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": 426.18035888671875,
+      "learning_rate": 5e-06,
+      "loss": 12.446,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 133.50704956054688,
+      "learning_rate": 1e-05,
+      "loss": 10.5775,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 24.645387649536133,
+      "learning_rate": 1.5e-05,
+      "loss": 9.3468,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.871162414550781,
+      "learning_rate": 2e-05,
+      "loss": 8.5442,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.278607368469238,
+      "learning_rate": 2.5e-05,
+      "loss": 8.1103,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.399601459503174,
+      "learning_rate": 3e-05,
+      "loss": 7.8631,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.213372230529785,
+      "learning_rate": 3.5e-05,
+      "loss": 7.7065,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.212835788726807,
+      "learning_rate": 4e-05,
+      "loss": 7.5715,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.848857402801514,
+      "learning_rate": 4.5e-05,
+      "loss": 7.2805,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9707300662994385,
+      "learning_rate": 5e-05,
+      "loss": 7.089,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.6102747917175293,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 6.7527,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.939392566680908,
+      "learning_rate": 6e-05,
+      "loss": 6.4982,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.5908989906311035,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 6.4366,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.695846080780029,
+      "learning_rate": 7e-05,
+      "loss": 6.2118,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.324390888214111,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 6.0004,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.662642002105713,
+      "learning_rate": 8e-05,
+      "loss": 5.8238,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.4017486572265625,
+      "learning_rate": 8.5e-05,
+      "loss": 5.7451,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5080506801605225,
+      "learning_rate": 9e-05,
+      "loss": 5.4592,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.715005874633789,
+      "learning_rate": 9.5e-05,
+      "loss": 5.4808,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.401458263397217,
+      "learning_rate": 0.0001,
+      "loss": 5.3641,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7545948028564453,
+      "learning_rate": 9.999999582259295e-05,
+      "loss": 5.2415,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.698465824127197,
+      "learning_rate": 9.999998329037248e-05,
+      "loss": 5.0305,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.618136167526245,
+      "learning_rate": 9.999996240334068e-05,
+      "loss": 5.0326,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.638108491897583,
+      "learning_rate": 9.999993316150106e-05,
+      "loss": 5.0168,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.8629040718078613,
+      "learning_rate": 9.999989556485852e-05,
+      "loss": 4.8621,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.249255180358887,
+      "learning_rate": 9.999984961341928e-05,
+      "loss": 4.9185,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0278563499450684,
+      "learning_rate": 9.999979530719108e-05,
+      "loss": 4.8108,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.35263729095459,
+      "learning_rate": 9.999973264618297e-05,
+      "loss": 4.6944,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1974549293518066,
+      "learning_rate": 9.999966163040541e-05,
+      "loss": 4.6346,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.496695041656494,
+      "learning_rate": 9.99995822598703e-05,
+      "loss": 4.6871,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.7935426235198975,
+      "learning_rate": 9.999949453459088e-05,
+      "loss": 4.6717,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.644333600997925,
+      "learning_rate": 9.999939845458179e-05,
+      "loss": 4.5868,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5917253494262695,
+      "learning_rate": 9.999929401985911e-05,
+      "loss": 4.629,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.8092305660247803,
+      "learning_rate": 9.99991812304403e-05,
+      "loss": 4.4057,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9377129077911377,
+      "learning_rate": 9.999906008634416e-05,
+      "loss": 4.4272,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.021524906158447,
+      "learning_rate": 9.9998930587591e-05,
+      "loss": 4.4972,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.621630907058716,
+      "learning_rate": 9.99987927342024e-05,
+      "loss": 4.4616,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0628721714019775,
+      "learning_rate": 9.999864652620143e-05,
+      "loss": 4.2793,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7971071004867554,
+      "learning_rate": 9.999849196361251e-05,
+      "loss": 4.3384,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0656657218933105,
+      "learning_rate": 9.999832904646147e-05,
+      "loss": 4.3016,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.393131971359253,
+      "learning_rate": 9.99981577747755e-05,
+      "loss": 4.3037,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4283273220062256,
+      "learning_rate": 9.999797814858327e-05,
+      "loss": 4.3435,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5568203926086426,
+      "learning_rate": 9.999779016791477e-05,
+      "loss": 4.3479,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8993186950683594,
+      "learning_rate": 9.999759383280141e-05,
+      "loss": 4.2961,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1974663734436035,
+      "learning_rate": 9.9997389143276e-05,
+      "loss": 4.224,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.119879722595215,
+      "learning_rate": 9.999717609937275e-05,
+      "loss": 4.2174,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.360037088394165,
+      "learning_rate": 9.999695470112725e-05,
+      "loss": 4.1384,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7669646739959717,
+      "learning_rate": 9.99967249485765e-05,
+      "loss": 4.1489,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0529701709747314,
+      "learning_rate": 9.999648684175888e-05,
+      "loss": 4.1061,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.922174096107483,
+      "learning_rate": 9.999624038071418e-05,
+      "loss": 4.0857,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9776692390441895,
+      "learning_rate": 9.99959855654836e-05,
+      "loss": 4.1507,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.774624228477478,
+      "learning_rate": 9.999572239610968e-05,
+      "loss": 4.1731,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.029313802719116,
+      "learning_rate": 9.999545087263645e-05,
+      "loss": 4.152,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8924164772033691,
+      "learning_rate": 9.999517099510923e-05,
+      "loss": 4.204,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8299586772918701,
+      "learning_rate": 9.999488276357481e-05,
+      "loss": 4.0556,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6953365802764893,
+      "learning_rate": 9.999458617808137e-05,
+      "loss": 4.0019,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6515867710113525,
+      "learning_rate": 9.999428123867843e-05,
+      "loss": 4.0755,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.808777093887329,
+      "learning_rate": 9.999396794541695e-05,
+      "loss": 4.0334,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6550085544586182,
+      "learning_rate": 9.999364629834931e-05,
+      "loss": 4.1135,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9163920879364014,
+      "learning_rate": 9.999331629752924e-05,
+      "loss": 3.9906,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9714016914367676,
+      "learning_rate": 9.999297794301188e-05,
+      "loss": 4.0637,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7115269899368286,
+      "learning_rate": 9.999263123485376e-05,
+      "loss": 3.9657,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7634260654449463,
+      "learning_rate": 9.999227617311282e-05,
+      "loss": 4.0182,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6388550996780396,
+      "learning_rate": 9.99919127578484e-05,
+      "loss": 3.9817,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.863453984260559,
+      "learning_rate": 9.999154098912122e-05,
+      "loss": 3.9667,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6266491413116455,
+      "learning_rate": 9.999116086699338e-05,
+      "loss": 4.0099,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.661559820175171,
+      "learning_rate": 9.999077239152843e-05,
+      "loss": 3.9082,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8201313018798828,
+      "learning_rate": 9.999037556279127e-05,
+      "loss": 3.8129,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7540106773376465,
+      "learning_rate": 9.998997038084818e-05,
+      "loss": 3.8648,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6878774166107178,
+      "learning_rate": 9.998955684576692e-05,
+      "loss": 3.8323,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.452054500579834,
+      "learning_rate": 9.998913495761654e-05,
+      "loss": 3.8726,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8333476781845093,
+      "learning_rate": 9.998870471646757e-05,
+      "loss": 3.7823,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8102201223373413,
+      "learning_rate": 9.99882661223919e-05,
+      "loss": 3.8086,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5616700649261475,
+      "learning_rate": 9.998781917546278e-05,
+      "loss": 3.8216,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7297526597976685,
+      "learning_rate": 9.998736387575493e-05,
+      "loss": 3.7636,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.594663381576538,
+      "learning_rate": 9.998690022334442e-05,
+      "loss": 3.831,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6152981519699097,
+      "learning_rate": 9.998642821830873e-05,
+      "loss": 3.9026,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6784144639968872,
+      "learning_rate": 9.99859478607267e-05,
+      "loss": 3.7546,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4909480810165405,
+      "learning_rate": 9.998545915067864e-05,
+      "loss": 3.7448,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6328327655792236,
+      "learning_rate": 9.998496208824618e-05,
+      "loss": 3.8096,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6126734018325806,
+      "learning_rate": 9.99844566735124e-05,
+      "loss": 3.6974,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4800206422805786,
+      "learning_rate": 9.998394290656172e-05,
+      "loss": 3.815,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.491275668144226,
+      "learning_rate": 9.998342078748002e-05,
+      "loss": 3.7098,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5322000980377197,
+      "learning_rate": 9.998289031635453e-05,
+      "loss": 3.8152,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5692301988601685,
+      "learning_rate": 9.998235149327389e-05,
+      "loss": 3.756,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4518071413040161,
+      "learning_rate": 9.998180431832816e-05,
+      "loss": 3.7329,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.467740535736084,
+      "learning_rate": 9.998124879160872e-05,
+      "loss": 3.6498,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4529855251312256,
+      "learning_rate": 9.998068491320844e-05,
+      "loss": 3.7261,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5686273574829102,
+      "learning_rate": 9.998011268322152e-05,
+      "loss": 3.7537,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4502928256988525,
+      "learning_rate": 9.997953210174361e-05,
+      "loss": 3.7691,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4596298933029175,
+      "learning_rate": 9.997894316887167e-05,
+      "loss": 3.6356,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4464672803878784,
+      "learning_rate": 9.997834588470414e-05,
+      "loss": 3.5731,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5296144485473633,
+      "learning_rate": 9.997774024934082e-05,
+      "loss": 3.7799,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7960543632507324,
+      "learning_rate": 9.997712626288292e-05,
+      "loss": 3.583,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5159963369369507,
+      "learning_rate": 9.997650392543303e-05,
+      "loss": 3.7108,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3995882272720337,
+      "learning_rate": 9.997587323709512e-05,
+      "loss": 3.6631,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3247661590576172,
+      "learning_rate": 9.997523419797459e-05,
+      "loss": 3.7176,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.460199236869812,
+      "learning_rate": 9.997458680817824e-05,
+      "loss": 3.6528,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5060099363327026,
+      "learning_rate": 9.997393106781421e-05,
+      "loss": 3.6827,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.497809886932373,
+      "learning_rate": 9.99732669769921e-05,
+      "loss": 3.592,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.390540599822998,
+      "learning_rate": 9.997259453582286e-05,
+      "loss": 3.6004,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4243422746658325,
+      "learning_rate": 9.997191374441887e-05,
+      "loss": 3.6976,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4172883033752441,
+      "learning_rate": 9.997122460289387e-05,
+      "loss": 3.5944,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.426273226737976,
+      "learning_rate": 9.997052711136301e-05,
+      "loss": 3.6531,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4944050312042236,
+      "learning_rate": 9.996982126994287e-05,
+      "loss": 3.5637,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3679972887039185,
+      "learning_rate": 9.996910707875136e-05,
+      "loss": 3.6406,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3345797061920166,
+      "learning_rate": 9.996838453790784e-05,
+      "loss": 3.66,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4048354625701904,
+      "learning_rate": 9.996765364753302e-05,
+      "loss": 3.4909,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5919451713562012,
+      "learning_rate": 9.996691440774906e-05,
+      "loss": 3.5142,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5102155208587646,
+      "learning_rate": 9.996616681867945e-05,
+      "loss": 3.5907,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4334536790847778,
+      "learning_rate": 9.996541088044913e-05,
+      "loss": 3.5902,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4496665000915527,
+      "learning_rate": 9.996464659318442e-05,
+      "loss": 3.6334,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3563491106033325,
+      "learning_rate": 9.996387395701302e-05,
+      "loss": 3.5022,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6198639869689941,
+      "learning_rate": 9.996309297206403e-05,
+      "loss": 3.6494,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.323764443397522,
+      "learning_rate": 9.996230363846797e-05,
+      "loss": 3.6278,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3152966499328613,
+      "learning_rate": 9.996150595635671e-05,
+      "loss": 3.5054,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3538353443145752,
+      "learning_rate": 9.996069992586355e-05,
+      "loss": 3.5348,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3252787590026855,
+      "learning_rate": 9.995988554712318e-05,
+      "loss": 3.5337,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2623413801193237,
+      "learning_rate": 9.995906282027166e-05,
+      "loss": 3.4686,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2985036373138428,
+      "learning_rate": 9.995823174544649e-05,
+      "loss": 3.4308,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4442095756530762,
+      "learning_rate": 9.995739232278654e-05,
+      "loss": 3.4891,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4478293657302856,
+      "learning_rate": 9.995654455243204e-05,
+      "loss": 3.4328,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2850241661071777,
+      "learning_rate": 9.995568843452468e-05,
+      "loss": 3.409,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4163511991500854,
+      "learning_rate": 9.99548239692075e-05,
+      "loss": 3.4389,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2994258403778076,
+      "learning_rate": 9.995395115662498e-05,
+      "loss": 3.4672,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.49166738986969,
+      "learning_rate": 9.995306999692293e-05,
+      "loss": 3.4624,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4814167022705078,
+      "learning_rate": 9.995218049024858e-05,
+      "loss": 3.5354,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4671299457550049,
+      "learning_rate": 9.995128263675058e-05,
+      "loss": 3.4293,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2286376953125,
+      "learning_rate": 9.995037643657899e-05,
+      "loss": 3.5113,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3620476722717285,
+      "learning_rate": 9.994946188988516e-05,
+      "loss": 3.3887,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4385720491409302,
+      "learning_rate": 9.994853899682197e-05,
+      "loss": 3.4533,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5871965885162354,
+      "learning_rate": 9.994760775754362e-05,
+      "loss": 3.5667,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4593559503555298,
+      "learning_rate": 9.994666817220571e-05,
+      "loss": 3.4628,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3421152830123901,
+      "learning_rate": 9.994572024096521e-05,
+      "loss": 3.5004,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4086111783981323,
+      "learning_rate": 9.994476396398055e-05,
+      "loss": 3.4311,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2984883785247803,
+      "learning_rate": 9.994379934141153e-05,
+      "loss": 3.5144,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.317713737487793,
+      "learning_rate": 9.994282637341932e-05,
+      "loss": 3.472,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2906763553619385,
+      "learning_rate": 9.994184506016651e-05,
+      "loss": 3.4089,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1714425086975098,
+      "learning_rate": 9.994085540181703e-05,
+      "loss": 3.4539,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.336357593536377,
+      "learning_rate": 9.99398573985363e-05,
+      "loss": 3.4997,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4577925205230713,
+      "learning_rate": 9.993885105049107e-05,
+      "loss": 3.4119,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4263298511505127,
+      "learning_rate": 9.993783635784947e-05,
+      "loss": 3.4416,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6232032775878906,
+      "learning_rate": 9.993681332078108e-05,
+      "loss": 3.3947,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.410841941833496,
+      "learning_rate": 9.993578193945685e-05,
+      "loss": 3.355,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4003833532333374,
+      "learning_rate": 9.99347422140491e-05,
+      "loss": 3.4094,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.203757405281067,
+      "learning_rate": 9.993369414473157e-05,
+      "loss": 3.3542,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.323467493057251,
+      "learning_rate": 9.993263773167939e-05,
+      "loss": 3.4024,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2646758556365967,
+      "learning_rate": 9.993157297506907e-05,
+      "loss": 3.4129,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.267346739768982,
+      "learning_rate": 9.993049987507856e-05,
+      "loss": 3.3932,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2627936601638794,
+      "learning_rate": 9.992941843188715e-05,
+      "loss": 3.3173,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2175451517105103,
+      "learning_rate": 9.992832864567554e-05,
+      "loss": 3.2085,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2749602794647217,
+      "learning_rate": 9.992723051662584e-05,
+      "loss": 3.4163,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.393596649169922,
+      "learning_rate": 9.992612404492153e-05,
+      "loss": 3.3345,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2145110368728638,
+      "learning_rate": 9.992500923074752e-05,
+      "loss": 3.3136,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2380290031433105,
+      "learning_rate": 9.992388607429006e-05,
+      "loss": 3.3217,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1945409774780273,
+      "learning_rate": 9.992275457573685e-05,
+      "loss": 3.2349,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3474513292312622,
+      "learning_rate": 9.992161473527695e-05,
+      "loss": 3.2629,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3624639511108398,
+      "learning_rate": 9.992046655310083e-05,
+      "loss": 3.3477,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2872951030731201,
+      "learning_rate": 9.991931002940034e-05,
+      "loss": 3.3691,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.257765531539917,
+      "learning_rate": 9.991814516436873e-05,
+      "loss": 3.3897,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2471275329589844,
+      "learning_rate": 9.991697195820065e-05,
+      "loss": 3.4198,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2760255336761475,
