{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.030597377367654,
  "eval_steps": 500,
  "global_step": 1950,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007770762506070908,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0001,
      "loss": 0.3922,
      "step": 5
    },
    {
      "epoch": 0.015541525012141816,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0001,
      "loss": 0.2477,
      "step": 10
    },
    {
      "epoch": 0.023312287518212724,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0001,
      "loss": 0.1374,
      "step": 15
    },
    {
      "epoch": 0.03108305002428363,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0001,
      "loss": 0.1301,
      "step": 20
    },
    {
      "epoch": 0.03885381253035454,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0001,
      "loss": 0.1581,
      "step": 25
    },
    {
      "epoch": 0.04662457503642545,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001,
      "loss": 0.1942,
      "step": 30
    },
    {
      "epoch": 0.054395337542496355,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0001,
      "loss": 0.2204,
      "step": 35
    },
    {
      "epoch": 0.06216610004856726,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0001,
      "loss": 0.2906,
      "step": 40
    },
    {
      "epoch": 0.06993686255463817,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001,
      "loss": 0.2077,
      "step": 45
    },
    {
      "epoch": 0.07770762506070908,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0001,
      "loss": 0.2592,
      "step": 50
    },
    {
      "epoch": 0.08547838756677999,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0001,
      "loss": 0.1449,
      "step": 55
    },
    {
      "epoch": 0.0932491500728509,
      "grad_norm": 0.0361328125,
      "learning_rate": 0.0001,
      "loss": 0.0462,
      "step": 60
    },
    {
      "epoch": 0.1010199125789218,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0001,
      "loss": 0.0439,
      "step": 65
    },
    {
      "epoch": 0.10879067508499271,
      "grad_norm": 0.04248046875,
      "learning_rate": 0.0001,
      "loss": 0.0318,
      "step": 70
    },
    {
      "epoch": 0.11656143759106362,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0001,
      "loss": 0.0901,
      "step": 75
    },
    {
      "epoch": 0.12433220009713453,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0001,
      "loss": 0.1907,
      "step": 80
    },
    {
      "epoch": 0.13210296260320545,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001,
      "loss": 0.1495,
      "step": 85
    },
    {
      "epoch": 0.13987372510927634,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0001,
      "loss": 0.1917,
      "step": 90
    },
    {
      "epoch": 0.14764448761534726,
      "grad_norm": 0.125,
      "learning_rate": 0.0001,
      "loss": 0.1706,
      "step": 95
    },
    {
      "epoch": 0.15541525012141816,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0001,
      "loss": 0.2569,
      "step": 100
    },
    {
      "epoch": 0.16318601262748908,
      "grad_norm": 0.04052734375,
      "learning_rate": 0.0001,
      "loss": 0.1343,
      "step": 105
    },
    {
      "epoch": 0.17095677513355997,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0001,
      "loss": 0.0482,
      "step": 110
    },
    {
      "epoch": 0.1787275376396309,
      "grad_norm": 0.0281982421875,
      "learning_rate": 0.0001,
      "loss": 0.0397,
      "step": 115
    },
    {
      "epoch": 0.1864983001457018,
      "grad_norm": 0.025390625,
      "learning_rate": 0.0001,
      "loss": 0.0529,
      "step": 120
    },
    {
      "epoch": 0.1942690626517727,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0001,
      "loss": 0.083,
      "step": 125
    },
    {
      "epoch": 0.2020398251578436,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0001,
      "loss": 0.1048,
      "step": 130
    },
    {
      "epoch": 0.20981058766391453,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0001,
      "loss": 0.1576,
      "step": 135
    },
    {
      "epoch": 0.21758135016998542,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0001,
      "loss": 0.1435,
      "step": 140
    },
    {
      "epoch": 0.22535211267605634,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001,
      "loss": 0.1928,
      "step": 145
    },
    {
      "epoch": 0.23312287518212724,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0001,
      "loss": 0.2278,
      "step": 150
    },
    {
      "epoch": 0.24089363768819816,
      "grad_norm": 0.0439453125,
      "learning_rate": 0.0001,
      "loss": 0.0949,
      "step": 155
    },
    {
      "epoch": 0.24866440019426905,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0001,
      "loss": 0.0518,
      "step": 160
    },
    {
      "epoch": 0.25643516270033995,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0001,
      "loss": 0.0378,
      "step": 165
    },
    {
      "epoch": 0.2642059252064109,
      "grad_norm": 0.041259765625,
      "learning_rate": 0.0001,
      "loss": 0.032,
      "step": 170
    },
    {
      "epoch": 0.2719766877124818,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0001,
      "loss": 0.0801,
      "step": 175
    },
    {
      "epoch": 0.2797474502185527,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0001,
      "loss": 0.1381,
      "step": 180
    },
    {
      "epoch": 0.2875182127246236,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0001,
      "loss": 0.1337,
      "step": 185
    },
    {
      "epoch": 0.29528897523069453,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0001,
      "loss": 0.1276,
      "step": 190
    },
    {
      "epoch": 0.3030597377367654,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0001,
      "loss": 0.155,
      "step": 195
    },
    {
      "epoch": 0.3108305002428363,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0001,
      "loss": 0.1496,
      "step": 200
    },
    {
      "epoch": 0.3186012627489072,
      "grad_norm": 0.0269775390625,
      "learning_rate": 0.0001,
      "loss": 0.1061,
      "step": 205
    },
    {
      "epoch": 0.32637202525497816,
      "grad_norm": 0.021484375,
      "learning_rate": 0.0001,
      "loss": 0.0486,
      "step": 210
    },
    {
      "epoch": 0.33414278776104905,
      "grad_norm": 0.0296630859375,
      "learning_rate": 0.0001,
      "loss": 0.0279,
      "step": 215
    },
    {
      "epoch": 0.34191355026711995,
      "grad_norm": 0.019775390625,
      "learning_rate": 0.0001,
      "loss": 0.0284,
      "step": 220
    },
    {
      "epoch": 0.34968431277319084,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0001,
      "loss": 0.0874,
      "step": 225
    },
    {
      "epoch": 0.3574550752792618,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0001,
      "loss": 0.1743,
      "step": 230
    },
    {
      "epoch": 0.3652258377853327,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0001,
      "loss": 0.1282,
      "step": 235
    },
    {
      "epoch": 0.3729966002914036,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0001,
      "loss": 0.1091,
      "step": 240
    },
    {
