diff --git "a/checkpoint-10000/trainer_state.json" "b/checkpoint-10000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10000/trainer_state.json"
@@ -0,0 +1,14821 @@
+{
+  "best_metric": 0.6908321380615234,
+  "best_model_checkpoint": "saves/starcoder2-7b/lora/sft/checkpoint-10000",
+  "epoch": 1.596169193934557,
+  "eval_steps": 100,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.628385603427887,
+      "learning_rate": 4.999999126897802e-05,
+      "loss": 1.2582,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0855119228363037,
+      "learning_rate": 4.999996507591817e-05,
+      "loss": 0.801,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5689586400985718,
+      "learning_rate": 4.9999921420838745e-05,
+      "loss": 1.067,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0851330757141113,
+      "learning_rate": 4.999986030377024e-05,
+      "loss": 1.2953,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.397479772567749,
+      "learning_rate": 4.999978172475535e-05,
+      "loss": 0.9826,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.344118595123291,
+      "learning_rate": 4.9999685683848954e-05,
+      "loss": 0.9485,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.158163070678711,
+      "learning_rate": 4.9999596278606616e-05,
+      "loss": 0.8103,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.602233648300171,
+      "learning_rate": 4.999946880647276e-05,
+      "loss": 0.8648,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.557242751121521,
+      "learning_rate": 4.999932387266596e-05,
+      "loss": 1.0198,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.36068856716156,
+      "learning_rate": 4.999916147728746e-05,
+      "loss": 0.9367,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3263639211654663,
+      "learning_rate": 4.999898162045068e-05,
+      "loss": 0.9695,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.333601474761963,
+      "learning_rate": 4.999878430228126e-05,
+      "loss": 1.1509,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4753800630569458,
+      "learning_rate": 4.999856952291702e-05,
+      "loss": 1.1461,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5096240043640137,
+      "learning_rate": 4.9998337282507965e-05,
+      "loss": 1.1722,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.189892053604126,
+      "learning_rate": 4.999808758121633e-05,
+      "loss": 1.1834,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9292634725570679,
+      "learning_rate": 4.999782041921651e-05,
+      "loss": 0.9498,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1775777339935303,
+      "learning_rate": 4.9997535796695134e-05,
+      "loss": 0.9346,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6854296922683716,
+      "learning_rate": 4.999723371385099e-05,
+      "loss": 1.119,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4571490287780762,
+      "learning_rate": 4.999691417089507e-05,
+      "loss": 0.8671,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.277044653892517,
+      "learning_rate": 4.999657716805059e-05,
+      "loss": 1.2469,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 0.8478816747665405,
+      "eval_runtime": 96.2736,
+      "eval_samples_per_second": 7.24,
+      "eval_steps_per_second": 7.24,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6687743067741394,
+      "learning_rate": 4.9996222705552933e-05,
+      "loss": 0.735,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3488354682922363,
+      "learning_rate": 4.9995850783649665e-05,
+      "loss": 0.8344,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1043323278427124,
+      "learning_rate": 4.9995461402600593e-05,
+      "loss": 0.8254,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9382895827293396,
+      "learning_rate": 4.9995054562677684e-05,
+      "loss": 0.9179,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2824612855911255,
+      "learning_rate": 4.9994630264165107e-05,
+      "loss": 0.8663,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0491925477981567,
+      "learning_rate": 4.999418850735923e-05,
+      "loss": 0.9247,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3642233610153198,
+      "learning_rate": 4.99937292925686e-05,
+      "loss": 0.8253,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.747757911682129,
+      "learning_rate": 4.9993252620113976e-05,
+      "loss": 1.0245,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.299494981765747,
+      "learning_rate": 4.999275849032832e-05,
+      "loss": 0.8723,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7195830345153809,
+      "learning_rate": 4.999224690355675e-05,
+      "loss": 1.0524,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9922987222671509,
+      "learning_rate": 4.9991717860156616e-05,
+      "loss": 0.9502,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0577458143234253,
+      "learning_rate": 4.9991171360497437e-05,
+      "loss": 1.0115,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0001195669174194,
+      "learning_rate": 4.999060740496093e-05,
+      "loss": 1.1999,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2456804513931274,
+      "learning_rate": 4.999002599394102e-05,
+      "loss": 0.8882,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0445325374603271,
+      "learning_rate": 4.9989427127843814e-05,
+      "loss": 1.0615,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2410887479782104,
+      "learning_rate": 4.9988810807087584e-05,
+      "loss": 1.1068,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8935971260070801,
+      "learning_rate": 4.998817703210285e-05,
+      "loss": 0.6683,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1614488363265991,
+      "learning_rate": 4.9987525803332265e-05,
+      "loss": 0.7446,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9392004013061523,
+      "learning_rate": 4.998685712123072e-05,
+      "loss": 0.7397,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0314444303512573,
+      "learning_rate": 4.9986170986265266e-05,
+      "loss": 1.3584,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 0.8368077278137207,
+      "eval_runtime": 96.5262,
+      "eval_samples_per_second": 7.221,
+      "eval_steps_per_second": 7.221,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8964811563491821,
+      "learning_rate": 4.998546739891516e-05,
+      "loss": 0.9546,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0679796934127808,
+      "learning_rate": 4.998474635967185e-05,
+      "loss": 0.864,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2340985536575317,
+      "learning_rate": 4.998400786903896e-05,
+      "loss": 0.885,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7219617366790771,
+      "learning_rate": 4.9983251927532315e-05,
+      "loss": 1.1069,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1480705738067627,
+      "learning_rate": 4.9982478535679924e-05,
+      "loss": 1.0416,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.515589714050293,
+      "learning_rate": 4.9981687694021996e-05,
+      "loss": 1.1844,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6687963008880615,
+      "learning_rate": 4.998087940311091e-05,
+      "loss": 0.8664,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9256645441055298,
+      "learning_rate": 4.998005366351125e-05,
+      "loss": 1.0125,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2500052452087402,
+      "learning_rate": 4.997921047579978e-05,
+      "loss": 1.1374,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0543216466903687,
+      "learning_rate": 4.9978349840565434e-05,
+      "loss": 0.8502,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3009012937545776,
+      "learning_rate": 4.997747175840937e-05,
+      "loss": 1.0357,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8456661105155945,
+      "learning_rate": 4.997657622994491e-05,
+      "loss": 0.6883,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5856515765190125,
+      "learning_rate": 4.9975663255797555e-05,
+      "loss": 0.7656,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.973818302154541,
+      "learning_rate": 4.997473283660501e-05,
+      "loss": 0.823,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9960187673568726,
+      "learning_rate": 4.997378497301715e-05,
+      "loss": 0.8726,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2900679111480713,
+      "learning_rate": 4.997281966569604e-05,
+      "loss": 0.9781,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.828894853591919,
+      "learning_rate": 4.9971836915315926e-05,
+      "loss": 0.8932,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.239621877670288,
+      "learning_rate": 4.9970836722563256e-05,
+      "loss": 1.2022,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0117149353027344,
+      "learning_rate": 4.996981908813664e-05,
+      "loss": 0.8032,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8861119747161865,
+      "learning_rate": 4.996878401274687e-05,
+      "loss": 1.0651,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.8281473517417908,
+      "eval_runtime": 96.5283,
+      "eval_samples_per_second": 7.221,
+      "eval_steps_per_second": 7.221,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8583046197891235,
+      "learning_rate": 4.996773149711693e-05,
+      "loss": 0.8784,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5717499256134033,
+      "learning_rate": 4.9966661541981984e-05,
+      "loss": 0.8395,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.982342004776001,
+      "learning_rate": 4.9965574148089376e-05,
+      "loss": 0.9869,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9000777006149292,
+      "learning_rate": 4.9964469316198633e-05,
+      "loss": 0.8435,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8733209371566772,
+      "learning_rate": 4.9963347047081464e-05,
+      "loss": 0.7281,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.323739767074585,
+      "learning_rate": 4.9962207341521746e-05,
+      "loss": 1.1013,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7102876901626587,
+      "learning_rate": 4.996105020031554e-05,
+      "loss": 0.8276,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9196123480796814,
+      "learning_rate": 4.995987562427109e-05,
+      "loss": 0.8274,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.210099458694458,
+      "learning_rate": 4.995868361420883e-05,
+      "loss": 1.3257,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8923581838607788,
+      "learning_rate": 4.9957474170961335e-05,
+      "loss": 0.6815,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9576735496520996,
+      "learning_rate": 4.9956247295373396e-05,
+      "loss": 1.23,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3774089813232422,
+      "learning_rate": 4.995500298830196e-05,
+      "loss": 1.0556,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1523677110671997,
+      "learning_rate": 4.995374125061614e-05,
+      "loss": 1.1787,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8310608863830566,
+      "learning_rate": 4.9952462083197246e-05,
+      "loss": 0.8525,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9814196825027466,
+      "learning_rate": 4.9951165486938765e-05,
+      "loss": 0.8522,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9878122210502625,
+      "learning_rate": 4.994985146274633e-05,
+      "loss": 0.6618,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2652586698532104,
+      "learning_rate": 4.994852001153777e-05,
+      "loss": 1.0489,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2940975427627563,
+      "learning_rate": 4.994717113424307e-05,
+      "loss": 1.104,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9636249542236328,
+      "learning_rate": 4.99458048318044e-05,
+      "loss": 0.9228,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8122813105583191,
+      "learning_rate": 4.994442110517611e-05,
+      "loss": 0.9209,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.8184689879417419,
+      "eval_runtime": 96.4572,
+      "eval_samples_per_second": 7.226,
+      "eval_steps_per_second": 7.226,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8742052912712097,
+      "learning_rate": 4.99430199553247e-05,
+      "loss": 0.9608,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5679522752761841,
+      "learning_rate": 4.9941601383228835e-05,
+      "loss": 0.5963,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0234627723693848,
+      "learning_rate": 4.994016538987938e-05,
+      "loss": 0.8642,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8581897616386414,
+      "learning_rate": 4.993871197627934e-05,
+      "loss": 0.8993,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4666485786437988,
+      "learning_rate": 4.9937241143443904e-05,
+      "loss": 0.8565,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1166578531265259,
+      "learning_rate": 4.993575289240041e-05,
+      "loss": 0.881,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.303992748260498,
+      "learning_rate": 4.9934247224188393e-05,
+      "loss": 0.9962,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9011989235877991,
+      "learning_rate": 4.993272413985952e-05,
+      "loss": 0.9316,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8321458101272583,
+      "learning_rate": 4.993118364047764e-05,
+      "loss": 0.7889,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7780352234840393,
+      "learning_rate": 4.992962572711877e-05,
+      "loss": 0.8287,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9090210199356079,
+      "learning_rate": 4.992805040087108e-05,
+      "loss": 0.7018,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8694137334823608,
+      "learning_rate": 4.9926457662834906e-05,
+      "loss": 0.8484,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6327371001243591,
+      "learning_rate": 4.992484751412274e-05,
+      "loss": 0.716,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.200668215751648,
+      "learning_rate": 4.9923219955859254e-05,
+      "loss": 0.9525,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8530198931694031,
+      "learning_rate": 4.9921574989181266e-05,
+      "loss": 0.744,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.168479323387146,
+      "learning_rate": 4.991991261523775e-05,
+      "loss": 0.729,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9499714970588684,
+      "learning_rate": 4.9918232835189834e-05,
+      "loss": 0.7725,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8434467911720276,
+      "learning_rate": 4.991653565021084e-05,
+      "loss": 1.1558,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7665804624557495,
+      "learning_rate": 4.99148210614862e-05,
+      "loss": 1.0208,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5782546401023865,
+      "learning_rate": 4.991308907021353e-05,
+      "loss": 0.8306,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 0.8132078051567078,
+      "eval_runtime": 96.433,
+      "eval_samples_per_second": 7.228,
+      "eval_steps_per_second": 7.228,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0821778774261475,
+      "learning_rate": 4.9911339677602584e-05,
+      "loss": 0.9503,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5409029126167297,
+      "learning_rate": 4.99095728848753e-05,
+      "loss": 0.8586,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9011789560317993,
+      "learning_rate": 4.990778869326575e-05,
+      "loss": 0.7981,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0092263221740723,
+      "learning_rate": 4.990598710402013e-05,
+      "loss": 1.0174,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4362307786941528,
+      "learning_rate": 4.9904168118396844e-05,
+      "loss": 0.8373,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1772639751434326,
+      "learning_rate": 4.9902331737666414e-05,
+      "loss": 0.9599,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9610542058944702,
+      "learning_rate": 4.990047796311151e-05,
+      "loss": 0.6895,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9922348260879517,
+      "learning_rate": 4.989860679602698e-05,
+      "loss": 0.7315,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2409151792526245,
+      "learning_rate": 4.9896718237719785e-05,
+      "loss": 0.8574,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.016333818435669,
+      "learning_rate": 4.9894812289509046e-05,
+      "loss": 1.1248,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9131489396095276,
+      "learning_rate": 4.989288895272604e-05,
+      "loss": 0.9847,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.215469479560852,
+      "learning_rate": 4.989094822871419e-05,
+      "loss": 0.912,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0536105632781982,
+      "learning_rate": 4.988899011882903e-05,
+      "loss": 0.8425,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9705311059951782,
+      "learning_rate": 4.988701462443829e-05,
+      "loss": 0.9385,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2488442659378052,
+      "learning_rate": 4.98850217469218e-05,
+      "loss": 0.7865,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7318600416183472,
+      "learning_rate": 4.988301148767157e-05,
+      "loss": 0.8231,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8247858881950378,
+      "learning_rate": 4.9880983848091704e-05,
+      "loss": 0.8553,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.858172595500946,
+      "learning_rate": 4.987893882959849e-05,
+      "loss": 1.3952,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2286418676376343,
+      "learning_rate": 4.987687643362033e-05,
+      "loss": 0.837,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.034350872039795,
+      "learning_rate": 4.9874796661597765e-05,
+      "loss": 0.9175,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.8063747882843018,
+      "eval_runtime": 96.4224,
+      "eval_samples_per_second": 7.229,
+      "eval_steps_per_second": 7.229,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7192366123199463,
+      "learning_rate": 4.987269951498348e-05,
+      "loss": 0.8563,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2645854949951172,
+      "learning_rate": 4.98705849952423e-05,
+      "loss": 0.6663,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0610381364822388,
+      "learning_rate": 4.9868453103851176e-05,
+      "loss": 0.8452,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8550002574920654,
+      "learning_rate": 4.986630384229919e-05,
+      "loss": 0.8894,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7490519285202026,
+      "learning_rate": 4.986413721208757e-05,
+      "loss": 0.9106,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.557860255241394,
+      "learning_rate": 4.986195321472965e-05,
+      "loss": 0.685,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7450752258300781,
+      "learning_rate": 4.9859751851750934e-05,
+      "loss": 0.8472,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.176376461982727,
+      "learning_rate": 4.985753312468903e-05,
+      "loss": 1.0197,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0625300407409668,
+      "learning_rate": 4.985529703509367e-05,
+      "loss": 0.9685,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8808372616767883,
+      "learning_rate": 4.985304358452672e-05,
+      "loss": 0.8612,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8110201954841614,
+      "learning_rate": 4.985077277456218e-05,
+      "loss": 0.8401,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9364888072013855,
+      "learning_rate": 4.984848460678618e-05,
+      "loss": 0.6197,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0113518238067627,
+      "learning_rate": 4.984617908279694e-05,
+      "loss": 0.9889,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1148868799209595,
+      "learning_rate": 4.984385620420485e-05,
+      "loss": 0.9558,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9506175518035889,
+      "learning_rate": 4.984151597263238e-05,
+      "loss": 0.7323,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0044193267822266,
+      "learning_rate": 4.983915838971415e-05,
+      "loss": 0.7504,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2674214839935303,
+      "learning_rate": 4.9836783457096875e-05,
+      "loss": 1.032,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4945333003997803,
+      "learning_rate": 4.983439117643942e-05,
+      "loss": 1.0359,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9860715866088867,
+      "learning_rate": 4.9831981549412744e-05,
+      "loss": 1.1152,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8287227153778076,
+      "learning_rate": 4.982955457769992e-05,
+      "loss": 0.8157,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.8022791743278503,
+      "eval_runtime": 96.5324,
+      "eval_samples_per_second": 7.22,
+      "eval_steps_per_second": 7.22,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9216273427009583,
+      "learning_rate": 4.9827110262996144e-05,
+      "loss": 0.8395,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7642357349395752,
+      "learning_rate": 4.982464860700874e-05,
+      "loss": 0.8817,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8851175308227539,
+      "learning_rate": 4.982216961145711e-05,
+      "loss": 0.8558,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44226109981536865,
+      "learning_rate": 4.98196732780728e-05,
+      "loss": 0.882,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8005027174949646,
+      "learning_rate": 4.981715960859945e-05,
+      "loss": 0.8835,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7451304793357849,
+      "learning_rate": 4.981462860479281e-05,
+      "loss": 0.8551,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1069347858428955,
+      "learning_rate": 4.9812080268420745e-05,
+      "loss": 0.999,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8892244100570679,
+      "learning_rate": 4.980951460126322e-05,
+      "loss": 1.012,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8935977816581726,
+      "learning_rate": 4.9806931605112305e-05,
+      "loss": 0.9911,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8456961512565613,
+      "learning_rate": 4.9804331281772176e-05,
+      "loss": 0.7595,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.78443443775177,
+      "learning_rate": 4.980171363305911e-05,
+      "loss": 0.8308,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0028038024902344,
+      "learning_rate": 4.979907866080149e-05,
+      "loss": 0.9637,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1801577806472778,
+      "learning_rate": 4.9796426366839786e-05,
+      "loss": 0.6159,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8370681405067444,
+      "learning_rate": 4.979375675302659e-05,
+      "loss": 0.9276,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8605382442474365,
+      "learning_rate": 4.979106982122658e-05,
+      "loss": 1.1077,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7788259387016296,
+      "learning_rate": 4.978836557331652e-05,
+      "loss": 0.8172,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4312686920166016,
+      "learning_rate": 4.978564401118528e-05,
+      "loss": 0.8759,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9109662175178528,
+      "learning_rate": 4.978290513673381e-05,
+      "loss": 0.947,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1819065809249878,
+      "learning_rate": 4.9780148951875195e-05,
+      "loss": 0.7364,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9400575160980225,
+      "learning_rate": 4.977737545853455e-05,
+      "loss": 0.9469,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.7995806932449341,
+      "eval_runtime": 96.5877,
+      "eval_samples_per_second": 7.216,
+      "eval_steps_per_second": 7.216,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.693812370300293,
+      "learning_rate": 4.9774584658649126e-05,
+      "loss": 0.9433,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0892895460128784,
+      "learning_rate": 4.9771776554168234e-05,
+      "loss": 0.7027,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9118362665176392,
+      "learning_rate": 4.976895114705329e-05,
+      "loss": 0.9468,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8032681345939636,
+      "learning_rate": 4.976610843927779e-05,
+      "loss": 0.7927,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.168225646018982,
+      "learning_rate": 4.976324843282732e-05,
+      "loss": 0.9673,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.077602744102478,
+      "learning_rate": 4.976037112969953e-05,
+      "loss": 0.9156,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8643108606338501,
+      "learning_rate": 4.9757476531904165e-05,
+      "loss": 0.6999,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.933397650718689,
+      "learning_rate": 4.975456464146306e-05,
+      "loss": 0.8828,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7036295533180237,
+      "learning_rate": 4.975163546041011e-05,
+      "loss": 0.8709,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5974694490432739,
+      "learning_rate": 4.974868899079128e-05,
+      "loss": 0.7594,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7244943380355835,
+      "learning_rate": 4.974572523466465e-05,
+      "loss": 0.8714,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5783522725105286,
+      "learning_rate": 4.9742744194100345e-05,
+      "loss": 0.8941,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7480617761611938,
+      "learning_rate": 4.973974587118055e-05,
+      "loss": 0.9798,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7548874020576477,
+      "learning_rate": 4.973673026799956e-05,
+      "loss": 0.7767,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7075071930885315,
+      "learning_rate": 4.97336973866637e-05,
+      "loss": 0.7779,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7042987942695618,
+      "learning_rate": 4.97306472292914e-05,
+      "loss": 0.8249,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0242459774017334,
+      "learning_rate": 4.972757979801313e-05,
+      "loss": 0.9223,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6138095259666443,
+      "learning_rate": 4.9724495094971436e-05,
+      "loss": 0.9842,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7905042767524719,
+      "learning_rate": 4.9721393122320925e-05,
+      "loss": 0.8738,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9658048748970032,
+      "learning_rate": 4.9718273882228265e-05,
+      "loss": 0.8872,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.7954564690589905,
+      "eval_runtime": 96.643,
+      "eval_samples_per_second": 7.212,
+      "eval_steps_per_second": 7.212,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8425014019012451,
+      "learning_rate": 4.97151373768722e-05,
+      "loss": 0.778,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5527231693267822,
+      "learning_rate": 4.971198360844351e-05,
+      "loss": 0.8332,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7870334386825562,
+      "learning_rate": 4.9708812579145056e-05,
+      "loss": 0.9265,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9935321807861328,
+      "learning_rate": 4.970562429119173e-05,
+      "loss": 0.7243,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9546892046928406,
+      "learning_rate": 4.970241874681051e-05,
+      "loss": 0.9908,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7340118885040283,
+      "learning_rate": 4.969919594824039e-05,
+      "loss": 0.7932,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.1686015129089355,
+      "learning_rate": 4.9695955897732453e-05,
+      "loss": 0.9842,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9721456170082092,
+      "learning_rate": 4.9692698597549815e-05,
+      "loss": 0.9271,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6477334499359131,
+      "learning_rate": 4.9689424049967623e-05,
+      "loss": 0.934,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0759055614471436,
+      "learning_rate": 4.968613225727311e-05,
+      "loss": 1.0465,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7222158908843994,
+      "learning_rate": 4.968282322176552e-05,
+      "loss": 0.7732,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8591343760490417,
+      "learning_rate": 4.9679496945756155e-05,
+      "loss": 0.9062,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8495111465454102,
+      "learning_rate": 4.967615343156837e-05,
+      "loss": 0.8861,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6847331523895264,
+      "learning_rate": 4.967279268153753e-05,
+      "loss": 0.8001,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.690113365650177,
+      "learning_rate": 4.9669414698011074e-05,
+      "loss": 0.7378,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8349626064300537,
+      "learning_rate": 4.9666019483348456e-05,
+      "loss": 0.7193,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6444108486175537,
+      "learning_rate": 4.966260703992116e-05,
+      "loss": 0.8729,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9515655040740967,
+      "learning_rate": 4.965917737011274e-05,
+      "loss": 0.7532,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8138986229896545,
+      "learning_rate": 4.965573047631873e-05,
+      "loss": 1.0124,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0182080268859863,
+      "learning_rate": 4.9652266360946745e-05,
+      "loss": 0.8842,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.7912728190422058,
+      "eval_runtime": 96.5004,
+      "eval_samples_per_second": 7.223,
+      "eval_steps_per_second": 7.223,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9665297269821167,
+      "learning_rate": 4.96487850264164e-05,
+      "loss": 1.0155,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1356585025787354,
+      "learning_rate": 4.964528647515933e-05,
+      "loss": 0.8705,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5548833608627319,
+      "learning_rate": 4.9641770709619234e-05,
+      "loss": 0.9634,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8028444647789001,
+      "learning_rate": 4.9638237732251794e-05,
+      "loss": 0.8722,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.934234082698822,
+      "learning_rate": 4.9634687545524724e-05,
+      "loss": 0.9731,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7293463349342346,
+      "learning_rate": 4.963112015191778e-05,
+      "loss": 1.0237,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6442769169807434,
+      "learning_rate": 4.962753555392271e-05,
+      "loss": 1.1331,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7877534031867981,
+      "learning_rate": 4.962393375404331e-05,
+      "loss": 1.0737,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5739997625350952,
+      "learning_rate": 4.9620314754795343e-05,
+      "loss": 0.8836,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7318402528762817,
+      "learning_rate": 4.9616678558706634e-05,
+      "loss": 0.9981,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5463365316390991,
+      "learning_rate": 4.961302516831699e-05,
+      "loss": 0.7336,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7839176654815674,
+      "learning_rate": 4.960935458617824e-05,
+      "loss": 1.025,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7076404690742493,
+      "learning_rate": 4.9605666814854225e-05,
+      "loss": 0.833,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.732940673828125,
+      "learning_rate": 4.960196185692077e-05,
+      "loss": 0.5103,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7256388068199158,
+      "learning_rate": 4.959823971496574e-05,
+      "loss": 0.8617,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1714242696762085,
+      "learning_rate": 4.959450039158898e-05,
+      "loss": 1.0345,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5849193930625916,
+      "learning_rate": 4.9590743889402325e-05,
+      "loss": 0.729,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6283109784126282,
+      "learning_rate": 4.958697021102963e-05,
+      "loss": 0.8527,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6387770175933838,
+      "learning_rate": 4.9583179359106746e-05,
+      "loss": 0.7411,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5853758454322815,
+      "learning_rate": 4.957937133628151e-05,
+      "loss": 0.7909,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.7863278985023499,
+      "eval_runtime": 96.3784,
+      "eval_samples_per_second": 7.232,
+      "eval_steps_per_second": 7.232,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9301708936691284,
+      "learning_rate": 4.9575546145213755e-05,
+      "loss": 0.7149,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.125088095664978,
+      "learning_rate": 4.9571703788575314e-05,
+      "loss": 0.8034,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0697988271713257,
+      "learning_rate": 4.956784426905e-05,
+      "loss": 0.8874,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7094873189926147,
+      "learning_rate": 4.956396758933361e-05,
+      "loss": 0.6612,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8048680424690247,
+      "learning_rate": 4.956007375213393e-05,
+      "loss": 0.9558,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8820949196815491,
+      "learning_rate": 4.9556162760170756e-05,
+      "loss": 0.9442,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7214958071708679,
+      "learning_rate": 4.955223461617583e-05,
+      "loss": 0.8392,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8364250063896179,
+      "learning_rate": 4.954828932289288e-05,
+      "loss": 0.9834,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8735854625701904,
+      "learning_rate": 4.954432688307764e-05,
+      "loss": 0.8817,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.810013473033905,
+      "learning_rate": 4.9540347299497805e-05,
+      "loss": 0.7723,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8791002035140991,
+      "learning_rate": 4.953635057493302e-05,
+      "loss": 0.706,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7556783556938171,
+      "learning_rate": 4.953233671217493e-05,
+      "loss": 0.8145,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3251086473464966,
+      "learning_rate": 4.952830571402716e-05,
+      "loss": 0.8413,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8531173467636108,
+      "learning_rate": 4.952425758330527e-05,
+      "loss": 0.8236,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0738744735717773,
+      "learning_rate": 4.952019232283681e-05,
+      "loss": 0.8357,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7908213138580322,
+      "learning_rate": 4.9516109935461306e-05,
+      "loss": 0.6165,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9802565574645996,
+      "learning_rate": 4.951201042403021e-05,
+      "loss": 0.7203,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7866708636283875,
+      "learning_rate": 4.9507893791406974e-05,
+      "loss": 0.8479,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6721138954162598,
+      "learning_rate": 4.950376004046698e-05,
+      "loss": 0.8871,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1981366872787476,
+      "learning_rate": 4.9499609174097574e-05,
+      "loss": 0.8196,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.7843652367591858,
+      "eval_runtime": 96.5411,
+      "eval_samples_per_second": 7.22,
+      "eval_steps_per_second": 7.22,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7013841867446899,
+      "learning_rate": 4.9495441195198064e-05,
+      "loss": 1.0009,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8476290702819824,
+      "learning_rate": 4.949125610667972e-05,
+      "loss": 0.5127,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7680797576904297,
+      "learning_rate": 4.9487053911465735e-05,
+      "loss": 0.7003,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9771925806999207,
+      "learning_rate": 4.948283461249127e-05,
+      "loss": 1.1135,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4247405529022217,
+      "learning_rate": 4.947859821270342e-05,
+      "loss": 0.8253,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.184887409210205,
+      "learning_rate": 4.947434471506125e-05,
+      "loss": 1.1208,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7579745054244995,
+      "learning_rate": 4.9470074122535745e-05,
+      "loss": 1.1363,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8529625535011292,
+      "learning_rate": 4.9465786438109826e-05,
+      "loss": 0.8699,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.810576319694519,
+      "learning_rate": 4.9461481664778374e-05,
+      "loss": 1.0166,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8605110049247742,
+      "learning_rate": 4.9457159805548187e-05,
+      "loss": 0.9427,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.59971684217453,
+      "learning_rate": 4.945282086343801e-05,
+      "loss": 0.6536,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0233818292617798,
+      "learning_rate": 4.9448464841478506e-05,
+      "loss": 0.9505,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8945149779319763,
+      "learning_rate": 4.9444091742712293e-05,
+      "loss": 0.8416,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.702805757522583,
+      "learning_rate": 4.9439701570193886e-05,
+      "loss": 0.9419,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7464181184768677,
+      "learning_rate": 4.9435294326989745e-05,
+      "loss": 0.7972,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1765002012252808,
+      "learning_rate": 4.943175624360097e-05,
+      "loss": 0.9914,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6549853682518005,
+      "learning_rate": 4.9427318280928034e-05,
+      "loss": 0.8924,
+      "step": 1285
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5978650450706482,
+      "learning_rate": 4.942286325621888e-05,
+      "loss": 0.6224,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7752617597579956,
+      "learning_rate": 4.941839117258523e-05,
+      "loss": 0.8666,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6919072866439819,
+      "learning_rate": 4.941390203315078e-05,
+      "loss": 0.9341,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.7824844717979431,
+      "eval_runtime": 96.8874,
+      "eval_samples_per_second": 7.194,
+      "eval_steps_per_second": 7.194,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7222729325294495,
+      "learning_rate": 4.94093958410511e-05,
+      "loss": 0.9925,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9575716853141785,
+      "learning_rate": 4.9404872599433686e-05,
+      "loss": 0.8623,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7721400260925293,
+      "learning_rate": 4.940033231145793e-05,
+      "loss": 1.0061,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7019990682601929,
+      "learning_rate": 4.9395774980295165e-05,
+      "loss": 0.8697,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7828916907310486,
+      "learning_rate": 4.939120060912858e-05,
+      "loss": 1.0066,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0238871574401855,
+      "learning_rate": 4.93866092011533e-05,
+      "loss": 1.0285,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.48669734597206116,
+      "learning_rate": 4.938200075957634e-05,
+      "loss": 0.7454,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8834619522094727,
+      "learning_rate": 4.93773752876166e-05,
+      "loss": 0.9998,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6462609767913818,
+      "learning_rate": 4.9372732788504905e-05,
+      "loss": 0.7278,
+      "step": 1345
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7309257388114929,
+      "learning_rate": 4.936807326548395e-05,
+      "loss": 0.7301,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8515027165412903,
+      "learning_rate": 4.936339672180833e-05,
+      "loss": 0.8307,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.913206934928894,
+      "learning_rate": 4.935870316074451e-05,
+      "loss": 0.9467,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6705841422080994,
+      "learning_rate": 4.935399258557088e-05,
+      "loss": 0.7124,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.676695704460144,
+      "learning_rate": 4.934926499957767e-05,
+      "loss": 0.9318,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0529104471206665,
+      "learning_rate": 4.934452040606703e-05,
+      "loss": 1.0307,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7150225639343262,
+      "learning_rate": 4.933975880835296e-05,
+      "loss": 0.8718,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7180047035217285,
+      "learning_rate": 4.933498020976135e-05,
