diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11208 @@
+{
+  "best_metric": 0.9374278867489128,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-hotel_classifier_v1/checkpoint-7920",
+  "epoch": 9.99684443041969,
+  "eval_steps": 500,
+  "global_step": 7920,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.535375595092773,
+      "learning_rate": 3.1565656565656567e-07,
+      "loss": 0.3943,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.275211811065674,
+      "learning_rate": 6.313131313131313e-07,
+      "loss": 0.396,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.508435249328613,
+      "learning_rate": 9.469696969696971e-07,
+      "loss": 0.3975,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.307165145874023,
+      "learning_rate": 1.2626262626262627e-06,
+      "loss": 0.3366,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.061122894287109,
+      "learning_rate": 1.5782828282828285e-06,
+      "loss": 0.4097,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.293136596679688,
+      "learning_rate": 1.8939393939393941e-06,
+      "loss": 0.4201,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.586406230926514,
+      "learning_rate": 2.2095959595959595e-06,
+      "loss": 0.4283,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.168486595153809,
+      "learning_rate": 2.5252525252525253e-06,
+      "loss": 0.3727,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.197160720825195,
+      "learning_rate": 2.840909090909091e-06,
+      "loss": 0.3542,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.844666957855225,
+      "learning_rate": 3.156565656565657e-06,
+      "loss": 0.3935,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.720086574554443,
+      "learning_rate": 3.4722222222222224e-06,
+      "loss": 0.3884,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.447674751281738,
+      "learning_rate": 3.7878787878787882e-06,
+      "loss": 0.4302,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.148931980133057,
+      "learning_rate": 4.103535353535354e-06,
+      "loss": 0.4136,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.811440944671631,
+      "learning_rate": 4.419191919191919e-06,
+      "loss": 0.3814,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.132671356201172,
+      "learning_rate": 4.734848484848485e-06,
+      "loss": 0.3308,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.768515586853027,
+      "learning_rate": 5.050505050505051e-06,
+      "loss": 0.3848,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.829127550125122,
+      "learning_rate": 5.3661616161616165e-06,
+      "loss": 0.3663,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 20.981216430664062,
+      "learning_rate": 5.681818181818182e-06,
+      "loss": 0.3536,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.633984088897705,
+      "learning_rate": 5.997474747474747e-06,
+      "loss": 0.4212,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.5228352546691895,
+      "learning_rate": 6.313131313131314e-06,
+      "loss": 0.4307,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.808162689208984,
+      "learning_rate": 6.628787878787879e-06,
+      "loss": 0.3652,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.340692520141602,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.3759,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.8118767738342285,
+      "learning_rate": 7.26010101010101e-06,
+      "loss": 0.3847,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.361588478088379,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.4047,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.00014591217041,
+      "learning_rate": 7.891414141414141e-06,
+      "loss": 0.3308,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.148036956787109,
+      "learning_rate": 8.207070707070708e-06,
+      "loss": 0.3464,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.50632905960083,
+      "learning_rate": 8.522727272727273e-06,
+      "loss": 0.3367,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.759868621826172,
+      "learning_rate": 8.838383838383838e-06,
+      "loss": 0.3583,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.8641631603240967,
+      "learning_rate": 9.154040404040405e-06,
+      "loss": 0.3426,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.213869571685791,
+      "learning_rate": 9.46969696969697e-06,
+      "loss": 0.372,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.771566390991211,
+      "learning_rate": 9.785353535353536e-06,
+      "loss": 0.331,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.640277862548828,
+      "learning_rate": 1.0101010101010101e-05,
+      "loss": 0.3799,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.302839756011963,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.3462,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.186686038970947,
+      "learning_rate": 1.0732323232323233e-05,
+      "loss": 0.3582,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.541094779968262,
+      "learning_rate": 1.10479797979798e-05,
+      "loss": 0.3635,
+      "step": 175
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.478331565856934,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 0.4136,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.366847991943359,
+      "learning_rate": 1.167929292929293e-05,
+      "loss": 0.45,
+      "step": 185
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.444547653198242,
+      "learning_rate": 1.1994949494949495e-05,
+      "loss": 0.4138,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.427088260650635,
+      "learning_rate": 1.2310606060606061e-05,
+      "loss": 0.3632,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.618551254272461,
+      "learning_rate": 1.2626262626262628e-05,
+      "loss": 0.3806,
+      "step": 200
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.3855743408203125,
+      "learning_rate": 1.2941919191919191e-05,
+      "loss": 0.3745,
+      "step": 205
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.827747821807861,
+      "learning_rate": 1.3257575757575758e-05,
+      "loss": 0.3834,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.268397331237793,
+      "learning_rate": 1.3573232323232325e-05,
+      "loss": 0.3853,
+      "step": 215
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.605027198791504,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.3575,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.3819899559021,
+      "learning_rate": 1.4204545454545456e-05,
+      "loss": 0.3496,
+      "step": 225
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.890773773193359,
+      "learning_rate": 1.452020202020202e-05,
+      "loss": 0.3919,
+      "step": 230
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.219390869140625,
+      "learning_rate": 1.4835858585858586e-05,
+      "loss": 0.4109,
+      "step": 235
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.1185407638549805,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.4155,
+      "step": 240
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.402219295501709,
+      "learning_rate": 1.5467171717171718e-05,
+      "loss": 0.3612,
+      "step": 245
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.047497272491455,
+      "learning_rate": 1.5782828282828283e-05,
+      "loss": 0.3361,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.719345569610596,
+      "learning_rate": 1.6098484848484848e-05,
+      "loss": 0.3424,
+      "step": 255
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.2764410972595215,
+      "learning_rate": 1.6414141414141416e-05,
+      "loss": 0.3623,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.838797092437744,
+      "learning_rate": 1.672979797979798e-05,
+      "loss": 0.3693,
+      "step": 265
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.54249382019043,
+      "learning_rate": 1.7045454545454546e-05,
+      "loss": 0.361,
+      "step": 270
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.195765495300293,
+      "learning_rate": 1.736111111111111e-05,
+      "loss": 0.3968,
+      "step": 275
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.6201019287109375,
+      "learning_rate": 1.7676767676767676e-05,
+      "loss": 0.3452,
+      "step": 280
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.951904296875,
+      "learning_rate": 1.7992424242424244e-05,
+      "loss": 0.3683,
+      "step": 285
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.5732245445251465,
+      "learning_rate": 1.830808080808081e-05,
+      "loss": 0.3386,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.06243371963501,
+      "learning_rate": 1.8623737373737374e-05,
+      "loss": 0.3522,
+      "step": 295
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.905794620513916,
+      "learning_rate": 1.893939393939394e-05,
+      "loss": 0.3975,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.875278949737549,
+      "learning_rate": 1.9255050505050508e-05,
+      "loss": 0.3815,
+      "step": 305
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.854613304138184,
+      "learning_rate": 1.9570707070707073e-05,
+      "loss": 0.3909,
+      "step": 310
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.744078159332275,
+      "learning_rate": 1.9886363636363638e-05,
+      "loss": 0.3325,
+      "step": 315
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.597081661224365,
+      "learning_rate": 2.0202020202020203e-05,
+      "loss": 0.3719,
+      "step": 320
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.01826286315918,
+      "learning_rate": 2.0517676767676768e-05,
+      "loss": 0.3838,
+      "step": 325
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 7.755288124084473,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.4001,
+      "step": 330
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.653243064880371,
+      "learning_rate": 2.1148989898989898e-05,
+      "loss": 0.3736,
+      "step": 335
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.260344505310059,
+      "learning_rate": 2.1464646464646466e-05,
+      "loss": 0.3709,
+      "step": 340
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.140988349914551,
+      "learning_rate": 2.178030303030303e-05,
+      "loss": 0.355,
+      "step": 345
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.586170196533203,
+      "learning_rate": 2.20959595959596e-05,
+      "loss": 0.3408,
+      "step": 350
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 7.47233247756958,
+      "learning_rate": 2.2411616161616164e-05,
+      "loss": 0.3549,
+      "step": 355
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.570106506347656,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.3181,
+      "step": 360
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 9.143569946289062,
+      "learning_rate": 2.3042929292929294e-05,
+      "loss": 0.3356,
+      "step": 365
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 8.9073486328125,
+      "learning_rate": 2.335858585858586e-05,
+      "loss": 0.3568,
+      "step": 370
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.186656475067139,
+      "learning_rate": 2.3674242424242428e-05,
+      "loss": 0.3685,
+      "step": 375
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 8.131010055541992,
+      "learning_rate": 2.398989898989899e-05,
+      "loss": 0.3729,
+      "step": 380
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.3320746421813965,
+      "learning_rate": 2.4305555555555558e-05,
+      "loss": 0.403,
+      "step": 385
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.896105766296387,
+      "learning_rate": 2.4621212121212123e-05,
+      "loss": 0.3482,
+      "step": 390
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.726195335388184,
+      "learning_rate": 2.4936868686868688e-05,
+      "loss": 0.3356,
+      "step": 395
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.583104610443115,
+      "learning_rate": 2.5252525252525256e-05,
+      "loss": 0.3746,
+      "step": 400
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.950831890106201,
+      "learning_rate": 2.5568181818181817e-05,
+      "loss": 0.3745,
+      "step": 405
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.2157697677612305,
+      "learning_rate": 2.5883838383838382e-05,
+      "loss": 0.336,
+      "step": 410
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.632233142852783,
+      "learning_rate": 2.619949494949495e-05,
+      "loss": 0.3522,
+      "step": 415
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 7.407590389251709,
+      "learning_rate": 2.6515151515151516e-05,
+      "loss": 0.3595,
+      "step": 420
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.182541847229004,
+      "learning_rate": 2.6830808080808084e-05,
+      "loss": 0.3956,
+      "step": 425
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.410284996032715,
+      "learning_rate": 2.714646464646465e-05,
+      "loss": 0.3799,
+      "step": 430
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 9.689289093017578,
+      "learning_rate": 2.746212121212121e-05,
+      "loss": 0.4463,
+      "step": 435
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.971395492553711,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.3616,
+      "step": 440
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 7.5525054931640625,
+      "learning_rate": 2.8093434343434344e-05,
+      "loss": 0.3643,
+      "step": 445
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.672715187072754,
+      "learning_rate": 2.8409090909090912e-05,
+      "loss": 0.4039,
+      "step": 450
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.12246561050415,
+      "learning_rate": 2.8724747474747477e-05,
+      "loss": 0.3334,
+      "step": 455
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.249082088470459,
+      "learning_rate": 2.904040404040404e-05,
+      "loss": 0.39,
+      "step": 460
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.411304950714111,
+      "learning_rate": 2.935606060606061e-05,
+      "loss": 0.3378,
+      "step": 465
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.509771347045898,
+      "learning_rate": 2.9671717171717172e-05,
+      "loss": 0.3743,
+      "step": 470
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.500698566436768,
+      "learning_rate": 2.9987373737373737e-05,
+      "loss": 0.356,
+      "step": 475
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.153441429138184,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.3384,
+      "step": 480
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.439070701599121,
+      "learning_rate": 3.061868686868687e-05,
+      "loss": 0.3442,
+      "step": 485
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.8518900871276855,
+      "learning_rate": 3.0934343434343436e-05,
+      "loss": 0.3638,
+      "step": 490
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.123161315917969,
+      "learning_rate": 3.125e-05,
+      "loss": 0.3523,
+      "step": 495
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.859007358551025,
+      "learning_rate": 3.1565656565656566e-05,
+      "loss": 0.436,
+      "step": 500
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.138620853424072,
+      "learning_rate": 3.188131313131314e-05,
+      "loss": 0.339,
+      "step": 505
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.137459754943848,
+      "learning_rate": 3.2196969696969696e-05,
+      "loss": 0.3705,
+      "step": 510
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.6870880126953125,
+      "learning_rate": 3.251262626262627e-05,
+      "loss": 0.3422,
+      "step": 515
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 10.836819648742676,
+      "learning_rate": 3.282828282828283e-05,
+      "loss": 0.3844,
+      "step": 520
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 7.4189324378967285,
+      "learning_rate": 3.314393939393939e-05,
+      "loss": 0.3733,
+      "step": 525
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 6.844618320465088,
+      "learning_rate": 3.345959595959596e-05,
+      "loss": 0.3402,
+      "step": 530
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.5922932624816895,
+      "learning_rate": 3.377525252525253e-05,
+      "loss": 0.3738,
+      "step": 535
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.552920341491699,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 0.3493,
+      "step": 540
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 7.970205783843994,
+      "learning_rate": 3.440656565656566e-05,
+      "loss": 0.3124,
+      "step": 545
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.992683410644531,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.3272,
+      "step": 550
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.078422546386719,
+      "learning_rate": 3.5037878787878794e-05,
+      "loss": 0.321,
+      "step": 555
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.718451023101807,
+      "learning_rate": 3.535353535353535e-05,
+      "loss": 0.2823,
+      "step": 560
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.938317775726318,
+      "learning_rate": 3.566919191919192e-05,
+      "loss": 0.3317,
+      "step": 565
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.020884990692139,
+      "learning_rate": 3.598484848484849e-05,
+      "loss": 0.3149,
+      "step": 570
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.855194568634033,
+      "learning_rate": 3.6300505050505054e-05,
+      "loss": 0.3093,
+      "step": 575
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.711932182312012,
+      "learning_rate": 3.661616161616162e-05,
+      "loss": 0.3436,
+      "step": 580
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.580200672149658,
+      "learning_rate": 3.6931818181818184e-05,
+      "loss": 0.3601,
+      "step": 585
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.678188323974609,
+      "learning_rate": 3.724747474747475e-05,
+      "loss": 0.3181,
+      "step": 590
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.502762317657471,
+      "learning_rate": 3.7563131313131314e-05,
+      "loss": 0.3434,
+      "step": 595
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.576380729675293,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 0.316,
+      "step": 600
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.164270877838135,
+      "learning_rate": 3.8194444444444444e-05,
+      "loss": 0.3284,
+      "step": 605
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.344931125640869,
+      "learning_rate": 3.8510101010101015e-05,
+      "loss": 0.2597,
+      "step": 610
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 8.411974906921387,
+      "learning_rate": 3.8825757575757574e-05,
+      "loss": 0.3743,
+      "step": 615
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.547687530517578,
+      "learning_rate": 3.9141414141414145e-05,
+      "loss": 0.326,
+      "step": 620
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.556216239929199,
+      "learning_rate": 3.945707070707071e-05,
+      "loss": 0.2833,
+      "step": 625
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.882171630859375,
+      "learning_rate": 3.9772727272727275e-05,
+      "loss": 0.3326,
+      "step": 630
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.282110214233398,
+      "learning_rate": 4.008838383838384e-05,
+      "loss": 0.3164,
+      "step": 635
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.85945463180542,
+      "learning_rate": 4.0404040404040405e-05,
+      "loss": 0.3376,
+      "step": 640
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.612574577331543,
+      "learning_rate": 4.071969696969698e-05,
+      "loss": 0.3251,
+      "step": 645
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.971954345703125,
+      "learning_rate": 4.1035353535353535e-05,
+      "loss": 0.3217,
+      "step": 650
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.100465774536133,
+      "learning_rate": 4.13510101010101e-05,
+      "loss": 0.2818,
+      "step": 655
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.470828533172607,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.3824,
+      "step": 660
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.753726959228516,
+      "learning_rate": 4.198232323232324e-05,
+      "loss": 0.3121,
+      "step": 665
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.886876106262207,
+      "learning_rate": 4.2297979797979795e-05,
+      "loss": 0.3579,
+      "step": 670
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.900394916534424,
+      "learning_rate": 4.261363636363637e-05,
+      "loss": 0.3588,
+      "step": 675
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.405203342437744,
+      "learning_rate": 4.292929292929293e-05,
+      "loss": 0.3274,
+      "step": 680
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 7.3850884437561035,
+      "learning_rate": 4.32449494949495e-05,
+      "loss": 0.3266,
+      "step": 685
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 11.002800941467285,
+      "learning_rate": 4.356060606060606e-05,
+      "loss": 0.3474,
+      "step": 690
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.883533954620361,
+      "learning_rate": 4.387626262626263e-05,
+      "loss": 0.299,
+      "step": 695
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 8.321223258972168,
+      "learning_rate": 4.41919191919192e-05,
+      "loss": 0.3322,
+      "step": 700
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.380652904510498,
+      "learning_rate": 4.450757575757576e-05,
+      "loss": 0.3067,
+      "step": 705
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 10.384994506835938,
+      "learning_rate": 4.482323232323233e-05,
+      "loss": 0.3191,
+      "step": 710
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 8.773216247558594,
+      "learning_rate": 4.5138888888888894e-05,
+      "loss": 0.3478,
+      "step": 715
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.572368144989014,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.3499,
+      "step": 720
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 8.111650466918945,
+      "learning_rate": 4.5770202020202024e-05,
+      "loss": 0.3425,
+      "step": 725
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.066849231719971,
+      "learning_rate": 4.608585858585859e-05,
+      "loss": 0.2808,
+      "step": 730
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.649580001831055,
+      "learning_rate": 4.6401515151515154e-05,
+      "loss": 0.3322,
+      "step": 735
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.393223762512207,
+      "learning_rate": 4.671717171717172e-05,
+      "loss": 0.3113,
+      "step": 740
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.742656230926514,
+      "learning_rate": 4.7032828282828283e-05,
+      "loss": 0.3641,
+      "step": 745
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.521514415740967,
+      "learning_rate": 4.7348484848484855e-05,
+      "loss": 0.3556,
+      "step": 750
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.544399738311768,
+      "learning_rate": 4.7664141414141413e-05,
+      "loss": 0.3603,
+      "step": 755
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.904191017150879,
+      "learning_rate": 4.797979797979798e-05,
+      "loss": 0.3105,
+      "step": 760
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.711209774017334,
+      "learning_rate": 4.829545454545455e-05,
+      "loss": 0.3257,
+      "step": 765
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.6685709953308105,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.3354,
+      "step": 770
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.159426212310791,
+      "learning_rate": 4.892676767676767e-05,
+      "loss": 0.3052,
+      "step": 775
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.870561122894287,
+      "learning_rate": 4.9242424242424245e-05,
+      "loss": 0.2651,
+      "step": 780
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.169192314147949,
+      "learning_rate": 4.955808080808081e-05,
+      "loss": 0.2628,
+      "step": 785
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.3205976486206055,
+      "learning_rate": 4.9873737373737375e-05,
+      "loss": 0.3281,
+      "step": 790
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9074287742966184,
+      "eval_loss": 0.2726101577281952,
+      "eval_runtime": 30.2701,
+      "eval_samples_per_second": 372.215,
+      "eval_steps_per_second": 11.662,
+      "step": 792
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 8.79980182647705,
+      "learning_rate": 4.997895622895623e-05,
+      "loss": 0.3758,
+      "step": 795
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 8.34786319732666,
+      "learning_rate": 4.994388327721661e-05,
+      "loss": 0.2892,
+      "step": 800
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.995574951171875,
+      "learning_rate": 4.9908810325477e-05,
+      "loss": 0.309,
+      "step": 805
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.120898246765137,
+      "learning_rate": 4.9873737373737375e-05,
+      "loss": 0.2644,
+      "step": 810
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 7.833585262298584,
+      "learning_rate": 4.983866442199775e-05,
+      "loss": 0.3759,
+      "step": 815
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.475259304046631,
+      "learning_rate": 4.980359147025814e-05,
+      "loss": 0.3299,
+      "step": 820
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.855483055114746,
+      "learning_rate": 4.976851851851852e-05,
+      "loss": 0.3579,
+      "step": 825
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.96250581741333,
+      "learning_rate": 4.97334455667789e-05,
+      "loss": 0.2873,
+      "step": 830
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.715710639953613,
+      "learning_rate": 4.9698372615039285e-05,
+      "loss": 0.2978,
+      "step": 835
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 6.888841152191162,
+      "learning_rate": 4.966329966329967e-05,
+      "loss": 0.2915,
+      "step": 840
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.960631370544434,
+      "learning_rate": 4.962822671156005e-05,
+      "loss": 0.3697,
+      "step": 845
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.9937028884887695,
+      "learning_rate": 4.9593153759820425e-05,
+      "loss": 0.2814,
+      "step": 850
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.158823013305664,
+      "learning_rate": 4.955808080808081e-05,
+      "loss": 0.2447,
+      "step": 855
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 7.404542446136475,
+      "learning_rate": 4.9523007856341195e-05,
+      "loss": 0.3188,
+      "step": 860
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 5.870089054107666,
+      "learning_rate": 4.948793490460157e-05,
+      "loss": 0.2989,
+      "step": 865
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.1226983070373535,
+      "learning_rate": 4.945286195286196e-05,
+      "loss": 0.2756,
+      "step": 870
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 6.2829203605651855,
+      "learning_rate": 4.941778900112234e-05,
+      "loss": 0.2453,
+      "step": 875
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 5.74476432800293,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 0.341,
+      "step": 880
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.578733444213867,
+      "learning_rate": 4.93476430976431e-05,
+      "loss": 0.2674,
+      "step": 885
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.299071788787842,
+      "learning_rate": 4.931257014590348e-05,
+      "loss": 0.3993,
+      "step": 890
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.260754585266113,
+      "learning_rate": 4.927749719416386e-05,
+      "loss": 0.2653,
+      "step": 895
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.061776638031006,
+      "learning_rate": 4.9242424242424245e-05,
+      "loss": 0.2809,
+      "step": 900
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.618584156036377,
+      "learning_rate": 4.920735129068463e-05,
+      "loss": 0.3077,
+      "step": 905
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.5875964164733887,
+      "learning_rate": 4.917227833894501e-05,
+      "loss": 0.2913,
+      "step": 910
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 4.226754188537598,
+      "learning_rate": 4.9137205387205386e-05,
+      "loss": 0.2864,
+      "step": 915
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.963589668273926,
+      "learning_rate": 4.910213243546577e-05,
+      "loss": 0.3157,
+      "step": 920
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.765893936157227,
+      "learning_rate": 4.9067059483726155e-05,
+      "loss": 0.3482,
+      "step": 925
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 6.1433587074279785,
+      "learning_rate": 4.903198653198653e-05,
+      "loss": 0.2988,
+      "step": 930
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.580607891082764,
+      "learning_rate": 4.899691358024692e-05,
+      "loss": 0.3024,
+      "step": 935
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.167281150817871,
+      "learning_rate": 4.89618406285073e-05,
+      "loss": 0.2796,
+      "step": 940
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.730119705200195,
+      "learning_rate": 4.892676767676767e-05,
+      "loss": 0.2802,
+      "step": 945
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 5.232539176940918,
+      "learning_rate": 4.889169472502806e-05,
+      "loss": 0.2917,
+      "step": 950
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 9.396648406982422,
+      "learning_rate": 4.885662177328844e-05,
+      "loss": 0.3061,
+      "step": 955
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 5.580398082733154,
+      "learning_rate": 4.882154882154882e-05,
+      "loss": 0.2816,
+      "step": 960
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 5.00685453414917,
+      "learning_rate": 4.8786475869809205e-05,
+      "loss": 0.2377,
+      "step": 965
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 4.347193717956543,
+      "learning_rate": 4.875140291806959e-05,
+      "loss": 0.2854,
+      "step": 970
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 8.344078063964844,
+      "learning_rate": 4.871632996632997e-05,
+      "loss": 0.3654,
+      "step": 975
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 3.9968340396881104,
+      "learning_rate": 4.8681257014590346e-05,
+      "loss": 0.2709,
+      "step": 980
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.086044788360596,
+      "learning_rate": 4.864618406285073e-05,
+      "loss": 0.2583,
+      "step": 985
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 6.1204962730407715,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.3022,
+      "step": 990
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 4.514072895050049,
+      "learning_rate": 4.857603815937149e-05,
+      "loss": 0.2892,
+      "step": 995
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 4.766295909881592,
+      "learning_rate": 4.854096520763188e-05,
+      "loss": 0.3233,
+      "step": 1000
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 6.514324188232422,
+      "learning_rate": 4.850589225589226e-05,
+      "loss": 0.352,
+      "step": 1005
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.8759655952453613,
+      "learning_rate": 4.847081930415264e-05,
+      "loss": 0.2878,
+      "step": 1010
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 6.272017478942871,
+      "learning_rate": 4.843574635241302e-05,
+      "loss": 0.299,
+      "step": 1015
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 6.9353790283203125,
+      "learning_rate": 4.84006734006734e-05,
+      "loss": 0.3158,
+      "step": 1020
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 6.130936622619629,
+      "learning_rate": 4.836560044893379e-05,
+      "loss": 0.2587,
+      "step": 1025
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.3950865268707275,
+      "learning_rate": 4.8330527497194165e-05,
+      "loss": 0.2754,
