diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8421 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.15573493913359462,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 16.694459915161133,
+      "learning_rate": 5e-06,
+      "loss": 7.0517,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.555038452148438,
+      "learning_rate": 1e-05,
+      "loss": 6.8494,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 29.299560546875,
+      "learning_rate": 1.5e-05,
+      "loss": 6.571,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.319277286529541,
+      "learning_rate": 2e-05,
+      "loss": 6.2651,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.005606174468994,
+      "learning_rate": 2.5e-05,
+      "loss": 5.9421,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.155610084533691,
+      "learning_rate": 3e-05,
+      "loss": 5.6719,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.178586959838867,
+      "learning_rate": 3.5e-05,
+      "loss": 5.3734,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.331709384918213,
+      "learning_rate": 4e-05,
+      "loss": 4.9898,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 32.85832977294922,
+      "learning_rate": 4.5e-05,
+      "loss": 4.7119,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7961432933807373,
+      "learning_rate": 5e-05,
+      "loss": 4.518,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7457516193389893,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.3472,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8961405754089355,
+      "learning_rate": 6e-05,
+      "loss": 4.1619,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.1687309741973877,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.0246,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0345044136047363,
+      "learning_rate": 7e-05,
+      "loss": 3.9169,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9801828861236572,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.9073,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7653589248657227,
+      "learning_rate": 8e-05,
+      "loss": 3.7324,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9164013862609863,
+      "learning_rate": 8.5e-05,
+      "loss": 3.8318,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.860090732574463,
+      "learning_rate": 9e-05,
+      "loss": 3.7374,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.771846771240234,
+      "learning_rate": 9.5e-05,
+      "loss": 3.5654,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.3580846786499023,
+      "learning_rate": 0.0001,
+      "loss": 3.6788,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.8400185108184814,
+      "learning_rate": 9.999999582259295e-05,
+      "loss": 3.545,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.759910821914673,
+      "learning_rate": 9.999998329037248e-05,
+      "loss": 3.4873,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.470923900604248,
+      "learning_rate": 9.999996240334068e-05,
+      "loss": 3.5124,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.7801690101623535,
+      "learning_rate": 9.999993316150106e-05,
+      "loss": 3.4615,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2033209800720215,
+      "learning_rate": 9.999989556485852e-05,
+      "loss": 3.4743,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.036935329437256,
+      "learning_rate": 9.999984961341928e-05,
+      "loss": 3.4325,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.526097297668457,
+      "learning_rate": 9.999979530719108e-05,
+      "loss": 3.413,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.5993170738220215,
+      "learning_rate": 9.999973264618297e-05,
+      "loss": 3.3975,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3761978149414062,
+      "learning_rate": 9.999966163040541e-05,
+      "loss": 3.4034,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.243953227996826,
+      "learning_rate": 9.99995822598703e-05,
+      "loss": 3.3272,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1254262924194336,
+      "learning_rate": 9.999949453459088e-05,
+      "loss": 3.3863,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0516600608825684,
+      "learning_rate": 9.999939845458179e-05,
+      "loss": 3.3217,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.231921434402466,
+      "learning_rate": 9.999929401985911e-05,
+      "loss": 3.3736,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1770987510681152,
+      "learning_rate": 9.99991812304403e-05,
+      "loss": 3.247,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.155881881713867,
+      "learning_rate": 9.999906008634416e-05,
+      "loss": 3.2366,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1822783946990967,
+      "learning_rate": 9.9998930587591e-05,
+      "loss": 3.1996,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9985451698303223,
+      "learning_rate": 9.99987927342024e-05,
+      "loss": 3.1917,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9846060276031494,
+      "learning_rate": 9.999864652620143e-05,
+      "loss": 3.1866,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.978461503982544,
+      "learning_rate": 9.999849196361251e-05,
+      "loss": 3.2095,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.951501488685608,
+      "learning_rate": 9.999832904646147e-05,
+      "loss": 3.2103,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8836382627487183,
+      "learning_rate": 9.99981577747755e-05,
+      "loss": 3.14,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9959434270858765,
+      "learning_rate": 9.999797814858327e-05,
+      "loss": 3.2671,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0655322074890137,
+      "learning_rate": 9.999779016791477e-05,
+      "loss": 3.1047,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0735039710998535,
+      "learning_rate": 9.999759383280141e-05,
+      "loss": 3.2484,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.947547197341919,
+      "learning_rate": 9.9997389143276e-05,
+      "loss": 3.2312,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.10322904586792,
+      "learning_rate": 9.999717609937275e-05,
+      "loss": 3.2057,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0274617671966553,
+      "learning_rate": 9.999695470112725e-05,
+      "loss": 3.1822,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.832457423210144,
+      "learning_rate": 9.99967249485765e-05,
+      "loss": 3.1618,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8780583143234253,
+      "learning_rate": 9.999648684175888e-05,
+      "loss": 3.1652,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9842203855514526,
+      "learning_rate": 9.999624038071418e-05,
+      "loss": 3.2369,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9477784633636475,
+      "learning_rate": 9.99959855654836e-05,
+      "loss": 3.1294,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.213186025619507,
+      "learning_rate": 9.999572239610968e-05,
+      "loss": 3.0936,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7027944326400757,
+      "learning_rate": 9.999545087263645e-05,
+      "loss": 3.1281,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8745968341827393,
+      "learning_rate": 9.999517099510923e-05,
+      "loss": 3.1393,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.82529878616333,
+      "learning_rate": 9.999488276357481e-05,
+      "loss": 3.0725,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6872284412384033,
+      "learning_rate": 9.999458617808137e-05,
+      "loss": 3.0868,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7451159954071045,
+      "learning_rate": 9.999428123867843e-05,
+      "loss": 3.1431,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6614208221435547,
+      "learning_rate": 9.999396794541695e-05,
+      "loss": 3.031,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8496636152267456,
+      "learning_rate": 9.999364629834931e-05,
+      "loss": 3.1492,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8741862773895264,
+      "learning_rate": 9.999331629752924e-05,
+      "loss": 3.0386,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6683368682861328,
+      "learning_rate": 9.999297794301188e-05,
+      "loss": 2.9959,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7165924310684204,
+      "learning_rate": 9.999263123485376e-05,
+      "loss": 3.0575,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6788839101791382,
+      "learning_rate": 9.999227617311282e-05,
+      "loss": 3.0931,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.258460760116577,
+      "learning_rate": 9.99919127578484e-05,
+      "loss": 3.0206,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5884743928909302,
+      "learning_rate": 9.999154098912122e-05,
+      "loss": 2.9483,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.593381643295288,
+      "learning_rate": 9.999116086699338e-05,
+      "loss": 3.103,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6749074459075928,
+      "learning_rate": 9.999077239152843e-05,
+      "loss": 3.0349,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5321128368377686,
+      "learning_rate": 9.999037556279127e-05,
+      "loss": 3.1176,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7349498271942139,
+      "learning_rate": 9.998997038084818e-05,
+      "loss": 2.9748,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5452516078948975,
+      "learning_rate": 9.998955684576692e-05,
+      "loss": 3.0371,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6734381914138794,
+      "learning_rate": 9.998913495761654e-05,
+      "loss": 3.0451,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6274564266204834,
+      "learning_rate": 9.998870471646757e-05,
+      "loss": 2.9161,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4812135696411133,
+      "learning_rate": 9.99882661223919e-05,
+      "loss": 2.9783,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9705489873886108,
+      "learning_rate": 9.998781917546278e-05,
+      "loss": 3.004,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.530045509338379,
+      "learning_rate": 9.998736387575493e-05,
+      "loss": 2.9929,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5938191413879395,
+      "learning_rate": 9.998690022334442e-05,
+      "loss": 2.9642,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7416777610778809,
+      "learning_rate": 9.998642821830873e-05,
+      "loss": 2.9631,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4602569341659546,
+      "learning_rate": 9.99859478607267e-05,
+      "loss": 2.9334,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7514872550964355,
+      "learning_rate": 9.998545915067864e-05,
+      "loss": 3.0424,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5700337886810303,
+      "learning_rate": 9.998496208824618e-05,
+      "loss": 2.8512,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7854647636413574,
+      "learning_rate": 9.99844566735124e-05,
+      "loss": 2.9332,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.721405029296875,
+      "learning_rate": 9.998394290656172e-05,
+      "loss": 2.9415,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4982709884643555,
+      "learning_rate": 9.998342078748002e-05,
+      "loss": 2.9164,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4483710527420044,
+      "learning_rate": 9.998289031635453e-05,
+      "loss": 2.9553,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6196078062057495,
+      "learning_rate": 9.998235149327389e-05,
+      "loss": 3.0263,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9381450414657593,
+      "learning_rate": 9.998180431832816e-05,
+      "loss": 2.8707,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.689458966255188,
+      "learning_rate": 9.998124879160872e-05,
+      "loss": 3.0073,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4204018115997314,
+      "learning_rate": 9.998068491320844e-05,
+      "loss": 2.9653,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4616014957427979,
+      "learning_rate": 9.998011268322152e-05,
+      "loss": 2.809,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.962567925453186,
+      "learning_rate": 9.997953210174361e-05,
+      "loss": 2.9659,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.573438286781311,
+      "learning_rate": 9.997894316887167e-05,
+      "loss": 2.9086,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5592458248138428,
+      "learning_rate": 9.997834588470414e-05,
+      "loss": 2.9147,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5808589458465576,
+      "learning_rate": 9.997774024934082e-05,
+      "loss": 2.9209,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4629749059677124,
+      "learning_rate": 9.997712626288292e-05,
+      "loss": 2.9284,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.589766025543213,
+      "learning_rate": 9.997650392543303e-05,
+      "loss": 2.9241,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5819960832595825,
+      "learning_rate": 9.997587323709512e-05,
+      "loss": 2.9587,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6048158407211304,
+      "learning_rate": 9.997523419797459e-05,
+      "loss": 2.9088,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6197158098220825,
+      "learning_rate": 9.997458680817824e-05,
+      "loss": 2.8723,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4653692245483398,
+      "learning_rate": 9.997393106781421e-05,
+      "loss": 2.9394,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4455420970916748,
+      "learning_rate": 9.99732669769921e-05,
+      "loss": 2.8423,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.731045126914978,
+      "learning_rate": 9.997259453582286e-05,
+      "loss": 2.8767,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4818463325500488,
+      "learning_rate": 9.997191374441887e-05,
+      "loss": 2.8679,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5123176574707031,
+      "learning_rate": 9.997122460289387e-05,
+      "loss": 2.8927,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3707941770553589,
+      "learning_rate": 9.997052711136301e-05,
+      "loss": 2.8869,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4157047271728516,
+      "learning_rate": 9.996982126994287e-05,
+      "loss": 2.8204,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.498525619506836,
+      "learning_rate": 9.996910707875136e-05,
+      "loss": 2.8487,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.434728741645813,
+      "learning_rate": 9.996838453790784e-05,
+      "loss": 2.9148,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5604894161224365,
+      "learning_rate": 9.996765364753302e-05,
+      "loss": 2.8488,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5228891372680664,
+      "learning_rate": 9.996691440774906e-05,
+      "loss": 2.9394,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4666147232055664,
+      "learning_rate": 9.996616681867945e-05,
+      "loss": 2.7372,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.326279640197754,
+      "learning_rate": 9.996541088044913e-05,
+      "loss": 2.7743,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5345813035964966,
+      "learning_rate": 9.996464659318442e-05,
+      "loss": 2.9155,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3781248331069946,
+      "learning_rate": 9.996387395701302e-05,
+      "loss": 2.8862,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3769475221633911,
+      "learning_rate": 9.996309297206403e-05,
+      "loss": 2.9665,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 93.51887512207031,
+      "learning_rate": 9.996230363846797e-05,
+      "loss": 2.766,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.470819354057312,
+      "learning_rate": 9.996150595635671e-05,
+      "loss": 2.8976,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.455989956855774,
+      "learning_rate": 9.996069992586355e-05,
+      "loss": 2.8753,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7001138925552368,
+      "learning_rate": 9.995988554712318e-05,
+      "loss": 2.8228,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3552448749542236,
+      "learning_rate": 9.995906282027166e-05,
+      "loss": 2.8087,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3833279609680176,
+      "learning_rate": 9.995823174544649e-05,
+      "loss": 2.8641,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3860262632369995,
+      "learning_rate": 9.995739232278654e-05,
+      "loss": 2.7923,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5044357776641846,
+      "learning_rate": 9.995654455243204e-05,
+      "loss": 2.9038,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.378627061843872,
+      "learning_rate": 9.995568843452468e-05,
+      "loss": 2.8108,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3770326375961304,
+      "learning_rate": 9.99548239692075e-05,
+      "loss": 2.78,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.398013949394226,
+      "learning_rate": 9.995395115662498e-05,
+      "loss": 2.8295,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4935119152069092,
+      "learning_rate": 9.995306999692293e-05,
+      "loss": 2.8256,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3398939371109009,
+      "learning_rate": 9.995218049024858e-05,
+      "loss": 2.7482,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4971387386322021,
+      "learning_rate": 9.995128263675058e-05,
+      "loss": 2.7946,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5211292505264282,
+      "learning_rate": 9.995037643657899e-05,
+      "loss": 2.874,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3577936887741089,
+      "learning_rate": 9.994946188988516e-05,
+      "loss": 2.8901,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.396310567855835,
+      "learning_rate": 9.994853899682197e-05,
+      "loss": 2.832,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3555642366409302,
+      "learning_rate": 9.994760775754362e-05,
+      "loss": 2.7835,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.388643741607666,
+      "learning_rate": 9.994666817220571e-05,
+      "loss": 2.8192,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3072245121002197,
+      "learning_rate": 9.994572024096521e-05,
+      "loss": 2.7829,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4088162183761597,
+      "learning_rate": 9.994476396398055e-05,
+      "loss": 2.8474,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4542754888534546,
+      "learning_rate": 9.994379934141153e-05,
+      "loss": 2.8988,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3547110557556152,
+      "learning_rate": 9.994282637341932e-05,
+      "loss": 2.8288,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3031294345855713,
+      "learning_rate": 9.994184506016651e-05,
+      "loss": 2.7699,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3675835132598877,
+      "learning_rate": 9.994085540181703e-05,
+      "loss": 2.7619,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4714610576629639,
+      "learning_rate": 9.99398573985363e-05,
+      "loss": 2.7984,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3885003328323364,
+      "learning_rate": 9.993885105049107e-05,
+      "loss": 2.784,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2875152826309204,
+      "learning_rate": 9.993783635784947e-05,
+      "loss": 2.7799,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4449490308761597,
+      "learning_rate": 9.993681332078108e-05,
+      "loss": 2.7587,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2932398319244385,
+      "learning_rate": 9.993578193945685e-05,
+      "loss": 2.7646,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3415541648864746,
+      "learning_rate": 9.99347422140491e-05,
+      "loss": 2.7156,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3976376056671143,
+      "learning_rate": 9.993369414473157e-05,
+      "loss": 2.8197,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4145207405090332,
+      "learning_rate": 9.993263773167939e-05,
+      "loss": 2.7966,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3587496280670166,
+      "learning_rate": 9.993157297506907e-05,
+      "loss": 2.7801,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4527009725570679,
+      "learning_rate": 9.993049987507856e-05,
+      "loss": 2.7119,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4694035053253174,
+      "learning_rate": 9.992941843188715e-05,
+      "loss": 2.7307,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4610073566436768,
+      "learning_rate": 9.992832864567554e-05,
+      "loss": 2.8147,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3035316467285156,
+      "learning_rate": 9.992723051662584e-05,
+      "loss": 2.819,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5005671977996826,
+      "learning_rate": 9.992612404492153e-05,
+      "loss": 2.763,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4197033643722534,
+      "learning_rate": 9.992500923074752e-05,
+      "loss": 2.8168,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3414314985275269,
+      "learning_rate": 9.992388607429006e-05,
+      "loss": 2.7943,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3533778190612793,
+      "learning_rate": 9.992275457573685e-05,
+      "loss": 2.6949,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4098223447799683,
+      "learning_rate": 9.992161473527695e-05,
+      "loss": 2.8062,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2892730236053467,
+      "learning_rate": 9.992046655310083e-05,
+      "loss": 2.742,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3262995481491089,
+      "learning_rate": 9.991931002940034e-05,
+      "loss": 2.7459,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6442351341247559,
+      "learning_rate": 9.991814516436873e-05,
+      "loss": 2.6766,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3681560754776,
+      "learning_rate": 9.991697195820065e-05,