      "epoch": 0.38076736279747453,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0001,
      "loss": 0.1333,
      "step": 245
    },
    {
      "epoch": 0.3885381253035454,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0001,
      "loss": 0.1941,
      "step": 250
    },
    {
      "epoch": 0.3963088878096163,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0001,
      "loss": 0.1179,
      "step": 255
    },
    {
      "epoch": 0.4040796503156872,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0001,
      "loss": 0.0754,
      "step": 260
    },
    {
      "epoch": 0.41185041282175816,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0001,
      "loss": 0.0429,
      "step": 265
    },
    {
      "epoch": 0.41962117532782905,
      "grad_norm": 0.047607421875,
      "learning_rate": 0.0001,
      "loss": 0.0354,
      "step": 270
    },
    {
      "epoch": 0.42739193783389995,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0001,
      "loss": 0.1228,
      "step": 275
    },
    {
      "epoch": 0.43516270033997084,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0001,
      "loss": 0.1233,
      "step": 280
    },
    {
      "epoch": 0.4429334628460418,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0001,
      "loss": 0.1454,
      "step": 285
    },
    {
      "epoch": 0.4507042253521127,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.1248,
      "step": 290
    },
    {
      "epoch": 0.4584749878581836,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0001,
      "loss": 0.1154,
      "step": 295
    },
    {
      "epoch": 0.4662457503642545,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.1596,
      "step": 300
    },
    {
      "epoch": 0.4740165128703254,
      "grad_norm": 0.048583984375,
      "learning_rate": 0.0001,
      "loss": 0.124,
      "step": 305
    },
    {
      "epoch": 0.4817872753763963,
      "grad_norm": 0.0205078125,
      "learning_rate": 0.0001,
      "loss": 0.0428,
      "step": 310
    },
    {
      "epoch": 0.4895580378824672,
      "grad_norm": 0.034423828125,
      "learning_rate": 0.0001,
      "loss": 0.016,
      "step": 315
    },
    {
      "epoch": 0.4973288003885381,
      "grad_norm": 0.0361328125,
      "learning_rate": 0.0001,
      "loss": 0.0196,
      "step": 320
    },
    {
      "epoch": 0.505099562894609,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0001,
      "loss": 0.1089,
      "step": 325
    },
    {
      "epoch": 0.5128703254006799,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0001,
      "loss": 0.132,
      "step": 330
    },
    {
      "epoch": 0.5206410879067509,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0001,
      "loss": 0.1718,
      "step": 335
    },
    {
      "epoch": 0.5284118504128218,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0001,
      "loss": 0.1096,
      "step": 340
    },
    {
      "epoch": 0.5361826129188927,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0001,
      "loss": 0.1342,
      "step": 345
    },
    {
      "epoch": 0.5439533754249636,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0001,
      "loss": 0.1532,
      "step": 350
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0001,
      "loss": 0.112,
      "step": 355
    },
    {
      "epoch": 0.5594949004371054,
      "grad_norm": 0.051513671875,
      "learning_rate": 0.0001,
      "loss": 0.0371,
      "step": 360
    },
    {
      "epoch": 0.5672656629431763,
      "grad_norm": 0.0264892578125,
      "learning_rate": 0.0001,
      "loss": 0.0321,
      "step": 365
    },
    {
      "epoch": 0.5750364254492472,
      "grad_norm": 0.032470703125,
      "learning_rate": 0.0001,
      "loss": 0.0473,
      "step": 370
    },
    {
      "epoch": 0.5828071879553182,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0001,
      "loss": 0.0738,
      "step": 375
    },
    {
      "epoch": 0.5905779504613891,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0001,
      "loss": 0.1384,
      "step": 380
    },
    {
      "epoch": 0.59834871296746,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0001,
      "loss": 0.1191,
      "step": 385
    },
    {
      "epoch": 0.6061194754735308,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0001,
      "loss": 0.1188,
      "step": 390
    },
    {
      "epoch": 0.6138902379796017,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0001,
      "loss": 0.1327,
      "step": 395
    },
    {
      "epoch": 0.6216610004856726,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0001,
      "loss": 0.1695,
      "step": 400
    },
    {
      "epoch": 0.6294317629917435,
      "grad_norm": 0.046142578125,
      "learning_rate": 0.0001,
      "loss": 0.11,
      "step": 405
    },
    {
      "epoch": 0.6372025254978144,
      "grad_norm": 0.042236328125,
      "learning_rate": 0.0001,
      "loss": 0.052,
      "step": 410
    },
    {
      "epoch": 0.6449732880038854,
      "grad_norm": 0.0205078125,
      "learning_rate": 0.0001,
      "loss": 0.02,
      "step": 415
    },
    {
      "epoch": 0.6527440505099563,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0001,
      "loss": 0.0554,
      "step": 420
    },
    {
      "epoch": 0.6605148130160272,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0001,
      "loss": 0.063,
      "step": 425
    },
    {
      "epoch": 0.6682855755220981,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0001,
      "loss": 0.1229,
      "step": 430
    },
    {
      "epoch": 0.676056338028169,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0001,
      "loss": 0.1218,
      "step": 435
    },
    {
      "epoch": 0.6838271005342399,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0001,
      "loss": 0.11,
      "step": 440
    },
    {
      "epoch": 0.6915978630403108,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0001,
      "loss": 0.1338,
      "step": 445
    },
    {
      "epoch": 0.6993686255463817,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001,
      "loss": 0.2131,
      "step": 450
    },
    {
      "epoch": 0.7071393880524527,
      "grad_norm": 0.04541015625,
      "learning_rate": 0.0001,
      "loss": 0.111,
      "step": 455
    },
    {
      "epoch": 0.7149101505585236,
      "grad_norm": 0.0264892578125,
      "learning_rate": 0.0001,
      "loss": 0.0656,
      "step": 460
    },
    {
      "epoch": 0.7226809130645945,
      "grad_norm": 0.023193359375,
      "learning_rate": 0.0001,
      "loss": 0.0216,
      "step": 465
    },
    {
      "epoch": 0.7304516755706654,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0001,
      "loss": 0.042,
      "step": 470
    },
    {
      "epoch": 0.7382224380767363,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0001,
      "loss": 0.0533,
      "step": 475
    },
    {
      "epoch": 0.7459932005828072,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0001,
      "loss": 0.0643,
      "step": 480
    },
    {
      "epoch": 0.753763963088878,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0001,
      "loss": 0.08,
      "step": 485
    },
    {
      "epoch": 0.7615347255949491,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0001,
