{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 594,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003367003367003367,
      "grad_norm": 4.149549460211877,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 2.1262,
      "step": 1
    },
    {
      "epoch": 0.016835016835016835,
      "grad_norm": 4.010411470158707,
      "learning_rate": 8.333333333333333e-07,
      "loss": 2.0953,
      "step": 5
    },
    {
      "epoch": 0.03367003367003367,
      "grad_norm": 2.095167650626825,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.8934,
      "step": 10
    },
    {
      "epoch": 0.050505050505050504,
      "grad_norm": 3.1062734406420245,
      "learning_rate": 2.5e-06,
      "loss": 1.4715,
      "step": 15
    },
    {
      "epoch": 0.06734006734006734,
      "grad_norm": 0.5160710231319299,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.9113,
      "step": 20
    },
    {
      "epoch": 0.08417508417508418,
      "grad_norm": 0.42528861790572803,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.7253,
      "step": 25
    },
    {
      "epoch": 0.10101010101010101,
      "grad_norm": 0.27763208880764356,
      "learning_rate": 5e-06,
      "loss": 0.664,
      "step": 30
    },
    {
      "epoch": 0.11784511784511785,
      "grad_norm": 0.22702639326519608,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.5673,
      "step": 35
    },
    {
      "epoch": 0.13468013468013468,
      "grad_norm": 0.20027531882369276,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5462,
      "step": 40
    },
    {
      "epoch": 0.15151515151515152,
      "grad_norm": 0.22331242368395698,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.5208,
      "step": 45
    },
    {
      "epoch": 0.16835016835016836,
      "grad_norm": 0.2495290013519025,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.5063,
      "step": 50
    },
    {
      "epoch": 0.18518518518518517,
      "grad_norm": 0.22997291660332314,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.4763,
      "step": 55
    },
    {
      "epoch": 0.20202020202020202,
      "grad_norm": 0.1854483270058641,
      "learning_rate": 1e-05,
      "loss": 0.4797,
      "step": 60
    },
    {
      "epoch": 0.21885521885521886,
      "grad_norm": 0.19584468158412918,
      "learning_rate": 9.997836953115927e-06,
      "loss": 0.4705,
      "step": 65
    },
    {
      "epoch": 0.2356902356902357,
      "grad_norm": 0.20152991740174855,
      "learning_rate": 9.991349683972435e-06,
      "loss": 0.4539,
      "step": 70
    },
    {
      "epoch": 0.25252525252525254,
      "grad_norm": 0.16770527024092638,
      "learning_rate": 9.980543805476447e-06,
      "loss": 0.4611,
      "step": 75
    },
    {
      "epoch": 0.26936026936026936,
      "grad_norm": 0.1345303934796462,
      "learning_rate": 9.965428667076687e-06,
      "loss": 0.4208,
      "step": 80
    },
    {
      "epoch": 0.28619528619528617,
      "grad_norm": 0.14467263254757398,
      "learning_rate": 9.946017346674362e-06,
      "loss": 0.4291,
      "step": 85
    },
    {
      "epoch": 0.30303030303030304,
      "grad_norm": 0.11675624115717066,
      "learning_rate": 9.922326639307918e-06,
      "loss": 0.4118,
      "step": 90
    },
    {
      "epoch": 0.31986531986531985,
      "grad_norm": 0.12419897003228939,
      "learning_rate": 9.894377042621654e-06,
      "loss": 0.4169,
      "step": 95
    },
    {
      "epoch": 0.3367003367003367,
      "grad_norm": 0.1411171642205521,
      "learning_rate": 9.86219273913078e-06,