+      "learning_rate": 9.991579041109212e-05,
+      "loss": 3.2262,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.210864543914795,
+      "learning_rate": 9.991460052324061e-05,
+      "loss": 3.3433,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3055309057235718,
+      "learning_rate": 9.991340229484491e-05,
+      "loss": 3.3011,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.254165530204773,
+      "learning_rate": 9.991219572610526e-05,
+      "loss": 3.2136,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3235994577407837,
+      "learning_rate": 9.991098081722325e-05,
+      "loss": 3.3157,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2104732990264893,
+      "learning_rate": 9.990975756840193e-05,
+      "loss": 3.3373,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.163535237312317,
+      "learning_rate": 9.990852597984566e-05,
+      "loss": 3.2427,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4858397245407104,
+      "learning_rate": 9.990728605176025e-05,
+      "loss": 3.4172,
+      "step": 845
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2087734937667847,
+      "learning_rate": 9.990603778435288e-05,
+      "loss": 3.3381,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1765812635421753,
+      "learning_rate": 9.990478117783214e-05,
+      "loss": 3.2524,
+      "step": 855
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4474873542785645,
+      "learning_rate": 9.990351623240799e-05,
+      "loss": 3.3397,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.129799485206604,
+      "learning_rate": 9.990224294829181e-05,
+      "loss": 3.2058,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.261946201324463,
+      "learning_rate": 9.990096132569636e-05,
+      "loss": 3.2694,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1320353746414185,
+      "learning_rate": 9.98996713648358e-05,
+      "loss": 3.2668,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2273974418640137,
+      "learning_rate": 9.989837306592567e-05,
+      "loss": 3.2865,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1889970302581787,
+      "learning_rate": 9.989706642918291e-05,
+      "loss": 3.2292,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4396411180496216,
+      "learning_rate": 9.989575145482583e-05,
+      "loss": 3.1921,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2249482870101929,
+      "learning_rate": 9.989442814307422e-05,
+      "loss": 3.2454,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.168701171875,
+      "learning_rate": 9.989309649414914e-05,
+      "loss": 3.2479,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.205196499824524,
+      "learning_rate": 9.989175650827314e-05,
+      "loss": 3.2383,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1792218685150146,
+      "learning_rate": 9.98904081856701e-05,
+      "loss": 3.2414,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3853870630264282,
+      "learning_rate": 9.988905152656534e-05,
+      "loss": 3.187,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1717991828918457,
+      "learning_rate": 9.988768653118554e-05,
+      "loss": 3.2797,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2416123151779175,
+      "learning_rate": 9.988631319975881e-05,
+      "loss": 3.1849,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1639971733093262,
+      "learning_rate": 9.988493153251459e-05,
+      "loss": 3.2093,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1045247316360474,
+      "learning_rate": 9.988354152968379e-05,
+      "loss": 3.2056,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1998716592788696,
+      "learning_rate": 9.988214319149865e-05,
+      "loss": 3.1229,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2485584020614624,
+      "learning_rate": 9.988073651819281e-05,
+      "loss": 3.2585,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1756173372268677,
+      "learning_rate": 9.987932151000137e-05,
+      "loss": 3.1428,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1936028003692627,
+      "learning_rate": 9.987789816716073e-05,
+      "loss": 3.1707,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.148327350616455,
+      "learning_rate": 9.987646648990876e-05,
+      "loss": 3.2643,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2649089097976685,
+      "learning_rate": 9.987502647848466e-05,
+      "loss": 3.2811,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1567057371139526,
+      "learning_rate": 9.987357813312905e-05,
+      "loss": 3.1877,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1892539262771606,
+      "learning_rate": 9.987212145408395e-05,
+      "loss": 3.1995,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2463796138763428,
+      "learning_rate": 9.987065644159278e-05,
+      "loss": 3.2275,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.171724557876587,
+      "learning_rate": 9.986918309590031e-05,
+      "loss": 3.1717,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1639827489852905,
+      "learning_rate": 9.986770141725277e-05,
+      "loss": 3.1898,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1524816751480103,
+      "learning_rate": 9.986621140589772e-05,
+      "loss": 3.2314,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1690089702606201,
+      "learning_rate": 9.986471306208412e-05,
+      "loss": 3.2588,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1586463451385498,
+      "learning_rate": 9.986320638606235e-05,
+      "loss": 3.2041,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1429603099822998,
+      "learning_rate": 9.986169137808419e-05,
+      "loss": 3.1679,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1966091394424438,
+      "learning_rate": 9.986016803840275e-05,
+      "loss": 3.1918,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0967084169387817,
+      "learning_rate": 9.985863636727262e-05,
+      "loss": 3.136,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1640881299972534,
+      "learning_rate": 9.985709636494971e-05,
+      "loss": 3.1794,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2081552743911743,
+      "learning_rate": 9.985554803169134e-05,
+      "loss": 3.1594,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1216572523117065,
+      "learning_rate": 9.985399136775627e-05,
+      "loss": 3.1647,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1249446868896484,
+      "learning_rate": 9.985242637340458e-05,
+      "loss": 3.2107,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2786773443222046,
+      "learning_rate": 9.985085304889778e-05,
+      "loss": 3.1852,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1727296113967896,
+      "learning_rate": 9.984927139449877e-05,
+      "loss": 3.073,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1632459163665771,
+      "learning_rate": 9.984768141047184e-05,
+      "loss": 3.2066,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1269068717956543,
+      "learning_rate": 9.984608309708266e-05,
+      "loss": 3.1533,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1445410251617432,
+      "learning_rate": 9.984447645459831e-05,
+      "loss": 3.1572,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1868129968643188,
+      "learning_rate": 9.984286148328728e-05,
+      "loss": 3.1172,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2418878078460693,
+      "learning_rate": 9.984123818341937e-05,
+      "loss": 3.24,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.129860758781433,
+      "learning_rate": 9.983960655526587e-05,
+      "loss": 3.1601,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3935043811798096,
+      "learning_rate": 9.983796659909941e-05,
+      "loss": 3.23,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1915266513824463,
+      "learning_rate": 9.983631831519401e-05,
+      "loss": 3.1316,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.090220332145691,
+      "learning_rate": 9.983466170382511e-05,
+      "loss": 3.0414,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.199776530265808,
+      "learning_rate": 9.983299676526949e-05,
+      "loss": 3.1456,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2279752492904663,
+      "learning_rate": 9.98313234998054e-05,
+      "loss": 3.16,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.152394413948059,
+      "learning_rate": 9.982964190771242e-05,
+      "loss": 3.0318,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1189119815826416,
+      "learning_rate": 9.98279519892715e-05,
+      "loss": 3.2038,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1378631591796875,
+      "learning_rate": 9.982625374476508e-05,
+      "loss": 3.161,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.188965082168579,
+      "learning_rate": 9.98245471744769e-05,
+      "loss": 3.074,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1226409673690796,
+      "learning_rate": 9.982283227869211e-05,
+      "loss": 2.9548,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.189955234527588,
+      "learning_rate": 9.982110905769729e-05,
+      "loss": 3.2099,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1701760292053223,
+      "learning_rate": 9.981937751178037e-05,
+      "loss": 3.0603,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0948868989944458,
+      "learning_rate": 9.981763764123067e-05,
+      "loss": 3.1611,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0805236101150513,
+      "learning_rate": 9.981588944633893e-05,
+      "loss": 3.2236,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1839979887008667,
+      "learning_rate": 9.981413292739727e-05,
+      "loss": 3.1761,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.116890788078308,
+      "learning_rate": 9.981236808469922e-05,
+      "loss": 3.0368,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0786408185958862,
+      "learning_rate": 9.981059491853964e-05,
+      "loss": 3.0299,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1373776197433472,
+      "learning_rate": 9.980881342921482e-05,
+      "loss": 3.1921,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1015180349349976,
+      "learning_rate": 9.980702361702246e-05,
+      "loss": 3.0895,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1974120140075684,
+      "learning_rate": 9.980522548226162e-05,
+      "loss": 3.1037,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0535948276519775,
+      "learning_rate": 9.980341902523279e-05,
+      "loss": 3.1369,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0819313526153564,
+      "learning_rate": 9.980160424623776e-05,
+      "loss": 3.0054,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2494381666183472,
+      "learning_rate": 9.979978114557984e-05,
+      "loss": 3.1208,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0917394161224365,
+      "learning_rate": 9.979794972356362e-05,
+      "loss": 3.128,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0876154899597168,
+      "learning_rate": 9.979610998049516e-05,
+      "loss": 3.072,
+      "step": 1205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.162632703781128,
+      "learning_rate": 9.979426191668182e-05,
+      "loss": 3.0458,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1051195859909058,
+      "learning_rate": 9.979240553243246e-05,
+      "loss": 3.0632,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.030842661857605,
+      "learning_rate": 9.979054082805726e-05,
+      "loss": 3.078,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1635804176330566,
+      "learning_rate": 9.978866780386777e-05,
+      "loss": 3.1124,
+      "step": 1225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1035135984420776,
+      "learning_rate": 9.978678646017701e-05,
+      "loss": 2.9975,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.128629207611084,
+      "learning_rate": 9.978489679729934e-05,
+      "loss": 3.1404,
+      "step": 1235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2434566020965576,
+      "learning_rate": 9.978299881555048e-05,
+      "loss": 3.0941,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0948474407196045,
+      "learning_rate": 9.97810925152476e-05,
+      "loss": 3.064,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.141882300376892,
+      "learning_rate": 9.977917789670925e-05,
+      "loss": 3.0667,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.645033836364746,
+      "learning_rate": 9.977725496025535e-05,
+      "loss": 3.0299,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.086485743522644,
+      "learning_rate": 9.977532370620718e-05,
+      "loss": 3.1131,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.088101863861084,
+      "learning_rate": 9.977338413488748e-05,
+      "loss": 3.0244,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0710957050323486,
+      "learning_rate": 9.977143624662035e-05,
+      "loss": 3.1098,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1186764240264893,
+      "learning_rate": 9.976948004173125e-05,
+      "loss": 3.0069,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0686196088790894,
+      "learning_rate": 9.976751552054706e-05,
+      "loss": 2.9599,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.171991229057312,
+      "learning_rate": 9.976554268339607e-05,
+      "loss": 3.0856,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1365917921066284,
+      "learning_rate": 9.97635615306079e-05,
+      "loss": 3.0759,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1249431371688843,
+      "learning_rate": 9.976157206251361e-05,
+      "loss": 3.0795,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1358364820480347,
+      "learning_rate": 9.975957427944563e-05,
+      "loss": 3.0917,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.114884853363037,
+      "learning_rate": 9.975756818173778e-05,
+      "loss": 3.1097,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1120790243148804,
+      "learning_rate": 9.975555376972529e-05,
+      "loss": 3.0639,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1494477987289429,
+      "learning_rate": 9.975353104374474e-05,
+      "loss": 3.1468,
+      "step": 1315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1145358085632324,
+      "learning_rate": 9.975150000413411e-05,
+      "loss": 3.0082,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0778875350952148,
+      "learning_rate": 9.97494606512328e-05,
+      "loss": 2.9986,
+      "step": 1325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0747805833816528,
+      "learning_rate": 9.974741298538158e-05,
+      "loss": 3.0235,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.131702184677124,
+      "learning_rate": 9.974535700692257e-05,
+      "loss": 3.0582,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0232104063034058,
+      "learning_rate": 9.974329271619938e-05,
+      "loss": 3.0176,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0874016284942627,
+      "learning_rate": 9.97412201135569e-05,
+      "loss": 3.0286,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.143131971359253,
+      "learning_rate": 9.973913919934147e-05,
+      "loss": 3.1015,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1062005758285522,
+      "learning_rate": 9.973704997390079e-05,
+      "loss": 3.1263,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0381218194961548,
+      "learning_rate": 9.973495243758398e-05,
+      "loss": 3.0049,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1106137037277222,
+      "learning_rate": 9.973284659074151e-05,
+      "loss": 3.1603,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0703545808792114,
+      "learning_rate": 9.973073243372528e-05,
+      "loss": 2.9916,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0879706144332886,
+      "learning_rate": 9.972860996688853e-05,
+      "loss": 2.9712,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0254627466201782,
+      "learning_rate": 9.972647919058595e-05,
+      "loss": 2.9599,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.025583028793335,
+      "learning_rate": 9.972434010517358e-05,
+      "loss": 3.1044,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0460535287857056,
+      "learning_rate": 9.972219271100882e-05,
+      "loss": 3.0289,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0902493000030518,
+      "learning_rate": 9.972003700845052e-05,
+      "loss": 3.1197,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0334908962249756,
+      "learning_rate": 9.971787299785888e-05,
+      "loss": 3.0422,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3371566534042358,
+      "learning_rate": 9.97157006795955e-05,
+      "loss": 3.0107,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0712171792984009,
+      "learning_rate": 9.971352005402338e-05,
+      "loss": 3.019,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0898765325546265,
+      "learning_rate": 9.971133112150688e-05,
+      "loss": 3.0046,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0404963493347168,
+      "learning_rate": 9.970913388241177e-05,
+      "loss": 2.9385,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0246516466140747,
+      "learning_rate": 9.970692833710517e-05,
+      "loss": 2.9883,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0641276836395264,
+      "learning_rate": 9.970471448595566e-05,
+      "loss": 3.0216,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0950161218643188,
+      "learning_rate": 9.970249232933315e-05,
+      "loss": 2.9638,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0462632179260254,
+      "learning_rate": 9.970026186760896e-05,
+      "loss": 2.9371,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0741990804672241,
+      "learning_rate": 9.969802310115577e-05,
+      "loss": 2.9691,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0775219202041626,
+      "learning_rate": 9.96957760303477e-05,
+      "loss": 2.9662,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9930628538131714,
+      "learning_rate": 9.969352065556022e-05,
+      "loss": 2.9606,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0325313806533813,
+      "learning_rate": 9.969125697717017e-05,
+      "loss": 2.9754,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1127303838729858,
+      "learning_rate": 9.968898499555582e-05,
+      "loss": 2.9961,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1072659492492676,
+      "learning_rate": 9.968670471109682e-05,
+      "loss": 2.9944,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9964501261711121,
+      "learning_rate": 9.968441612417418e-05,
+      "loss": 2.9855,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9971349835395813,
+      "learning_rate": 9.968211923517032e-05,
+      "loss": 2.9822,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0763752460479736,
+      "learning_rate": 9.967981404446905e-05,
+      "loss": 3.0237,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0490317344665527,
+      "learning_rate": 9.967750055245555e-05,
+      "loss": 2.9854,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0169031620025635,
+      "learning_rate": 9.967517875951638e-05,
+      "loss": 3.0106,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0954837799072266,
+      "learning_rate": 9.967284866603955e-05,
+      "loss": 3.0582,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1878273487091064,
+      "learning_rate": 9.967051027241436e-05,
+      "loss": 3.0586,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0284348726272583,
+      "learning_rate": 9.966816357903155e-05,
+      "loss": 2.8988,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0099921226501465,
+      "learning_rate": 9.96658085862833e-05,
+      "loss": 3.0007,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1277384757995605,
+      "learning_rate": 9.966344529456305e-05,
+      "loss": 2.9734,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0432116985321045,
+      "learning_rate": 9.966107370426573e-05,
+      "loss": 3.0413,
+      "step": 1525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0945227146148682,
+      "learning_rate": 9.965869381578765e-05,
+      "loss": 3.0244,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.060196876525879,
+      "learning_rate": 9.965630562952643e-05,
+      "loss": 3.0701,
+      "step": 1535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0392168760299683,
+      "learning_rate": 9.965390914588115e-05,
+      "loss": 2.9761,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1757158041000366,
+      "learning_rate": 9.965150436525225e-05,
+      "loss": 2.938,
+      "step": 1545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0493308305740356,
+      "learning_rate": 9.964909128804159e-05,
+      "loss": 3.0117,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0868642330169678,
+      "learning_rate": 9.964666991465234e-05,
+      "loss": 2.9415,
+      "step": 1555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0857800245285034,
+      "learning_rate": 9.964424024548914e-05,
+      "loss": 2.9739,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0179953575134277,
+      "learning_rate": 9.964180228095793e-05,
+      "loss": 2.9885,
+      "step": 1565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0234651565551758,
+      "learning_rate": 9.963935602146612e-05,
+      "loss": 2.9714,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1033720970153809,
+      "learning_rate": 9.963690146742248e-05,
+      "loss": 2.9075,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.068599820137024,
+      "learning_rate": 9.963443861923715e-05,
+      "loss": 3.0289,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9899237751960754,
+      "learning_rate": 9.963196747732164e-05,
+      "loss": 2.894,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0521947145462036,
+      "learning_rate": 9.96294880420889e-05,
+      "loss": 2.9584,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3821214437484741,
+      "learning_rate": 9.96270003139532e-05,
+      "loss": 2.9169,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9883657693862915,
