diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.2076465855114595,
+  "eval_steps": 500,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 22.853303909301758,
+      "learning_rate": 5e-06,
+      "loss": 12.1145,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.43573570251465,
+      "learning_rate": 1e-05,
+      "loss": 11.9676,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 19.94394874572754,
+      "learning_rate": 1.5e-05,
+      "loss": 11.5155,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.082651138305664,
+      "learning_rate": 2e-05,
+      "loss": 10.7436,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.632811546325684,
+      "learning_rate": 2.5e-05,
+      "loss": 9.9719,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.842703819274902,
+      "learning_rate": 3e-05,
+      "loss": 9.4029,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.429098129272461,
+      "learning_rate": 3.5e-05,
+      "loss": 8.9413,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.7367329597473145,
+      "learning_rate": 4e-05,
+      "loss": 8.5153,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8989975452423096,
+      "learning_rate": 4.5e-05,
+      "loss": 8.2129,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.700549602508545,
+      "learning_rate": 5e-05,
+      "loss": 7.9862,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4647518396377563,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 7.9207,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.956396758556366,
+      "learning_rate": 6e-05,
+      "loss": 7.8615,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3899364471435547,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 7.8017,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3081855773925781,
+      "learning_rate": 7e-05,
+      "loss": 7.7464,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0955415964126587,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 7.7372,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2173418998718262,
+      "learning_rate": 8e-05,
+      "loss": 7.6077,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0822601318359375,
+      "learning_rate": 8.5e-05,
+      "loss": 7.4808,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3264280557632446,
+      "learning_rate": 9e-05,
+      "loss": 7.3525,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0716509819030762,
+      "learning_rate": 9.5e-05,
+      "loss": 7.208,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5171961784362793,
+      "learning_rate": 0.0001,
+      "loss": 7.006,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.525022268295288,
+      "learning_rate": 9.999999582259295e-05,
+      "loss": 6.7849,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.4417152404785156,
+      "learning_rate": 9.999998329037248e-05,
+      "loss": 6.6431,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5786519050598145,
+      "learning_rate": 9.999996240334068e-05,
+      "loss": 6.4465,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0681309700012207,
+      "learning_rate": 9.999993316150106e-05,
+      "loss": 6.3265,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.026639938354492,
+      "learning_rate": 9.999989556485852e-05,
+      "loss": 6.2183,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6160576343536377,
+      "learning_rate": 9.999984961341928e-05,
+      "loss": 6.0676,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.067312479019165,
+      "learning_rate": 9.999979530719108e-05,
+      "loss": 5.9099,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.035238027572632,
+      "learning_rate": 9.999973264618297e-05,
+      "loss": 5.7629,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.245169162750244,
+      "learning_rate": 9.999966163040541e-05,
+      "loss": 5.622,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.640432119369507,
+      "learning_rate": 9.99995822598703e-05,
+      "loss": 5.5972,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0888869762420654,
+      "learning_rate": 9.999949453459088e-05,
+      "loss": 5.4544,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.006242275238037,
+      "learning_rate": 9.999939845458179e-05,
+      "loss": 5.212,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.634506344795227,
+      "learning_rate": 9.999929401985911e-05,
+      "loss": 5.2706,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9156520366668701,
+      "learning_rate": 9.99991812304403e-05,
+      "loss": 5.0865,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.466616153717041,
+      "learning_rate": 9.999906008634416e-05,
+      "loss": 5.1106,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7136033773422241,
+      "learning_rate": 9.9998930587591e-05,
+      "loss": 5.0478,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7738032341003418,
+      "learning_rate": 9.99987927342024e-05,
+      "loss": 4.8517,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7486941814422607,
+      "learning_rate": 9.999864652620143e-05,
+      "loss": 4.7968,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6300857067108154,
+      "learning_rate": 9.999849196361251e-05,
+      "loss": 4.7883,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.781404972076416,
+      "learning_rate": 9.999832904646147e-05,
+      "loss": 4.7904,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1201188564300537,
+      "learning_rate": 9.99981577747755e-05,
+      "loss": 4.6775,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.80289888381958,
+      "learning_rate": 9.999797814858327e-05,
+      "loss": 4.6004,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6210174560546875,
+      "learning_rate": 9.999779016791477e-05,
+      "loss": 4.5648,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5090669393539429,
+      "learning_rate": 9.999759383280141e-05,
+      "loss": 4.5799,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8481578826904297,
+      "learning_rate": 9.9997389143276e-05,
+      "loss": 4.6043,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7592591047286987,
+      "learning_rate": 9.999717609937275e-05,
+      "loss": 4.4139,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6663849353790283,
+      "learning_rate": 9.999695470112725e-05,
+      "loss": 4.3353,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4900798797607422,
+      "learning_rate": 9.99967249485765e-05,
+      "loss": 4.3518,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7587333917617798,
+      "learning_rate": 9.999648684175888e-05,
+      "loss": 4.2212,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.715404748916626,
+      "learning_rate": 9.999624038071418e-05,
+      "loss": 4.2822,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.564717173576355,
+      "learning_rate": 9.99959855654836e-05,
+      "loss": 4.3598,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5458120107650757,
+      "learning_rate": 9.999572239610968e-05,
+      "loss": 4.2799,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7090470790863037,
+      "learning_rate": 9.999545087263645e-05,
+      "loss": 4.3046,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3876731395721436,
+      "learning_rate": 9.999517099510923e-05,
+      "loss": 4.1524,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3959612846374512,
+      "learning_rate": 9.999488276357481e-05,
+      "loss": 4.1775,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4238133430480957,
+      "learning_rate": 9.999458617808137e-05,
+      "loss": 4.1218,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3902766704559326,
+      "learning_rate": 9.999428123867843e-05,
+      "loss": 4.1117,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.559579849243164,
+      "learning_rate": 9.999396794541695e-05,
+      "loss": 4.0118,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7723685503005981,
+      "learning_rate": 9.999364629834931e-05,
+      "loss": 4.1028,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3731063604354858,
+      "learning_rate": 9.999331629752924e-05,
+      "loss": 4.0173,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5322914123535156,
+      "learning_rate": 9.999297794301188e-05,
+      "loss": 3.9811,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4149526357650757,
+      "learning_rate": 9.999263123485376e-05,
+      "loss": 4.0336,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4588673114776611,
+      "learning_rate": 9.999227617311282e-05,
+      "loss": 3.9312,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5408458709716797,
+      "learning_rate": 9.99919127578484e-05,
+      "loss": 4.0194,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3926481008529663,
+      "learning_rate": 9.999154098912122e-05,
+      "loss": 3.9101,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4581048488616943,
+      "learning_rate": 9.999116086699338e-05,
+      "loss": 3.9549,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3182730674743652,
+      "learning_rate": 9.999077239152843e-05,
+      "loss": 3.9235,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1929956674575806,
+      "learning_rate": 9.999037556279127e-05,
+      "loss": 3.9121,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2226181030273438,
+      "learning_rate": 9.998997038084818e-05,
+      "loss": 3.971,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.223927617073059,
+      "learning_rate": 9.998955684576692e-05,
+      "loss": 3.8719,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.295952320098877,
+      "learning_rate": 9.998913495761654e-05,
+      "loss": 3.7629,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.415813684463501,
+      "learning_rate": 9.998870471646757e-05,
+      "loss": 3.9207,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4144867658615112,
+      "learning_rate": 9.99882661223919e-05,
+      "loss": 3.8548,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2493077516555786,
+      "learning_rate": 9.998781917546278e-05,
+      "loss": 3.7518,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2064306735992432,
+      "learning_rate": 9.998736387575493e-05,
+      "loss": 3.8492,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1887041330337524,
+      "learning_rate": 9.998690022334442e-05,
+      "loss": 3.6611,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.199149489402771,
+      "learning_rate": 9.998642821830873e-05,
+      "loss": 3.6549,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2293339967727661,
+      "learning_rate": 9.99859478607267e-05,
+      "loss": 3.7655,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2537128925323486,
+      "learning_rate": 9.998545915067864e-05,
+      "loss": 3.7447,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2480026483535767,
+      "learning_rate": 9.998496208824618e-05,
+      "loss": 3.6993,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.243604302406311,
+      "learning_rate": 9.99844566735124e-05,
+      "loss": 3.8235,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0544401407241821,
+      "learning_rate": 9.998394290656172e-05,
+      "loss": 3.6816,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0956451892852783,
+      "learning_rate": 9.998342078748002e-05,
+      "loss": 3.689,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2277470827102661,
+      "learning_rate": 9.998289031635453e-05,
+      "loss": 3.7383,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2064182758331299,
+      "learning_rate": 9.998235149327389e-05,
+      "loss": 3.6562,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.196374773979187,
+      "learning_rate": 9.998180431832816e-05,
+      "loss": 3.7557,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1541024446487427,
+      "learning_rate": 9.998124879160872e-05,
+      "loss": 3.6842,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1177842617034912,
+      "learning_rate": 9.998068491320844e-05,
+      "loss": 3.5725,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0798002481460571,
+      "learning_rate": 9.998011268322152e-05,
+      "loss": 3.7061,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0827932357788086,
+      "learning_rate": 9.997953210174361e-05,
+      "loss": 3.5582,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1261018514633179,
+      "learning_rate": 9.997894316887167e-05,
+      "loss": 3.5588,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0506893396377563,
+      "learning_rate": 9.997834588470414e-05,
+      "loss": 3.6082,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1189738512039185,
+      "learning_rate": 9.997774024934082e-05,
+      "loss": 3.6589,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2031960487365723,
+      "learning_rate": 9.997712626288292e-05,
+      "loss": 3.6575,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0222091674804688,
+      "learning_rate": 9.997650392543303e-05,
+      "loss": 3.549,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.14531672000885,
+      "learning_rate": 9.997587323709512e-05,
+      "loss": 3.6673,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.097542405128479,
+      "learning_rate": 9.997523419797459e-05,
+      "loss": 3.61,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1833710670471191,
+      "learning_rate": 9.997458680817824e-05,
+      "loss": 3.575,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0786690711975098,
+      "learning_rate": 9.997393106781421e-05,
+      "loss": 3.59,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.235292673110962,
+      "learning_rate": 9.99732669769921e-05,
+      "loss": 3.5331,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9819581508636475,
+      "learning_rate": 9.997259453582286e-05,
+      "loss": 3.4546,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0492055416107178,
+      "learning_rate": 9.997191374441887e-05,
+      "loss": 3.4808,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1332746744155884,
+      "learning_rate": 9.997122460289387e-05,
+      "loss": 3.5452,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9693100452423096,
+      "learning_rate": 9.997052711136301e-05,
+      "loss": 3.4994,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0119966268539429,
+      "learning_rate": 9.996982126994287e-05,
+      "loss": 3.4818,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6381217241287231,
+      "learning_rate": 9.996910707875136e-05,
+      "loss": 3.5326,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0752264261245728,
+      "learning_rate": 9.996838453790784e-05,
+      "loss": 3.5052,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9675344228744507,
+      "learning_rate": 9.996765364753302e-05,
+      "loss": 3.4253,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0243892669677734,
+      "learning_rate": 9.996691440774906e-05,
+      "loss": 3.4922,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9779052734375,
+      "learning_rate": 9.996616681867945e-05,
+      "loss": 3.4225,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.985131025314331,
+      "learning_rate": 9.996541088044913e-05,
+      "loss": 3.4707,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9479859471321106,
+      "learning_rate": 9.996464659318442e-05,
+      "loss": 3.4492,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0502394437789917,
+      "learning_rate": 9.996387395701302e-05,
+      "loss": 3.4477,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9307985305786133,
+      "learning_rate": 9.996309297206403e-05,
+      "loss": 3.5132,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.967714250087738,
+      "learning_rate": 9.996230363846797e-05,
+      "loss": 3.5393,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.960995614528656,
+      "learning_rate": 9.996150595635671e-05,
+      "loss": 3.434,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8325942754745483,
+      "learning_rate": 9.996069992586355e-05,
+      "loss": 3.4219,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9604061841964722,
+      "learning_rate": 9.995988554712318e-05,
+      "loss": 3.4472,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8947728872299194,
+      "learning_rate": 9.995906282027166e-05,
+      "loss": 3.4022,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9560570120811462,
+      "learning_rate": 9.995823174544649e-05,
+      "loss": 3.4306,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8619065880775452,
+      "learning_rate": 9.995739232278654e-05,
+      "loss": 3.4754,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.920760989189148,
+      "learning_rate": 9.995654455243204e-05,
+      "loss": 3.4462,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8838915824890137,
+      "learning_rate": 9.995568843452468e-05,
+      "loss": 3.3967,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9878486394882202,
+      "learning_rate": 9.99548239692075e-05,
+      "loss": 3.4574,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8555717468261719,
+      "learning_rate": 9.995395115662498e-05,
+      "loss": 3.3,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9583438634872437,
+      "learning_rate": 9.995306999692293e-05,
+      "loss": 3.4206,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8627120852470398,
+      "learning_rate": 9.995218049024858e-05,
+      "loss": 3.4418,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0740196704864502,
+      "learning_rate": 9.995128263675058e-05,
+      "loss": 3.3928,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9919221997261047,
+      "learning_rate": 9.995037643657899e-05,
+      "loss": 3.4763,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9224211573600769,
+      "learning_rate": 9.994946188988516e-05,
+      "loss": 3.3116,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9968213438987732,
+      "learning_rate": 9.994853899682197e-05,
+      "loss": 3.4198,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8939771056175232,
+      "learning_rate": 9.994760775754362e-05,
+      "loss": 3.3994,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8807076811790466,
+      "learning_rate": 9.994666817220571e-05,
+      "loss": 3.3369,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8917527198791504,
+      "learning_rate": 9.994572024096521e-05,
+      "loss": 3.333,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8234641551971436,
+      "learning_rate": 9.994476396398055e-05,
+      "loss": 3.3612,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9144483804702759,
+      "learning_rate": 9.994379934141153e-05,
+      "loss": 3.2792,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9296488761901855,
+      "learning_rate": 9.994282637341932e-05,
+      "loss": 3.3628,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9157776832580566,
+      "learning_rate": 9.994184506016651e-05,
+      "loss": 3.3198,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9763591289520264,
+      "learning_rate": 9.994085540181703e-05,
+      "loss": 3.3131,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8278407454490662,
+      "learning_rate": 9.99398573985363e-05,
+      "loss": 3.2838,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8521916270256042,
+      "learning_rate": 9.993885105049107e-05,
+      "loss": 3.3399,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.845905601978302,
+      "learning_rate": 9.993783635784947e-05,
+      "loss": 3.3636,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9214987754821777,
+      "learning_rate": 9.993681332078108e-05,
+      "loss": 3.3432,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8530499339103699,
+      "learning_rate": 9.993578193945685e-05,
+      "loss": 3.2439,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.850612998008728,
+      "learning_rate": 9.99347422140491e-05,
+      "loss": 3.3615,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8508695363998413,
+      "learning_rate": 9.993369414473157e-05,
+      "loss": 3.3102,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.876903772354126,
+      "learning_rate": 9.993263773167939e-05,
+      "loss": 3.3395,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.807232677936554,
+      "learning_rate": 9.993157297506907e-05,
+      "loss": 3.3262,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8869733810424805,
+      "learning_rate": 9.993049987507856e-05,
+      "loss": 3.2959,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9176018238067627,
+      "learning_rate": 9.992941843188715e-05,
+      "loss": 3.3596,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8725350499153137,
+      "learning_rate": 9.992832864567554e-05,
+      "loss": 3.2677,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8401938676834106,
+      "learning_rate": 9.992723051662584e-05,
+      "loss": 3.2402,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8018854856491089,
+      "learning_rate": 9.992612404492153e-05,
+      "loss": 3.251,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7587234377861023,
+      "learning_rate": 9.992500923074752e-05,
+      "loss": 3.2547,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8637373447418213,
+      "learning_rate": 9.992388607429006e-05,
+      "loss": 3.2784,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8591487407684326,
+      "learning_rate": 9.992275457573685e-05,
+      "loss": 3.1925,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9411582350730896,
+      "learning_rate": 9.992161473527695e-05,
+      "loss": 3.3802,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8033069968223572,
+      "learning_rate": 9.992046655310083e-05,
+      "loss": 3.2041,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8022111058235168,
+      "learning_rate": 9.991931002940034e-05,
+      "loss": 3.2322,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8955051302909851,
+      "learning_rate": 9.991814516436873e-05,
+      "loss": 3.2662,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.801121711730957,
+      "learning_rate": 9.991697195820065e-05,
+      "loss": 3.2897,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.787140429019928,
+      "learning_rate": 9.991579041109212e-05,
+      "loss": 3.1742,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8072747588157654,
+      "learning_rate": 9.991460052324061e-05,
+      "loss": 3.3023,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8016116619110107,
+      "learning_rate": 9.991340229484491e-05,
+      "loss": 3.2717,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8292195200920105,
+      "learning_rate": 9.991219572610526e-05,
+      "loss": 3.2082,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7627029418945312,
+      "learning_rate": 9.991098081722325e-05,
+      "loss": 3.2383,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7964347004890442,
+      "learning_rate": 9.990975756840193e-05,
+      "loss": 3.2019,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7604963779449463,
+      "learning_rate": 9.990852597984566e-05,
+      "loss": 3.2389,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7262404561042786,
+      "learning_rate": 9.990728605176025e-05,
+      "loss": 3.2274,
+      "step": 845
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7462863922119141,
+      "learning_rate": 9.990603778435288e-05,
+      "loss": 3.2469,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.771310031414032,
+      "learning_rate": 9.990478117783214e-05,
+      "loss": 3.2769,
+      "step": 855
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7902434468269348,
+      "learning_rate": 9.990351623240799e-05,
+      "loss": 3.2449,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8581898808479309,
+      "learning_rate": 9.990224294829181e-05,
+      "loss": 3.188,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8014155030250549,
+      "learning_rate": 9.990096132569636e-05,
+      "loss": 3.1406,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8136110901832581,
+      "learning_rate": 9.98996713648358e-05,
+      "loss": 3.1561,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7171638011932373,
+      "learning_rate": 9.989837306592567e-05,
+      "loss": 3.2144,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7497773766517639,
+      "learning_rate": 9.989706642918291e-05,
+      "loss": 3.2477,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7564408779144287,
+      "learning_rate": 9.989575145482583e-05,
+      "loss": 3.1529,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7447835803031921,
+      "learning_rate": 9.989442814307422e-05,
+      "loss": 3.2437,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7729227542877197,
+      "learning_rate": 9.989309649414914e-05,
+      "loss": 3.0955,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7931673526763916,
+      "learning_rate": 9.989175650827314e-05,
+      "loss": 3.1797,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7773964405059814,
+      "learning_rate": 9.98904081856701e-05,
+      "loss": 3.1585,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7937477827072144,
+      "learning_rate": 9.988905152656534e-05,
+      "loss": 3.1189,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7267899513244629,
+      "learning_rate": 9.988768653118554e-05,
+      "loss": 3.1991,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7704606652259827,
+      "learning_rate": 9.988631319975881e-05,
+      "loss": 3.2208,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7747164368629456,
+      "learning_rate": 9.988493153251459e-05,
+      "loss": 3.1474,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8747826814651489,
+      "learning_rate": 9.988354152968379e-05,
+      "loss": 3.196,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7874693870544434,
+      "learning_rate": 9.988214319149865e-05,
+      "loss": 3.1748,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7538564801216125,
+      "learning_rate": 9.988073651819281e-05,
+      "loss": 3.2051,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7071685791015625,
+      "learning_rate": 9.987932151000137e-05,
+      "loss": 3.2217,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7620505094528198,
+      "learning_rate": 9.987789816716073e-05,
+      "loss": 3.0676,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6933626532554626,
+      "learning_rate": 9.987646648990876e-05,
+      "loss": 3.2543,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7517188191413879,
+      "learning_rate": 9.987502647848466e-05,
+      "loss": 3.1505,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7286583185195923,
+      "learning_rate": 9.987357813312905e-05,
+      "loss": 3.1992,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7523605823516846,
+      "learning_rate": 9.987212145408395e-05,
+      "loss": 3.1227,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8072115182876587,
+      "learning_rate": 9.987065644159278e-05,
+      "loss": 3.2161,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7103545069694519,
+      "learning_rate": 9.986918309590031e-05,
+      "loss": 3.0493,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6771014332771301,
+      "learning_rate": 9.986770141725277e-05,
+      "loss": 3.1441,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6824824810028076,
+      "learning_rate": 9.986621140589772e-05,
+      "loss": 3.1546,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6801164150238037,
+      "learning_rate": 9.986471306208412e-05,
+      "loss": 3.1402,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7732380032539368,
+      "learning_rate": 9.986320638606235e-05,
+      "loss": 3.0469,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7445975542068481,
+      "learning_rate": 9.986169137808419e-05,
+      "loss": 3.0466,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7410663366317749,
+      "learning_rate": 9.986016803840275e-05,
+      "loss": 3.0762,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.730871319770813,
+      "learning_rate": 9.985863636727262e-05,
+      "loss": 3.0699,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7168570160865784,
+      "learning_rate": 9.985709636494971e-05,
+      "loss": 3.0877,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9159319400787354,
+      "learning_rate": 9.985554803169134e-05,
+      "loss": 3.0868,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7057777643203735,
+      "learning_rate": 9.985399136775627e-05,
+      "loss": 3.1236,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8357502818107605,
+      "learning_rate": 9.985242637340458e-05,
+      "loss": 3.0401,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7837785482406616,
+      "learning_rate": 9.985085304889778e-05,
+      "loss": 3.1543,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7093569040298462,
+      "learning_rate": 9.984927139449877e-05,
+      "loss": 3.1902,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8165985941886902,
+      "learning_rate": 9.984768141047184e-05,
+      "loss": 3.1477,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7882426381111145,
+      "learning_rate": 9.984608309708266e-05,
+      "loss": 3.1079,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6674025058746338,
+      "learning_rate": 9.984447645459831e-05,
+      "loss": 3.1376,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6377187371253967,
+      "learning_rate": 9.984286148328728e-05,
+      "loss": 3.0991,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9796004295349121,
+      "learning_rate": 9.984123818341937e-05,
+      "loss": 3.0937,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9811123609542847,
+      "learning_rate": 9.983960655526587e-05,
+      "loss": 3.0612,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6583461761474609,
+      "learning_rate": 9.983796659909941e-05,
+      "loss": 3.1111,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6885143518447876,
+      "learning_rate": 9.983631831519401e-05,
+      "loss": 3.1157,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6264825463294983,
+      "learning_rate": 9.983466170382511e-05,