      "loss": 0.4336,
      "step": 100
    },
    {
      "epoch": 0.35353535353535354,
      "grad_norm": 0.10846956134773478,
      "learning_rate": 9.825801575298248e-06,
      "loss": 0.3873,
      "step": 105
    },
    {
      "epoch": 0.37037037037037035,
      "grad_norm": 0.12002423610117473,
      "learning_rate": 9.785235037441473e-06,
      "loss": 0.3619,
      "step": 110
    },
    {
      "epoch": 0.3872053872053872,
      "grad_norm": 0.11343783522484456,
      "learning_rate": 9.74052822448978e-06,
      "loss": 0.3993,
      "step": 115
    },
    {
      "epoch": 0.40404040404040403,
      "grad_norm": 0.11192347591145078,
      "learning_rate": 9.691719817616148e-06,
      "loss": 0.4268,
      "step": 120
    },
    {
      "epoch": 0.4208754208754209,
      "grad_norm": 0.1139058653749672,
      "learning_rate": 9.63885204676954e-06,
      "loss": 0.3725,
      "step": 125
    },
    {
      "epoch": 0.4377104377104377,
      "grad_norm": 0.1254583279635675,
      "learning_rate": 9.581970654136752e-06,
      "loss": 0.3645,
      "step": 130
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.11137112662931582,
      "learning_rate": 9.521124854565425e-06,
      "loss": 0.3796,
      "step": 135
    },
    {
      "epoch": 0.4713804713804714,
      "grad_norm": 0.12134068097374133,
      "learning_rate": 9.45636729298243e-06,
      "loss": 0.4135,
      "step": 140
    },
    {
      "epoch": 0.4882154882154882,
      "grad_norm": 0.10390787212244738,
      "learning_rate": 9.387753998844482e-06,
      "loss": 0.4136,
      "step": 145
    },
    {
      "epoch": 0.5050505050505051,
      "grad_norm": 0.10067346916725926,
      "learning_rate": 9.315344337660422e-06,
      "loss": 0.3814,
      "step": 150
    },
    {
      "epoch": 0.5218855218855218,
      "grad_norm": 0.12175671218216305,
      "learning_rate": 9.239200959627048e-06,
      "loss": 0.4118,
      "step": 155
    },
    {
      "epoch": 0.5387205387205387,
      "grad_norm": 0.10919894991390798,
      "learning_rate": 9.159389745423003e-06,
      "loss": 0.4231,
      "step": 160
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.10398902410046555,
      "learning_rate": 9.07597974920756e-06,
      "loss": 0.4062,
      "step": 165
    },
    {
      "epoch": 0.5723905723905723,
      "grad_norm": 0.100265725065611,
      "learning_rate": 8.98904313887369e-06,
      "loss": 0.3677,
      "step": 170
    },
    {
      "epoch": 0.5892255892255892,
      "grad_norm": 0.099307367621639,
      "learning_rate": 8.89865513360703e-06,
      "loss": 0.3758,
      "step": 175
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 0.09507905089725636,
      "learning_rate": 8.804893938804839e-06,
      "loss": 0.3703,
      "step": 180
    },
    {
      "epoch": 0.622895622895623,
      "grad_norm": 0.11024852756485287,
      "learning_rate": 8.707840678411223e-06,
      "loss": 0.4052,
      "step": 185
    },
    {
      "epoch": 0.6397306397306397,
      "grad_norm": 0.10365136825762392,
      "learning_rate": 8.607579324727175e-06,
      "loss": 0.3955,
      "step": 190
    },
    {
      "epoch": 0.6565656565656566,
      "grad_norm": 0.11091339739436984,
      "learning_rate": 8.504196625756166e-06,
      "loss": 0.3884,
      "step": 195
    },
    {
      "epoch": 0.6734006734006734,
      "grad_norm": 0.10685048356906915,
      "learning_rate": 8.397782030148147e-06,
      "loss": 0.3702,
      "step": 200
    },
    {
      "epoch": 0.6902356902356902,
      "grad_norm": 0.1246306899349896,