+      "loss": 0.7515,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0961759090423584,
+      "learning_rate": 4.933018461362997e-05,
+      "loss": 0.8797,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.830609142780304,
+      "learning_rate": 4.9325372023308446e-05,
+      "loss": 0.6927,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5277318358421326,
+      "learning_rate": 4.9320542442158305e-05,
+      "loss": 0.8801,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.7787255644798279,
+      "eval_runtime": 96.8812,
+      "eval_samples_per_second": 7.194,
+      "eval_steps_per_second": 7.194,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3845161199569702,
+      "learning_rate": 4.931569587355289e-05,
+      "loss": 0.8782,
+      "step": 1405
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8579941987991333,
+      "learning_rate": 4.9310832320877476e-05,
+      "loss": 0.713,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2643532454967499,
+      "learning_rate": 4.930595178752914e-05,
+      "loss": 0.9781,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4968445897102356,
+      "learning_rate": 4.930105427691685e-05,
+      "loss": 0.93,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9254417419433594,
+      "learning_rate": 4.929613979246144e-05,
+      "loss": 0.6353,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9814417958259583,
+      "learning_rate": 4.9291208337595574e-05,
+      "loss": 0.9672,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7159338593482971,
+      "learning_rate": 4.928625991576379e-05,
+      "loss": 0.9482,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.623866617679596,
+      "learning_rate": 4.9281294530422476e-05,
+      "loss": 0.623,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8750379681587219,
+      "learning_rate": 4.927631218503985e-05,
+      "loss": 0.772,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5593128800392151,
+      "learning_rate": 4.9271312883096e-05,
+      "loss": 0.6579,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6411569714546204,
+      "learning_rate": 4.9266296628082834e-05,
+      "loss": 0.9239,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9317705631256104,
+      "learning_rate": 4.9261263423504135e-05,
+      "loss": 0.9315,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8312699198722839,
+      "learning_rate": 4.9256213272875486e-05,
+      "loss": 0.7334,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6170663833618164,
+      "learning_rate": 4.925114617972433e-05,
+      "loss": 0.8603,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7176920771598816,
+      "learning_rate": 4.924606214758995e-05,
+      "loss": 0.8738,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8957033157348633,
+      "learning_rate": 4.924096118002343e-05,
+      "loss": 0.8861,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5490685701370239,
+      "learning_rate": 4.923584328058772e-05,
+      "loss": 0.712,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7401763796806335,
+      "learning_rate": 4.923070845285757e-05,
+      "loss": 0.8118,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7380841374397278,
+      "learning_rate": 4.922555670041957e-05,
+      "loss": 0.8476,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0009427070617676,
+      "learning_rate": 4.922038802687212e-05,
+      "loss": 0.9109,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.777683675289154,
+      "eval_runtime": 96.9147,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 7.192,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7970065474510193,
+      "learning_rate": 4.921520243582545e-05,
+      "loss": 0.616,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6530303955078125,
+      "learning_rate": 4.92099999309016e-05,
+      "loss": 0.9223,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.48044708371162415,
+      "learning_rate": 4.9204780515734406e-05,
+      "loss": 0.6762,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7560244798660278,
+      "learning_rate": 4.919954419396956e-05,
+      "loss": 0.8726,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8580659031867981,
+      "learning_rate": 4.919429096926453e-05,
+      "loss": 0.7654,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1246473789215088,
+      "learning_rate": 4.918902084528859e-05,
+      "loss": 0.9123,
+      "step": 1530
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0745307207107544,
+      "learning_rate": 4.918373382572283e-05,
+      "loss": 0.79,
+      "step": 1535
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.9591856598854065,
+      "learning_rate": 4.917842991426014e-05,
+      "loss": 1.1778,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0233389139175415,
+      "learning_rate": 4.91731091146052e-05,
+      "loss": 0.8827,
+      "step": 1545
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.648965060710907,
+      "learning_rate": 4.91677714304745e-05,
+      "loss": 0.8634,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6523327231407166,
+      "learning_rate": 4.91624168655963e-05,
+      "loss": 0.9916,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8029198050498962,
+      "learning_rate": 4.915704542371068e-05,
+      "loss": 0.7867,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6397082805633545,
+      "learning_rate": 4.915165710856948e-05,
+      "loss": 0.7738,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5862845778465271,
+      "learning_rate": 4.914625192393636e-05,
+      "loss": 0.7026,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5333505868911743,
+      "learning_rate": 4.914082987358673e-05,
+      "loss": 0.8623,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5689602494239807,
+      "learning_rate": 4.913539096130779e-05,
+      "loss": 0.7619,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7333836555480957,
+      "learning_rate": 4.912993519089853e-05,
+      "loss": 0.8116,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7610496282577515,
+      "learning_rate": 4.91244625661697e-05,
+      "loss": 0.74,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6331669092178345,
+      "learning_rate": 4.9118973090943835e-05,
+      "loss": 1.0445,
+      "step": 1595
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7263479828834534,
+      "learning_rate": 4.911346676905521e-05,
+      "loss": 0.8964,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.7759388089179993,
+      "eval_runtime": 96.8818,
+      "eval_samples_per_second": 7.194,
+      "eval_steps_per_second": 7.194,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6523721814155579,
+      "learning_rate": 4.910794360434993e-05,
+      "loss": 1.0127,
+      "step": 1605
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.055384874343872,
+      "learning_rate": 4.9102403600685796e-05,
+      "loss": 0.9855,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7640814185142517,
+      "learning_rate": 4.9096846761932414e-05,
+      "loss": 0.7963,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5843799710273743,
+      "learning_rate": 4.9091273091971124e-05,
+      "loss": 0.8854,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9825207591056824,
+      "learning_rate": 4.9085682594695036e-05,
+      "loss": 0.8086,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9490563869476318,
+      "learning_rate": 4.908007527400901e-05,
+      "loss": 0.6838,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9472922682762146,
+      "learning_rate": 4.907445113382966e-05,
+      "loss": 0.8732,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6690593957901001,
+      "learning_rate": 4.9068810178085344e-05,
+      "loss": 0.8551,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7245538830757141,
+      "learning_rate": 4.906315241071616e-05,
+      "loss": 0.7639,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8342815041542053,
+      "learning_rate": 4.905747783567397e-05,
+      "loss": 0.9417,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6241989135742188,
+      "learning_rate": 4.9051786456922354e-05,
+      "loss": 0.9394,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5671687126159668,
+      "learning_rate": 4.904607827843663e-05,
+      "loss": 0.6381,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.795868456363678,
+      "learning_rate": 4.9040353304203864e-05,
+      "loss": 0.7676,
+      "step": 1665
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.9995182156562805,
+      "learning_rate": 4.9034611538222844e-05,
+      "loss": 1.0327,
+      "step": 1670
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7473803758621216,
+      "learning_rate": 4.902885298450409e-05,
+      "loss": 0.8835,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5757468938827515,
+      "learning_rate": 4.902307764706984e-05,
+      "loss": 0.7548,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8357987403869629,
+      "learning_rate": 4.901728552995407e-05,
+      "loss": 0.9184,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6664137244224548,
+      "learning_rate": 4.901147663720247e-05,
+      "loss": 0.9872,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.861997663974762,
+      "learning_rate": 4.900565097287243e-05,
+      "loss": 0.8541,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7566475868225098,
+      "learning_rate": 4.8999808541033086e-05,
+      "loss": 0.9265,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.7741928696632385,
+      "eval_runtime": 96.9038,
+      "eval_samples_per_second": 7.193,
+      "eval_steps_per_second": 7.193,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.45475611090660095,
+      "learning_rate": 4.8993949345765266e-05,
+      "loss": 0.7186,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8672823905944824,
+      "learning_rate": 4.8988073391161515e-05,
+      "loss": 0.919,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7782495617866516,
+      "learning_rate": 4.8982180681326074e-05,
+      "loss": 0.6618,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6640329957008362,
+      "learning_rate": 4.897627122037489e-05,
+      "loss": 0.6662,
+      "step": 1720
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8019454479217529,
+      "learning_rate": 4.897034501243561e-05,
+      "loss": 0.9459,
+      "step": 1725
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8336368799209595,
+      "learning_rate": 4.896440206164761e-05,
+      "loss": 0.8058,
+      "step": 1730
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6316781044006348,
+      "learning_rate": 4.8958442372161906e-05,
+      "loss": 0.9132,
+      "step": 1735
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7768308520317078,
+      "learning_rate": 4.895246594814124e-05,
+      "loss": 0.7512,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9891632795333862,
+      "learning_rate": 4.894647279376002e-05,
+      "loss": 0.843,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6162430047988892,
+      "learning_rate": 4.894046291320439e-05,
+      "loss": 0.8233,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6184887290000916,
+      "learning_rate": 4.893443631067211e-05,
+      "loss": 0.7428,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7117312550544739,
+      "learning_rate": 4.892839299037267e-05,
+      "loss": 0.8707,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7165163159370422,
+      "learning_rate": 4.892233295652721e-05,
+      "loss": 1.0485,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8377657532691956,
+      "learning_rate": 4.891625621336855e-05,
+      "loss": 0.7368,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6349939703941345,
+      "learning_rate": 4.89101627651412e-05,
+      "loss": 0.7357,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.969137191772461,
+      "learning_rate": 4.890405261610131e-05,
+      "loss": 0.7605,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5980018377304077,
+      "learning_rate": 4.889792577051671e-05,
+      "loss": 0.9253,
+      "step": 1785
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.681398332118988,
+      "learning_rate": 4.889178223266688e-05,
+      "loss": 0.7235,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6999421715736389,
+      "learning_rate": 4.888562200684299e-05,
+      "loss": 0.8521,
+      "step": 1795
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7693730592727661,
+      "learning_rate": 4.887944509734783e-05,
+      "loss": 0.8632,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 0.76987224817276,
+      "eval_runtime": 96.9052,
+      "eval_samples_per_second": 7.193,
+      "eval_steps_per_second": 7.193,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7641138434410095,
+      "learning_rate": 4.8873251508495865e-05,
+      "loss": 0.7074,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.732545018196106,
+      "learning_rate": 4.886704124461321e-05,
+      "loss": 0.6901,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0327179431915283,
+      "learning_rate": 4.88608143100376e-05,
+      "loss": 0.8256,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7066757082939148,
+      "learning_rate": 4.885457070911845e-05,
+      "loss": 0.6635,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.809877336025238,
+      "learning_rate": 4.8848310446216806e-05,
+      "loss": 0.795,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.738153338432312,
+      "learning_rate": 4.8842033525705335e-05,
+      "loss": 0.9089,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.754896879196167,
+      "learning_rate": 4.883573995196836e-05,
+      "loss": 0.7103,
+      "step": 1835
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0111182928085327,
+      "learning_rate": 4.8829429729401826e-05,
+      "loss": 1.046,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6233395934104919,
+      "learning_rate": 4.8823102862413306e-05,
+      "loss": 0.761,
+      "step": 1845
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3443419933319092,
+      "learning_rate": 4.8816759355422e-05,
+      "loss": 0.8436,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6685923337936401,
+      "learning_rate": 4.8810399212858736e-05,
+      "loss": 0.8956,
+      "step": 1855
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0405924320220947,
+      "learning_rate": 4.880402243916596e-05,
+      "loss": 1.1458,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8413107991218567,
+      "learning_rate": 4.879762903879772e-05,
+      "loss": 0.8133,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7151504158973694,
+      "learning_rate": 4.8791219016219705e-05,
+      "loss": 0.9207,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6887856125831604,
+      "learning_rate": 4.878479237590918e-05,
+      "loss": 0.8185,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5687748193740845,
+      "learning_rate": 4.877834912235506e-05,
+      "loss": 0.9035,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9966350793838501,
+      "learning_rate": 4.877188926005782e-05,
+      "loss": 0.7764,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0459462404251099,
+      "learning_rate": 4.8765412793529574e-05,
+      "loss": 0.6658,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8338847160339355,
+      "learning_rate": 4.8758919727293995e-05,
+      "loss": 0.7363,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7602768540382385,
+      "learning_rate": 4.875241006588638e-05,
+      "loss": 1.0081,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.7692809700965881,
+      "eval_runtime": 96.4899,
+      "eval_samples_per_second": 7.224,
+      "eval_steps_per_second": 7.224,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5455746054649353,
+      "learning_rate": 4.874588381385362e-05,
+      "loss": 0.7855,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8574795126914978,
+      "learning_rate": 4.8739340975754165e-05,
+      "loss": 1.068,
+      "step": 1910
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.0321904420852661,
+      "learning_rate": 4.873278155615808e-05,
+      "loss": 0.8239,
+      "step": 1915
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2484744787216187,
+      "learning_rate": 4.8726205559646996e-05,
+      "loss": 0.9307,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7140147686004639,
+      "learning_rate": 4.871961299081412e-05,
+      "loss": 0.9876,
+      "step": 1925
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8003590106964111,
+      "learning_rate": 4.871300385426426e-05,
+      "loss": 0.8615,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7282931208610535,
+      "learning_rate": 4.870637815461376e-05,
+      "loss": 0.8734,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6800629496574402,
+      "learning_rate": 4.869973589649055e-05,
+      "loss": 0.7718,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8813210129737854,
+      "learning_rate": 4.869307708453413e-05,
+      "loss": 0.7943,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6612805724143982,
+      "learning_rate": 4.868640172339557e-05,
+      "loss": 0.6807,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.653191328048706,
+      "learning_rate": 4.867970981773748e-05,
+      "loss": 0.8948,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7479822635650635,
+      "learning_rate": 4.8673001372234025e-05,
+      "loss": 0.8583,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": NaN,
+      "learning_rate": 4.8667622710291026e-05,
+      "loss": 0.7443,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5788535475730896,
+      "learning_rate": 4.866088450488172e-05,
+      "loss": 0.7249,
+      "step": 1970
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7408040165901184,
+      "learning_rate": 4.86541297727762e-05,
+      "loss": 0.7115,
+      "step": 1975
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6549968719482422,
+      "learning_rate": 4.864735851869251e-05,
+      "loss": 0.9095,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4595119059085846,
+      "learning_rate": 4.864057074736026e-05,
+      "loss": 1.2808,
+      "step": 1985
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5746715068817139,
+      "learning_rate": 4.863376646352058e-05,
+      "loss": 0.8139,
+      "step": 1990
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6972643136978149,
+      "learning_rate": 4.862694567192614e-05,
+      "loss": 0.9797,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6935243010520935,
+      "learning_rate": 4.8620108377341124e-05,
+      "loss": 0.7651,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.766412615776062,
+      "eval_runtime": 96.4555,
+      "eval_samples_per_second": 7.226,
+      "eval_steps_per_second": 7.226,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9983006715774536,
+      "learning_rate": 4.861325458454128e-05,
+      "loss": 0.8256,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6732650995254517,
+      "learning_rate": 4.860638429831384e-05,
+      "loss": 0.8136,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6780042052268982,
+      "learning_rate": 4.859949752345758e-05,
+      "loss": 0.8911,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9892123937606812,
+      "learning_rate": 4.8592594264782794e-05,
+      "loss": 0.7907,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9327254295349121,
+      "learning_rate": 4.8585674527111266e-05,
+      "loss": 0.8712,
+      "step": 2025
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0295612812042236,
+      "learning_rate": 4.857873831527632e-05,
+      "loss": 0.9188,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3071186542510986,
+      "learning_rate": 4.8571785634122766e-05,
+      "loss": 0.8801,
+      "step": 2035
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9625150561332703,
+      "learning_rate": 4.856481648850694e-05,
+      "loss": 0.8333,
+      "step": 2040
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6674854159355164,
+      "learning_rate": 4.855783088329664e-05,
+      "loss": 1.0388,
+      "step": 2045
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5447000861167908,
+      "learning_rate": 4.8550828823371196e-05,
+      "loss": 0.7893,
+      "step": 2050
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9970148801803589,
+      "learning_rate": 4.854381031362142e-05,
+      "loss": 0.8198,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7657136917114258,
+      "learning_rate": 4.853677535894961e-05,
+      "loss": 0.5977,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4694065451622009,
+      "learning_rate": 4.852972396426956e-05,
+      "loss": 0.5965,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8955700993537903,
+      "learning_rate": 4.852265613450653e-05,
+      "loss": 0.6938,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9884099960327148,
+      "learning_rate": 4.851557187459727e-05,
+      "loss": 0.8946,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6793637871742249,
+      "learning_rate": 4.850847118949002e-05,
+      "loss": 0.841,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7438017725944519,
+      "learning_rate": 4.850135408414447e-05,
+      "loss": 0.8843,
+      "step": 2085
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7632609009742737,
+      "learning_rate": 4.849422056353178e-05,
+      "loss": 0.8263,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7281492352485657,
+      "learning_rate": 4.84870706326346e-05,
+      "loss": 0.8989,
+      "step": 2095
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6480591893196106,
+      "learning_rate": 4.847990429644702e-05,
+      "loss": 1.0037,
+      "step": 2100
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.7653521299362183,
+      "eval_runtime": 96.4452,
+      "eval_samples_per_second": 7.227,
+      "eval_steps_per_second": 7.227,
+      "step": 2100
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5578673481941223,
+      "learning_rate": 4.8472721559974584e-05,
+      "loss": 0.911,
+      "step": 2105
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5615595579147339,
+      "learning_rate": 4.846552242823433e-05,
+      "loss": 0.6938,
+      "step": 2110
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.588246762752533,
+      "learning_rate": 4.845830690625469e-05,
+      "loss": 0.7898,
+      "step": 2115
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8140611052513123,
+      "learning_rate": 4.8451074999075595e-05,
+      "loss": 0.7702,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9400056600570679,
+      "learning_rate": 4.8443826711748385e-05,
+      "loss": 0.7959,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7187873721122742,
+      "learning_rate": 4.8436562049335874e-05,
+      "loss": 0.7223,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7627830505371094,
+      "learning_rate": 4.8429281016912275e-05,
+      "loss": 0.793,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6755004525184631,
+      "learning_rate": 4.842198361956328e-05,
+      "loss": 0.7665,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6032254695892334,
+      "learning_rate": 4.8414669862385966e-05,
+      "loss": 0.7952,
+      "step": 2145
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8377916216850281,
+      "learning_rate": 4.840733975048887e-05,
+      "loss": 1.0016,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7361429929733276,
+      "learning_rate": 4.839999328899194e-05,
+      "loss": 0.8773,
+      "step": 2155
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8006517887115479,
+      "learning_rate": 4.8392630483026546e-05,
+      "loss": 0.9334,
+      "step": 2160
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9716467261314392,
+      "learning_rate": 4.8385251337735473e-05,
+      "loss": 1.0359,
+      "step": 2165
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6826418042182922,
+      "learning_rate": 4.8377855858272925e-05,
+      "loss": 0.6841,
+      "step": 2170
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4519975781440735,
+      "learning_rate": 4.8370444049804494e-05,
+      "loss": 0.8326,
+      "step": 2175
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.677891731262207,
+      "learning_rate": 4.836301591750721e-05,
+      "loss": 1.0841,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5161852836608887,
+      "learning_rate": 4.835557146656948e-05,
+      "loss": 0.8701,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6586780548095703,
+      "learning_rate": 4.834811070219112e-05,
+      "loss": 0.8261,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.48046165704727173,
+      "learning_rate": 4.834063362958333e-05,
+      "loss": 0.6375,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0315968990325928,
+      "learning_rate": 4.833314025396872e-05,
+      "loss": 0.8768,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.7641988396644592,
+      "eval_runtime": 96.3923,
+      "eval_samples_per_second": 7.231,
+      "eval_steps_per_second": 7.231,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7704123258590698,
+      "learning_rate": 4.8325630580581263e-05,
+      "loss": 0.8849,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.087425708770752,
+      "learning_rate": 4.831810461466634e-05,
+      "loss": 0.9828,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4766077995300293,
+      "learning_rate": 4.83105623614807e-05,
+      "loss": 0.7103,
+      "step": 2215
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6079148054122925,
+      "learning_rate": 4.830300382629247e-05,
+      "loss": 0.7253,
+      "step": 2220
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6767585873603821,
+      "learning_rate": 4.829542901438115e-05,
+      "loss": 0.7852,
+      "step": 2225
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7065784335136414,
+      "learning_rate": 4.8287837931037585e-05,
+      "loss": 0.8047,
+      "step": 2230
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8305274248123169,
+      "learning_rate": 4.828023058156404e-05,
+      "loss": 0.7912,
+      "step": 2235
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8435990810394287,
+      "learning_rate": 4.827260697127409e-05,
+      "loss": 0.826,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8484389185905457,
+      "learning_rate": 4.8264967105492705e-05,
+      "loss": 0.706,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7461299300193787,
+      "learning_rate": 4.825731098955617e-05,
+      "loss": 0.763,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7928741574287415,
+      "learning_rate": 4.824963862881216e-05,
+      "loss": 0.8125,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7152695059776306,
+      "learning_rate": 4.824195002861968e-05,
+      "loss": 1.129,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8594226241111755,
+      "learning_rate": 4.8234245194349056e-05,
+      "loss": 0.8873,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9760085940361023,
+      "learning_rate": 4.822652413138199e-05,
+      "loss": 0.9713,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7297483682632446,
+      "learning_rate": 4.8218786845111505e-05,
+      "loss": 0.6953,
+      "step": 2275
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8251492381095886,
+      "learning_rate": 4.8211033340941956e-05,
+      "loss": 0.7649,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.742917537689209,
+      "learning_rate": 4.820326362428901e-05,
+      "loss": 0.9756,
+      "step": 2285
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7784115076065063,
+      "learning_rate": 4.819547770057969e-05,
+      "loss": 0.6937,
+      "step": 2290
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.782772183418274,
+      "learning_rate": 4.8187675575252314e-05,
+      "loss": 0.9062,
+      "step": 2295
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7802585363388062,
+      "learning_rate": 4.8179857253756514e-05,
+      "loss": 0.8052,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.7618402242660522,
+      "eval_runtime": 96.4079,
+      "eval_samples_per_second": 7.23,
+      "eval_steps_per_second": 7.23,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.799985945224762,
+      "learning_rate": 4.8172022741553255e-05,
+      "loss": 0.9046,
+      "step": 2305
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.026978850364685,
+      "learning_rate": 4.816417204411481e-05,
+      "loss": 0.7195,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8067365884780884,
+      "learning_rate": 4.8156305166924734e-05,
+      "loss": 0.8193,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.247164249420166,
+      "learning_rate": 4.81484221154779e-05,
+      "loss": 0.6138,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8662647604942322,
+      "learning_rate": 4.814052289528047e-05,
+      "loss": 0.7763,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9020537734031677,
+      "learning_rate": 4.813260751184992e-05,
+      "loss": 0.9236,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6113781929016113,
+      "learning_rate": 4.812467597071499e-05,
+      "loss": 0.8753,
+      "step": 2335
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6988622546195984,
+      "learning_rate": 4.811672827741572e-05,
+      "loss": 0.6747,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9095928072929382,
+      "learning_rate": 4.810876443750344e-05,
+      "loss": 1.0578,
+      "step": 2345
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.643699049949646,
+      "learning_rate": 4.8100784456540724e-05,
+      "loss": 0.8177,
+      "step": 2350
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7084022760391235,
+      "learning_rate": 4.809278834010146e-05,
+      "loss": 0.9345,
+      "step": 2355
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5328305959701538,
+      "learning_rate": 4.808477609377078e-05,
+      "loss": 0.6781,
+      "step": 2360
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8238436579704285,
+      "learning_rate": 4.80767477231451e-05,
+      "loss": 0.7306,
+      "step": 2365
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.0184216499328613,
+      "learning_rate": 4.806870323383208e-05,
+      "loss": 1.0288,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8620426654815674,
+      "learning_rate": 4.806064263145066e-05,
+      "loss": 0.7925,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6541377305984497,
+      "learning_rate": 4.805256592163102e-05,
+      "loss": 0.8629,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8664489984512329,
+      "learning_rate": 4.8044473110014594e-05,
+      "loss": 0.8184,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7283564209938049,
+      "learning_rate": 4.803636420225406e-05,
+      "loss": 0.9444,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7168800234794617,
+      "learning_rate": 4.802823920401335e-05,
+      "loss": 0.8118,
+      "step": 2395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8198531866073608,
+      "learning_rate": 4.802009812096762e-05,
+      "loss": 0.7271,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.7595117688179016,
+      "eval_runtime": 96.4847,
+      "eval_samples_per_second": 7.224,
+      "eval_steps_per_second": 7.224,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5693966150283813,
+      "learning_rate": 4.801194095880327e-05,
+      "loss": 0.7801,
+      "step": 2405
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7175332307815552,
+      "learning_rate": 4.800376772321793e-05,
+      "loss": 0.7873,
+      "step": 2410
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7779633402824402,
+      "learning_rate": 4.799557841992046e-05,
+      "loss": 0.894,
+      "step": 2415
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7832231521606445,
+      "learning_rate": 4.798737305463092e-05,
+      "loss": 0.8035,
+      "step": 2420
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5115272998809814,
+      "learning_rate": 4.797915163308064e-05,
+      "loss": 0.8885,
+      "step": 2425
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9534878730773926,
+      "learning_rate": 4.79709141610121e-05,
+      "loss": 0.8175,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7053850889205933,
+      "learning_rate": 4.796266064417905e-05,
+      "loss": 0.6971,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.236257791519165,
+      "learning_rate": 4.795439108834641e-05,
+      "loss": 1.0832,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6936543583869934,
+      "learning_rate": 4.794610549929031e-05,
+      "loss": 0.858,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8064691424369812,
+      "learning_rate": 4.793780388279809e-05,
+      "loss": 0.6951,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7180449962615967,
+      "learning_rate": 4.792948624466827e-05,
+      "loss": 0.6779,
+      "step": 2455
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6903377175331116,
+      "learning_rate": 4.792115259071058e-05,
+      "loss": 0.8281,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9112733006477356,
+      "learning_rate": 4.791280292674591e-05,
+      "loss": 0.938,
+      "step": 2465
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8657469153404236,
+      "learning_rate": 4.790443725860636e-05,
+      "loss": 0.8063,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9260883927345276,
+      "learning_rate": 4.7896055592135194e-05,
+      "loss": 1.0093,
+      "step": 2475
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7651245594024658,
+      "learning_rate": 4.788765793318685e-05,
+      "loss": 0.6686,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6063816547393799,
+      "learning_rate": 4.7879244287626945e-05,
+      "loss": 0.8516,
+      "step": 2485
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9127621650695801,
+      "learning_rate": 4.787081466133225e-05,
+      "loss": 0.7992,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.061246633529663,
+      "learning_rate": 4.7862369060190716e-05,
+      "loss": 0.8232,
+      "step": 2495
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7100695967674255,
+      "learning_rate": 4.785390749010143e-05,
+      "loss": 0.9615,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.7581596970558167,
+      "eval_runtime": 96.5797,
+      "eval_samples_per_second": 7.217,
+      "eval_steps_per_second": 7.217,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 16.361513137817383,
+      "learning_rate": 4.784542995697464e-05,
+      "loss": 0.7725,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7746205925941467,
+      "learning_rate": 4.7836936466731764e-05,
+      "loss": 0.8464,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7703484892845154,
+      "learning_rate": 4.7828427025305345e-05,
+      "loss": 0.8596,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7838412523269653,
+      "learning_rate": 4.7819901638639066e-05,
+      "loss": 0.666,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5832842588424683,
+      "learning_rate": 4.781136031268776e-05,
+      "loss": 0.4995,
+      "step": 2525
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.798271894454956,
+      "learning_rate": 4.780280305341739e-05,
+      "loss": 1.0017,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.463828444480896,
+      "learning_rate": 4.779422986680503e-05,
+      "loss": 0.5894,
+      "step": 2535
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.761908233165741,
+      "learning_rate": 4.7785640758838916e-05,
+      "loss": 0.9198,
+      "step": 2540
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8427887558937073,
+      "learning_rate": 4.777703573551837e-05,
+      "loss": 0.8572,
+      "step": 2545
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6188894510269165,
+      "learning_rate": 4.776841480285384e-05,
+      "loss": 0.9102,
+      "step": 2550
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7198623418807983,
+      "learning_rate": 4.775977796686691e-05,
+      "loss": 0.8472,
+      "step": 2555
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0144587755203247,
+      "learning_rate": 4.775112523359023e-05,
+      "loss": 0.7059,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9784219861030579,
+      "learning_rate": 4.77424566090676e-05,
+      "loss": 0.7417,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5349156856536865,
+      "learning_rate": 4.773377209935387e-05,
+      "loss": 0.7287,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7715370655059814,
+      "learning_rate": 4.772507171051502e-05,
+      "loss": 0.8393,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8483054637908936,
+      "learning_rate": 4.771635544862813e-05,
+      "loss": 0.8938,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8196272253990173,
+      "learning_rate": 4.770762331978132e-05,
+      "loss": 0.8321,
+      "step": 2585
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6155353784561157,
+      "learning_rate": 4.769887533007384e-05,
+      "loss": 0.9291,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8897277116775513,
+      "learning_rate": 4.769011148561601e-05,
+      "loss": 0.7098,
+      "step": 2595
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2256160974502563,
+      "learning_rate": 4.768133179252921e-05,
+      "loss": 0.8284,
+      "step": 2600
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.7554901838302612,
+      "eval_runtime": 96.5279,
+      "eval_samples_per_second": 7.221,
+      "eval_steps_per_second": 7.221,
+      "step": 2600
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6943432688713074,
+      "learning_rate": 4.767253625694588e-05,
+      "loss": 0.8785,
+      "step": 2605
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6707726120948792,
+      "learning_rate": 4.7663724885009556e-05,
+      "loss": 0.7949,
+      "step": 2610
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5595915913581848,
+      "learning_rate": 4.765489768287481e-05,
+      "loss": 0.8796,
+      "step": 2615
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9889727234840393,
+      "learning_rate": 4.7646054656707306e-05,
+      "loss": 1.0676,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8624396324157715,
+      "learning_rate": 4.763719581268371e-05,
+      "loss": 0.709,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7466241121292114,
+      "learning_rate": 4.7628321156991767e-05,
+      "loss": 0.8084,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6439360976219177,
+      "learning_rate": 4.761943069583027e-05,
+      "loss": 0.8831,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9999917149543762,
+      "learning_rate": 4.761052443540904e-05,
+      "loss": 0.6372,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.688369870185852,
+      "learning_rate": 4.760160238194894e-05,
+      "loss": 0.7938,
+      "step": 2645
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6920734643936157,
+      "learning_rate": 4.759266454168186e-05,
+      "loss": 0.7378,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7592100501060486,
+      "learning_rate": 4.758371092085073e-05,
+      "loss": 1.097,
+      "step": 2655
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9243403077125549,
+      "learning_rate": 4.757474152570946e-05,
+      "loss": 1.0404,
+      "step": 2660
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8212980031967163,
+      "learning_rate": 4.756575636252304e-05,
+      "loss": 0.6179,
+      "step": 2665
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6905696392059326,
+      "learning_rate": 4.755675543756744e-05,
+      "loss": 0.8398,
+      "step": 2670
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8420882821083069,
+      "learning_rate": 4.754773875712961e-05,
+      "loss": 0.7552,
+      "step": 2675
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6216087341308594,
+      "learning_rate": 4.7538706327507575e-05,
+      "loss": 0.8345,
+      "step": 2680
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7430551648139954,
+      "learning_rate": 4.75296581550103e-05,
+      "loss": 0.8277,
+      "step": 2685
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7866222262382507,
+      "learning_rate": 4.752059424595778e-05,
+      "loss": 0.9178,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6548468470573425,
+      "learning_rate": 4.7511514606680985e-05,
+      "loss": 0.745,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6956586837768555,
+      "learning_rate": 4.750241924352187e-05,
+      "loss": 0.8631,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.7539612650871277,
+      "eval_runtime": 96.4433,
+      "eval_samples_per_second": 7.227,
+      "eval_steps_per_second": 7.227,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6508235335350037,
+      "learning_rate": 4.7493308162833394e-05,
+      "loss": 0.9936,
+      "step": 2705
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8658422827720642,