+      "learning_rate": 9.962450429333026e-05,
+      "loss": 2.9626,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0212249755859375,
+      "learning_rate": 9.962199998063715e-05,
+      "loss": 3.0065,
+      "step": 1605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0280909538269043,
+      "learning_rate": 9.961948737629231e-05,
+      "loss": 2.9553,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0603508949279785,
+      "learning_rate": 9.961696648071561e-05,
+      "loss": 2.9652,
+      "step": 1615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1211047172546387,
+      "learning_rate": 9.961443729432828e-05,
+      "loss": 2.9174,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9653599858283997,
+      "learning_rate": 9.961189981755294e-05,
+      "loss": 2.9223,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0482845306396484,
+      "learning_rate": 9.960935405081357e-05,
+      "loss": 2.9028,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9978487491607666,
+      "learning_rate": 9.960679999453559e-05,
+      "loss": 2.9068,
+      "step": 1635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.010831356048584,
+      "learning_rate": 9.960423764914575e-05,
+      "loss": 2.9448,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0715936422348022,
+      "learning_rate": 9.960166701507221e-05,
+      "loss": 2.8968,
+      "step": 1645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0137978792190552,
+      "learning_rate": 9.959908809274453e-05,
+      "loss": 2.9403,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1693183183670044,
+      "learning_rate": 9.959650088259362e-05,
+      "loss": 3.0026,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0348877906799316,
+      "learning_rate": 9.959390538505181e-05,
+      "loss": 2.9192,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0360805988311768,
+      "learning_rate": 9.959130160055279e-05,
+      "loss": 2.9323,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0036619901657104,
+      "learning_rate": 9.958868952953163e-05,
+      "loss": 2.8693,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.053857684135437,
+      "learning_rate": 9.95860691724248e-05,
+      "loss": 2.9377,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0208983421325684,
+      "learning_rate": 9.958344052967018e-05,
+      "loss": 2.9409,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9855614900588989,
+      "learning_rate": 9.958080360170698e-05,
+      "loss": 2.8709,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4475946426391602,
+      "learning_rate": 9.957815838897585e-05,
+      "loss": 2.9105,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.201923131942749,
+      "learning_rate": 9.957550489191874e-05,
+      "loss": 2.9473,
+      "step": 1695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9978868961334229,
+      "learning_rate": 9.957284311097908e-05,
+      "loss": 2.9694,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0527559518814087,
+      "learning_rate": 9.957017304660164e-05,
+      "loss": 2.8897,
+      "step": 1705
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9843922853469849,
+      "learning_rate": 9.956749469923258e-05,
+      "loss": 2.9307,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0339102745056152,
+      "learning_rate": 9.956480806931942e-05,
+      "loss": 2.953,
+      "step": 1715
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9908878207206726,
+      "learning_rate": 9.956211315731112e-05,
+      "loss": 2.927,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0224506855010986,
+      "learning_rate": 9.955940996365796e-05,
+      "loss": 2.8849,
+      "step": 1725
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9599247574806213,
+      "learning_rate": 9.955669848881166e-05,
+      "loss": 2.9145,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9522779583930969,
+      "learning_rate": 9.955397873322527e-05,
+      "loss": 2.8941,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0172940492630005,
+      "learning_rate": 9.955125069735327e-05,
+      "loss": 2.7805,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1202576160430908,
+      "learning_rate": 9.95485143816515e-05,
+      "loss": 2.9176,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0471433401107788,
+      "learning_rate": 9.95457697865772e-05,
+      "loss": 2.8996,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9848760366439819,
+      "learning_rate": 9.954301691258894e-05,
+      "loss": 2.8674,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1284005641937256,
+      "learning_rate": 9.954025576014674e-05,
+      "loss": 2.9183,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0246782302856445,
+      "learning_rate": 9.953748632971201e-05,
+      "loss": 2.9124,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0608782768249512,
+      "learning_rate": 9.953470862174748e-05,
+      "loss": 2.8387,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0004770755767822,
+      "learning_rate": 9.953192263671728e-05,
+      "loss": 2.9515,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9922521710395813,
+      "learning_rate": 9.952912837508697e-05,
+      "loss": 2.8403,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9911460280418396,
+      "learning_rate": 9.952632583732343e-05,
+      "loss": 2.9618,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.007582187652588,
+      "learning_rate": 9.952351502389498e-05,
+      "loss": 2.9109,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0116487741470337,
+      "learning_rate": 9.95206959352713e-05,
+      "loss": 2.9067,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9460859298706055,
+      "learning_rate": 9.951786857192341e-05,
+      "loss": 2.7994,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9887266755104065,
+      "learning_rate": 9.951503293432378e-05,
+      "loss": 2.9541,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0387400388717651,
+      "learning_rate": 9.951218902294625e-05,
+      "loss": 2.9019,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0255929231643677,
+      "learning_rate": 9.9509336838266e-05,
+      "loss": 2.8885,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0374118089675903,
+      "learning_rate": 9.950647638075963e-05,
+      "loss": 2.9043,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9710773825645447,
+      "learning_rate": 9.950360765090511e-05,
+      "loss": 2.8977,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0649971961975098,
+      "learning_rate": 9.950073064918179e-05,
+      "loss": 2.8915,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0160247087478638,
+      "learning_rate": 9.949784537607042e-05,
+      "loss": 2.8783,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.115307092666626,
+      "learning_rate": 9.949495183205311e-05,
+      "loss": 2.962,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9603940844535828,
+      "learning_rate": 9.949205001761334e-05,
+      "loss": 2.8992,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0170562267303467,
+      "learning_rate": 9.948913993323603e-05,
+      "loss": 2.8606,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9743481874465942,
+      "learning_rate": 9.948622157940744e-05,
+      "loss": 2.9336,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9810060262680054,
+      "learning_rate": 9.948329495661517e-05,
+      "loss": 2.8288,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9882898330688477,
+      "learning_rate": 9.94803600653483e-05,
+      "loss": 2.9572,
+      "step": 1865
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0242362022399902,
+      "learning_rate": 9.947741690609722e-05,
+      "loss": 2.8965,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9902169704437256,
+      "learning_rate": 9.94744654793537e-05,
+      "loss": 2.8731,
+      "step": 1875
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9611408114433289,
+      "learning_rate": 9.947150578561095e-05,
+      "loss": 2.8655,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9836825728416443,
+      "learning_rate": 9.946853782536351e-05,
+      "loss": 2.8838,
+      "step": 1885
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0155102014541626,
+      "learning_rate": 9.946556159910732e-05,
+      "loss": 2.8826,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.954998791217804,
+      "learning_rate": 9.946257710733966e-05,
+      "loss": 2.8323,
+      "step": 1895
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9763340950012207,
+      "learning_rate": 9.945958435055927e-05,
+      "loss": 2.8122,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.003547191619873,
+      "learning_rate": 9.945658332926622e-05,
+      "loss": 2.9441,
+      "step": 1905
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9589858651161194,
+      "learning_rate": 9.945357404396197e-05,
+      "loss": 2.902,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0141637325286865,
+      "learning_rate": 9.945055649514934e-05,
+      "loss": 2.9434,
+      "step": 1915
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0157129764556885,
+      "learning_rate": 9.944753068333256e-05,
+      "loss": 2.8803,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9476221203804016,
+      "learning_rate": 9.944449660901725e-05,
+      "loss": 2.88,
+      "step": 1925
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9949897527694702,
+      "learning_rate": 9.944145427271036e-05,
+      "loss": 2.8042,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0566439628601074,
+      "learning_rate": 9.94384036749203e-05,
+      "loss": 2.8467,
+      "step": 1935
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.107778787612915,
+      "learning_rate": 9.943534481615677e-05,
+      "loss": 2.935,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9934400916099548,
+      "learning_rate": 9.943227769693091e-05,
+      "loss": 2.7954,
+      "step": 1945
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9271550178527832,
+      "learning_rate": 9.942920231775524e-05,
+      "loss": 2.8843,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0349804162979126,
+      "learning_rate": 9.942611867914363e-05,
+      "loss": 2.8866,
+      "step": 1955
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9543700814247131,
+      "learning_rate": 9.942302678161133e-05,
+      "loss": 2.7701,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.033215045928955,
+      "learning_rate": 9.9419926625675e-05,
+      "loss": 2.8383,
+      "step": 1965
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.969088077545166,
+      "learning_rate": 9.941681821185265e-05,
+      "loss": 2.7915,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9457157254219055,
+      "learning_rate": 9.941370154066371e-05,
+      "loss": 2.8717,
+      "step": 1975
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9877985119819641,
+      "learning_rate": 9.941057661262895e-05,
+      "loss": 2.9107,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.091912031173706,
+      "learning_rate": 9.940744342827055e-05,
+      "loss": 2.8685,
+      "step": 1985
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9706595540046692,
+      "learning_rate": 9.940430198811202e-05,
+      "loss": 2.8178,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.005304217338562,
+      "learning_rate": 9.940115229267831e-05,
+      "loss": 2.9224,
+      "step": 1995
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.002151608467102,
+      "learning_rate": 9.939799434249571e-05,
+      "loss": 2.9313,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.065636157989502,
+      "learning_rate": 9.93948281380919e-05,
+      "loss": 2.7868,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0273174047470093,
+      "learning_rate": 9.939165367999597e-05,
+      "loss": 2.8666,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0852288007736206,
+      "learning_rate": 9.938847096873831e-05,
+      "loss": 2.8292,
+      "step": 2015
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0099509954452515,
+      "learning_rate": 9.938528000485078e-05,
+      "loss": 2.8784,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0168004035949707,
+      "learning_rate": 9.938208078886655e-05,
+      "loss": 2.8441,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.016646146774292,
+      "learning_rate": 9.937887332132023e-05,
+      "loss": 2.8802,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9762295484542847,
+      "learning_rate": 9.937565760274776e-05,
+      "loss": 2.7822,
+      "step": 2035
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9930585026741028,
+      "learning_rate": 9.937243363368645e-05,
+      "loss": 2.9224,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9699833989143372,
+      "learning_rate": 9.936920141467505e-05,
+      "loss": 2.8538,
+      "step": 2045
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0288692712783813,
+      "learning_rate": 9.936596094625364e-05,
+      "loss": 2.8906,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0242335796356201,
+      "learning_rate": 9.936271222896367e-05,
+      "loss": 2.8683,
+      "step": 2055
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0041627883911133,
+      "learning_rate": 9.935945526334803e-05,
+      "loss": 2.8595,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0165358781814575,
+      "learning_rate": 9.935619004995089e-05,
+      "loss": 2.9446,
+      "step": 2065
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.028939127922058,
+      "learning_rate": 9.935291658931791e-05,
+      "loss": 2.8069,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0034905672073364,
+      "learning_rate": 9.934963488199604e-05,
+      "loss": 2.8484,
+      "step": 2075
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9936198592185974,
+      "learning_rate": 9.934634492853366e-05,
+      "loss": 2.7948,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9549576640129089,
+      "learning_rate": 9.93430467294805e-05,
+      "loss": 2.809,
+      "step": 2085
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9358664751052856,
+      "learning_rate": 9.933974028538768e-05,
+      "loss": 2.8372,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9265405535697937,
+      "learning_rate": 9.933642559680769e-05,
+      "loss": 2.8702,
+      "step": 2095
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0869392156600952,
+      "learning_rate": 9.93331026642944e-05,
+      "loss": 2.8338,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9829980731010437,
+      "learning_rate": 9.932977148840307e-05,
+      "loss": 2.8661,
+      "step": 2105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0239920616149902,
+      "learning_rate": 9.93264320696903e-05,
+      "loss": 2.8775,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9715605974197388,
+      "learning_rate": 9.932308440871413e-05,
+      "loss": 2.8514,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.974958598613739,
+      "learning_rate": 9.931972850603394e-05,
+      "loss": 2.8397,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9619662165641785,
+      "learning_rate": 9.931636436221048e-05,
+      "loss": 2.8135,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9940944910049438,
+      "learning_rate": 9.931299197780586e-05,
+      "loss": 2.9036,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2264835834503174,
+      "learning_rate": 9.930961135338361e-05,
+      "loss": 2.8158,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9642705917358398,
+      "learning_rate": 9.930622248950864e-05,
+      "loss": 2.8381,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9516462683677673,
+      "learning_rate": 9.93028253867472e-05,
+      "loss": 2.7937,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9463425278663635,
+      "learning_rate": 9.929942004566695e-05,
+      "loss": 2.8532,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9425057768821716,
+      "learning_rate": 9.929600646683686e-05,
+      "loss": 2.7654,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9857633113861084,
+      "learning_rate": 9.929258465082738e-05,
+      "loss": 2.7952,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9640198349952698,
+      "learning_rate": 9.928915459821027e-05,
+      "loss": 2.8419,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9426899552345276,
+      "learning_rate": 9.928571630955865e-05,
+      "loss": 2.8464,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0421459674835205,
+      "learning_rate": 9.928226978544708e-05,
+      "loss": 2.8661,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9406557679176331,
+      "learning_rate": 9.927881502645146e-05,
+      "loss": 2.7382,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0043518543243408,
+      "learning_rate": 9.927535203314904e-05,
+      "loss": 2.7779,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.065354585647583,
+      "learning_rate": 9.927188080611849e-05,
+      "loss": 2.8238,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9444592595100403,
+      "learning_rate": 9.926840134593984e-05,
+      "loss": 2.9026,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9701180458068848,
+      "learning_rate": 9.92649136531945e-05,
+      "loss": 2.8057,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9374419450759888,
+      "learning_rate": 9.926141772846525e-05,
+      "loss": 2.7791,
+      "step": 2205
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9759231805801392,
+      "learning_rate": 9.925791357233623e-05,
+      "loss": 2.7605,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.112991452217102,
+      "learning_rate": 9.925440118539298e-05,
+      "loss": 2.8001,
+      "step": 2215
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.983928918838501,
+      "learning_rate": 9.925088056822241e-05,
+      "loss": 2.885,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0053462982177734,
+      "learning_rate": 9.924735172141281e-05,
+      "loss": 2.8472,
+      "step": 2225
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.935817301273346,
+      "learning_rate": 9.924381464555381e-05,
+      "loss": 2.7499,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0035237073898315,
+      "learning_rate": 9.924026934123647e-05,
+      "loss": 2.8148,
+      "step": 2235
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9368577599525452,
+      "learning_rate": 9.92367158090532e-05,
+      "loss": 2.7976,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9868404865264893,
+      "learning_rate": 9.923315404959775e-05,
+      "loss": 2.8149,
+      "step": 2245
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9926549792289734,
+      "learning_rate": 9.92295840634653e-05,
+      "loss": 2.806,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9526712894439697,
+      "learning_rate": 9.922600585125237e-05,
+      "loss": 2.7992,
+      "step": 2255
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9616160988807678,
+      "learning_rate": 9.922241941355688e-05,
+      "loss": 2.899,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9389507174491882,
+      "learning_rate": 9.921882475097811e-05,
+      "loss": 2.8195,
+      "step": 2265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9922882914543152,
+      "learning_rate": 9.921522186411671e-05,
+      "loss": 2.8199,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9568200707435608,
+      "learning_rate": 9.92116107535747e-05,
+      "loss": 2.7601,
+      "step": 2275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9939841032028198,
+      "learning_rate": 9.920799141995549e-05,
+      "loss": 2.7975,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9421373605728149,
+      "learning_rate": 9.920436386386387e-05,
+      "loss": 2.9141,
+      "step": 2285
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9713159203529358,
+      "learning_rate": 9.920072808590596e-05,
+      "loss": 2.8758,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9179216623306274,
+      "learning_rate": 9.91970840866893e-05,
+      "loss": 2.877,
+      "step": 2295
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8952556252479553,
+      "learning_rate": 9.919343186682282e-05,
+      "loss": 2.7489,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9589419960975647,
+      "learning_rate": 9.918977142691675e-05,
+      "loss": 2.7736,
+      "step": 2305
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9862581491470337,
+      "learning_rate": 9.918610276758273e-05,
+      "loss": 2.8109,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9847877621650696,
+      "learning_rate": 9.918242588943383e-05,
+      "loss": 2.8705,
+      "step": 2315
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9045746326446533,
+      "learning_rate": 9.91787407930844e-05,
+      "loss": 2.659,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0158355236053467,
+      "learning_rate": 9.917504747915023e-05,
+      "loss": 2.8525,
+      "step": 2325
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9902061223983765,
+      "learning_rate": 9.917134594824843e-05,
+      "loss": 2.822,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0260533094406128,
+      "learning_rate": 9.916763620099754e-05,
+      "loss": 2.7306,
+      "step": 2335
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0259482860565186,
+      "learning_rate": 9.916391823801743e-05,
+      "loss": 2.8628,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9793617725372314,
+      "learning_rate": 9.916019205992937e-05,
+      "loss": 2.7452,
+      "step": 2345
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2318428754806519,
+      "learning_rate": 9.915645766735597e-05,
+      "loss": 2.7405,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.922905683517456,
+      "learning_rate": 9.915271506092126e-05,
+      "loss": 2.8019,
+      "step": 2355
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9939602017402649,
+      "learning_rate": 9.91489642412506e-05,
+      "loss": 2.8141,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9771950840950012,
