      "step": 908
+    },
+    {
+      "epoch": 4.179310344827586,
+      "grad_norm": 8.303007780105153,
+      "learning_rate": 5e-06,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 82050236,
+      "step": 909
+    },
+    {
+      "epoch": 4.179310344827586,
+      "loss": 0.04546171426773071,
+      "loss_ce": 2.1037711121607572e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 82050236,
+      "step": 909
+    },
+    {
+      "epoch": 4.183908045977011,
+      "grad_norm": 11.886180061278875,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 82140600,
+      "step": 910
+    },
+    {
+      "epoch": 4.183908045977011,
+      "loss": 0.10049276053905487,
+      "loss_ce": 2.888959352276288e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 82140600,
+      "step": 910
+    },
+    {
+      "epoch": 4.188505747126436,
+      "grad_norm": 8.89001202277643,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 82231004,
+      "step": 911
+    },
+    {
+      "epoch": 4.188505747126436,
+      "loss": 0.156759113073349,
+      "loss_ce": 2.0833926100749522e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 82231004,
+      "step": 911
+    },
+    {
+      "epoch": 4.1931034482758625,
+      "grad_norm": 12.14512131743274,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 82321456,
+      "step": 912
+    },
+    {
+      "epoch": 4.1931034482758625,
+      "loss": 0.0638713464140892,
+      "loss_ce": 2.8572279916261323e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 82321456,
+      "step": 912
+    },
+    {
+      "epoch": 4.197701149425288,
+      "grad_norm": 6.273360573385607,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 82411704,
+      "step": 913
+    },
+    {
+      "epoch": 4.197701149425288,
+      "loss": 0.08992119133472443,
+      "loss_ce": 1.1499548691062955e-06,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 82411704,
+      "step": 913
+    },
+    {
+      "epoch": 4.202298850574713,
+      "grad_norm": 4.863327392449604,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 82500532,
+      "step": 914
+    },
+    {
+      "epoch": 4.202298850574713,
+      "loss": 0.15036578476428986,
+      "loss_ce": 5.677235094481148e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 82500532,
+      "step": 914
+    },
+    {
+      "epoch": 4.206896551724138,
+      "grad_norm": 6.0320903674371795,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 82591044,
+      "step": 915
+    },
+    {
+      "epoch": 4.206896551724138,
+      "loss": 0.10052984952926636,
+      "loss_ce": 3.5464396205497906e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 82591044,
+      "step": 915
+    },
+    {
+      "epoch": 4.211494252873563,
+      "grad_norm": 15.82805235612327,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 82681520,
+      "step": 916
+    },
+    {
+      "epoch": 4.211494252873563,
+      "loss": 0.09116413444280624,
+      "loss_ce": 2.3386701286653988e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 82681520,
+      "step": 916
+    },
+    {
+      "epoch": 4.216091954022988,
+      "grad_norm": 2.3437945707030305,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 82771904,
+      "step": 917
+    },
+    {
+      "epoch": 4.216091954022988,
+      "loss": 0.09512968361377716,
+      "loss_ce": 6.396087883331347e-06,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 82771904,
+      "step": 917
+    },
+    {
+      "epoch": 4.220689655172414,
+      "grad_norm": 8.364208607807933,
+      "learning_rate": 5e-06,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 82862244,
+      "step": 918
+    },
+    {
+      "epoch": 4.220689655172414,
+      "loss": 0.08399280905723572,
+      "loss_ce": 8.435705240117386e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 82862244,
+      "step": 918
+    },
+    {
+      "epoch": 4.225287356321839,
+      "grad_norm": 3.326406479061023,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 82952644,
+      "step": 919
+    },
+    {
+      "epoch": 4.225287356321839,
+      "loss": 0.06023194640874863,
+      "loss_ce": 2.0768720787600614e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 82952644,
+      "step": 919
+    },
+    {
+      "epoch": 4.2298850574712645,
+      "grad_norm": 3.4665111443776944,
+      "learning_rate": 5e-06,
+      "loss": 0.131,
+      "num_input_tokens_seen": 83042996,
+      "step": 920
+    },
+    {
+      "epoch": 4.2298850574712645,
+      "loss": 0.19263532757759094,
+      "loss_ce": 8.370232535526156e-06,
+      "loss_iou": 0.40625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 83042996,
+      "step": 920
+    },
+    {
+      "epoch": 4.23448275862069,
+      "grad_norm": 9.069959650004307,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 83133396,
+      "step": 921
+    },
+    {
+      "epoch": 4.23448275862069,
+      "loss": 0.11099687963724136,
+      "loss_ce": 3.496557837934233e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 83133396,
+      "step": 921
+    },
+    {
+      "epoch": 4.239080459770115,
+      "grad_norm": 2.5690493136384003,
+      "learning_rate": 5e-06,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 83223872,
+      "step": 922
+    },
+    {
+      "epoch": 4.239080459770115,
+      "loss": 0.11521795392036438,
+      "loss_ce": 1.4101822671364062e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 83223872,
+      "step": 922
+    },
+    {
+      "epoch": 4.24367816091954,
+      "grad_norm": 7.69995972428386,
+      "learning_rate": 5e-06,
+      "loss": 0.1505,
+      "num_input_tokens_seen": 83314296,
+      "step": 923
+    },
+    {
+      "epoch": 4.24367816091954,
+      "loss": 0.125550776720047,
+      "loss_ce": 1.4667482446384383e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 83314296,
+      "step": 923
+    },
+    {
+      "epoch": 4.248275862068965,
+      "grad_norm": 6.582981564440316,
+      "learning_rate": 5e-06,
+      "loss": 0.1091,
+      "num_input_tokens_seen": 83404576,
+      "step": 924
+    },
+    {
+      "epoch": 4.248275862068965,
+      "loss": 0.11165105551481247,
+      "loss_ce": 1.7746649973560125e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 83404576,
+      "step": 924
+    },
+    {
+      "epoch": 4.252873563218391,
+      "grad_norm": 3.1233405856262006,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 83494884,
+      "step": 925
+    },
+    {
+      "epoch": 4.252873563218391,
+      "loss": 0.13092860579490662,
+      "loss_ce": 8.193834219127893e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 83494884,
+      "step": 925
+    },
+    {
+      "epoch": 4.257471264367816,
+      "grad_norm": 11.295201522679497,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 83585372,
+      "step": 926
+    },
+    {
+      "epoch": 4.257471264367816,
+      "loss": 0.09147991985082626,
+      "loss_ce": 3.3995776902884245e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 83585372,
+      "step": 926
+    },
+    {
+      "epoch": 4.2620689655172415,
+      "grad_norm": 8.74331179817824,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 83675776,
+      "step": 927
+    },
+    {
+      "epoch": 4.2620689655172415,
+      "loss": 0.11818952858448029,
+      "loss_ce": 1.0204122190771159e-05,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 83675776,
+      "step": 927
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 36.691461336314475,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 83766024,
+      "step": 928
+    },
+    {
+      "epoch": 4.266666666666667,
+      "loss": 0.11256686598062515,
+      "loss_ce": 2.7873388717125636e-06,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 83766024,
+      "step": 928
+    },
+    {
+      "epoch": 4.271264367816092,
+      "grad_norm": 20.28534876927393,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 83856372,
+      "step": 929
+    },
+    {
+      "epoch": 4.271264367816092,
+      "loss": 0.1561427265405655,
+      "loss_ce": 0.0033259547781199217,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 83856372,
+      "step": 929
+    },
+    {
+      "epoch": 4.275862068965517,
+      "grad_norm": 21.074305399488686,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 83946688,
+      "step": 930
+    },
+    {
+      "epoch": 4.275862068965517,
+      "loss": 0.10977804660797119,
+      "loss_ce": 6.307951480266638e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 83946688,
+      "step": 930
+    },
+    {
+      "epoch": 4.280459770114943,
+      "grad_norm": 18.474400113113084,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 84037100,
+      "step": 931
+    },
+    {
+      "epoch": 4.280459770114943,
+      "loss": 0.08449871093034744,
+      "loss_ce": 5.657064320985228e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 84037100,
+      "step": 931
+    },
+    {
+      "epoch": 4.285057471264368,
+      "grad_norm": 6.907766753439544,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 84127468,
+      "step": 932
+    },
+    {
+      "epoch": 4.285057471264368,
+      "loss": 0.10681600868701935,
+      "loss_ce": 4.487107617023867e-06,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 84127468,
+      "step": 932
+    },
+    {
+      "epoch": 4.289655172413793,
+      "grad_norm": 11.040747277806908,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 84216928,
+      "step": 933
+    },
+    {
+      "epoch": 4.289655172413793,
+      "loss": 0.1147032380104065,
+      "loss_ce": 4.869817348662764e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 84216928,
+      "step": 933
+    },
+    {
+      "epoch": 4.294252873563218,
+      "grad_norm": 5.536586522856837,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 84307292,
+      "step": 934
+    },
+    {
+      "epoch": 4.294252873563218,
+      "loss": 0.14283494651317596,
+      "loss_ce": 0.0001957785279955715,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 84307292,
+      "step": 934
+    },
+    {
+      "epoch": 4.2988505747126435,
+      "grad_norm": 2.131997523716501,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 84396872,
+      "step": 935
+    },
+    {
+      "epoch": 4.2988505747126435,
+      "loss": 0.12867018580436707,
+      "loss_ce": 8.078463906713296e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 84396872,
+      "step": 935
+    },
+    {
+      "epoch": 4.303448275862069,
+      "grad_norm": 11.27288052453101,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 84487316,
+      "step": 936
+    },
+    {
+      "epoch": 4.303448275862069,
+      "loss": 0.06220962479710579,
+      "loss_ce": 0.0010371371172368526,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 84487316,
+      "step": 936
+    },
+    {
+      "epoch": 4.308045977011494,
+      "grad_norm": 3.4892119545715974,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 84577824,
+      "step": 937
+    },
+    {
+      "epoch": 4.308045977011494,
+      "loss": 0.1438092142343521,
+      "loss_ce": 1.0385462701378856e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 84577824,
+      "step": 937
+    },
+    {
+      "epoch": 4.31264367816092,
+      "grad_norm": 6.088518946016204,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 84668208,
+      "step": 938
+    },
+    {
+      "epoch": 4.31264367816092,
+      "loss": 0.08154906332492828,
+      "loss_ce": 2.134743226633873e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 84668208,
+      "step": 938
+    },
+    {
+      "epoch": 4.317241379310345,
+      "grad_norm": 10.741692080234259,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 84758388,
+      "step": 939
+    },
+    {
+      "epoch": 4.317241379310345,
+      "loss": 0.12125536799430847,
+      "loss_ce": 9.027476153278258e-06,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 84758388,
+      "step": 939
+    },
+    {
+      "epoch": 4.32183908045977,
+      "grad_norm": 12.104640117183845,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 84848688,
+      "step": 940
+    },
+    {
+      "epoch": 4.32183908045977,
+      "loss": 0.125244140625,
+      "loss_ce": 0.0001831004919949919,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 84848688,
+      "step": 940
+    },
+    {
+      "epoch": 4.326436781609195,
+      "grad_norm": 13.842704032927974,
+      "learning_rate": 5e-06,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 84938896,
+      "step": 941
+    },
+    {
+      "epoch": 4.326436781609195,
+      "loss": 0.08458052575588226,
+      "loss_ce": 1.0556854022070183e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 84938896,
+      "step": 941
+    },
+    {
+      "epoch": 4.3310344827586205,
+      "grad_norm": 27.64969062986627,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 85029124,
+      "step": 942
+    },
+    {
+      "epoch": 4.3310344827586205,
+      "loss": 0.11936682462692261,
+      "loss_ce": 0.00025671368348412216,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 85029124,
+      "step": 942
+    },
+    {
+      "epoch": 4.335632183908046,
+      "grad_norm": 7.051941352640006,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 85119572,
+      "step": 943
+    },
+    {
+      "epoch": 4.335632183908046,
+      "loss": 0.08214512467384338,
+      "loss_ce": 7.063375960569829e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 85119572,
+      "step": 943
+    },
+    {
+      "epoch": 4.340229885057472,
+      "grad_norm": 5.222428783369038,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 85209984,
+      "step": 944
+    },
+    {
+      "epoch": 4.340229885057472,
+      "loss": 0.13096053898334503,
+      "loss_ce": 9.61720070336014e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 85209984,
+      "step": 944
+    },
+    {
+      "epoch": 4.344827586206897,
+      "grad_norm": 8.261994222444807,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 85300436,
+      "step": 945
+    },
+    {
+      "epoch": 4.344827586206897,
+      "loss": 0.11582186818122864,
+      "loss_ce": 0.0001144730849773623,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 85300436,
+      "step": 945
+    },
+    {
+      "epoch": 4.349425287356322,
+      "grad_norm": 4.239609844753722,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 85390688,
+      "step": 946
+    },
+    {
+      "epoch": 4.349425287356322,
+      "loss": 0.08635582774877548,
+      "loss_ce": 6.338353159662802e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 85390688,
+      "step": 946
+    },
+    {
+      "epoch": 4.354022988505747,
+      "grad_norm": 15.411419922368433,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 85481212,
+      "step": 947
+    },
+    {
+      "epoch": 4.354022988505747,
+      "loss": 0.11237078905105591,
+      "loss_ce": 5.069013241154607e-06,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 85481212,
+      "step": 947
+    },
+    {
+      "epoch": 4.358620689655172,
+      "grad_norm": 3.634268474402103,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 85570804,
+      "step": 948
+    },
+    {
+      "epoch": 4.358620689655172,
+      "loss": 0.1522293984889984,
+      "loss_ce": 7.724566785327625e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 85570804,
+      "step": 948
+    },
+    {
+      "epoch": 4.363218390804597,
+      "grad_norm": 8.013548594881062,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 85661312,
+      "step": 949
+    },
+    {
+      "epoch": 4.363218390804597,
+      "loss": 0.14923954010009766,
+      "loss_ce": 8.487531158607453e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 85661312,
+      "step": 949
+    },
+    {
+      "epoch": 4.3678160919540225,
+      "grad_norm": 2.762105405490039,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 85751636,
+      "step": 950
+    },
+    {
+      "epoch": 4.3678160919540225,
+      "loss": 0.1016370952129364,
+      "loss_ce": 1.3553465578297619e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 85751636,
+      "step": 950
+    },
+    {
+      "epoch": 4.372413793103449,
+      "grad_norm": 5.105256776246609,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 85842060,
+      "step": 951
+    },
+    {
+      "epoch": 4.372413793103449,
+      "loss": 0.10755231976509094,
+      "loss_ce": 8.379087375942618e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 85842060,
+      "step": 951
+    },
+    {
+      "epoch": 4.377011494252874,
+      "grad_norm": 5.570764796509669,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 85932444,
+      "step": 952
+    },
+    {
+      "epoch": 4.377011494252874,
+      "loss": 0.0883231908082962,
+      "loss_ce": 3.58409270120319e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 85932444,
+      "step": 952
+    },
+    {
+      "epoch": 4.381609195402299,
+      "grad_norm": 10.856984323349382,
+      "learning_rate": 5e-06,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 86022780,
+      "step": 953
+    },
+    {
+      "epoch": 4.381609195402299,
+      "loss": 0.18059954047203064,
+      "loss_ce": 5.753596997237764e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 86022780,
+      "step": 953
+    },
+    {
+      "epoch": 4.386206896551724,
+      "grad_norm": 13.39000045516796,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 86113048,
+      "step": 954
+    },
+    {
+      "epoch": 4.386206896551724,
+      "loss": 0.10757958889007568,
+      "loss_ce": 3.564245707821101e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 86113048,
+      "step": 954
+    },
+    {
+      "epoch": 4.390804597701149,
+      "grad_norm": 3.0587557281723505,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 86203536,
+      "step": 955
+    },
+    {
+      "epoch": 4.390804597701149,
+      "loss": 0.08970493823289871,
+      "loss_ce": 1.377277476422023e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 86203536,
+      "step": 955
+    },
+    {
+      "epoch": 4.395402298850574,
+      "grad_norm": 7.714066456128423,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 86293828,
+      "step": 956
+    },
+    {
+      "epoch": 4.395402298850574,
+      "loss": 0.14037680625915527,
+      "loss_ce": 8.751240238780156e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 86293828,
+      "step": 956
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 6.43769065897863,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 86384244,
+      "step": 957
+    },
+    {
+      "epoch": 4.4,
+      "loss": 0.12558409571647644,
+      "loss_ce": 0.0003399590204935521,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 86384244,
+      "step": 957
+    },
+    {
+      "epoch": 4.4045977011494255,
+      "grad_norm": 7.325715877251544,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 86474564,
+      "step": 958
+    },
+    {
+      "epoch": 4.4045977011494255,
+      "loss": 0.07348855584859848,
+      "loss_ce": 3.274789924034849e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 86474564,
+      "step": 958
+    },
+    {
+      "epoch": 4.409195402298851,
+      "grad_norm": 3.1430755970159763,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 86565012,
+      "step": 959
+    },
+    {
+      "epoch": 4.409195402298851,
+      "loss": 0.10845954716205597,
+      "loss_ce": 9.163121285382658e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 86565012,
+      "step": 959
+    },
+    {
+      "epoch": 4.413793103448276,
+      "grad_norm": 2.479253070842102,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 86655508,
+      "step": 960
+    },
+    {
+      "epoch": 4.413793103448276,
+      "loss": 0.09556898474693298,
+      "loss_ce": 3.370657577761449e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 86655508,
+      "step": 960
+    },
+    {
+      "epoch": 4.418390804597701,
+      "grad_norm": 8.037612054391559,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 86744356,
+      "step": 961
+    },
+    {
+      "epoch": 4.418390804597701,
+      "loss": 0.13194900751113892,
+      "loss_ce": 2.151706939912401e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 86744356,
+      "step": 961
+    },
+    {
+      "epoch": 4.422988505747126,
+      "grad_norm": 4.558636962872201,
+      "learning_rate": 5e-06,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 86834692,
+      "step": 962
+    },
+    {
+      "epoch": 4.422988505747126,
+      "loss": 0.05887310206890106,
+      "loss_ce": 4.69345968667767e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 86834692,
+      "step": 962
+    },
+    {
+      "epoch": 4.427586206896552,
+      "grad_norm": 11.446145366052058,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 86925088,
+      "step": 963
+    },
+    {
+      "epoch": 4.427586206896552,
+      "loss": 0.0905739963054657,
+      "loss_ce": 1.3087726983940229e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 86925088,
+      "step": 963
+    },
+    {
+      "epoch": 4.432183908045977,
+      "grad_norm": 9.136099668207656,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 87015388,
+      "step": 964
+    },
+    {
+      "epoch": 4.432183908045977,
+      "loss": 0.16606704890727997,
+      "loss_ce": 2.0905536075588316e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 87015388,
+      "step": 964
+    },
+    {
+      "epoch": 4.436781609195402,
+      "grad_norm": 5.4148068009345645,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 87104260,
+      "step": 965
+    },
+    {
+      "epoch": 4.436781609195402,
+      "loss": 0.1504197120666504,
+      "loss_ce": 5.959868576610461e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 87104260,
+      "step": 965
+    },
+    {
+      "epoch": 4.441379310344828,
+      "grad_norm": 7.551140001822664,
+      "learning_rate": 5e-06,
+      "loss": 0.072,
+      "num_input_tokens_seen": 87194688,
+      "step": 966
+    },
+    {
+      "epoch": 4.441379310344828,
+      "loss": 0.0605931356549263,
+      "loss_ce": 0.0001225552405230701,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 87194688,
+      "step": 966
+    },
+    {
+      "epoch": 4.445977011494253,
+      "grad_norm": 5.492004326107786,
+      "learning_rate": 5e-06,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 87285044,
+      "step": 967
+    },
+    {
+      "epoch": 4.445977011494253,
+      "loss": 0.13285866379737854,
+      "loss_ce": 4.616224396158941e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 87285044,
+      "step": 967
+    },
+    {
+      "epoch": 4.450574712643678,
+      "grad_norm": 7.815560936649822,
+      "learning_rate": 5e-06,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 87375332,
+      "step": 968
+    },
+    {
+      "epoch": 4.450574712643678,
+      "loss": 0.12036024034023285,
+      "loss_ce": 2.943439358205069e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 87375332,
+      "step": 968
+    },
+    {
+      "epoch": 4.455172413793104,
+      "grad_norm": 19.127660857216004,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 87465720,
+      "step": 969
+    },
+    {
+      "epoch": 4.455172413793104,
+      "loss": 0.09645415097475052,
+      "loss_ce": 3.339698650961509e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 87465720,
+      "step": 969
+    },
+    {
+      "epoch": 4.459770114942529,
+      "grad_norm": 2.8889092802079226,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 87555976,
+      "step": 970
+    },
+    {
+      "epoch": 4.459770114942529,
+      "loss": 0.07163398712873459,
+      "loss_ce": 9.234347089659423e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 87555976,
+      "step": 970
+    },
+    {
+      "epoch": 4.464367816091954,
+      "grad_norm": 4.010750930380829,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 87646328,
+      "step": 971
+    },
+    {
+      "epoch": 4.464367816091954,
+      "loss": 0.11213794350624084,
+      "loss_ce": 1.108814331018948e-06,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 87646328,
+      "step": 971
+    },
+    {
+      "epoch": 4.468965517241379,
+      "grad_norm": 8.210323216736478,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 87736608,
+      "step": 972
+    },
+    {
+      "epoch": 4.468965517241379,
+      "loss": 0.18182888627052307,
+      "loss_ce": 5.140444045537151e-06,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 87736608,
+      "step": 972
+    },
+    {
+      "epoch": 4.4735632183908045,
+      "grad_norm": 9.865783278700588,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 87826276,
+      "step": 973
+    },
+    {
+      "epoch": 4.4735632183908045,
+      "loss": 0.06133665144443512,
+      "loss_ce": 1.1581903891055845e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 87826276,
+      "step": 973
+    },
+    {
+      "epoch": 4.47816091954023,
+      "grad_norm": 4.060783363175244,
+      "learning_rate": 5e-06,
+      "loss": 0.1517,
+      "num_input_tokens_seen": 87915748,
+      "step": 974
+    },
+    {
+      "epoch": 4.47816091954023,
+      "loss": 0.1866350769996643,
+      "loss_ce": 2.0090221369173378e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 87915748,
+      "step": 974
+    },
+    {
+      "epoch": 4.482758620689655,
+      "grad_norm": 4.078469036538909,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 88006180,
+      "step": 975
+    },
+    {
+      "epoch": 4.482758620689655,
+      "loss": 0.16530264914035797,
+      "loss_ce": 1.9446521037025377e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 88006180,
+      "step": 975
+    },
+    {
+      "epoch": 4.487356321839081,
+      "grad_norm": 14.551219328912012,
+      "learning_rate": 5e-06,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 88096576,
+      "step": 976
+    },
+    {
+      "epoch": 4.487356321839081,
+      "loss": 0.09992989897727966,
+      "loss_ce": 1.5359226381406188e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 88096576,
+      "step": 976
+    },
+    {
+      "epoch": 4.491954022988506,
+      "grad_norm": 15.074496002244894,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 88186864,
+      "step": 977
+    },
+    {
+      "epoch": 4.491954022988506,
+      "loss": 0.098409004509449,
+      "loss_ce": 5.076651632407447e-06,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 88186864,
+      "step": 977
+    },
+    {
+      "epoch": 4.496551724137931,
+      "grad_norm": 8.357602751730505,
+      "learning_rate": 5e-06,
+      "loss": 0.094,
+      "num_input_tokens_seen": 88277108,
+      "step": 978
+    },
+    {
+      "epoch": 4.496551724137931,
+      "loss": 0.12131184339523315,
+      "loss_ce": 4.475433343031909e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 88277108,
+      "step": 978
+    },
+    {
+      "epoch": 4.501149425287356,
+      "grad_norm": 4.402022300916744,
+      "learning_rate": 5e-06,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 88367416,
+      "step": 979
+    },
+    {
+      "epoch": 4.501149425287356,
+      "loss": 0.07329022884368896,
+      "loss_ce": 7.855774310883135e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 88367416,
+      "step": 979
+    },
+    {
+      "epoch": 4.505747126436781,
+      "grad_norm": 14.341101022137071,
+      "learning_rate": 5e-06,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 88457876,
+      "step": 980
+    },
+    {
+      "epoch": 4.505747126436781,
+      "loss": 0.053125277161598206,
+      "loss_ce": 9.428293196833692e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 88457876,
+      "step": 980
+    },
+    {
+      "epoch": 4.510344827586207,
+      "grad_norm": 10.397148286106711,
+      "learning_rate": 5e-06,
+      "loss": 0.1395,
+      "num_input_tokens_seen": 88548152,
+      "step": 981
+    },
+    {
+      "epoch": 4.510344827586207,
+      "loss": 0.17081327736377716,
+      "loss_ce": 2.164947181881871e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 88548152,
+      "step": 981
+    },
+    {
+      "epoch": 4.514942528735633,
+      "grad_norm": 4.078102449108354,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 88638536,
+      "step": 982
+    },
+    {
+      "epoch": 4.514942528735633,
+      "loss": 0.1220250129699707,
+      "loss_ce": 7.67759484006092e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 88638536,
+      "step": 982
+    },
+    {
+      "epoch": 4.519540229885058,
+      "grad_norm": 11.283949648161077,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 88728900,
+      "step": 983
+    },
+    {
+      "epoch": 4.519540229885058,
+      "loss": 0.09133091568946838,
+      "loss_ce": 7.060511507006595e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 88728900,
+      "step": 983
+    },
+    {
+      "epoch": 4.524137931034483,
+      "grad_norm": 9.40893126127556,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 88819192,
+      "step": 984
+    },
+    {
+      "epoch": 4.524137931034483,
+      "loss": 0.08349698781967163,
+      "loss_ce": 8.912971338759235e-07,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 88819192,
+      "step": 984
+    },
+    {
+      "epoch": 4.528735632183908,
+      "grad_norm": 7.932363219291443,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 88909616,
+      "step": 985
+    },
+    {
+      "epoch": 4.528735632183908,
+      "loss": 0.10644952207803726,
+      "loss_ce": 3.472416938166134e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 88909616,
+      "step": 985
+    },
+    {
+      "epoch": 4.533333333333333,
+      "grad_norm": 2.4934351424602568,
+      "learning_rate": 5e-06,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 88999228,
+      "step": 986
+    },
+    {
+      "epoch": 4.533333333333333,
+      "loss": 0.09027735888957977,
+      "loss_ce": 6.3577076616638806e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 88999228,
+      "step": 986
+    },
+    {
+      "epoch": 4.537931034482758,
+      "grad_norm": 4.191824583419648,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 89089628,
+      "step": 987
+    },
+    {
+      "epoch": 4.537931034482758,
+      "loss": 0.05755450576543808,
+      "loss_ce": 5.9386413340689614e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 89089628,
+      "step": 987
+    },
+    {
+      "epoch": 4.5425287356321835,
+      "grad_norm": 4.980888074280528,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 89179956,
+      "step": 988
+    },
+    {
+      "epoch": 4.5425287356321835,
+      "loss": 0.06999941170215607,
+      "loss_ce": 7.347244263655739e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 89179956,
+      "step": 988
+    },
+    {
+      "epoch": 4.5471264367816095,