+      "loss": 3.0442,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7234529256820679,
+      "learning_rate": 9.983299676526949e-05,
+      "loss": 3.0376,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7701594829559326,
+      "learning_rate": 9.98313234998054e-05,
+      "loss": 3.1793,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.721459686756134,
+      "learning_rate": 9.982964190771242e-05,
+      "loss": 3.0806,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7231050729751587,
+      "learning_rate": 9.98279519892715e-05,
+      "loss": 3.1389,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6259997487068176,
+      "learning_rate": 9.982625374476508e-05,
+      "loss": 3.0385,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0630403757095337,
+      "learning_rate": 9.98245471744769e-05,
+      "loss": 3.1913,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6463804244995117,
+      "learning_rate": 9.982283227869211e-05,
+      "loss": 3.0785,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6704946756362915,
+      "learning_rate": 9.982110905769729e-05,
+      "loss": 3.0769,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7171010375022888,
+      "learning_rate": 9.981937751178037e-05,
+      "loss": 3.0616,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6447939276695251,
+      "learning_rate": 9.981763764123067e-05,
+      "loss": 3.0625,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6780021786689758,
+      "learning_rate": 9.981588944633893e-05,
+      "loss": 3.127,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6806884407997131,
+      "learning_rate": 9.981413292739727e-05,
+      "loss": 3.0475,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6623690128326416,
+      "learning_rate": 9.981236808469922e-05,
+      "loss": 3.1024,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6666389107704163,
+      "learning_rate": 9.981059491853964e-05,
+      "loss": 3.0464,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7587738037109375,
+      "learning_rate": 9.980881342921482e-05,
+      "loss": 3.095,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.644595742225647,
+      "learning_rate": 9.980702361702246e-05,
+      "loss": 3.0969,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7200291752815247,
+      "learning_rate": 9.980522548226162e-05,
+      "loss": 3.1315,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6549781560897827,
+      "learning_rate": 9.980341902523279e-05,
+      "loss": 2.938,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7735085487365723,
+      "learning_rate": 9.980160424623776e-05,
+      "loss": 3.0731,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6460607647895813,
+      "learning_rate": 9.979978114557984e-05,
+      "loss": 3.0199,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6810137033462524,
+      "learning_rate": 9.979794972356362e-05,
+      "loss": 3.1603,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8146769404411316,
+      "learning_rate": 9.979610998049516e-05,
+      "loss": 3.0765,
+      "step": 1205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6214923858642578,
+      "learning_rate": 9.979426191668182e-05,
+      "loss": 3.0428,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.742673397064209,
+      "learning_rate": 9.979240553243246e-05,
+      "loss": 3.1155,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7062981128692627,
+      "learning_rate": 9.979054082805726e-05,
+      "loss": 3.1619,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6651098132133484,
+      "learning_rate": 9.978866780386777e-05,
+      "loss": 3.0153,
+      "step": 1225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6889267563819885,
+      "learning_rate": 9.978678646017701e-05,
+      "loss": 3.0224,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7883036732673645,
+      "learning_rate": 9.978489679729934e-05,
+      "loss": 3.1105,
+      "step": 1235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7156437039375305,
+      "learning_rate": 9.978299881555048e-05,
+      "loss": 3.0585,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0810874700546265,
+      "learning_rate": 9.97810925152476e-05,
+      "loss": 3.0122,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6139539480209351,
+      "learning_rate": 9.977917789670925e-05,
+      "loss": 2.9331,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6547234058380127,
+      "learning_rate": 9.977725496025535e-05,
+      "loss": 3.0056,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7181740999221802,
+      "learning_rate": 9.977532370620718e-05,
+      "loss": 3.0179,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6752457618713379,
+      "learning_rate": 9.977338413488748e-05,
+      "loss": 3.0591,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7583520412445068,
+      "learning_rate": 9.977143624662035e-05,
+      "loss": 3.0328,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7005865573883057,
+      "learning_rate": 9.976948004173125e-05,
+      "loss": 3.0539,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7303656339645386,
+      "learning_rate": 9.976751552054706e-05,
+      "loss": 2.9923,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6937984824180603,
+      "learning_rate": 9.976554268339607e-05,
+      "loss": 2.9828,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7232158184051514,
+      "learning_rate": 9.97635615306079e-05,
+      "loss": 3.0522,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7259365916252136,
+      "learning_rate": 9.976157206251361e-05,
+      "loss": 3.0295,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6522563695907593,
+      "learning_rate": 9.975957427944563e-05,
+      "loss": 3.0231,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.674528181552887,
+      "learning_rate": 9.975756818173778e-05,
+      "loss": 2.9418,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.662665069103241,
+      "learning_rate": 9.975555376972529e-05,
+      "loss": 3.0541,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6916167140007019,
+      "learning_rate": 9.975353104374474e-05,
+      "loss": 3.0789,
+      "step": 1315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.618000328540802,
+      "learning_rate": 9.975150000413411e-05,
+      "loss": 3.0446,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6587952971458435,
+      "learning_rate": 9.97494606512328e-05,
+      "loss": 3.0223,
+      "step": 1325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6588654518127441,
+      "learning_rate": 9.974741298538158e-05,
+      "loss": 2.9947,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7426931858062744,
+      "learning_rate": 9.974535700692257e-05,
+      "loss": 2.9983,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7117341756820679,
+      "learning_rate": 9.974329271619938e-05,
+      "loss": 3.0,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7186235189437866,
+      "learning_rate": 9.97412201135569e-05,
+      "loss": 3.0373,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.689609944820404,
+      "learning_rate": 9.973913919934147e-05,
+      "loss": 3.0417,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.671654224395752,
+      "learning_rate": 9.973704997390079e-05,
+      "loss": 2.98,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.638454020023346,
+      "learning_rate": 9.973495243758398e-05,
+      "loss": 3.0236,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.682000994682312,
+      "learning_rate": 9.973284659074151e-05,
+      "loss": 3.0175,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6575376391410828,
+      "learning_rate": 9.973073243372528e-05,
+      "loss": 3.0828,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6904576420783997,
+      "learning_rate": 9.972860996688853e-05,
+      "loss": 3.0477,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.659777045249939,
+      "learning_rate": 9.972647919058595e-05,
+      "loss": 3.0369,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.608909010887146,
+      "learning_rate": 9.972434010517358e-05,
+      "loss": 2.9291,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.62169349193573,
+      "learning_rate": 9.972219271100882e-05,
+      "loss": 2.9552,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.694141149520874,
+      "learning_rate": 9.972003700845052e-05,
+      "loss": 3.0171,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7307689189910889,
+      "learning_rate": 9.971787299785888e-05,
+      "loss": 2.9232,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6323487162590027,
+      "learning_rate": 9.97157006795955e-05,
+      "loss": 2.9982,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6147862076759338,
+      "learning_rate": 9.971352005402338e-05,
+      "loss": 2.9525,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7860469222068787,
+      "learning_rate": 9.971133112150688e-05,
+      "loss": 3.0276,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6670712828636169,
+      "learning_rate": 9.970913388241177e-05,
+      "loss": 2.9067,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6455975770950317,
+      "learning_rate": 9.970692833710517e-05,
+      "loss": 2.9451,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1323728561401367,
+      "learning_rate": 9.970471448595566e-05,
+      "loss": 2.9645,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6510455012321472,
+      "learning_rate": 9.970249232933315e-05,
+      "loss": 2.9124,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.654278576374054,
+      "learning_rate": 9.970026186760896e-05,
+      "loss": 3.0103,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6372249722480774,
+      "learning_rate": 9.969802310115577e-05,
+      "loss": 3.0373,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6383979320526123,
+      "learning_rate": 9.96957760303477e-05,
+      "loss": 2.9313,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6379492878913879,
+      "learning_rate": 9.969352065556022e-05,
+      "loss": 2.9868,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6993206143379211,
+      "learning_rate": 9.969125697717017e-05,
+      "loss": 2.9362,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6615366339683533,
+      "learning_rate": 9.968898499555582e-05,
+      "loss": 3.0651,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6392012238502502,
+      "learning_rate": 9.968670471109682e-05,
+      "loss": 3.0256,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6886845231056213,
+      "learning_rate": 9.968441612417418e-05,
+      "loss": 3.0021,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6394385695457458,
+      "learning_rate": 9.968211923517032e-05,
+      "loss": 2.9713,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5927764773368835,
+      "learning_rate": 9.967981404446905e-05,
+      "loss": 2.9737,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6014785766601562,
+      "learning_rate": 9.967750055245555e-05,
+      "loss": 2.945,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6324195861816406,
+      "learning_rate": 9.967517875951638e-05,
+      "loss": 2.9703,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6089346408843994,
+      "learning_rate": 9.967284866603955e-05,
+      "loss": 2.9977,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5694971084594727,
+      "learning_rate": 9.967051027241436e-05,
+      "loss": 3.0373,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5827173590660095,
+      "learning_rate": 9.966816357903155e-05,
+      "loss": 2.9348,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6285317540168762,
+      "learning_rate": 9.96658085862833e-05,
+      "loss": 3.0382,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6293942332267761,
+      "learning_rate": 9.966344529456305e-05,
+      "loss": 3.0137,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5975614786148071,
+      "learning_rate": 9.966107370426573e-05,
+      "loss": 2.9306,
+      "step": 1525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5978628396987915,
+      "learning_rate": 9.965869381578765e-05,
+      "loss": 3.0425,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6369414329528809,
+      "learning_rate": 9.965630562952643e-05,
+      "loss": 2.996,
+      "step": 1535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6182513236999512,
+      "learning_rate": 9.965390914588115e-05,
+      "loss": 3.0531,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6268611550331116,
+      "learning_rate": 9.965150436525225e-05,
+      "loss": 2.8913,
+      "step": 1545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5867034792900085,
+      "learning_rate": 9.964909128804159e-05,
+      "loss": 2.8619,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5972966551780701,
+      "learning_rate": 9.964666991465234e-05,
+      "loss": 2.8913,
+      "step": 1555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6169000267982483,
+      "learning_rate": 9.964424024548914e-05,
+      "loss": 2.9683,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5815173387527466,
+      "learning_rate": 9.964180228095793e-05,
+      "loss": 3.0152,
+      "step": 1565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6301456093788147,
+      "learning_rate": 9.963935602146612e-05,
+      "loss": 2.9586,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5690643787384033,
+      "learning_rate": 9.963690146742248e-05,
+      "loss": 2.8759,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6407469511032104,
+      "learning_rate": 9.963443861923715e-05,
+      "loss": 2.9338,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.663486659526825,
+      "learning_rate": 9.963196747732164e-05,
+      "loss": 2.9995,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7133648991584778,
+      "learning_rate": 9.96294880420889e-05,
+      "loss": 2.9646,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6082223057746887,
+      "learning_rate": 9.96270003139532e-05,
+      "loss": 2.9112,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6178692579269409,
+      "learning_rate": 9.962450429333026e-05,
+      "loss": 2.9247,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6287717819213867,
+      "learning_rate": 9.962199998063715e-05,
+      "loss": 2.9557,
+      "step": 1605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6333958506584167,
+      "learning_rate": 9.961948737629231e-05,
+      "loss": 2.9913,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5590129494667053,
+      "learning_rate": 9.961696648071561e-05,
+      "loss": 2.9374,
+      "step": 1615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6144304871559143,
+      "learning_rate": 9.961443729432828e-05,
+      "loss": 2.9766,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6085737943649292,
+      "learning_rate": 9.961189981755294e-05,
+      "loss": 2.9565,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5823155641555786,
+      "learning_rate": 9.960935405081357e-05,
+      "loss": 2.973,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.616106390953064,
+      "learning_rate": 9.960679999453559e-05,
+      "loss": 2.948,
+      "step": 1635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5598835349082947,
+      "learning_rate": 9.960423764914575e-05,
+      "loss": 2.9017,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5778008103370667,
+      "learning_rate": 9.960166701507221e-05,
+      "loss": 2.9766,
+      "step": 1645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5873135328292847,
+      "learning_rate": 9.959908809274453e-05,
+      "loss": 2.9497,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6301279664039612,
+      "learning_rate": 9.959650088259362e-05,
+      "loss": 2.9397,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.658209502696991,
+      "learning_rate": 9.959390538505181e-05,
+      "loss": 2.9985,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5934017300605774,
+      "learning_rate": 9.959130160055279e-05,
+      "loss": 2.9025,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.589269757270813,
+      "learning_rate": 9.958868952953163e-05,
+      "loss": 2.9439,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5899887681007385,
+      "learning_rate": 9.95860691724248e-05,
+      "loss": 2.9371,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5731266736984253,
+      "learning_rate": 9.958344052967018e-05,
+      "loss": 2.9451,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6009959578514099,
+      "learning_rate": 9.958080360170698e-05,
+      "loss": 2.951,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9744573831558228,
+      "learning_rate": 9.957815838897585e-05,
+      "loss": 2.9517,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5741543173789978,
+      "learning_rate": 9.957550489191874e-05,
+      "loss": 3.0085,
+      "step": 1695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6033324003219604,
+      "learning_rate": 9.957284311097908e-05,
+      "loss": 2.9703,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5774264931678772,
+      "learning_rate": 9.957017304660164e-05,
+      "loss": 2.8817,
+      "step": 1705
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5756993293762207,
+      "learning_rate": 9.956749469923258e-05,
+      "loss": 2.8569,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6163966059684753,
+      "learning_rate": 9.956480806931942e-05,
+      "loss": 2.918,
+      "step": 1715
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.770514726638794,
+      "learning_rate": 9.956211315731112e-05,
+      "loss": 2.8842,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6141473054885864,
+      "learning_rate": 9.955940996365796e-05,
+      "loss": 2.9396,
+      "step": 1725
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5673449039459229,
+      "learning_rate": 9.955669848881166e-05,
+      "loss": 2.9406,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5737307071685791,
+      "learning_rate": 9.955397873322527e-05,
+      "loss": 2.9081,
+      "step": 1735
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.594089150428772,
+      "learning_rate": 9.955125069735327e-05,
+      "loss": 2.9658,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.566038191318512,
+      "learning_rate": 9.95485143816515e-05,
+      "loss": 2.9663,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5605973601341248,
+      "learning_rate": 9.95457697865772e-05,
+      "loss": 2.9176,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.559630811214447,
+      "learning_rate": 9.954301691258894e-05,
+      "loss": 2.9069,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5856837630271912,
+      "learning_rate": 9.954025576014674e-05,
+      "loss": 2.9068,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5786495208740234,
+      "learning_rate": 9.953748632971201e-05,
+      "loss": 2.8596,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5448077321052551,
+      "learning_rate": 9.953470862174748e-05,
+      "loss": 2.9493,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5594565868377686,
+      "learning_rate": 9.953192263671728e-05,
+      "loss": 2.8856,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5783025622367859,
+      "learning_rate": 9.952912837508697e-05,
+      "loss": 2.8718,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6230905652046204,
+      "learning_rate": 9.952632583732343e-05,
+      "loss": 2.8847,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5969368815422058,
+      "learning_rate": 9.952351502389498e-05,
+      "loss": 2.9528,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5886838436126709,
+      "learning_rate": 9.95206959352713e-05,
+      "loss": 2.8894,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5576210618019104,
+      "learning_rate": 9.951786857192341e-05,
+      "loss": 2.8797,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6236076354980469,
+      "learning_rate": 9.951503293432378e-05,
+      "loss": 2.8924,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6171234250068665,
+      "learning_rate": 9.951218902294625e-05,
+      "loss": 2.9156,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5546749830245972,
+      "learning_rate": 9.9509336838266e-05,
+      "loss": 2.855,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6169042587280273,
+      "learning_rate": 9.950647638075963e-05,
+      "loss": 2.8412,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5753995180130005,
+      "learning_rate": 9.950360765090511e-05,
+      "loss": 2.8897,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6147080659866333,
+      "learning_rate": 9.950073064918179e-05,
+      "loss": 2.8935,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6217177510261536,
+      "learning_rate": 9.949784537607042e-05,
+      "loss": 2.97,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5595420598983765,
+      "learning_rate": 9.949495183205311e-05,
+      "loss": 2.8961,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5867671370506287,
+      "learning_rate": 9.949205001761334e-05,
+      "loss": 2.9235,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5483968257904053,
+      "learning_rate": 9.948913993323603e-05,
+      "loss": 2.9206,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7083644866943359,
+      "learning_rate": 9.948622157940744e-05,
+      "loss": 2.9148,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9476220607757568,
+      "learning_rate": 9.948329495661517e-05,
+      "loss": 2.77,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5603005886077881,
+      "learning_rate": 9.94803600653483e-05,
+      "loss": 2.9148,
+      "step": 1865
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5355193018913269,
+      "learning_rate": 9.947741690609722e-05,
+      "loss": 2.8512,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7880305647850037,
+      "learning_rate": 9.94744654793537e-05,
+      "loss": 2.9223,
+      "step": 1875
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5976160168647766,
+      "learning_rate": 9.947150578561095e-05,
+      "loss": 2.8519,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5794243812561035,
+      "learning_rate": 9.946853782536351e-05,
+      "loss": 2.9047,
+      "step": 1885
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.556047260761261,
+      "learning_rate": 9.946556159910732e-05,
+      "loss": 2.8099,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5811231136322021,
+      "learning_rate": 9.946257710733966e-05,
+      "loss": 2.9445,
+      "step": 1895
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1472445726394653,
+      "learning_rate": 9.945958435055927e-05,
+      "loss": 2.9373,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6591916084289551,
+      "learning_rate": 9.945658332926622e-05,
+      "loss": 2.8804,
+      "step": 1905
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6377168297767639,
+      "learning_rate": 9.945357404396197e-05,
+      "loss": 2.9638,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6102290153503418,
+      "learning_rate": 9.945055649514934e-05,
+      "loss": 2.8822,
+      "step": 1915
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5673019886016846,
+      "learning_rate": 9.944753068333256e-05,
+      "loss": 2.874,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5595168471336365,
+      "learning_rate": 9.944449660901725e-05,
+      "loss": 2.9435,
+      "step": 1925
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5644038915634155,
+      "learning_rate": 9.944145427271036e-05,
+      "loss": 2.8599,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5800265669822693,
+      "learning_rate": 9.94384036749203e-05,
+      "loss": 2.8535,
+      "step": 1935
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5838456153869629,
+      "learning_rate": 9.943534481615677e-05,
+      "loss": 2.9013,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5913266539573669,
+      "learning_rate": 9.943227769693091e-05,
+      "loss": 2.8963,
+      "step": 1945
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5595894455909729,
+      "learning_rate": 9.942920231775524e-05,
+      "loss": 2.8754,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5802275538444519,
+      "learning_rate": 9.942611867914363e-05,
+      "loss": 2.8892,
+      "step": 1955
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5698996782302856,
+      "learning_rate": 9.942302678161133e-05,
+      "loss": 2.8585,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5160923004150391,
+      "learning_rate": 9.9419926625675e-05,
+      "loss": 2.8061,
+      "step": 1965
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5342493057250977,
+      "learning_rate": 9.941681821185265e-05,
+      "loss": 2.804,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.581838846206665,
+      "learning_rate": 9.941370154066371e-05,
+      "loss": 2.8983,
+      "step": 1975
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5829259157180786,
+      "learning_rate": 9.941057661262895e-05,
+      "loss": 2.882,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5753056406974792,
+      "learning_rate": 9.940744342827055e-05,
+      "loss": 2.9156,
+      "step": 1985
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5727913975715637,
+      "learning_rate": 9.940430198811202e-05,
+      "loss": 2.7908,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6199966669082642,
+      "learning_rate": 9.940115229267831e-05,
+      "loss": 2.8781,
+      "step": 1995
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.549972653388977,
+      "learning_rate": 9.939799434249571e-05,
+      "loss": 2.8531,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5726840496063232,
+      "learning_rate": 9.93948281380919e-05,
+      "loss": 2.8616,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5470526814460754,
+      "learning_rate": 9.939165367999597e-05,
+      "loss": 2.8873,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5901708602905273,
+      "learning_rate": 9.938847096873831e-05,
+      "loss": 2.8515,
+      "step": 2015
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.675588071346283,
+      "learning_rate": 9.938528000485078e-05,
+      "loss": 2.9438,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6412743330001831,
+      "learning_rate": 9.938208078886655e-05,
+      "loss": 2.8129,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6289473176002502,
+      "learning_rate": 9.937887332132023e-05,
+      "loss": 2.8433,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5613908171653748,
+      "learning_rate": 9.937565760274776e-05,
+      "loss": 2.8569,
+      "step": 2035
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5247993469238281,
+      "learning_rate": 9.937243363368645e-05,
+      "loss": 2.8519,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5785958170890808,
+      "learning_rate": 9.936920141467505e-05,
+      "loss": 2.7757,
+      "step": 2045
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9428956508636475,
+      "learning_rate": 9.936596094625364e-05,
+      "loss": 2.8119,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5532321929931641,
+      "learning_rate": 9.936271222896367e-05,
+      "loss": 2.924,
+      "step": 2055
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5641935467720032,
+      "learning_rate": 9.935945526334803e-05,
+      "loss": 2.886,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6113528609275818,
+      "learning_rate": 9.935619004995089e-05,
+      "loss": 2.7854,
+      "step": 2065
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5929989218711853,
+      "learning_rate": 9.935291658931791e-05,
+      "loss": 2.8403,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5220435261726379,
+      "learning_rate": 9.934963488199604e-05,
+      "loss": 2.8937,
+      "step": 2075
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7125425934791565,
+      "learning_rate": 9.934634492853366e-05,
+      "loss": 2.7198,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5659758448600769,
+      "learning_rate": 9.93430467294805e-05,
+      "loss": 2.7804,
+      "step": 2085
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6278787851333618,
+      "learning_rate": 9.933974028538768e-05,
+      "loss": 2.7919,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5576902627944946,
+      "learning_rate": 9.933642559680769e-05,
+      "loss": 2.8423,
+      "step": 2095
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.591615617275238,
+      "learning_rate": 9.93331026642944e-05,
+      "loss": 2.9117,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5419116020202637,
+      "learning_rate": 9.932977148840307e-05,
+      "loss": 2.897,
+      "step": 2105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5935481786727905,
+      "learning_rate": 9.93264320696903e-05,
+      "loss": 2.8017,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5284276008605957,
+      "learning_rate": 9.932308440871413e-05,
+      "loss": 2.8424,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5392113924026489,
+      "learning_rate": 9.931972850603394e-05,
+      "loss": 2.817,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5365735292434692,
+      "learning_rate": 9.931636436221048e-05,
+      "loss": 2.7876,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5741152763366699,
+      "learning_rate": 9.931299197780586e-05,
+      "loss": 2.8058,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5733075141906738,
+      "learning_rate": 9.930961135338361e-05,
+      "loss": 2.8374,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.546199381351471,
+      "learning_rate": 9.930622248950864e-05,
+      "loss": 2.8094,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5660104751586914,
+      "learning_rate": 9.93028253867472e-05,
+      "loss": 2.7896,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5546613335609436,
+      "learning_rate": 9.929942004566695e-05,
+      "loss": 2.791,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5801413655281067,