      "loss": 0.1347,
      "step": 490
    },
    {
      "epoch": 0.76930548810102,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0001,
      "loss": 0.0792,
      "step": 495
    },
    {
      "epoch": 0.7770762506070908,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0001,
      "loss": 0.1212,
      "step": 500
    },
    {
      "epoch": 0.7848470131131617,
      "grad_norm": 0.05615234375,
      "learning_rate": 0.0001,
      "loss": 0.1122,
      "step": 505
    },
    {
      "epoch": 0.7926177756192326,
      "grad_norm": 0.006927490234375,
      "learning_rate": 0.0001,
      "loss": 0.0504,
      "step": 510
    },
    {
      "epoch": 0.8003885381253035,
      "grad_norm": 0.01495361328125,
      "learning_rate": 0.0001,
      "loss": 0.0143,
      "step": 515
    },
    {
      "epoch": 0.8081593006313744,
      "grad_norm": 0.03125,
      "learning_rate": 0.0001,
      "loss": 0.0229,
      "step": 520
    },
    {
      "epoch": 0.8159300631374453,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0001,
      "loss": 0.0618,
      "step": 525
    },
    {
      "epoch": 0.8237008256435163,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0001,
      "loss": 0.1189,
      "step": 530
    },
    {
      "epoch": 0.8314715881495872,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0001,
      "loss": 0.104,
      "step": 535
    },
    {
      "epoch": 0.8392423506556581,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0001,
      "loss": 0.0988,
      "step": 540
    },
    {
      "epoch": 0.847013113161729,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001,
      "loss": 0.1413,
      "step": 545
    },
    {
      "epoch": 0.8547838756677999,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0001,
      "loss": 0.1027,
      "step": 550
    },
    {
      "epoch": 0.8625546381738708,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0001,
      "loss": 0.1324,
      "step": 555
    },
    {
      "epoch": 0.8703254006799417,
      "grad_norm": 0.0062255859375,
      "learning_rate": 0.0001,
      "loss": 0.0265,
      "step": 560
    },
    {
      "epoch": 0.8780961631860126,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0001,
      "loss": 0.02,
      "step": 565
    },
    {
      "epoch": 0.8858669256920836,
      "grad_norm": 0.0296630859375,
      "learning_rate": 0.0001,
      "loss": 0.0326,
      "step": 570
    },
    {
      "epoch": 0.8936376881981545,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0001,
      "loss": 0.0746,
      "step": 575
    },
    {
      "epoch": 0.9014084507042254,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0001,
      "loss": 0.1296,
      "step": 580
    },
    {
      "epoch": 0.9091792132102963,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0001,
      "loss": 0.1037,
      "step": 585
    },
    {
      "epoch": 0.9169499757163672,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0001,
      "loss": 0.1419,
      "step": 590
    },
    {
      "epoch": 0.924720738222438,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.1326,
      "step": 595
    },
    {
      "epoch": 0.932491500728509,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0001,
      "loss": 0.154,
      "step": 600
    },
    {
      "epoch": 0.9402622632345798,
      "grad_norm": 0.05908203125,
      "learning_rate": 0.0001,
      "loss": 0.1029,
      "step": 605
    },
    {
      "epoch": 0.9480330257406508,
      "grad_norm": 0.032958984375,
      "learning_rate": 0.0001,
      "loss": 0.0223,
      "step": 610
    },
    {
      "epoch": 0.9558037882467217,
      "grad_norm": 0.0361328125,
      "learning_rate": 0.0001,
      "loss": 0.0326,
      "step": 615
    },
    {
      "epoch": 0.9635745507527926,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0001,
      "loss": 0.0454,
      "step": 620
    },
    {
      "epoch": 0.9713453132588635,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0001,
      "loss": 0.1382,
      "step": 625
    },
    {
      "epoch": 0.9791160757649344,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0001,
      "loss": 0.1148,
      "step": 630
    },
    {
      "epoch": 0.9868868382710053,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0001,
      "loss": 0.1029,
      "step": 635
    },
    {
      "epoch": 0.9946576007770762,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0001,
      "loss": 0.1802,
      "step": 640
    },
    {
      "epoch": 1.0024283632831472,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0001,
      "loss": 0.1323,
      "step": 645
    },
    {
      "epoch": 1.010199125789218,
      "grad_norm": 0.053955078125,
      "learning_rate": 0.0001,
      "loss": 0.0703,
      "step": 650
    },
    {
      "epoch": 1.017969888295289,
      "grad_norm": 0.00970458984375,
      "learning_rate": 0.0001,
      "loss": 0.0169,
      "step": 655
    },
    {
      "epoch": 1.0257406508013598,
      "grad_norm": 0.0311279296875,
      "learning_rate": 0.0001,
      "loss": 0.0106,
      "step": 660
    },
    {
      "epoch": 1.0335114133074308,
      "grad_norm": 0.04736328125,
      "learning_rate": 0.0001,
      "loss": 0.0357,
      "step": 665
    },
    {
      "epoch": 1.0412821758135018,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0001,
      "loss": 0.0458,
      "step": 670
    },
    {
      "epoch": 1.0490529383195726,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0001,
      "loss": 0.073,
      "step": 675
    },
    {
      "epoch": 1.0568237008256436,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0001,
      "loss": 0.059,
      "step": 680
    },
    {
      "epoch": 1.0645944633317144,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0001,
      "loss": 0.0613,
      "step": 685
    },
    {
      "epoch": 1.0723652258377854,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0001,
      "loss": 0.0741,
      "step": 690
    },
    {
      "epoch": 1.0801359883438562,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0001,
      "loss": 0.0879,
      "step": 695
    },
    {
      "epoch": 1.0879067508499272,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0001,
      "loss": 0.0678,
      "step": 700
    },
    {
      "epoch": 1.095677513355998,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0001,
      "loss": 0.0388,
      "step": 705
    },
    {
      "epoch": 1.103448275862069,
      "grad_norm": 0.03662109375,
      "learning_rate": 0.0001,
      "loss": 0.0128,
      "step": 710
    },
    {
      "epoch": 1.11121903836814,
      "grad_norm": 0.020263671875,
      "learning_rate": 0.0001,
      "loss": 0.0178,
      "step": 715
    },
    {
      "epoch": 1.1189898008742107,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0001,
      "loss": 0.0538,
      "step": 720
    },
    {
      "epoch": 1.1267605633802817,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0001,
      "loss": 0.1117,
      "step": 725
    },
    {
      "epoch": 1.1345313258863525,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001,
      "loss": 0.0767,
      "step": 730
    },
    {
      "epoch": 1.1423020883924235,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0001,
      "loss": 0.061,