+      "grad_norm": 6.574375021382182,
+      "learning_rate": 5e-06,
+      "loss": 0.1534,
+      "num_input_tokens_seen": 89270268,
+      "step": 989
+    },
+    {
+      "epoch": 4.5471264367816095,
+      "loss": 0.1247449517250061,
+      "loss_ce": 4.3468216972541995e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 89270268,
+      "step": 989
+    },
+    {
+      "epoch": 4.551724137931035,
+      "grad_norm": 5.770679845977342,
+      "learning_rate": 5e-06,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 89360680,
+      "step": 990
+    },
+    {
+      "epoch": 4.551724137931035,
+      "loss": 0.05505823343992233,
+      "loss_ce": 4.520721631706692e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 89360680,
+      "step": 990
+    },
+    {
+      "epoch": 4.55632183908046,
+      "grad_norm": 3.4680196649545714,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 89451124,
+      "step": 991
+    },
+    {
+      "epoch": 4.55632183908046,
+      "loss": 0.1168176680803299,
+      "loss_ce": 2.6902296667685732e-05,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 89451124,
+      "step": 991
+    },
+    {
+      "epoch": 4.560919540229885,
+      "grad_norm": 6.00004334718162,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 89541568,
+      "step": 992
+    },
+    {
+      "epoch": 4.560919540229885,
+      "loss": 0.11234864592552185,
+      "loss_ce": 1.3441143892123364e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 89541568,
+      "step": 992
+    },
+    {
+      "epoch": 4.56551724137931,
+      "grad_norm": 6.73223448714399,
+      "learning_rate": 5e-06,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 89632052,
+      "step": 993
+    },
+    {
+      "epoch": 4.56551724137931,
+      "loss": 0.07431768625974655,
+      "loss_ce": 7.383117917925119e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 89632052,
+      "step": 993
+    },
+    {
+      "epoch": 4.570114942528735,
+      "grad_norm": 11.420047670032142,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 89722536,
+      "step": 994
+    },
+    {
+      "epoch": 4.570114942528735,
+      "loss": 0.09373007714748383,
+      "loss_ce": 0.008082492277026176,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 89722536,
+      "step": 994
+    },
+    {
+      "epoch": 4.574712643678161,
+      "grad_norm": 14.176346167444665,
+      "learning_rate": 5e-06,
+      "loss": 0.113,
+      "num_input_tokens_seen": 89812932,
+      "step": 995
+    },
+    {
+      "epoch": 4.574712643678161,
+      "loss": 0.12351539731025696,
+      "loss_ce": 1.0755041330412496e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 89812932,
+      "step": 995
+    },
+    {
+      "epoch": 4.5793103448275865,
+      "grad_norm": 10.667470789331928,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 89903292,
+      "step": 996
+    },
+    {
+      "epoch": 4.5793103448275865,
+      "loss": 0.11320923268795013,
+      "loss_ce": 1.1899982382601593e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 89903292,
+      "step": 996
+    },
+    {
+      "epoch": 4.583908045977012,
+      "grad_norm": 7.2249750864413205,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 89993576,
+      "step": 997
+    },
+    {
+      "epoch": 4.583908045977012,
+      "loss": 0.11794456839561462,
+      "loss_ce": 2.465084253344685e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 89993576,
+      "step": 997
+    },
+    {
+      "epoch": 4.588505747126437,
+      "grad_norm": 10.042877155900863,
+      "learning_rate": 5e-06,
+      "loss": 0.1231,
+      "num_input_tokens_seen": 90083884,
+      "step": 998
+    },
+    {
+      "epoch": 4.588505747126437,
+      "loss": 0.1203087866306305,
+      "loss_ce": 8.497871931467671e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 90083884,
+      "step": 998
+    },
+    {
+      "epoch": 4.593103448275862,
+      "grad_norm": 4.373260926369358,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 90174300,
+      "step": 999
+    },
+    {
+      "epoch": 4.593103448275862,
+      "loss": 0.08949369937181473,
+      "loss_ce": 1.6161524399649352e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 90174300,
+      "step": 999
+    },
+    {
+      "epoch": 4.597701149425287,
+      "grad_norm": 25.49877787319248,
+      "learning_rate": 5e-06,
+      "loss": 0.1146,
+      "num_input_tokens_seen": 90264624,
+      "step": 1000
+    },
+    {
+      "epoch": 4.597701149425287,
+      "eval_seeclick_CIoU": 0.4661611318588257,
+      "eval_seeclick_GIoU": 0.4498722702264786,
+      "eval_seeclick_IoU": 0.5080912709236145,
+      "eval_seeclick_MAE_all": 0.060889746993780136,
+      "eval_seeclick_MAE_h": 0.05166458524763584,
+      "eval_seeclick_MAE_w": 0.10992535948753357,
+      "eval_seeclick_MAE_x_boxes": 0.1068052388727665,
+      "eval_seeclick_MAE_y_boxes": 0.054549889639019966,
+      "eval_seeclick_NUM_probability": 0.9999993443489075,
+      "eval_seeclick_inside_bbox": 0.7698863744735718,
+      "eval_seeclick_loss": 0.3709671199321747,
+      "eval_seeclick_loss_ce": 0.07325447350740433,
+      "eval_seeclick_loss_iou": 0.47247314453125,
+      "eval_seeclick_loss_num": 0.06253814697265625,
+      "eval_seeclick_loss_xval": 0.3125,
+      "eval_seeclick_runtime": 75.293,
+      "eval_seeclick_samples_per_second": 0.571,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 90264624,
+      "step": 1000
+    },
+    {
+      "epoch": 4.597701149425287,
+      "eval_icons_CIoU": 0.5760022103786469,
+      "eval_icons_GIoU": 0.576844722032547,
+      "eval_icons_IoU": 0.613362729549408,
+      "eval_icons_MAE_all": 0.04183553345501423,
+      "eval_icons_MAE_h": 0.07954326272010803,
+      "eval_icons_MAE_w": 0.06392070464789867,
+      "eval_icons_MAE_x_boxes": 0.059913450852036476,
+      "eval_icons_MAE_y_boxes": 0.07867859303951263,
+      "eval_icons_NUM_probability": 0.9999994337558746,
+      "eval_icons_inside_bbox": 0.8107638955116272,
+      "eval_icons_loss": 0.20698395371437073,
+      "eval_icons_loss_ce": 1.277818611811199e-06,
+      "eval_icons_loss_iou": 0.42694091796875,
+      "eval_icons_loss_num": 0.04441070556640625,
+      "eval_icons_loss_xval": 0.2221221923828125,
+      "eval_icons_runtime": 88.2999,
+      "eval_icons_samples_per_second": 0.566,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 90264624,
+      "step": 1000
+    },
+    {
+      "epoch": 4.597701149425287,
+      "eval_screenspot_CIoU": 0.4215618173281352,
+      "eval_screenspot_GIoU": 0.4092308084170024,
+      "eval_screenspot_IoU": 0.48668718338012695,
+      "eval_screenspot_MAE_all": 0.08650621399283409,
+      "eval_screenspot_MAE_h": 0.08026436219612758,
+      "eval_screenspot_MAE_w": 0.17552465697129568,
+      "eval_screenspot_MAE_x_boxes": 0.16668692231178284,
+      "eval_screenspot_MAE_y_boxes": 0.07600387185811996,
+      "eval_screenspot_NUM_probability": 0.9999993443489075,
+      "eval_screenspot_inside_bbox": 0.7637499968210856,
+      "eval_screenspot_loss": 0.43140459060668945,
+      "eval_screenspot_loss_ce": 0.00010789081594945553,
+      "eval_screenspot_loss_iou": 0.4072265625,
+      "eval_screenspot_loss_num": 0.08810933430989583,
+      "eval_screenspot_loss_xval": 0.4403483072916667,
+      "eval_screenspot_runtime": 149.3508,
+      "eval_screenspot_samples_per_second": 0.596,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 90264624,
+      "step": 1000
+    },
+    {
+      "epoch": 4.597701149425287,
+      "eval_compot_CIoU": 0.47925493121147156,
+      "eval_compot_GIoU": 0.4560560882091522,
+      "eval_compot_IoU": 0.5442279279232025,
+      "eval_compot_MAE_all": 0.0558595210313797,
+      "eval_compot_MAE_h": 0.07545717805624008,
+      "eval_compot_MAE_w": 0.11159718781709671,
+      "eval_compot_MAE_x_boxes": 0.10114440321922302,
+      "eval_compot_MAE_y_boxes": 0.07544495910406113,
+      "eval_compot_NUM_probability": 0.9999988079071045,
+      "eval_compot_inside_bbox": 0.8072916567325592,
+      "eval_compot_loss": 0.3027111887931824,
+      "eval_compot_loss_ce": 0.01121709169819951,
+      "eval_compot_loss_iou": 0.4888916015625,
+      "eval_compot_loss_num": 0.05005645751953125,
+      "eval_compot_loss_xval": 0.250244140625,
+      "eval_compot_runtime": 86.8458,
+      "eval_compot_samples_per_second": 0.576,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 90264624,
+      "step": 1000
+    },
+    {
+      "epoch": 4.597701149425287,
+      "loss": 0.18884405493736267,
+      "loss_ce": 0.006959293968975544,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 90264624,
+      "step": 1000
+    },
+    {
+      "epoch": 4.602298850574712,
+      "grad_norm": 9.40244134173418,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 90355016,
+      "step": 1001
+    },
+    {
+      "epoch": 4.602298850574712,
+      "loss": 0.08515298366546631,
+      "loss_ce": 5.4719315812690184e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 90355016,
+      "step": 1001
+    },
+    {
+      "epoch": 4.606896551724138,
+      "grad_norm": 11.782183850460743,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 90445512,
+      "step": 1002
+    },
+    {
+      "epoch": 4.606896551724138,
+      "loss": 0.08575557172298431,
+      "loss_ce": 1.6436308214906603e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 90445512,
+      "step": 1002
+    },
+    {
+      "epoch": 4.611494252873563,
+      "grad_norm": 17.076982393187414,
+      "learning_rate": 5e-06,
+      "loss": 0.1481,
+      "num_input_tokens_seen": 90536000,
+      "step": 1003
+    },
+    {
+      "epoch": 4.611494252873563,
+      "loss": 0.10569582879543304,
+      "loss_ce": 4.39713076048065e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 90536000,
+      "step": 1003
+    },
+    {
+      "epoch": 4.6160919540229886,
+      "grad_norm": 16.92181833640432,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 90626368,
+      "step": 1004
+    },
+    {
+      "epoch": 4.6160919540229886,
+      "loss": 0.10335642099380493,
+      "loss_ce": 8.644953595648985e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 90626368,
+      "step": 1004
+    },
+    {
+      "epoch": 4.620689655172414,
+      "grad_norm": 5.272669970898714,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 90716756,
+      "step": 1005
+    },
+    {
+      "epoch": 4.620689655172414,
+      "loss": 0.12034394592046738,
+      "loss_ce": 1.3136214874975849e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 90716756,
+      "step": 1005
+    },
+    {
+      "epoch": 4.625287356321839,
+      "grad_norm": 5.5450374658240476,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 90806388,
+      "step": 1006
+    },
+    {
+      "epoch": 4.625287356321839,
+      "loss": 0.12717095017433167,
+      "loss_ce": 4.190186700725462e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 90806388,
+      "step": 1006
+    },
+    {
+      "epoch": 4.629885057471264,
+      "grad_norm": 10.252060177606076,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 90896652,
+      "step": 1007
+    },
+    {
+      "epoch": 4.629885057471264,
+      "loss": 0.10986798256635666,
+      "loss_ce": 4.700279987446265e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 90896652,
+      "step": 1007
+    },
+    {
+      "epoch": 4.63448275862069,
+      "grad_norm": 18.179346667786145,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 90986972,
+      "step": 1008
+    },
+    {
+      "epoch": 4.63448275862069,
+      "loss": 0.09801249206066132,
+      "loss_ce": 5.289793534757337e-06,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 90986972,
+      "step": 1008
+    },
+    {
+      "epoch": 4.639080459770115,
+      "grad_norm": 7.486242267110328,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 91077320,
+      "step": 1009
+    },
+    {
+      "epoch": 4.639080459770115,
+      "loss": 0.0817301869392395,
+      "loss_ce": 4.111347152502276e-06,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 91077320,
+      "step": 1009
+    },
+    {
+      "epoch": 4.64367816091954,
+      "grad_norm": 7.378317449558524,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 91167668,
+      "step": 1010
+    },
+    {
+      "epoch": 4.64367816091954,
+      "loss": 0.11923195421695709,
+      "loss_ce": 1.5039631762192585e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 91167668,
+      "step": 1010
+    },
+    {
+      "epoch": 4.6482758620689655,
+      "grad_norm": 6.873925032428975,
+      "learning_rate": 5e-06,
+      "loss": 0.1342,
+      "num_input_tokens_seen": 91257984,
+      "step": 1011
+    },
+    {
+      "epoch": 4.6482758620689655,
+      "loss": 0.13492737710475922,
+      "loss_ce": 9.166550626105163e-06,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 91257984,
+      "step": 1011
+    },
+    {
+      "epoch": 4.652873563218391,
+      "grad_norm": 13.994177721226423,
+      "learning_rate": 5e-06,
+      "loss": 0.087,
+      "num_input_tokens_seen": 91348380,
+      "step": 1012
+    },
+    {
+      "epoch": 4.652873563218391,
+      "loss": 0.07507447898387909,
+      "loss_ce": 1.2354627187960432e-06,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 91348380,
+      "step": 1012
+    },
+    {
+      "epoch": 4.657471264367816,
+      "grad_norm": 3.014366999503986,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 91438616,
+      "step": 1013
+    },
+    {
+      "epoch": 4.657471264367816,
+      "loss": 0.16335362195968628,
+      "loss_ce": 2.354383104830049e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 91438616,
+      "step": 1013
+    },
+    {
+      "epoch": 4.662068965517241,
+      "grad_norm": 17.72787254515268,
+      "learning_rate": 5e-06,
+      "loss": 0.1255,
+      "num_input_tokens_seen": 91529032,
+      "step": 1014
+    },
+    {
+      "epoch": 4.662068965517241,
+      "loss": 0.13172364234924316,
+      "loss_ce": 9.771740224095993e-06,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 91529032,
+      "step": 1014
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 3.6520684848160605,
+      "learning_rate": 5e-06,
+      "loss": 0.126,
+      "num_input_tokens_seen": 91618688,
+      "step": 1015
+    },
+    {
+      "epoch": 4.666666666666667,
+      "loss": 0.12861014902591705,
+      "loss_ce": 9.073386536329053e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 91618688,
+      "step": 1015
+    },
+    {
+      "epoch": 4.671264367816092,
+      "grad_norm": 6.527601418047689,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 91709076,
+      "step": 1016
+    },
+    {
+      "epoch": 4.671264367816092,
+      "loss": 0.09489475935697556,
+      "loss_ce": 3.086655124207027e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 91709076,
+      "step": 1016
+    },
+    {
+      "epoch": 4.675862068965517,
+      "grad_norm": 14.192420949264795,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 91799384,
+      "step": 1017
+    },
+    {
+      "epoch": 4.675862068965517,
+      "loss": 0.08747270703315735,
+      "loss_ce": 9.329753083875403e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 91799384,
+      "step": 1017
+    },
+    {
+      "epoch": 4.680459770114942,
+      "grad_norm": 5.653695210589391,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 91889776,
+      "step": 1018
+    },
+    {
+      "epoch": 4.680459770114942,
+      "loss": 0.08493378758430481,
+      "loss_ce": 3.372960463821073e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 91889776,
+      "step": 1018
+    },
+    {
+      "epoch": 4.685057471264368,
+      "grad_norm": 4.63777017993729,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 91980096,
+      "step": 1019
+    },
+    {
+      "epoch": 4.685057471264368,
+      "loss": 0.07615326344966888,
+      "loss_ce": 1.1908843589480966e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 91980096,
+      "step": 1019
+    },
+    {
+      "epoch": 4.689655172413794,
+      "grad_norm": 4.472710831706082,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 92070584,
+      "step": 1020
+    },
+    {
+      "epoch": 4.689655172413794,
+      "loss": 0.11327888071537018,
+      "loss_ce": 2.8146971089881845e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 92070584,
+      "step": 1020
+    },
+    {
+      "epoch": 4.694252873563219,
+      "grad_norm": 3.4544586422872907,
+      "learning_rate": 5e-06,
+      "loss": 0.096,
+      "num_input_tokens_seen": 92161012,
+      "step": 1021
+    },
+    {
+      "epoch": 4.694252873563219,
+      "loss": 0.06871644407510757,
+      "loss_ce": 6.117389602877665e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 92161012,
+      "step": 1021
+    },
+    {
+      "epoch": 4.698850574712644,
+      "grad_norm": 7.42414792402292,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 92251412,
+      "step": 1022
+    },
+    {
+      "epoch": 4.698850574712644,
+      "loss": 0.10006897151470184,
+      "loss_ce": 9.33870833250694e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 92251412,
+      "step": 1022
+    },
+    {
+      "epoch": 4.703448275862069,
+      "grad_norm": 6.69514367017083,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 92341956,
+      "step": 1023
+    },
+    {
+      "epoch": 4.703448275862069,
+      "loss": 0.14697983860969543,
+      "loss_ce": 6.821977149229497e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 92341956,
+      "step": 1023
+    },
+    {
+      "epoch": 4.708045977011494,
+      "grad_norm": 14.417454722090794,
+      "learning_rate": 5e-06,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 92432144,
+      "step": 1024
+    },
+    {
+      "epoch": 4.708045977011494,
+      "loss": 0.13804002106189728,
+      "loss_ce": 9.01172825251706e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 92432144,
+      "step": 1024
+    },
+    {
+      "epoch": 4.712643678160919,
+      "grad_norm": 7.595947463221086,
+      "learning_rate": 5e-06,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 92522620,
+      "step": 1025
+    },
+    {
+      "epoch": 4.712643678160919,
+      "loss": 0.10795444995164871,
+      "loss_ce": 2.1407777239801362e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 92522620,
+      "step": 1025
+    },
+    {
+      "epoch": 4.7172413793103445,
+      "grad_norm": 3.0300131913282713,
+      "learning_rate": 5e-06,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 92612984,
+      "step": 1026
+    },
+    {
+      "epoch": 4.7172413793103445,
+      "loss": 0.06870204210281372,
+      "loss_ce": 6.9793131842743605e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 92612984,
+      "step": 1026
+    },
+    {
+      "epoch": 4.72183908045977,
+      "grad_norm": 7.887427930687986,
+      "learning_rate": 5e-06,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 92702504,
+      "step": 1027
+    },
+    {
+      "epoch": 4.72183908045977,
+      "loss": 0.07237892597913742,
+      "loss_ce": 2.1748670405941084e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 92702504,
+      "step": 1027
+    },
+    {
+      "epoch": 4.726436781609196,
+      "grad_norm": 9.408243473561752,
+      "learning_rate": 5e-06,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 92792888,
+      "step": 1028
+    },
+    {
+      "epoch": 4.726436781609196,
+      "loss": 0.07289677858352661,
+      "loss_ce": 5.5426976359740365e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 92792888,
+      "step": 1028
+    },
+    {
+      "epoch": 4.731034482758621,
+      "grad_norm": 9.742419905000565,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 92883336,
+      "step": 1029
+    },
+    {
+      "epoch": 4.731034482758621,
+      "loss": 0.10518839955329895,
+      "loss_ce": 2.4827564629958943e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 92883336,
+      "step": 1029
+    },
+    {
+      "epoch": 4.735632183908046,
+      "grad_norm": 2.7055194664846183,
+      "learning_rate": 5e-06,
+      "loss": 0.1231,
+      "num_input_tokens_seen": 92972936,
+      "step": 1030
+    },
+    {
+      "epoch": 4.735632183908046,
+      "loss": 0.10875724256038666,
+      "loss_ce": 8.414130570599809e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 92972936,
+      "step": 1030
+    },
+    {
+      "epoch": 4.740229885057471,
+      "grad_norm": 11.304101815777114,
+      "learning_rate": 5e-06,
+      "loss": 0.1415,
+      "num_input_tokens_seen": 93063392,
+      "step": 1031
+    },
+    {
+      "epoch": 4.740229885057471,
+      "loss": 0.09561444818973541,
+      "loss_ce": 2.8712015591736417e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 93063392,
+      "step": 1031
+    },
+    {
+      "epoch": 4.744827586206896,
+      "grad_norm": 15.258461645144374,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 93152916,
+      "step": 1032
+    },
+    {
+      "epoch": 4.744827586206896,
+      "loss": 0.11910735070705414,
+      "loss_ce": 4.878862000623485e-06,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 93152916,
+      "step": 1032
+    },
+    {
+      "epoch": 4.749425287356322,
+      "grad_norm": 2.356961429838356,
+      "learning_rate": 5e-06,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 93243212,
+      "step": 1033
+    },
+    {
+      "epoch": 4.749425287356322,
+      "loss": 0.11762569844722748,
+      "loss_ce": 0.0005450131138786674,
+      "loss_iou": 0.375,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 93243212,
+      "step": 1033
+    },
+    {
+      "epoch": 4.7540229885057474,
+      "grad_norm": 15.1692410364505,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 93333636,
+      "step": 1034
+    },
+    {
+      "epoch": 4.7540229885057474,
+      "loss": 0.147009015083313,
+      "loss_ce": 5.839167897647712e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 93333636,
+      "step": 1034
+    },
+    {
+      "epoch": 4.758620689655173,
+      "grad_norm": 7.946502842662023,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 93423992,
+      "step": 1035
+    },
+    {
+      "epoch": 4.758620689655173,
+      "loss": 0.1617765724658966,
+      "loss_ce": 3.341192496009171e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 93423992,
+      "step": 1035
+    },
+    {
+      "epoch": 4.763218390804598,
+      "grad_norm": 15.109850534366604,
+      "learning_rate": 5e-06,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 93514320,
+      "step": 1036
+    },
+    {
+      "epoch": 4.763218390804598,
+      "loss": 0.0966891348361969,
+      "loss_ce": 9.446171134186443e-06,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 93514320,
+      "step": 1036
+    },
+    {
+      "epoch": 4.767816091954023,
+      "grad_norm": 4.842981275277206,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 93603144,
+      "step": 1037
+    },
+    {
+      "epoch": 4.767816091954023,
+      "loss": 0.11073972284793854,
+      "loss_ce": 2.194262560806237e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 93603144,
+      "step": 1037
+    },
+    {
+      "epoch": 4.772413793103448,
+      "grad_norm": 21.01549868516104,
+      "learning_rate": 5e-06,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 93693660,
+      "step": 1038
+    },
+    {
+      "epoch": 4.772413793103448,
+      "loss": 0.08537042886018753,
+      "loss_ce": 1.2760566278302576e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 93693660,
+      "step": 1038
+    },
+    {
+      "epoch": 4.777011494252873,
+      "grad_norm": 10.712563478328072,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 93783296,
+      "step": 1039
+    },
+    {
+      "epoch": 4.777011494252873,
+      "loss": 0.11916904151439667,
+      "loss_ce": 2.8420639864634722e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 93783296,
+      "step": 1039
+    },
+    {
+      "epoch": 4.781609195402299,
+      "grad_norm": 7.345903871032101,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 93873768,
+      "step": 1040
+    },
+    {
+      "epoch": 4.781609195402299,
+      "loss": 0.05897822231054306,
+      "loss_ce": 3.0012429306225386e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 93873768,
+      "step": 1040
+    },
+    {
+      "epoch": 4.786206896551724,
+      "grad_norm": 4.133426960663791,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 93964260,
+      "step": 1041
+    },
+    {
+      "epoch": 4.786206896551724,
+      "loss": 0.0858168751001358,
+      "loss_ce": 1.6706742826499976e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 93964260,
+      "step": 1041
+    },
+    {
+      "epoch": 4.7908045977011495,
+      "grad_norm": 7.686995682812649,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 94054556,
+      "step": 1042
+    },
+    {
+      "epoch": 4.7908045977011495,
+      "loss": 0.12363208830356598,
+      "loss_ce": 5.381094069889514e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 94054556,
+      "step": 1042
+    },
+    {
+      "epoch": 4.795402298850575,
+      "grad_norm": 1.6904479508606451,
+      "learning_rate": 5e-06,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 94145004,
+      "step": 1043
+    },
+    {
+      "epoch": 4.795402298850575,
+      "loss": 0.06091611459851265,
+      "loss_ce": 3.0297642297227867e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 94145004,
+      "step": 1043
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 3.669144596101602,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 94235372,
+      "step": 1044
+    },
+    {
+      "epoch": 4.8,
+      "loss": 0.13778972625732422,
+      "loss_ce": 2.869950094464002e-06,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 94235372,
+      "step": 1044
+    },
+    {
+      "epoch": 4.804597701149425,
+      "grad_norm": 6.97279172429298,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 94325696,
+      "step": 1045
+    },
+    {
+      "epoch": 4.804597701149425,
+      "loss": 0.05801108479499817,
+      "loss_ce": 0.0023165077436715364,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 94325696,
+      "step": 1045
+    },
+    {
+      "epoch": 4.809195402298851,
+      "grad_norm": 11.58305315850037,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 94416004,
+      "step": 1046
+    },
+    {
+      "epoch": 4.809195402298851,
+      "loss": 0.1216389536857605,
+      "loss_ce": 1.1142959920107387e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 94416004,
+      "step": 1046
+    },
+    {
+      "epoch": 4.813793103448276,
+      "grad_norm": 16.0940836142544,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 94506316,
+      "step": 1047
+    },
+    {
+      "epoch": 4.813793103448276,
+      "loss": 0.10546806454658508,
+      "loss_ce": 2.9831333449692465e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 94506316,
+      "step": 1047
+    },
+    {
+      "epoch": 4.818390804597701,
+      "grad_norm": 10.955820630731282,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 94596544,
+      "step": 1048
+    },
+    {
+      "epoch": 4.818390804597701,
+      "loss": 0.16584840416908264,
+      "loss_ce": 1.5889574569882825e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 94596544,
+      "step": 1048
+    },
+    {
+      "epoch": 4.8229885057471265,
+      "grad_norm": 6.034273447193758,
+      "learning_rate": 5e-06,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 94686864,
+      "step": 1049
+    },
+    {
+      "epoch": 4.8229885057471265,
+      "loss": 0.06973493099212646,
+      "loss_ce": 2.2640974748355802e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 94686864,
+      "step": 1049
+    },
+    {
+      "epoch": 4.827586206896552,
+      "grad_norm": 4.252496465368457,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 94777296,
+      "step": 1050
+    },
+    {
+      "epoch": 4.827586206896552,
+      "loss": 0.10622747987508774,
+      "loss_ce": 7.20810130587779e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 94777296,
+      "step": 1050
+    },
+    {
+      "epoch": 4.832183908045977,
+      "grad_norm": 3.0187179996735334,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 94867600,
+      "step": 1051
+    },
+    {
+      "epoch": 4.832183908045977,
+      "loss": 0.1049966886639595,
+      "loss_ce": 9.584927056494053e-07,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 94867600,
+      "step": 1051
+    },
+    {
+      "epoch": 4.836781609195402,
+      "grad_norm": 16.087126169900518,
+      "learning_rate": 5e-06,
+      "loss": 0.082,
+      "num_input_tokens_seen": 94958008,
+      "step": 1052
+    },
+    {
+      "epoch": 4.836781609195402,
+      "loss": 0.1101231724023819,
+      "loss_ce": 1.5750989405205473e-05,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 94958008,
+      "step": 1052
+    },
+    {
+      "epoch": 4.841379310344828,
+      "grad_norm": 12.550203705796418,
+      "learning_rate": 5e-06,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 95048356,
+      "step": 1053
+    },
+    {
+      "epoch": 4.841379310344828,
+      "loss": 0.08209192752838135,
+      "loss_ce": 1.4905896023265086e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 95048356,
+      "step": 1053
+    },
+    {
+      "epoch": 4.845977011494253,
+      "grad_norm": 4.5783225747939555,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 95138768,
+      "step": 1054
+    },
+    {
+      "epoch": 4.845977011494253,
+      "loss": 0.1782418191432953,
+      "loss_ce": 3.904753612005152e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 95138768,
+      "step": 1054
+    },
+    {
+      "epoch": 4.850574712643678,
+      "grad_norm": 17.243051689458277,
+      "learning_rate": 5e-06,
+      "loss": 0.1833,
+      "num_input_tokens_seen": 95229088,
+      "step": 1055
+    },
+    {
+      "epoch": 4.850574712643678,
+      "loss": 0.11458714306354523,
+      "loss_ce": 2.4146735086105764e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 95229088,
+      "step": 1055
+    },
+    {
+      "epoch": 4.855172413793103,
+      "grad_norm": 6.0951803316443876,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 95318604,
+      "step": 1056
+    },
+    {
+      "epoch": 4.855172413793103,
+      "loss": 0.09546582400798798,
+      "loss_ce": 6.840703463240061e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 95318604,
+      "step": 1056
+    },
+    {
+      "epoch": 4.8597701149425285,
+      "grad_norm": 2.513374225059268,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 95409032,
+      "step": 1057
+    },
+    {
+      "epoch": 4.8597701149425285,
+      "loss": 0.105290487408638,