+      "learning_rate": 9.929600646683686e-05,
+      "loss": 2.7856,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5390631556510925,
+      "learning_rate": 9.929258465082738e-05,
+      "loss": 2.8316,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5845677852630615,
+      "learning_rate": 9.928915459821027e-05,
+      "loss": 2.9162,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5651169419288635,
+      "learning_rate": 9.928571630955865e-05,
+      "loss": 2.7643,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5238113403320312,
+      "learning_rate": 9.928226978544708e-05,
+      "loss": 2.8771,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5257121920585632,
+      "learning_rate": 9.927881502645146e-05,
+      "loss": 2.7709,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5365587472915649,
+      "learning_rate": 9.927535203314904e-05,
+      "loss": 2.8009,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5348502397537231,
+      "learning_rate": 9.927188080611849e-05,
+      "loss": 2.8163,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5342689752578735,
+      "learning_rate": 9.926840134593984e-05,
+      "loss": 2.7546,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5506367683410645,
+      "learning_rate": 9.92649136531945e-05,
+      "loss": 2.7844,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5489587187767029,
+      "learning_rate": 9.926141772846525e-05,
+      "loss": 2.8751,
+      "step": 2205
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5317386388778687,
+      "learning_rate": 9.925791357233623e-05,
+      "loss": 2.8411,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5903447866439819,
+      "learning_rate": 9.925440118539298e-05,
+      "loss": 2.7004,
+      "step": 2215
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5535102486610413,
+      "learning_rate": 9.925088056822241e-05,
+      "loss": 2.9187,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5045163035392761,
+      "learning_rate": 9.924735172141281e-05,
+      "loss": 2.8802,
+      "step": 2225
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5693023800849915,
+      "learning_rate": 9.924381464555381e-05,
+      "loss": 2.7818,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5721497535705566,
+      "learning_rate": 9.924026934123647e-05,
+      "loss": 2.9065,
+      "step": 2235
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5432178378105164,
+      "learning_rate": 9.92367158090532e-05,
+      "loss": 2.8294,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5397220849990845,
+      "learning_rate": 9.923315404959775e-05,
+      "loss": 2.8706,
+      "step": 2245
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5166792869567871,
+      "learning_rate": 9.92295840634653e-05,
+      "loss": 2.8328,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5213223099708557,
+      "learning_rate": 9.922600585125237e-05,
+      "loss": 2.7479,
+      "step": 2255
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5308531522750854,
+      "learning_rate": 9.922241941355688e-05,
+      "loss": 2.8448,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.534652829170227,
+      "learning_rate": 9.921882475097811e-05,
+      "loss": 2.7837,
+      "step": 2265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5321337580680847,
+      "learning_rate": 9.921522186411671e-05,
+      "loss": 2.8529,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5570808053016663,
+      "learning_rate": 9.92116107535747e-05,
+      "loss": 2.8583,
+      "step": 2275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5451025366783142,
+      "learning_rate": 9.920799141995549e-05,
+      "loss": 2.8421,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5665819048881531,
+      "learning_rate": 9.920436386386387e-05,
+      "loss": 2.8888,
+      "step": 2285
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5931339859962463,
+      "learning_rate": 9.920072808590596e-05,
+      "loss": 2.8165,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6514272689819336,
+      "learning_rate": 9.91970840866893e-05,
+      "loss": 2.7878,
+      "step": 2295
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5498843193054199,
+      "learning_rate": 9.919343186682282e-05,
+      "loss": 2.8296,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5507535338401794,
+      "learning_rate": 9.918977142691675e-05,
+      "loss": 2.9395,
+      "step": 2305
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5743236541748047,
+      "learning_rate": 9.918610276758273e-05,
+      "loss": 2.8723,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.571609616279602,
+      "learning_rate": 9.918242588943383e-05,
+      "loss": 2.8247,
+      "step": 2315
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5714886784553528,
+      "learning_rate": 9.91787407930844e-05,
+      "loss": 2.745,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5083488821983337,
+      "learning_rate": 9.917504747915023e-05,
+      "loss": 2.8432,
+      "step": 2325
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5156746506690979,
+      "learning_rate": 9.917134594824843e-05,
+      "loss": 2.7739,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5860164761543274,
+      "learning_rate": 9.916763620099754e-05,
+      "loss": 2.8171,
+      "step": 2335
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5663352608680725,
+      "learning_rate": 9.916391823801743e-05,
+      "loss": 2.7924,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5711392164230347,
+      "learning_rate": 9.916019205992937e-05,
+      "loss": 2.8257,
+      "step": 2345
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5605615377426147,
+      "learning_rate": 9.915645766735597e-05,
+      "loss": 2.8102,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.525581419467926,
+      "learning_rate": 9.915271506092126e-05,
+      "loss": 2.7617,
+      "step": 2355
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5562646389007568,
+      "learning_rate": 9.91489642412506e-05,
+      "loss": 2.8065,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5505369901657104,
+      "learning_rate": 9.914520520897072e-05,
+      "loss": 2.7898,
+      "step": 2365
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5090936422348022,
+      "learning_rate": 9.914143796470978e-05,
+      "loss": 2.7875,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5600467920303345,
+      "learning_rate": 9.913766250909726e-05,
+      "loss": 2.737,
+      "step": 2375
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5364627838134766,
+      "learning_rate": 9.9133878842764e-05,
+      "loss": 2.8018,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5347198843955994,
+      "learning_rate": 9.913008696634226e-05,
+      "loss": 2.9207,
+      "step": 2385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.554924726486206,
+      "learning_rate": 9.912628688046564e-05,
+      "loss": 2.8066,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5463462471961975,
+      "learning_rate": 9.912247858576914e-05,
+      "loss": 2.8526,
+      "step": 2395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5783821940422058,
+      "learning_rate": 9.911866208288906e-05,
+      "loss": 2.7679,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5517280697822571,
+      "learning_rate": 9.911483737246319e-05,
+      "loss": 2.7647,
+      "step": 2405
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5631377696990967,
+      "learning_rate": 9.911100445513058e-05,
+      "loss": 2.8823,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5921874642372131,
+      "learning_rate": 9.910716333153169e-05,
+      "loss": 2.798,
+      "step": 2415
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5638368725776672,
+      "learning_rate": 9.910331400230841e-05,
+      "loss": 2.8741,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4942677915096283,
+      "learning_rate": 9.90994564681039e-05,
+      "loss": 2.8442,
+      "step": 2425
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5413124561309814,
+      "learning_rate": 9.909559072956275e-05,
+      "loss": 2.7035,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49930527806282043,
+      "learning_rate": 9.909171678733091e-05,
+      "loss": 2.6748,
+      "step": 2435
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49008217453956604,
+      "learning_rate": 9.908783464205573e-05,
+      "loss": 2.799,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5212256908416748,
+      "learning_rate": 9.908394429438585e-05,
+      "loss": 2.7156,
+      "step": 2445
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5115553736686707,
+      "learning_rate": 9.908004574497139e-05,
+      "loss": 2.6909,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.58125901222229,
+      "learning_rate": 9.907613899446372e-05,
+      "loss": 2.7898,
+      "step": 2455
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5332221984863281,
+      "learning_rate": 9.90722240435157e-05,
+      "loss": 2.8059,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5360547304153442,
+      "learning_rate": 9.906830089278148e-05,
+      "loss": 2.6719,
+      "step": 2465
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5108291506767273,
+      "learning_rate": 9.906436954291661e-05,
+      "loss": 2.7546,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5774298906326294,
+      "learning_rate": 9.906042999457798e-05,
+      "loss": 2.8152,
+      "step": 2475
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48830124735832214,
+      "learning_rate": 9.905648224842391e-05,
+      "loss": 2.7579,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5295991897583008,
+      "learning_rate": 9.905252630511405e-05,
+      "loss": 2.7803,
+      "step": 2485
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5608240962028503,
+      "learning_rate": 9.904856216530938e-05,
+      "loss": 2.7776,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49888157844543457,
+      "learning_rate": 9.904458982967235e-05,
+      "loss": 2.7999,
+      "step": 2495
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5546042919158936,
+      "learning_rate": 9.904060929886668e-05,
+      "loss": 2.7343,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.520534336566925,
+      "learning_rate": 9.903662057355751e-05,
+      "loss": 2.7562,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.49917396903038025,
+      "learning_rate": 9.903262365441137e-05,
+      "loss": 2.8137,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.495328813791275,
+      "learning_rate": 9.90286185420961e-05,
+      "loss": 2.7253,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.53936767578125,
+      "learning_rate": 9.902460523728094e-05,
+      "loss": 2.7622,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5512412190437317,
+      "learning_rate": 9.902058374063651e-05,
+      "loss": 2.7985,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5316717028617859,
+      "learning_rate": 9.90165540528348e-05,
+      "loss": 2.8292,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.515882134437561,
+      "learning_rate": 9.901251617454913e-05,
+      "loss": 2.6853,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5248692035675049,
+      "learning_rate": 9.900847010645422e-05,
+      "loss": 2.7325,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5220763683319092,
+      "learning_rate": 9.900441584922616e-05,
+      "loss": 2.8392,
+      "step": 2545
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5244632363319397,
+      "learning_rate": 9.90003534035424e-05,
+      "loss": 2.7418,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5263069272041321,
+      "learning_rate": 9.899628277008175e-05,
+      "loss": 2.7767,
+      "step": 2555
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5132372975349426,
+      "learning_rate": 9.899220394952441e-05,
+      "loss": 2.7909,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.537783145904541,
+      "learning_rate": 9.898811694255195e-05,
+      "loss": 2.7593,
+      "step": 2565
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.525317907333374,
+      "learning_rate": 9.898402174984726e-05,
+      "loss": 2.7297,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5169928073883057,
+      "learning_rate": 9.897991837209464e-05,
+      "loss": 2.7129,
+      "step": 2575
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.501149594783783,
+      "learning_rate": 9.897580680997977e-05,
+      "loss": 2.7714,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5354222059249878,
+      "learning_rate": 9.897168706418965e-05,
+      "loss": 2.8513,
+      "step": 2585
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4864729344844818,
+      "learning_rate": 9.89675591354127e-05,
+      "loss": 2.7689,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4846419095993042,
+      "learning_rate": 9.896342302433866e-05,
+      "loss": 2.6412,
+      "step": 2595
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4884699285030365,
+      "learning_rate": 9.895927873165867e-05,
+      "loss": 2.6938,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5016258955001831,
+      "learning_rate": 9.895512625806522e-05,
+      "loss": 2.8369,
+      "step": 2605
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5517599582672119,
+      "learning_rate": 9.895096560425218e-05,
+      "loss": 2.7567,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9112642407417297,
+      "learning_rate": 9.894679677091477e-05,
+      "loss": 2.7887,
+      "step": 2615
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5439919829368591,
+      "learning_rate": 9.89426197587496e-05,
+      "loss": 2.7397,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5502176284790039,
+      "learning_rate": 9.893843456845462e-05,
+      "loss": 2.7208,
+      "step": 2625
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6637817025184631,
+      "learning_rate": 9.893424120072916e-05,
+      "loss": 2.7284,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5237710475921631,
+      "learning_rate": 9.893003965627394e-05,
+      "loss": 2.789,
+      "step": 2635
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.49424341320991516,
+      "learning_rate": 9.8925829935791e-05,
+      "loss": 2.783,
+      "step": 2640
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4912557005882263,
+      "learning_rate": 9.892161203998375e-05,
+      "loss": 2.7168,
+      "step": 2645
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5104436278343201,
+      "learning_rate": 9.891738596955703e-05,
+      "loss": 2.804,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5361624360084534,
+      "learning_rate": 9.891315172521696e-05,
+      "loss": 2.7201,
+      "step": 2655
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5073333382606506,
+      "learning_rate": 9.890890930767111e-05,
+      "loss": 2.727,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5058367252349854,
+      "learning_rate": 9.890465871762832e-05,
+      "loss": 2.7644,
+      "step": 2665
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4916439354419708,
+      "learning_rate": 9.89003999557989e-05,
+      "loss": 2.6892,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.47397249937057495,
+      "learning_rate": 9.889613302289443e-05,
+      "loss": 2.7098,
+      "step": 2675
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5411562323570251,
+      "learning_rate": 9.889185791962792e-05,
+      "loss": 2.7663,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5167352557182312,
+      "learning_rate": 9.888757464671372e-05,
+      "loss": 2.7211,
+      "step": 2685
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5874642133712769,
+      "learning_rate": 9.888328320486755e-05,
+      "loss": 2.775,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5036665797233582,
+      "learning_rate": 9.88789835948065e-05,
+      "loss": 2.7649,
+      "step": 2695
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4973278045654297,
+      "learning_rate": 9.887467581724901e-05,
+      "loss": 2.7809,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5413213968276978,
+      "learning_rate": 9.887035987291488e-05,
+      "loss": 2.7553,
+      "step": 2705
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.563840389251709,
+      "learning_rate": 9.88660357625253e-05,
+      "loss": 2.7394,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5571768879890442,
+      "learning_rate": 9.886170348680284e-05,
+      "loss": 2.7773,
+      "step": 2715
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5031925439834595,
+      "learning_rate": 9.885736304647139e-05,
+      "loss": 2.792,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4850078821182251,
+      "learning_rate": 9.885301444225619e-05,
+      "loss": 2.7825,
+      "step": 2725
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4734047055244446,
+      "learning_rate": 9.884865767488391e-05,
+      "loss": 2.6549,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.48243555426597595,
+      "learning_rate": 9.884429274508256e-05,
+      "loss": 2.727,
+      "step": 2735
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4956502914428711,
+      "learning_rate": 9.883991965358146e-05,
+      "loss": 2.7425,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5262529850006104,
+      "learning_rate": 9.883553840111136e-05,
+      "loss": 2.6797,
+      "step": 2745
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.49567079544067383,
+      "learning_rate": 9.883114898840437e-05,
+      "loss": 2.7348,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.46910232305526733,
+      "learning_rate": 9.882675141619391e-05,
+      "loss": 2.758,
+      "step": 2755
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4964003562927246,
+      "learning_rate": 9.882234568521482e-05,
+      "loss": 2.7468,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5286035537719727,
+      "learning_rate": 9.881793179620328e-05,
+      "loss": 2.7085,
+      "step": 2765
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.48203030228614807,
+      "learning_rate": 9.881350974989682e-05,
+      "loss": 2.8009,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5055955052375793,
+      "learning_rate": 9.880907954703437e-05,
+      "loss": 2.7423,
+      "step": 2775
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.500762939453125,
+      "learning_rate": 9.880464118835619e-05,
+      "loss": 2.7162,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5581158995628357,
+      "learning_rate": 9.880019467460391e-05,
+      "loss": 2.7653,
+      "step": 2785
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5193862915039062,
+      "learning_rate": 9.879574000652052e-05,
+      "loss": 2.7621,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5194688439369202,
+      "learning_rate": 9.87912771848504e-05,
+      "loss": 2.7168,
+      "step": 2795
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.512714684009552,
+      "learning_rate": 9.878680621033925e-05,
+      "loss": 2.7876,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4905962646007538,
+      "learning_rate": 9.878232708373415e-05,
+      "loss": 2.7724,
+      "step": 2805
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4945067763328552,
+      "learning_rate": 9.877783980578357e-05,
+      "loss": 2.818,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.482650488615036,
+      "learning_rate": 9.877334437723728e-05,
+      "loss": 2.7113,
+      "step": 2815
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5028238892555237,
+      "learning_rate": 9.87688407988465e-05,
+      "loss": 2.7551,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5313793420791626,
+      "learning_rate": 9.876432907136372e-05,
+      "loss": 2.8015,
+      "step": 2825
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.47801080346107483,
+      "learning_rate": 9.875980919554287e-05,
+      "loss": 2.7167,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5323198437690735,
+      "learning_rate": 9.875528117213915e-05,
+      "loss": 2.8153,
+      "step": 2835
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.511436939239502,
+      "learning_rate": 9.875074500190922e-05,
+      "loss": 2.7059,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5409953594207764,
+      "learning_rate": 9.874620068561104e-05,
+      "loss": 2.7779,
+      "step": 2845
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4929635226726532,
+      "learning_rate": 9.874164822400396e-05,
+      "loss": 2.6824,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.47751227021217346,
+      "learning_rate": 9.873708761784866e-05,
+      "loss": 2.6997,
+      "step": 2855
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.48180460929870605,
+      "learning_rate": 9.873251886790722e-05,
+      "loss": 2.7791,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5525875091552734,
+      "learning_rate": 9.872794197494307e-05,
+      "loss": 2.761,
+      "step": 2865
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5176541209220886,
+      "learning_rate": 9.872335693972094e-05,
+      "loss": 2.7704,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5033114552497864,
+      "learning_rate": 9.871876376300702e-05,
+      "loss": 2.7601,
+      "step": 2875
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.48789456486701965,
+      "learning_rate": 9.871416244556881e-05,
+      "loss": 2.738,
+      "step": 2880
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5415577292442322,
+      "learning_rate": 9.870955298817517e-05,
+      "loss": 2.8266,
+      "step": 2885
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4857690632343292,
+      "learning_rate": 9.87049353915963e-05,
+      "loss": 2.7934,
+      "step": 2890
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5184076428413391,
+      "learning_rate": 9.87003096566038e-05,
+      "loss": 2.7348,
+      "step": 2895
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5099581480026245,
+      "learning_rate": 9.869567578397062e-05,
+      "loss": 2.6813,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46895912289619446,
+      "learning_rate": 9.869103377447105e-05,
+      "loss": 2.7081,
+      "step": 2905
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4918888807296753,
+      "learning_rate": 9.868638362888076e-05,
+      "loss": 2.676,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4882889688014984,
+      "learning_rate": 9.868172534797678e-05,
+      "loss": 2.7535,
+      "step": 2915
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5414462089538574,
+      "learning_rate": 9.867705893253747e-05,
+      "loss": 2.667,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48702454566955566,
+      "learning_rate": 9.867238438334259e-05,
+      "loss": 2.7499,
+      "step": 2925
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5037529468536377,
+      "learning_rate": 9.866770170117324e-05,
+      "loss": 2.7512,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48932573199272156,
+      "learning_rate": 9.866301088681185e-05,
+      "loss": 2.7325,
+      "step": 2935
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5030003190040588,
+      "learning_rate": 9.865831194104227e-05,
+      "loss": 2.789,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5003800392150879,
+      "learning_rate": 9.865360486464967e-05,
+      "loss": 2.786,
+      "step": 2945
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4700438976287842,
+      "learning_rate": 9.864888965842056e-05,
+      "loss": 2.6937,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5254529714584351,
+      "learning_rate": 9.864416632314287e-05,
+      "loss": 2.7087,
+      "step": 2955
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5321142077445984,
+      "learning_rate": 9.863943485960583e-05,
+      "loss": 2.763,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47648805379867554,
+      "learning_rate": 9.863469526860005e-05,
+      "loss": 2.7007,
+      "step": 2965
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4922126829624176,
+      "learning_rate": 9.862994755091752e-05,
+      "loss": 2.76,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5004321932792664,
+      "learning_rate": 9.862519170735154e-05,
+      "loss": 2.7557,
+      "step": 2975
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5046461224555969,
+      "learning_rate": 9.86204277386968e-05,
+      "loss": 2.629,
+      "step": 2980
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5086774230003357,
+      "learning_rate": 9.861565564574935e-05,
+      "loss": 2.7379,
+      "step": 2985
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.49016687273979187,
+      "learning_rate": 9.861087542930658e-05,
+      "loss": 2.6622,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5240083336830139,
+      "learning_rate": 9.860608709016726e-05,
+      "loss": 2.6404,
+      "step": 2995
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4846411645412445,
+      "learning_rate": 9.860129062913147e-05,
+      "loss": 2.7735,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48679450154304504,
+      "learning_rate": 9.859648604700072e-05,
+      "loss": 2.7901,
+      "step": 3005
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4719792604446411,
+      "learning_rate": 9.859167334457781e-05,
+      "loss": 2.7074,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5059725046157837,
+      "learning_rate": 9.858685252266696e-05,
+      "loss": 2.776,
+      "step": 3015
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45968782901763916,
+      "learning_rate": 9.858202358207367e-05,
+      "loss": 2.719,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4833773970603943,
+      "learning_rate": 9.857718652360486e-05,
+      "loss": 2.6993,
+      "step": 3025
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4649290144443512,
+      "learning_rate": 9.857234134806878e-05,
+      "loss": 2.7421,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4785956144332886,
+      "learning_rate": 9.856748805627503e-05,
+      "loss": 2.6739,
+      "step": 3035
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5102661848068237,
+      "learning_rate": 9.856262664903462e-05,
+      "loss": 2.6735,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47354817390441895,
+      "learning_rate": 9.855775712715981e-05,
+      "loss": 2.7199,
+      "step": 3045
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5386574268341064,
+      "learning_rate": 9.855287949146432e-05,
+      "loss": 2.7656,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48266148567199707,
+      "learning_rate": 9.854799374276318e-05,
+      "loss": 2.7509,
+      "step": 3055
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4888915717601776,
+      "learning_rate": 9.854309988187277e-05,
+      "loss": 2.8177,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4666389524936676,
+      "learning_rate": 9.853819790961086e-05,
+      "loss": 2.7875,
+      "step": 3065
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47116002440452576,
+      "learning_rate": 9.853328782679652e-05,
+      "loss": 2.7678,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46211445331573486,
+      "learning_rate": 9.852836963425021e-05,
+      "loss": 2.6319,
+      "step": 3075
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4894236624240875,
+      "learning_rate": 9.852344333279375e-05,
+      "loss": 2.7572,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4708699882030487,
+      "learning_rate": 9.851850892325032e-05,
+      "loss": 2.7437,
+      "step": 3085
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4778779149055481,
+      "learning_rate": 9.851356640644443e-05,
+      "loss": 2.6449,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4887005686759949,
+      "learning_rate": 9.850861578320196e-05,
+      "loss": 2.7334,
+      "step": 3095
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5115506649017334,
+      "learning_rate": 9.850365705435013e-05,
+      "loss": 2.6913,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5078063607215881,
+      "learning_rate": 9.849869022071753e-05,
+      "loss": 2.7388,
+      "step": 3105
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47528019547462463,
+      "learning_rate": 9.849371528313411e-05,
+      "loss": 2.7922,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5062806606292725,
+      "learning_rate": 9.848873224243114e-05,
+      "loss": 2.7267,
+      "step": 3115
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5993074178695679,
+      "learning_rate": 9.848374109944128e-05,
+      "loss": 2.7237,
+      "step": 3120
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4584481418132782,
+      "learning_rate": 9.847874185499855e-05,
+      "loss": 2.6344,
+      "step": 3125
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48799651861190796,
+      "learning_rate": 9.847373450993828e-05,
+      "loss": 2.8466,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4959023892879486,
+      "learning_rate": 9.846871906509719e-05,
+      "loss": 2.762,
+      "step": 3135
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4655931890010834,
+      "learning_rate": 9.846369552131334e-05,
+      "loss": 2.6703,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4945942759513855,
+      "learning_rate": 9.845866387942614e-05,
+      "loss": 2.7516,
+      "step": 3145
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.49716705083847046,
+      "learning_rate": 9.845362414027637e-05,
+      "loss": 2.6776,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4992823600769043,
+      "learning_rate": 9.844857630470614e-05,
+      "loss": 2.6524,
+      "step": 3155
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48290660977363586,
+      "learning_rate": 9.844352037355893e-05,
+      "loss": 2.7595,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5002864003181458,
+      "learning_rate": 9.843845634767957e-05,
+      "loss": 2.6566,
+      "step": 3165
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5041113495826721,