+      "step": 1030
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 8.806388854980469,
+      "learning_rate": 4.829545454545455e-05,
+      "loss": 0.3292,
+      "step": 1035
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 7.561717987060547,
+      "learning_rate": 4.8260381593714935e-05,
+      "loss": 0.2389,
+      "step": 1040
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 8.65743637084961,
+      "learning_rate": 4.8225308641975306e-05,
+      "loss": 0.3279,
+      "step": 1045
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 5.86659049987793,
+      "learning_rate": 4.819023569023569e-05,
+      "loss": 0.3095,
+      "step": 1050
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 7.73600959777832,
+      "learning_rate": 4.8155162738496075e-05,
+      "loss": 0.3485,
+      "step": 1055
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 4.956167697906494,
+      "learning_rate": 4.812008978675645e-05,
+      "loss": 0.3488,
+      "step": 1060
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 6.985816955566406,
+      "learning_rate": 4.808501683501684e-05,
+      "loss": 0.3873,
+      "step": 1065
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 4.710123062133789,
+      "learning_rate": 4.804994388327722e-05,
+      "loss": 0.4012,
+      "step": 1070
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 6.593132495880127,
+      "learning_rate": 4.80148709315376e-05,
+      "loss": 0.3882,
+      "step": 1075
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 5.515847682952881,
+      "learning_rate": 4.797979797979798e-05,
+      "loss": 0.3156,
+      "step": 1080
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 6.208699703216553,
+      "learning_rate": 4.794472502805836e-05,
+      "loss": 0.3626,
+      "step": 1085
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 6.190163612365723,
+      "learning_rate": 4.790965207631875e-05,
+      "loss": 0.3481,
+      "step": 1090
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 11.208972930908203,
+      "learning_rate": 4.7874579124579126e-05,
+      "loss": 0.4056,
+      "step": 1095
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 5.851810455322266,
+      "learning_rate": 4.783950617283951e-05,
+      "loss": 0.3477,
+      "step": 1100
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 4.882748603820801,
+      "learning_rate": 4.7804433221099895e-05,
+      "loss": 0.4436,
+      "step": 1105
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 17.56443214416504,
+      "learning_rate": 4.7769360269360266e-05,
+      "loss": 0.4184,
+      "step": 1110
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 5.368799209594727,
+      "learning_rate": 4.773428731762065e-05,
+      "loss": 0.3565,
+      "step": 1115
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 8.128335952758789,
+      "learning_rate": 4.7699214365881036e-05,
+      "loss": 0.3948,
+      "step": 1120
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 5.376011848449707,
+      "learning_rate": 4.7664141414141413e-05,
+      "loss": 0.3788,
+      "step": 1125
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 5.1336989402771,
+      "learning_rate": 4.76290684624018e-05,
+      "loss": 0.373,
+      "step": 1130
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 8.23189640045166,
+      "learning_rate": 4.759399551066218e-05,
+      "loss": 0.4303,
+      "step": 1135
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 8.911417007446289,
+      "learning_rate": 4.755892255892256e-05,
+      "loss": 0.4153,
+      "step": 1140
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 4.826540946960449,
+      "learning_rate": 4.752384960718294e-05,
+      "loss": 0.3791,
+      "step": 1145
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 5.428144931793213,
+      "learning_rate": 4.748877665544332e-05,
+      "loss": 0.3209,
+      "step": 1150
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 6.460602283477783,
+      "learning_rate": 4.745370370370371e-05,
+      "loss": 0.3716,
+      "step": 1155
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 6.3134074211120605,
+      "learning_rate": 4.7418630751964086e-05,
+      "loss": 0.4052,
+      "step": 1160
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 5.615858554840088,
+      "learning_rate": 4.738355780022447e-05,
+      "loss": 0.3763,
+      "step": 1165
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 4.606671333312988,
+      "learning_rate": 4.7348484848484855e-05,
+      "loss": 0.4,
+      "step": 1170
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 5.418331146240234,
+      "learning_rate": 4.731341189674523e-05,
+      "loss": 0.3651,
+      "step": 1175
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 6.259594917297363,
+      "learning_rate": 4.727833894500561e-05,
+      "loss": 0.3761,
+      "step": 1180
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 5.668846607208252,
+      "learning_rate": 4.7243265993265996e-05,
+      "loss": 0.3686,
+      "step": 1185
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 4.353462219238281,
+      "learning_rate": 4.720819304152638e-05,
+      "loss": 0.4522,
+      "step": 1190
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 6.253231048583984,
+      "learning_rate": 4.717312008978676e-05,
+      "loss": 0.3652,
+      "step": 1195
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 5.421621799468994,
+      "learning_rate": 4.713804713804714e-05,
+      "loss": 0.3651,
+      "step": 1200
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 6.4615254402160645,
+      "learning_rate": 4.710297418630752e-05,
+      "loss": 0.3915,
+      "step": 1205
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 5.112613677978516,
+      "learning_rate": 4.70679012345679e-05,
+      "loss": 0.3875,
+      "step": 1210
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 4.873692035675049,
+      "learning_rate": 4.7032828282828283e-05,
+      "loss": 0.3579,
+      "step": 1215
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 5.282414436340332,
+      "learning_rate": 4.699775533108867e-05,
+      "loss": 0.3637,
+      "step": 1220
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 6.1139326095581055,
+      "learning_rate": 4.6962682379349046e-05,
+      "loss": 0.3375,
+      "step": 1225
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 5.3503570556640625,
+      "learning_rate": 4.692760942760943e-05,
+      "loss": 0.3735,
+      "step": 1230
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 4.9948320388793945,
+      "learning_rate": 4.6892536475869815e-05,
+      "loss": 0.3671,
+      "step": 1235
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 4.220916271209717,
+      "learning_rate": 4.685746352413019e-05,
+      "loss": 0.3288,
+      "step": 1240
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 7.0633320808410645,
+      "learning_rate": 4.682239057239057e-05,
+      "loss": 0.4333,
+      "step": 1245
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 5.053625583648682,
+      "learning_rate": 4.6787317620650956e-05,
+      "loss": 0.3685,
+      "step": 1250
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 5.34464168548584,
+      "learning_rate": 4.675224466891134e-05,
+      "loss": 0.38,
+      "step": 1255
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 5.092491626739502,
+      "learning_rate": 4.671717171717172e-05,
+      "loss": 0.3958,
+      "step": 1260
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 5.202730655670166,
+      "learning_rate": 4.66820987654321e-05,
+      "loss": 0.3605,
+      "step": 1265
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 6.6122565269470215,
+      "learning_rate": 4.664702581369249e-05,
+      "loss": 0.389,
+      "step": 1270
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 5.779764175415039,
+      "learning_rate": 4.661195286195286e-05,
+      "loss": 0.3243,
+      "step": 1275
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 5.279807090759277,
+      "learning_rate": 4.6576879910213244e-05,
+      "loss": 0.3811,
+      "step": 1280
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 9.88245964050293,
+      "learning_rate": 4.654180695847363e-05,
+      "loss": 0.3707,
+      "step": 1285
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 4.916264533996582,
+      "learning_rate": 4.6506734006734006e-05,
+      "loss": 0.3429,
+      "step": 1290
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 8.445237159729004,
+      "learning_rate": 4.647166105499439e-05,
+      "loss": 0.3736,
+      "step": 1295
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.9717190265655518,
+      "learning_rate": 4.6436588103254776e-05,
+      "loss": 0.3817,
+      "step": 1300
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 5.730360984802246,
+      "learning_rate": 4.6401515151515154e-05,
+      "loss": 0.3613,
+      "step": 1305
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 5.815140247344971,
+      "learning_rate": 4.636644219977553e-05,
+      "loss": 0.3751,
+      "step": 1310
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 5.798765659332275,
+      "learning_rate": 4.6331369248035916e-05,
+      "loss": 0.4397,
+      "step": 1315
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 11.663105964660645,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.3845,
+      "step": 1320
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 7.130865573883057,
+      "learning_rate": 4.626122334455668e-05,
+      "loss": 0.3788,
+      "step": 1325
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 4.88721227645874,
+      "learning_rate": 4.622615039281706e-05,
+      "loss": 0.3824,
+      "step": 1330
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 6.343611240386963,
+      "learning_rate": 4.619107744107745e-05,
+      "loss": 0.4382,
+      "step": 1335
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 4.633550643920898,
+      "learning_rate": 4.6156004489337826e-05,
+      "loss": 0.3251,
+      "step": 1340
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 5.174993515014648,
+      "learning_rate": 4.6120931537598204e-05,
+      "loss": 0.3692,
+      "step": 1345
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 5.309804916381836,
+      "learning_rate": 4.608585858585859e-05,
+      "loss": 0.3935,
+      "step": 1350
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 4.699385643005371,
+      "learning_rate": 4.605078563411897e-05,
+      "loss": 0.3077,
+      "step": 1355
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 4.771191120147705,
+      "learning_rate": 4.601571268237935e-05,
+      "loss": 0.3601,
+      "step": 1360
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 4.814445972442627,
+      "learning_rate": 4.5980639730639736e-05,
+      "loss": 0.3333,
+      "step": 1365
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 7.081927299499512,
+      "learning_rate": 4.5945566778900114e-05,
+      "loss": 0.3544,
+      "step": 1370
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 5.424753665924072,
+      "learning_rate": 4.591049382716049e-05,
+      "loss": 0.3482,
+      "step": 1375
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 4.949700832366943,
+      "learning_rate": 4.5875420875420876e-05,
+      "loss": 0.3574,
+      "step": 1380
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 3.7646026611328125,
+      "learning_rate": 4.584034792368126e-05,
+      "loss": 0.3712,
+      "step": 1385
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 4.071770668029785,
+      "learning_rate": 4.580527497194164e-05,
+      "loss": 0.4441,
+      "step": 1390
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.3949429988861084,
+      "learning_rate": 4.5770202020202024e-05,
+      "loss": 0.3856,
+      "step": 1395
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 4.872361183166504,
+      "learning_rate": 4.573512906846241e-05,
+      "loss": 0.3636,
+      "step": 1400
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 4.493046760559082,
+      "learning_rate": 4.5700056116722786e-05,
+      "loss": 0.3666,
+      "step": 1405
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 4.998297691345215,
+      "learning_rate": 4.5664983164983164e-05,
+      "loss": 0.3263,
+      "step": 1410
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 4.83442497253418,
+      "learning_rate": 4.562991021324355e-05,
+      "loss": 0.3429,
+      "step": 1415
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 4.50236701965332,
+      "learning_rate": 4.5594837261503933e-05,
+      "loss": 0.2945,
+      "step": 1420
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.233889579772949,
+      "learning_rate": 4.555976430976431e-05,
+      "loss": 0.3994,
+      "step": 1425
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.307263374328613,
+      "learning_rate": 4.5524691358024696e-05,
+      "loss": 0.3399,
+      "step": 1430
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.249146461486816,
+      "learning_rate": 4.5489618406285074e-05,
+      "loss": 0.3312,
+      "step": 1435
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 4.6194305419921875,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.3909,
+      "step": 1440
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 7.478885650634766,
+      "learning_rate": 4.5419472502805836e-05,
+      "loss": 0.3273,
+      "step": 1445
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 5.683533668518066,
+      "learning_rate": 4.538439955106622e-05,
+      "loss": 0.3942,
+      "step": 1450
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 4.889841079711914,
+      "learning_rate": 4.5349326599326606e-05,
+      "loss": 0.3861,
+      "step": 1455
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 6.0156097412109375,
+      "learning_rate": 4.5314253647586984e-05,
+      "loss": 0.3326,
+      "step": 1460
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 5.2260966300964355,
+      "learning_rate": 4.527918069584737e-05,
+      "loss": 0.3788,
+      "step": 1465
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 5.414533615112305,
+      "learning_rate": 4.5244107744107746e-05,
+      "loss": 0.3502,
+      "step": 1470
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 5.083827018737793,
+      "learning_rate": 4.5209034792368124e-05,
+      "loss": 0.3708,
+      "step": 1475
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 4.513807773590088,
+      "learning_rate": 4.517396184062851e-05,
+      "loss": 0.342,
+      "step": 1480
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 5.6209306716918945,
+      "learning_rate": 4.5138888888888894e-05,
+      "loss": 0.3859,
+      "step": 1485
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 5.69753360748291,
+      "learning_rate": 4.510381593714927e-05,
+      "loss": 0.3575,
+      "step": 1490
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 4.411763668060303,
+      "learning_rate": 4.5068742985409656e-05,
+      "loss": 0.3641,
+      "step": 1495
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 5.696104049682617,
+      "learning_rate": 4.5033670033670034e-05,
+      "loss": 0.4081,
+      "step": 1500
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 4.2068328857421875,
+      "learning_rate": 4.499859708193042e-05,
+      "loss": 0.3609,
+      "step": 1505
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 6.309601306915283,
+      "learning_rate": 4.49635241301908e-05,
+      "loss": 0.401,
+      "step": 1510
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 5.353550910949707,
+      "learning_rate": 4.492845117845118e-05,
+      "loss": 0.4102,
+      "step": 1515
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 3.678487539291382,
+      "learning_rate": 4.4893378226711566e-05,
+      "loss": 0.3771,
+      "step": 1520
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 5.291224956512451,
+      "learning_rate": 4.4858305274971944e-05,
+      "loss": 0.3758,
+      "step": 1525
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 4.863537311553955,
+      "learning_rate": 4.482323232323233e-05,
+      "loss": 0.4096,
+      "step": 1530
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 5.359092712402344,
+      "learning_rate": 4.4788159371492707e-05,
+      "loss": 0.3885,
+      "step": 1535
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 3.60457444190979,
+      "learning_rate": 4.4753086419753084e-05,
+      "loss": 0.3486,
+      "step": 1540
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 4.0034098625183105,
+      "learning_rate": 4.471801346801347e-05,
+      "loss": 0.3616,
+      "step": 1545
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.7283554077148438,
+      "learning_rate": 4.4682940516273854e-05,
+      "loss": 0.3479,
+      "step": 1550
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 4.062889099121094,
+      "learning_rate": 4.464786756453423e-05,
+      "loss": 0.3356,
+      "step": 1555
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 5.248371601104736,
+      "learning_rate": 4.4612794612794616e-05,
+      "loss": 0.417,
+      "step": 1560
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 4.094751834869385,
+      "learning_rate": 4.4577721661054994e-05,
+      "loss": 0.3633,
+      "step": 1565
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 5.946717739105225,
+      "learning_rate": 4.454264870931538e-05,
+      "loss": 0.3865,
+      "step": 1570
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 4.753849506378174,
+      "learning_rate": 4.450757575757576e-05,
+      "loss": 0.3755,
+      "step": 1575
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 5.561149597167969,
+      "learning_rate": 4.447250280583614e-05,
+      "loss": 0.3858,
+      "step": 1580
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9191444040117156,
+      "eval_loss": 0.238821342587471,
+      "eval_runtime": 30.6428,
+      "eval_samples_per_second": 367.688,
+      "eval_steps_per_second": 11.52,
+      "step": 1584
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 4.399683475494385,
+      "learning_rate": 4.4437429854096526e-05,
+      "loss": 0.3895,
+      "step": 1585
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 4.277946472167969,
+      "learning_rate": 4.4402356902356904e-05,
+      "loss": 0.3646,
+      "step": 1590
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 4.514025688171387,
+      "learning_rate": 4.436728395061729e-05,
+      "loss": 0.3665,
+      "step": 1595
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 5.112893104553223,
+      "learning_rate": 4.433221099887767e-05,
+      "loss": 0.3455,
+      "step": 1600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 3.7568562030792236,
+      "learning_rate": 4.429713804713805e-05,
+      "loss": 0.2892,
+      "step": 1605
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 5.433623790740967,
+      "learning_rate": 4.426206509539843e-05,
+      "loss": 0.3422,
+      "step": 1610
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 5.30995512008667,
+      "learning_rate": 4.4226992143658814e-05,
+      "loss": 0.3403,
+      "step": 1615
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 4.3577189445495605,
+      "learning_rate": 4.41919191919192e-05,
+      "loss": 0.3074,
+      "step": 1620
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 5.388636589050293,
+      "learning_rate": 4.4156846240179577e-05,
+      "loss": 0.395,
+      "step": 1625
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 6.779383659362793,
+      "learning_rate": 4.4121773288439954e-05,
+      "loss": 0.3433,
+      "step": 1630
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 4.6266655921936035,
+      "learning_rate": 4.408670033670034e-05,
+      "loss": 0.3681,
+      "step": 1635
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 4.8226470947265625,
+      "learning_rate": 4.405162738496072e-05,
+      "loss": 0.4178,
+      "step": 1640
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 4.542765140533447,
+      "learning_rate": 4.40165544332211e-05,
+      "loss": 0.3342,
+      "step": 1645
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 5.656178951263428,
+      "learning_rate": 4.3981481481481486e-05,
+      "loss": 0.4088,
+      "step": 1650
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 6.368711948394775,
+      "learning_rate": 4.3946408529741864e-05,
+      "loss": 0.3735,
+      "step": 1655
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 5.6990532875061035,
+      "learning_rate": 4.391133557800225e-05,
+      "loss": 0.3299,
+      "step": 1660
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 4.243890285491943,
+      "learning_rate": 4.387626262626263e-05,
+      "loss": 0.3423,
+      "step": 1665
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 4.525588035583496,
+      "learning_rate": 4.384118967452301e-05,
+      "loss": 0.3774,
+      "step": 1670
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 5.508600234985352,
+      "learning_rate": 4.380611672278339e-05,
+      "loss": 0.3774,
+      "step": 1675
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 5.072585105895996,
+      "learning_rate": 4.3771043771043774e-05,
+      "loss": 0.3367,
+      "step": 1680
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 5.3470916748046875,
+      "learning_rate": 4.373597081930416e-05,
+      "loss": 0.3801,
+      "step": 1685
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 4.230000972747803,
+      "learning_rate": 4.370089786756454e-05,
+      "loss": 0.3349,
+      "step": 1690
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 5.275707244873047,
+      "learning_rate": 4.3665824915824915e-05,
+      "loss": 0.3061,
+      "step": 1695
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 8.197766304016113,
+      "learning_rate": 4.36307519640853e-05,
+      "loss": 0.3691,
+      "step": 1700
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 3.7935025691986084,
+      "learning_rate": 4.359567901234568e-05,
+      "loss": 0.2861,
+      "step": 1705
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 8.349021911621094,
+      "learning_rate": 4.356060606060606e-05,
+      "loss": 0.3493,
+      "step": 1710
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 4.536565780639648,
+      "learning_rate": 4.3525533108866447e-05,
+      "loss": 0.3424,
+      "step": 1715
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 4.19216251373291,
+      "learning_rate": 4.3490460157126825e-05,
+      "loss": 0.3153,
+      "step": 1720
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 3.8481414318084717,
+      "learning_rate": 4.345538720538721e-05,
+      "loss": 0.3553,
+      "step": 1725
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 4.058391094207764,
+      "learning_rate": 4.342031425364759e-05,
+      "loss": 0.3507,
+      "step": 1730
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 6.941283226013184,
+      "learning_rate": 4.338524130190797e-05,
+      "loss": 0.3657,
+      "step": 1735
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 5.315598011016846,
+      "learning_rate": 4.335016835016835e-05,
+      "loss": 0.3934,
+      "step": 1740
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 3.5970749855041504,
+      "learning_rate": 4.3315095398428734e-05,
+      "loss": 0.3391,
+      "step": 1745
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.7733476161956787,
+      "learning_rate": 4.328002244668912e-05,
+      "loss": 0.2927,
+      "step": 1750
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 3.066009283065796,
+      "learning_rate": 4.32449494949495e-05,
+      "loss": 0.3159,
+      "step": 1755
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 4.86247444152832,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 0.3831,
+      "step": 1760
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 4.143043518066406,
+      "learning_rate": 4.317480359147026e-05,
+      "loss": 0.319,
+      "step": 1765
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 4.56455135345459,
+      "learning_rate": 4.3139730639730644e-05,
+      "loss": 0.3157,
+      "step": 1770
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 4.469295024871826,
+      "learning_rate": 4.310465768799102e-05,
+      "loss": 0.3114,
+      "step": 1775
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 4.398653507232666,
+      "learning_rate": 4.306958473625141e-05,
+      "loss": 0.3309,
+      "step": 1780
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 4.164375305175781,
+      "learning_rate": 4.303451178451179e-05,
+      "loss": 0.3012,
+      "step": 1785
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 5.0801191329956055,
+      "learning_rate": 4.299943883277217e-05,
+      "loss": 0.2919,
+      "step": 1790
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 4.642740726470947,
+      "learning_rate": 4.296436588103255e-05,
+      "loss": 0.3578,
+      "step": 1795
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 4.617292404174805,
+      "learning_rate": 4.292929292929293e-05,
+      "loss": 0.332,
+      "step": 1800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 4.861376762390137,
+      "learning_rate": 4.289421997755331e-05,
+      "loss": 0.35,
+      "step": 1805
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 4.150514602661133,
+      "learning_rate": 4.2859147025813695e-05,
+      "loss": 0.3394,
+      "step": 1810
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 4.65651798248291,
+      "learning_rate": 4.282407407407408e-05,
+      "loss": 0.2993,
+      "step": 1815
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 5.09665060043335,
+      "learning_rate": 4.278900112233446e-05,
+      "loss": 0.3343,
+      "step": 1820
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 5.433923244476318,
+      "learning_rate": 4.2753928170594835e-05,
+      "loss": 0.3599,
+      "step": 1825
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 3.9490609169006348,
+      "learning_rate": 4.271885521885522e-05,
+      "loss": 0.2961,
+      "step": 1830
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 4.98903751373291,
+      "learning_rate": 4.2683782267115604e-05,
+      "loss": 0.3604,
+      "step": 1835
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 6.297445297241211,
+      "learning_rate": 4.264870931537598e-05,
+      "loss": 0.3621,
+      "step": 1840
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 4.120467185974121,
+      "learning_rate": 4.261363636363637e-05,
+      "loss": 0.3558,
+      "step": 1845
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 5.103471755981445,
+      "learning_rate": 4.257856341189675e-05,
+      "loss": 0.2985,
+      "step": 1850
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 4.877329349517822,
+      "learning_rate": 4.254349046015713e-05,
+      "loss": 0.3268,
+      "step": 1855
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 4.503664016723633,
+      "learning_rate": 4.250841750841751e-05,
+      "loss": 0.3332,
+      "step": 1860
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 4.7574286460876465,
+      "learning_rate": 4.247334455667789e-05,
+      "loss": 0.3568,
+      "step": 1865
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 5.113763809204102,
+      "learning_rate": 4.243827160493827e-05,
+      "loss": 0.3517,
+      "step": 1870
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 5.123772144317627,
+      "learning_rate": 4.2403198653198655e-05,
+      "loss": 0.3457,
+      "step": 1875
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 6.1239542961120605,
+      "learning_rate": 4.236812570145904e-05,
+      "loss": 0.3828,
+      "step": 1880
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 4.464895248413086,
+      "learning_rate": 4.233305274971942e-05,
+      "loss": 0.3287,
+      "step": 1885
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 4.614039421081543,
+      "learning_rate": 4.2297979797979795e-05,
+      "loss": 0.2941,
+      "step": 1890
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 4.918231964111328,
+      "learning_rate": 4.226290684624018e-05,
+      "loss": 0.3533,
+      "step": 1895
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 5.08254337310791,
+      "learning_rate": 4.2227833894500565e-05,
+      "loss": 0.3486,
+      "step": 1900
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 5.221201419830322,
+      "learning_rate": 4.219276094276094e-05,
+      "loss": 0.3594,
+      "step": 1905
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 3.0806403160095215,
+      "learning_rate": 4.215768799102133e-05,
+      "loss": 0.3374,
+      "step": 1910
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 4.533454895019531,
+      "learning_rate": 4.212261503928171e-05,
+      "loss": 0.335,
+      "step": 1915
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 3.4555981159210205,
+      "learning_rate": 4.208754208754209e-05,
+      "loss": 0.3093,
+      "step": 1920
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 4.895094871520996,
+      "learning_rate": 4.205246913580247e-05,
+      "loss": 0.3278,
+      "step": 1925
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 3.9753971099853516,
+      "learning_rate": 4.201739618406285e-05,
+      "loss": 0.2992,
+      "step": 1930
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 3.6200075149536133,
+      "learning_rate": 4.198232323232324e-05,
+      "loss": 0.2834,
+      "step": 1935
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 3.9631054401397705,
+      "learning_rate": 4.1947250280583615e-05,
+      "loss": 0.2866,
+      "step": 1940
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 5.465904712677002,
+      "learning_rate": 4.1912177328844e-05,
+      "loss": 0.3507,
+      "step": 1945
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 5.171625137329102,
+      "learning_rate": 4.1877104377104384e-05,
+      "loss": 0.3361,
+      "step": 1950
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 4.916808128356934,
+      "learning_rate": 4.1842031425364755e-05,
+      "loss": 0.2809,
+      "step": 1955
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 4.764858245849609,
+      "learning_rate": 4.180695847362514e-05,
+      "loss": 0.3078,
+      "step": 1960
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 6.535092830657959,
+      "learning_rate": 4.1771885521885525e-05,
+      "loss": 0.3821,
+      "step": 1965
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 3.9484145641326904,
+      "learning_rate": 4.17368125701459e-05,
+      "loss": 0.3501,
+      "step": 1970
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 3.7241628170013428,
+      "learning_rate": 4.170173961840629e-05,
+      "loss": 0.3211,
+      "step": 1975
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 5.4349493980407715,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.3098,
+      "step": 1980
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 5.487954139709473,
+      "learning_rate": 4.163159371492705e-05,
+      "loss": 0.2918,
+      "step": 1985
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 4.490516662597656,
+      "learning_rate": 4.159652076318743e-05,
+      "loss": 0.3514,
+      "step": 1990
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 3.850192070007324,
+      "learning_rate": 4.156144781144781e-05,
+      "loss": 0.2702,
+      "step": 1995
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 4.7108154296875,
+      "learning_rate": 4.15263748597082e-05,
+      "loss": 0.3058,
+      "step": 2000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 5.35396146774292,
+      "learning_rate": 4.1491301907968575e-05,
+      "loss": 0.3224,
+      "step": 2005
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 4.2288408279418945,
+      "learning_rate": 4.145622895622896e-05,
+      "loss": 0.3917,
+      "step": 2010
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 3.220329523086548,
+      "learning_rate": 4.1421156004489344e-05,
+      "loss": 0.3292,
+      "step": 2015
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 4.9558892250061035,
+      "learning_rate": 4.1386083052749716e-05,
+      "loss": 0.2801,
+      "step": 2020