+      "loss_ce": 0.00047786219511181116,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 95409032,
+      "step": 1057
+    },
+    {
+      "epoch": 4.864367816091954,
+      "grad_norm": 3.376425029688418,
+      "learning_rate": 5e-06,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 95499352,
+      "step": 1058
+    },
+    {
+      "epoch": 4.864367816091954,
+      "loss": 0.1294291913509369,
+      "loss_ce": 0.0005839776713401079,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 95499352,
+      "step": 1058
+    },
+    {
+      "epoch": 4.86896551724138,
+      "grad_norm": 4.718684401574836,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 95589880,
+      "step": 1059
+    },
+    {
+      "epoch": 4.86896551724138,
+      "loss": 0.15736906230449677,
+      "loss_ce": 2.0424929971341044e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 95589880,
+      "step": 1059
+    },
+    {
+      "epoch": 4.873563218390805,
+      "grad_norm": 8.564949517103082,
+      "learning_rate": 5e-06,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 95680168,
+      "step": 1060
+    },
+    {
+      "epoch": 4.873563218390805,
+      "loss": 0.07507706433534622,
+      "loss_ce": 3.8181360650924034e-06,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 95680168,
+      "step": 1060
+    },
+    {
+      "epoch": 4.87816091954023,
+      "grad_norm": 10.678428697285344,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 95770520,
+      "step": 1061
+    },
+    {
+      "epoch": 4.87816091954023,
+      "loss": 0.13752196729183197,
+      "loss_ce": 4.027618342661299e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 95770520,
+      "step": 1061
+    },
+    {
+      "epoch": 4.882758620689655,
+      "grad_norm": 5.476416486972618,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 95860856,
+      "step": 1062
+    },
+    {
+      "epoch": 4.882758620689655,
+      "loss": 0.10755479335784912,
+      "loss_ce": 1.0849686077563092e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 95860856,
+      "step": 1062
+    },
+    {
+      "epoch": 4.88735632183908,
+      "grad_norm": 7.341716557148394,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 95951252,
+      "step": 1063
+    },
+    {
+      "epoch": 4.88735632183908,
+      "loss": 0.08713214844465256,
+      "loss_ce": 4.465159690880682e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 95951252,
+      "step": 1063
+    },
+    {
+      "epoch": 4.8919540229885055,
+      "grad_norm": 17.624799280413747,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 96041592,
+      "step": 1064
+    },
+    {
+      "epoch": 4.8919540229885055,
+      "loss": 0.1143837422132492,
+      "loss_ce": 3.4374650567770004e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 96041592,
+      "step": 1064
+    },
+    {
+      "epoch": 4.896551724137931,
+      "grad_norm": 9.416499244614185,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 96132008,
+      "step": 1065
+    },
+    {
+      "epoch": 4.896551724137931,
+      "loss": 0.11976835876703262,
+      "loss_ce": 4.7893059672787786e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 96132008,
+      "step": 1065
+    },
+    {
+      "epoch": 4.901149425287357,
+      "grad_norm": 6.806788764833692,
+      "learning_rate": 5e-06,
+      "loss": 0.1071,
+      "num_input_tokens_seen": 96222480,
+      "step": 1066
+    },
+    {
+      "epoch": 4.901149425287357,
+      "loss": 0.09189382195472717,
+      "loss_ce": 5.3979856602381915e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 96222480,
+      "step": 1066
+    },
+    {
+      "epoch": 4.905747126436782,
+      "grad_norm": 5.549592637186866,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 96312836,
+      "step": 1067
+    },
+    {
+      "epoch": 4.905747126436782,
+      "loss": 0.08161412179470062,
+      "loss_ce": 4.0634346078149974e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 96312836,
+      "step": 1067
+    },
+    {
+      "epoch": 4.910344827586207,
+      "grad_norm": 6.49206654790828,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 96403204,
+      "step": 1068
+    },
+    {
+      "epoch": 4.910344827586207,
+      "loss": 0.10242622345685959,
+      "loss_ce": 9.231689546140842e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 96403204,
+      "step": 1068
+    },
+    {
+      "epoch": 4.914942528735632,
+      "grad_norm": 3.007842215921885,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 96493592,
+      "step": 1069
+    },
+    {
+      "epoch": 4.914942528735632,
+      "loss": 0.09680457413196564,
+      "loss_ce": 6.385076267179102e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 96493592,
+      "step": 1069
+    },
+    {
+      "epoch": 4.919540229885057,
+      "grad_norm": 10.319391804852176,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 96583944,
+      "step": 1070
+    },
+    {
+      "epoch": 4.919540229885057,
+      "loss": 0.08242271840572357,
+      "loss_ce": 5.577308911597356e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 96583944,
+      "step": 1070
+    },
+    {
+      "epoch": 4.924137931034482,
+      "grad_norm": 2.879146983290818,
+      "learning_rate": 5e-06,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 96674328,
+      "step": 1071
+    },
+    {
+      "epoch": 4.924137931034482,
+      "loss": 0.10010585933923721,
+      "loss_ce": 8.201654054573737e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 96674328,
+      "step": 1071
+    },
+    {
+      "epoch": 4.928735632183908,
+      "grad_norm": 15.403924468504222,
+      "learning_rate": 5e-06,
+      "loss": 0.0968,
+      "num_input_tokens_seen": 96764696,
+      "step": 1072
+    },
+    {
+      "epoch": 4.928735632183908,
+      "loss": 0.12627822160720825,
+      "loss_ce": 2.7013094950234517e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 96764696,
+      "step": 1072
+    },
+    {
+      "epoch": 4.933333333333334,
+      "grad_norm": 24.679178097861215,
+      "learning_rate": 5e-06,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 96855144,
+      "step": 1073
+    },
+    {
+      "epoch": 4.933333333333334,
+      "loss": 0.11346793919801712,
+      "loss_ce": 3.5857883631251752e-06,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 96855144,
+      "step": 1073
+    },
+    {
+      "epoch": 4.937931034482759,
+      "grad_norm": 17.443099366886536,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 96945600,
+      "step": 1074
+    },
+    {
+      "epoch": 4.937931034482759,
+      "loss": 0.09844504296779633,
+      "loss_ce": 5.636948117171414e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 96945600,
+      "step": 1074
+    },
+    {
+      "epoch": 4.942528735632184,
+      "grad_norm": 11.649607579048189,
+      "learning_rate": 5e-06,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 97035976,
+      "step": 1075
+    },
+    {
+      "epoch": 4.942528735632184,
+      "loss": 0.09174925833940506,
+      "loss_ce": 1.3420964933175128e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 97035976,
+      "step": 1075
+    },
+    {
+      "epoch": 4.947126436781609,
+      "grad_norm": 8.321494891514785,
+      "learning_rate": 5e-06,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 97126184,
+      "step": 1076
+    },
+    {
+      "epoch": 4.947126436781609,
+      "loss": 0.1108294129371643,
+      "loss_ce": 4.831304522667779e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 97126184,
+      "step": 1076
+    },
+    {
+      "epoch": 4.951724137931034,
+      "grad_norm": 2.7753841701876714,
+      "learning_rate": 5e-06,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 97215208,
+      "step": 1077
+    },
+    {
+      "epoch": 4.951724137931034,
+      "loss": 0.10865768790245056,
+      "loss_ce": 7.61427145334892e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 97215208,
+      "step": 1077
+    },
+    {
+      "epoch": 4.956321839080459,
+      "grad_norm": 3.502619253449482,
+      "learning_rate": 5e-06,
+      "loss": 0.1241,
+      "num_input_tokens_seen": 97305660,
+      "step": 1078
+    },
+    {
+      "epoch": 4.956321839080459,
+      "loss": 0.07795646786689758,
+      "loss_ce": 1.4574313354387414e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 97305660,
+      "step": 1078
+    },
+    {
+      "epoch": 4.960919540229885,
+      "grad_norm": 2.214432026018128,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 97395964,
+      "step": 1079
+    },
+    {
+      "epoch": 4.960919540229885,
+      "loss": 0.09012555330991745,
+      "loss_ce": 3.7664180126739666e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 97395964,
+      "step": 1079
+    },
+    {
+      "epoch": 4.9655172413793105,
+      "grad_norm": 6.4074181213192105,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 97486300,
+      "step": 1080
+    },
+    {
+      "epoch": 4.9655172413793105,
+      "loss": 0.07362109422683716,
+      "loss_ce": 1.2694945326074958e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 97486300,
+      "step": 1080
+    },
+    {
+      "epoch": 4.970114942528736,
+      "grad_norm": 10.501486543841219,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 97576584,
+      "step": 1081
+    },
+    {
+      "epoch": 4.970114942528736,
+      "loss": 0.07947252690792084,
+      "loss_ce": 4.7581602302670944e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 97576584,
+      "step": 1081
+    },
+    {
+      "epoch": 4.974712643678161,
+      "grad_norm": 5.4733017757114615,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 97666176,
+      "step": 1082
+    },
+    {
+      "epoch": 4.974712643678161,
+      "loss": 0.2110956907272339,
+      "loss_ce": 5.60448916075984e-06,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 97666176,
+      "step": 1082
+    },
+    {
+      "epoch": 4.979310344827586,
+      "grad_norm": 4.717726419007518,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 97756556,
+      "step": 1083
+    },
+    {
+      "epoch": 4.979310344827586,
+      "loss": 0.12327490001916885,
+      "loss_ce": 1.4395311154657975e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 97756556,
+      "step": 1083
+    },
+    {
+      "epoch": 4.983908045977012,
+      "grad_norm": 6.386820886831165,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "num_input_tokens_seen": 97845408,
+      "step": 1084
+    },
+    {
+      "epoch": 4.983908045977012,
+      "loss": 0.1393074244260788,
+      "loss_ce": 0.00013201337424106896,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 97845408,
+      "step": 1084
+    },
+    {
+      "epoch": 4.988505747126437,
+      "grad_norm": 17.923854404624937,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 97935700,
+      "step": 1085
+    },
+    {
+      "epoch": 4.988505747126437,
+      "loss": 0.1198144182562828,
+      "loss_ce": 2.406018211331684e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 97935700,
+      "step": 1085
+    },
+    {
+      "epoch": 4.993103448275862,
+      "grad_norm": 8.89371007698411,
+      "learning_rate": 5e-06,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 98026076,
+      "step": 1086
+    },
+    {
+      "epoch": 4.993103448275862,
+      "loss": 0.08179621398448944,
+      "loss_ce": 3.9620990719413385e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 98026076,
+      "step": 1086
+    },
+    {
+      "epoch": 4.997701149425287,
+      "grad_norm": 3.2738674103409777,
+      "learning_rate": 5e-06,
+      "loss": 0.1316,
+      "num_input_tokens_seen": 98116464,
+      "step": 1087
+    },
+    {
+      "epoch": 4.997701149425287,
+      "loss": 0.1348174810409546,
+      "loss_ce": 2.13292059925152e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 98116464,
+      "step": 1087
+    },
+    {
+      "epoch": 4.997701149425287,
+      "loss": 0.11705206334590912,
+      "loss_ce": 0.00010870777623495087,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 98160852,
+      "step": 1087
+    },
+    {
+      "epoch": 5.002298850574713,
+      "grad_norm": 3.097439221083733,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 98206020,
+      "step": 1088
+    },
+    {
+      "epoch": 5.002298850574713,
+      "loss": 0.09875764697790146,
+      "loss_ce": 6.379698606906459e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 98206020,
+      "step": 1088
+    },
+    {
+      "epoch": 5.006896551724138,
+      "grad_norm": 12.127549215108248,
+      "learning_rate": 5e-06,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 98296448,
+      "step": 1089
+    },
+    {
+      "epoch": 5.006896551724138,
+      "loss": 0.1051742285490036,
+      "loss_ce": 5.6425316870445386e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 98296448,
+      "step": 1089
+    },
+    {
+      "epoch": 5.011494252873563,
+      "grad_norm": 3.0110884831761777,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 98386704,
+      "step": 1090
+    },
+    {
+      "epoch": 5.011494252873563,
+      "loss": 0.117917500436306,
+      "loss_ce": 1.2841372154070996e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 98386704,
+      "step": 1090
+    },
+    {
+      "epoch": 5.016091954022989,
+      "grad_norm": 8.800434556379146,
+      "learning_rate": 5e-06,
+      "loss": 0.1628,
+      "num_input_tokens_seen": 98475456,
+      "step": 1091
+    },
+    {
+      "epoch": 5.016091954022989,
+      "loss": 0.15691357851028442,
+      "loss_ce": 0.0011213450925424695,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 98475456,
+      "step": 1091
+    },
+    {
+      "epoch": 5.020689655172414,
+      "grad_norm": 17.875153503898837,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 98565748,
+      "step": 1092
+    },
+    {
+      "epoch": 5.020689655172414,
+      "loss": 0.08784059435129166,
+      "loss_ce": 2.6260444428771734e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 98565748,
+      "step": 1092
+    },
+    {
+      "epoch": 5.025287356321839,
+      "grad_norm": 1.9310524258242443,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 98656216,
+      "step": 1093
+    },
+    {
+      "epoch": 5.025287356321839,
+      "loss": 0.10792776942253113,
+      "loss_ce": 1.7617947378312238e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 98656216,
+      "step": 1093
+    },
+    {
+      "epoch": 5.029885057471264,
+      "grad_norm": 5.2811306139564245,
+      "learning_rate": 5e-06,
+      "loss": 0.082,
+      "num_input_tokens_seen": 98746572,
+      "step": 1094
+    },
+    {
+      "epoch": 5.029885057471264,
+      "loss": 0.05932975560426712,
+      "loss_ce": 3.58494685315236e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 98746572,
+      "step": 1094
+    },
+    {
+      "epoch": 5.0344827586206895,
+      "grad_norm": 14.24076067000812,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 98836948,
+      "step": 1095
+    },
+    {
+      "epoch": 5.0344827586206895,
+      "loss": 0.08336181938648224,
+      "loss_ce": 4.8835227062227204e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 98836948,
+      "step": 1095
+    },
+    {
+      "epoch": 5.039080459770115,
+      "grad_norm": 7.5844239875355415,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 98927324,
+      "step": 1096
+    },
+    {
+      "epoch": 5.039080459770115,
+      "loss": 0.16165432333946228,
+      "loss_ce": 3.3243813959416e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 98927324,
+      "step": 1096
+    },
+    {
+      "epoch": 5.04367816091954,
+      "grad_norm": 8.539898456692784,
+      "learning_rate": 5e-06,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 99017760,
+      "step": 1097
+    },
+    {
+      "epoch": 5.04367816091954,
+      "loss": 0.08533644676208496,
+      "loss_ce": 2.455884896335192e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 99017760,
+      "step": 1097
+    },
+    {
+      "epoch": 5.048275862068966,
+      "grad_norm": 8.297830161901864,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 99108192,
+      "step": 1098
+    },
+    {
+      "epoch": 5.048275862068966,
+      "loss": 0.084136001765728,
+      "loss_ce": 2.95531572191976e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 99108192,
+      "step": 1098
+    },
+    {
+      "epoch": 5.052873563218391,
+      "grad_norm": 10.472390357500974,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 99198528,
+      "step": 1099
+    },
+    {
+      "epoch": 5.052873563218391,
+      "loss": 0.10827698558568954,
+      "loss_ce": 3.113792627118528e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 99198528,
+      "step": 1099
+    },
+    {
+      "epoch": 5.057471264367816,
+      "grad_norm": 9.292431953931642,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 99288908,
+      "step": 1100
+    },
+    {
+      "epoch": 5.057471264367816,
+      "loss": 0.14652323722839355,
+      "loss_ce": 9.990337275667116e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 99288908,
+      "step": 1100
+    },
+    {
+      "epoch": 5.062068965517241,
+      "grad_norm": 6.538074511107595,
+      "learning_rate": 5e-06,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 99379280,
+      "step": 1101
+    },
+    {
+      "epoch": 5.062068965517241,
+      "loss": 0.07154995948076248,
+      "loss_ce": 3.2016840123105794e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 99379280,
+      "step": 1101
+    },
+    {
+      "epoch": 5.066666666666666,
+      "grad_norm": 11.332646756842689,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 99469616,
+      "step": 1102
+    },
+    {
+      "epoch": 5.066666666666666,
+      "loss": 0.13791075348854065,
+      "loss_ce": 1.82047119778872e-06,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 99469616,
+      "step": 1102
+    },
+    {
+      "epoch": 5.071264367816092,
+      "grad_norm": 4.595074957103007,
+      "learning_rate": 5e-06,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 99559192,
+      "step": 1103
+    },
+    {
+      "epoch": 5.071264367816092,
+      "loss": 0.06943254172801971,
+      "loss_ce": 0.0002797079214360565,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 99559192,
+      "step": 1103
+    },
+    {
+      "epoch": 5.075862068965518,
+      "grad_norm": 9.00228465896505,
+      "learning_rate": 5e-06,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 99649544,
+      "step": 1104
+    },
+    {
+      "epoch": 5.075862068965518,
+      "loss": 0.09139812737703323,
+      "loss_ce": 5.6064582167891786e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 99649544,
+      "step": 1104
+    },
+    {
+      "epoch": 5.080459770114943,
+      "grad_norm": 6.434548797647199,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 99739820,
+      "step": 1105
+    },
+    {
+      "epoch": 5.080459770114943,
+      "loss": 0.07129809260368347,
+      "loss_ce": 2.4284672690555453e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 99739820,
+      "step": 1105
+    },
+    {
+      "epoch": 5.085057471264368,
+      "grad_norm": 4.039364108519724,
+      "learning_rate": 5e-06,
+      "loss": 0.086,
+      "num_input_tokens_seen": 99830196,
+      "step": 1106
+    },
+    {
+      "epoch": 5.085057471264368,
+      "loss": 0.07135014235973358,
+      "loss_ce": 1.5301206076401286e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 99830196,
+      "step": 1106
+    },
+    {
+      "epoch": 5.089655172413793,
+      "grad_norm": 3.655688996317076,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 99920560,
+      "step": 1107
+    },
+    {
+      "epoch": 5.089655172413793,
+      "loss": 0.10698340833187103,
+      "loss_ce": 1.9302660803077742e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 99920560,
+      "step": 1107
+    },
+    {
+      "epoch": 5.094252873563218,
+      "grad_norm": 2.852608379852805,
+      "learning_rate": 5e-06,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 100010896,
+      "step": 1108
+    },
+    {
+      "epoch": 5.094252873563218,
+      "loss": 0.10475088655948639,
+      "loss_ce": 1.4554200788552407e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 100010896,
+      "step": 1108
+    },
+    {
+      "epoch": 5.098850574712643,
+      "grad_norm": 4.647429106957139,
+      "learning_rate": 5e-06,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 100101200,
+      "step": 1109
+    },
+    {
+      "epoch": 5.098850574712643,
+      "loss": 0.0705321729183197,
+      "loss_ce": 6.04609158472158e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 100101200,
+      "step": 1109
+    },
+    {
+      "epoch": 5.103448275862069,
+      "grad_norm": 4.542358695843599,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 100191484,
+      "step": 1110
+    },
+    {
+      "epoch": 5.103448275862069,
+      "loss": 0.11061207950115204,
+      "loss_ce": 1.637555760680698e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 100191484,
+      "step": 1110
+    },
+    {
+      "epoch": 5.1080459770114945,
+      "grad_norm": 3.826753741184423,
+      "learning_rate": 5e-06,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 100281800,
+      "step": 1111
+    },
+    {
+      "epoch": 5.1080459770114945,
+      "loss": 0.06089954450726509,
+      "loss_ce": 1.7182098872581264e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 100281800,
+      "step": 1111
+    },
+    {
+      "epoch": 5.11264367816092,
+      "grad_norm": 6.265815731921294,
+      "learning_rate": 5e-06,
+      "loss": 0.1078,
+      "num_input_tokens_seen": 100371368,
+      "step": 1112
+    },
+    {
+      "epoch": 5.11264367816092,
+      "loss": 0.12157793343067169,
+      "loss_ce": 1.1157572771480773e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 100371368,
+      "step": 1112
+    },
+    {
+      "epoch": 5.117241379310345,
+      "grad_norm": 15.357245102652293,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 100461748,
+      "step": 1113
+    },
+    {
+      "epoch": 5.117241379310345,
+      "loss": 0.07990702241659164,
+      "loss_ce": 1.2005009921267629e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 100461748,
+      "step": 1113
+    },
+    {
+      "epoch": 5.12183908045977,
+      "grad_norm": 18.749188171843112,
+      "learning_rate": 5e-06,
+      "loss": 0.0803,
+      "num_input_tokens_seen": 100552144,
+      "step": 1114
+    },
+    {
+      "epoch": 5.12183908045977,
+      "loss": 0.08655305206775665,
+      "loss_ce": 5.20814774063183e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 100552144,
+      "step": 1114
+    },
+    {
+      "epoch": 5.126436781609195,
+      "grad_norm": 11.884415815340965,
+      "learning_rate": 5e-06,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 100642552,
+      "step": 1115
+    },
+    {
+      "epoch": 5.126436781609195,
+      "loss": 0.09320227056741714,
+      "loss_ce": 1.5868474747549044e-06,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 100642552,
+      "step": 1115
+    },
+    {
+      "epoch": 5.13103448275862,
+      "grad_norm": 8.417856730574274,
+      "learning_rate": 5e-06,
+      "loss": 0.1048,
+      "num_input_tokens_seen": 100732936,
+      "step": 1116
+    },
+    {
+      "epoch": 5.13103448275862,
+      "loss": 0.10708998143672943,
+      "loss_ce": 1.9059289115830325e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 100732936,
+      "step": 1116
+    },
+    {
+      "epoch": 5.135632183908046,
+      "grad_norm": 10.398734680131996,
+      "learning_rate": 5e-06,
+      "loss": 0.062,
+      "num_input_tokens_seen": 100823412,
+      "step": 1117
+    },
+    {
+      "epoch": 5.135632183908046,
+      "loss": 0.0529075488448143,
+      "loss_ce": 3.5846733226208016e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 100823412,
+      "step": 1117
+    },
+    {
+      "epoch": 5.1402298850574715,
+      "grad_norm": 15.371538486553407,
+      "learning_rate": 5e-06,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 100913968,
+      "step": 1118
+    },
+    {
+      "epoch": 5.1402298850574715,
+      "loss": 0.05358627811074257,
+      "loss_ce": 8.896154758986086e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 100913968,
+      "step": 1118
+    },
+    {
+      "epoch": 5.144827586206897,
+      "grad_norm": 3.823708765476094,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 101004252,
+      "step": 1119
+    },
+    {
+      "epoch": 5.144827586206897,
+      "loss": 0.08534368127584457,
+      "loss_ce": 1.6531796063645743e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 101004252,
+      "step": 1119
+    },
+    {
+      "epoch": 5.149425287356322,
+      "grad_norm": 3.7000663900486743,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 101094664,
+      "step": 1120
+    },
+    {
+      "epoch": 5.149425287356322,
+      "loss": 0.09906667470932007,
+      "loss_ce": 0.0002965327585116029,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 101094664,
+      "step": 1120
+    },
+    {
+      "epoch": 5.154022988505747,
+      "grad_norm": 10.501783275375674,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 101185136,
+      "step": 1121
+    },
+    {
+      "epoch": 5.154022988505747,
+      "loss": 0.06211956962943077,
+      "loss_ce": 1.6294376109726727e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 101185136,
+      "step": 1121
+    },
+    {
+      "epoch": 5.158620689655172,
+      "grad_norm": 5.293184104736612,
+      "learning_rate": 5e-06,
+      "loss": 0.1864,
+      "num_input_tokens_seen": 101275444,
+      "step": 1122
+    },
+    {
+      "epoch": 5.158620689655172,
+      "loss": 0.13397647440433502,
+      "loss_ce": 4.307346898713149e-06,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 101275444,
+      "step": 1122
+    },
+    {
+      "epoch": 5.163218390804598,
+      "grad_norm": 7.75756649950046,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 101365108,
+      "step": 1123
+    },
+    {
+      "epoch": 5.163218390804598,
+      "loss": 0.07096749544143677,
+      "loss_ce": 1.4126431779004633e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 101365108,
+      "step": 1123
+    },
+    {
+      "epoch": 5.167816091954023,
+      "grad_norm": 7.295865334996816,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 101455576,
+      "step": 1124
+    },
+    {
+      "epoch": 5.167816091954023,
+      "loss": 0.07797445356845856,
+      "loss_ce": 0.00015463109593838453,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 101455576,
+      "step": 1124
+    },
+    {
+      "epoch": 5.172413793103448,
+      "grad_norm": 13.03580556396787,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 101546000,
+      "step": 1125
+    },
+    {
+      "epoch": 5.172413793103448,
+      "loss": 0.15164653956890106,
+      "loss_ce": 4.69675296699279e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 101546000,
+      "step": 1125
+    },
+    {
+      "epoch": 5.1770114942528735,
+      "grad_norm": 6.396126737755581,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 101636448,
+      "step": 1126
+    },
+    {
+      "epoch": 5.1770114942528735,
+      "loss": 0.07902251929044724,
+      "loss_ce": 7.354409899562597e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 101636448,
+      "step": 1126
+    },
+    {
+      "epoch": 5.181609195402299,
+      "grad_norm": 10.026494787186246,
+      "learning_rate": 5e-06,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 101725992,
+      "step": 1127
+    },
+    {
+      "epoch": 5.181609195402299,
+      "loss": 0.15227390825748444,
+      "loss_ce": 6.44809642835753e-06,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 101725992,
+      "step": 1127
+    },
+    {
+      "epoch": 5.186206896551724,
+      "grad_norm": 4.0752860409330935,
+      "learning_rate": 5e-06,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 101816380,
+      "step": 1128
+    },
+    {
+      "epoch": 5.186206896551724,
+      "loss": 0.06123349815607071,
+      "loss_ce": 1.5233559679472819e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 101816380,
+      "step": 1128
+    },
+    {
+      "epoch": 5.190804597701149,
+      "grad_norm": 2.938814507290908,
+      "learning_rate": 5e-06,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 101906756,
+      "step": 1129
+    },
+    {
+      "epoch": 5.190804597701149,
+      "loss": 0.06633394211530685,
+      "loss_ce": 1.9246750525780953e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 101906756,
+      "step": 1129
+    },
+    {
+      "epoch": 5.195402298850575,
+      "grad_norm": 4.357895106909357,
+      "learning_rate": 5e-06,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 101997112,
+      "step": 1130
+    },
+    {
+      "epoch": 5.195402298850575,
+      "loss": 0.062284551560878754,
+      "loss_ce": 1.343094845651649e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 101997112,
+      "step": 1130
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 3.8775406744142624,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 102087380,
+      "step": 1131
+    },
+    {
+      "epoch": 5.2,
+      "loss": 0.08111874014139175,
+      "loss_ce": 3.01617683362565e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 102087380,
+      "step": 1131
+    },
+    {
+      "epoch": 5.204597701149425,
+      "grad_norm": 4.692014408620994,
+      "learning_rate": 5e-06,
+      "loss": 0.087,
+      "num_input_tokens_seen": 102177880,
+      "step": 1132
+    },
+    {
+      "epoch": 5.204597701149425,
+      "loss": 0.11302285641431808,
+      "loss_ce": 3.152359931846149e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 102177880,
+      "step": 1132
+    },
+    {
+      "epoch": 5.2091954022988505,
+      "grad_norm": 3.3544978398050005,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 102267436,
+      "step": 1133
+    },
+    {
+      "epoch": 5.2091954022988505,
+      "loss": 0.15970894694328308,
+      "loss_ce": 1.0458235919941217e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 102267436,
+      "step": 1133
+    },
+    {
+      "epoch": 5.213793103448276,
+      "grad_norm": 2.4937616664035356,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 102357772,
+      "step": 1134
+    },
+    {
+      "epoch": 5.213793103448276,
+      "loss": 0.07588402926921844,
+      "loss_ce": 4.784332122653723e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 102357772,
+      "step": 1134
+    },
+    {
+      "epoch": 5.218390804597701,
+      "grad_norm": 11.78713407810388,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 102448228,
