diff --git "a/checkpoint-6384/trainer_state.json" "b/checkpoint-6384/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-6384/trainer_state.json"
@@ -0,0 +1,4487 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6384,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.129890453834116e-07,
+      "loss": 1.701,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.259780907668232e-07,
+      "loss": 1.7127,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.389671361502348e-07,
+      "loss": 1.7125,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.2519561815336464e-06,
+      "loss": 1.7616,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.5649452269170582e-06,
+      "loss": 1.7915,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7578125,
+      "learning_rate": 1.8779342723004696e-06,
+      "loss": 1.6313,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.190923317683881e-06,
+      "loss": 1.6517,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.498046875,
+      "learning_rate": 2.503912363067293e-06,
+      "loss": 1.6164,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 1.7285,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.1298904538341164e-06,
+      "loss": 1.7422,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.451171875,
+      "learning_rate": 3.4428794992175273e-06,
+      "loss": 1.8099,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.76171875,
+      "learning_rate": 3.755868544600939e-06,
+      "loss": 1.6931,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.734375,
+      "learning_rate": 4.068857589984351e-06,
+      "loss": 1.7755,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.56640625,
+      "learning_rate": 4.381846635367762e-06,
+      "loss": 1.7773,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.453125,
+      "learning_rate": 4.694835680751174e-06,
+      "loss": 1.6329,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.140625,
+      "learning_rate": 5.007824726134586e-06,
+      "loss": 1.5937,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.046875,
+      "learning_rate": 5.320813771517997e-06,
+      "loss": 1.7887,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 1.6871,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.946791862284821e-06,
+      "loss": 1.7644,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.03125,
+      "learning_rate": 6.259780907668233e-06,
+      "loss": 1.6919,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.66015625,
+      "learning_rate": 6.572769953051644e-06,
+      "loss": 1.7546,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.885758998435055e-06,
+      "loss": 1.5878,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.15625,
+      "learning_rate": 7.198748043818467e-06,
+      "loss": 1.6308,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.511737089201878e-06,
+      "loss": 1.7152,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.875,
+      "learning_rate": 7.82472613458529e-06,
+      "loss": 1.6832,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9140625,
+      "learning_rate": 8.137715179968702e-06,
+      "loss": 1.6066,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2578125,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 1.5553,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.763693270735524e-06,
+      "loss": 1.6204,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.076682316118937e-06,
+      "loss": 1.5727,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.389671361502349e-06,
+      "loss": 1.4473,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.70266040688576e-06,
+      "loss": 1.5953,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0015649452269172e-05,
+      "loss": 1.6369,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.0328638497652584e-05,
+      "loss": 1.5789,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0641627543035994e-05,
+      "loss": 1.4574,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.0954616588419406e-05,
+      "loss": 1.3869,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.25,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 1.5444,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.158059467918623e-05,
+      "loss": 1.5301,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.1893583724569641e-05,
+      "loss": 1.3601,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.2206572769953053e-05,
+      "loss": 1.4415,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.2519561815336466e-05,
+      "loss": 1.4813,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2832550860719876e-05,
+      "loss": 1.3704,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.3145539906103288e-05,
+      "loss": 1.51,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.34585289514867e-05,
+      "loss": 1.4507,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.377151799687011e-05,
+      "loss": 1.377,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 1.4595,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.609375,
+      "learning_rate": 1.4397496087636934e-05,
+      "loss": 1.3629,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.4710485133020344e-05,
+      "loss": 1.3975,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.5023474178403756e-05,
+      "loss": 1.4047,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.533646322378717e-05,
+      "loss": 1.4089,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.28125,
+      "learning_rate": 1.564945226917058e-05,
+      "loss": 1.4803,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.44921875,
+      "learning_rate": 1.5962441314553993e-05,
+      "loss": 1.3644,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.349609375,
+      "learning_rate": 1.6275430359937403e-05,
+      "loss": 1.344,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.6588419405320814e-05,
+      "loss": 1.3061,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 1.2939,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.721439749608764e-05,
+      "loss": 1.2941,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.752738654147105e-05,
+      "loss": 1.3217,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.7840375586854463e-05,
+      "loss": 1.2587,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.8153364632237873e-05,
+      "loss": 1.2699,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.8466353677621284e-05,
+      "loss": 1.3925,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.498046875,
+      "learning_rate": 1.8779342723004698e-05,
+      "loss": 1.1709,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6640625,
+      "learning_rate": 1.9092331768388108e-05,
+      "loss": 1.1771,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.765625,
+      "learning_rate": 1.940532081377152e-05,
+      "loss": 1.2695,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 1.3151,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.9999998504833103e-05,
+      "loss": 1.3924,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.9999819085346603e-05,
+      "loss": 1.3382,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.9999340638628598e-05,
+      "loss": 1.3295,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.9998563178986208e-05,
+      "loss": 1.2471,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.9997486729668013e-05,
+      "loss": 1.3491,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.9996111322863365e-05,
+      "loss": 1.3386,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.9994436999701415e-05,
+      "loss": 1.2892,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.999246381024989e-05,
+      "loss": 1.2936,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.9990191813513596e-05,
+      "loss": 1.2746,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 1.9987621077432647e-05,
+      "loss": 1.2508,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.9984751678880453e-05,
+      "loss": 1.4038,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.310546875,
+      "learning_rate": 1.998158370366139e-05,
+      "loss": 1.2358,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.61328125,
+      "learning_rate": 1.997811724650826e-05,
+      "loss": 1.3496,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.28125,
+      "learning_rate": 1.9974352411079447e-05,
+      "loss": 1.2574,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2890625,
+      "learning_rate": 1.9970289309955814e-05,
+      "loss": 1.2745,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.9965928064637353e-05,
+      "loss": 1.2888,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.9961268805539535e-05,
+      "loss": 1.3366,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5,
+      "learning_rate": 1.9956311671989412e-05,
+      "loss": 1.2548,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6875,
+      "learning_rate": 1.995105681222146e-05,
+      "loss": 1.289,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.994550438337314e-05,
+      "loss": 1.2034,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.275390625,
+      "learning_rate": 1.9939654551480196e-05,
+      "loss": 1.2457,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.9933507491471698e-05,
+      "loss": 1.2896,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.99270633871648e-05,
+      "loss": 1.2492,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.9920322431259266e-05,
+      "loss": 1.2831,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 1.991328482533167e-05,
+      "loss": 1.1767,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.546875,
+      "learning_rate": 1.9905950779829406e-05,
+      "loss": 1.3898,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.9898320514064374e-05,
+      "loss": 1.2236,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.989039425620643e-05,
+      "loss": 1.2442,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.9882172243276544e-05,
+      "loss": 1.2188,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.9873654721139758e-05,