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 5.528378486633301,
+      "learning_rate": 4.13510101010101e-05,
+      "loss": 0.3651,
+      "step": 2025
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 4.349493503570557,
+      "learning_rate": 4.1315937149270485e-05,
+      "loss": 0.3316,
+      "step": 2030
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 6.729834079742432,
+      "learning_rate": 4.128086419753087e-05,
+      "loss": 0.306,
+      "step": 2035
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 6.802405834197998,
+      "learning_rate": 4.124579124579125e-05,
+      "loss": 0.3359,
+      "step": 2040
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 5.517313003540039,
+      "learning_rate": 4.121071829405163e-05,
+      "loss": 0.4124,
+      "step": 2045
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 5.775959491729736,
+      "learning_rate": 4.117564534231202e-05,
+      "loss": 0.2944,
+      "step": 2050
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 4.595028400421143,
+      "learning_rate": 4.114057239057239e-05,
+      "loss": 0.3424,
+      "step": 2055
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 4.881396770477295,
+      "learning_rate": 4.110549943883277e-05,
+      "loss": 0.3831,
+      "step": 2060
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 5.669928550720215,
+      "learning_rate": 4.107042648709316e-05,
+      "loss": 0.3419,
+      "step": 2065
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 8.649759292602539,
+      "learning_rate": 4.1035353535353535e-05,
+      "loss": 0.334,
+      "step": 2070
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 4.799255847930908,
+      "learning_rate": 4.100028058361392e-05,
+      "loss": 0.3617,
+      "step": 2075
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 4.407303810119629,
+      "learning_rate": 4.0965207631874305e-05,
+      "loss": 0.3098,
+      "step": 2080
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.6626803874969482,
+      "learning_rate": 4.093013468013468e-05,
+      "loss": 0.3043,
+      "step": 2085
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 3.779773473739624,
+      "learning_rate": 4.089506172839506e-05,
+      "loss": 0.2603,
+      "step": 2090
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 5.3514580726623535,
+      "learning_rate": 4.0859988776655445e-05,
+      "loss": 0.3647,
+      "step": 2095
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 3.579667568206787,
+      "learning_rate": 4.082491582491583e-05,
+      "loss": 0.3261,
+      "step": 2100
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 4.847380638122559,
+      "learning_rate": 4.078984287317621e-05,
+      "loss": 0.3698,
+      "step": 2105
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 5.92960786819458,
+      "learning_rate": 4.075476992143659e-05,
+      "loss": 0.3651,
+      "step": 2110
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 5.301548004150391,
+      "learning_rate": 4.071969696969698e-05,
+      "loss": 0.304,
+      "step": 2115
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 4.4866414070129395,
+      "learning_rate": 4.068462401795735e-05,
+      "loss": 0.3712,
+      "step": 2120
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 5.734927654266357,
+      "learning_rate": 4.064955106621773e-05,
+      "loss": 0.3436,
+      "step": 2125
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 4.9071784019470215,
+      "learning_rate": 4.061447811447812e-05,
+      "loss": 0.3203,
+      "step": 2130
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 4.999136924743652,
+      "learning_rate": 4.0579405162738496e-05,
+      "loss": 0.3208,
+      "step": 2135
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 6.468421459197998,
+      "learning_rate": 4.054433221099888e-05,
+      "loss": 0.3767,
+      "step": 2140
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 4.205317497253418,
+      "learning_rate": 4.0509259259259265e-05,
+      "loss": 0.3008,
+      "step": 2145
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 4.611363410949707,
+      "learning_rate": 4.047418630751964e-05,
+      "loss": 0.337,
+      "step": 2150
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 6.029399871826172,
+      "learning_rate": 4.043911335578002e-05,
+      "loss": 0.4424,
+      "step": 2155
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 5.025997638702393,
+      "learning_rate": 4.0404040404040405e-05,
+      "loss": 0.3843,
+      "step": 2160
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 3.0415127277374268,
+      "learning_rate": 4.036896745230079e-05,
+      "loss": 0.2424,
+      "step": 2165
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 4.070464134216309,
+      "learning_rate": 4.033389450056117e-05,
+      "loss": 0.2705,
+      "step": 2170
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 4.6594157218933105,
+      "learning_rate": 4.029882154882155e-05,
+      "loss": 0.3259,
+      "step": 2175
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 4.543966770172119,
+      "learning_rate": 4.026374859708194e-05,
+      "loss": 0.3605,
+      "step": 2180
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.8322956562042236,
+      "learning_rate": 4.0228675645342315e-05,
+      "loss": 0.3231,
+      "step": 2185
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 4.966674327850342,
+      "learning_rate": 4.019360269360269e-05,
+      "loss": 0.3303,
+      "step": 2190
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 5.007614612579346,
+      "learning_rate": 4.015852974186308e-05,
+      "loss": 0.3462,
+      "step": 2195
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 5.209838390350342,
+      "learning_rate": 4.012345679012346e-05,
+      "loss": 0.3329,
+      "step": 2200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 3.9229366779327393,
+      "learning_rate": 4.008838383838384e-05,
+      "loss": 0.3379,
+      "step": 2205
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 4.192971229553223,
+      "learning_rate": 4.0053310886644225e-05,
+      "loss": 0.3328,
+      "step": 2210
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 3.172508478164673,
+      "learning_rate": 4.00182379349046e-05,
+      "loss": 0.3011,
+      "step": 2215
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 5.599798679351807,
+      "learning_rate": 3.998316498316498e-05,
+      "loss": 0.3056,
+      "step": 2220
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 4.3995161056518555,
+      "learning_rate": 3.9948092031425366e-05,
+      "loss": 0.2975,
+      "step": 2225
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 3.838740110397339,
+      "learning_rate": 3.991301907968575e-05,
+      "loss": 0.3177,
+      "step": 2230
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 5.497195720672607,
+      "learning_rate": 3.987794612794613e-05,
+      "loss": 0.3512,
+      "step": 2235
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 3.737442970275879,
+      "learning_rate": 3.984287317620651e-05,
+      "loss": 0.333,
+      "step": 2240
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 5.394541263580322,
+      "learning_rate": 3.98078002244669e-05,
+      "loss": 0.3512,
+      "step": 2245
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 4.133081436157227,
+      "learning_rate": 3.9772727272727275e-05,
+      "loss": 0.3893,
+      "step": 2250
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 3.7995636463165283,
+      "learning_rate": 3.973765432098765e-05,
+      "loss": 0.3247,
+      "step": 2255
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 4.582999229431152,
+      "learning_rate": 3.970258136924804e-05,
+      "loss": 0.3163,
+      "step": 2260
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 4.94778299331665,
+      "learning_rate": 3.966750841750842e-05,
+      "loss": 0.3005,
+      "step": 2265
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 4.156577110290527,
+      "learning_rate": 3.96324354657688e-05,
+      "loss": 0.3287,
+      "step": 2270
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 4.105434894561768,
+      "learning_rate": 3.9597362514029185e-05,
+      "loss": 0.3999,
+      "step": 2275
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 4.0938520431518555,
+      "learning_rate": 3.956228956228956e-05,
+      "loss": 0.323,
+      "step": 2280
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 3.5186057090759277,
+      "learning_rate": 3.952721661054994e-05,
+      "loss": 0.3274,
+      "step": 2285
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 4.522500991821289,
+      "learning_rate": 3.9492143658810326e-05,
+      "loss": 0.3259,
+      "step": 2290
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 4.503625869750977,
+      "learning_rate": 3.945707070707071e-05,
+      "loss": 0.339,
+      "step": 2295
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 4.543713569641113,
+      "learning_rate": 3.942199775533109e-05,
+      "loss": 0.3772,
+      "step": 2300
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 4.134209632873535,
+      "learning_rate": 3.938692480359147e-05,
+      "loss": 0.3633,
+      "step": 2305
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 3.960190773010254,
+      "learning_rate": 3.935185185185186e-05,
+      "loss": 0.3175,
+      "step": 2310
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 5.128618240356445,
+      "learning_rate": 3.9316778900112236e-05,
+      "loss": 0.3517,
+      "step": 2315
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 4.038329124450684,
+      "learning_rate": 3.9281705948372613e-05,
+      "loss": 0.3352,
+      "step": 2320
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 6.29808235168457,
+      "learning_rate": 3.9246632996633e-05,
+      "loss": 0.3404,
+      "step": 2325
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 4.927892208099365,
+      "learning_rate": 3.921156004489338e-05,
+      "loss": 0.3254,
+      "step": 2330
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 4.7702956199646,
+      "learning_rate": 3.917648709315376e-05,
+      "loss": 0.3489,
+      "step": 2335
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 5.17894983291626,
+      "learning_rate": 3.9141414141414145e-05,
+      "loss": 0.316,
+      "step": 2340
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 3.4703924655914307,
+      "learning_rate": 3.910634118967452e-05,
+      "loss": 0.2962,
+      "step": 2345
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 3.9641339778900146,
+      "learning_rate": 3.907126823793491e-05,
+      "loss": 0.3585,
+      "step": 2350
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 4.66275691986084,
+      "learning_rate": 3.9036195286195286e-05,
+      "loss": 0.3098,
+      "step": 2355
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 5.101861953735352,
+      "learning_rate": 3.900112233445567e-05,
+      "loss": 0.3469,
+      "step": 2360
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 4.622920036315918,
+      "learning_rate": 3.8966049382716055e-05,
+      "loss": 0.2948,
+      "step": 2365
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 4.948495864868164,
+      "learning_rate": 3.893097643097643e-05,
+      "loss": 0.3751,
+      "step": 2370
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 2.8351807594299316,
+      "learning_rate": 3.889590347923682e-05,
+      "loss": 0.3153,
+      "step": 2375
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9269548238217804,
+      "eval_loss": 0.21232399344444275,
+      "eval_runtime": 30.35,
+      "eval_samples_per_second": 371.236,
+      "eval_steps_per_second": 11.631,
+      "step": 2376
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 7.791706562042236,
+      "learning_rate": 3.8860830527497196e-05,
+      "loss": 0.2925,
+      "step": 2380
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 5.178705215454102,
+      "learning_rate": 3.8825757575757574e-05,
+      "loss": 0.2774,
+      "step": 2385
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 3.320344924926758,
+      "learning_rate": 3.879068462401796e-05,
+      "loss": 0.312,
+      "step": 2390
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 6.275354385375977,
+      "learning_rate": 3.875561167227834e-05,
+      "loss": 0.2897,
+      "step": 2395
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 5.21083402633667,
+      "learning_rate": 3.872053872053872e-05,
+      "loss": 0.2648,
+      "step": 2400
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 5.03237771987915,
+      "learning_rate": 3.8685465768799106e-05,
+      "loss": 0.3339,
+      "step": 2405
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 4.449357509613037,
+      "learning_rate": 3.8650392817059484e-05,
+      "loss": 0.2813,
+      "step": 2410
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 4.017120361328125,
+      "learning_rate": 3.861531986531987e-05,
+      "loss": 0.3165,
+      "step": 2415
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 4.37885856628418,
+      "learning_rate": 3.8580246913580246e-05,
+      "loss": 0.2724,
+      "step": 2420
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 4.221261978149414,
+      "learning_rate": 3.854517396184063e-05,
+      "loss": 0.2831,
+      "step": 2425
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 4.640631675720215,
+      "learning_rate": 3.8510101010101015e-05,
+      "loss": 0.3903,
+      "step": 2430
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 5.1502790451049805,
+      "learning_rate": 3.8475028058361393e-05,
+      "loss": 0.2986,
+      "step": 2435
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 5.26745080947876,
+      "learning_rate": 3.843995510662178e-05,
+      "loss": 0.3286,
+      "step": 2440
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 5.3141608238220215,
+      "learning_rate": 3.8404882154882156e-05,
+      "loss": 0.3251,
+      "step": 2445
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 4.887467384338379,
+      "learning_rate": 3.8369809203142534e-05,
+      "loss": 0.3233,
+      "step": 2450
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 5.6849870681762695,
+      "learning_rate": 3.833473625140292e-05,
+      "loss": 0.2832,
+      "step": 2455
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 6.014511585235596,
+      "learning_rate": 3.82996632996633e-05,
+      "loss": 0.2874,
+      "step": 2460
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 3.2397172451019287,
+      "learning_rate": 3.826459034792368e-05,
+      "loss": 0.262,
+      "step": 2465
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 3.949284315109253,
+      "learning_rate": 3.8229517396184066e-05,
+      "loss": 0.2735,
+      "step": 2470
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 4.78221321105957,
+      "learning_rate": 3.8194444444444444e-05,
+      "loss": 0.3558,
+      "step": 2475
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 3.865053415298462,
+      "learning_rate": 3.815937149270483e-05,
+      "loss": 0.2894,
+      "step": 2480
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 4.015622615814209,
+      "learning_rate": 3.8124298540965206e-05,
+      "loss": 0.2962,
+      "step": 2485
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 4.832337856292725,
+      "learning_rate": 3.808922558922559e-05,
+      "loss": 0.306,
+      "step": 2490
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 4.834969520568848,
+      "learning_rate": 3.8054152637485976e-05,
+      "loss": 0.3422,
+      "step": 2495
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 4.97349739074707,
+      "learning_rate": 3.8019079685746354e-05,
+      "loss": 0.3503,
+      "step": 2500
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 2.8850040435791016,
+      "learning_rate": 3.798400673400674e-05,
+      "loss": 0.313,
+      "step": 2505
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 2.962406635284424,
+      "learning_rate": 3.7948933782267116e-05,
+      "loss": 0.3092,
+      "step": 2510
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 4.120716094970703,
+      "learning_rate": 3.79138608305275e-05,
+      "loss": 0.3108,
+      "step": 2515
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 4.813681125640869,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 0.3014,
+      "step": 2520
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 3.768829107284546,
+      "learning_rate": 3.7843714927048263e-05,
+      "loss": 0.2768,
+      "step": 2525
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 7.496875286102295,
+      "learning_rate": 3.780864197530865e-05,
+      "loss": 0.308,
+      "step": 2530
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 4.922712326049805,
+      "learning_rate": 3.7773569023569026e-05,
+      "loss": 0.3271,
+      "step": 2535
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 4.2658257484436035,
+      "learning_rate": 3.7738496071829404e-05,
+      "loss": 0.3013,
+      "step": 2540
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 4.059297561645508,
+      "learning_rate": 3.770342312008979e-05,
+      "loss": 0.3059,
+      "step": 2545
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 3.8374571800231934,
+      "learning_rate": 3.7668350168350167e-05,
+      "loss": 0.2321,
+      "step": 2550
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 5.786901950836182,
+      "learning_rate": 3.763327721661055e-05,
+      "loss": 0.3373,
+      "step": 2555
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 4.773471832275391,
+      "learning_rate": 3.7598204264870936e-05,
+      "loss": 0.3583,
+      "step": 2560
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 6.408568382263184,
+      "learning_rate": 3.7563131313131314e-05,
+      "loss": 0.2985,
+      "step": 2565
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 4.450567722320557,
+      "learning_rate": 3.75280583613917e-05,
+      "loss": 0.2836,
+      "step": 2570
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 4.5761518478393555,
+      "learning_rate": 3.7492985409652076e-05,
+      "loss": 0.3189,
+      "step": 2575
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 3.900135040283203,
+      "learning_rate": 3.745791245791246e-05,
+      "loss": 0.2739,
+      "step": 2580
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 6.129603862762451,
+      "learning_rate": 3.742283950617284e-05,
+      "loss": 0.2963,
+      "step": 2585
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 4.674997329711914,
+      "learning_rate": 3.7387766554433224e-05,
+      "loss": 0.2901,
+      "step": 2590
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 5.454425811767578,
+      "learning_rate": 3.735269360269361e-05,
+      "loss": 0.3981,
+      "step": 2595
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 5.348542213439941,
+      "learning_rate": 3.7317620650953986e-05,
+      "loss": 0.3473,
+      "step": 2600
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 4.9225993156433105,
+      "learning_rate": 3.7282547699214364e-05,
+      "loss": 0.3197,
+      "step": 2605
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 3.92462420463562,
+      "learning_rate": 3.724747474747475e-05,
+      "loss": 0.3024,
+      "step": 2610
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 4.231691360473633,
+      "learning_rate": 3.721240179573513e-05,
+      "loss": 0.3212,
+      "step": 2615
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 5.571329116821289,
+      "learning_rate": 3.717732884399551e-05,
+      "loss": 0.3108,
+      "step": 2620
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 5.001312732696533,
+      "learning_rate": 3.7142255892255896e-05,
+      "loss": 0.2861,
+      "step": 2625
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 3.9836950302124023,
+      "learning_rate": 3.710718294051628e-05,
+      "loss": 0.2966,
+      "step": 2630
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 4.538575649261475,
+      "learning_rate": 3.707210998877666e-05,
+      "loss": 0.3082,
+      "step": 2635
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 4.797281742095947,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.3267,
+      "step": 2640
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 8.28226089477539,
+      "learning_rate": 3.700196408529742e-05,
+      "loss": 0.2942,
+      "step": 2645
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 5.338452339172363,
+      "learning_rate": 3.69668911335578e-05,
+      "loss": 0.3044,
+      "step": 2650
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 4.229073524475098,
+      "learning_rate": 3.6931818181818184e-05,
+      "loss": 0.2909,
+      "step": 2655
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 4.318957328796387,
+      "learning_rate": 3.689674523007857e-05,
+      "loss": 0.3422,
+      "step": 2660
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 3.900193452835083,
+      "learning_rate": 3.6861672278338946e-05,
+      "loss": 0.2796,
+      "step": 2665
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 5.6975016593933105,
+      "learning_rate": 3.6826599326599324e-05,
+      "loss": 0.2995,
+      "step": 2670
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 4.920116901397705,
+      "learning_rate": 3.679152637485971e-05,
+      "loss": 0.2879,
+      "step": 2675
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 5.000295162200928,
+      "learning_rate": 3.6756453423120094e-05,
+      "loss": 0.32,
+      "step": 2680
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 5.2279744148254395,
+      "learning_rate": 3.672138047138047e-05,
+      "loss": 0.2981,
+      "step": 2685
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 3.1572046279907227,
+      "learning_rate": 3.6686307519640856e-05,
+      "loss": 0.293,
+      "step": 2690
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 5.613211631774902,
+      "learning_rate": 3.665123456790124e-05,
+      "loss": 0.3369,
+      "step": 2695
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 4.345326900482178,
+      "learning_rate": 3.661616161616162e-05,
+      "loss": 0.2925,
+      "step": 2700
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 4.713896751403809,
+      "learning_rate": 3.6581088664422e-05,
+      "loss": 0.3267,
+      "step": 2705
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 4.135952472686768,
+      "learning_rate": 3.654601571268238e-05,
+      "loss": 0.3134,
+      "step": 2710
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 5.108008861541748,
+      "learning_rate": 3.651094276094276e-05,
+      "loss": 0.306,
+      "step": 2715
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 3.937014102935791,
+      "learning_rate": 3.6475869809203144e-05,
+      "loss": 0.383,
+      "step": 2720
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 5.377154350280762,
+      "learning_rate": 3.644079685746353e-05,
+      "loss": 0.2877,
+      "step": 2725
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 4.1062397956848145,
+      "learning_rate": 3.6405723905723907e-05,
+      "loss": 0.297,
+      "step": 2730
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 4.515702724456787,
+      "learning_rate": 3.6370650953984285e-05,
+      "loss": 0.3075,
+      "step": 2735
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 4.550343990325928,
+      "learning_rate": 3.633557800224467e-05,
+      "loss": 0.3604,
+      "step": 2740
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 3.844510793685913,
+      "learning_rate": 3.6300505050505054e-05,
+      "loss": 0.2656,
+      "step": 2745
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 6.030447483062744,
+      "learning_rate": 3.626543209876543e-05,
+      "loss": 0.3203,
+      "step": 2750
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 5.013733863830566,
+      "learning_rate": 3.6230359147025816e-05,
+      "loss": 0.3293,
+      "step": 2755
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 4.251259803771973,
+      "learning_rate": 3.61952861952862e-05,
+      "loss": 0.2754,
+      "step": 2760
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 4.327945709228516,
+      "learning_rate": 3.616021324354658e-05,
+      "loss": 0.3405,
+      "step": 2765
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 3.967000722885132,
+      "learning_rate": 3.612514029180696e-05,
+      "loss": 0.3288,
+      "step": 2770
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 4.833152770996094,
+      "learning_rate": 3.609006734006734e-05,
+      "loss": 0.3181,
+      "step": 2775
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 3.159540891647339,
+      "learning_rate": 3.6054994388327726e-05,
+      "loss": 0.3068,
+      "step": 2780
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 5.1113786697387695,
+      "learning_rate": 3.6019921436588104e-05,
+      "loss": 0.2819,
+      "step": 2785
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 3.6158957481384277,
+      "learning_rate": 3.598484848484849e-05,
+      "loss": 0.2748,
+      "step": 2790
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 4.457354545593262,
+      "learning_rate": 3.5949775533108874e-05,
+      "loss": 0.3064,
+      "step": 2795
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 5.5556321144104,
+      "learning_rate": 3.5914702581369245e-05,
+      "loss": 0.347,
+      "step": 2800
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 5.094475269317627,
+      "learning_rate": 3.587962962962963e-05,
+      "loss": 0.33,
+      "step": 2805
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 4.055764675140381,
+      "learning_rate": 3.5844556677890014e-05,
+      "loss": 0.2988,
+      "step": 2810
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 2.9098262786865234,
+      "learning_rate": 3.580948372615039e-05,
+      "loss": 0.3241,
+      "step": 2815
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 5.585843086242676,
+      "learning_rate": 3.577441077441078e-05,
+      "loss": 0.3274,
+      "step": 2820
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 4.769870758056641,
+      "learning_rate": 3.573933782267116e-05,
+      "loss": 0.2939,
+      "step": 2825
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 4.0134806632995605,
+      "learning_rate": 3.570426487093154e-05,
+      "loss": 0.2857,
+      "step": 2830
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 4.413273334503174,
+      "learning_rate": 3.566919191919192e-05,
+      "loss": 0.2621,
+      "step": 2835
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 4.19149923324585,
+      "learning_rate": 3.56341189674523e-05,
+      "loss": 0.3622,
+      "step": 2840
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 3.729079484939575,
+      "learning_rate": 3.5599046015712686e-05,
+      "loss": 0.2431,
+      "step": 2845
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 3.8650453090667725,
+      "learning_rate": 3.5563973063973064e-05,
+      "loss": 0.3272,
+      "step": 2850
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 3.7103049755096436,
+      "learning_rate": 3.552890011223345e-05,
+      "loss": 0.2923,
+      "step": 2855
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 2.58754301071167,
+      "learning_rate": 3.5493827160493834e-05,
+      "loss": 0.3168,
+      "step": 2860
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 3.4648077487945557,
+      "learning_rate": 3.5458754208754205e-05,
+      "loss": 0.3168,
+      "step": 2865
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 3.0660760402679443,
+      "learning_rate": 3.542368125701459e-05,
+      "loss": 0.2783,
+      "step": 2870
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 2.9443509578704834,
+      "learning_rate": 3.5388608305274974e-05,
+      "loss": 0.2383,
+      "step": 2875
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 4.880836486816406,
+      "learning_rate": 3.535353535353535e-05,
+      "loss": 0.2943,
+      "step": 2880
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 4.978140354156494,
+      "learning_rate": 3.531846240179574e-05,
+      "loss": 0.3028,
+      "step": 2885
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 5.634300231933594,
+      "learning_rate": 3.528338945005612e-05,
+      "loss": 0.2654,
+      "step": 2890
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 3.940747022628784,
+      "learning_rate": 3.52483164983165e-05,
+      "loss": 0.2823,
+      "step": 2895
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 6.629083633422852,
+      "learning_rate": 3.521324354657688e-05,
+      "loss": 0.3178,
+      "step": 2900
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 3.8521242141723633,
+      "learning_rate": 3.517817059483726e-05,
+      "loss": 0.3095,
+      "step": 2905
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 4.260024070739746,
+      "learning_rate": 3.514309764309765e-05,
+      "loss": 0.3074,
+      "step": 2910
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 4.263473033905029,
+      "learning_rate": 3.5108024691358025e-05,
+      "loss": 0.3457,
+      "step": 2915
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 3.278444766998291,
+      "learning_rate": 3.507295173961841e-05,
+      "loss": 0.2981,
+      "step": 2920
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 5.176063537597656,
+      "learning_rate": 3.5037878787878794e-05,
+      "loss": 0.2897,
+      "step": 2925
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 4.075413703918457,
+      "learning_rate": 3.500280583613917e-05,
+      "loss": 0.3,
+      "step": 2930
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 4.786322116851807,
+      "learning_rate": 3.496773288439955e-05,
+      "loss": 0.3551,
+      "step": 2935
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 4.71600341796875,
+      "learning_rate": 3.4932659932659934e-05,
+      "loss": 0.3114,
+      "step": 2940
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 3.7457523345947266,
+      "learning_rate": 3.489758698092032e-05,
+      "loss": 0.3826,
+      "step": 2945
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 4.5677266120910645,
+      "learning_rate": 3.48625140291807e-05,
+      "loss": 0.2986,
+      "step": 2950
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 4.330593585968018,
+      "learning_rate": 3.482744107744108e-05,
+      "loss": 0.3378,
+      "step": 2955
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 3.9780690670013428,
+      "learning_rate": 3.4792368125701466e-05,
+      "loss": 0.2854,
+      "step": 2960
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 3.9562747478485107,
+      "learning_rate": 3.475729517396184e-05,
+      "loss": 0.2592,
+      "step": 2965
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 3.7066001892089844,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.2898,
+      "step": 2970
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 3.942182779312134,
+      "learning_rate": 3.468714927048261e-05,
+      "loss": 0.2752,
+      "step": 2975
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 4.99793815612793,
+      "learning_rate": 3.4652076318742985e-05,
+      "loss": 0.3511,
+      "step": 2980
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 3.5230560302734375,
+      "learning_rate": 3.461700336700337e-05,
+      "loss": 0.333,
+      "step": 2985
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 3.2714133262634277,
+      "learning_rate": 3.4581930415263754e-05,
+      "loss": 0.2815,
+      "step": 2990
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 3.4161295890808105,
+      "learning_rate": 3.454685746352413e-05,
+      "loss": 0.2857,
+      "step": 2995
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 4.5021281242370605,
+      "learning_rate": 3.451178451178451e-05,
+      "loss": 0.2956,
+      "step": 3000
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 3.383610486984253,
+      "learning_rate": 3.4476711560044895e-05,
+      "loss": 0.2898,
+      "step": 3005
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 4.295190334320068,
+      "learning_rate": 3.444163860830528e-05,
+      "loss": 0.3056,
+      "step": 3010
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 4.418600559234619,