+      "step": 1135
+    },
+    {
+      "epoch": 5.218390804597701,
+      "loss": 0.12686412036418915,
+      "loss_ce": 2.5471808839938603e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 102448228,
+      "step": 1135
+    },
+    {
+      "epoch": 5.222988505747127,
+      "grad_norm": 7.345884023163392,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 102538620,
+      "step": 1136
+    },
+    {
+      "epoch": 5.222988505747127,
+      "loss": 0.07740141451358795,
+      "loss_ce": 2.40960180235561e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 102538620,
+      "step": 1136
+    },
+    {
+      "epoch": 5.227586206896552,
+      "grad_norm": 9.68121463931108,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 102628180,
+      "step": 1137
+    },
+    {
+      "epoch": 5.227586206896552,
+      "loss": 0.09852880239486694,
+      "loss_ce": 1.8055738109978847e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 102628180,
+      "step": 1137
+    },
+    {
+      "epoch": 5.232183908045977,
+      "grad_norm": 3.7617569064718013,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 102717736,
+      "step": 1138
+    },
+    {
+      "epoch": 5.232183908045977,
+      "loss": 0.08675423264503479,
+      "loss_ce": 2.3268486984306946e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 102717736,
+      "step": 1138
+    },
+    {
+      "epoch": 5.236781609195402,
+      "grad_norm": 3.9125840477144767,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 102808252,
+      "step": 1139
+    },
+    {
+      "epoch": 5.236781609195402,
+      "loss": 0.1034986600279808,
+      "loss_ce": 1.3551202755479608e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 102808252,
+      "step": 1139
+    },
+    {
+      "epoch": 5.241379310344827,
+      "grad_norm": 2.097538565665048,
+      "learning_rate": 5e-06,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 102898628,
+      "step": 1140
+    },
+    {
+      "epoch": 5.241379310344827,
+      "loss": 0.06704915314912796,
+      "loss_ce": 2.035472562056384e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 102898628,
+      "step": 1140
+    },
+    {
+      "epoch": 5.2459770114942526,
+      "grad_norm": 10.818635533541947,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 102988948,
+      "step": 1141
+    },
+    {
+      "epoch": 5.2459770114942526,
+      "loss": 0.07962171733379364,
+      "loss_ce": 2.4237215257016942e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 102988948,
+      "step": 1141
+    },
+    {
+      "epoch": 5.250574712643678,
+      "grad_norm": 7.192797833007836,
+      "learning_rate": 5e-06,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 103079352,
+      "step": 1142
+    },
+    {
+      "epoch": 5.250574712643678,
+      "loss": 0.08450967818498611,
+      "loss_ce": 6.502830729004927e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 103079352,
+      "step": 1142
+    },
+    {
+      "epoch": 5.255172413793104,
+      "grad_norm": 6.327504848695889,
+      "learning_rate": 5e-06,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 103169612,
+      "step": 1143
+    },
+    {
+      "epoch": 5.255172413793104,
+      "loss": 0.12516465783119202,
+      "loss_ce": 1.2068423529854044e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 103169612,
+      "step": 1143
+    },
+    {
+      "epoch": 5.259770114942529,
+      "grad_norm": 3.5430313631820884,
+      "learning_rate": 5e-06,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 103260048,
+      "step": 1144
+    },
+    {
+      "epoch": 5.259770114942529,
+      "loss": 0.06030872091650963,
+      "loss_ce": 5.9857538872165605e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 103260048,
+      "step": 1144
+    },
+    {
+      "epoch": 5.264367816091954,
+      "grad_norm": 15.453671778683573,
+      "learning_rate": 5e-06,
+      "loss": 0.0821,
+      "num_input_tokens_seen": 103350296,
+      "step": 1145
+    },
+    {
+      "epoch": 5.264367816091954,
+      "loss": 0.08191235363483429,
+      "loss_ce": 3.175610572725418e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 103350296,
+      "step": 1145
+    },
+    {
+      "epoch": 5.268965517241379,
+      "grad_norm": 6.718717304064819,
+      "learning_rate": 5e-06,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 103440532,
+      "step": 1146
+    },
+    {
+      "epoch": 5.268965517241379,
+      "loss": 0.06786315143108368,
+      "loss_ce": 2.2574939066544175e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 103440532,
+      "step": 1146
+    },
+    {
+      "epoch": 5.273563218390804,
+      "grad_norm": 17.691788722797366,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 103530900,
+      "step": 1147
+    },
+    {
+      "epoch": 5.273563218390804,
+      "loss": 0.11429747939109802,
+      "loss_ce": 9.143472198047675e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 103530900,
+      "step": 1147
+    },
+    {
+      "epoch": 5.2781609195402295,
+      "grad_norm": 18.892872568107364,
+      "learning_rate": 5e-06,
+      "loss": 0.0927,
+      "num_input_tokens_seen": 103621200,
+      "step": 1148
+    },
+    {
+      "epoch": 5.2781609195402295,
+      "loss": 0.05986202508211136,
+      "loss_ce": 1.7945828858501045e-06,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 103621200,
+      "step": 1148
+    },
+    {
+      "epoch": 5.2827586206896555,
+      "grad_norm": 19.67550320719364,
+      "learning_rate": 5e-06,
+      "loss": 0.1304,
+      "num_input_tokens_seen": 103711424,
+      "step": 1149
+    },
+    {
+      "epoch": 5.2827586206896555,
+      "loss": 0.15571549534797668,
+      "loss_ce": 1.4810606444370933e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 103711424,
+      "step": 1149
+    },
+    {
+      "epoch": 5.287356321839081,
+      "grad_norm": 10.959254538309226,
+      "learning_rate": 5e-06,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 103801748,
+      "step": 1150
+    },
+    {
+      "epoch": 5.287356321839081,
+      "loss": 0.08585269004106522,
+      "loss_ce": 6.745824521203758e-06,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 103801748,
+      "step": 1150
+    },
+    {
+      "epoch": 5.291954022988506,
+      "grad_norm": 6.5980592228912345,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 103892096,
+      "step": 1151
+    },
+    {
+      "epoch": 5.291954022988506,
+      "loss": 0.13581782579421997,
+      "loss_ce": 1.4601598195440602e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 103892096,
+      "step": 1151
+    },
+    {
+      "epoch": 5.296551724137931,
+      "grad_norm": 4.876191398306122,
+      "learning_rate": 5e-06,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 103982468,
+      "step": 1152
+    },
+    {
+      "epoch": 5.296551724137931,
+      "loss": 0.03845709189772606,
+      "loss_ce": 4.945307409798261e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 103982468,
+      "step": 1152
+    },
+    {
+      "epoch": 5.301149425287356,
+      "grad_norm": 8.328380265990786,
+      "learning_rate": 5e-06,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 104072856,
+      "step": 1153
+    },
+    {
+      "epoch": 5.301149425287356,
+      "loss": 0.07374840974807739,
+      "loss_ce": 2.674235474842135e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 104072856,
+      "step": 1153
+    },
+    {
+      "epoch": 5.305747126436781,
+      "grad_norm": 8.532773650525401,
+      "learning_rate": 5e-06,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 104162432,
+      "step": 1154
+    },
+    {
+      "epoch": 5.305747126436781,
+      "loss": 0.10141049325466156,
+      "loss_ce": 1.5837054888834246e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 104162432,
+      "step": 1154
+    },
+    {
+      "epoch": 5.310344827586207,
+      "grad_norm": 3.6361735465088287,
+      "learning_rate": 5e-06,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 104252752,
+      "step": 1155
+    },
+    {
+      "epoch": 5.310344827586207,
+      "loss": 0.06680650264024734,
+      "loss_ce": 1.8778815501718782e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 104252752,
+      "step": 1155
+    },
+    {
+      "epoch": 5.314942528735632,
+      "grad_norm": 27.691441920507174,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 104343184,
+      "step": 1156
+    },
+    {
+      "epoch": 5.314942528735632,
+      "loss": 0.11418163776397705,
+      "loss_ce": 1.5387857274618e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 104343184,
+      "step": 1156
+    },
+    {
+      "epoch": 5.319540229885058,
+      "grad_norm": 8.129126905990978,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 104433600,
+      "step": 1157
+    },
+    {
+      "epoch": 5.319540229885058,
+      "loss": 0.1396959125995636,
+      "loss_ce": 4.747589264297858e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 104433600,
+      "step": 1157
+    },
+    {
+      "epoch": 5.324137931034483,
+      "grad_norm": 2.5246901713346284,
+      "learning_rate": 5e-06,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 104524052,
+      "step": 1158
+    },
+    {
+      "epoch": 5.324137931034483,
+      "loss": 0.08805723488330841,
+      "loss_ce": 1.4025717064214405e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 104524052,
+      "step": 1158
+    },
+    {
+      "epoch": 5.328735632183908,
+      "grad_norm": 6.032560857969416,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 104613016,
+      "step": 1159
+    },
+    {
+      "epoch": 5.328735632183908,
+      "loss": 0.10382804274559021,
+      "loss_ce": 2.2498687030747533e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 104613016,
+      "step": 1159
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 3.524592869861034,
+      "learning_rate": 5e-06,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 104703380,
+      "step": 1160
+    },
+    {
+      "epoch": 5.333333333333333,
+      "loss": 0.0802207663655281,
+      "loss_ce": 5.31496243638685e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 104703380,
+      "step": 1160
+    },
+    {
+      "epoch": 5.337931034482759,
+      "grad_norm": 4.969952758545786,
+      "learning_rate": 5e-06,
+      "loss": 0.0857,
+      "num_input_tokens_seen": 104793836,
+      "step": 1161
+    },
+    {
+      "epoch": 5.337931034482759,
+      "loss": 0.09672747552394867,
+      "loss_ce": 2.0068077901669312e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 104793836,
+      "step": 1161
+    },
+    {
+      "epoch": 5.342528735632184,
+      "grad_norm": 6.83000765916347,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 104884288,
+      "step": 1162
+    },
+    {
+      "epoch": 5.342528735632184,
+      "loss": 0.061434537172317505,
+      "loss_ce": 1.7906917491927743e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 104884288,
+      "step": 1162
+    },
+    {
+      "epoch": 5.347126436781609,
+      "grad_norm": 15.595761412741526,
+      "learning_rate": 5e-06,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 104974708,
+      "step": 1163
+    },
+    {
+      "epoch": 5.347126436781609,
+      "loss": 0.09743352234363556,
+      "loss_ce": 6.154959919513203e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 104974708,
+      "step": 1163
+    },
+    {
+      "epoch": 5.3517241379310345,
+      "grad_norm": 9.98309693591081,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 105064256,
+      "step": 1164
+    },
+    {
+      "epoch": 5.3517241379310345,
+      "loss": 0.06634046137332916,
+      "loss_ce": 7.154278864618391e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 105064256,
+      "step": 1164
+    },
+    {
+      "epoch": 5.35632183908046,
+      "grad_norm": 14.072679950902984,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 105154564,
+      "step": 1165
+    },
+    {
+      "epoch": 5.35632183908046,
+      "loss": 0.07407604902982712,
+      "loss_ce": 9.88530428003287e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 105154564,
+      "step": 1165
+    },
+    {
+      "epoch": 5.360919540229885,
+      "grad_norm": 18.28267664671943,
+      "learning_rate": 5e-06,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 105244844,
+      "step": 1166
+    },
+    {
+      "epoch": 5.360919540229885,
+      "loss": 0.12162074446678162,
+      "loss_ce": 8.195744158001617e-06,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 105244844,
+      "step": 1166
+    },
+    {
+      "epoch": 5.36551724137931,
+      "grad_norm": 20.34315476487668,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 105335136,
+      "step": 1167
+    },
+    {
+      "epoch": 5.36551724137931,
+      "loss": 0.08798626065254211,
+      "loss_ce": 1.933791281771846e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 105335136,
+      "step": 1167
+    },
+    {
+      "epoch": 5.370114942528736,
+      "grad_norm": 3.8998763710651003,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 105425568,
+      "step": 1168
+    },
+    {
+      "epoch": 5.370114942528736,
+      "loss": 0.13412779569625854,
+      "loss_ce": 3.038228669538512e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 105425568,
+      "step": 1168
+    },
+    {
+      "epoch": 5.374712643678161,
+      "grad_norm": 3.9686217500508056,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 105515932,
+      "step": 1169
+    },
+    {
+      "epoch": 5.374712643678161,
+      "loss": 0.12314353883266449,
+      "loss_ce": 5.114484793011798e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 105515932,
+      "step": 1169
+    },
+    {
+      "epoch": 5.379310344827586,
+      "grad_norm": 9.188483207270677,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 105606460,
+      "step": 1170
+    },
+    {
+      "epoch": 5.379310344827586,
+      "loss": 0.0674906075000763,
+      "loss_ce": 3.1501404009759426e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 105606460,
+      "step": 1170
+    },
+    {
+      "epoch": 5.3839080459770114,
+      "grad_norm": 13.62728698765477,
+      "learning_rate": 5e-06,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 105696844,
+      "step": 1171
+    },
+    {
+      "epoch": 5.3839080459770114,
+      "loss": 0.05981367826461792,
+      "loss_ce": 1.448656985303387e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 105696844,
+      "step": 1171
+    },
+    {
+      "epoch": 5.388505747126437,
+      "grad_norm": 7.114022345551658,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 105786460,
+      "step": 1172
+    },
+    {
+      "epoch": 5.388505747126437,
+      "loss": 0.09999861568212509,
+      "loss_ce": 2.3021595552563667e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 105786460,
+      "step": 1172
+    },
+    {
+      "epoch": 5.393103448275862,
+      "grad_norm": 4.472914999467795,
+      "learning_rate": 5e-06,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 105876844,
+      "step": 1173
+    },
+    {
+      "epoch": 5.393103448275862,
+      "loss": 0.0646452009677887,
+      "loss_ce": 8.972834621090442e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 105876844,
+      "step": 1173
+    },
+    {
+      "epoch": 5.397701149425288,
+      "grad_norm": 10.796359253655176,
+      "learning_rate": 5e-06,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 105967272,
+      "step": 1174
+    },
+    {
+      "epoch": 5.397701149425288,
+      "loss": 0.07185741513967514,
+      "loss_ce": 3.7756042274850188e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 105967272,
+      "step": 1174
+    },
+    {
+      "epoch": 5.402298850574713,
+      "grad_norm": 3.5477662420576417,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 106057604,
+      "step": 1175
+    },
+    {
+      "epoch": 5.402298850574713,
+      "loss": 0.07314296066761017,
+      "loss_ce": 5.335675086826086e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 106057604,
+      "step": 1175
+    },
+    {
+      "epoch": 5.406896551724138,
+      "grad_norm": 3.654190105765733,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 106148012,
+      "step": 1176
+    },
+    {
+      "epoch": 5.406896551724138,
+      "loss": 0.0644906684756279,
+      "loss_ce": 7.029044809314655e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 106148012,
+      "step": 1176
+    },
+    {
+      "epoch": 5.411494252873563,
+      "grad_norm": 2.0857683176392814,
+      "learning_rate": 5e-06,
+      "loss": 0.0829,
+      "num_input_tokens_seen": 106236844,
+      "step": 1177
+    },
+    {
+      "epoch": 5.411494252873563,
+      "loss": 0.10485847294330597,
+      "loss_ce": 1.5335266652982682e-05,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 106236844,
+      "step": 1177
+    },
+    {
+      "epoch": 5.416091954022988,
+      "grad_norm": 4.7865908582620245,
+      "learning_rate": 5e-06,
+      "loss": 0.058,
+      "num_input_tokens_seen": 106327196,
+      "step": 1178
+    },
+    {
+      "epoch": 5.416091954022988,
+      "loss": 0.07083334028720856,
+      "loss_ce": 2.0433958525245544e-06,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 106327196,
+      "step": 1178
+    },
+    {
+      "epoch": 5.4206896551724135,
+      "grad_norm": 4.381050617741422,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 106417488,
+      "step": 1179
+    },
+    {
+      "epoch": 5.4206896551724135,
+      "loss": 0.1030469685792923,
+      "loss_ce": 4.370722763269441e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 106417488,
+      "step": 1179
+    },
+    {
+      "epoch": 5.425287356321839,
+      "grad_norm": 13.90099727079854,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 106507752,
+      "step": 1180
+    },
+    {
+      "epoch": 5.425287356321839,
+      "loss": 0.11128388345241547,
+      "loss_ce": 1.5395814898511162e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 106507752,
+      "step": 1180
+    },
+    {
+      "epoch": 5.429885057471265,
+      "grad_norm": 3.514125111179275,
+      "learning_rate": 5e-06,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 106598104,
+      "step": 1181
+    },
+    {
+      "epoch": 5.429885057471265,
+      "loss": 0.05190886929631233,
+      "loss_ce": 1.372814949718304e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 106598104,
+      "step": 1181
+    },
+    {
+      "epoch": 5.43448275862069,
+      "grad_norm": 7.033128839020526,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 106688460,
+      "step": 1182
+    },
+    {
+      "epoch": 5.43448275862069,
+      "loss": 0.10943891108036041,
+      "loss_ce": 2.8767924504791154e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 106688460,
+      "step": 1182
+    },
+    {
+      "epoch": 5.439080459770115,
+      "grad_norm": 7.6220799806368005,
+      "learning_rate": 5e-06,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 106778892,
+      "step": 1183
+    },
+    {
+      "epoch": 5.439080459770115,
+      "loss": 0.13684214651584625,
+      "loss_ce": 3.1839987059356645e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 106778892,
+      "step": 1183
+    },
+    {
+      "epoch": 5.44367816091954,
+      "grad_norm": 3.201233161347327,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 106869232,
+      "step": 1184
+    },
+    {
+      "epoch": 5.44367816091954,
+      "loss": 0.11746550351381302,
+      "loss_ce": 3.350429324200377e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 106869232,
+      "step": 1184
+    },
+    {
+      "epoch": 5.448275862068965,
+      "grad_norm": 27.107960993966334,
+      "learning_rate": 5e-06,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 106959604,
+      "step": 1185
+    },
+    {
+      "epoch": 5.448275862068965,
+      "loss": 0.08980852365493774,
+      "loss_ce": 2.580695945653133e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 106959604,
+      "step": 1185
+    },
+    {
+      "epoch": 5.4528735632183905,
+      "grad_norm": 5.056052589774106,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 107049260,
+      "step": 1186
+    },
+    {
+      "epoch": 5.4528735632183905,
+      "loss": 0.0906330794095993,
+      "loss_ce": 1.1129596714454237e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 107049260,
+      "step": 1186
+    },
+    {
+      "epoch": 5.4574712643678165,
+      "grad_norm": 12.077013452361433,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 107139660,
+      "step": 1187
+    },
+    {
+      "epoch": 5.4574712643678165,
+      "loss": 0.08940693736076355,
+      "loss_ce": 2.0952680642949417e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 107139660,
+      "step": 1187
+    },
+    {
+      "epoch": 5.462068965517242,
+      "grad_norm": 5.117358275086491,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 107229976,
+      "step": 1188
+    },
+    {
+      "epoch": 5.462068965517242,
+      "loss": 0.1243157610297203,
+      "loss_ce": 0.0007195670041255653,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 107229976,
+      "step": 1188
+    },
+    {
+      "epoch": 5.466666666666667,
+      "grad_norm": 3.84883008143984,
+      "learning_rate": 5e-06,
+      "loss": 0.063,
+      "num_input_tokens_seen": 107320400,
+      "step": 1189
+    },
+    {
+      "epoch": 5.466666666666667,
+      "loss": 0.04762866720557213,
+      "loss_ce": 5.176388003746979e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 107320400,
+      "step": 1189
+    },
+    {
+      "epoch": 5.471264367816092,
+      "grad_norm": 11.839810250471555,
+      "learning_rate": 5e-06,
+      "loss": 0.1302,
+      "num_input_tokens_seen": 107410776,
+      "step": 1190
+    },
+    {
+      "epoch": 5.471264367816092,
+      "loss": 0.15773043036460876,
+      "loss_ce": 1.558406620461028e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 107410776,
+      "step": 1190
+    },
+    {
+      "epoch": 5.475862068965517,
+      "grad_norm": 3.3411639885164677,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 107500968,
+      "step": 1191
+    },
+    {
+      "epoch": 5.475862068965517,
+      "loss": 0.08212100714445114,
+      "loss_ce": 0.00012027601769659668,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 107500968,
+      "step": 1191
+    },
+    {
+      "epoch": 5.480459770114942,
+      "grad_norm": 2.668518928020011,
+      "learning_rate": 5e-06,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 107591340,
+      "step": 1192
+    },
+    {
+      "epoch": 5.480459770114942,
+      "loss": 0.11270363628864288,
+      "loss_ce": 6.325638969428837e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 107591340,
+      "step": 1192
+    },
+    {
+      "epoch": 5.485057471264367,
+      "grad_norm": 3.423089746700433,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 107681688,
+      "step": 1193
+    },
+    {
+      "epoch": 5.485057471264367,
+      "loss": 0.06189355626702309,
+      "loss_ce": 3.9072983781807125e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 107681688,
+      "step": 1193
+    },
+    {
+      "epoch": 5.489655172413793,
+      "grad_norm": 10.782127348669686,
+      "learning_rate": 5e-06,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 107772020,
+      "step": 1194
+    },
+    {
+      "epoch": 5.489655172413793,
+      "loss": 0.10825426876544952,
+      "loss_ce": 3.8937610952416435e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 107772020,
+      "step": 1194
+    },
+    {
+      "epoch": 5.494252873563219,
+      "grad_norm": 10.321202311466154,
+      "learning_rate": 5e-06,
+      "loss": 0.058,
+      "num_input_tokens_seen": 107862420,
+      "step": 1195
+    },
+    {
+      "epoch": 5.494252873563219,
+      "loss": 0.06930500268936157,
+      "loss_ce": 1.4843457392998971e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 107862420,
+      "step": 1195
+    },
+    {
+      "epoch": 5.498850574712644,
+      "grad_norm": 1.7063662050581758,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 107952696,
+      "step": 1196
+    },
+    {
+      "epoch": 5.498850574712644,
+      "loss": 0.09334798157215118,
+      "loss_ce": 9.968431186280213e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 107952696,
+      "step": 1196
+    },
+    {
+      "epoch": 5.503448275862069,
+      "grad_norm": 5.082277526279771,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 108043016,
+      "step": 1197
+    },
+    {
+      "epoch": 5.503448275862069,
+      "loss": 0.0524345263838768,
+      "loss_ce": 5.3265985116013326e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 108043016,
+      "step": 1197
+    },
+    {
+      "epoch": 5.508045977011494,
+      "grad_norm": 3.926403547610472,
+      "learning_rate": 5e-06,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 108133392,
+      "step": 1198
+    },
+    {
+      "epoch": 5.508045977011494,
+      "loss": 0.055503346025943756,
+      "loss_ce": 7.1315121203952e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 108133392,
+      "step": 1198
+    },
+    {
+      "epoch": 5.512643678160919,
+      "grad_norm": 19.466891089230415,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 108223928,
+      "step": 1199
+    },
+    {
+      "epoch": 5.512643678160919,
+      "loss": 0.1550014615058899,
+      "loss_ce": 3.3197462471434847e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 108223928,
+      "step": 1199
+    },
+    {
+      "epoch": 5.517241379310345,
+      "grad_norm": 8.883212548201756,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 108314296,
+      "step": 1200
+    },
+    {
+      "epoch": 5.517241379310345,
+      "loss": 0.11142435669898987,
+      "loss_ce": 1.9941233404097147e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 108314296,
+      "step": 1200
+    },
+    {
+      "epoch": 5.52183908045977,
+      "grad_norm": 4.261788928644466,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 108404724,
+      "step": 1201
+    },
+    {
+      "epoch": 5.52183908045977,
+      "loss": 0.09745465219020844,
+      "loss_ce": 1.2026385775243398e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 108404724,
+      "step": 1201
+    },
+    {
+      "epoch": 5.5264367816091955,
+      "grad_norm": 2.987130600386024,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 108495024,
+      "step": 1202
+    },
+    {
+      "epoch": 5.5264367816091955,
+      "loss": 0.15418361127376556,
+      "loss_ce": 6.983517960179597e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 108495024,
+      "step": 1202
+    },
+    {
+      "epoch": 5.531034482758621,
+      "grad_norm": 2.3653856974799323,
+      "learning_rate": 5e-06,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 108585416,
+      "step": 1203
+    },
+    {
+      "epoch": 5.531034482758621,
+      "loss": 0.06593763828277588,
+      "loss_ce": 1.96732235053787e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 108585416,
+      "step": 1203
+    },
+    {
+      "epoch": 5.535632183908046,
+      "grad_norm": 3.330630153861718,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 108675664,
+      "step": 1204
+    },
+    {
+      "epoch": 5.535632183908046,
+      "loss": 0.11529748141765594,
+      "loss_ce": 2.066693014057819e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 108675664,
+      "step": 1204
+    },
+    {
+      "epoch": 5.540229885057471,
+      "grad_norm": 6.486089157651149,
+      "learning_rate": 5e-06,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 108765884,
+      "step": 1205
+    },
+    {
+      "epoch": 5.540229885057471,
+      "loss": 0.07939766347408295,
+      "loss_ce": 6.185707206896041e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 108765884,
+      "step": 1205
+    },
+    {
+      "epoch": 5.544827586206896,
+      "grad_norm": 5.30690182772674,
+      "learning_rate": 5e-06,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 108856308,
+      "step": 1206
+    },
+    {
+      "epoch": 5.544827586206896,
+      "loss": 0.07844828069210052,
+      "loss_ce": 1.810871071938891e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 108856308,
+      "step": 1206
+    },
+    {
+      "epoch": 5.549425287356322,
+      "grad_norm": 3.5054972022904183,
+      "learning_rate": 5e-06,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 108946648,
+      "step": 1207
+    },
+    {
+      "epoch": 5.549425287356322,
+      "loss": 0.07729589194059372,
+      "loss_ce": 1.0129077963938471e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 108946648,
+      "step": 1207
+    },
+    {
+      "epoch": 5.554022988505747,
+      "grad_norm": 9.78898145858432,
+      "learning_rate": 5e-06,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 109037112,
+      "step": 1208
+    },
+    {
+      "epoch": 5.554022988505747,
+      "loss": 0.07796503603458405,
+      "loss_ce": 7.88130364526296e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 109037112,
+      "step": 1208
+    },
+    {
+      "epoch": 5.558620689655172,
+      "grad_norm": 4.85707186606737,
+      "learning_rate": 5e-06,
+      "loss": 0.109,
+      "num_input_tokens_seen": 109126120,
+      "step": 1209
+    },
+    {
+      "epoch": 5.558620689655172,
+      "loss": 0.12425880879163742,
+      "loss_ce": 2.1751711756223813e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 109126120,
+      "step": 1209
+    },
+    {
+      "epoch": 5.563218390804598,
+      "grad_norm": 8.786051137164725,
+      "learning_rate": 5e-06,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 109216380,
+      "step": 1210
+    },
+    {
+      "epoch": 5.563218390804598,
+      "loss": 0.0687236338853836,
+      "loss_ce": 2.855926322808955e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 109216380,
+      "step": 1210
+    },
+    {
+      "epoch": 5.567816091954023,
+      "grad_norm": 3.482256798617991,
+      "learning_rate": 5e-06,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 109306644,
+      "step": 1211
+    },
+    {
+      "epoch": 5.567816091954023,
+      "loss": 0.0845988318324089,
+      "loss_ce": 4.1002945181389805e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 109306644,
+      "step": 1211
+    },
+    {
+      "epoch": 5.572413793103449,
+      "grad_norm": 4.264015760408713,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 109397144,
+      "step": 1212
+    },
+    {
+      "epoch": 5.572413793103449,
+      "loss": 0.07220568507909775,
+      "loss_ce": 1.6351415979443118e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 109397144,
+      "step": 1212
