{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.1946587537091988,
  "eval_steps": 126,
  "global_step": 252,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_loss": 3.176351308822632,
      "eval_runtime": 160.501,
      "eval_samples_per_second": 11.19,
      "eval_steps_per_second": 0.704,
      "step": 0
    },
    {
      "epoch": 0.004747774480712166,
      "grad_norm": 14.394155502319336,
      "learning_rate": 0.0,
      "loss": 3.5173,
      "step": 1
    },
    {
      "epoch": 0.009495548961424332,
      "grad_norm": 15.094048500061035,
      "learning_rate": 3.174603174603175e-07,
      "loss": 3.7637,
      "step": 2
    },
    {
      "epoch": 0.014243323442136498,
      "grad_norm": 15.361183166503906,
      "learning_rate": 6.34920634920635e-07,
      "loss": 3.7578,
      "step": 3
    },
    {
      "epoch": 0.018991097922848664,
      "grad_norm": 15.74487018585205,
      "learning_rate": 9.523809523809525e-07,
      "loss": 3.7525,
      "step": 4
    },
    {
      "epoch": 0.02373887240356083,
      "grad_norm": 13.999384880065918,
      "learning_rate": 1.26984126984127e-06,
      "loss": 3.6421,
      "step": 5
    },
    {
      "epoch": 0.028486646884272996,
      "grad_norm": 14.453660011291504,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 3.656,
      "step": 6
    },
    {
      "epoch": 0.03323442136498516,
      "grad_norm": 14.581559181213379,
      "learning_rate": 1.904761904761905e-06,
      "loss": 3.7368,
      "step": 7
    },
    {
      "epoch": 0.03798219584569733,
      "grad_norm": 13.600449562072754,
      "learning_rate": 2.222222222222222e-06,
      "loss": 3.5658,
      "step": 8
    },
    {
      "epoch": 0.042729970326409496,
      "grad_norm": 11.520191192626953,
      "learning_rate": 2.53968253968254e-06,
      "loss": 3.3966,
      "step": 9
    },
    {
      "epoch": 0.04747774480712166,
      "grad_norm": 11.848709106445312,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 3.4279,
      "step": 10
    },
    {
      "epoch": 0.05222551928783383,
      "grad_norm": 11.579791069030762,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 3.4126,
      "step": 11
    },
    {
      "epoch": 0.05697329376854599,
      "grad_norm": 10.861617088317871,
      "learning_rate": 3.492063492063492e-06,
      "loss": 3.1806,
      "step": 12
    },
    {
      "epoch": 0.06172106824925816,
      "grad_norm": 11.614352226257324,
      "learning_rate": 3.80952380952381e-06,
      "loss": 3.0254,
      "step": 13
    },
    {
      "epoch": 0.06646884272997032,
      "grad_norm": 11.77472972869873,
      "learning_rate": 4.126984126984127e-06,
      "loss": 2.9602,
      "step": 14
    },
    {
      "epoch": 0.0712166172106825,
      "grad_norm": 11.360806465148926,
      "learning_rate": 4.444444444444444e-06,
      "loss": 2.9155,
      "step": 15
    },
    {
      "epoch": 0.07596439169139466,
      "grad_norm": 12.338720321655273,
      "learning_rate": 4.761904761904762e-06,
      "loss": 2.4742,
      "step": 16
    },
    {
      "epoch": 0.08071216617210683,
      "grad_norm": 13.61253833770752,
      "learning_rate": 5.07936507936508e-06,
      "loss": 2.4264,
      "step": 17
    },
    {
      "epoch": 0.08545994065281899,
      "grad_norm": 13.436763763427734,
      "learning_rate": 5.396825396825397e-06,
      "loss": 2.323,
      "step": 18
    },
    {
      "epoch": 0.09020771513353115,
      "grad_norm": 11.5920991897583,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 2.0951,
      "step": 19
    },
    {
      "epoch": 0.09495548961424333,
      "grad_norm": 10.541767120361328,
      "learning_rate": 6.031746031746032e-06,
      "loss": 1.9129,
      "step": 20
    },
    {
      "epoch": 0.09970326409495549,
      "grad_norm": 10.552443504333496,
      "learning_rate": 6.349206349206349e-06,