+      "loss": 1.2307,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.986484194449777e-05,
+      "loss": 1.1559,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.985573417688138e-05,
+      "loss": 1.2001,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.9846331690642562e-05,
+      "loss": 1.1715,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.62890625,
+      "learning_rate": 1.9836634766946347e-05,
+      "loss": 1.1591,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.9826643695762406e-05,
+      "loss": 1.1985,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3125,
+      "learning_rate": 1.9816358775856383e-05,
+      "loss": 1.3536,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.9805780314780948e-05,
+      "loss": 1.2464,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.9794908628866623e-05,
+      "loss": 1.1834,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.296875,
+      "learning_rate": 1.97837440432123e-05,
+      "loss": 1.265,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.9772286891675533e-05,
+      "loss": 1.2534,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2890625,
+      "learning_rate": 1.976053751686255e-05,
+      "loss": 1.2437,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.9748496270118012e-05,
+      "loss": 1.2886,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 1.97361635115145e-05,
+      "loss": 1.2001,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.46875,
+      "learning_rate": 1.9723539609841755e-05,
+      "loss": 1.247,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.248046875,
+      "learning_rate": 1.9710624942595638e-05,
+      "loss": 1.1803,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.63671875,
+      "learning_rate": 1.9697419895966855e-05,
+      "loss": 1.3192,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5234375,
+      "learning_rate": 1.968392486482941e-05,
+      "loss": 1.287,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.967014025272878e-05,
+      "loss": 1.2746,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.9656066471869858e-05,
+      "loss": 1.2616,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.9641703943104637e-05,
+      "loss": 1.266,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.9627053095919605e-05,
+      "loss": 1.2531,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.9612114368422913e-05,
+      "loss": 1.3275,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.61328125,
+      "learning_rate": 1.9596888207331283e-05,
+      "loss": 1.29,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.958137506795662e-05,
+      "loss": 1.2133,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.956557541419244e-05,
+      "loss": 1.2453,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.9549489718499956e-05,
+      "loss": 1.1542,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.953311846189398e-05,
+      "loss": 1.2101,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.9516462133928518e-05,
+      "loss": 1.1516,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.949952123268215e-05,
+      "loss": 1.2818,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.9482296264743113e-05,
+      "loss": 1.1855,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.302734375,
+      "learning_rate": 1.946478774519418e-05,
+      "loss": 1.2746,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.21875,
+      "learning_rate": 1.9446996197597236e-05,
+      "loss": 1.25,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.25390625,
+      "learning_rate": 1.9428922153977627e-05,
+      "loss": 1.2361,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.326171875,
+      "learning_rate": 1.9410566154808248e-05,
+      "loss": 1.2371,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.93919287489934e-05,
+      "loss": 1.2412,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.9373010493852344e-05,
+      "loss": 1.1178,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.9353811955102663e-05,
+      "loss": 1.2023,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2890625,
+      "learning_rate": 1.933433370684334e-05,
+      "loss": 1.2467,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.498046875,
+      "learning_rate": 1.9314576331537566e-05,
+      "loss": 1.1468,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3359375,
+      "learning_rate": 1.929454041999536e-05,
+      "loss": 1.2449,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.265625,
+      "learning_rate": 1.9274226571355877e-05,
+      "loss": 1.2869,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.224609375,
+      "learning_rate": 1.9253635393069495e-05,
+      "loss": 1.2359,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.923276750087966e-05,
+      "loss": 1.1137,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.224609375,
+      "learning_rate": 1.921162351880447e-05,
+      "loss": 1.1493,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.9190204079118e-05,
+      "loss": 1.138,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.9168509822331406e-05,
+      "loss": 1.2596,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.9146541397173792e-05,
+      "loss": 1.2008,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.912429946057277e-05,
+      "loss": 1.2667,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.910178467763485e-05,
+      "loss": 1.1383,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.9078997721625528e-05,
+      "loss": 1.248,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.9055939273949176e-05,
+      "loss": 1.2668,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.903261002412864e-05,
+      "loss": 1.2421,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.900901066978464e-05,
+      "loss": 1.1993,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 1.8985141916614906e-05,
+      "loss": 1.2946,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8961004478373063e-05,
+      "loss": 1.1856,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.8936599076847306e-05,
+      "loss": 1.2292,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.8911926441838805e-05,
+      "loss": 1.2673,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.330078125,
+      "learning_rate": 1.888698731113988e-05,
+      "loss": 1.1434,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.8861782430511946e-05,
+      "loss": 1.2079,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.8836312553663202e-05,
+      "loss": 1.3075,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 1.8810578442226106e-05,
+      "loss": 1.1881,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.8784580865734586e-05,
+      "loss": 1.1529,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.8758320601601045e-05,
+      "loss": 1.2133,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.8731798435093093e-05,
+      "loss": 1.237,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 1.8705015159310084e-05,
+      "loss": 1.1524,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.8677971575159387e-05,
+      "loss": 1.1887,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.275390625,
+      "learning_rate": 1.8650668491332445e-05,
+      "loss": 1.2928,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.862310672428058e-05,
+      "loss": 1.2284,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.30078125,
+      "learning_rate": 1.8595287098190593e-05,
+      "loss": 1.2392,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.8567210444960117e-05,
+      "loss": 1.2275,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.8538877604172728e-05,
+      "loss": 1.2713,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.8510289423072843e-05,
+      "loss": 1.1952,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.84814467565404e-05,
+      "loss": 1.2008,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.845235046706527e-05,
+      "loss": 1.206,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.8423001424721482e-05,
+      "loss": 1.2427,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 1.8393400507141207e-05,
+      "loss": 1.2823,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.8363548599488496e-05,
+      "loss": 1.2011,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 1.833344659443283e-05,
+      "loss": 1.2452,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.28125,
+      "learning_rate": 1.8303095392122425e-05,
+      "loss": 1.209,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.827249590015729e-05,
+      "loss": 1.1991,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4921875,
+      "learning_rate": 1.8241649033562124e-05,
+      "loss": 1.1159,
+      "step": 1740
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.8210555714758925e-05,
+      "loss": 1.1678,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.817921687353941e-05,
+      "loss": 1.2747,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5,
+      "learning_rate": 1.814763344703724e-05,
+      "loss": 1.1928,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.31640625,
+      "learning_rate": 1.8115806379699954e-05,
+      "loss": 1.2759,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.28515625,
+      "learning_rate": 1.8083736623260747e-05,
+      "loss": 1.2477,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.8051425136710026e-05,
+      "loss": 1.2446,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 1.8018872886266705e-05,
+      "loss": 1.1644,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.7986080845349326e-05,
+      "loss": 1.2178,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.28125,
+      "learning_rate": 1.7953049994546954e-05,
+      "loss": 1.2681,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.29296875,
+      "learning_rate": 1.7919781321589844e-05,
+      "loss": 1.2823,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.7886275821319904e-05,
+      "loss": 1.3387,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.785253449566097e-05,
+      "loss": 1.3572,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.310546875,