+    },
+    {
+      "epoch": 5.577011494252874,
+      "grad_norm": 3.1255568749212816,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 109487456,
+      "step": 1213
+    },
+    {
+      "epoch": 5.577011494252874,
+      "loss": 0.052804555743932724,
+      "loss_ce": 9.145278454525396e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 109487456,
+      "step": 1213
+    },
+    {
+      "epoch": 5.581609195402299,
+      "grad_norm": 3.94664764195231,
+      "learning_rate": 5e-06,
+      "loss": 0.067,
+      "num_input_tokens_seen": 109577820,
+      "step": 1214
+    },
+    {
+      "epoch": 5.581609195402299,
+      "loss": 0.04463765025138855,
+      "loss_ce": 5.693935236195102e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 109577820,
+      "step": 1214
+    },
+    {
+      "epoch": 5.586206896551724,
+      "grad_norm": 8.212489399334382,
+      "learning_rate": 5e-06,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 109668260,
+      "step": 1215
+    },
+    {
+      "epoch": 5.586206896551724,
+      "loss": 0.06496123224496841,
+      "loss_ce": 5.034160494687967e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 109668260,
+      "step": 1215
+    },
+    {
+      "epoch": 5.590804597701149,
+      "grad_norm": 9.692768417013069,
+      "learning_rate": 5e-06,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 109758576,
+      "step": 1216
+    },
+    {
+      "epoch": 5.590804597701149,
+      "loss": 0.07001283019781113,
+      "loss_ce": 5.50604636373464e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 109758576,
+      "step": 1216
+    },
+    {
+      "epoch": 5.5954022988505745,
+      "grad_norm": 8.222517521217215,
+      "learning_rate": 5e-06,
+      "loss": 0.0974,
+      "num_input_tokens_seen": 109848992,
+      "step": 1217
+    },
+    {
+      "epoch": 5.5954022988505745,
+      "loss": 0.09305752068758011,
+      "loss_ce": 2.4687436962267384e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 109848992,
+      "step": 1217
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 14.945824448106285,
+      "learning_rate": 5e-06,
+      "loss": 0.0871,
+      "num_input_tokens_seen": 109939364,
+      "step": 1218
+    },
+    {
+      "epoch": 5.6,
+      "loss": 0.09159138798713684,
+      "loss_ce": 8.140663339872845e-06,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 109939364,
+      "step": 1218
+    },
+    {
+      "epoch": 5.604597701149425,
+      "grad_norm": 28.416530398349355,
+      "learning_rate": 5e-06,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 110029792,
+      "step": 1219
+    },
+    {
+      "epoch": 5.604597701149425,
+      "loss": 0.046073831617832184,
+      "loss_ce": 7.547159839305095e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 110029792,
+      "step": 1219
+    },
+    {
+      "epoch": 5.609195402298851,
+      "grad_norm": 12.008787286117224,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 110120336,
+      "step": 1220
+    },
+    {
+      "epoch": 5.609195402298851,
+      "loss": 0.11564863473176956,
+      "loss_ce": 4.804190029972233e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 110120336,
+      "step": 1220
+    },
+    {
+      "epoch": 5.613793103448276,
+      "grad_norm": 22.46676513633078,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 110210540,
+      "step": 1221
+    },
+    {
+      "epoch": 5.613793103448276,
+      "loss": 0.07196103781461716,
+      "loss_ce": 5.928850441705436e-07,
+      "loss_iou": 0.390625,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 110210540,
+      "step": 1221
+    },
+    {
+      "epoch": 5.618390804597701,
+      "grad_norm": 23.841340782969,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 110299420,
+      "step": 1222
+    },
+    {
+      "epoch": 5.618390804597701,
+      "loss": 0.11307486146688461,
+      "loss_ce": 7.235145858430769e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 110299420,
+      "step": 1222
+    },
+    {
+      "epoch": 5.622988505747126,
+      "grad_norm": 7.337714001096004,
+      "learning_rate": 5e-06,
+      "loss": 0.088,
+      "num_input_tokens_seen": 110389696,
+      "step": 1223
+    },
+    {
+      "epoch": 5.622988505747126,
+      "loss": 0.07472731918096542,
+      "loss_ce": 6.606188981095329e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 110389696,
+      "step": 1223
+    },
+    {
+      "epoch": 5.627586206896551,
+      "grad_norm": 11.251552468198575,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 110480088,
+      "step": 1224
+    },
+    {
+      "epoch": 5.627586206896551,
+      "loss": 0.0974973812699318,
+      "loss_ce": 1.3505153901860467e-06,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 110480088,
+      "step": 1224
+    },
+    {
+      "epoch": 5.6321839080459775,
+      "grad_norm": 5.577183353924994,
+      "learning_rate": 5e-06,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 110570532,
+      "step": 1225
+    },
+    {
+      "epoch": 5.6321839080459775,
+      "loss": 0.10643593221902847,
+      "loss_ce": 2.1139007003512233e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 110570532,
+      "step": 1225
+    },
+    {
+      "epoch": 5.636781609195403,
+      "grad_norm": 6.701001289894439,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 110659360,
+      "step": 1226
+    },
+    {
+      "epoch": 5.636781609195403,
+      "loss": 0.1065259799361229,
+      "loss_ce": 5.0150600145570934e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 110659360,
+      "step": 1226
+    },
+    {
+      "epoch": 5.641379310344828,
+      "grad_norm": 5.386629642258452,
+      "learning_rate": 5e-06,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 110749656,
+      "step": 1227
+    },
+    {
+      "epoch": 5.641379310344828,
+      "loss": 0.07365281879901886,
+      "loss_ce": 1.3907001630286686e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 110749656,
+      "step": 1227
+    },
+    {
+      "epoch": 5.645977011494253,
+      "grad_norm": 5.0480129670553175,
+      "learning_rate": 5e-06,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 110839976,
+      "step": 1228
+    },
+    {
+      "epoch": 5.645977011494253,
+      "loss": 0.09352147579193115,
+      "loss_ce": 1.5612586139468476e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 110839976,
+      "step": 1228
+    },
+    {
+      "epoch": 5.650574712643678,
+      "grad_norm": 3.2142834138747904,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 110930364,
+      "step": 1229
+    },
+    {
+      "epoch": 5.650574712643678,
+      "loss": 0.1116681694984436,
+      "loss_ce": 4.347142748883925e-06,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 110930364,
+      "step": 1229
+    },
+    {
+      "epoch": 5.655172413793103,
+      "grad_norm": 4.682890336305807,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 111020076,
+      "step": 1230
+    },
+    {
+      "epoch": 5.655172413793103,
+      "loss": 0.14498014748096466,
+      "loss_ce": 6.385195320035564e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 111020076,
+      "step": 1230
+    },
+    {
+      "epoch": 5.659770114942528,
+      "grad_norm": 7.235632320302035,
+      "learning_rate": 5e-06,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 111110316,
+      "step": 1231
+    },
+    {
+      "epoch": 5.659770114942528,
+      "loss": 0.0729096531867981,
+      "loss_ce": 3.36788289132528e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 111110316,
+      "step": 1231
+    },
+    {
+      "epoch": 5.664367816091954,
+      "grad_norm": 2.3169355950035992,
+      "learning_rate": 5e-06,
+      "loss": 0.0644,
+      "num_input_tokens_seen": 111200728,
+      "step": 1232
+    },
+    {
+      "epoch": 5.664367816091954,
+      "loss": 0.07670523226261139,
+      "loss_ce": 1.4556415408151224e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 111200728,
+      "step": 1232
+    },
+    {
+      "epoch": 5.6689655172413795,
+      "grad_norm": 9.054517929993095,
+      "learning_rate": 5e-06,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 111290364,
+      "step": 1233
+    },
+    {
+      "epoch": 5.6689655172413795,
+      "loss": 0.13384370505809784,
+      "loss_ce": 2.4130109522957355e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 111290364,
+      "step": 1233
+    },
+    {
+      "epoch": 5.673563218390805,
+      "grad_norm": 2.4147250325583536,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 111380760,
+      "step": 1234
+    },
+    {
+      "epoch": 5.673563218390805,
+      "loss": 0.08226744830608368,
+      "loss_ce": 7.31718364477274e-06,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 111380760,
+      "step": 1234
+    },
+    {
+      "epoch": 5.67816091954023,
+      "grad_norm": 4.243893457772928,
+      "learning_rate": 5e-06,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 111471116,
+      "step": 1235
+    },
+    {
+      "epoch": 5.67816091954023,
+      "loss": 0.07256424427032471,
+      "loss_ce": 8.707445886102505e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 111471116,
+      "step": 1235
+    },
+    {
+      "epoch": 5.682758620689655,
+      "grad_norm": 1.8993669782662472,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 111561436,
+      "step": 1236
+    },
+    {
+      "epoch": 5.682758620689655,
+      "loss": 0.13686691224575043,
+      "loss_ce": 3.2013495001592673e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 111561436,
+      "step": 1236
+    },
+    {
+      "epoch": 5.68735632183908,
+      "grad_norm": 4.878685087840704,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 111651816,
+      "step": 1237
+    },
+    {
+      "epoch": 5.68735632183908,
+      "loss": 0.07835354655981064,
+      "loss_ce": 1.4925600225978997e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 111651816,
+      "step": 1237
+    },
+    {
+      "epoch": 5.691954022988506,
+      "grad_norm": 8.928325096054786,
+      "learning_rate": 5e-06,
+      "loss": 0.1009,
+      "num_input_tokens_seen": 111742112,
+      "step": 1238
+    },
+    {
+      "epoch": 5.691954022988506,
+      "loss": 0.12431655079126358,
+      "loss_ce": 1.8453329175827093e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 111742112,
+      "step": 1238
+    },
+    {
+      "epoch": 5.696551724137931,
+      "grad_norm": 4.857562908474219,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 111832440,
+      "step": 1239
+    },
+    {
+      "epoch": 5.696551724137931,
+      "loss": 0.10367751866579056,
+      "loss_ce": 9.310015229857527e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 111832440,
+      "step": 1239
+    },
+    {
+      "epoch": 5.7011494252873565,
+      "grad_norm": 2.1347164816602997,
+      "learning_rate": 5e-06,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 111922904,
+      "step": 1240
+    },
+    {
+      "epoch": 5.7011494252873565,
+      "loss": 0.06786856055259705,
+      "loss_ce": 1.2722593055514153e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 111922904,
+      "step": 1240
+    },
+    {
+      "epoch": 5.705747126436782,
+      "grad_norm": 8.531354475094487,
+      "learning_rate": 5e-06,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 112013212,
+      "step": 1241
+    },
+    {
+      "epoch": 5.705747126436782,
+      "loss": 0.061662279069423676,
+      "loss_ce": 1.5145051293075085e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 112013212,
+      "step": 1241
+    },
+    {
+      "epoch": 5.710344827586207,
+      "grad_norm": 18.82697007656036,
+      "learning_rate": 5e-06,
+      "loss": 0.1572,
+      "num_input_tokens_seen": 112103552,
+      "step": 1242
+    },
+    {
+      "epoch": 5.710344827586207,
+      "loss": 0.1454518437385559,
+      "loss_ce": 5.055922429164639e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 112103552,
+      "step": 1242
+    },
+    {
+      "epoch": 5.714942528735632,
+      "grad_norm": 5.439005001756043,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 112193936,
+      "step": 1243
+    },
+    {
+      "epoch": 5.714942528735632,
+      "loss": 0.04381529614329338,
+      "loss_ce": 3.7829264329047874e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 112193936,
+      "step": 1243
+    },
+    {
+      "epoch": 5.719540229885057,
+      "grad_norm": 18.879016825135775,
+      "learning_rate": 5e-06,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 112284184,
+      "step": 1244
+    },
+    {
+      "epoch": 5.719540229885057,
+      "loss": 0.10435809940099716,
+      "loss_ce": 1.8497697965358384e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 112284184,
+      "step": 1244
+    },
+    {
+      "epoch": 5.724137931034483,
+      "grad_norm": 4.028482212688753,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 112374556,
+      "step": 1245
+    },
+    {
+      "epoch": 5.724137931034483,
+      "loss": 0.1041383296251297,
+      "loss_ce": 1.2350090401014313e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 112374556,
+      "step": 1245
+    },
+    {
+      "epoch": 5.728735632183908,
+      "grad_norm": 6.123310827199504,
+      "learning_rate": 5e-06,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 112464124,
+      "step": 1246
+    },
+    {
+      "epoch": 5.728735632183908,
+      "loss": 0.0810786560177803,
+      "loss_ce": 8.710509973752778e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 112464124,
+      "step": 1246
+    },
+    {
+      "epoch": 5.733333333333333,
+      "grad_norm": 6.108386978174192,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 112554496,
+      "step": 1247
+    },
+    {
+      "epoch": 5.733333333333333,
+      "loss": 0.06712520867586136,
+      "loss_ce": 1.7052239854820073e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 112554496,
+      "step": 1247
+    },
+    {
+      "epoch": 5.7379310344827585,
+      "grad_norm": 6.640331304354758,
+      "learning_rate": 5e-06,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 112644988,
+      "step": 1248
+    },
+    {
+      "epoch": 5.7379310344827585,
+      "loss": 0.10246039181947708,
+      "loss_ce": 2.814024264807813e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 112644988,
+      "step": 1248
+    },
+    {
+      "epoch": 5.742528735632184,
+      "grad_norm": 14.848833163560919,
+      "learning_rate": 5e-06,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 112735364,
+      "step": 1249
+    },
+    {
+      "epoch": 5.742528735632184,
+      "loss": 0.0761798620223999,
+      "loss_ce": 7.98174187366385e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 112735364,
+      "step": 1249
+    },
+    {
+      "epoch": 5.747126436781609,
+      "grad_norm": 11.72709348916867,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 112825776,
+      "step": 1250
+    },
+    {
+      "epoch": 5.747126436781609,
+      "eval_seeclick_CIoU": 0.4573807418346405,
+      "eval_seeclick_GIoU": 0.4410470873117447,
+      "eval_seeclick_IoU": 0.49894072115421295,
+      "eval_seeclick_MAE_all": 0.061043232679367065,
+      "eval_seeclick_MAE_h": 0.053627705201506615,
+      "eval_seeclick_MAE_w": 0.11304067447781563,
+      "eval_seeclick_MAE_x_boxes": 0.10517356544733047,
+      "eval_seeclick_MAE_y_boxes": 0.05592040531337261,
+      "eval_seeclick_NUM_probability": 0.9999991357326508,
+      "eval_seeclick_inside_bbox": 0.7542613744735718,
+      "eval_seeclick_loss": 0.37355467677116394,
+      "eval_seeclick_loss_ce": 0.07825379073619843,
+      "eval_seeclick_loss_iou": 0.5123291015625,
+      "eval_seeclick_loss_num": 0.06250762939453125,
+      "eval_seeclick_loss_xval": 0.31231689453125,
+      "eval_seeclick_runtime": 85.2206,
+      "eval_seeclick_samples_per_second": 0.505,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 112825776,
+      "step": 1250
+    },
+    {
+      "epoch": 5.747126436781609,
+      "eval_icons_CIoU": 0.5825834572315216,
+      "eval_icons_GIoU": 0.5834551155567169,
+      "eval_icons_IoU": 0.616531640291214,
+      "eval_icons_MAE_all": 0.043770069256424904,
+      "eval_icons_MAE_h": 0.07383041456341743,
+      "eval_icons_MAE_w": 0.06421982683241367,
+      "eval_icons_MAE_x_boxes": 0.05823229439556599,
+      "eval_icons_MAE_y_boxes": 0.07248594611883163,
+      "eval_icons_NUM_probability": 0.999999612569809,
+      "eval_icons_inside_bbox": 0.7673611044883728,
+      "eval_icons_loss": 0.21254034340381622,
+      "eval_icons_loss_ce": 4.21214093648814e-06,
+      "eval_icons_loss_iou": 0.447265625,
+      "eval_icons_loss_num": 0.04521942138671875,
+      "eval_icons_loss_xval": 0.226104736328125,
+      "eval_icons_runtime": 85.9172,
+      "eval_icons_samples_per_second": 0.582,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 112825776,
+      "step": 1250
+    },
+    {
+      "epoch": 5.747126436781609,
+      "eval_screenspot_CIoU": 0.4307680130004883,
+      "eval_screenspot_GIoU": 0.4167415201663971,
+      "eval_screenspot_IoU": 0.4970965087413788,
+      "eval_screenspot_MAE_all": 0.08507336676120758,
+      "eval_screenspot_MAE_h": 0.07819427798191707,
+      "eval_screenspot_MAE_w": 0.17445524781942368,
+      "eval_screenspot_MAE_x_boxes": 0.1695632884899775,
+      "eval_screenspot_MAE_y_boxes": 0.0742349624633789,
+      "eval_screenspot_NUM_probability": 0.9999984900156657,
+      "eval_screenspot_inside_bbox": 0.7637499968210856,
+      "eval_screenspot_loss": 0.4254446029663086,
+      "eval_screenspot_loss_ce": 6.250914157135412e-05,
+      "eval_screenspot_loss_iou": 0.3890787760416667,
+      "eval_screenspot_loss_num": 0.087158203125,
+      "eval_screenspot_loss_xval": 0.4358723958333333,
+      "eval_screenspot_runtime": 160.2493,
+      "eval_screenspot_samples_per_second": 0.555,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 112825776,
+      "step": 1250
+    },
+    {
+      "epoch": 5.747126436781609,
+      "eval_compot_CIoU": 0.48233287036418915,
+      "eval_compot_GIoU": 0.4589613825082779,
+      "eval_compot_IoU": 0.5473670959472656,
+      "eval_compot_MAE_all": 0.05806807801127434,
+      "eval_compot_MAE_h": 0.07524906471371651,
+      "eval_compot_MAE_w": 0.11501751467585564,
+      "eval_compot_MAE_x_boxes": 0.10463830083608627,
+      "eval_compot_MAE_y_boxes": 0.07366756349802017,
+      "eval_compot_NUM_probability": 0.999998927116394,
+      "eval_compot_inside_bbox": 0.7760416567325592,
+      "eval_compot_loss": 0.3192625343799591,
+      "eval_compot_loss_ce": 0.0109380844514817,
+      "eval_compot_loss_iou": 0.501953125,
+      "eval_compot_loss_num": 0.0531005859375,
+      "eval_compot_loss_xval": 0.265533447265625,
+      "eval_compot_runtime": 86.8999,
+      "eval_compot_samples_per_second": 0.575,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 112825776,
+      "step": 1250
+    },
+    {
+      "epoch": 5.747126436781609,
+      "loss": 0.1921016126871109,
+      "loss_ce": 0.006859918590635061,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 112825776,
+      "step": 1250
+    },
+    {
+      "epoch": 5.751724137931035,
+      "grad_norm": 48.39229529580768,
+      "learning_rate": 5e-06,
+      "loss": 0.1628,
+      "num_input_tokens_seen": 112916064,
+      "step": 1251
+    },
+    {
+      "epoch": 5.751724137931035,
+      "loss": 0.14167295396327972,
+      "loss_ce": 4.0873368561733514e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 112916064,
+      "step": 1251
+    },
+    {
+      "epoch": 5.75632183908046,
+      "grad_norm": 7.309880590381852,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 113006264,
+      "step": 1252
+    },
+    {
+      "epoch": 5.75632183908046,
+      "loss": 0.1379764974117279,
+      "loss_ce": 6.526359356939793e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 113006264,
+      "step": 1252
+    },
+    {
+      "epoch": 5.760919540229885,
+      "grad_norm": 4.7842464607453,
+      "learning_rate": 5e-06,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 113096588,
+      "step": 1253
+    },
+    {
+      "epoch": 5.760919540229885,
+      "loss": 0.06593406945466995,
+      "loss_ce": 8.451766007055994e-07,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 113096588,
+      "step": 1253
+    },
+    {
+      "epoch": 5.76551724137931,
+      "grad_norm": 9.30963816664603,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 113186932,
+      "step": 1254
+    },
+    {
+      "epoch": 5.76551724137931,
+      "loss": 0.10627120733261108,
+      "loss_ce": 9.002267688629217e-06,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 113186932,
+      "step": 1254
+    },
+    {
+      "epoch": 5.7701149425287355,
+      "grad_norm": 5.032701026633435,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 113277316,
+      "step": 1255
+    },
+    {
+      "epoch": 5.7701149425287355,
+      "loss": 0.1125558465719223,
+      "loss_ce": 7.017050393187674e-06,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 113277316,
+      "step": 1255
+    },
+    {
+      "epoch": 5.774712643678161,
+      "grad_norm": 7.348457813587205,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 113367640,
+      "step": 1256
+    },
+    {
+      "epoch": 5.774712643678161,
+      "loss": 0.06398795545101166,
+      "loss_ce": 0.00011466428986750543,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 113367640,
+      "step": 1256
+    },
+    {
+      "epoch": 5.779310344827586,
+      "grad_norm": 3.6430091269186304,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 113457940,
+      "step": 1257
+    },
+    {
+      "epoch": 5.779310344827586,
+      "loss": 0.07985039055347443,
+      "loss_ce": 1.6411700926255435e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 113457940,
+      "step": 1257
+    },
+    {
+      "epoch": 5.783908045977012,
+      "grad_norm": 3.3598636033304454,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 113548324,
+      "step": 1258
+    },
+    {
+      "epoch": 5.783908045977012,
+      "loss": 0.11753726005554199,
+      "loss_ce": 4.458225521375425e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 113548324,
+      "step": 1258
+    },
+    {
+      "epoch": 5.788505747126437,
+      "grad_norm": 4.068043795707238,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 113638644,
+      "step": 1259
+    },
+    {
+      "epoch": 5.788505747126437,
+      "loss": 0.138657346367836,
+      "loss_ce": 7.249969939948642e-07,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 113638644,
+      "step": 1259
+    },
+    {
+      "epoch": 5.793103448275862,
+      "grad_norm": 2.279662032054951,
+      "learning_rate": 5e-06,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 113729080,
+      "step": 1260
+    },
+    {
+      "epoch": 5.793103448275862,
+      "loss": 0.10126355290412903,
+      "loss_ce": 6.22008246864425e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 113729080,
+      "step": 1260
+    },
+    {
+      "epoch": 5.797701149425287,
+      "grad_norm": 5.649226738355132,
+      "learning_rate": 5e-06,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 113819484,
+      "step": 1261
+    },
+    {
+      "epoch": 5.797701149425287,
+      "loss": 0.08105180412530899,
+      "loss_ce": 4.743245881400071e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 113819484,
+      "step": 1261
+    },
+    {
+      "epoch": 5.802298850574712,
+      "grad_norm": 1.5641332197845963,
+      "learning_rate": 5e-06,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 113909888,
+      "step": 1262
+    },
+    {
+      "epoch": 5.802298850574712,
+      "loss": 0.061422187834978104,
+      "loss_ce": 5.559354576689657e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 113909888,
+      "step": 1262
+    },
+    {
+      "epoch": 5.8068965517241375,
+      "grad_norm": 15.670084350710225,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 114000272,
+      "step": 1263
+    },
+    {
+      "epoch": 5.8068965517241375,
+      "loss": 0.11508636921644211,
+      "loss_ce": 4.585368969856063e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 114000272,
+      "step": 1263
+    },
+    {
+      "epoch": 5.811494252873564,
+      "grad_norm": 3.835335458461329,
+      "learning_rate": 5e-06,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 114090524,
+      "step": 1264
+    },
+    {
+      "epoch": 5.811494252873564,
+      "loss": 0.08618977665901184,
+      "loss_ce": 8.132398761517834e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 114090524,
+      "step": 1264
+    },
+    {
+      "epoch": 5.816091954022989,
+      "grad_norm": 9.99731943967783,
+      "learning_rate": 5e-06,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 114180868,
+      "step": 1265
+    },
+    {
+      "epoch": 5.816091954022989,
+      "loss": 0.1053665354847908,
+      "loss_ce": 4.596740836859681e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 114180868,
+      "step": 1265
+    },
+    {
+      "epoch": 5.820689655172414,
+      "grad_norm": 2.911511418981655,
+      "learning_rate": 5e-06,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 114271204,
+      "step": 1266
+    },
+    {
+      "epoch": 5.820689655172414,
+      "loss": 0.07806918770074844,
+      "loss_ce": 5.221021638135426e-06,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 114271204,
+      "step": 1266
+    },
+    {
+      "epoch": 5.825287356321839,
+      "grad_norm": 6.2344029082542,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 114361536,
+      "step": 1267
+    },
+    {
+      "epoch": 5.825287356321839,
+      "loss": 0.05028009042143822,
+      "loss_ce": 2.380430260018329e-06,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 114361536,
+      "step": 1267
+    },
+    {
+      "epoch": 5.829885057471264,
+      "grad_norm": 3.7663498969437175,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 114451868,
+      "step": 1268
+    },
+    {
+      "epoch": 5.829885057471264,
+      "loss": 0.08314789086580276,
+      "loss_ce": 2.7451824280433357e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 114451868,
+      "step": 1268
+    },
+    {
+      "epoch": 5.834482758620689,
+      "grad_norm": 7.1687713447544406,
+      "learning_rate": 5e-06,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 114542224,
+      "step": 1269
+    },
+    {
+      "epoch": 5.834482758620689,
+      "loss": 0.08698830008506775,
+      "loss_ce": 4.3723342969315127e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 114542224,
+      "step": 1269
+    },
+    {
+      "epoch": 5.8390804597701145,
+      "grad_norm": 7.785551994387596,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 114632548,
+      "step": 1270
+    },
+    {
+      "epoch": 5.8390804597701145,
+      "loss": 0.04482455551624298,
+      "loss_ce": 9.493617653788533e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 114632548,
+      "step": 1270
+    },
+    {
+      "epoch": 5.8436781609195405,
+      "grad_norm": 4.9792230982606815,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 114722960,
+      "step": 1271
+    },
+    {
+      "epoch": 5.8436781609195405,
+      "loss": 0.0937369242310524,
+      "loss_ce": 2.1818657387484564e-06,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 114722960,
+      "step": 1271
+    },
+    {
+      "epoch": 5.848275862068966,
+      "grad_norm": 7.05750958333928,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 114813360,
+      "step": 1272
+    },
+    {
+      "epoch": 5.848275862068966,
+      "loss": 0.08838266134262085,
+      "loss_ce": 1.9016588339582086e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 114813360,
+      "step": 1272
+    },
+    {
+      "epoch": 5.852873563218391,
+      "grad_norm": 2.2929149537355933,
+      "learning_rate": 5e-06,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 114903664,
+      "step": 1273
+    },
+    {
+      "epoch": 5.852873563218391,
+      "loss": 0.08243682235479355,
+      "loss_ce": 8.84367273101816e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 114903664,
+      "step": 1273
+    },
+    {
+      "epoch": 5.857471264367816,
+      "grad_norm": 11.633034929488161,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 114994008,
+      "step": 1274
+    },
+    {
+      "epoch": 5.857471264367816,
+      "loss": 0.1149655431509018,
+      "loss_ce": 5.821747436129954e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 114994008,
+      "step": 1274
+    },
+    {
+      "epoch": 5.862068965517241,
+      "grad_norm": 13.85967700936422,
+      "learning_rate": 5e-06,
+      "loss": 0.0762,
+      "num_input_tokens_seen": 115084404,
+      "step": 1275
+    },
+    {
+      "epoch": 5.862068965517241,
+      "loss": 0.07548947632312775,
+      "loss_ce": 4.2469814616197255e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 115084404,
+      "step": 1275
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 8.591339370910877,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 115174752,
+      "step": 1276
+    },
+    {
+      "epoch": 5.866666666666667,
+      "loss": 0.08044753968715668,
+      "loss_ce": 3.201591425749939e-06,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 115174752,
+      "step": 1276
+    },
+    {
+      "epoch": 5.871264367816092,
+      "grad_norm": 10.263498686615506,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 115265232,
+      "step": 1277
+    },
+    {
+      "epoch": 5.871264367816092,
+      "loss": 0.12569460272789001,
+      "loss_ce": 7.947778613015544e-06,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 115265232,
+      "step": 1277
+    },
+    {
+      "epoch": 5.875862068965517,
+      "grad_norm": 11.400124255336879,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 115355708,
+      "step": 1278
+    },
+    {
+      "epoch": 5.875862068965517,
+      "loss": 0.1197689026594162,
+      "loss_ce": 2.66920051217312e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 115355708,
+      "step": 1278
+    },
+    {
+      "epoch": 5.880459770114943,
+      "grad_norm": 4.3793822993822715,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 115446204,
+      "step": 1279
+    },
+    {
+      "epoch": 5.880459770114943,
+      "loss": 0.1235429123044014,
+      "loss_ce": 7.75079388404265e-06,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 115446204,
+      "step": 1279
+    },
+    {
+      "epoch": 5.885057471264368,
+      "grad_norm": 8.059241246157443,
+      "learning_rate": 5e-06,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 115536632,
+      "step": 1280
+    },
+    {
+      "epoch": 5.885057471264368,