+      "learning_rate": 9.843338422791424e-05,
+      "loss": 2.7158,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4819050133228302,
+      "learning_rate": 9.842830401511048e-05,
+      "loss": 2.7051,
+      "step": 3175
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4830915331840515,
+      "learning_rate": 9.842321571011714e-05,
+      "loss": 2.6264,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47560450434684753,
+      "learning_rate": 9.841811931378448e-05,
+      "loss": 2.7182,
+      "step": 3185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5044006705284119,
+      "learning_rate": 9.841301482696413e-05,
+      "loss": 2.7406,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48234134912490845,
+      "learning_rate": 9.840790225050896e-05,
+      "loss": 2.6935,
+      "step": 3195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47623756527900696,
+      "learning_rate": 9.840278158527328e-05,
+      "loss": 2.5854,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5063531994819641,
+      "learning_rate": 9.839765283211278e-05,
+      "loss": 2.7055,
+      "step": 3205
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4794359803199768,
+      "learning_rate": 9.83925159918844e-05,
+      "loss": 2.6928,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.490084707736969,
+      "learning_rate": 9.838737106544651e-05,
+      "loss": 2.657,
+      "step": 3215
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4782032072544098,
+      "learning_rate": 9.838221805365881e-05,
+      "loss": 2.7051,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5030024647712708,
+      "learning_rate": 9.837705695738236e-05,
+      "loss": 2.6931,
+      "step": 3225
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5247048735618591,
+      "learning_rate": 9.837188777747954e-05,
+      "loss": 2.6958,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.53005051612854,
+      "learning_rate": 9.836671051481411e-05,
+      "loss": 2.7533,
+      "step": 3235
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4887393116950989,
+      "learning_rate": 9.836152517025116e-05,
+      "loss": 2.6884,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.522157609462738,
+      "learning_rate": 9.835633174465715e-05,
+      "loss": 2.7079,
+      "step": 3245
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47654110193252563,
+      "learning_rate": 9.83511302388999e-05,
+      "loss": 2.7148,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.467733770608902,
+      "learning_rate": 9.834592065384853e-05,
+      "loss": 2.7792,
+      "step": 3255
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4853402078151703,
+      "learning_rate": 9.834070299037357e-05,
+      "loss": 2.7667,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4571678936481476,
+      "learning_rate": 9.833547724934687e-05,
+      "loss": 2.619,
+      "step": 3265
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4815673530101776,
+      "learning_rate": 9.833024343164159e-05,
+      "loss": 2.7381,
+      "step": 3270
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.517046332359314,
+      "learning_rate": 9.832500153813234e-05,
+      "loss": 2.6987,
+      "step": 3275
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.49250227212905884,
+      "learning_rate": 9.831975156969498e-05,
+      "loss": 2.6335,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.49026939272880554,
+      "learning_rate": 9.831449352720677e-05,
+      "loss": 2.6859,
+      "step": 3285
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.49874892830848694,
+      "learning_rate": 9.830922741154631e-05,
+      "loss": 2.6655,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4574241638183594,
+      "learning_rate": 9.830395322359357e-05,
+      "loss": 2.6863,
+      "step": 3295
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47240307927131653,
+      "learning_rate": 9.829867096422983e-05,
+      "loss": 2.702,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47739139199256897,
+      "learning_rate": 9.829338063433773e-05,
+      "loss": 2.6996,
+      "step": 3305
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46502938866615295,
+      "learning_rate": 9.828808223480126e-05,
+      "loss": 2.7402,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4774216115474701,
+      "learning_rate": 9.828277576650578e-05,
+      "loss": 2.6645,
+      "step": 3315
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4785415232181549,
+      "learning_rate": 9.827746123033796e-05,
+      "loss": 2.7052,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4438062012195587,
+      "learning_rate": 9.827213862718587e-05,
+      "loss": 2.7874,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47122207283973694,
+      "learning_rate": 9.826680795793887e-05,
+      "loss": 2.5818,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5015072226524353,
+      "learning_rate": 9.826146922348773e-05,
+      "loss": 2.755,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46495261788368225,
+      "learning_rate": 9.825612242472448e-05,
+      "loss": 2.7493,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5018097162246704,
+      "learning_rate": 9.82507675625426e-05,
+      "loss": 2.6804,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4791370630264282,
+      "learning_rate": 9.824540463783682e-05,
+      "loss": 2.6961,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4463869631290436,
+      "learning_rate": 9.824003365150329e-05,
+      "loss": 2.6513,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4927726984024048,
+      "learning_rate": 9.82346546044395e-05,
+      "loss": 2.7968,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4694409966468811,
+      "learning_rate": 9.822926749754424e-05,
+      "loss": 2.7208,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.48208680748939514,
+      "learning_rate": 9.822387233171771e-05,
+      "loss": 2.6943,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46970227360725403,
+      "learning_rate": 9.821846910786136e-05,
+      "loss": 2.6824,
+      "step": 3375
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4563305974006653,
+      "learning_rate": 9.821305782687812e-05,
+      "loss": 2.6542,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4760410487651825,
+      "learning_rate": 9.820763848967215e-05,
+      "loss": 2.6817,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4696536958217621,
+      "learning_rate": 9.820221109714902e-05,
+      "loss": 2.7677,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46862608194351196,
+      "learning_rate": 9.81967756502156e-05,
+      "loss": 2.7377,
+      "step": 3395
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4967498481273651,
+      "learning_rate": 9.819133214978017e-05,
+      "loss": 2.7366,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4818565845489502,
+      "learning_rate": 9.818588059675229e-05,
+      "loss": 2.7036,
+      "step": 3405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.466817706823349,
+      "learning_rate": 9.818042099204292e-05,
+      "loss": 2.6388,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4831058382987976,
+      "learning_rate": 9.817495333656433e-05,
+      "loss": 2.6477,
+      "step": 3415
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46210381388664246,
+      "learning_rate": 9.816947763123013e-05,
+      "loss": 2.6227,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4551011323928833,
+      "learning_rate": 9.816399387695529e-05,
+      "loss": 2.662,
+      "step": 3425
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.49353134632110596,
+      "learning_rate": 9.815850207465616e-05,
+      "loss": 2.6875,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46854260563850403,
+      "learning_rate": 9.815300222525035e-05,
+      "loss": 2.6816,
+      "step": 3435
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.48094117641448975,
+      "learning_rate": 9.814749432965691e-05,
+      "loss": 2.5564,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.48280051350593567,
+      "learning_rate": 9.814197838879615e-05,
+      "loss": 2.6996,
+      "step": 3445
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4708018898963928,
+      "learning_rate": 9.813645440358981e-05,
+      "loss": 2.6743,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4690140187740326,
+      "learning_rate": 9.813092237496087e-05,
+      "loss": 2.7559,
+      "step": 3455
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4613482654094696,
+      "learning_rate": 9.812538230383376e-05,
+      "loss": 2.6355,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4627975523471832,
+      "learning_rate": 9.811983419113418e-05,
+      "loss": 2.6983,
+      "step": 3465
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8663000464439392,
+      "learning_rate": 9.81142780377892e-05,
+      "loss": 2.7452,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4756057858467102,
+      "learning_rate": 9.810871384472724e-05,
+      "loss": 2.7169,
+      "step": 3475
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4795553386211395,
+      "learning_rate": 9.810314161287806e-05,
+      "loss": 2.675,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7104564905166626,
+      "learning_rate": 9.809756134317276e-05,
+      "loss": 2.704,
+      "step": 3485
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4432218670845032,
+      "learning_rate": 9.809197303654376e-05,
+      "loss": 2.7469,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46970778703689575,
+      "learning_rate": 9.808637669392487e-05,
+      "loss": 2.6462,
+      "step": 3495
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.48928606510162354,
+      "learning_rate": 9.808077231625119e-05,
+      "loss": 2.684,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4625703990459442,
+      "learning_rate": 9.807515990445922e-05,
+      "loss": 2.6199,
+      "step": 3505
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4593104422092438,
+      "learning_rate": 9.806953945948676e-05,
+      "loss": 2.6489,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45209789276123047,
+      "learning_rate": 9.806391098227297e-05,
+      "loss": 2.709,
+      "step": 3515
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44082218408584595,
+      "learning_rate": 9.805827447375835e-05,
+      "loss": 2.7346,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4719259440898895,
+      "learning_rate": 9.805262993488473e-05,
+      "loss": 2.6525,
+      "step": 3525
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45654940605163574,
+      "learning_rate": 9.804697736659532e-05,
+      "loss": 2.7006,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44832736253738403,
+      "learning_rate": 9.804131676983458e-05,
+      "loss": 2.7157,
+      "step": 3535
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5046043992042542,
+      "learning_rate": 9.803564814554844e-05,
+      "loss": 2.6532,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4667973518371582,
+      "learning_rate": 9.802997149468408e-05,
+      "loss": 2.6358,
+      "step": 3545
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4885671138763428,
+      "learning_rate": 9.802428681819003e-05,
+      "loss": 2.6677,
+      "step": 3550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.48970192670822144,
+      "learning_rate": 9.801859411701621e-05,
+      "loss": 2.7361,
+      "step": 3555
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44877639412879944,
+      "learning_rate": 9.801289339211384e-05,
+      "loss": 2.6483,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4444518983364105,
+      "learning_rate": 9.800718464443547e-05,
+      "loss": 2.6258,
+      "step": 3565
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4697633385658264,
+      "learning_rate": 9.800146787493503e-05,
+      "loss": 2.7095,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5028236508369446,
+      "learning_rate": 9.799574308456775e-05,
+      "loss": 2.5583,
+      "step": 3575
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4589617848396301,
+      "learning_rate": 9.799001027429026e-05,
+      "loss": 2.719,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.451840877532959,
+      "learning_rate": 9.798426944506048e-05,
+      "loss": 2.694,
+      "step": 3585
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4443571865558624,
+      "learning_rate": 9.797852059783766e-05,
+      "loss": 2.812,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4731528162956238,
+      "learning_rate": 9.79727637335824e-05,
+      "loss": 2.6827,
+      "step": 3595
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4516463577747345,
+      "learning_rate": 9.79669988532567e-05,
+      "loss": 2.6045,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46725428104400635,
+      "learning_rate": 9.79612259578238e-05,
+      "loss": 2.6305,
+      "step": 3605
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4663410782814026,
+      "learning_rate": 9.795544504824836e-05,
+      "loss": 2.6642,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45506349205970764,
+      "learning_rate": 9.794965612549635e-05,
+      "loss": 2.6397,
+      "step": 3615
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4761527478694916,
+      "learning_rate": 9.794385919053506e-05,
+      "loss": 2.6289,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4458506405353546,
+      "learning_rate": 9.793805424433313e-05,
+      "loss": 2.6625,
+      "step": 3625
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47145402431488037,
+      "learning_rate": 9.793224128786058e-05,
+      "loss": 2.7404,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.49709179997444153,
+      "learning_rate": 9.79264203220887e-05,
+      "loss": 2.6449,
+      "step": 3635
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45615118741989136,
+      "learning_rate": 9.792059134799016e-05,
+      "loss": 2.6077,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4632767140865326,
+      "learning_rate": 9.791475436653896e-05,
+      "loss": 2.671,
+      "step": 3645
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46377870440483093,
+      "learning_rate": 9.790890937871046e-05,
+      "loss": 2.638,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43299993872642517,
+      "learning_rate": 9.79030563854813e-05,
+      "loss": 2.7455,
+      "step": 3655
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44504308700561523,
+      "learning_rate": 9.789719538782952e-05,
+      "loss": 2.7091,
+      "step": 3660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.49651631712913513,
+      "learning_rate": 9.789132638673445e-05,
+      "loss": 2.6545,
+      "step": 3665
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45269301533699036,
+      "learning_rate": 9.78854493831768e-05,
+      "loss": 2.6075,
+      "step": 3670
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47577160596847534,
+      "learning_rate": 9.787956437813859e-05,
+      "loss": 2.6639,
+      "step": 3675
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.455028235912323,
+      "learning_rate": 9.787367137260317e-05,
+      "loss": 2.6678,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4394490718841553,
+      "learning_rate": 9.786777036755525e-05,
+      "loss": 2.5793,
+      "step": 3685
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4365720748901367,
+      "learning_rate": 9.786186136398085e-05,
+      "loss": 2.5489,
+      "step": 3690
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.48585519194602966,
+      "learning_rate": 9.785594436286738e-05,
+      "loss": 2.6967,
+      "step": 3695
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44858038425445557,
+      "learning_rate": 9.78500193652035e-05,
+      "loss": 2.6162,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4587553143501282,
+      "learning_rate": 9.784408637197929e-05,
+      "loss": 2.7097,
+      "step": 3705
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47037503123283386,
+      "learning_rate": 9.783814538418612e-05,
+      "loss": 2.6305,
+      "step": 3710
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5158719420433044,
+      "learning_rate": 9.783219640281671e-05,
+      "loss": 2.7348,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4614676535129547,
+      "learning_rate": 9.78262394288651e-05,
+      "loss": 2.7363,
+      "step": 3720
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44795313477516174,
+      "learning_rate": 9.782027446332671e-05,
+      "loss": 2.6296,
+      "step": 3725
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4497576951980591,
+      "learning_rate": 9.781430150719822e-05,
+      "loss": 2.6117,
+      "step": 3730
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4469236433506012,
+      "learning_rate": 9.780832056147772e-05,
+      "loss": 2.6916,
+      "step": 3735
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4670577645301819,
+      "learning_rate": 9.78023316271646e-05,
+      "loss": 2.6915,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4494103193283081,
+      "learning_rate": 9.779633470525957e-05,
+      "loss": 2.6948,
+      "step": 3745
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6939617395401001,
+      "learning_rate": 9.77903297967647e-05,
+      "loss": 2.602,
+      "step": 3750
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47299855947494507,
+      "learning_rate": 9.778431690268341e-05,
+      "loss": 2.6834,
+      "step": 3755
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4501034915447235,
+      "learning_rate": 9.777829602402043e-05,
+      "loss": 2.6095,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46022433042526245,
+      "learning_rate": 9.777226716178178e-05,
+      "loss": 2.6128,
+      "step": 3765
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.451348215341568,
+      "learning_rate": 9.776623031697492e-05,
+      "loss": 2.6788,
+      "step": 3770
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4470846951007843,
+      "learning_rate": 9.776018549060855e-05,
+      "loss": 2.63,
+      "step": 3775
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44781848788261414,
+      "learning_rate": 9.775413268369274e-05,
+      "loss": 2.6488,
+      "step": 3780
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4540098309516907,
+      "learning_rate": 9.77480718972389e-05,
+      "loss": 2.7276,
+      "step": 3785
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4752447009086609,
+      "learning_rate": 9.774200313225976e-05,
+      "loss": 2.632,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43620023131370544,
+      "learning_rate": 9.773592638976939e-05,
+      "loss": 2.6545,
+      "step": 3795
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43904945254325867,
+      "learning_rate": 9.77298416707832e-05,
+      "loss": 2.6761,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4729931056499481,
+      "learning_rate": 9.772374897631791e-05,
+      "loss": 2.6566,
+      "step": 3805
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46864768862724304,
+      "learning_rate": 9.771764830739159e-05,
+      "loss": 2.6555,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4787927567958832,
+      "learning_rate": 9.771153966502365e-05,
+      "loss": 2.7037,
+      "step": 3815
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4806594252586365,
+      "learning_rate": 9.77054230502348e-05,
+      "loss": 2.6394,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.48692581057548523,
+      "learning_rate": 9.769929846404711e-05,
+      "loss": 2.5801,
+      "step": 3825
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4725556969642639,
+      "learning_rate": 9.769316590748401e-05,
+      "loss": 2.6879,
+      "step": 3830
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4904841482639313,
+      "learning_rate": 9.768702538157018e-05,
+      "loss": 2.6225,
+      "step": 3835
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.328408718109131,
+      "learning_rate": 9.76808768873317e-05,
+      "loss": 2.6563,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45041215419769287,
+      "learning_rate": 9.767472042579595e-05,
+      "loss": 2.757,
+      "step": 3845
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4601709246635437,
+      "learning_rate": 9.766855599799168e-05,
+      "loss": 2.5673,
+      "step": 3850
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47665128111839294,
+      "learning_rate": 9.76623836049489e-05,
+      "loss": 2.6578,
+      "step": 3855
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.538678765296936,
+      "learning_rate": 9.765620324769903e-05,
+      "loss": 2.6603,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4757884442806244,
+      "learning_rate": 9.765001492727476e-05,
+      "loss": 2.6839,
+      "step": 3865
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5321572422981262,
+      "learning_rate": 9.764381864471017e-05,
+      "loss": 2.6712,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4851582944393158,
+      "learning_rate": 9.76376144010406e-05,
+      "loss": 2.6123,
+      "step": 3875
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4795767366886139,
+      "learning_rate": 9.763140219730278e-05,
+      "loss": 2.6749,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46437713503837585,
+      "learning_rate": 9.762518203453472e-05,
+      "loss": 2.6988,
+      "step": 3885
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44960570335388184,
+      "learning_rate": 9.761895391377581e-05,
+      "loss": 2.6925,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4717967212200165,
+      "learning_rate": 9.761271783606674e-05,
+      "loss": 2.6331,
+      "step": 3895
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4433802366256714,
+      "learning_rate": 9.760647380244952e-05,
+      "loss": 2.5515,
+      "step": 3900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4423994719982147,
+      "learning_rate": 9.760022181396753e-05,
+      "loss": 2.6673,
+      "step": 3905
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4311283528804779,
+      "learning_rate": 9.759396187166543e-05,
+      "loss": 2.6342,
+      "step": 3910
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.48175472021102905,
+      "learning_rate": 9.758769397658926e-05,
+      "loss": 2.6015,
+      "step": 3915
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4609222412109375,
+      "learning_rate": 9.758141812978634e-05,
+      "loss": 2.6753,
+      "step": 3920
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.472646027803421,
+      "learning_rate": 9.757513433230534e-05,
+      "loss": 2.6857,
+      "step": 3925
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45973503589630127,
+      "learning_rate": 9.756884258519627e-05,
+      "loss": 2.6661,
+      "step": 3930
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4948663115501404,
+      "learning_rate": 9.756254288951046e-05,
+      "loss": 2.6909,
+      "step": 3935
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4636407494544983,
+      "learning_rate": 9.755623524630055e-05,
+      "loss": 2.5633,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4298071563243866,
+      "learning_rate": 9.754991965662055e-05,
+      "loss": 2.5557,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4571687877178192,
+      "learning_rate": 9.754359612152574e-05,
+      "loss": 2.6234,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44791626930236816,
+      "learning_rate": 9.753726464207278e-05,
+      "loss": 2.644,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4616393744945526,
+      "learning_rate": 9.753092521931963e-05,
+      "loss": 2.627,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.494600772857666,
+      "learning_rate": 9.752457785432558e-05,
+      "loss": 2.6378,
+      "step": 3965
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0571376085281372,
+      "learning_rate": 9.751822254815126e-05,
+      "loss": 2.6758,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42886582016944885,
+      "learning_rate": 9.751185930185861e-05,
+      "loss": 2.6015,
+      "step": 3975
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44739213585853577,
+      "learning_rate": 9.750548811651091e-05,
+      "loss": 2.714,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4368375539779663,
+      "learning_rate": 9.749910899317276e-05,
+      "loss": 2.6899,
+      "step": 3985
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46134066581726074,
+      "learning_rate": 9.74927219329101e-05,
+      "loss": 2.6674,
+      "step": 3990
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4207320809364319,
+      "learning_rate": 9.748632693679016e-05,
+      "loss": 2.6575,
+      "step": 3995
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43614500761032104,
+      "learning_rate": 9.747992400588153e-05,
+      "loss": 2.6039,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42727071046829224,
+      "learning_rate": 9.747351314125413e-05,
+      "loss": 2.6236,
+      "step": 4005
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4672296941280365,
+      "learning_rate": 9.746709434397916e-05,
+      "loss": 2.654,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4414095878601074,
+      "learning_rate": 9.746066761512921e-05,
+      "loss": 2.7373,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44531089067459106,
+      "learning_rate": 9.745423295577815e-05,
+      "loss": 2.7208,
+      "step": 4020
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4542682468891144,
+      "learning_rate": 9.744779036700118e-05,
+      "loss": 2.6198,
+      "step": 4025
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4374953806400299,
+      "learning_rate": 9.744133984987485e-05,
+      "loss": 2.688,
+      "step": 4030
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4767053425312042,
+      "learning_rate": 9.743488140547701e-05,
+      "loss": 2.6254,
+      "step": 4035
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.455422580242157,
+      "learning_rate": 9.742841503488682e-05,
+      "loss": 2.6225,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41789522767066956,
+      "learning_rate": 9.742194073918482e-05,
+      "loss": 2.667,
+      "step": 4045
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4357880651950836,
+      "learning_rate": 9.741545851945282e-05,
+      "loss": 2.5513,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48044779896736145,
+      "learning_rate": 9.740896837677398e-05,
+      "loss": 2.632,
+      "step": 4055
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6033629775047302,
+      "learning_rate": 9.740247031223279e-05,
+      "loss": 2.6417,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4413008689880371,
+      "learning_rate": 9.739596432691502e-05,
+      "loss": 2.6632,
+      "step": 4065
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48309609293937683,
+      "learning_rate": 9.738945042190785e-05,
+      "loss": 2.6018,
+      "step": 4070
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4507386088371277,
+      "learning_rate": 9.738292859829968e-05,
+      "loss": 2.5908,
+      "step": 4075
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4712318181991577,
+      "learning_rate": 9.73763988571803e-05,
+      "loss": 2.5956,
+      "step": 4080
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4821806848049164,
+      "learning_rate": 9.73698611996408e-05,
+      "loss": 2.6004,
+      "step": 4085
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4476467967033386,
+      "learning_rate": 9.736331562677361e-05,
+      "loss": 2.6794,
+      "step": 4090
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4711425304412842,
+      "learning_rate": 9.735676213967247e-05,
+      "loss": 2.688,
+      "step": 4095
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4471677541732788,
+      "learning_rate": 9.735020073943244e-05,
+      "loss": 2.6259,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4406508207321167,
+      "learning_rate": 9.734363142714988e-05,
+      "loss": 2.6744,
+      "step": 4105
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4376254677772522,
+      "learning_rate": 9.733705420392254e-05,
+      "loss": 2.6459,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45931753516197205,
+      "learning_rate": 9.733046907084944e-05,
+      "loss": 2.647,
+      "step": 4115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45559918880462646,
+      "learning_rate": 9.732387602903091e-05,
+      "loss": 2.7104,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4632091522216797,
+      "learning_rate": 9.731727507956862e-05,
+      "loss": 2.6075,
+      "step": 4125
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46415290236473083,
+      "learning_rate": 9.73106662235656e-05,
+      "loss": 2.579,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.453960657119751,
+      "learning_rate": 9.730404946212614e-05,
+      "loss": 2.5816,
+      "step": 4135
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45433130860328674,
+      "learning_rate": 9.729742479635589e-05,
+      "loss": 2.5565,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.460023432970047,
+      "learning_rate": 9.729079222736177e-05,
+      "loss": 2.5765,
+      "step": 4145
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4378867447376251,
+      "learning_rate": 9.728415175625211e-05,
+      "loss": 2.6229,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46070396900177,
+      "learning_rate": 9.727750338413647e-05,
+      "loss": 2.6524,
+      "step": 4155
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43714022636413574,
+      "learning_rate": 9.727084711212579e-05,
+      "loss": 2.6102,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4564690887928009,
+      "learning_rate": 9.726418294133229e-05,
+      "loss": 2.5983,
+      "step": 4165
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4709508419036865,
+      "learning_rate": 9.725751087286954e-05,