      "step": 735
    },
    {
      "epoch": 1.1500728508984945,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.0714,
      "step": 740
    },
    {
      "epoch": 1.1578436134045653,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0001,
      "loss": 0.0921,
      "step": 745
    },
    {
      "epoch": 1.1656143759106363,
      "grad_norm": 0.050537109375,
      "learning_rate": 0.0001,
      "loss": 0.0691,
      "step": 750
    },
    {
      "epoch": 1.173385138416707,
      "grad_norm": 0.028076171875,
      "learning_rate": 0.0001,
      "loss": 0.0239,
      "step": 755
    },
    {
      "epoch": 1.1811559009227781,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0001,
      "loss": 0.0202,
      "step": 760
    },
    {
      "epoch": 1.188926663428849,
      "grad_norm": 0.0654296875,
      "learning_rate": 0.0001,
      "loss": 0.0296,
      "step": 765
    },
    {
      "epoch": 1.19669742593492,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0001,
      "loss": 0.0632,
      "step": 770
    },
    {
      "epoch": 1.2044681884409907,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0001,
      "loss": 0.0675,
      "step": 775
    },
    {
      "epoch": 1.2122389509470617,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0001,
      "loss": 0.0772,
      "step": 780
    },
    {
      "epoch": 1.2200097134531327,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0001,
      "loss": 0.0628,
      "step": 785
    },
    {
      "epoch": 1.2277804759592035,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0001,
      "loss": 0.0662,
      "step": 790
    },
    {
      "epoch": 1.2355512384652745,
      "grad_norm": 0.05078125,
      "learning_rate": 0.0001,
      "loss": 0.0962,
      "step": 795
    },
    {
      "epoch": 1.2433220009713453,
      "grad_norm": 0.0302734375,
      "learning_rate": 0.0001,
      "loss": 0.0606,
      "step": 800
    },
    {
      "epoch": 1.2510927634774163,
      "grad_norm": 0.0361328125,
      "learning_rate": 0.0001,
      "loss": 0.0119,
      "step": 805
    },
    {
      "epoch": 1.258863525983487,
      "grad_norm": 0.03515625,
      "learning_rate": 0.0001,
      "loss": 0.0156,
      "step": 810
    },
    {
      "epoch": 1.266634288489558,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0001,
      "loss": 0.0393,
      "step": 815
    },
    {
      "epoch": 1.2744050509956288,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001,
      "loss": 0.1104,
      "step": 820
    },
    {
      "epoch": 1.2821758135016998,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0001,
      "loss": 0.0592,
      "step": 825
    },
    {
      "epoch": 1.2899465760077709,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0001,
      "loss": 0.0604,
      "step": 830
    },
    {
      "epoch": 1.2977173385138416,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0001,
      "loss": 0.0507,
      "step": 835
    },
    {
      "epoch": 1.3054881010199126,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.079,
      "step": 840
    },
    {
      "epoch": 1.3132588635259834,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0001,
      "loss": 0.1041,
      "step": 845
    },
    {
      "epoch": 1.3210296260320544,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0001,
      "loss": 0.0751,
      "step": 850
    },
    {
      "epoch": 1.3288003885381254,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0001,
      "loss": 0.0338,
      "step": 855
    },
    {
      "epoch": 1.3365711510441962,
      "grad_norm": 0.0546875,
      "learning_rate": 0.0001,
      "loss": 0.0144,
      "step": 860
    },
    {
      "epoch": 1.344341913550267,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0001,
      "loss": 0.0359,
      "step": 865
    },
    {
      "epoch": 1.352112676056338,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0001,
      "loss": 0.0498,
      "step": 870
    },
    {
      "epoch": 1.359883438562409,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0001,
      "loss": 0.0422,
      "step": 875
    },
    {
      "epoch": 1.3676542010684798,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0001,
      "loss": 0.046,
      "step": 880
    },
    {
      "epoch": 1.3754249635745508,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0001,
      "loss": 0.059,
      "step": 885
    },
    {
      "epoch": 1.3831957260806216,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0001,
      "loss": 0.0545,
      "step": 890
    },
    {
      "epoch": 1.3909664885866926,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0001,
      "loss": 0.117,
      "step": 895
    },
    {
      "epoch": 1.3987372510927636,
      "grad_norm": 0.03857421875,
      "learning_rate": 0.0001,
      "loss": 0.0588,
      "step": 900
    },
    {
      "epoch": 1.4065080135988344,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0001,
      "loss": 0.0243,
      "step": 905
    },
    {
      "epoch": 1.4142787761049052,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0001,
      "loss": 0.0204,
      "step": 910
    },
    {
      "epoch": 1.4220495386109762,
      "grad_norm": 0.040283203125,
      "learning_rate": 0.0001,
      "loss": 0.0323,
      "step": 915
    },
    {
      "epoch": 1.4298203011170472,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0001,
      "loss": 0.0851,
      "step": 920
    },
    {
      "epoch": 1.437591063623118,
      "grad_norm": 0.078125,
      "learning_rate": 0.0001,
      "loss": 0.0821,
      "step": 925
    },
    {
      "epoch": 1.445361826129189,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0001,
      "loss": 0.0597,
      "step": 930
    },
    {
      "epoch": 1.4531325886352597,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.0617,
      "step": 935
    },
    {
      "epoch": 1.4609033511413307,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.047,
      "step": 940
    },
    {
      "epoch": 1.4686741136474017,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0001,
      "loss": 0.0853,
      "step": 945
    },
    {
      "epoch": 1.4764448761534725,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0001,
      "loss": 0.0803,
      "step": 950
    },
    {
      "epoch": 1.4842156386595435,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0001,
      "loss": 0.0302,
      "step": 955
    },
    {
      "epoch": 1.4919864011656143,
      "grad_norm": 0.0284423828125,
      "learning_rate": 0.0001,
      "loss": 0.0173,
      "step": 960
    },
    {
      "epoch": 1.4997571636716853,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0001,
      "loss": 0.0262,
      "step": 965
    },
    {
      "epoch": 1.5075279261777563,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0001,
      "loss": 0.0533,
      "step": 970
    },
    {
      "epoch": 1.515298688683827,
      "grad_norm": 0.039794921875,
      "learning_rate": 0.0001,
      "loss": 0.0651,
      "step": 975
    },
    {
      "epoch": 1.523069451189898,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0001,
      "loss": 0.0682,
      "step": 980
    },
    {
      "epoch": 1.530840213695969,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0001,