+      "loss": 0.08218298107385635,
+      "loss_ce": 1.439858169760555e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 115536632,
+      "step": 1280
+    },
+    {
+      "epoch": 5.889655172413793,
+      "grad_norm": 9.48342209591454,
+      "learning_rate": 5e-06,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 115627096,
+      "step": 1281
+    },
+    {
+      "epoch": 5.889655172413793,
+      "loss": 0.08530169725418091,
+      "loss_ce": 5.068644895800389e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 115627096,
+      "step": 1281
+    },
+    {
+      "epoch": 5.894252873563218,
+      "grad_norm": 4.958627136670461,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 115717508,
+      "step": 1282
+    },
+    {
+      "epoch": 5.894252873563218,
+      "loss": 0.10856582224369049,
+      "loss_ce": 1.4800659300817642e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 115717508,
+      "step": 1282
+    },
+    {
+      "epoch": 5.898850574712643,
+      "grad_norm": 2.619956686440511,
+      "learning_rate": 5e-06,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 115807116,
+      "step": 1283
+    },
+    {
+      "epoch": 5.898850574712643,
+      "loss": 0.1267540454864502,
+      "loss_ce": 1.4542956705554388e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 115807116,
+      "step": 1283
+    },
+    {
+      "epoch": 5.903448275862069,
+      "grad_norm": 1.9178286629575005,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 115897536,
+      "step": 1284
+    },
+    {
+      "epoch": 5.903448275862069,
+      "loss": 0.06888516247272491,
+      "loss_ce": 6.993210263317451e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 115897536,
+      "step": 1284
+    },
+    {
+      "epoch": 5.908045977011494,
+      "grad_norm": 4.687249341290416,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 115987980,
+      "step": 1285
+    },
+    {
+      "epoch": 5.908045977011494,
+      "loss": 0.05946702882647514,
+      "loss_ce": 1.87891364475945e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 115987980,
+      "step": 1285
+    },
+    {
+      "epoch": 5.9126436781609195,
+      "grad_norm": 9.032450106052789,
+      "learning_rate": 5e-06,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 116078456,
+      "step": 1286
+    },
+    {
+      "epoch": 5.9126436781609195,
+      "loss": 0.051251400262117386,
+      "loss_ce": 1.238779896084452e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 116078456,
+      "step": 1286
+    },
+    {
+      "epoch": 5.917241379310345,
+      "grad_norm": 6.2264189550968805,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 116168752,
+      "step": 1287
+    },
+    {
+      "epoch": 5.917241379310345,
+      "loss": 0.1135876327753067,
+      "loss_ce": 1.2120165138185257e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 116168752,
+      "step": 1287
+    },
+    {
+      "epoch": 5.92183908045977,
+      "grad_norm": 4.579123122057683,
+      "learning_rate": 5e-06,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 116259044,
+      "step": 1288
+    },
+    {
+      "epoch": 5.92183908045977,
+      "loss": 0.09734739363193512,
+      "loss_ce": 2.6833829906536266e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 116259044,
+      "step": 1288
+    },
+    {
+      "epoch": 5.926436781609196,
+      "grad_norm": 6.37974017894556,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 116349420,
+      "step": 1289
+    },
+    {
+      "epoch": 5.926436781609196,
+      "loss": 0.12337080389261246,
+      "loss_ce": 3.4989259347639745e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 116349420,
+      "step": 1289
+    },
+    {
+      "epoch": 5.931034482758621,
+      "grad_norm": 5.236407477085609,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 116439780,
+      "step": 1290
+    },
+    {
+      "epoch": 5.931034482758621,
+      "loss": 0.11725269258022308,
+      "loss_ce": 4.161834567639744e-06,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 116439780,
+      "step": 1290
+    },
+    {
+      "epoch": 5.935632183908046,
+      "grad_norm": 6.939019502168552,
+      "learning_rate": 5e-06,
+      "loss": 0.0819,
+      "num_input_tokens_seen": 116530192,
+      "step": 1291
+    },
+    {
+      "epoch": 5.935632183908046,
+      "loss": 0.10045474022626877,
+      "loss_ce": 5.191009404370561e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 116530192,
+      "step": 1291
+    },
+    {
+      "epoch": 5.940229885057471,
+      "grad_norm": 5.923858548372062,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 116620592,
+      "step": 1292
+    },
+    {
+      "epoch": 5.940229885057471,
+      "loss": 0.061371780931949615,
+      "loss_ce": 9.32706370804226e-07,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 116620592,
+      "step": 1292
+    },
+    {
+      "epoch": 5.944827586206896,
+      "grad_norm": 5.436359048849817,
+      "learning_rate": 5e-06,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 116710964,
+      "step": 1293
+    },
+    {
+      "epoch": 5.944827586206896,
+      "loss": 0.05396192520856857,
+      "loss_ce": 6.846322776254965e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 116710964,
+      "step": 1293
+    },
+    {
+      "epoch": 5.949425287356322,
+      "grad_norm": 7.23604737625205,
+      "learning_rate": 5e-06,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 116800480,
+      "step": 1294
+    },
+    {
+      "epoch": 5.949425287356322,
+      "loss": 0.08210571110248566,
+      "loss_ce": 1.3423375094134826e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 116800480,
+      "step": 1294
+    },
+    {
+      "epoch": 5.954022988505747,
+      "grad_norm": 2.028553713875407,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 116890808,
+      "step": 1295
+    },
+    {
+      "epoch": 5.954022988505747,
+      "loss": 0.09970621019601822,
+      "loss_ce": 3.580232078093104e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 116890808,
+      "step": 1295
+    },
+    {
+      "epoch": 5.958620689655173,
+      "grad_norm": 8.871343329486912,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 116981168,
+      "step": 1296
+    },
+    {
+      "epoch": 5.958620689655173,
+      "loss": 0.0988428071141243,
+      "loss_ce": 2.6887415515375324e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 116981168,
+      "step": 1296
+    },
+    {
+      "epoch": 5.963218390804598,
+      "grad_norm": 6.717350736332018,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 117071416,
+      "step": 1297
+    },
+    {
+      "epoch": 5.963218390804598,
+      "loss": 0.10025615990161896,
+      "loss_ce": 5.9157691794098355e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 117071416,
+      "step": 1297
+    },
+    {
+      "epoch": 5.967816091954023,
+      "grad_norm": 8.58100221704676,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 117161736,
+      "step": 1298
+    },
+    {
+      "epoch": 5.967816091954023,
+      "loss": 0.10571430623531342,
+      "loss_ce": 3.193915108568035e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 117161736,
+      "step": 1298
+    },
+    {
+      "epoch": 5.972413793103448,
+      "grad_norm": 11.414603962669956,
+      "learning_rate": 5e-06,
+      "loss": 0.0951,
+      "num_input_tokens_seen": 117252180,
+      "step": 1299
+    },
+    {
+      "epoch": 5.972413793103448,
+      "loss": 0.09330064058303833,
+      "loss_ce": 3.891930464305915e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 117252180,
+      "step": 1299
+    },
+    {
+      "epoch": 5.977011494252873,
+      "grad_norm": 14.302282290509867,
+      "learning_rate": 5e-06,
+      "loss": 0.1022,
+      "num_input_tokens_seen": 117342672,
+      "step": 1300
+    },
+    {
+      "epoch": 5.977011494252873,
+      "loss": 0.09649240970611572,
+      "loss_ce": 1.108789183490444e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 117342672,
+      "step": 1300
+    },
+    {
+      "epoch": 5.9816091954022985,
+      "grad_norm": 4.2991078616550515,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 117433108,
+      "step": 1301
+    },
+    {
+      "epoch": 5.9816091954022985,
+      "loss": 0.11118273437023163,
+      "loss_ce": 7.203209406725364e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 117433108,
+      "step": 1301
+    },
+    {
+      "epoch": 5.9862068965517246,
+      "grad_norm": 8.51577704791888,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 117523460,
+      "step": 1302
+    },
+    {
+      "epoch": 5.9862068965517246,
+      "loss": 0.07513649761676788,
+      "loss_ce": 2.222008106400608e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 117523460,
+      "step": 1302
+    },
+    {
+      "epoch": 5.99080459770115,
+      "grad_norm": 2.677824187892606,
+      "learning_rate": 5e-06,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 117613652,
+      "step": 1303
+    },
+    {
+      "epoch": 5.99080459770115,
+      "loss": 0.12823861837387085,
+      "loss_ce": 1.9021113985218108e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 117613652,
+      "step": 1303
+    },
+    {
+      "epoch": 5.995402298850575,
+      "grad_norm": 7.857525256311372,
+      "learning_rate": 5e-06,
+      "loss": 0.0986,
+      "num_input_tokens_seen": 117703216,
+      "step": 1304
+    },
+    {
+      "epoch": 5.995402298850575,
+      "loss": 0.12224046885967255,
+      "loss_ce": 3.282908437540755e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 117703216,
+      "step": 1304
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 3.121211959039489,
+      "learning_rate": 5e-06,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 117793536,
+      "step": 1305
+    },
+    {
+      "epoch": 6.0,
+      "loss": 0.0870618224143982,
+      "loss_ce": 2.569025309639983e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 117793536,
+      "step": 1305
+    },
+    {
+      "epoch": 6.004597701149425,
+      "grad_norm": 10.240533439865134,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 117883776,
+      "step": 1306
+    },
+    {
+      "epoch": 6.004597701149425,
+      "loss": 0.1209506243467331,
+      "loss_ce": 1.8270541204401525e-06,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 117883776,
+      "step": 1306
+    },
+    {
+      "epoch": 6.00919540229885,
+      "grad_norm": 9.764972826473027,
+      "learning_rate": 5e-06,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 117974056,
+      "step": 1307
+    },
+    {
+      "epoch": 6.00919540229885,
+      "loss": 0.05394062399864197,
+      "loss_ce": 8.061311973506236e-07,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 117974056,
+      "step": 1307
+    },
+    {
+      "epoch": 6.0137931034482754,
+      "grad_norm": 2.2633345565968166,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 118064428,
+      "step": 1308
+    },
+    {
+      "epoch": 6.0137931034482754,
+      "loss": 0.09164643287658691,
+      "loss_ce": 2.152237811969826e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 118064428,
+      "step": 1308
+    },
+    {
+      "epoch": 6.0183908045977015,
+      "grad_norm": 16.306990357156884,
+      "learning_rate": 5e-06,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 118154820,
+      "step": 1309
+    },
+    {
+      "epoch": 6.0183908045977015,
+      "loss": 0.0783190131187439,
+      "loss_ce": 2.6165927920374088e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 118154820,
+      "step": 1309
+    },
+    {
+      "epoch": 6.022988505747127,
+      "grad_norm": 5.540727316669466,
+      "learning_rate": 5e-06,
+      "loss": 0.093,
+      "num_input_tokens_seen": 118245100,
+      "step": 1310
+    },
+    {
+      "epoch": 6.022988505747127,
+      "loss": 0.07801572233438492,
+      "loss_ce": 1.2792523193638772e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 118245100,
+      "step": 1310
+    },
+    {
+      "epoch": 6.027586206896552,
+      "grad_norm": 6.643253537988018,
+      "learning_rate": 5e-06,
+      "loss": 0.0683,
+      "num_input_tokens_seen": 118335420,
+      "step": 1311
+    },
+    {
+      "epoch": 6.027586206896552,
+      "loss": 0.07764066010713577,
+      "loss_ce": 3.9446958908229135e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 118335420,
+      "step": 1311
+    },
+    {
+      "epoch": 6.032183908045977,
+      "grad_norm": 6.788294653006372,
+      "learning_rate": 5e-06,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 118425852,
+      "step": 1312
+    },
+    {
+      "epoch": 6.032183908045977,
+      "loss": 0.09127424657344818,
+      "loss_ce": 2.6687364879762754e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 118425852,
+      "step": 1312
+    },
+    {
+      "epoch": 6.036781609195402,
+      "grad_norm": 3.573469877891761,
+      "learning_rate": 5e-06,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 118516176,
+      "step": 1313
+    },
+    {
+      "epoch": 6.036781609195402,
+      "loss": 0.10177914798259735,
+      "loss_ce": 3.0242531465773936e-06,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 118516176,
+      "step": 1313
+    },
+    {
+      "epoch": 6.041379310344827,
+      "grad_norm": 2.587490346214488,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 118606480,
+      "step": 1314
+    },
+    {
+      "epoch": 6.041379310344827,
+      "loss": 0.07138314843177795,
+      "loss_ce": 2.536031388444826e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 118606480,
+      "step": 1314
+    },
+    {
+      "epoch": 6.045977011494253,
+      "grad_norm": 8.936376542422272,
+      "learning_rate": 5e-06,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 118696880,
+      "step": 1315
+    },
+    {
+      "epoch": 6.045977011494253,
+      "loss": 0.07341183722019196,
+      "loss_ce": 1.8055545751849422e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 118696880,
+      "step": 1315
+    },
+    {
+      "epoch": 6.050574712643678,
+      "grad_norm": 12.767342588452228,
+      "learning_rate": 5e-06,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 118787220,
+      "step": 1316
+    },
+    {
+      "epoch": 6.050574712643678,
+      "loss": 0.062339168041944504,
+      "loss_ce": 0.00011382619413780048,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 118787220,
+      "step": 1316
+    },
+    {
+      "epoch": 6.055172413793104,
+      "grad_norm": 10.482681799720183,
+      "learning_rate": 5e-06,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 118877572,
+      "step": 1317
+    },
+    {
+      "epoch": 6.055172413793104,
+      "loss": 0.04859533905982971,
+      "loss_ce": 1.1354484740877524e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 118877572,
+      "step": 1317
+    },
+    {
+      "epoch": 6.059770114942529,
+      "grad_norm": 12.105797073827482,
+      "learning_rate": 5e-06,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 118967880,
+      "step": 1318
+    },
+    {
+      "epoch": 6.059770114942529,
+      "loss": 0.1294127106666565,
+      "loss_ce": 2.920265615102835e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 118967880,
+      "step": 1318
+    },
+    {
+      "epoch": 6.064367816091954,
+      "grad_norm": 1.9154409626850923,
+      "learning_rate": 5e-06,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 119058296,
+      "step": 1319
+    },
+    {
+      "epoch": 6.064367816091954,
+      "loss": 0.09125930815935135,
+      "loss_ce": 2.701348603295628e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 119058296,
+      "step": 1319
+    },
+    {
+      "epoch": 6.068965517241379,
+      "grad_norm": 1.9985880290089337,
+      "learning_rate": 5e-06,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 119148684,
+      "step": 1320
+    },
+    {
+      "epoch": 6.068965517241379,
+      "loss": 0.1152973622083664,
+      "loss_ce": 1.952744241862092e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 119148684,
+      "step": 1320
+    },
+    {
+      "epoch": 6.073563218390804,
+      "grad_norm": 6.069434025871692,
+      "learning_rate": 5e-06,
+      "loss": 0.0762,
+      "num_input_tokens_seen": 119239004,
+      "step": 1321
+    },
+    {
+      "epoch": 6.073563218390804,
+      "loss": 0.060495294630527496,
+      "loss_ce": 5.5230746511369944e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 119239004,
+      "step": 1321
+    },
+    {
+      "epoch": 6.07816091954023,
+      "grad_norm": 10.274039091733039,
+      "learning_rate": 5e-06,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 119329324,
+      "step": 1322
+    },
+    {
+      "epoch": 6.07816091954023,
+      "loss": 0.10678257048130035,
+      "loss_ce": 1.5705069245086634e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 119329324,
+      "step": 1322
+    },
+    {
+      "epoch": 6.082758620689655,
+      "grad_norm": 3.0466261855959544,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 119419688,
+      "step": 1323
+    },
+    {
+      "epoch": 6.082758620689655,
+      "loss": 0.06655121594667435,
+      "loss_ce": 7.638142960786354e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 119419688,
+      "step": 1323
+    },
+    {
+      "epoch": 6.0873563218390805,
+      "grad_norm": 3.872617007458599,
+      "learning_rate": 5e-06,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 119510040,
+      "step": 1324
+    },
+    {
+      "epoch": 6.0873563218390805,
+      "loss": 0.06537115573883057,
+      "loss_ce": 4.8276036977767944e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 119510040,
+      "step": 1324
+    },
+    {
+      "epoch": 6.091954022988506,
+      "grad_norm": 14.262285949730552,
+      "learning_rate": 5e-06,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 119600444,
+      "step": 1325
+    },
+    {
+      "epoch": 6.091954022988506,
+      "loss": 0.06211470812559128,
+      "loss_ce": 1.1435970009188168e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 119600444,
+      "step": 1325
+    },
+    {
+      "epoch": 6.096551724137931,
+      "grad_norm": 5.582731893935351,
+      "learning_rate": 5e-06,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 119690784,
+      "step": 1326
+    },
+    {
+      "epoch": 6.096551724137931,
+      "loss": 0.08985207229852676,
+      "loss_ce": 6.9361602072604e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 119690784,
+      "step": 1326
+    },
+    {
+      "epoch": 6.101149425287356,
+      "grad_norm": 8.165861222533081,
+      "learning_rate": 5e-06,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 119781104,
+      "step": 1327
+    },
+    {
+      "epoch": 6.101149425287356,
+      "loss": 0.06276784837245941,
+      "loss_ce": 8.212633133553027e-07,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 119781104,
+      "step": 1327
+    },
+    {
+      "epoch": 6.105747126436782,
+      "grad_norm": 11.7019496158795,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 119871456,
+      "step": 1328
+    },
+    {
+      "epoch": 6.105747126436782,
+      "loss": 0.06695958971977234,
+      "loss_ce": 4.02033947466407e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 119871456,
+      "step": 1328
+    },
+    {
+      "epoch": 6.110344827586207,
+      "grad_norm": 11.09087965893546,
+      "learning_rate": 5e-06,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 119961804,
+      "step": 1329
+    },
+    {
+      "epoch": 6.110344827586207,
+      "loss": 0.06232057511806488,
+      "loss_ce": 3.6832166188105475e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 119961804,
+      "step": 1329
+    },
+    {
+      "epoch": 6.114942528735632,
+      "grad_norm": 1.7273225707646134,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 120052288,
+      "step": 1330
+    },
+    {
+      "epoch": 6.114942528735632,
+      "loss": 0.05487591400742531,
+      "loss_ce": 5.3095800467417575e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 120052288,
+      "step": 1330
+    },
+    {
+      "epoch": 6.119540229885057,
+      "grad_norm": 1.895873348755958,
+      "learning_rate": 5e-06,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 120142748,
+      "step": 1331
+    },
+    {
+      "epoch": 6.119540229885057,
+      "loss": 0.08318766951560974,
+      "loss_ce": 1.2013872037641704e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 120142748,
+      "step": 1331
+    },
+    {
+      "epoch": 6.124137931034483,
+      "grad_norm": 12.365279177414369,
+      "learning_rate": 5e-06,
+      "loss": 0.0879,
+      "num_input_tokens_seen": 120231640,
+      "step": 1332
+    },
+    {
+      "epoch": 6.124137931034483,
+      "loss": 0.11951163411140442,
+      "loss_ce": 2.006320391956251e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 120231640,
+      "step": 1332
+    },
+    {
+      "epoch": 6.128735632183908,
+      "grad_norm": 8.100321593651932,
+      "learning_rate": 5e-06,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 120322024,
+      "step": 1333
+    },
+    {
+      "epoch": 6.128735632183908,
+      "loss": 0.07466816902160645,
+      "loss_ce": 6.915668109286344e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 120322024,
+      "step": 1333
+    },
+    {
+      "epoch": 6.133333333333334,
+      "grad_norm": 10.20027536647147,
+      "learning_rate": 5e-06,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 120412328,
+      "step": 1334
+    },
+    {
+      "epoch": 6.133333333333334,
+      "loss": 0.07674149423837662,
+      "loss_ce": 5.042035354563268e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 120412328,
+      "step": 1334
+    },
+    {
+      "epoch": 6.137931034482759,
+      "grad_norm": 15.927575829987184,
+      "learning_rate": 5e-06,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 120502728,
+      "step": 1335
+    },
+    {
+      "epoch": 6.137931034482759,
+      "loss": 0.07090835273265839,
+      "loss_ce": 1.6013153071980923e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 120502728,
+      "step": 1335
+    },
+    {
+      "epoch": 6.142528735632184,
+      "grad_norm": 6.003633344367404,
+      "learning_rate": 5e-06,
+      "loss": 0.093,
+      "num_input_tokens_seen": 120592968,
+      "step": 1336
+    },
+    {
+      "epoch": 6.142528735632184,
+      "loss": 0.08331505209207535,
+      "loss_ce": 2.065743046841817e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 120592968,
+      "step": 1336
+    },
+    {
+      "epoch": 6.147126436781609,
+      "grad_norm": 16.256143484292767,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 120683224,
+      "step": 1337
+    },
+    {
+      "epoch": 6.147126436781609,
+      "loss": 0.05023811012506485,
+      "loss_ce": 6.1769642343278974e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 120683224,
+      "step": 1337
+    },
+    {
+      "epoch": 6.151724137931034,
+      "grad_norm": 18.84221449609671,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 120773604,
+      "step": 1338
+    },
+    {
+      "epoch": 6.151724137931034,
+      "loss": 0.1148335188627243,
+      "loss_ce": 1.1128912774438504e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 120773604,
+      "step": 1338
+    },
+    {
+      "epoch": 6.1563218390804595,
+      "grad_norm": 3.6340664937584357,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 120863972,
+      "step": 1339
+    },
+    {
+      "epoch": 6.1563218390804595,
+      "loss": 0.10637908428907394,
+      "loss_ce": 2.5327462935820222e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 120863972,
+      "step": 1339
+    },
+    {
+      "epoch": 6.160919540229885,
+      "grad_norm": 12.35950915110001,
+      "learning_rate": 5e-06,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 120954292,
+      "step": 1340
+    },
+    {
+      "epoch": 6.160919540229885,
+      "loss": 0.050802938640117645,
+      "loss_ce": 2.1687461412511766e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 120954292,
+      "step": 1340
+    },
+    {
+      "epoch": 6.165517241379311,
+      "grad_norm": 13.05285924563968,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 121044680,
+      "step": 1341
+    },
+    {
+      "epoch": 6.165517241379311,
+      "loss": 0.09323087334632874,
+      "loss_ce": 1.4931548321328592e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 121044680,
+      "step": 1341
+    },
+    {
+      "epoch": 6.170114942528736,
+      "grad_norm": 3.75282762212465,
+      "learning_rate": 5e-06,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 121135088,
+      "step": 1342
+    },
+    {
+      "epoch": 6.170114942528736,
+      "loss": 0.06280642002820969,
+      "loss_ce": 3.1762905564391986e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 121135088,
+      "step": 1342
+    },
+    {
+      "epoch": 6.174712643678161,
+      "grad_norm": 9.367037578077188,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 121225428,
+      "step": 1343
+    },
+    {
+      "epoch": 6.174712643678161,
+      "loss": 0.1399770975112915,
+      "loss_ce": 8.234508641180582e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 121225428,
+      "step": 1343
+    },
+    {
+      "epoch": 6.179310344827586,
+      "grad_norm": 19.171060340321993,
+      "learning_rate": 5e-06,
+      "loss": 0.087,
+      "num_input_tokens_seen": 121315824,
+      "step": 1344
+    },
+    {
+      "epoch": 6.179310344827586,
+      "loss": 0.06587206572294235,
+      "loss_ce": 1.513374263595324e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 121315824,
+      "step": 1344
+    },
+    {
+      "epoch": 6.183908045977011,
+      "grad_norm": 4.876284313063547,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 121406176,
+      "step": 1345
+    },
+    {
+      "epoch": 6.183908045977011,
+      "loss": 0.06390294432640076,
+      "loss_ce": 6.762678367522312e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 121406176,
+      "step": 1345
+    },
+    {
+      "epoch": 6.188505747126436,
+      "grad_norm": 16.38884750049181,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 121496544,
+      "step": 1346
+    },
+    {
+      "epoch": 6.188505747126436,
+      "loss": 0.1316826045513153,
+      "loss_ce": 2.9771730623906478e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 121496544,
+      "step": 1346
+    },
+    {
+      "epoch": 6.1931034482758625,
+      "grad_norm": 11.544406525213137,
+      "learning_rate": 5e-06,
+      "loss": 0.0773,
+      "num_input_tokens_seen": 121587024,
+      "step": 1347
+    },
+    {
+      "epoch": 6.1931034482758625,
+      "loss": 0.08613596856594086,
+      "loss_ce": 1.535906267235987e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 121587024,
+      "step": 1347
+    },
+    {
+      "epoch": 6.197701149425288,
+      "grad_norm": 2.445337746033652,
+      "learning_rate": 5e-06,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 121677320,
+      "step": 1348
+    },
+    {
+      "epoch": 6.197701149425288,
+      "loss": 0.04452810809016228,
+      "loss_ce": 2.960706069643493e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 121677320,
+      "step": 1348
+    },
+    {
+      "epoch": 6.202298850574713,
+      "grad_norm": 6.030279032225266,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 121767660,
+      "step": 1349
+    },
+    {
+      "epoch": 6.202298850574713,
+      "loss": 0.10142619907855988,
+      "loss_ce": 1.0309655635865056e-06,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 121767660,
+      "step": 1349
+    },
+    {
+      "epoch": 6.206896551724138,
+      "grad_norm": 13.218267366174958,
+      "learning_rate": 5e-06,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 121857952,
+      "step": 1350
+    },
+    {
+      "epoch": 6.206896551724138,
+      "loss": 0.08160283416509628,
+      "loss_ce": 1.4093273421167396e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 121857952,
+      "step": 1350
+    },
+    {
+      "epoch": 6.211494252873563,
+      "grad_norm": 7.3910687621686115,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 121948348,
+      "step": 1351
+    },
+    {
+      "epoch": 6.211494252873563,
+      "loss": 0.1298474818468094,
+      "loss_ce": 4.096345946891233e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 121948348,
+      "step": 1351
+    },
+    {
+      "epoch": 6.216091954022988,
+      "grad_norm": 3.477775072702562,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 122038552,
+      "step": 1352
+    },
+    {
+      "epoch": 6.216091954022988,
+      "loss": 0.06851699203252792,
+      "loss_ce": 5.026361122872913e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 122038552,
+      "step": 1352
+    },
+    {
+      "epoch": 6.220689655172414,
+      "grad_norm": 4.659650631220517,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 122128880,
+      "step": 1353
+    },
+    {
+      "epoch": 6.220689655172414,
+      "loss": 0.09352925419807434,
+      "loss_ce": 2.3392831280943938e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 122128880,
+      "step": 1353
+    },
+    {
+      "epoch": 6.225287356321839,
+      "grad_norm": 7.501436100973512,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 122219280,
+      "step": 1354
+    },
+    {
+      "epoch": 6.225287356321839,
+      "loss": 0.11285711824893951,
+      "loss_ce": 1.0735830983321648e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 122219280,
+      "step": 1354
+    },
+    {
+      "epoch": 6.2298850574712645,
+      "grad_norm": 7.168147699957736,
+      "learning_rate": 5e-06,
+      "loss": 0.0745,
+      "num_input_tokens_seen": 122308872,
+      "step": 1355
+    },
+    {
+      "epoch": 6.2298850574712645,
+      "loss": 0.07650821655988693,
+      "loss_ce": 6.474406859524606e-07,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 122308872,
+      "step": 1355
+    },
+    {
+      "epoch": 6.23448275862069,
+      "grad_norm": 2.803107500761448,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 122399160,
+      "step": 1356
+    },
+    {
+      "epoch": 6.23448275862069,
+      "loss": 0.17111368477344513,
+      "loss_ce": 1.622533659428882e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 122399160,
+      "step": 1356
+    },
+    {
+      "epoch": 6.239080459770115,
+      "grad_norm": 1.8395032972537242,
+      "learning_rate": 5e-06,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 122489652,
+      "step": 1357
+    },
+    {
+      "epoch": 6.239080459770115,
+      "loss": 0.07030150294303894,