+      "loss": 2.7022,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4441114366054535,
+      "learning_rate": 9.725083090785242e-05,
+      "loss": 2.6998,
+      "step": 4175
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4681941568851471,
+      "learning_rate": 9.724414304739712e-05,
+      "loss": 2.591,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4607504904270172,
+      "learning_rate": 9.723744729262116e-05,
+      "loss": 2.6352,
+      "step": 4185
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4364299774169922,
+      "learning_rate": 9.723074364464337e-05,
+      "loss": 2.6441,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.440005898475647,
+      "learning_rate": 9.722403210458393e-05,
+      "loss": 2.6804,
+      "step": 4195
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4373852014541626,
+      "learning_rate": 9.721731267356426e-05,
+      "loss": 2.5563,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4685074985027313,
+      "learning_rate": 9.721058535270722e-05,
+      "loss": 2.6828,
+      "step": 4205
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4627612829208374,
+      "learning_rate": 9.720385014313686e-05,
+      "loss": 2.6066,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4650856554508209,
+      "learning_rate": 9.719710704597865e-05,
+      "loss": 2.653,
+      "step": 4215
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46641984581947327,
+      "learning_rate": 9.719035606235931e-05,
+      "loss": 2.6144,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46783891320228577,
+      "learning_rate": 9.718359719340693e-05,
+      "loss": 2.5703,
+      "step": 4225
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46737152338027954,
+      "learning_rate": 9.717683044025086e-05,
+      "loss": 2.676,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4733370542526245,
+      "learning_rate": 9.717005580402184e-05,
+      "loss": 2.615,
+      "step": 4235
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44572392106056213,
+      "learning_rate": 9.716327328585183e-05,
+      "loss": 2.6043,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42774394154548645,
+      "learning_rate": 9.715648288687421e-05,
+      "loss": 2.5733,
+      "step": 4245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45602691173553467,
+      "learning_rate": 9.714968460822363e-05,
+      "loss": 2.6572,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4395467936992645,
+      "learning_rate": 9.714287845103602e-05,
+      "loss": 2.6027,
+      "step": 4255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4580719769001007,
+      "learning_rate": 9.71360644164487e-05,
+      "loss": 2.6521,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44203102588653564,
+      "learning_rate": 9.712924250560025e-05,
+      "loss": 2.6278,
+      "step": 4265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44138801097869873,
+      "learning_rate": 9.712241271963059e-05,
+      "loss": 2.6583,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44238343834877014,
+      "learning_rate": 9.711557505968097e-05,
+      "loss": 2.5736,
+      "step": 4275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46433305740356445,
+      "learning_rate": 9.710872952689391e-05,
+      "loss": 2.6362,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46467325091362,
+      "learning_rate": 9.71018761224133e-05,
+      "loss": 2.6514,
+      "step": 4285
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4357128143310547,
+      "learning_rate": 9.709501484738429e-05,
+      "loss": 2.5921,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42506784200668335,
+      "learning_rate": 9.70881457029534e-05,
+      "loss": 2.6226,
+      "step": 4295
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43655288219451904,
+      "learning_rate": 9.708126869026841e-05,
+      "loss": 2.6587,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.436148464679718,
+      "learning_rate": 9.707438381047846e-05,
+      "loss": 2.7148,
+      "step": 4305
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45658332109451294,
+      "learning_rate": 9.7067491064734e-05,
+      "loss": 2.5335,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5743585824966431,
+      "learning_rate": 9.706059045418676e-05,
+      "loss": 2.6984,
+      "step": 4315
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4573557376861572,
+      "learning_rate": 9.705368197998982e-05,
+      "loss": 2.6159,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46645599603652954,
+      "learning_rate": 9.704676564329756e-05,
+      "loss": 2.5614,
+      "step": 4325
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4437452256679535,
+      "learning_rate": 9.703984144526565e-05,
+      "loss": 2.6626,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5487425327301025,
+      "learning_rate": 9.703290938705114e-05,
+      "loss": 2.6374,
+      "step": 4335
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4546929597854614,
+      "learning_rate": 9.702596946981232e-05,
+      "loss": 2.5978,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4414508044719696,
+      "learning_rate": 9.701902169470883e-05,
+      "loss": 2.5674,
+      "step": 4345
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43854445219039917,
+      "learning_rate": 9.701206606290164e-05,
+      "loss": 2.5672,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43143099546432495,
+      "learning_rate": 9.700510257555295e-05,
+      "loss": 2.6724,
+      "step": 4355
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2194734811782837,
+      "learning_rate": 9.69981312338264e-05,
+      "loss": 2.5872,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44143447279930115,
+      "learning_rate": 9.699115203888686e-05,
+      "loss": 2.6795,
+      "step": 4365
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4504442811012268,
+      "learning_rate": 9.69841649919005e-05,
+      "loss": 2.6197,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44425585865974426,
+      "learning_rate": 9.697717009403485e-05,
+      "loss": 2.6262,
+      "step": 4375
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4563242793083191,
+      "learning_rate": 9.697016734645873e-05,
+      "loss": 2.5885,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4582779109477997,
+      "learning_rate": 9.696315675034227e-05,
+      "loss": 2.5492,
+      "step": 4385
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6134490966796875,
+      "learning_rate": 9.69561383068569e-05,
+      "loss": 2.6515,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44351518154144287,
+      "learning_rate": 9.694911201717541e-05,
+      "loss": 2.6398,
+      "step": 4395
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.47304531931877136,
+      "learning_rate": 9.694207788247186e-05,
+      "loss": 2.6666,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4427909553050995,
+      "learning_rate": 9.693503590392159e-05,
+      "loss": 2.6025,
+      "step": 4405
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48961570858955383,
+      "learning_rate": 9.692798608270132e-05,
+      "loss": 2.6154,
+      "step": 4410
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48413747549057007,
+      "learning_rate": 9.692092841998904e-05,
+      "loss": 2.6933,
+      "step": 4415
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45368140935897827,
+      "learning_rate": 9.691386291696407e-05,
+      "loss": 2.7363,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.434784471988678,
+      "learning_rate": 9.690678957480702e-05,
+      "loss": 2.6139,
+      "step": 4425
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42487633228302,
+      "learning_rate": 9.689970839469981e-05,
+      "loss": 2.6677,
+      "step": 4430
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4479994475841522,
+      "learning_rate": 9.689261937782571e-05,
+      "loss": 2.6045,
+      "step": 4435
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.429272323846817,
+      "learning_rate": 9.688552252536924e-05,
+      "loss": 2.6238,
+      "step": 4440
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42785945534706116,
+      "learning_rate": 9.687841783851626e-05,
+      "loss": 2.6271,
+      "step": 4445
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4483790993690491,
+      "learning_rate": 9.687130531845394e-05,
+      "loss": 2.6301,
+      "step": 4450
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4295051395893097,
+      "learning_rate": 9.686418496637076e-05,
+      "loss": 2.6144,
+      "step": 4455
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5474414229393005,
+      "learning_rate": 9.685705678345652e-05,
+      "loss": 2.6079,
+      "step": 4460
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4479367733001709,
+      "learning_rate": 9.684992077090227e-05,
+      "loss": 2.6305,
+      "step": 4465
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47512468695640564,
+      "learning_rate": 9.684277692990045e-05,
+      "loss": 2.6108,
+      "step": 4470
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4287829101085663,
+      "learning_rate": 9.683562526164475e-05,
+      "loss": 2.5836,
+      "step": 4475
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45058339834213257,
+      "learning_rate": 9.68284657673302e-05,
+      "loss": 2.542,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.452127605676651,
+      "learning_rate": 9.682129844815311e-05,
+      "loss": 2.6147,
+      "step": 4485
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4142586588859558,
+      "learning_rate": 9.681412330531112e-05,
+      "loss": 2.5931,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44630104303359985,
+      "learning_rate": 9.680694034000318e-05,
+      "loss": 2.5587,
+      "step": 4495
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4468352496623993,
+      "learning_rate": 9.67997495534295e-05,
+      "loss": 2.5974,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4442676901817322,
+      "learning_rate": 9.679255094679169e-05,
+      "loss": 2.5082,
+      "step": 4505
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43427979946136475,
+      "learning_rate": 9.678534452129256e-05,
+      "loss": 2.5954,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4230964481830597,
+      "learning_rate": 9.677813027813632e-05,
+      "loss": 2.6305,
+      "step": 4515
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44261035323143005,
+      "learning_rate": 9.677090821852842e-05,
+      "loss": 2.7107,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4132828116416931,
+      "learning_rate": 9.676367834367562e-05,
+      "loss": 2.5846,
+      "step": 4525
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45867741107940674,
+      "learning_rate": 9.675644065478602e-05,
+      "loss": 2.7023,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43606331944465637,
+      "learning_rate": 9.674919515306904e-05,
+      "loss": 2.5448,
+      "step": 4535
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4151594340801239,
+      "learning_rate": 9.674194183973533e-05,
+      "loss": 2.5913,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43477490544319153,
+      "learning_rate": 9.673468071599694e-05,
+      "loss": 2.7136,
+      "step": 4545
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43338918685913086,
+      "learning_rate": 9.672741178306713e-05,
+      "loss": 2.6662,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5320208668708801,
+      "learning_rate": 9.672013504216054e-05,
+      "loss": 2.5576,
+      "step": 4555
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40916314721107483,
+      "learning_rate": 9.671285049449307e-05,
+      "loss": 2.6193,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.49342936277389526,
+      "learning_rate": 9.670555814128197e-05,
+      "loss": 2.5804,
+      "step": 4565
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47213032841682434,
+      "learning_rate": 9.669825798374574e-05,
+      "loss": 2.6344,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43034955859184265,
+      "learning_rate": 9.66909500231042e-05,
+      "loss": 2.5364,
+      "step": 4575
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43315377831459045,
+      "learning_rate": 9.668363426057852e-05,
+      "loss": 2.5594,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4056445062160492,
+      "learning_rate": 9.66763106973911e-05,
+      "loss": 2.6513,
+      "step": 4585
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4561117887496948,
+      "learning_rate": 9.666897933476568e-05,
+      "loss": 2.631,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42313024401664734,
+      "learning_rate": 9.666164017392734e-05,
+      "loss": 2.5258,
+      "step": 4595
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8057920932769775,
+      "learning_rate": 9.665429321610239e-05,
+      "loss": 2.6251,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44359615445137024,
+      "learning_rate": 9.66469384625185e-05,
+      "loss": 2.5413,
+      "step": 4605
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42278745770454407,
+      "learning_rate": 9.66395759144046e-05,
+      "loss": 2.6051,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44322043657302856,
+      "learning_rate": 9.6632205572991e-05,
+      "loss": 2.7094,
+      "step": 4615
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4160989224910736,
+      "learning_rate": 9.662482743950918e-05,
+      "loss": 2.576,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4590180814266205,
+      "learning_rate": 9.661744151519206e-05,
+      "loss": 2.6747,
+      "step": 4625
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4923911988735199,
+      "learning_rate": 9.661004780127373e-05,
+      "loss": 2.5757,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.46068045496940613,
+      "learning_rate": 9.660264629898974e-05,
+      "loss": 2.6323,
+      "step": 4635
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45393189787864685,
+      "learning_rate": 9.65952370095768e-05,
+      "loss": 2.6051,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4184952676296234,
+      "learning_rate": 9.6587819934273e-05,
+      "loss": 2.4936,
+      "step": 4645
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42835912108421326,
+      "learning_rate": 9.658039507431768e-05,
+      "loss": 2.6475,
+      "step": 4650
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4460197687149048,
+      "learning_rate": 9.657296243095151e-05,
+      "loss": 2.5666,
+      "step": 4655
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4295268654823303,
+      "learning_rate": 9.656552200541648e-05,
+      "loss": 2.5816,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4362679123878479,
+      "learning_rate": 9.655807379895583e-05,
+      "loss": 2.5206,
+      "step": 4665
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4387598931789398,
+      "learning_rate": 9.655061781281413e-05,
+      "loss": 2.6061,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42419132590293884,
+      "learning_rate": 9.654315404823729e-05,
+      "loss": 2.6148,
+      "step": 4675
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45118197798728943,
+      "learning_rate": 9.653568250647242e-05,
+      "loss": 2.569,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42909640073776245,
+      "learning_rate": 9.652820318876802e-05,
+      "loss": 2.6687,
+      "step": 4685
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4224296808242798,
+      "learning_rate": 9.652071609637385e-05,
+      "loss": 2.6229,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45114147663116455,
+      "learning_rate": 9.651322123054097e-05,
+      "loss": 2.525,
+      "step": 4695
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42290931940078735,
+      "learning_rate": 9.650571859252175e-05,
+      "loss": 2.6203,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4412761330604553,
+      "learning_rate": 9.649820818356985e-05,
+      "loss": 2.6164,
+      "step": 4705
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4252312183380127,
+      "learning_rate": 9.649069000494025e-05,
+      "loss": 2.5934,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43498319387435913,
+      "learning_rate": 9.648316405788916e-05,
+      "loss": 2.5539,
+      "step": 4715
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4346221089363098,
+      "learning_rate": 9.647563034367419e-05,
+      "loss": 2.5706,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4147849380970001,
+      "learning_rate": 9.646808886355417e-05,
+      "loss": 2.6134,
+      "step": 4725
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4281368553638458,
+      "learning_rate": 9.646053961878925e-05,
+      "loss": 2.476,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42818668484687805,
+      "learning_rate": 9.64529826106409e-05,
+      "loss": 2.6142,
+      "step": 4735
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4243021607398987,
+      "learning_rate": 9.644541784037186e-05,
+      "loss": 2.6014,
+      "step": 4740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44032031297683716,
+      "learning_rate": 9.643784530924616e-05,
+      "loss": 2.5568,
+      "step": 4745
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.41509535908699036,
+      "learning_rate": 9.643026501852916e-05,
+      "loss": 2.6213,
+      "step": 4750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4436090886592865,
+      "learning_rate": 9.64226769694875e-05,
+      "loss": 2.5149,
+      "step": 4755
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4368179738521576,
+      "learning_rate": 9.641508116338908e-05,
+      "loss": 2.5847,
+      "step": 4760
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.431039959192276,
+      "learning_rate": 9.640747760150319e-05,
+      "loss": 2.5417,
+      "step": 4765
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4300715923309326,
+      "learning_rate": 9.639986628510032e-05,
+      "loss": 2.6377,
+      "step": 4770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4452112317085266,
+      "learning_rate": 9.639224721545229e-05,
+      "loss": 2.5228,
+      "step": 4775
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4239508807659149,
+      "learning_rate": 9.638462039383223e-05,
+      "loss": 2.5218,
+      "step": 4780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4282490015029907,
+      "learning_rate": 9.637698582151454e-05,
+      "loss": 2.5999,
+      "step": 4785
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4104209840297699,
+      "learning_rate": 9.636934349977493e-05,
+      "loss": 2.6569,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4533073902130127,
+      "learning_rate": 9.636169342989043e-05,
+      "loss": 2.5486,
+      "step": 4795
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45976290106773376,
+      "learning_rate": 9.635403561313932e-05,
+      "loss": 2.5855,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42636388540267944,
+      "learning_rate": 9.634637005080119e-05,
+      "loss": 2.5965,
+      "step": 4805
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40920981764793396,
+      "learning_rate": 9.633869674415693e-05,
+      "loss": 2.5985,
+      "step": 4810
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4292509853839874,
+      "learning_rate": 9.633101569448873e-05,
+      "loss": 2.667,
+      "step": 4815
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4465123414993286,
+      "learning_rate": 9.632332690308004e-05,
+      "loss": 2.6471,
+      "step": 4820
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4507482647895813,
+      "learning_rate": 9.631563037121565e-05,
+      "loss": 2.5746,
+      "step": 4825
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42116397619247437,
+      "learning_rate": 9.63079261001816e-05,
+      "loss": 2.571,
+      "step": 4830
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4215540289878845,
+      "learning_rate": 9.630021409126529e-05,
+      "loss": 2.5717,
+      "step": 4835
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4463818371295929,
+      "learning_rate": 9.629249434575531e-05,
+      "loss": 2.6347,
+      "step": 4840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43039610981941223,
+      "learning_rate": 9.628476686494164e-05,
+      "loss": 2.6388,
+      "step": 4845
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4435877203941345,
+      "learning_rate": 9.627703165011551e-05,
+      "loss": 2.5859,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4407738149166107,
+      "learning_rate": 9.626928870256944e-05,
+      "loss": 2.669,
+      "step": 4855
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43019118905067444,
+      "learning_rate": 9.626153802359724e-05,
+      "loss": 2.58,
+      "step": 4860
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4782212972640991,
+      "learning_rate": 9.625377961449402e-05,
+      "loss": 2.6199,
+      "step": 4865
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4269406199455261,
+      "learning_rate": 9.62460134765562e-05,
+      "loss": 2.6848,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4501302242279053,
+      "learning_rate": 9.623823961108145e-05,
+      "loss": 2.5115,
+      "step": 4875
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42027103900909424,
+      "learning_rate": 9.623045801936876e-05,
+      "loss": 2.5616,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.49340420961380005,
+      "learning_rate": 9.622266870271842e-05,
+      "loss": 2.5795,
+      "step": 4885
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4740692377090454,
+      "learning_rate": 9.6214871662432e-05,
+      "loss": 2.5894,
+      "step": 4890
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4227573573589325,
+      "learning_rate": 9.620706689981232e-05,
+      "loss": 2.5845,
+      "step": 4895
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.459430992603302,
+      "learning_rate": 9.619925441616354e-05,
+      "loss": 2.6363,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.46070754528045654,
+      "learning_rate": 9.619143421279112e-05,
+      "loss": 2.6723,
+      "step": 4905
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41950926184654236,
+      "learning_rate": 9.618360629100176e-05,
+      "loss": 2.6047,
+      "step": 4910
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4319295585155487,
+      "learning_rate": 9.61757706521035e-05,
+      "loss": 2.6677,
+      "step": 4915
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4238024652004242,
+      "learning_rate": 9.616792729740564e-05,
+      "loss": 2.5574,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.40642815828323364,
+      "learning_rate": 9.616007622821877e-05,
+      "loss": 2.5506,
+      "step": 4925
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.45434537529945374,
+      "learning_rate": 9.615221744585476e-05,
+      "loss": 2.6201,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43558722734451294,
+      "learning_rate": 9.614435095162679e-05,
+      "loss": 2.514,
+      "step": 4935
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.44114238023757935,
+      "learning_rate": 9.613647674684936e-05,
+      "loss": 2.6691,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43615633249282837,
+      "learning_rate": 9.612859483283818e-05,
+      "loss": 2.5694,
+      "step": 4945
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.44374772906303406,
+      "learning_rate": 9.612070521091029e-05,
+      "loss": 2.6251,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41793927550315857,
+      "learning_rate": 9.611280788238403e-05,
+      "loss": 2.5372,
+      "step": 4955
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4168284833431244,
+      "learning_rate": 9.610490284857901e-05,
+      "loss": 2.6401,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4371614158153534,
+      "learning_rate": 9.609699011081613e-05,
+      "loss": 2.5374,
+      "step": 4965
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3999212682247162,
+      "learning_rate": 9.608906967041757e-05,
+      "loss": 2.5409,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4286386966705322,
+      "learning_rate": 9.608114152870684e-05,
+      "loss": 2.5899,
+      "step": 4975
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4027004539966583,
+      "learning_rate": 9.607320568700866e-05,
+      "loss": 2.6604,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.40868592262268066,
+      "learning_rate": 9.606526214664911e-05,
+      "loss": 2.4927,
+      "step": 4985
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4261837899684906,
+      "learning_rate": 9.605731090895551e-05,
+      "loss": 2.4716,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42090514302253723,
+      "learning_rate": 9.604935197525649e-05,
+      "loss": 2.6167,
+      "step": 4995
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43358609080314636,
+      "learning_rate": 9.604138534688196e-05,
+      "loss": 2.5361,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4550504684448242,
+      "learning_rate": 9.60334110251631e-05,
+      "loss": 2.4966,
+      "step": 5005
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4262298047542572,
+      "learning_rate": 9.602542901143241e-05,
+      "loss": 2.545,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.802013635635376,
+      "learning_rate": 9.601743930702363e-05,
+      "loss": 2.5712,
+      "step": 5015
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4280243515968323,
+      "learning_rate": 9.600944191327184e-05,
+      "loss": 2.5474,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4164045751094818,
+      "learning_rate": 9.600143683151335e-05,
+      "loss": 2.6035,
+      "step": 5025
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4340657889842987,
+      "learning_rate": 9.59934240630858e-05,
+      "loss": 2.6125,
+      "step": 5030
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43794602155685425,
+      "learning_rate": 9.598540360932806e-05,
+      "loss": 2.5988,
+      "step": 5035
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4344055950641632,
+      "learning_rate": 9.597737547158035e-05,
+      "loss": 2.5557,
+      "step": 5040
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41289204359054565,
+      "learning_rate": 9.596933965118413e-05,
+      "loss": 2.592,
+      "step": 5045
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43079742789268494,
+      "learning_rate": 9.596129614948217e-05,
+      "loss": 2.6173,
+      "step": 5050
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4369302988052368,
+      "learning_rate": 9.595324496781849e-05,
+      "loss": 2.6261,
+      "step": 5055
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4181274473667145,
+      "learning_rate": 9.594518610753841e-05,
+      "loss": 2.5244,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4111604690551758,
+      "learning_rate": 9.593711956998855e-05,
+      "loss": 2.5089,
+      "step": 5065
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42932844161987305,
+      "learning_rate": 9.59290453565168e-05,
+      "loss": 2.5177,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4276784062385559,
+      "learning_rate": 9.592096346847232e-05,
+      "loss": 2.5896,
+      "step": 5075
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4142734408378601,
+      "learning_rate": 9.591287390720555e-05,
+      "loss": 2.5454,
+      "step": 5080
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43754541873931885,
+      "learning_rate": 9.590477667406828e-05,
+      "loss": 2.5082,
+      "step": 5085
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.432293564081192,
+      "learning_rate": 9.589667177041347e-05,
+      "loss": 2.6108,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42626193165779114,
+      "learning_rate": 9.588855919759544e-05,
+      "loss": 2.5278,
+      "step": 5095
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4210931360721588,
+      "learning_rate": 9.588043895696978e-05,
+      "loss": 2.5151,
+      "step": 5100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43070048093795776,
+      "learning_rate": 9.587231104989334e-05,
+      "loss": 2.6216,
+      "step": 5105
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4684944748878479,
+      "learning_rate": 9.586417547772426e-05,
+      "loss": 2.5659,
+      "step": 5110
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4296283721923828,
+      "learning_rate": 9.585603224182196e-05,
+      "loss": 2.5886,
+      "step": 5115
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4142526388168335,
+      "learning_rate": 9.584788134354718e-05,
+      "loss": 2.6322,
+      "step": 5120
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.45040595531463623,
+      "learning_rate": 9.583972278426186e-05,
+      "loss": 2.5103,
+      "step": 5125
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43056461215019226,
+      "learning_rate": 9.583155656532929e-05,
+      "loss": 2.611,
+      "step": 5130
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.419399619102478,
+      "learning_rate": 9.582338268811402e-05,
+      "loss": 2.5528,
+      "step": 5135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3907436430454254,
+      "learning_rate": 9.581520115398185e-05,
+      "loss": 2.5383,
+      "step": 5140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4345782697200775,
+      "learning_rate": 9.580701196429991e-05,
+      "loss": 2.5638,
+      "step": 5145
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4163147211074829,
+      "learning_rate": 9.579881512043655e-05,
+      "loss": 2.5274,
+      "step": 5150
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4311186671257019,
+      "learning_rate": 9.579061062376146e-05,
+      "loss": 2.5877,
+      "step": 5155
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42464789748191833,
+      "learning_rate": 9.578239847564559e-05,