+      "learning_rate": 1.7818558353588812e-05,
+      "loss": 1.1462,
+      "step": 1870
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.7784348411100976e-05,
+      "loss": 1.1801,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.7749905691186415e-05,
+      "loss": 1.1731,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.7715231223794877e-05,
+      "loss": 1.2395,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.768032604580612e-05,
+      "loss": 1.2627,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.7645191200998903e-05,
+      "loss": 1.2788,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.760982774001977e-05,
+      "loss": 1.0834,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2890625,
+      "learning_rate": 1.7574236720351635e-05,
+      "loss": 1.2172,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 1.753841920628216e-05,
+      "loss": 1.2457,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.7502376268871932e-05,
+      "loss": 1.2578,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.306640625,
+      "learning_rate": 1.7466108985922424e-05,
+      "loss": 1.324,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.228515625,
+      "learning_rate": 1.7429618441943783e-05,
+      "loss": 1.2364,
+      "step": 1980
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.25,
+      "learning_rate": 1.7392905728122384e-05,
+      "loss": 1.2368,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5,
+      "learning_rate": 1.7355971942288197e-05,
+      "loss": 1.3144,
+      "step": 2000
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.7318818188881988e-05,
+      "loss": 1.2544,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.7281445578922252e-05,
+      "loss": 1.1518,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.279296875,
+      "learning_rate": 1.724385522997202e-05,
+      "loss": 1.1567,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.302734375,
+      "learning_rate": 1.720604826610542e-05,
+      "loss": 1.1804,
+      "step": 2040
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.7168025817874086e-05,
+      "loss": 1.2257,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.30859375,
+      "learning_rate": 1.7129789022273327e-05,
+      "loss": 1.1921,
+      "step": 2060
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.7091339022708144e-05,
+      "loss": 1.1847,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.23828125,
+      "learning_rate": 1.7052676968959025e-05,
+      "loss": 1.1852,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.291015625,
+      "learning_rate": 1.7013804017147583e-05,
+      "loss": 1.3184,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.306640625,
+      "learning_rate": 1.6974721329701963e-05,
+      "loss": 1.1809,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.6935430075322082e-05,
+      "loss": 1.2397,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.6895931428944708e-05,
+      "loss": 1.3251,
+      "step": 2120
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.6856226571708297e-05,
+      "loss": 1.2728,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.296875,
+      "learning_rate": 1.6816316690917678e-05,
+      "loss": 1.1343,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.677620298000856e-05,
+      "loss": 1.2037,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.6735886638511835e-05,
+      "loss": 1.2551,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 1.6695368872017717e-05,
+      "loss": 1.3094,
+      "step": 2170
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.279296875,
+      "learning_rate": 1.6654650892139672e-05,
+      "loss": 1.2483,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.333984375,
+      "learning_rate": 1.6613733916478214e-05,
+      "loss": 1.267,
+      "step": 2190
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.294921875,
+      "learning_rate": 1.6572619168584473e-05,
+      "loss": 1.1888,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.6531307877923613e-05,
+      "loss": 1.2813,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3828125,
+      "learning_rate": 1.6489801279838075e-05,
+      "loss": 1.2645,
+      "step": 2220
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.644810061551062e-05,
+      "loss": 1.1858,
+      "step": 2230
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.640620713192723e-05,
+      "loss": 1.2411,
+      "step": 2240
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.25390625,
+      "learning_rate": 1.6364122081839813e-05,
+      "loss": 1.1877,
+      "step": 2250
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.632184672372874e-05,
+      "loss": 1.2581,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.6279382321765207e-05,
+      "loss": 1.3208,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.6236730145773444e-05,
+      "loss": 1.2014,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6193891471192732e-05,
+      "loss": 1.1917,
+      "step": 2290
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.615086757903928e-05,
+      "loss": 1.2693,
+      "step": 2300
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.482421875,
+      "learning_rate": 1.6107659755867895e-05,
+      "loss": 1.1737,
+      "step": 2310
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.606426929373353e-05,
+      "loss": 1.1647,
+      "step": 2320
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.326171875,
+      "learning_rate": 1.602069749015263e-05,
+      "loss": 1.2467,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.5976945648064348e-05,
+      "loss": 1.1257,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.593301507579158e-05,
+      "loss": 1.2191,
+      "step": 2350
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 1.588890708700183e-05,
+      "loss": 1.1931,
+      "step": 2360
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.5844623000667937e-05,
+      "loss": 1.2469,
+      "step": 2370
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.5800164141028627e-05,
+      "loss": 1.2154,
+      "step": 2380
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.5755531837548933e-05,
+      "loss": 1.2458,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.224609375,
+      "learning_rate": 1.571072742488041e-05,
+      "loss": 1.1664,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.5665752242821243e-05,
+      "loss": 1.2026,
+      "step": 2410
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.228515625,
+      "learning_rate": 1.5620607636276185e-05,
+      "loss": 1.2434,
+      "step": 2420
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5859375,
+      "learning_rate": 1.5575294955216327e-05,
+      "loss": 1.3785,
+      "step": 2430
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.5529815554638742e-05,
+      "loss": 1.2455,
+      "step": 2440
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.548417079452596e-05,
+      "loss": 1.2574,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.283203125,
+      "learning_rate": 1.5438362039805294e-05,
+      "loss": 1.1683,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.65234375,
+      "learning_rate": 1.5392390660308042e-05,
+      "loss": 1.3087,
+      "step": 2470
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.25390625,
+      "learning_rate": 1.5346258030728506e-05,
+      "loss": 1.1486,
+      "step": 2480
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.5299965530582895e-05,
+      "loss": 1.0666,
+      "step": 2490
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.525351454416807e-05,
+      "loss": 1.2205,
+      "step": 2500
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.520690646052014e-05,
+      "loss": 1.1699,
+      "step": 2510
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.5160142673372954e-05,
+      "loss": 1.2787,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.314453125,
+      "learning_rate": 1.5113224581116376e-05,
+      "loss": 1.2602,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.5066153586754527e-05,
+      "loss": 1.1701,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 1.5018931097863775e-05,
+      "loss": 1.1812,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.4971558526550682e-05,
+      "loss": 1.19,
+      "step": 2560
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.765625,
+      "learning_rate": 1.4924037289409763e-05,
+      "loss": 1.144,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.29296875,
+      "learning_rate": 1.4876368807481123e-05,
+      "loss": 1.1487,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.298828125,
+      "learning_rate": 1.4828554506207967e-05,
+      "loss": 1.2094,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.220703125,
+      "learning_rate": 1.4780595815393976e-05,
+      "loss": 1.1933,
+      "step": 2600
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.4732494169160555e-05,
+      "loss": 1.1843,
+      "step": 2610
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.30859375,
+      "learning_rate": 1.4684251005903932e-05,
+      "loss": 1.2875,
+      "step": 2620
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.478515625,
+      "learning_rate": 1.4635867768252159e-05,
+      "loss": 1.2586,
+      "step": 2630
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4587345903021974e-05,
+      "loss": 1.2581,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.4538686861175535e-05,
+      "loss": 1.2135,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5,
+      "learning_rate": 1.4489892097777014e-05,
+      "loss": 1.1416,
+      "step": 2660
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 1.4440963071949116e-05,
+      "loss": 1.2433,
+      "step": 2670
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.236328125,
+      "learning_rate": 1.439190124682942e-05,
+      "loss": 1.2693,
+      "step": 2680
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 1.4342708089526652e-05,
+      "loss": 1.1214,
+      "step": 2690