+      "learning_rate": 9.914520520897072e-05,
+      "loss": 2.7399,
+      "step": 2365
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9652144908905029,
+      "learning_rate": 9.914143796470978e-05,
+      "loss": 2.8629,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9644772410392761,
+      "learning_rate": 9.913766250909726e-05,
+      "loss": 2.821,
+      "step": 2375
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.967097282409668,
+      "learning_rate": 9.9133878842764e-05,
+      "loss": 2.8817,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9556108713150024,
+      "learning_rate": 9.913008696634226e-05,
+      "loss": 2.7719,
+      "step": 2385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9388400912284851,
+      "learning_rate": 9.912628688046564e-05,
+      "loss": 2.7504,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.002897024154663,
+      "learning_rate": 9.912247858576914e-05,
+      "loss": 2.7832,
+      "step": 2395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9746267199516296,
+      "learning_rate": 9.911866208288906e-05,
+      "loss": 2.781,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0600833892822266,
+      "learning_rate": 9.911483737246319e-05,
+      "loss": 2.7016,
+      "step": 2405
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.995445966720581,
+      "learning_rate": 9.911100445513058e-05,
+      "loss": 2.6595,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9815908074378967,
+      "learning_rate": 9.910716333153169e-05,
+      "loss": 2.7907,
+      "step": 2415
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9226474761962891,
+      "learning_rate": 9.910331400230841e-05,
+      "loss": 2.8152,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9193379878997803,
+      "learning_rate": 9.90994564681039e-05,
+      "loss": 2.7424,
+      "step": 2425
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.184956431388855,
+      "learning_rate": 9.909559072956275e-05,
+      "loss": 2.7992,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9895349740982056,
+      "learning_rate": 9.909171678733091e-05,
+      "loss": 2.8171,
+      "step": 2435
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9701690077781677,
+      "learning_rate": 9.908783464205573e-05,
+      "loss": 2.792,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9654098153114319,
+      "learning_rate": 9.908394429438585e-05,
+      "loss": 2.8027,
+      "step": 2445
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9834014773368835,
+      "learning_rate": 9.908004574497139e-05,
+      "loss": 2.7914,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9326025247573853,
+      "learning_rate": 9.907613899446372e-05,
+      "loss": 2.7933,
+      "step": 2455
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9691859483718872,
+      "learning_rate": 9.90722240435157e-05,
+      "loss": 2.6898,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9826489686965942,
+      "learning_rate": 9.906830089278148e-05,
+      "loss": 2.8018,
+      "step": 2465
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9426137804985046,
+      "learning_rate": 9.906436954291661e-05,
+      "loss": 2.6029,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0089119672775269,
+      "learning_rate": 9.906042999457798e-05,
+      "loss": 2.8499,
+      "step": 2475
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9421010613441467,
+      "learning_rate": 9.905648224842391e-05,
+      "loss": 2.7623,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0579719543457031,
+      "learning_rate": 9.905252630511405e-05,
+      "loss": 2.7838,
+      "step": 2485
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9042184948921204,
+      "learning_rate": 9.904856216530938e-05,
+      "loss": 2.8165,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.01488196849823,
+      "learning_rate": 9.904458982967235e-05,
+      "loss": 2.742,
+      "step": 2495
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9507584571838379,
+      "learning_rate": 9.904060929886668e-05,
+      "loss": 2.7197,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9410552382469177,
+      "learning_rate": 9.903662057355751e-05,
+      "loss": 2.6619,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.25792396068573,
+      "learning_rate": 9.903262365441137e-05,
+      "loss": 2.7888,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0204368829727173,
+      "learning_rate": 9.90286185420961e-05,
+      "loss": 2.824,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9386840462684631,
+      "learning_rate": 9.902460523728094e-05,
+      "loss": 2.8246,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9271748661994934,
+      "learning_rate": 9.902058374063651e-05,
+      "loss": 2.6852,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9196599721908569,
+      "learning_rate": 9.90165540528348e-05,
+      "loss": 2.7834,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1706078052520752,
+      "learning_rate": 9.901251617454913e-05,
+      "loss": 2.6639,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.943121075630188,
+      "learning_rate": 9.900847010645422e-05,
+      "loss": 2.7387,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.083099365234375,
+      "learning_rate": 9.900441584922616e-05,
+      "loss": 2.7291,
+      "step": 2545
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9541768431663513,
+      "learning_rate": 9.90003534035424e-05,
+      "loss": 2.7774,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9969542622566223,
+      "learning_rate": 9.899628277008175e-05,
+      "loss": 2.7335,
+      "step": 2555
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9491851925849915,
+      "learning_rate": 9.899220394952441e-05,
+      "loss": 2.7639,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9778962135314941,
+      "learning_rate": 9.898811694255195e-05,
+      "loss": 2.7343,
+      "step": 2565
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9081137180328369,
+      "learning_rate": 9.898402174984726e-05,
+      "loss": 2.7505,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9772841930389404,
+      "learning_rate": 9.897991837209464e-05,
+      "loss": 2.7681,
+      "step": 2575
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9568387866020203,
+      "learning_rate": 9.897580680997977e-05,
+      "loss": 2.8041,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9198315739631653,
+      "learning_rate": 9.897168706418965e-05,
+      "loss": 2.7374,
+      "step": 2585
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9598795175552368,
+      "learning_rate": 9.89675591354127e-05,
+      "loss": 2.8483,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9452329874038696,
+      "learning_rate": 9.896342302433866e-05,
+      "loss": 2.8131,
+      "step": 2595
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0038601160049438,
+      "learning_rate": 9.895927873165867e-05,
+      "loss": 2.7686,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9627488851547241,
+      "learning_rate": 9.895512625806522e-05,
+      "loss": 2.8559,
+      "step": 2605
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9338017702102661,
+      "learning_rate": 9.895096560425218e-05,
+      "loss": 2.7203,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9379633665084839,
+      "learning_rate": 9.894679677091477e-05,
+      "loss": 2.7543,
+      "step": 2615
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0294241905212402,
+      "learning_rate": 9.89426197587496e-05,
+      "loss": 2.673,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9556803703308105,
+      "learning_rate": 9.893843456845462e-05,
+      "loss": 2.7599,
+      "step": 2625
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.996230959892273,
+      "learning_rate": 9.893424120072916e-05,
+      "loss": 2.7511,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9652789235115051,
+      "learning_rate": 9.893003965627394e-05,
+      "loss": 2.8098,
+      "step": 2635
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.020904302597046,
+      "learning_rate": 9.8925829935791e-05,
+      "loss": 2.7081,
+      "step": 2640
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9489056468009949,
+      "learning_rate": 9.892161203998375e-05,
+      "loss": 2.8232,
+      "step": 2645
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9602557420730591,
+      "learning_rate": 9.891738596955703e-05,
+      "loss": 2.7007,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9232204556465149,
+      "learning_rate": 9.891315172521696e-05,
+      "loss": 2.7668,
+      "step": 2655
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.890339195728302,
+      "learning_rate": 9.890890930767111e-05,
+      "loss": 2.7568,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9281946420669556,
+      "learning_rate": 9.890465871762832e-05,
+      "loss": 2.7579,
+      "step": 2665
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0171369314193726,
+      "learning_rate": 9.89003999557989e-05,
+      "loss": 2.7579,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9164352416992188,
+      "learning_rate": 9.889613302289443e-05,
+      "loss": 2.7982,
+      "step": 2675
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9846821427345276,
+      "learning_rate": 9.889185791962792e-05,
+      "loss": 2.7966,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8969522714614868,
+      "learning_rate": 9.888757464671372e-05,
+      "loss": 2.8282,
+      "step": 2685
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9448633193969727,
+      "learning_rate": 9.888328320486755e-05,
+      "loss": 2.7211,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9905846118927002,
+      "learning_rate": 9.88789835948065e-05,
+      "loss": 2.7425,
+      "step": 2695
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.867183268070221,
+      "learning_rate": 9.887467581724901e-05,
+      "loss": 2.6803,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9273882508277893,
+      "learning_rate": 9.887035987291488e-05,
+      "loss": 2.8092,
+      "step": 2705
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0473154783248901,
+      "learning_rate": 9.88660357625253e-05,
+      "loss": 2.6823,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9323872327804565,
+      "learning_rate": 9.886170348680284e-05,
+      "loss": 2.8402,
+      "step": 2715
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.931232750415802,
+      "learning_rate": 9.885736304647139e-05,
+      "loss": 2.6604,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9706621170043945,
+      "learning_rate": 9.885301444225619e-05,
+      "loss": 2.7445,
+      "step": 2725
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0844476222991943,
+      "learning_rate": 9.884865767488391e-05,
+      "loss": 2.6966,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.036224603652954,
+      "learning_rate": 9.884429274508256e-05,
+      "loss": 2.6997,
+      "step": 2735
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.001693606376648,
+      "learning_rate": 9.883991965358146e-05,
+      "loss": 2.6475,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9305307269096375,
+      "learning_rate": 9.883553840111136e-05,
+      "loss": 2.8103,
+      "step": 2745
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9967879056930542,
+      "learning_rate": 9.883114898840437e-05,
+      "loss": 2.7731,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9027132391929626,
+      "learning_rate": 9.882675141619391e-05,
+      "loss": 2.7005,
+      "step": 2755
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9116165041923523,
+      "learning_rate": 9.882234568521482e-05,
+      "loss": 2.7546,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9303196668624878,
+      "learning_rate": 9.881793179620328e-05,
+      "loss": 2.7153,
+      "step": 2765
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.912118136882782,
+      "learning_rate": 9.881350974989682e-05,
+      "loss": 2.6853,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.256979942321777,
+      "learning_rate": 9.880907954703437e-05,
+      "loss": 2.7084,
+      "step": 2775
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8896024823188782,
+      "learning_rate": 9.880464118835619e-05,
+      "loss": 2.7418,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9137628078460693,
+      "learning_rate": 9.880019467460391e-05,
+      "loss": 2.7236,
+      "step": 2785
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9052730798721313,
+      "learning_rate": 9.879574000652052e-05,
+      "loss": 2.6729,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8999108076095581,
+      "learning_rate": 9.87912771848504e-05,
+      "loss": 2.7737,
+      "step": 2795
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8724818825721741,
+      "learning_rate": 9.878680621033925e-05,
+      "loss": 2.6645,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9051921963691711,
+      "learning_rate": 9.878232708373415e-05,
+      "loss": 2.7398,
+      "step": 2805
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8917181491851807,
+      "learning_rate": 9.877783980578357e-05,
+      "loss": 2.7334,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9198272824287415,
+      "learning_rate": 9.877334437723728e-05,
+      "loss": 2.7152,
+      "step": 2815
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9171764850616455,
+      "learning_rate": 9.87688407988465e-05,
+      "loss": 2.8138,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9169588088989258,
+      "learning_rate": 9.876432907136372e-05,
+      "loss": 2.6297,
+      "step": 2825
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9290556311607361,
+      "learning_rate": 9.875980919554287e-05,
+      "loss": 2.7047,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8996796607971191,
+      "learning_rate": 9.875528117213915e-05,
+      "loss": 2.7318,
+      "step": 2835
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9041152000427246,
+      "learning_rate": 9.875074500190922e-05,
+      "loss": 2.6888,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9060922861099243,
+      "learning_rate": 9.874620068561104e-05,
+      "loss": 2.7527,
+      "step": 2845
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0037927627563477,
+      "learning_rate": 9.874164822400396e-05,
+      "loss": 2.6982,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.888512909412384,
+      "learning_rate": 9.873708761784866e-05,
+      "loss": 2.7739,
+      "step": 2855
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0384182929992676,
+      "learning_rate": 9.873251886790722e-05,
+      "loss": 2.7278,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9704871773719788,
+      "learning_rate": 9.872794197494307e-05,
+      "loss": 2.757,
+      "step": 2865
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9436751008033752,
+      "learning_rate": 9.872335693972094e-05,
+      "loss": 2.7235,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8816616535186768,
+      "learning_rate": 9.871876376300702e-05,
+      "loss": 2.7398,
+      "step": 2875
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9220442771911621,
+      "learning_rate": 9.871416244556881e-05,
+      "loss": 2.7167,
+      "step": 2880
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8692771792411804,
+      "learning_rate": 9.870955298817517e-05,
+      "loss": 2.726,
+      "step": 2885
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8928543925285339,
+      "learning_rate": 9.87049353915963e-05,
+      "loss": 2.7556,
+      "step": 2890
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.907241702079773,
+      "learning_rate": 9.87003096566038e-05,
+      "loss": 2.6646,
+      "step": 2895
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.978085458278656,
+      "learning_rate": 9.869567578397062e-05,
+      "loss": 2.8021,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8571372032165527,
+      "learning_rate": 9.869103377447105e-05,
+      "loss": 2.7259,
+      "step": 2905
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8874258995056152,
+      "learning_rate": 9.868638362888076e-05,
+      "loss": 2.6799,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9457389116287231,
+      "learning_rate": 9.868172534797678e-05,
+      "loss": 2.7191,
+      "step": 2915
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9386129975318909,
+      "learning_rate": 9.867705893253747e-05,
+      "loss": 2.7276,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9299934506416321,
+      "learning_rate": 9.867238438334259e-05,
+      "loss": 2.7491,
+      "step": 2925
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9067992568016052,
+      "learning_rate": 9.866770170117324e-05,
+      "loss": 2.808,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9418742060661316,
+      "learning_rate": 9.866301088681185e-05,
+      "loss": 2.6524,
+      "step": 2935
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9391916990280151,
+      "learning_rate": 9.865831194104227e-05,
+      "loss": 2.7127,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8920215368270874,
+      "learning_rate": 9.865360486464967e-05,
+      "loss": 2.6937,
+      "step": 2945
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9285803437232971,
+      "learning_rate": 9.864888965842056e-05,
+      "loss": 2.7352,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9339229464530945,
+      "learning_rate": 9.864416632314287e-05,
+      "loss": 2.7267,
+      "step": 2955
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9018287062644958,
+      "learning_rate": 9.863943485960583e-05,
+      "loss": 2.7197,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9664990901947021,
+      "learning_rate": 9.863469526860005e-05,
+      "loss": 2.5212,
+      "step": 2965
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8801552057266235,
+      "learning_rate": 9.862994755091752e-05,
+      "loss": 2.8365,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9511743783950806,
+      "learning_rate": 9.862519170735154e-05,
+      "loss": 2.7717,
+      "step": 2975
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9073875546455383,
+      "learning_rate": 9.86204277386968e-05,
+      "loss": 2.741,
+      "step": 2980
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9927611947059631,
+      "learning_rate": 9.861565564574935e-05,
+      "loss": 2.6818,
+      "step": 2985
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9015472531318665,
+      "learning_rate": 9.861087542930658e-05,
+      "loss": 2.7042,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9496403336524963,
+      "learning_rate": 9.860608709016726e-05,
+      "loss": 2.7583,
+      "step": 2995
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9233843684196472,
+      "learning_rate": 9.860129062913147e-05,
+      "loss": 2.7403,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9176088571548462,
+      "learning_rate": 9.859648604700072e-05,
+      "loss": 2.6649,
+      "step": 3005
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9728446006774902,
+      "learning_rate": 9.859167334457781e-05,
+      "loss": 2.7618,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9472501873970032,
+      "learning_rate": 9.858685252266696e-05,
+      "loss": 2.7938,
+      "step": 3015
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9505561590194702,
+      "learning_rate": 9.858202358207367e-05,
+      "loss": 2.6653,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9159867167472839,
+      "learning_rate": 9.857718652360486e-05,
+      "loss": 2.7031,
+      "step": 3025
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9321017861366272,
+      "learning_rate": 9.857234134806878e-05,
+      "loss": 2.7369,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9328396320343018,
+      "learning_rate": 9.856748805627503e-05,
+      "loss": 2.6564,
+      "step": 3035
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9047075510025024,
+      "learning_rate": 9.856262664903462e-05,
+      "loss": 2.7192,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9433457851409912,
+      "learning_rate": 9.855775712715981e-05,
+      "loss": 2.6187,
+      "step": 3045
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8693333864212036,
+      "learning_rate": 9.855287949146432e-05,
+      "loss": 2.6618,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9327419996261597,
+      "learning_rate": 9.854799374276318e-05,
+      "loss": 2.686,
+      "step": 3055
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8990132808685303,
+      "learning_rate": 9.854309988187277e-05,
+      "loss": 2.68,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9077532291412354,
+      "learning_rate": 9.853819790961086e-05,
+      "loss": 2.6599,
+      "step": 3065
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.847281277179718,
+      "learning_rate": 9.853328782679652e-05,
+      "loss": 2.6408,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8919866681098938,
+      "learning_rate": 9.852836963425021e-05,
+      "loss": 2.7691,
+      "step": 3075
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9604905843734741,
+      "learning_rate": 9.852344333279375e-05,
+      "loss": 2.6686,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9188917875289917,
+      "learning_rate": 9.851850892325032e-05,
+      "loss": 2.7263,
+      "step": 3085
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9484882950782776,
+      "learning_rate": 9.851356640644443e-05,
+      "loss": 2.6038,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9097116589546204,
+      "learning_rate": 9.850861578320196e-05,
+      "loss": 2.6905,
+      "step": 3095
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9270111918449402,
+      "learning_rate": 9.850365705435013e-05,
+      "loss": 2.7394,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8859333395957947,
+      "learning_rate": 9.849869022071753e-05,