+      "loss": 2.7411,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.375429391860962,
+      "learning_rate": 9.991579041109212e-05,
+      "loss": 2.7042,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4320073127746582,
+      "learning_rate": 9.991460052324061e-05,
+      "loss": 2.7202,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3459241390228271,
+      "learning_rate": 9.991340229484491e-05,
+      "loss": 2.817,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2648005485534668,
+      "learning_rate": 9.991219572610526e-05,
+      "loss": 2.7615,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3283494710922241,
+      "learning_rate": 9.991098081722325e-05,
+      "loss": 2.8405,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3819454908370972,
+      "learning_rate": 9.990975756840193e-05,
+      "loss": 2.7977,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.178113341331482,
+      "learning_rate": 9.990852597984566e-05,
+      "loss": 2.7456,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.30787193775177,
+      "learning_rate": 9.990728605176025e-05,
+      "loss": 2.7424,
+      "step": 845
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3064510822296143,
+      "learning_rate": 9.990603778435288e-05,
+      "loss": 2.7829,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3056379556655884,
+      "learning_rate": 9.990478117783214e-05,
+      "loss": 2.7477,
+      "step": 855
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2564347982406616,
+      "learning_rate": 9.990351623240799e-05,
+      "loss": 2.7946,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3979121446609497,
+      "learning_rate": 9.990224294829181e-05,
+      "loss": 2.7785,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3233988285064697,
+      "learning_rate": 9.990096132569636e-05,
+      "loss": 2.7582,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.27315354347229,
+      "learning_rate": 9.98996713648358e-05,
+      "loss": 2.7432,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2971699237823486,
+      "learning_rate": 9.989837306592567e-05,
+      "loss": 2.713,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2790080308914185,
+      "learning_rate": 9.989706642918291e-05,
+      "loss": 2.6743,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2432963848114014,
+      "learning_rate": 9.989575145482583e-05,
+      "loss": 2.7015,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.248884916305542,
+      "learning_rate": 9.989442814307422e-05,
+      "loss": 2.6055,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2589277029037476,
+      "learning_rate": 9.989309649414914e-05,
+      "loss": 2.823,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3967111110687256,
+      "learning_rate": 9.989175650827314e-05,
+      "loss": 2.696,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2264184951782227,
+      "learning_rate": 9.98904081856701e-05,
+      "loss": 2.7369,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2057732343673706,
+      "learning_rate": 9.988905152656534e-05,
+      "loss": 2.685,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3205448389053345,
+      "learning_rate": 9.988768653118554e-05,
+      "loss": 2.7085,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2320032119750977,
+      "learning_rate": 9.988631319975881e-05,
+      "loss": 2.7486,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2307050228118896,
+      "learning_rate": 9.988493153251459e-05,
+      "loss": 2.6843,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2744957208633423,
+      "learning_rate": 9.988354152968379e-05,
+      "loss": 2.7852,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.229554295539856,
+      "learning_rate": 9.988214319149865e-05,
+      "loss": 2.7506,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2029807567596436,
+      "learning_rate": 9.988073651819281e-05,
+      "loss": 2.619,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2296098470687866,
+      "learning_rate": 9.987932151000137e-05,
+      "loss": 2.7678,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2401868104934692,
+      "learning_rate": 9.987789816716073e-05,
+      "loss": 2.7287,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2568613290786743,
+      "learning_rate": 9.987646648990876e-05,
+      "loss": 2.7619,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2899633646011353,
+      "learning_rate": 9.987502647848466e-05,
+      "loss": 2.7561,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2250640392303467,
+      "learning_rate": 9.987357813312905e-05,
+      "loss": 2.6726,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2166798114776611,
+      "learning_rate": 9.987212145408395e-05,
+      "loss": 2.637,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1826984882354736,
+      "learning_rate": 9.987065644159278e-05,
+      "loss": 2.7008,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2532957792282104,
+      "learning_rate": 9.986918309590031e-05,
+      "loss": 2.6961,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2717292308807373,
+      "learning_rate": 9.986770141725277e-05,
+      "loss": 2.6593,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2488884925842285,
+      "learning_rate": 9.986621140589772e-05,
+      "loss": 2.688,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2200218439102173,
+      "learning_rate": 9.986471306208412e-05,
+      "loss": 2.7618,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2542004585266113,
+      "learning_rate": 9.986320638606235e-05,
+      "loss": 2.778,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2081526517868042,
+      "learning_rate": 9.986169137808419e-05,
+      "loss": 2.7338,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1915208101272583,
+      "learning_rate": 9.986016803840275e-05,
+      "loss": 2.7009,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3669586181640625,
+      "learning_rate": 9.985863636727262e-05,
+      "loss": 2.6911,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.187229871749878,
+      "learning_rate": 9.985709636494971e-05,
+      "loss": 2.7024,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.280336856842041,
+      "learning_rate": 9.985554803169134e-05,
+      "loss": 2.7368,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3382422924041748,
+      "learning_rate": 9.985399136775627e-05,
+      "loss": 2.6641,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3108211755752563,
+      "learning_rate": 9.985242637340458e-05,
+      "loss": 2.7826,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.210227370262146,
+      "learning_rate": 9.985085304889778e-05,
+      "loss": 2.6693,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.290107011795044,
+      "learning_rate": 9.984927139449877e-05,
+      "loss": 2.6264,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1667602062225342,
+      "learning_rate": 9.984768141047184e-05,
+      "loss": 2.6979,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1986876726150513,
+      "learning_rate": 9.984608309708266e-05,
+      "loss": 2.7929,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3489216566085815,
+      "learning_rate": 9.984447645459831e-05,
+      "loss": 2.6875,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2335636615753174,
+      "learning_rate": 9.984286148328728e-05,
+      "loss": 2.7285,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1312981843948364,
+      "learning_rate": 9.984123818341937e-05,
+      "loss": 2.6781,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2616697549819946,
+      "learning_rate": 9.983960655526587e-05,
+      "loss": 2.6815,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2799261808395386,
+      "learning_rate": 9.983796659909941e-05,
+      "loss": 2.7166,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1821049451828003,
+      "learning_rate": 9.983631831519401e-05,
+      "loss": 2.7518,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2306567430496216,
+      "learning_rate": 9.983466170382511e-05,
+      "loss": 2.6697,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2967737913131714,
+      "learning_rate": 9.983299676526949e-05,
+      "loss": 2.6899,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3925654888153076,
+      "learning_rate": 9.98313234998054e-05,
+      "loss": 2.718,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2064905166625977,
+      "learning_rate": 9.982964190771242e-05,
+      "loss": 2.6919,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.822039246559143,
+      "learning_rate": 9.98279519892715e-05,
+      "loss": 2.6436,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1734588146209717,
+      "learning_rate": 9.982625374476508e-05,
+      "loss": 2.7434,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.198256492614746,
+      "learning_rate": 9.98245471744769e-05,
+      "loss": 2.7056,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.246696949005127,
+      "learning_rate": 9.982283227869211e-05,
+      "loss": 2.6653,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3072293996810913,
+      "learning_rate": 9.982110905769729e-05,
+      "loss": 2.6307,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2789084911346436,
+      "learning_rate": 9.981937751178037e-05,
+      "loss": 2.733,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2261282205581665,
+      "learning_rate": 9.981763764123067e-05,
+      "loss": 2.7011,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2762025594711304,
+      "learning_rate": 9.981588944633893e-05,
+      "loss": 2.6645,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2454729080200195,
+      "learning_rate": 9.981413292739727e-05,
+      "loss": 2.6511,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1904339790344238,
+      "learning_rate": 9.981236808469922e-05,
+      "loss": 2.6525,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1482722759246826,
+      "learning_rate": 9.981059491853964e-05,
+      "loss": 2.651,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1825352907180786,
+      "learning_rate": 9.980881342921482e-05,
+      "loss": 2.6883,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3523434400558472,
+      "learning_rate": 9.980702361702246e-05,
+      "loss": 2.6506,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.233847975730896,
+      "learning_rate": 9.980522548226162e-05,
+      "loss": 2.652,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1804288625717163,
+      "learning_rate": 9.980341902523279e-05,
+      "loss": 2.6791,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.250654697418213,
+      "learning_rate": 9.980160424623776e-05,
+      "loss": 2.7,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1734504699707031,
+      "learning_rate": 9.979978114557984e-05,
+      "loss": 2.6861,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1893103122711182,
+      "learning_rate": 9.979794972356362e-05,
+      "loss": 2.6489,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3656939268112183,
+      "learning_rate": 9.979610998049516e-05,
+      "loss": 2.6323,
+      "step": 1205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2057812213897705,
+      "learning_rate": 9.979426191668182e-05,
+      "loss": 2.6566,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.164214015007019,
+      "learning_rate": 9.979240553243246e-05,
+      "loss": 2.6822,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1863861083984375,
+      "learning_rate": 9.979054082805726e-05,
+      "loss": 2.6155,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2101305723190308,
+      "learning_rate": 9.978866780386777e-05,
+      "loss": 2.6826,
+      "step": 1225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1750643253326416,
+      "learning_rate": 9.978678646017701e-05,
+      "loss": 2.7178,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.174863338470459,
+      "learning_rate": 9.978489679729934e-05,
+      "loss": 2.6536,
+      "step": 1235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1917805671691895,
+      "learning_rate": 9.978299881555048e-05,
+      "loss": 2.6622,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.185265064239502,
+      "learning_rate": 9.97810925152476e-05,
+      "loss": 2.6336,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2117754220962524,
+      "learning_rate": 9.977917789670925e-05,
+      "loss": 2.6547,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2526233196258545,
+      "learning_rate": 9.977725496025535e-05,
+      "loss": 2.6936,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3354440927505493,
+      "learning_rate": 9.977532370620718e-05,
+      "loss": 2.6691,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1586819887161255,
+      "learning_rate": 9.977338413488748e-05,
+      "loss": 2.6894,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1898832321166992,
+      "learning_rate": 9.977143624662035e-05,
+      "loss": 2.5902,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1117446422576904,
+      "learning_rate": 9.976948004173125e-05,
+      "loss": 2.5829,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2026795148849487,
+      "learning_rate": 9.976751552054706e-05,
+      "loss": 2.6944,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2496997117996216,
+      "learning_rate": 9.976554268339607e-05,
+      "loss": 2.6832,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1364150047302246,
+      "learning_rate": 9.97635615306079e-05,
+      "loss": 2.6488,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1486976146697998,
+      "learning_rate": 9.976157206251361e-05,
+      "loss": 2.7186,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1302496194839478,
+      "learning_rate": 9.975957427944563e-05,
+      "loss": 2.5963,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2616629600524902,
+      "learning_rate": 9.975756818173778e-05,
+      "loss": 2.6154,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1366959810256958,
+      "learning_rate": 9.975555376972529e-05,
+      "loss": 2.6481,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1197172403335571,
+      "learning_rate": 9.975353104374474e-05,
+      "loss": 2.6264,
+      "step": 1315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1398710012435913,
+      "learning_rate": 9.975150000413411e-05,
+      "loss": 2.6111,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1890826225280762,
+      "learning_rate": 9.97494606512328e-05,
+      "loss": 2.6475,
+      "step": 1325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2538880109786987,
+      "learning_rate": 9.974741298538158e-05,
+      "loss": 2.6265,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1973954439163208,
+      "learning_rate": 9.974535700692257e-05,
+      "loss": 2.6508,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1694157123565674,
+      "learning_rate": 9.974329271619938e-05,
+      "loss": 2.6307,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2422724962234497,
+      "learning_rate": 9.97412201135569e-05,
+      "loss": 2.6052,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1929125785827637,
+      "learning_rate": 9.973913919934147e-05,
+      "loss": 2.6468,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1389029026031494,
+      "learning_rate": 9.973704997390079e-05,
+      "loss": 2.6284,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2522790431976318,
+      "learning_rate": 9.973495243758398e-05,
+      "loss": 2.6394,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1373947858810425,
+      "learning_rate": 9.973284659074151e-05,
+      "loss": 2.6192,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1404366493225098,
+      "learning_rate": 9.973073243372528e-05,
+      "loss": 2.7104,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1740000247955322,
+      "learning_rate": 9.972860996688853e-05,
+      "loss": 2.5994,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.206592082977295,
+      "learning_rate": 9.972647919058595e-05,
+      "loss": 2.6076,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1794425249099731,
+      "learning_rate": 9.972434010517358e-05,
+      "loss": 2.5993,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.219059705734253,
+      "learning_rate": 9.972219271100882e-05,
+      "loss": 2.547,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2484699487686157,
+      "learning_rate": 9.972003700845052e-05,
+      "loss": 2.6569,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2481240034103394,
+      "learning_rate": 9.971787299785888e-05,
+      "loss": 2.6824,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1482815742492676,
+      "learning_rate": 9.97157006795955e-05,
+      "loss": 2.659,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1680371761322021,
+      "learning_rate": 9.971352005402338e-05,
+      "loss": 2.6117,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2397286891937256,
+      "learning_rate": 9.971133112150688e-05,
+      "loss": 2.7112,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1797263622283936,
+      "learning_rate": 9.970913388241177e-05,
+      "loss": 2.6673,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.109689474105835,
+      "learning_rate": 9.970692833710517e-05,
+      "loss": 2.5973,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1781117916107178,
+      "learning_rate": 9.970471448595566e-05,
+      "loss": 2.5703,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1917835474014282,
+      "learning_rate": 9.970249232933315e-05,
+      "loss": 2.5825,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.191479206085205,
+      "learning_rate": 9.970026186760896e-05,
+      "loss": 2.6062,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2497385740280151,
+      "learning_rate": 9.969802310115577e-05,
+      "loss": 2.6155,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1715008020401,
+      "learning_rate": 9.96957760303477e-05,
+      "loss": 2.6717,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1210248470306396,
+      "learning_rate": 9.969352065556022e-05,
+      "loss": 2.685,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1478345394134521,
+      "learning_rate": 9.969125697717017e-05,
+      "loss": 2.698,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1940054893493652,
+      "learning_rate": 9.968898499555582e-05,
+      "loss": 2.6145,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1371153593063354,
+      "learning_rate": 9.968670471109682e-05,
+      "loss": 2.5919,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2233850955963135,
+      "learning_rate": 9.968441612417418e-05,
+      "loss": 2.6782,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.101838231086731,
+      "learning_rate": 9.968211923517032e-05,
+      "loss": 2.6082,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1706115007400513,
+      "learning_rate": 9.967981404446905e-05,
+      "loss": 2.6303,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0932798385620117,
+      "learning_rate": 9.967750055245555e-05,
+      "loss": 2.6398,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.125888705253601,
+      "learning_rate": 9.967517875951638e-05,
+      "loss": 2.6324,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0839251279830933,
+      "learning_rate": 9.967284866603955e-05,
+      "loss": 2.5443,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.099361538887024,
+      "learning_rate": 9.967051027241436e-05,
+      "loss": 2.7641,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2205827236175537,
+      "learning_rate": 9.966816357903155e-05,
+      "loss": 2.7047,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3490623235702515,
+      "learning_rate": 9.96658085862833e-05,
+      "loss": 2.5456,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1421146392822266,
+      "learning_rate": 9.966344529456305e-05,
+      "loss": 2.6994,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2021700143814087,
+      "learning_rate": 9.966107370426573e-05,
+      "loss": 2.6377,
+      "step": 1525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.165802001953125,
+      "learning_rate": 9.965869381578765e-05,
+      "loss": 2.6826,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1552740335464478,
+      "learning_rate": 9.965630562952643e-05,
+      "loss": 2.6222,
+      "step": 1535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.077202558517456,
+      "learning_rate": 9.965390914588115e-05,
+      "loss": 2.6252,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1232484579086304,
+      "learning_rate": 9.965150436525225e-05,
+      "loss": 2.5909,
+      "step": 1545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.157147765159607,
+      "learning_rate": 9.964909128804159e-05,
+      "loss": 2.6624,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1162078380584717,
+      "learning_rate": 9.964666991465234e-05,
+      "loss": 2.6058,
+      "step": 1555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1494001150131226,
+      "learning_rate": 9.964424024548914e-05,
+      "loss": 2.6556,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1770685911178589,
+      "learning_rate": 9.964180228095793e-05,
+      "loss": 2.6416,
+      "step": 1565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.107396125793457,
+      "learning_rate": 9.963935602146612e-05,
+      "loss": 2.6283,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1803878545761108,
+      "learning_rate": 9.963690146742248e-05,
+      "loss": 2.6393,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1271448135375977,
+      "learning_rate": 9.963443861923715e-05,
+      "loss": 2.6155,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0880908966064453,
+      "learning_rate": 9.963196747732164e-05,
+      "loss": 2.6,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1506364345550537,
+      "learning_rate": 9.96294880420889e-05,
+      "loss": 2.6165,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1723263263702393,
+      "learning_rate": 9.96270003139532e-05,