      "loss": 1.7977,
      "step": 21
    },
    {
      "epoch": 0.10445103857566766,
      "grad_norm": 10.990846633911133,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.584,
      "step": 22
    },
    {
      "epoch": 0.10919881305637982,
      "grad_norm": 9.763855934143066,
      "learning_rate": 6.984126984126984e-06,
      "loss": 1.401,
      "step": 23
    },
    {
      "epoch": 0.11394658753709198,
      "grad_norm": 7.580389499664307,
      "learning_rate": 7.301587301587301e-06,
      "loss": 1.2337,
      "step": 24
    },
    {
      "epoch": 0.11869436201780416,
      "grad_norm": 7.879193305969238,
      "learning_rate": 7.61904761904762e-06,
      "loss": 1.185,
      "step": 25
    },
    {
      "epoch": 0.12344213649851632,
      "grad_norm": 7.337639331817627,
      "learning_rate": 7.936507936507936e-06,
      "loss": 1.0778,
      "step": 26
    },
    {
      "epoch": 0.1281899109792285,
      "grad_norm": 5.66098690032959,
      "learning_rate": 8.253968253968254e-06,
      "loss": 0.9721,
      "step": 27
    },
    {
      "epoch": 0.13293768545994064,
      "grad_norm": 4.907323360443115,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.9584,
      "step": 28
    },
    {
      "epoch": 0.13768545994065282,
      "grad_norm": 3.510972023010254,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.8819,
      "step": 29
    },
    {
      "epoch": 0.142433234421365,
      "grad_norm": 3.220461368560791,
      "learning_rate": 9.206349206349207e-06,
      "loss": 0.8042,
      "step": 30
    },
    {
      "epoch": 0.14718100890207717,
      "grad_norm": 2.63944935798645,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.7757,
      "step": 31
    },
    {
      "epoch": 0.1519287833827893,
      "grad_norm": 11.566150665283203,
      "learning_rate": 9.841269841269842e-06,
      "loss": 0.6714,
      "step": 32
    },
    {
      "epoch": 0.1566765578635015,
      "grad_norm": 2.1101737022399902,
      "learning_rate": 1.015873015873016e-05,
      "loss": 0.6488,
      "step": 33
    },
    {
      "epoch": 0.16142433234421366,
      "grad_norm": 3.368835926055908,
      "learning_rate": 1.0476190476190477e-05,
      "loss": 0.6551,
      "step": 34
    },
    {
      "epoch": 0.1661721068249258,
      "grad_norm": 2.465441942214966,
      "learning_rate": 1.0793650793650794e-05,
      "loss": 0.6039,
      "step": 35
    },
    {
      "epoch": 0.17091988130563798,
      "grad_norm": 1.3464806079864502,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 0.5671,
      "step": 36
    },
    {
      "epoch": 0.17566765578635016,
      "grad_norm": 2.223472833633423,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.5821,
      "step": 37
    },
    {
      "epoch": 0.1804154302670623,
      "grad_norm": 1.5176966190338135,
      "learning_rate": 1.1746031746031748e-05,
      "loss": 0.5397,
      "step": 38
    },
    {
      "epoch": 0.18516320474777448,
      "grad_norm": 2.944708824157715,
      "learning_rate": 1.2063492063492064e-05,
      "loss": 0.5455,
      "step": 39
    },
    {
      "epoch": 0.18991097922848665,
      "grad_norm": 1.5701731443405151,
      "learning_rate": 1.2380952380952383e-05,
      "loss": 0.5256,
      "step": 40
    },
    {
      "epoch": 0.1946587537091988,
      "grad_norm": 1.0416580438613892,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 0.4746,
      "step": 41
    },
    {
      "epoch": 0.19940652818991098,
      "grad_norm": 1.0439484119415283,
      "learning_rate": 1.3015873015873018e-05,
      "loss": 0.4844,
      "step": 42
    },
    {
      "epoch": 0.20415430267062315,
      "grad_norm": 2.974339485168457,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.509,
      "step": 43
    },
    {
      "epoch": 0.20890207715133532,
      "grad_norm": 1.0552765130996704,
      "learning_rate": 1.3650793650793652e-05,