+      "loss": 2.6556,
+      "step": 3105
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9107666015625,
+      "learning_rate": 9.849371528313411e-05,
+      "loss": 2.7394,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9303011894226074,
+      "learning_rate": 9.848873224243114e-05,
+      "loss": 2.5549,
+      "step": 3115
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9335408210754395,
+      "learning_rate": 9.848374109944128e-05,
+      "loss": 2.5988,
+      "step": 3120
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8851681351661682,
+      "learning_rate": 9.847874185499855e-05,
+      "loss": 2.6995,
+      "step": 3125
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.949263334274292,
+      "learning_rate": 9.847373450993828e-05,
+      "loss": 2.7589,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9172579050064087,
+      "learning_rate": 9.846871906509719e-05,
+      "loss": 2.6774,
+      "step": 3135
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.918825089931488,
+      "learning_rate": 9.846369552131334e-05,
+      "loss": 2.6942,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9259763956069946,
+      "learning_rate": 9.845866387942614e-05,
+      "loss": 2.7608,
+      "step": 3145
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8987737894058228,
+      "learning_rate": 9.845362414027637e-05,
+      "loss": 2.6132,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9737030267715454,
+      "learning_rate": 9.844857630470614e-05,
+      "loss": 2.7465,
+      "step": 3155
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9079709053039551,
+      "learning_rate": 9.844352037355893e-05,
+      "loss": 2.7223,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9806636571884155,
+      "learning_rate": 9.843845634767957e-05,
+      "loss": 2.678,
+      "step": 3165
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8805148601531982,
+      "learning_rate": 9.843338422791424e-05,
+      "loss": 2.6418,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9657164812088013,
+      "learning_rate": 9.842830401511048e-05,
+      "loss": 2.8044,
+      "step": 3175
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.867219090461731,
+      "learning_rate": 9.842321571011714e-05,
+      "loss": 2.6218,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9496291875839233,
+      "learning_rate": 9.841811931378448e-05,
+      "loss": 2.6626,
+      "step": 3185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9704864621162415,
+      "learning_rate": 9.841301482696413e-05,
+      "loss": 2.7197,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8749892115592957,
+      "learning_rate": 9.840790225050896e-05,
+      "loss": 2.75,
+      "step": 3195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8840216398239136,
+      "learning_rate": 9.840278158527328e-05,
+      "loss": 2.6349,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9313614964485168,
+      "learning_rate": 9.839765283211278e-05,
+      "loss": 2.7177,
+      "step": 3205
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9082131385803223,
+      "learning_rate": 9.83925159918844e-05,
+      "loss": 2.7346,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8978071808815002,
+      "learning_rate": 9.838737106544651e-05,
+      "loss": 2.7223,
+      "step": 3215
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9321641325950623,
+      "learning_rate": 9.838221805365881e-05,
+      "loss": 2.7754,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8767082095146179,
+      "learning_rate": 9.837705695738236e-05,
+      "loss": 2.773,
+      "step": 3225
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.866547703742981,
+      "learning_rate": 9.837188777747954e-05,
+      "loss": 2.6626,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8942350745201111,
+      "learning_rate": 9.836671051481411e-05,
+      "loss": 2.6688,
+      "step": 3235
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9156157970428467,
+      "learning_rate": 9.836152517025116e-05,
+      "loss": 2.64,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9996711611747742,
+      "learning_rate": 9.835633174465715e-05,
+      "loss": 2.7502,
+      "step": 3245
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.992365300655365,
+      "learning_rate": 9.83511302388999e-05,
+      "loss": 2.7685,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9065532088279724,
+      "learning_rate": 9.834592065384853e-05,
+      "loss": 2.7477,
+      "step": 3255
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0436832904815674,
+      "learning_rate": 9.834070299037357e-05,
+      "loss": 2.6386,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9090085625648499,
+      "learning_rate": 9.833547724934687e-05,
+      "loss": 2.6495,
+      "step": 3265
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8861886858940125,
+      "learning_rate": 9.833024343164159e-05,
+      "loss": 2.6555,
+      "step": 3270
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.897297739982605,
+      "learning_rate": 9.832500153813234e-05,
+      "loss": 2.6979,
+      "step": 3275
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8549093008041382,
+      "learning_rate": 9.831975156969498e-05,
+      "loss": 2.6331,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8714961409568787,
+      "learning_rate": 9.831449352720677e-05,
+      "loss": 2.7421,
+      "step": 3285
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8711605072021484,
+      "learning_rate": 9.830922741154631e-05,
+      "loss": 2.6881,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9907476902008057,
+      "learning_rate": 9.830395322359357e-05,
+      "loss": 2.6823,
+      "step": 3295
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9561863541603088,
+      "learning_rate": 9.829867096422983e-05,
+      "loss": 2.7065,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8654948472976685,
+      "learning_rate": 9.829338063433773e-05,
+      "loss": 2.6353,
+      "step": 3305
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9002318978309631,
+      "learning_rate": 9.828808223480126e-05,
+      "loss": 2.5643,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.982883870601654,
+      "learning_rate": 9.828277576650578e-05,
+      "loss": 2.6818,
+      "step": 3315
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9662113785743713,
+      "learning_rate": 9.827746123033796e-05,
+      "loss": 2.6858,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.878825843334198,
+      "learning_rate": 9.827213862718587e-05,
+      "loss": 2.7031,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9228978157043457,
+      "learning_rate": 9.826680795793887e-05,
+      "loss": 2.682,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.883462131023407,
+      "learning_rate": 9.826146922348773e-05,
+      "loss": 2.7801,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3158867359161377,
+      "learning_rate": 9.825612242472448e-05,
+      "loss": 2.7218,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9186383485794067,
+      "learning_rate": 9.82507675625426e-05,
+      "loss": 2.6559,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8798549175262451,
+      "learning_rate": 9.824540463783682e-05,
+      "loss": 2.702,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9441373944282532,
+      "learning_rate": 9.824003365150329e-05,
+      "loss": 2.6571,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9047520756721497,
+      "learning_rate": 9.82346546044395e-05,
+      "loss": 2.6296,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8344783782958984,
+      "learning_rate": 9.822926749754424e-05,
+      "loss": 2.7719,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9019176959991455,
+      "learning_rate": 9.822387233171771e-05,
+      "loss": 2.6919,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8871980905532837,
+      "learning_rate": 9.821846910786136e-05,
+      "loss": 2.6547,
+      "step": 3375
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8814390897750854,
+      "learning_rate": 9.821305782687812e-05,
+      "loss": 2.6632,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9737427830696106,
+      "learning_rate": 9.820763848967215e-05,
+      "loss": 2.6727,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8877798914909363,
+      "learning_rate": 9.820221109714902e-05,
+      "loss": 2.6185,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8861896991729736,
+      "learning_rate": 9.81967756502156e-05,
+      "loss": 2.6623,
+      "step": 3395
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8764777779579163,
+      "learning_rate": 9.819133214978017e-05,
+      "loss": 2.6221,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8637602925300598,
+      "learning_rate": 9.818588059675229e-05,
+      "loss": 2.6344,
+      "step": 3405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.854915201663971,
+      "learning_rate": 9.818042099204292e-05,
+      "loss": 2.6734,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8507201075553894,
+      "learning_rate": 9.817495333656433e-05,
+      "loss": 2.6481,
+      "step": 3415
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.914390504360199,
+      "learning_rate": 9.816947763123013e-05,
+      "loss": 2.6175,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8820226192474365,
+      "learning_rate": 9.816399387695529e-05,
+      "loss": 2.6468,
+      "step": 3425
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.923869788646698,
+      "learning_rate": 9.815850207465616e-05,
+      "loss": 2.6911,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8830671310424805,
+      "learning_rate": 9.815300222525035e-05,
+      "loss": 2.6725,
+      "step": 3435
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9591666460037231,
+      "learning_rate": 9.814749432965691e-05,
+      "loss": 2.7198,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8644976019859314,
+      "learning_rate": 9.814197838879615e-05,
+      "loss": 2.7346,
+      "step": 3445
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8714803457260132,
+      "learning_rate": 9.813645440358981e-05,
+      "loss": 2.5632,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9447192549705505,
+      "learning_rate": 9.813092237496087e-05,
+      "loss": 2.6271,
+      "step": 3455
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.863563597202301,
+      "learning_rate": 9.812538230383376e-05,
+      "loss": 2.5901,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9166259765625,
+      "learning_rate": 9.811983419113418e-05,
+      "loss": 2.6283,
+      "step": 3465
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9546032547950745,
+      "learning_rate": 9.81142780377892e-05,
+      "loss": 2.7022,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8953493237495422,
+      "learning_rate": 9.810871384472724e-05,
+      "loss": 2.7038,
+      "step": 3475
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8548471927642822,
+      "learning_rate": 9.810314161287806e-05,
+      "loss": 2.6846,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9003151655197144,
+      "learning_rate": 9.809756134317276e-05,
+      "loss": 2.628,
+      "step": 3485
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9344114661216736,
+      "learning_rate": 9.809197303654376e-05,
+      "loss": 2.6244,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0213067531585693,
+      "learning_rate": 9.808637669392487e-05,
+      "loss": 2.6757,
+      "step": 3495
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9519891738891602,
+      "learning_rate": 9.808077231625119e-05,
+      "loss": 2.8113,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.828662633895874,
+      "learning_rate": 9.807515990445922e-05,
+      "loss": 2.6669,
+      "step": 3505
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9194101095199585,
+      "learning_rate": 9.806953945948676e-05,
+      "loss": 2.7813,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8909932374954224,
+      "learning_rate": 9.806391098227297e-05,
+      "loss": 2.654,
+      "step": 3515
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9151848554611206,
+      "learning_rate": 9.805827447375835e-05,
+      "loss": 2.5737,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9467875361442566,
+      "learning_rate": 9.805262993488473e-05,
+      "loss": 2.672,
+      "step": 3525
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9194997549057007,
+      "learning_rate": 9.804697736659532e-05,
+      "loss": 2.6176,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.898655891418457,
+      "learning_rate": 9.804131676983458e-05,
+      "loss": 2.6835,
+      "step": 3535
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8706089854240417,
+      "learning_rate": 9.803564814554844e-05,
+      "loss": 2.617,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9053091406822205,
+      "learning_rate": 9.802997149468408e-05,
+      "loss": 2.6776,
+      "step": 3545
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.880883514881134,
+      "learning_rate": 9.802428681819003e-05,
+      "loss": 2.6045,
+      "step": 3550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8876532316207886,
+      "learning_rate": 9.801859411701621e-05,
+      "loss": 2.5734,
+      "step": 3555
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8963689208030701,
+      "learning_rate": 9.801289339211384e-05,
+      "loss": 2.5846,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8962706327438354,
+      "learning_rate": 9.800718464443547e-05,
+      "loss": 2.6435,
+      "step": 3565
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9086889028549194,
+      "learning_rate": 9.800146787493503e-05,
+      "loss": 2.674,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8865189552307129,
+      "learning_rate": 9.799574308456775e-05,
+      "loss": 2.6793,
+      "step": 3575
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8926461338996887,
+      "learning_rate": 9.799001027429026e-05,
+      "loss": 2.5838,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8744246959686279,
+      "learning_rate": 9.798426944506048e-05,
+      "loss": 2.6417,
+      "step": 3585
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8832759857177734,
+      "learning_rate": 9.797852059783766e-05,
+      "loss": 2.7798,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8582252860069275,
+      "learning_rate": 9.79727637335824e-05,
+      "loss": 2.6746,
+      "step": 3595
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8821295499801636,
+      "learning_rate": 9.79669988532567e-05,
+      "loss": 2.6595,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0266629457473755,
+      "learning_rate": 9.79612259578238e-05,
+      "loss": 2.5805,
+      "step": 3605
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9103145003318787,
+      "learning_rate": 9.795544504824836e-05,
+      "loss": 2.6765,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8983671069145203,
+      "learning_rate": 9.794965612549635e-05,
+      "loss": 2.6009,
+      "step": 3615
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9042502641677856,
+      "learning_rate": 9.794385919053506e-05,
+      "loss": 2.6381,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9195305109024048,
+      "learning_rate": 9.793805424433313e-05,
+      "loss": 2.7178,
+      "step": 3625
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8771228194236755,
+      "learning_rate": 9.793224128786058e-05,
+      "loss": 2.6587,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8741174936294556,
+      "learning_rate": 9.79264203220887e-05,
+      "loss": 2.6741,
+      "step": 3635
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8881059885025024,
+      "learning_rate": 9.792059134799016e-05,
+      "loss": 2.6871,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8525225520133972,
+      "learning_rate": 9.791475436653896e-05,
+      "loss": 2.6658,
+      "step": 3645
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9314354658126831,
+      "learning_rate": 9.790890937871046e-05,
+      "loss": 2.6072,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.873634934425354,
+      "learning_rate": 9.79030563854813e-05,
+      "loss": 2.7086,
+      "step": 3655
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8968808650970459,
+      "learning_rate": 9.789719538782952e-05,
+      "loss": 2.6597,
+      "step": 3660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.925771176815033,
+      "learning_rate": 9.789132638673445e-05,
+      "loss": 2.6986,
+      "step": 3665
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8554425239562988,
+      "learning_rate": 9.78854493831768e-05,
+      "loss": 2.6748,
+      "step": 3670
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9309921860694885,
+      "learning_rate": 9.787956437813859e-05,
+      "loss": 2.6289,
+      "step": 3675
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8908752202987671,
+      "learning_rate": 9.787367137260317e-05,
+      "loss": 2.5814,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9342464804649353,
+      "learning_rate": 9.786777036755525e-05,
+      "loss": 2.6986,
+      "step": 3685
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9137042164802551,
+      "learning_rate": 9.786186136398085e-05,
+      "loss": 2.5418,
+      "step": 3690
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.906597912311554,
+      "learning_rate": 9.785594436286738e-05,
+      "loss": 2.6815,
+      "step": 3695
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8371202349662781,
+      "learning_rate": 9.78500193652035e-05,
+      "loss": 2.6548,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9031906723976135,
+      "learning_rate": 9.784408637197929e-05,
+      "loss": 2.7005,
+      "step": 3705
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.855553150177002,
+      "learning_rate": 9.783814538418612e-05,
+      "loss": 2.7043,
+      "step": 3710
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.857261598110199,
+      "learning_rate": 9.783219640281671e-05,
+      "loss": 2.6461,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8530116081237793,
+      "learning_rate": 9.78262394288651e-05,
+      "loss": 2.5952,
+      "step": 3720
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8632864356040955,
+      "learning_rate": 9.782027446332671e-05,
+      "loss": 2.6953,
+      "step": 3725
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2306029796600342,
+      "learning_rate": 9.781430150719822e-05,
+      "loss": 2.6686,
+      "step": 3730
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9154161214828491,
+      "learning_rate": 9.780832056147772e-05,
+      "loss": 2.6715,
+      "step": 3735
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8505373001098633,
+      "learning_rate": 9.78023316271646e-05,
+      "loss": 2.644,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8341196775436401,
+      "learning_rate": 9.779633470525957e-05,
+      "loss": 2.7057,
+      "step": 3745
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8755388259887695,
+      "learning_rate": 9.77903297967647e-05,
+      "loss": 2.6308,
+      "step": 3750
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8660781979560852,
+      "learning_rate": 9.778431690268341e-05,
+      "loss": 2.7434,
+      "step": 3755
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9707303047180176,
+      "learning_rate": 9.777829602402043e-05,
+      "loss": 2.6893,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9136419296264648,
+      "learning_rate": 9.777226716178178e-05,
+      "loss": 2.6329,
+      "step": 3765
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.880149245262146,
+      "learning_rate": 9.776623031697492e-05,
+      "loss": 2.5684,
+      "step": 3770
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9061768054962158,
+      "learning_rate": 9.776018549060855e-05,
+      "loss": 2.6096,
+      "step": 3775
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9377713799476624,
+      "learning_rate": 9.775413268369274e-05,
+      "loss": 2.5674,
+      "step": 3780
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9229673743247986,
+      "learning_rate": 9.77480718972389e-05,
+      "loss": 2.7171,
+      "step": 3785
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8514954447746277,
+      "learning_rate": 9.774200313225976e-05,
+      "loss": 2.6015,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8883507251739502,
+      "learning_rate": 9.773592638976939e-05,
+      "loss": 2.5579,
+      "step": 3795
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0416260957717896,
+      "learning_rate": 9.77298416707832e-05,
+      "loss": 2.7123,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9112457036972046,
+      "learning_rate": 9.772374897631791e-05,
+      "loss": 2.554,
+      "step": 3805
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9341439604759216,
+      "learning_rate": 9.771764830739159e-05,
+      "loss": 2.6602,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8770835995674133,
+      "learning_rate": 9.771153966502365e-05,
+      "loss": 2.6043,
+      "step": 3815
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8952957391738892,
+      "learning_rate": 9.77054230502348e-05,
+      "loss": 2.7295,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.863966166973114,
+      "learning_rate": 9.769929846404711e-05,
+      "loss": 2.6714,
+      "step": 3825
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8968183398246765,
+      "learning_rate": 9.769316590748401e-05,
+      "loss": 2.4955,
+      "step": 3830
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1003830432891846,
+      "learning_rate": 9.768702538157018e-05,
+      "loss": 2.6911,
+      "step": 3835
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8449302911758423,
+      "learning_rate": 9.76808768873317e-05,
+      "loss": 2.6048,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8563681840896606,