      "learning_rate": 8.288427609806899e-06,
      "loss": 0.4024,
      "step": 205
    },
    {
      "epoch": 0.7070707070707071,
      "grad_norm": 0.1066849900460222,
      "learning_rate": 8.176227980227693e-06,
      "loss": 0.4052,
      "step": 210
    },
    {
      "epoch": 0.7239057239057239,
      "grad_norm": 0.11875691065211648,
      "learning_rate": 8.061280218634192e-06,
      "loss": 0.3896,
      "step": 215
    },
    {
      "epoch": 0.7407407407407407,
      "grad_norm": 0.09762435583125909,
      "learning_rate": 7.943683779985412e-06,
      "loss": 0.3868,
      "step": 220
    },
    {
      "epoch": 0.7575757575757576,
      "grad_norm": 0.10054336444833502,
      "learning_rate": 7.823540410925434e-06,
      "loss": 0.3674,
      "step": 225
    },
    {
      "epoch": 0.7744107744107744,
      "grad_norm": 0.10842059524327032,
      "learning_rate": 7.700954061750295e-06,
      "loss": 0.3918,
      "step": 230
    },
    {
      "epoch": 0.7912457912457912,
      "grad_norm": 0.09745155947498828,
      "learning_rate": 7.576030796468233e-06,
      "loss": 0.3679,
      "step": 235
    },
    {
      "epoch": 0.8080808080808081,
      "grad_norm": 0.10372602136364313,
      "learning_rate": 7.4488787010311425e-06,
      "loss": 0.386,
      "step": 240
    },
    {
      "epoch": 0.8249158249158249,
      "grad_norm": 0.10925288274096943,
      "learning_rate": 7.319607789816555e-06,
      "loss": 0.3822,
      "step": 245
    },
    {
      "epoch": 0.8417508417508418,
      "grad_norm": 0.10360616006701923,
      "learning_rate": 7.188329910441154e-06,
      "loss": 0.3656,
      "step": 250
    },
    {
      "epoch": 0.8585858585858586,
      "grad_norm": 0.09128323758171268,
      "learning_rate": 7.05515864698811e-06,
      "loss": 0.3801,
      "step": 255
    },
    {
      "epoch": 0.8754208754208754,
      "grad_norm": 0.09321485064307464,
      "learning_rate": 6.920209221732007e-06,
      "loss": 0.3453,
      "step": 260
    },
    {
      "epoch": 0.8922558922558923,
      "grad_norm": 0.11568566960803582,
      "learning_rate": 6.783598395446371e-06,
      "loss": 0.4071,
      "step": 265
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.09087682600845605,
      "learning_rate": 6.64544436638005e-06,
      "loss": 0.3685,
      "step": 270
    },
    {
      "epoch": 0.9259259259259259,
      "grad_norm": 0.08403968322909508,
      "learning_rate": 6.505866667989884e-06,
      "loss": 0.3535,
      "step": 275
    },
    {
      "epoch": 0.9427609427609428,
      "grad_norm": 0.09338895709806748,
      "learning_rate": 6.364986065518106e-06,
      "loss": 0.3425,
      "step": 280
    },
    {
      "epoch": 0.9595959595959596,
      "grad_norm": 0.11009895402833714,
      "learning_rate": 6.222924451504001e-06,
      "loss": 0.3903,
      "step": 285
    },
    {
      "epoch": 0.9764309764309764,
      "grad_norm": 0.0962540505203609,
      "learning_rate": 6.079804740320181e-06,
      "loss": 0.3743,
      "step": 290
    },
    {
      "epoch": 0.9932659932659933,
      "grad_norm": 0.11859376799293515,
      "learning_rate": 5.935750761824777e-06,
      "loss": 0.3911,
      "step": 295
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.3753235638141632,
      "eval_runtime": 28.7233,
      "eval_samples_per_second": 20.506,
      "eval_steps_per_second": 5.153,
      "step": 297
    },
    {
      "epoch": 1.0101010101010102,
      "grad_norm": 0.09638846774704153,
      "learning_rate": 5.790887154221521e-06,
      "loss": 0.3592,