      "loss": 0.5047,
      "step": 44
    },
    {
      "epoch": 0.21364985163204747,
      "grad_norm": 3.1413824558258057,
      "learning_rate": 1.3968253968253968e-05,
      "loss": 0.4924,
      "step": 45
    },
    {
      "epoch": 0.21839762611275965,
      "grad_norm": 1.9047034978866577,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.456,
      "step": 46
    },
    {
      "epoch": 0.22314540059347182,
      "grad_norm": 1.2171484231948853,
      "learning_rate": 1.4603174603174603e-05,
      "loss": 0.4459,
      "step": 47
    },
    {
      "epoch": 0.22789317507418397,
      "grad_norm": 4.339561462402344,
      "learning_rate": 1.4920634920634922e-05,
      "loss": 0.4565,
      "step": 48
    },
    {
      "epoch": 0.23264094955489614,
      "grad_norm": 2.3773863315582275,
      "learning_rate": 1.523809523809524e-05,
      "loss": 0.4568,
      "step": 49
    },
    {
      "epoch": 0.23738872403560832,
      "grad_norm": 1.8882341384887695,
      "learning_rate": 1.555555555555556e-05,
      "loss": 0.4452,
      "step": 50
    },
    {
      "epoch": 0.24213649851632046,
      "grad_norm": 1.1871311664581299,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 0.4238,
      "step": 51
    },
    {
      "epoch": 0.24688427299703264,
      "grad_norm": 2.8959860801696777,
      "learning_rate": 1.6190476190476193e-05,
      "loss": 0.426,
      "step": 52
    },
    {
      "epoch": 0.2516320474777448,
      "grad_norm": 0.9582217335700989,
      "learning_rate": 1.6507936507936507e-05,
      "loss": 0.4246,
      "step": 53
    },
    {
      "epoch": 0.256379821958457,
      "grad_norm": 0.8898813128471375,
      "learning_rate": 1.6825396825396828e-05,
      "loss": 0.4072,
      "step": 54
    },
    {
      "epoch": 0.26112759643916916,
      "grad_norm": 0.9023370742797852,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.4083,
      "step": 55
    },
    {
      "epoch": 0.2658753709198813,
      "grad_norm": 1.1952933073043823,
      "learning_rate": 1.7460317460317463e-05,
      "loss": 0.401,
      "step": 56
    },
    {
      "epoch": 0.27062314540059346,
      "grad_norm": 1.1986902952194214,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.4006,
      "step": 57
    },
    {
      "epoch": 0.27537091988130563,
      "grad_norm": 0.6974486112594604,
      "learning_rate": 1.8095238095238097e-05,
      "loss": 0.4173,
      "step": 58
    },
    {
      "epoch": 0.2801186943620178,
      "grad_norm": 1.2794380187988281,
      "learning_rate": 1.8412698412698415e-05,
      "loss": 0.3893,
      "step": 59
    },
    {
      "epoch": 0.28486646884273,
      "grad_norm": 0.7234415411949158,
      "learning_rate": 1.8730158730158732e-05,
      "loss": 0.3904,
      "step": 60
    },
    {
      "epoch": 0.28961424332344216,
      "grad_norm": 0.6361059546470642,
      "learning_rate": 1.904761904761905e-05,
      "loss": 0.4136,
      "step": 61
    },
    {
      "epoch": 0.29436201780415433,
      "grad_norm": 1.7151949405670166,
      "learning_rate": 1.9365079365079367e-05,
      "loss": 0.3943,
      "step": 62
    },
    {
      "epoch": 0.29910979228486645,
      "grad_norm": 0.6753223538398743,
      "learning_rate": 1.9682539682539684e-05,
      "loss": 0.3938,
      "step": 63
    },
    {
      "epoch": 0.3038575667655786,
      "grad_norm": 2.0604114532470703,
      "learning_rate": 2e-05,
      "loss": 0.3909,
      "step": 64
    },
    {
      "epoch": 0.3086053412462908,
      "grad_norm": 0.7779602408409119,
      "learning_rate": 1.9999846502070808e-05,
      "loss": 0.3661,
      "step": 65
    },
    {
      "epoch": 0.313353115727003,
      "grad_norm": 0.7002611756324768,
      "learning_rate": 1.9999386012995554e-05,
      "loss": 0.3522,
      "step": 66
    },
    {
      "epoch": 0.31810089020771515,