+      "loss": 2.5844,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1875972747802734,
+      "learning_rate": 9.962450429333026e-05,
+      "loss": 2.6243,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1350229978561401,
+      "learning_rate": 9.962199998063715e-05,
+      "loss": 2.5884,
+      "step": 1605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1161363124847412,
+      "learning_rate": 9.961948737629231e-05,
+      "loss": 2.5777,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1104639768600464,
+      "learning_rate": 9.961696648071561e-05,
+      "loss": 2.5815,
+      "step": 1615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1496812105178833,
+      "learning_rate": 9.961443729432828e-05,
+      "loss": 2.6008,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1391973495483398,
+      "learning_rate": 9.961189981755294e-05,
+      "loss": 2.561,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1404366493225098,
+      "learning_rate": 9.960935405081357e-05,
+      "loss": 2.6299,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1675031185150146,
+      "learning_rate": 9.960679999453559e-05,
+      "loss": 2.5978,
+      "step": 1635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1145226955413818,
+      "learning_rate": 9.960423764914575e-05,
+      "loss": 2.6079,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.149471640586853,
+      "learning_rate": 9.960166701507221e-05,
+      "loss": 2.5546,
+      "step": 1645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1248434782028198,
+      "learning_rate": 9.959908809274453e-05,
+      "loss": 2.6836,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2030333280563354,
+      "learning_rate": 9.959650088259362e-05,
+      "loss": 2.5319,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.083941102027893,
+      "learning_rate": 9.959390538505181e-05,
+      "loss": 2.645,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1261988878250122,
+      "learning_rate": 9.959130160055279e-05,
+      "loss": 2.6777,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1170322895050049,
+      "learning_rate": 9.958868952953163e-05,
+      "loss": 2.6576,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0682528018951416,
+      "learning_rate": 9.95860691724248e-05,
+      "loss": 2.5257,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.057816982269287,
+      "learning_rate": 9.958344052967018e-05,
+      "loss": 2.5689,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1180094480514526,
+      "learning_rate": 9.958080360170698e-05,
+      "loss": 2.6371,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0731226205825806,
+      "learning_rate": 9.957815838897585e-05,
+      "loss": 2.5501,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.168988585472107,
+      "learning_rate": 9.957550489191874e-05,
+      "loss": 2.7309,
+      "step": 1695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1264047622680664,
+      "learning_rate": 9.957284311097908e-05,
+      "loss": 2.5135,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1069132089614868,
+      "learning_rate": 9.957017304660164e-05,
+      "loss": 2.608,
+      "step": 1705
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.098832368850708,
+      "learning_rate": 9.956749469923258e-05,
+      "loss": 2.6549,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1627501249313354,
+      "learning_rate": 9.956480806931942e-05,
+      "loss": 2.6026,
+      "step": 1715
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1295748949050903,
+      "learning_rate": 9.956211315731112e-05,
+      "loss": 2.5852,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1540073156356812,
+      "learning_rate": 9.955940996365796e-05,
+      "loss": 2.5608,
+      "step": 1725
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0821458101272583,
+      "learning_rate": 9.955669848881166e-05,
+      "loss": 2.6053,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2134456634521484,
+      "learning_rate": 9.955397873322527e-05,
+      "loss": 2.6815,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0910789966583252,
+      "learning_rate": 9.955125069735327e-05,
+      "loss": 2.5816,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1496877670288086,
+      "learning_rate": 9.95485143816515e-05,
+      "loss": 2.5937,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1392103433609009,
+      "learning_rate": 9.95457697865772e-05,
+      "loss": 2.5294,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0626020431518555,
+      "learning_rate": 9.954301691258894e-05,
+      "loss": 2.5277,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.166907787322998,
+      "learning_rate": 9.954025576014674e-05,
+      "loss": 2.59,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2326178550720215,
+      "learning_rate": 9.953748632971201e-05,
+      "loss": 2.6237,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1073418855667114,
+      "learning_rate": 9.953470862174748e-05,
+      "loss": 2.6524,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1103116273880005,
+      "learning_rate": 9.953192263671728e-05,
+      "loss": 2.6037,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2522003650665283,
+      "learning_rate": 9.952912837508697e-05,
+      "loss": 2.5939,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1388449668884277,
+      "learning_rate": 9.952632583732343e-05,
+      "loss": 2.6656,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.131797194480896,
+      "learning_rate": 9.952351502389498e-05,
+      "loss": 2.5931,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1771594285964966,
+      "learning_rate": 9.95206959352713e-05,
+      "loss": 2.6178,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0900403261184692,
+      "learning_rate": 9.951786857192341e-05,
+      "loss": 2.5953,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.126647710800171,
+      "learning_rate": 9.951503293432378e-05,
+      "loss": 2.5609,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.715612888336182,
+      "learning_rate": 9.951218902294625e-05,
+      "loss": 2.5418,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.128297209739685,
+      "learning_rate": 9.9509336838266e-05,
+      "loss": 2.6196,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.048874855041504,
+      "learning_rate": 9.950647638075963e-05,
+      "loss": 2.5432,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0706759691238403,
+      "learning_rate": 9.950360765090511e-05,
+      "loss": 2.5859,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.179185390472412,
+      "learning_rate": 9.950073064918179e-05,
+      "loss": 2.5805,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0880879163742065,
+      "learning_rate": 9.949784537607042e-05,
+      "loss": 2.6247,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0795921087265015,
+      "learning_rate": 9.949495183205311e-05,
+      "loss": 2.5821,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0767638683319092,
+      "learning_rate": 9.949205001761334e-05,
+      "loss": 2.6091,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2812715768814087,
+      "learning_rate": 9.948913993323603e-05,
+      "loss": 2.5576,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1545400619506836,
+      "learning_rate": 9.948622157940744e-05,
+      "loss": 2.5421,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0433111190795898,
+      "learning_rate": 9.948329495661517e-05,
+      "loss": 2.6006,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0744532346725464,
+      "learning_rate": 9.94803600653483e-05,
+      "loss": 2.5866,
+      "step": 1865
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2122173309326172,
+      "learning_rate": 9.947741690609722e-05,
+      "loss": 2.5485,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0181015729904175,
+      "learning_rate": 9.94744654793537e-05,
+      "loss": 2.6409,
+      "step": 1875
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.117605447769165,
+      "learning_rate": 9.947150578561095e-05,
+      "loss": 2.6059,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1164997816085815,
+      "learning_rate": 9.946853782536351e-05,
+      "loss": 2.6042,
+      "step": 1885
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0538325309753418,
+      "learning_rate": 9.946556159910732e-05,
+      "loss": 2.594,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1567620038986206,
+      "learning_rate": 9.946257710733966e-05,
+      "loss": 2.5912,
+      "step": 1895
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0961169004440308,
+      "learning_rate": 9.945958435055927e-05,
+      "loss": 2.558,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1749523878097534,
+      "learning_rate": 9.945658332926622e-05,
+      "loss": 2.5591,
+      "step": 1905
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0897295475006104,
+      "learning_rate": 9.945357404396197e-05,
+      "loss": 2.6401,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1683534383773804,
+      "learning_rate": 9.945055649514934e-05,
+      "loss": 2.5055,
+      "step": 1915
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1230716705322266,
+      "learning_rate": 9.944753068333256e-05,
+      "loss": 2.5532,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0992531776428223,
+      "learning_rate": 9.944449660901725e-05,
+      "loss": 2.6044,
+      "step": 1925
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0720140933990479,
+      "learning_rate": 9.944145427271036e-05,
+      "loss": 2.5079,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0670244693756104,
+      "learning_rate": 9.94384036749203e-05,
+      "loss": 2.56,
+      "step": 1935
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1218425035476685,
+      "learning_rate": 9.943534481615677e-05,
+      "loss": 2.5734,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.082127571105957,
+      "learning_rate": 9.943227769693091e-05,
+      "loss": 2.6295,
+      "step": 1945
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.17219877243042,
+      "learning_rate": 9.942920231775524e-05,
+      "loss": 2.6056,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.162589192390442,
+      "learning_rate": 9.942611867914363e-05,
+      "loss": 2.6079,
+      "step": 1955
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1274863481521606,
+      "learning_rate": 9.942302678161133e-05,
+      "loss": 2.6269,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1556025743484497,
+      "learning_rate": 9.9419926625675e-05,
+      "loss": 2.5169,
+      "step": 1965
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.09199059009552,
+      "learning_rate": 9.941681821185265e-05,
+      "loss": 2.6053,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0204704999923706,
+      "learning_rate": 9.941370154066371e-05,
+      "loss": 2.5729,
+      "step": 1975
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.085049033164978,
+      "learning_rate": 9.941057661262895e-05,
+      "loss": 2.5681,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.032410740852356,
+      "learning_rate": 9.940744342827055e-05,
+      "loss": 2.5059,
+      "step": 1985
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1550447940826416,
+      "learning_rate": 9.940430198811202e-05,
+      "loss": 2.5686,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.024214506149292,
+      "learning_rate": 9.940115229267831e-05,
+      "loss": 2.5808,
+      "step": 1995
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0897711515426636,
+      "learning_rate": 9.939799434249571e-05,
+      "loss": 2.6292,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0533472299575806,
+      "learning_rate": 9.93948281380919e-05,
+      "loss": 2.5688,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1985580921173096,
+      "learning_rate": 9.939165367999597e-05,
+      "loss": 2.5858,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0592386722564697,
+      "learning_rate": 9.938847096873831e-05,
+      "loss": 2.6065,
+      "step": 2015
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0760610103607178,
+      "learning_rate": 9.938528000485078e-05,
+      "loss": 2.5727,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0417274236679077,
+      "learning_rate": 9.938208078886655e-05,
+      "loss": 2.6028,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2206109762191772,
+      "learning_rate": 9.937887332132023e-05,
+      "loss": 2.4845,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1777435541152954,
+      "learning_rate": 9.937565760274776e-05,
+      "loss": 2.5709,
+      "step": 2035
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.033681869506836,
+      "learning_rate": 9.937243363368645e-05,
+      "loss": 2.5036,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0503953695297241,
+      "learning_rate": 9.936920141467505e-05,
+      "loss": 2.4568,
+      "step": 2045
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1372333765029907,
+      "learning_rate": 9.936596094625364e-05,
+      "loss": 2.5203,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0466725826263428,
+      "learning_rate": 9.936271222896367e-05,
+      "loss": 2.5539,
+      "step": 2055
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0012574195861816,
+      "learning_rate": 9.935945526334803e-05,
+      "loss": 2.559,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0861599445343018,
+      "learning_rate": 9.935619004995089e-05,
+      "loss": 2.5607,
+      "step": 2065
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1400490999221802,
+      "learning_rate": 9.935291658931791e-05,
+      "loss": 2.5509,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.077583909034729,
+      "learning_rate": 9.934963488199604e-05,
+      "loss": 2.6342,
+      "step": 2075
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1083285808563232,
+      "learning_rate": 9.934634492853366e-05,
+      "loss": 2.5543,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0388864278793335,
+      "learning_rate": 9.93430467294805e-05,
+      "loss": 2.5475,
+      "step": 2085
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0358593463897705,
+      "learning_rate": 9.933974028538768e-05,
+      "loss": 2.511,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.082077145576477,
+      "learning_rate": 9.933642559680769e-05,
+      "loss": 2.6436,
+      "step": 2095
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0648292303085327,
+      "learning_rate": 9.93331026642944e-05,
+      "loss": 2.5245,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1852469444274902,
+      "learning_rate": 9.932977148840307e-05,
+      "loss": 2.5699,
+      "step": 2105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0260061025619507,
+      "learning_rate": 9.93264320696903e-05,
+      "loss": 2.5785,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1791861057281494,
+      "learning_rate": 9.932308440871413e-05,
+      "loss": 2.622,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.137078046798706,
+      "learning_rate": 9.931972850603394e-05,
+      "loss": 2.5737,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0763417482376099,
+      "learning_rate": 9.931636436221048e-05,
+      "loss": 2.5374,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.076704502105713,
+      "learning_rate": 9.931299197780586e-05,
+      "loss": 2.5295,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0001014471054077,
+      "learning_rate": 9.930961135338361e-05,
+      "loss": 2.5311,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0862035751342773,
+      "learning_rate": 9.930622248950864e-05,
+      "loss": 2.5832,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0613610744476318,
+      "learning_rate": 9.93028253867472e-05,
+      "loss": 2.5399,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0755136013031006,
+      "learning_rate": 9.929942004566695e-05,
+      "loss": 2.5693,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0707213878631592,
+      "learning_rate": 9.929600646683686e-05,
+      "loss": 2.5055,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1070294380187988,
+      "learning_rate": 9.929258465082738e-05,
+      "loss": 2.5259,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0437445640563965,
+      "learning_rate": 9.928915459821027e-05,
+      "loss": 2.5405,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1321793794631958,
+      "learning_rate": 9.928571630955865e-05,
+      "loss": 2.5938,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0642640590667725,
+      "learning_rate": 9.928226978544708e-05,
+      "loss": 2.5843,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.035362720489502,
+      "learning_rate": 9.927881502645146e-05,
+      "loss": 2.4824,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.045406460762024,
+      "learning_rate": 9.927535203314904e-05,
+      "loss": 2.5346,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0597273111343384,
+      "learning_rate": 9.927188080611849e-05,
+      "loss": 2.5173,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.048155426979065,
+      "learning_rate": 9.926840134593984e-05,
+      "loss": 2.5794,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1493093967437744,
+      "learning_rate": 9.92649136531945e-05,
+      "loss": 2.6028,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0265226364135742,
+      "learning_rate": 9.926141772846525e-05,
+      "loss": 2.5747,
+      "step": 2205
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.208702564239502,
+      "learning_rate": 9.925791357233623e-05,
+      "loss": 2.531,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0254908800125122,
+      "learning_rate": 9.925440118539298e-05,
+      "loss": 2.4746,
+      "step": 2215
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1062318086624146,
+      "learning_rate": 9.925088056822241e-05,
+      "loss": 2.5029,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0477129220962524,
+      "learning_rate": 9.924735172141281e-05,
+      "loss": 2.6436,
+      "step": 2225
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9932832717895508,
+      "learning_rate": 9.924381464555381e-05,
+      "loss": 2.5093,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0758368968963623,
+      "learning_rate": 9.924026934123647e-05,
+      "loss": 2.577,
+      "step": 2235
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2045130729675293,
+      "learning_rate": 9.92367158090532e-05,
+      "loss": 2.5723,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0155096054077148,
+      "learning_rate": 9.923315404959775e-05,
+      "loss": 2.4082,
+      "step": 2245
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0391439199447632,
+      "learning_rate": 9.92295840634653e-05,
+      "loss": 2.5649,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.091220736503601,
+      "learning_rate": 9.922600585125237e-05,
+      "loss": 2.5247,
+      "step": 2255
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0505359172821045,
+      "learning_rate": 9.922241941355688e-05,
+      "loss": 2.5203,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.045560598373413,
+      "learning_rate": 9.921882475097811e-05,
+      "loss": 2.5181,
+      "step": 2265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9981916546821594,
+      "learning_rate": 9.921522186411671e-05,
+      "loss": 2.5296,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0471158027648926,
+      "learning_rate": 9.92116107535747e-05,
+      "loss": 2.5203,
+      "step": 2275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.221785545349121,
+      "learning_rate": 9.920799141995549e-05,
+      "loss": 2.5248,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1808770895004272,
+      "learning_rate": 9.920436386386387e-05,
+      "loss": 2.5805,
+      "step": 2285
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0634390115737915,
+      "learning_rate": 9.920072808590596e-05,
+      "loss": 2.4238,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0285111665725708,
+      "learning_rate": 9.91970840866893e-05,
+      "loss": 2.4982,
+      "step": 2295
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0531038045883179,
+      "learning_rate": 9.919343186682282e-05,
+      "loss": 2.4398,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9787712097167969,
+      "learning_rate": 9.918977142691675e-05,
+      "loss": 2.5401,
+      "step": 2305
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9837114214897156,
+      "learning_rate": 9.918610276758273e-05,
+      "loss": 2.5029,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0753235816955566,
+      "learning_rate": 9.918242588943383e-05,
+      "loss": 2.4836,
+      "step": 2315
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9955765008926392,
+      "learning_rate": 9.91787407930844e-05,
+      "loss": 2.5107,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0494532585144043,
+      "learning_rate": 9.917504747915023e-05,
+      "loss": 2.5548,
+      "step": 2325
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1216340065002441,
+      "learning_rate": 9.917134594824843e-05,
+      "loss": 2.5433,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9878633618354797,
+      "learning_rate": 9.916763620099754e-05,
+      "loss": 2.5708,
+      "step": 2335
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.014472246170044,
+      "learning_rate": 9.916391823801743e-05,
+      "loss": 2.4668,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0498167276382446,
+      "learning_rate": 9.916019205992937e-05,
+      "loss": 2.4979,
+      "step": 2345
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1067233085632324,
+      "learning_rate": 9.915645766735597e-05,
+      "loss": 2.523,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9927934408187866,
+      "learning_rate": 9.915271506092126e-05,
+      "loss": 2.5109,
+      "step": 2355
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1743513345718384,
+      "learning_rate": 9.91489642412506e-05,