      "loss": 0.0532,
      "step": 985
    },
    {
      "epoch": 1.53861097620204,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0001,
      "loss": 0.0736,
      "step": 990
    },
    {
      "epoch": 1.5463817387081107,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0001,
      "loss": 0.0748,
      "step": 995
    },
    {
      "epoch": 1.5541525012141817,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0001,
      "loss": 0.0824,
      "step": 1000
    },
    {
      "epoch": 1.5619232637202525,
      "grad_norm": 0.016357421875,
      "learning_rate": 0.0001,
      "loss": 0.0252,
      "step": 1005
    },
    {
      "epoch": 1.5696940262263235,
      "grad_norm": 0.025390625,
      "learning_rate": 0.0001,
      "loss": 0.0179,
      "step": 1010
    },
    {
      "epoch": 1.5774647887323945,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0001,
      "loss": 0.0342,
      "step": 1015
    },
    {
      "epoch": 1.5852355512384653,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.0657,
      "step": 1020
    },
    {
      "epoch": 1.593006313744536,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0001,
      "loss": 0.0668,
      "step": 1025
    },
    {
      "epoch": 1.600777076250607,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0001,
      "loss": 0.0769,
      "step": 1030
    },
    {
      "epoch": 1.608547838756678,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0001,
      "loss": 0.0369,
      "step": 1035
    },
    {
      "epoch": 1.616318601262749,
      "grad_norm": 0.125,
      "learning_rate": 0.0001,
      "loss": 0.0803,
      "step": 1040
    },
    {
      "epoch": 1.6240893637688198,
      "grad_norm": 0.046875,
      "learning_rate": 0.0001,
      "loss": 0.083,
      "step": 1045
    },
    {
      "epoch": 1.6318601262748906,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0001,
      "loss": 0.0768,
      "step": 1050
    },
    {
      "epoch": 1.6396308887809616,
      "grad_norm": 0.03173828125,
      "learning_rate": 0.0001,
      "loss": 0.0226,
      "step": 1055
    },
    {
      "epoch": 1.6474016512870326,
      "grad_norm": 0.037353515625,
      "learning_rate": 0.0001,
      "loss": 0.0145,
      "step": 1060
    },
    {
      "epoch": 1.6551724137931034,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0001,
      "loss": 0.022,
      "step": 1065
    },
    {
      "epoch": 1.6629431762991742,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0001,
      "loss": 0.0627,
      "step": 1070
    },
    {
      "epoch": 1.6707139388052452,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0001,
      "loss": 0.0709,
      "step": 1075
    },
    {
      "epoch": 1.6784847013113162,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0001,
      "loss": 0.0613,
      "step": 1080
    },
    {
      "epoch": 1.6862554638173872,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0001,
      "loss": 0.0543,
      "step": 1085
    },
    {
      "epoch": 1.694026226323458,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0001,
      "loss": 0.064,
      "step": 1090
    },
    {
      "epoch": 1.7017969888295288,
      "grad_norm": 0.055908203125,
      "learning_rate": 0.0001,
      "loss": 0.0977,
      "step": 1095
    },
    {
      "epoch": 1.7095677513355998,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0001,
      "loss": 0.0519,
      "step": 1100
    },
    {
      "epoch": 1.7173385138416708,
      "grad_norm": 0.02734375,
      "learning_rate": 0.0001,
      "loss": 0.0124,
      "step": 1105
    },
    {
      "epoch": 1.7251092763477416,
      "grad_norm": 0.05224609375,
      "learning_rate": 0.0001,
      "loss": 0.0103,
      "step": 1110
    },
    {
      "epoch": 1.7328800388538124,
      "grad_norm": 0.09375,
      "learning_rate": 0.0001,
      "loss": 0.0276,
      "step": 1115
    },
    {
      "epoch": 1.7406508013598834,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0001,
      "loss": 0.045,
      "step": 1120
    },
    {
      "epoch": 1.7484215638659544,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0001,
      "loss": 0.0708,
      "step": 1125
    },
    {
      "epoch": 1.7561923263720254,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0001,
      "loss": 0.0618,
      "step": 1130
    },
    {
      "epoch": 1.7639630888780962,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001,
      "loss": 0.0687,
      "step": 1135
    },
    {
      "epoch": 1.771733851384167,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0001,
      "loss": 0.0451,
      "step": 1140
    },
    {
      "epoch": 1.779504613890238,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0001,
      "loss": 0.1109,
      "step": 1145
    },
    {
      "epoch": 1.787275376396309,
      "grad_norm": 0.032958984375,
      "learning_rate": 0.0001,
      "loss": 0.0728,
      "step": 1150
    },
    {
      "epoch": 1.79504613890238,
      "grad_norm": 0.044921875,
      "learning_rate": 0.0001,
      "loss": 0.0254,
      "step": 1155
    },
    {
      "epoch": 1.8028169014084507,
      "grad_norm": 0.0458984375,
      "learning_rate": 0.0001,
      "loss": 0.0137,
      "step": 1160
    },
    {
      "epoch": 1.8105876639145215,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0001,
      "loss": 0.0242,
      "step": 1165
    },
    {
      "epoch": 1.8183584264205925,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0001,
      "loss": 0.0717,
      "step": 1170
    },
    {
      "epoch": 1.8261291889266635,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0001,
      "loss": 0.0648,
      "step": 1175
    },
    {
      "epoch": 1.8338999514327343,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001,
      "loss": 0.0778,
      "step": 1180
    },
    {
      "epoch": 1.841670713938805,
      "grad_norm": 0.08154296875,
      "learning_rate": 0.0001,
      "loss": 0.0435,
      "step": 1185
    },
    {
      "epoch": 1.849441476444876,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0001,
      "loss": 0.0714,
      "step": 1190
    },
    {
      "epoch": 1.8572122389509471,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0001,
      "loss": 0.0979,
      "step": 1195
    },
    {
      "epoch": 1.8649830014570181,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0001,
      "loss": 0.0785,
      "step": 1200
    },
    {
      "epoch": 1.872753763963089,
      "grad_norm": 0.057861328125,
      "learning_rate": 0.0001,
      "loss": 0.0183,
      "step": 1205
    },
    {
      "epoch": 1.8805245264691597,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0001,
      "loss": 0.0167,
      "step": 1210
    },
    {
      "epoch": 1.8882952889752307,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0001,
      "loss": 0.0269,
      "step": 1215
    },
    {
      "epoch": 1.8960660514813017,
      "grad_norm": 0.06689453125,
      "learning_rate": 0.0001,
      "loss": 0.0532,
      "step": 1220
    },
    {
      "epoch": 1.9038368139873725,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001,
      "loss": 0.0654,
      "step": 1225
    },
    {
      "epoch": 1.9116075764934433,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0001,