+      "learning_rate": 4.7484181370979475e-05,
+      "loss": 0.8,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9571516513824463,
+      "learning_rate": 4.747503887433501e-05,
+      "loss": 0.7028,
+      "step": 2715
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7693742513656616,
+      "learning_rate": 4.7465880679285866e-05,
+      "loss": 0.7194,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.34340238571167,
+      "learning_rate": 4.745670679222888e-05,
+      "loss": 1.0445,
+      "step": 2725
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.71327805519104,
+      "learning_rate": 4.7447517219571834e-05,
+      "loss": 0.8088,
+      "step": 2730
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9449920058250427,
+      "learning_rate": 4.743831196773349e-05,
+      "loss": 0.7939,
+      "step": 2735
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8091790676116943,
+      "learning_rate": 4.742909104314353e-05,
+      "loss": 0.7816,
+      "step": 2740
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5790795087814331,
+      "learning_rate": 4.741985445224263e-05,
+      "loss": 0.8778,
+      "step": 2745
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1936956644058228,
+      "learning_rate": 4.741060220148236e-05,
+      "loss": 1.0242,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5158389806747437,
+      "learning_rate": 4.7401334297325244e-05,
+      "loss": 0.7954,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8950900435447693,
+      "learning_rate": 4.7392050746244754e-05,
+      "loss": 0.7603,
+      "step": 2760
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7289401888847351,
+      "learning_rate": 4.738275155472528e-05,
+      "loss": 0.879,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8410510420799255,
+      "learning_rate": 4.7373436729262145e-05,
+      "loss": 0.7399,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7992503643035889,
+      "learning_rate": 4.736410627636156e-05,
+      "loss": 0.6779,
+      "step": 2775
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6706194281578064,
+      "learning_rate": 4.73547602025407e-05,
+      "loss": 0.7878,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7177903652191162,
+      "learning_rate": 4.734539851432763e-05,
+      "loss": 0.6958,
+      "step": 2785
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6557692885398865,
+      "learning_rate": 4.73360212182613e-05,
+      "loss": 0.6695,
+      "step": 2790
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6754157543182373,
+      "learning_rate": 4.7326628320891586e-05,
+      "loss": 0.9057,
+      "step": 2795
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.1403777599334717,
+      "learning_rate": 4.731721982877926e-05,
+      "loss": 1.0507,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.7518497705459595,
+      "eval_runtime": 96.4525,
+      "eval_samples_per_second": 7.226,
+      "eval_steps_per_second": 7.226,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8268899321556091,
+      "learning_rate": 4.730779574849598e-05,
+      "loss": 0.7375,
+      "step": 2805
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5358712673187256,
+      "learning_rate": 4.72983560866243e-05,
+      "loss": 0.7839,
+      "step": 2810
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.0761948823928833,
+      "learning_rate": 4.7288900849757636e-05,
+      "loss": 0.7936,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7037429213523865,
+      "learning_rate": 4.7279430044500315e-05,
+      "loss": 0.6875,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6378889679908752,
+      "learning_rate": 4.726994367746751e-05,
+      "loss": 0.9209,
+      "step": 2825
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5508277416229248,
+      "learning_rate": 4.7260441755285284e-05,
+      "loss": 0.9402,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9046247005462646,
+      "learning_rate": 4.725092428459055e-05,
+      "loss": 0.6336,
+      "step": 2835
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8689594864845276,
+      "learning_rate": 4.7241391272031096e-05,
+      "loss": 1.1281,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8785949945449829,
+      "learning_rate": 4.723184272426555e-05,
+      "loss": 0.711,
+      "step": 2845
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9959015250205994,
+      "learning_rate": 4.722227864796339e-05,
+      "loss": 0.7432,
+      "step": 2850
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6438590884208679,
+      "learning_rate": 4.721269904980497e-05,
+      "loss": 0.883,
+      "step": 2855
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6714455485343933,
+      "learning_rate": 4.720310393648145e-05,
+      "loss": 1.065,
+      "step": 2860
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7378780245780945,
+      "learning_rate": 4.7193493314694846e-05,
+      "loss": 0.5352,
+      "step": 2865
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7698020935058594,
+      "learning_rate": 4.7183867191158006e-05,
+      "loss": 0.7016,
+      "step": 2870
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.952795684337616,
+      "learning_rate": 4.7174225572594586e-05,
+      "loss": 1.0659,
+      "step": 2875
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6401458978652954,
+      "learning_rate": 4.71645684657391e-05,
+      "loss": 0.7335,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8375076055526733,
+      "learning_rate": 4.715489587733685e-05,
+      "loss": 0.9264,
+      "step": 2885
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.693505048751831,
+      "learning_rate": 4.714520781414397e-05,
+      "loss": 1.0286,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0239859819412231,
+      "learning_rate": 4.7135504282927375e-05,
+      "loss": 0.6875,
+      "step": 2895
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.602035403251648,
+      "learning_rate": 4.712578529046483e-05,
+      "loss": 0.8247,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.7512397766113281,
+      "eval_runtime": 96.4745,
+      "eval_samples_per_second": 7.225,
+      "eval_steps_per_second": 7.225,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6859713196754456,
+      "learning_rate": 4.711605084354487e-05,
+      "loss": 0.7521,
+      "step": 2905
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7126486301422119,
+      "learning_rate": 4.7106300948966817e-05,
+      "loss": 0.7656,
+      "step": 2910
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4363511800765991,
+      "learning_rate": 4.70965356135408e-05,
+      "loss": 1.1595,
+      "step": 2915
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6381859183311462,
+      "learning_rate": 4.7086754844087724e-05,
+      "loss": 0.6949,
+      "step": 2920
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7931796312332153,
+      "learning_rate": 4.7076958647439284e-05,
+      "loss": 1.0821,
+      "step": 2925
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9333865642547607,
+      "learning_rate": 4.706714703043795e-05,
+      "loss": 0.7753,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8860915899276733,
+      "learning_rate": 4.705731999993694e-05,
+      "loss": 0.7257,
+      "step": 2935
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6868377327919006,
+      "learning_rate": 4.704747756280027e-05,
+      "loss": 0.8148,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5337914228439331,
+      "learning_rate": 4.7037619725902706e-05,
+      "loss": 0.7379,
+      "step": 2945
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4664730429649353,
+      "learning_rate": 4.7027746496129745e-05,
+      "loss": 0.6226,
+      "step": 2950
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7305762767791748,
+      "learning_rate": 4.701785788037768e-05,
+      "loss": 0.9018,
+      "step": 2955
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6576158404350281,
+      "learning_rate": 4.7007953885553525e-05,
+      "loss": 0.7777,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9728206396102905,
+      "learning_rate": 4.699803451857503e-05,
+      "loss": 0.8004,
+      "step": 2965
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6211077570915222,
+      "learning_rate": 4.69880997863707e-05,
+      "loss": 0.7407,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2564159631729126,
+      "learning_rate": 4.697814969587976e-05,
+      "loss": 0.7993,
+      "step": 2975
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.927930474281311,
+      "learning_rate": 4.696818425405217e-05,
+      "loss": 0.8803,
+      "step": 2980
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9062425494194031,
+      "learning_rate": 4.695820346784861e-05,
+      "loss": 0.8835,
+      "step": 2985
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6738875508308411,
+      "learning_rate": 4.694820734424047e-05,
+      "loss": 0.7817,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.326353669166565,
+      "learning_rate": 4.6938195890209866e-05,
+      "loss": 0.9213,
+      "step": 2995
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4853856563568115,
+      "learning_rate": 4.692816911274962e-05,
+      "loss": 0.9835,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.7496011257171631,
+      "eval_runtime": 96.515,
+      "eval_samples_per_second": 7.222,
+      "eval_steps_per_second": 7.222,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5411309003829956,
+      "learning_rate": 4.691812701886324e-05,
+      "loss": 0.7556,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7545793652534485,
+      "learning_rate": 4.6908069615564966e-05,
+      "loss": 0.8295,
+      "step": 3010
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.850104808807373,
+      "learning_rate": 4.6897996909879695e-05,
+      "loss": 1.0194,
+      "step": 3015
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.69708651304245,
+      "learning_rate": 4.6887908908843026e-05,
+      "loss": 0.7918,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1333253383636475,
+      "learning_rate": 4.687780561950126e-05,
+      "loss": 0.7287,
+      "step": 3025
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9223487973213196,
+      "learning_rate": 4.686768704891134e-05,
+      "loss": 0.9592,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7700949311256409,
+      "learning_rate": 4.685755320414091e-05,
+      "loss": 0.8572,
+      "step": 3035
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5573208332061768,
+      "learning_rate": 4.684740409226829e-05,
+      "loss": 0.9441,
+      "step": 3040
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6346720457077026,
+      "learning_rate": 4.6837239720382426e-05,
+      "loss": 0.8398,
+      "step": 3045
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8065741062164307,
+      "learning_rate": 4.682706009558297e-05,
+      "loss": 0.9325,
+      "step": 3050
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3001660406589508,
+      "learning_rate": 4.681686522498018e-05,
+      "loss": 0.8997,
+      "step": 3055
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.860211968421936,
+      "learning_rate": 4.680665511569501e-05,
+      "loss": 0.6883,
+      "step": 3060
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.722518265247345,
+      "learning_rate": 4.6796429774859015e-05,
+      "loss": 0.8607,
+      "step": 3065
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6525880694389343,
+      "learning_rate": 4.678618920961442e-05,
+      "loss": 0.9256,
+      "step": 3070
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7581719756126404,
+      "learning_rate": 4.6775933427114084e-05,
+      "loss": 0.662,
+      "step": 3075
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6604760885238647,
+      "learning_rate": 4.676566243452146e-05,
+      "loss": 0.734,
+      "step": 3080
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7573785781860352,
+      "learning_rate": 4.6755376239010665e-05,
+      "loss": 0.7113,
+      "step": 3085
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8933848738670349,
+      "learning_rate": 4.674507484776641e-05,
+      "loss": 0.8523,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5941946506500244,
+      "learning_rate": 4.6734758267984044e-05,
+      "loss": 0.7907,
+      "step": 3095
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7756261825561523,
+      "learning_rate": 4.672442650686949e-05,
+      "loss": 0.8407,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.7495761513710022,
+      "eval_runtime": 96.4482,
+      "eval_samples_per_second": 7.227,
+      "eval_steps_per_second": 7.227,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6407367587089539,
+      "learning_rate": 4.671407957163931e-05,
+      "loss": 0.6413,
+      "step": 3105
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.069754719734192,
+      "learning_rate": 4.670371746952063e-05,
+      "loss": 0.8934,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9698624610900879,
+      "learning_rate": 4.669334020775122e-05,
+      "loss": 0.7261,
+      "step": 3115
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6487118005752563,
+      "learning_rate": 4.668294779357938e-05,
+      "loss": 0.8951,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0640240907669067,
+      "learning_rate": 4.667254023426404e-05,
+      "loss": 0.8568,
+      "step": 3125
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5772892236709595,
+      "learning_rate": 4.666211753707468e-05,
+      "loss": 0.9798,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6915898323059082,
+      "learning_rate": 4.665167970929137e-05,
+      "loss": 0.8694,
+      "step": 3135
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5959879159927368,
+      "learning_rate": 4.664122675820474e-05,
+      "loss": 0.6521,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.833991289138794,
+      "learning_rate": 4.663075869111597e-05,
+      "loss": 0.9194,
+      "step": 3145
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9575549960136414,
+      "learning_rate": 4.662027551533685e-05,
+      "loss": 1.0088,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5501818656921387,
+      "learning_rate": 4.660977723818965e-05,
+      "loss": 0.5997,
+      "step": 3155
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6001989245414734,
+      "learning_rate": 4.659926386700725e-05,
+      "loss": 0.7643,
+      "step": 3160
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6806654930114746,
+      "learning_rate": 4.658873540913303e-05,
+      "loss": 0.899,
+      "step": 3165
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7098959684371948,
+      "learning_rate": 4.657819187192094e-05,
+      "loss": 1.0281,
+      "step": 3170
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9234817028045654,
+      "learning_rate": 4.6567633262735446e-05,
+      "loss": 0.9495,
+      "step": 3175
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.596527636051178,
+      "learning_rate": 4.655705958895153e-05,
+      "loss": 0.6352,
+      "step": 3180
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.154539704322815,
+      "learning_rate": 4.6546470857954736e-05,
+      "loss": 0.8939,
+      "step": 3185
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7502239942550659,
+      "learning_rate": 4.653586707714108e-05,
+      "loss": 0.692,
+      "step": 3190
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7868794202804565,
+      "learning_rate": 4.652524825391711e-05,
+      "loss": 0.908,
+      "step": 3195
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6436206102371216,
+      "learning_rate": 4.6514614395699886e-05,
+      "loss": 0.7417,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.7466740012168884,
+      "eval_runtime": 96.4309,
+      "eval_samples_per_second": 7.228,
+      "eval_steps_per_second": 7.228,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8566870093345642,
+      "learning_rate": 4.6503965509916956e-05,
+      "loss": 0.8041,
+      "step": 3205
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5982272028923035,
+      "learning_rate": 4.649330160400639e-05,
+      "loss": 0.4528,
+      "step": 3210
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6511960029602051,
+      "learning_rate": 4.648262268541671e-05,
+      "loss": 0.877,
+      "step": 3215
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8476071357727051,
+      "learning_rate": 4.6471928761606965e-05,
+      "loss": 0.7145,
+      "step": 3220
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0408881902694702,
+      "learning_rate": 4.6461219840046654e-05,
+      "loss": 0.5539,
+      "step": 3225
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7445903420448303,
+      "learning_rate": 4.645049592821577e-05,
+      "loss": 0.8306,
+      "step": 3230
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.9672279357910156,
+      "learning_rate": 4.6439757033604756e-05,
+      "loss": 0.8645,
+      "step": 3235
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7082134485244751,
+      "learning_rate": 4.6429003163714556e-05,
+      "loss": 0.8188,
+      "step": 3240
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8803107142448425,
+      "learning_rate": 4.641823432605654e-05,
+      "loss": 0.7956,
+      "step": 3245
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7926101088523865,
+      "learning_rate": 4.640745052815254e-05,
+      "loss": 0.715,
+      "step": 3250
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.890519380569458,
+      "learning_rate": 4.639665177753485e-05,
+      "loss": 0.8825,
+      "step": 3255
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.9909971952438354,
+      "learning_rate": 4.638583808174619e-05,
+      "loss": 0.7843,
+      "step": 3260
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7450726628303528,
+      "learning_rate": 4.6375009448339743e-05,
+      "loss": 0.9714,
+      "step": 3265
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8220781683921814,
+      "learning_rate": 4.636416588487911e-05,
+      "loss": 0.8467,
+      "step": 3270
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.025499701499939,
+      "learning_rate": 4.63533073989383e-05,
+      "loss": 0.9301,
+      "step": 3275
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8067827820777893,
+      "learning_rate": 4.634243399810181e-05,
+      "loss": 0.7078,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8833619952201843,
+      "learning_rate": 4.6331545689964475e-05,
+      "loss": 0.699,
+      "step": 3285
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0600448846817017,
+      "learning_rate": 4.632064248213159e-05,
+      "loss": 0.7849,
+      "step": 3290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0503095388412476,
+      "learning_rate": 4.630972438221885e-05,
+      "loss": 0.6215,
+      "step": 3295
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5159885287284851,
+      "learning_rate": 4.629879139785235e-05,
+      "loss": 0.7449,
+      "step": 3300
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.7472941279411316,
+      "eval_runtime": 96.4994,
+      "eval_samples_per_second": 7.223,
+      "eval_steps_per_second": 7.223,
+      "step": 3300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.072464108467102,
+      "learning_rate": 4.6287843536668575e-05,
+      "loss": 0.8511,
+      "step": 3305
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.9016098976135254,
+      "learning_rate": 4.62768808063144e-05,
+      "loss": 0.7373,
+      "step": 3310
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0161947011947632,
+      "learning_rate": 4.626590321444712e-05,
+      "loss": 0.9035,
+      "step": 3315
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7459146976470947,
+      "learning_rate": 4.625491076873435e-05,
+      "loss": 0.6468,
+      "step": 3320
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.950080394744873,
+      "learning_rate": 4.624390347685413e-05,
+      "loss": 0.7211,
+      "step": 3325
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7308927774429321,
+      "learning_rate": 4.623288134649485e-05,
+      "loss": 0.9238,
+      "step": 3330
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7227129340171814,
+      "learning_rate": 4.622184438535527e-05,
+      "loss": 0.9773,
+      "step": 3335
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7054020166397095,
+      "learning_rate": 4.62107926011445e-05,
+      "loss": 0.7783,
+      "step": 3340
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6535981297492981,
+      "learning_rate": 4.619972600158201e-05,
+      "loss": 0.6559,
+      "step": 3345
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7245693206787109,
+      "learning_rate": 4.618864459439762e-05,
+      "loss": 0.8352,
+      "step": 3350
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9683626890182495,
+      "learning_rate": 4.6177548387331485e-05,
+      "loss": 0.9397,
+      "step": 3355
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.1847660541534424,
+      "learning_rate": 4.616643738813411e-05,
+      "loss": 0.7383,
+      "step": 3360
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8566804528236389,
+      "learning_rate": 4.615531160456633e-05,
+      "loss": 0.8066,
+      "step": 3365
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7312522530555725,
+      "learning_rate": 4.61441710443993e-05,
+      "loss": 0.7974,
+      "step": 3370
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6620572209358215,
+      "learning_rate": 4.6133015715414484e-05,
+      "loss": 0.9136,
+      "step": 3375
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5405072569847107,
+      "learning_rate": 4.612184562540369e-05,
+      "loss": 0.6921,
+      "step": 3380
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7474086284637451,
+      "learning_rate": 4.611066078216901e-05,
+      "loss": 0.8463,
+      "step": 3385
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9125152230262756,
+      "learning_rate": 4.609946119352287e-05,
+      "loss": 0.8508,
+      "step": 3390
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9998400211334229,
+      "learning_rate": 4.608824686728797e-05,
+      "loss": 0.8735,
+      "step": 3395
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5990025401115417,
+      "learning_rate": 4.6077017811297304e-05,
+      "loss": 0.8562,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.743736743927002,
+      "eval_runtime": 96.3748,
+      "eval_samples_per_second": 7.232,
+      "eval_steps_per_second": 7.232,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.35676899552345276,
+      "learning_rate": 4.606577403339418e-05,
+      "loss": 0.8914,
+      "step": 3405
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.772233784198761,
+      "learning_rate": 4.605451554143216e-05,
+      "loss": 0.779,
+      "step": 3410
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7336989641189575,
+      "learning_rate": 4.604324234327509e-05,
+      "loss": 0.7678,
+      "step": 3415
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7039794325828552,
+      "learning_rate": 4.603195444679711e-05,
+      "loss": 0.8783,
+      "step": 3420
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6955629587173462,
+      "learning_rate": 4.602065185988259e-05,
+      "loss": 0.818,
+      "step": 3425
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7369412779808044,
+      "learning_rate": 4.60093345904262e-05,
+      "loss": 0.6942,
+      "step": 3430
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6824669241905212,
+      "learning_rate": 4.5998002646332835e-05,
+      "loss": 0.9274,
+      "step": 3435
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.355720043182373,
+      "learning_rate": 4.598665603551765e-05,
+      "loss": 0.7219,
+      "step": 3440
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8629677295684814,
+      "learning_rate": 4.597529476590605e-05,
+      "loss": 0.8023,
+      "step": 3445
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2956135272979736,
+      "learning_rate": 4.596391884543368e-05,
+      "loss": 0.9574,
+      "step": 3450
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.2683722972869873,
+      "learning_rate": 4.59525282820464e-05,
+      "loss": 0.6996,
+      "step": 3455
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7317371368408203,
+      "learning_rate": 4.594112308370032e-05,
+      "loss": 1.03,
+      "step": 3460
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0310641527175903,
+      "learning_rate": 4.5929703258361756e-05,
+      "loss": 0.6917,
+      "step": 3465
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9479489326477051,
+      "learning_rate": 4.591826881400726e-05,
+      "loss": 0.9939,
+      "step": 3470
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9485552310943604,
+      "learning_rate": 4.5906819758623576e-05,
+      "loss": 1.0317,
+      "step": 3475
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.724987268447876,
+      "learning_rate": 4.589535610020765e-05,
+      "loss": 0.6915,
+      "step": 3480
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7091718316078186,
+      "learning_rate": 4.5883877846766654e-05,
+      "loss": 0.8673,
+      "step": 3485
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8297457098960876,
+      "learning_rate": 4.587238500631793e-05,
+      "loss": 0.8114,
+      "step": 3490
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7213541269302368,
+      "learning_rate": 4.586087758688903e-05,
+      "loss": 0.863,
+      "step": 3495
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.1096009016036987,
+      "learning_rate": 4.584935559651765e-05,
+      "loss": 0.9222,
+      "step": 3500
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.7428527474403381,
+      "eval_runtime": 96.3993,
+      "eval_samples_per_second": 7.23,
+      "eval_steps_per_second": 7.23,
+      "step": 3500
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5834380984306335,
+      "learning_rate": 4.583781904325172e-05,
+      "loss": 0.6609,
+      "step": 3505
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5797068476676941,
+      "learning_rate": 4.5826267935149285e-05,
+      "loss": 0.7933,
+      "step": 3510
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6027450561523438,
+      "learning_rate": 4.581470228027861e-05,
+      "loss": 0.7841,
+      "step": 3515
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5674509406089783,
+      "learning_rate": 4.5803122086718077e-05,
+      "loss": 0.7721,
+      "step": 3520
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7398461103439331,
+      "learning_rate": 4.5791527362556235e-05,
+      "loss": 0.7651,
+      "step": 3525
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6617181301116943,
+      "learning_rate": 4.577991811589181e-05,
+      "loss": 0.9359,
+      "step": 3530
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.49279505014419556,
+      "learning_rate": 4.576829435483362e-05,
+      "loss": 0.6278,
+      "step": 3535
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5201964378356934,
+      "learning_rate": 4.575665608750067e-05,
+      "loss": 0.853,
+      "step": 3540
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7188725471496582,
+      "learning_rate": 4.5745003322022084e-05,
+      "loss": 0.8338,
+      "step": 3545
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0798031091690063,
+      "learning_rate": 4.573333606653708e-05,
+      "loss": 0.9776,
+      "step": 3550
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6439509987831116,
+      "learning_rate": 4.5721654329195046e-05,
+      "loss": 0.9331,
+      "step": 3555
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7663920521736145,
+      "learning_rate": 4.570995811815545e-05,
+      "loss": 1.0533,
+      "step": 3560
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7230969071388245,
+      "learning_rate": 4.569824744158789e-05,
+      "loss": 0.6966,
+      "step": 3565
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.016112208366394,
+      "learning_rate": 4.568652230767205e-05,
+      "loss": 0.8393,
+      "step": 3570
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0165222883224487,
+      "learning_rate": 4.567478272459773e-05,
+      "loss": 1.0218,
+      "step": 3575
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.709685742855072,
+      "learning_rate": 4.5663028700564826e-05,
+      "loss": 0.7273,
+      "step": 3580
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5664321780204773,
+      "learning_rate": 4.565126024378328e-05,
+      "loss": 0.9079,
+      "step": 3585
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7938306927680969,
+      "learning_rate": 4.5639477362473173e-05,
+      "loss": 0.976,
+      "step": 3590
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6710417866706848,
+      "learning_rate": 4.5627680064864606e-05,
+      "loss": 1.1969,
+      "step": 3595
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9886580109596252,
+      "learning_rate": 4.5615868359197796e-05,
+      "loss": 0.9242,
+      "step": 3600
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.7412505149841309,
+      "eval_runtime": 96.4,
+      "eval_samples_per_second": 7.23,
+      "eval_steps_per_second": 7.23,
+      "step": 3600
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8157562613487244,
+      "learning_rate": 4.5604042253723014e-05,
+      "loss": 0.8398,
+      "step": 3605
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.2368131875991821,
+      "learning_rate": 4.559220175670054e-05,
+      "loss": 0.8742,
+      "step": 3610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6060155034065247,
+      "learning_rate": 4.558034687640078e-05,
+      "loss": 0.6993,
+      "step": 3615
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.1366558074951172,
+      "learning_rate": 4.556847762110415e-05,
+      "loss": 0.9328,
+      "step": 3620
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7205525636672974,
+      "learning_rate": 4.555659399910108e-05,
+      "loss": 0.827,
+      "step": 3625
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6944175958633423,
+      "learning_rate": 4.554469601869209e-05,
+      "loss": 0.7805,
+      "step": 3630
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6939406394958496,
+      "learning_rate": 4.55327836881877e-05,
+      "loss": 0.7996,
+      "step": 3635
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.592650830745697,
+      "learning_rate": 4.552085701590844e-05,
+      "loss": 0.6599,
+      "step": 3640
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5287877321243286,
+      "learning_rate": 4.5508916010184884e-05,
+      "loss": 0.6856,
+      "step": 3645
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6414081454277039,
+      "learning_rate": 4.549696067935762e-05,
+      "loss": 0.7622,
+      "step": 3650
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.2272289991378784,
+      "learning_rate": 4.548499103177719e-05,
+      "loss": 1.0834,
+      "step": 3655
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5912505388259888,
+      "learning_rate": 4.547300707580422e-05,
+      "loss": 0.8738,
+      "step": 3660
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6686813235282898,
+      "learning_rate": 4.5461008819809246e-05,
+      "loss": 0.6221,
+      "step": 3665
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.891153872013092,
+      "learning_rate": 4.544899627217286e-05,
+      "loss": 0.9009,
+      "step": 3670
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.1651557683944702,
+      "learning_rate": 4.543696944128559e-05,
+      "loss": 0.8448,
+      "step": 3675
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7525443434715271,
+      "learning_rate": 4.5424928335547964e-05,
+      "loss": 0.6654,
+      "step": 3680
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6798614859580994,
+      "learning_rate": 4.541287296337048e-05,
+      "loss": 0.9244,
+      "step": 3685
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.498735249042511,
+      "learning_rate": 4.540080333317358e-05,
+      "loss": 0.6815,
+      "step": 3690
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6097673773765564,
+      "learning_rate": 4.5388719453387694e-05,
+      "loss": 0.8536,
+      "step": 3695
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6685522198677063,
+      "learning_rate": 4.537662133245319e-05,
+      "loss": 0.8092,
+      "step": 3700
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.7402560114860535,
+      "eval_runtime": 96.4998,
+      "eval_samples_per_second": 7.223,
+      "eval_steps_per_second": 7.223,
+      "step": 3700
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.162788987159729,
+      "learning_rate": 4.5364508978820375e-05,
+      "loss": 0.6143,
+      "step": 3705
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8281823992729187,
+      "learning_rate": 4.5352382400949524e-05,
+      "loss": 0.8143,
+      "step": 3710
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6465135812759399,
+      "learning_rate": 4.534024160731082e-05,
+      "loss": 0.9152,
+      "step": 3715
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5903899669647217,
+      "learning_rate": 4.532808660638438e-05,
+      "loss": 0.7229,
+      "step": 3720
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6988681554794312,
+      "learning_rate": 4.5315917406660265e-05,
+      "loss": 0.6863,
+      "step": 3725
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7910459637641907,
+      "learning_rate": 4.530373401663843e-05,
+      "loss": 0.8762,
+      "step": 3730
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7580087184906006,
+      "learning_rate": 4.529153644482875e-05,
+      "loss": 0.9896,
+      "step": 3735
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6871665716171265,
+      "learning_rate": 4.5279324699751005e-05,
+      "loss": 0.8831,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0093677043914795,
+      "learning_rate": 4.526709878993488e-05,
+      "loss": 0.742,
+      "step": 3745
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9898921847343445,
+      "learning_rate": 4.525485872391996e-05,
+      "loss": 0.766,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8706837296485901,
+      "learning_rate": 4.524260451025569e-05,
+      "loss": 0.7545,
+      "step": 3755
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.1715607643127441,
+      "learning_rate": 4.523033615750142e-05,
+      "loss": 0.84,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.017062783241272,
+      "learning_rate": 4.521805367422638e-05,
+      "loss": 0.7477,
+      "step": 3765
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6071624159812927,
+      "learning_rate": 4.520575706900965e-05,
+      "loss": 0.793,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5821404457092285,
+      "learning_rate": 4.519344635044018e-05,
+      "loss": 0.7514,
+      "step": 3775
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6849238872528076,
+      "learning_rate": 4.51811215271168e-05,
+      "loss": 0.862,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8808868527412415,
+      "learning_rate": 4.5168782607648166e-05,
+      "loss": 0.7189,
+      "step": 3785
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7080340385437012,
+      "learning_rate": 4.5156429600652774e-05,
+      "loss": 0.6987,
+      "step": 3790
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.705869734287262,
+      "learning_rate": 4.5144062514759e-05,
+      "loss": 0.6482,
+      "step": 3795
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6345694065093994,
+      "learning_rate": 4.5131681358605007e-05,
+      "loss": 0.7279,
+      "step": 3800
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.7394095063209534,
+      "eval_runtime": 96.4977,
+      "eval_samples_per_second": 7.223,
+      "eval_steps_per_second": 7.223,
+      "step": 3800
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.748913586139679,
+      "learning_rate": 4.511928614083881e-05,
+      "loss": 0.7474,
+      "step": 3805
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6260043382644653,
+      "learning_rate": 4.5106876870118255e-05,
+      "loss": 0.7469,
+      "step": 3810
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.596367597579956,
+      "learning_rate": 4.509445355511098e-05,
+      "loss": 0.8437,
+      "step": 3815
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3925014734268188,
+      "learning_rate": 4.5082016204494445e-05,
+      "loss": 1.0928,
+      "step": 3820
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.1370338201522827,
+      "learning_rate": 4.506956482695592e-05,
+      "loss": 0.8908,
+      "step": 3825
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6746950149536133,
+      "learning_rate": 4.505709943119246e-05,
+      "loss": 0.7121,
+      "step": 3830
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6608826518058777,
+      "learning_rate": 4.504462002591091e-05,
+      "loss": 0.9397,
+      "step": 3835
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6542508006095886,
+      "learning_rate": 4.5032126619827916e-05,
+      "loss": 0.6942,
+      "step": 3840
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5825070738792419,
+      "learning_rate": 4.5019619221669895e-05,
+      "loss": 0.7083,
+      "step": 3845
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8596588373184204,
+      "learning_rate": 4.500709784017303e-05,
+      "loss": 0.839,
+      "step": 3850
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.641009509563446,
+      "learning_rate": 4.499456248408328e-05,
+      "loss": 0.72,
+      "step": 3855
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.213782548904419,
+      "learning_rate": 4.498201316215635e-05,
+      "loss": 0.7116,
+      "step": 3860
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.1411411762237549,
+      "learning_rate": 4.496944988315775e-05,
+      "loss": 1.0208,
+      "step": 3865
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8265553712844849,
+      "learning_rate": 4.495687265586266e-05,
+      "loss": 0.7664,
+      "step": 3870
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.9309681057929993,
+      "learning_rate": 4.4944281489056065e-05,
+      "loss": 0.9126,
+      "step": 3875
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.49171608686447144,
+      "learning_rate": 4.493167639153266e-05,
+      "loss": 0.6271,
+      "step": 3880
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.743669867515564,
+      "learning_rate": 4.491905737209688e-05,
+      "loss": 0.7965,
+      "step": 3885
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6191633939743042,
+      "learning_rate": 4.490642443956287e-05,
+      "loss": 0.5884,
+      "step": 3890
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5481441020965576,
+      "learning_rate": 4.489377760275452e-05,
+      "loss": 0.6281,
+      "step": 3895
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7155417203903198,
+      "learning_rate": 4.488111687050539e-05,
+      "loss": 0.7774,
+      "step": 3900
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.738506019115448,
+      "eval_runtime": 96.7667,
+      "eval_samples_per_second": 7.203,
+      "eval_steps_per_second": 7.203,
+      "step": 3900
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.032523274421692,
+      "learning_rate": 4.4868442251658795e-05,
+      "loss": 0.7621,
+      "step": 3905
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.584082841873169,
+      "learning_rate": 4.4855753755067703e-05,
+      "loss": 0.6617,
+      "step": 3910
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7214722037315369,
+      "learning_rate": 4.4843051389594814e-05,
+      "loss": 0.8669,
+      "step": 3915
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6019904613494873,
+      "learning_rate": 4.4830335164112504e-05,
+      "loss": 0.736,
+      "step": 3920
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8038384318351746,
+      "learning_rate": 4.48176050875028e-05,
+      "loss": 0.637,
+      "step": 3925
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9631878733634949,
+      "learning_rate": 4.4804861168657455e-05,
+      "loss": 0.9722,
+      "step": 3930
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5342935919761658,
+      "learning_rate": 4.4792103416477836e-05,
+      "loss": 0.8081,
+      "step": 3935
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5893488526344299,
+      "learning_rate": 4.477933183987503e-05,
+      "loss": 0.61,
+      "step": 3940
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.388850212097168,
+      "learning_rate": 4.476654644776973e-05,
+      "loss": 0.8454,
+      "step": 3945
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6928623914718628,
+      "learning_rate": 4.4753747249092305e-05,
+      "loss": 0.7209,
+      "step": 3950