      "step": 300
    },
    {
      "epoch": 1.026936026936027,
      "grad_norm": 0.11218182778453092,
      "learning_rate": 5.645339256220427e-06,
      "loss": 0.3346,
      "step": 305
    },
    {
      "epoch": 1.0437710437710437,
      "grad_norm": 0.0894851822596739,
      "learning_rate": 5.499232998592399e-06,
      "loss": 0.3261,
      "step": 310
    },
    {
      "epoch": 1.0606060606060606,
      "grad_norm": 0.11093920612376884,
      "learning_rate": 5.352694795211555e-06,
      "loss": 0.3812,
      "step": 315
    },
    {
      "epoch": 1.0774410774410774,
      "grad_norm": 0.10664119581590888,
      "learning_rate": 5.20585143367959e-06,
      "loss": 0.3603,
      "step": 320
    },
    {
      "epoch": 1.0942760942760943,
      "grad_norm": 0.09997813791060199,
      "learning_rate": 5.058829965626742e-06,
      "loss": 0.3335,
      "step": 325
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 0.09409979391826565,
      "learning_rate": 4.911757596784358e-06,
      "loss": 0.3611,
      "step": 330
    },
    {
      "epoch": 1.127946127946128,
      "grad_norm": 0.10458387692519078,
      "learning_rate": 4.7647615769241e-06,
      "loss": 0.3335,
      "step": 335
    },
    {
      "epoch": 1.144781144781145,
      "grad_norm": 0.09111079710967974,
      "learning_rate": 4.617969089759066e-06,
      "loss": 0.3428,
      "step": 340
    },
    {
      "epoch": 1.1616161616161615,
      "grad_norm": 0.08443699111109511,
      "learning_rate": 4.471507142902036e-06,
      "loss": 0.3223,
      "step": 345
    },
    {
      "epoch": 1.1784511784511784,
      "grad_norm": 0.09133102611827909,
      "learning_rate": 4.325502457976126e-06,
      "loss": 0.3435,
      "step": 350
    },
    {
      "epoch": 1.1952861952861953,
      "grad_norm": 0.09066367143881884,
      "learning_rate": 4.180081360972852e-06,
      "loss": 0.3651,
      "step": 355
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 0.10084499213383416,
      "learning_rate": 4.035369672952516e-06,
      "loss": 0.3956,
      "step": 360
    },
    {
      "epoch": 1.228956228956229,
      "grad_norm": 0.09491527150886968,
      "learning_rate": 3.891492601181462e-06,
      "loss": 0.3453,
      "step": 365
    },
    {
      "epoch": 1.2457912457912457,
      "grad_norm": 0.1019704740198795,
      "learning_rate": 3.7485746308004013e-06,
      "loss": 0.3479,
      "step": 370
    },
    {
      "epoch": 1.2626262626262625,
      "grad_norm": 0.0968882851124518,
      "learning_rate": 3.6067394171175397e-06,
      "loss": 0.3303,
      "step": 375
    },
    {
      "epoch": 1.2794612794612794,
      "grad_norm": 0.08550746429570652,
      "learning_rate": 3.466109678619681e-06,
      "loss": 0.3269,
      "step": 380
    },
    {
      "epoch": 1.2962962962962963,
      "grad_norm": 0.10176082406674497,
      "learning_rate": 3.3268070907938915e-06,
      "loss": 0.3401,
      "step": 385
    },
    {
      "epoch": 1.3131313131313131,
      "grad_norm": 0.09276368484156704,
      "learning_rate": 3.1889521808515888e-06,
      "loss": 0.3308,
      "step": 390
    },
    {
      "epoch": 1.32996632996633,
      "grad_norm": 0.10581834870886321,
      "learning_rate": 3.0526642234461313e-06,
      "loss": 0.3238,
      "step": 395
    },
    {
      "epoch": 1.3468013468013469,
      "grad_norm": 0.10428102445335462,
      "learning_rate": 2.9180611374741623e-06,
      "loss": 0.3595,
      "step": 400
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 0.0899593621588282,