      "grad_norm": 0.6202103495597839,
      "learning_rate": 1.999861854691106e-05,
      "loss": 0.3831,
      "step": 67
    },
    {
      "epoch": 0.3228486646884273,
      "grad_norm": 0.44983768463134766,
      "learning_rate": 1.9997544127378217e-05,
      "loss": 0.37,
      "step": 68
    },
    {
      "epoch": 0.32759643916913944,
      "grad_norm": 0.6840509176254272,
      "learning_rate": 1.999616278738126e-05,
      "loss": 0.3751,
      "step": 69
    },
    {
      "epoch": 0.3323442136498516,
      "grad_norm": 0.416205495595932,
      "learning_rate": 1.999447456932676e-05,
      "loss": 0.3535,
      "step": 70
    },
    {
      "epoch": 0.3370919881305638,
      "grad_norm": 0.5574557185173035,
      "learning_rate": 1.9992479525042305e-05,
      "loss": 0.3792,
      "step": 71
    },
    {
      "epoch": 0.34183976261127597,
      "grad_norm": 0.46201279759407043,
      "learning_rate": 1.9990177715774927e-05,
      "loss": 0.3519,
      "step": 72
    },
    {
      "epoch": 0.34658753709198814,
      "grad_norm": 0.4977426826953888,
      "learning_rate": 1.9987569212189224e-05,
      "loss": 0.3488,
      "step": 73
    },
    {
      "epoch": 0.3513353115727003,
      "grad_norm": 0.45472055673599243,
      "learning_rate": 1.9984654094365175e-05,
      "loss": 0.3602,
      "step": 74
    },
    {
      "epoch": 0.3560830860534125,
      "grad_norm": 0.4730567932128906,
      "learning_rate": 1.9981432451795687e-05,
      "loss": 0.3583,
      "step": 75
    },
    {
      "epoch": 0.3608308605341246,
      "grad_norm": 0.49239611625671387,
      "learning_rate": 1.997790438338385e-05,
      "loss": 0.3474,
      "step": 76
    },
    {
      "epoch": 0.3655786350148368,
      "grad_norm": 0.3924044370651245,
      "learning_rate": 1.997406999743991e-05,
      "loss": 0.3538,
      "step": 77
    },
    {
      "epoch": 0.37032640949554896,
      "grad_norm": 0.4983470141887665,
      "learning_rate": 1.996992941167792e-05,
      "loss": 0.3699,
      "step": 78
    },
    {
      "epoch": 0.37507418397626113,
      "grad_norm": 0.4259602129459381,
      "learning_rate": 1.9965482753212154e-05,
      "loss": 0.3529,
      "step": 79
    },
    {
      "epoch": 0.3798219584569733,
      "grad_norm": 0.3953029215335846,
      "learning_rate": 1.9960730158553186e-05,
      "loss": 0.3514,
      "step": 80
    },
    {
      "epoch": 0.3845697329376855,
      "grad_norm": 0.7401227951049805,
      "learning_rate": 1.99556717736037e-05,
      "loss": 0.3795,
      "step": 81
    },
    {
      "epoch": 0.3893175074183976,
      "grad_norm": 2.0949933528900146,
      "learning_rate": 1.9950307753654016e-05,
      "loss": 0.3747,
      "step": 82
    },
    {
      "epoch": 0.3940652818991098,
      "grad_norm": 0.48020580410957336,
      "learning_rate": 1.9944638263377332e-05,
      "loss": 0.3678,
      "step": 83
    },
    {
      "epoch": 0.39881305637982195,
      "grad_norm": 0.4131617248058319,
      "learning_rate": 1.9938663476824646e-05,
      "loss": 0.3787,
      "step": 84
    },
    {
      "epoch": 0.4035608308605341,
      "grad_norm": 0.4024046063423157,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 0.3402,
      "step": 85
    },
    {
      "epoch": 0.4083086053412463,
      "grad_norm": 0.35955172777175903,
      "learning_rate": 1.9925798757952003e-05,
      "loss": 0.3605,
      "step": 86
    },
    {
      "epoch": 0.4130563798219585,
      "grad_norm": 0.4338579773902893,
      "learning_rate": 1.9918909220573588e-05,
      "loss": 0.357,
      "step": 87
    },
    {
      "epoch": 0.41780415430267065,
      "grad_norm": 0.3916882872581482,
      "learning_rate": 1.991171517679013e-05,
      "loss": 0.354,
      "step": 88
    },
    {
      "epoch": 0.42255192878338277,
      "grad_norm": 0.39583539962768555,
      "learning_rate": 1.9904216847455795e-05,