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.2383430004119873,
+      "learning_rate": 4.4740934252782757e-05,
+      "loss": 0.8205,
+      "step": 3955
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6005001664161682,
+      "learning_rate": 4.472810746779074e-05,
+      "loss": 0.6083,
+      "step": 3960
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7928474545478821,
+      "learning_rate": 4.471526690307552e-05,
+      "loss": 0.9735,
+      "step": 3965
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8710891008377075,
+      "learning_rate": 4.4702412567606014e-05,
+      "loss": 0.7573,
+      "step": 3970
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6327987313270569,
+      "learning_rate": 4.468954447036071e-05,
+      "loss": 0.8563,
+      "step": 3975
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7048762440681458,
+      "learning_rate": 4.467666262032777e-05,
+      "loss": 0.9176,
+      "step": 3980
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6058861017227173,
+      "learning_rate": 4.466376702650492e-05,
+      "loss": 0.5525,
+      "step": 3985
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.637993574142456,
+      "learning_rate": 4.465085769789949e-05,
+      "loss": 0.7256,
+      "step": 3990
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6992897987365723,
+      "learning_rate": 4.463793464352842e-05,
+      "loss": 0.8824,
+      "step": 3995
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7812734246253967,
+      "learning_rate": 4.462499787241822e-05,
+      "loss": 0.8942,
+      "step": 4000
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.7363680601119995,
+      "eval_runtime": 96.9231,
+      "eval_samples_per_second": 7.191,
+      "eval_steps_per_second": 7.191,
+      "step": 4000
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.907598078250885,
+      "learning_rate": 4.4612047393605e-05,
+      "loss": 0.867,
+      "step": 4005
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9081722497940063,
+      "learning_rate": 4.459908321613442e-05,
+      "loss": 0.8757,
+      "step": 4010
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5538048148155212,
+      "learning_rate": 4.4586105349061726e-05,
+      "loss": 0.6709,
+      "step": 4015
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6632833480834961,
+      "learning_rate": 4.457311380145173e-05,
+      "loss": 0.8362,
+      "step": 4020
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8646539449691772,
+      "learning_rate": 4.4560108582378766e-05,
+      "loss": 0.8527,
+      "step": 4025
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6309005618095398,
+      "learning_rate": 4.454708970092678e-05,
+      "loss": 0.595,
+      "step": 4030
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5711541175842285,
+      "learning_rate": 4.45340571661892e-05,
+      "loss": 0.8069,
+      "step": 4035
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.1379880905151367,
+      "learning_rate": 4.4521010987269006e-05,
+      "loss": 0.8464,
+      "step": 4040
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6005469560623169,
+      "learning_rate": 4.450795117327874e-05,
+      "loss": 0.5801,
+      "step": 4045
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7842866778373718,
+      "learning_rate": 4.449487773334042e-05,
+      "loss": 0.6238,
+      "step": 4050
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7519890069961548,
+      "learning_rate": 4.448179067658563e-05,
+      "loss": 1.1255,
+      "step": 4055
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5955212712287903,
+      "learning_rate": 4.446869001215542e-05,
+      "loss": 0.7738,
+      "step": 4060
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5085921287536621,
+      "learning_rate": 4.4455575749200364e-05,
+      "loss": 0.6239,
+      "step": 4065
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.779778003692627,
+      "learning_rate": 4.444244789688056e-05,
+      "loss": 0.9719,
+      "step": 4070
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7279208898544312,
+      "learning_rate": 4.442930646436554e-05,
+      "loss": 0.9854,
+      "step": 4075
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9218065738677979,
+      "learning_rate": 4.4416151460834376e-05,
+      "loss": 0.8096,
+      "step": 4080
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7595914006233215,
+      "learning_rate": 4.44029828954756e-05,
+      "loss": 0.7955,
+      "step": 4085
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.785493016242981,
+      "learning_rate": 4.43898007774872e-05,
+      "loss": 0.8598,
+      "step": 4090
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5540453195571899,
+      "learning_rate": 4.437660511607666e-05,
+      "loss": 0.8485,
+      "step": 4095
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7215760350227356,
+      "learning_rate": 4.43633959204609e-05,
+      "loss": 0.9286,
+      "step": 4100
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.7347923517227173,
+      "eval_runtime": 96.8658,
+      "eval_samples_per_second": 7.196,
+      "eval_steps_per_second": 7.196,
+      "step": 4100
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7934743762016296,
+      "learning_rate": 4.435017319986631e-05,
+      "loss": 0.7829,
+      "step": 4105
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.503614068031311,
+      "learning_rate": 4.43369369635287e-05,
+      "loss": 0.7203,
+      "step": 4110
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6292420625686646,
+      "learning_rate": 4.4323687220693365e-05,
+      "loss": 0.7556,
+      "step": 4115
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6981114149093628,
+      "learning_rate": 4.431042398061499e-05,
+      "loss": 0.6953,
+      "step": 4120
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8554514050483704,
+      "learning_rate": 4.4297147252557715e-05,
+      "loss": 0.7731,
+      "step": 4125
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.1464003324508667,
+      "learning_rate": 4.428385704579509e-05,
+      "loss": 0.7761,
+      "step": 4130
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6772524118423462,
+      "learning_rate": 4.427055336961008e-05,
+      "loss": 0.7529,
+      "step": 4135
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5949820280075073,
+      "learning_rate": 4.425723623329507e-05,
+      "loss": 0.9164,
+      "step": 4140
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.848900318145752,
+      "learning_rate": 4.4243905646151825e-05,
+      "loss": 0.8385,
+      "step": 4145
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7119936943054199,
+      "learning_rate": 4.4230561617491514e-05,
+      "loss": 0.6342,
+      "step": 4150
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4240078628063202,
+      "learning_rate": 4.421720415663472e-05,
+      "loss": 0.9921,
+      "step": 4155
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.166399359703064,
+      "learning_rate": 4.4203833272911355e-05,
+      "loss": 0.6751,
+      "step": 4160
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7882303595542908,
+      "learning_rate": 4.4190448975660756e-05,
+      "loss": 0.8711,
+      "step": 4165
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7739405632019043,
+      "learning_rate": 4.417705127423162e-05,
+      "loss": 0.7635,
+      "step": 4170
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6729245781898499,
+      "learning_rate": 4.416364017798197e-05,
+      "loss": 1.0083,
+      "step": 4175
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7291648983955383,
+      "learning_rate": 4.4150215696279233e-05,
+      "loss": 0.9355,
+      "step": 4180
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.569436252117157,
+      "learning_rate": 4.413677783850015e-05,
+      "loss": 0.5718,
+      "step": 4185
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7857233285903931,
+      "learning_rate": 4.412332661403085e-05,
+      "loss": 0.6356,
+      "step": 4190
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.124894618988037,
+      "learning_rate": 4.410986203226672e-05,
+      "loss": 0.9911,
+      "step": 4195
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7088748216629028,
+      "learning_rate": 4.409638410261256e-05,
+      "loss": 0.7703,
+      "step": 4200
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.7353793978691101,
+      "eval_runtime": 96.9146,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 7.192,
+      "step": 4200
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8883334398269653,
+      "learning_rate": 4.4082892834482456e-05,
+      "loss": 0.7829,
+      "step": 4205
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5809643864631653,
+      "learning_rate": 4.406938823729979e-05,
+      "loss": 0.79,
+      "step": 4210
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2371888160705566,
+      "learning_rate": 4.405587032049731e-05,
+      "loss": 0.9394,
+      "step": 4215
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6468964219093323,
+      "learning_rate": 4.4042339093517e-05,
+      "loss": 0.7621,
+      "step": 4220
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8613569736480713,
+      "learning_rate": 4.4028794565810194e-05,
+      "loss": 0.9303,
+      "step": 4225
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8210548162460327,
+      "learning_rate": 4.4015236746837505e-05,
+      "loss": 1.04,
+      "step": 4230
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8066801428794861,
+      "learning_rate": 4.4001665646068804e-05,
+      "loss": 0.9942,
+      "step": 4235
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6841477751731873,
+      "learning_rate": 4.3988081272983263e-05,
+      "loss": 0.6893,
+      "step": 4240
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7812705636024475,
+      "learning_rate": 4.3974483637069333e-05,
+      "loss": 0.9125,
+      "step": 4245
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7913382649421692,
+      "learning_rate": 4.3960872747824686e-05,
+      "loss": 0.9298,
+      "step": 4250
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6053805947303772,
+      "learning_rate": 4.394724861475631e-05,
+      "loss": 0.7055,
+      "step": 4255
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6879487633705139,
+      "learning_rate": 4.393361124738039e-05,
+      "loss": 0.605,
+      "step": 4260
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7929925918579102,
+      "learning_rate": 4.3919960655222394e-05,
+      "loss": 0.8569,
+      "step": 4265
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5888631939888,
+      "learning_rate": 4.390629684781701e-05,
+      "loss": 0.6246,
+      "step": 4270
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9546008706092834,
+      "learning_rate": 4.389261983470815e-05,
+      "loss": 0.7964,
+      "step": 4275
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.620267391204834,
+      "learning_rate": 4.387892962544896e-05,
+      "loss": 0.7127,
+      "step": 4280
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7655039429664612,
+      "learning_rate": 4.3865226229601805e-05,
+      "loss": 0.6936,
+      "step": 4285
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5404471158981323,
+      "learning_rate": 4.3851509656738264e-05,
+      "loss": 0.6141,
+      "step": 4290
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9140282273292542,
+      "learning_rate": 4.38377799164391e-05,
+      "loss": 1.152,
+      "step": 4295
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4845621585845947,
+      "learning_rate": 4.382403701829429e-05,
+      "loss": 0.8322,
+      "step": 4300
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.733027458190918,
+      "eval_runtime": 96.886,
+      "eval_samples_per_second": 7.194,
+      "eval_steps_per_second": 7.194,
+      "step": 4300
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.598147988319397,
+      "learning_rate": 4.381028097190299e-05,
+      "loss": 0.772,
+      "step": 4305
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5572992563247681,
+      "learning_rate": 4.3796511786873574e-05,
+      "loss": 0.7232,
+      "step": 4310
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7913936376571655,
+      "learning_rate": 4.378272947282354e-05,
+      "loss": 0.6972,
+      "step": 4315
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4532865583896637,
+      "learning_rate": 4.376893403937959e-05,
+      "loss": 0.7454,
+      "step": 4320
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8871356844902039,
+      "learning_rate": 4.375512549617759e-05,
+      "loss": 0.6946,
+      "step": 4325
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7564520835876465,
+      "learning_rate": 4.374130385286255e-05,
+      "loss": 0.9257,
+      "step": 4330
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7280387282371521,
+      "learning_rate": 4.3727469119088624e-05,
+      "loss": 0.756,
+      "step": 4335
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6494055986404419,
+      "learning_rate": 4.3713621304519144e-05,
+      "loss": 0.6358,
+      "step": 4340
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6048948764801025,
+      "learning_rate": 4.369976041882654e-05,
+      "loss": 0.6705,
+      "step": 4345
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6458585858345032,
+      "learning_rate": 4.36858864716924e-05,
+      "loss": 0.7999,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.837872326374054,
+      "learning_rate": 4.36719994728074e-05,
+      "loss": 0.7671,
+      "step": 4355
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6451572775840759,
+      "learning_rate": 4.365809943187138e-05,
+      "loss": 0.8672,
+      "step": 4360
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6438645124435425,
+      "learning_rate": 4.364418635859326e-05,
+      "loss": 0.78,
+      "step": 4365
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7427099347114563,
+      "learning_rate": 4.363026026269106e-05,
+      "loss": 0.8977,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7844499945640564,
+      "learning_rate": 4.36163211538919e-05,
+      "loss": 0.7586,
+      "step": 4375
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8544999361038208,
+      "learning_rate": 4.360236904193201e-05,
+      "loss": 0.7085,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.431629180908203,
+      "learning_rate": 4.358840393655668e-05,
+      "loss": 0.8572,
+      "step": 4385
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6864097118377686,
+      "learning_rate": 4.357442584752027e-05,
+      "loss": 0.6848,
+      "step": 4390
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7158388495445251,
+      "learning_rate": 4.356043478458623e-05,
+      "loss": 1.0071,
+      "step": 4395
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7883514165878296,
+      "learning_rate": 4.3546430757527066e-05,
+      "loss": 0.9851,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.732368528842926,
+      "eval_runtime": 96.9109,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 7.192,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.673925518989563,
+      "learning_rate": 4.353241377612433e-05,
+      "loss": 0.7076,
+      "step": 4405
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.9540270566940308,
+      "learning_rate": 4.351838385016862e-05,
+      "loss": 0.8989,
+      "step": 4410
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8137551546096802,
+      "learning_rate": 4.35043409894596e-05,
+      "loss": 0.7633,
+      "step": 4415
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.779330313205719,
+      "learning_rate": 4.349028520380594e-05,
+      "loss": 0.7013,
+      "step": 4420
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7883580327033997,
+      "learning_rate": 4.347621650302535e-05,
+      "loss": 0.9788,
+      "step": 4425
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7106336951255798,
+      "learning_rate": 4.3462134896944565e-05,
+      "loss": 0.8399,
+      "step": 4430
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6303668022155762,
+      "learning_rate": 4.344804039539933e-05,
+      "loss": 0.5943,
+      "step": 4435
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.2975471019744873,
+      "learning_rate": 4.3433933008234395e-05,
+      "loss": 0.8917,
+      "step": 4440
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.68232661485672,
+      "learning_rate": 4.341981274530351e-05,
+      "loss": 0.7756,
+      "step": 4445
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6689594984054565,
+      "learning_rate": 4.340567961646943e-05,
+      "loss": 0.772,
+      "step": 4450
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.102365493774414,
+      "learning_rate": 4.339153363160388e-05,
+      "loss": 0.738,
+      "step": 4455
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6535090804100037,
+      "learning_rate": 4.337737480058758e-05,
+      "loss": 0.9096,
+      "step": 4460
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.676058292388916,
+      "learning_rate": 4.3363203133310206e-05,
+      "loss": 0.9634,
+      "step": 4465
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9258711934089661,
+      "learning_rate": 4.3349018639670415e-05,
+      "loss": 0.8025,
+      "step": 4470
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5786353349685669,
+      "learning_rate": 4.333482132957581e-05,
+      "loss": 0.7638,
+      "step": 4475
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7258582711219788,
+      "learning_rate": 4.332061121294296e-05,
+      "loss": 1.3538,
+      "step": 4480
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9384926557540894,
+      "learning_rate": 4.330638829969738e-05,
+      "loss": 0.8485,
+      "step": 4485
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5252525806427002,
+      "learning_rate": 4.3292152599773494e-05,
+      "loss": 0.8547,
+      "step": 4490
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7551200985908508,
+      "learning_rate": 4.32779041231147e-05,
+      "loss": 0.7435,
+      "step": 4495
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7492663264274597,
+      "learning_rate": 4.3263642879673286e-05,
+      "loss": 0.8712,
+      "step": 4500
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.7316818237304688,
+      "eval_runtime": 96.9418,
+      "eval_samples_per_second": 7.19,
+      "eval_steps_per_second": 7.19,
+      "step": 4500
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7490917444229126,
+      "learning_rate": 4.3249368879410475e-05,
+      "loss": 0.7598,
+      "step": 4505
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7305790781974792,
+      "learning_rate": 4.323508213229639e-05,
+      "loss": 0.8315,
+      "step": 4510
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7009093165397644,
+      "learning_rate": 4.3220782648310075e-05,
+      "loss": 0.7482,
+      "step": 4515
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7155885100364685,
+      "learning_rate": 4.320647043743945e-05,
+      "loss": 0.8385,
+      "step": 4520
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6159176826477051,
+      "learning_rate": 4.319214550968133e-05,
+      "loss": 0.6507,
+      "step": 4525
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7776069045066833,
+      "learning_rate": 4.3177807875041424e-05,
+      "loss": 0.855,
+      "step": 4530
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6204195618629456,
+      "learning_rate": 4.316345754353432e-05,
+      "loss": 0.7169,
+      "step": 4535
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7233458757400513,
+      "learning_rate": 4.3149094525183426e-05,
+      "loss": 0.5399,
+      "step": 4540
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7921779155731201,
+      "learning_rate": 4.313471883002108e-05,
+      "loss": 0.9124,
+      "step": 4545
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9145547747612,
+      "learning_rate": 4.3120330468088435e-05,
+      "loss": 1.2346,
+      "step": 4550
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8733106255531311,
+      "learning_rate": 4.310592944943549e-05,
+      "loss": 0.6737,
+      "step": 4555
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6620619297027588,
+      "learning_rate": 4.3091515784121107e-05,
+      "loss": 0.8041,
+      "step": 4560
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7026892900466919,
+      "learning_rate": 4.307708948221296e-05,
+      "loss": 0.9422,
+      "step": 4565
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7953292727470398,
+      "learning_rate": 4.3062650553787566e-05,
+      "loss": 0.7398,
+      "step": 4570
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6465870141983032,
+      "learning_rate": 4.304819900893024e-05,
+      "loss": 0.8175,
+      "step": 4575
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.3427163362503052,
+      "learning_rate": 4.303373485773513e-05,
+      "loss": 0.7331,
+      "step": 4580
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6665405035018921,
+      "learning_rate": 4.3019258110305186e-05,
+      "loss": 0.7529,
+      "step": 4585
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.796320915222168,
+      "learning_rate": 4.300476877675215e-05,
+      "loss": 0.915,
+      "step": 4590
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.551832377910614,
+      "learning_rate": 4.299026686719655e-05,
+      "loss": 0.7693,
+      "step": 4595
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.75690096616745,
+      "learning_rate": 4.297575239176771e-05,
+      "loss": 0.7871,
+      "step": 4600
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.730965256690979,
+      "eval_runtime": 96.8803,
+      "eval_samples_per_second": 7.194,
+      "eval_steps_per_second": 7.194,
+      "step": 4600
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7093445062637329,
+      "learning_rate": 4.296122536060373e-05,
+      "loss": 0.6279,
+      "step": 4605
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6522731781005859,
+      "learning_rate": 4.294668578385147e-05,
+      "loss": 0.5442,
+      "step": 4610
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7964634299278259,
+      "learning_rate": 4.2932133671666565e-05,
+      "loss": 1.0221,
+      "step": 4615
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7733820676803589,
+      "learning_rate": 4.2917569034213395e-05,
+      "loss": 0.7152,
+      "step": 4620
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8039364218711853,
+      "learning_rate": 4.2902991881665097e-05,
+      "loss": 1.0939,
+      "step": 4625
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.756020724773407,
+      "learning_rate": 4.2888402224203536e-05,
+      "loss": 0.7539,
+      "step": 4630
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5059025287628174,
+      "learning_rate": 4.2873800072019345e-05,
+      "loss": 0.8716,
+      "step": 4635
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8273636102676392,
+      "learning_rate": 4.285918543531183e-05,
+      "loss": 0.687,
+      "step": 4640
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6505921483039856,
+      "learning_rate": 4.2844558324289076e-05,
+      "loss": 1.0697,
+      "step": 4645
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6481053829193115,
+      "learning_rate": 4.282991874916784e-05,
+      "loss": 0.884,
+      "step": 4650
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8193663954734802,
+      "learning_rate": 4.28152667201736e-05,
+      "loss": 0.962,
+      "step": 4655
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7153398990631104,
+      "learning_rate": 4.280060224754053e-05,
+      "loss": 0.7705,
+      "step": 4660
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7197556495666504,
+      "learning_rate": 4.278592534151149e-05,
+      "loss": 0.8521,
+      "step": 4665
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.009098768234253,
+      "learning_rate": 4.2771236012338044e-05,
+      "loss": 0.8425,
+      "step": 4670
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.598564088344574,
+      "learning_rate": 4.275653427028041e-05,
+      "loss": 0.7072,
+      "step": 4675
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8035867810249329,
+      "learning_rate": 4.2741820125607504e-05,
+      "loss": 0.6689,
+      "step": 4680
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6109891533851624,
+      "learning_rate": 4.2727093588596866e-05,
+      "loss": 0.7776,
+      "step": 4685
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7504151463508606,
+      "learning_rate": 4.271235466953473e-05,
+      "loss": 0.7481,
+      "step": 4690
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9492344260215759,
+      "learning_rate": 4.269760337871594e-05,
+      "loss": 0.7261,
+      "step": 4695
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5792133212089539,
+      "learning_rate": 4.2682839726444035e-05,
+      "loss": 0.7156,
+      "step": 4700
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.7283556461334229,
+      "eval_runtime": 96.8998,
+      "eval_samples_per_second": 7.193,
+      "eval_steps_per_second": 7.193,
+      "step": 4700
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8895491361618042,
+      "learning_rate": 4.266806372303113e-05,
+      "loss": 0.8466,
+      "step": 4705
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7995960712432861,
+      "learning_rate": 4.2653275378798005e-05,
+      "loss": 0.7823,
+      "step": 4710
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6673771739006042,
+      "learning_rate": 4.263847470407405e-05,
+      "loss": 0.5461,
+      "step": 4715
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6228974461555481,
+      "learning_rate": 4.262366170919726e-05,
+      "loss": 0.7611,
+      "step": 4720
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8050612807273865,
+      "learning_rate": 4.2608836404514255e-05,
+      "loss": 0.6524,
+      "step": 4725
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8815121650695801,
+      "learning_rate": 4.2593998800380216e-05,
+      "loss": 0.9997,
+      "step": 4730
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0408731698989868,
+      "learning_rate": 4.257914890715897e-05,
+      "loss": 0.7031,
+      "step": 4735
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6744192838668823,
+      "learning_rate": 4.256428673522287e-05,
+      "loss": 0.6587,
+      "step": 4740
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.014369249343872,
+      "learning_rate": 4.254941229495289e-05,
+      "loss": 0.7726,
+      "step": 4745
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7497864365577698,
+      "learning_rate": 4.2534525596738526e-05,
+      "loss": 0.7327,
+      "step": 4750
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6479122042655945,
+      "learning_rate": 4.2519626650977905e-05,
+      "loss": 0.7071,
+      "step": 4755
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6300268173217773,
+      "learning_rate": 4.250471546807765e-05,
+      "loss": 0.9479,
+      "step": 4760
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8272077441215515,
+      "learning_rate": 4.248979205845294e-05,
+      "loss": 0.9013,
+      "step": 4765
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7070410847663879,
+      "learning_rate": 4.2474856432527524e-05,
+      "loss": 0.713,
+      "step": 4770
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7199767231941223,
+      "learning_rate": 4.2459908600733654e-05,
+      "loss": 0.9308,
+      "step": 4775
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6886048316955566,
+      "learning_rate": 4.244494857351212e-05,
+      "loss": 0.8008,
+      "step": 4780
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6097077131271362,
+      "learning_rate": 4.242997636131222e-05,
+      "loss": 0.9639,
+      "step": 4785
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0947343111038208,
+      "learning_rate": 4.241499197459178e-05,
+      "loss": 0.9012,
+      "step": 4790
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6965738534927368,
+      "learning_rate": 4.239999542381712e-05,
+      "loss": 0.6745,
+      "step": 4795
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8290371894836426,
+      "learning_rate": 4.238498671946306e-05,
+      "loss": 0.7856,
+      "step": 4800
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.7277354598045349,
+      "eval_runtime": 96.9165,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 7.192,
+      "step": 4800
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8061904907226562,
+      "learning_rate": 4.2369965872012904e-05,
+      "loss": 0.7034,
+      "step": 4805
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6652625799179077,
+      "learning_rate": 4.2354932891958434e-05,
+      "loss": 0.5825,
+      "step": 4810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6529026627540588,
+      "learning_rate": 4.2339887789799916e-05,
+      "loss": 0.7407,
+      "step": 4815
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7802493572235107,
+      "learning_rate": 4.232483057604607e-05,
+      "loss": 0.8906,
+      "step": 4820
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7709060907363892,
+      "learning_rate": 4.230976126121411e-05,
+      "loss": 0.863,
+      "step": 4825
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.2582249641418457,
+      "learning_rate": 4.229467985582966e-05,
+      "loss": 1.065,
+      "step": 4830
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5523508191108704,
+      "learning_rate": 4.22795863704268e-05,
+      "loss": 0.5925,
+      "step": 4835
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.3535953760147095,
+      "learning_rate": 4.2264480815548076e-05,
+      "loss": 0.7993,
+      "step": 4840
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.067133903503418,
+      "learning_rate": 4.2249363201744425e-05,
+      "loss": 0.7921,
+      "step": 4845
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6478603482246399,
+      "learning_rate": 4.223423353957523e-05,
+      "loss": 0.6769,
+      "step": 4850
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6439855694770813,
+      "learning_rate": 4.2219091839608276e-05,
+      "loss": 0.9018,
+      "step": 4855
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5302556753158569,
+      "learning_rate": 4.2203938112419786e-05,
+      "loss": 0.837,
+      "step": 4860
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8129810690879822,
+      "learning_rate": 4.218877236859433e-05,
+      "loss": 0.9195,
+      "step": 4865
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6531801819801331,
+      "learning_rate": 4.217359461872493e-05,
+      "loss": 0.6829,
+      "step": 4870
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7695423364639282,
+      "learning_rate": 4.215840487341296e-05,
+      "loss": 0.7739,
+      "step": 4875
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.3163946866989136,
+      "learning_rate": 4.2143203143268184e-05,
+      "loss": 0.9678,
+      "step": 4880
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.1124577522277832,
+      "learning_rate": 4.212798943890871e-05,
+      "loss": 0.9327,
+      "step": 4885
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8979106545448303,
+      "learning_rate": 4.2112763770961074e-05,
+      "loss": 0.7043,
+      "step": 4890
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9879763126373291,
+      "learning_rate": 4.2097526150060085e-05,
+      "loss": 0.8129,
+      "step": 4895
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7016007304191589,
+      "learning_rate": 4.208227658684898e-05,
+      "loss": 0.7906,
+      "step": 4900
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 0.7254941463470459,
+      "eval_runtime": 96.9328,
+      "eval_samples_per_second": 7.191,
+      "eval_steps_per_second": 7.191,
+      "step": 4900
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5404706597328186,
+      "learning_rate": 4.206701509197927e-05,
+      "loss": 0.7769,
+      "step": 4905
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7096789479255676,
+      "learning_rate": 4.205174167611085e-05,
+      "loss": 0.5985,
+      "step": 4910
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8139373660087585,
+      "learning_rate": 4.20364563499119e-05,
+      "loss": 0.75,
+      "step": 4915
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2196255922317505,
+      "learning_rate": 4.202115912405897e-05,
+      "loss": 0.8441,
+      "step": 4920
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6912347674369812,
+      "learning_rate": 4.200585000923689e-05,
+      "loss": 0.8885,
+      "step": 4925
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7879334688186646,
+      "learning_rate": 4.199052901613878e-05,
+      "loss": 0.6353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.0302627086639404,
+      "learning_rate": 4.197519615546608e-05,
+      "loss": 0.7704,
+      "step": 4935
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5351320505142212,
+      "learning_rate": 4.195985143792851e-05,
+      "loss": 0.8094,
+      "step": 4940
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7632457613945007,
+      "learning_rate": 4.194449487424409e-05,
+      "loss": 0.9625,
+      "step": 4945
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7706131935119629,
+      "learning_rate": 4.1929126475139096e-05,
+      "loss": 0.7007,
+      "step": 4950
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6262048482894897,
+      "learning_rate": 4.191374625134806e-05,
+      "loss": 0.7768,
+      "step": 4955
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8302519917488098,
+      "learning_rate": 4.189835421361381e-05,
+      "loss": 0.8281,
+      "step": 4960
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5914260149002075,
+      "learning_rate": 4.188295037268738e-05,
+      "loss": 0.8554,
+      "step": 4965
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7599936127662659,
+      "learning_rate": 4.1867534739328085e-05,
+      "loss": 0.9547,
+      "step": 4970
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4832470417022705,
+      "learning_rate": 4.1852107324303455e-05,
+      "loss": 0.5212,
+      "step": 4975
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8040557503700256,
+      "learning_rate": 4.183666813838927e-05,
+      "loss": 0.8939,
+      "step": 4980
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7503822445869446,
+      "learning_rate": 4.182121719236952e-05,
+      "loss": 0.9279,
+      "step": 4985
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7608035206794739,
+      "learning_rate": 4.180575449703639e-05,
+      "loss": 0.7965,
+      "step": 4990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5479308366775513,
+      "learning_rate": 4.1790280063190315e-05,
+      "loss": 0.7478,
+      "step": 4995
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7714606523513794,
+      "learning_rate": 4.177479390163989e-05,
+      "loss": 0.7917,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.7249829173088074,
+      "eval_runtime": 96.9791,
+      "eval_samples_per_second": 7.187,
+      "eval_steps_per_second": 7.187,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7731960415840149,
+      "learning_rate": 4.175929602320192e-05,
+      "loss": 0.5968,
+      "step": 5005
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.525674045085907,
+      "learning_rate": 4.174378643870138e-05,
+      "loss": 0.6748,
+      "step": 5010
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8199644088745117,
+      "learning_rate": 4.172826515897146e-05,
+      "loss": 0.7478,
+      "step": 5015
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5651956796646118,
+      "learning_rate": 4.1712732194853464e-05,
+      "loss": 0.5784,
+      "step": 5020
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8027454614639282,
+      "learning_rate": 4.1697187557196896e-05,
+      "loss": 0.9719,
+      "step": 5025
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8870482444763184,
+      "learning_rate": 4.168163125685939e-05,
+      "loss": 0.7407,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6260176301002502,
+      "learning_rate": 4.1666063304706756e-05,
+      "loss": 0.8039,
+      "step": 5035
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5424874424934387,
+      "learning_rate": 4.165048371161291e-05,
+      "loss": 0.8169,
+      "step": 5040
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8967744708061218,
+      "learning_rate": 4.163489248845992e-05,
+      "loss": 0.7546,
+      "step": 5045
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.871800422668457,
+      "learning_rate": 4.161928964613797e-05,
+      "loss": 0.7789,
+      "step": 5050
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9717918634414673,
+      "learning_rate": 4.1603675195545356e-05,
+      "loss": 0.7955,
+      "step": 5055
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6605653762817383,
+      "learning_rate": 4.15880491475885e-05,
+      "loss": 0.8416,
+      "step": 5060
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9123837351799011,
+      "learning_rate": 4.1572411513181896e-05,
+      "loss": 0.6594,
+      "step": 5065
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8939911723136902,
+      "learning_rate": 4.155676230324816e-05,
+      "loss": 0.858,
+      "step": 5070
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6447166204452515,
+      "learning_rate": 4.154110152871797e-05,
+      "loss": 0.7425,
+      "step": 5075
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9701817631721497,
+      "learning_rate": 4.152542920053009e-05,
+      "loss": 0.9332,
+      "step": 5080
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7782546281814575,
+      "learning_rate": 4.1509745329631365e-05,
+      "loss": 0.8372,
+      "step": 5085
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6977235078811646,
+      "learning_rate": 4.149404992697669e-05,
+      "loss": 0.6689,
+      "step": 5090
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7119914293289185,
+      "learning_rate": 4.147834300352901e-05,
+      "loss": 0.6255,
+      "step": 5095
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7148675322532654,
+      "learning_rate": 4.146262457025933e-05,
+      "loss": 0.6395,
+      "step": 5100
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.7237269282341003,
+      "eval_runtime": 96.898,
+      "eval_samples_per_second": 7.193,
+      "eval_steps_per_second": 7.193,
+      "step": 5100
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7843590378761292,
+      "learning_rate": 4.144689463814669e-05,
+      "loss": 0.6286,
+      "step": 5105
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8296339511871338,
+      "learning_rate": 4.143115321817815e-05,
+      "loss": 0.8533,
+      "step": 5110
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6789836287498474,
+      "learning_rate": 4.1415400321348827e-05,
+      "loss": 0.7821,
+      "step": 5115
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.9112969040870667,
+      "learning_rate": 4.139963595866182e-05,
+      "loss": 0.9232,
+      "step": 5120
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5057934522628784,
+      "learning_rate": 4.138386014112824e-05,
+      "loss": 0.5756,
+      "step": 5125
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7892030477523804,
+      "learning_rate": 4.136807287976721e-05,
+      "loss": 0.6673,
+      "step": 5130
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6285967826843262,
+      "learning_rate": 4.135227418560585e-05,
+      "loss": 0.843,
+      "step": 5135
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5548838973045349,
+      "learning_rate": 4.133646406967927e-05,
+      "loss": 0.8089,
+      "step": 5140
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6649813055992126,
+      "learning_rate": 4.132064254303053e-05,
+      "loss": 0.8778,
+      "step": 5145
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6559309959411621,
+      "learning_rate": 4.1304809616710685e-05,
+      "loss": 0.6655,
+      "step": 5150
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.839788019657135,
+      "learning_rate": 4.1288965301778745e-05,
+      "loss": 0.8372,
+      "step": 5155
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5987736582756042,
+      "learning_rate": 4.1273109609301686e-05,
+      "loss": 0.6379,
+      "step": 5160
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7344145774841309,
+      "learning_rate": 4.125724255035439e-05,
+      "loss": 0.7199,
+      "step": 5165
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.2459057569503784,
+      "learning_rate": 4.124136413601973e-05,
+      "loss": 0.7656,
+      "step": 5170
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.2029509544372559,