      "learning_rate": 2.785259384049959e-06,
      "loss": 0.3298,
      "step": 405
    },
    {
      "epoch": 1.3804713804713804,
      "grad_norm": 0.08797336350474357,
      "learning_rate": 2.6543738657411033e-06,
      "loss": 0.354,
      "step": 410
    },
    {
      "epoch": 1.3973063973063973,
      "grad_norm": 0.09021559950024842,
      "learning_rate": 2.525517827152614e-06,
      "loss": 0.3298,
      "step": 415
    },
    {
      "epoch": 1.4141414141414141,
      "grad_norm": 0.10219971409103493,
      "learning_rate": 2.3988027569455895e-06,
      "loss": 0.3468,
      "step": 420
    },
    {
      "epoch": 1.430976430976431,
      "grad_norm": 0.10290620157032898,
      "learning_rate": 2.274338291375147e-06,
      "loss": 0.3577,
      "step": 425
    },
    {
      "epoch": 1.4478114478114479,
      "grad_norm": 0.08826604210110597,
      "learning_rate": 2.1522321194310577e-06,
      "loss": 0.3473,
      "step": 430
    },
    {
      "epoch": 1.4646464646464645,
      "grad_norm": 0.08903591800358818,
      "learning_rate": 2.0325898896632178e-06,
      "loss": 0.3339,
      "step": 435
    },
    {
      "epoch": 1.4814814814814814,
      "grad_norm": 0.08848973658266114,
      "learning_rate": 1.915515118772555e-06,
      "loss": 0.3215,
      "step": 440
    },
    {
      "epoch": 1.4983164983164983,
      "grad_norm": 0.09450509559743611,
      "learning_rate": 1.8011091020464138e-06,
      "loss": 0.3519,
      "step": 445
    },
    {
      "epoch": 1.5151515151515151,
      "grad_norm": 0.08781947425372004,
      "learning_rate": 1.689470825715998e-06,
      "loss": 0.3344,
      "step": 450
    },
    {
      "epoch": 1.531986531986532,
      "grad_norm": 0.09293777777558654,
      "learning_rate": 1.580696881311611e-06,
      "loss": 0.3429,
      "step": 455
    },
    {
      "epoch": 1.5488215488215489,
      "grad_norm": 0.10333592478818118,
      "learning_rate": 1.4748813820898554e-06,
      "loss": 0.3583,
      "step": 460
    },
    {
      "epoch": 1.5656565656565657,
      "grad_norm": 0.10312555456577105,
      "learning_rate": 1.3721158816050872e-06,
      "loss": 0.3245,
      "step": 465
    },
    {
      "epoch": 1.5824915824915826,
      "grad_norm": 0.09235484309875117,
      "learning_rate": 1.272489294495548e-06,
      "loss": 0.3203,
      "step": 470
    },
    {
      "epoch": 1.5993265993265995,
      "grad_norm": 0.09652201115483773,
      "learning_rate": 1.1760878195527642e-06,
      "loss": 0.3346,
      "step": 475
    },
    {
      "epoch": 1.6161616161616161,
      "grad_norm": 0.10091941077870167,
      "learning_rate": 1.0829948651407374e-06,
      "loss": 0.3212,
      "step": 480
    },
    {
      "epoch": 1.632996632996633,
      "grad_norm": 0.08814459685515268,
      "learning_rate": 9.932909770294542e-07,
      "loss": 0.3278,
      "step": 485
    },
    {
      "epoch": 1.6498316498316499,
      "grad_norm": 0.09440301955370818,
      "learning_rate": 9.070537687051817e-07,
      "loss": 0.3194,
      "step": 490
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.08616003726483284,
      "learning_rate": 8.243578542178227e-07,
      "loss": 0.3066,
      "step": 495
    },
    {
      "epoch": 1.6835016835016834,
      "grad_norm": 0.09770744203946605,
      "learning_rate": 7.452747836234392e-07,
      "loss": 0.319,
      "step": 500
    },
    {
      "epoch": 1.7003367003367003,
      "grad_norm": 0.10679236845336801,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.3533,
      "step": 505
    },
    {