      "loss": 0.0592,
      "step": 1230
    },
    {
      "epoch": 1.9193783389995143,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0001,
      "loss": 0.0627,
      "step": 1235
    },
    {
      "epoch": 1.9271491015055853,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0001,
      "loss": 0.0654,
      "step": 1240
    },
    {
      "epoch": 1.9349198640116563,
      "grad_norm": 0.052734375,
      "learning_rate": 0.0001,
      "loss": 0.0763,
      "step": 1245
    },
    {
      "epoch": 1.942690626517727,
      "grad_norm": 0.04296875,
      "learning_rate": 0.0001,
      "loss": 0.0551,
      "step": 1250
    },
    {
      "epoch": 1.9504613890237978,
      "grad_norm": 0.0244140625,
      "learning_rate": 0.0001,
      "loss": 0.0299,
      "step": 1255
    },
    {
      "epoch": 1.9582321515298688,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0001,
      "loss": 0.0362,
      "step": 1260
    },
    {
      "epoch": 1.9660029140359399,
      "grad_norm": 0.06494140625,
      "learning_rate": 0.0001,
      "loss": 0.0285,
      "step": 1265
    },
    {
      "epoch": 1.9737736765420106,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0001,
      "loss": 0.0653,
      "step": 1270
    },
    {
      "epoch": 1.9815444390480816,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0001,
      "loss": 0.0591,
      "step": 1275
    },
    {
      "epoch": 1.9893152015541524,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0001,
      "loss": 0.0599,
      "step": 1280
    },
    {
      "epoch": 1.9970859640602234,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0001,
      "loss": 0.0703,
      "step": 1285
    },
    {
      "epoch": 2.0048567265662944,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0001,
      "loss": 0.0841,
      "step": 1290
    },
    {
      "epoch": 2.0126274890723654,
      "grad_norm": 0.04345703125,
      "learning_rate": 0.0001,
      "loss": 0.0243,
      "step": 1295
    },
    {
      "epoch": 2.020398251578436,
      "grad_norm": 0.0260009765625,
      "learning_rate": 0.0001,
      "loss": 0.0117,
      "step": 1300
    },
    {
      "epoch": 2.028169014084507,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0001,
      "loss": 0.0329,
      "step": 1305
    },
    {
      "epoch": 2.035939776590578,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0001,
      "loss": 0.0344,
      "step": 1310
    },
    {
      "epoch": 2.043710539096649,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001,
      "loss": 0.0517,
      "step": 1315
    },
    {
      "epoch": 2.0514813016027196,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0001,
      "loss": 0.0209,
      "step": 1320
    },
    {
      "epoch": 2.0592520641087906,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0001,
      "loss": 0.0464,
      "step": 1325
    },
    {
      "epoch": 2.0670228266148616,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0001,
      "loss": 0.0241,
      "step": 1330
    },
    {
      "epoch": 2.0747935891209326,
      "grad_norm": 0.043701171875,
      "learning_rate": 0.0001,
      "loss": 0.0356,
      "step": 1335
    },
    {
      "epoch": 2.0825643516270036,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0001,
      "loss": 0.068,
      "step": 1340
    },
    {
      "epoch": 2.090335114133074,
      "grad_norm": 0.03515625,
      "learning_rate": 0.0001,
      "loss": 0.0273,
      "step": 1345
    },
    {
      "epoch": 2.098105876639145,
      "grad_norm": 0.0157470703125,
      "learning_rate": 0.0001,
      "loss": 0.0066,
      "step": 1350
    },
    {
      "epoch": 2.105876639145216,
      "grad_norm": 0.05419921875,
      "learning_rate": 0.0001,
      "loss": 0.0076,
      "step": 1355
    },
    {
      "epoch": 2.113647401651287,
      "grad_norm": 0.0595703125,
      "learning_rate": 0.0001,
      "loss": 0.0128,
      "step": 1360
    },
    {
      "epoch": 2.1214181641573577,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0001,
      "loss": 0.0345,
      "step": 1365
    },
    {
      "epoch": 2.1291889266634287,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0001,
      "loss": 0.0319,
      "step": 1370
    },
    {
      "epoch": 2.1369596891694997,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0001,
      "loss": 0.0345,
      "step": 1375
    },
    {
      "epoch": 2.1447304516755707,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0001,
      "loss": 0.0193,
      "step": 1380
    },
    {
      "epoch": 2.1525012141816418,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0001,
      "loss": 0.0348,
      "step": 1385
    },
    {
      "epoch": 2.1602719766877123,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0001,
      "loss": 0.0704,
      "step": 1390
    },
    {
      "epoch": 2.1680427391937833,
      "grad_norm": 0.0185546875,
      "learning_rate": 0.0001,
      "loss": 0.0289,
      "step": 1395
    },
    {
      "epoch": 2.1758135016998543,
      "grad_norm": 0.03466796875,
      "learning_rate": 0.0001,
      "loss": 0.0092,
      "step": 1400
    },
    {
      "epoch": 2.1835842642059253,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0001,
      "loss": 0.0246,
      "step": 1405
    },
    {
      "epoch": 2.191355026711996,
      "grad_norm": 0.0257568359375,
      "learning_rate": 0.0001,
      "loss": 0.0137,
      "step": 1410
    },
    {
      "epoch": 2.199125789218067,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.0364,
      "step": 1415
    },
    {
      "epoch": 2.206896551724138,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0001,
      "loss": 0.0496,
      "step": 1420
    },
    {
      "epoch": 2.214667314230209,
      "grad_norm": 0.06005859375,
      "learning_rate": 0.0001,
      "loss": 0.0228,
      "step": 1425
    },
    {
      "epoch": 2.22243807673628,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0001,
      "loss": 0.0163,
      "step": 1430
    },
    {
      "epoch": 2.2302088392423505,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0001,
      "loss": 0.0283,
      "step": 1435
    },
    {
      "epoch": 2.2379796017484215,
      "grad_norm": 0.078125,
      "learning_rate": 0.0001,
      "loss": 0.0626,
      "step": 1440
    },
    {
      "epoch": 2.2457503642544925,
      "grad_norm": 0.035888671875,
      "learning_rate": 0.0001,
      "loss": 0.0391,
      "step": 1445
    },
    {
      "epoch": 2.2535211267605635,
      "grad_norm": 0.034912109375,
      "learning_rate": 0.0001,
      "loss": 0.0111,
      "step": 1450
    },
    {
      "epoch": 2.2612918892666345,
      "grad_norm": 0.038818359375,
      "learning_rate": 0.0001,
      "loss": 0.0083,
      "step": 1455
    },
    {
      "epoch": 2.269062651772705,
      "grad_norm": 0.061279296875,
      "learning_rate": 0.0001,
      "loss": 0.0177,
      "step": 1460
    },
    {
      "epoch": 2.276833414278776,
      "grad_norm": 0.057373046875,
      "learning_rate": 0.0001,
      "loss": 0.0448,
      "step": 1465
    },
    {
      "epoch": 2.284604176784847,
      "grad_norm": 0.109375,
      "learning_rate": 0.0001,
      "loss": 0.0218,
      "step": 1470
    },
    {