+      "learning_rate": 4.12254743773885e-05,
+      "loss": 0.6369,
+      "step": 5175
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8527836799621582,
+      "learning_rate": 4.1209573285559376e-05,
+      "loss": 0.7597,
+      "step": 5180
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6795602440834045,
+      "learning_rate": 4.1193660871639006e-05,
+      "loss": 0.8141,
+      "step": 5185
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8437166810035706,
+      "learning_rate": 4.1177737146741904e-05,
+      "loss": 0.7443,
+      "step": 5190
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7290692329406738,
+      "learning_rate": 4.1161802121990513e-05,
+      "loss": 0.5944,
+      "step": 5195
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.624458372592926,
+      "learning_rate": 4.114585580851515e-05,
+      "loss": 0.7567,
+      "step": 5200
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.7232300043106079,
+      "eval_runtime": 96.871,
+      "eval_samples_per_second": 7.195,
+      "eval_steps_per_second": 7.195,
+      "step": 5200
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9863276481628418,
+      "learning_rate": 4.1129898217454034e-05,
+      "loss": 0.7102,
+      "step": 5205
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8225752115249634,
+      "learning_rate": 4.111392935995324e-05,
+      "loss": 0.8148,
+      "step": 5210
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.687274694442749,
+      "learning_rate": 4.109794924716673e-05,
+      "loss": 0.7285,
+      "step": 5215
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.306779146194458,
+      "learning_rate": 4.108195789025632e-05,
+      "loss": 1.0412,
+      "step": 5220
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.872298002243042,
+      "learning_rate": 4.1065955300391676e-05,
+      "loss": 0.8104,
+      "step": 5225
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41228190064430237,
+      "learning_rate": 4.104994148875032e-05,
+      "loss": 0.8437,
+      "step": 5230
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.0037802457809448,
+      "learning_rate": 4.103391646651761e-05,
+      "loss": 0.579,
+      "step": 5235
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6584619879722595,
+      "learning_rate": 4.101788024488673e-05,
+      "loss": 0.8698,
+      "step": 5240
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8180784583091736,
+      "learning_rate": 4.1001832835058685e-05,
+      "loss": 0.9177,
+      "step": 5245
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7600345015525818,
+      "learning_rate": 4.0985774248242296e-05,
+      "loss": 0.6935,
+      "step": 5250
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9433395266532898,
+      "learning_rate": 4.09697044956542e-05,
+      "loss": 0.642,
+      "step": 5255
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.625467836856842,
+      "learning_rate": 4.095362358851883e-05,
+      "loss": 0.8038,
+      "step": 5260
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5945066213607788,
+      "learning_rate": 4.0937531538068387e-05,
+      "loss": 0.7096,
+      "step": 5265
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9545826315879822,
+      "learning_rate": 4.092142835554289e-05,
+      "loss": 0.6535,
+      "step": 5270
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8320907354354858,
+      "learning_rate": 4.090531405219012e-05,
+      "loss": 0.7521,
+      "step": 5275
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7830407619476318,
+      "learning_rate": 4.088918863926562e-05,
+      "loss": 0.5991,
+      "step": 5280
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6662408709526062,
+      "learning_rate": 4.0873052128032684e-05,
+      "loss": 0.9341,
+      "step": 5285
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.2364461421966553,
+      "learning_rate": 4.0856904529762394e-05,
+      "loss": 0.6156,
+      "step": 5290
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5708659291267395,
+      "learning_rate": 4.084074585573354e-05,
+      "loss": 0.6609,
+      "step": 5295
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6155332326889038,
+      "learning_rate": 4.082457611723266e-05,
+      "loss": 0.8551,
+      "step": 5300
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.7219573855400085,
+      "eval_runtime": 96.9322,
+      "eval_samples_per_second": 7.191,
+      "eval_steps_per_second": 7.191,
+      "step": 5300
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5997724533081055,
+      "learning_rate": 4.0808395325554023e-05,
+      "loss": 0.7243,
+      "step": 5305
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5939134359359741,
+      "learning_rate": 4.079220349199962e-05,
+      "loss": 0.7455,
+      "step": 5310
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5447105765342712,
+      "learning_rate": 4.077600062787915e-05,
+      "loss": 0.8424,
+      "step": 5315
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.741060197353363,
+      "learning_rate": 4.075978674451001e-05,
+      "loss": 0.6837,
+      "step": 5320
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8776602745056152,
+      "learning_rate": 4.074356185321732e-05,
+      "loss": 0.6541,
+      "step": 5325
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5684154629707336,
+      "learning_rate": 4.072732596533385e-05,
+      "loss": 0.6843,
+      "step": 5330
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9258201718330383,
+      "learning_rate": 4.071107909220009e-05,
+      "loss": 0.7829,
+      "step": 5335
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6690560579299927,
+      "learning_rate": 4.0694821245164165e-05,
+      "loss": 0.7726,
+      "step": 5340
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9005510807037354,
+      "learning_rate": 4.0678552435581905e-05,
+      "loss": 0.845,
+      "step": 5345
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6624660491943359,
+      "learning_rate": 4.066227267481676e-05,
+      "loss": 0.5798,
+      "step": 5350
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5465379953384399,
+      "learning_rate": 4.064598197423985e-05,
+      "loss": 0.6488,
+      "step": 5355
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7643209099769592,
+      "learning_rate": 4.062968034522995e-05,
+      "loss": 0.6886,
+      "step": 5360
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7964931130409241,
+      "learning_rate": 4.061336779917342e-05,
+      "loss": 1.0865,
+      "step": 5365
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6491959095001221,
+      "learning_rate": 4.05970443474643e-05,
+      "loss": 0.5869,
+      "step": 5370
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.0309019088745117,
+      "learning_rate": 4.058071000150421e-05,
+      "loss": 0.8846,
+      "step": 5375
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6856257915496826,
+      "learning_rate": 4.05643647727024e-05,
+      "loss": 0.774,
+      "step": 5380
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6670073866844177,
+      "learning_rate": 4.0548008672475714e-05,
+      "loss": 0.6854,
+      "step": 5385
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.9411470890045166,
+      "learning_rate": 4.053164171224858e-05,
+      "loss": 0.9237,
+      "step": 5390
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7178419232368469,
+      "learning_rate": 4.051526390345303e-05,
+      "loss": 0.6262,
+      "step": 5395
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.69013512134552,
+      "learning_rate": 4.049887525752867e-05,
+      "loss": 0.7392,
+      "step": 5400
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.7226108908653259,
+      "eval_runtime": 96.9511,
+      "eval_samples_per_second": 7.189,
+      "eval_steps_per_second": 7.189,
+      "step": 5400
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.5154650211334229,
+      "learning_rate": 4.048247578592266e-05,
+      "loss": 1.0026,
+      "step": 5405
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6985073685646057,
+      "learning_rate": 4.046606550008973e-05,
+      "loss": 0.4776,
+      "step": 5410
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.687282383441925,
+      "learning_rate": 4.044964441149217e-05,
+      "loss": 0.7665,
+      "step": 5415
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7575299143791199,
+      "learning_rate": 4.043321253159981e-05,
+      "loss": 0.8202,
+      "step": 5420
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.0001776218414307,
+      "learning_rate": 4.041676987189003e-05,
+      "loss": 0.7952,
+      "step": 5425
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.240440845489502,
+      "learning_rate": 4.040031644384771e-05,
+      "loss": 0.7086,
+      "step": 5430
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7027778029441833,
+      "learning_rate": 4.0383852258965274e-05,
+      "loss": 0.8268,
+      "step": 5435
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7208477854728699,
+      "learning_rate": 4.036737732874266e-05,
+      "loss": 0.7095,
+      "step": 5440
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6870366334915161,
+      "learning_rate": 4.035089166468731e-05,
+      "loss": 0.6324,
+      "step": 5445
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6353491544723511,
+      "learning_rate": 4.033439527831415e-05,
+      "loss": 0.6463,
+      "step": 5450
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8659310340881348,
+      "learning_rate": 4.03178881811456e-05,
+      "loss": 0.5445,
+      "step": 5455
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.102812647819519,
+      "learning_rate": 4.030137038471158e-05,
+      "loss": 1.0779,
+      "step": 5460
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6689164638519287,
+      "learning_rate": 4.028484190054947e-05,
+      "loss": 0.7816,
+      "step": 5465
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7742553949356079,
+      "learning_rate": 4.02683027402041e-05,
+      "loss": 0.7534,
+      "step": 5470
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6275122761726379,
+      "learning_rate": 4.025175291522779e-05,
+      "loss": 0.8127,
+      "step": 5475
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7048559188842773,
+      "learning_rate": 4.0235192437180276e-05,
+      "loss": 0.7884,
+      "step": 5480
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6857985854148865,
+      "learning_rate": 4.0218621317628755e-05,
+      "loss": 0.6286,
+      "step": 5485
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.750784158706665,
+      "learning_rate": 4.0202039568147857e-05,
+      "loss": 0.76,
+      "step": 5490
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.906353235244751,
+      "learning_rate": 4.018544720031962e-05,
+      "loss": 0.7227,
+      "step": 5495
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.001240611076355,
+      "learning_rate": 4.0168844225733516e-05,
+      "loss": 0.9312,
+      "step": 5500
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.7204533815383911,
+      "eval_runtime": 96.9186,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 7.192,
+      "step": 5500
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6060788035392761,
+      "learning_rate": 4.015223065598642e-05,
+      "loss": 0.7476,
+      "step": 5505
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9126180410385132,
+      "learning_rate": 4.013560650268261e-05,
+      "loss": 0.9137,
+      "step": 5510
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7585060000419617,
+      "learning_rate": 4.0118971777433755e-05,
+      "loss": 0.7926,
+      "step": 5515
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7748726606369019,
+      "learning_rate": 4.01023264918589e-05,
+      "loss": 0.8049,
+      "step": 5520
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6712713241577148,
+      "learning_rate": 4.008567065758449e-05,
+      "loss": 0.6242,
+      "step": 5525
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9764819741249084,
+      "learning_rate": 4.006900428624429e-05,
+      "loss": 0.7445,
+      "step": 5530
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6748983263969421,
+      "learning_rate": 4.005232738947948e-05,
+      "loss": 0.9529,
+      "step": 5535
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9980877637863159,
+      "learning_rate": 4.003563997893857e-05,
+      "loss": 1.0637,
+      "step": 5540
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8130014538764954,
+      "learning_rate": 4.0018942066277406e-05,
+      "loss": 0.7479,
+      "step": 5545
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5256005525588989,
+      "learning_rate": 4.000223366315917e-05,
+      "loss": 0.5274,
+      "step": 5550
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6466362476348877,
+      "learning_rate": 3.9985514781254385e-05,
+      "loss": 0.7175,
+      "step": 5555
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8849642276763916,
+      "learning_rate": 3.9968785432240874e-05,
+      "loss": 0.7762,
+      "step": 5560
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7992335557937622,
+      "learning_rate": 3.9952045627803795e-05,
+      "loss": 0.6103,
+      "step": 5565
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6376257538795471,
+      "learning_rate": 3.993529537963559e-05,
+      "loss": 0.7441,
+      "step": 5570
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6867454648017883,
+      "learning_rate": 3.9918534699436e-05,
+      "loss": 0.8815,
+      "step": 5575
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8116294741630554,
+      "learning_rate": 3.990176359891206e-05,
+      "loss": 0.9572,
+      "step": 5580
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8836712837219238,
+      "learning_rate": 3.988498208977808e-05,
+      "loss": 0.8463,
+      "step": 5585
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.835573673248291,
+      "learning_rate": 3.9868190183755624e-05,
+      "loss": 0.7096,
+      "step": 5590
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9869277477264404,
+      "learning_rate": 3.9851387892573545e-05,
+      "loss": 0.9144,
+      "step": 5595
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6233755350112915,
+      "learning_rate": 3.983457522796793e-05,
+      "loss": 0.8323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.719563901424408,
+      "eval_runtime": 96.851,
+      "eval_samples_per_second": 7.197,
+      "eval_steps_per_second": 7.197,
+      "step": 5600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6275632977485657,
+      "learning_rate": 3.981775220168212e-05,
+      "loss": 0.751,
+      "step": 5605
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7892312407493591,
+      "learning_rate": 3.980091882546669e-05,
+      "loss": 0.932,
+      "step": 5610
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6969819068908691,
+      "learning_rate": 3.9784075111079456e-05,
+      "loss": 0.7515,
+      "step": 5615
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7002953290939331,
+      "learning_rate": 3.976722107028544e-05,
+      "loss": 0.8556,
+      "step": 5620
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.1970090866088867,
+      "learning_rate": 3.9750356714856876e-05,
+      "loss": 0.7594,
+      "step": 5625
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7774887084960938,
+      "learning_rate": 3.9733482056573214e-05,
+      "loss": 0.788,
+      "step": 5630
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5987717509269714,
+      "learning_rate": 3.971659710722109e-05,
+      "loss": 0.7458,
+      "step": 5635
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7544295191764832,
+      "learning_rate": 3.969970187859435e-05,
+      "loss": 0.9236,
+      "step": 5640
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.217820644378662,
+      "learning_rate": 3.968279638249398e-05,
+      "loss": 0.7796,
+      "step": 5645
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.211364507675171,
+      "learning_rate": 3.966588063072817e-05,
+      "loss": 0.7593,
+      "step": 5650
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.2978931665420532,
+      "learning_rate": 3.964895463511227e-05,
+      "loss": 0.8755,
+      "step": 5655
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6411646008491516,
+      "learning_rate": 3.963201840746877e-05,
+      "loss": 0.8999,
+      "step": 5660
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8105179667472839,
+      "learning_rate": 3.961507195962732e-05,
+      "loss": 0.758,
+      "step": 5665
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8283591270446777,
+      "learning_rate": 3.959811530342471e-05,
+      "loss": 0.8578,
+      "step": 5670
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8040729761123657,
+      "learning_rate": 3.958114845070485e-05,
+      "loss": 0.7062,
+      "step": 5675
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.696398138999939,
+      "learning_rate": 3.956417141331878e-05,
+      "loss": 0.8301,
+      "step": 5680
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7806740403175354,
+      "learning_rate": 3.9547184203124636e-05,
+      "loss": 0.7002,
+      "step": 5685
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6160992980003357,
+      "learning_rate": 3.9530186831987695e-05,
+      "loss": 0.7433,
+      "step": 5690
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.9394809603691101,
+      "learning_rate": 3.95131793117803e-05,
+      "loss": 0.8878,
+      "step": 5695
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7936028838157654,
+      "learning_rate": 3.94961616543819e-05,
+      "loss": 0.7312,
+      "step": 5700
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.7196908593177795,
+      "eval_runtime": 96.9199,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 7.192,
+      "step": 5700
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7855328321456909,
+      "learning_rate": 3.947913387167901e-05,
+      "loss": 0.8013,
+      "step": 5705
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7000160217285156,
+      "learning_rate": 3.9462095975565226e-05,
+      "loss": 0.7631,
+      "step": 5710
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5747112035751343,
+      "learning_rate": 3.944504797794122e-05,
+      "loss": 0.6641,
+      "step": 5715
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7016122341156006,
+      "learning_rate": 3.942798989071469e-05,
+      "loss": 0.8245,
+      "step": 5720
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6330326795578003,
+      "learning_rate": 3.941092172580042e-05,
+      "loss": 0.6604,
+      "step": 5725
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6938309669494629,
+      "learning_rate": 3.939384349512019e-05,
+      "loss": 0.891,
+      "step": 5730
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6257404685020447,
+      "learning_rate": 3.937675521060284e-05,
+      "loss": 0.6852,
+      "step": 5735
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.243001937866211,
+      "learning_rate": 3.9359656884184234e-05,
+      "loss": 0.7764,
+      "step": 5740
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5932753086090088,
+      "learning_rate": 3.9342548527807225e-05,
+      "loss": 0.6175,
+      "step": 5745
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0795527696609497,
+      "learning_rate": 3.9325430153421706e-05,
+      "loss": 0.8249,
+      "step": 5750
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9690996408462524,
+      "learning_rate": 3.9308301772984535e-05,
+      "loss": 0.7415,
+      "step": 5755
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7389626502990723,
+      "learning_rate": 3.9291163398459585e-05,
+      "loss": 0.8866,
+      "step": 5760
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9238413572311401,
+      "learning_rate": 3.927401504181768e-05,
+      "loss": 0.9837,
+      "step": 5765
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.18929922580719,
+      "learning_rate": 3.925685671503665e-05,
+      "loss": 0.7166,
+      "step": 5770
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.717707633972168,
+      "learning_rate": 3.923968843010127e-05,
+      "loss": 0.8331,
+      "step": 5775
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5547253489494324,
+      "learning_rate": 3.9222510199003276e-05,
+      "loss": 0.5933,
+      "step": 5780
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.73494952917099,
+      "learning_rate": 3.920532203374134e-05,
+      "loss": 0.7895,
+      "step": 5785
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8860884308815002,
+      "learning_rate": 3.91881239463211e-05,
+      "loss": 0.757,
+      "step": 5790
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8135414719581604,
+      "learning_rate": 3.917091594875508e-05,
+      "loss": 0.8091,
+      "step": 5795
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7246249318122864,
+      "learning_rate": 3.915369805306278e-05,
+      "loss": 1.0,
+      "step": 5800
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.7181897163391113,
+      "eval_runtime": 96.9106,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 7.192,
+      "step": 5800
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6413865089416504,
+      "learning_rate": 3.913647027127057e-05,
+      "loss": 0.8138,
+      "step": 5805
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7799732089042664,
+      "learning_rate": 3.9119232615411724e-05,
+      "loss": 0.74,
+      "step": 5810
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5068621635437012,
+      "learning_rate": 3.910198509752647e-05,
+      "loss": 0.6272,
+      "step": 5815
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6522389054298401,
+      "learning_rate": 3.9084727729661864e-05,
+      "loss": 0.747,
+      "step": 5820
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.615208625793457,
+      "learning_rate": 3.9067460523871855e-05,
+      "loss": 1.0008,
+      "step": 5825
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7739670872688293,
+      "learning_rate": 3.9050183492217284e-05,
+      "loss": 0.959,
+      "step": 5830
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7364481687545776,
+      "learning_rate": 3.9032896646765837e-05,
+      "loss": 0.7106,
+      "step": 5835
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9992987513542175,
+      "learning_rate": 3.901559999959207e-05,
+      "loss": 0.6638,
+      "step": 5840
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9391655921936035,
+      "learning_rate": 3.8998293562777353e-05,
+      "loss": 0.8529,
+      "step": 5845
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5818854570388794,
+      "learning_rate": 3.8980977348409936e-05,
+      "loss": 0.8186,
+      "step": 5850
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7900370359420776,
+      "learning_rate": 3.8963651368584876e-05,
+      "loss": 0.6693,
+      "step": 5855
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6716821193695068,
+      "learning_rate": 3.8946315635404054e-05,
+      "loss": 0.7926,
+      "step": 5860
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6300959587097168,
+      "learning_rate": 3.892897016097615e-05,
+      "loss": 0.8177,
+      "step": 5865
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5477313995361328,
+      "learning_rate": 3.8911614957416675e-05,
+      "loss": 0.6463,
+      "step": 5870
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8673714995384216,
+      "learning_rate": 3.889425003684793e-05,
+      "loss": 0.7662,
+      "step": 5875
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9277421236038208,
+      "learning_rate": 3.8876875411398975e-05,
+      "loss": 0.7654,
+      "step": 5880
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.111858606338501,
+      "learning_rate": 3.885949109320567e-05,
+      "loss": 0.7782,
+      "step": 5885
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8440257906913757,
+      "learning_rate": 3.884209709441066e-05,
+      "loss": 0.7054,
+      "step": 5890
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8476880192756653,
+      "learning_rate": 3.882469342716333e-05,
+      "loss": 0.7382,
+      "step": 5895
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5822775959968567,
+      "learning_rate": 3.8807280103619826e-05,
+      "loss": 0.6164,
+      "step": 5900
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.7177068591117859,
+      "eval_runtime": 96.8009,
+      "eval_samples_per_second": 7.2,
+      "eval_steps_per_second": 7.2,
+      "step": 5900
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6871796250343323,
+      "learning_rate": 3.878985713594302e-05,
+      "loss": 0.6928,
+      "step": 5905
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7812449932098389,
+      "learning_rate": 3.8772424536302564e-05,
+      "loss": 0.8246,
+      "step": 5910
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8643015623092651,
+      "learning_rate": 3.87549823168748e-05,
+      "loss": 0.6893,
+      "step": 5915
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6011418104171753,
+      "learning_rate": 3.873753048984278e-05,
+      "loss": 0.5402,
+      "step": 5920
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6913355588912964,
+      "learning_rate": 3.8720069067396315e-05,
+      "loss": 0.6997,
+      "step": 5925
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.840412437915802,
+      "learning_rate": 3.870259806173188e-05,
+      "loss": 0.7524,
+      "step": 5930
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7205111980438232,
+      "learning_rate": 3.8685117485052645e-05,
+      "loss": 0.7316,
+      "step": 5935
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5926095843315125,
+      "learning_rate": 3.8667627349568494e-05,
+      "loss": 0.6616,
+      "step": 5940
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8431223630905151,
+      "learning_rate": 3.865012766749595e-05,
+      "loss": 0.9906,
+      "step": 5945
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8787121176719666,
+      "learning_rate": 3.863261845105822e-05,
+      "loss": 0.7688,
+      "step": 5950
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5568470358848572,
+      "learning_rate": 3.861509971248519e-05,
+      "loss": 0.7319,
+      "step": 5955
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6905276775360107,
+      "learning_rate": 3.859757146401336e-05,
+      "loss": 0.8506,
+      "step": 5960
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.733594536781311,
+      "learning_rate": 3.858003371788591e-05,
+      "loss": 0.65,
+      "step": 5965
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9627974033355713,
+      "learning_rate": 3.8562486486352614e-05,
+      "loss": 1.0121,
+      "step": 5970
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7356416583061218,
+      "learning_rate": 3.854492978166991e-05,
+      "loss": 0.698,
+      "step": 5975
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6784470081329346,
+      "learning_rate": 3.8527363616100844e-05,
+      "loss": 0.6361,
+      "step": 5980
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8226351141929626,
+      "learning_rate": 3.8509788001915035e-05,
+      "loss": 0.7528,
+      "step": 5985
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.026175618171692,
+      "learning_rate": 3.8492202951388744e-05,
+      "loss": 1.0142,
+      "step": 5990
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.860149085521698,
+      "learning_rate": 3.847460847680482e-05,
+      "loss": 0.6081,
+      "step": 5995
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7153595089912415,
+      "learning_rate": 3.845700459045266e-05,
+      "loss": 0.7484,
+      "step": 6000
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.7146974802017212,
+      "eval_runtime": 96.5929,
+      "eval_samples_per_second": 7.216,
+      "eval_steps_per_second": 7.216,
+      "step": 6000
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8637629151344299,
+      "learning_rate": 3.843939130462827e-05,
+      "loss": 0.9717,
+      "step": 6005
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39873212575912476,
+      "learning_rate": 3.84217686316342e-05,
+      "loss": 0.64,
+      "step": 6010
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5839079022407532,
+      "learning_rate": 3.8404136583779585e-05,
+      "loss": 0.8288,
+      "step": 6015
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.2063088417053223,
+      "learning_rate": 3.8386495173380064e-05,
+      "loss": 0.7722,
+      "step": 6020
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6402775049209595,
+      "learning_rate": 3.836884441275786e-05,
+      "loss": 0.7645,
+      "step": 6025
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8476734161376953,
+      "learning_rate": 3.83511843142417e-05,
+      "loss": 0.6249,
+      "step": 6030
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7386351823806763,
+      "learning_rate": 3.833351489016684e-05,
+      "loss": 0.6188,
+      "step": 6035
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.661152184009552,
+      "learning_rate": 3.831583615287504e-05,
+      "loss": 0.9277,
+      "step": 6040
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6577073335647583,
+      "learning_rate": 3.82981481147146e-05,
+      "loss": 0.7678,
+      "step": 6045
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9322777986526489,
+      "learning_rate": 3.828045078804027e-05,
+      "loss": 0.6959,
+      "step": 6050
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.2478550672531128,
+      "learning_rate": 3.826274418521332e-05,
+      "loss": 0.9087,
+      "step": 6055
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5843325257301331,
+      "learning_rate": 3.824502831860148e-05,
+      "loss": 0.5186,
+      "step": 6060
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4977940618991852,
+      "learning_rate": 3.822730320057898e-05,
+      "loss": 0.7431,
+      "step": 6065
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7447279691696167,
+      "learning_rate": 3.8209568843526475e-05,
+      "loss": 0.7302,
+      "step": 6070
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5054610967636108,
+      "learning_rate": 3.8191825259831095e-05,
+      "loss": 0.7878,
+      "step": 6075
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.2159337997436523,
+      "learning_rate": 3.8174072461886396e-05,
+      "loss": 0.663,
+      "step": 6080
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8180432915687561,
+      "learning_rate": 3.815631046209241e-05,
+      "loss": 0.7184,
+      "step": 6085
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8640487790107727,
+      "learning_rate": 3.8138539272855544e-05,
+      "loss": 0.7011,
+      "step": 6090
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.569340705871582,
+      "learning_rate": 3.812075890658867e-05,
+      "loss": 0.734,
+      "step": 6095
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.774334728717804,
+      "learning_rate": 3.810296937571103e-05,
+      "loss": 0.7924,
+      "step": 6100
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.7144004106521606,
+      "eval_runtime": 96.5311,
+      "eval_samples_per_second": 7.22,
+      "eval_steps_per_second": 7.22,
+      "step": 6100
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6256681680679321,
+      "learning_rate": 3.80851706926483e-05,
+      "loss": 0.6543,
+      "step": 6105
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6679534912109375,
+      "learning_rate": 3.806736286983254e-05,
+      "loss": 0.6674,
+      "step": 6110
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.696412980556488,
+      "learning_rate": 3.804954591970218e-05,
+      "loss": 0.8335,
+      "step": 6115
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8444018363952637,
+      "learning_rate": 3.803171985470203e-05,
+      "loss": 0.7172,
+      "step": 6120
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8965733051300049,
+      "learning_rate": 3.8013884687283274e-05,
+      "loss": 1.0342,
+      "step": 6125
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7367607355117798,
+      "learning_rate": 3.799604042990347e-05,
+      "loss": 0.6862,
+      "step": 6130
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8707432746887207,
+      "learning_rate": 3.797818709502647e-05,
+      "loss": 0.8098,
+      "step": 6135
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5112609267234802,
+      "learning_rate": 3.7960324695122544e-05,
+      "loss": 0.5907,
+      "step": 6140
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39674660563468933,
+      "learning_rate": 3.794245324266823e-05,
+      "loss": 0.6887,
+      "step": 6145
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7135210037231445,
+      "learning_rate": 3.79245727501464e-05,
+      "loss": 0.7683,
+      "step": 6150
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6788827776908875,
+      "learning_rate": 3.790668323004628e-05,
+      "loss": 0.9064,
+      "step": 6155
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5234334468841553,
+      "learning_rate": 3.7888784694863355e-05,
+      "loss": 0.7373,
+      "step": 6160
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6748809218406677,
+      "learning_rate": 3.7870877157099425e-05,
+      "loss": 0.6998,
+      "step": 6165
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7746409773826599,
+      "learning_rate": 3.785296062926259e-05,
+      "loss": 0.7789,
+      "step": 6170
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7480994462966919,
+      "learning_rate": 3.783503512386721e-05,
+      "loss": 0.7114,
+      "step": 6175
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2271349430084229,
+      "learning_rate": 3.781710065343392e-05,
+      "loss": 0.7294,
+      "step": 6180
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7802749872207642,
+      "learning_rate": 3.779915723048963e-05,
+      "loss": 0.8348,
+      "step": 6185
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6718894243240356,
+      "learning_rate": 3.778120486756749e-05,
+      "loss": 0.7428,
+      "step": 6190
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7281386852264404,
+      "learning_rate": 3.776324357720688e-05,
+      "loss": 0.8076,
+      "step": 6195
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7086895704269409,
+      "learning_rate": 3.7745273371953464e-05,
+      "loss": 0.9389,
+      "step": 6200
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.7145183682441711,
+      "eval_runtime": 96.6703,
+      "eval_samples_per_second": 7.21,
+      "eval_steps_per_second": 7.21,
+      "step": 6200
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8768818974494934,
+      "learning_rate": 3.7727294264359095e-05,
+      "loss": 0.8837,
+      "step": 6205
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5558096170425415,
+      "learning_rate": 3.7709306266981835e-05,
+      "loss": 0.6894,
+      "step": 6210
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7635754942893982,
+      "learning_rate": 3.769130939238598e-05,
+      "loss": 0.7869,
+      "step": 6215
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8129850625991821,
+      "learning_rate": 3.7673303653142036e-05,
+      "loss": 0.8049,
+      "step": 6220
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7782718539237976,
+      "learning_rate": 3.765528906182666e-05,
+      "loss": 1.1669,
+      "step": 6225
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8600115180015564,
+      "learning_rate": 3.763726563102272e-05,
+      "loss": 0.7453,
+      "step": 6230
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6311535239219666,
+      "learning_rate": 3.761923337331926e-05,
+      "loss": 0.7537,
+      "step": 6235
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7079928517341614,
+      "learning_rate": 3.7601192301311486e-05,
+      "loss": 0.8818,
+      "step": 6240
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6088127493858337,
+      "learning_rate": 3.7583142427600735e-05,
+      "loss": 0.8239,
+      "step": 6245
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9693393707275391,
+      "learning_rate": 3.756508376479455e-05,
+      "loss": 0.4772,
+      "step": 6250
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9866142868995667,
+      "learning_rate": 3.754701632550656e-05,
+      "loss": 0.9703,
+      "step": 6255
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7974732518196106,
+      "learning_rate": 3.7528940122356534e-05,
+      "loss": 0.7359,
+      "step": 6260
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7444310188293457,
+      "learning_rate": 3.751085516797038e-05,
+      "loss": 0.6785,
+      "step": 6265
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.586471438407898,
+      "learning_rate": 3.7492761474980096e-05,
+      "loss": 0.6079,
+      "step": 6270
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7271062731742859,
+      "learning_rate": 3.747465905602382e-05,
+      "loss": 0.7284,
+      "step": 6275
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7004055380821228,
+      "learning_rate": 3.7456547923745745e-05,
+      "loss": 0.7364,
+      "step": 6280
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7666735053062439,
+      "learning_rate": 3.7438428090796166e-05,
+      "loss": 1.0261,
+      "step": 6285
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7594092488288879,
+      "learning_rate": 3.742029956983146e-05,
+      "loss": 0.7293,
+      "step": 6290
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5505672097206116,
+      "learning_rate": 3.7402162373514074e-05,
+      "loss": 1.1323,
+      "step": 6295
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6808300018310547,
+      "learning_rate": 3.73840165145125e-05,
+      "loss": 0.7108,
+      "step": 6300
+    },
+    {
+      "epoch": 1.01,
+      "eval_loss": 0.7136371731758118,
+      "eval_runtime": 96.6091,
+      "eval_samples_per_second": 7.215,
+      "eval_steps_per_second": 7.215,
+      "step": 6300
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.147871494293213,
+      "learning_rate": 3.736586200550129e-05,
+      "loss": 1.0234,
+      "step": 6305
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7230010628700256,
+      "learning_rate": 3.734769885916104e-05,
+      "loss": 0.6727,
+      "step": 6310
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5923539400100708,
+      "learning_rate": 3.7329527088178376e-05,
+      "loss": 0.5398,
+      "step": 6315
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.5129786729812622,
+      "learning_rate": 3.731134670524594e-05,
+      "loss": 0.8927,
+      "step": 6320
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6973879933357239,
+      "learning_rate": 3.7293157723062425e-05,
+      "loss": 0.5394,
+      "step": 6325
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6135405898094177,
+      "learning_rate": 3.727496015433246e-05,
+      "loss": 0.6949,
+      "step": 6330
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7291958928108215,
+      "learning_rate": 3.725675401176675e-05,
+      "loss": 0.7333,
+      "step": 6335
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.8721144795417786,
+      "learning_rate": 3.7238539308081956e-05,
+      "loss": 0.7178,
+      "step": 6340
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6904857754707336,
+      "learning_rate": 3.722031605600069e-05,
+      "loss": 0.5423,
+      "step": 6345
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6890765428543091,
+      "learning_rate": 3.720208426825157e-05,
+      "loss": 0.6649,
+      "step": 6350
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7029244899749756,
+      "learning_rate": 3.7183843957569174e-05,
+      "loss": 0.6566,
+      "step": 6355
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7533852458000183,
+      "learning_rate": 3.716559513669403e-05,
+      "loss": 0.6503,
+      "step": 6360
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8862756490707397,
+      "learning_rate": 3.7147337818372595e-05,
+      "loss": 0.6742,
+      "step": 6365
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.6946579217910767,
+      "learning_rate": 3.712907201535728e-05,
+      "loss": 0.6658,
+      "step": 6370
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7204101085662842,
+      "learning_rate": 3.711079774040641e-05,
+      "loss": 0.7723,
+      "step": 6375
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5266463756561279,
+      "learning_rate": 3.7092515006284245e-05,
+      "loss": 0.5676,
+      "step": 6380
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7050403952598572,
+      "learning_rate": 3.707422382576092e-05,
+      "loss": 0.6753,
+      "step": 6385
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.6561117768287659,