+      "loss": 2.5388,
+      "step": 5160
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41310375928878784,
+      "learning_rate": 9.577417867746113e-05,
+      "loss": 2.5988,
+      "step": 5165
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41435274481773376,
+      "learning_rate": 9.576595123058158e-05,
+      "loss": 2.5967,
+      "step": 5170
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.413984090089798,
+      "learning_rate": 9.575771613638175e-05,
+      "loss": 2.6272,
+      "step": 5175
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.48375368118286133,
+      "learning_rate": 9.574947339623765e-05,
+      "loss": 2.5781,
+      "step": 5180
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4446849524974823,
+      "learning_rate": 9.574122301152665e-05,
+      "loss": 2.4589,
+      "step": 5185
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4144495725631714,
+      "learning_rate": 9.573296498362733e-05,
+      "loss": 2.6018,
+      "step": 5190
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41857171058654785,
+      "learning_rate": 9.572469931391959e-05,
+      "loss": 2.5456,
+      "step": 5195
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43397587537765503,
+      "learning_rate": 9.571642600378459e-05,
+      "loss": 2.5585,
+      "step": 5200
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43935874104499817,
+      "learning_rate": 9.570814505460475e-05,
+      "loss": 2.6245,
+      "step": 5205
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4225093126296997,
+      "learning_rate": 9.569985646776382e-05,
+      "loss": 2.5474,
+      "step": 5210
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4171465039253235,
+      "learning_rate": 9.569156024464677e-05,
+      "loss": 2.518,
+      "step": 5215
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4292967915534973,
+      "learning_rate": 9.568325638663989e-05,
+      "loss": 2.6068,
+      "step": 5220
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.39570608735084534,
+      "learning_rate": 9.567494489513068e-05,
+      "loss": 2.5812,
+      "step": 5225
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43639835715293884,
+      "learning_rate": 9.5666625771508e-05,
+      "loss": 2.5516,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44418323040008545,
+      "learning_rate": 9.565829901716194e-05,
+      "loss": 2.5485,
+      "step": 5235
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43961599469184875,
+      "learning_rate": 9.564996463348384e-05,
+      "loss": 2.5873,
+      "step": 5240
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4714978337287903,
+      "learning_rate": 9.564162262186638e-05,
+      "loss": 2.609,
+      "step": 5245
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4296232759952545,
+      "learning_rate": 9.563327298370347e-05,
+      "loss": 2.4804,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44850480556488037,
+      "learning_rate": 9.56249157203903e-05,
+      "loss": 2.6686,
+      "step": 5255
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44591453671455383,
+      "learning_rate": 9.561655083332333e-05,
+      "loss": 2.5126,
+      "step": 5260
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.406211256980896,
+      "learning_rate": 9.56081783239003e-05,
+      "loss": 2.542,
+      "step": 5265
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4236679971218109,
+      "learning_rate": 9.559979819352024e-05,
+      "loss": 2.6027,
+      "step": 5270
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42918193340301514,
+      "learning_rate": 9.559141044358342e-05,
+      "loss": 2.5232,
+      "step": 5275
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4205743968486786,
+      "learning_rate": 9.558301507549143e-05,
+      "loss": 2.594,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4558124542236328,
+      "learning_rate": 9.557461209064708e-05,
+      "loss": 2.5495,
+      "step": 5285
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41778597235679626,
+      "learning_rate": 9.556620149045447e-05,
+      "loss": 2.6671,
+      "step": 5290
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4571208357810974,
+      "learning_rate": 9.555778327631903e-05,
+      "loss": 2.5489,
+      "step": 5295
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4185440242290497,
+      "learning_rate": 9.554935744964734e-05,
+      "loss": 2.5423,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40534549951553345,
+      "learning_rate": 9.55409240118474e-05,
+      "loss": 2.6528,
+      "step": 5305
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.402639776468277,
+      "learning_rate": 9.553248296432835e-05,
+      "loss": 2.5484,
+      "step": 5310
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41017353534698486,
+      "learning_rate": 9.552403430850068e-05,
+      "loss": 2.5546,
+      "step": 5315
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43276363611221313,
+      "learning_rate": 9.551557804577612e-05,
+      "loss": 2.6295,
+      "step": 5320
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41769182682037354,
+      "learning_rate": 9.550711417756767e-05,
+      "loss": 2.5561,
+      "step": 5325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40523388981819153,
+      "learning_rate": 9.549864270528965e-05,
+      "loss": 2.5533,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4437076151371002,
+      "learning_rate": 9.549016363035758e-05,
+      "loss": 2.4891,
+      "step": 5335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4482382833957672,
+      "learning_rate": 9.548167695418828e-05,
+      "loss": 2.616,
+      "step": 5340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43537673354148865,
+      "learning_rate": 9.547318267819986e-05,
+      "loss": 2.6128,
+      "step": 5345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44061940908432007,
+      "learning_rate": 9.546468080381167e-05,
+      "loss": 2.495,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4103207588195801,
+      "learning_rate": 9.545617133244436e-05,
+      "loss": 2.594,
+      "step": 5355
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.417971134185791,
+      "learning_rate": 9.544765426551981e-05,
+      "loss": 2.5634,
+      "step": 5360
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.455618679523468,
+      "learning_rate": 9.543912960446117e-05,
+      "loss": 2.5698,
+      "step": 5365
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42610639333724976,
+      "learning_rate": 9.543059735069293e-05,
+      "loss": 2.5372,
+      "step": 5370
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3974537253379822,
+      "learning_rate": 9.542205750564077e-05,
+      "loss": 2.5772,
+      "step": 5375
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43186962604522705,
+      "learning_rate": 9.541351007073167e-05,
+      "loss": 2.5044,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3905530869960785,
+      "learning_rate": 9.540495504739387e-05,
+      "loss": 2.5846,
+      "step": 5385
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4255884885787964,
+      "learning_rate": 9.53963924370569e-05,
+      "loss": 2.5759,
+      "step": 5390
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40423619747161865,
+      "learning_rate": 9.538782224115151e-05,
+      "loss": 2.4566,
+      "step": 5395
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4104406237602234,
+      "learning_rate": 9.537924446110976e-05,
+      "loss": 2.5948,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40407249331474304,
+      "learning_rate": 9.537065909836497e-05,
+      "loss": 2.4443,
+      "step": 5405
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41802841424942017,
+      "learning_rate": 9.536206615435174e-05,
+      "loss": 2.5375,
+      "step": 5410
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42932572960853577,
+      "learning_rate": 9.535346563050591e-05,
+      "loss": 2.5331,
+      "step": 5415
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4369787871837616,
+      "learning_rate": 9.534485752826456e-05,
+      "loss": 2.5378,
+      "step": 5420
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4627120792865753,
+      "learning_rate": 9.53362418490661e-05,
+      "loss": 2.4881,
+      "step": 5425
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4559946060180664,
+      "learning_rate": 9.53276185943502e-05,
+      "loss": 2.5768,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40449872612953186,
+      "learning_rate": 9.531898776555773e-05,
+      "loss": 2.6286,
+      "step": 5435
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4265064001083374,
+      "learning_rate": 9.531034936413091e-05,
+      "loss": 2.4764,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3989698886871338,
+      "learning_rate": 9.530170339151314e-05,
+      "loss": 2.5481,
+      "step": 5445
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4114220142364502,
+      "learning_rate": 9.529304984914917e-05,
+      "loss": 2.492,
+      "step": 5450
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3953469395637512,
+      "learning_rate": 9.528438873848496e-05,
+      "loss": 2.625,
+      "step": 5455
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4363243579864502,
+      "learning_rate": 9.527572006096773e-05,
+      "loss": 2.4935,
+      "step": 5460
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41872188448905945,
+      "learning_rate": 9.526704381804601e-05,
+      "loss": 2.5175,
+      "step": 5465
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4172491729259491,
+      "learning_rate": 9.525836001116957e-05,
+      "loss": 2.5538,
+      "step": 5470
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42436566948890686,
+      "learning_rate": 9.524966864178943e-05,
+      "loss": 2.5735,
+      "step": 5475
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41646748781204224,
+      "learning_rate": 9.524096971135789e-05,
+      "loss": 2.5593,
+      "step": 5480
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42362192273139954,
+      "learning_rate": 9.52322632213285e-05,
+      "loss": 2.5247,
+      "step": 5485
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43437501788139343,
+      "learning_rate": 9.52235491731561e-05,
+      "loss": 2.583,
+      "step": 5490
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4016006588935852,
+      "learning_rate": 9.521482756829675e-05,
+      "loss": 2.5819,
+      "step": 5495
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42257416248321533,
+      "learning_rate": 9.520609840820782e-05,
+      "loss": 2.4648,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40284213423728943,
+      "learning_rate": 9.51973616943479e-05,
+      "loss": 2.547,
+      "step": 5505
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3993910253047943,
+      "learning_rate": 9.518861742817689e-05,
+      "loss": 2.5611,
+      "step": 5510
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40198683738708496,
+      "learning_rate": 9.51798656111559e-05,
+      "loss": 2.5085,
+      "step": 5515
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.48461267352104187,
+      "learning_rate": 9.517110624474733e-05,
+      "loss": 2.6064,
+      "step": 5520
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4318106770515442,
+      "learning_rate": 9.516233933041485e-05,
+      "loss": 2.5556,
+      "step": 5525
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4365288019180298,
+      "learning_rate": 9.515356486962337e-05,
+      "loss": 2.4389,
+      "step": 5530
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44906938076019287,
+      "learning_rate": 9.514478286383907e-05,
+      "loss": 2.5958,
+      "step": 5535
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3984452784061432,
+      "learning_rate": 9.513599331452939e-05,
+      "loss": 2.5969,
+      "step": 5540
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41292133927345276,
+      "learning_rate": 9.512719622316304e-05,
+      "loss": 2.4199,
+      "step": 5545
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.38791704177856445,
+      "learning_rate": 9.511839159120997e-05,
+      "loss": 2.5272,
+      "step": 5550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40555277466773987,
+      "learning_rate": 9.51095794201414e-05,
+      "loss": 2.4644,
+      "step": 5555
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44232016801834106,
+      "learning_rate": 9.510075971142983e-05,
+      "loss": 2.5922,
+      "step": 5560
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41886869072914124,
+      "learning_rate": 9.509193246654896e-05,
+      "loss": 2.4728,
+      "step": 5565
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44592100381851196,
+      "learning_rate": 9.508309768697384e-05,
+      "loss": 2.5868,
+      "step": 5570
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.45704740285873413,
+      "learning_rate": 9.507425537418068e-05,
+      "loss": 2.5979,
+      "step": 5575
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5030524730682373,
+      "learning_rate": 9.506540552964704e-05,
+      "loss": 2.4954,
+      "step": 5580
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4133281409740448,
+      "learning_rate": 9.505654815485168e-05,
+      "loss": 2.575,
+      "step": 5585
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4267963767051697,
+      "learning_rate": 9.504768325127464e-05,
+      "loss": 2.5366,
+      "step": 5590
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4224865734577179,
+      "learning_rate": 9.503881082039719e-05,
+      "loss": 2.6053,
+      "step": 5595
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3982297480106354,
+      "learning_rate": 9.50299308637019e-05,
+      "loss": 2.5306,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4445144236087799,
+      "learning_rate": 9.502104338267257e-05,
+      "loss": 2.566,
+      "step": 5605
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4519396126270294,
+      "learning_rate": 9.501214837879428e-05,
+      "loss": 2.5387,
+      "step": 5610
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4724132716655731,
+      "learning_rate": 9.500324585355333e-05,
+      "loss": 2.4905,
+      "step": 5615
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42017862200737,
+      "learning_rate": 9.499433580843732e-05,
+      "loss": 2.563,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4018830955028534,
+      "learning_rate": 9.498541824493506e-05,
+      "loss": 2.5532,
+      "step": 5625
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4024451971054077,
+      "learning_rate": 9.497649316453667e-05,
+      "loss": 2.5599,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39384689927101135,
+      "learning_rate": 9.496756056873348e-05,
+      "loss": 2.5233,
+      "step": 5635
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6134846210479736,
+      "learning_rate": 9.49586204590181e-05,
+      "loss": 2.5745,
+      "step": 5640
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4114353358745575,
+      "learning_rate": 9.494967283688438e-05,
+      "loss": 2.5409,
+      "step": 5645
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4241526424884796,
+      "learning_rate": 9.494071770382745e-05,
+      "loss": 2.5238,
+      "step": 5650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42969533801078796,
+      "learning_rate": 9.493175506134367e-05,
+      "loss": 2.5693,
+      "step": 5655
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4156467020511627,
+      "learning_rate": 9.492278491093067e-05,
+      "loss": 2.5525,
+      "step": 5660
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42757707834243774,
+      "learning_rate": 9.491380725408731e-05,
+      "loss": 2.6104,
+      "step": 5665
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4465058147907257,
+      "learning_rate": 9.490482209231374e-05,
+      "loss": 2.5648,
+      "step": 5670
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41913658380508423,
+      "learning_rate": 9.489582942711135e-05,
+      "loss": 2.5937,
+      "step": 5675
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4023391306400299,
+      "learning_rate": 9.488682925998277e-05,
+      "loss": 2.5246,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5977442860603333,
+      "learning_rate": 9.48778215924319e-05,
+      "loss": 2.4813,
+      "step": 5685
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41153383255004883,
+      "learning_rate": 9.48688064259639e-05,
+      "loss": 2.5116,
+      "step": 5690
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40170493721961975,
+      "learning_rate": 9.485978376208513e-05,
+      "loss": 2.5328,
+      "step": 5695
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4177417457103729,
+      "learning_rate": 9.485075360230328e-05,
+      "loss": 2.5872,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3866337835788727,
+      "learning_rate": 9.484171594812728e-05,
+      "loss": 2.5794,
+      "step": 5705
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42438727617263794,
+      "learning_rate": 9.483267080106723e-05,
+      "loss": 2.6115,
+      "step": 5710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4217809736728668,
+      "learning_rate": 9.482361816263456e-05,
+      "loss": 2.6132,
+      "step": 5715
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4019116461277008,
+      "learning_rate": 9.481455803434195e-05,
+      "loss": 2.5553,
+      "step": 5720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40675032138824463,
+      "learning_rate": 9.48054904177033e-05,
+      "loss": 2.5529,
+      "step": 5725
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40133029222488403,
+      "learning_rate": 9.479641531423377e-05,
+      "loss": 2.5836,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4816568195819855,
+      "learning_rate": 9.47873327254498e-05,
+      "loss": 2.4856,
+      "step": 5735
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4664642810821533,
+      "learning_rate": 9.477824265286904e-05,
+      "loss": 2.5506,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4399329721927643,
+      "learning_rate": 9.476914509801039e-05,
+      "loss": 2.3854,
+      "step": 5745
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.43458667397499084,
+      "learning_rate": 9.476004006239406e-05,
+      "loss": 2.4873,
+      "step": 5750
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4179019331932068,
+      "learning_rate": 9.475092754754142e-05,
+      "loss": 2.5509,
+      "step": 5755
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4346717894077301,
+      "learning_rate": 9.474180755497517e-05,
+      "loss": 2.5133,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4267875850200653,
+      "learning_rate": 9.473268008621922e-05,
+      "loss": 2.5054,
+      "step": 5765
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4317276179790497,
+      "learning_rate": 9.472354514279874e-05,
+      "loss": 2.4919,
+      "step": 5770
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4079444408416748,
+      "learning_rate": 9.471440272624012e-05,
+      "loss": 2.4883,
+      "step": 5775
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41009727120399475,
+      "learning_rate": 9.470525283807105e-05,
+      "loss": 2.5695,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.397983580827713,
+      "learning_rate": 9.469609547982044e-05,
+      "loss": 2.467,
+      "step": 5785
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41284382343292236,
+      "learning_rate": 9.468693065301843e-05,
+      "loss": 2.5015,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4381202459335327,
+      "learning_rate": 9.467775835919646e-05,
+      "loss": 2.5712,
+      "step": 5795
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4211287796497345,
+      "learning_rate": 9.466857859988715e-05,
+      "loss": 2.5747,
+      "step": 5800
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4178813099861145,
+      "learning_rate": 9.465939137662444e-05,
+      "loss": 2.5033,
+      "step": 5805
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4161866307258606,
+      "learning_rate": 9.465019669094346e-05,
+      "loss": 2.534,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.45467084646224976,
+      "learning_rate": 9.464099454438062e-05,
+      "loss": 2.5733,
+      "step": 5815
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.43376684188842773,
+      "learning_rate": 9.463178493847355e-05,
+      "loss": 2.5451,
+      "step": 5820
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.38424524664878845,
+      "learning_rate": 9.462256787476113e-05,
+      "loss": 2.577,
+      "step": 5825
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4212976098060608,
+      "learning_rate": 9.461334335478354e-05,
+      "loss": 2.4922,
+      "step": 5830
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3950728178024292,
+      "learning_rate": 9.460411138008214e-05,
+      "loss": 2.4477,
+      "step": 5835
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40212520956993103,
+      "learning_rate": 9.459487195219954e-05,
+      "loss": 2.5425,
+      "step": 5840
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4139125347137451,
+      "learning_rate": 9.458562507267963e-05,
+      "loss": 2.5773,
+      "step": 5845
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39388182759284973,
+      "learning_rate": 9.457637074306754e-05,
+      "loss": 2.5544,
+      "step": 5850
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4139162302017212,
+      "learning_rate": 9.456710896490961e-05,
+      "loss": 2.5498,
+      "step": 5855
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39500972628593445,
+      "learning_rate": 9.455783973975348e-05,
+      "loss": 2.5518,
+      "step": 5860
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41567158699035645,
+      "learning_rate": 9.454856306914797e-05,
+      "loss": 2.5741,
+      "step": 5865
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4042838215827942,
+      "learning_rate": 9.453927895464321e-05,
+      "loss": 2.4405,
+      "step": 5870
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4088144302368164,
+      "learning_rate": 9.452998739779051e-05,
+      "loss": 2.5267,
+      "step": 5875
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4243738651275635,
+      "learning_rate": 9.452068840014246e-05,
+      "loss": 2.6396,
+      "step": 5880
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.43451276421546936,
+      "learning_rate": 9.451138196325291e-05,
+      "loss": 2.5851,
+      "step": 5885
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4059518575668335,
+      "learning_rate": 9.45020680886769e-05,
+      "loss": 2.5406,
+      "step": 5890
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39090514183044434,
+      "learning_rate": 9.449274677797078e-05,
+      "loss": 2.5177,
+      "step": 5895
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4177246689796448,
+      "learning_rate": 9.448341803269208e-05,
+      "loss": 2.5224,
+      "step": 5900
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.400373250246048,
+      "learning_rate": 9.44740818543996e-05,
+      "loss": 2.4936,
+      "step": 5905
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4037063717842102,
+      "learning_rate": 9.44647382446534e-05,
+      "loss": 2.5234,
+      "step": 5910
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39929255843162537,
+      "learning_rate": 9.445538720501474e-05,
+      "loss": 2.4908,
+      "step": 5915
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4081108570098877,
+      "learning_rate": 9.444602873704616e-05,
+      "loss": 2.5233,
+      "step": 5920
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4456443190574646,
+      "learning_rate": 9.443666284231141e-05,
+      "loss": 2.4664,
+      "step": 5925
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41113734245300293,
+      "learning_rate": 9.44272895223755e-05,
+      "loss": 2.6345,
+      "step": 5930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4244571030139923,
+      "learning_rate": 9.441790877880469e-05,
+      "loss": 2.6155,
+      "step": 5935
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40785330533981323,
+      "learning_rate": 9.440852061316645e-05,
+      "loss": 2.604,
+      "step": 5940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39475783705711365,
+      "learning_rate": 9.439912502702952e-05,
+      "loss": 2.5231,
+      "step": 5945
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4153846502304077,
+      "learning_rate": 9.438972202196387e-05,
+      "loss": 2.5882,
+      "step": 5950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4280529022216797,
+      "learning_rate": 9.43803115995407e-05,
+      "loss": 2.5006,
+      "step": 5955
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4195101857185364,
+      "learning_rate": 9.437089376133246e-05,
+      "loss": 2.4476,
+      "step": 5960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41208332777023315,
+      "learning_rate": 9.436146850891283e-05,
+      "loss": 2.4982,
+      "step": 5965
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42996230721473694,
+      "learning_rate": 9.435203584385673e-05,
+      "loss": 2.5181,
+      "step": 5970
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41841280460357666,
+      "learning_rate": 9.434259576774034e-05,
+      "loss": 2.4857,
+      "step": 5975
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4292512834072113,
+      "learning_rate": 9.433314828214103e-05,
+      "loss": 2.5192,
+      "step": 5980
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3976883590221405,
+      "learning_rate": 9.43236933886375e-05,
+      "loss": 2.4644,
+      "step": 5985
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.413585901260376,
+      "learning_rate": 9.431423108880957e-05,
+      "loss": 2.5587,
+      "step": 5990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41920366883277893,
+      "learning_rate": 9.430476138423838e-05,
+      "loss": 2.5074,
+      "step": 5995
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4178551435470581,
+      "learning_rate": 9.429528427650627e-05,
+      "loss": 2.5476,
+      "step": 6000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40292230248451233,
+      "learning_rate": 9.428579976719684e-05,
+      "loss": 2.542,
+      "step": 6005
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42206066846847534,
+      "learning_rate": 9.427630785789491e-05,
+      "loss": 2.4536,
+      "step": 6010
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39787808060646057,
+      "learning_rate": 9.426680855018654e-05,
+      "loss": 2.5108,
+      "step": 6015
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4292732775211334,
+      "learning_rate": 9.425730184565907e-05,
+      "loss": 2.4867,
+      "step": 6020
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39383068680763245,
+      "learning_rate": 9.424778774590096e-05,
+      "loss": 2.5065,
+      "step": 6025
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.48230046033859253,
+      "learning_rate": 9.423826625250204e-05,
+      "loss": 2.5631,
+      "step": 6030
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4167843461036682,
+      "learning_rate": 9.422873736705329e-05,
+      "loss": 2.6068,
+      "step": 6035
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41564056277275085,
+      "learning_rate": 9.421920109114696e-05,
+      "loss": 2.5275,
+      "step": 6040
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3906739056110382,
+      "learning_rate": 9.420965742637653e-05,
+      "loss": 2.4517,
+      "step": 6045
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3962468206882477,
+      "learning_rate": 9.420010637433669e-05,
+      "loss": 2.4874,
+      "step": 6050
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39459890127182007,
+      "learning_rate": 9.419054793662342e-05,
+      "loss": 2.4959,
+      "step": 6055
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4024867117404938,
+      "learning_rate": 9.418098211483387e-05,
+      "loss": 2.5535,
+      "step": 6060
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3957017660140991,
+      "learning_rate": 9.417140891056646e-05,
+      "loss": 2.509,
+      "step": 6065
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39824315905570984,
+      "learning_rate": 9.416182832542085e-05,
+      "loss": 2.5002,
+      "step": 6070
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40759459137916565,
+      "learning_rate": 9.41522403609979e-05,
+      "loss": 2.463,
+      "step": 6075
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4107415974140167,
+      "learning_rate": 9.414264501889975e-05,
+      "loss": 2.5316,
+      "step": 6080
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4148827791213989,
+      "learning_rate": 9.41330423007297e-05,
+      "loss": 2.5347,
+      "step": 6085
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4169529676437378,
+      "learning_rate": 9.412343220809239e-05,
+      "loss": 2.4815,
+      "step": 6090
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4059647023677826,
+      "learning_rate": 9.41138147425936e-05,
+      "loss": 2.51,
+      "step": 6095
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41517674922943115,
+      "learning_rate": 9.410418990584034e-05,
+      "loss": 2.5417,
+      "step": 6100
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.38337206840515137,
+      "learning_rate": 9.409455769944093e-05,
+      "loss": 2.5227,
+      "step": 6105
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.44654256105422974,
+      "learning_rate": 9.408491812500486e-05,
+      "loss": 2.5933,
+      "step": 6110
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39182335138320923,
+      "learning_rate": 9.407527118414287e-05,
+      "loss": 2.5118,
+      "step": 6115
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4115958511829376,