+      "loss": 2.5416,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1398380994796753,
+      "learning_rate": 9.914520520897072e-05,
+      "loss": 2.5681,
+      "step": 2365
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9992568492889404,
+      "learning_rate": 9.914143796470978e-05,
+      "loss": 2.5094,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0729209184646606,
+      "learning_rate": 9.913766250909726e-05,
+      "loss": 2.5173,
+      "step": 2375
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9504898190498352,
+      "learning_rate": 9.9133878842764e-05,
+      "loss": 2.5074,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0302964448928833,
+      "learning_rate": 9.913008696634226e-05,
+      "loss": 2.5985,
+      "step": 2385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0379302501678467,
+      "learning_rate": 9.912628688046564e-05,
+      "loss": 2.5735,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.08954918384552,
+      "learning_rate": 9.912247858576914e-05,
+      "loss": 2.5455,
+      "step": 2395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0359597206115723,
+      "learning_rate": 9.911866208288906e-05,
+      "loss": 2.4796,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0189626216888428,
+      "learning_rate": 9.911483737246319e-05,
+      "loss": 2.53,
+      "step": 2405
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0307023525238037,
+      "learning_rate": 9.911100445513058e-05,
+      "loss": 2.5547,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0170398950576782,
+      "learning_rate": 9.910716333153169e-05,
+      "loss": 2.5463,
+      "step": 2415
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0645707845687866,
+      "learning_rate": 9.910331400230841e-05,
+      "loss": 2.5964,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0609711408615112,
+      "learning_rate": 9.90994564681039e-05,
+      "loss": 2.6244,
+      "step": 2425
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0642567873001099,
+      "learning_rate": 9.909559072956275e-05,
+      "loss": 2.5779,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0323892831802368,
+      "learning_rate": 9.909171678733091e-05,
+      "loss": 2.4837,
+      "step": 2435
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1155509948730469,
+      "learning_rate": 9.908783464205573e-05,
+      "loss": 2.5502,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.963847815990448,
+      "learning_rate": 9.908394429438585e-05,
+      "loss": 2.4727,
+      "step": 2445
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0375754833221436,
+      "learning_rate": 9.908004574497139e-05,
+      "loss": 2.5011,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.971430242061615,
+      "learning_rate": 9.907613899446372e-05,
+      "loss": 2.5695,
+      "step": 2455
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0261136293411255,
+      "learning_rate": 9.90722240435157e-05,
+      "loss": 2.5326,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0324265956878662,
+      "learning_rate": 9.906830089278148e-05,
+      "loss": 2.5056,
+      "step": 2465
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.000641942024231,
+      "learning_rate": 9.906436954291661e-05,
+      "loss": 2.4449,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.076046109199524,
+      "learning_rate": 9.906042999457798e-05,
+      "loss": 2.5485,
+      "step": 2475
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0487278699874878,
+      "learning_rate": 9.905648224842391e-05,
+      "loss": 2.5664,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0127536058425903,
+      "learning_rate": 9.905252630511405e-05,
+      "loss": 2.4641,
+      "step": 2485
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.059358835220337,
+      "learning_rate": 9.904856216530938e-05,
+      "loss": 2.5029,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9418269395828247,
+      "learning_rate": 9.904458982967235e-05,
+      "loss": 2.482,
+      "step": 2495
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.052708387374878,
+      "learning_rate": 9.904060929886668e-05,
+      "loss": 2.4913,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9937246441841125,
+      "learning_rate": 9.903662057355751e-05,
+      "loss": 2.529,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9859743118286133,
+      "learning_rate": 9.903262365441137e-05,
+      "loss": 2.3473,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9854574203491211,
+      "learning_rate": 9.90286185420961e-05,
+      "loss": 2.4444,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0326589345932007,
+      "learning_rate": 9.902460523728094e-05,
+      "loss": 2.5487,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.10690438747406,
+      "learning_rate": 9.902058374063651e-05,
+      "loss": 2.4787,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1171677112579346,
+      "learning_rate": 9.90165540528348e-05,
+      "loss": 2.4576,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0410051345825195,
+      "learning_rate": 9.901251617454913e-05,
+      "loss": 2.4366,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.087079405784607,
+      "learning_rate": 9.900847010645422e-05,
+      "loss": 2.5504,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0660760402679443,
+      "learning_rate": 9.900441584922616e-05,
+      "loss": 2.5075,
+      "step": 2545
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.050161600112915,
+      "learning_rate": 9.90003534035424e-05,
+      "loss": 2.4918,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0482395887374878,
+      "learning_rate": 9.899628277008175e-05,
+      "loss": 2.4436,
+      "step": 2555
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0029594898223877,
+      "learning_rate": 9.899220394952441e-05,
+      "loss": 2.5277,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0098901987075806,
+      "learning_rate": 9.898811694255195e-05,
+      "loss": 2.4827,
+      "step": 2565
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1033732891082764,
+      "learning_rate": 9.898402174984726e-05,
+      "loss": 2.5772,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9989816546440125,
+      "learning_rate": 9.897991837209464e-05,
+      "loss": 2.517,
+      "step": 2575
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1884549856185913,
+      "learning_rate": 9.897580680997977e-05,
+      "loss": 2.5246,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9615799188613892,
+      "learning_rate": 9.897168706418965e-05,
+      "loss": 2.4685,
+      "step": 2585
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0535063743591309,
+      "learning_rate": 9.89675591354127e-05,
+      "loss": 2.4825,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0240947008132935,
+      "learning_rate": 9.896342302433866e-05,
+      "loss": 2.5427,
+      "step": 2595
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0623677968978882,
+      "learning_rate": 9.895927873165867e-05,
+      "loss": 2.4895,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0304031372070312,
+      "learning_rate": 9.895512625806522e-05,
+      "loss": 2.5037,
+      "step": 2605
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0591580867767334,
+      "learning_rate": 9.895096560425218e-05,
+      "loss": 2.5483,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.059604525566101,
+      "learning_rate": 9.894679677091477e-05,
+      "loss": 2.57,
+      "step": 2615
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0299981832504272,
+      "learning_rate": 9.89426197587496e-05,
+      "loss": 2.4304,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0912493467330933,
+      "learning_rate": 9.893843456845462e-05,
+      "loss": 2.5913,
+      "step": 2625
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9758343696594238,
+      "learning_rate": 9.893424120072916e-05,
+      "loss": 2.5686,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9720059037208557,
+      "learning_rate": 9.893003965627394e-05,
+      "loss": 2.5197,
+      "step": 2635
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0174733400344849,
+      "learning_rate": 9.8925829935791e-05,
+      "loss": 2.4461,
+      "step": 2640
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0520356893539429,
+      "learning_rate": 9.892161203998375e-05,
+      "loss": 2.5178,
+      "step": 2645
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9675366282463074,
+      "learning_rate": 9.891738596955703e-05,
+      "loss": 2.4992,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.003288745880127,
+      "learning_rate": 9.891315172521696e-05,
+      "loss": 2.5163,
+      "step": 2655
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9630966782569885,
+      "learning_rate": 9.890890930767111e-05,
+      "loss": 2.5159,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9560485482215881,
+      "learning_rate": 9.890465871762832e-05,
+      "loss": 2.4631,
+      "step": 2665
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.061198353767395,
+      "learning_rate": 9.89003999557989e-05,
+      "loss": 2.4701,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0241599082946777,
+      "learning_rate": 9.889613302289443e-05,
+      "loss": 2.4786,
+      "step": 2675
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9885386824607849,
+      "learning_rate": 9.889185791962792e-05,
+      "loss": 2.5212,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9871917963027954,
+      "learning_rate": 9.888757464671372e-05,
+      "loss": 2.5568,
+      "step": 2685
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.950515866279602,
+      "learning_rate": 9.888328320486755e-05,
+      "loss": 2.4996,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.945688784122467,
+      "learning_rate": 9.88789835948065e-05,
+      "loss": 2.4747,
+      "step": 2695
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0324536561965942,
+      "learning_rate": 9.887467581724901e-05,
+      "loss": 2.5204,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9816248416900635,
+      "learning_rate": 9.887035987291488e-05,
+      "loss": 2.4452,
+      "step": 2705
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0598872900009155,
+      "learning_rate": 9.88660357625253e-05,
+      "loss": 2.5088,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0420717000961304,
+      "learning_rate": 9.886170348680284e-05,
+      "loss": 2.5659,
+      "step": 2715
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0138999223709106,
+      "learning_rate": 9.885736304647139e-05,
+      "loss": 2.4567,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9980942010879517,
+      "learning_rate": 9.885301444225619e-05,
+      "loss": 2.5009,
+      "step": 2725
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0251879692077637,
+      "learning_rate": 9.884865767488391e-05,
+      "loss": 2.543,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0415172576904297,
+      "learning_rate": 9.884429274508256e-05,
+      "loss": 2.5281,
+      "step": 2735
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9737923741340637,
+      "learning_rate": 9.883991965358146e-05,
+      "loss": 2.5367,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9333593845367432,
+      "learning_rate": 9.883553840111136e-05,
+      "loss": 2.4772,
+      "step": 2745
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0520544052124023,
+      "learning_rate": 9.883114898840437e-05,
+      "loss": 2.5889,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.020678162574768,
+      "learning_rate": 9.882675141619391e-05,
+      "loss": 2.4955,
+      "step": 2755
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9737017154693604,
+      "learning_rate": 9.882234568521482e-05,
+      "loss": 2.5924,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9897264242172241,
+      "learning_rate": 9.881793179620328e-05,
+      "loss": 2.4564,
+      "step": 2765
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0046138763427734,
+      "learning_rate": 9.881350974989682e-05,
+      "loss": 2.626,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.064558744430542,
+      "learning_rate": 9.880907954703437e-05,
+      "loss": 2.5432,
+      "step": 2775
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9963716864585876,
+      "learning_rate": 9.880464118835619e-05,
+      "loss": 2.4694,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9534794688224792,
+      "learning_rate": 9.880019467460391e-05,
+      "loss": 2.515,
+      "step": 2785
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0217405557632446,
+      "learning_rate": 9.879574000652052e-05,
+      "loss": 2.5118,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0339412689208984,
+      "learning_rate": 9.87912771848504e-05,
+      "loss": 2.5101,
+      "step": 2795
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.055714726448059,
+      "learning_rate": 9.878680621033925e-05,
+      "loss": 2.4237,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9570962190628052,
+      "learning_rate": 9.878232708373415e-05,
+      "loss": 2.4589,
+      "step": 2805
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9586167335510254,
+      "learning_rate": 9.877783980578357e-05,
+      "loss": 2.513,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0818899869918823,
+      "learning_rate": 9.877334437723728e-05,
+      "loss": 2.588,
+      "step": 2815
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.970255970954895,
+      "learning_rate": 9.87688407988465e-05,
+      "loss": 2.4591,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0507453680038452,
+      "learning_rate": 9.876432907136372e-05,
+      "loss": 2.4895,
+      "step": 2825
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0207104682922363,
+      "learning_rate": 9.875980919554287e-05,
+      "loss": 2.4132,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9952415227890015,
+      "learning_rate": 9.875528117213915e-05,
+      "loss": 2.4723,
+      "step": 2835
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0037434101104736,
+      "learning_rate": 9.875074500190922e-05,
+      "loss": 2.4826,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9465364813804626,
+      "learning_rate": 9.874620068561104e-05,
+      "loss": 2.4688,
+      "step": 2845
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0243892669677734,
+      "learning_rate": 9.874164822400396e-05,
+      "loss": 2.4506,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0115193128585815,
+      "learning_rate": 9.873708761784866e-05,
+      "loss": 2.5321,
+      "step": 2855
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0065405368804932,
+      "learning_rate": 9.873251886790722e-05,
+      "loss": 2.4068,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9861264228820801,
+      "learning_rate": 9.872794197494307e-05,
+      "loss": 2.5082,
+      "step": 2865
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.051580548286438,
+      "learning_rate": 9.872335693972094e-05,
+      "loss": 2.5909,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0542223453521729,
+      "learning_rate": 9.871876376300702e-05,
+      "loss": 2.5026,
+      "step": 2875
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1372771263122559,
+      "learning_rate": 9.871416244556881e-05,
+      "loss": 2.4612,
+      "step": 2880
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.022678256034851,
+      "learning_rate": 9.870955298817517e-05,
+      "loss": 2.5265,
+      "step": 2885
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9577171802520752,
+      "learning_rate": 9.87049353915963e-05,
+      "loss": 2.5249,
+      "step": 2890
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9479445815086365,
+      "learning_rate": 9.87003096566038e-05,
+      "loss": 2.5136,
+      "step": 2895
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9847010374069214,
+      "learning_rate": 9.869567578397062e-05,
+      "loss": 2.4214,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9892215728759766,
+      "learning_rate": 9.869103377447105e-05,
+      "loss": 2.556,
+      "step": 2905
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.006055474281311,
+      "learning_rate": 9.868638362888076e-05,
+      "loss": 2.4785,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.975849986076355,
+      "learning_rate": 9.868172534797678e-05,
+      "loss": 2.5206,
+      "step": 2915
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9815483689308167,
+      "learning_rate": 9.867705893253747e-05,
+      "loss": 2.4531,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9992381930351257,
+      "learning_rate": 9.867238438334259e-05,
+      "loss": 2.4902,
+      "step": 2925
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0498075485229492,
+      "learning_rate": 9.866770170117324e-05,
+      "loss": 2.4622,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9191021919250488,
+      "learning_rate": 9.866301088681185e-05,
+      "loss": 2.3964,
+      "step": 2935
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9569787383079529,
+      "learning_rate": 9.865831194104227e-05,
+      "loss": 2.4257,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9479259252548218,
+      "learning_rate": 9.865360486464967e-05,
+      "loss": 2.4037,
+      "step": 2945
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0182474851608276,
+      "learning_rate": 9.864888965842056e-05,
+      "loss": 2.4884,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9924219846725464,
+      "learning_rate": 9.864416632314287e-05,
+      "loss": 2.4639,
+      "step": 2955
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9995209574699402,
+      "learning_rate": 9.863943485960583e-05,
+      "loss": 2.5196,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9799997210502625,
+      "learning_rate": 9.863469526860005e-05,
+      "loss": 2.4846,
+      "step": 2965
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9801844358444214,
+      "learning_rate": 9.862994755091752e-05,
+      "loss": 2.5074,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.01804518699646,
+      "learning_rate": 9.862519170735154e-05,
+      "loss": 2.4073,
+      "step": 2975
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.982128918170929,
+      "learning_rate": 9.86204277386968e-05,
+      "loss": 2.5648,
+      "step": 2980
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0903494358062744,
+      "learning_rate": 9.861565564574935e-05,
+      "loss": 2.4034,
+      "step": 2985
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9478833675384521,
+      "learning_rate": 9.861087542930658e-05,
+      "loss": 2.48,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0135399103164673,
+      "learning_rate": 9.860608709016726e-05,
+      "loss": 2.465,
+      "step": 2995
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0020098686218262,
+      "learning_rate": 9.860129062913147e-05,
+      "loss": 2.4947,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9754078388214111,
+      "learning_rate": 9.859648604700072e-05,
+      "loss": 2.4997,
+      "step": 3005
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9789213538169861,
+      "learning_rate": 9.859167334457781e-05,
+      "loss": 2.4775,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9456048011779785,
+      "learning_rate": 9.858685252266696e-05,
+      "loss": 2.5746,
+      "step": 3015
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9769589900970459,
+      "learning_rate": 9.858202358207367e-05,
+      "loss": 2.4112,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0837334394454956,
+      "learning_rate": 9.857718652360486e-05,
+      "loss": 2.5154,
+      "step": 3025
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0221744775772095,
+      "learning_rate": 9.857234134806878e-05,
+      "loss": 2.4581,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0674024820327759,
+      "learning_rate": 9.856748805627503e-05,
+      "loss": 2.3895,
+      "step": 3035
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0216269493103027,
+      "learning_rate": 9.856262664903462e-05,
+      "loss": 2.4782,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0229367017745972,
+      "learning_rate": 9.855775712715981e-05,
+      "loss": 2.4997,
+      "step": 3045
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.058603286743164,
+      "learning_rate": 9.855287949146432e-05,
+      "loss": 2.4946,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9665271639823914,
+      "learning_rate": 9.854799374276318e-05,
+      "loss": 2.4733,
+      "step": 3055
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9766295552253723,
+      "learning_rate": 9.854309988187277e-05,
+      "loss": 2.5086,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9893321394920349,
+      "learning_rate": 9.853819790961086e-05,
+      "loss": 2.4585,
+      "step": 3065
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.995509922504425,
+      "learning_rate": 9.853328782679652e-05,
+      "loss": 2.5517,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9660744667053223,
+      "learning_rate": 9.852836963425021e-05,
+      "loss": 2.4862,
+      "step": 3075
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9365461468696594,
+      "learning_rate": 9.852344333279375e-05,
+      "loss": 2.4671,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9948537945747375,
+      "learning_rate": 9.851850892325032e-05,
+      "loss": 2.5716,
+      "step": 3085
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9941670298576355,
+      "learning_rate": 9.851356640644443e-05,
+      "loss": 2.3602,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9700287580490112,
+      "learning_rate": 9.850861578320196e-05,
+      "loss": 2.4559,
+      "step": 3095
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9778099656105042,
+      "learning_rate": 9.850365705435013e-05,
+      "loss": 2.4516,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0385972261428833,
+      "learning_rate": 9.849869022071753e-05,