+      "loss_ce": 4.265766619937494e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 122489652,
+      "step": 1357
+    },
+    {
+      "epoch": 6.24367816091954,
+      "grad_norm": 1.8235275088698975,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 122580024,
+      "step": 1358
+    },
+    {
+      "epoch": 6.24367816091954,
+      "loss": 0.05960645154118538,
+      "loss_ce": 5.619423973257653e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 122580024,
+      "step": 1358
+    },
+    {
+      "epoch": 6.248275862068965,
+      "grad_norm": 2.3308796814869392,
+      "learning_rate": 5e-06,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 122670360,
+      "step": 1359
+    },
+    {
+      "epoch": 6.248275862068965,
+      "loss": 0.08280164748430252,
+      "loss_ce": 7.4597710408852436e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 122670360,
+      "step": 1359
+    },
+    {
+      "epoch": 6.252873563218391,
+      "grad_norm": 9.546289417744838,
+      "learning_rate": 5e-06,
+      "loss": 0.0745,
+      "num_input_tokens_seen": 122760712,
+      "step": 1360
+    },
+    {
+      "epoch": 6.252873563218391,
+      "loss": 0.06621609628200531,
+      "loss_ce": 8.209115549107082e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 122760712,
+      "step": 1360
+    },
+    {
+      "epoch": 6.257471264367816,
+      "grad_norm": 21.086995926043166,
+      "learning_rate": 5e-06,
+      "loss": 0.083,
+      "num_input_tokens_seen": 122850908,
+      "step": 1361
+    },
+    {
+      "epoch": 6.257471264367816,
+      "loss": 0.08541100472211838,
+      "loss_ce": 7.5566276791505516e-06,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 122850908,
+      "step": 1361
+    },
+    {
+      "epoch": 6.2620689655172415,
+      "grad_norm": 9.01280263169444,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 122941356,
+      "step": 1362
+    },
+    {
+      "epoch": 6.2620689655172415,
+      "loss": 0.1276589035987854,
+      "loss_ce": 3.869728971039876e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 122941356,
+      "step": 1362
+    },
+    {
+      "epoch": 6.266666666666667,
+      "grad_norm": 20.135580773469314,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 123031748,
+      "step": 1363
+    },
+    {
+      "epoch": 6.266666666666667,
+      "loss": 0.10605251789093018,
+      "loss_ce": 3.935991117032245e-06,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 123031748,
+      "step": 1363
+    },
+    {
+      "epoch": 6.271264367816092,
+      "grad_norm": 13.434332845560562,
+      "learning_rate": 5e-06,
+      "loss": 0.0927,
+      "num_input_tokens_seen": 123121280,
+      "step": 1364
+    },
+    {
+      "epoch": 6.271264367816092,
+      "loss": 0.07541698217391968,
+      "loss_ce": 8.042437002586666e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 123121280,
+      "step": 1364
+    },
+    {
+      "epoch": 6.275862068965517,
+      "grad_norm": 4.409333354948716,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 123211620,
+      "step": 1365
+    },
+    {
+      "epoch": 6.275862068965517,
+      "loss": 0.10695058107376099,
+      "loss_ce": 1.728949655444012e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 123211620,
+      "step": 1365
+    },
+    {
+      "epoch": 6.280459770114943,
+      "grad_norm": 3.154737625505015,
+      "learning_rate": 5e-06,
+      "loss": 0.075,
+      "num_input_tokens_seen": 123301896,
+      "step": 1366
+    },
+    {
+      "epoch": 6.280459770114943,
+      "loss": 0.07635073363780975,
+      "loss_ce": 1.1014501978934277e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 123301896,
+      "step": 1366
+    },
+    {
+      "epoch": 6.285057471264368,
+      "grad_norm": 2.211275058736632,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 123392472,
+      "step": 1367
+    },
+    {
+      "epoch": 6.285057471264368,
+      "loss": 0.07074186205863953,
+      "loss_ce": 1.7377531548845582e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 123392472,
+      "step": 1367
+    },
+    {
+      "epoch": 6.289655172413793,
+      "grad_norm": 10.219890906499858,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 123482900,
+      "step": 1368
+    },
+    {
+      "epoch": 6.289655172413793,
+      "loss": 0.07744231820106506,
+      "loss_ce": 3.968204964621691e-06,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 123482900,
+      "step": 1368
+    },
+    {
+      "epoch": 6.294252873563218,
+      "grad_norm": 5.327344530549374,
+      "learning_rate": 5e-06,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 123573300,
+      "step": 1369
+    },
+    {
+      "epoch": 6.294252873563218,
+      "loss": 0.09067036956548691,
+      "loss_ce": 9.419472189620137e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 123573300,
+      "step": 1369
+    },
+    {
+      "epoch": 6.2988505747126435,
+      "grad_norm": 9.772421156940164,
+      "learning_rate": 5e-06,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 123663696,
+      "step": 1370
+    },
+    {
+      "epoch": 6.2988505747126435,
+      "loss": 0.0640704333782196,
+      "loss_ce": 2.9297052606125362e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 123663696,
+      "step": 1370
+    },
+    {
+      "epoch": 6.303448275862069,
+      "grad_norm": 18.114247114460188,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 123754076,
+      "step": 1371
+    },
+    {
+      "epoch": 6.303448275862069,
+      "loss": 0.07899777591228485,
+      "loss_ce": 3.0194664759619627e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 123754076,
+      "step": 1371
+    },
+    {
+      "epoch": 6.308045977011494,
+      "grad_norm": 3.332469449720361,
+      "learning_rate": 5e-06,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 123844508,
+      "step": 1372
+    },
+    {
+      "epoch": 6.308045977011494,
+      "loss": 0.06758559495210648,
+      "loss_ce": 4.417397576617077e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 123844508,
+      "step": 1372
+    },
+    {
+      "epoch": 6.31264367816092,
+      "grad_norm": 6.634405744544336,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 123934812,
+      "step": 1373
+    },
+    {
+      "epoch": 6.31264367816092,
+      "loss": 0.08992957323789597,
+      "loss_ce": 9.527670954412315e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 123934812,
+      "step": 1373
+    },
+    {
+      "epoch": 6.317241379310345,
+      "grad_norm": 10.333106801799953,
+      "learning_rate": 5e-06,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 124025200,
+      "step": 1374
+    },
+    {
+      "epoch": 6.317241379310345,
+      "loss": 0.04593438282608986,
+      "loss_ce": 2.068851244985126e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 124025200,
+      "step": 1374
+    },
+    {
+      "epoch": 6.32183908045977,
+      "grad_norm": 6.513003814868987,
+      "learning_rate": 5e-06,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 124115436,
+      "step": 1375
+    },
+    {
+      "epoch": 6.32183908045977,
+      "loss": 0.07307623326778412,
+      "loss_ce": 1.8943998156828457e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 124115436,
+      "step": 1375
+    },
+    {
+      "epoch": 6.326436781609195,
+      "grad_norm": 7.020049530325385,
+      "learning_rate": 5e-06,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 124205816,
+      "step": 1376
+    },
+    {
+      "epoch": 6.326436781609195,
+      "loss": 0.062137503176927567,
+      "loss_ce": 3.71252508557518e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 124205816,
+      "step": 1376
+    },
+    {
+      "epoch": 6.3310344827586205,
+      "grad_norm": 6.7984756030733005,
+      "learning_rate": 5e-06,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 124296160,
+      "step": 1377
+    },
+    {
+      "epoch": 6.3310344827586205,
+      "loss": 0.09830499440431595,
+      "loss_ce": 6.89133012201637e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 124296160,
+      "step": 1377
+    },
+    {
+      "epoch": 6.335632183908046,
+      "grad_norm": 9.550154500204966,
+      "learning_rate": 5e-06,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 124386652,
+      "step": 1378
+    },
+    {
+      "epoch": 6.335632183908046,
+      "loss": 0.032549649477005005,
+      "loss_ce": 2.6519669518165756e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 124386652,
+      "step": 1378
+    },
+    {
+      "epoch": 6.340229885057472,
+      "grad_norm": 15.868279771186451,
+      "learning_rate": 5e-06,
+      "loss": 0.1227,
+      "num_input_tokens_seen": 124477136,
+      "step": 1379
+    },
+    {
+      "epoch": 6.340229885057472,
+      "loss": 0.10919815301895142,
+      "loss_ce": 6.263924660743214e-06,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 124477136,
+      "step": 1379
+    },
+    {
+      "epoch": 6.344827586206897,
+      "grad_norm": 9.149745216189178,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 124567584,
+      "step": 1380
+    },
+    {
+      "epoch": 6.344827586206897,
+      "loss": 0.043360911309719086,
+      "loss_ce": 1.0688753718568478e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 124567584,
+      "step": 1380
+    },
+    {
+      "epoch": 6.349425287356322,
+      "grad_norm": 6.148379301187082,
+      "learning_rate": 5e-06,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 124657880,
+      "step": 1381
+    },
+    {
+      "epoch": 6.349425287356322,
+      "loss": 0.07773515582084656,
+      "loss_ce": 6.885257789690513e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 124657880,
+      "step": 1381
+    },
+    {
+      "epoch": 6.354022988505747,
+      "grad_norm": 6.06320096249193,
+      "learning_rate": 5e-06,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 124748200,
+      "step": 1382
+    },
+    {
+      "epoch": 6.354022988505747,
+      "loss": 0.08277206867933273,
+      "loss_ce": 8.399106263823342e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 124748200,
+      "step": 1382
+    },
+    {
+      "epoch": 6.358620689655172,
+      "grad_norm": 6.051287795431972,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 124838752,
+      "step": 1383
+    },
+    {
+      "epoch": 6.358620689655172,
+      "loss": 0.056419774889945984,
+      "loss_ce": 8.03236889623804e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 124838752,
+      "step": 1383
+    },
+    {
+      "epoch": 6.363218390804597,
+      "grad_norm": 13.7747801267887,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 124929112,
+      "step": 1384
+    },
+    {
+      "epoch": 6.363218390804597,
+      "loss": 0.07443118095397949,
+      "loss_ce": 7.509520219173282e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 124929112,
+      "step": 1384
+    },
+    {
+      "epoch": 6.3678160919540225,
+      "grad_norm": 2.09029899827099,
+      "learning_rate": 5e-06,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 125019604,
+      "step": 1385
+    },
+    {
+      "epoch": 6.3678160919540225,
+      "loss": 0.08999022096395493,
+      "loss_ce": 9.14329484658083e-06,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 125019604,
+      "step": 1385
+    },
+    {
+      "epoch": 6.372413793103449,
+      "grad_norm": 4.966748140294823,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 125109988,
+      "step": 1386
+    },
+    {
+      "epoch": 6.372413793103449,
+      "loss": 0.08931561559438705,
+      "loss_ce": 5.922461241425481e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 125109988,
+      "step": 1386
+    },
+    {
+      "epoch": 6.377011494252874,
+      "grad_norm": 1.8678820550766893,
+      "learning_rate": 5e-06,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 125200440,
+      "step": 1387
+    },
+    {
+      "epoch": 6.377011494252874,
+      "loss": 0.0770668312907219,
+      "loss_ce": 5.572132067754865e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 125200440,
+      "step": 1387
+    },
+    {
+      "epoch": 6.381609195402299,
+      "grad_norm": 16.6653828543424,
+      "learning_rate": 5e-06,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 125290784,
+      "step": 1388
+    },
+    {
+      "epoch": 6.381609195402299,
+      "loss": 0.09468194842338562,
+      "loss_ce": 1.167914092548017e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 125290784,
+      "step": 1388
+    },
+    {
+      "epoch": 6.386206896551724,
+      "grad_norm": 9.34884327902768,
+      "learning_rate": 5e-06,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 125381168,
+      "step": 1389
+    },
+    {
+      "epoch": 6.386206896551724,
+      "loss": 0.060101285576820374,
+      "loss_ce": 2.7430016416474245e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 125381168,
+      "step": 1389
+    },
+    {
+      "epoch": 6.390804597701149,
+      "grad_norm": 4.121494614520851,
+      "learning_rate": 5e-06,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 125471516,
+      "step": 1390
+    },
+    {
+      "epoch": 6.390804597701149,
+      "loss": 0.09581901133060455,
+      "loss_ce": 2.433588269923348e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 125471516,
+      "step": 1390
+    },
+    {
+      "epoch": 6.395402298850574,
+      "grad_norm": 14.675052033868381,
+      "learning_rate": 5e-06,
+      "loss": 0.1022,
+      "num_input_tokens_seen": 125561880,
+      "step": 1391
+    },
+    {
+      "epoch": 6.395402298850574,
+      "loss": 0.1025380939245224,
+      "loss_ce": 4.480792267713696e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 125561880,
+      "step": 1391
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 7.391101398441527,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 125651460,
+      "step": 1392
+    },
+    {
+      "epoch": 6.4,
+      "loss": 0.1120477020740509,
+      "loss_ce": 2.4182809283956885e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 125651460,
+      "step": 1392
+    },
+    {
+      "epoch": 6.4045977011494255,
+      "grad_norm": 4.598424061793132,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 125741836,
+      "step": 1393
+    },
+    {
+      "epoch": 6.4045977011494255,
+      "loss": 0.08984687924385071,
+      "loss_ce": 3.1254223813448334e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 125741836,
+      "step": 1393
+    },
+    {
+      "epoch": 6.409195402298851,
+      "grad_norm": 10.543183735469755,
+      "learning_rate": 5e-06,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 125832188,
+      "step": 1394
+    },
+    {
+      "epoch": 6.409195402298851,
+      "loss": 0.06207028031349182,
+      "loss_ce": 1.2784106729668565e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 125832188,
+      "step": 1394
+    },
+    {
+      "epoch": 6.413793103448276,
+      "grad_norm": 11.282419963710336,
+      "learning_rate": 5e-06,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 125922616,
+      "step": 1395
+    },
+    {
+      "epoch": 6.413793103448276,
+      "loss": 0.09065688401460648,
+      "loss_ce": 4.416605861479184e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 125922616,
+      "step": 1395
+    },
+    {
+      "epoch": 6.418390804597701,
+      "grad_norm": 6.53067476157625,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 126012968,
+      "step": 1396
+    },
+    {
+      "epoch": 6.418390804597701,
+      "loss": 0.11000474542379379,
+      "loss_ce": 4.139963493798859e-06,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 126012968,
+      "step": 1396
+    },
+    {
+      "epoch": 6.422988505747126,
+      "grad_norm": 5.213448872545618,
+      "learning_rate": 5e-06,
+      "loss": 0.0968,
+      "num_input_tokens_seen": 126102532,
+      "step": 1397
+    },
+    {
+      "epoch": 6.422988505747126,
+      "loss": 0.10766983777284622,
+      "loss_ce": 3.8197222238522954e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 126102532,
+      "step": 1397
+    },
+    {
+      "epoch": 6.427586206896552,
+      "grad_norm": 19.99194183396297,
+      "learning_rate": 5e-06,
+      "loss": 0.069,
+      "num_input_tokens_seen": 126192844,
+      "step": 1398
+    },
+    {
+      "epoch": 6.427586206896552,
+      "loss": 0.07072758674621582,
+      "loss_ce": 3.0986652745923493e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 126192844,
+      "step": 1398
+    },
+    {
+      "epoch": 6.432183908045977,
+      "grad_norm": 2.972490756323674,
+      "learning_rate": 5e-06,
+      "loss": 0.0783,
+      "num_input_tokens_seen": 126282472,
+      "step": 1399
+    },
+    {
+      "epoch": 6.432183908045977,
+      "loss": 0.0742267370223999,
+      "loss_ce": 5.376310582505539e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 126282472,
+      "step": 1399
+    },
+    {
+      "epoch": 6.436781609195402,
+      "grad_norm": 3.093510331762982,
+      "learning_rate": 5e-06,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 126372972,
+      "step": 1400
+    },
+    {
+      "epoch": 6.436781609195402,
+      "loss": 0.06427352130413055,
+      "loss_ce": 1.876192618510686e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 126372972,
+      "step": 1400
+    },
+    {
+      "epoch": 6.441379310344828,
+      "grad_norm": 8.92440544924096,
+      "learning_rate": 5e-06,
+      "loss": 0.092,
+      "num_input_tokens_seen": 126463292,
+      "step": 1401
+    },
+    {
+      "epoch": 6.441379310344828,
+      "loss": 0.07695218920707703,
+      "loss_ce": 2.11300380215107e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 126463292,
+      "step": 1401
+    },
+    {
+      "epoch": 6.445977011494253,
+      "grad_norm": 5.049167412556098,
+      "learning_rate": 5e-06,
+      "loss": 0.096,
+      "num_input_tokens_seen": 126553568,
+      "step": 1402
+    },
+    {
+      "epoch": 6.445977011494253,
+      "loss": 0.12945988774299622,
+      "loss_ce": 4.316266313253436e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 126553568,
+      "step": 1402
+    },
+    {
+      "epoch": 6.450574712643678,
+      "grad_norm": 7.016975178744314,
+      "learning_rate": 5e-06,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 126643956,
+      "step": 1403
+    },
+    {
+      "epoch": 6.450574712643678,
+      "loss": 0.0685601681470871,
+      "loss_ce": 2.4308599222422345e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 126643956,
+      "step": 1403
+    },
+    {
+      "epoch": 6.455172413793104,
+      "grad_norm": 4.991828966505337,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 126734232,
+      "step": 1404
+    },
+    {
+      "epoch": 6.455172413793104,
+      "loss": 0.13235914707183838,
+      "loss_ce": 4.41625525127165e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 126734232,
+      "step": 1404
+    },
+    {
+      "epoch": 6.459770114942529,
+      "grad_norm": 4.41915443276764,
+      "learning_rate": 5e-06,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 126824688,
+      "step": 1405
+    },
+    {
+      "epoch": 6.459770114942529,
+      "loss": 0.05354461073875427,
+      "loss_ce": 1.52462337155157e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 126824688,
+      "step": 1405
+    },
+    {
+      "epoch": 6.464367816091954,
+      "grad_norm": 27.333175354486055,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 126914992,
+      "step": 1406
+    },
+    {
+      "epoch": 6.464367816091954,
+      "loss": 0.08278225362300873,
+      "loss_ce": 3.324659246572992e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 126914992,
+      "step": 1406
+    },
+    {
+      "epoch": 6.468965517241379,
+      "grad_norm": 15.2120482775101,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 127005376,
+      "step": 1407
+    },
+    {
+      "epoch": 6.468965517241379,
+      "loss": 0.07776758074760437,
+      "loss_ce": 8.793870620138478e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 127005376,
+      "step": 1407
+    },
+    {
+      "epoch": 6.4735632183908045,
+      "grad_norm": 5.4209480783540585,
+      "learning_rate": 5e-06,
+      "loss": 0.0926,
+      "num_input_tokens_seen": 127094892,
+      "step": 1408
+    },
+    {
+      "epoch": 6.4735632183908045,
+      "loss": 0.11467273533344269,
+      "loss_ce": 2.9248494683997706e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 127094892,
+      "step": 1408
+    },
+    {
+      "epoch": 6.47816091954023,
+      "grad_norm": 6.961408738662496,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 127184420,
+      "step": 1409
+    },
+    {
+      "epoch": 6.47816091954023,
+      "loss": 0.08804384618997574,
+      "loss_ce": 6.312624805104861e-07,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 127184420,
+      "step": 1409
+    },
+    {
+      "epoch": 6.482758620689655,
+      "grad_norm": 5.090185336665533,
+      "learning_rate": 5e-06,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 127274716,
+      "step": 1410
+    },
+    {
+      "epoch": 6.482758620689655,
+      "loss": 0.09357266128063202,
+      "loss_ce": 2.103199039993342e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 127274716,
+      "step": 1410
+    },
+    {
+      "epoch": 6.487356321839081,
+      "grad_norm": 3.1966959428594626,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 127365044,
+      "step": 1411
+    },
+    {
+      "epoch": 6.487356321839081,
+      "loss": 0.04377942159771919,
+      "loss_ce": 1.9543890630302485e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 127365044,
+      "step": 1411
+    },
+    {
+      "epoch": 6.491954022988506,
+      "grad_norm": 8.572838699176348,
+      "learning_rate": 5e-06,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 127455440,
+      "step": 1412
+    },
+    {
+      "epoch": 6.491954022988506,
+      "loss": 0.09225938469171524,
+      "loss_ce": 4.743013505503768e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 127455440,
+      "step": 1412
+    },
+    {
+      "epoch": 6.496551724137931,
+      "grad_norm": 13.582941616327785,
+      "learning_rate": 5e-06,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 127545812,
+      "step": 1413
+    },
+    {
+      "epoch": 6.496551724137931,
+      "loss": 0.044346556067466736,
+      "loss_ce": 4.514195552474121e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 127545812,
+      "step": 1413
+    },
+    {
+      "epoch": 6.501149425287356,
+      "grad_norm": 8.548817159191486,
+      "learning_rate": 5e-06,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 127636132,
+      "step": 1414
+    },
+    {
+      "epoch": 6.501149425287356,
+      "loss": 0.10972800850868225,
+      "loss_ce": 2.04993671104603e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 127636132,
+      "step": 1414
+    },
+    {
+      "epoch": 6.505747126436781,
+      "grad_norm": 5.849875972289943,
+      "learning_rate": 5e-06,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 127724944,
+      "step": 1415
+    },
+    {
+      "epoch": 6.505747126436781,
+      "loss": 0.04938147962093353,
+      "loss_ce": 4.03994590669754e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 127724944,
+      "step": 1415
+    },
+    {
+      "epoch": 6.510344827586207,
+      "grad_norm": 2.6655195062146237,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 127815308,
+      "step": 1416
+    },
+    {
+      "epoch": 6.510344827586207,
+      "loss": 0.17061986029148102,
+      "loss_ce": 5.711633275495842e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 127815308,
+      "step": 1416
+    },
+    {
+      "epoch": 6.514942528735633,
+      "grad_norm": 11.77741222539578,
+      "learning_rate": 5e-06,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 127905652,
+      "step": 1417
+    },
+    {
+      "epoch": 6.514942528735633,
+      "loss": 0.06571365892887115,
+      "loss_ce": 9.316358045907691e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 127905652,
+      "step": 1417
+    },
+    {
+      "epoch": 6.519540229885058,
+      "grad_norm": 7.004727703721243,
+      "learning_rate": 5e-06,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 127996084,
+      "step": 1418
+    },
+    {
+      "epoch": 6.519540229885058,
+      "loss": 0.0499795600771904,
+      "loss_ce": 7.023472790024243e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 127996084,
+      "step": 1418
+    },
+    {
+      "epoch": 6.524137931034483,
+      "grad_norm": 11.550186444604932,
+      "learning_rate": 5e-06,
+      "loss": 0.0998,
+      "num_input_tokens_seen": 128086452,
+      "step": 1419
+    },
+    {
+      "epoch": 6.524137931034483,
+      "loss": 0.1050664559006691,
+      "loss_ce": 2.4955639673862606e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 128086452,
+      "step": 1419
+    },
+    {
+      "epoch": 6.528735632183908,
+      "grad_norm": 8.09948814876527,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 128176808,
+      "step": 1420
+    },
+    {
+      "epoch": 6.528735632183908,
+      "loss": 0.09395314753055573,
+      "loss_ce": 4.7787834773771465e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 128176808,
+      "step": 1420
+    },
+    {
+      "epoch": 6.533333333333333,
+      "grad_norm": 23.81012046657335,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 128267156,
+      "step": 1421
+    },
+    {
+      "epoch": 6.533333333333333,
+      "loss": 0.1023278459906578,
+      "loss_ce": 2.4068172024271917e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 128267156,
+      "step": 1421
+    },
+    {
+      "epoch": 6.537931034482758,
+      "grad_norm": 4.4144379527625315,
+      "learning_rate": 5e-06,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 128357584,
+      "step": 1422
+    },
+    {
+      "epoch": 6.537931034482758,
+      "loss": 0.0876457542181015,
+      "loss_ce": 2.978431984956842e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 128357584,
+      "step": 1422
+    },
+    {
+      "epoch": 6.5425287356321835,
+      "grad_norm": 3.0109543272903045,
+      "learning_rate": 5e-06,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 128447236,
+      "step": 1423
+    },
+    {
+      "epoch": 6.5425287356321835,
+      "loss": 0.06799499690532684,
+      "loss_ce": 6.287074938882142e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 128447236,
+      "step": 1423
+    },
+    {
+      "epoch": 6.5471264367816095,
+      "grad_norm": 3.6963053002280675,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 128537560,
+      "step": 1424
+    },
+    {
+      "epoch": 6.5471264367816095,
+      "loss": 0.06803764402866364,
+      "loss_ce": 1.3963182937004603e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 128537560,
+      "step": 1424
+    },
+    {
+      "epoch": 6.551724137931035,
+      "grad_norm": 3.0893766848228514,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 128627956,
+      "step": 1425
+    },
+    {
+      "epoch": 6.551724137931035,
+      "loss": 0.0690060555934906,
+      "loss_ce": 2.1069548893137835e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 128627956,
+      "step": 1425
+    },
+    {
+      "epoch": 6.55632183908046,
+      "grad_norm": 2.559876796681711,
+      "learning_rate": 5e-06,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 128718388,
+      "step": 1426
+    },
+    {
+      "epoch": 6.55632183908046,
+      "loss": 0.08054003119468689,
+      "loss_ce": 4.145399088884005e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 128718388,
+      "step": 1426
+    },
+    {
+      "epoch": 6.560919540229885,
+      "grad_norm": 13.711469163719162,
+      "learning_rate": 5e-06,
+      "loss": 0.1533,
+      "num_input_tokens_seen": 128808732,
+      "step": 1427
+    },
+    {
+      "epoch": 6.560919540229885,
+      "loss": 0.1875627636909485,
+      "loss_ce": 3.2246465707430616e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 128808732,
+      "step": 1427
+    },
+    {
+      "epoch": 6.56551724137931,
+      "grad_norm": 8.158037791630148,
+      "learning_rate": 5e-06,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 128898284,
+      "step": 1428
+    },
+    {
+      "epoch": 6.56551724137931,
+      "loss": 0.08450591564178467,
+      "loss_ce": 2.7392322863306617e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 128898284,
+      "step": 1428
+    },
+    {
+      "epoch": 6.570114942528735,
+      "grad_norm": 16.342416907286246,
+      "learning_rate": 5e-06,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 128988672,
+      "step": 1429
+    },
+    {
+      "epoch": 6.570114942528735,
+      "loss": 0.06996608525514603,
+      "loss_ce": 4.540259851637529e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 128988672,
+      "step": 1429
+    },
+    {
+      "epoch": 6.574712643678161,
+      "grad_norm": 2.9446395510222763,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 129079076,
+      "step": 1430
+    },
+    {
+      "epoch": 6.574712643678161,
+      "loss": 0.07431189715862274,
+      "loss_ce": 6.262610259000212e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 129079076,
+      "step": 1430
+    },
+    {
+      "epoch": 6.5793103448275865,
+      "grad_norm": 5.022862929220284,
+      "learning_rate": 5e-06,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 129169404,
+      "step": 1431
+    },
+    {
+      "epoch": 6.5793103448275865,
+      "loss": 0.042931437492370605,
+      "loss_ce": 8.465913197142072e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 129169404,
+      "step": 1431
+    },
+    {
+      "epoch": 6.583908045977012,
+      "grad_norm": 5.866439214352878,
+      "learning_rate": 5e-06,
+      "loss": 0.0915,
+      "num_input_tokens_seen": 129258248,
+      "step": 1432
+    },
+    {
+      "epoch": 6.583908045977012,
+      "loss": 0.11402902007102966,
+      "loss_ce": 1.534522198198829e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 129258248,
+      "step": 1432
+    },
+    {
+      "epoch": 6.588505747126437,
+      "grad_norm": 2.4783513803938586,
+      "learning_rate": 5e-06,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 129348620,
+      "step": 1433
+    },
+    {
+      "epoch": 6.588505747126437,
+      "loss": 0.0649535059928894,
+      "loss_ce": 4.467387498152675e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 129348620,
+      "step": 1433
+    },
+    {
+      "epoch": 6.593103448275862,
+      "grad_norm": 44.36894368647196,
+      "learning_rate": 5e-06,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 129439104,
+      "step": 1434
+    },
+    {
+      "epoch": 6.593103448275862,