+      "learning_rate": 3.440656565656566e-05,
+      "loss": 0.3353,
+      "step": 3015
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 3.2519736289978027,
+      "learning_rate": 3.437149270482604e-05,
+      "loss": 0.2821,
+      "step": 3020
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 4.863473415374756,
+      "learning_rate": 3.4336419753086427e-05,
+      "loss": 0.2791,
+      "step": 3025
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 5.4529571533203125,
+      "learning_rate": 3.43013468013468e-05,
+      "loss": 0.284,
+      "step": 3030
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 3.654813289642334,
+      "learning_rate": 3.426627384960718e-05,
+      "loss": 0.3713,
+      "step": 3035
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 3.9698240756988525,
+      "learning_rate": 3.423120089786757e-05,
+      "loss": 0.3097,
+      "step": 3040
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 5.582406997680664,
+      "learning_rate": 3.4196127946127945e-05,
+      "loss": 0.3094,
+      "step": 3045
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 3.9481139183044434,
+      "learning_rate": 3.416105499438833e-05,
+      "loss": 0.3442,
+      "step": 3050
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 4.35070276260376,
+      "learning_rate": 3.4125982042648714e-05,
+      "loss": 0.3439,
+      "step": 3055
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 3.3845298290252686,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 0.2651,
+      "step": 3060
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 5.312651634216309,
+      "learning_rate": 3.405583613916947e-05,
+      "loss": 0.3117,
+      "step": 3065
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 5.3635687828063965,
+      "learning_rate": 3.4020763187429855e-05,
+      "loss": 0.3314,
+      "step": 3070
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 3.2859911918640137,
+      "learning_rate": 3.398569023569024e-05,
+      "loss": 0.3148,
+      "step": 3075
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 3.283454656600952,
+      "learning_rate": 3.395061728395062e-05,
+      "loss": 0.2948,
+      "step": 3080
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 3.952256441116333,
+      "learning_rate": 3.3915544332211e-05,
+      "loss": 0.3107,
+      "step": 3085
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 3.161120891571045,
+      "learning_rate": 3.388047138047139e-05,
+      "loss": 0.2975,
+      "step": 3090
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 4.279951572418213,
+      "learning_rate": 3.3845398428731765e-05,
+      "loss": 0.3192,
+      "step": 3095
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 4.90150260925293,
+      "learning_rate": 3.381032547699214e-05,
+      "loss": 0.3295,
+      "step": 3100
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 3.666090488433838,
+      "learning_rate": 3.377525252525253e-05,
+      "loss": 0.3414,
+      "step": 3105
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 4.42512321472168,
+      "learning_rate": 3.374017957351291e-05,
+      "loss": 0.3293,
+      "step": 3110
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 3.514129161834717,
+      "learning_rate": 3.370510662177329e-05,
+      "loss": 0.3253,
+      "step": 3115
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 3.3121232986450195,
+      "learning_rate": 3.3670033670033675e-05,
+      "loss": 0.3109,
+      "step": 3120
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 3.626138925552368,
+      "learning_rate": 3.363496071829405e-05,
+      "loss": 0.3139,
+      "step": 3125
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 3.8030307292938232,
+      "learning_rate": 3.359988776655443e-05,
+      "loss": 0.325,
+      "step": 3130
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 3.4151670932769775,
+      "learning_rate": 3.3564814814814815e-05,
+      "loss": 0.3,
+      "step": 3135
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 4.360004901885986,
+      "learning_rate": 3.35297418630752e-05,
+      "loss": 0.3233,
+      "step": 3140
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 4.586578369140625,
+      "learning_rate": 3.349466891133558e-05,
+      "loss": 0.3339,
+      "step": 3145
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 3.7512130737304688,
+      "learning_rate": 3.345959595959596e-05,
+      "loss": 0.3004,
+      "step": 3150
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 5.146670818328857,
+      "learning_rate": 3.342452300785635e-05,
+      "loss": 0.2649,
+      "step": 3155
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 4.754834175109863,
+      "learning_rate": 3.3389450056116725e-05,
+      "loss": 0.2952,
+      "step": 3160
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 5.98429012298584,
+      "learning_rate": 3.33543771043771e-05,
+      "loss": 0.3438,
+      "step": 3165
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9289074287742967,
+      "eval_loss": 0.20633326470851898,
+      "eval_runtime": 30.6786,
+      "eval_samples_per_second": 367.26,
+      "eval_steps_per_second": 11.506,
+      "step": 3169
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 6.001084327697754,
+      "learning_rate": 3.331930415263749e-05,
+      "loss": 0.2967,
+      "step": 3170
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 4.173760890960693,
+      "learning_rate": 3.328423120089787e-05,
+      "loss": 0.2739,
+      "step": 3175
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 4.663118362426758,
+      "learning_rate": 3.324915824915825e-05,
+      "loss": 0.3153,
+      "step": 3180
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 4.8701653480529785,
+      "learning_rate": 3.3214085297418635e-05,
+      "loss": 0.3125,
+      "step": 3185
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 4.151704788208008,
+      "learning_rate": 3.317901234567901e-05,
+      "loss": 0.2958,
+      "step": 3190
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 3.546596050262451,
+      "learning_rate": 3.314393939393939e-05,
+      "loss": 0.2878,
+      "step": 3195
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 3.466984272003174,
+      "learning_rate": 3.3108866442199775e-05,
+      "loss": 0.3219,
+      "step": 3200
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 4.747018814086914,
+      "learning_rate": 3.307379349046016e-05,
+      "loss": 0.2471,
+      "step": 3205
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 4.214364528656006,
+      "learning_rate": 3.3038720538720545e-05,
+      "loss": 0.2699,
+      "step": 3210
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 5.498867988586426,
+      "learning_rate": 3.300364758698092e-05,
+      "loss": 0.3006,
+      "step": 3215
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 3.139638662338257,
+      "learning_rate": 3.296857463524131e-05,
+      "loss": 0.2823,
+      "step": 3220
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 4.400967597961426,
+      "learning_rate": 3.2933501683501685e-05,
+      "loss": 0.314,
+      "step": 3225
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 3.398244619369507,
+      "learning_rate": 3.289842873176206e-05,
+      "loss": 0.2826,
+      "step": 3230
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 4.529625415802002,
+      "learning_rate": 3.286335578002245e-05,
+      "loss": 0.2466,
+      "step": 3235
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 5.697416305541992,
+      "learning_rate": 3.282828282828283e-05,
+      "loss": 0.3115,
+      "step": 3240
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 4.382936954498291,
+      "learning_rate": 3.279320987654321e-05,
+      "loss": 0.2974,
+      "step": 3245
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 4.81410551071167,
+      "learning_rate": 3.2758136924803595e-05,
+      "loss": 0.2962,
+      "step": 3250
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 3.9076387882232666,
+      "learning_rate": 3.272306397306397e-05,
+      "loss": 0.2288,
+      "step": 3255
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 4.519957065582275,
+      "learning_rate": 3.268799102132436e-05,
+      "loss": 0.2485,
+      "step": 3260
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 3.5229649543762207,
+      "learning_rate": 3.2652918069584735e-05,
+      "loss": 0.3127,
+      "step": 3265
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 3.927300453186035,
+      "learning_rate": 3.261784511784512e-05,
+      "loss": 0.3054,
+      "step": 3270
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 3.727557897567749,
+      "learning_rate": 3.2582772166105505e-05,
+      "loss": 0.2959,
+      "step": 3275
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 4.2276411056518555,
+      "learning_rate": 3.254769921436588e-05,
+      "loss": 0.2484,
+      "step": 3280
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 6.399277687072754,
+      "learning_rate": 3.251262626262627e-05,
+      "loss": 0.3026,
+      "step": 3285
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 3.5500333309173584,
+      "learning_rate": 3.2477553310886645e-05,
+      "loss": 0.2635,
+      "step": 3290
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 4.322241306304932,
+      "learning_rate": 3.244248035914702e-05,
+      "loss": 0.2649,
+      "step": 3295
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 4.4791646003723145,
+      "learning_rate": 3.240740740740741e-05,
+      "loss": 0.2709,
+      "step": 3300
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 3.112387180328369,
+      "learning_rate": 3.237233445566779e-05,
+      "loss": 0.3416,
+      "step": 3305
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 5.389982223510742,
+      "learning_rate": 3.233726150392817e-05,
+      "loss": 0.3331,
+      "step": 3310
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 2.948604106903076,
+      "learning_rate": 3.2302188552188555e-05,
+      "loss": 0.2616,
+      "step": 3315
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 7.787486553192139,
+      "learning_rate": 3.226711560044893e-05,
+      "loss": 0.2587,
+      "step": 3320
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 3.437147855758667,
+      "learning_rate": 3.223204264870932e-05,
+      "loss": 0.3404,
+      "step": 3325
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 4.956410884857178,
+      "learning_rate": 3.2196969696969696e-05,
+      "loss": 0.2494,
+      "step": 3330
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 4.321473598480225,
+      "learning_rate": 3.216189674523008e-05,
+      "loss": 0.2784,
+      "step": 3335
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 3.953444719314575,
+      "learning_rate": 3.2126823793490465e-05,
+      "loss": 0.2725,
+      "step": 3340
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 4.455861568450928,
+      "learning_rate": 3.209175084175084e-05,
+      "loss": 0.2897,
+      "step": 3345
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 3.8986704349517822,
+      "learning_rate": 3.205667789001123e-05,
+      "loss": 0.2698,
+      "step": 3350
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 7.029972553253174,
+      "learning_rate": 3.2021604938271605e-05,
+      "loss": 0.2804,
+      "step": 3355
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 3.9815454483032227,
+      "learning_rate": 3.198653198653199e-05,
+      "loss": 0.2517,
+      "step": 3360
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 3.934307098388672,
+      "learning_rate": 3.195145903479237e-05,
+      "loss": 0.2983,
+      "step": 3365
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 4.12862491607666,
+      "learning_rate": 3.191638608305275e-05,
+      "loss": 0.3046,
+      "step": 3370
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 3.9412386417388916,
+      "learning_rate": 3.188131313131314e-05,
+      "loss": 0.2791,
+      "step": 3375
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 4.196868896484375,
+      "learning_rate": 3.1846240179573515e-05,
+      "loss": 0.2794,
+      "step": 3380
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 3.9001171588897705,
+      "learning_rate": 3.181116722783389e-05,
+      "loss": 0.2795,
+      "step": 3385
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 5.460897445678711,
+      "learning_rate": 3.177609427609428e-05,
+      "loss": 0.2511,
+      "step": 3390
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 3.990734100341797,
+      "learning_rate": 3.1741021324354656e-05,
+      "loss": 0.2945,
+      "step": 3395
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 4.729461193084717,
+      "learning_rate": 3.170594837261504e-05,
+      "loss": 0.2503,
+      "step": 3400
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 3.912752628326416,
+      "learning_rate": 3.1670875420875425e-05,
+      "loss": 0.2746,
+      "step": 3405
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 3.8767850399017334,
+      "learning_rate": 3.16358024691358e-05,
+      "loss": 0.2945,
+      "step": 3410
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 8.053336143493652,
+      "learning_rate": 3.160072951739619e-05,
+      "loss": 0.2717,
+      "step": 3415
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 5.382170677185059,
+      "learning_rate": 3.1565656565656566e-05,
+      "loss": 0.3345,
+      "step": 3420
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 4.81245231628418,
+      "learning_rate": 3.153058361391695e-05,
+      "loss": 0.2573,
+      "step": 3425
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 3.6458661556243896,
+      "learning_rate": 3.149551066217733e-05,
+      "loss": 0.2949,
+      "step": 3430
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 4.600330829620361,
+      "learning_rate": 3.146043771043771e-05,
+      "loss": 0.3283,
+      "step": 3435
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 5.574528217315674,
+      "learning_rate": 3.14253647586981e-05,
+      "loss": 0.3561,
+      "step": 3440
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 3.4262194633483887,
+      "learning_rate": 3.1390291806958475e-05,
+      "loss": 0.2758,
+      "step": 3445
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 4.168435573577881,
+      "learning_rate": 3.135521885521885e-05,
+      "loss": 0.3236,
+      "step": 3450
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 3.577876567840576,
+      "learning_rate": 3.132014590347924e-05,
+      "loss": 0.3166,
+      "step": 3455
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 4.944897651672363,
+      "learning_rate": 3.1285072951739616e-05,
+      "loss": 0.2965,
+      "step": 3460
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 5.118645191192627,
+      "learning_rate": 3.125e-05,
+      "loss": 0.3527,
+      "step": 3465
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 4.388352394104004,
+      "learning_rate": 3.1214927048260385e-05,
+      "loss": 0.315,
+      "step": 3470
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 3.719374895095825,
+      "learning_rate": 3.117985409652076e-05,
+      "loss": 0.3096,
+      "step": 3475
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 3.322565793991089,
+      "learning_rate": 3.114478114478115e-05,
+      "loss": 0.2812,
+      "step": 3480
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 7.674746036529541,
+      "learning_rate": 3.1109708193041526e-05,
+      "loss": 0.2881,
+      "step": 3485
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 3.8031699657440186,
+      "learning_rate": 3.107463524130191e-05,
+      "loss": 0.3029,
+      "step": 3490
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 4.542752265930176,
+      "learning_rate": 3.103956228956229e-05,
+      "loss": 0.3444,
+      "step": 3495
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 3.4702298641204834,
+      "learning_rate": 3.100448933782267e-05,
+      "loss": 0.2512,
+      "step": 3500
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 4.33992862701416,
+      "learning_rate": 3.096941638608306e-05,
+      "loss": 0.2764,
+      "step": 3505
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 4.737889766693115,
+      "learning_rate": 3.0934343434343436e-05,
+      "loss": 0.2661,
+      "step": 3510
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 4.014540195465088,
+      "learning_rate": 3.0899270482603814e-05,
+      "loss": 0.2963,
+      "step": 3515
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 5.9853739738464355,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 0.2476,
+      "step": 3520
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 5.3976922035217285,
+      "learning_rate": 3.082912457912458e-05,
+      "loss": 0.32,
+      "step": 3525
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 4.2126994132995605,
+      "learning_rate": 3.079405162738496e-05,
+      "loss": 0.2863,
+      "step": 3530
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 5.164499282836914,
+      "learning_rate": 3.0758978675645346e-05,
+      "loss": 0.2959,
+      "step": 3535
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 3.448996067047119,
+      "learning_rate": 3.072390572390573e-05,
+      "loss": 0.2765,
+      "step": 3540
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 4.592435359954834,
+      "learning_rate": 3.068883277216611e-05,
+      "loss": 0.3242,
+      "step": 3545
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 5.053036212921143,
+      "learning_rate": 3.0653759820426486e-05,
+      "loss": 0.3191,
+      "step": 3550
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 5.953499794006348,
+      "learning_rate": 3.061868686868687e-05,
+      "loss": 0.2928,
+      "step": 3555
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 3.3007402420043945,
+      "learning_rate": 3.058361391694725e-05,
+      "loss": 0.2843,
+      "step": 3560
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 5.7247538566589355,
+      "learning_rate": 3.054854096520763e-05,
+      "loss": 0.2888,
+      "step": 3565
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 4.04235315322876,
+      "learning_rate": 3.0513468013468015e-05,
+      "loss": 0.2843,
+      "step": 3570
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 4.666456699371338,
+      "learning_rate": 3.04783950617284e-05,
+      "loss": 0.2964,
+      "step": 3575
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 3.578707218170166,
+      "learning_rate": 3.0443322109988777e-05,
+      "loss": 0.301,
+      "step": 3580
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 4.544867038726807,
+      "learning_rate": 3.040824915824916e-05,
+      "loss": 0.294,
+      "step": 3585
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 4.963395595550537,
+      "learning_rate": 3.037317620650954e-05,
+      "loss": 0.3305,
+      "step": 3590
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 5.475175380706787,
+      "learning_rate": 3.0338103254769924e-05,
+      "loss": 0.3334,
+      "step": 3595
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 4.5566182136535645,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.2817,
+      "step": 3600
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 3.9185471534729004,
+      "learning_rate": 3.0267957351290687e-05,
+      "loss": 0.266,
+      "step": 3605
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 3.6879515647888184,
+      "learning_rate": 3.023288439955107e-05,
+      "loss": 0.2653,
+      "step": 3610
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 4.596480369567871,
+      "learning_rate": 3.0197811447811446e-05,
+      "loss": 0.2927,
+      "step": 3615
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 6.306631088256836,
+      "learning_rate": 3.016273849607183e-05,
+      "loss": 0.273,
+      "step": 3620
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 3.5089361667633057,
+      "learning_rate": 3.0127665544332212e-05,
+      "loss": 0.2919,
+      "step": 3625
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 2.887002468109131,
+      "learning_rate": 3.0092592592592593e-05,
+      "loss": 0.3032,
+      "step": 3630
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 4.098235130310059,
+      "learning_rate": 3.0057519640852978e-05,
+      "loss": 0.2942,
+      "step": 3635
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 3.0776865482330322,
+      "learning_rate": 3.002244668911336e-05,
+      "loss": 0.3071,
+      "step": 3640
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 4.976138591766357,
+      "learning_rate": 2.9987373737373737e-05,
+      "loss": 0.3216,
+      "step": 3645
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 5.015041828155518,
+      "learning_rate": 2.995230078563412e-05,
+      "loss": 0.3226,
+      "step": 3650
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 5.457238674163818,
+      "learning_rate": 2.99172278338945e-05,
+      "loss": 0.3281,
+      "step": 3655
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 7.7326154708862305,
+      "learning_rate": 2.9882154882154885e-05,
+      "loss": 0.2475,
+      "step": 3660
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 5.758991718292236,
+      "learning_rate": 2.9847081930415266e-05,
+      "loss": 0.2407,
+      "step": 3665
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 4.086374759674072,
+      "learning_rate": 2.9812008978675647e-05,
+      "loss": 0.2912,
+      "step": 3670
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 5.025475978851318,
+      "learning_rate": 2.9776936026936032e-05,
+      "loss": 0.2748,
+      "step": 3675
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 5.168757915496826,
+      "learning_rate": 2.9741863075196406e-05,
+      "loss": 0.289,
+      "step": 3680
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 4.065808296203613,
+      "learning_rate": 2.970679012345679e-05,
+      "loss": 0.2596,
+      "step": 3685
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 4.855968952178955,
+      "learning_rate": 2.9671717171717172e-05,
+      "loss": 0.2994,
+      "step": 3690
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 5.673643112182617,
+      "learning_rate": 2.9636644219977554e-05,
+      "loss": 0.3035,
+      "step": 3695
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 5.468099117279053,
+      "learning_rate": 2.960157126823794e-05,
+      "loss": 0.3282,
+      "step": 3700
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 4.7444539070129395,
+      "learning_rate": 2.956649831649832e-05,
+      "loss": 0.314,
+      "step": 3705
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 5.548402786254883,
+      "learning_rate": 2.9531425364758698e-05,
+      "loss": 0.3139,
+      "step": 3710
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 2.8443384170532227,
+      "learning_rate": 2.949635241301908e-05,
+      "loss": 0.2505,
+      "step": 3715
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 4.943212985992432,
+      "learning_rate": 2.946127946127946e-05,
+      "loss": 0.3039,
+      "step": 3720
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 2.9205162525177,
+      "learning_rate": 2.9426206509539845e-05,
+      "loss": 0.2505,
+      "step": 3725
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 4.9782819747924805,
+      "learning_rate": 2.9391133557800226e-05,
+      "loss": 0.2869,
+      "step": 3730
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 4.202998161315918,
+      "learning_rate": 2.935606060606061e-05,
+      "loss": 0.2964,
+      "step": 3735
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 2.9494664669036865,
+      "learning_rate": 2.9320987654320992e-05,
+      "loss": 0.2614,
+      "step": 3740
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 4.606307506561279,
+      "learning_rate": 2.928591470258137e-05,
+      "loss": 0.2911,
+      "step": 3745
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 3.4345922470092773,
+      "learning_rate": 2.925084175084175e-05,
+      "loss": 0.2904,
+      "step": 3750
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 5.4425225257873535,
+      "learning_rate": 2.9215768799102133e-05,
+      "loss": 0.236,
+      "step": 3755
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 4.596513748168945,
+      "learning_rate": 2.9180695847362517e-05,
+      "loss": 0.2712,
+      "step": 3760
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 4.600802421569824,
+      "learning_rate": 2.91456228956229e-05,
+      "loss": 0.2535,
+      "step": 3765
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 4.583037853240967,
+      "learning_rate": 2.911054994388328e-05,
+      "loss": 0.3521,
+      "step": 3770
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 4.230092525482178,
+      "learning_rate": 2.9075476992143658e-05,
+      "loss": 0.2912,
+      "step": 3775
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 5.071386814117432,
+      "learning_rate": 2.904040404040404e-05,
+      "loss": 0.3251,
+      "step": 3780
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 3.691915512084961,
+      "learning_rate": 2.9005331088664424e-05,
+      "loss": 0.2609,
+      "step": 3785
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 4.946557521820068,
+      "learning_rate": 2.8970258136924805e-05,
+      "loss": 0.3196,
+      "step": 3790
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 3.6709282398223877,
+      "learning_rate": 2.8935185185185186e-05,
+      "loss": 0.2614,
+      "step": 3795
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 4.0705156326293945,
+      "learning_rate": 2.890011223344557e-05,
+      "loss": 0.2454,
+      "step": 3800
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 2.649872064590454,
+      "learning_rate": 2.8865039281705952e-05,
+      "loss": 0.2605,
+      "step": 3805
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 3.158909559249878,
+      "learning_rate": 2.882996632996633e-05,
+      "loss": 0.2613,
+      "step": 3810
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 4.526869773864746,
+      "learning_rate": 2.879489337822671e-05,
+      "loss": 0.3054,
+      "step": 3815
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 4.478386402130127,
+      "learning_rate": 2.8759820426487093e-05,
+      "loss": 0.3222,
+      "step": 3820
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 4.932674884796143,
+      "learning_rate": 2.8724747474747477e-05,
+      "loss": 0.3065,
+      "step": 3825
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 4.305761814117432,
+      "learning_rate": 2.868967452300786e-05,
+      "loss": 0.2936,
+      "step": 3830
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 4.24973201751709,
+      "learning_rate": 2.865460157126824e-05,
+      "loss": 0.3374,
+      "step": 3835
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 4.847403049468994,
+      "learning_rate": 2.8619528619528618e-05,
+      "loss": 0.2892,
+      "step": 3840
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 3.8974742889404297,
+      "learning_rate": 2.8584455667789e-05,
+      "loss": 0.2794,
+      "step": 3845
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 3.9178075790405273,
+      "learning_rate": 2.8549382716049384e-05,
+      "loss": 0.3159,
+      "step": 3850
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 9.098495483398438,
+      "learning_rate": 2.8514309764309765e-05,
+      "loss": 0.3046,
+      "step": 3855
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 4.239152431488037,
+      "learning_rate": 2.8479236812570146e-05,
+      "loss": 0.3068,
+      "step": 3860
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 4.6306233406066895,
+      "learning_rate": 2.844416386083053e-05,
+      "loss": 0.2958,
+      "step": 3865
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 3.943925380706787,
+      "learning_rate": 2.8409090909090912e-05,
+      "loss": 0.2473,
+      "step": 3870
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 4.4443230628967285,
+      "learning_rate": 2.837401795735129e-05,
+      "loss": 0.2509,
+      "step": 3875
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 4.1649627685546875,
+      "learning_rate": 2.833894500561167e-05,
+      "loss": 0.3133,
+      "step": 3880
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 4.602564811706543,
+      "learning_rate": 2.8303872053872056e-05,
+      "loss": 0.2862,
+      "step": 3885
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 2.6787569522857666,
+      "learning_rate": 2.8268799102132438e-05,
+      "loss": 0.2361,
+      "step": 3890
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 4.606193542480469,
+      "learning_rate": 2.823372615039282e-05,
+      "loss": 0.257,
+      "step": 3895
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 4.961560249328613,
+      "learning_rate": 2.8198653198653204e-05,
+      "loss": 0.2647,
+      "step": 3900
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 5.335364818572998,
+      "learning_rate": 2.8163580246913578e-05,
+      "loss": 0.2821,
+      "step": 3905
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 4.9541144371032715,
+      "learning_rate": 2.8128507295173963e-05,
+      "loss": 0.2529,
+      "step": 3910
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 9.019654273986816,
+      "learning_rate": 2.8093434343434344e-05,
+      "loss": 0.2928,
+      "step": 3915
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 3.9855988025665283,
+      "learning_rate": 2.8058361391694725e-05,
+      "loss": 0.2618,
+      "step": 3920
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 5.474127292633057,
+      "learning_rate": 2.802328843995511e-05,
+      "loss": 0.2638,
+      "step": 3925
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 3.234879493713379,
+      "learning_rate": 2.798821548821549e-05,
+      "loss": 0.2727,
+      "step": 3930
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 3.7750728130340576,
+      "learning_rate": 2.7953142536475873e-05,
+      "loss": 0.2758,
+      "step": 3935
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 5.3055500984191895,
+      "learning_rate": 2.791806958473625e-05,
+      "loss": 0.2931,
+      "step": 3940
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 4.582919120788574,
+      "learning_rate": 2.7882996632996632e-05,
+      "loss": 0.2383,
+      "step": 3945
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 3.664515733718872,
+      "learning_rate": 2.7847923681257017e-05,
+      "loss": 0.2898,
+      "step": 3950
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 5.255746364593506,
+      "learning_rate": 2.7812850729517398e-05,
+      "loss": 0.298,
+      "step": 3955
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 5.309757709503174,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.3219,
+      "step": 3960
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9282861453803142,
+      "eval_loss": 0.2060965746641159,
+      "eval_runtime": 31.0537,
+      "eval_samples_per_second": 362.824,
+      "eval_steps_per_second": 11.367,
+      "step": 3961
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 4.868314743041992,
+      "learning_rate": 2.7742704826038164e-05,
+      "loss": 0.2529,
+      "step": 3965
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 4.233950138092041,
+      "learning_rate": 2.770763187429854e-05,
+      "loss": 0.2707,
+      "step": 3970
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 4.4386420249938965,
+      "learning_rate": 2.7672558922558923e-05,
+      "loss": 0.288,
+      "step": 3975
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 4.4777655601501465,
+      "learning_rate": 2.7637485970819304e-05,
+      "loss": 0.296,
+      "step": 3980
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 3.8996291160583496,
+      "learning_rate": 2.7602413019079686e-05,
+      "loss": 0.2209,
+      "step": 3985
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 4.081080913543701,
+      "learning_rate": 2.756734006734007e-05,
+      "loss": 0.2999,
+      "step": 3990
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 3.330458879470825,
+      "learning_rate": 2.753226711560045e-05,