      "epoch": 2.292374939290918,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0001,
      "loss": 0.0324,
      "step": 1475
    },
    {
      "epoch": 2.300145701796989,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.0296,
      "step": 1480
    },
    {
      "epoch": 2.3079164643030596,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0001,
      "loss": 0.0303,
      "step": 1485
    },
    {
      "epoch": 2.3156872268091306,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0001,
      "loss": 0.0771,
      "step": 1490
    },
    {
      "epoch": 2.3234579893152016,
      "grad_norm": 0.03466796875,
      "learning_rate": 0.0001,
      "loss": 0.018,
      "step": 1495
    },
    {
      "epoch": 2.3312287518212726,
      "grad_norm": 0.03662109375,
      "learning_rate": 0.0001,
      "loss": 0.0117,
      "step": 1500
    },
    {
      "epoch": 2.338999514327343,
      "grad_norm": 0.052001953125,
      "learning_rate": 0.0001,
      "loss": 0.0169,
      "step": 1505
    },
    {
      "epoch": 2.346770276833414,
      "grad_norm": 0.068359375,
      "learning_rate": 0.0001,
      "loss": 0.0227,
      "step": 1510
    },
    {
      "epoch": 2.354541039339485,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0001,
      "loss": 0.034,
      "step": 1515
    },
    {
      "epoch": 2.3623118018455562,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0001,
      "loss": 0.031,
      "step": 1520
    },
    {
      "epoch": 2.370082564351627,
      "grad_norm": 0.08544921875,
      "learning_rate": 0.0001,
      "loss": 0.0278,
      "step": 1525
    },
    {
      "epoch": 2.377853326857698,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001,
      "loss": 0.0325,
      "step": 1530
    },
    {
      "epoch": 2.385624089363769,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0001,
      "loss": 0.0277,
      "step": 1535
    },
    {
      "epoch": 2.39339485186984,
      "grad_norm": 0.0634765625,
      "learning_rate": 0.0001,
      "loss": 0.0537,
      "step": 1540
    },
    {
      "epoch": 2.401165614375911,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0001,
      "loss": 0.04,
      "step": 1545
    },
    {
      "epoch": 2.4089363768819814,
      "grad_norm": 0.01531982421875,
      "learning_rate": 0.0001,
      "loss": 0.0078,
      "step": 1550
    },
    {
      "epoch": 2.4167071393880524,
      "grad_norm": 0.024169921875,
      "learning_rate": 0.0001,
      "loss": 0.0086,
      "step": 1555
    },
    {
      "epoch": 2.4244779018941234,
      "grad_norm": 0.047119140625,
      "learning_rate": 0.0001,
      "loss": 0.0212,
      "step": 1560
    },
    {
      "epoch": 2.4322486644001944,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001,
      "loss": 0.0399,
      "step": 1565
    },
    {
      "epoch": 2.4400194269062654,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0001,
      "loss": 0.0285,
      "step": 1570
    },
    {
      "epoch": 2.447790189412336,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001,
      "loss": 0.0261,
      "step": 1575
    },
    {
      "epoch": 2.455560951918407,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0001,
      "loss": 0.0237,
      "step": 1580
    },
    {
      "epoch": 2.463331714424478,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0001,
      "loss": 0.0401,
      "step": 1585
    },
    {
      "epoch": 2.471102476930549,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0001,
      "loss": 0.058,
      "step": 1590
    },
    {
      "epoch": 2.4788732394366195,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0001,
      "loss": 0.0378,
      "step": 1595
    },
    {
      "epoch": 2.4866440019426905,
      "grad_norm": 0.03662109375,
      "learning_rate": 0.0001,
      "loss": 0.0066,
      "step": 1600
    },
    {
      "epoch": 2.4944147644487615,
      "grad_norm": 0.0478515625,
      "learning_rate": 0.0001,
      "loss": 0.0128,
      "step": 1605
    },
    {
      "epoch": 2.5021855269548325,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0001,
      "loss": 0.0217,
      "step": 1610
    },
    {
      "epoch": 2.509956289460903,
      "grad_norm": 0.060302734375,
      "learning_rate": 0.0001,
      "loss": 0.0329,
      "step": 1615
    },
    {
      "epoch": 2.517727051966974,
      "grad_norm": 0.125,
      "learning_rate": 0.0001,
      "loss": 0.0318,
      "step": 1620
    },
    {
      "epoch": 2.525497814473045,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001,
      "loss": 0.0288,
      "step": 1625
    },
    {
      "epoch": 2.533268576979116,
      "grad_norm": 0.0556640625,
      "learning_rate": 0.0001,
      "loss": 0.0258,
      "step": 1630
    },
    {
      "epoch": 2.541039339485187,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0001,
      "loss": 0.03,
      "step": 1635
    },
    {
      "epoch": 2.5488101019912577,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0001,
      "loss": 0.0755,
      "step": 1640
    },
    {
      "epoch": 2.5565808644973287,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0001,
      "loss": 0.029,
      "step": 1645
    },
    {
      "epoch": 2.5643516270033997,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0001,
      "loss": 0.0105,
      "step": 1650
    },
    {
      "epoch": 2.5721223895094707,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0001,
      "loss": 0.0113,
      "step": 1655
    },
    {
      "epoch": 2.5798931520155417,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0001,
      "loss": 0.0124,
      "step": 1660
    },
    {
      "epoch": 2.5876639145216123,
      "grad_norm": 0.064453125,
      "learning_rate": 0.0001,
      "loss": 0.0364,
      "step": 1665
    },
    {
      "epoch": 2.5954346770276833,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0001,
      "loss": 0.0363,
      "step": 1670
    },
    {
      "epoch": 2.6032054395337543,
      "grad_norm": 0.109375,
      "learning_rate": 0.0001,
      "loss": 0.0296,
      "step": 1675
    },
    {
      "epoch": 2.6109762020398253,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0001,
      "loss": 0.0218,
      "step": 1680
    },
    {
      "epoch": 2.6187469645458963,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0001,
      "loss": 0.0346,
      "step": 1685
    },
    {
      "epoch": 2.626517727051967,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0001,
      "loss": 0.0738,
      "step": 1690
    },
    {
      "epoch": 2.634288489558038,
      "grad_norm": 0.037841796875,
      "learning_rate": 0.0001,
      "loss": 0.0247,
      "step": 1695
    },
    {
      "epoch": 2.642059252064109,
      "grad_norm": 0.00836181640625,
      "learning_rate": 0.0001,
      "loss": 0.0104,
      "step": 1700
    },
    {
      "epoch": 2.64983001457018,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0001,
      "loss": 0.017,
      "step": 1705
    },
    {
      "epoch": 2.657600777076251,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0001,
      "loss": 0.029,
      "step": 1710
    },
    {
      "epoch": 2.6653715395823214,
      "grad_norm": 0.0693359375,
      "learning_rate": 0.0001,
      "loss": 0.0319,
      "step": 1715
    },