      "loss": 0.3284,
      "step": 89
    },
    {
      "epoch": 0.42729970326409494,
      "grad_norm": 0.43609434366226196,
      "learning_rate": 1.9896414462766188e-05,
      "loss": 0.3424,
      "step": 90
    },
    {
      "epoch": 0.4320474777448071,
      "grad_norm": 0.4122212529182434,
      "learning_rate": 1.9888308262251286e-05,
      "loss": 0.3386,
      "step": 91
    },
    {
      "epoch": 0.4367952522255193,
      "grad_norm": 0.3249572515487671,
      "learning_rate": 1.9879898494768093e-05,
      "loss": 0.3157,
      "step": 92
    },
    {
      "epoch": 0.44154302670623147,
      "grad_norm": 0.745229959487915,
      "learning_rate": 1.9871185418492978e-05,
      "loss": 0.3738,
      "step": 93
    },
    {
      "epoch": 0.44629080118694364,
      "grad_norm": 0.39437487721443176,
      "learning_rate": 1.9862169300913784e-05,
      "loss": 0.3386,
      "step": 94
    },
    {
      "epoch": 0.45103857566765576,
      "grad_norm": 0.3622061312198639,
      "learning_rate": 1.985285041882158e-05,
      "loss": 0.3234,
      "step": 95
    },
    {
      "epoch": 0.45578635014836794,
      "grad_norm": 0.4053417146205902,
      "learning_rate": 1.9843229058302192e-05,
      "loss": 0.3608,
      "step": 96
    },
    {
      "epoch": 0.4605341246290801,
      "grad_norm": 0.4641744792461395,
      "learning_rate": 1.9833305514727396e-05,
      "loss": 0.3431,
      "step": 97
    },
    {
      "epoch": 0.4652818991097923,
      "grad_norm": 0.39851275086402893,
      "learning_rate": 1.9823080092745878e-05,
      "loss": 0.3531,
      "step": 98
    },
    {
      "epoch": 0.47002967359050446,
      "grad_norm": 0.3545701205730438,
      "learning_rate": 1.9812553106273848e-05,
      "loss": 0.3842,
      "step": 99
    },
    {
      "epoch": 0.47477744807121663,
      "grad_norm": 0.37058180570602417,
      "learning_rate": 1.9801724878485438e-05,
      "loss": 0.3407,
      "step": 100
    },
    {
      "epoch": 0.4795252225519288,
      "grad_norm": 0.33017316460609436,
      "learning_rate": 1.9790595741802757e-05,
      "loss": 0.3369,
      "step": 101
    },
    {
      "epoch": 0.48427299703264093,
      "grad_norm": 0.3639221787452698,
      "learning_rate": 1.9779166037885692e-05,
      "loss": 0.3427,
      "step": 102
    },
    {
      "epoch": 0.4890207715133531,
      "grad_norm": 0.43298575282096863,
      "learning_rate": 1.9767436117621416e-05,
      "loss": 0.3693,
      "step": 103
    },
    {
      "epoch": 0.4937685459940653,
      "grad_norm": 0.33556169271469116,
      "learning_rate": 1.9755406341113622e-05,
      "loss": 0.3266,
      "step": 104
    },
    {
      "epoch": 0.49851632047477745,
      "grad_norm": 0.33984947204589844,
      "learning_rate": 1.974307707767147e-05,
      "loss": 0.3309,
      "step": 105
    },
    {
      "epoch": 0.5032640949554896,
      "grad_norm": 0.35068896412849426,
      "learning_rate": 1.973044870579824e-05,
      "loss": 0.3516,
      "step": 106
    },
    {
      "epoch": 0.5080118694362018,
      "grad_norm": 0.3809346556663513,
      "learning_rate": 1.971752161317972e-05,
      "loss": 0.3361,
      "step": 107
    },
    {
      "epoch": 0.512759643916914,
      "grad_norm": 0.359453946352005,
      "learning_rate": 1.9704296196672298e-05,
      "loss": 0.3737,
      "step": 108
    },
    {
      "epoch": 0.5175074183976262,
      "grad_norm": 0.3776707947254181,
      "learning_rate": 1.969077286229078e-05,
      "loss": 0.336,
      "step": 109
    },
    {
      "epoch": 0.5222551928783383,
      "grad_norm": 0.37052029371261597,
      "learning_rate": 1.9676952025195937e-05,
      "loss": 0.3278,
      "step": 110
    },
    {
      "epoch": 0.5270029673590505,
      "grad_norm": 0.37363845109939575,
      "learning_rate": 1.966283410968174e-05,