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.4293385071076776e-05,
+      "loss": 1.2386,
+      "step": 2700
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.424393366639905e-05,
+      "loss": 1.1962,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.4194355354251883e-05,
+      "loss": 1.2677,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.478515625,
+      "learning_rate": 1.4144651617188649e-05,
+      "loss": 1.2425,
+      "step": 2730
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.4094823941513313e-05,
+      "loss": 1.1145,
+      "step": 2740
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.4044873817236035e-05,
+      "loss": 1.2104,
+      "step": 2750
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.244140625,
+      "learning_rate": 1.399480273802857e-05,
+      "loss": 1.1633,
+      "step": 2760
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.3944612201179629e-05,
+      "loss": 1.2205,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.3894303707550085e-05,
+      "loss": 1.1349,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.251953125,
+      "learning_rate": 1.3843878761528115e-05,
+      "loss": 1.2719,
+      "step": 2790
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 1.3793338870984206e-05,
+      "loss": 1.0525,
+      "step": 2800
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.3742685547226046e-05,
+      "loss": 1.2179,
+      "step": 2810
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.287109375,
+      "learning_rate": 1.3691920304953362e-05,
+      "loss": 1.2462,
+      "step": 2820
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.3641044662212595e-05,
+      "loss": 1.1888,
+      "step": 2830
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.48828125,
+      "learning_rate": 1.3590060140351538e-05,
+      "loss": 1.2698,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.3538968263973806e-05,
+      "loss": 1.1812,
+      "step": 2850
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5859375,
+      "learning_rate": 1.348777056089328e-05,
+      "loss": 1.2201,
+      "step": 2860
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.3436468562088399e-05,
+      "loss": 1.152,
+      "step": 2870
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.3385063801656383e-05,
+      "loss": 1.2649,
+      "step": 2880
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.296875,
+      "learning_rate": 1.333355781676736e-05,
+      "loss": 1.2261,
+      "step": 2890
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.3281952147618403e-05,
+      "loss": 1.2537,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.3230248337387464e-05,
+      "loss": 1.2657,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.31640625,
+      "learning_rate": 1.3178447932187244e-05,
+      "loss": 1.1331,
+      "step": 2920
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.3126552481018938e-05,
+      "loss": 1.1538,
+      "step": 2930
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.228515625,
+      "learning_rate": 1.3074563535725932e-05,
+      "loss": 1.2196,
+      "step": 2940
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.302248265094739e-05,
+      "loss": 1.1369,
+      "step": 2950
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 1.2970311384071764e-05,
+      "loss": 1.2216,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.294921875,
+      "learning_rate": 1.2918051295190228e-05,
+      "loss": 1.2409,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.224609375,
+      "learning_rate": 1.286570394705003e-05,
+      "loss": 1.3215,
+      "step": 2980
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 1.2813270905007742e-05,
+      "loss": 1.1936,
+      "step": 2990
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.302734375,
+      "learning_rate": 1.2760753736982474e-05,
+      "loss": 1.2732,
+      "step": 3000
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.270815401340897e-05,
+      "loss": 1.292,
+      "step": 3010
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.2655473307190658e-05,
+      "loss": 1.3339,
+      "step": 3020
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.251953125,
+      "learning_rate": 1.2602713193652616e-05,
+      "loss": 1.1758,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.2549875250494454e-05,
+      "loss": 1.1572,
+      "step": 3040
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.249696105774314e-05,
+      "loss": 1.2257,
+      "step": 3050
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.2443972197705758e-05,
+      "loss": 1.2389,
+      "step": 3060
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.30859375,
+      "learning_rate": 1.2390910254922188e-05,
+      "loss": 1.1787,
+      "step": 3070
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.609375,
+      "learning_rate": 1.2337776816117719e-05,
+      "loss": 1.2179,
+      "step": 3080
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.244140625,
+      "learning_rate": 1.2284573470155607e-05,
+      "loss": 1.217,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.2231301807989564e-05,
+      "loss": 1.1811,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 1.2177963422616169e-05,
+      "loss": 1.2079,
+      "step": 3110
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.2124559909027247e-05,
+      "loss": 1.2469,
+      "step": 3120
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.207109286416218e-05,
+      "loss": 1.0678,
+      "step": 3130
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.2017563886860129e-05,
+      "loss": 1.2444,
+      "step": 3140
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1963974577812239e-05,
+      "loss": 1.1752,
+      "step": 3150
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.1910326539513776e-05,
+      "loss": 1.2623,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.1856621376216204e-05,
+      "loss": 1.2113,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.1802860693879201e-05,
+      "loss": 1.1681,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.1749046100122652e-05,
+      "loss": 1.168,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.255859375,
+      "learning_rate": 1.1695179204178573e-05,
+      "loss": 1.1828,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.30859375,
+      "learning_rate": 1.164126161684298e-05,
+      "loss": 1.2672,
+      "step": 3210
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.1587294950427728e-05,
+      "loss": 1.2852,
+      "step": 3220
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.314453125,
+      "learning_rate": 1.15332808187123e-05,
+      "loss": 1.1258,
+      "step": 3230
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.147922083689554e-05,
+      "loss": 1.1663,
+      "step": 3240
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 1.1425116621547368e-05,
+      "loss": 1.2313,
+      "step": 3250
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5234375,
+      "learning_rate": 1.1370969790560424e-05,
+      "loss": 1.1852,
+      "step": 3260
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.1316781963101704e-05,
+      "loss": 1.2804,
+      "step": 3270
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.1262554759564117e-05,
+      "loss": 1.1496,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.859375,
+      "learning_rate": 1.120828980151806e-05,
+      "loss": 1.2723,
+      "step": 3290
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1153988711662902e-05,
+      "loss": 1.1763,
+      "step": 3300
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.28125,
+      "learning_rate": 1.1099653113778483e-05,
+      "loss": 1.1622,
+      "step": 3310
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.287109375,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 1.2021,
+      "step": 3320
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.0990884894152112e-05,
+      "loss": 1.2304,
+      "step": 3330
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 1.0936455524934962e-05,
+      "loss": 1.2693,
+      "step": 3340
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.088199815264089e-05,
+      "loss": 1.2332,
+      "step": 3350
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 1.082751440572308e-05,
+      "loss": 1.1947,
+      "step": 3360
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.077300591342342e-05,
+      "loss": 1.1761,
+      "step": 3370
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.0718474305723743e-05,
+      "loss": 1.1481,
+      "step": 3380
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 1.0663921213297122e-05,
+      "loss": 1.1735,
+      "step": 3390
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.24609375,
+      "learning_rate": 1.0609348267459088e-05,
+      "loss": 1.1684,
+      "step": 3400
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.22265625,
+      "learning_rate": 1.0554757100118856e-05,
+      "loss": 1.1962,
+      "step": 3410
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.0500149343730524e-05,
+      "loss": 1.2755,
+      "step": 3420
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.25390625,
+      "learning_rate": 1.0445526631244255e-05,
+      "loss": 1.2307,
+      "step": 3430
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0390890596057448e-05,
+      "loss": 1.2283,
+      "step": 3440
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.322265625,
+      "learning_rate": 1.0336242871965893e-05,
+      "loss": 1.2655,
+      "step": 3450
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.326171875,
+      "learning_rate": 1.0281585093114927e-05,
+      "loss": 1.1062,
+      "step": 3460
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.283203125,
+      "learning_rate": 1.0226918893950537e-05,
+      "loss": 1.1902,
+      "step": 3470
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 1.0172245909170524e-05,
+      "loss": 1.1236,
+      "step": 3480
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 1.0117567773675592e-05,