    {
      "epoch": 2.6731423020883924,
      "grad_norm": 0.0498046875,
      "learning_rate": 0.0001,
      "loss": 0.0267,
      "step": 1720
    },
    {
      "epoch": 2.6809130645944634,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0001,
      "loss": 0.0352,
      "step": 1725
    },
    {
      "epoch": 2.688683827100534,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0001,
      "loss": 0.0296,
      "step": 1730
    },
    {
      "epoch": 2.696454589606605,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001,
      "loss": 0.0476,
      "step": 1735
    },
    {
      "epoch": 2.704225352112676,
      "grad_norm": 0.0712890625,
      "learning_rate": 0.0001,
      "loss": 0.0693,
      "step": 1740
    },
    {
      "epoch": 2.711996114618747,
      "grad_norm": 0.01177978515625,
      "learning_rate": 0.0001,
      "loss": 0.0263,
      "step": 1745
    },
    {
      "epoch": 2.719766877124818,
      "grad_norm": 0.048095703125,
      "learning_rate": 0.0001,
      "loss": 0.0117,
      "step": 1750
    },
    {
      "epoch": 2.7275376396308886,
      "grad_norm": 0.0390625,
      "learning_rate": 0.0001,
      "loss": 0.0121,
      "step": 1755
    },
    {
      "epoch": 2.7353084021369596,
      "grad_norm": 0.07861328125,
      "learning_rate": 0.0001,
      "loss": 0.0207,
      "step": 1760
    },
    {
      "epoch": 2.7430791646430306,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0001,
      "loss": 0.0459,
      "step": 1765
    },
    {
      "epoch": 2.7508499271491016,
      "grad_norm": 0.0576171875,
      "learning_rate": 0.0001,
      "loss": 0.0356,
      "step": 1770
    },
    {
      "epoch": 2.7586206896551726,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0001,
      "loss": 0.0385,
      "step": 1775
    },
    {
      "epoch": 2.766391452161243,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0001,
      "loss": 0.0476,
      "step": 1780
    },
    {
      "epoch": 2.774162214667314,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.034,
      "step": 1785
    },
    {
      "epoch": 2.781932977173385,
      "grad_norm": 0.06884765625,
      "learning_rate": 0.0001,
      "loss": 0.0657,
      "step": 1790
    },
    {
      "epoch": 2.789703739679456,
      "grad_norm": 0.07080078125,
      "learning_rate": 0.0001,
      "loss": 0.035,
      "step": 1795
    },
    {
      "epoch": 2.797474502185527,
      "grad_norm": 0.042724609375,
      "learning_rate": 0.0001,
      "loss": 0.0238,
      "step": 1800
    },
    {
      "epoch": 2.8052452646915977,
      "grad_norm": 0.0673828125,
      "learning_rate": 0.0001,
      "loss": 0.0119,
      "step": 1805
    },
    {
      "epoch": 2.8130160271976687,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0001,
      "loss": 0.0143,
      "step": 1810
    },
    {
      "epoch": 2.8207867897037397,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0001,
      "loss": 0.0368,
      "step": 1815
    },
    {
      "epoch": 2.8285575522098103,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0001,
      "loss": 0.0289,
      "step": 1820
    },
    {
      "epoch": 2.8363283147158818,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0001,
      "loss": 0.0301,
      "step": 1825
    },
    {
      "epoch": 2.8440990772219523,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0001,
      "loss": 0.0344,
      "step": 1830
    },
    {
      "epoch": 2.8518698397280233,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0001,
      "loss": 0.0246,
      "step": 1835
    },
    {
      "epoch": 2.8596406022340943,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0001,
      "loss": 0.0867,
      "step": 1840
    },
    {
      "epoch": 2.867411364740165,
      "grad_norm": 0.033203125,
      "learning_rate": 0.0001,
      "loss": 0.0426,
      "step": 1845
    },
    {
      "epoch": 2.875182127246236,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0001,
      "loss": 0.0329,
      "step": 1850
    },
    {
      "epoch": 2.882952889752307,
      "grad_norm": 0.0341796875,
      "learning_rate": 0.0001,
      "loss": 0.0092,
      "step": 1855
    },
    {
      "epoch": 2.890723652258378,
      "grad_norm": 0.0400390625,
      "learning_rate": 0.0001,
      "loss": 0.0262,
      "step": 1860
    },
    {
      "epoch": 2.898494414764449,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0001,
      "loss": 0.0433,
      "step": 1865
    },
    {
      "epoch": 2.9062651772705195,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0001,
      "loss": 0.0282,
      "step": 1870
    },
    {
      "epoch": 2.9140359397765905,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0001,
      "loss": 0.0286,
      "step": 1875
    },
    {
      "epoch": 2.9218067022826615,
      "grad_norm": 0.109375,
      "learning_rate": 0.0001,
      "loss": 0.0278,
      "step": 1880
    },
    {
      "epoch": 2.9295774647887325,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0001,
      "loss": 0.0216,
      "step": 1885
    },
    {
      "epoch": 2.9373482272948035,
      "grad_norm": 0.05517578125,
      "learning_rate": 0.0001,
      "loss": 0.0828,
      "step": 1890
    },
    {
      "epoch": 2.945118989800874,
      "grad_norm": 0.025390625,
      "learning_rate": 0.0001,
      "loss": 0.0286,
      "step": 1895
    },
    {
      "epoch": 2.952889752306945,
      "grad_norm": 0.01043701171875,
      "learning_rate": 0.0001,
      "loss": 0.0267,
      "step": 1900
    },
    {
      "epoch": 2.960660514813016,
      "grad_norm": 0.06201171875,
      "learning_rate": 0.0001,
      "loss": 0.0149,
      "step": 1905
    },
    {
      "epoch": 2.968431277319087,
      "grad_norm": 0.0615234375,
      "learning_rate": 0.0001,
      "loss": 0.0528,
      "step": 1910
    },
    {
      "epoch": 2.976202039825158,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0001,
      "loss": 0.0393,
      "step": 1915
    },
    {
      "epoch": 2.9839728023312286,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0001,
      "loss": 0.0304,
      "step": 1920
    },
    {
      "epoch": 2.9917435648372996,
      "grad_norm": 0.058349609375,
      "learning_rate": 0.0001,
      "loss": 0.0279,
      "step": 1925
    },
    {
      "epoch": 2.9995143273433706,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001,
      "loss": 0.0327,
      "step": 1930
    },
    {
      "epoch": 3.0072850898494417,
      "grad_norm": 0.058837890625,
      "learning_rate": 0.0001,
      "loss": 0.0456,
      "step": 1935
    },
    {
      "epoch": 3.015055852355512,
      "grad_norm": 0.041748046875,
      "learning_rate": 0.0001,
      "loss": 0.0142,
      "step": 1940
    },
    {
      "epoch": 3.022826614861583,
      "grad_norm": 0.052490234375,
      "learning_rate": 0.0001,
      "loss": 0.0105,
      "step": 1945
    },
    {
      "epoch": 3.030597377367654,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0001,
      "loss": 0.0103,
      "step": 1950
    }
  ],
  "logging_steps": 5,
  "max_steps": 1950,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 90,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.358681688807424e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}