+      "loss": 0.2358,
+      "step": 3995
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 3.4479591846466064,
+      "learning_rate": 2.7497194163860833e-05,
+      "loss": 0.2842,
+      "step": 4000
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 3.1991751194000244,
+      "learning_rate": 2.746212121212121e-05,
+      "loss": 0.2246,
+      "step": 4005
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 3.404226779937744,
+      "learning_rate": 2.7427048260381592e-05,
+      "loss": 0.2825,
+      "step": 4010
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 3.8928279876708984,
+      "learning_rate": 2.7391975308641977e-05,
+      "loss": 0.2674,
+      "step": 4015
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 3.8274989128112793,
+      "learning_rate": 2.7356902356902358e-05,
+      "loss": 0.2823,
+      "step": 4020
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 4.07806921005249,
+      "learning_rate": 2.7321829405162743e-05,
+      "loss": 0.2861,
+      "step": 4025
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 4.24616003036499,
+      "learning_rate": 2.7286756453423124e-05,
+      "loss": 0.263,
+      "step": 4030
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 4.640658855438232,
+      "learning_rate": 2.7251683501683502e-05,
+      "loss": 0.2288,
+      "step": 4035
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 4.159529685974121,
+      "learning_rate": 2.7216610549943883e-05,
+      "loss": 0.3265,
+      "step": 4040
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 4.500327110290527,
+      "learning_rate": 2.7181537598204264e-05,
+      "loss": 0.2981,
+      "step": 4045
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 3.5905261039733887,
+      "learning_rate": 2.714646464646465e-05,
+      "loss": 0.2129,
+      "step": 4050
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 4.6054768562316895,
+      "learning_rate": 2.711139169472503e-05,
+      "loss": 0.265,
+      "step": 4055
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 4.793301105499268,
+      "learning_rate": 2.7076318742985412e-05,
+      "loss": 0.2604,
+      "step": 4060
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 4.09229040145874,
+      "learning_rate": 2.7041245791245796e-05,
+      "loss": 0.2351,
+      "step": 4065
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 2.998105049133301,
+      "learning_rate": 2.700617283950617e-05,
+      "loss": 0.2736,
+      "step": 4070
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 3.2948195934295654,
+      "learning_rate": 2.6971099887766556e-05,
+      "loss": 0.2334,
+      "step": 4075
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 8.048471450805664,
+      "learning_rate": 2.6936026936026937e-05,
+      "loss": 0.2639,
+      "step": 4080
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 5.577630519866943,
+      "learning_rate": 2.6900953984287318e-05,
+      "loss": 0.2802,
+      "step": 4085
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 5.575340747833252,
+      "learning_rate": 2.6865881032547703e-05,
+      "loss": 0.2753,
+      "step": 4090
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 4.186885833740234,
+      "learning_rate": 2.6830808080808084e-05,
+      "loss": 0.318,
+      "step": 4095
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 3.725076198577881,
+      "learning_rate": 2.6795735129068462e-05,
+      "loss": 0.2317,
+      "step": 4100
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 4.088349342346191,
+      "learning_rate": 2.6760662177328843e-05,
+      "loss": 0.3188,
+      "step": 4105
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 4.296757698059082,
+      "learning_rate": 2.6725589225589225e-05,
+      "loss": 0.2829,
+      "step": 4110
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 3.276031494140625,
+      "learning_rate": 2.669051627384961e-05,
+      "loss": 0.2975,
+      "step": 4115
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 4.0198822021484375,
+      "learning_rate": 2.665544332210999e-05,
+      "loss": 0.2875,
+      "step": 4120
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 3.913980722427368,
+      "learning_rate": 2.6620370370370372e-05,
+      "loss": 0.2846,
+      "step": 4125
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 3.347252607345581,
+      "learning_rate": 2.6585297418630757e-05,
+      "loss": 0.3131,
+      "step": 4130
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 4.632914066314697,
+      "learning_rate": 2.655022446689113e-05,
+      "loss": 0.2782,
+      "step": 4135
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 4.8889994621276855,
+      "learning_rate": 2.6515151515151516e-05,
+      "loss": 0.3266,
+      "step": 4140
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 3.2841732501983643,
+      "learning_rate": 2.6480078563411897e-05,
+      "loss": 0.2397,
+      "step": 4145
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 3.5024795532226562,
+      "learning_rate": 2.644500561167228e-05,
+      "loss": 0.2792,
+      "step": 4150
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 3.300025224685669,
+      "learning_rate": 2.6409932659932663e-05,
+      "loss": 0.3101,
+      "step": 4155
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 5.32696533203125,
+      "learning_rate": 2.6374859708193044e-05,
+      "loss": 0.2954,
+      "step": 4160
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 3.7689497470855713,
+      "learning_rate": 2.6339786756453422e-05,
+      "loss": 0.2261,
+      "step": 4165
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 3.969750165939331,
+      "learning_rate": 2.6304713804713804e-05,
+      "loss": 0.2868,
+      "step": 4170
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 4.76958703994751,
+      "learning_rate": 2.6269640852974188e-05,
+      "loss": 0.2719,
+      "step": 4175
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 3.5822479724884033,
+      "learning_rate": 2.623456790123457e-05,
+      "loss": 0.2558,
+      "step": 4180
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 4.568985939025879,
+      "learning_rate": 2.619949494949495e-05,
+      "loss": 0.2863,
+      "step": 4185
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 3.336733341217041,
+      "learning_rate": 2.6164421997755336e-05,
+      "loss": 0.2341,
+      "step": 4190
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 4.853896617889404,
+      "learning_rate": 2.6129349046015717e-05,
+      "loss": 0.3089,
+      "step": 4195
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 3.807572364807129,
+      "learning_rate": 2.6094276094276095e-05,
+      "loss": 0.2481,
+      "step": 4200
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 3.8648457527160645,
+      "learning_rate": 2.6059203142536476e-05,
+      "loss": 0.2698,
+      "step": 4205
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 4.929937362670898,
+      "learning_rate": 2.6024130190796857e-05,
+      "loss": 0.2856,
+      "step": 4210
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 5.259349822998047,
+      "learning_rate": 2.5989057239057242e-05,
+      "loss": 0.2458,
+      "step": 4215
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 3.1494500637054443,
+      "learning_rate": 2.5953984287317623e-05,
+      "loss": 0.2607,
+      "step": 4220
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 4.356003284454346,
+      "learning_rate": 2.5918911335578005e-05,
+      "loss": 0.3131,
+      "step": 4225
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 4.362011909484863,
+      "learning_rate": 2.5883838383838382e-05,
+      "loss": 0.2444,
+      "step": 4230
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 3.3773274421691895,
+      "learning_rate": 2.5848765432098764e-05,
+      "loss": 0.2645,
+      "step": 4235
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 4.698849678039551,
+      "learning_rate": 2.581369248035915e-05,
+      "loss": 0.2883,
+      "step": 4240
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 3.2968344688415527,
+      "learning_rate": 2.577861952861953e-05,
+      "loss": 0.2687,
+      "step": 4245
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 5.072206974029541,
+      "learning_rate": 2.574354657687991e-05,
+      "loss": 0.2277,
+      "step": 4250
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 3.3548755645751953,
+      "learning_rate": 2.5708473625140296e-05,
+      "loss": 0.2301,
+      "step": 4255
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 3.633535623550415,
+      "learning_rate": 2.5673400673400677e-05,
+      "loss": 0.2947,
+      "step": 4260
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 3.377603769302368,
+      "learning_rate": 2.5638327721661055e-05,
+      "loss": 0.3107,
+      "step": 4265
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 4.4706292152404785,
+      "learning_rate": 2.5603254769921436e-05,
+      "loss": 0.2703,
+      "step": 4270
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 4.663482189178467,
+      "learning_rate": 2.5568181818181817e-05,
+      "loss": 0.2663,
+      "step": 4275
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 6.979432582855225,
+      "learning_rate": 2.5533108866442202e-05,
+      "loss": 0.2655,
+      "step": 4280
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 4.990533351898193,
+      "learning_rate": 2.5498035914702583e-05,
+      "loss": 0.2869,
+      "step": 4285
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 3.7890074253082275,
+      "learning_rate": 2.5462962962962965e-05,
+      "loss": 0.2838,
+      "step": 4290
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 5.708714008331299,
+      "learning_rate": 2.5427890011223343e-05,
+      "loss": 0.3285,
+      "step": 4295
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 2.7534244060516357,
+      "learning_rate": 2.5392817059483724e-05,
+      "loss": 0.2415,
+      "step": 4300
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 3.5662901401519775,
+      "learning_rate": 2.535774410774411e-05,
+      "loss": 0.243,
+      "step": 4305
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 3.50412917137146,
+      "learning_rate": 2.532267115600449e-05,
+      "loss": 0.2914,
+      "step": 4310
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 3.6044745445251465,
+      "learning_rate": 2.5287598204264875e-05,
+      "loss": 0.2389,
+      "step": 4315
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 4.193859577178955,
+      "learning_rate": 2.5252525252525256e-05,
+      "loss": 0.2341,
+      "step": 4320
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 3.331221103668213,
+      "learning_rate": 2.5217452300785637e-05,
+      "loss": 0.2793,
+      "step": 4325
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 4.2629218101501465,
+      "learning_rate": 2.5182379349046015e-05,
+      "loss": 0.2569,
+      "step": 4330
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 3.203535318374634,
+      "learning_rate": 2.5147306397306396e-05,
+      "loss": 0.2371,
+      "step": 4335
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 3.9185290336608887,
+      "learning_rate": 2.511223344556678e-05,
+      "loss": 0.2407,
+      "step": 4340
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 4.686899662017822,
+      "learning_rate": 2.5077160493827162e-05,
+      "loss": 0.257,
+      "step": 4345
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 4.1918044090271,
+      "learning_rate": 2.5042087542087544e-05,
+      "loss": 0.2633,
+      "step": 4350
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 5.8334197998046875,
+      "learning_rate": 2.500701459034793e-05,
+      "loss": 0.2313,
+      "step": 4355
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 2.523483991622925,
+      "learning_rate": 2.4971941638608306e-05,
+      "loss": 0.2662,
+      "step": 4360
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 3.8091158866882324,
+      "learning_rate": 2.4936868686868688e-05,
+      "loss": 0.2552,
+      "step": 4365
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 4.169099807739258,
+      "learning_rate": 2.490179573512907e-05,
+      "loss": 0.2415,
+      "step": 4370
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 4.776482582092285,
+      "learning_rate": 2.486672278338945e-05,
+      "loss": 0.3552,
+      "step": 4375
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 3.383645534515381,
+      "learning_rate": 2.4831649831649835e-05,
+      "loss": 0.2465,
+      "step": 4380
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 5.437908172607422,
+      "learning_rate": 2.4796576879910213e-05,
+      "loss": 0.2071,
+      "step": 4385
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 6.401407718658447,
+      "learning_rate": 2.4761503928170597e-05,
+      "loss": 0.2453,
+      "step": 4390
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 3.6610829830169678,
+      "learning_rate": 2.472643097643098e-05,
+      "loss": 0.3037,
+      "step": 4395
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 4.050108432769775,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 0.3081,
+      "step": 4400
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 4.4872846603393555,
+      "learning_rate": 2.465628507295174e-05,
+      "loss": 0.2714,
+      "step": 4405
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 4.275237560272217,
+      "learning_rate": 2.4621212121212123e-05,
+      "loss": 0.2813,
+      "step": 4410
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 3.3150930404663086,
+      "learning_rate": 2.4586139169472504e-05,
+      "loss": 0.2749,
+      "step": 4415
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 5.4988112449646,
+      "learning_rate": 2.4551066217732885e-05,
+      "loss": 0.2812,
+      "step": 4420
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 4.53728723526001,
+      "learning_rate": 2.4515993265993266e-05,
+      "loss": 0.2989,
+      "step": 4425
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 3.4260852336883545,
+      "learning_rate": 2.448092031425365e-05,
+      "loss": 0.2321,
+      "step": 4430
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 4.183116912841797,
+      "learning_rate": 2.444584736251403e-05,
+      "loss": 0.2656,
+      "step": 4435
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 3.943293571472168,
+      "learning_rate": 2.441077441077441e-05,
+      "loss": 0.2478,
+      "step": 4440
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 5.351258277893066,
+      "learning_rate": 2.4375701459034795e-05,
+      "loss": 0.3055,
+      "step": 4445
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 5.555354118347168,
+      "learning_rate": 2.4340628507295173e-05,
+      "loss": 0.2892,
+      "step": 4450
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 3.2171051502227783,
+      "learning_rate": 2.4305555555555558e-05,
+      "loss": 0.2577,
+      "step": 4455
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 3.921525239944458,
+      "learning_rate": 2.427048260381594e-05,
+      "loss": 0.2564,
+      "step": 4460
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 3.5476150512695312,
+      "learning_rate": 2.423540965207632e-05,
+      "loss": 0.2562,
+      "step": 4465
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 4.040847301483154,
+      "learning_rate": 2.42003367003367e-05,
+      "loss": 0.3094,
+      "step": 4470
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 2.7773661613464355,
+      "learning_rate": 2.4165263748597083e-05,
+      "loss": 0.2124,
+      "step": 4475
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 2.3389170169830322,
+      "learning_rate": 2.4130190796857467e-05,
+      "loss": 0.2226,
+      "step": 4480
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 5.925633430480957,
+      "learning_rate": 2.4095117845117845e-05,
+      "loss": 0.2561,
+      "step": 4485
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 2.9923391342163086,
+      "learning_rate": 2.4060044893378227e-05,
+      "loss": 0.2495,
+      "step": 4490
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 3.6197221279144287,
+      "learning_rate": 2.402497194163861e-05,
+      "loss": 0.2802,
+      "step": 4495
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 5.358989238739014,
+      "learning_rate": 2.398989898989899e-05,
+      "loss": 0.28,
+      "step": 4500
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 4.6347222328186035,
+      "learning_rate": 2.3954826038159374e-05,
+      "loss": 0.3067,
+      "step": 4505
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 4.678347110748291,
+      "learning_rate": 2.3919753086419755e-05,
+      "loss": 0.2787,
+      "step": 4510
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 4.654605388641357,
+      "learning_rate": 2.3884680134680133e-05,
+      "loss": 0.2626,
+      "step": 4515
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 5.828289031982422,
+      "learning_rate": 2.3849607182940518e-05,
+      "loss": 0.2885,
+      "step": 4520
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 3.468040704727173,
+      "learning_rate": 2.38145342312009e-05,
+      "loss": 0.2434,
+      "step": 4525
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 4.807915210723877,
+      "learning_rate": 2.377946127946128e-05,
+      "loss": 0.2721,
+      "step": 4530
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 4.4735307693481445,
+      "learning_rate": 2.374438832772166e-05,
+      "loss": 0.2604,
+      "step": 4535
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 4.641149997711182,
+      "learning_rate": 2.3709315375982043e-05,
+      "loss": 0.3029,
+      "step": 4540
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 3.3757479190826416,
+      "learning_rate": 2.3674242424242428e-05,
+      "loss": 0.2298,
+      "step": 4545
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 4.867739200592041,
+      "learning_rate": 2.3639169472502806e-05,
+      "loss": 0.2497,
+      "step": 4550
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 4.057545185089111,
+      "learning_rate": 2.360409652076319e-05,
+      "loss": 0.2428,
+      "step": 4555
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 3.220677137374878,
+      "learning_rate": 2.356902356902357e-05,
+      "loss": 0.28,
+      "step": 4560
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 4.784502983093262,
+      "learning_rate": 2.353395061728395e-05,
+      "loss": 0.2452,
+      "step": 4565
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 3.487222194671631,
+      "learning_rate": 2.3498877665544334e-05,
+      "loss": 0.2951,
+      "step": 4570
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 4.065348148345947,
+      "learning_rate": 2.3463804713804715e-05,
+      "loss": 0.2959,
+      "step": 4575
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 3.666316509246826,
+      "learning_rate": 2.3428731762065097e-05,
+      "loss": 0.274,
+      "step": 4580
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 3.5543367862701416,
+      "learning_rate": 2.3393658810325478e-05,
+      "loss": 0.2848,
+      "step": 4585
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 3.669658899307251,
+      "learning_rate": 2.335858585858586e-05,
+      "loss": 0.2976,
+      "step": 4590
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 3.8172848224639893,
+      "learning_rate": 2.3323512906846244e-05,
+      "loss": 0.2515,
+      "step": 4595
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 5.17378044128418,
+      "learning_rate": 2.3288439955106622e-05,
+      "loss": 0.2708,
+      "step": 4600
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 4.316329479217529,
+      "learning_rate": 2.3253367003367003e-05,
+      "loss": 0.2889,
+      "step": 4605
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 3.2597568035125732,
+      "learning_rate": 2.3218294051627388e-05,
+      "loss": 0.267,
+      "step": 4610
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 4.561779022216797,
+      "learning_rate": 2.3183221099887766e-05,
+      "loss": 0.2302,
+      "step": 4615
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 3.4911928176879883,
+      "learning_rate": 2.314814814814815e-05,
+      "loss": 0.2201,
+      "step": 4620
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 3.8962509632110596,
+      "learning_rate": 2.311307519640853e-05,
+      "loss": 0.2912,
+      "step": 4625
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 3.9013776779174805,
+      "learning_rate": 2.3078002244668913e-05,
+      "loss": 0.272,
+      "step": 4630
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 4.102755546569824,
+      "learning_rate": 2.3042929292929294e-05,
+      "loss": 0.2707,
+      "step": 4635
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 3.7835803031921387,
+      "learning_rate": 2.3007856341189676e-05,
+      "loss": 0.2607,
+      "step": 4640
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 5.810889720916748,
+      "learning_rate": 2.2972783389450057e-05,
+      "loss": 0.2643,
+      "step": 4645
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 3.618899345397949,
+      "learning_rate": 2.2937710437710438e-05,
+      "loss": 0.2641,
+      "step": 4650
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 4.813767433166504,
+      "learning_rate": 2.290263748597082e-05,
+      "loss": 0.2822,
+      "step": 4655
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 3.811750650405884,
+      "learning_rate": 2.2867564534231204e-05,
+      "loss": 0.2552,
+      "step": 4660
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 4.628886699676514,
+      "learning_rate": 2.2832491582491582e-05,
+      "loss": 0.3075,
+      "step": 4665
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 4.643641471862793,
+      "learning_rate": 2.2797418630751967e-05,
+      "loss": 0.2424,
+      "step": 4670
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 3.5601320266723633,
+      "learning_rate": 2.2762345679012348e-05,
+      "loss": 0.2516,
+      "step": 4675
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 5.497618675231934,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.2789,
+      "step": 4680
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 4.102509021759033,
+      "learning_rate": 2.269219977553311e-05,
+      "loss": 0.2455,
+      "step": 4685
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 3.8160579204559326,
+      "learning_rate": 2.2657126823793492e-05,
+      "loss": 0.275,
+      "step": 4690
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 4.764002323150635,
+      "learning_rate": 2.2622053872053873e-05,
+      "loss": 0.3483,
+      "step": 4695
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 6.374079704284668,
+      "learning_rate": 2.2586980920314254e-05,
+      "loss": 0.2756,
+      "step": 4700
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 4.590950965881348,
+      "learning_rate": 2.2551907968574636e-05,
+      "loss": 0.2835,
+      "step": 4705
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 2.721900463104248,
+      "learning_rate": 2.2516835016835017e-05,
+      "loss": 0.2431,
+      "step": 4710
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 4.414608478546143,
+      "learning_rate": 2.24817620650954e-05,
+      "loss": 0.2905,
+      "step": 4715
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 4.187251091003418,
+      "learning_rate": 2.2446689113355783e-05,
+      "loss": 0.268,
+      "step": 4720
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 4.103115558624268,
+      "learning_rate": 2.2411616161616164e-05,
+      "loss": 0.2733,
+      "step": 4725
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 5.582839488983154,
+      "learning_rate": 2.2376543209876542e-05,
+      "loss": 0.2885,
+      "step": 4730
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 5.9075846672058105,
+      "learning_rate": 2.2341470258136927e-05,
+      "loss": 0.2761,
+      "step": 4735
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 3.8202288150787354,
+      "learning_rate": 2.2306397306397308e-05,
+      "loss": 0.2638,
+      "step": 4740
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 4.397262096405029,
+      "learning_rate": 2.227132435465769e-05,
+      "loss": 0.2418,
+      "step": 4745
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 4.079167366027832,
+      "learning_rate": 2.223625140291807e-05,
+      "loss": 0.2293,
+      "step": 4750
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9333451673027425,
+      "eval_loss": 0.19650360941886902,
+      "eval_runtime": 30.786,
+      "eval_samples_per_second": 365.979,
+      "eval_steps_per_second": 11.466,
+      "step": 4753
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 3.4745070934295654,
+      "learning_rate": 2.2201178451178452e-05,
+      "loss": 0.237,
+      "step": 4755
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 3.574777364730835,
+      "learning_rate": 2.2166105499438833e-05,
+      "loss": 0.2915,
+      "step": 4760
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 3.827456474304199,
+      "learning_rate": 2.2131032547699215e-05,
+      "loss": 0.256,
+      "step": 4765
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 4.654812335968018,
+      "learning_rate": 2.20959595959596e-05,
+      "loss": 0.2466,
+      "step": 4770
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 3.977865695953369,
+      "learning_rate": 2.2060886644219977e-05,
+      "loss": 0.2456,
+      "step": 4775
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 4.178462028503418,
+      "learning_rate": 2.202581369248036e-05,
+      "loss": 0.2269,
+      "step": 4780
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 3.395653486251831,
+      "learning_rate": 2.1990740740740743e-05,
+      "loss": 0.2495,
+      "step": 4785
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 7.679159164428711,
+      "learning_rate": 2.1955667789001124e-05,
+      "loss": 0.2434,
+      "step": 4790
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 5.095306873321533,
+      "learning_rate": 2.1920594837261506e-05,
+      "loss": 0.2586,
+      "step": 4795
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 3.722534656524658,
+      "learning_rate": 2.1885521885521887e-05,
+      "loss": 0.3154,
+      "step": 4800
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 3.9734911918640137,
+      "learning_rate": 2.185044893378227e-05,
+      "loss": 0.256,
+      "step": 4805
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 5.107233047485352,
+      "learning_rate": 2.181537598204265e-05,
+      "loss": 0.2283,
+      "step": 4810
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 4.56982946395874,
+      "learning_rate": 2.178030303030303e-05,
+      "loss": 0.2581,
+      "step": 4815
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 5.576462268829346,
+      "learning_rate": 2.1745230078563412e-05,
+      "loss": 0.2505,
+      "step": 4820
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 3.6051368713378906,
+      "learning_rate": 2.1710157126823794e-05,
+      "loss": 0.2267,
+      "step": 4825
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 4.3116888999938965,
+      "learning_rate": 2.1675084175084175e-05,
+      "loss": 0.2321,
+      "step": 4830
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 5.356314659118652,
+      "learning_rate": 2.164001122334456e-05,
+      "loss": 0.2258,
+      "step": 4835
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 5.7071638107299805,
+      "learning_rate": 2.1604938271604937e-05,
+      "loss": 0.2359,
+      "step": 4840
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 4.500997066497803,
+      "learning_rate": 2.1569865319865322e-05,
+      "loss": 0.3082,
+      "step": 4845
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 4.977388858795166,
+      "learning_rate": 2.1534792368125703e-05,
+      "loss": 0.249,
+      "step": 4850
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 4.279355525970459,
+      "learning_rate": 2.1499719416386085e-05,
+      "loss": 0.2502,
+      "step": 4855
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 6.30715799331665,
+      "learning_rate": 2.1464646464646466e-05,
+      "loss": 0.1995,
+      "step": 4860
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 6.237218856811523,
+      "learning_rate": 2.1429573512906847e-05,
+      "loss": 0.2367,
+      "step": 4865
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 4.428788661956787,
+      "learning_rate": 2.139450056116723e-05,
+      "loss": 0.2602,
+      "step": 4870
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 6.145692825317383,
+      "learning_rate": 2.135942760942761e-05,
+      "loss": 0.3016,
+      "step": 4875
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 4.525232791900635,
+      "learning_rate": 2.132435465768799e-05,
+      "loss": 0.2944,
+      "step": 4880
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 5.535518169403076,
+      "learning_rate": 2.1289281705948376e-05,
+      "loss": 0.2569,
+      "step": 4885
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 5.3942694664001465,
+      "learning_rate": 2.1254208754208754e-05,
+      "loss": 0.2662,
+      "step": 4890
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 3.7821834087371826,
+      "learning_rate": 2.1219135802469135e-05,
+      "loss": 0.2459,
+      "step": 4895
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 3.851886510848999,
+      "learning_rate": 2.118406285072952e-05,
+      "loss": 0.2634,
+      "step": 4900
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 4.465198040008545,
+      "learning_rate": 2.1148989898989898e-05,
+      "loss": 0.3029,
+      "step": 4905
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 4.11469030380249,
+      "learning_rate": 2.1113916947250282e-05,
+      "loss": 0.2845,
+      "step": 4910
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 5.311576843261719,
+      "learning_rate": 2.1078843995510664e-05,
+      "loss": 0.268,
+      "step": 4915
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 3.6384809017181396,
+      "learning_rate": 2.1043771043771045e-05,
+      "loss": 0.2588,
+      "step": 4920
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 3.4926209449768066,
+      "learning_rate": 2.1008698092031426e-05,
+      "loss": 0.2553,
+      "step": 4925
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 4.46201229095459,
+      "learning_rate": 2.0973625140291807e-05,
+      "loss": 0.2713,
+      "step": 4930
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 3.5251691341400146,
+      "learning_rate": 2.0938552188552192e-05,
+      "loss": 0.2274,
+      "step": 4935
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 5.804261207580566,
+      "learning_rate": 2.090347923681257e-05,
+      "loss": 0.2522,
+      "step": 4940
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 3.7517952919006348,
+      "learning_rate": 2.086840628507295e-05,
+      "loss": 0.2236,
+      "step": 4945
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 5.249544143676758,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.2713,
+      "step": 4950
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 4.673359394073486,
+      "learning_rate": 2.0798260381593714e-05,
+      "loss": 0.2539,
+      "step": 4955
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 5.284143447875977,
+      "learning_rate": 2.07631874298541e-05,
+      "loss": 0.2733,
+      "step": 4960
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 4.652414798736572,
+      "learning_rate": 2.072811447811448e-05,
+      "loss": 0.2911,
+      "step": 4965
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 4.203030586242676,
+      "learning_rate": 2.0693041526374858e-05,
+      "loss": 0.3251,
+      "step": 4970
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 6.154104232788086,
+      "learning_rate": 2.0657968574635242e-05,
+      "loss": 0.2777,
+      "step": 4975
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 4.065311431884766,
+      "learning_rate": 2.0622895622895624e-05,
+      "loss": 0.2843,
+      "step": 4980