+      "learning_rate": 3.705592421161251e-05,
+      "loss": 0.6988,
+      "step": 6390
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5257726311683655,
+      "learning_rate": 3.7037616176620955e-05,
+      "loss": 0.6265,
+      "step": 6395
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.6152855753898621,
+      "learning_rate": 3.7019299733574074e-05,
+      "loss": 0.8076,
+      "step": 6400
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 0.7153558731079102,
+      "eval_runtime": 96.493,
+      "eval_samples_per_second": 7.223,
+      "eval_steps_per_second": 7.223,
+      "step": 6400
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.9646801948547363,
+      "learning_rate": 3.700097489526559e-05,
+      "loss": 0.8186,
+      "step": 6405
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8193671703338623,
+      "learning_rate": 3.6982641674495046e-05,
+      "loss": 0.5955,
+      "step": 6410
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5945185422897339,
+      "learning_rate": 3.696430008406788e-05,
+      "loss": 0.7449,
+      "step": 6415
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8218982219696045,
+      "learning_rate": 3.6945950136795345e-05,
+      "loss": 0.6192,
+      "step": 6420
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.723145604133606,
+      "learning_rate": 3.692759184549455e-05,
+      "loss": 0.7703,
+      "step": 6425
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8936192989349365,
+      "learning_rate": 3.690922522298843e-05,
+      "loss": 0.6683,
+      "step": 6430
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.6989235281944275,
+      "learning_rate": 3.6890850282105725e-05,
+      "loss": 0.5886,
+      "step": 6435
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.9722846150398254,
+      "learning_rate": 3.687246703568101e-05,
+      "loss": 0.6229,
+      "step": 6440
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8043028116226196,
+      "learning_rate": 3.6854075496554636e-05,
+      "loss": 0.8865,
+      "step": 6445
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7527907490730286,
+      "learning_rate": 3.683567567757276e-05,
+      "loss": 0.6773,
+      "step": 6450
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.6083170175552368,
+      "learning_rate": 3.681726759158733e-05,
+      "loss": 0.6311,
+      "step": 6455
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7950381636619568,
+      "learning_rate": 3.679885125145605e-05,
+      "loss": 0.8398,
+      "step": 6460
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8302904367446899,
+      "learning_rate": 3.678042667004239e-05,
+      "loss": 0.4981,
+      "step": 6465
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7814335227012634,
+      "learning_rate": 3.67619938602156e-05,
+      "loss": 0.8345,
+      "step": 6470
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.1472254991531372,
+      "learning_rate": 3.6743552834850656e-05,
+      "loss": 0.7503,
+      "step": 6475
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7045092582702637,
+      "learning_rate": 3.672510360682827e-05,
+      "loss": 0.9974,
+      "step": 6480
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6515605449676514,
+      "learning_rate": 3.670664618903489e-05,
+      "loss": 0.7965,
+      "step": 6485
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6281563639640808,
+      "learning_rate": 3.66881805943627e-05,
+      "loss": 0.6854,
+      "step": 6490
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6973702907562256,
+      "learning_rate": 3.666970683570957e-05,
+      "loss": 0.6939,
+      "step": 6495
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7053531408309937,
+      "learning_rate": 3.6651224925979086e-05,
+      "loss": 0.7232,
+      "step": 6500
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": 0.7147144079208374,
+      "eval_runtime": 96.4093,
+      "eval_samples_per_second": 7.23,
+      "eval_steps_per_second": 7.23,
+      "step": 6500
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.0588470697402954,
+      "learning_rate": 3.663273487808052e-05,
+      "loss": 0.7061,
+      "step": 6505
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7132044434547424,
+      "learning_rate": 3.661423670492885e-05,
+      "loss": 1.0585,
+      "step": 6510
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7809985280036926,
+      "learning_rate": 3.65957304194447e-05,
+      "loss": 0.7325,
+      "step": 6515
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.9116477966308594,
+      "learning_rate": 3.6577216034554373e-05,
+      "loss": 0.6707,
+      "step": 6520
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6966428160667419,
+      "learning_rate": 3.655869356318982e-05,
+      "loss": 0.7778,
+      "step": 6525
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5731261968612671,
+      "learning_rate": 3.654016301828867e-05,
+      "loss": 0.7336,
+      "step": 6530
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6520384550094604,
+      "learning_rate": 3.6521624412794167e-05,
+      "loss": 0.6992,
+      "step": 6535
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7069156169891357,
+      "learning_rate": 3.6503077759655166e-05,
+      "loss": 0.608,
+      "step": 6540
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7179171442985535,
+      "learning_rate": 3.648452307182618e-05,
+      "loss": 0.6944,
+      "step": 6545
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.7859477996826172,
+      "learning_rate": 3.6465960362267335e-05,
+      "loss": 0.6541,
+      "step": 6550
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.8490133881568909,
+      "learning_rate": 3.6447389643944326e-05,
+      "loss": 0.6887,
+      "step": 6555
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6254005432128906,
+      "learning_rate": 3.6428810929828466e-05,
+      "loss": 0.6112,
+      "step": 6560
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.42452389001846313,
+      "learning_rate": 3.6410224232896654e-05,
+      "loss": 0.9964,
+      "step": 6565
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.1462258100509644,
+      "learning_rate": 3.639162956613135e-05,
+      "loss": 0.896,
+      "step": 6570
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9457197785377502,
+      "learning_rate": 3.637302694252058e-05,
+      "loss": 0.8939,
+      "step": 6575
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.888251006603241,
+      "learning_rate": 3.6354416375057965e-05,
+      "loss": 0.5634,
+      "step": 6580
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6899970173835754,
+      "learning_rate": 3.6335797876742625e-05,
+      "loss": 0.814,
+      "step": 6585
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.56333327293396,
+      "learning_rate": 3.631717146057925e-05,
+      "loss": 0.5733,
+      "step": 6590
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9747548699378967,
+      "learning_rate": 3.629853713957806e-05,
+      "loss": 0.7239,
+      "step": 6595
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6559699177742004,
+      "learning_rate": 3.627989492675477e-05,
+      "loss": 0.6456,
+      "step": 6600
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 0.7122019529342651,
+      "eval_runtime": 96.4526,
+      "eval_samples_per_second": 7.226,
+      "eval_steps_per_second": 7.226,
+      "step": 6600
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.7743479609489441,
+      "learning_rate": 3.626124483513063e-05,
+      "loss": 0.8055,
+      "step": 6605
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6561269164085388,
+      "learning_rate": 3.6242586877732395e-05,
+      "loss": 0.7325,
+      "step": 6610
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6919766664505005,
+      "learning_rate": 3.6223921067592316e-05,
+      "loss": 0.6057,
+      "step": 6615
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7156655192375183,
+      "learning_rate": 3.6205247417748103e-05,
+      "loss": 0.9774,
+      "step": 6620
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.8031120896339417,
+      "learning_rate": 3.618656594124297e-05,
+      "loss": 0.8455,
+      "step": 6625
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.9751262664794922,
+      "learning_rate": 3.616787665112558e-05,
+      "loss": 0.7065,
+      "step": 6630
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.6306493282318115,
+      "learning_rate": 3.6149179560450075e-05,
+      "loss": 0.6623,
+      "step": 6635
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7967116832733154,
+      "learning_rate": 3.613047468227601e-05,
+      "loss": 0.755,
+      "step": 6640
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.1591620445251465,
+      "learning_rate": 3.6111762029668425e-05,
+      "loss": 0.7066,
+      "step": 6645
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.8396896123886108,
+      "learning_rate": 3.609304161569776e-05,
+      "loss": 0.6199,
+      "step": 6650
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.650834858417511,
+      "learning_rate": 3.607431345343987e-05,
+      "loss": 0.7941,
+      "step": 6655
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.0054168701171875,
+      "learning_rate": 3.6055577555976046e-05,
+      "loss": 0.6738,
+      "step": 6660
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.663850724697113,
+      "learning_rate": 3.603683393639297e-05,
+      "loss": 0.7506,
+      "step": 6665
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.1505427360534668,
+      "learning_rate": 3.6018082607782715e-05,
+      "loss": 0.9257,
+      "step": 6670
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.731639564037323,
+      "learning_rate": 3.599932358324274e-05,
+      "loss": 0.7686,
+      "step": 6675
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.7264305949211121,
+      "learning_rate": 3.598055687587588e-05,
+      "loss": 0.777,
+      "step": 6680
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6651543378829956,
+      "learning_rate": 3.5961782498790356e-05,
+      "loss": 0.6417,
+      "step": 6685
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6642833948135376,
+      "learning_rate": 3.594300046509971e-05,
+      "loss": 0.7173,
+      "step": 6690
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.7369661927223206,
+      "learning_rate": 3.592421078792285e-05,
+      "loss": 0.6174,
+      "step": 6695
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6418754458427429,
+      "learning_rate": 3.590541348038403e-05,
+      "loss": 0.5862,
+      "step": 6700
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 0.713315486907959,
+      "eval_runtime": 96.4112,
+      "eval_samples_per_second": 7.229,
+      "eval_steps_per_second": 7.229,
+      "step": 6700
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.8054782152175903,
+      "learning_rate": 3.588660855561282e-05,
+      "loss": 0.9726,
+      "step": 6705
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6561216115951538,
+      "learning_rate": 3.586779602674413e-05,
+      "loss": 0.7284,
+      "step": 6710
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.7178167700767517,
+      "learning_rate": 3.584897590691815e-05,
+      "loss": 0.7098,
+      "step": 6715
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6914810538291931,
+      "learning_rate": 3.58301482092804e-05,
+      "loss": 0.6683,
+      "step": 6720
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.605938196182251,
+      "learning_rate": 3.5811312946981676e-05,
+      "loss": 0.7859,
+      "step": 6725
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.7012625336647034,
+      "learning_rate": 3.579247013317809e-05,
+      "loss": 0.5396,
+      "step": 6730
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8261893391609192,
+      "learning_rate": 3.5773619781030965e-05,
+      "loss": 0.7595,
+      "step": 6735
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7922738194465637,
+      "learning_rate": 3.5754761903706965e-05,
+      "loss": 0.6672,
+      "step": 6740
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5855022668838501,
+      "learning_rate": 3.573589651437795e-05,
+      "loss": 0.8951,
+      "step": 6745
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.1443228721618652,
+      "learning_rate": 3.571702362622106e-05,
+      "loss": 0.834,
+      "step": 6750
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.4697206020355225,
+      "learning_rate": 3.5698143252418645e-05,
+      "loss": 0.7024,
+      "step": 6755
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.629658043384552,
+      "learning_rate": 3.567925540615833e-05,
+      "loss": 0.7174,
+      "step": 6760
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.3336237668991089,
+      "learning_rate": 3.5660360100632914e-05,
+      "loss": 0.7339,
+      "step": 6765
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8946998715400696,
+      "learning_rate": 3.564145734904044e-05,
+      "loss": 0.7148,
+      "step": 6770
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.2121728658676147,
+      "learning_rate": 3.562254716458411e-05,
+      "loss": 0.9227,
+      "step": 6775
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6504545211791992,
+      "learning_rate": 3.560362956047235e-05,
+      "loss": 0.7391,
+      "step": 6780
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6672701239585876,
+      "learning_rate": 3.5584704549918776e-05,
+      "loss": 0.6783,
+      "step": 6785
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8841736316680908,
+      "learning_rate": 3.556577214614216e-05,
+      "loss": 0.6379,
+      "step": 6790
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7543352842330933,
+      "learning_rate": 3.554683236236642e-05,
+      "loss": 0.921,
+      "step": 6795
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6553550958633423,
+      "learning_rate": 3.552788521182067e-05,
+      "loss": 0.6935,
+      "step": 6800
+    },
+    {
+      "epoch": 1.09,
+      "eval_loss": 0.7111696600914001,
+      "eval_runtime": 96.419,
+      "eval_samples_per_second": 7.229,
+      "eval_steps_per_second": 7.229,
+      "step": 6800
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7093417644500732,
+      "learning_rate": 3.550893070773914e-05,
+      "loss": 0.5619,
+      "step": 6805
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6494777798652649,
+      "learning_rate": 3.5489968863361215e-05,
+      "loss": 0.7891,
+      "step": 6810
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6380666494369507,
+      "learning_rate": 3.547099969193137e-05,
+      "loss": 0.6761,
+      "step": 6815
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7495108842849731,
+      "learning_rate": 3.545202320669927e-05,
+      "loss": 0.69,
+      "step": 6820
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.9093177914619446,
+      "learning_rate": 3.543303942091961e-05,
+      "loss": 0.7455,
+      "step": 6825
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7216375470161438,
+      "learning_rate": 3.541404834785222e-05,
+      "loss": 0.8079,
+      "step": 6830
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.1701329946517944,
+      "learning_rate": 3.539505000076203e-05,
+      "loss": 1.0064,
+      "step": 6835
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6244146227836609,
+      "learning_rate": 3.5376044392919037e-05,
+      "loss": 0.8856,
+      "step": 6840
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7619176506996155,
+      "learning_rate": 3.535703153759832e-05,
+      "loss": 0.615,
+      "step": 6845
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5909354090690613,
+      "learning_rate": 3.533801144807998e-05,
+      "loss": 0.6205,
+      "step": 6850
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5859089493751526,
+      "learning_rate": 3.531898413764923e-05,
+      "loss": 0.6251,
+      "step": 6855
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7175759673118591,
+      "learning_rate": 3.52999496195963e-05,
+      "loss": 0.8056,
+      "step": 6860
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.8045479655265808,
+      "learning_rate": 3.528090790721643e-05,
+      "loss": 0.6239,
+      "step": 6865
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5341171622276306,
+      "learning_rate": 3.5261859013809934e-05,
+      "loss": 0.5879,
+      "step": 6870
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7800705432891846,
+      "learning_rate": 3.5242802952682106e-05,
+      "loss": 0.7499,
+      "step": 6875
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6275398135185242,
+      "learning_rate": 3.5223739737143256e-05,
+      "loss": 0.578,
+      "step": 6880
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.110107660293579,
+      "learning_rate": 3.52046693805087e-05,
+      "loss": 0.7843,
+      "step": 6885
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6027718782424927,
+      "learning_rate": 3.5185591896098726e-05,
+      "loss": 0.5948,
+      "step": 6890
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.8754519820213318,
+      "learning_rate": 3.5166507297238615e-05,
+      "loss": 0.6245,
+      "step": 6895
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.006706953048706,
+      "learning_rate": 3.51474155972586e-05,
+      "loss": 0.7522,
+      "step": 6900
+    },
+    {
+      "epoch": 1.1,
+      "eval_loss": 0.7102616429328918,
+      "eval_runtime": 96.4458,
+      "eval_samples_per_second": 7.227,
+      "eval_steps_per_second": 7.227,
+      "step": 6900
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7887657880783081,
+      "learning_rate": 3.51283168094939e-05,
+      "loss": 0.8036,
+      "step": 6905
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7712429761886597,
+      "learning_rate": 3.5109210947284656e-05,
+      "loss": 0.7136,
+      "step": 6910
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6748402118682861,
+      "learning_rate": 3.509009802397598e-05,
+      "loss": 0.881,
+      "step": 6915
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6241844296455383,
+      "learning_rate": 3.5070978052917885e-05,
+      "loss": 0.8814,
+      "step": 6920
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.8334391117095947,
+      "learning_rate": 3.5051851047465336e-05,
+      "loss": 0.7423,
+      "step": 6925
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6817840337753296,
+      "learning_rate": 3.503271702097818e-05,
+      "loss": 0.5209,
+      "step": 6930
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.9585073590278625,
+      "learning_rate": 3.50135759868212e-05,
+      "loss": 0.9416,
+      "step": 6935
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7557898163795471,
+      "learning_rate": 3.499442795836405e-05,
+      "loss": 0.6085,
+      "step": 6940
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.8940297365188599,
+      "learning_rate": 3.497527294898129e-05,
+      "loss": 0.6232,
+      "step": 6945
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6422560214996338,
+      "learning_rate": 3.495611097205232e-05,
+      "loss": 0.6165,
+      "step": 6950
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.92333984375,
+      "learning_rate": 3.493694204096147e-05,
+      "loss": 0.7409,
+      "step": 6955
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.8350429534912109,
+      "learning_rate": 3.491776616909786e-05,
+      "loss": 0.6304,
+      "step": 6960
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.8454934358596802,
+      "learning_rate": 3.48985833698555e-05,
+      "loss": 0.8221,
+      "step": 6965
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6815975904464722,
+      "learning_rate": 3.487939365663321e-05,
+      "loss": 0.7426,
+      "step": 6970
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5304132699966431,
+      "learning_rate": 3.4860197042834674e-05,
+      "loss": 0.6932,
+      "step": 6975
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.9002396464347839,
+      "learning_rate": 3.4840993541868364e-05,
+      "loss": 0.752,
+      "step": 6980
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.8047484755516052,
+      "learning_rate": 3.482178316714758e-05,
+      "loss": 0.8861,
+      "step": 6985
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.8372194170951843,
+      "learning_rate": 3.4802565932090405e-05,
+      "loss": 0.6855,
+      "step": 6990
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.7594023942947388,
+      "learning_rate": 3.4783341850119754e-05,
+      "loss": 0.7424,
+      "step": 6995
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.8143925070762634,
+      "learning_rate": 3.476411093466327e-05,
+      "loss": 1.0525,
+      "step": 7000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 0.7106046080589294,
+      "eval_runtime": 96.364,
+      "eval_samples_per_second": 7.233,
+      "eval_steps_per_second": 7.233,
+      "step": 7000
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2525312900543213,
+      "learning_rate": 3.474487319915342e-05,
+      "loss": 0.7996,
+      "step": 7005
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.9011229276657104,
+      "learning_rate": 3.4725628657027386e-05,
+      "loss": 0.7672,
+      "step": 7010
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.46732136607170105,
+      "learning_rate": 3.4706377321727156e-05,
+      "loss": 0.9557,
+      "step": 7015
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.7189437747001648,
+      "learning_rate": 3.468711920669942e-05,
+      "loss": 0.6853,
+      "step": 7020
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.7930631041526794,
+      "learning_rate": 3.466785432539562e-05,
+      "loss": 0.9078,
+      "step": 7025
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.3144121170043945,
+      "learning_rate": 3.464858269127193e-05,
+      "loss": 0.8064,
+      "step": 7030
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.7118679881095886,
+      "learning_rate": 3.462930431778924e-05,
+      "loss": 0.7595,
+      "step": 7035
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.0305006504058838,
+      "learning_rate": 3.461001921841314e-05,
+      "loss": 0.8286,
+      "step": 7040
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.4701656997203827,
+      "learning_rate": 3.45907274066139e-05,
+      "loss": 0.6739,
+      "step": 7045
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7981716990470886,
+      "learning_rate": 3.457142889586653e-05,
+      "loss": 0.8902,
+      "step": 7050
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.9665462970733643,
+      "learning_rate": 3.455212369965066e-05,
+      "loss": 0.8668,
+      "step": 7055
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5912620425224304,
+      "learning_rate": 3.453281183145063e-05,
+      "loss": 0.515,
+      "step": 7060
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.9274890422821045,
+      "learning_rate": 3.451349330475543e-05,
+      "loss": 0.7517,
+      "step": 7065
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.968731164932251,
+      "learning_rate": 3.449416813305869e-05,
+      "loss": 0.5942,
+      "step": 7070
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.8442440032958984,
+      "learning_rate": 3.447483632985871e-05,
+      "loss": 0.9199,
+      "step": 7075
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7139914035797119,
+      "learning_rate": 3.445549790865838e-05,
+      "loss": 0.7511,
+      "step": 7080
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.8142327666282654,
+      "learning_rate": 3.443615288296524e-05,
+      "loss": 1.0244,
+      "step": 7085
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7862592935562134,
+      "learning_rate": 3.441680126629144e-05,
+      "loss": 0.4681,
+      "step": 7090
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.774732232093811,
+      "learning_rate": 3.439744307215374e-05,
+      "loss": 0.7968,
+      "step": 7095
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7824143767356873,
+      "learning_rate": 3.437807831407346e-05,
+      "loss": 0.8285,
+      "step": 7100
+    },
+    {
+      "epoch": 1.13,
+      "eval_loss": 0.7099367380142212,
+      "eval_runtime": 96.408,
+      "eval_samples_per_second": 7.23,
+      "eval_steps_per_second": 7.23,
+      "step": 7100
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.6416094303131104,
+      "learning_rate": 3.435870700557655e-05,
+      "loss": 0.8833,
+      "step": 7105
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.0388530492782593,
+      "learning_rate": 3.433932916019352e-05,
+      "loss": 0.8625,
+      "step": 7110
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.665253221988678,
+      "learning_rate": 3.431994479145944e-05,
+      "loss": 0.7174,
+      "step": 7115
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.6002438068389893,
+      "learning_rate": 3.4300553912913916e-05,
+      "loss": 0.5811,
+      "step": 7120
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.6728983521461487,
+      "learning_rate": 3.428115653810114e-05,
+      "loss": 0.7784,
+      "step": 7125
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.487757921218872,
+      "learning_rate": 3.4261752680569824e-05,
+      "loss": 0.8581,
+      "step": 7130
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.7253740429878235,
+      "learning_rate": 3.424234235387322e-05,
+      "loss": 0.6423,
+      "step": 7135
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0124573707580566,
+      "learning_rate": 3.422292557156907e-05,
+      "loss": 0.951,
+      "step": 7140
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.8678650856018066,
+      "learning_rate": 3.4203502347219626e-05,
+      "loss": 0.7357,
+      "step": 7145
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.793936550617218,
+      "learning_rate": 3.41840726943917e-05,
+      "loss": 0.631,
+      "step": 7150
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.7116626501083374,
+      "learning_rate": 3.416463662665652e-05,
+      "loss": 0.8185,
+      "step": 7155
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.9050395488739014,
+      "learning_rate": 3.414519415758983e-05,
+      "loss": 0.6293,
+      "step": 7160
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.674711287021637,
+      "learning_rate": 3.412574530077185e-05,
+      "loss": 0.6727,
+      "step": 7165
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.682181179523468,
+      "learning_rate": 3.4106290069787236e-05,
+      "loss": 0.7405,
+      "step": 7170
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8667547702789307,
+      "learning_rate": 3.408682847822512e-05,
+      "loss": 0.6902,
+      "step": 7175
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8550475835800171,
+      "learning_rate": 3.406736053967907e-05,
+      "loss": 0.7606,
+      "step": 7180
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8812040686607361,
+      "learning_rate": 3.404788626774708e-05,
+      "loss": 0.9635,
+      "step": 7185
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.7503578066825867,
+      "learning_rate": 3.402840567603158e-05,
+      "loss": 0.8821,
+      "step": 7190
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.757117748260498,
+      "learning_rate": 3.400891877813941e-05,
+      "loss": 0.8054,
+      "step": 7195
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6797623038291931,
+      "learning_rate": 3.39894255876818e-05,
+      "loss": 0.6116,
+      "step": 7200
+    },
+    {
+      "epoch": 1.15,
+      "eval_loss": 0.7078786492347717,
+      "eval_runtime": 96.401,
+      "eval_samples_per_second": 7.23,
+      "eval_steps_per_second": 7.23,
+      "step": 7200
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.7826797366142273,
+      "learning_rate": 3.39699261182744e-05,
+      "loss": 0.6275,
+      "step": 7205
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.9613910913467407,
+      "learning_rate": 3.395042038353723e-05,
+      "loss": 0.696,
+      "step": 7210
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.7311391234397888,
+      "learning_rate": 3.3930908397094696e-05,
+      "loss": 0.7455,
+      "step": 7215
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8681967854499817,
+      "learning_rate": 3.391139017257555e-05,
+      "loss": 0.617,
+      "step": 7220
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.2949939966201782,
+      "learning_rate": 3.3891865723612926e-05,
+      "loss": 0.8595,
+      "step": 7225
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.1373263597488403,
+      "learning_rate": 3.3872335063844295e-05,
+      "loss": 0.5906,
+      "step": 7230
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6243600845336914,
+      "learning_rate": 3.385279820691147e-05,
+      "loss": 0.8558,
+      "step": 7235
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7297492623329163,
+      "learning_rate": 3.383325516646058e-05,
+      "loss": 0.6055,
+      "step": 7240
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8958369493484497,
+      "learning_rate": 3.3813705956142085e-05,
+      "loss": 0.714,
+      "step": 7245
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8226301670074463,
+      "learning_rate": 3.3794150589610763e-05,
+      "loss": 0.7724,
+      "step": 7250
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7973611354827881,
+      "learning_rate": 3.377458908052566e-05,
+      "loss": 0.696,
+      "step": 7255
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.0526561737060547,
+      "learning_rate": 3.3755021442550146e-05,
+      "loss": 0.7198,
+      "step": 7260
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7639563083648682,
+      "learning_rate": 3.373544768935186e-05,
+      "loss": 0.6681,
+      "step": 7265
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8041620850563049,
+      "learning_rate": 3.3715867834602715e-05,
+      "loss": 0.7158,
+      "step": 7270
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.0435208082199097,
+      "learning_rate": 3.369628189197887e-05,
+      "loss": 0.719,
+      "step": 7275
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.6697341203689575,
+      "learning_rate": 3.3676689875160755e-05,
+      "loss": 0.5694,
+      "step": 7280
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7368444800376892,
+      "learning_rate": 3.365709179783304e-05,
+      "loss": 0.646,
+      "step": 7285
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.7546617984771729,
+      "learning_rate": 3.363748767368463e-05,
+      "loss": 0.7492,
+      "step": 7290
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8340469002723694,
+      "learning_rate": 3.3617877516408625e-05,
+      "loss": 0.753,
+      "step": 7295
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6987440586090088,
+      "learning_rate": 3.3598261339702374e-05,
+      "loss": 0.5617,
+      "step": 7300
+    },
+    {
+      "epoch": 1.17,
+      "eval_loss": 0.708717405796051,
+      "eval_runtime": 96.3901,
+      "eval_samples_per_second": 7.231,
+      "eval_steps_per_second": 7.231,
+      "step": 7300
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.7812116742134094,
+      "learning_rate": 3.357863915726743e-05,
+      "loss": 0.5933,
+      "step": 7305
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.8569033741950989,
+      "learning_rate": 3.3559010982809526e-05,
+      "loss": 0.7012,
+      "step": 7310
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5553389191627502,
+      "learning_rate": 3.3539376830038567e-05,
+      "loss": 0.5819,
+      "step": 7315
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.8315708637237549,
+      "learning_rate": 3.351973671266867e-05,
+      "loss": 0.6821,
+      "step": 7320
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.828441321849823,
+      "learning_rate": 3.350009064441809e-05,
+      "loss": 0.6926,
+      "step": 7325
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5075239539146423,
+      "learning_rate": 3.348043863900924e-05,
+      "loss": 0.64,
+      "step": 7330
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.7042556405067444,
+      "learning_rate": 3.34607807101687e-05,
+      "loss": 0.7332,
+      "step": 7335
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.8633832931518555,
+      "learning_rate": 3.344111687162718e-05,
+      "loss": 1.0268,
+      "step": 7340
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.3040523529052734,
+      "learning_rate": 3.342144713711949e-05,
+      "loss": 0.9377,
+      "step": 7345
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.7739453315734863,
+      "learning_rate": 3.34017715203846e-05,
+      "loss": 0.863,
+      "step": 7350
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5962020754814148,
+      "learning_rate": 3.338209003516556e-05,
+      "loss": 0.5611,
+      "step": 7355
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6497868299484253,
+      "learning_rate": 3.336240269520953e-05,
+      "loss": 0.8599,
+      "step": 7360
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.0249778032302856,
+      "learning_rate": 3.334270951426776e-05,
+      "loss": 0.7472,
+      "step": 7365
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.7308823466300964,
+      "learning_rate": 3.332301050609558e-05,
+      "loss": 0.692,
+      "step": 7370
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.7353125214576721,
+      "learning_rate": 3.3303305684452377e-05,
+      "loss": 0.6405,
+      "step": 7375
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.055648684501648,
+      "learning_rate": 3.328359506310162e-05,
+      "loss": 0.5795,
+      "step": 7380
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.2591193914413452,
+      "learning_rate": 3.326387865581082e-05,
+      "loss": 0.7434,
+      "step": 7385
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.751158595085144,
+      "learning_rate": 3.3244156476351525e-05,
+      "loss": 0.7315,
+      "step": 7390
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.7036099433898926,
+      "learning_rate": 3.322442853849931e-05,
+      "loss": 0.7636,
+      "step": 7395
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6783479452133179,
+      "learning_rate": 3.320469485603378e-05,
+      "loss": 0.6514,
+      "step": 7400
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 0.7071788311004639,
+      "eval_runtime": 96.4058,
+      "eval_samples_per_second": 7.23,
+      "eval_steps_per_second": 7.23,
+      "step": 7400
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.9339218735694885,
+      "learning_rate": 3.318495544273857e-05,
+      "loss": 0.7816,
+      "step": 7405
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6175532937049866,
+      "learning_rate": 3.316521031240128e-05,
+      "loss": 0.5474,
+      "step": 7410
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.8843727707862854,
+      "learning_rate": 3.314545947881353e-05,
+      "loss": 0.7105,
+      "step": 7415
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.0864967107772827,
+      "learning_rate": 3.312570295577093e-05,
+      "loss": 0.8522,
+      "step": 7420
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6682702302932739,
+      "learning_rate": 3.310594075707303e-05,
+      "loss": 0.5584,
+      "step": 7425
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.801595151424408,
+      "learning_rate": 3.308617289652337e-05,
+      "loss": 0.8889,
+      "step": 7430
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.8389090299606323,
+      "learning_rate": 3.306639938792945e-05,
+      "loss": 0.7226,
+      "step": 7435
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.5336723327636719,
+      "learning_rate": 3.30466202451027e-05,
+      "loss": 0.8794,
+      "step": 7440
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.8237048387527466,
+      "learning_rate": 3.30268354818585e-05,
+      "loss": 0.7764,
+      "step": 7445
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.9008681178092957,
+      "learning_rate": 3.3007045112016124e-05,
+      "loss": 0.7336,
+      "step": 7450
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6193007826805115,
+      "learning_rate": 3.29872491493988e-05,
+      "loss": 0.8906,
+      "step": 7455
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.7680274248123169,
+      "learning_rate": 3.296744760783365e-05,
+      "loss": 0.7991,
+      "step": 7460
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": NaN,
+      "learning_rate": 3.2951602367033386e-05,
+      "loss": 0.8565,
+      "step": 7465
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6190903782844543,
+      "learning_rate": 3.293179081821891e-05,
+      "loss": 0.8935,
+      "step": 7470
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.7620663046836853,
+      "learning_rate": 3.291197372919322e-05,
+      "loss": 0.7041,
+      "step": 7475
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.7184949517250061,
+      "learning_rate": 3.289215111379821e-05,
+      "loss": 0.6436,
+      "step": 7480
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.8207274079322815,
+      "learning_rate": 3.287232298587961e-05,
+      "loss": 0.8094,
+      "step": 7485
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8164545297622681,
+      "learning_rate": 3.2852489359287016e-05,
+      "loss": 0.8277,
+      "step": 7490
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.776091992855072,
+      "learning_rate": 3.283265024787383e-05,
+      "loss": 0.9329,
+      "step": 7495
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8750385046005249,
+      "learning_rate": 3.281280566549733e-05,
+      "loss": 0.6729,
+      "step": 7500
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 0.7052386403083801,
+      "eval_runtime": 96.3606,
+      "eval_samples_per_second": 7.233,
+      "eval_steps_per_second": 7.233,
+      "step": 7500
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6827027797698975,
+      "learning_rate": 3.27929556260186e-05,
+      "loss": 0.5928,
+      "step": 7505
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7305205464363098,
+      "learning_rate": 3.2773100143302504e-05,
+      "loss": 0.8407,
+      "step": 7510
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.0960073471069336,
+      "learning_rate": 3.2753239231217745e-05,
+      "loss": 0.6898,
+      "step": 7515
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7878029942512512,
+      "learning_rate": 3.273337290363683e-05,
+      "loss": 0.654,
+      "step": 7520
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5136789679527283,
+      "learning_rate": 3.2713501174436e-05,
+      "loss": 0.5858,
+      "step": 7525
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.817991316318512,
+      "learning_rate": 3.269362405749531e-05,
+      "loss": 0.5891,
+      "step": 7530
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8485821485519409,
+      "learning_rate": 3.267374156669855e-05,
+      "loss": 0.7298,
+      "step": 7535
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.862983226776123,
+      "learning_rate": 3.265385371593329e-05,
+      "loss": 0.9098,
+      "step": 7540
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7313618063926697,
+      "learning_rate": 3.263396051909083e-05,
+      "loss": 0.9118,
+      "step": 7545
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6416789293289185,
+      "learning_rate": 3.2614061990066193e-05,
+      "loss": 0.6785,
+      "step": 7550
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.7634716033935547,
+      "learning_rate": 3.2594158142758146e-05,
+      "loss": 0.7872,
+      "step": 7555
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.9261329174041748,
+      "learning_rate": 3.257424899106917e-05,
+      "loss": 0.7336,
+      "step": 7560
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.7905421257019043,
+      "learning_rate": 3.255433454890544e-05,
+      "loss": 0.4871,
+      "step": 7565
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.7929773926734924,
+      "learning_rate": 3.2534414830176825e-05,
+      "loss": 0.7026,
+      "step": 7570
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.7353596091270447,
+      "learning_rate": 3.251448984879689e-05,
+      "loss": 0.657,
+      "step": 7575
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5401962399482727,
+      "learning_rate": 3.2494559618682867e-05,
+      "loss": 0.9139,
+      "step": 7580
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.8317365646362305,
+      "learning_rate": 3.2474624153755665e-05,
+      "loss": 0.5765,
+      "step": 7585
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.9745128750801086,
+      "learning_rate": 3.2454683467939834e-05,
+      "loss": 0.7377,
+      "step": 7590
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.9197364449501038,
+      "learning_rate": 3.2434737575163585e-05,
+      "loss": 0.8045,
+      "step": 7595
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5326982140541077,