+      "loss": 2.4677,
+      "step": 3105
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.038212537765503,
+      "learning_rate": 9.849371528313411e-05,
+      "loss": 2.586,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0120850801467896,
+      "learning_rate": 9.848873224243114e-05,
+      "loss": 2.4363,
+      "step": 3115
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9313243627548218,
+      "learning_rate": 9.848374109944128e-05,
+      "loss": 2.4811,
+      "step": 3120
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9424246549606323,
+      "learning_rate": 9.847874185499855e-05,
+      "loss": 2.4652,
+      "step": 3125
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0624967813491821,
+      "learning_rate": 9.847373450993828e-05,
+      "loss": 2.5134,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0194240808486938,
+      "learning_rate": 9.846871906509719e-05,
+      "loss": 2.4193,
+      "step": 3135
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9781209826469421,
+      "learning_rate": 9.846369552131334e-05,
+      "loss": 2.5048,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9597983956336975,
+      "learning_rate": 9.845866387942614e-05,
+      "loss": 2.4352,
+      "step": 3145
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9718809723854065,
+      "learning_rate": 9.845362414027637e-05,
+      "loss": 2.4466,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0027637481689453,
+      "learning_rate": 9.844857630470614e-05,
+      "loss": 2.5808,
+      "step": 3155
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9830366373062134,
+      "learning_rate": 9.844352037355893e-05,
+      "loss": 2.4467,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.020050048828125,
+      "learning_rate": 9.843845634767957e-05,
+      "loss": 2.471,
+      "step": 3165
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9836028814315796,
+      "learning_rate": 9.843338422791424e-05,
+      "loss": 2.5503,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0021939277648926,
+      "learning_rate": 9.842830401511048e-05,
+      "loss": 2.4773,
+      "step": 3175
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.023052453994751,
+      "learning_rate": 9.842321571011714e-05,
+      "loss": 2.3922,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.957406222820282,
+      "learning_rate": 9.841811931378448e-05,
+      "loss": 2.4769,
+      "step": 3185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9420168399810791,
+      "learning_rate": 9.841301482696413e-05,
+      "loss": 2.4828,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9661071300506592,
+      "learning_rate": 9.840790225050896e-05,
+      "loss": 2.3626,
+      "step": 3195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9690583944320679,
+      "learning_rate": 9.840278158527328e-05,
+      "loss": 2.4542,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.907796323299408,
+      "learning_rate": 9.839765283211278e-05,
+      "loss": 2.4928,
+      "step": 3205
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9589967131614685,
+      "learning_rate": 9.83925159918844e-05,
+      "loss": 2.4723,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9484946131706238,
+      "learning_rate": 9.838737106544651e-05,
+      "loss": 2.4992,
+      "step": 3215
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0401115417480469,
+      "learning_rate": 9.838221805365881e-05,
+      "loss": 2.4724,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9925845265388489,
+      "learning_rate": 9.837705695738236e-05,
+      "loss": 2.3688,
+      "step": 3225
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9755850434303284,
+      "learning_rate": 9.837188777747954e-05,
+      "loss": 2.4867,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9387975335121155,
+      "learning_rate": 9.836671051481411e-05,
+      "loss": 2.461,
+      "step": 3235
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9345282912254333,
+      "learning_rate": 9.836152517025116e-05,
+      "loss": 2.5068,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0869733095169067,
+      "learning_rate": 9.835633174465715e-05,
+      "loss": 2.5028,
+      "step": 3245
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0614070892333984,
+      "learning_rate": 9.83511302388999e-05,
+      "loss": 2.3522,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9533520936965942,
+      "learning_rate": 9.834592065384853e-05,
+      "loss": 2.4049,
+      "step": 3255
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0925136804580688,
+      "learning_rate": 9.834070299037357e-05,
+      "loss": 2.4904,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0727174282073975,
+      "learning_rate": 9.833547724934687e-05,
+      "loss": 2.425,
+      "step": 3265
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9756423830986023,
+      "learning_rate": 9.833024343164159e-05,
+      "loss": 2.4972,
+      "step": 3270
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9418448805809021,
+      "learning_rate": 9.832500153813234e-05,
+      "loss": 2.4574,
+      "step": 3275
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9274198412895203,
+      "learning_rate": 9.831975156969498e-05,
+      "loss": 2.4227,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9617379903793335,
+      "learning_rate": 9.831449352720677e-05,
+      "loss": 2.4033,
+      "step": 3285
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9729361534118652,
+      "learning_rate": 9.830922741154631e-05,
+      "loss": 2.4718,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.011651873588562,
+      "learning_rate": 9.830395322359357e-05,
+      "loss": 2.5037,
+      "step": 3295
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1149110794067383,
+      "learning_rate": 9.829867096422983e-05,
+      "loss": 2.5295,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1510682106018066,
+      "learning_rate": 9.829338063433773e-05,
+      "loss": 2.4923,
+      "step": 3305
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0033206939697266,
+      "learning_rate": 9.828808223480126e-05,
+      "loss": 2.478,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9701094627380371,
+      "learning_rate": 9.828277576650578e-05,
+      "loss": 2.3896,
+      "step": 3315
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9452363848686218,
+      "learning_rate": 9.827746123033796e-05,
+      "loss": 2.4513,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9493092894554138,
+      "learning_rate": 9.827213862718587e-05,
+      "loss": 2.5528,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9727687239646912,
+      "learning_rate": 9.826680795793887e-05,
+      "loss": 2.4901,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.007023811340332,
+      "learning_rate": 9.826146922348773e-05,
+      "loss": 2.5563,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9239678382873535,
+      "learning_rate": 9.825612242472448e-05,
+      "loss": 2.4295,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.042544960975647,
+      "learning_rate": 9.82507675625426e-05,
+      "loss": 2.4131,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0116506814956665,
+      "learning_rate": 9.824540463783682e-05,
+      "loss": 2.516,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9312745332717896,
+      "learning_rate": 9.824003365150329e-05,
+      "loss": 2.4893,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9113055467605591,
+      "learning_rate": 9.82346546044395e-05,
+      "loss": 2.4056,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.961380124092102,
+      "learning_rate": 9.822926749754424e-05,
+      "loss": 2.5197,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9460198283195496,
+      "learning_rate": 9.822387233171771e-05,
+      "loss": 2.4089,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.909464955329895,
+      "learning_rate": 9.821846910786136e-05,
+      "loss": 2.4964,
+      "step": 3375
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9589513540267944,
+      "learning_rate": 9.821305782687812e-05,
+      "loss": 2.3827,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9687663912773132,
+      "learning_rate": 9.820763848967215e-05,
+      "loss": 2.5384,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9451897144317627,
+      "learning_rate": 9.820221109714902e-05,
+      "loss": 2.3377,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9563037753105164,
+      "learning_rate": 9.81967756502156e-05,
+      "loss": 2.515,
+      "step": 3395
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8919141292572021,
+      "learning_rate": 9.819133214978017e-05,
+      "loss": 2.43,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9723004698753357,
+      "learning_rate": 9.818588059675229e-05,
+      "loss": 2.5428,
+      "step": 3405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9102802276611328,
+      "learning_rate": 9.818042099204292e-05,
+      "loss": 2.4936,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.96049964427948,
+      "learning_rate": 9.817495333656433e-05,
+      "loss": 2.4819,
+      "step": 3415
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.958992600440979,
+      "learning_rate": 9.816947763123013e-05,
+      "loss": 2.4658,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0817489624023438,
+      "learning_rate": 9.816399387695529e-05,
+      "loss": 2.459,
+      "step": 3425
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.945145845413208,
+      "learning_rate": 9.815850207465616e-05,
+      "loss": 2.4695,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9507633447647095,
+      "learning_rate": 9.815300222525035e-05,
+      "loss": 2.4209,
+      "step": 3435
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0667479038238525,
+      "learning_rate": 9.814749432965691e-05,
+      "loss": 2.4587,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9662715792655945,
+      "learning_rate": 9.814197838879615e-05,
+      "loss": 2.4872,
+      "step": 3445
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0137649774551392,
+      "learning_rate": 9.813645440358981e-05,
+      "loss": 2.495,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9914734959602356,
+      "learning_rate": 9.813092237496087e-05,
+      "loss": 2.4763,
+      "step": 3455
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.963979959487915,
+      "learning_rate": 9.812538230383376e-05,
+      "loss": 2.4035,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.956588864326477,
+      "learning_rate": 9.811983419113418e-05,
+      "loss": 2.4477,
+      "step": 3465
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9696434140205383,
+      "learning_rate": 9.81142780377892e-05,
+      "loss": 2.4391,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9527379274368286,
+      "learning_rate": 9.810871384472724e-05,
+      "loss": 2.4328,
+      "step": 3475
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9704457521438599,
+      "learning_rate": 9.810314161287806e-05,
+      "loss": 2.4171,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9139119386672974,
+      "learning_rate": 9.809756134317276e-05,
+      "loss": 2.3619,
+      "step": 3485
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9535418152809143,
+      "learning_rate": 9.809197303654376e-05,
+      "loss": 2.4865,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9711298942565918,
+      "learning_rate": 9.808637669392487e-05,
+      "loss": 2.4289,
+      "step": 3495
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9339505434036255,
+      "learning_rate": 9.808077231625119e-05,
+      "loss": 2.4111,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9773271083831787,
+      "learning_rate": 9.807515990445922e-05,
+      "loss": 2.4486,
+      "step": 3505
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0019117593765259,
+      "learning_rate": 9.806953945948676e-05,
+      "loss": 2.4267,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9486785531044006,
+      "learning_rate": 9.806391098227297e-05,
+      "loss": 2.4769,
+      "step": 3515
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9409202933311462,
+      "learning_rate": 9.805827447375835e-05,
+      "loss": 2.4697,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.928676187992096,
+      "learning_rate": 9.805262993488473e-05,
+      "loss": 2.4648,
+      "step": 3525
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.949226975440979,
+      "learning_rate": 9.804697736659532e-05,
+      "loss": 2.4599,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9870195388793945,
+      "learning_rate": 9.804131676983458e-05,
+      "loss": 2.3966,
+      "step": 3535
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.024148941040039,
+      "learning_rate": 9.803564814554844e-05,
+      "loss": 2.4154,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9488735795021057,
+      "learning_rate": 9.802997149468408e-05,
+      "loss": 2.4451,
+      "step": 3545
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9772784113883972,
+      "learning_rate": 9.802428681819003e-05,
+      "loss": 2.3846,
+      "step": 3550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0094990730285645,
+      "learning_rate": 9.801859411701621e-05,
+      "loss": 2.4446,
+      "step": 3555
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0590399503707886,
+      "learning_rate": 9.801289339211384e-05,
+      "loss": 2.5499,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 9.800718464443547e-05,
+      "loss": 2.5048,
+      "step": 3565
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0001592636108398,
+      "learning_rate": 9.800146787493503e-05,
+      "loss": 2.424,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0542467832565308,
+      "learning_rate": 9.799574308456775e-05,
+      "loss": 2.4615,
+      "step": 3575
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9968492388725281,
+      "learning_rate": 9.799001027429026e-05,
+      "loss": 2.4325,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9719383120536804,
+      "learning_rate": 9.798426944506048e-05,
+      "loss": 2.5351,
+      "step": 3585
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9848989844322205,
+      "learning_rate": 9.797852059783766e-05,
+      "loss": 2.4584,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0010429620742798,
+      "learning_rate": 9.79727637335824e-05,
+      "loss": 2.5851,
+      "step": 3595
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.021357536315918,
+      "learning_rate": 9.79669988532567e-05,
+      "loss": 2.4036,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9273085594177246,
+      "learning_rate": 9.79612259578238e-05,
+      "loss": 2.4378,
+      "step": 3605
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0058780908584595,
+      "learning_rate": 9.795544504824836e-05,
+      "loss": 2.4723,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.915285050868988,
+      "learning_rate": 9.794965612549635e-05,
+      "loss": 2.4909,
+      "step": 3615
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9782754778862,
+      "learning_rate": 9.794385919053506e-05,
+      "loss": 2.3628,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9310125708580017,
+      "learning_rate": 9.793805424433313e-05,
+      "loss": 2.4546,
+      "step": 3625
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9507859945297241,
+      "learning_rate": 9.793224128786058e-05,
+      "loss": 2.4996,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0298519134521484,
+      "learning_rate": 9.79264203220887e-05,
+      "loss": 2.3714,
+      "step": 3635
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1216673851013184,
+      "learning_rate": 9.792059134799016e-05,
+      "loss": 2.4238,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9437344670295715,
+      "learning_rate": 9.791475436653896e-05,
+      "loss": 2.4045,
+      "step": 3645
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9950194954872131,
+      "learning_rate": 9.790890937871046e-05,
+      "loss": 2.4571,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9664518237113953,
+      "learning_rate": 9.79030563854813e-05,
+      "loss": 2.4943,
+      "step": 3655
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9285597205162048,
+      "learning_rate": 9.789719538782952e-05,
+      "loss": 2.4504,
+      "step": 3660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9740208387374878,
+      "learning_rate": 9.789132638673445e-05,
+      "loss": 2.4484,
+      "step": 3665
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 168.87538146972656,
+      "learning_rate": 9.78854493831768e-05,
+      "loss": 2.4598,
+      "step": 3670
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9382103681564331,
+      "learning_rate": 9.787956437813859e-05,
+      "loss": 2.495,
+      "step": 3675
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9842873215675354,
+      "learning_rate": 9.787367137260317e-05,
+      "loss": 2.4197,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.952576220035553,
+      "learning_rate": 9.786777036755525e-05,
+      "loss": 2.4391,
+      "step": 3685
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9747008085250854,
+      "learning_rate": 9.786186136398085e-05,
+      "loss": 2.4659,
+      "step": 3690
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.972825288772583,
+      "learning_rate": 9.785594436286738e-05,
+      "loss": 2.4165,
+      "step": 3695
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0265997648239136,
+      "learning_rate": 9.78500193652035e-05,
+      "loss": 2.4413,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9106327891349792,
+      "learning_rate": 9.784408637197929e-05,
+      "loss": 2.4337,
+      "step": 3705
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9530829191207886,
+      "learning_rate": 9.783814538418612e-05,
+      "loss": 2.5054,
+      "step": 3710
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9083680510520935,
+      "learning_rate": 9.783219640281671e-05,
+      "loss": 2.4544,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9191516637802124,
+      "learning_rate": 9.78262394288651e-05,
+      "loss": 2.4776,
+      "step": 3720
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9617584943771362,
+      "learning_rate": 9.782027446332671e-05,
+      "loss": 2.3777,
+      "step": 3725
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.000213623046875,
+      "learning_rate": 9.781430150719822e-05,
+      "loss": 2.5021,
+      "step": 3730
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9784063696861267,
+      "learning_rate": 9.780832056147772e-05,
+      "loss": 2.422,
+      "step": 3735
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9529116749763489,
+      "learning_rate": 9.78023316271646e-05,
+      "loss": 2.4064,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.01068115234375,
+      "learning_rate": 9.779633470525957e-05,
+      "loss": 2.4303,
+      "step": 3745
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0238860845565796,
+      "learning_rate": 9.77903297967647e-05,
+      "loss": 2.4397,
+      "step": 3750
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9882420301437378,
+      "learning_rate": 9.778431690268341e-05,
+      "loss": 2.3933,
+      "step": 3755
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9538924098014832,
+      "learning_rate": 9.777829602402043e-05,
+      "loss": 2.3385,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9392457008361816,
+      "learning_rate": 9.777226716178178e-05,
+      "loss": 2.4382,
+      "step": 3765
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9855114221572876,
+      "learning_rate": 9.776623031697492e-05,
+      "loss": 2.4051,
+      "step": 3770
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9578977823257446,
+      "learning_rate": 9.776018549060855e-05,
+      "loss": 2.4168,
+      "step": 3775
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9765996336936951,
+      "learning_rate": 9.775413268369274e-05,
+      "loss": 2.4126,
+      "step": 3780
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9845826625823975,
+      "learning_rate": 9.77480718972389e-05,
+      "loss": 2.434,
+      "step": 3785
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9247323274612427,
+      "learning_rate": 9.774200313225976e-05,
+      "loss": 2.4746,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9562143683433533,
+      "learning_rate": 9.773592638976939e-05,
+      "loss": 2.4637,
+      "step": 3795
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9440853595733643,
+      "learning_rate": 9.77298416707832e-05,
+      "loss": 2.4316,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9409793615341187,
+      "learning_rate": 9.772374897631791e-05,
+      "loss": 2.4483,
+      "step": 3805
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9646011590957642,
+      "learning_rate": 9.771764830739159e-05,
+      "loss": 2.4489,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8940223455429077,
+      "learning_rate": 9.771153966502365e-05,
+      "loss": 2.4302,
+      "step": 3815
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9929117560386658,
+      "learning_rate": 9.77054230502348e-05,
+      "loss": 2.4735,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.000969409942627,
+      "learning_rate": 9.769929846404711e-05,
+      "loss": 2.3971,
+      "step": 3825
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9429438710212708,
+      "learning_rate": 9.769316590748401e-05,
+      "loss": 2.4022,
+      "step": 3830
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9274194836616516,
+      "learning_rate": 9.768702538157018e-05,
+      "loss": 2.4446,
+      "step": 3835
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9822215437889099,
+      "learning_rate": 9.76808768873317e-05,
+      "loss": 2.4341,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8823676705360413,
+      "learning_rate": 9.767472042579595e-05,