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 3.8225202560424805,
+      "learning_rate": 2.058782267115601e-05,
+      "loss": 0.2391,
+      "step": 4985
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 4.593325138092041,
+      "learning_rate": 2.0552749719416386e-05,
+      "loss": 0.3011,
+      "step": 4990
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 5.887181758880615,
+      "learning_rate": 2.0517676767676768e-05,
+      "loss": 0.2212,
+      "step": 4995
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 4.735381603240967,
+      "learning_rate": 2.0482603815937152e-05,
+      "loss": 0.2143,
+      "step": 5000
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 4.427429676055908,
+      "learning_rate": 2.044753086419753e-05,
+      "loss": 0.2591,
+      "step": 5005
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 4.791842460632324,
+      "learning_rate": 2.0412457912457915e-05,
+      "loss": 0.3011,
+      "step": 5010
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 4.0126519203186035,
+      "learning_rate": 2.0377384960718296e-05,
+      "loss": 0.2545,
+      "step": 5015
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 2.8306500911712646,
+      "learning_rate": 2.0342312008978674e-05,
+      "loss": 0.1874,
+      "step": 5020
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 2.80130672454834,
+      "learning_rate": 2.030723905723906e-05,
+      "loss": 0.308,
+      "step": 5025
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 4.053665637969971,
+      "learning_rate": 2.027216610549944e-05,
+      "loss": 0.2487,
+      "step": 5030
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 5.102134704589844,
+      "learning_rate": 2.023709315375982e-05,
+      "loss": 0.2623,
+      "step": 5035
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 4.028321743011475,
+      "learning_rate": 2.0202020202020203e-05,
+      "loss": 0.2384,
+      "step": 5040
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 5.456782817840576,
+      "learning_rate": 2.0166947250280584e-05,
+      "loss": 0.2789,
+      "step": 5045
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 5.348091125488281,
+      "learning_rate": 2.013187429854097e-05,
+      "loss": 0.3073,
+      "step": 5050
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 3.7774267196655273,
+      "learning_rate": 2.0096801346801347e-05,
+      "loss": 0.2165,
+      "step": 5055
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 4.179872989654541,
+      "learning_rate": 2.006172839506173e-05,
+      "loss": 0.2697,
+      "step": 5060
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 4.155890941619873,
+      "learning_rate": 2.0026655443322113e-05,
+      "loss": 0.2153,
+      "step": 5065
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 3.982516050338745,
+      "learning_rate": 1.999158249158249e-05,
+      "loss": 0.2258,
+      "step": 5070
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 4.4911417961120605,
+      "learning_rate": 1.9956509539842875e-05,
+      "loss": 0.2268,
+      "step": 5075
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 4.529824256896973,
+      "learning_rate": 1.9921436588103256e-05,
+      "loss": 0.2646,
+      "step": 5080
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 3.3216142654418945,
+      "learning_rate": 1.9886363636363638e-05,
+      "loss": 0.2395,
+      "step": 5085
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 6.945095062255859,
+      "learning_rate": 1.985129068462402e-05,
+      "loss": 0.1904,
+      "step": 5090
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 3.2466161251068115,
+      "learning_rate": 1.98162177328844e-05,
+      "loss": 0.28,
+      "step": 5095
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 5.192864418029785,
+      "learning_rate": 1.978114478114478e-05,
+      "loss": 0.2273,
+      "step": 5100
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 3.710413932800293,
+      "learning_rate": 1.9746071829405163e-05,
+      "loss": 0.2306,
+      "step": 5105
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 3.9969325065612793,
+      "learning_rate": 1.9710998877665544e-05,
+      "loss": 0.2511,
+      "step": 5110
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 6.11897087097168,
+      "learning_rate": 1.967592592592593e-05,
+      "loss": 0.3264,
+      "step": 5115
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 3.841963052749634,
+      "learning_rate": 1.9640852974186307e-05,
+      "loss": 0.2446,
+      "step": 5120
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 5.392006874084473,
+      "learning_rate": 1.960578002244669e-05,
+      "loss": 0.2914,
+      "step": 5125
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 4.254004955291748,
+      "learning_rate": 1.9570707070707073e-05,
+      "loss": 0.3384,
+      "step": 5130
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 3.3631274700164795,
+      "learning_rate": 1.9535634118967454e-05,
+      "loss": 0.2298,
+      "step": 5135
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 3.9269559383392334,
+      "learning_rate": 1.9500561167227835e-05,
+      "loss": 0.2669,
+      "step": 5140
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 4.241915225982666,
+      "learning_rate": 1.9465488215488217e-05,
+      "loss": 0.237,
+      "step": 5145
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 4.429218292236328,
+      "learning_rate": 1.9430415263748598e-05,
+      "loss": 0.2256,
+      "step": 5150
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 4.974701881408691,
+      "learning_rate": 1.939534231200898e-05,
+      "loss": 0.2571,
+      "step": 5155
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 4.843481540679932,
+      "learning_rate": 1.936026936026936e-05,
+      "loss": 0.2828,
+      "step": 5160
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 3.2197484970092773,
+      "learning_rate": 1.9325196408529742e-05,
+      "loss": 0.2486,
+      "step": 5165
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 4.035525321960449,
+      "learning_rate": 1.9290123456790123e-05,
+      "loss": 0.2425,
+      "step": 5170
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 4.07498836517334,
+      "learning_rate": 1.9255050505050508e-05,
+      "loss": 0.2727,
+      "step": 5175
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 3.8623833656311035,
+      "learning_rate": 1.921997755331089e-05,
+      "loss": 0.2062,
+      "step": 5180
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 4.7320170402526855,
+      "learning_rate": 1.9184904601571267e-05,
+      "loss": 0.2759,
+      "step": 5185
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 5.4494805335998535,
+      "learning_rate": 1.914983164983165e-05,
+      "loss": 0.2888,
+      "step": 5190
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 2.981865644454956,
+      "learning_rate": 1.9114758698092033e-05,
+      "loss": 0.2751,
+      "step": 5195
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 3.272700548171997,
+      "learning_rate": 1.9079685746352414e-05,
+      "loss": 0.2644,
+      "step": 5200
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 4.498180389404297,
+      "learning_rate": 1.9044612794612795e-05,
+      "loss": 0.2034,
+      "step": 5205
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 4.897580623626709,
+      "learning_rate": 1.9009539842873177e-05,
+      "loss": 0.2711,
+      "step": 5210
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 3.8099071979522705,
+      "learning_rate": 1.8974466891133558e-05,
+      "loss": 0.2655,
+      "step": 5215
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 3.4621076583862305,
+      "learning_rate": 1.893939393939394e-05,
+      "loss": 0.2897,
+      "step": 5220
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 4.474539756774902,
+      "learning_rate": 1.8904320987654324e-05,
+      "loss": 0.2543,
+      "step": 5225
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 3.4364736080169678,
+      "learning_rate": 1.8869248035914702e-05,
+      "loss": 0.2317,
+      "step": 5230
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 4.023273944854736,
+      "learning_rate": 1.8834175084175083e-05,
+      "loss": 0.2592,
+      "step": 5235
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 4.372448921203613,
+      "learning_rate": 1.8799102132435468e-05,
+      "loss": 0.2864,
+      "step": 5240
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 6.084567546844482,
+      "learning_rate": 1.876402918069585e-05,
+      "loss": 0.2646,
+      "step": 5245
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 5.141995906829834,
+      "learning_rate": 1.872895622895623e-05,
+      "loss": 0.2363,
+      "step": 5250
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 4.04919958114624,
+      "learning_rate": 1.8693883277216612e-05,
+      "loss": 0.2776,
+      "step": 5255
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 2.956270694732666,
+      "learning_rate": 1.8658810325476993e-05,
+      "loss": 0.239,
+      "step": 5260
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 2.664314031600952,
+      "learning_rate": 1.8623737373737374e-05,
+      "loss": 0.2209,
+      "step": 5265
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 4.711406707763672,
+      "learning_rate": 1.8588664421997756e-05,
+      "loss": 0.2655,
+      "step": 5270
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 4.861990928649902,
+      "learning_rate": 1.855359147025814e-05,
+      "loss": 0.1961,
+      "step": 5275
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 3.5957541465759277,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.2659,
+      "step": 5280
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 3.0312016010284424,
+      "learning_rate": 1.84834455667789e-05,
+      "loss": 0.2362,
+      "step": 5285
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 5.4257612228393555,
+      "learning_rate": 1.8448372615039284e-05,
+      "loss": 0.2533,
+      "step": 5290
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 6.223615646362305,
+      "learning_rate": 1.8413299663299662e-05,
+      "loss": 0.2499,
+      "step": 5295
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 8.294111251831055,
+      "learning_rate": 1.8378226711560047e-05,
+      "loss": 0.2661,
+      "step": 5300
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 3.717557430267334,
+      "learning_rate": 1.8343153759820428e-05,
+      "loss": 0.273,
+      "step": 5305
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 4.496303081512451,
+      "learning_rate": 1.830808080808081e-05,
+      "loss": 0.2157,
+      "step": 5310
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 3.0501766204833984,
+      "learning_rate": 1.827300785634119e-05,
+      "loss": 0.2549,
+      "step": 5315
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 6.245630264282227,
+      "learning_rate": 1.8237934904601572e-05,
+      "loss": 0.2824,
+      "step": 5320
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 5.474722385406494,
+      "learning_rate": 1.8202861952861953e-05,
+      "loss": 0.2663,
+      "step": 5325
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 4.296999454498291,
+      "learning_rate": 1.8167789001122335e-05,
+      "loss": 0.252,
+      "step": 5330
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 3.865527629852295,
+      "learning_rate": 1.8132716049382716e-05,
+      "loss": 0.2702,
+      "step": 5335
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 5.470089435577393,
+      "learning_rate": 1.80976430976431e-05,
+      "loss": 0.261,
+      "step": 5340
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 9.061738014221191,
+      "learning_rate": 1.806257014590348e-05,
+      "loss": 0.2576,
+      "step": 5345
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 6.673803329467773,
+      "learning_rate": 1.8027497194163863e-05,
+      "loss": 0.2223,
+      "step": 5350
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 4.0333757400512695,
+      "learning_rate": 1.7992424242424244e-05,
+      "loss": 0.2649,
+      "step": 5355
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 3.765423059463501,
+      "learning_rate": 1.7957351290684622e-05,
+      "loss": 0.2732,
+      "step": 5360
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 3.0370352268218994,
+      "learning_rate": 1.7922278338945007e-05,
+      "loss": 0.2282,
+      "step": 5365
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 5.002012252807617,
+      "learning_rate": 1.788720538720539e-05,
+      "loss": 0.2534,
+      "step": 5370
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 3.9884979724884033,
+      "learning_rate": 1.785213243546577e-05,
+      "loss": 0.2456,
+      "step": 5375
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 4.451979637145996,
+      "learning_rate": 1.781705948372615e-05,
+      "loss": 0.236,
+      "step": 5380
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 4.7708821296691895,
+      "learning_rate": 1.7781986531986532e-05,
+      "loss": 0.2723,
+      "step": 5385
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 4.162632942199707,
+      "learning_rate": 1.7746913580246917e-05,
+      "loss": 0.2559,
+      "step": 5390
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 3.7455341815948486,
+      "learning_rate": 1.7711840628507295e-05,
+      "loss": 0.2428,
+      "step": 5395
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 5.373408317565918,
+      "learning_rate": 1.7676767676767676e-05,
+      "loss": 0.3197,
+      "step": 5400
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 4.155795097351074,
+      "learning_rate": 1.764169472502806e-05,
+      "loss": 0.2593,
+      "step": 5405
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 4.299027442932129,
+      "learning_rate": 1.760662177328844e-05,
+      "loss": 0.2721,
+      "step": 5410
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 4.297277450561523,
+      "learning_rate": 1.7571548821548823e-05,
+      "loss": 0.2158,
+      "step": 5415
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 6.006078243255615,
+      "learning_rate": 1.7536475869809205e-05,
+      "loss": 0.28,
+      "step": 5420
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 3.393371343612671,
+      "learning_rate": 1.7501402918069586e-05,
+      "loss": 0.2354,
+      "step": 5425
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 4.720546722412109,
+      "learning_rate": 1.7466329966329967e-05,
+      "loss": 0.2221,
+      "step": 5430
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 3.9563212394714355,
+      "learning_rate": 1.743125701459035e-05,
+      "loss": 0.2655,
+      "step": 5435
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 3.566749095916748,
+      "learning_rate": 1.7396184062850733e-05,
+      "loss": 0.2167,
+      "step": 5440
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 4.543542861938477,
+      "learning_rate": 1.736111111111111e-05,
+      "loss": 0.2479,
+      "step": 5445
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 5.372675895690918,
+      "learning_rate": 1.7326038159371492e-05,
+      "loss": 0.2707,
+      "step": 5450
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 2.8858323097229004,
+      "learning_rate": 1.7290965207631877e-05,
+      "loss": 0.2729,
+      "step": 5455
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 3.638657569885254,
+      "learning_rate": 1.7255892255892255e-05,
+      "loss": 0.2336,
+      "step": 5460
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 5.720593452453613,
+      "learning_rate": 1.722081930415264e-05,
+      "loss": 0.2598,
+      "step": 5465
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 3.5562641620635986,
+      "learning_rate": 1.718574635241302e-05,
+      "loss": 0.204,
+      "step": 5470
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 4.291893005371094,
+      "learning_rate": 1.71506734006734e-05,
+      "loss": 0.2583,
+      "step": 5475
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 3.839714765548706,
+      "learning_rate": 1.7115600448933784e-05,
+      "loss": 0.2395,
+      "step": 5480
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 4.324798107147217,
+      "learning_rate": 1.7080527497194165e-05,
+      "loss": 0.2329,
+      "step": 5485
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 4.8277482986450195,
+      "learning_rate": 1.7045454545454546e-05,
+      "loss": 0.2224,
+      "step": 5490
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 6.462353706359863,
+      "learning_rate": 1.7010381593714927e-05,
+      "loss": 0.2772,
+      "step": 5495
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 5.617993354797363,
+      "learning_rate": 1.697530864197531e-05,
+      "loss": 0.2225,
+      "step": 5500
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 4.734283447265625,
+      "learning_rate": 1.6940235690235693e-05,
+      "loss": 0.2187,
+      "step": 5505
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 4.4282307624816895,
+      "learning_rate": 1.690516273849607e-05,
+      "loss": 0.2566,
+      "step": 5510
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 3.390618324279785,
+      "learning_rate": 1.6870089786756456e-05,
+      "loss": 0.2824,
+      "step": 5515
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 4.822201251983643,
+      "learning_rate": 1.6835016835016837e-05,
+      "loss": 0.2689,
+      "step": 5520
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 3.144552707672119,
+      "learning_rate": 1.6799943883277215e-05,
+      "loss": 0.3142,
+      "step": 5525
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 2.439606189727783,
+      "learning_rate": 1.67648709315376e-05,
+      "loss": 0.2189,
+      "step": 5530
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 3.6068406105041504,
+      "learning_rate": 1.672979797979798e-05,
+      "loss": 0.2547,
+      "step": 5535
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 4.948526859283447,
+      "learning_rate": 1.6694725028058362e-05,
+      "loss": 0.2645,
+      "step": 5540
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 7.270456790924072,
+      "learning_rate": 1.6659652076318744e-05,
+      "loss": 0.264,
+      "step": 5545
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9360078104198101,
+      "eval_loss": 0.19658659398555756,
+      "eval_runtime": 30.5973,
+      "eval_samples_per_second": 368.235,
+      "eval_steps_per_second": 11.537,
+      "step": 5545
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 4.378459453582764,
+      "learning_rate": 1.6624579124579125e-05,
+      "loss": 0.2207,
+      "step": 5550
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 2.6183969974517822,
+      "learning_rate": 1.6589506172839506e-05,
+      "loss": 0.2584,
+      "step": 5555
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 5.326751232147217,
+      "learning_rate": 1.6554433221099888e-05,
+      "loss": 0.2925,
+      "step": 5560
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 3.954202890396118,
+      "learning_rate": 1.6519360269360272e-05,
+      "loss": 0.2606,
+      "step": 5565
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 4.499910354614258,
+      "learning_rate": 1.6484287317620654e-05,
+      "loss": 0.3058,
+      "step": 5570
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 4.521121978759766,
+      "learning_rate": 1.644921436588103e-05,
+      "loss": 0.2819,
+      "step": 5575
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 5.335522174835205,
+      "learning_rate": 1.6414141414141416e-05,
+      "loss": 0.2528,
+      "step": 5580
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 3.814232587814331,
+      "learning_rate": 1.6379068462401797e-05,
+      "loss": 0.2193,
+      "step": 5585
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 5.540006637573242,
+      "learning_rate": 1.634399551066218e-05,
+      "loss": 0.262,
+      "step": 5590
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 4.523319721221924,
+      "learning_rate": 1.630892255892256e-05,
+      "loss": 0.2386,
+      "step": 5595
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 3.219916343688965,
+      "learning_rate": 1.627384960718294e-05,
+      "loss": 0.2565,
+      "step": 5600
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 4.485598564147949,
+      "learning_rate": 1.6238776655443323e-05,
+      "loss": 0.2365,
+      "step": 5605
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 4.061976909637451,
+      "learning_rate": 1.6203703703703704e-05,
+      "loss": 0.2371,
+      "step": 5610
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 5.030536651611328,
+      "learning_rate": 1.6168630751964085e-05,
+      "loss": 0.2003,
+      "step": 5615
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 3.8906493186950684,
+      "learning_rate": 1.6133557800224467e-05,
+      "loss": 0.2918,
+      "step": 5620
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 3.8694236278533936,
+      "learning_rate": 1.6098484848484848e-05,
+      "loss": 0.2723,
+      "step": 5625
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 5.940452575683594,
+      "learning_rate": 1.6063411896745232e-05,
+      "loss": 0.2855,
+      "step": 5630
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 3.4204578399658203,
+      "learning_rate": 1.6028338945005614e-05,
+      "loss": 0.2282,
+      "step": 5635
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 3.2205560207366943,
+      "learning_rate": 1.5993265993265995e-05,
+      "loss": 0.2388,
+      "step": 5640
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 3.111553907394409,
+      "learning_rate": 1.5958193041526376e-05,
+      "loss": 0.2118,
+      "step": 5645
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 4.048334121704102,
+      "learning_rate": 1.5923120089786758e-05,
+      "loss": 0.219,
+      "step": 5650
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 4.817101955413818,
+      "learning_rate": 1.588804713804714e-05,
+      "loss": 0.2814,
+      "step": 5655
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 3.2638566493988037,
+      "learning_rate": 1.585297418630752e-05,
+      "loss": 0.2466,
+      "step": 5660
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 3.2386622428894043,
+      "learning_rate": 1.58179012345679e-05,
+      "loss": 0.2243,
+      "step": 5665
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 4.803056240081787,
+      "learning_rate": 1.5782828282828283e-05,
+      "loss": 0.2663,
+      "step": 5670
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 3.9072980880737305,
+      "learning_rate": 1.5747755331088664e-05,
+      "loss": 0.2469,
+      "step": 5675
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 3.86452054977417,
+      "learning_rate": 1.571268237934905e-05,
+      "loss": 0.2452,
+      "step": 5680
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 4.956637382507324,
+      "learning_rate": 1.5677609427609427e-05,
+      "loss": 0.2648,
+      "step": 5685
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 4.016686916351318,
+      "learning_rate": 1.5642536475869808e-05,
+      "loss": 0.2004,
+      "step": 5690
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 3.1592161655426025,
+      "learning_rate": 1.5607463524130193e-05,
+      "loss": 0.2631,
+      "step": 5695
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 4.588578701019287,
+      "learning_rate": 1.5572390572390574e-05,
+      "loss": 0.2253,
+      "step": 5700
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 3.7832539081573486,
+      "learning_rate": 1.5537317620650955e-05,
+      "loss": 0.2003,
+      "step": 5705
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 4.404537200927734,
+      "learning_rate": 1.5502244668911337e-05,
+      "loss": 0.2359,
+      "step": 5710
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 4.948363780975342,
+      "learning_rate": 1.5467171717171718e-05,
+      "loss": 0.2465,
+      "step": 5715
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 4.368085861206055,
+      "learning_rate": 1.54320987654321e-05,
+      "loss": 0.2179,
+      "step": 5720
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 4.845777988433838,
+      "learning_rate": 1.539702581369248e-05,
+      "loss": 0.2548,
+      "step": 5725
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 8.992183685302734,
+      "learning_rate": 1.5361952861952865e-05,
+      "loss": 0.2025,
+      "step": 5730
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 3.767132520675659,
+      "learning_rate": 1.5326879910213243e-05,
+      "loss": 0.2935,
+      "step": 5735
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 3.4275033473968506,
+      "learning_rate": 1.5291806958473624e-05,
+      "loss": 0.2561,
+      "step": 5740
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 4.892788887023926,
+      "learning_rate": 1.5256734006734007e-05,
+      "loss": 0.2097,
+      "step": 5745
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 3.0885932445526123,
+      "learning_rate": 1.5221661054994389e-05,
+      "loss": 0.2691,
+      "step": 5750
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 4.2935638427734375,
+      "learning_rate": 1.518658810325477e-05,
+      "loss": 0.2446,
+      "step": 5755
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 4.711670875549316,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.2522,
+      "step": 5760
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 5.017344951629639,
+      "learning_rate": 1.5116442199775536e-05,
+      "loss": 0.2281,
+      "step": 5765
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 3.443549156188965,
+      "learning_rate": 1.5081369248035915e-05,
+      "loss": 0.2475,
+      "step": 5770
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 4.226526737213135,
+      "learning_rate": 1.5046296296296297e-05,
+      "loss": 0.2141,
+      "step": 5775
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 4.959639549255371,
+      "learning_rate": 1.501122334455668e-05,
+      "loss": 0.2701,
+      "step": 5780
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 5.746207237243652,
+      "learning_rate": 1.497615039281706e-05,
+      "loss": 0.2597,
+      "step": 5785
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 5.84602689743042,
+      "learning_rate": 1.4941077441077442e-05,
+      "loss": 0.2926,
+      "step": 5790
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 3.8785650730133057,
+      "learning_rate": 1.4906004489337824e-05,
+      "loss": 0.2758,
+      "step": 5795
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 5.078735828399658,
+      "learning_rate": 1.4870931537598203e-05,
+      "loss": 0.2638,
+      "step": 5800
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 4.840407371520996,
+      "learning_rate": 1.4835858585858586e-05,
+      "loss": 0.2448,
+      "step": 5805
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 3.1513493061065674,
+      "learning_rate": 1.480078563411897e-05,
+      "loss": 0.2848,
+      "step": 5810
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 3.3741588592529297,
+      "learning_rate": 1.4765712682379349e-05,
+      "loss": 0.1672,
+      "step": 5815
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 4.370522499084473,
+      "learning_rate": 1.473063973063973e-05,
+      "loss": 0.2356,
+      "step": 5820
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 5.121838569641113,
+      "learning_rate": 1.4695566778900113e-05,
+      "loss": 0.2645,
+      "step": 5825
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 5.192338943481445,
+      "learning_rate": 1.4660493827160496e-05,
+      "loss": 0.2491,
+      "step": 5830
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 4.110005855560303,
+      "learning_rate": 1.4625420875420876e-05,
+      "loss": 0.1904,
+      "step": 5835
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 5.020746231079102,
+      "learning_rate": 1.4590347923681259e-05,
+      "loss": 0.227,
+      "step": 5840
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 6.311166763305664,
+      "learning_rate": 1.455527497194164e-05,
+      "loss": 0.2609,
+      "step": 5845
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 4.739572048187256,
+      "learning_rate": 1.452020202020202e-05,
+      "loss": 0.2533,
+      "step": 5850
+    },
+    {
+      "epoch": 7.39,
+      "grad_norm": 2.9139962196350098,
+      "learning_rate": 1.4485129068462402e-05,
+      "loss": 0.2313,
+      "step": 5855
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 3.835667133331299,
+      "learning_rate": 1.4450056116722785e-05,
+      "loss": 0.2426,
+      "step": 5860
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 3.7305521965026855,
+      "learning_rate": 1.4414983164983165e-05,
+      "loss": 0.2111,
+      "step": 5865
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 4.593957901000977,
+      "learning_rate": 1.4379910213243546e-05,
+      "loss": 0.2086,
+      "step": 5870
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 5.136810779571533,
+      "learning_rate": 1.434483726150393e-05,
+      "loss": 0.2535,
+      "step": 5875
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 5.4645891189575195,
+      "learning_rate": 1.4309764309764309e-05,
+      "loss": 0.2316,
+      "step": 5880
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 4.172373294830322,
+      "learning_rate": 1.4274691358024692e-05,
+      "loss": 0.2368,
+      "step": 5885
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 3.749448537826538,
+      "learning_rate": 1.4239618406285073e-05,
+      "loss": 0.2411,
+      "step": 5890
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 5.06339168548584,
+      "learning_rate": 1.4204545454545456e-05,
+      "loss": 0.2343,
+      "step": 5895
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 4.473503112792969,
+      "learning_rate": 1.4169472502805836e-05,
+      "loss": 0.234,
+      "step": 5900
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 4.5991973876953125,
+      "learning_rate": 1.4134399551066219e-05,
+      "loss": 0.2515,
+      "step": 5905
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 5.038834095001221,
+      "learning_rate": 1.4099326599326602e-05,
+      "loss": 0.2594,
+      "step": 5910
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 4.8332319259643555,
+      "learning_rate": 1.4064253647586981e-05,
+      "loss": 0.1814,
+      "step": 5915
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 4.062576770782471,
+      "learning_rate": 1.4029180695847363e-05,
+      "loss": 0.2401,
+      "step": 5920
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 5.763060092926025,
+      "learning_rate": 1.3994107744107746e-05,
+      "loss": 0.2677,
+      "step": 5925
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 3.7424607276916504,
+      "learning_rate": 1.3959034792368125e-05,
+      "loss": 0.2129,
+      "step": 5930
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 3.7639622688293457,
+      "learning_rate": 1.3923961840628508e-05,
+      "loss": 0.2557,
+      "step": 5935
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 4.351307392120361,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.2599,
+      "step": 5940
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 3.547727346420288,
+      "learning_rate": 1.385381593714927e-05,
+      "loss": 0.2395,
+      "step": 5945
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 4.094538688659668,
+      "learning_rate": 1.3818742985409652e-05,
+      "loss": 0.2235,
+      "step": 5950
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 4.936663627624512,
+      "learning_rate": 1.3783670033670035e-05,
+      "loss": 0.2762,
+      "step": 5955
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 3.174994945526123,
+      "learning_rate": 1.3748597081930416e-05,
+      "loss": 0.201,
+      "step": 5960
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 8.465005874633789,
+      "learning_rate": 1.3713524130190796e-05,
+      "loss": 0.2336,
+      "step": 5965
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 4.044407367706299,