+      "learning_rate": 9.767472042579595e-05,
+      "loss": 2.5762,
+      "step": 3845
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0113301277160645,
+      "learning_rate": 9.766855599799168e-05,
+      "loss": 2.6386,
+      "step": 3850
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8555015921592712,
+      "learning_rate": 9.76623836049489e-05,
+      "loss": 2.6446,
+      "step": 3855
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8985573649406433,
+      "learning_rate": 9.765620324769903e-05,
+      "loss": 2.5649,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8131394982337952,
+      "learning_rate": 9.765001492727476e-05,
+      "loss": 2.5887,
+      "step": 3865
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8711910843849182,
+      "learning_rate": 9.764381864471017e-05,
+      "loss": 2.6545,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8688917756080627,
+      "learning_rate": 9.76376144010406e-05,
+      "loss": 2.6525,
+      "step": 3875
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9521529674530029,
+      "learning_rate": 9.763140219730278e-05,
+      "loss": 2.5494,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8395829200744629,
+      "learning_rate": 9.762518203453472e-05,
+      "loss": 2.602,
+      "step": 3885
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8899918794631958,
+      "learning_rate": 9.761895391377581e-05,
+      "loss": 2.5693,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8677209615707397,
+      "learning_rate": 9.761271783606674e-05,
+      "loss": 2.6311,
+      "step": 3895
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.879787266254425,
+      "learning_rate": 9.760647380244952e-05,
+      "loss": 2.6324,
+      "step": 3900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8869214057922363,
+      "learning_rate": 9.760022181396753e-05,
+      "loss": 2.6626,
+      "step": 3905
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8692914843559265,
+      "learning_rate": 9.759396187166543e-05,
+      "loss": 2.6632,
+      "step": 3910
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8648121356964111,
+      "learning_rate": 9.758769397658926e-05,
+      "loss": 2.6965,
+      "step": 3915
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9258065819740295,
+      "learning_rate": 9.758141812978634e-05,
+      "loss": 2.5958,
+      "step": 3920
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.96610027551651,
+      "learning_rate": 9.757513433230534e-05,
+      "loss": 2.5795,
+      "step": 3925
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8762859106063843,
+      "learning_rate": 9.756884258519627e-05,
+      "loss": 2.6072,
+      "step": 3930
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8678105473518372,
+      "learning_rate": 9.756254288951046e-05,
+      "loss": 2.5776,
+      "step": 3935
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8727731704711914,
+      "learning_rate": 9.755623524630055e-05,
+      "loss": 2.626,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9462859034538269,
+      "learning_rate": 9.754991965662055e-05,
+      "loss": 2.6043,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9287548661231995,
+      "learning_rate": 9.754359612152574e-05,
+      "loss": 2.6365,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9011033177375793,
+      "learning_rate": 9.753726464207278e-05,
+      "loss": 2.6301,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9051899909973145,
+      "learning_rate": 9.753092521931963e-05,
+      "loss": 2.6536,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8769954442977905,
+      "learning_rate": 9.752457785432558e-05,
+      "loss": 2.6209,
+      "step": 3965
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.847991406917572,
+      "learning_rate": 9.751822254815126e-05,
+      "loss": 2.5809,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8471715450286865,
+      "learning_rate": 9.751185930185861e-05,
+      "loss": 2.5716,
+      "step": 3975
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8967334628105164,
+      "learning_rate": 9.750548811651091e-05,
+      "loss": 2.5771,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.880821168422699,
+      "learning_rate": 9.749910899317276e-05,
+      "loss": 2.6172,
+      "step": 3985
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9148368239402771,
+      "learning_rate": 9.74927219329101e-05,
+      "loss": 2.6083,
+      "step": 3990
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9552202820777893,
+      "learning_rate": 9.748632693679016e-05,
+      "loss": 2.5633,
+      "step": 3995
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.883283257484436,
+      "learning_rate": 9.747992400588153e-05,
+      "loss": 2.5733,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8497976660728455,
+      "learning_rate": 9.747351314125413e-05,
+      "loss": 2.6082,
+      "step": 4005
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8496754765510559,
+      "learning_rate": 9.746709434397916e-05,
+      "loss": 2.5787,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9008999466896057,
+      "learning_rate": 9.746066761512921e-05,
+      "loss": 2.5925,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9060966372489929,
+      "learning_rate": 9.745423295577815e-05,
+      "loss": 2.6035,
+      "step": 4020
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9370459318161011,
+      "learning_rate": 9.744779036700118e-05,
+      "loss": 2.6155,
+      "step": 4025
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9507484436035156,
+      "learning_rate": 9.744133984987485e-05,
+      "loss": 2.6359,
+      "step": 4030
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8699339032173157,
+      "learning_rate": 9.743488140547701e-05,
+      "loss": 2.582,
+      "step": 4035
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8462199568748474,
+      "learning_rate": 9.742841503488682e-05,
+      "loss": 2.5009,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.889661967754364,
+      "learning_rate": 9.742194073918482e-05,
+      "loss": 2.6188,
+      "step": 4045
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.795385479927063,
+      "learning_rate": 9.741545851945282e-05,
+      "loss": 2.5178,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8574442267417908,
+      "learning_rate": 9.740896837677398e-05,
+      "loss": 2.5263,
+      "step": 4055
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9187895059585571,
+      "learning_rate": 9.740247031223279e-05,
+      "loss": 2.7314,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8459597229957581,
+      "learning_rate": 9.739596432691502e-05,
+      "loss": 2.578,
+      "step": 4065
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8605881929397583,
+      "learning_rate": 9.738945042190785e-05,
+      "loss": 2.4828,
+      "step": 4070
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8366894125938416,
+      "learning_rate": 9.738292859829968e-05,
+      "loss": 2.629,
+      "step": 4075
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0206564664840698,
+      "learning_rate": 9.73763988571803e-05,
+      "loss": 2.6191,
+      "step": 4080
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8625677227973938,
+      "learning_rate": 9.73698611996408e-05,
+      "loss": 2.6617,
+      "step": 4085
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9392462968826294,
+      "learning_rate": 9.736331562677361e-05,
+      "loss": 2.6755,
+      "step": 4090
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8422070741653442,
+      "learning_rate": 9.735676213967247e-05,
+      "loss": 2.5503,
+      "step": 4095
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8981051445007324,
+      "learning_rate": 9.735020073943244e-05,
+      "loss": 2.5982,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9680556654930115,
+      "learning_rate": 9.734363142714988e-05,
+      "loss": 2.5611,
+      "step": 4105
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9095462560653687,
+      "learning_rate": 9.733705420392254e-05,
+      "loss": 2.6577,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.893800675868988,
+      "learning_rate": 9.733046907084944e-05,
+      "loss": 2.6089,
+      "step": 4115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8759180903434753,
+      "learning_rate": 9.732387602903091e-05,
+      "loss": 2.5665,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9333301186561584,
+      "learning_rate": 9.731727507956862e-05,
+      "loss": 2.5782,
+      "step": 4125
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8407547473907471,
+      "learning_rate": 9.73106662235656e-05,
+      "loss": 2.6662,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9685622453689575,
+      "learning_rate": 9.730404946212614e-05,
+      "loss": 2.6501,
+      "step": 4135
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8434611558914185,
+      "learning_rate": 9.729742479635589e-05,
+      "loss": 2.6027,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9312947392463684,
+      "learning_rate": 9.729079222736177e-05,
+      "loss": 2.6895,
+      "step": 4145
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9935230016708374,
+      "learning_rate": 9.728415175625211e-05,
+      "loss": 2.5057,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.830369770526886,
+      "learning_rate": 9.727750338413647e-05,
+      "loss": 2.7466,
+      "step": 4155
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8677718043327332,
+      "learning_rate": 9.727084711212579e-05,
+      "loss": 2.4992,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8497341871261597,
+      "learning_rate": 9.726418294133229e-05,
+      "loss": 2.6372,
+      "step": 4165
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9312416315078735,
+      "learning_rate": 9.725751087286954e-05,
+      "loss": 2.5805,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9182408452033997,
+      "learning_rate": 9.725083090785242e-05,
+      "loss": 2.5602,
+      "step": 4175
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9026691317558289,
+      "learning_rate": 9.724414304739712e-05,
+      "loss": 2.5619,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.011706829071045,
+      "learning_rate": 9.723744729262116e-05,
+      "loss": 2.638,
+      "step": 4185
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8374316692352295,
+      "learning_rate": 9.723074364464337e-05,
+      "loss": 2.6765,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9110986590385437,
+      "learning_rate": 9.722403210458393e-05,
+      "loss": 2.6213,
+      "step": 4195
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9063001275062561,
+      "learning_rate": 9.721731267356426e-05,
+      "loss": 2.6603,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8161987066268921,
+      "learning_rate": 9.721058535270722e-05,
+      "loss": 2.5362,
+      "step": 4205
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8989723324775696,
+      "learning_rate": 9.720385014313686e-05,
+      "loss": 2.6161,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1583442687988281,
+      "learning_rate": 9.719710704597865e-05,
+      "loss": 2.609,
+      "step": 4215
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.878730058670044,
+      "learning_rate": 9.719035606235931e-05,
+      "loss": 2.5894,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.80543452501297,
+      "learning_rate": 9.718359719340693e-05,
+      "loss": 2.6045,
+      "step": 4225
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.887678325176239,
+      "learning_rate": 9.717683044025086e-05,
+      "loss": 2.6354,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8895277380943298,
+      "learning_rate": 9.717005580402184e-05,
+      "loss": 2.5541,
+      "step": 4235
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8329252004623413,
+      "learning_rate": 9.716327328585183e-05,
+      "loss": 2.6044,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.848848283290863,
+      "learning_rate": 9.715648288687421e-05,
+      "loss": 2.5584,
+      "step": 4245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8922110795974731,
+      "learning_rate": 9.714968460822363e-05,
+      "loss": 2.6417,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8554126024246216,
+      "learning_rate": 9.714287845103602e-05,
+      "loss": 2.6612,
+      "step": 4255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9240444302558899,
+      "learning_rate": 9.71360644164487e-05,
+      "loss": 2.6067,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8754482269287109,
+      "learning_rate": 9.712924250560025e-05,
+      "loss": 2.6154,
+      "step": 4265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9094959497451782,
+      "learning_rate": 9.712241271963059e-05,
+      "loss": 2.6372,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9169852137565613,
+      "learning_rate": 9.711557505968097e-05,
+      "loss": 2.6219,
+      "step": 4275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.830064058303833,
+      "learning_rate": 9.710872952689391e-05,
+      "loss": 2.5718,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8268505930900574,
+      "learning_rate": 9.71018761224133e-05,
+      "loss": 2.6308,
+      "step": 4285
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8337271809577942,
+      "learning_rate": 9.709501484738429e-05,
+      "loss": 2.5819,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.862708568572998,
+      "learning_rate": 9.70881457029534e-05,
+      "loss": 2.5541,
+      "step": 4295
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9032387137413025,
+      "learning_rate": 9.708126869026841e-05,
+      "loss": 2.5672,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8866088390350342,
+      "learning_rate": 9.707438381047846e-05,
+      "loss": 2.647,
+      "step": 4305
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8936301469802856,
+      "learning_rate": 9.7067491064734e-05,
+      "loss": 2.6116,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8633494973182678,
+      "learning_rate": 9.706059045418676e-05,
+      "loss": 2.5728,
+      "step": 4315
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8519980311393738,
+      "learning_rate": 9.705368197998982e-05,
+      "loss": 2.5943,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.857424259185791,
+      "learning_rate": 9.704676564329756e-05,
+      "loss": 2.5973,
+      "step": 4325
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8597893714904785,
+      "learning_rate": 9.703984144526565e-05,
+      "loss": 2.6598,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7669423222541809,
+      "learning_rate": 9.703290938705114e-05,
+      "loss": 2.4878,
+      "step": 4335
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9712740182876587,
+      "learning_rate": 9.702596946981232e-05,
+      "loss": 2.6038,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8552388548851013,
+      "learning_rate": 9.701902169470883e-05,
+      "loss": 2.6088,
+      "step": 4345
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8856068253517151,
+      "learning_rate": 9.701206606290164e-05,
+      "loss": 2.672,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8588581681251526,
+      "learning_rate": 9.700510257555295e-05,
+      "loss": 2.5714,
+      "step": 4355
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8580461740493774,
+      "learning_rate": 9.69981312338264e-05,
+      "loss": 2.5614,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8324874639511108,
+      "learning_rate": 9.699115203888686e-05,
+      "loss": 2.6712,
+      "step": 4365
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8572141528129578,
+      "learning_rate": 9.69841649919005e-05,
+      "loss": 2.5101,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9788317680358887,
+      "learning_rate": 9.697717009403485e-05,
+      "loss": 2.5436,
+      "step": 4375
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9205249547958374,
+      "learning_rate": 9.697016734645873e-05,
+      "loss": 2.6321,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2291226387023926,
+      "learning_rate": 9.696315675034227e-05,
+      "loss": 2.6545,
+      "step": 4385
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8951936364173889,
+      "learning_rate": 9.69561383068569e-05,
+      "loss": 2.5656,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8908535242080688,
+      "learning_rate": 9.694911201717541e-05,
+      "loss": 2.5434,
+      "step": 4395
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9164063334465027,
+      "learning_rate": 9.694207788247186e-05,
+      "loss": 2.5592,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8768287301063538,
+      "learning_rate": 9.693503590392159e-05,
+      "loss": 2.5555,
+      "step": 4405
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9219667911529541,
+      "learning_rate": 9.692798608270132e-05,
+      "loss": 2.5082,
+      "step": 4410
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9322551488876343,
+      "learning_rate": 9.692092841998904e-05,
+      "loss": 2.5783,
+      "step": 4415
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8495707511901855,
+      "learning_rate": 9.691386291696407e-05,
+      "loss": 2.5951,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8649889826774597,
+      "learning_rate": 9.690678957480702e-05,
+      "loss": 2.6198,
+      "step": 4425
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8151428699493408,
+      "learning_rate": 9.689970839469981e-05,
+      "loss": 2.6141,
+      "step": 4430
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8521438241004944,
+      "learning_rate": 9.689261937782571e-05,
+      "loss": 2.5601,
+      "step": 4435
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8708034157752991,
+      "learning_rate": 9.688552252536924e-05,
+      "loss": 2.6124,
+      "step": 4440
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8403300642967224,
+      "learning_rate": 9.687841783851626e-05,
+      "loss": 2.6103,
+      "step": 4445
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8076593279838562,
+      "learning_rate": 9.687130531845394e-05,
+      "loss": 2.6496,
+      "step": 4450
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9915675520896912,
+      "learning_rate": 9.686418496637076e-05,
+      "loss": 2.7353,
+      "step": 4455
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8794858455657959,
+      "learning_rate": 9.685705678345652e-05,
+      "loss": 2.5538,
+      "step": 4460
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8467981219291687,
+      "learning_rate": 9.684992077090227e-05,
+      "loss": 2.5615,
+      "step": 4465
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8367045521736145,
+      "learning_rate": 9.684277692990045e-05,
+      "loss": 2.6861,
+      "step": 4470
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.844404935836792,
+      "learning_rate": 9.683562526164475e-05,
+      "loss": 2.5874,
+      "step": 4475
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8487129211425781,
+      "learning_rate": 9.68284657673302e-05,
+      "loss": 2.6135,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.887482762336731,
+      "learning_rate": 9.682129844815311e-05,
+      "loss": 2.5206,
+      "step": 4485
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8892886638641357,
+      "learning_rate": 9.681412330531112e-05,
+      "loss": 2.5615,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8832013010978699,
+      "learning_rate": 9.680694034000318e-05,
+      "loss": 2.599,
+      "step": 4495
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9030696749687195,
+      "learning_rate": 9.67997495534295e-05,
+      "loss": 2.5834,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8113387227058411,
+      "learning_rate": 9.679255094679169e-05,
+      "loss": 2.5838,
+      "step": 4505
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8178723454475403,
+      "learning_rate": 9.678534452129256e-05,
+      "loss": 2.574,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8456236720085144,
+      "learning_rate": 9.677813027813632e-05,
+      "loss": 2.6009,
+      "step": 4515
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8258169889450073,
+      "learning_rate": 9.677090821852842e-05,
+      "loss": 2.5212,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8239529132843018,
+      "learning_rate": 9.676367834367562e-05,
+      "loss": 2.6102,
+      "step": 4525
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8398469686508179,
+      "learning_rate": 9.675644065478602e-05,
+      "loss": 2.6733,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.846358060836792,
+      "learning_rate": 9.674919515306904e-05,
+      "loss": 2.5892,
+      "step": 4535
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8715657591819763,
+      "learning_rate": 9.674194183973533e-05,
+      "loss": 2.5927,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8649322986602783,
+      "learning_rate": 9.673468071599694e-05,
+      "loss": 2.6216,
+      "step": 4545
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8926638960838318,
+      "learning_rate": 9.672741178306713e-05,
+      "loss": 2.5253,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8999990820884705,
+      "learning_rate": 9.672013504216054e-05,
+      "loss": 2.5345,
+      "step": 4555
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8736711144447327,
+      "learning_rate": 9.671285049449307e-05,
+      "loss": 2.6288,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9069123864173889,
+      "learning_rate": 9.670555814128197e-05,
+      "loss": 2.5743,
+      "step": 4565
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8568558692932129,
+      "learning_rate": 9.669825798374574e-05,
+      "loss": 2.6456,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.83430016040802,
+      "learning_rate": 9.66909500231042e-05,
+      "loss": 2.5464,
+      "step": 4575