+      "loss": 2.372,
+      "step": 3845
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9482948184013367,
+      "learning_rate": 9.766855599799168e-05,
+      "loss": 2.4077,
+      "step": 3850
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9701356291770935,
+      "learning_rate": 9.76623836049489e-05,
+      "loss": 2.4383,
+      "step": 3855
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.919223964214325,
+      "learning_rate": 9.765620324769903e-05,
+      "loss": 2.4483,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.935723066329956,
+      "learning_rate": 9.765001492727476e-05,
+      "loss": 2.3922,
+      "step": 3865
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9847173690795898,
+      "learning_rate": 9.764381864471017e-05,
+      "loss": 2.4067,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6558985710144043,
+      "learning_rate": 9.76376144010406e-05,
+      "loss": 2.4297,
+      "step": 3875
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9964712858200073,
+      "learning_rate": 9.763140219730278e-05,
+      "loss": 2.3822,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9748810529708862,
+      "learning_rate": 9.762518203453472e-05,
+      "loss": 2.4836,
+      "step": 3885
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9918500781059265,
+      "learning_rate": 9.761895391377581e-05,
+      "loss": 2.3835,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9291976690292358,
+      "learning_rate": 9.761271783606674e-05,
+      "loss": 2.3923,
+      "step": 3895
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9225846529006958,
+      "learning_rate": 9.760647380244952e-05,
+      "loss": 2.3407,
+      "step": 3900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9817166328430176,
+      "learning_rate": 9.760022181396753e-05,
+      "loss": 2.4411,
+      "step": 3905
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9370828866958618,
+      "learning_rate": 9.759396187166543e-05,
+      "loss": 2.4813,
+      "step": 3910
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9917733073234558,
+      "learning_rate": 9.758769397658926e-05,
+      "loss": 2.4315,
+      "step": 3915
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9605938196182251,
+      "learning_rate": 9.758141812978634e-05,
+      "loss": 2.4632,
+      "step": 3920
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9316677451133728,
+      "learning_rate": 9.757513433230534e-05,
+      "loss": 2.3509,
+      "step": 3925
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9785593152046204,
+      "learning_rate": 9.756884258519627e-05,
+      "loss": 2.4633,
+      "step": 3930
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.081371784210205,
+      "learning_rate": 9.756254288951046e-05,
+      "loss": 2.4635,
+      "step": 3935
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9920836091041565,
+      "learning_rate": 9.755623524630055e-05,
+      "loss": 2.5126,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8962555527687073,
+      "learning_rate": 9.754991965662055e-05,
+      "loss": 2.4347,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9276800155639648,
+      "learning_rate": 9.754359612152574e-05,
+      "loss": 2.5463,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9453554153442383,
+      "learning_rate": 9.753726464207278e-05,
+      "loss": 2.477,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.917123019695282,
+      "learning_rate": 9.753092521931963e-05,
+      "loss": 2.4019,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9330175518989563,
+      "learning_rate": 9.752457785432558e-05,
+      "loss": 2.4103,
+      "step": 3965
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9581329226493835,
+      "learning_rate": 9.751822254815126e-05,
+      "loss": 2.3869,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.968804657459259,
+      "learning_rate": 9.751185930185861e-05,
+      "loss": 2.3455,
+      "step": 3975
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9856200218200684,
+      "learning_rate": 9.750548811651091e-05,
+      "loss": 2.4164,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0376811027526855,
+      "learning_rate": 9.749910899317276e-05,
+      "loss": 2.4661,
+      "step": 3985
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.920340359210968,
+      "learning_rate": 9.74927219329101e-05,
+      "loss": 2.4717,
+      "step": 3990
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9537367820739746,
+      "learning_rate": 9.748632693679016e-05,
+      "loss": 2.3571,
+      "step": 3995
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9520323276519775,
+      "learning_rate": 9.747992400588153e-05,
+      "loss": 2.3711,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9458045363426208,
+      "learning_rate": 9.747351314125413e-05,
+      "loss": 2.4095,
+      "step": 4005
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9521518349647522,
+      "learning_rate": 9.746709434397916e-05,
+      "loss": 2.4112,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9129638671875,
+      "learning_rate": 9.746066761512921e-05,
+      "loss": 2.4116,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9636595249176025,
+      "learning_rate": 9.745423295577815e-05,
+      "loss": 2.4009,
+      "step": 4020
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8987670540809631,
+      "learning_rate": 9.744779036700118e-05,
+      "loss": 2.384,
+      "step": 4025
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9577127695083618,
+      "learning_rate": 9.744133984987485e-05,
+      "loss": 2.3504,
+      "step": 4030
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9470349550247192,
+      "learning_rate": 9.743488140547701e-05,
+      "loss": 2.3952,
+      "step": 4035
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9211857318878174,
+      "learning_rate": 9.742841503488682e-05,
+      "loss": 2.3886,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9081634283065796,
+      "learning_rate": 9.742194073918482e-05,
+      "loss": 2.4524,
+      "step": 4045
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9131794571876526,
+      "learning_rate": 9.741545851945282e-05,
+      "loss": 2.446,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9620927572250366,
+      "learning_rate": 9.740896837677398e-05,
+      "loss": 2.4299,
+      "step": 4055
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9445740580558777,
+      "learning_rate": 9.740247031223279e-05,
+      "loss": 2.4553,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9932169914245605,
+      "learning_rate": 9.739596432691502e-05,
+      "loss": 2.3283,
+      "step": 4065
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9901602864265442,
+      "learning_rate": 9.738945042190785e-05,
+      "loss": 2.4504,
+      "step": 4070
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9628769159317017,
+      "learning_rate": 9.738292859829968e-05,
+      "loss": 2.3281,
+      "step": 4075
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9448304772377014,
+      "learning_rate": 9.73763988571803e-05,
+      "loss": 2.4356,
+      "step": 4080
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9509755969047546,
+      "learning_rate": 9.73698611996408e-05,
+      "loss": 2.413,
+      "step": 4085
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0834165811538696,
+      "learning_rate": 9.736331562677361e-05,
+      "loss": 2.4798,
+      "step": 4090
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9410833120346069,
+      "learning_rate": 9.735676213967247e-05,
+      "loss": 2.4502,
+      "step": 4095
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.015580415725708,
+      "learning_rate": 9.735020073943244e-05,
+      "loss": 2.4024,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.933193027973175,
+      "learning_rate": 9.734363142714988e-05,
+      "loss": 2.3695,
+      "step": 4105
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8989652991294861,
+      "learning_rate": 9.733705420392254e-05,
+      "loss": 2.4065,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9334768056869507,
+      "learning_rate": 9.733046907084944e-05,
+      "loss": 2.4496,
+      "step": 4115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9185383915901184,
+      "learning_rate": 9.732387602903091e-05,
+      "loss": 2.4073,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9374887347221375,
+      "learning_rate": 9.731727507956862e-05,
+      "loss": 2.4111,
+      "step": 4125
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9488128423690796,
+      "learning_rate": 9.73106662235656e-05,
+      "loss": 2.4784,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9355990290641785,
+      "learning_rate": 9.730404946212614e-05,
+      "loss": 2.3758,
+      "step": 4135
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.003129482269287,
+      "learning_rate": 9.729742479635589e-05,
+      "loss": 2.4134,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8866252899169922,
+      "learning_rate": 9.729079222736177e-05,
+      "loss": 2.4188,
+      "step": 4145
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9120451211929321,
+      "learning_rate": 9.728415175625211e-05,
+      "loss": 2.5029,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9351089596748352,
+      "learning_rate": 9.727750338413647e-05,
+      "loss": 2.3642,
+      "step": 4155
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9402521252632141,
+      "learning_rate": 9.727084711212579e-05,
+      "loss": 2.4079,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9316896796226501,
+      "learning_rate": 9.726418294133229e-05,
+      "loss": 2.4645,
+      "step": 4165
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9427815079689026,
+      "learning_rate": 9.725751087286954e-05,
+      "loss": 2.4452,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9455908536911011,
+      "learning_rate": 9.725083090785242e-05,
+      "loss": 2.3438,
+      "step": 4175
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9462844729423523,
+      "learning_rate": 9.724414304739712e-05,
+      "loss": 2.4545,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0262565612792969,
+      "learning_rate": 9.723744729262116e-05,
+      "loss": 2.4366,
+      "step": 4185
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8880996704101562,
+      "learning_rate": 9.723074364464337e-05,
+      "loss": 2.4417,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9354972243309021,
+      "learning_rate": 9.722403210458393e-05,
+      "loss": 2.4213,
+      "step": 4195
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9378357529640198,
+      "learning_rate": 9.721731267356426e-05,
+      "loss": 2.2891,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9203493595123291,
+      "learning_rate": 9.721058535270722e-05,
+      "loss": 2.4043,
+      "step": 4205
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9648817181587219,
+      "learning_rate": 9.720385014313686e-05,
+      "loss": 2.418,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9651432037353516,
+      "learning_rate": 9.719710704597865e-05,
+      "loss": 2.4086,
+      "step": 4215
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9522579312324524,
+      "learning_rate": 9.719035606235931e-05,
+      "loss": 2.3964,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9644230604171753,
+      "learning_rate": 9.718359719340693e-05,
+      "loss": 2.4268,
+      "step": 4225
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8973716497421265,
+      "learning_rate": 9.717683044025086e-05,
+      "loss": 2.2791,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8778495192527771,
+      "learning_rate": 9.717005580402184e-05,
+      "loss": 2.4634,
+      "step": 4235
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9130176305770874,
+      "learning_rate": 9.716327328585183e-05,
+      "loss": 2.4629,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.963130533695221,
+      "learning_rate": 9.715648288687421e-05,
+      "loss": 2.4705,
+      "step": 4245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0115249156951904,
+      "learning_rate": 9.714968460822363e-05,
+      "loss": 2.486,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9198908805847168,
+      "learning_rate": 9.714287845103602e-05,
+      "loss": 2.4086,
+      "step": 4255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9920523762702942,
+      "learning_rate": 9.71360644164487e-05,
+      "loss": 2.4953,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9044355750083923,
+      "learning_rate": 9.712924250560025e-05,
+      "loss": 2.3152,
+      "step": 4265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9670037627220154,
+      "learning_rate": 9.712241271963059e-05,
+      "loss": 2.3416,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9109953045845032,
+      "learning_rate": 9.711557505968097e-05,
+      "loss": 2.3695,
+      "step": 4275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8977879285812378,
+      "learning_rate": 9.710872952689391e-05,
+      "loss": 2.4446,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9965835809707642,
+      "learning_rate": 9.71018761224133e-05,
+      "loss": 2.424,
+      "step": 4285
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9184681177139282,
+      "learning_rate": 9.709501484738429e-05,
+      "loss": 2.3769,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9263006448745728,
+      "learning_rate": 9.70881457029534e-05,
+      "loss": 2.4598,
+      "step": 4295
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8972083330154419,
+      "learning_rate": 9.708126869026841e-05,
+      "loss": 2.4464,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8948732018470764,
+      "learning_rate": 9.707438381047846e-05,
+      "loss": 2.4345,
+      "step": 4305
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9369457364082336,
+      "learning_rate": 9.7067491064734e-05,
+      "loss": 2.3738,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9304699301719666,
+      "learning_rate": 9.706059045418676e-05,
+      "loss": 2.3378,
+      "step": 4315
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8773072361946106,
+      "learning_rate": 9.705368197998982e-05,
+      "loss": 2.4176,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9393570423126221,
+      "learning_rate": 9.704676564329756e-05,
+      "loss": 2.439,
+      "step": 4325
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9274637699127197,
+      "learning_rate": 9.703984144526565e-05,
+      "loss": 2.3999,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9372665882110596,
+      "learning_rate": 9.703290938705114e-05,
+      "loss": 2.369,
+      "step": 4335
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9269606471061707,
+      "learning_rate": 9.702596946981232e-05,
+      "loss": 2.3853,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9954174757003784,
+      "learning_rate": 9.701902169470883e-05,
+      "loss": 2.3756,
+      "step": 4345
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9097785353660583,
+      "learning_rate": 9.701206606290164e-05,
+      "loss": 2.3708,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9885306358337402,
+      "learning_rate": 9.700510257555295e-05,
+      "loss": 2.5199,
+      "step": 4355
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9939241409301758,
+      "learning_rate": 9.69981312338264e-05,
+      "loss": 2.4033,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9005853533744812,
+      "learning_rate": 9.699115203888686e-05,
+      "loss": 2.3802,
+      "step": 4365
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9160817861557007,
+      "learning_rate": 9.69841649919005e-05,
+      "loss": 2.4785,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.921023428440094,
+      "learning_rate": 9.697717009403485e-05,
+      "loss": 2.4694,
+      "step": 4375
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9178997874259949,
+      "learning_rate": 9.697016734645873e-05,
+      "loss": 2.4071,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9311355352401733,
+      "learning_rate": 9.696315675034227e-05,
+      "loss": 2.3469,
+      "step": 4385
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9294891357421875,
+      "learning_rate": 9.69561383068569e-05,
+      "loss": 2.408,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9211428165435791,
+      "learning_rate": 9.694911201717541e-05,
+      "loss": 2.4584,
+      "step": 4395
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9221304655075073,
+      "learning_rate": 9.694207788247186e-05,
+      "loss": 2.4154,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9390175342559814,
+      "learning_rate": 9.693503590392159e-05,
+      "loss": 2.4211,
+      "step": 4405
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9208308458328247,
+      "learning_rate": 9.692798608270132e-05,
+      "loss": 2.4262,
+      "step": 4410
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9476896524429321,
+      "learning_rate": 9.692092841998904e-05,
+      "loss": 2.3832,
+      "step": 4415
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.961846113204956,
+      "learning_rate": 9.691386291696407e-05,
+      "loss": 2.3995,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9665421843528748,
+      "learning_rate": 9.690678957480702e-05,
+      "loss": 2.3773,
+      "step": 4425
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9126176238059998,
+      "learning_rate": 9.689970839469981e-05,
+      "loss": 2.4987,
+      "step": 4430
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.914466917514801,
+      "learning_rate": 9.689261937782571e-05,
+      "loss": 2.3921,
+      "step": 4435
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.919625997543335,
+      "learning_rate": 9.688552252536924e-05,
+      "loss": 2.4546,
+      "step": 4440
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8936855792999268,
+      "learning_rate": 9.687841783851626e-05,
+      "loss": 2.4703,
+      "step": 4445
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9013928174972534,
+      "learning_rate": 9.687130531845394e-05,
+      "loss": 2.3124,
+      "step": 4450
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9058140516281128,
+      "learning_rate": 9.686418496637076e-05,
+      "loss": 2.4755,
+      "step": 4455
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8965856432914734,
+      "learning_rate": 9.685705678345652e-05,
+      "loss": 2.5435,
+      "step": 4460
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9851200580596924,
+      "learning_rate": 9.684992077090227e-05,
+      "loss": 2.4775,
+      "step": 4465
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9580357074737549,
+      "learning_rate": 9.684277692990045e-05,
+      "loss": 2.4441,
+      "step": 4470
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0136078596115112,
+      "learning_rate": 9.683562526164475e-05,
+      "loss": 2.4326,
+      "step": 4475
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9018086791038513,
+      "learning_rate": 9.68284657673302e-05,
+      "loss": 2.3944,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9321445822715759,
+      "learning_rate": 9.682129844815311e-05,
+      "loss": 2.4115,
+      "step": 4485
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0092363357543945,
+      "learning_rate": 9.681412330531112e-05,
+      "loss": 2.4297,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9574854373931885,
+      "learning_rate": 9.680694034000318e-05,
+      "loss": 2.4465,
+      "step": 4495
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.913598895072937,
+      "learning_rate": 9.67997495534295e-05,
+      "loss": 2.4751,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9203943610191345,
+      "learning_rate": 9.679255094679169e-05,
+      "loss": 2.397,
+      "step": 4505
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8859915137290955,
+      "learning_rate": 9.678534452129256e-05,
+      "loss": 2.4967,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.907242476940155,
+      "learning_rate": 9.677813027813632e-05,
+      "loss": 2.4613,
+      "step": 4515
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9645996689796448,
+      "learning_rate": 9.677090821852842e-05,
+      "loss": 2.427,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9643423557281494,
+      "learning_rate": 9.676367834367562e-05,
+      "loss": 2.4261,
+      "step": 4525
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9031646251678467,
+      "learning_rate": 9.675644065478602e-05,
+      "loss": 2.4376,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8973751664161682,
+      "learning_rate": 9.674919515306904e-05,
+      "loss": 2.4113,
+      "step": 4535
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9448693990707397,
+      "learning_rate": 9.674194183973533e-05,
+      "loss": 2.4086,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0016571283340454,
+      "learning_rate": 9.673468071599694e-05,
+      "loss": 2.412,
+      "step": 4545
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9797172546386719,
+      "learning_rate": 9.672741178306713e-05,
+      "loss": 2.328,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.943922221660614,
+      "learning_rate": 9.672013504216054e-05,
+      "loss": 2.4524,
+      "step": 4555
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9353291392326355,
+      "learning_rate": 9.671285049449307e-05,
+      "loss": 2.415,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.927184522151947,
+      "learning_rate": 9.670555814128197e-05,
+      "loss": 2.4034,
+      "step": 4565
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0380089282989502,
+      "learning_rate": 9.669825798374574e-05,
+      "loss": 2.3605,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9893558621406555,
+      "learning_rate": 9.66909500231042e-05,
+      "loss": 2.3742,
+      "step": 4575
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9415997266769409,