+      "loss": 1.1946,
+      "step": 3490
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.006288612252047e-05,
+      "loss": 1.3188,
+      "step": 3500
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.0008202590865013e-05,
+      "loss": 1.2047,
+      "step": 3510
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.953518813925314e-06,
+      "loss": 1.2158,
+      "step": 3520
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.898836426924803e-06,
+      "loss": 1.1857,
+      "step": 3530
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2109375,
+      "learning_rate": 9.844157065045338e-06,
+      "loss": 1.1521,
+      "step": 3540
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.789482363378323e-06,
+      "loss": 1.1706,
+      "step": 3550
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.251953125,
+      "learning_rate": 9.7348139568758e-06,
+      "loss": 1.1956,
+      "step": 3560
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.283203125,
+      "learning_rate": 9.680153480301572e-06,
+      "loss": 1.2862,
+      "step": 3570
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.625502568182312e-06,
+      "loss": 1.1988,
+      "step": 3580
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 9.570862854758672e-06,
+      "loss": 1.2345,
+      "step": 3590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.263671875,
+      "learning_rate": 9.516235973936448e-06,
+      "loss": 1.2219,
+      "step": 3600
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.25,
+      "learning_rate": 9.46162355923767e-06,
+      "loss": 1.1842,
+      "step": 3610
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.24609375,
+      "learning_rate": 9.407027243751812e-06,
+      "loss": 1.1988,
+      "step": 3620
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.640625,
+      "learning_rate": 9.352448660086916e-06,
+      "loss": 1.2242,
+      "step": 3630
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.283203125,
+      "learning_rate": 9.297889440320783e-06,
+      "loss": 1.3356,
+      "step": 3640
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.243351215952178e-06,
+      "loss": 1.1629,
+      "step": 3650
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.188835617852022e-06,
+      "loss": 1.2277,
+      "step": 3660
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.212890625,
+      "learning_rate": 9.13434427621465e-06,
+      "loss": 1.1474,
+      "step": 3670
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.07987882050905e-06,
+      "loss": 1.2247,
+      "step": 3680
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.259765625,
+      "learning_rate": 9.025440879430115e-06,
+      "loss": 1.2345,
+      "step": 3690
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.23828125,
+      "learning_rate": 8.971032080849995e-06,
+      "loss": 1.1974,
+      "step": 3700
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3203125,
+      "learning_rate": 8.916654051769354e-06,
+      "loss": 1.1557,
+      "step": 3710
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.89453125,
+      "learning_rate": 8.862308418268764e-06,
+      "loss": 1.1153,
+      "step": 3720
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.27734375,
+      "learning_rate": 8.807996805460057e-06,
+      "loss": 1.2041,
+      "step": 3730
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.753720837437735e-06,
+      "loss": 1.1255,
+      "step": 3740
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.79296875,
+      "learning_rate": 8.699482137230411e-06,
+      "loss": 1.249,
+      "step": 3750
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.28125,
+      "learning_rate": 8.64528232675226e-06,
+      "loss": 1.1329,
+      "step": 3760
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6328125,
+      "learning_rate": 8.591123026754535e-06,
+      "loss": 1.1503,
+      "step": 3770
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.267578125,
+      "learning_rate": 8.537005856777082e-06,
+      "loss": 1.268,
+      "step": 3780
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.26953125,
+      "learning_rate": 8.48293243509994e-06,
+      "loss": 1.3126,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.609375,
+      "learning_rate": 8.428904378694915e-06,
+      "loss": 1.3095,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.69921875,
+      "learning_rate": 8.374923303177249e-06,
+      "loss": 1.2363,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.310546875,
+      "learning_rate": 8.320990822757305e-06,
+      "loss": 1.2064,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2578125,
+      "learning_rate": 8.267108550192287e-06,
+      "loss": 1.2578,
+      "step": 3830
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.55859375,
+      "learning_rate": 8.213278096738025e-06,
+      "loss": 1.2356,
+      "step": 3840
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 8.15950107210079e-06,
+      "loss": 1.2283,
+      "step": 3850
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.27734375,
+      "learning_rate": 8.105779084389148e-06,
+      "loss": 1.2084,
+      "step": 3860
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.251953125,
+      "learning_rate": 8.0521137400659e-06,
+      "loss": 1.1478,
+      "step": 3870
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.244140625,
+      "learning_rate": 7.9985066439e-06,
+      "loss": 1.2908,
+      "step": 3880
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.944959398918617e-06,
+      "loss": 1.0884,
+      "step": 3890
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.64453125,
+      "learning_rate": 7.891473606359149e-06,
+      "loss": 1.2915,
+      "step": 3900
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.838050865621385e-06,
+      "loss": 1.241,
+      "step": 3910
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.59375,
+      "learning_rate": 7.784692774219657e-06,
+      "loss": 1.1474,
+      "step": 3920
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.59375,
+      "learning_rate": 7.731400927735058e-06,
+      "loss": 1.2211,
+      "step": 3930
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 7.67817691976776e-06,
+      "loss": 1.2347,
+      "step": 3940
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.251953125,
+      "learning_rate": 7.62502234188932e-06,
+      "loss": 1.189,
+      "step": 3950
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 7.571938783595126e-06,
+      "loss": 1.1896,
+      "step": 3960
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.255859375,
+      "learning_rate": 7.518927832256851e-06,
+      "loss": 1.1365,
+      "step": 3970
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.465991073074963e-06,
+      "loss": 1.1745,
+      "step": 3980
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.462890625,
+      "learning_rate": 7.413130089031364e-06,
+      "loss": 1.2028,
+      "step": 3990
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3359375,
+      "learning_rate": 7.3603464608420186e-06,
+      "loss": 1.1725,
+      "step": 4000
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.287109375,
+      "learning_rate": 7.307641766909702e-06,
+      "loss": 1.1516,
+      "step": 4010
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3046875,
+      "learning_rate": 7.2550175832768035e-06,
+      "loss": 1.254,
+      "step": 4020
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.24609375,
+      "learning_rate": 7.202475483578181e-06,
+      "loss": 1.2421,
+      "step": 4030
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6484375,
+      "learning_rate": 7.150017038994125e-06,
+      "loss": 1.1862,
+      "step": 4040
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5703125,
+      "learning_rate": 7.097643818203356e-06,
+      "loss": 1.1795,
+      "step": 4050
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3359375,
+      "learning_rate": 7.045357387336133e-06,
+      "loss": 1.2763,
+      "step": 4060
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.2265625,
+      "learning_rate": 6.993159309927406e-06,
+      "loss": 1.2483,
+      "step": 4070
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.259765625,
+      "learning_rate": 6.941051146870075e-06,
+      "loss": 1.1744,
+      "step": 4080
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.302734375,
+      "learning_rate": 6.889034456368304e-06,
+      "loss": 1.1285,
+      "step": 4090
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.837110793890927e-06,
+      "loss": 1.2007,
+      "step": 4100
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.7852817121249405e-06,
+      "loss": 1.201,
+      "step": 4110
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.310546875,
+      "learning_rate": 6.733548760929063e-06,
+      "loss": 1.1394,
+      "step": 4120
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.208984375,
+      "learning_rate": 6.6819134872874015e-06,
+      "loss": 1.1848,
+      "step": 4130
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.248046875,
+      "learning_rate": 6.6303774352631855e-06,
+      "loss": 1.2268,
+      "step": 4140
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.279296875,
+      "learning_rate": 6.578942145952585e-06,
+      "loss": 1.1888,
+      "step": 4150
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.527609157438652e-06,
+      "loss": 1.1486,
+      "step": 4160
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.2421875,
+      "learning_rate": 6.476380004745291e-06,
+      "loss": 1.179,
+      "step": 4170
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.30078125,
+      "learning_rate": 6.425256219791391e-06,
+      "loss": 1.2605,
+      "step": 4180
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.41796875,
+      "learning_rate": 6.374239331345004e-06,
+      "loss": 1.1136,
+      "step": 4190
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24609375,
+      "learning_rate": 6.323330864977612e-06,
+      "loss": 1.2338,
+      "step": 4200