+      "loss": 0.06779413670301437,
+      "loss_ce": 6.966247383388691e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 129439104,
+      "step": 1434
+    },
+    {
+      "epoch": 6.597701149425287,
+      "grad_norm": 6.9878597014692225,
+      "learning_rate": 5e-06,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 129529500,
+      "step": 1435
+    },
+    {
+      "epoch": 6.597701149425287,
+      "loss": 0.06153077632188797,
+      "loss_ce": 7.337920578720514e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 129529500,
+      "step": 1435
+    },
+    {
+      "epoch": 6.602298850574712,
+      "grad_norm": 14.338290986997505,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 129619932,
+      "step": 1436
+    },
+    {
+      "epoch": 6.602298850574712,
+      "loss": 0.09894341975450516,
+      "loss_ce": 3.59489640686661e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 129619932,
+      "step": 1436
+    },
+    {
+      "epoch": 6.606896551724138,
+      "grad_norm": 8.118572650309352,
+      "learning_rate": 5e-06,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 129710368,
+      "step": 1437
+    },
+    {
+      "epoch": 6.606896551724138,
+      "loss": 0.05395686253905296,
+      "loss_ce": 1.782671802175173e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 129710368,
+      "step": 1437
+    },
+    {
+      "epoch": 6.611494252873563,
+      "grad_norm": 13.176137538777503,
+      "learning_rate": 5e-06,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 129800728,
+      "step": 1438
+    },
+    {
+      "epoch": 6.611494252873563,
+      "loss": 0.10741189867258072,
+      "loss_ce": 2.0546713130897842e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 129800728,
+      "step": 1438
+    },
+    {
+      "epoch": 6.6160919540229886,
+      "grad_norm": 2.9757478665850132,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 129891160,
+      "step": 1439
+    },
+    {
+      "epoch": 6.6160919540229886,
+      "loss": 0.11056976020336151,
+      "loss_ce": 4.582754172588466e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 129891160,
+      "step": 1439
+    },
+    {
+      "epoch": 6.620689655172414,
+      "grad_norm": 2.0836625866660023,
+      "learning_rate": 5e-06,
+      "loss": 0.093,
+      "num_input_tokens_seen": 129981596,
+      "step": 1440
+    },
+    {
+      "epoch": 6.620689655172414,
+      "loss": 0.07989851385354996,
+      "loss_ce": 3.400920468266122e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 129981596,
+      "step": 1440
+    },
+    {
+      "epoch": 6.625287356321839,
+      "grad_norm": 11.346287982435365,
+      "learning_rate": 5e-06,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 130070368,
+      "step": 1441
+    },
+    {
+      "epoch": 6.625287356321839,
+      "loss": 0.1127094179391861,
+      "loss_ce": 8.00615543994354e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 130070368,
+      "step": 1441
+    },
+    {
+      "epoch": 6.629885057471264,
+      "grad_norm": 11.945688660639014,
+      "learning_rate": 5e-06,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 130160760,
+      "step": 1442
+    },
+    {
+      "epoch": 6.629885057471264,
+      "loss": 0.06225637346506119,
+      "loss_ce": 1.5771218386362307e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 130160760,
+      "step": 1442
+    },
+    {
+      "epoch": 6.63448275862069,
+      "grad_norm": 5.975406754271047,
+      "learning_rate": 5e-06,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 130251268,
+      "step": 1443
+    },
+    {
+      "epoch": 6.63448275862069,
+      "loss": 0.0756722092628479,
+      "loss_ce": 1.9129147403873503e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 130251268,
+      "step": 1443
+    },
+    {
+      "epoch": 6.639080459770115,
+      "grad_norm": 4.4111769643184005,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 130341548,
+      "step": 1444
+    },
+    {
+      "epoch": 6.639080459770115,
+      "loss": 0.09213773906230927,
+      "loss_ce": 5.1713404900510795e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 130341548,
+      "step": 1444
+    },
+    {
+      "epoch": 6.64367816091954,
+      "grad_norm": 1.9533172759588555,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 130431868,
+      "step": 1445
+    },
+    {
+      "epoch": 6.64367816091954,
+      "loss": 0.03545193374156952,
+      "loss_ce": 5.765972673543729e-06,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 130431868,
+      "step": 1445
+    },
+    {
+      "epoch": 6.6482758620689655,
+      "grad_norm": 1.6643815600000895,
+      "learning_rate": 5e-06,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 130522236,
+      "step": 1446
+    },
+    {
+      "epoch": 6.6482758620689655,
+      "loss": 0.06866565346717834,
+      "loss_ce": 1.09890947896929e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 130522236,
+      "step": 1446
+    },
+    {
+      "epoch": 6.652873563218391,
+      "grad_norm": 8.868765611159223,
+      "learning_rate": 5e-06,
+      "loss": 0.0745,
+      "num_input_tokens_seen": 130612552,
+      "step": 1447
+    },
+    {
+      "epoch": 6.652873563218391,
+      "loss": 0.07027009129524231,
+      "loss_ce": 3.366212013133918e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 130612552,
+      "step": 1447
+    },
+    {
+      "epoch": 6.657471264367816,
+      "grad_norm": 8.270525030687544,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 130702956,
+      "step": 1448
+    },
+    {
+      "epoch": 6.657471264367816,
+      "loss": 0.09026643633842468,
+      "loss_ce": 1.069144764187513e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 130702956,
+      "step": 1448
+    },
+    {
+      "epoch": 6.662068965517241,
+      "grad_norm": 3.3647597113536123,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 130793328,
+      "step": 1449
+    },
+    {
+      "epoch": 6.662068965517241,
+      "loss": 0.09332942962646484,
+      "loss_ce": 6.675434633507393e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 130793328,
+      "step": 1449
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 131.97107940993814,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 130883796,
+      "step": 1450
+    },
+    {
+      "epoch": 6.666666666666667,
+      "loss": 0.057534217834472656,
+      "loss_ce": 2.383916944381781e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 130883796,
+      "step": 1450
+    },
+    {
+      "epoch": 6.671264367816092,
+      "grad_norm": 23.1470764725479,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 130974220,
+      "step": 1451
+    },
+    {
+      "epoch": 6.671264367816092,
+      "loss": 0.09790711104869843,
+      "loss_ce": 6.722117177559994e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 130974220,
+      "step": 1451
+    },
+    {
+      "epoch": 6.675862068965517,
+      "grad_norm": 10.16468406745285,
+      "learning_rate": 5e-06,
+      "loss": 0.0803,
+      "num_input_tokens_seen": 131064624,
+      "step": 1452
+    },
+    {
+      "epoch": 6.675862068965517,
+      "loss": 0.061509158462285995,
+      "loss_ce": 3.912944157491438e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 131064624,
+      "step": 1452
+    },
+    {
+      "epoch": 6.680459770114942,
+      "grad_norm": 4.911042024533096,
+      "learning_rate": 5e-06,
+      "loss": 0.0926,
+      "num_input_tokens_seen": 131154952,
+      "step": 1453
+    },
+    {
+      "epoch": 6.680459770114942,
+      "loss": 0.09288729727268219,
+      "loss_ce": 2.2301146600511856e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 131154952,
+      "step": 1453
+    },
+    {
+      "epoch": 6.685057471264368,
+      "grad_norm": 2.22095720475405,
+      "learning_rate": 5e-06,
+      "loss": 0.0656,
+      "num_input_tokens_seen": 131245408,
+      "step": 1454
+    },
+    {
+      "epoch": 6.685057471264368,
+      "loss": 0.07592545449733734,
+      "loss_ce": 5.346221314539434e-06,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 131245408,
+      "step": 1454
+    },
+    {
+      "epoch": 6.689655172413794,
+      "grad_norm": 11.830161052137443,
+      "learning_rate": 5e-06,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 131335812,
+      "step": 1455
+    },
+    {
+      "epoch": 6.689655172413794,
+      "loss": 0.06331153213977814,
+      "loss_ce": 3.3334035833831877e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 131335812,
+      "step": 1455
+    },
+    {
+      "epoch": 6.694252873563219,
+      "grad_norm": 3.0578099363890328,
+      "learning_rate": 5e-06,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 131426204,
+      "step": 1456
+    },
+    {
+      "epoch": 6.694252873563219,
+      "loss": 0.07001975178718567,
+      "loss_ce": 1.2430735296220519e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 131426204,
+      "step": 1456
+    },
+    {
+      "epoch": 6.698850574712644,
+      "grad_norm": 5.750992935778732,
+      "learning_rate": 5e-06,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 131516528,
+      "step": 1457
+    },
+    {
+      "epoch": 6.698850574712644,
+      "loss": 0.05864344537258148,
+      "loss_ce": 3.921055849787081e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 131516528,
+      "step": 1457
+    },
+    {
+      "epoch": 6.703448275862069,
+      "grad_norm": 10.893446380591408,
+      "learning_rate": 5e-06,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 131606848,
+      "step": 1458
+    },
+    {
+      "epoch": 6.703448275862069,
+      "loss": 0.11121590435504913,
+      "loss_ce": 9.852826224232558e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 131606848,
+      "step": 1458
+    },
+    {
+      "epoch": 6.708045977011494,
+      "grad_norm": 27.414007106230482,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 131697268,
+      "step": 1459
+    },
+    {
+      "epoch": 6.708045977011494,
+      "loss": 0.11673033237457275,
+      "loss_ce": 1.5854706362006254e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 131697268,
+      "step": 1459
+    },
+    {
+      "epoch": 6.712643678160919,
+      "grad_norm": 5.804991202555291,
+      "learning_rate": 5e-06,
+      "loss": 0.0791,
+      "num_input_tokens_seen": 131787672,
+      "step": 1460
+    },
+    {
+      "epoch": 6.712643678160919,
+      "loss": 0.09766550362110138,
+      "loss_ce": 9.254023098037578e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 131787672,
+      "step": 1460
+    },
+    {
+      "epoch": 6.7172413793103445,
+      "grad_norm": 3.1386099033424135,
+      "learning_rate": 5e-06,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 131878152,
+      "step": 1461
+    },
+    {
+      "epoch": 6.7172413793103445,
+      "loss": 0.13510626554489136,
+      "loss_ce": 4.960803835274419e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 131878152,
+      "step": 1461
+    },
+    {
+      "epoch": 6.72183908045977,
+      "grad_norm": 9.852881694264854,
+      "learning_rate": 5e-06,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 131968480,
+      "step": 1462
+    },
+    {
+      "epoch": 6.72183908045977,
+      "loss": 0.06898908317089081,
+      "loss_ce": 4.097350029041991e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 131968480,
+      "step": 1462
+    },
+    {
+      "epoch": 6.726436781609196,
+      "grad_norm": 2.7071823032354185,
+      "learning_rate": 5e-06,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 132059024,
+      "step": 1463
+    },
+    {
+      "epoch": 6.726436781609196,
+      "loss": 0.059280022978782654,
+      "loss_ce": 1.489080568717327e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 132059024,
+      "step": 1463
+    },
+    {
+      "epoch": 6.731034482758621,
+      "grad_norm": 7.777141629876652,
+      "learning_rate": 5e-06,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 132149448,
+      "step": 1464
+    },
+    {
+      "epoch": 6.731034482758621,
+      "loss": 0.07515916228294373,
+      "loss_ce": 2.0007742023153696e-06,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 132149448,
+      "step": 1464
+    },
+    {
+      "epoch": 6.735632183908046,
+      "grad_norm": 10.178285238489115,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 132239800,
+      "step": 1465
+    },
+    {
+      "epoch": 6.735632183908046,
+      "loss": 0.08457478135824203,
+      "loss_ce": 7.161090616136789e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 132239800,
+      "step": 1465
+    },
+    {
+      "epoch": 6.740229885057471,
+      "grad_norm": 6.325965805360473,
+      "learning_rate": 5e-06,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 132330264,
+      "step": 1466
+    },
+    {
+      "epoch": 6.740229885057471,
+      "loss": 0.0843062549829483,
+      "loss_ce": 1.4451411516347434e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 132330264,
+      "step": 1466
+    },
+    {
+      "epoch": 6.744827586206896,
+      "grad_norm": 6.120131274499989,
+      "learning_rate": 5e-06,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 132420728,
+      "step": 1467
+    },
+    {
+      "epoch": 6.744827586206896,
+      "loss": 0.07916628569364548,
+      "loss_ce": 3.6849833122687414e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 132420728,
+      "step": 1467
+    },
+    {
+      "epoch": 6.749425287356322,
+      "grad_norm": 19.37810950613584,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 132511056,
+      "step": 1468
+    },
+    {
+      "epoch": 6.749425287356322,
+      "loss": 0.06741497665643692,
+      "loss_ce": 1.6471570916110068e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 132511056,
+      "step": 1468
+    },
+    {
+      "epoch": 6.7540229885057474,
+      "grad_norm": 1.9425313486502551,
+      "learning_rate": 5e-06,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 132601416,
+      "step": 1469
+    },
+    {
+      "epoch": 6.7540229885057474,
+      "loss": 0.09817390143871307,
+      "loss_ce": 1.4117615137365647e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 132601416,
+      "step": 1469
+    },
+    {
+      "epoch": 6.758620689655173,
+      "grad_norm": 2.6250049604837926,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 132691676,
+      "step": 1470
+    },
+    {
+      "epoch": 6.758620689655173,
+      "loss": 0.08522357046604156,
+      "loss_ce": 3.2353859751310665e-06,
+      "loss_iou": 0.421875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 132691676,
+      "step": 1470
+    },
+    {
+      "epoch": 6.763218390804598,
+      "grad_norm": 2.724148470573844,
+      "learning_rate": 5e-06,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 132782044,
+      "step": 1471
+    },
+    {
+      "epoch": 6.763218390804598,
+      "loss": 0.06568065285682678,
+      "loss_ce": 6.820980161137413e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 132782044,
+      "step": 1471
+    },
+    {
+      "epoch": 6.767816091954023,
+      "grad_norm": 1.3975437384035634,
+      "learning_rate": 5e-06,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 132872492,
+      "step": 1472
+    },
+    {
+      "epoch": 6.767816091954023,
+      "loss": 0.03942399471998215,
+      "loss_ce": 1.0540796210989356e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 132872492,
+      "step": 1472
+    },
+    {
+      "epoch": 6.772413793103448,
+      "grad_norm": 3.6657098912800854,
+      "learning_rate": 5e-06,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 132962916,
+      "step": 1473
+    },
+    {
+      "epoch": 6.772413793103448,
+      "loss": 0.09079530835151672,
+      "loss_ce": 5.5107075240812264e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 132962916,
+      "step": 1473
+    },
+    {
+      "epoch": 6.777011494252873,
+      "grad_norm": 6.399156287473962,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 133053228,
+      "step": 1474
+    },
+    {
+      "epoch": 6.777011494252873,
+      "loss": 0.10846640169620514,
+      "loss_ce": 6.926821697561536e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 133053228,
+      "step": 1474
+    },
+    {
+      "epoch": 6.781609195402299,
+      "grad_norm": 13.807814737472036,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 133143584,
+      "step": 1475
+    },
+    {
+      "epoch": 6.781609195402299,
+      "loss": 0.10274454951286316,
+      "loss_ce": 9.868255438050255e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 133143584,
+      "step": 1475
+    },
+    {
+      "epoch": 6.786206896551724,
+      "grad_norm": 4.909907187391009,
+      "learning_rate": 5e-06,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 133233944,
+      "step": 1476
+    },
+    {
+      "epoch": 6.786206896551724,
+      "loss": 0.037097539752721786,
+      "loss_ce": 3.4241477351315552e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 133233944,
+      "step": 1476
+    },
+    {
+      "epoch": 6.7908045977011495,
+      "grad_norm": 10.088133006314749,
+      "learning_rate": 5e-06,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 133324236,
+      "step": 1477
+    },
+    {
+      "epoch": 6.7908045977011495,
+      "loss": 0.07998788356781006,
+      "loss_ce": 1.318014255957678e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 133324236,
+      "step": 1477
+    },
+    {
+      "epoch": 6.795402298850575,
+      "grad_norm": 23.2424717637074,
+      "learning_rate": 5e-06,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 133414528,
+      "step": 1478
+    },
+    {
+      "epoch": 6.795402298850575,
+      "loss": 0.060220953077077866,
+      "loss_ce": 9.768824384082109e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 133414528,
+      "step": 1478
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 5.280653577103845,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 133504924,
+      "step": 1479
+    },
+    {
+      "epoch": 6.8,
+      "loss": 0.09979541599750519,
+      "loss_ce": 2.9406905923679005e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 133504924,
+      "step": 1479
+    },
+    {
+      "epoch": 6.804597701149425,
+      "grad_norm": 5.817883669287115,
+      "learning_rate": 5e-06,
+      "loss": 0.078,
+      "num_input_tokens_seen": 133595204,
+      "step": 1480
+    },
+    {
+      "epoch": 6.804597701149425,
+      "loss": 0.05428166314959526,
+      "loss_ce": 6.149261480459245e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 133595204,
+      "step": 1480
+    },
+    {
+      "epoch": 6.809195402298851,
+      "grad_norm": 23.814326518568098,
+      "learning_rate": 5e-06,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 133685616,
+      "step": 1481
+    },
+    {
+      "epoch": 6.809195402298851,
+      "loss": 0.11151270568370819,
+      "loss_ce": 1.4743751535206684e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 133685616,
+      "step": 1481
+    },
+    {
+      "epoch": 6.813793103448276,
+      "grad_norm": 8.881491653313477,
+      "learning_rate": 5e-06,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 133775972,
+      "step": 1482
+    },
+    {
+      "epoch": 6.813793103448276,
+      "loss": 0.04635809361934662,
+      "loss_ce": 9.521063475403935e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 133775972,
+      "step": 1482
+    },
+    {
+      "epoch": 6.818390804597701,
+      "grad_norm": 11.628206791707466,
+      "learning_rate": 5e-06,
+      "loss": 0.076,
+      "num_input_tokens_seen": 133866336,
+      "step": 1483
+    },
+    {
+      "epoch": 6.818390804597701,
+      "loss": 0.08055493235588074,
+      "loss_ce": 3.7818542750756023e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 133866336,
+      "step": 1483
+    },
+    {
+      "epoch": 6.8229885057471265,
+      "grad_norm": 2.105574615398993,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 133956784,
+      "step": 1484
+    },
+    {
+      "epoch": 6.8229885057471265,
+      "loss": 0.09039635956287384,
+      "loss_ce": 3.2979687603074126e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 133956784,
+      "step": 1484
+    },
+    {
+      "epoch": 6.827586206896552,
+      "grad_norm": 3.8079419854599075,
+      "learning_rate": 5e-06,
+      "loss": 0.1018,
+      "num_input_tokens_seen": 134047040,
+      "step": 1485
+    },
+    {
+      "epoch": 6.827586206896552,
+      "loss": 0.09102049469947815,
+      "loss_ce": 0.00044431857531890273,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 134047040,
+      "step": 1485
+    },
+    {
+      "epoch": 6.832183908045977,
+      "grad_norm": 21.116070086676018,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 134137528,
+      "step": 1486
+    },
+    {
+      "epoch": 6.832183908045977,
+      "loss": 0.0954207181930542,
+      "loss_ce": 7.509706392738735e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 134137528,
+      "step": 1486
+    },
+    {
+      "epoch": 6.836781609195402,
+      "grad_norm": 14.756310820481287,
+      "learning_rate": 5e-06,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 134227828,
+      "step": 1487
+    },
+    {
+      "epoch": 6.836781609195402,
+      "loss": 0.06289245933294296,
+      "loss_ce": 1.098936172638787e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 134227828,
+      "step": 1487
+    },
+    {
+      "epoch": 6.841379310344828,
+      "grad_norm": 2.276241824617552,
+      "learning_rate": 5e-06,
+      "loss": 0.073,
+      "num_input_tokens_seen": 134318180,
+      "step": 1488
+    },
+    {
+      "epoch": 6.841379310344828,
+      "loss": 0.07049933075904846,
+      "loss_ce": 3.7283075471350458e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 134318180,
+      "step": 1488
+    },
+    {
+      "epoch": 6.845977011494253,
+      "grad_norm": 9.670203919462505,
+      "learning_rate": 5e-06,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 134408616,
+      "step": 1489
+    },
+    {
+      "epoch": 6.845977011494253,
+      "loss": 0.05651029944419861,
+      "loss_ce": 7.007898602751084e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 134408616,
+      "step": 1489
+    },
+    {
+      "epoch": 6.850574712643678,
+      "grad_norm": 4.106681680803513,
+      "learning_rate": 5e-06,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 134498968,
+      "step": 1490
+    },
+    {
+      "epoch": 6.850574712643678,
+      "loss": 0.055772751569747925,
+      "loss_ce": 1.8806914567903732e-06,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 134498968,
+      "step": 1490
+    },
+    {
+      "epoch": 6.855172413793103,
+      "grad_norm": 8.347874999340169,
+      "learning_rate": 5e-06,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 134589296,
+      "step": 1491
+    },
+    {
+      "epoch": 6.855172413793103,
+      "loss": 0.10776175558567047,
+      "loss_ce": 1.9452994820312597e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 134589296,
+      "step": 1491
+    },
+    {
+      "epoch": 6.8597701149425285,
+      "grad_norm": 11.405237946200163,
+      "learning_rate": 5e-06,
+      "loss": 0.0905,
+      "num_input_tokens_seen": 134679776,
+      "step": 1492
+    },
+    {
+      "epoch": 6.8597701149425285,
+      "loss": 0.09717725962400436,
+      "loss_ce": 9.286872227676213e-06,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 134679776,
+      "step": 1492
+    },
+    {
+      "epoch": 6.864367816091954,
+      "grad_norm": 3.675422132416597,
+      "learning_rate": 5e-06,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 134769980,
+      "step": 1493
+    },
+    {
+      "epoch": 6.864367816091954,
+      "loss": 0.0675213634967804,
+      "loss_ce": 1.2203744290673058e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 134769980,
+      "step": 1493
+    },
+    {
+      "epoch": 6.86896551724138,
+      "grad_norm": 5.2386125622777415,
+      "learning_rate": 5e-06,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 134860216,
+      "step": 1494
+    },
+    {
+      "epoch": 6.86896551724138,
+      "loss": 0.07779578119516373,
+      "loss_ce": 2.9361290216911584e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 134860216,
+      "step": 1494
+    },
+    {
+      "epoch": 6.873563218390805,
+      "grad_norm": 3.015206476190183,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 134950708,
+      "step": 1495
+    },
+    {
+      "epoch": 6.873563218390805,
+      "loss": 0.06314820051193237,
+      "loss_ce": 7.3293504101457074e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 134950708,
+      "step": 1495
+    },
+    {
+      "epoch": 6.87816091954023,
+      "grad_norm": 2.4409371869924272,
+      "learning_rate": 5e-06,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 135041104,
+      "step": 1496
+    },
+    {
+      "epoch": 6.87816091954023,
+      "loss": 0.046701833605766296,
+      "loss_ce": 9.940345989889465e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 135041104,
+      "step": 1496
+    },
+    {
+      "epoch": 6.882758620689655,
+      "grad_norm": 4.738944993160565,
+      "learning_rate": 5e-06,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 135131460,
+      "step": 1497
+    },
+    {
+      "epoch": 6.882758620689655,
+      "loss": 0.04962436109781265,
+      "loss_ce": 2.7832118121295935e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 135131460,
+      "step": 1497
+    },
+    {
+      "epoch": 6.88735632183908,
+      "grad_norm": 8.49862390193789,
+      "learning_rate": 5e-06,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 135221032,
+      "step": 1498
+    },
+    {
+      "epoch": 6.88735632183908,
+      "loss": 0.041058339178562164,
+      "loss_ce": 4.569673365040217e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 135221032,
+      "step": 1498
+    },
+    {
+      "epoch": 6.8919540229885055,
+      "grad_norm": 15.242863888922875,
+      "learning_rate": 5e-06,
+      "loss": 0.064,
+      "num_input_tokens_seen": 135311548,
+      "step": 1499
+    },
+    {
+      "epoch": 6.8919540229885055,
+      "loss": 0.06402122974395752,
+      "loss_ce": 1.0608761840558145e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 135311548,
+      "step": 1499
+    },
+    {
+      "epoch": 6.896551724137931,
+      "grad_norm": 2.704452207506696,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 135401836,
+      "step": 1500
+    },
+    {
+      "epoch": 6.896551724137931,
+      "eval_seeclick_CIoU": 0.46876952052116394,
+      "eval_seeclick_GIoU": 0.4556391090154648,
+      "eval_seeclick_IoU": 0.5108004212379456,
+      "eval_seeclick_MAE_all": 0.061132827773690224,
+      "eval_seeclick_MAE_h": 0.04886363446712494,
+      "eval_seeclick_MAE_w": 0.12054737657308578,
+      "eval_seeclick_MAE_x_boxes": 0.10729708895087242,
+      "eval_seeclick_MAE_y_boxes": 0.052372606471180916,
+      "eval_seeclick_NUM_probability": 0.9999991655349731,
+      "eval_seeclick_inside_bbox": 0.7542613744735718,
+      "eval_seeclick_loss": 0.37735113501548767,
+      "eval_seeclick_loss_ce": 0.07988087832927704,
+      "eval_seeclick_loss_iou": 0.4791259765625,
+      "eval_seeclick_loss_num": 0.06346893310546875,
+      "eval_seeclick_loss_xval": 0.317535400390625,
+      "eval_seeclick_runtime": 75.9161,
+      "eval_seeclick_samples_per_second": 0.566,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 135401836,
+      "step": 1500
+    },
+    {
+      "epoch": 6.896551724137931,
+      "eval_icons_CIoU": 0.6196990609169006,
+      "eval_icons_GIoU": 0.6126499474048615,
+      "eval_icons_IoU": 0.6482784748077393,
+      "eval_icons_MAE_all": 0.03874216973781586,
+      "eval_icons_MAE_h": 0.0633212048560381,
+      "eval_icons_MAE_w": 0.06230618245899677,
+      "eval_icons_MAE_x_boxes": 0.056700803339481354,
+      "eval_icons_MAE_y_boxes": 0.06309299729764462,
+      "eval_icons_NUM_probability": 0.9999997913837433,
+      "eval_icons_inside_bbox": 0.8107638955116272,
+      "eval_icons_loss": 0.1911141723394394,
+      "eval_icons_loss_ce": 1.2011703915959515e-06,
+      "eval_icons_loss_iou": 0.45794677734375,
+      "eval_icons_loss_num": 0.04096221923828125,
+      "eval_icons_loss_xval": 0.2046356201171875,
+      "eval_icons_runtime": 99.4274,
+      "eval_icons_samples_per_second": 0.503,
+      "eval_icons_steps_per_second": 0.02,
+      "num_input_tokens_seen": 135401836,
+      "step": 1500
+    },
+    {
+      "epoch": 6.896551724137931,
+      "eval_screenspot_CIoU": 0.4273842175801595,
+      "eval_screenspot_GIoU": 0.4152764479319255,
+      "eval_screenspot_IoU": 0.4939347803592682,
+      "eval_screenspot_MAE_all": 0.083914448817571,
+      "eval_screenspot_MAE_h": 0.07696965336799622,
+      "eval_screenspot_MAE_w": 0.17438126603762308,
+      "eval_screenspot_MAE_x_boxes": 0.1727352738380432,
+      "eval_screenspot_MAE_y_boxes": 0.0719092587629954,
+      "eval_screenspot_NUM_probability": 0.9999995628992716,
+      "eval_screenspot_inside_bbox": 0.7504166762034098,
+      "eval_screenspot_loss": 0.41863083839416504,
+      "eval_screenspot_loss_ce": 4.884489499090705e-05,
+      "eval_screenspot_loss_iou": 0.3917236328125,
+      "eval_screenspot_loss_num": 0.08594767252604167,
+      "eval_screenspot_loss_xval": 0.4297281901041667,
+      "eval_screenspot_runtime": 164.7584,
+      "eval_screenspot_samples_per_second": 0.54,
+      "eval_screenspot_steps_per_second": 0.018,
+      "num_input_tokens_seen": 135401836,
+      "step": 1500
+    },
+    {
+      "epoch": 6.896551724137931,
+      "eval_compot_CIoU": 0.4806292653083801,
+      "eval_compot_GIoU": 0.4541057199239731,
+      "eval_compot_IoU": 0.546671450138092,
+      "eval_compot_MAE_all": 0.05766832269728184,
+      "eval_compot_MAE_h": 0.07456954568624496,
+      "eval_compot_MAE_w": 0.11397556215524673,
+      "eval_compot_MAE_x_boxes": 0.10434515029191971,
+      "eval_compot_MAE_y_boxes": 0.07447323948144913,
+      "eval_compot_NUM_probability": 0.9999996423721313,
+      "eval_compot_inside_bbox": 0.7604166567325592,
+      "eval_compot_loss": 0.31393927335739136,
+      "eval_compot_loss_ce": 0.01060745446011424,
+      "eval_compot_loss_iou": 0.49609375,
+      "eval_compot_loss_num": 0.05194854736328125,
+      "eval_compot_loss_xval": 0.259796142578125,
+      "eval_compot_runtime": 92.9548,
+      "eval_compot_samples_per_second": 0.538,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 135401836,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 135401836,
+  "num_train_epochs": 47,
+  "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": 839231021613056.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}