+      "learning_rate": 3.241478648935876e-05,
+      "loss": 0.6401,
+      "step": 7600
+    },
+    {
+      "epoch": 1.21,
+      "eval_loss": 0.7055444121360779,
+      "eval_runtime": 96.2986,
+      "eval_samples_per_second": 7.238,
+      "eval_steps_per_second": 7.238,
+      "step": 7600
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.8797016143798828,
+      "learning_rate": 3.2394830224460837e-05,
+      "loss": 0.6938,
+      "step": 7605
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5764625668525696,
+      "learning_rate": 3.2374868794408884e-05,
+      "loss": 0.5537,
+      "step": 7610
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.7063139081001282,
+      "learning_rate": 3.235490221314562e-05,
+      "loss": 0.4823,
+      "step": 7615
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6020994186401367,
+      "learning_rate": 3.23349304946173e-05,
+      "loss": 0.4869,
+      "step": 7620
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.8765579462051392,
+      "learning_rate": 3.231495365277385e-05,
+      "loss": 0.9822,
+      "step": 7625
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.7133936882019043,
+      "learning_rate": 3.22949717015687e-05,
+      "loss": 0.7893,
+      "step": 7630
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.1999497413635254,
+      "learning_rate": 3.227498465495888e-05,
+      "loss": 0.7949,
+      "step": 7635
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.8327472805976868,
+      "learning_rate": 3.225499252690499e-05,
+      "loss": 0.8728,
+      "step": 7640
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.9887890219688416,
+      "learning_rate": 3.2234995331371166e-05,
+      "loss": 0.764,
+      "step": 7645
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.9890260696411133,
+      "learning_rate": 3.2214993082325073e-05,
+      "loss": 0.8756,
+      "step": 7650
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.651091456413269,
+      "learning_rate": 3.2194985793737924e-05,
+      "loss": 0.6986,
+      "step": 7655
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.0091456174850464,
+      "learning_rate": 3.2174973479584446e-05,
+      "loss": 0.6858,
+      "step": 7660
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.0353959798812866,
+      "learning_rate": 3.215495615384287e-05,
+      "loss": 0.7436,
+      "step": 7665
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.9490379691123962,
+      "learning_rate": 3.213493383049494e-05,
+      "loss": 0.6652,
+      "step": 7670
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.9487096667289734,
+      "learning_rate": 3.211490652352589e-05,
+      "loss": 0.7113,
+      "step": 7675
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8836408853530884,
+      "learning_rate": 3.20948742469244e-05,
+      "loss": 0.7104,
+      "step": 7680
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.9050471782684326,
+      "learning_rate": 3.207483701468268e-05,
+      "loss": 0.7832,
+      "step": 7685
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.7933790683746338,
+      "learning_rate": 3.205479484079635e-05,
+      "loss": 0.7119,
+      "step": 7690
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8188216090202332,
+      "learning_rate": 3.203474773926453e-05,
+      "loss": 0.7536,
+      "step": 7695
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.7438432574272156,
+      "learning_rate": 3.201469572408973e-05,
+      "loss": 0.8089,
+      "step": 7700
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.7051511406898499,
+      "eval_runtime": 96.3295,
+      "eval_samples_per_second": 7.236,
+      "eval_steps_per_second": 7.236,
+      "step": 7700
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.9201672673225403,
+      "learning_rate": 3.1994638809277915e-05,
+      "loss": 1.0177,
+      "step": 7705
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.7813202738761902,
+      "learning_rate": 3.197457700883849e-05,
+      "loss": 0.476,
+      "step": 7710
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8475176692008972,
+      "learning_rate": 3.195451033678425e-05,
+      "loss": 0.7233,
+      "step": 7715
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8281943202018738,
+      "learning_rate": 3.19344388071314e-05,
+      "loss": 0.7427,
+      "step": 7720
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8039165735244751,
+      "learning_rate": 3.191436243389954e-05,
+      "loss": 0.6422,
+      "step": 7725
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5459123849868774,
+      "learning_rate": 3.1894281231111644e-05,
+      "loss": 0.6259,
+      "step": 7730
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.7360697984695435,
+      "learning_rate": 3.187419521279407e-05,
+      "loss": 0.6194,
+      "step": 7735
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6768085956573486,
+      "learning_rate": 3.185410439297653e-05,
+      "loss": 0.6967,
+      "step": 7740
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7461891174316406,
+      "learning_rate": 3.1834008785692104e-05,
+      "loss": 0.7128,
+      "step": 7745
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7571334838867188,
+      "learning_rate": 3.18139084049772e-05,
+      "loss": 0.6859,
+      "step": 7750
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.8840231895446777,
+      "learning_rate": 3.1793803264871575e-05,
+      "loss": 0.7524,
+      "step": 7755
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5408929586410522,
+      "learning_rate": 3.1773693379418296e-05,
+      "loss": 0.55,
+      "step": 7760
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5577660202980042,
+      "learning_rate": 3.175357876266375e-05,
+      "loss": 0.6207,
+      "step": 7765
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.8528269529342651,
+      "learning_rate": 3.1733459428657637e-05,
+      "loss": 0.513,
+      "step": 7770
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.616848886013031,
+      "learning_rate": 3.171333539145294e-05,
+      "loss": 0.8487,
+      "step": 7775
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7524570226669312,
+      "learning_rate": 3.169320666510593e-05,
+      "loss": 0.7275,
+      "step": 7780
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6796889305114746,
+      "learning_rate": 3.167307326367616e-05,
+      "loss": 0.4541,
+      "step": 7785
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7805665135383606,
+      "learning_rate": 3.165293520122643e-05,
+      "loss": 0.9711,
+      "step": 7790
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7790519595146179,
+      "learning_rate": 3.1632792491822835e-05,
+      "loss": 0.5531,
+      "step": 7795
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.93754643201828,
+      "learning_rate": 3.161264514953467e-05,
+      "loss": 0.8166,
+      "step": 7800
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 0.7040937542915344,
+      "eval_runtime": 96.3461,
+      "eval_samples_per_second": 7.234,
+      "eval_steps_per_second": 7.234,
+      "step": 7800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.0540165901184082,
+      "learning_rate": 3.1592493188434494e-05,
+      "loss": 0.7175,
+      "step": 7805
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6427468061447144,
+      "learning_rate": 3.157233662259808e-05,
+      "loss": 0.7531,
+      "step": 7810
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.8710203766822815,
+      "learning_rate": 3.155217546610442e-05,
+      "loss": 0.5609,
+      "step": 7815
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5983574390411377,
+      "learning_rate": 3.153200973303573e-05,
+      "loss": 0.8761,
+      "step": 7820
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.892261803150177,
+      "learning_rate": 3.151183943747738e-05,
+      "loss": 0.8162,
+      "step": 7825
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.8922862410545349,
+      "learning_rate": 3.149166459351797e-05,
+      "loss": 0.6916,
+      "step": 7830
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7080487608909607,
+      "learning_rate": 3.1471485215249264e-05,
+      "loss": 0.7539,
+      "step": 7835
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 20.36235809326172,
+      "learning_rate": 3.145130131676618e-05,
+      "loss": 0.7404,
+      "step": 7840
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.768466055393219,
+      "learning_rate": 3.1431112912166804e-05,
+      "loss": 0.6678,
+      "step": 7845
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.8763453960418701,
+      "learning_rate": 3.141092001555238e-05,
+      "loss": 0.5503,
+      "step": 7850
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.8134922385215759,
+      "learning_rate": 3.139072264102727e-05,
+      "loss": 0.698,
+      "step": 7855
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1443438529968262,
+      "learning_rate": 3.1370520802698953e-05,
+      "loss": 1.0277,
+      "step": 7860
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.806175172328949,
+      "learning_rate": 3.135031451467808e-05,
+      "loss": 0.7382,
+      "step": 7865
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5716249346733093,
+      "learning_rate": 3.1330103791078345e-05,
+      "loss": 0.6695,
+      "step": 7870
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7769394516944885,
+      "learning_rate": 3.130988864601659e-05,
+      "loss": 0.6633,
+      "step": 7875
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5917829871177673,
+      "learning_rate": 3.1289669093612714e-05,
+      "loss": 0.4479,
+      "step": 7880
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8687866926193237,
+      "learning_rate": 3.1269445147989706e-05,
+      "loss": 0.7416,
+      "step": 7885
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8341571092605591,
+      "learning_rate": 3.124921682327363e-05,
+      "loss": 0.6651,
+      "step": 7890
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7416822910308838,
+      "learning_rate": 3.1228984133593594e-05,
+      "loss": 0.6556,
+      "step": 7895
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8006006479263306,
+      "learning_rate": 3.1208747093081765e-05,
+      "loss": 0.8685,
+      "step": 7900
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 0.702598512172699,
+      "eval_runtime": 96.335,
+      "eval_samples_per_second": 7.235,
+      "eval_steps_per_second": 7.235,
+      "step": 7900
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8203070759773254,
+      "learning_rate": 3.118850571587335e-05,
+      "loss": 0.6639,
+      "step": 7905
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7723727226257324,
+      "learning_rate": 3.116826001610658e-05,
+      "loss": 0.7805,
+      "step": 7910
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.870233952999115,
+      "learning_rate": 3.114801000792271e-05,
+      "loss": 0.7065,
+      "step": 7915
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.6729398965835571,
+      "learning_rate": 3.112775570546599e-05,
+      "loss": 0.9504,
+      "step": 7920
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7119274735450745,
+      "learning_rate": 3.110749712288369e-05,
+      "loss": 0.6274,
+      "step": 7925
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.7193456888198853,
+      "learning_rate": 3.1087234274326056e-05,
+      "loss": 0.5857,
+      "step": 7930
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.8700212836265564,
+      "learning_rate": 3.106696717394633e-05,
+      "loss": 0.6635,
+      "step": 7935
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5653272271156311,
+      "learning_rate": 3.104669583590069e-05,
+      "loss": 0.6838,
+      "step": 7940
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.8067308068275452,
+      "learning_rate": 3.102642027434832e-05,
+      "loss": 0.7206,
+      "step": 7945
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.267585277557373,
+      "learning_rate": 3.100614050345131e-05,
+      "loss": 0.7372,
+      "step": 7950
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.7224301099777222,
+      "learning_rate": 3.098585653737473e-05,
+      "loss": 0.6712,
+      "step": 7955
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.1728746891021729,
+      "learning_rate": 3.096556839028654e-05,
+      "loss": 0.6384,
+      "step": 7960
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.8433157801628113,
+      "learning_rate": 3.0945276076357646e-05,
+      "loss": 0.7598,
+      "step": 7965
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.8825304508209229,
+      "learning_rate": 3.092497960976187e-05,
+      "loss": 0.6488,
+      "step": 7970
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.8050782084465027,
+      "learning_rate": 3.0904679004675905e-05,
+      "loss": 0.5749,
+      "step": 7975
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.9096771478652954,
+      "learning_rate": 3.088437427527937e-05,
+      "loss": 0.7139,
+      "step": 7980
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.7473983764648438,
+      "learning_rate": 3.086406543575475e-05,
+      "loss": 0.7281,
+      "step": 7985
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.1370782852172852,
+      "learning_rate": 3.084375250028739e-05,
+      "loss": 0.6314,
+      "step": 7990
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.675596296787262,
+      "learning_rate": 3.08234354830655e-05,
+      "loss": 0.7009,
+      "step": 7995
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.1450175046920776,
+      "learning_rate": 3.080311439828016e-05,
+      "loss": 0.6945,
+      "step": 8000
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 0.7042596936225891,
+      "eval_runtime": 96.3147,
+      "eval_samples_per_second": 7.237,
+      "eval_steps_per_second": 7.237,
+      "step": 8000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.6750172972679138,
+      "learning_rate": 3.0782789260125264e-05,
+      "loss": 0.6747,
+      "step": 8005
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.9685637354850769,
+      "learning_rate": 3.0762460082797565e-05,
+      "loss": 0.8169,
+      "step": 8010
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.6871526837348938,
+      "learning_rate": 3.07421268804966e-05,
+      "loss": 0.7907,
+      "step": 8015
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2020535469055176,
+      "learning_rate": 3.072178966742476e-05,
+      "loss": 0.5841,
+      "step": 8020
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.7885614633560181,
+      "learning_rate": 3.070144845778721e-05,
+      "loss": 0.7879,
+      "step": 8025
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.8481292724609375,
+      "learning_rate": 3.0681103265791913e-05,
+      "loss": 0.7492,
+      "step": 8030
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.7926096320152283,
+      "learning_rate": 3.066075410564962e-05,
+      "loss": 0.7373,
+      "step": 8035
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.6218678951263428,
+      "learning_rate": 3.064040099157384e-05,
+      "loss": 0.5947,
+      "step": 8040
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.7932231426239014,
+      "learning_rate": 3.062004393778086e-05,
+      "loss": 0.6297,
+      "step": 8045
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.7588164806365967,
+      "learning_rate": 3.05996829584897e-05,
+      "loss": 0.5988,
+      "step": 8050
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.5323835611343384,
+      "learning_rate": 3.057931806792214e-05,
+      "loss": 0.8176,
+      "step": 8055
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.672051727771759,
+      "learning_rate": 3.055894928030269e-05,
+      "loss": 0.4785,
+      "step": 8060
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.862082839012146,
+      "learning_rate": 3.0538576609858566e-05,
+      "loss": 0.6622,
+      "step": 8065
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.8672384023666382,
+      "learning_rate": 3.0518200070819715e-05,
+      "loss": 0.5951,
+      "step": 8070
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.7039456367492676,
+      "learning_rate": 3.0497819677418767e-05,
+      "loss": 0.5597,
+      "step": 8075
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0149037837982178,
+      "learning_rate": 3.0477435443891068e-05,
+      "loss": 0.8546,
+      "step": 8080
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.7860717177391052,
+      "learning_rate": 3.0457047384474623e-05,
+      "loss": 0.5704,
+      "step": 8085
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0836188793182373,
+      "learning_rate": 3.0436655513410135e-05,
+      "loss": 0.7824,
+      "step": 8090
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5916305780410767,
+      "learning_rate": 3.0416259844940932e-05,
+      "loss": 0.5636,
+      "step": 8095
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.9937427639961243,
+      "learning_rate": 3.0395860393313026e-05,
+      "loss": 0.6955,
+      "step": 8100
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 0.7009605765342712,
+      "eval_runtime": 96.2791,
+      "eval_samples_per_second": 7.239,
+      "eval_steps_per_second": 7.239,
+      "step": 8100
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.223351001739502,
+      "learning_rate": 3.0375457172775067e-05,
+      "loss": 0.7957,
+      "step": 8105
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.7758160829544067,
+      "learning_rate": 3.0355050197578333e-05,
+      "loss": 0.7482,
+      "step": 8110
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.0522717237472534,
+      "learning_rate": 3.033463948197671e-05,
+      "loss": 0.5511,
+      "step": 8115
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8289081454277039,
+      "learning_rate": 3.0314225040226723e-05,
+      "loss": 0.8692,
+      "step": 8120
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7654734253883362,
+      "learning_rate": 3.029380688658749e-05,
+      "loss": 0.7779,
+      "step": 8125
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8430918455123901,
+      "learning_rate": 3.0273385035320708e-05,
+      "loss": 0.6381,
+      "step": 8130
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5839015245437622,
+      "learning_rate": 3.025295950069067e-05,
+      "loss": 0.7259,
+      "step": 8135
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.9747032523155212,
+      "learning_rate": 3.0232530296964247e-05,
+      "loss": 0.7543,
+      "step": 8140
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7823505401611328,
+      "learning_rate": 3.0212097438410864e-05,
+      "loss": 0.603,
+      "step": 8145
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.9050419926643372,
+      "learning_rate": 3.0191660939302484e-05,
+      "loss": 0.8558,
+      "step": 8150
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8175522685050964,
+      "learning_rate": 3.017122081391365e-05,
+      "loss": 0.9133,
+      "step": 8155
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7007612586021423,
+      "learning_rate": 3.0150777076521403e-05,
+      "loss": 0.5508,
+      "step": 8160
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.9761927127838135,
+      "learning_rate": 3.0130329741405334e-05,
+      "loss": 0.597,
+      "step": 8165
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7669122815132141,
+      "learning_rate": 3.0109878822847515e-05,
+      "loss": 0.7017,
+      "step": 8170
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.3604693412780762,
+      "learning_rate": 3.008942433513255e-05,
+      "loss": 0.6975,
+      "step": 8175
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6270090937614441,
+      "learning_rate": 3.0068966292547528e-05,
+      "loss": 0.6954,
+      "step": 8180
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.9544779062271118,
+      "learning_rate": 3.004850470938201e-05,
+      "loss": 0.5711,
+      "step": 8185
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7997902035713196,
+      "learning_rate": 3.0028039599928047e-05,
+      "loss": 0.6225,
+      "step": 8190
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.1297340393066406,
+      "learning_rate": 3.000757097848015e-05,
+      "loss": 0.9648,
+      "step": 8195
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.915552020072937,
+      "learning_rate": 2.998709885933526e-05,
+      "loss": 0.734,
+      "step": 8200
+    },
+    {
+      "epoch": 1.31,
+      "eval_loss": 0.7022165060043335,
+      "eval_runtime": 96.3376,
+      "eval_samples_per_second": 7.235,
+      "eval_steps_per_second": 7.235,
+      "step": 8200
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.8937833309173584,
+      "learning_rate": 2.996662325679279e-05,
+      "loss": 0.711,
+      "step": 8205
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.8294236063957214,
+      "learning_rate": 2.9946144185154573e-05,
+      "loss": 0.6208,
+      "step": 8210
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7142773866653442,
+      "learning_rate": 2.992566165872487e-05,
+      "loss": 0.63,
+      "step": 8215
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.2502591609954834,
+      "learning_rate": 2.9905175691810346e-05,
+      "loss": 0.6856,
+      "step": 8220
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7072376012802124,
+      "learning_rate": 2.9884686298720084e-05,
+      "loss": 0.7738,
+      "step": 8225
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7068510055541992,
+      "learning_rate": 2.9864193493765536e-05,
+      "loss": 0.5731,
+      "step": 8230
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0625722408294678,
+      "learning_rate": 2.984369729126057e-05,
+      "loss": 0.6232,
+      "step": 8235
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.9145768284797668,
+      "learning_rate": 2.98231977055214e-05,
+      "loss": 0.831,
+      "step": 8240
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.9636755585670471,
+      "learning_rate": 2.9802694750866615e-05,
+      "loss": 0.8864,
+      "step": 8245
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.1421067714691162,
+      "learning_rate": 2.978218844161715e-05,
+      "loss": 0.7562,
+      "step": 8250
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.645041286945343,
+      "learning_rate": 2.976167879209629e-05,
+      "loss": 0.5159,
+      "step": 8255
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5683246850967407,
+      "learning_rate": 2.9741165816629662e-05,
+      "loss": 0.7033,
+      "step": 8260
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.5764894485473633,
+      "learning_rate": 2.9720649529545197e-05,
+      "loss": 0.6767,
+      "step": 8265
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.7696818709373474,
+      "learning_rate": 2.970012994517314e-05,
+      "loss": 0.6647,
+      "step": 8270
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.7804857492446899,
+      "learning_rate": 2.967960707784605e-05,
+      "loss": 0.6663,
+      "step": 8275
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.8747091889381409,
+      "learning_rate": 2.9659080941898788e-05,
+      "loss": 0.7183,
+      "step": 8280
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6558755040168762,
+      "learning_rate": 2.963855155166847e-05,
+      "loss": 0.5533,
+      "step": 8285
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.0046048164367676,
+      "learning_rate": 2.9618018921494505e-05,
+      "loss": 0.6945,
+      "step": 8290
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.8384654521942139,
+      "learning_rate": 2.959748306571857e-05,
+      "loss": 0.6501,
+      "step": 8295
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5712169408798218,
+      "learning_rate": 2.9576943998684577e-05,
+      "loss": 0.5586,
+      "step": 8300
+    },
+    {
+      "epoch": 1.32,
+      "eval_loss": 0.701697826385498,
+      "eval_runtime": 96.3495,
+      "eval_samples_per_second": 7.234,
+      "eval_steps_per_second": 7.234,
+      "step": 8300
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.8932040929794312,
+      "learning_rate": 2.9556401734738682e-05,
+      "loss": 0.776,
+      "step": 8305
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.7956164479255676,
+      "learning_rate": 2.9535856288229295e-05,
+      "loss": 0.7136,
+      "step": 8310
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.699739396572113,
+      "learning_rate": 2.951530767350703e-05,
+      "loss": 0.4986,
+      "step": 8315
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.7867723107337952,
+      "learning_rate": 2.949475590492473e-05,
+      "loss": 0.6532,
+      "step": 8320
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6946756839752197,
+      "learning_rate": 2.947420099683741e-05,
+      "loss": 0.7284,
+      "step": 8325
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.097390055656433,
+      "learning_rate": 2.9453642963602306e-05,
+      "loss": 1.0162,
+      "step": 8330
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6341527104377747,
+      "learning_rate": 2.9433081819578843e-05,
+      "loss": 0.5738,
+      "step": 8335
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.7065830826759338,
+      "learning_rate": 2.941251757912859e-05,
+      "loss": 0.5473,
+      "step": 8340
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6362321376800537,
+      "learning_rate": 2.939195025661529e-05,
+      "loss": 0.6538,
+      "step": 8345
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.0062533617019653,
+      "learning_rate": 2.9371379866404862e-05,
+      "loss": 0.6313,
+      "step": 8350
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.0622398853302002,
+      "learning_rate": 2.935080642286534e-05,
+      "loss": 0.7483,
+      "step": 8355
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.726993978023529,
+      "learning_rate": 2.933022994036689e-05,
+      "loss": 0.6649,
+      "step": 8360
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5122197866439819,
+      "learning_rate": 2.930965043328181e-05,
+      "loss": 0.6865,
+      "step": 8365
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.8607394099235535,
+      "learning_rate": 2.9289067915984526e-05,
+      "loss": 0.5822,
+      "step": 8370
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7147690653800964,
+      "learning_rate": 2.926848240285154e-05,
+      "loss": 0.559,
+      "step": 8375
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7905888557434082,
+      "learning_rate": 2.9247893908261458e-05,
+      "loss": 0.8028,
+      "step": 8380
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.8332535624504089,
+      "learning_rate": 2.9227302446594963e-05,
+      "loss": 0.7625,
+      "step": 8385
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.8591724038124084,
+      "learning_rate": 2.9206708032234826e-05,
+      "loss": 0.7804,
+      "step": 8390
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6741548776626587,
+      "learning_rate": 2.918611067956586e-05,
+      "loss": 0.8665,
+      "step": 8395
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5031669735908508,
+      "learning_rate": 2.916551040297495e-05,
+      "loss": 0.7299,
+      "step": 8400
+    },
+    {
+      "epoch": 1.34,
+      "eval_loss": 0.6999276876449585,
+      "eval_runtime": 96.3216,
+      "eval_samples_per_second": 7.236,
+      "eval_steps_per_second": 7.236,
+      "step": 8400
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6288011074066162,
+      "learning_rate": 2.9144907216851004e-05,
+      "loss": 0.6782,
+      "step": 8405
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.8632309436798096,
+      "learning_rate": 2.9124301135584975e-05,
+      "loss": 0.6778,
+      "step": 8410
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6853322982788086,
+      "learning_rate": 2.910369217356984e-05,
+      "loss": 0.7244,
+      "step": 8415
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6843327283859253,
+      "learning_rate": 2.908308034520058e-05,
+      "loss": 0.7216,
+      "step": 8420
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.206913948059082,
+      "learning_rate": 2.906246566487417e-05,
+      "loss": 0.6733,
+      "step": 8425
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.9815182685852051,
+      "learning_rate": 2.904184814698961e-05,
+      "loss": 0.6827,
+      "step": 8430
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.9608283042907715,
+      "learning_rate": 2.9021227805947844e-05,
+      "loss": 0.8357,
+      "step": 8435
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.2694772481918335,
+      "learning_rate": 2.90006046561518e-05,
+      "loss": 0.6741,
+      "step": 8440
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.78603595495224,
+      "learning_rate": 2.897997871200639e-05,
+      "loss": 0.7701,
+      "step": 8445
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.9763398170471191,
+      "learning_rate": 2.895934998791845e-05,
+      "loss": 0.9359,
+      "step": 8450
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8996787667274475,
+      "learning_rate": 2.8938718498296768e-05,
+      "loss": 0.7951,
+      "step": 8455
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5990316867828369,
+      "learning_rate": 2.891808425755206e-05,
+      "loss": 0.7176,
+      "step": 8460
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.936537265777588,
+      "learning_rate": 2.889744728009696e-05,
+      "loss": 0.8185,
+      "step": 8465
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5997591614723206,
+      "learning_rate": 2.8876807580346044e-05,
+      "loss": 0.6776,
+      "step": 8470
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.9869585633277893,
+      "learning_rate": 2.8856165172715745e-05,
+      "loss": 0.9732,
+      "step": 8475
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8746089935302734,
+      "learning_rate": 2.883552007162441e-05,
+      "loss": 0.7027,
+      "step": 8480
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.7140242457389832,
+      "learning_rate": 2.8814872291492278e-05,
+      "loss": 0.8119,
+      "step": 8485
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6302705407142639,
+      "learning_rate": 2.8794221846741436e-05,
+      "loss": 0.7862,
+      "step": 8490
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.8000974655151367,
+      "learning_rate": 2.8773568751795843e-05,
+      "loss": 0.7948,
+      "step": 8495
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.35248124599456787,
+      "learning_rate": 2.875291302108131e-05,
+      "loss": 1.089,
+      "step": 8500
+    },
+    {
+      "epoch": 1.36,
+      "eval_loss": 0.6993909478187561,
+      "eval_runtime": 96.3537,
+      "eval_samples_per_second": 7.234,
+      "eval_steps_per_second": 7.234,
+      "step": 8500
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.1372653245925903,
+      "learning_rate": 2.8732254669025497e-05,
+      "loss": 0.7275,
+      "step": 8505
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6813796162605286,
+      "learning_rate": 2.8711593710057883e-05,
+      "loss": 0.6084,
+      "step": 8510
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.1797153949737549,
+      "learning_rate": 2.8690930158609757e-05,
+      "loss": 0.8862,
+      "step": 8515
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6880937814712524,
+      "learning_rate": 2.8670264029114247e-05,
+      "loss": 0.7201,
+      "step": 8520
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.8386674523353577,
+      "learning_rate": 2.864959533600626e-05,
+      "loss": 0.7845,
+      "step": 8525
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.9347553849220276,
+      "learning_rate": 2.862892409372251e-05,
+      "loss": 0.7399,
+      "step": 8530
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.7480734586715698,
+      "learning_rate": 2.8608250316701474e-05,
+      "loss": 0.6557,
+      "step": 8535
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.7449476718902588,
+      "learning_rate": 2.8587574019383413e-05,
+      "loss": 0.6119,
+      "step": 8540
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.7354079484939575,
+      "learning_rate": 2.856689521621034e-05,
+      "loss": 0.7845,
+      "step": 8545
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.656905472278595,
+      "learning_rate": 2.854621392162602e-05,
+      "loss": 0.6436,
+      "step": 8550
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.025376558303833,
+      "learning_rate": 2.8525530150075968e-05,
+      "loss": 0.8306,
+      "step": 8555
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6368470191955566,
+      "learning_rate": 2.8504843916007412e-05,
+      "loss": 0.7914,
+      "step": 8560
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.7506736516952515,
+      "learning_rate": 2.8484155233869315e-05,
+      "loss": 0.854,
+      "step": 8565
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6676509380340576,
+      "learning_rate": 2.846346411811235e-05,
+      "loss": 0.7416,
+      "step": 8570
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.8177019953727722,
+      "learning_rate": 2.8442770583188872e-05,
+      "loss": 0.5851,
+      "step": 8575
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.9827241897583008,
+      "learning_rate": 2.842207464355294e-05,
+      "loss": 0.9374,
+      "step": 8580
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.079214334487915,
+      "learning_rate": 2.8401376313660295e-05,
+      "loss": 0.7287,
+      "step": 8585
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5321961641311646,
+      "learning_rate": 2.8380675607968343e-05,
+      "loss": 0.6194,
+      "step": 8590
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.8180492520332336,
+      "learning_rate": 2.835997254093614e-05,
+      "loss": 0.4846,
+      "step": 8595
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6867579221725464,
+      "learning_rate": 2.8339267127024416e-05,
+      "loss": 0.5733,
+      "step": 8600
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 0.6994162201881409,
+      "eval_runtime": 96.4595,
+      "eval_samples_per_second": 7.226,
+      "eval_steps_per_second": 7.226,
+      "step": 8600
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.8024903535842896,
+      "learning_rate": 2.8318559380695514e-05,
+      "loss": 0.679,
+      "step": 8605
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.9138768911361694,
+      "learning_rate": 2.8297849316413422e-05,
+      "loss": 0.6586,
+      "step": 8610
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.6323714256286621,
+      "learning_rate": 2.8277136948643734e-05,
+      "loss": 0.5754,
+      "step": 8615
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.9891794323921204,
+      "learning_rate": 2.825642229185367e-05,
+      "loss": 0.7783,
+      "step": 8620
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.8239473700523376,
+      "learning_rate": 2.8235705360512034e-05,
+      "loss": 0.7095,
+      "step": 8625
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.0162553787231445,
+      "learning_rate": 2.8214986169089232e-05,
+      "loss": 0.7844,
+      "step": 8630
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.9242910146713257,
+      "learning_rate": 2.819426473205723e-05,
+      "loss": 0.7479,
+      "step": 8635
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.6635788679122925,
+      "learning_rate": 2.8173541063889587e-05,
+      "loss": 0.7058,
+      "step": 8640
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.7192090153694153,
+      "learning_rate": 2.8152815179061396e-05,
+      "loss": 0.8564,
+      "step": 8645
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.3398587703704834,
+      "learning_rate": 2.8132087092049316e-05,
+      "loss": 0.8072,
+      "step": 8650
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.8398715257644653,
+      "learning_rate": 2.811135681733153e-05,
+      "loss": 0.7366,
+      "step": 8655
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.9099376201629639,
+      "learning_rate": 2.8090624369387776e-05,
+      "loss": 0.661,
+      "step": 8660
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.099500298500061,
+      "learning_rate": 2.806988976269927e-05,
+      "loss": 0.8471,
+      "step": 8665
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.8880857229232788,
+      "learning_rate": 2.8049153011748757e-05,
+      "loss": 0.8047,
+      "step": 8670
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.6709883809089661,
+      "learning_rate": 2.8028414131020492e-05,
+      "loss": 0.6804,
+      "step": 8675
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7524588108062744,
+      "learning_rate": 2.80076731350002e-05,
+      "loss": 0.6399,
+      "step": 8680
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.844389021396637,
+      "learning_rate": 2.7986930038175084e-05,
+      "loss": 0.5184,
+      "step": 8685
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.387044072151184,
+      "learning_rate": 2.7966184855033827e-05,
+      "loss": 0.7384,
+      "step": 8690
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.4819062054157257,
+      "learning_rate": 2.7945437600066556e-05,
+      "loss": 0.8486,
+      "step": 8695
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5254135727882385,
+      "learning_rate": 2.792468828776484e-05,
+      "loss": 0.5409,
+      "step": 8700
+    },
+    {
+      "epoch": 1.39,
+      "eval_loss": 0.6987185478210449,
+      "eval_runtime": 96.4354,
+      "eval_samples_per_second": 7.228,
+      "eval_steps_per_second": 7.228,
+      "step": 8700
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.8113055229187012,
+      "learning_rate": 2.7903936932621717e-05,
+      "loss": 0.4735,
+      "step": 8705
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7469605207443237,
+      "learning_rate": 2.7883183549131613e-05,
+      "loss": 0.6068,
+      "step": 8710
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.925462007522583,
+      "learning_rate": 2.7862428151790394e-05,
+      "loss": 0.7301,
+      "step": 8715
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.9389353394508362,
+      "learning_rate": 2.7841670755095323e-05,
+      "loss": 0.7569,
+      "step": 8720
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7033401131629944,
+      "learning_rate": 2.7820911373545074e-05,
+      "loss": 0.8086,
+      "step": 8725
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.9423335194587708,
+      "learning_rate": 2.7800150021639674e-05,
+      "loss": 0.666,
+      "step": 8730
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5969627499580383,
+      "learning_rate": 2.7779386713880566e-05,
+      "loss": 0.8106,
+      "step": 8735
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.9117034673690796,
+      "learning_rate": 2.7758621464770552e-05,
+      "loss": 0.6877,
+      "step": 8740
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.0278637409210205,
+      "learning_rate": 2.7737854288813757e-05,
+      "loss": 0.6363,
+      "step": 8745
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8942778706550598,
+      "learning_rate": 2.771708520051568e-05,
+      "loss": 0.8134,
+      "step": 8750
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6770984530448914,
+      "learning_rate": 2.769631421438315e-05,
+      "loss": 0.6516,
+      "step": 8755
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7041003108024597,
+      "learning_rate": 2.7675541344924337e-05,
+      "loss": 0.5647,
+      "step": 8760
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8790050745010376,
+      "learning_rate": 2.7654766606648702e-05,
+      "loss": 0.8915,
+      "step": 8765
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7827969193458557,
+      "learning_rate": 2.7633990014067008e-05,
+      "loss": 0.6071,
+      "step": 8770
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.996577262878418,
+      "learning_rate": 2.761321158169134e-05,
+      "loss": 0.604,
+      "step": 8775
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.9607555270195007,
+      "learning_rate": 2.759243132403505e-05,
+      "loss": 0.8923,
+      "step": 8780
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7468242049217224,
+      "learning_rate": 2.7571649255612774e-05,
+      "loss": 0.7277,
+      "step": 8785
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6365399956703186,
+      "learning_rate": 2.7550865390940395e-05,
+      "loss": 0.7278,
+      "step": 8790
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8831495046615601,
+      "learning_rate": 2.753007974453508e-05,
+      "loss": 0.5988,
+      "step": 8795
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8448741436004639,
+      "learning_rate": 2.7509292330915216e-05,
+      "loss": 0.8848,
+      "step": 8800
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 0.6976245045661926,
+      "eval_runtime": 96.6807,
+      "eval_samples_per_second": 7.209,