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.90625,
+      "learning_rate": 6.272532343018537e-06,
+      "loss": 1.1881,
+      "step": 4210
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.57421875,
+      "learning_rate": 6.221845284509396e-06,
+      "loss": 1.2123,
+      "step": 4220
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.244140625,
+      "learning_rate": 6.171271205158691e-06,
+      "loss": 1.1596,
+      "step": 4230
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.80859375,
+      "learning_rate": 6.120811617296473e-06,
+      "loss": 1.2606,
+      "step": 4240
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.070468029829126e-06,
+      "loss": 1.1965,
+      "step": 4250
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.287109375,
+      "learning_rate": 6.020241948194253e-06,
+      "loss": 1.27,
+      "step": 4260
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.255859375,
+      "learning_rate": 5.970134874315627e-06,
+      "loss": 1.2202,
+      "step": 4270
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40234375,
+      "learning_rate": 5.920148306558322e-06,
+      "loss": 1.2765,
+      "step": 4280
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.2890625,
+      "learning_rate": 5.870283739683875e-06,
+      "loss": 1.1402,
+      "step": 4290
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.34375,
+      "learning_rate": 5.820542664805597e-06,
+      "loss": 1.1105,
+      "step": 4300
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.279296875,
+      "learning_rate": 5.770926569344001e-06,
+      "loss": 1.172,
+      "step": 4310
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.251953125,
+      "learning_rate": 5.7214369369822805e-06,
+      "loss": 1.2084,
+      "step": 4320
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.275390625,
+      "learning_rate": 5.672075247621993e-06,
+      "loss": 1.1865,
+      "step": 4330
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.271484375,
+      "learning_rate": 5.622842977338772e-06,
+      "loss": 1.1764,
+      "step": 4340
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.51171875,
+      "learning_rate": 5.573741598338199e-06,
+      "loss": 1.1887,
+      "step": 4350
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.255859375,
+      "learning_rate": 5.5247725789117855e-06,
+      "loss": 1.2179,
+      "step": 4360
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2734375,
+      "learning_rate": 5.475937383393046e-06,
+      "loss": 1.1983,
+      "step": 4370
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 5.427237472113737e-06,
+      "loss": 1.238,
+      "step": 4380
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.378674301360167e-06,
+      "loss": 1.2155,
+      "step": 4390
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.34375,
+      "learning_rate": 5.330249323329664e-06,
+      "loss": 1.225,
+      "step": 4400
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.32421875,
+      "learning_rate": 5.281963986087124e-06,
+      "loss": 1.227,
+      "step": 4410
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.59375,
+      "learning_rate": 5.233819733521755e-06,
+      "loss": 1.0978,
+      "step": 4420
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 5.1858180053038565e-06,
+      "loss": 1.1948,
+      "step": 4430
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.32421875,
+      "learning_rate": 5.1379602368417804e-06,
+      "loss": 1.3205,
+      "step": 4440
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.265625,
+      "learning_rate": 5.090247859239022e-06,
+      "loss": 1.2407,
+      "step": 4450
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.265625,
+      "learning_rate": 5.042682299251409e-06,
+      "loss": 1.2151,
+      "step": 4460
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.30859375,
+      "learning_rate": 4.99526497924445e-06,
+      "loss": 1.1847,
+      "step": 4470
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.318359375,
+      "learning_rate": 4.947997317150791e-06,
+      "loss": 1.191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.900880726427807e-06,
+      "loss": 1.2072,
+      "step": 4490
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.8539166160153715e-06,
+      "loss": 1.3071,
+      "step": 4500
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 7.4375,
+      "learning_rate": 4.807106390293671e-06,
+      "loss": 1.1894,
+      "step": 4510
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.375,
+      "learning_rate": 4.760451449041255e-06,
+      "loss": 1.1538,
+      "step": 4520
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5390625,
+      "learning_rate": 4.7139531873931535e-06,
+      "loss": 1.1113,
+      "step": 4530
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.69140625,
+      "learning_rate": 4.667612995799168e-06,
+      "loss": 1.2497,
+      "step": 4540
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.621432259982293e-06,
+      "loss": 1.1855,
+      "step": 4550
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 4.57541236089726e-06,
+      "loss": 1.2383,
+      "step": 4560
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.84375,
+      "learning_rate": 4.529554674689273e-06,
+      "loss": 1.1982,
+      "step": 4570
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.251953125,
+      "learning_rate": 4.483860572652835e-06,
+      "loss": 1.168,
+      "step": 4580
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.353515625,
+      "learning_rate": 4.438331421190749e-06,
+      "loss": 1.212,
+      "step": 4590
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.3929685817732585e-06,
+      "loss": 1.1425,
+      "step": 4600
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.267578125,
+      "learning_rate": 4.3477734108973315e-06,
+      "loss": 1.2936,
+      "step": 4610
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.77734375,
+      "learning_rate": 4.302747260046104e-06,
+      "loss": 1.3108,
+      "step": 4620
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6015625,
+      "learning_rate": 4.257891475648452e-06,
+      "loss": 1.2638,
+      "step": 4630
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.213207399038749e-06,
+      "loss": 1.2374,
+      "step": 4640
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.16869636641674e-06,
+      "loss": 1.1836,
+      "step": 4650
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5625,
+      "learning_rate": 4.124359708807591e-06,
+      "loss": 1.2344,
+      "step": 4660
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.080198752022089e-06,
+      "loss": 1.1874,
+      "step": 4670
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.39453125,
+      "learning_rate": 4.036214816616979e-06,
+      "loss": 1.319,
+      "step": 4680
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.22265625,
+      "learning_rate": 3.992409217855511e-06,
+      "loss": 1.1896,
+      "step": 4690
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.248046875,
+      "learning_rate": 3.948783265668074e-06,
+      "loss": 1.2851,
+      "step": 4700
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.271484375,
+      "learning_rate": 3.9053382646130354e-06,
+      "loss": 1.1189,
+      "step": 4710
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.291015625,
+      "learning_rate": 3.8620755138377396e-06,
+      "loss": 1.0757,
+      "step": 4720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.818996307039649e-06,
+      "loss": 1.2775,
+      "step": 4730
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 3.7761019324276703e-06,
+      "loss": 1.2485,
+      "step": 4740
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.7333936726836086e-06,
+      "loss": 1.2393,
+      "step": 4750
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.2578125,
+      "learning_rate": 3.690872804923836e-06,
+      "loss": 1.0362,
+      "step": 4760
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.263671875,
+      "learning_rate": 3.6485406006611048e-06,
+      "loss": 1.1587,
+      "step": 4770
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.306640625,
+      "learning_rate": 3.606398325766489e-06,
+      "loss": 1.2127,
+      "step": 4780
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.28125,
+      "learning_rate": 3.5644472404315743e-06,
+      "loss": 1.1429,
+      "step": 4790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.296875,
+      "learning_rate": 3.522688599130737e-06,
+      "loss": 1.0943,
+      "step": 4800
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.283203125,
+      "learning_rate": 3.4811236505836655e-06,
+      "loss": 1.1858,
+      "step": 4810
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.4397536377179984e-06,
+      "loss": 1.1966,
+      "step": 4820
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.39857979763215e-06,
+      "loss": 1.2747,
+      "step": 4830
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.28125,
+      "learning_rate": 3.357603361558339e-06,
+      "loss": 1.281,
+      "step": 4840
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 3.3168255548257588e-06,
+      "loss": 1.1035,
+      "step": 4850
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.453125,
+      "learning_rate": 3.276247596823934e-06,
+      "loss": 1.2398,
+      "step": 4860
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.235870700966265e-06,
+      "loss": 1.1781,
+      "step": 4870
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.23828125,
+      "learning_rate": 3.1956960746537236e-06,
+      "loss": 1.1593,
+      "step": 4880
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.275390625,
+      "learning_rate": 3.155724919238785e-06,
+      "loss": 1.2204,
+      "step": 4890
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.24609375,
+      "learning_rate": 3.1159584299894596e-06,
+      "loss": 1.2988,
+      "step": 4900
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.291015625,
+      "learning_rate": 3.076397796053582e-06,