+      "learning_rate": 1.3678451178451179e-05,
+      "loss": 0.2409,
+      "step": 5970
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 3.7791435718536377,
+      "learning_rate": 1.3643378226711562e-05,
+      "loss": 0.2239,
+      "step": 5975
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 3.899807929992676,
+      "learning_rate": 1.3608305274971942e-05,
+      "loss": 0.2047,
+      "step": 5980
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 3.9796271324157715,
+      "learning_rate": 1.3573232323232325e-05,
+      "loss": 0.2508,
+      "step": 5985
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 4.528881549835205,
+      "learning_rate": 1.3538159371492706e-05,
+      "loss": 0.2265,
+      "step": 5990
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 5.062260150909424,
+      "learning_rate": 1.3503086419753085e-05,
+      "loss": 0.2737,
+      "step": 5995
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 3.58141827583313,
+      "learning_rate": 1.3468013468013468e-05,
+      "loss": 0.214,
+      "step": 6000
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 5.861845970153809,
+      "learning_rate": 1.3432940516273851e-05,
+      "loss": 0.2699,
+      "step": 6005
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 4.987063407897949,
+      "learning_rate": 1.3397867564534231e-05,
+      "loss": 0.2445,
+      "step": 6010
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 4.735630512237549,
+      "learning_rate": 1.3362794612794612e-05,
+      "loss": 0.2429,
+      "step": 6015
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 4.741174697875977,
+      "learning_rate": 1.3327721661054995e-05,
+      "loss": 0.23,
+      "step": 6020
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 4.381152153015137,
+      "learning_rate": 1.3292648709315378e-05,
+      "loss": 0.2805,
+      "step": 6025
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 4.9460225105285645,
+      "learning_rate": 1.3257575757575758e-05,
+      "loss": 0.2029,
+      "step": 6030
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 3.5638632774353027,
+      "learning_rate": 1.322250280583614e-05,
+      "loss": 0.2448,
+      "step": 6035
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 3.8605401515960693,
+      "learning_rate": 1.3187429854096522e-05,
+      "loss": 0.2503,
+      "step": 6040
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 4.938667297363281,
+      "learning_rate": 1.3152356902356902e-05,
+      "loss": 0.2737,
+      "step": 6045
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 4.475386619567871,
+      "learning_rate": 1.3117283950617285e-05,
+      "loss": 0.2499,
+      "step": 6050
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 6.9248433113098145,
+      "learning_rate": 1.3082210998877668e-05,
+      "loss": 0.2775,
+      "step": 6055
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 3.5063343048095703,
+      "learning_rate": 1.3047138047138047e-05,
+      "loss": 0.2482,
+      "step": 6060
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 3.5089409351348877,
+      "learning_rate": 1.3012065095398429e-05,
+      "loss": 0.2667,
+      "step": 6065
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 3.6856486797332764,
+      "learning_rate": 1.2976992143658812e-05,
+      "loss": 0.2353,
+      "step": 6070
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 4.717007160186768,
+      "learning_rate": 1.2941919191919191e-05,
+      "loss": 0.3042,
+      "step": 6075
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 5.4124274253845215,
+      "learning_rate": 1.2906846240179574e-05,
+      "loss": 0.2233,
+      "step": 6080
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 5.383946895599365,
+      "learning_rate": 1.2871773288439956e-05,
+      "loss": 0.2953,
+      "step": 6085
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 3.441504716873169,
+      "learning_rate": 1.2836700336700338e-05,
+      "loss": 0.2632,
+      "step": 6090
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 4.523129463195801,
+      "learning_rate": 1.2801627384960718e-05,
+      "loss": 0.2465,
+      "step": 6095
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 4.589935779571533,
+      "learning_rate": 1.2766554433221101e-05,
+      "loss": 0.2839,
+      "step": 6100
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 3.677739143371582,
+      "learning_rate": 1.2731481481481482e-05,
+      "loss": 0.241,
+      "step": 6105
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 4.748384952545166,
+      "learning_rate": 1.2696408529741862e-05,
+      "loss": 0.2094,
+      "step": 6110
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 3.1519172191619873,
+      "learning_rate": 1.2661335578002245e-05,
+      "loss": 0.2024,
+      "step": 6115
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 3.67272686958313,
+      "learning_rate": 1.2626262626262628e-05,
+      "loss": 0.2358,
+      "step": 6120
+    },
+    {
+      "epoch": 7.73,
+      "grad_norm": 5.061676502227783,
+      "learning_rate": 1.2591189674523008e-05,
+      "loss": 0.2828,
+      "step": 6125
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 4.950560092926025,
+      "learning_rate": 1.255611672278339e-05,
+      "loss": 0.2636,
+      "step": 6130
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 2.415231466293335,
+      "learning_rate": 1.2521043771043772e-05,
+      "loss": 0.2496,
+      "step": 6135
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 4.400854587554932,
+      "learning_rate": 1.2485970819304153e-05,
+      "loss": 0.2073,
+      "step": 6140
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 3.8155202865600586,
+      "learning_rate": 1.2450897867564534e-05,
+      "loss": 0.269,
+      "step": 6145
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 3.3153977394104004,
+      "learning_rate": 1.2415824915824917e-05,
+      "loss": 0.2143,
+      "step": 6150
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 4.120780944824219,
+      "learning_rate": 1.2380751964085299e-05,
+      "loss": 0.2568,
+      "step": 6155
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 3.529374837875366,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 0.1837,
+      "step": 6160
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 4.4600419998168945,
+      "learning_rate": 1.2310606060606061e-05,
+      "loss": 0.2657,
+      "step": 6165
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 6.646722793579102,
+      "learning_rate": 1.2275533108866443e-05,
+      "loss": 0.2236,
+      "step": 6170
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 5.6444220542907715,
+      "learning_rate": 1.2240460157126826e-05,
+      "loss": 0.1963,
+      "step": 6175
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 3.797537326812744,
+      "learning_rate": 1.2205387205387205e-05,
+      "loss": 0.2058,
+      "step": 6180
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 4.484612941741943,
+      "learning_rate": 1.2170314253647586e-05,
+      "loss": 0.2159,
+      "step": 6185
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 3.293269634246826,
+      "learning_rate": 1.213524130190797e-05,
+      "loss": 0.1979,
+      "step": 6190
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 3.680863857269287,
+      "learning_rate": 1.210016835016835e-05,
+      "loss": 0.2406,
+      "step": 6195
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 5.308590412139893,
+      "learning_rate": 1.2065095398428734e-05,
+      "loss": 0.2228,
+      "step": 6200
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 6.3553032875061035,
+      "learning_rate": 1.2030022446689113e-05,
+      "loss": 0.218,
+      "step": 6205
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 3.920201301574707,
+      "learning_rate": 1.1994949494949495e-05,
+      "loss": 0.212,
+      "step": 6210
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 3.0987179279327393,
+      "learning_rate": 1.1959876543209878e-05,
+      "loss": 0.2444,
+      "step": 6215
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 4.677021026611328,
+      "learning_rate": 1.1924803591470259e-05,
+      "loss": 0.2197,
+      "step": 6220
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 3.304238796234131,
+      "learning_rate": 1.188973063973064e-05,
+      "loss": 0.2784,
+      "step": 6225
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 4.291449546813965,
+      "learning_rate": 1.1854657687991021e-05,
+      "loss": 0.2329,
+      "step": 6230
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 4.73038911819458,
+      "learning_rate": 1.1819584736251403e-05,
+      "loss": 0.246,
+      "step": 6235
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 3.7323665618896484,
+      "learning_rate": 1.1784511784511786e-05,
+      "loss": 0.2441,
+      "step": 6240
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 4.738189697265625,
+      "learning_rate": 1.1749438832772167e-05,
+      "loss": 0.2484,
+      "step": 6245
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 6.337230682373047,
+      "learning_rate": 1.1714365881032548e-05,
+      "loss": 0.2331,
+      "step": 6250
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 3.0896143913269043,
+      "learning_rate": 1.167929292929293e-05,
+      "loss": 0.2343,
+      "step": 6255
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 5.781737327575684,
+      "learning_rate": 1.1644219977553311e-05,
+      "loss": 0.3054,
+      "step": 6260
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 4.307906150817871,
+      "learning_rate": 1.1609147025813694e-05,
+      "loss": 0.263,
+      "step": 6265
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 5.129984378814697,
+      "learning_rate": 1.1574074074074075e-05,
+      "loss": 0.2449,
+      "step": 6270
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 4.1641645431518555,
+      "learning_rate": 1.1539001122334456e-05,
+      "loss": 0.2198,
+      "step": 6275
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 4.486323356628418,
+      "learning_rate": 1.1503928170594838e-05,
+      "loss": 0.2521,
+      "step": 6280
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 4.946229934692383,
+      "learning_rate": 1.1468855218855219e-05,
+      "loss": 0.2334,
+      "step": 6285
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 4.629194736480713,
+      "learning_rate": 1.1433782267115602e-05,
+      "loss": 0.2692,
+      "step": 6290
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 4.051533222198486,
+      "learning_rate": 1.1398709315375983e-05,
+      "loss": 0.2114,
+      "step": 6295
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 3.8015079498291016,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 0.2008,
+      "step": 6300
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 4.475102424621582,
+      "learning_rate": 1.1328563411896746e-05,
+      "loss": 0.2452,
+      "step": 6305
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 5.8379740715026855,
+      "learning_rate": 1.1293490460157127e-05,
+      "loss": 0.1917,
+      "step": 6310
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 6.705021381378174,
+      "learning_rate": 1.1258417508417509e-05,
+      "loss": 0.1956,
+      "step": 6315
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 4.986030578613281,
+      "learning_rate": 1.1223344556677892e-05,
+      "loss": 0.2262,
+      "step": 6320
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 3.0361666679382324,
+      "learning_rate": 1.1188271604938271e-05,
+      "loss": 0.2653,
+      "step": 6325
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 3.705697774887085,
+      "learning_rate": 1.1153198653198654e-05,
+      "loss": 0.2214,
+      "step": 6330
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 3.4059932231903076,
+      "learning_rate": 1.1118125701459035e-05,
+      "loss": 0.2112,
+      "step": 6335
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9339664506967249,
+      "eval_loss": 0.19636236131191254,
+      "eval_runtime": 30.6925,
+      "eval_samples_per_second": 367.093,
+      "eval_steps_per_second": 11.501,
+      "step": 6338
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 3.763967514038086,
+      "learning_rate": 1.1083052749719417e-05,
+      "loss": 0.2139,
+      "step": 6340
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 3.921048641204834,
+      "learning_rate": 1.10479797979798e-05,
+      "loss": 0.2053,
+      "step": 6345
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 4.496042728424072,
+      "learning_rate": 1.101290684624018e-05,
+      "loss": 0.2398,
+      "step": 6350
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 5.23396635055542,
+      "learning_rate": 1.0977833894500562e-05,
+      "loss": 0.2395,
+      "step": 6355
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 3.9729983806610107,
+      "learning_rate": 1.0942760942760944e-05,
+      "loss": 0.2863,
+      "step": 6360
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 4.526211261749268,
+      "learning_rate": 1.0907687991021325e-05,
+      "loss": 0.2382,
+      "step": 6365
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 3.4816806316375732,
+      "learning_rate": 1.0872615039281706e-05,
+      "loss": 0.1844,
+      "step": 6370
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 3.5667312145233154,
+      "learning_rate": 1.0837542087542087e-05,
+      "loss": 0.1889,
+      "step": 6375
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 3.7927794456481934,
+      "learning_rate": 1.0802469135802469e-05,
+      "loss": 0.2186,
+      "step": 6380
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 3.523933172225952,
+      "learning_rate": 1.0767396184062852e-05,
+      "loss": 0.2387,
+      "step": 6385
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 6.282043933868408,
+      "learning_rate": 1.0732323232323233e-05,
+      "loss": 0.2677,
+      "step": 6390
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 4.3757429122924805,
+      "learning_rate": 1.0697250280583614e-05,
+      "loss": 0.2183,
+      "step": 6395
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 4.603021144866943,
+      "learning_rate": 1.0662177328843996e-05,
+      "loss": 0.2065,
+      "step": 6400
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 5.131260395050049,
+      "learning_rate": 1.0627104377104377e-05,
+      "loss": 0.2392,
+      "step": 6405
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 3.176405191421509,
+      "learning_rate": 1.059203142536476e-05,
+      "loss": 0.2095,
+      "step": 6410
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 4.559404373168945,
+      "learning_rate": 1.0556958473625141e-05,
+      "loss": 0.2014,
+      "step": 6415
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 3.7030699253082275,
+      "learning_rate": 1.0521885521885522e-05,
+      "loss": 0.1986,
+      "step": 6420
+    },
+    {
+      "epoch": 8.11,
+      "grad_norm": 4.664905071258545,
+      "learning_rate": 1.0486812570145904e-05,
+      "loss": 0.2277,
+      "step": 6425
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 4.896106719970703,
+      "learning_rate": 1.0451739618406285e-05,
+      "loss": 0.2134,
+      "step": 6430
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 4.307590484619141,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.2161,
+      "step": 6435
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 4.684503555297852,
+      "learning_rate": 1.038159371492705e-05,
+      "loss": 0.1894,
+      "step": 6440
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 4.339939594268799,
+      "learning_rate": 1.0346520763187429e-05,
+      "loss": 0.2469,
+      "step": 6445
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 4.4233598709106445,
+      "learning_rate": 1.0311447811447812e-05,
+      "loss": 0.1936,
+      "step": 6450
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 6.015918254852295,
+      "learning_rate": 1.0276374859708193e-05,
+      "loss": 0.2699,
+      "step": 6455
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 5.27017879486084,
+      "learning_rate": 1.0241301907968576e-05,
+      "loss": 0.2233,
+      "step": 6460
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 4.850338459014893,
+      "learning_rate": 1.0206228956228957e-05,
+      "loss": 0.2465,
+      "step": 6465
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 6.081806182861328,
+      "learning_rate": 1.0171156004489337e-05,
+      "loss": 0.2478,
+      "step": 6470
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 4.162698745727539,
+      "learning_rate": 1.013608305274972e-05,
+      "loss": 0.2141,
+      "step": 6475
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 5.176547527313232,
+      "learning_rate": 1.0101010101010101e-05,
+      "loss": 0.2155,
+      "step": 6480
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 3.4331626892089844,
+      "learning_rate": 1.0065937149270484e-05,
+      "loss": 0.2084,
+      "step": 6485
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 4.974352836608887,
+      "learning_rate": 1.0030864197530866e-05,
+      "loss": 0.2105,
+      "step": 6490
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 2.469017744064331,
+      "learning_rate": 9.995791245791245e-06,
+      "loss": 0.203,
+      "step": 6495
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 3.5954174995422363,
+      "learning_rate": 9.960718294051628e-06,
+      "loss": 0.2053,
+      "step": 6500
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 4.489570617675781,
+      "learning_rate": 9.92564534231201e-06,
+      "loss": 0.2119,
+      "step": 6505
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 3.7584943771362305,
+      "learning_rate": 9.89057239057239e-06,
+      "loss": 0.2301,
+      "step": 6510
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 4.778498649597168,
+      "learning_rate": 9.855499438832772e-06,
+      "loss": 0.2515,
+      "step": 6515
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 4.494474411010742,
+      "learning_rate": 9.820426487093153e-06,
+      "loss": 0.2281,
+      "step": 6520
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 3.4955692291259766,
+      "learning_rate": 9.785353535353536e-06,
+      "loss": 0.1935,
+      "step": 6525
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 5.162160396575928,
+      "learning_rate": 9.750280583613918e-06,
+      "loss": 0.2538,
+      "step": 6530
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 5.552359580993652,
+      "learning_rate": 9.715207631874299e-06,
+      "loss": 0.2136,
+      "step": 6535
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 4.608691215515137,
+      "learning_rate": 9.68013468013468e-06,
+      "loss": 0.2472,
+      "step": 6540
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 3.7796683311462402,
+      "learning_rate": 9.645061728395062e-06,
+      "loss": 0.2413,
+      "step": 6545
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 3.775463819503784,
+      "learning_rate": 9.609988776655445e-06,
+      "loss": 0.2054,
+      "step": 6550
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 3.659821033477783,
+      "learning_rate": 9.574915824915826e-06,
+      "loss": 0.2081,
+      "step": 6555
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 6.806145668029785,
+      "learning_rate": 9.539842873176207e-06,
+      "loss": 0.2037,
+      "step": 6560
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 4.858828067779541,
+      "learning_rate": 9.504769921436588e-06,
+      "loss": 0.2141,
+      "step": 6565
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 4.524445533752441,
+      "learning_rate": 9.46969696969697e-06,
+      "loss": 0.2022,
+      "step": 6570
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 2.968122959136963,
+      "learning_rate": 9.434624017957351e-06,
+      "loss": 0.181,
+      "step": 6575
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 3.753946542739868,
+      "learning_rate": 9.399551066217734e-06,
+      "loss": 0.2077,
+      "step": 6580
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 5.309144973754883,
+      "learning_rate": 9.364478114478115e-06,
+      "loss": 0.216,
+      "step": 6585
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 6.1520915031433105,
+      "learning_rate": 9.329405162738497e-06,
+      "loss": 0.2135,
+      "step": 6590
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 3.9922449588775635,
+      "learning_rate": 9.294332210998878e-06,
+      "loss": 0.2449,
+      "step": 6595
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 5.85723352432251,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.2171,
+      "step": 6600
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 4.760324954986572,
+      "learning_rate": 9.224186307519642e-06,
+      "loss": 0.2476,
+      "step": 6605
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 4.929134368896484,
+      "learning_rate": 9.189113355780023e-06,
+      "loss": 0.2313,
+      "step": 6610
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 4.692222595214844,
+      "learning_rate": 9.154040404040405e-06,
+      "loss": 0.2596,
+      "step": 6615
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 3.903709888458252,
+      "learning_rate": 9.118967452300786e-06,
+      "loss": 0.2269,
+      "step": 6620
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 3.900644063949585,
+      "learning_rate": 9.083894500561167e-06,
+      "loss": 0.267,
+      "step": 6625
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 6.073973655700684,
+      "learning_rate": 9.04882154882155e-06,
+      "loss": 0.2433,
+      "step": 6630
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 3.9931607246398926,
+      "learning_rate": 9.013748597081932e-06,
+      "loss": 0.2311,
+      "step": 6635
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 3.873122453689575,
+      "learning_rate": 8.978675645342311e-06,
+      "loss": 0.1948,
+      "step": 6640
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 4.756349563598633,
+      "learning_rate": 8.943602693602694e-06,
+      "loss": 0.2286,
+      "step": 6645
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 3.753122091293335,
+      "learning_rate": 8.908529741863075e-06,
+      "loss": 0.2294,
+      "step": 6650
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 3.135073184967041,
+      "learning_rate": 8.873456790123458e-06,
+      "loss": 0.2049,
+      "step": 6655
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 6.234099388122559,
+      "learning_rate": 8.838383838383838e-06,
+      "loss": 0.189,
+      "step": 6660
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 4.082734107971191,
+      "learning_rate": 8.80331088664422e-06,
+      "loss": 0.241,
+      "step": 6665
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 3.6797385215759277,
+      "learning_rate": 8.768237934904602e-06,
+      "loss": 0.2375,
+      "step": 6670
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 4.899085998535156,
+      "learning_rate": 8.733164983164984e-06,
+      "loss": 0.2462,
+      "step": 6675
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 4.697770118713379,
+      "learning_rate": 8.698092031425367e-06,
+      "loss": 0.2699,
+      "step": 6680
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 3.9699809551239014,
+      "learning_rate": 8.663019079685746e-06,
+      "loss": 0.2228,
+      "step": 6685
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 4.850160121917725,
+      "learning_rate": 8.627946127946127e-06,
+      "loss": 0.2258,
+      "step": 6690
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 4.5991435050964355,
+      "learning_rate": 8.59287317620651e-06,
+      "loss": 0.2292,
+      "step": 6695
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 4.3139729499816895,
+      "learning_rate": 8.557800224466892e-06,
+      "loss": 0.2515,
+      "step": 6700
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 4.702777862548828,
+      "learning_rate": 8.522727272727273e-06,
+      "loss": 0.2783,
+      "step": 6705
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 4.994075775146484,
+      "learning_rate": 8.487654320987654e-06,
+      "loss": 0.2642,
+      "step": 6710
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 5.382604598999023,
+      "learning_rate": 8.452581369248036e-06,
+      "loss": 0.2481,
+      "step": 6715
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 4.709623336791992,
+      "learning_rate": 8.417508417508419e-06,
+      "loss": 0.2354,
+      "step": 6720
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 4.874531269073486,
+      "learning_rate": 8.3824354657688e-06,
+      "loss": 0.2405,
+      "step": 6725
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 2.4052963256835938,
+      "learning_rate": 8.347362514029181e-06,
+      "loss": 0.2146,
+      "step": 6730
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 4.033393859863281,
+      "learning_rate": 8.312289562289563e-06,
+      "loss": 0.2354,
+      "step": 6735
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 5.043489456176758,
+      "learning_rate": 8.277216610549944e-06,
+      "loss": 0.2828,
+      "step": 6740
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 2.287036418914795,
+      "learning_rate": 8.242143658810327e-06,
+      "loss": 0.215,
+      "step": 6745
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 4.0818257331848145,
+      "learning_rate": 8.207070707070708e-06,
+      "loss": 0.2178,
+      "step": 6750
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 3.8903939723968506,
+      "learning_rate": 8.17199775533109e-06,
+      "loss": 0.2404,
+      "step": 6755
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 4.213070392608643,
+      "learning_rate": 8.13692480359147e-06,
+      "loss": 0.2031,
+      "step": 6760
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 6.595643997192383,
+      "learning_rate": 8.101851851851852e-06,
+      "loss": 0.2362,
+      "step": 6765
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 7.220956802368164,
+      "learning_rate": 8.066778900112233e-06,
+      "loss": 0.1888,
+      "step": 6770
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 3.939709186553955,
+      "learning_rate": 8.031705948372616e-06,
+      "loss": 0.2153,
+      "step": 6775
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 3.0982513427734375,
+      "learning_rate": 7.996632996632998e-06,
+      "loss": 0.2229,
+      "step": 6780
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 4.180152416229248,
+      "learning_rate": 7.961560044893379e-06,
+      "loss": 0.2441,
+      "step": 6785
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 4.217372894287109,
+      "learning_rate": 7.92648709315376e-06,
+      "loss": 0.2147,
+      "step": 6790
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 8.51630973815918,
+      "learning_rate": 7.891414141414141e-06,
+      "loss": 0.249,
+      "step": 6795
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 3.5144059658050537,
+      "learning_rate": 7.856341189674524e-06,
+      "loss": 0.2197,
+      "step": 6800
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 4.652500152587891,
+      "learning_rate": 7.821268237934904e-06,
+      "loss": 0.1941,
+      "step": 6805
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 3.596229076385498,
+      "learning_rate": 7.786195286195287e-06,
+      "loss": 0.2038,
+      "step": 6810
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 4.538928031921387,
+      "learning_rate": 7.751122334455668e-06,
+      "loss": 0.2451,
+      "step": 6815
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 4.045384883880615,
+      "learning_rate": 7.71604938271605e-06,
+      "loss": 0.2267,
+      "step": 6820
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 4.464452266693115,
+      "learning_rate": 7.680976430976433e-06,
+      "loss": 0.2692,
+      "step": 6825
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 3.7945337295532227,
+      "learning_rate": 7.645903479236812e-06,
+      "loss": 0.2383,
+      "step": 6830
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 3.41070818901062,
+      "learning_rate": 7.610830527497194e-06,
+      "loss": 0.2674,
+      "step": 6835
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 3.7921557426452637,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.1982,
+      "step": 6840
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 3.909104824066162,
+      "learning_rate": 7.540684624017958e-06,
+      "loss": 0.2757,
+      "step": 6845
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 5.502076148986816,
+      "learning_rate": 7.50561167227834e-06,
+      "loss": 0.2631,
+      "step": 6850
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 3.835015296936035,
+      "learning_rate": 7.470538720538721e-06,
+      "loss": 0.2224,
+      "step": 6855
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 4.8051981925964355,
+      "learning_rate": 7.435465768799102e-06,
+      "loss": 0.2218,
+      "step": 6860
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 3.487741708755493,
+      "learning_rate": 7.400392817059485e-06,
+      "loss": 0.2544,
+      "step": 6865
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 4.873679161071777,
+      "learning_rate": 7.365319865319865e-06,
+      "loss": 0.2444,
+      "step": 6870
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 3.686678409576416,
+      "learning_rate": 7.330246913580248e-06,
+      "loss": 0.2339,
+      "step": 6875
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 5.513541221618652,
+      "learning_rate": 7.295173961840629e-06,
+      "loss": 0.2217,
+      "step": 6880
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 3.6160805225372314,
+      "learning_rate": 7.26010101010101e-06,
+      "loss": 0.2233,
+      "step": 6885
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 4.07836389541626,
+      "learning_rate": 7.225028058361393e-06,
+      "loss": 0.227,
+      "step": 6890
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 4.27037239074707,
+      "learning_rate": 7.189955106621773e-06,
+      "loss": 0.2357,
+      "step": 6895
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 4.137974739074707,
+      "learning_rate": 7.1548821548821545e-06,
+      "loss": 0.232,
+      "step": 6900
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 4.392239570617676,
+      "learning_rate": 7.119809203142537e-06,
+      "loss": 0.2315,
+      "step": 6905
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 3.3127002716064453,
+      "learning_rate": 7.084736251402918e-06,
+      "loss": 0.1945,
+      "step": 6910
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 3.7440578937530518,
+      "learning_rate": 7.049663299663301e-06,
+      "loss": 0.1997,
+      "step": 6915
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 6.077698707580566,
+      "learning_rate": 7.014590347923681e-06,
+      "loss": 0.2577,
+      "step": 6920
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 4.052228927612305,
+      "learning_rate": 6.979517396184063e-06,
+      "loss": 0.2184,
+      "step": 6925
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 3.750011920928955,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.1776,
+      "step": 6930
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 3.9637081623077393,
+      "learning_rate": 6.909371492704826e-06,
+      "loss": 0.2195,
+      "step": 6935
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 4.1828694343566895,
+      "learning_rate": 6.874298540965208e-06,
+      "loss": 0.2448,
+      "step": 6940
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 5.8484416007995605,
+      "learning_rate": 6.8392255892255895e-06,
+      "loss": 0.299,
+      "step": 6945
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 5.047119140625,
+      "learning_rate": 6.804152637485971e-06,
+      "loss": 0.2631,
+      "step": 6950
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 4.096033573150635,
+      "learning_rate": 6.769079685746353e-06,
+      "loss": 0.2167,
+      "step": 6955
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 4.195711135864258,