+      "eval_steps_per_second": 7.209,
+      "step": 8800
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.0347315073013306,
+      "learning_rate": 2.7488503164600427e-05,
+      "loss": 0.7309,
+      "step": 8805
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.0705949068069458,
+      "learning_rate": 2.7467712260111567e-05,
+      "loss": 0.635,
+      "step": 8810
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.0177220106124878,
+      "learning_rate": 2.7446919631970718e-05,
+      "loss": 0.689,
+      "step": 8815
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.7912683486938477,
+      "learning_rate": 2.742612529470114e-05,
+      "loss": 0.893,
+      "step": 8820
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.7678771615028381,
+      "learning_rate": 2.7405329262827296e-05,
+      "loss": 0.6563,
+      "step": 8825
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6450463533401489,
+      "learning_rate": 2.738453155087483e-05,
+      "loss": 0.6637,
+      "step": 8830
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.9644352197647095,
+      "learning_rate": 2.7363732173370586e-05,
+      "loss": 0.7377,
+      "step": 8835
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.7947851419448853,
+      "learning_rate": 2.7342931144842533e-05,
+      "loss": 0.6275,
+      "step": 8840
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.9577487111091614,
+      "learning_rate": 2.7322128479819815e-05,
+      "loss": 0.7928,
+      "step": 8845
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.8619856834411621,
+      "learning_rate": 2.7301324192832707e-05,
+      "loss": 0.6892,
+      "step": 8850
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6695562601089478,
+      "learning_rate": 2.7280518298412628e-05,
+      "loss": 0.5982,
+      "step": 8855
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6930350065231323,
+      "learning_rate": 2.7259710811092126e-05,
+      "loss": 0.8486,
+      "step": 8860
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7074466347694397,
+      "learning_rate": 2.7238901745404844e-05,
+      "loss": 0.6394,
+      "step": 8865
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7025793194770813,
+      "learning_rate": 2.7218091115885523e-05,
+      "loss": 0.7898,
+      "step": 8870
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8207305073738098,
+      "learning_rate": 2.719727893707002e-05,
+      "loss": 0.5337,
+      "step": 8875
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.752471923828125,
+      "learning_rate": 2.7176465223495268e-05,
+      "loss": 0.8052,
+      "step": 8880
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8493356704711914,
+      "learning_rate": 2.7155649989699254e-05,
+      "loss": 0.7002,
+      "step": 8885
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.0340080261230469,
+      "learning_rate": 2.713483325022104e-05,
+      "loss": 0.9414,
+      "step": 8890
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7286359071731567,
+      "learning_rate": 2.711401501960075e-05,
+      "loss": 0.6413,
+      "step": 8895
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7379997372627258,
+      "learning_rate": 2.7093195312379537e-05,
+      "loss": 0.5739,
+      "step": 8900
+    },
+    {
+      "epoch": 1.42,
+      "eval_loss": 0.6970831155776978,
+      "eval_runtime": 96.7643,
+      "eval_samples_per_second": 7.203,
+      "eval_steps_per_second": 7.203,
+      "step": 8900
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.2503465414047241,
+      "learning_rate": 2.7072374143099573e-05,
+      "loss": 0.8559,
+      "step": 8905
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.6661009788513184,
+      "learning_rate": 2.705155152630407e-05,
+      "loss": 0.788,
+      "step": 8910
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.6985836625099182,
+      "learning_rate": 2.7030727476537256e-05,
+      "loss": 0.7824,
+      "step": 8915
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7565920948982239,
+      "learning_rate": 2.7009902008344344e-05,
+      "loss": 0.6379,
+      "step": 8920
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.0957528352737427,
+      "learning_rate": 2.6989075136271535e-05,
+      "loss": 0.7973,
+      "step": 8925
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.3178807497024536,
+      "learning_rate": 2.6968246874866026e-05,
+      "loss": 0.8141,
+      "step": 8930
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.6984952092170715,
+      "learning_rate": 2.6947417238675987e-05,
+      "loss": 0.7697,
+      "step": 8935
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.9059376120567322,
+      "learning_rate": 2.6926586242250527e-05,
+      "loss": 0.6771,
+      "step": 8940
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0285694599151611,
+      "learning_rate": 2.690575390013971e-05,
+      "loss": 0.7465,
+      "step": 8945
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7756787538528442,
+      "learning_rate": 2.6884920226894572e-05,
+      "loss": 0.7743,
+      "step": 8950
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7653606534004211,
+      "learning_rate": 2.6864085237067037e-05,
+      "loss": 0.8444,
+      "step": 8955
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0834332704544067,
+      "learning_rate": 2.684324894520996e-05,
+      "loss": 0.6053,
+      "step": 8960
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7883355021476746,
+      "learning_rate": 2.682241136587712e-05,
+      "loss": 0.6907,
+      "step": 8965
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7430680394172668,
+      "learning_rate": 2.6801572513623193e-05,
+      "loss": 0.833,
+      "step": 8970
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.8970258235931396,
+      "learning_rate": 2.678073240300372e-05,
+      "loss": 0.777,
+      "step": 8975
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7223610281944275,
+      "learning_rate": 2.6759891048575147e-05,
+      "loss": 0.579,
+      "step": 8980
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.9050683379173279,
+      "learning_rate": 2.673904846489478e-05,
+      "loss": 0.7313,
+      "step": 8985
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.8505553603172302,
+      "learning_rate": 2.6718204666520786e-05,
+      "loss": 0.6838,
+      "step": 8990
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7363889813423157,
+      "learning_rate": 2.669735966801217e-05,
+      "loss": 0.7082,
+      "step": 8995
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7967817783355713,
+      "learning_rate": 2.6676513483928788e-05,
+      "loss": 0.728,
+      "step": 9000
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 0.6963269114494324,
+      "eval_runtime": 96.6643,
+      "eval_samples_per_second": 7.211,
+      "eval_steps_per_second": 7.211,
+      "step": 9000
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.0329532623291016,
+      "learning_rate": 2.665566612883132e-05,
+      "loss": 0.5401,
+      "step": 9005
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6207738518714905,
+      "learning_rate": 2.6634817617281262e-05,
+      "loss": 0.735,
+      "step": 9010
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.8492811918258667,
+      "learning_rate": 2.6613967963840925e-05,
+      "loss": 0.6334,
+      "step": 9015
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7099390029907227,
+      "learning_rate": 2.6593117183073406e-05,
+      "loss": 0.6005,
+      "step": 9020
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7716400623321533,
+      "learning_rate": 2.657226528954259e-05,
+      "loss": 0.5958,
+      "step": 9025
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6929382681846619,
+      "learning_rate": 2.655141229781315e-05,
+      "loss": 0.7592,
+      "step": 9030
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6694722771644592,
+      "learning_rate": 2.6530558222450525e-05,
+      "loss": 0.6927,
+      "step": 9035
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.8594262599945068,
+      "learning_rate": 2.6509703078020892e-05,
+      "loss": 0.9045,
+      "step": 9040
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7283955812454224,
+      "learning_rate": 2.64888468790912e-05,
+      "loss": 0.7548,
+      "step": 9045
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.9150108098983765,
+      "learning_rate": 2.646798964022913e-05,
+      "loss": 0.6196,
+      "step": 9050
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.8249070644378662,
+      "learning_rate": 2.6447131376003076e-05,
+      "loss": 0.7259,
+      "step": 9055
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.8211736679077148,
+      "learning_rate": 2.642627210098214e-05,
+      "loss": 0.7903,
+      "step": 9060
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.9737334251403809,
+      "learning_rate": 2.6405411829736165e-05,
+      "loss": 0.8098,
+      "step": 9065
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.6321923136711121,
+      "learning_rate": 2.638455057683567e-05,
+      "loss": 0.6646,
+      "step": 9070
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.0885649919509888,
+      "learning_rate": 2.6363688356851847e-05,
+      "loss": 0.671,
+      "step": 9075
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.0221432447433472,
+      "learning_rate": 2.634282518435658e-05,
+      "loss": 0.8984,
+      "step": 9080
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.6629742980003357,
+      "learning_rate": 2.632196107392242e-05,
+      "loss": 0.7947,
+      "step": 9085
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.847280740737915,
+      "learning_rate": 2.6301096040122565e-05,
+      "loss": 0.7693,
+      "step": 9090
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.0186293125152588,
+      "learning_rate": 2.628023009753086e-05,
+      "loss": 0.8952,
+      "step": 9095
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.9672142863273621,
+      "learning_rate": 2.6259363260721776e-05,
+      "loss": 0.7503,
+      "step": 9100
+    },
+    {
+      "epoch": 1.45,
+      "eval_loss": 0.6952997446060181,
+      "eval_runtime": 96.6663,
+      "eval_samples_per_second": 7.21,
+      "eval_steps_per_second": 7.21,
+      "step": 9100
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.8531110286712646,
+      "learning_rate": 2.6238495544270435e-05,
+      "loss": 0.5878,
+      "step": 9105
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.7663474082946777,
+      "learning_rate": 2.6217626962752555e-05,
+      "loss": 0.7022,
+      "step": 9110
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.7978513836860657,
+      "learning_rate": 2.6196757530744443e-05,
+      "loss": 0.6604,
+      "step": 9115
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.7127459645271301,
+      "learning_rate": 2.6175887262823023e-05,
+      "loss": 0.7602,
+      "step": 9120
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.6364534497261047,
+      "learning_rate": 2.615501617356581e-05,
+      "loss": 0.8812,
+      "step": 9125
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9094502329826355,
+      "learning_rate": 2.613414427755086e-05,
+      "loss": 0.655,
+      "step": 9130
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9474580883979797,
+      "learning_rate": 2.611327158935683e-05,
+      "loss": 0.9989,
+      "step": 9135
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.2897037267684937,
+      "learning_rate": 2.6092398123562895e-05,
+      "loss": 0.9347,
+      "step": 9140
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.690949022769928,
+      "learning_rate": 2.60715238947488e-05,
+      "loss": 0.7953,
+      "step": 9145
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8102849721908569,
+      "learning_rate": 2.6050648917494808e-05,
+      "loss": 0.716,
+      "step": 9150
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8799526691436768,
+      "learning_rate": 2.6029773206381718e-05,
+      "loss": 0.7998,
+      "step": 9155
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.7595265507698059,
+      "learning_rate": 2.600889677599082e-05,
+      "loss": 0.7642,
+      "step": 9160
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.819495439529419,
+      "learning_rate": 2.5988019640903927e-05,
+      "loss": 0.7664,
+      "step": 9165
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9698195457458496,
+      "learning_rate": 2.596714181570335e-05,
+      "loss": 0.6731,
+      "step": 9170
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.1988072395324707,
+      "learning_rate": 2.5946263314971842e-05,
+      "loss": 0.8619,
+      "step": 9175
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.3569873571395874,
+      "learning_rate": 2.5925384153292682e-05,
+      "loss": 0.6456,
+      "step": 9180
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.0610495805740356,
+      "learning_rate": 2.5904504345249565e-05,
+      "loss": 0.8024,
+      "step": 9185
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7540451288223267,
+      "learning_rate": 2.5883623905426664e-05,
+      "loss": 1.0539,
+      "step": 9190
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7257586121559143,
+      "learning_rate": 2.5862742848408583e-05,
+      "loss": 0.7662,
+      "step": 9195
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8342102766036987,
+      "learning_rate": 2.5841861188780355e-05,
+      "loss": 0.706,
+      "step": 9200
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 0.6950993537902832,
+      "eval_runtime": 96.6378,
+      "eval_samples_per_second": 7.213,
+      "eval_steps_per_second": 7.213,
+      "step": 9200
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7215315699577332,
+      "learning_rate": 2.582097894112745e-05,
+      "loss": 0.5098,
+      "step": 9205
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.9047237038612366,
+      "learning_rate": 2.5800096120035732e-05,
+      "loss": 0.7546,
+      "step": 9210
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8365469574928284,
+      "learning_rate": 2.577921274009146e-05,
+      "loss": 0.9019,
+      "step": 9215
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.236525535583496,
+      "learning_rate": 2.5758328815881312e-05,
+      "loss": 0.5778,
+      "step": 9220
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.10982346534729,
+      "learning_rate": 2.5737444361992312e-05,
+      "loss": 0.642,
+      "step": 9225
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8745054602622986,
+      "learning_rate": 2.5716559393011885e-05,
+      "loss": 0.8715,
+      "step": 9230
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7256401181221008,
+      "learning_rate": 2.569567392352778e-05,
+      "loss": 0.6683,
+      "step": 9235
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7779073715209961,
+      "learning_rate": 2.5674787968128143e-05,
+      "loss": 0.8675,
+      "step": 9240
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.9540502429008484,
+      "learning_rate": 2.5653901541401415e-05,
+      "loss": 0.797,
+      "step": 9245
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5401924848556519,
+      "learning_rate": 2.5633014657936394e-05,
+      "loss": 0.5741,
+      "step": 9250
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8894802331924438,
+      "learning_rate": 2.5612127332322168e-05,
+      "loss": 0.597,
+      "step": 9255
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8277390599250793,
+      "learning_rate": 2.5591239579148173e-05,
+      "loss": 0.6365,
+      "step": 9260
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8742364645004272,
+      "learning_rate": 2.5570351413004125e-05,
+      "loss": 0.7009,
+      "step": 9265
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.7853513360023499,
+      "learning_rate": 2.5549462848480004e-05,
+      "loss": 0.6336,
+      "step": 9270
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.785223662853241,
+      "learning_rate": 2.5528573900166115e-05,
+      "loss": 0.6963,
+      "step": 9275
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6762696504592896,
+      "learning_rate": 2.5507684582653e-05,
+      "loss": 0.8248,
+      "step": 9280
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.2204557657241821,
+      "learning_rate": 2.5486794910531458e-05,
+      "loss": 0.6296,
+      "step": 9285
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8152850270271301,
+      "learning_rate": 2.5465904898392564e-05,
+      "loss": 0.7962,
+      "step": 9290
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.834222137928009,
+      "learning_rate": 2.5445014560827578e-05,
+      "loss": 0.7165,
+      "step": 9295
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.7482254505157471,
+      "learning_rate": 2.5424123912428054e-05,
+      "loss": 0.7926,
+      "step": 9300
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.6945102214813232,
+      "eval_runtime": 96.5254,
+      "eval_samples_per_second": 7.221,
+      "eval_steps_per_second": 7.221,
+      "step": 9300
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.741940975189209,
+      "learning_rate": 2.5403232967785705e-05,
+      "loss": 0.5465,
+      "step": 9305
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.8186244368553162,
+      "learning_rate": 2.5382341741492494e-05,
+      "loss": 0.6283,
+      "step": 9310
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.7384134531021118,
+      "learning_rate": 2.536145024814054e-05,
+      "loss": 0.6346,
+      "step": 9315
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6944209337234497,
+      "learning_rate": 2.5340558502322187e-05,
+      "loss": 0.561,
+      "step": 9320
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6684926152229309,
+      "learning_rate": 2.531966651862993e-05,
+      "loss": 0.6977,
+      "step": 9325
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.822957456111908,
+      "learning_rate": 2.5298774311656443e-05,
+      "loss": 0.618,
+      "step": 9330
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.0287977457046509,
+      "learning_rate": 2.5277881895994547e-05,
+      "loss": 0.6704,
+      "step": 9335
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.7576120495796204,
+      "learning_rate": 2.5256989286237216e-05,
+      "loss": 0.5916,
+      "step": 9340
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6524910926818848,
+      "learning_rate": 2.5236096496977564e-05,
+      "loss": 0.6364,
+      "step": 9345
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.776617705821991,
+      "learning_rate": 2.5215203542808803e-05,
+      "loss": 0.5652,
+      "step": 9350
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.645124614238739,
+      "learning_rate": 2.5194310438324298e-05,
+      "loss": 0.7498,
+      "step": 9355
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.037840723991394,
+      "learning_rate": 2.5173417198117498e-05,
+      "loss": 0.6456,
+      "step": 9360
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.8559253215789795,
+      "learning_rate": 2.515252383678195e-05,
+      "loss": 0.8085,
+      "step": 9365
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.9479286670684814,
+      "learning_rate": 2.513163036891128e-05,
+      "loss": 0.8607,
+      "step": 9370
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6962096691131592,
+      "learning_rate": 2.51107368090992e-05,
+      "loss": 0.7528,
+      "step": 9375
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.7785203456878662,
+      "learning_rate": 2.508984317193948e-05,
+      "loss": 0.508,
+      "step": 9380
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.7632963061332703,
+      "learning_rate": 2.5068949472025943e-05,
+      "loss": 0.7146,
+      "step": 9385
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.7676249742507935,
+      "learning_rate": 2.504805572395245e-05,
+      "loss": 0.7835,
+      "step": 9390
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6752475500106812,
+      "learning_rate": 2.5027161942312922e-05,
+      "loss": 0.6139,
+      "step": 9395
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.7635406255722046,
+      "learning_rate": 2.5006268141701274e-05,
+      "loss": 0.6019,
+      "step": 9400
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 0.6944969296455383,
+      "eval_runtime": 96.5995,
+      "eval_samples_per_second": 7.215,
+      "eval_steps_per_second": 7.215,
+      "step": 9400
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8899182081222534,
+      "learning_rate": 2.4985374336711438e-05,
+      "loss": 0.738,
+      "step": 9405
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.17216157913208,
+      "learning_rate": 2.496448054193735e-05,
+      "loss": 0.6101,
+      "step": 9410
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8314846754074097,
+      "learning_rate": 2.4943586771972965e-05,
+      "loss": 0.69,
+      "step": 9415
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.7142031192779541,
+      "learning_rate": 2.4922693041412202e-05,
+      "loss": 0.5057,
+      "step": 9420
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.2739388942718506,
+      "learning_rate": 2.4901799364848925e-05,
+      "loss": 0.4485,
+      "step": 9425
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.9823365211486816,
+      "learning_rate": 2.4880905756877008e-05,
+      "loss": 0.7312,
+      "step": 9430
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.3676027059555054,
+      "learning_rate": 2.486001223209024e-05,
+      "loss": 0.6295,
+      "step": 9435
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8690130710601807,
+      "learning_rate": 2.4839118805082377e-05,
+      "loss": 0.6172,
+      "step": 9440
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7541719079017639,
+      "learning_rate": 2.4818225490447097e-05,
+      "loss": 0.6258,
+      "step": 9445
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.6225134134292603,
+      "learning_rate": 2.4797332302777993e-05,
+      "loss": 0.5882,
+      "step": 9450
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.886218786239624,
+      "learning_rate": 2.4776439256668576e-05,
+      "loss": 0.5042,
+      "step": 9455
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7720217108726501,
+      "learning_rate": 2.4755546366712272e-05,
+      "loss": 0.8231,
+      "step": 9460
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7719532251358032,
+      "learning_rate": 2.4734653647502363e-05,
+      "loss": 0.5943,
+      "step": 9465
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7584754228591919,
+      "learning_rate": 2.4713761113632032e-05,
+      "loss": 0.6572,
+      "step": 9470
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8645241856575012,
+      "learning_rate": 2.4692868779694358e-05,
+      "loss": 0.6461,
+      "step": 9475
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8362308740615845,
+      "learning_rate": 2.467197666028223e-05,
+      "loss": 0.8308,
+      "step": 9480
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.881805956363678,
+      "learning_rate": 2.465108476998842e-05,
+      "loss": 0.84,
+      "step": 9485
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7092866897583008,
+      "learning_rate": 2.4630193123405542e-05,
+      "loss": 0.6114,
+      "step": 9490
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6957134008407593,
+      "learning_rate": 2.4609301735126015e-05,
+      "loss": 0.7143,
+      "step": 9495
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.1079517602920532,
+      "learning_rate": 2.45884106197421e-05,
+      "loss": 0.6707,
+      "step": 9500
+    },
+    {
+      "epoch": 1.52,
+      "eval_loss": 0.694333553314209,
+      "eval_runtime": 96.6934,
+      "eval_samples_per_second": 7.208,
+      "eval_steps_per_second": 7.208,
+      "step": 9500
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9612390398979187,
+      "learning_rate": 2.4567519791845857e-05,
+      "loss": 0.776,
+      "step": 9505
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.011856198310852,
+      "learning_rate": 2.4546629266029155e-05,
+      "loss": 0.8204,
+      "step": 9510
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9229696393013,
+      "learning_rate": 2.4525739056883644e-05,
+      "loss": 0.7631,
+      "step": 9515
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7902365922927856,
+      "learning_rate": 2.4504849179000746e-05,
+      "loss": 0.7414,
+      "step": 9520
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5655250549316406,
+      "learning_rate": 2.448395964697166e-05,
+      "loss": 0.7102,
+      "step": 9525
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6864548921585083,
+      "learning_rate": 2.4463070475387353e-05,
+      "loss": 0.6807,
+      "step": 9530
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9175466299057007,
+      "learning_rate": 2.444218167883852e-05,
+      "loss": 0.6063,
+      "step": 9535
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7826316356658936,
+      "learning_rate": 2.4421293271915608e-05,
+      "loss": 0.8181,
+      "step": 9540
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6215065121650696,
+      "learning_rate": 2.4400405269208797e-05,
+      "loss": 0.5949,
+      "step": 9545
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.8049483895301819,
+      "learning_rate": 2.437951768530796e-05,
+      "loss": 0.6341,
+      "step": 9550
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.45780548453330994,
+      "learning_rate": 2.43586305348027e-05,
+      "loss": 1.0438,
+      "step": 9555
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7412901520729065,
+      "learning_rate": 2.4337743832282323e-05,
+      "loss": 0.7247,
+      "step": 9560
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.0728024244308472,
+      "learning_rate": 2.431685759233579e-05,
+      "loss": 0.5836,
+      "step": 9565
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.903080701828003,
+      "learning_rate": 2.4295971829551766e-05,
+      "loss": 0.8739,
+      "step": 9570
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8286559581756592,
+      "learning_rate": 2.427508655851859e-05,
+      "loss": 0.8085,
+      "step": 9575
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8452445864677429,
+      "learning_rate": 2.4254201793824216e-05,
+      "loss": 0.749,
+      "step": 9580
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.992088258266449,
+      "learning_rate": 2.423331755005629e-05,
+      "loss": 0.635,
+      "step": 9585
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5679229497909546,
+      "learning_rate": 2.4212433841802067e-05,
+      "loss": 0.8887,
+      "step": 9590
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7643200159072876,
+      "learning_rate": 2.419155068364844e-05,
+      "loss": 0.7613,
+      "step": 9595
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7898627519607544,
+      "learning_rate": 2.4170668090181915e-05,
+      "loss": 0.7269,
+      "step": 9600
+    },
+    {
+      "epoch": 1.53,
+      "eval_loss": 0.6939559578895569,
+      "eval_runtime": 96.732,
+      "eval_samples_per_second": 7.205,
+      "eval_steps_per_second": 7.205,
+      "step": 9600
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8673202395439148,
+      "learning_rate": 2.4149786075988602e-05,
+      "loss": 0.8336,
+      "step": 9605
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8820798993110657,
+      "learning_rate": 2.41289046556542e-05,
+      "loss": 0.631,
+      "step": 9610
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.9562051296234131,
+      "learning_rate": 2.410802384376402e-05,
+      "loss": 0.637,
+      "step": 9615
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8073970675468445,
+      "learning_rate": 2.4087143654902893e-05,
+      "loss": 0.7949,
+      "step": 9620
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.629291296005249,
+      "learning_rate": 2.4066264103655283e-05,
+      "loss": 0.8292,
+      "step": 9625
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.6791005730628967,
+      "learning_rate": 2.404538520460517e-05,
+      "loss": 0.867,
+      "step": 9630
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8496805429458618,
+      "learning_rate": 2.402450697233607e-05,
+      "loss": 0.7857,
+      "step": 9635
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.7444857954978943,
+      "learning_rate": 2.4003629421431052e-05,
+      "loss": 0.658,
+      "step": 9640
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.0270099639892578,
+      "learning_rate": 2.398275256647272e-05,
+      "loss": 0.6766,
+      "step": 9645
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.839475691318512,
+      "learning_rate": 2.3961876422043154e-05,
+      "loss": 0.5771,
+      "step": 9650
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.7087832689285278,
+      "learning_rate": 2.394100100272397e-05,
+      "loss": 0.7797,
+      "step": 9655
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8930874466896057,
+      "learning_rate": 2.3920126323096275e-05,
+      "loss": 0.7951,
+      "step": 9660
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.2568659782409668,
+      "learning_rate": 2.389925239774064e-05,
+      "loss": 0.7785,
+      "step": 9665
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.0711442232131958,
+      "learning_rate": 2.3878379241237136e-05,
+      "loss": 0.974,
+      "step": 9670
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.7370679974555969,
+      "learning_rate": 2.385750686816526e-05,
+      "loss": 0.7111,
+      "step": 9675
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8300942182540894,
+      "learning_rate": 2.384080954357563e-05,
+      "loss": 0.9931,
+      "step": 9680
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.7908097505569458,
+      "learning_rate": 2.3819938617419347e-05,
+      "loss": 0.729,
+      "step": 9685
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8966140747070312,
+      "learning_rate": 2.379906851551441e-05,
+      "loss": 0.7695,
+      "step": 9690
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5384413003921509,
+      "learning_rate": 2.3778199252438203e-05,
+      "loss": 0.6235,
+      "step": 9695
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.7455211877822876,
+      "learning_rate": 2.3757330842767546e-05,
+      "loss": 0.7216,
+      "step": 9700
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 0.6923454999923706,
+      "eval_runtime": 96.5713,
+      "eval_samples_per_second": 7.217,
+      "eval_steps_per_second": 7.217,
+      "step": 9700
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.0986971855163574,
+      "learning_rate": 2.3736463301078614e-05,
+      "loss": 0.6697,
+      "step": 9705
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8946139216423035,
+      "learning_rate": 2.3715596641947022e-05,
+      "loss": 0.6864,
+      "step": 9710
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8130088448524475,
+      "learning_rate": 2.3694730879947754e-05,
+      "loss": 0.6585,
+      "step": 9715
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8894361853599548,
+      "learning_rate": 2.367386602965515e-05,
+      "loss": 0.6961,
+      "step": 9720
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.7765643000602722,
+      "learning_rate": 2.3653002105642922e-05,
+      "loss": 0.6075,
+      "step": 9725
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.9015631079673767,
+      "learning_rate": 2.3632139122484173e-05,
+      "loss": 0.6439,
+      "step": 9730
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.1747032403945923,
+      "learning_rate": 2.3611277094751283e-05,
+      "loss": 0.6192,
+      "step": 9735
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.7363355159759521,
+      "learning_rate": 2.359041603701601e-05,
+      "loss": 1.036,
+      "step": 9740
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9469441771507263,
+      "learning_rate": 2.356955596384943e-05,
+      "loss": 0.6255,
+      "step": 9745
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9195448756217957,
+      "learning_rate": 2.3548696889821907e-05,
+      "loss": 0.7055,
+      "step": 9750
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.0644798278808594,
+      "learning_rate": 2.352783882950315e-05,
+      "loss": 0.7532,
+      "step": 9755
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.6971336603164673,
+      "learning_rate": 2.3506981797462094e-05,
+      "loss": 0.5817,
+      "step": 9760
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7396060824394226,
+      "learning_rate": 2.3486125808267025e-05,
+      "loss": 0.6482,
+      "step": 9765
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7229952216148376,
+      "learning_rate": 2.3465270876485474e-05,
+      "loss": 0.7125,
+      "step": 9770
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7994909286499023,
+      "learning_rate": 2.344441701668421e-05,
+      "loss": 0.5666,
+      "step": 9775
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.6988499760627747,
+      "learning_rate": 2.3423564243429282e-05,
+      "loss": 0.8044,
+      "step": 9780
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.8466933369636536,
+      "learning_rate": 2.340271257128598e-05,
+      "loss": 0.6553,
+      "step": 9785
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5719871520996094,
+      "learning_rate": 2.3381862014818807e-05,
+      "loss": 0.7292,
+      "step": 9790
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9753498435020447,
+      "learning_rate": 2.33610125885915e-05,
+      "loss": 0.8397,
+      "step": 9795
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.608966052532196,
+      "learning_rate": 2.3340164307167007e-05,
+      "loss": 0.6394,
+      "step": 9800
+    },
+    {
+      "epoch": 1.56,
+      "eval_loss": 0.6920195817947388,
+      "eval_runtime": 96.5737,
+      "eval_samples_per_second": 7.217,
+      "eval_steps_per_second": 7.217,
+      "step": 9800
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.151438593864441,
+      "learning_rate": 2.331931718510746e-05,
+      "loss": 0.6101,
+      "step": 9805
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.9338546395301819,
+      "learning_rate": 2.3298471236974202e-05,
+      "loss": 0.8272,
+      "step": 9810
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.1721985340118408,
+      "learning_rate": 2.327762647732776e-05,
+      "loss": 0.6039,
+      "step": 9815
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.8202813863754272,
+      "learning_rate": 2.3256782920727787e-05,
+      "loss": 0.6222,
+      "step": 9820
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5816913843154907,
+      "learning_rate": 2.323594058173315e-05,
+      "loss": 0.6952,
+      "step": 9825
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.4169448614120483,
+      "learning_rate": 2.321509947490185e-05,
+      "loss": 0.7452,
+      "step": 9830
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.9609310626983643,
+      "learning_rate": 2.3194259614790998e-05,
+      "loss": 0.6724,
+      "step": 9835
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7673792243003845,
+      "learning_rate": 2.317342101595687e-05,
+      "loss": 0.9891,
+      "step": 9840
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.8118376135826111,
+      "learning_rate": 2.3152583692954835e-05,
+      "loss": 0.6174,
+      "step": 9845
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.775577187538147,
+      "learning_rate": 2.3131747660339394e-05,
+      "loss": 0.6912,
+      "step": 9850
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.860774040222168,
+      "learning_rate": 2.3110912932664138e-05,
+      "loss": 0.8078,
+      "step": 9855
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7636804580688477,
+      "learning_rate": 2.3090079524481734e-05,
+      "loss": 0.6305,
+      "step": 9860
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.1903213262557983,
+      "learning_rate": 2.3069247450343943e-05,
+      "loss": 0.7058,
+      "step": 9865
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.9413545727729797,
+      "learning_rate": 2.30484167248016e-05,
+      "loss": 0.7658,
+      "step": 9870
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.8364406228065491,
+      "learning_rate": 2.3027587362404558e-05,
+      "loss": 0.6607,
+      "step": 9875
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.1050459146499634,
+      "learning_rate": 2.3006759377701765e-05,
+      "loss": 0.8299,
+      "step": 9880
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.7848144173622131,
+      "learning_rate": 2.2985932785241203e-05,
+      "loss": 0.7717,
+      "step": 9885
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.9436622858047485,
+      "learning_rate": 2.2965107599569836e-05,
+      "loss": 0.8987,
+      "step": 9890
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.8566385507583618,
+      "learning_rate": 2.2944283835233697e-05,
+      "loss": 0.6131,
+      "step": 9895
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.8907533288002014,
+      "learning_rate": 2.29234615067778e-05,
+      "loss": 0.7608,
+      "step": 9900
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.6909372210502625,
+      "eval_runtime": 96.6101,
+      "eval_samples_per_second": 7.215,
+      "eval_steps_per_second": 7.215,
+      "step": 9900
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.7827823758125305,
+      "learning_rate": 2.290264062874616e-05,
+      "loss": 0.7079,
+      "step": 9905
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.2197152376174927,
+      "learning_rate": 2.288182121568178e-05,
+      "loss": 0.9356,
+      "step": 9910
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.8879411816596985,
+      "learning_rate": 2.286100328212665e-05,
+      "loss": 0.6318,
+      "step": 9915
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6319537162780762,
+      "learning_rate": 2.28401868426217e-05,
+      "loss": 0.5469,
+      "step": 9920
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.0840219259262085,
+      "learning_rate": 2.281937191170686e-05,
+      "loss": 0.5847,
+      "step": 9925
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4195910692214966,
+      "learning_rate": 2.2798558503920948e-05,
+      "loss": 0.7168,
+      "step": 9930
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.0517491102218628,
+      "learning_rate": 2.2777746633801764e-05,
+      "loss": 0.5673,
+      "step": 9935
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.8244277834892273,
+      "learning_rate": 2.2756936315886024e-05,
+      "loss": 0.7266,
+      "step": 9940
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.8401049375534058,
+      "learning_rate": 2.2736127564709346e-05,
+      "loss": 0.8369,
+      "step": 9945
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.9072527289390564,
+      "learning_rate": 2.2715320394806265e-05,
+      "loss": 0.664,
+      "step": 9950
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6931943893432617,
+      "learning_rate": 2.269451482071022e-05,
+      "loss": 1.0078,
+      "step": 9955
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.7260577082633972,
+      "learning_rate": 2.2673710856953512e-05,
+      "loss": 0.5779,
+      "step": 9960
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.784848690032959,
+      "learning_rate": 2.2652908518067336e-05,
+      "loss": 0.7065,
+      "step": 9965
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.9373456239700317,
+      "learning_rate": 2.2632107818581757e-05,
+      "loss": 0.8002,
+      "step": 9970
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6380440592765808,
+      "learning_rate": 2.261130877302566e-05,
+      "loss": 0.7027,
+      "step": 9975
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6546103954315186,
+      "learning_rate": 2.259051139592681e-05,
+      "loss": 0.6839,
+      "step": 9980
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6351701021194458,
+      "learning_rate": 2.2569715701811814e-05,
+      "loss": 0.7078,
+      "step": 9985
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.8696810603141785,
+      "learning_rate": 2.2548921705206057e-05,
+      "loss": 0.7586,
+      "step": 9990
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.7104160785675049,
+      "learning_rate": 2.2528129420633785e-05,
+      "loss": 0.7514,
+      "step": 9995
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.8413506746292114,
+      "learning_rate": 2.2507338862618012e-05,
+      "loss": 1.034,
+      "step": 10000
+    },
+    {
+      "epoch": 1.6,
+      "eval_loss": 0.6908321380615234,
+      "eval_runtime": 96.5988,
+      "eval_samples_per_second": 7.215,
+      "eval_steps_per_second": 7.215,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 18795,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 4.3155317587968e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}