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.863539457321167,
+      "learning_rate": 9.668363426057852e-05,
+      "loss": 2.5426,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8580342531204224,
+      "learning_rate": 9.66763106973911e-05,
+      "loss": 2.4622,
+      "step": 4585
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8713130354881287,
+      "learning_rate": 9.666897933476568e-05,
+      "loss": 2.5456,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8637117147445679,
+      "learning_rate": 9.666164017392734e-05,
+      "loss": 2.4993,
+      "step": 4595
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8373559713363647,
+      "learning_rate": 9.665429321610239e-05,
+      "loss": 2.5849,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8939230442047119,
+      "learning_rate": 9.66469384625185e-05,
+      "loss": 2.5976,
+      "step": 4605
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8490365147590637,
+      "learning_rate": 9.66395759144046e-05,
+      "loss": 2.584,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9264221787452698,
+      "learning_rate": 9.6632205572991e-05,
+      "loss": 2.6107,
+      "step": 4615
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7929337024688721,
+      "learning_rate": 9.662482743950918e-05,
+      "loss": 2.5931,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8125439882278442,
+      "learning_rate": 9.661744151519206e-05,
+      "loss": 2.5662,
+      "step": 4625
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9339861869812012,
+      "learning_rate": 9.661004780127373e-05,
+      "loss": 2.5943,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8792482614517212,
+      "learning_rate": 9.660264629898974e-05,
+      "loss": 2.6323,
+      "step": 4635
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8501409292221069,
+      "learning_rate": 9.65952370095768e-05,
+      "loss": 2.6656,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8210235238075256,
+      "learning_rate": 9.6587819934273e-05,
+      "loss": 2.65,
+      "step": 4645
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8683012127876282,
+      "learning_rate": 9.658039507431768e-05,
+      "loss": 2.5358,
+      "step": 4650
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8194971084594727,
+      "learning_rate": 9.657296243095151e-05,
+      "loss": 2.471,
+      "step": 4655
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9473135471343994,
+      "learning_rate": 9.656552200541648e-05,
+      "loss": 2.4583,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8621748685836792,
+      "learning_rate": 9.655807379895583e-05,
+      "loss": 2.565,
+      "step": 4665
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8269180059432983,
+      "learning_rate": 9.655061781281413e-05,
+      "loss": 2.6074,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5623159408569336,
+      "learning_rate": 9.654315404823729e-05,
+      "loss": 2.4922,
+      "step": 4675
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8667492866516113,
+      "learning_rate": 9.653568250647242e-05,
+      "loss": 2.5392,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9075419902801514,
+      "learning_rate": 9.652820318876802e-05,
+      "loss": 2.5615,
+      "step": 4685
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8343729972839355,
+      "learning_rate": 9.652071609637385e-05,
+      "loss": 2.5566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9610413908958435,
+      "learning_rate": 9.651322123054097e-05,
+      "loss": 2.58,
+      "step": 4695
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.852506697177887,
+      "learning_rate": 9.650571859252175e-05,
+      "loss": 2.4487,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.905195415019989,
+      "learning_rate": 9.649820818356985e-05,
+      "loss": 2.5839,
+      "step": 4705
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1879503726959229,
+      "learning_rate": 9.649069000494025e-05,
+      "loss": 2.5757,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8989693522453308,
+      "learning_rate": 9.648316405788916e-05,
+      "loss": 2.5529,
+      "step": 4715
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8502875566482544,
+      "learning_rate": 9.647563034367419e-05,
+      "loss": 2.5711,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8935533165931702,
+      "learning_rate": 9.646808886355417e-05,
+      "loss": 2.53,
+      "step": 4725
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8671066761016846,
+      "learning_rate": 9.646053961878925e-05,
+      "loss": 2.5382,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.878388524055481,
+      "learning_rate": 9.64529826106409e-05,
+      "loss": 2.5296,
+      "step": 4735
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9976630806922913,
+      "learning_rate": 9.644541784037186e-05,
+      "loss": 2.4583,
+      "step": 4740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.842018723487854,
+      "learning_rate": 9.643784530924616e-05,
+      "loss": 2.5627,
+      "step": 4745
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8966639637947083,
+      "learning_rate": 9.643026501852916e-05,
+      "loss": 2.5443,
+      "step": 4750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.910148561000824,
+      "learning_rate": 9.64226769694875e-05,
+      "loss": 2.5689,
+      "step": 4755
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8362109065055847,
+      "learning_rate": 9.641508116338908e-05,
+      "loss": 2.5138,
+      "step": 4760
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8682191967964172,
+      "learning_rate": 9.640747760150319e-05,
+      "loss": 2.6001,
+      "step": 4765
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9326234459877014,
+      "learning_rate": 9.639986628510032e-05,
+      "loss": 2.5912,
+      "step": 4770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9297434091567993,
+      "learning_rate": 9.639224721545229e-05,
+      "loss": 2.5961,
+      "step": 4775
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8584567904472351,
+      "learning_rate": 9.638462039383223e-05,
+      "loss": 2.6148,
+      "step": 4780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8481149673461914,
+      "learning_rate": 9.637698582151454e-05,
+      "loss": 2.5483,
+      "step": 4785
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.882465124130249,
+      "learning_rate": 9.636934349977493e-05,
+      "loss": 2.5742,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8507612347602844,
+      "learning_rate": 9.636169342989043e-05,
+      "loss": 2.4774,
+      "step": 4795
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9036900997161865,
+      "learning_rate": 9.635403561313932e-05,
+      "loss": 2.581,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8621731400489807,
+      "learning_rate": 9.634637005080119e-05,
+      "loss": 2.6321,
+      "step": 4805
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8277601003646851,
+      "learning_rate": 9.633869674415693e-05,
+      "loss": 2.6218,
+      "step": 4810
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8086833357810974,
+      "learning_rate": 9.633101569448873e-05,
+      "loss": 2.5187,
+      "step": 4815
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8093054294586182,
+      "learning_rate": 9.632332690308004e-05,
+      "loss": 2.5494,
+      "step": 4820
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8423508405685425,
+      "learning_rate": 9.631563037121565e-05,
+      "loss": 2.5583,
+      "step": 4825
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.924947202205658,
+      "learning_rate": 9.63079261001816e-05,
+      "loss": 2.5407,
+      "step": 4830
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9315409660339355,
+      "learning_rate": 9.630021409126529e-05,
+      "loss": 2.5831,
+      "step": 4835
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8367403149604797,
+      "learning_rate": 9.629249434575531e-05,
+      "loss": 2.5988,
+      "step": 4840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8889833092689514,
+      "learning_rate": 9.628476686494164e-05,
+      "loss": 2.5474,
+      "step": 4845
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8519444465637207,
+      "learning_rate": 9.627703165011551e-05,
+      "loss": 2.5637,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8834317922592163,
+      "learning_rate": 9.626928870256944e-05,
+      "loss": 2.5171,
+      "step": 4855
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0047601461410522,
+      "learning_rate": 9.626153802359724e-05,
+      "loss": 2.5084,
+      "step": 4860
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8508792519569397,
+      "learning_rate": 9.625377961449402e-05,
+      "loss": 2.5835,
+      "step": 4865
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8897415995597839,
+      "learning_rate": 9.62460134765562e-05,
+      "loss": 2.6173,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8602465391159058,
+      "learning_rate": 9.623823961108145e-05,
+      "loss": 2.5928,
+      "step": 4875
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9230908751487732,
+      "learning_rate": 9.623045801936876e-05,
+      "loss": 2.5874,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9149127006530762,
+      "learning_rate": 9.622266870271842e-05,
+      "loss": 2.6345,
+      "step": 4885
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8511772155761719,
+      "learning_rate": 9.6214871662432e-05,
+      "loss": 2.5543,
+      "step": 4890
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9300504326820374,
+      "learning_rate": 9.620706689981232e-05,
+      "loss": 2.5291,
+      "step": 4895
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8304131627082825,
+      "learning_rate": 9.619925441616354e-05,
+      "loss": 2.6192,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.919185996055603,
+      "learning_rate": 9.619143421279112e-05,
+      "loss": 2.549,
+      "step": 4905
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8824974894523621,
+      "learning_rate": 9.618360629100176e-05,
+      "loss": 2.5356,
+      "step": 4910
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8055275082588196,
+      "learning_rate": 9.61757706521035e-05,
+      "loss": 2.5401,
+      "step": 4915
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8731273412704468,
+      "learning_rate": 9.616792729740564e-05,
+      "loss": 2.5297,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.854189932346344,
+      "learning_rate": 9.616007622821877e-05,
+      "loss": 2.5158,
+      "step": 4925
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9523569345474243,
+      "learning_rate": 9.615221744585476e-05,
+      "loss": 2.5233,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.856011152267456,
+      "learning_rate": 9.614435095162679e-05,
+      "loss": 2.5444,
+      "step": 4935
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.837837815284729,
+      "learning_rate": 9.613647674684936e-05,
+      "loss": 2.4838,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8889691829681396,
+      "learning_rate": 9.612859483283818e-05,
+      "loss": 2.5201,
+      "step": 4945
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8023569583892822,
+      "learning_rate": 9.612070521091029e-05,
+      "loss": 2.5872,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.817022442817688,
+      "learning_rate": 9.611280788238403e-05,
+      "loss": 2.6694,
+      "step": 4955
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8649120926856995,
+      "learning_rate": 9.610490284857901e-05,
+      "loss": 2.5821,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8711997270584106,
+      "learning_rate": 9.609699011081613e-05,
+      "loss": 2.585,
+      "step": 4965
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8163708448410034,
+      "learning_rate": 9.608906967041757e-05,
+      "loss": 2.57,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8397497534751892,
+      "learning_rate": 9.608114152870684e-05,
+      "loss": 2.503,
+      "step": 4975
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8428658843040466,
+      "learning_rate": 9.607320568700866e-05,
+      "loss": 2.5595,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8402503132820129,
+      "learning_rate": 9.606526214664911e-05,
+      "loss": 2.584,
+      "step": 4985
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8311858773231506,
+      "learning_rate": 9.605731090895551e-05,
+      "loss": 2.5462,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.86878901720047,
+      "learning_rate": 9.604935197525649e-05,
+      "loss": 2.4943,
+      "step": 4995
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8355313539505005,
+      "learning_rate": 9.604138534688196e-05,
+      "loss": 2.6392,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9247829914093018,
+      "learning_rate": 9.60334110251631e-05,
+      "loss": 2.5364,
+      "step": 5005
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.786386251449585,
+      "learning_rate": 9.602542901143241e-05,
+      "loss": 2.49,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8346070051193237,
+      "learning_rate": 9.601743930702363e-05,
+      "loss": 2.515,
+      "step": 5015
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8351532816886902,
+      "learning_rate": 9.600944191327184e-05,
+      "loss": 2.658,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.861367404460907,
+      "learning_rate": 9.600143683151335e-05,
+      "loss": 2.5745,
+      "step": 5025
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9311974048614502,
+      "learning_rate": 9.59934240630858e-05,
+      "loss": 2.3812,
+      "step": 5030
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9864075779914856,
+      "learning_rate": 9.598540360932806e-05,
+      "loss": 2.6545,
+      "step": 5035
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8862329125404358,
+      "learning_rate": 9.597737547158035e-05,
+      "loss": 2.6021,
+      "step": 5040
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8298649787902832,
+      "learning_rate": 9.596933965118413e-05,
+      "loss": 2.5961,
+      "step": 5045
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8990307450294495,
+      "learning_rate": 9.596129614948217e-05,
+      "loss": 2.5433,
+      "step": 5050
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8748336434364319,
+      "learning_rate": 9.595324496781849e-05,
+      "loss": 2.5281,
+      "step": 5055
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8545122742652893,
+      "learning_rate": 9.594518610753841e-05,
+      "loss": 2.5098,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8890294432640076,
+      "learning_rate": 9.593711956998855e-05,
+      "loss": 2.6042,
+      "step": 5065
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8146799802780151,
+      "learning_rate": 9.59290453565168e-05,
+      "loss": 2.55,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8873039484024048,
+      "learning_rate": 9.592096346847232e-05,
+      "loss": 2.5832,
+      "step": 5075
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8284848928451538,
+      "learning_rate": 9.591287390720555e-05,
+      "loss": 2.5372,
+      "step": 5080
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8028897643089294,
+      "learning_rate": 9.590477667406828e-05,
+      "loss": 2.5178,
+      "step": 5085
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8129159808158875,
+      "learning_rate": 9.589667177041347e-05,
+      "loss": 2.4817,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9105582237243652,
+      "learning_rate": 9.588855919759544e-05,
+      "loss": 2.5446,
+      "step": 5095
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8480199575424194,
+      "learning_rate": 9.588043895696978e-05,
+      "loss": 2.5992,
+      "step": 5100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8487927317619324,
+      "learning_rate": 9.587231104989334e-05,
+      "loss": 2.5023,
+      "step": 5105
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8280501961708069,
+      "learning_rate": 9.586417547772426e-05,
+      "loss": 2.5985,
+      "step": 5110
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8131060600280762,
+      "learning_rate": 9.585603224182196e-05,
+      "loss": 2.5782,
+      "step": 5115
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9216845035552979,
+      "learning_rate": 9.584788134354718e-05,
+      "loss": 2.5754,
+      "step": 5120
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8948925733566284,
+      "learning_rate": 9.583972278426186e-05,
+      "loss": 2.5467,
+      "step": 5125
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9335658550262451,
+      "learning_rate": 9.583155656532929e-05,
+      "loss": 2.5987,
+      "step": 5130
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.855377197265625,
+      "learning_rate": 9.582338268811402e-05,
+      "loss": 2.5197,
+      "step": 5135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8376838564872742,
+      "learning_rate": 9.581520115398185e-05,
+      "loss": 2.5988,
+      "step": 5140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.821534276008606,
+      "learning_rate": 9.580701196429991e-05,
+      "loss": 2.5337,
+      "step": 5145
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9049367308616638,
+      "learning_rate": 9.579881512043655e-05,
+      "loss": 2.4811,
+      "step": 5150
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8443823456764221,
+      "learning_rate": 9.579061062376146e-05,
+      "loss": 2.5282,
+      "step": 5155
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8637499809265137,
+      "learning_rate": 9.578239847564559e-05,
+      "loss": 2.5045,
+      "step": 5160
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8997322916984558,
+      "learning_rate": 9.577417867746113e-05,
+      "loss": 2.4567,
+      "step": 5165
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8250539898872375,
+      "learning_rate": 9.576595123058158e-05,
+      "loss": 2.6141,
+      "step": 5170
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8978786468505859,
+      "learning_rate": 9.575771613638175e-05,
+      "loss": 2.5219,
+      "step": 5175
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8355889916419983,
+      "learning_rate": 9.574947339623765e-05,
+      "loss": 2.5735,
+      "step": 5180
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8650621771812439,
+      "learning_rate": 9.574122301152665e-05,
+      "loss": 2.4939,
+      "step": 5185
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7968794107437134,
+      "learning_rate": 9.573296498362733e-05,
+      "loss": 2.5174,
+      "step": 5190
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.835128664970398,
+      "learning_rate": 9.572469931391959e-05,
+      "loss": 2.4664,
+      "step": 5195
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7810455560684204,
+      "learning_rate": 9.571642600378459e-05,
+      "loss": 2.5401,
+      "step": 5200
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9348500967025757,
+      "learning_rate": 9.570814505460475e-05,
+      "loss": 2.5762,
+      "step": 5205
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8303199410438538,
+      "learning_rate": 9.569985646776382e-05,
+      "loss": 2.6497,
+      "step": 5210
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8318626880645752,
+      "learning_rate": 9.569156024464677e-05,
+      "loss": 2.5392,
+      "step": 5215
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7986493706703186,
+      "learning_rate": 9.568325638663989e-05,
+      "loss": 2.5787,
+      "step": 5220
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.072327971458435,
+      "learning_rate": 9.567494489513068e-05,
+      "loss": 2.5907,
+      "step": 5225
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7918429374694824,
+      "learning_rate": 9.5666625771508e-05,
+      "loss": 2.6192,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3978387117385864,
+      "learning_rate": 9.565829901716194e-05,
+      "loss": 2.5824,
+      "step": 5235
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8879933953285217,
+      "learning_rate": 9.564996463348384e-05,
+      "loss": 2.4868,
+      "step": 5240
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8487280607223511,
+      "learning_rate": 9.564162262186638e-05,
+      "loss": 2.5147,
+      "step": 5245
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.83644038438797,
+      "learning_rate": 9.563327298370347e-05,
+      "loss": 2.5149,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8397255539894104,
+      "learning_rate": 9.56249157203903e-05,
+      "loss": 2.5266,
+      "step": 5255
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.865634024143219,
+      "learning_rate": 9.561655083332333e-05,
+      "loss": 2.517,
+      "step": 5260
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.855461597442627,
+      "learning_rate": 9.56081783239003e-05,
+      "loss": 2.5692,
+      "step": 5265
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8657575845718384,
+      "learning_rate": 9.559979819352024e-05,
+      "loss": 2.5369,
+      "step": 5270
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.317520260810852,
+      "learning_rate": 9.559141044358342e-05,
+      "loss": 2.5019,
+      "step": 5275
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8395318984985352,
+      "learning_rate": 9.558301507549143e-05,
+      "loss": 2.5625,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8784891366958618,
+      "learning_rate": 9.557461209064708e-05,
+      "loss": 2.5085,
+      "step": 5285
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7839368581771851,
+      "learning_rate": 9.556620149045447e-05,
+      "loss": 2.5198,
+      "step": 5290
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8340033888816833,
+      "learning_rate": 9.555778327631903e-05,
+      "loss": 2.4993,