+      "learning_rate": 6.734006734006734e-06,
+      "loss": 0.2297,
+      "step": 6960
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 3.8289475440979004,
+      "learning_rate": 6.6989337822671155e-06,
+      "loss": 0.2179,
+      "step": 6965
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 3.401686429977417,
+      "learning_rate": 6.663860830527498e-06,
+      "loss": 0.2043,
+      "step": 6970
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 4.296989917755127,
+      "learning_rate": 6.628787878787879e-06,
+      "loss": 0.2303,
+      "step": 6975
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 4.737481594085693,
+      "learning_rate": 6.593714927048261e-06,
+      "loss": 0.2366,
+      "step": 6980
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 3.570190191268921,
+      "learning_rate": 6.558641975308642e-06,
+      "loss": 0.2343,
+      "step": 6985
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 3.296848773956299,
+      "learning_rate": 6.523569023569024e-06,
+      "loss": 0.2027,
+      "step": 6990
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 3.2430880069732666,
+      "learning_rate": 6.488496071829406e-06,
+      "loss": 0.2393,
+      "step": 6995
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 2.606722354888916,
+      "learning_rate": 6.453423120089787e-06,
+      "loss": 0.2182,
+      "step": 7000
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 2.176530599594116,
+      "learning_rate": 6.418350168350169e-06,
+      "loss": 0.1735,
+      "step": 7005
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 3.334801197052002,
+      "learning_rate": 6.3832772166105505e-06,
+      "loss": 0.2543,
+      "step": 7010
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 5.0535712242126465,
+      "learning_rate": 6.348204264870931e-06,
+      "loss": 0.2458,
+      "step": 7015
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 4.359066009521484,
+      "learning_rate": 6.313131313131314e-06,
+      "loss": 0.2181,
+      "step": 7020
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 3.5706419944763184,
+      "learning_rate": 6.278058361391695e-06,
+      "loss": 0.2164,
+      "step": 7025
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 4.484238147735596,
+      "learning_rate": 6.2429854096520766e-06,
+      "loss": 0.2284,
+      "step": 7030
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 3.5250422954559326,
+      "learning_rate": 6.207912457912459e-06,
+      "loss": 0.2315,
+      "step": 7035
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 6.015649795532227,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 0.2277,
+      "step": 7040
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 5.115208625793457,
+      "learning_rate": 6.137766554433221e-06,
+      "loss": 0.2247,
+      "step": 7045
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 4.147441387176514,
+      "learning_rate": 6.1026936026936026e-06,
+      "loss": 0.2206,
+      "step": 7050
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 4.62777042388916,
+      "learning_rate": 6.067620650953985e-06,
+      "loss": 0.2347,
+      "step": 7055
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 5.681117057800293,
+      "learning_rate": 6.032547699214367e-06,
+      "loss": 0.2375,
+      "step": 7060
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 3.498126268386841,
+      "learning_rate": 5.997474747474747e-06,
+      "loss": 0.2592,
+      "step": 7065
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 4.412965297698975,
+      "learning_rate": 5.9624017957351294e-06,
+      "loss": 0.2171,
+      "step": 7070
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 4.652202606201172,
+      "learning_rate": 5.927328843995511e-06,
+      "loss": 0.2692,
+      "step": 7075
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 4.772882461547852,
+      "learning_rate": 5.892255892255893e-06,
+      "loss": 0.2087,
+      "step": 7080
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 2.609529495239258,
+      "learning_rate": 5.857182940516274e-06,
+      "loss": 0.2169,
+      "step": 7085
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 5.504788398742676,
+      "learning_rate": 5.8221099887766555e-06,
+      "loss": 0.227,
+      "step": 7090
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 6.363174915313721,
+      "learning_rate": 5.787037037037038e-06,
+      "loss": 0.2547,
+      "step": 7095
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 3.957911729812622,
+      "learning_rate": 5.751964085297419e-06,
+      "loss": 0.2115,
+      "step": 7100
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 4.692325592041016,
+      "learning_rate": 5.716891133557801e-06,
+      "loss": 0.2466,
+      "step": 7105
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 4.724496841430664,
+      "learning_rate": 5.681818181818182e-06,
+      "loss": 0.2277,
+      "step": 7110
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 4.038902282714844,
+      "learning_rate": 5.646745230078564e-06,
+      "loss": 0.2101,
+      "step": 7115
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 3.702584981918335,
+      "learning_rate": 5.611672278338946e-06,
+      "loss": 0.2237,
+      "step": 7120
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 3.163513660430908,
+      "learning_rate": 5.576599326599327e-06,
+      "loss": 0.2489,
+      "step": 7125
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 5.134917736053467,
+      "learning_rate": 5.541526374859708e-06,
+      "loss": 0.2716,
+      "step": 7130
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9350315079435519,
+      "eval_loss": 0.19689346849918365,
+      "eval_runtime": 30.592,
+      "eval_samples_per_second": 368.298,
+      "eval_steps_per_second": 11.539,
+      "step": 7130
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 3.693969964981079,
+      "learning_rate": 5.50645342312009e-06,
+      "loss": 0.2216,
+      "step": 7135
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 4.428549766540527,
+      "learning_rate": 5.471380471380472e-06,
+      "loss": 0.2392,
+      "step": 7140
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 4.35985803604126,
+      "learning_rate": 5.436307519640853e-06,
+      "loss": 0.2004,
+      "step": 7145
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 3.040268898010254,
+      "learning_rate": 5.401234567901234e-06,
+      "loss": 0.1953,
+      "step": 7150
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 3.617007255554199,
+      "learning_rate": 5.3661616161616165e-06,
+      "loss": 0.2437,
+      "step": 7155
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 4.108349800109863,
+      "learning_rate": 5.331088664421998e-06,
+      "loss": 0.2216,
+      "step": 7160
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 3.3783700466156006,
+      "learning_rate": 5.29601571268238e-06,
+      "loss": 0.2061,
+      "step": 7165
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 4.007216930389404,
+      "learning_rate": 5.260942760942761e-06,
+      "loss": 0.2064,
+      "step": 7170
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 8.224294662475586,
+      "learning_rate": 5.2258698092031425e-06,
+      "loss": 0.2101,
+      "step": 7175
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 3.3103721141815186,
+      "learning_rate": 5.190796857463525e-06,
+      "loss": 0.2194,
+      "step": 7180
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 3.5700058937072754,
+      "learning_rate": 5.155723905723906e-06,
+      "loss": 0.1869,
+      "step": 7185
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 3.387657642364502,
+      "learning_rate": 5.120650953984288e-06,
+      "loss": 0.2251,
+      "step": 7190
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 3.176109790802002,
+      "learning_rate": 5.0855780022446685e-06,
+      "loss": 0.2031,
+      "step": 7195
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 6.948771953582764,
+      "learning_rate": 5.050505050505051e-06,
+      "loss": 0.2642,
+      "step": 7200
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 6.0026936531066895,
+      "learning_rate": 5.015432098765433e-06,
+      "loss": 0.2378,
+      "step": 7205
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 2.5248005390167236,
+      "learning_rate": 4.980359147025814e-06,
+      "loss": 0.2127,
+      "step": 7210
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 3.726877212524414,
+      "learning_rate": 4.945286195286195e-06,
+      "loss": 0.2236,
+      "step": 7215
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 3.999799966812134,
+      "learning_rate": 4.910213243546577e-06,
+      "loss": 0.24,
+      "step": 7220
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 7.589893341064453,
+      "learning_rate": 4.875140291806959e-06,
+      "loss": 0.2157,
+      "step": 7225
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 5.575037479400635,
+      "learning_rate": 4.84006734006734e-06,
+      "loss": 0.27,
+      "step": 7230
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 3.9892919063568115,
+      "learning_rate": 4.804994388327722e-06,
+      "loss": 0.2545,
+      "step": 7235
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 5.8539204597473145,
+      "learning_rate": 4.7699214365881036e-06,
+      "loss": 0.2339,
+      "step": 7240
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 4.872870922088623,
+      "learning_rate": 4.734848484848485e-06,
+      "loss": 0.2257,
+      "step": 7245
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 6.796045780181885,
+      "learning_rate": 4.699775533108867e-06,
+      "loss": 0.2549,
+      "step": 7250
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 4.447669982910156,
+      "learning_rate": 4.664702581369248e-06,
+      "loss": 0.2361,
+      "step": 7255
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 7.638195037841797,
+      "learning_rate": 4.6296296296296296e-06,
+      "loss": 0.1995,
+      "step": 7260
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 3.5221500396728516,
+      "learning_rate": 4.594556677890012e-06,
+      "loss": 0.1817,
+      "step": 7265
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 4.982468605041504,
+      "learning_rate": 4.559483726150393e-06,
+      "loss": 0.1782,
+      "step": 7270
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 3.861450672149658,
+      "learning_rate": 4.524410774410775e-06,
+      "loss": 0.2261,
+      "step": 7275
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 5.906322479248047,
+      "learning_rate": 4.489337822671156e-06,
+      "loss": 0.2211,
+      "step": 7280
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 3.9798150062561035,
+      "learning_rate": 4.454264870931538e-06,
+      "loss": 0.2121,
+      "step": 7285
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 4.5946574211120605,
+      "learning_rate": 4.419191919191919e-06,
+      "loss": 0.1983,
+      "step": 7290
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 2.7848196029663086,
+      "learning_rate": 4.384118967452301e-06,
+      "loss": 0.2093,
+      "step": 7295
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 4.086716651916504,
+      "learning_rate": 4.349046015712683e-06,
+      "loss": 0.2656,
+      "step": 7300
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 3.784971237182617,
+      "learning_rate": 4.313973063973064e-06,
+      "loss": 0.1955,
+      "step": 7305
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 4.028592109680176,
+      "learning_rate": 4.278900112233446e-06,
+      "loss": 0.2244,
+      "step": 7310
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 4.5212860107421875,
+      "learning_rate": 4.243827160493827e-06,
+      "loss": 0.2344,
+      "step": 7315
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 5.080355644226074,
+      "learning_rate": 4.208754208754209e-06,
+      "loss": 0.234,
+      "step": 7320
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 5.625095844268799,
+      "learning_rate": 4.173681257014591e-06,
+      "loss": 0.229,
+      "step": 7325
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 3.1823513507843018,
+      "learning_rate": 4.138608305274972e-06,
+      "loss": 0.1844,
+      "step": 7330
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 3.4595417976379395,
+      "learning_rate": 4.103535353535354e-06,
+      "loss": 0.2185,
+      "step": 7335
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 4.2038092613220215,
+      "learning_rate": 4.068462401795735e-06,
+      "loss": 0.2608,
+      "step": 7340
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 4.772618770599365,
+      "learning_rate": 4.033389450056117e-06,
+      "loss": 0.1781,
+      "step": 7345
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 3.737285852432251,
+      "learning_rate": 3.998316498316499e-06,
+      "loss": 0.2849,
+      "step": 7350
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 3.705977201461792,
+      "learning_rate": 3.96324354657688e-06,
+      "loss": 0.215,
+      "step": 7355
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 2.9728264808654785,
+      "learning_rate": 3.928170594837262e-06,
+      "loss": 0.1991,
+      "step": 7360
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 5.008718490600586,
+      "learning_rate": 3.8930976430976435e-06,
+      "loss": 0.1982,
+      "step": 7365
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 3.1645705699920654,
+      "learning_rate": 3.858024691358025e-06,
+      "loss": 0.1939,
+      "step": 7370
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 4.226245880126953,
+      "learning_rate": 3.822951739618406e-06,
+      "loss": 0.2325,
+      "step": 7375
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 3.5430185794830322,
+      "learning_rate": 3.7878787878787882e-06,
+      "loss": 0.2228,
+      "step": 7380
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 4.020698070526123,
+      "learning_rate": 3.75280583613917e-06,
+      "loss": 0.2776,
+      "step": 7385
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 4.468031406402588,
+      "learning_rate": 3.717732884399551e-06,
+      "loss": 0.2172,
+      "step": 7390
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 2.731644868850708,
+      "learning_rate": 3.6826599326599325e-06,
+      "loss": 0.247,
+      "step": 7395
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 4.782643795013428,
+      "learning_rate": 3.6475869809203147e-06,
+      "loss": 0.2133,
+      "step": 7400
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 5.192986488342285,
+      "learning_rate": 3.6125140291806964e-06,
+      "loss": 0.2052,
+      "step": 7405
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 4.026821136474609,
+      "learning_rate": 3.5774410774410772e-06,
+      "loss": 0.2063,
+      "step": 7410
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 3.378051996231079,
+      "learning_rate": 3.542368125701459e-06,
+      "loss": 0.2619,
+      "step": 7415
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 4.427807807922363,
+      "learning_rate": 3.5072951739618407e-06,
+      "loss": 0.2071,
+      "step": 7420
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 5.541877269744873,
+      "learning_rate": 3.4722222222222224e-06,
+      "loss": 0.2285,
+      "step": 7425
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 5.540366172790527,
+      "learning_rate": 3.437149270482604e-06,
+      "loss": 0.205,
+      "step": 7430
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 4.237565517425537,
+      "learning_rate": 3.4020763187429854e-06,
+      "loss": 0.2446,
+      "step": 7435
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 4.471877098083496,
+      "learning_rate": 3.367003367003367e-06,
+      "loss": 0.2052,
+      "step": 7440
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 3.985455274581909,
+      "learning_rate": 3.331930415263749e-06,
+      "loss": 0.225,
+      "step": 7445
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 4.494751930236816,
+      "learning_rate": 3.2968574635241305e-06,
+      "loss": 0.2327,
+      "step": 7450
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 3.907879114151001,
+      "learning_rate": 3.261784511784512e-06,
+      "loss": 0.2233,
+      "step": 7455
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 3.090066432952881,
+      "learning_rate": 3.2267115600448936e-06,
+      "loss": 0.183,
+      "step": 7460
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 2.8341875076293945,
+      "learning_rate": 3.1916386083052753e-06,
+      "loss": 0.2292,
+      "step": 7465
+    },
+    {
+      "epoch": 9.43,
+      "grad_norm": 5.143249034881592,
+      "learning_rate": 3.156565656565657e-06,
+      "loss": 0.1985,
+      "step": 7470
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 4.155516624450684,
+      "learning_rate": 3.1214927048260383e-06,
+      "loss": 0.232,
+      "step": 7475
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 4.792262077331543,
+      "learning_rate": 3.0864197530864196e-06,
+      "loss": 0.2638,
+      "step": 7480
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 4.730192184448242,
+      "learning_rate": 3.0513468013468013e-06,
+      "loss": 0.2117,
+      "step": 7485
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 3.7558236122131348,
+      "learning_rate": 3.0162738496071834e-06,
+      "loss": 0.1972,
+      "step": 7490
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 3.852829933166504,
+      "learning_rate": 2.9812008978675647e-06,
+      "loss": 0.2209,
+      "step": 7495
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 2.8247456550598145,
+      "learning_rate": 2.9461279461279464e-06,
+      "loss": 0.2484,
+      "step": 7500
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 3.4103751182556152,
+      "learning_rate": 2.9110549943883277e-06,
+      "loss": 0.2088,
+      "step": 7505
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 5.389248847961426,
+      "learning_rate": 2.8759820426487094e-06,
+      "loss": 0.2363,
+      "step": 7510
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 1.9696253538131714,
+      "learning_rate": 2.840909090909091e-06,
+      "loss": 0.1722,
+      "step": 7515
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 5.2949957847595215,
+      "learning_rate": 2.805836139169473e-06,
+      "loss": 0.2351,
+      "step": 7520
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 4.246339797973633,
+      "learning_rate": 2.770763187429854e-06,
+      "loss": 0.2195,
+      "step": 7525
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 6.840764045715332,
+      "learning_rate": 2.735690235690236e-06,
+      "loss": 0.28,
+      "step": 7530
+    },
+    {
+      "epoch": 9.51,
+      "grad_norm": 3.876375198364258,
+      "learning_rate": 2.700617283950617e-06,
+      "loss": 0.2086,
+      "step": 7535
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 4.135307312011719,
+      "learning_rate": 2.665544332210999e-06,
+      "loss": 0.2252,
+      "step": 7540
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 3.7665975093841553,
+      "learning_rate": 2.6304713804713806e-06,
+      "loss": 0.2262,
+      "step": 7545
+    },
+    {
+      "epoch": 9.53,
+      "grad_norm": 5.482239246368408,
+      "learning_rate": 2.5953984287317623e-06,
+      "loss": 0.2329,
+      "step": 7550
+    },
+    {
+      "epoch": 9.54,
+      "grad_norm": 4.878592491149902,
+      "learning_rate": 2.560325476992144e-06,
+      "loss": 0.2065,
+      "step": 7555
+    },
+    {
+      "epoch": 9.54,
+      "grad_norm": 4.114917755126953,
+      "learning_rate": 2.5252525252525253e-06,
+      "loss": 0.222,
+      "step": 7560
+    },
+    {
+      "epoch": 9.55,
+      "grad_norm": 3.008373737335205,
+      "learning_rate": 2.490179573512907e-06,
+      "loss": 0.2433,
+      "step": 7565
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 3.457642078399658,
+      "learning_rate": 2.4551066217732883e-06,
+      "loss": 0.1743,
+      "step": 7570
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 3.45147705078125,
+      "learning_rate": 2.42003367003367e-06,
+      "loss": 0.206,
+      "step": 7575
+    },
+    {
+      "epoch": 9.57,
+      "grad_norm": 4.917750835418701,
+      "learning_rate": 2.3849607182940518e-06,
+      "loss": 0.2543,
+      "step": 7580
+    },
+    {
+      "epoch": 9.57,
+      "grad_norm": 4.047665596008301,
+      "learning_rate": 2.3498877665544335e-06,
+      "loss": 0.2035,
+      "step": 7585
+    },
+    {
+      "epoch": 9.58,
+      "grad_norm": 4.722668170928955,
+      "learning_rate": 2.3148148148148148e-06,
+      "loss": 0.2181,
+      "step": 7590
+    },
+    {
+      "epoch": 9.59,
+      "grad_norm": 3.802427053451538,
+      "learning_rate": 2.2797418630751965e-06,
+      "loss": 0.2177,
+      "step": 7595
+    },
+    {
+      "epoch": 9.59,
+      "grad_norm": 3.1694014072418213,
+      "learning_rate": 2.244668911335578e-06,
+      "loss": 0.1918,
+      "step": 7600
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 4.125642776489258,
+      "learning_rate": 2.2095959595959595e-06,
+      "loss": 0.2271,
+      "step": 7605
+    },
+    {
+      "epoch": 9.61,
+      "grad_norm": 3.856869697570801,
+      "learning_rate": 2.1745230078563416e-06,
+      "loss": 0.227,
+      "step": 7610
+    },
+    {
+      "epoch": 9.61,
+      "grad_norm": 3.888904094696045,
+      "learning_rate": 2.139450056116723e-06,
+      "loss": 0.2026,
+      "step": 7615
+    },
+    {
+      "epoch": 9.62,
+      "grad_norm": 4.0479559898376465,
+      "learning_rate": 2.1043771043771047e-06,
+      "loss": 0.2405,
+      "step": 7620
+    },
+    {
+      "epoch": 9.62,
+      "grad_norm": 4.43900203704834,
+      "learning_rate": 2.069304152637486e-06,
+      "loss": 0.1956,
+      "step": 7625
+    },
+    {
+      "epoch": 9.63,
+      "grad_norm": 3.870103359222412,
+      "learning_rate": 2.0342312008978677e-06,
+      "loss": 0.2279,
+      "step": 7630
+    },
+    {
+      "epoch": 9.64,
+      "grad_norm": 4.76093053817749,
+      "learning_rate": 1.9991582491582494e-06,
+      "loss": 0.2402,
+      "step": 7635
+    },
+    {
+      "epoch": 9.64,
+      "grad_norm": 5.051716327667236,
+      "learning_rate": 1.964085297418631e-06,
+      "loss": 0.2125,
+      "step": 7640
+    },
+    {
+      "epoch": 9.65,
+      "grad_norm": 3.5321850776672363,
+      "learning_rate": 1.9290123456790124e-06,
+      "loss": 0.2201,
+      "step": 7645
+    },
+    {
+      "epoch": 9.66,
+      "grad_norm": 3.247424840927124,
+      "learning_rate": 1.8939393939393941e-06,
+      "loss": 0.2298,
+      "step": 7650
+    },
+    {
+      "epoch": 9.66,
+      "grad_norm": 3.185377597808838,
+      "learning_rate": 1.8588664421997754e-06,
+      "loss": 0.2168,
+      "step": 7655
+    },
+    {
+      "epoch": 9.67,
+      "grad_norm": 3.3158586025238037,
+      "learning_rate": 1.8237934904601573e-06,
+      "loss": 0.1928,
+      "step": 7660
+    },
+    {
+      "epoch": 9.67,
+      "grad_norm": 6.204581260681152,
+      "learning_rate": 1.7887205387205386e-06,
+      "loss": 0.2298,
+      "step": 7665
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 6.48652982711792,
+      "learning_rate": 1.7536475869809203e-06,
+      "loss": 0.2061,
+      "step": 7670
+    },
+    {
+      "epoch": 9.69,
+      "grad_norm": 4.5980610847473145,
+      "learning_rate": 1.718574635241302e-06,
+      "loss": 0.2036,
+      "step": 7675
+    },
+    {
+      "epoch": 9.69,
+      "grad_norm": 3.362778663635254,
+      "learning_rate": 1.6835016835016836e-06,
+      "loss": 0.189,
+      "step": 7680
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 3.340287208557129,
+      "learning_rate": 1.6484287317620653e-06,
+      "loss": 0.2157,
+      "step": 7685
+    },
+    {
+      "epoch": 9.71,
+      "grad_norm": 5.877618789672852,
+      "learning_rate": 1.6133557800224468e-06,
+      "loss": 0.2393,
+      "step": 7690
+    },
+    {
+      "epoch": 9.71,
+      "grad_norm": 3.4454119205474854,
+      "learning_rate": 1.5782828282828285e-06,
+      "loss": 0.2319,
+      "step": 7695
+    },
+    {
+      "epoch": 9.72,
+      "grad_norm": 3.299503803253174,
+      "learning_rate": 1.5432098765432098e-06,
+      "loss": 0.1892,
+      "step": 7700
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 4.933961868286133,
+      "learning_rate": 1.5081369248035917e-06,
+      "loss": 0.2198,
+      "step": 7705
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 3.560870885848999,
+      "learning_rate": 1.4730639730639732e-06,
+      "loss": 0.2641,
+      "step": 7710
+    },
+    {
+      "epoch": 9.74,
+      "grad_norm": 3.664213180541992,
+      "learning_rate": 1.4379910213243547e-06,
+      "loss": 0.2243,
+      "step": 7715
+    },
+    {
+      "epoch": 9.74,
+      "grad_norm": 3.95974063873291,
+      "learning_rate": 1.4029180695847364e-06,
+      "loss": 0.1927,
+      "step": 7720
+    },
+    {
+      "epoch": 9.75,
+      "grad_norm": 3.3959004878997803,
+      "learning_rate": 1.367845117845118e-06,
+      "loss": 0.2236,
+      "step": 7725
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 5.863026142120361,
+      "learning_rate": 1.3327721661054994e-06,
+      "loss": 0.2025,
+      "step": 7730
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 3.299866199493408,
+      "learning_rate": 1.2976992143658812e-06,
+      "loss": 0.2239,
+      "step": 7735
+    },
+    {
+      "epoch": 9.77,
+      "grad_norm": 4.295718193054199,
+      "learning_rate": 1.2626262626262627e-06,
+      "loss": 0.2186,
+      "step": 7740
+    },
+    {
+      "epoch": 9.78,
+      "grad_norm": 3.7850561141967773,
+      "learning_rate": 1.2275533108866442e-06,
+      "loss": 0.223,
+      "step": 7745
+    },
+    {
+      "epoch": 9.78,
+      "grad_norm": 4.560152053833008,
+      "learning_rate": 1.1924803591470259e-06,
+      "loss": 0.2405,
+      "step": 7750
+    },
+    {
+      "epoch": 9.79,
+      "grad_norm": 3.047708511352539,
+      "learning_rate": 1.1574074074074074e-06,
+      "loss": 0.2084,
+      "step": 7755
+    },
+    {
+      "epoch": 9.79,
+      "grad_norm": 5.037156105041504,
+      "learning_rate": 1.122334455667789e-06,
+      "loss": 0.2129,
+      "step": 7760
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 5.772583484649658,
+      "learning_rate": 1.0872615039281708e-06,
+      "loss": 0.2092,
+      "step": 7765
+    },
+    {
+      "epoch": 9.81,
+      "grad_norm": 4.003391265869141,
+      "learning_rate": 1.0521885521885523e-06,
+      "loss": 0.2305,
+      "step": 7770
+    },
+    {
+      "epoch": 9.81,
+      "grad_norm": 2.626347303390503,
+      "learning_rate": 1.0171156004489338e-06,
+      "loss": 0.2035,
+      "step": 7775
+    },
+    {
+      "epoch": 9.82,
+      "grad_norm": 3.2241694927215576,
+      "learning_rate": 9.820426487093155e-07,
+      "loss": 0.2329,
+      "step": 7780
+    },
+    {
+      "epoch": 9.83,
+      "grad_norm": 4.2929840087890625,
+      "learning_rate": 9.469696969696971e-07,
+      "loss": 0.1965,
+      "step": 7785
+    },
+    {
+      "epoch": 9.83,
+      "grad_norm": 3.6872854232788086,
+      "learning_rate": 9.118967452300787e-07,
+      "loss": 0.165,
+      "step": 7790
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 4.689639091491699,
+      "learning_rate": 8.768237934904602e-07,
+      "loss": 0.2551,
+      "step": 7795
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 3.661680221557617,
+      "learning_rate": 8.417508417508418e-07,
+      "loss": 0.2764,
+      "step": 7800
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 2.970412492752075,
+      "learning_rate": 8.066778900112234e-07,
+      "loss": 0.2205,
+      "step": 7805
+    },
+    {
+      "epoch": 9.86,
+      "grad_norm": 4.266114234924316,
+      "learning_rate": 7.716049382716049e-07,
+      "loss": 0.2475,
+      "step": 7810
+    },
+    {
+      "epoch": 9.86,
+      "grad_norm": 5.572594165802002,
+      "learning_rate": 7.365319865319866e-07,
+      "loss": 0.2269,
+      "step": 7815
+    },
+    {
+      "epoch": 9.87,
+      "grad_norm": 4.087586879730225,
+      "learning_rate": 7.014590347923682e-07,
+      "loss": 0.2639,
+      "step": 7820
+    },
+    {
+      "epoch": 9.88,
+      "grad_norm": 3.8800711631774902,
+      "learning_rate": 6.663860830527497e-07,
+      "loss": 0.2137,
+      "step": 7825
+    },
+    {
+      "epoch": 9.88,
+      "grad_norm": 3.1962506771087646,
+      "learning_rate": 6.313131313131313e-07,
+      "loss": 0.2075,
+      "step": 7830
+    },
+    {
+      "epoch": 9.89,
+      "grad_norm": 5.0696330070495605,
+      "learning_rate": 5.962401795735129e-07,
+      "loss": 0.1933,
+      "step": 7835
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 3.3314061164855957,
+      "learning_rate": 5.611672278338944e-07,
+      "loss": 0.192,
+      "step": 7840
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 3.919238328933716,
+      "learning_rate": 5.260942760942762e-07,
+      "loss": 0.2143,
+      "step": 7845
+    },
+    {
+      "epoch": 9.91,
+      "grad_norm": 5.7739338874816895,
+      "learning_rate": 4.910213243546578e-07,
+      "loss": 0.2476,
+      "step": 7850
+    },
+    {
+      "epoch": 9.91,
+      "grad_norm": 4.4509100914001465,
+      "learning_rate": 4.5594837261503933e-07,
+      "loss": 0.2342,
+      "step": 7855
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 5.685267925262451,
+      "learning_rate": 4.208754208754209e-07,
+      "loss": 0.2409,
+      "step": 7860
+    },
+    {
+      "epoch": 9.93,
+      "grad_norm": 5.558847904205322,
+      "learning_rate": 3.8580246913580245e-07,
+      "loss": 0.2162,
+      "step": 7865
+    },
+    {
+      "epoch": 9.93,
+      "grad_norm": 3.85164737701416,
+      "learning_rate": 3.507295173961841e-07,
+      "loss": 0.2148,
+      "step": 7870
+    },
+    {
+      "epoch": 9.94,
+      "grad_norm": 4.685579776763916,
+      "learning_rate": 3.1565656565656567e-07,
+      "loss": 0.2376,
+      "step": 7875
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 3.917475700378418,
+      "learning_rate": 2.805836139169472e-07,
+      "loss": 0.1544,
+      "step": 7880
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 3.2414307594299316,
+      "learning_rate": 2.455106621773289e-07,
+      "loss": 0.2079,
+      "step": 7885
+    },
+    {
+      "epoch": 9.96,
+      "grad_norm": 3.448442220687866,
+      "learning_rate": 2.1043771043771044e-07,
+      "loss": 0.2294,
+      "step": 7890
+    },
+    {
+      "epoch": 9.97,
+      "grad_norm": 4.516743183135986,
+      "learning_rate": 1.7536475869809205e-07,
+      "loss": 0.2418,
+      "step": 7895
+    },
+    {
+      "epoch": 9.97,
+      "grad_norm": 4.4012837409973145,
+      "learning_rate": 1.402918069584736e-07,
+      "loss": 0.2044,
+      "step": 7900
+    },
+    {
+      "epoch": 9.98,
+      "grad_norm": 3.965578317642212,
+      "learning_rate": 1.0521885521885522e-07,
+      "loss": 0.2348,
+      "step": 7905
+    },
+    {
+      "epoch": 9.98,
+      "grad_norm": 4.180352687835693,
+      "learning_rate": 7.01459034792368e-08,
+      "loss": 0.2494,
+      "step": 7910
+    },
+    {
+      "epoch": 9.99,
+      "grad_norm": 5.506495952606201,
+      "learning_rate": 3.50729517396184e-08,
+      "loss": 0.2019,
+      "step": 7915
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 3.8127410411834717,
+      "learning_rate": 0.0,
+      "loss": 0.1938,
+      "step": 7920
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.9374278867489128,
+      "eval_loss": 0.1932387501001358,
+      "eval_runtime": 30.9585,
+      "eval_samples_per_second": 363.939,
+      "eval_steps_per_second": 11.402,
+      "step": 7920
+    },
+    {
+      "epoch": 10.0,
+      "step": 7920,
+      "total_flos": 2.5200889503926428e+19,
+      "train_loss": 0.285455356154478,
+      "train_runtime": 5480.7713,
+      "train_samples_per_second": 185.014,
+      "train_steps_per_second": 1.445
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 7920,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 2.5200889503926428e+19,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}