+      "learning_rate": 9.406561687846693e-05,
+      "loss": 2.514,
+      "step": 6120
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4028056263923645,
+      "learning_rate": 9.405595520959025e-05,
+      "loss": 2.499,
+      "step": 6125
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4078702926635742,
+      "learning_rate": 9.404628617912723e-05,
+      "loss": 2.5611,
+      "step": 6130
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.37058547139167786,
+      "learning_rate": 9.403660978869353e-05,
+      "loss": 2.5408,
+      "step": 6135
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4222530126571655,
+      "learning_rate": 9.402692603990608e-05,
+      "loss": 2.4928,
+      "step": 6140
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4302789568901062,
+      "learning_rate": 9.401723493438295e-05,
+      "loss": 2.5524,
+      "step": 6145
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4333960711956024,
+      "learning_rate": 9.400753647374352e-05,
+      "loss": 2.4823,
+      "step": 6150
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4021005928516388,
+      "learning_rate": 9.399783065960835e-05,
+      "loss": 2.4623,
+      "step": 6155
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4088895916938782,
+      "learning_rate": 9.398811749359924e-05,
+      "loss": 2.5608,
+      "step": 6160
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4114287793636322,
+      "learning_rate": 9.397839697733924e-05,
+      "loss": 2.5372,
+      "step": 6165
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39912447333335876,
+      "learning_rate": 9.396866911245261e-05,
+      "loss": 2.4929,
+      "step": 6170
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4255404770374298,
+      "learning_rate": 9.395893390056484e-05,
+      "loss": 2.6007,
+      "step": 6175
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41881871223449707,
+      "learning_rate": 9.394919134330264e-05,
+      "loss": 2.4438,
+      "step": 6180
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41193875670433044,
+      "learning_rate": 9.393944144229396e-05,
+      "loss": 2.5344,
+      "step": 6185
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3886967599391937,
+      "learning_rate": 9.392968419916797e-05,
+      "loss": 2.4502,
+      "step": 6190
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.43345052003860474,
+      "learning_rate": 9.391991961555507e-05,
+      "loss": 2.4591,
+      "step": 6195
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.38647568225860596,
+      "learning_rate": 9.39101476930869e-05,
+      "loss": 2.4758,
+      "step": 6200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4379158020019531,
+      "learning_rate": 9.390036843339629e-05,
+      "loss": 2.4912,
+      "step": 6205
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3930405080318451,
+      "learning_rate": 9.389058183811732e-05,
+      "loss": 2.5021,
+      "step": 6210
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4169285297393799,
+      "learning_rate": 9.38807879088853e-05,
+      "loss": 2.5984,
+      "step": 6215
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40941083431243896,
+      "learning_rate": 9.387098664733676e-05,
+      "loss": 2.5662,
+      "step": 6220
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4065259099006653,
+      "learning_rate": 9.386117805510945e-05,
+      "loss": 2.4408,
+      "step": 6225
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42847809195518494,
+      "learning_rate": 9.385136213384237e-05,
+      "loss": 2.498,
+      "step": 6230
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40564921498298645,
+      "learning_rate": 9.38415388851757e-05,
+      "loss": 2.5569,
+      "step": 6235
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4025172293186188,
+      "learning_rate": 9.383170831075086e-05,
+      "loss": 2.6448,
+      "step": 6240
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41033682227134705,
+      "learning_rate": 9.382187041221054e-05,
+      "loss": 2.5675,
+      "step": 6245
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4393289387226105,
+      "learning_rate": 9.381202519119858e-05,
+      "loss": 2.5282,
+      "step": 6250
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4484148919582367,
+      "learning_rate": 9.38021726493601e-05,
+      "loss": 2.5626,
+      "step": 6255
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4018334150314331,
+      "learning_rate": 9.37923127883414e-05,
+      "loss": 2.4866,
+      "step": 6260
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4280560612678528,
+      "learning_rate": 9.378244560979007e-05,
+      "loss": 2.4265,
+      "step": 6265
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4088088870048523,
+      "learning_rate": 9.377257111535482e-05,
+      "loss": 2.4522,
+      "step": 6270
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3919166922569275,
+      "learning_rate": 9.376268930668569e-05,
+      "loss": 2.4887,
+      "step": 6275
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3895576596260071,
+      "learning_rate": 9.375280018543387e-05,
+      "loss": 2.547,
+      "step": 6280
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40554216504096985,
+      "learning_rate": 9.374290375325181e-05,
+      "loss": 2.4847,
+      "step": 6285
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40252646803855896,
+      "learning_rate": 9.373300001179316e-05,
+      "loss": 2.4715,
+      "step": 6290
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40641286969184875,
+      "learning_rate": 9.37230889627128e-05,
+      "loss": 2.5199,
+      "step": 6295
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.43575456738471985,
+      "learning_rate": 9.371317060766682e-05,
+      "loss": 2.4653,
+      "step": 6300
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.537861168384552,
+      "learning_rate": 9.370324494831254e-05,
+      "loss": 2.4904,
+      "step": 6305
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.43924617767333984,
+      "learning_rate": 9.369331198630852e-05,
+      "loss": 2.5255,
+      "step": 6310
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5511380434036255,
+      "learning_rate": 9.36833717233145e-05,
+      "loss": 2.5078,
+      "step": 6315
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4236999452114105,
+      "learning_rate": 9.367342416099147e-05,
+      "loss": 2.5718,
+      "step": 6320
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42077475786209106,
+      "learning_rate": 9.366346930100163e-05,
+      "loss": 2.4992,
+      "step": 6325
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4007509648799896,
+      "learning_rate": 9.36535071450084e-05,
+      "loss": 2.5035,
+      "step": 6330
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4051034152507782,
+      "learning_rate": 9.364353769467643e-05,
+      "loss": 2.5027,
+      "step": 6335
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41992536187171936,
+      "learning_rate": 9.363356095167156e-05,
+      "loss": 2.5054,
+      "step": 6340
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40460205078125,
+      "learning_rate": 9.362357691766088e-05,
+      "loss": 2.5792,
+      "step": 6345
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4221157431602478,
+      "learning_rate": 9.361358559431266e-05,
+      "loss": 2.459,
+      "step": 6350
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.44148164987564087,
+      "learning_rate": 9.360358698329646e-05,
+      "loss": 2.5903,
+      "step": 6355
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39422789216041565,
+      "learning_rate": 9.359358108628298e-05,
+      "loss": 2.474,
+      "step": 6360
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41729024052619934,
+      "learning_rate": 9.358356790494416e-05,
+      "loss": 2.3876,
+      "step": 6365
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41571033000946045,
+      "learning_rate": 9.357354744095317e-05,
+      "loss": 2.521,
+      "step": 6370
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39619484543800354,
+      "learning_rate": 9.35635196959844e-05,
+      "loss": 2.4225,
+      "step": 6375
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41480034589767456,
+      "learning_rate": 9.355348467171347e-05,
+      "loss": 2.5214,
+      "step": 6380
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40892118215560913,
+      "learning_rate": 9.354344236981716e-05,
+      "loss": 2.5877,
+      "step": 6385
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4124443829059601,
+      "learning_rate": 9.353339279197351e-05,
+      "loss": 2.5091,
+      "step": 6390
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4155789613723755,
+      "learning_rate": 9.352333593986176e-05,
+      "loss": 2.5137,
+      "step": 6395
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39372697472572327,
+      "learning_rate": 9.35132718151624e-05,
+      "loss": 2.5872,
+      "step": 6400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4239814281463623,
+      "learning_rate": 9.35032004195571e-05,
+      "loss": 2.4162,
+      "step": 6405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4162452518939972,
+      "learning_rate": 9.349312175472873e-05,
+      "loss": 2.4695,
+      "step": 6410
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.38392263650894165,
+      "learning_rate": 9.348303582236142e-05,
+      "loss": 2.4147,
+      "step": 6415
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42444008588790894,
+      "learning_rate": 9.347294262414047e-05,
+      "loss": 2.4569,
+      "step": 6420
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40727120637893677,
+      "learning_rate": 9.346284216175245e-05,
+      "loss": 2.4777,
+      "step": 6425
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4201492369174957,
+      "learning_rate": 9.345273443688509e-05,
+      "loss": 2.5476,
+      "step": 6430
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3959483504295349,
+      "learning_rate": 9.344261945122734e-05,
+      "loss": 2.4167,
+      "step": 6435
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4152466058731079,
+      "learning_rate": 9.343249720646939e-05,
+      "loss": 2.4635,
+      "step": 6440
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42645207047462463,
+      "learning_rate": 9.342236770430264e-05,
+      "loss": 2.4596,
+      "step": 6445
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4253899157047272,
+      "learning_rate": 9.341223094641967e-05,
+      "loss": 2.449,
+      "step": 6450
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4127592146396637,
+      "learning_rate": 9.340208693451431e-05,
+      "loss": 2.5218,
+      "step": 6455
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4086507260799408,
+      "learning_rate": 9.339193567028157e-05,
+      "loss": 2.5817,
+      "step": 6460
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.385835736989975,
+      "learning_rate": 9.338177715541772e-05,
+      "loss": 2.4178,
+      "step": 6465
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42638924717903137,
+      "learning_rate": 9.337161139162019e-05,
+      "loss": 2.468,
+      "step": 6470
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7627649307250977,
+      "learning_rate": 9.336143838058761e-05,
+      "loss": 2.5243,
+      "step": 6475
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41333797574043274,
+      "learning_rate": 9.335125812401993e-05,
+      "loss": 2.481,
+      "step": 6480
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42390263080596924,
+      "learning_rate": 9.334107062361816e-05,
+      "loss": 2.5773,
+      "step": 6485
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42286890745162964,
+      "learning_rate": 9.333087588108462e-05,
+      "loss": 2.5103,
+      "step": 6490
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42975831031799316,
+      "learning_rate": 9.332067389812282e-05,
+      "loss": 2.5229,
+      "step": 6495
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4115741550922394,
+      "learning_rate": 9.331046467643746e-05,
+      "loss": 2.496,
+      "step": 6500
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41413912177085876,
+      "learning_rate": 9.330024821773448e-05,
+      "loss": 2.494,
+      "step": 6505
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4010534882545471,
+      "learning_rate": 9.329002452372102e-05,
+      "loss": 2.5275,
+      "step": 6510
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4117087721824646,
+      "learning_rate": 9.327979359610537e-05,
+      "loss": 2.4417,
+      "step": 6515
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4015790820121765,
+      "learning_rate": 9.326955543659714e-05,
+      "loss": 2.5132,
+      "step": 6520
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3718120753765106,
+      "learning_rate": 9.325931004690706e-05,
+      "loss": 2.503,
+      "step": 6525
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39360469579696655,
+      "learning_rate": 9.324905742874709e-05,
+      "loss": 2.4649,
+      "step": 6530
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4089715778827667,
+      "learning_rate": 9.323879758383042e-05,
+      "loss": 2.5323,
+      "step": 6535
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39839860796928406,
+      "learning_rate": 9.322853051387141e-05,
+      "loss": 2.4598,
+      "step": 6540
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40426933765411377,
+      "learning_rate": 9.321825622058568e-05,
+      "loss": 2.5219,
+      "step": 6545
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3851563334465027,
+      "learning_rate": 9.320797470569001e-05,
+      "loss": 2.4376,
+      "step": 6550
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4088975489139557,
+      "learning_rate": 9.319768597090242e-05,
+      "loss": 2.4521,
+      "step": 6555
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42461198568344116,
+      "learning_rate": 9.318739001794208e-05,
+      "loss": 2.5222,
+      "step": 6560
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4081784784793854,
+      "learning_rate": 9.317708684852942e-05,
+      "loss": 2.4052,
+      "step": 6565
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39663317799568176,
+      "learning_rate": 9.31667764643861e-05,
+      "loss": 2.4898,
+      "step": 6570
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41101670265197754,
+      "learning_rate": 9.31564588672349e-05,
+      "loss": 2.565,
+      "step": 6575
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41281595826148987,
+      "learning_rate": 9.314613405879986e-05,
+      "loss": 2.4857,
+      "step": 6580
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4000832736492157,
+      "learning_rate": 9.313580204080623e-05,
+      "loss": 2.4475,
+      "step": 6585
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4148818552494049,
+      "learning_rate": 9.312546281498047e-05,
+      "loss": 2.4633,
+      "step": 6590
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4158025085926056,
+      "learning_rate": 9.311511638305018e-05,
+      "loss": 2.4929,
+      "step": 6595
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41840317845344543,
+      "learning_rate": 9.310476274674425e-05,
+      "loss": 2.4317,
+      "step": 6600
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.455756813287735,
+      "learning_rate": 9.30944019077927e-05,
+      "loss": 2.4391,
+      "step": 6605
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4126374125480652,
+      "learning_rate": 9.30840338679268e-05,
+      "loss": 2.4472,
+      "step": 6610
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4023590683937073,
+      "learning_rate": 9.307365862887905e-05,
+      "loss": 2.5058,
+      "step": 6615
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41333967447280884,
+      "learning_rate": 9.306327619238306e-05,
+      "loss": 2.5446,
+      "step": 6620
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40408971905708313,
+      "learning_rate": 9.305288656017372e-05,
+      "loss": 2.5163,
+      "step": 6625
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4060266613960266,
+      "learning_rate": 9.304248973398711e-05,
+      "loss": 2.5263,
+      "step": 6630
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39323556423187256,
+      "learning_rate": 9.303208571556048e-05,
+      "loss": 2.5271,
+      "step": 6635
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4106907546520233,
+      "learning_rate": 9.30216745066323e-05,
+      "loss": 2.5516,
+      "step": 6640
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39245516061782837,
+      "learning_rate": 9.301125610894226e-05,
+      "loss": 2.4968,
+      "step": 6645
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41882941126823425,
+      "learning_rate": 9.300083052423124e-05,
+      "loss": 2.525,
+      "step": 6650
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4115983843803406,
+      "learning_rate": 9.29903977542413e-05,
+      "loss": 2.4387,
+      "step": 6655
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5863580107688904,
+      "learning_rate": 9.297995780071571e-05,
+      "loss": 2.4415,
+      "step": 6660
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3929155468940735,
+      "learning_rate": 9.296951066539899e-05,
+      "loss": 2.5542,
+      "step": 6665
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.38427308201789856,
+      "learning_rate": 9.295905635003677e-05,
+      "loss": 2.4886,
+      "step": 6670
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3891828656196594,
+      "learning_rate": 9.294859485637595e-05,
+      "loss": 2.5036,
+      "step": 6675
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42718061804771423,
+      "learning_rate": 9.29381261861646e-05,
+      "loss": 2.5096,
+      "step": 6680
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4247053563594818,
+      "learning_rate": 9.292765034115198e-05,
+      "loss": 2.5098,
+      "step": 6685
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39936766028404236,
+      "learning_rate": 9.291716732308862e-05,
+      "loss": 2.4672,
+      "step": 6690
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4083389937877655,
+      "learning_rate": 9.290667713372613e-05,
+      "loss": 2.535,
+      "step": 6695
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4084083139896393,
+      "learning_rate": 9.289617977481741e-05,
+      "loss": 2.4817,
+      "step": 6700
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3895389437675476,
+      "learning_rate": 9.288567524811654e-05,
+      "loss": 2.5241,
+      "step": 6705
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4299841523170471,
+      "learning_rate": 9.287516355537876e-05,
+      "loss": 2.5264,
+      "step": 6710
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4136175513267517,
+      "learning_rate": 9.286464469836056e-05,
+      "loss": 2.4677,
+      "step": 6715
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4231582581996918,
+      "learning_rate": 9.28541186788196e-05,
+      "loss": 2.561,
+      "step": 6720
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3930548429489136,
+      "learning_rate": 9.284358549851471e-05,
+      "loss": 2.4701,
+      "step": 6725
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.43152880668640137,
+      "learning_rate": 9.283304515920598e-05,
+      "loss": 2.489,
+      "step": 6730
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41661301255226135,
+      "learning_rate": 9.282249766265464e-05,
+      "loss": 2.5104,
+      "step": 6735
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3891202509403229,
+      "learning_rate": 9.281194301062315e-05,
+      "loss": 2.4418,
+      "step": 6740
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.42730244994163513,
+      "learning_rate": 9.280138120487514e-05,
+      "loss": 2.4082,
+      "step": 6745
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41398555040359497,
+      "learning_rate": 9.279081224717547e-05,
+      "loss": 2.5146,
+      "step": 6750
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38694801926612854,
+      "learning_rate": 9.278023613929014e-05,
+      "loss": 2.5598,
+      "step": 6755
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39369094371795654,
+      "learning_rate": 9.276965288298642e-05,
+      "loss": 2.4753,
+      "step": 6760
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41138920187950134,
+      "learning_rate": 9.27590624800327e-05,
+      "loss": 2.4213,
+      "step": 6765
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3744901120662689,
+      "learning_rate": 9.27484649321986e-05,
+      "loss": 2.5475,
+      "step": 6770
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38795945048332214,
+      "learning_rate": 9.273786024125496e-05,
+      "loss": 2.4909,
+      "step": 6775
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.42375415563583374,
+      "learning_rate": 9.272724840897375e-05,
+      "loss": 2.4714,
+      "step": 6780
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4080946743488312,
+      "learning_rate": 9.271662943712818e-05,
+      "loss": 2.4646,
+      "step": 6785
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3873465061187744,
+      "learning_rate": 9.270600332749264e-05,
+      "loss": 2.5567,
+      "step": 6790
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4572586417198181,
+      "learning_rate": 9.269537008184273e-05,
+      "loss": 2.3747,
+      "step": 6795
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41849571466445923,
+      "learning_rate": 9.268472970195519e-05,
+      "loss": 2.5106,
+      "step": 6800
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41194024682044983,
+      "learning_rate": 9.267408218960803e-05,
+      "loss": 2.5059,
+      "step": 6805
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2143361568450928,
+      "learning_rate": 9.266342754658036e-05,
+      "loss": 2.5071,
+      "step": 6810
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38484567403793335,
+      "learning_rate": 9.265276577465259e-05,
+      "loss": 2.4169,
+      "step": 6815
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4275811016559601,
+      "learning_rate": 9.264209687560621e-05,
+      "loss": 2.4105,
+      "step": 6820
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4666740596294403,
+      "learning_rate": 9.263142085122399e-05,
+      "loss": 2.5108,
+      "step": 6825
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4433337152004242,
+      "learning_rate": 9.262073770328983e-05,
+      "loss": 2.5213,
+      "step": 6830
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40659528970718384,
+      "learning_rate": 9.261004743358886e-05,
+      "loss": 2.5441,
+      "step": 6835
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41464483737945557,
+      "learning_rate": 9.259935004390737e-05,
+      "loss": 2.5513,
+      "step": 6840
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41493332386016846,
+      "learning_rate": 9.258864553603286e-05,
+      "loss": 2.4474,
+      "step": 6845
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3970091640949249,
+      "learning_rate": 9.257793391175402e-05,
+      "loss": 2.4878,
+      "step": 6850
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.054415464401245,
+      "learning_rate": 9.256721517286073e-05,
+      "loss": 2.4728,
+      "step": 6855
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3904416561126709,
+      "learning_rate": 9.255648932114403e-05,
+      "loss": 2.5059,
+      "step": 6860
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.402787446975708,
+      "learning_rate": 9.254575635839618e-05,
+      "loss": 2.4252,
+      "step": 6865
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40585678815841675,
+      "learning_rate": 9.253501628641061e-05,
+      "loss": 2.4996,
+      "step": 6870
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4090687036514282,
+      "learning_rate": 9.252426910698198e-05,
+      "loss": 2.5401,
+      "step": 6875
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39720550179481506,
+      "learning_rate": 9.251351482190604e-05,
+      "loss": 2.446,
+      "step": 6880
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4272613227367401,
+      "learning_rate": 9.250275343297985e-05,
+      "loss": 2.5987,
+      "step": 6885
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39912158250808716,
+      "learning_rate": 9.249198494200158e-05,
+      "loss": 2.4138,
+      "step": 6890
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39067527651786804,
+      "learning_rate": 9.24812093507706e-05,
+      "loss": 2.4561,
+      "step": 6895
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40372803807258606,
+      "learning_rate": 9.247042666108747e-05,
+      "loss": 2.3964,
+      "step": 6900
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38236290216445923,
+      "learning_rate": 9.245963687475395e-05,
+      "loss": 2.4697,
+      "step": 6905
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4087802767753601,
+      "learning_rate": 9.244883999357293e-05,
+      "loss": 2.4561,
+      "step": 6910
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3982706069946289,
+      "learning_rate": 9.24380360193486e-05,
+      "loss": 2.4659,
+      "step": 6915
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40191465616226196,
+      "learning_rate": 9.242722495388621e-05,
+      "loss": 2.4952,
+      "step": 6920
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4031904339790344,
+      "learning_rate": 9.241640679899226e-05,
+      "loss": 2.4736,
+      "step": 6925
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4011722803115845,
+      "learning_rate": 9.240558155647443e-05,
+      "loss": 2.5239,
+      "step": 6930
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.37850308418273926,
+      "learning_rate": 9.239474922814157e-05,
+      "loss": 2.4478,
+      "step": 6935
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38924354314804077,
+      "learning_rate": 9.238390981580375e-05,
+      "loss": 2.4364,
+      "step": 6940
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.431418776512146,
+      "learning_rate": 9.237306332127215e-05,
+      "loss": 2.5644,
+      "step": 6945
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4281787872314453,
+      "learning_rate": 9.236220974635923e-05,
+      "loss": 2.4618,
+      "step": 6950
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3839815557003021,
+      "learning_rate": 9.235134909287853e-05,
+      "loss": 2.4899,
+      "step": 6955
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4136161506175995,
+      "learning_rate": 9.234048136264485e-05,
+      "loss": 2.4401,
+      "step": 6960
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40508943796157837,
+      "learning_rate": 9.232960655747416e-05,
+      "loss": 2.5467,
+      "step": 6965
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3965204358100891,
+      "learning_rate": 9.231872467918358e-05,
+      "loss": 2.4957,
+      "step": 6970
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5859538912773132,
+      "learning_rate": 9.230783572959143e-05,
+      "loss": 2.5033,
+      "step": 6975
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38892123103141785,
+      "learning_rate": 9.229693971051723e-05,
+      "loss": 2.5409,
+      "step": 6980
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.43363049626350403,
+      "learning_rate": 9.228603662378166e-05,
+      "loss": 2.4627,
+      "step": 6985
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3987354636192322,
+      "learning_rate": 9.227512647120656e-05,
+      "loss": 2.4305,
+      "step": 6990
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39116621017456055,
+      "learning_rate": 9.2264209254615e-05,
+      "loss": 2.5124,
+      "step": 6995
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3910858929157257,
+      "learning_rate": 9.225328497583122e-05,
+      "loss": 2.4893,
+      "step": 7000
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39687788486480713,
+      "learning_rate": 9.224235363668058e-05,
+      "loss": 2.454,
+      "step": 7005
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4088541567325592,
+      "learning_rate": 9.223141523898972e-05,
+      "loss": 2.5393,
+      "step": 7010
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4005780816078186,
+      "learning_rate": 9.222046978458639e-05,
+      "loss": 2.4435,
+      "step": 7015
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4463717043399811,
+      "learning_rate": 9.22095172752995e-05,
+      "loss": 2.492,
+      "step": 7020
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4956721067428589,
+      "learning_rate": 9.21985577129592e-05,
+      "loss": 2.5178,
+      "step": 7025
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3967847228050232,
+      "learning_rate": 9.218759109939681e-05,
+      "loss": 2.4929,
+      "step": 7030
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3810076117515564,
+      "learning_rate": 9.217661743644477e-05,
+      "loss": 2.4482,
+      "step": 7035
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3981812000274658,
+      "learning_rate": 9.216563672593677e-05,
+      "loss": 2.4768,
+      "step": 7040
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.422014057636261,
+      "learning_rate": 9.215464896970763e-05,
+      "loss": 2.5484,
+      "step": 7045
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3719334006309509,
+      "learning_rate": 9.214365416959336e-05,
+      "loss": 2.4238,
+      "step": 7050
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40755343437194824,
+      "learning_rate": 9.213265232743119e-05,
+      "loss": 2.4359,
+      "step": 7055
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38690686225891113,
+      "learning_rate": 9.212164344505942e-05,
+      "loss": 2.4695,
+      "step": 7060
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4379394054412842,