+      "step": 5295
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7913752198219299,
+      "learning_rate": 9.554935744964734e-05,
+      "loss": 2.5455,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8236285448074341,
+      "learning_rate": 9.55409240118474e-05,
+      "loss": 2.5467,
+      "step": 5305
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9026365876197815,
+      "learning_rate": 9.553248296432835e-05,
+      "loss": 2.6252,
+      "step": 5310
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8163199424743652,
+      "learning_rate": 9.552403430850068e-05,
+      "loss": 2.5245,
+      "step": 5315
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8554624915122986,
+      "learning_rate": 9.551557804577612e-05,
+      "loss": 2.4677,
+      "step": 5320
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9797208309173584,
+      "learning_rate": 9.550711417756767e-05,
+      "loss": 2.5746,
+      "step": 5325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8176595568656921,
+      "learning_rate": 9.549864270528965e-05,
+      "loss": 2.5455,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8998626470565796,
+      "learning_rate": 9.549016363035758e-05,
+      "loss": 2.5405,
+      "step": 5335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8350285291671753,
+      "learning_rate": 9.548167695418828e-05,
+      "loss": 2.5523,
+      "step": 5340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8097554445266724,
+      "learning_rate": 9.547318267819986e-05,
+      "loss": 2.499,
+      "step": 5345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.848049521446228,
+      "learning_rate": 9.546468080381167e-05,
+      "loss": 2.5055,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8338345885276794,
+      "learning_rate": 9.545617133244436e-05,
+      "loss": 2.5705,
+      "step": 5355
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8206494450569153,
+      "learning_rate": 9.544765426551981e-05,
+      "loss": 2.5566,
+      "step": 5360
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9236202836036682,
+      "learning_rate": 9.543912960446117e-05,
+      "loss": 2.5447,
+      "step": 5365
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8098321557044983,
+      "learning_rate": 9.543059735069293e-05,
+      "loss": 2.6103,
+      "step": 5370
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8408270478248596,
+      "learning_rate": 9.542205750564077e-05,
+      "loss": 2.4778,
+      "step": 5375
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8300800919532776,
+      "learning_rate": 9.541351007073167e-05,
+      "loss": 2.5336,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8288559913635254,
+      "learning_rate": 9.540495504739387e-05,
+      "loss": 2.4927,
+      "step": 5385
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8477457761764526,
+      "learning_rate": 9.53963924370569e-05,
+      "loss": 2.4736,
+      "step": 5390
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.85541832447052,
+      "learning_rate": 9.538782224115151e-05,
+      "loss": 2.5475,
+      "step": 5395
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8414666056632996,
+      "learning_rate": 9.537924446110976e-05,
+      "loss": 2.5914,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8404938578605652,
+      "learning_rate": 9.537065909836497e-05,
+      "loss": 2.5006,
+      "step": 5405
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8333709836006165,
+      "learning_rate": 9.536206615435174e-05,
+      "loss": 2.5423,
+      "step": 5410
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8228434324264526,
+      "learning_rate": 9.535346563050591e-05,
+      "loss": 2.5268,
+      "step": 5415
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8130061626434326,
+      "learning_rate": 9.534485752826456e-05,
+      "loss": 2.536,
+      "step": 5420
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8475875854492188,
+      "learning_rate": 9.53362418490661e-05,
+      "loss": 2.6006,
+      "step": 5425
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8787614107131958,
+      "learning_rate": 9.53276185943502e-05,
+      "loss": 2.5887,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7820987105369568,
+      "learning_rate": 9.531898776555773e-05,
+      "loss": 2.46,
+      "step": 5435
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8649749159812927,
+      "learning_rate": 9.531034936413091e-05,
+      "loss": 2.6428,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8620099425315857,
+      "learning_rate": 9.530170339151314e-05,
+      "loss": 2.5249,
+      "step": 5445
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.842795193195343,
+      "learning_rate": 9.529304984914917e-05,
+      "loss": 2.609,
+      "step": 5450
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8269608616828918,
+      "learning_rate": 9.528438873848496e-05,
+      "loss": 2.5465,
+      "step": 5455
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8167161345481873,
+      "learning_rate": 9.527572006096773e-05,
+      "loss": 2.6045,
+      "step": 5460
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9392737746238708,
+      "learning_rate": 9.526704381804601e-05,
+      "loss": 2.6612,
+      "step": 5465
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8064609169960022,
+      "learning_rate": 9.525836001116957e-05,
+      "loss": 2.5031,
+      "step": 5470
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.881822407245636,
+      "learning_rate": 9.524966864178943e-05,
+      "loss": 2.4596,
+      "step": 5475
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.835156261920929,
+      "learning_rate": 9.524096971135789e-05,
+      "loss": 2.4937,
+      "step": 5480
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8730606436729431,
+      "learning_rate": 9.52322632213285e-05,
+      "loss": 2.5675,
+      "step": 5485
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8560152649879456,
+      "learning_rate": 9.52235491731561e-05,
+      "loss": 2.547,
+      "step": 5490
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8217793107032776,
+      "learning_rate": 9.521482756829675e-05,
+      "loss": 2.5065,
+      "step": 5495
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8012948036193848,
+      "learning_rate": 9.520609840820782e-05,
+      "loss": 2.5431,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7998265027999878,
+      "learning_rate": 9.51973616943479e-05,
+      "loss": 2.4933,
+      "step": 5505
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8235401511192322,
+      "learning_rate": 9.518861742817689e-05,
+      "loss": 2.5527,
+      "step": 5510
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7956685423851013,
+      "learning_rate": 9.51798656111559e-05,
+      "loss": 2.5174,
+      "step": 5515
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7800649404525757,
+      "learning_rate": 9.517110624474733e-05,
+      "loss": 2.5969,
+      "step": 5520
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8804996013641357,
+      "learning_rate": 9.516233933041485e-05,
+      "loss": 2.5343,
+      "step": 5525
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8091599941253662,
+      "learning_rate": 9.515356486962337e-05,
+      "loss": 2.5541,
+      "step": 5530
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9333968162536621,
+      "learning_rate": 9.514478286383907e-05,
+      "loss": 2.4303,
+      "step": 5535
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8042595982551575,
+      "learning_rate": 9.513599331452939e-05,
+      "loss": 2.5206,
+      "step": 5540
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9287793636322021,
+      "learning_rate": 9.512719622316304e-05,
+      "loss": 2.482,
+      "step": 5545
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8644748330116272,
+      "learning_rate": 9.511839159120997e-05,
+      "loss": 2.552,
+      "step": 5550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2345731258392334,
+      "learning_rate": 9.51095794201414e-05,
+      "loss": 2.6796,
+      "step": 5555
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8384237289428711,
+      "learning_rate": 9.510075971142983e-05,
+      "loss": 2.6071,
+      "step": 5560
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8860867619514465,
+      "learning_rate": 9.509193246654896e-05,
+      "loss": 2.505,
+      "step": 5565
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9134503602981567,
+      "learning_rate": 9.508309768697384e-05,
+      "loss": 2.5178,
+      "step": 5570
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9127622842788696,
+      "learning_rate": 9.507425537418068e-05,
+      "loss": 2.5407,
+      "step": 5575
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9157562255859375,
+      "learning_rate": 9.506540552964704e-05,
+      "loss": 2.499,
+      "step": 5580
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8791778683662415,
+      "learning_rate": 9.505654815485168e-05,
+      "loss": 2.5127,
+      "step": 5585
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9766604900360107,
+      "learning_rate": 9.504768325127464e-05,
+      "loss": 2.5647,
+      "step": 5590
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8459810614585876,
+      "learning_rate": 9.503881082039719e-05,
+      "loss": 2.5528,
+      "step": 5595
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8299350738525391,
+      "learning_rate": 9.50299308637019e-05,
+      "loss": 2.5733,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8838971853256226,
+      "learning_rate": 9.502104338267257e-05,
+      "loss": 2.5487,
+      "step": 5605
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8144720196723938,
+      "learning_rate": 9.501214837879428e-05,
+      "loss": 2.4973,
+      "step": 5610
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8654022812843323,
+      "learning_rate": 9.500324585355333e-05,
+      "loss": 2.4937,
+      "step": 5615
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8445335030555725,
+      "learning_rate": 9.499433580843732e-05,
+      "loss": 2.5028,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7984415292739868,
+      "learning_rate": 9.498541824493506e-05,
+      "loss": 2.5314,
+      "step": 5625
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.887332022190094,
+      "learning_rate": 9.497649316453667e-05,
+      "loss": 2.5019,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9232797026634216,
+      "learning_rate": 9.496756056873348e-05,
+      "loss": 2.581,
+      "step": 5635
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8549322485923767,
+      "learning_rate": 9.49586204590181e-05,
+      "loss": 2.4772,
+      "step": 5640
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9022164344787598,
+      "learning_rate": 9.494967283688438e-05,
+      "loss": 2.4924,
+      "step": 5645
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7918215990066528,
+      "learning_rate": 9.494071770382745e-05,
+      "loss": 2.5032,
+      "step": 5650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8009668588638306,
+      "learning_rate": 9.493175506134367e-05,
+      "loss": 2.4126,
+      "step": 5655
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8071606755256653,
+      "learning_rate": 9.492278491093067e-05,
+      "loss": 2.5662,
+      "step": 5660
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8340517282485962,
+      "learning_rate": 9.491380725408731e-05,
+      "loss": 2.4342,
+      "step": 5665
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.882816731929779,
+      "learning_rate": 9.490482209231374e-05,
+      "loss": 2.4715,
+      "step": 5670
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8274314999580383,
+      "learning_rate": 9.489582942711135e-05,
+      "loss": 2.5624,
+      "step": 5675
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8811088800430298,
+      "learning_rate": 9.488682925998277e-05,
+      "loss": 2.4255,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.811182975769043,
+      "learning_rate": 9.48778215924319e-05,
+      "loss": 2.5682,
+      "step": 5685
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8752612471580505,
+      "learning_rate": 9.48688064259639e-05,
+      "loss": 2.56,
+      "step": 5690
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8651951551437378,
+      "learning_rate": 9.485978376208513e-05,
+      "loss": 2.4318,
+      "step": 5695
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7639158368110657,
+      "learning_rate": 9.485075360230328e-05,
+      "loss": 2.5785,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8580864071846008,
+      "learning_rate": 9.484171594812728e-05,
+      "loss": 2.562,
+      "step": 5705
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8065708875656128,
+      "learning_rate": 9.483267080106723e-05,
+      "loss": 2.4362,
+      "step": 5710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8630844354629517,
+      "learning_rate": 9.482361816263456e-05,
+      "loss": 2.5999,
+      "step": 5715
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8446179628372192,
+      "learning_rate": 9.481455803434195e-05,
+      "loss": 2.5417,
+      "step": 5720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9255685806274414,
+      "learning_rate": 9.48054904177033e-05,
+      "loss": 2.5215,
+      "step": 5725
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8238320350646973,
+      "learning_rate": 9.479641531423377e-05,
+      "loss": 2.5117,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8088456988334656,
+      "learning_rate": 9.47873327254498e-05,
+      "loss": 2.4002,
+      "step": 5735
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.874265730381012,
+      "learning_rate": 9.477824265286904e-05,
+      "loss": 2.5034,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8569507002830505,
+      "learning_rate": 9.476914509801039e-05,
+      "loss": 2.5784,
+      "step": 5745
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8233985900878906,
+      "learning_rate": 9.476004006239406e-05,
+      "loss": 2.5091,
+      "step": 5750
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8448342084884644,
+      "learning_rate": 9.475092754754142e-05,
+      "loss": 2.5758,
+      "step": 5755
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8199537992477417,
+      "learning_rate": 9.474180755497517e-05,
+      "loss": 2.5378,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9235616326332092,
+      "learning_rate": 9.473268008621922e-05,
+      "loss": 2.4539,
+      "step": 5765
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8268235921859741,
+      "learning_rate": 9.472354514279874e-05,
+      "loss": 2.4443,
+      "step": 5770
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8281229138374329,
+      "learning_rate": 9.471440272624012e-05,
+      "loss": 2.4912,
+      "step": 5775
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8137543201446533,
+      "learning_rate": 9.470525283807105e-05,
+      "loss": 2.4864,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8813849687576294,
+      "learning_rate": 9.469609547982044e-05,
+      "loss": 2.4846,
+      "step": 5785
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8242172002792358,
+      "learning_rate": 9.468693065301843e-05,
+      "loss": 2.4798,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8066756129264832,
+      "learning_rate": 9.467775835919646e-05,
+      "loss": 2.6005,
+      "step": 5795
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8584503531455994,
+      "learning_rate": 9.466857859988715e-05,
+      "loss": 2.4587,
+      "step": 5800
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9053176641464233,
+      "learning_rate": 9.465939137662444e-05,
+      "loss": 2.4792,
+      "step": 5805
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7943493127822876,
+      "learning_rate": 9.465019669094346e-05,
+      "loss": 2.5184,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9027978777885437,
+      "learning_rate": 9.464099454438062e-05,
+      "loss": 2.5248,
+      "step": 5815
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9596348404884338,
+      "learning_rate": 9.463178493847355e-05,
+      "loss": 2.5249,
+      "step": 5820
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8494065403938293,
+      "learning_rate": 9.462256787476113e-05,
+      "loss": 2.5189,
+      "step": 5825
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8080886602401733,
+      "learning_rate": 9.461334335478354e-05,
+      "loss": 2.5471,
+      "step": 5830
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8952863812446594,
+      "learning_rate": 9.460411138008214e-05,
+      "loss": 2.5034,
+      "step": 5835
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8741773962974548,
+      "learning_rate": 9.459487195219954e-05,
+      "loss": 2.4811,
+      "step": 5840
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8826447129249573,
+      "learning_rate": 9.458562507267963e-05,
+      "loss": 2.5462,
+      "step": 5845
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8280957937240601,
+      "learning_rate": 9.457637074306754e-05,
+      "loss": 2.4601,
+      "step": 5850
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1507675647735596,
+      "learning_rate": 9.456710896490961e-05,
+      "loss": 2.4643,
+      "step": 5855
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9381924867630005,
+      "learning_rate": 9.455783973975348e-05,
+      "loss": 2.4036,
+      "step": 5860
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8356437087059021,
+      "learning_rate": 9.454856306914797e-05,
+      "loss": 2.4806,
+      "step": 5865
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8351025581359863,
+      "learning_rate": 9.453927895464321e-05,
+      "loss": 2.5296,
+      "step": 5870
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8108028173446655,
+      "learning_rate": 9.452998739779051e-05,
+      "loss": 2.5027,
+      "step": 5875
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8971517086029053,
+      "learning_rate": 9.452068840014246e-05,
+      "loss": 2.4813,
+      "step": 5880
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7947912216186523,
+      "learning_rate": 9.451138196325291e-05,
+      "loss": 2.3876,
+      "step": 5885
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8771917819976807,
+      "learning_rate": 9.45020680886769e-05,
+      "loss": 2.3963,
+      "step": 5890
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8963824510574341,
+      "learning_rate": 9.449274677797078e-05,
+      "loss": 2.6016,
+      "step": 5895
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8207738995552063,
+      "learning_rate": 9.448341803269208e-05,
+      "loss": 2.5221,
+      "step": 5900
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8410871028900146,
+      "learning_rate": 9.44740818543996e-05,
+      "loss": 2.4755,
+      "step": 5905
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8241736888885498,
+      "learning_rate": 9.44647382446534e-05,
+      "loss": 2.6139,
+      "step": 5910
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9366174936294556,
+      "learning_rate": 9.445538720501474e-05,
+      "loss": 2.5018,
+      "step": 5915
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8756698966026306,
+      "learning_rate": 9.444602873704616e-05,
+      "loss": 2.611,
+      "step": 5920
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8006719946861267,
+      "learning_rate": 9.443666284231141e-05,
+      "loss": 2.5409,
+      "step": 5925
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8202231526374817,
+      "learning_rate": 9.44272895223755e-05,
+      "loss": 2.509,
+      "step": 5930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8358700275421143,
+      "learning_rate": 9.441790877880469e-05,
+      "loss": 2.5164,
+      "step": 5935
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8649426102638245,
+      "learning_rate": 9.440852061316645e-05,
+      "loss": 2.453,
+      "step": 5940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8377036452293396,
+      "learning_rate": 9.439912502702952e-05,
+      "loss": 2.4534,
+      "step": 5945
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8739152550697327,
+      "learning_rate": 9.438972202196387e-05,
+      "loss": 2.4939,
+      "step": 5950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8303829431533813,
+      "learning_rate": 9.43803115995407e-05,
+      "loss": 2.4998,
+      "step": 5955
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8236222863197327,
+      "learning_rate": 9.437089376133246e-05,
+      "loss": 2.5582,
+      "step": 5960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8027788400650024,
+      "learning_rate": 9.436146850891283e-05,
+      "loss": 2.5445,
+      "step": 5965
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9776527881622314,
+      "learning_rate": 9.435203584385673e-05,
+      "loss": 2.4772,
+      "step": 5970
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8954410552978516,
+      "learning_rate": 9.434259576774034e-05,
+      "loss": 2.5036,
+      "step": 5975
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8393428921699524,
+      "learning_rate": 9.433314828214103e-05,
+      "loss": 2.5546,
+      "step": 5980
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9841406941413879,
+      "learning_rate": 9.43236933886375e-05,
+      "loss": 2.5901,
+      "step": 5985
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8082411885261536,
+      "learning_rate": 9.431423108880957e-05,
+      "loss": 2.5326,
+      "step": 5990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.829982340335846,
+      "learning_rate": 9.430476138423838e-05,
+      "loss": 2.4184,
+      "step": 5995
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9179960489273071,
+      "learning_rate": 9.429528427650627e-05,
+      "loss": 2.5224,
+      "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
+}