+      "loss": 1.1869,
+      "step": 4910
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.25,
+      "learning_rate": 3.037044200423237e-06,
+      "loss": 1.1966,
+      "step": 4920
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.997898819899392e-06,
+      "loss": 1.198,
+      "step": 4930
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 2.9589628250567026e-06,
+      "loss": 1.0903,
+      "step": 4940
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.29296875,
+      "learning_rate": 2.9202373802085015e-06,
+      "loss": 1.2964,
+      "step": 4950
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.220703125,
+      "learning_rate": 2.8817236433719976e-06,
+      "loss": 1.1687,
+      "step": 4960
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.26953125,
+      "learning_rate": 2.843422766233638e-06,
+      "loss": 1.2173,
+      "step": 4970
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.28125,
+      "learning_rate": 2.805335894114668e-06,
+      "loss": 1.1603,
+      "step": 4980
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 2.7674641659368838e-06,
+      "loss": 1.098,
+      "step": 4990
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 2.7298087141885798e-06,
+      "loss": 1.1225,
+      "step": 5000
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.6923706648906756e-06,
+      "loss": 1.2247,
+      "step": 5010
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.828125,
+      "learning_rate": 2.6551511375630445e-06,
+      "loss": 1.1953,
+      "step": 5020
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.618151245191043e-06,
+      "loss": 1.1942,
+      "step": 5030
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.5813720941922347e-06,
+      "loss": 1.2025,
+      "step": 5040
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 2.544814784383278e-06,
+      "loss": 1.1174,
+      "step": 5050
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.306640625,
+      "learning_rate": 2.5084804089470716e-06,
+      "loss": 1.2186,
+      "step": 5060
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.375,
+      "learning_rate": 2.4723700544000306e-06,
+      "loss": 1.202,
+      "step": 5070
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 2.43648480055964e-06,
+      "loss": 1.2475,
+      "step": 5080
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.341796875,
+      "learning_rate": 2.400825720512112e-06,
+      "loss": 1.2315,
+      "step": 5090
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.291015625,
+      "learning_rate": 2.365393880580342e-06,
+      "loss": 1.2473,
+      "step": 5100
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3125,
+      "learning_rate": 2.3301903402919988e-06,
+      "loss": 1.1607,
+      "step": 5110
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.33984375,
+      "learning_rate": 2.2952161523478445e-06,
+      "loss": 1.201,
+      "step": 5120
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.609375,
+      "learning_rate": 2.2604723625902635e-06,
+      "loss": 1.2207,
+      "step": 5130
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.225960009971971e-06,
+      "loss": 1.3026,
+      "step": 5140
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.1916801265249667e-06,
+      "loss": 1.183,
+      "step": 5150
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.1576337373296663e-06,
+      "loss": 1.2809,
+      "step": 5160
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.283203125,
+      "learning_rate": 2.123821860484231e-06,
+      "loss": 1.1526,
+      "step": 5170
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.326171875,
+      "learning_rate": 2.0902455070741488e-06,
+      "loss": 1.2075,
+      "step": 5180
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.0569056811419842e-06,
+      "loss": 1.2174,
+      "step": 5190
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.0238033796573574e-06,
+      "loss": 1.1837,
+      "step": 5200
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.31640625,
+      "learning_rate": 1.9909395924871387e-06,
+      "loss": 1.1852,
+      "step": 5210
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.9583153023658286e-06,
+      "loss": 1.2049,
+      "step": 5220
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 1.925931484866197e-06,
+      "loss": 1.2049,
+      "step": 5230
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.59765625,
+      "learning_rate": 1.893789108370092e-06,
+      "loss": 1.0893,
+      "step": 5240
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.8618891340394896e-06,
+      "loss": 1.1856,
+      "step": 5250
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2890625,
+      "learning_rate": 1.830232515787742e-06,
+      "loss": 1.1596,
+      "step": 5260
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3359375,
+      "learning_rate": 1.7988202002510735e-06,
+      "loss": 1.2642,
+      "step": 5270
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3125,
+      "learning_rate": 1.7676531267602536e-06,
+      "loss": 1.1752,
+      "step": 5280
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.298828125,
+      "learning_rate": 1.7367322273125087e-06,
+      "loss": 1.2549,
+      "step": 5290
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.706058426543664e-06,
+      "loss": 1.2603,
+      "step": 5300
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4765625,
+      "learning_rate": 1.6756326417004865e-06,
+      "loss": 1.2556,
+      "step": 5310
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.6454557826132579e-06,
+      "loss": 1.2466,
+      "step": 5320
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.625,
+      "learning_rate": 1.615528751668568e-06,
+      "loss": 1.1597,
+      "step": 5330
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.5858524437823219e-06,
+      "loss": 1.2212,
+      "step": 5340
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.5564277463730027e-06,
+      "loss": 1.1966,
+      "step": 5350
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.5272555393351018e-06,
+      "loss": 1.1417,
+      "step": 5360
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.318359375,
+      "learning_rate": 1.4983366950128331e-06,
+      "loss": 1.1868,
+      "step": 5370
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.640625,
+      "learning_rate": 1.4696720781740392e-06,
+      "loss": 1.1643,
+      "step": 5380
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.68359375,
+      "learning_rate": 1.4412625459843288e-06,
+      "loss": 1.2583,
+      "step": 5390
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.4131089479814476e-06,
+      "loss": 1.1579,
+      "step": 5400
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.3852121260498708e-06,
+      "loss": 1.1645,
+      "step": 5410
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.349609375,
+      "learning_rate": 1.357572914395635e-06,
+      "loss": 1.2467,
+      "step": 5420
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.3301921395213858e-06,
+      "loss": 1.2517,
+      "step": 5430
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.3030706202016686e-06,
+      "loss": 1.2753,
+      "step": 5440
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 1.2762091674584386e-06,
+      "loss": 1.2371,
+      "step": 5450
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.2496085845368134e-06,
+      "loss": 1.2723,
+      "step": 5460
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.2232696668810507e-06,
+      "loss": 1.1452,
+      "step": 5470
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.1971932021107602e-06,
+      "loss": 1.1048,
+      "step": 5480
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.1713799699973571e-06,
+      "loss": 1.2114,
+      "step": 5490
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.302734375,
+      "learning_rate": 1.1458307424407388e-06,
+      "loss": 1.2246,
+      "step": 5500
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.1205462834462078e-06,
+      "loss": 1.1778,
+      "step": 5510
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.0955273491016206e-06,
+      "loss": 1.2014,
+      "step": 5520
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3203125,
+      "learning_rate": 1.0707746875547741e-06,
+      "loss": 1.3107,
+      "step": 5530
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 1.0462890389910507e-06,
+      "loss": 1.3038,
+      "step": 5540
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.0220711356112668e-06,
+      "loss": 1.2283,
+      "step": 5550
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.25,
+      "learning_rate": 9.9812170160978e-07,
+      "loss": 1.179,
+      "step": 5560
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 9.5,
+      "learning_rate": 9.744414531528433e-07,
+      "loss": 1.2472,
+      "step": 5570
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.244140625,
+      "learning_rate": 9.510310983571835e-07,
+      "loss": 1.2185,
+      "step": 5580
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.278913372688225e-07,
+      "loss": 1.257,
+      "step": 5590
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.050228618421464e-07,
+      "loss": 1.1373,
+      "step": 5600
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.824263559192158e-07,
+      "loss": 1.2337,
+      "step": 5610
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.51171875,
+      "learning_rate": 8.601024952093229e-07,
+      "loss": 1.1784,
+      "step": 5620
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.248046875,
+      "learning_rate": 8.380519472687632e-07,
+      "loss": 1.1264,
+      "step": 5630
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.26171875,
+      "learning_rate": 8.162753714809024e-07,
+      "loss": 1.1725,
+      "step": 5640
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.734375,
+      "learning_rate": 7.94773419036432e-07,
+      "loss": 1.2877,
+      "step": 5650
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.735467329139212e-07,