+      "learning_rate": 9.211062752431765e-05,
+      "loss": 2.4767,
+      "step": 7065
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38525328040122986,
+      "learning_rate": 9.209960456704656e-05,
+      "loss": 2.4227,
+      "step": 7070
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39550161361694336,
+      "learning_rate": 9.208857457508809e-05,
+      "loss": 2.4818,
+      "step": 7075
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41083836555480957,
+      "learning_rate": 9.207753755028526e-05,
+      "loss": 2.5052,
+      "step": 7080
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39531680941581726,
+      "learning_rate": 9.206649349448235e-05,
+      "loss": 2.5201,
+      "step": 7085
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4004659354686737,
+      "learning_rate": 9.205544240952477e-05,
+      "loss": 2.4212,
+      "step": 7090
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39139705896377563,
+      "learning_rate": 9.204438429725911e-05,
+      "loss": 2.5084,
+      "step": 7095
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3937572240829468,
+      "learning_rate": 9.203331915953314e-05,
+      "loss": 2.5023,
+      "step": 7100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5186988115310669,
+      "learning_rate": 9.202224699819584e-05,
+      "loss": 2.4963,
+      "step": 7105
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39096567034721375,
+      "learning_rate": 9.201116781509726e-05,
+      "loss": 2.5058,
+      "step": 7110
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4037856161594391,
+      "learning_rate": 9.200008161208875e-05,
+      "loss": 2.4755,
+      "step": 7115
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.430159330368042,
+      "learning_rate": 9.198898839102273e-05,
+      "loss": 2.5321,
+      "step": 7120
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38770103454589844,
+      "learning_rate": 9.197788815375287e-05,
+      "loss": 2.4797,
+      "step": 7125
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.386308878660202,
+      "learning_rate": 9.196678090213395e-05,
+      "loss": 2.4793,
+      "step": 7130
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3872789144515991,
+      "learning_rate": 9.195566663802198e-05,
+      "loss": 2.5062,
+      "step": 7135
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3953617811203003,
+      "learning_rate": 9.194454536327408e-05,
+      "loss": 2.5054,
+      "step": 7140
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4013359546661377,
+      "learning_rate": 9.19334170797486e-05,
+      "loss": 2.4864,
+      "step": 7145
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.42873141169548035,
+      "learning_rate": 9.192228178930503e-05,
+      "loss": 2.4277,
+      "step": 7150
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4245062470436096,
+      "learning_rate": 9.191113949380402e-05,
+      "loss": 2.4335,
+      "step": 7155
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.41321060061454773,
+      "learning_rate": 9.189999019510742e-05,
+      "loss": 2.5069,
+      "step": 7160
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39861875772476196,
+      "learning_rate": 9.188883389507823e-05,
+      "loss": 2.551,
+      "step": 7165
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39766925573349,
+      "learning_rate": 9.187767059558064e-05,
+      "loss": 2.4782,
+      "step": 7170
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3744603395462036,
+      "learning_rate": 9.186650029847998e-05,
+      "loss": 2.385,
+      "step": 7175
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3987163305282593,
+      "learning_rate": 9.185532300564276e-05,
+      "loss": 2.4393,
+      "step": 7180
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.459414541721344,
+      "learning_rate": 9.184413871893668e-05,
+      "loss": 2.5111,
+      "step": 7185
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39684823155403137,
+      "learning_rate": 9.183294744023058e-05,
+      "loss": 2.4736,
+      "step": 7190
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39512890577316284,
+      "learning_rate": 9.18217491713945e-05,
+      "loss": 2.5046,
+      "step": 7195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3997693955898285,
+      "learning_rate": 9.181054391429962e-05,
+      "loss": 2.4435,
+      "step": 7200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4011540710926056,
+      "learning_rate": 9.179933167081828e-05,
+      "loss": 2.5126,
+      "step": 7205
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.40439411997795105,
+      "learning_rate": 9.178811244282404e-05,
+      "loss": 2.43,
+      "step": 7210
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3938140869140625,
+      "learning_rate": 9.177688623219154e-05,
+      "loss": 2.4456,
+      "step": 7215
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.41280046105384827,
+      "learning_rate": 9.176565304079669e-05,
+      "loss": 2.5208,
+      "step": 7220
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.412489652633667,
+      "learning_rate": 9.175441287051648e-05,
+      "loss": 2.4181,
+      "step": 7225
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39930757880210876,
+      "learning_rate": 9.174316572322912e-05,
+      "loss": 2.4546,
+      "step": 7230
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3994155526161194,
+      "learning_rate": 9.173191160081395e-05,
+      "loss": 2.5252,
+      "step": 7235
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3978583514690399,
+      "learning_rate": 9.17206505051515e-05,
+      "loss": 2.5021,
+      "step": 7240
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3899782598018646,
+      "learning_rate": 9.170938243812349e-05,
+      "loss": 2.4929,
+      "step": 7245
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38845595717430115,
+      "learning_rate": 9.16981074016127e-05,
+      "loss": 2.4888,
+      "step": 7250
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4068017899990082,
+      "learning_rate": 9.16868253975032e-05,
+      "loss": 2.5208,
+      "step": 7255
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3715313971042633,
+      "learning_rate": 9.167553642768017e-05,
+      "loss": 2.4567,
+      "step": 7260
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39034759998321533,
+      "learning_rate": 9.166424049402993e-05,
+      "loss": 2.4485,
+      "step": 7265
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3878937363624573,
+      "learning_rate": 9.165293759844e-05,
+      "loss": 2.4706,
+      "step": 7270
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38909775018692017,
+      "learning_rate": 9.164162774279905e-05,
+      "loss": 2.5127,
+      "step": 7275
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3907211422920227,
+      "learning_rate": 9.163031092899693e-05,
+      "loss": 2.5612,
+      "step": 7280
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.40893805027008057,
+      "learning_rate": 9.161898715892463e-05,
+      "loss": 2.4466,
+      "step": 7285
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3992621600627899,
+      "learning_rate": 9.160765643447429e-05,
+      "loss": 2.4618,
+      "step": 7290
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39729106426239014,
+      "learning_rate": 9.159631875753925e-05,
+      "loss": 2.5035,
+      "step": 7295
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3885912001132965,
+      "learning_rate": 9.1584974130014e-05,
+      "loss": 2.4733,
+      "step": 7300
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39301106333732605,
+      "learning_rate": 9.157362255379419e-05,
+      "loss": 2.4502,
+      "step": 7305
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39136141538619995,
+      "learning_rate": 9.156226403077657e-05,
+      "loss": 2.5193,
+      "step": 7310
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38920316100120544,
+      "learning_rate": 9.155089856285917e-05,
+      "loss": 2.4854,
+      "step": 7315
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.009910583496094,
+      "learning_rate": 9.153952615194111e-05,
+      "loss": 2.4274,
+      "step": 7320
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3851626217365265,
+      "learning_rate": 9.152814679992265e-05,
+      "loss": 2.4179,
+      "step": 7325
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4013639986515045,
+      "learning_rate": 9.151676050870526e-05,
+      "loss": 2.4732,
+      "step": 7330
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4013362526893616,
+      "learning_rate": 9.150536728019154e-05,
+      "loss": 2.4821,
+      "step": 7335
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39503225684165955,
+      "learning_rate": 9.149396711628523e-05,
+      "loss": 2.4625,
+      "step": 7340
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.406381756067276,
+      "learning_rate": 9.14825600188913e-05,
+      "loss": 2.4566,
+      "step": 7345
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3946424126625061,
+      "learning_rate": 9.147114598991583e-05,
+      "loss": 2.4977,
+      "step": 7350
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39989611506462097,
+      "learning_rate": 9.145972503126603e-05,
+      "loss": 2.4461,
+      "step": 7355
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4167274236679077,
+      "learning_rate": 9.14482971448503e-05,
+      "loss": 2.4232,
+      "step": 7360
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4044479727745056,
+      "learning_rate": 9.143686233257824e-05,
+      "loss": 2.3864,
+      "step": 7365
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3942292332649231,
+      "learning_rate": 9.142542059636053e-05,
+      "loss": 2.5451,
+      "step": 7370
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4316862225532532,
+      "learning_rate": 9.141397193810903e-05,
+      "loss": 2.4552,
+      "step": 7375
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.40895283222198486,
+      "learning_rate": 9.14025163597368e-05,
+      "loss": 2.4712,
+      "step": 7380
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39713090658187866,
+      "learning_rate": 9.1391053863158e-05,
+      "loss": 2.4608,
+      "step": 7385
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.36548468470573425,
+      "learning_rate": 9.137958445028799e-05,
+      "loss": 2.4047,
+      "step": 7390
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.42381951212882996,
+      "learning_rate": 9.136810812304326e-05,
+      "loss": 2.4641,
+      "step": 7395
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38672852516174316,
+      "learning_rate": 9.135662488334147e-05,
+      "loss": 2.547,
+      "step": 7400
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4095192551612854,
+      "learning_rate": 9.13451347331014e-05,
+      "loss": 2.4975,
+      "step": 7405
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39217135310173035,
+      "learning_rate": 9.133363767424303e-05,
+      "loss": 2.4644,
+      "step": 7410
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39199957251548767,
+      "learning_rate": 9.132213370868748e-05,
+      "loss": 2.4391,
+      "step": 7415
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38853123784065247,
+      "learning_rate": 9.131062283835701e-05,
+      "loss": 2.4605,
+      "step": 7420
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38700804114341736,
+      "learning_rate": 9.129910506517503e-05,
+      "loss": 2.4024,
+      "step": 7425
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4261733591556549,
+      "learning_rate": 9.128758039106616e-05,
+      "loss": 2.4477,
+      "step": 7430
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38944190740585327,
+      "learning_rate": 9.127604881795609e-05,
+      "loss": 2.5221,
+      "step": 7435
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3824175000190735,
+      "learning_rate": 9.126451034777173e-05,
+      "loss": 2.5103,
+      "step": 7440
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3870241641998291,
+      "learning_rate": 9.12529649824411e-05,
+      "loss": 2.4718,
+      "step": 7445
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.37837105989456177,
+      "learning_rate": 9.124141272389337e-05,
+      "loss": 2.4636,
+      "step": 7450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4074797034263611,
+      "learning_rate": 9.122985357405892e-05,
+      "loss": 2.4479,
+      "step": 7455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3924579620361328,
+      "learning_rate": 9.121828753486921e-05,
+      "loss": 2.5197,
+      "step": 7460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.401430606842041,
+      "learning_rate": 9.120671460825691e-05,
+      "loss": 2.4797,
+      "step": 7465
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.43173182010650635,
+      "learning_rate": 9.119513479615577e-05,
+      "loss": 2.4647,
+      "step": 7470
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.426880419254303,
+      "learning_rate": 9.118354810050079e-05,
+      "loss": 2.4661,
+      "step": 7475
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3893256187438965,
+      "learning_rate": 9.117195452322801e-05,
+      "loss": 2.5352,
+      "step": 7480
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3918955624103546,
+      "learning_rate": 9.11603540662747e-05,
+      "loss": 2.4069,
+      "step": 7485
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4139378070831299,
+      "learning_rate": 9.114874673157927e-05,
+      "loss": 2.4466,
+      "step": 7490
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39826130867004395,
+      "learning_rate": 9.113713252108123e-05,
+      "loss": 2.3868,
+      "step": 7495
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.390816330909729,
+      "learning_rate": 9.112551143672127e-05,
+      "loss": 2.4641,
+      "step": 7500
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3829440474510193,
+      "learning_rate": 9.111388348044125e-05,
+      "loss": 2.4705,
+      "step": 7505
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4160729944705963,
+      "learning_rate": 9.110224865418417e-05,
+      "loss": 2.4675,
+      "step": 7510
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38804352283477783,
+      "learning_rate": 9.109060695989413e-05,
+      "loss": 2.5088,
+      "step": 7515
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38015854358673096,
+      "learning_rate": 9.107895839951643e-05,
+      "loss": 2.4145,
+      "step": 7520
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40144968032836914,
+      "learning_rate": 9.106730297499752e-05,
+      "loss": 2.464,
+      "step": 7525
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3918294906616211,
+      "learning_rate": 9.105564068828497e-05,
+      "loss": 2.5393,
+      "step": 7530
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40522143244743347,
+      "learning_rate": 9.104397154132747e-05,
+      "loss": 2.4013,
+      "step": 7535
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.46525633335113525,
+      "learning_rate": 9.103229553607493e-05,
+      "loss": 2.4965,
+      "step": 7540
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3851777911186218,
+      "learning_rate": 9.102061267447835e-05,
+      "loss": 2.4834,
+      "step": 7545
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3620539903640747,
+      "learning_rate": 9.100892295848989e-05,
+      "loss": 2.4963,
+      "step": 7550
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3865240216255188,
+      "learning_rate": 9.099722639006287e-05,
+      "loss": 2.4644,
+      "step": 7555
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4052528440952301,
+      "learning_rate": 9.098552297115175e-05,
+      "loss": 2.4452,
+      "step": 7560
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.37279537320137024,
+      "learning_rate": 9.097381270371212e-05,
+      "loss": 2.4679,
+      "step": 7565
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4291245639324188,
+      "learning_rate": 9.09620955897007e-05,
+      "loss": 2.3967,
+      "step": 7570
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4468008577823639,
+      "learning_rate": 9.09503716310754e-05,
+      "loss": 2.4279,
+      "step": 7575
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.45383358001709,
+      "learning_rate": 9.093864082979525e-05,
+      "loss": 2.5168,
+      "step": 7580
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.42808884382247925,
+      "learning_rate": 9.092690318782042e-05,
+      "loss": 2.4688,
+      "step": 7585
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39762333035469055,
+      "learning_rate": 9.091515870711221e-05,
+      "loss": 2.4416,
+      "step": 7590
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40360212326049805,
+      "learning_rate": 9.09034073896331e-05,
+      "loss": 2.4525,
+      "step": 7595
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4289168119430542,
+      "learning_rate": 9.08916492373467e-05,
+      "loss": 2.4929,
+      "step": 7600
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4005039930343628,
+      "learning_rate": 9.087988425221772e-05,
+      "loss": 2.4567,
+      "step": 7605
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38608285784721375,
+      "learning_rate": 9.086811243621206e-05,
+      "loss": 2.4835,
+      "step": 7610
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39694729447364807,
+      "learning_rate": 9.085633379129677e-05,
+      "loss": 2.571,
+      "step": 7615
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3838146924972534,
+      "learning_rate": 9.084454831943998e-05,
+      "loss": 2.4622,
+      "step": 7620
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39651983976364136,
+      "learning_rate": 9.083275602261101e-05,
+      "loss": 2.5446,
+      "step": 7625
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3944641053676605,
+      "learning_rate": 9.082095690278034e-05,
+      "loss": 2.4699,
+      "step": 7630
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3849901258945465,
+      "learning_rate": 9.080915096191951e-05,
+      "loss": 2.4743,
+      "step": 7635
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38443028926849365,
+      "learning_rate": 9.07973382020013e-05,
+      "loss": 2.3992,
+      "step": 7640
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38106220960617065,
+      "learning_rate": 9.078551862499953e-05,
+      "loss": 2.4435,
+      "step": 7645
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3706494867801666,
+      "learning_rate": 9.077369223288923e-05,
+      "loss": 2.4009,
+      "step": 7650
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3785257041454315,
+      "learning_rate": 9.076185902764656e-05,
+      "loss": 2.3583,
+      "step": 7655
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.41512757539749146,
+      "learning_rate": 9.075001901124878e-05,
+      "loss": 2.4419,
+      "step": 7660
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.37154436111450195,
+      "learning_rate": 9.073817218567431e-05,
+      "loss": 2.4839,
+      "step": 7665
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38510167598724365,
+      "learning_rate": 9.072631855290275e-05,
+      "loss": 2.359,
+      "step": 7670
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.42281782627105713,
+      "learning_rate": 9.071445811491475e-05,
+      "loss": 2.4507,
+      "step": 7675
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39192917943000793,
+      "learning_rate": 9.070259087369218e-05,
+      "loss": 2.4816,
+      "step": 7680
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3987870514392853,
+      "learning_rate": 9.069071683121799e-05,
+      "loss": 2.5078,
+      "step": 7685
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39897626638412476,
+      "learning_rate": 9.067883598947628e-05,
+      "loss": 2.5635,
+      "step": 7690
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39893895387649536,
+      "learning_rate": 9.066694835045233e-05,
+      "loss": 2.4309,
+      "step": 7695
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.36982956528663635,
+      "learning_rate": 9.06550539161325e-05,
+      "loss": 2.4978,
+      "step": 7700
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40385282039642334,
+      "learning_rate": 9.06431526885043e-05,
+      "loss": 2.5271,
+      "step": 7705
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3830477297306061,
+      "learning_rate": 9.063124466955639e-05,
+      "loss": 2.5021,
+      "step": 7710
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40946266055107117,
+      "learning_rate": 9.061932986127857e-05,
+      "loss": 2.3969,
+      "step": 7715
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39376917481422424,
+      "learning_rate": 9.060740826566172e-05,
+      "loss": 2.5365,
+      "step": 7720
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39925187826156616,
+      "learning_rate": 9.059547988469792e-05,
+      "loss": 2.4575,
+      "step": 7725
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.41565218567848206,
+      "learning_rate": 9.058354472038037e-05,
+      "loss": 2.537,
+      "step": 7730
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38379961252212524,
+      "learning_rate": 9.057160277470337e-05,
+      "loss": 2.4638,
+      "step": 7735
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.41947245597839355,
+      "learning_rate": 9.055965404966238e-05,
+      "loss": 2.4333,
+      "step": 7740
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.416264146566391,
+      "learning_rate": 9.054769854725399e-05,
+      "loss": 2.3659,
+      "step": 7745
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4641427993774414,
+      "learning_rate": 9.053573626947593e-05,
+      "loss": 2.4432,
+      "step": 7750
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3979499936103821,
+      "learning_rate": 9.052376721832702e-05,
+      "loss": 2.5049,
+      "step": 7755
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39330554008483887,
+      "learning_rate": 9.051179139580727e-05,
+      "loss": 2.4637,
+      "step": 7760
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3703835904598236,
+      "learning_rate": 9.049980880391782e-05,
+      "loss": 2.4912,
+      "step": 7765
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3897750675678253,
+      "learning_rate": 9.048781944466087e-05,
+      "loss": 2.5148,
+      "step": 7770
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3857094347476959,
+      "learning_rate": 9.04758233200398e-05,
+      "loss": 2.344,
+      "step": 7775
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4221190810203552,
+      "learning_rate": 9.046382043205914e-05,
+      "loss": 2.2714,
+      "step": 7780
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.43148499727249146,
+      "learning_rate": 9.045181078272453e-05,
+      "loss": 2.4572,
+      "step": 7785
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40261441469192505,
+      "learning_rate": 9.043979437404273e-05,
+      "loss": 2.3949,
+      "step": 7790
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39023756980895996,
+      "learning_rate": 9.042777120802164e-05,
+      "loss": 2.4435,
+      "step": 7795
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.423368364572525,
+      "learning_rate": 9.041574128667027e-05,
+      "loss": 2.4314,
+      "step": 7800
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.36904025077819824,
+      "learning_rate": 9.040370461199879e-05,
+      "loss": 2.4486,
+      "step": 7805
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4193035364151001,
+      "learning_rate": 9.039166118601849e-05,
+      "loss": 2.369,
+      "step": 7810
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.42742404341697693,
+      "learning_rate": 9.037961101074175e-05,
+      "loss": 2.4735,
+      "step": 7815
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38518020510673523,
+      "learning_rate": 9.036755408818217e-05,
+      "loss": 2.4769,
+      "step": 7820
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3878040015697479,
+      "learning_rate": 9.035549042035434e-05,
+      "loss": 2.4698,
+      "step": 7825
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38890600204467773,
+      "learning_rate": 9.03434200092741e-05,
+      "loss": 2.4951,
+      "step": 7830
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.41346439719200134,
+      "learning_rate": 9.033134285695837e-05,
+      "loss": 2.4539,
+      "step": 7835
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.37482950091362,
+      "learning_rate": 9.031925896542521e-05,
+      "loss": 2.4377,
+      "step": 7840
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.39379850029945374,
+      "learning_rate": 9.030716833669375e-05,
+      "loss": 2.4986,
+      "step": 7845
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38753172755241394,
+      "learning_rate": 9.029507097278433e-05,
+      "loss": 2.4886,
+      "step": 7850
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3896917998790741,
+      "learning_rate": 9.028296687571834e-05,
+      "loss": 2.4777,
+      "step": 7855
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38309529423713684,
+      "learning_rate": 9.027085604751835e-05,
+      "loss": 2.4763,
+      "step": 7860
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4089739918708801,
+      "learning_rate": 9.025873849020805e-05,
+      "loss": 2.4766,
+      "step": 7865
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4001132547855377,
+      "learning_rate": 9.024661420581222e-05,
+      "loss": 2.5231,
+      "step": 7870
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3836793899536133,
+      "learning_rate": 9.023448319635676e-05,
+      "loss": 2.4009,
+      "step": 7875
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.41857603192329407,
+      "learning_rate": 9.022234546386877e-05,
+      "loss": 2.4914,
+      "step": 7880
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3775434195995331,
+      "learning_rate": 9.021020101037637e-05,
+      "loss": 2.4404,
+      "step": 7885
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40075168013572693,
+      "learning_rate": 9.019804983790888e-05,
+      "loss": 2.541,
+      "step": 7890
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4062296748161316,
+      "learning_rate": 9.018589194849672e-05,
+      "loss": 2.4868,
+      "step": 7895
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.39095625281333923,
+      "learning_rate": 9.01737273441714e-05,
+      "loss": 2.4983,
+      "step": 7900
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3689084053039551,
+      "learning_rate": 9.016155602696559e-05,
+      "loss": 2.521,
+      "step": 7905
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3927435874938965,
+      "learning_rate": 9.01493779989131e-05,
+      "loss": 2.4496,
+      "step": 7910
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4094651937484741,
+      "learning_rate": 9.013719326204881e-05,
+      "loss": 2.4881,
+      "step": 7915
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.39562493562698364,
+      "learning_rate": 9.012500181840874e-05,
+      "loss": 2.3926,
+      "step": 7920
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.411185085773468,
+      "learning_rate": 9.011280367003003e-05,
+      "loss": 2.4627,
+      "step": 7925
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.38388654589653015,
+      "learning_rate": 9.010059881895099e-05,
+      "loss": 2.5106,
+      "step": 7930
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.38794299960136414,
+      "learning_rate": 9.008838726721094e-05,
+      "loss": 2.4842,
+      "step": 7935
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3904341161251068,
+      "learning_rate": 9.007616901685042e-05,
+      "loss": 2.3653,
+      "step": 7940
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.38047662377357483,
+      "learning_rate": 9.006394406991108e-05,
+      "loss": 2.402,
+      "step": 7945
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.38650602102279663,
+      "learning_rate": 9.00517124284356e-05,
+      "loss": 2.5092,
+      "step": 7950
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.400652140378952,
+      "learning_rate": 9.003947409446789e-05,
+      "loss": 2.4194,
+      "step": 7955
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.38847798109054565,
+      "learning_rate": 9.002722907005292e-05,
+      "loss": 2.5307,
+      "step": 7960
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3637595772743225,
+      "learning_rate": 9.001497735723679e-05,
+      "loss": 2.5119,
+      "step": 7965
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.42928528785705566,
+      "learning_rate": 9.000271895806669e-05,
+      "loss": 2.4995,
+      "step": 7970
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.41547733545303345,
+      "learning_rate": 8.9990453874591e-05,
+      "loss": 2.4895,
+      "step": 7975
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.40556448698043823,
+      "learning_rate": 8.997818210885912e-05,
+      "loss": 2.4254,
+      "step": 7980
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.37526944279670715,
+      "learning_rate": 8.996590366292164e-05,
+      "loss": 2.4894,
+      "step": 7985
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4072200357913971,
+      "learning_rate": 8.995361853883026e-05,
+      "loss": 2.4287,
+      "step": 7990
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44405511021614075,
+      "learning_rate": 8.994132673863774e-05,
+      "loss": 2.4783,
+      "step": 7995
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3991302251815796,
+      "learning_rate": 8.992902826439802e-05,
+      "loss": 2.5086,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 38527,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.126010100010189e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}