      "epoch": 1.7171717171717171,
      "grad_norm": 0.08984439281172861,
      "learning_rate": 5.982176856345445e-07,
      "loss": 0.3408,
      "step": 510
    },
    {
      "epoch": 1.734006734006734,
      "grad_norm": 0.0992116905666168,
      "learning_rate": 5.303708947990638e-07,
      "loss": 0.3397,
      "step": 515
    },
    {
      "epoch": 1.7508417508417509,
      "grad_norm": 0.08492852347673017,
      "learning_rate": 4.663913108871726e-07,
      "loss": 0.3661,
      "step": 520
    },
    {
      "epoch": 1.7676767676767677,
      "grad_norm": 0.08919997641532293,
      "learning_rate": 4.0633429023472004e-07,
      "loss": 0.3507,
      "step": 525
    },
    {
      "epoch": 1.7845117845117846,
      "grad_norm": 0.08884595301820054,
      "learning_rate": 3.5025179530225995e-07,
      "loss": 0.3475,
      "step": 530
    },
    {
      "epoch": 1.8013468013468015,
      "grad_norm": 0.10375781310984283,
      "learning_rate": 2.9819234971616154e-07,
      "loss": 0.3453,
      "step": 535
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.08617062897445908,
      "learning_rate": 2.5020099628504603e-07,
      "loss": 0.3225,
      "step": 540
    },
    {
      "epoch": 1.835016835016835,
      "grad_norm": 0.08459363441220882,
      "learning_rate": 2.0631925802791608e-07,
      "loss": 0.343,
      "step": 545
    },
    {
      "epoch": 1.8518518518518519,
      "grad_norm": 0.08738001505560129,
      "learning_rate": 1.6658510224765333e-07,
      "loss": 0.346,
      "step": 550
    },
    {
      "epoch": 1.8686868686868687,
      "grad_norm": 0.14091077594058707,
      "learning_rate": 1.3103290768099796e-07,
      "loss": 0.3525,
      "step": 555
    },
    {
      "epoch": 1.8855218855218854,
      "grad_norm": 0.10724350369660748,
      "learning_rate": 9.969343475342285e-08,
      "loss": 0.3415,
      "step": 560
    },
    {
      "epoch": 1.9023569023569022,
      "grad_norm": 0.10272663637676055,
      "learning_rate": 7.259379896463248e-08,
      "loss": 0.3283,
      "step": 565
    },
    {
      "epoch": 1.9191919191919191,
      "grad_norm": 0.10531925815392336,
      "learning_rate": 4.975744742772848e-08,
      "loss": 0.3437,
      "step": 570
    },
    {
      "epoch": 1.936026936026936,
      "grad_norm": 0.09578215650108557,
      "learning_rate": 3.120413858232474e-08,
      "loss": 0.3387,
      "step": 575
    },
    {
      "epoch": 1.9528619528619529,
      "grad_norm": 0.09866364512320212,
      "learning_rate": 1.69499250991767e-08,
      "loss": 0.327,
      "step": 580
    },
    {
      "epoch": 1.9696969696969697,
      "grad_norm": 0.09454492383929458,
      "learning_rate": 7.007139991108136e-09,
      "loss": 0.3122,
      "step": 585
    },
    {
      "epoch": 1.9865319865319866,
      "grad_norm": 0.10021678758376606,
      "learning_rate": 1.3843859422574269e-09,
      "loss": 0.3852,
      "step": 590
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.37048205733299255,
      "eval_runtime": 28.2507,
      "eval_samples_per_second": 20.849,
      "eval_steps_per_second": 5.239,
      "step": 594
    },
    {
      "epoch": 2.0,
      "step": 594,
      "total_flos": 1.5456743568084828e+18,
      "train_loss": 0.4185141025970279,
      "train_runtime": 4834.4452,
      "train_samples_per_second": 5.896,
      "train_steps_per_second": 0.123
    }
  ],
  "logging_steps": 5,
  "max_steps": 594,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5456743568084828e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}