+      "loss": 1.1887,
+      "step": 5660
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.330078125,
+      "learning_rate": 7.525959478605715e-07,
+      "loss": 1.1805,
+      "step": 5670
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.319216903732384e-07,
+      "loss": 1.1381,
+      "step": 5680
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.115245786797076e-07,
+      "loss": 1.2708,
+      "step": 5690
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.279296875,
+      "learning_rate": 6.914052227201984e-07,
+      "loss": 1.1155,
+      "step": 5700
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.240234375,
+      "learning_rate": 6.715642241291242e-07,
+      "loss": 1.2326,
+      "step": 5710
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.25,
+      "learning_rate": 6.520021762171092e-07,
+      "loss": 1.3097,
+      "step": 5720
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.327196639532352e-07,
+      "loss": 1.2124,
+      "step": 5730
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.259765625,
+      "learning_rate": 6.137172639475663e-07,
+      "loss": 1.2139,
+      "step": 5740
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.94995544433884e-07,
+      "loss": 1.2758,
+      "step": 5750
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.333984375,
+      "learning_rate": 5.76555065252713e-07,
+      "loss": 1.1812,
+      "step": 5760
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.171875,
+      "learning_rate": 5.583963778345747e-07,
+      "loss": 1.1077,
+      "step": 5770
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.405200251834941e-07,
+      "loss": 1.2181,
+      "step": 5780
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.26171875,
+      "learning_rate": 5.229265418607688e-07,
+      "loss": 1.241,
+      "step": 5790
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.33203125,
+      "learning_rate": 5.056164539689745e-07,
+      "loss": 1.262,
+      "step": 5800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.578125,
+      "learning_rate": 4.885902791362429e-07,
+      "loss": 1.1372,
+      "step": 5810
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.265625,
+      "learning_rate": 4.718485265007777e-07,
+      "loss": 1.2656,
+      "step": 5820
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.283203125,
+      "learning_rate": 4.5539169669563175e-07,
+      "loss": 1.3717,
+      "step": 5830
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.39220281833731e-07,
+      "loss": 1.1762,
+      "step": 5840
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.283203125,
+      "learning_rate": 4.2333476549316746e-07,
+      "loss": 1.24,
+      "step": 5850
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.077356227027329e-07,
+      "loss": 1.2425,
+      "step": 5860
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.9242331992771277e-07,
+      "loss": 1.1669,
+      "step": 5870
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6640625,
+      "learning_rate": 3.773983150559413e-07,
+      "loss": 1.1926,
+      "step": 5880
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.279296875,
+      "learning_rate": 3.626610573841094e-07,
+      "loss": 1.2356,
+      "step": 5890
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.55859375,
+      "learning_rate": 3.4821198760432553e-07,
+      "loss": 1.215,
+      "step": 5900
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.3405153779094144e-07,
+      "loss": 1.1959,
+      "step": 5910
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.296875,
+      "learning_rate": 3.2018013138762514e-07,
+      "loss": 1.2305,
+      "step": 5920
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 3.0659818319470956e-07,
+      "loss": 1.2097,
+      "step": 5930
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.93306099356776e-07,
+      "loss": 1.2443,
+      "step": 5940
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 2.8030427735052155e-07,
+      "loss": 1.2435,
+      "step": 5950
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.675931059728609e-07,
+      "loss": 1.1434,
+      "step": 5960
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 2.5517296532931226e-07,
+      "loss": 1.1753,
+      "step": 5970
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.4304422682262073e-07,
+      "loss": 1.2466,
+      "step": 5980
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.3120725314165758e-07,
+      "loss": 1.149,
+      "step": 5990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.196623982505719e-07,
+      "loss": 1.2448,
+      "step": 6000
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.084100073782136e-07,
+      "loss": 1.1645,
+      "step": 6010
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.97450417007794e-07,
+      "loss": 1.2789,
+      "step": 6020
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.2890625,
+      "learning_rate": 1.8678395486684153e-07,
+      "loss": 1.2842,
+      "step": 6030
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.7641093991739078e-07,
+      "loss": 1.0917,
+      "step": 6040
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 1.663316823464478e-07,
+      "loss": 1.2197,
+      "step": 6050
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.5654648355671432e-07,
+      "loss": 1.1653,
+      "step": 6060
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.279296875,
+      "learning_rate": 1.4705563615757258e-07,
+      "loss": 1.31,
+      "step": 6070
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.609375,
+      "learning_rate": 1.3785942395634132e-07,
+      "loss": 1.159,
+      "step": 6080
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3125,
+      "learning_rate": 1.2895812194978373e-07,
+      "loss": 1.2231,
+      "step": 6090
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.2035199631588278e-07,
+      "loss": 1.1684,
+      "step": 6100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.15625,
+      "learning_rate": 1.1204130440588656e-07,
+      "loss": 1.17,
+      "step": 6110
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.244140625,
+      "learning_rate": 1.040262947366133e-07,
+      "loss": 1.1736,
+      "step": 6120
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.203125,
+      "learning_rate": 9.630720698301177e-08,
+      "loss": 1.2581,
+      "step": 6130
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.66015625,
+      "learning_rate": 8.888427197100258e-08,
+      "loss": 1.1767,
+      "step": 6140
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.175771167057145e-08,
+      "loss": 1.2383,
+      "step": 6150
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.492773918913565e-08,
+      "loss": 1.2379,
+      "step": 6160
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.36328125,
+      "learning_rate": 6.839455876516354e-08,
+      "loss": 1.2352,
+      "step": 6170
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.215836576207834e-08,
+      "loss": 1.1953,
+      "step": 6180
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.67578125,
+      "learning_rate": 5.6219346662402807e-08,
+      "loss": 1.1986,
+      "step": 6190
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.326171875,
+      "learning_rate": 5.057767906219702e-08,
+      "loss": 1.2955,
+      "step": 6200
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.25390625,
+      "learning_rate": 4.5233531665735967e-08,
+      "loss": 1.1783,
+      "step": 6210
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.0187064280471364e-08,
+      "loss": 1.1547,
+      "step": 6220
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.287109375,
+      "learning_rate": 3.543842781224993e-08,
+      "loss": 1.2945,
+      "step": 6230
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.515625,
+      "learning_rate": 3.098776426080252e-08,
+      "loss": 1.336,
+      "step": 6240
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.6835206715497553e-08,
+      "loss": 1.1816,
+      "step": 6250
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 2.298087935135973e-08,
+      "loss": 1.2006,
+      "step": 6260
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.265625,
+      "learning_rate": 1.942489742535747e-08,
+      "loss": 1.1824,
+      "step": 6270
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28515625,
+      "learning_rate": 1.6167367272958977e-08,
+      "loss": 1.1481,
+      "step": 6280
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.320838630494703e-08,
+      "loss": 1.2337,
+      "step": 6290
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.0548043004513508e-08,
+      "loss": 1.2422,
+      "step": 6300
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.186416924605978e-09,
+      "loss": 1.1972,
+      "step": 6310
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.267578125,
+      "learning_rate": 6.1235786855518005e-09,
+      "loss": 1.1195,
+      "step": 6320
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.5078125,
+      "learning_rate": 4.359589972948719e-09,
+      "loss": 1.1938,
+      "step": 6330
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 2.894503535816329e-09,
+      "loss": 1.1405,
+      "step": 6340
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.7283631850228966e-09,
+      "loss": 1.2438,
+      "step": 6350
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2578125,
+      "learning_rate": 8.612037919697447e-10,
+      "loss": 1.1927,
+      "step": 6360
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.32421875,
+      "learning_rate": 2.930512875543023e-10,
+      "loss": 1.3042,
+      "step": 6370
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.51953125,
+      "learning_rate": 2.3922661390729517e-11,
+      "loss": 1.1158,
+      "step": 6380
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6384,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 9.362451321417892e+17,
+  "train_batch_size": 3,
+  "trial_name": null,
+  "trial_params": null
+}