+      "learning_rate": 9.668363426057852e-05,
+      "loss": 2.3593,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9507461786270142,
+      "learning_rate": 9.66763106973911e-05,
+      "loss": 2.4475,
+      "step": 4585
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8913505673408508,
+      "learning_rate": 9.666897933476568e-05,
+      "loss": 2.3505,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9150609374046326,
+      "learning_rate": 9.666164017392734e-05,
+      "loss": 2.3648,
+      "step": 4595
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9047843813896179,
+      "learning_rate": 9.665429321610239e-05,
+      "loss": 2.3748,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8874404430389404,
+      "learning_rate": 9.66469384625185e-05,
+      "loss": 2.4207,
+      "step": 4605
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9344981908798218,
+      "learning_rate": 9.66395759144046e-05,
+      "loss": 2.3732,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.858116626739502,
+      "learning_rate": 9.6632205572991e-05,
+      "loss": 2.3883,
+      "step": 4615
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9246544241905212,
+      "learning_rate": 9.662482743950918e-05,
+      "loss": 2.4142,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9574213624000549,
+      "learning_rate": 9.661744151519206e-05,
+      "loss": 2.4074,
+      "step": 4625
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9163877367973328,
+      "learning_rate": 9.661004780127373e-05,
+      "loss": 2.4177,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.906956672668457,
+      "learning_rate": 9.660264629898974e-05,
+      "loss": 2.4274,
+      "step": 4635
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8948305249214172,
+      "learning_rate": 9.65952370095768e-05,
+      "loss": 2.3675,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.923678457736969,
+      "learning_rate": 9.6587819934273e-05,
+      "loss": 2.4154,
+      "step": 4645
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9069342017173767,
+      "learning_rate": 9.658039507431768e-05,
+      "loss": 2.4141,
+      "step": 4650
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0293047428131104,
+      "learning_rate": 9.657296243095151e-05,
+      "loss": 2.4964,
+      "step": 4655
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9375176429748535,
+      "learning_rate": 9.656552200541648e-05,
+      "loss": 2.4301,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9422687888145447,
+      "learning_rate": 9.655807379895583e-05,
+      "loss": 2.3545,
+      "step": 4665
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.015133023262024,
+      "learning_rate": 9.655061781281413e-05,
+      "loss": 2.4336,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8931894898414612,
+      "learning_rate": 9.654315404823729e-05,
+      "loss": 2.3242,
+      "step": 4675
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9191193580627441,
+      "learning_rate": 9.653568250647242e-05,
+      "loss": 2.4335,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8859982490539551,
+      "learning_rate": 9.652820318876802e-05,
+      "loss": 2.3699,
+      "step": 4685
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8868136405944824,
+      "learning_rate": 9.652071609637385e-05,
+      "loss": 2.4315,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9325542449951172,
+      "learning_rate": 9.651322123054097e-05,
+      "loss": 2.5092,
+      "step": 4695
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8369110226631165,
+      "learning_rate": 9.650571859252175e-05,
+      "loss": 2.4071,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.930732786655426,
+      "learning_rate": 9.649820818356985e-05,
+      "loss": 2.4439,
+      "step": 4705
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9621731638908386,
+      "learning_rate": 9.649069000494025e-05,
+      "loss": 2.3203,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9342699646949768,
+      "learning_rate": 9.648316405788916e-05,
+      "loss": 2.4013,
+      "step": 4715
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8927175402641296,
+      "learning_rate": 9.647563034367419e-05,
+      "loss": 2.3867,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9639919400215149,
+      "learning_rate": 9.646808886355417e-05,
+      "loss": 2.4115,
+      "step": 4725
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9213064908981323,
+      "learning_rate": 9.646053961878925e-05,
+      "loss": 2.4272,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.933320939540863,
+      "learning_rate": 9.64529826106409e-05,
+      "loss": 2.2746,
+      "step": 4735
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9863835573196411,
+      "learning_rate": 9.644541784037186e-05,
+      "loss": 2.3839,
+      "step": 4740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8807384967803955,
+      "learning_rate": 9.643784530924616e-05,
+      "loss": 2.4419,
+      "step": 4745
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9401949644088745,
+      "learning_rate": 9.643026501852916e-05,
+      "loss": 2.4004,
+      "step": 4750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9072442650794983,
+      "learning_rate": 9.64226769694875e-05,
+      "loss": 2.4323,
+      "step": 4755
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9623330235481262,
+      "learning_rate": 9.641508116338908e-05,
+      "loss": 2.441,
+      "step": 4760
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0578004121780396,
+      "learning_rate": 9.640747760150319e-05,
+      "loss": 2.4555,
+      "step": 4765
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0271803140640259,
+      "learning_rate": 9.639986628510032e-05,
+      "loss": 2.561,
+      "step": 4770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9526053667068481,
+      "learning_rate": 9.639224721545229e-05,
+      "loss": 2.3601,
+      "step": 4775
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9829257726669312,
+      "learning_rate": 9.638462039383223e-05,
+      "loss": 2.2809,
+      "step": 4780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9243910312652588,
+      "learning_rate": 9.637698582151454e-05,
+      "loss": 2.4009,
+      "step": 4785
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8903430700302124,
+      "learning_rate": 9.636934349977493e-05,
+      "loss": 2.3912,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9060383439064026,
+      "learning_rate": 9.636169342989043e-05,
+      "loss": 2.4515,
+      "step": 4795
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.930600106716156,
+      "learning_rate": 9.635403561313932e-05,
+      "loss": 2.3118,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8614951372146606,
+      "learning_rate": 9.634637005080119e-05,
+      "loss": 2.3358,
+      "step": 4805
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9185161590576172,
+      "learning_rate": 9.633869674415693e-05,
+      "loss": 2.4221,
+      "step": 4810
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9345917701721191,
+      "learning_rate": 9.633101569448873e-05,
+      "loss": 2.3986,
+      "step": 4815
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9479703307151794,
+      "learning_rate": 9.632332690308004e-05,
+      "loss": 2.3668,
+      "step": 4820
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9155110716819763,
+      "learning_rate": 9.631563037121565e-05,
+      "loss": 2.4013,
+      "step": 4825
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9836328029632568,
+      "learning_rate": 9.63079261001816e-05,
+      "loss": 2.2863,
+      "step": 4830
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.933100700378418,
+      "learning_rate": 9.630021409126529e-05,
+      "loss": 2.3484,
+      "step": 4835
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.935150682926178,
+      "learning_rate": 9.629249434575531e-05,
+      "loss": 2.4107,
+      "step": 4840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9862618446350098,
+      "learning_rate": 9.628476686494164e-05,
+      "loss": 2.4107,
+      "step": 4845
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9364649057388306,
+      "learning_rate": 9.627703165011551e-05,
+      "loss": 2.4831,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9087908267974854,
+      "learning_rate": 9.626928870256944e-05,
+      "loss": 2.4281,
+      "step": 4855
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9340150356292725,
+      "learning_rate": 9.626153802359724e-05,
+      "loss": 2.4387,
+      "step": 4860
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9891103506088257,
+      "learning_rate": 9.625377961449402e-05,
+      "loss": 2.4083,
+      "step": 4865
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9188608527183533,
+      "learning_rate": 9.62460134765562e-05,
+      "loss": 2.3426,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.98333340883255,
+      "learning_rate": 9.623823961108145e-05,
+      "loss": 2.3484,
+      "step": 4875
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8739886283874512,
+      "learning_rate": 9.623045801936876e-05,
+      "loss": 2.337,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9066540598869324,
+      "learning_rate": 9.622266870271842e-05,
+      "loss": 2.3519,
+      "step": 4885
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9180181622505188,
+      "learning_rate": 9.6214871662432e-05,
+      "loss": 2.4442,
+      "step": 4890
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8962790966033936,
+      "learning_rate": 9.620706689981232e-05,
+      "loss": 2.3791,
+      "step": 4895
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8744876384735107,
+      "learning_rate": 9.619925441616354e-05,
+      "loss": 2.38,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8901191353797913,
+      "learning_rate": 9.619143421279112e-05,
+      "loss": 2.4539,
+      "step": 4905
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9480272531509399,
+      "learning_rate": 9.618360629100176e-05,
+      "loss": 2.3804,
+      "step": 4910
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8472768664360046,
+      "learning_rate": 9.61757706521035e-05,
+      "loss": 2.4269,
+      "step": 4915
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9040434956550598,
+      "learning_rate": 9.616792729740564e-05,
+      "loss": 2.3487,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.856554388999939,
+      "learning_rate": 9.616007622821877e-05,
+      "loss": 2.3798,
+      "step": 4925
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9677844643592834,
+      "learning_rate": 9.615221744585476e-05,
+      "loss": 2.3745,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9006856083869934,
+      "learning_rate": 9.614435095162679e-05,
+      "loss": 2.3225,
+      "step": 4935
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8769793510437012,
+      "learning_rate": 9.613647674684936e-05,
+      "loss": 2.4281,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9075840711593628,
+      "learning_rate": 9.612859483283818e-05,
+      "loss": 2.4083,
+      "step": 4945
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9165443181991577,
+      "learning_rate": 9.612070521091029e-05,
+      "loss": 2.4336,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9389075040817261,
+      "learning_rate": 9.611280788238403e-05,
+      "loss": 2.4472,
+      "step": 4955
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8993409872055054,
+      "learning_rate": 9.610490284857901e-05,
+      "loss": 2.3399,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8862504959106445,
+      "learning_rate": 9.609699011081613e-05,
+      "loss": 2.3442,
+      "step": 4965
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8736944198608398,
+      "learning_rate": 9.608906967041757e-05,
+      "loss": 2.2838,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8900834918022156,
+      "learning_rate": 9.608114152870684e-05,
+      "loss": 2.3632,
+      "step": 4975
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0188241004943848,
+      "learning_rate": 9.607320568700866e-05,
+      "loss": 2.4108,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8871243000030518,
+      "learning_rate": 9.606526214664911e-05,
+      "loss": 2.3564,
+      "step": 4985
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8526898622512817,
+      "learning_rate": 9.605731090895551e-05,
+      "loss": 2.3781,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9458948373794556,
+      "learning_rate": 9.604935197525649e-05,
+      "loss": 2.4574,
+      "step": 4995
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9441680312156677,
+      "learning_rate": 9.604138534688196e-05,
+      "loss": 2.3796,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0106290578842163,
+      "learning_rate": 9.60334110251631e-05,
+      "loss": 2.2789,
+      "step": 5005
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8974185585975647,
+      "learning_rate": 9.602542901143241e-05,
+      "loss": 2.3756,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8523945808410645,
+      "learning_rate": 9.601743930702363e-05,
+      "loss": 2.3871,
+      "step": 5015
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9893276691436768,
+      "learning_rate": 9.600944191327184e-05,
+      "loss": 2.3971,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8827396631240845,
+      "learning_rate": 9.600143683151335e-05,
+      "loss": 2.2318,
+      "step": 5025
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8643094301223755,
+      "learning_rate": 9.59934240630858e-05,
+      "loss": 2.3189,
+      "step": 5030
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8970931768417358,
+      "learning_rate": 9.598540360932806e-05,
+      "loss": 2.383,
+      "step": 5035
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9140662550926208,
+      "learning_rate": 9.597737547158035e-05,
+      "loss": 2.4223,
+      "step": 5040
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8942334651947021,
+      "learning_rate": 9.596933965118413e-05,
+      "loss": 2.4948,
+      "step": 5045
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9221407771110535,
+      "learning_rate": 9.596129614948217e-05,
+      "loss": 2.3774,
+      "step": 5050
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.945044994354248,
+      "learning_rate": 9.595324496781849e-05,
+      "loss": 2.3925,
+      "step": 5055
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9074574112892151,
+      "learning_rate": 9.594518610753841e-05,
+      "loss": 2.4114,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.915166437625885,
+      "learning_rate": 9.593711956998855e-05,
+      "loss": 2.3487,
+      "step": 5065
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8620178699493408,
+      "learning_rate": 9.59290453565168e-05,
+      "loss": 2.3773,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8804683685302734,
+      "learning_rate": 9.592096346847232e-05,
+      "loss": 2.3865,
+      "step": 5075
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.839989960193634,
+      "learning_rate": 9.591287390720555e-05,
+      "loss": 2.3571,
+      "step": 5080
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9044989347457886,
+      "learning_rate": 9.590477667406828e-05,
+      "loss": 2.3223,
+      "step": 5085
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.868333637714386,
+      "learning_rate": 9.589667177041347e-05,
+      "loss": 2.411,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.943976104259491,
+      "learning_rate": 9.588855919759544e-05,
+      "loss": 2.422,
+      "step": 5095
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9424667358398438,
+      "learning_rate": 9.588043895696978e-05,
+      "loss": 2.4196,
+      "step": 5100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8777533769607544,
+      "learning_rate": 9.587231104989334e-05,
+      "loss": 2.3739,
+      "step": 5105
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9545459747314453,
+      "learning_rate": 9.586417547772426e-05,
+      "loss": 2.4479,
+      "step": 5110
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9421502947807312,
+      "learning_rate": 9.585603224182196e-05,
+      "loss": 2.382,
+      "step": 5115
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9475576281547546,
+      "learning_rate": 9.584788134354718e-05,
+      "loss": 2.4039,
+      "step": 5120
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.928936243057251,
+      "learning_rate": 9.583972278426186e-05,
+      "loss": 2.4161,
+      "step": 5125
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9466306567192078,
+      "learning_rate": 9.583155656532929e-05,
+      "loss": 2.4013,
+      "step": 5130
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8597455024719238,
+      "learning_rate": 9.582338268811402e-05,
+      "loss": 2.4089,
+      "step": 5135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8554359078407288,
+      "learning_rate": 9.581520115398185e-05,
+      "loss": 2.4553,
+      "step": 5140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8894518613815308,
+      "learning_rate": 9.580701196429991e-05,
+      "loss": 2.4183,
+      "step": 5145
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9501975178718567,
+      "learning_rate": 9.579881512043655e-05,
+      "loss": 2.4335,
+      "step": 5150
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.871879518032074,
+      "learning_rate": 9.579061062376146e-05,
+      "loss": 2.3783,
+      "step": 5155
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9041570425033569,
+      "learning_rate": 9.578239847564559e-05,
+      "loss": 2.3275,
+      "step": 5160
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9120101928710938,
+      "learning_rate": 9.577417867746113e-05,
+      "loss": 2.3556,
+      "step": 5165
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8534817099571228,
+      "learning_rate": 9.576595123058158e-05,
+      "loss": 2.2598,
+      "step": 5170
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8799963593482971,
+      "learning_rate": 9.575771613638175e-05,
+      "loss": 2.4043,
+      "step": 5175
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9457526803016663,
+      "learning_rate": 9.574947339623765e-05,
+      "loss": 2.3628,
+      "step": 5180
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8919568657875061,
+      "learning_rate": 9.574122301152665e-05,
+      "loss": 2.4122,
+      "step": 5185
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.926529049873352,
+      "learning_rate": 9.573296498362733e-05,
+      "loss": 2.366,
+      "step": 5190
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9159050583839417,
+      "learning_rate": 9.572469931391959e-05,
+      "loss": 2.3643,
+      "step": 5195
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8460421562194824,
+      "learning_rate": 9.571642600378459e-05,
+      "loss": 2.4045,
+      "step": 5200
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.905124306678772,
+      "learning_rate": 9.570814505460475e-05,
+      "loss": 2.4061,
+      "step": 5205
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9014148116111755,
+      "learning_rate": 9.569985646776382e-05,
+      "loss": 2.4493,
+      "step": 5210
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8964940309524536,
+      "learning_rate": 9.569156024464677e-05,
+      "loss": 2.3615,
+      "step": 5215
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9067182540893555,
+      "learning_rate": 9.568325638663989e-05,
+      "loss": 2.4244,
+      "step": 5220
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8961549997329712,
+      "learning_rate": 9.567494489513068e-05,
+      "loss": 2.354,
+      "step": 5225
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9228354692459106,
+      "learning_rate": 9.5666625771508e-05,
+      "loss": 2.2951,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.00831937789917,
+      "learning_rate": 9.565829901716194e-05,
+      "loss": 2.3395,
+      "step": 5235
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9050048589706421,
+      "learning_rate": 9.564996463348384e-05,
+      "loss": 2.4149,
+      "step": 5240
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8947855830192566,
+      "learning_rate": 9.564162262186638e-05,
+      "loss": 2.421,
+      "step": 5245
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8860916495323181,
+      "learning_rate": 9.563327298370347e-05,
+      "loss": 2.3651,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8773351311683655,
+      "learning_rate": 9.56249157203903e-05,
+      "loss": 2.3462,
+      "step": 5255
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8875875473022461,
+      "learning_rate": 9.561655083332333e-05,
+      "loss": 2.3575,
+      "step": 5260
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9250591993331909,
+      "learning_rate": 9.56081783239003e-05,
+      "loss": 2.4625,
+      "step": 5265
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.88286954164505,
+      "learning_rate": 9.559979819352024e-05,
+      "loss": 2.3191,
+      "step": 5270
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9125967621803284,
+      "learning_rate": 9.559141044358342e-05,
+      "loss": 2.4135,
+      "step": 5275
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8564786911010742,
+      "learning_rate": 9.558301507549143e-05,
+      "loss": 2.3103,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.938084065914154,
+      "learning_rate": 9.557461209064708e-05,
+      "loss": 2.2852,
+      "step": 5285
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8967111110687256,
+      "learning_rate": 9.556620149045447e-05,
+      "loss": 2.421,
+      "step": 5290
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.04017174243927,
+      "learning_rate": 9.555778327631903e-05,
+      "loss": 2.4509,
+      "step": 5295
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0026618242263794,
+      "learning_rate": 9.554935744964734e-05,
+      "loss": 2.4177,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9193139672279358,
+      "learning_rate": 9.55409240118474e-05,
+      "loss": 2.3368,
+      "step": 5305
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9028599262237549,
+      "learning_rate": 9.553248296432835e-05,
+      "loss": 2.3671,
+      "step": 5310
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.97694331407547,
+      "learning_rate": 9.552403430850068e-05,
+      "loss": 2.3824,
+      "step": 5315
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9519087672233582,
+      "learning_rate": 9.551557804577612e-05,
+      "loss": 2.4141,
+      "step": 5320
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9448785781860352,
+      "learning_rate": 9.550711417756767e-05,
+      "loss": 2.4277,
+      "step": 5325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8522542715072632,
+      "learning_rate": 9.549864270528965e-05,
+      "loss": 2.4225,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.878156840801239,
+      "learning_rate": 9.549016363035758e-05,
+      "loss": 2.3591,
+      "step": 5335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0100148916244507,
+      "learning_rate": 9.548167695418828e-05,
+      "loss": 2.3569,
+      "step": 5340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9239642024040222,
+      "learning_rate": 9.547318267819986e-05,
+      "loss": 2.3878,
+      "step": 5345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8755254745483398,
+      "learning_rate": 9.546468080381167e-05,
+      "loss": 2.4089,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8972409963607788,
+      "learning_rate": 9.545617133244436e-05,
+      "loss": 2.4146,
+      "step": 5355
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.932225227355957,
+      "learning_rate": 9.544765426551981e-05,
+      "loss": 2.3644,
+      "step": 5360
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9268906116485596,
+      "learning_rate": 9.543912960446117e-05,
+      "loss": 2.4118,
+      "step": 5365
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8626202344894409,
+      "learning_rate": 9.543059735069293e-05,
+      "loss": 2.3805,
+      "step": 5370
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9206728339195251,
+      "learning_rate": 9.542205750564077e-05,
+      "loss": 2.363,
+      "step": 5375
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8917890787124634,
+      "learning_rate": 9.541351007073167e-05,
+      "loss": 2.4442,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8839393258094788,
+      "learning_rate": 9.540495504739387e-05,
+      "loss": 2.387,
+      "step": 5385
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9084489345550537,
+      "learning_rate": 9.53963924370569e-05,
+      "loss": 2.3601,
+      "step": 5390
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8639400005340576,
+      "learning_rate": 9.538782224115151e-05,
+      "loss": 2.3852,
+      "step": 5395
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8989722728729248,
+      "learning_rate": 9.537924446110976e-05,
+      "loss": 2.2873,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9546710848808289,
+      "learning_rate": 9.537065909836497e-05,
+      "loss": 2.4139,
+      "step": 5405
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.036780834197998,
+      "learning_rate": 9.536206615435174e-05,
+      "loss": 2.3641,
+      "step": 5410
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9554283618927002,
+      "learning_rate": 9.535346563050591e-05,
+      "loss": 2.3194,
+      "step": 5415
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.931952714920044,
+      "learning_rate": 9.534485752826456e-05,
+      "loss": 2.3533,
+      "step": 5420
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8716091513633728,
+      "learning_rate": 9.53362418490661e-05,
+      "loss": 2.374,
+      "step": 5425
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8770365715026855,
+      "learning_rate": 9.53276185943502e-05,
+      "loss": 2.4082,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8964621424674988,
+      "learning_rate": 9.531898776555773e-05,
+      "loss": 2.3201,
+      "step": 5435
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8662172555923462,
+      "learning_rate": 9.531034936413091e-05,
+      "loss": 2.3318,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8742765784263611,
+      "learning_rate": 9.530170339151314e-05,
+      "loss": 2.4078,
+      "step": 5445
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8629917502403259,
+      "learning_rate": 9.529304984914917e-05,
+      "loss": 2.3663,
+      "step": 5450
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9175959229469299,
+      "learning_rate": 9.528438873848496e-05,
+      "loss": 2.4359,
+      "step": 5455
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8681564927101135,
+      "learning_rate": 9.527572006096773e-05,
+      "loss": 2.4181,
+      "step": 5460
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8836453557014465,
+      "learning_rate": 9.526704381804601e-05,
+      "loss": 2.3333,
+      "step": 5465
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9177985787391663,
+      "learning_rate": 9.525836001116957e-05,
+      "loss": 2.3935,
+      "step": 5470
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9064120650291443,
+      "learning_rate": 9.524966864178943e-05,
+      "loss": 2.3506,
+      "step": 5475
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9788501858711243,
+      "learning_rate": 9.524096971135789e-05,
+      "loss": 2.475,
+      "step": 5480
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8929643034934998,
+      "learning_rate": 9.52322632213285e-05,
+      "loss": 2.3428,
+      "step": 5485
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8999564051628113,
+      "learning_rate": 9.52235491731561e-05,
+      "loss": 2.2941,
+      "step": 5490
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9700445532798767,
+      "learning_rate": 9.521482756829675e-05,
+      "loss": 2.3413,
+      "step": 5495
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9308359026908875,
+      "learning_rate": 9.520609840820782e-05,
+      "loss": 2.3305,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9669782519340515,
+      "learning_rate": 9.51973616943479e-05,
+      "loss": 2.285,
+      "step": 5505
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8696320652961731,
+      "learning_rate": 9.518861742817689e-05,
+      "loss": 2.3694,
+      "step": 5510
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.849438488483429,
+      "learning_rate": 9.51798656111559e-05,
+      "loss": 2.3765,
+      "step": 5515
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9462512731552124,
+      "learning_rate": 9.517110624474733e-05,
+      "loss": 2.3413,
+      "step": 5520
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9234495162963867,
+      "learning_rate": 9.516233933041485e-05,
+      "loss": 2.4403,
+      "step": 5525
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8750413060188293,
+      "learning_rate": 9.515356486962337e-05,
+      "loss": 2.3526,
+      "step": 5530
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.901748776435852,
+      "learning_rate": 9.514478286383907e-05,
+      "loss": 2.3699,
+      "step": 5535
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1687331199645996,
+      "learning_rate": 9.513599331452939e-05,
+      "loss": 2.2948,
+      "step": 5540
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9027603268623352,
+      "learning_rate": 9.512719622316304e-05,
+      "loss": 2.3891,
+      "step": 5545
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9732416868209839,
+      "learning_rate": 9.511839159120997e-05,
+      "loss": 2.3447,
+      "step": 5550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9370917081832886,
+      "learning_rate": 9.51095794201414e-05,
+      "loss": 2.3802,
+      "step": 5555
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9154049158096313,
+      "learning_rate": 9.510075971142983e-05,
+      "loss": 2.3071,
+      "step": 5560
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8721033930778503,
+      "learning_rate": 9.509193246654896e-05,
+      "loss": 2.403,
+      "step": 5565
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0244243144989014,
+      "learning_rate": 9.508309768697384e-05,
+      "loss": 2.3944,
+      "step": 5570
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8927170634269714,
+      "learning_rate": 9.507425537418068e-05,
+      "loss": 2.3246,
+      "step": 5575
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8788923025131226,
+      "learning_rate": 9.506540552964704e-05,
+      "loss": 2.4006,
+      "step": 5580
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8986176252365112,
+      "learning_rate": 9.505654815485168e-05,
+      "loss": 2.4444,
+      "step": 5585
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8998198509216309,
+      "learning_rate": 9.504768325127464e-05,
+      "loss": 2.3912,
+      "step": 5590
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8963303565979004,
+      "learning_rate": 9.503881082039719e-05,
+      "loss": 2.4216,
+      "step": 5595
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8949082493782043,
+      "learning_rate": 9.50299308637019e-05,
+      "loss": 2.277,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8327757716178894,
+      "learning_rate": 9.502104338267257e-05,
+      "loss": 2.3708,
+      "step": 5605
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8938215970993042,
+      "learning_rate": 9.501214837879428e-05,
+      "loss": 2.3129,
+      "step": 5610
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8860766291618347,
+      "learning_rate": 9.500324585355333e-05,
+      "loss": 2.2993,
+      "step": 5615
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8675596117973328,
+      "learning_rate": 9.499433580843732e-05,
+      "loss": 2.304,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9193153977394104,
+      "learning_rate": 9.498541824493506e-05,
+      "loss": 2.377,
+      "step": 5625
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8608167767524719,
+      "learning_rate": 9.497649316453667e-05,
+      "loss": 2.3717,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8738590478897095,
+      "learning_rate": 9.496756056873348e-05,
+      "loss": 2.3261,
+      "step": 5635
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9335272908210754,
+      "learning_rate": 9.49586204590181e-05,
+      "loss": 2.3961,
+      "step": 5640
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.942826509475708,
+      "learning_rate": 9.494967283688438e-05,
+      "loss": 2.383,
+      "step": 5645
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8700788021087646,
+      "learning_rate": 9.494071770382745e-05,
+      "loss": 2.4066,
+      "step": 5650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8861420750617981,
+      "learning_rate": 9.493175506134367e-05,
+      "loss": 2.4441,
+      "step": 5655
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9100940227508545,
+      "learning_rate": 9.492278491093067e-05,
+      "loss": 2.2915,
+      "step": 5660
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8735195398330688,
+      "learning_rate": 9.491380725408731e-05,
+      "loss": 2.3888,
+      "step": 5665
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9587289690971375,
+      "learning_rate": 9.490482209231374e-05,
+      "loss": 2.4957,
+      "step": 5670
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9370885491371155,
+      "learning_rate": 9.489582942711135e-05,
+      "loss": 2.4016,
+      "step": 5675
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8634048104286194,
+      "learning_rate": 9.488682925998277e-05,
+      "loss": 2.4169,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8932356238365173,
+      "learning_rate": 9.48778215924319e-05,
+      "loss": 2.3667,
+      "step": 5685
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8753283619880676,
+      "learning_rate": 9.48688064259639e-05,
+      "loss": 2.3942,
+      "step": 5690
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8889414072036743,
+      "learning_rate": 9.485978376208513e-05,
+      "loss": 2.4081,
+      "step": 5695
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9508289694786072,
+      "learning_rate": 9.485075360230328e-05,
+      "loss": 2.3999,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.948228657245636,
+      "learning_rate": 9.484171594812728e-05,
+      "loss": 2.3814,
+      "step": 5705
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9147594571113586,
+      "learning_rate": 9.483267080106723e-05,
+      "loss": 2.3166,
+      "step": 5710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9473706483840942,
+      "learning_rate": 9.482361816263456e-05,
+      "loss": 2.3472,
+      "step": 5715
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8958979845046997,
+      "learning_rate": 9.481455803434195e-05,
+      "loss": 2.4172,
+      "step": 5720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8935083150863647,
+      "learning_rate": 9.48054904177033e-05,
+      "loss": 2.3848,
+      "step": 5725
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8881231546401978,
+      "learning_rate": 9.479641531423377e-05,
+      "loss": 2.4037,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9401668310165405,
+      "learning_rate": 9.47873327254498e-05,
+      "loss": 2.3996,
+      "step": 5735
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8910726308822632,
+      "learning_rate": 9.477824265286904e-05,
+      "loss": 2.4338,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8702807426452637,
+      "learning_rate": 9.476914509801039e-05,
+      "loss": 2.3365,
+      "step": 5745
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8391140699386597,
+      "learning_rate": 9.476004006239406e-05,
+      "loss": 2.3767,
+      "step": 5750
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.802550733089447,
+      "learning_rate": 9.475092754754142e-05,
+      "loss": 2.4017,
+      "step": 5755
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8720895648002625,
+      "learning_rate": 9.474180755497517e-05,
+      "loss": 2.3925,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8401806950569153,
+      "learning_rate": 9.473268008621922e-05,
+      "loss": 2.2815,
+      "step": 5765
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8037576675415039,
+      "learning_rate": 9.472354514279874e-05,
+      "loss": 2.3053,
+      "step": 5770
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8823795318603516,
+      "learning_rate": 9.471440272624012e-05,
+      "loss": 2.3757,
+      "step": 5775
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.940277099609375,
+      "learning_rate": 9.470525283807105e-05,
+      "loss": 2.4429,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8633571863174438,
+      "learning_rate": 9.469609547982044e-05,
+      "loss": 2.3444,
+      "step": 5785
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8899670839309692,
+      "learning_rate": 9.468693065301843e-05,
+      "loss": 2.4284,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9663652181625366,
+      "learning_rate": 9.467775835919646e-05,
+      "loss": 2.3549,
+      "step": 5795
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8883543610572815,
+      "learning_rate": 9.466857859988715e-05,
+      "loss": 2.4788,
+      "step": 5800
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8142432570457458,
+      "learning_rate": 9.465939137662444e-05,
+      "loss": 2.3785,
+      "step": 5805
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9079815149307251,
+      "learning_rate": 9.465019669094346e-05,
+      "loss": 2.3809,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8642459511756897,
+      "learning_rate": 9.464099454438062e-05,
+      "loss": 2.3056,
+      "step": 5815
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9050517082214355,
+      "learning_rate": 9.463178493847355e-05,
+      "loss": 2.473,
+      "step": 5820
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.849720299243927,
+      "learning_rate": 9.462256787476113e-05,
+      "loss": 2.4178,
+      "step": 5825
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8839665055274963,
+      "learning_rate": 9.461334335478354e-05,
+      "loss": 2.2984,
+      "step": 5830
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8941105604171753,
+      "learning_rate": 9.460411138008214e-05,
+      "loss": 2.3722,
+      "step": 5835
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8784953951835632,
+      "learning_rate": 9.459487195219954e-05,
+      "loss": 2.3621,
+      "step": 5840
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8818730711936951,
+      "learning_rate": 9.458562507267963e-05,
+      "loss": 2.4061,
+      "step": 5845
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8525165915489197,
+      "learning_rate": 9.457637074306754e-05,
+      "loss": 2.3691,
+      "step": 5850
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9722347259521484,
+      "learning_rate": 9.456710896490961e-05,
+      "loss": 2.4946,
+      "step": 5855
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.930205225944519,
+      "learning_rate": 9.455783973975348e-05,
+      "loss": 2.3879,
+      "step": 5860
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9273691177368164,
+      "learning_rate": 9.454856306914797e-05,
+      "loss": 2.4156,
+      "step": 5865
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9482378363609314,
+      "learning_rate": 9.453927895464321e-05,
+      "loss": 2.3737,
+      "step": 5870
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9449193477630615,
+      "learning_rate": 9.452998739779051e-05,
+      "loss": 2.34,
+      "step": 5875
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8514248132705688,
+      "learning_rate": 9.452068840014246e-05,
+      "loss": 2.41,
+      "step": 5880
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8953546285629272,
+      "learning_rate": 9.451138196325291e-05,
+      "loss": 2.3523,
+      "step": 5885
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9075092077255249,
+      "learning_rate": 9.45020680886769e-05,
+      "loss": 2.3893,
+      "step": 5890
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9305874705314636,
+      "learning_rate": 9.449274677797078e-05,
+      "loss": 2.3725,
+      "step": 5895
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9045056104660034,
+      "learning_rate": 9.448341803269208e-05,
+      "loss": 2.3668,
+      "step": 5900
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9268478751182556,
+      "learning_rate": 9.44740818543996e-05,
+      "loss": 2.3532,
+      "step": 5905
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9682772159576416,
+      "learning_rate": 9.44647382446534e-05,
+      "loss": 2.4233,
+      "step": 5910
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9259551763534546,
+      "learning_rate": 9.445538720501474e-05,
+      "loss": 2.3546,
+      "step": 5915
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8874166011810303,
+      "learning_rate": 9.444602873704616e-05,
+      "loss": 2.3728,
+      "step": 5920
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.956397533416748,
+      "learning_rate": 9.443666284231141e-05,
+      "loss": 2.4158,
+      "step": 5925
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8724414110183716,
+      "learning_rate": 9.44272895223755e-05,
+      "loss": 2.3226,
+      "step": 5930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8658673167228699,
+      "learning_rate": 9.441790877880469e-05,
+      "loss": 2.3768,
+      "step": 5935
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.928284764289856,
+      "learning_rate": 9.440852061316645e-05,
+      "loss": 2.281,
+      "step": 5940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8784018754959106,
+      "learning_rate": 9.439912502702952e-05,
+      "loss": 2.332,
+      "step": 5945
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9238221049308777,
+      "learning_rate": 9.438972202196387e-05,
+      "loss": 2.4048,
+      "step": 5950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9103170037269592,
+      "learning_rate": 9.43803115995407e-05,
+      "loss": 2.3665,
+      "step": 5955
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8264444470405579,
+      "learning_rate": 9.437089376133246e-05,
+      "loss": 2.2835,
+      "step": 5960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8848369717597961,
+      "learning_rate": 9.436146850891283e-05,
+      "loss": 2.4472,
+      "step": 5965
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8830796480178833,
+      "learning_rate": 9.435203584385673e-05,
+      "loss": 2.4665,
+      "step": 5970
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8378769755363464,
+      "learning_rate": 9.434259576774034e-05,
+      "loss": 2.4061,
+      "step": 5975
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8402870893478394,
+      "learning_rate": 9.433314828214103e-05,
+      "loss": 2.4617,
+      "step": 5980
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8827956318855286,
+      "learning_rate": 9.43236933886375e-05,
+      "loss": 2.4319,
+      "step": 5985
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8267470598220825,
+      "learning_rate": 9.431423108880957e-05,
+      "loss": 2.3582,
+      "step": 5990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.874792218208313,
+      "learning_rate": 9.430476138423838e-05,
+      "loss": 2.2854,
+      "step": 5995
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8771827816963196,
+      "learning_rate": 9.429528427650627e-05,
+      "loss": 2.3077,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 38527,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.7134867590414336e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}