      "loss": 0.3481,
      "step": 111
    },
    {
      "epoch": 0.5317507418397626,
      "grad_norm": 0.3355433940887451,
      "learning_rate": 1.964841954916235e-05,
      "loss": 0.3379,
      "step": 112
    },
    {
      "epoch": 0.5364985163204747,
      "grad_norm": 0.3838210105895996,
      "learning_rate": 1.9633708786158803e-05,
      "loss": 0.3388,
      "step": 113
    },
    {
      "epoch": 0.5412462908011869,
      "grad_norm": 0.3923501670360565,
      "learning_rate": 1.9618702272285434e-05,
      "loss": 0.3749,
      "step": 114
    },
    {
      "epoch": 0.5459940652818991,
      "grad_norm": 0.4121781885623932,
      "learning_rate": 1.9603400468236e-05,
      "loss": 0.3149,
      "step": 115
    },
    {
      "epoch": 0.5507418397626113,
      "grad_norm": 0.31165602803230286,
      "learning_rate": 1.9587803843769547e-05,
      "loss": 0.3162,
      "step": 116
    },
    {
      "epoch": 0.5554896142433234,
      "grad_norm": 0.3426300883293152,
      "learning_rate": 1.9571912877695995e-05,
      "loss": 0.3408,
      "step": 117
    },
    {
      "epoch": 0.5602373887240356,
      "grad_norm": 0.33053115010261536,
      "learning_rate": 1.955572805786141e-05,
      "loss": 0.3397,
      "step": 118
    },
    {
      "epoch": 0.5649851632047478,
      "grad_norm": 0.30746808648109436,
      "learning_rate": 1.9539249881133062e-05,
      "loss": 0.3287,
      "step": 119
    },
    {
      "epoch": 0.56973293768546,
      "grad_norm": 0.3441820442676544,
      "learning_rate": 1.9522478853384154e-05,
      "loss": 0.3225,
      "step": 120
    },
    {
      "epoch": 0.5744807121661721,
      "grad_norm": 0.35982808470726013,
      "learning_rate": 1.9505415489478293e-05,
      "loss": 0.3429,
      "step": 121
    },
    {
      "epoch": 0.5792284866468843,
      "grad_norm": 0.3274133801460266,
      "learning_rate": 1.948806031325368e-05,
      "loss": 0.3436,
      "step": 122
    },
    {
      "epoch": 0.5839762611275965,
      "grad_norm": 0.38329532742500305,
      "learning_rate": 1.9470413857507036e-05,
      "loss": 0.3211,
      "step": 123
    },
    {
      "epoch": 0.5887240356083087,
      "grad_norm": 0.38898247480392456,
      "learning_rate": 1.945247666397725e-05,
      "loss": 0.3409,
      "step": 124
    },
    {
      "epoch": 0.5934718100890207,
      "grad_norm": 0.3356146216392517,
      "learning_rate": 1.943424928332873e-05,
      "loss": 0.3223,
      "step": 125
    },
    {
      "epoch": 0.5982195845697329,
      "grad_norm": 0.33519861102104187,
      "learning_rate": 1.9415732275134515e-05,
      "loss": 0.3369,
      "step": 126
    },
    {
      "epoch": 0.5982195845697329,
      "eval_loss": 0.15260468423366547,
      "eval_runtime": 142.1665,
      "eval_samples_per_second": 12.633,
      "eval_steps_per_second": 0.795,
      "step": 126
    },
    {
      "epoch": 0.6029673590504451,
      "grad_norm": 0.31465238332748413,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.3189,
      "step": 127
    },
    {
      "epoch": 0.6077151335311572,
      "grad_norm": 0.33517467975616455,
      "learning_rate": 1.937783165884092e-05,
      "loss": 0.3525,
      "step": 128
    },
    {
      "epoch": 0.6124629080118694,
      "grad_norm": 0.35138949751853943,
      "learning_rate": 1.9358449214274763e-05,
      "loss": 0.3466,
      "step": 129
    },
    {
      "epoch": 0.6172106824925816,
      "grad_norm": 0.3277546167373657,
      "learning_rate": 1.9338779469193638e-05,
      "loss": 0.3304,
      "step": 130
    },
    {
      "epoch": 0.6219584569732938,
      "grad_norm": 5.854161262512207,
      "learning_rate": 1.931882302745057e-05,
      "loss": 0.4189,
      "step": 131
    },
    {
      "epoch": 0.626706231454006,
      "grad_norm": 0.3688313961029053,
      "learning_rate": 1.9298580501700058e-05,
      "loss": 0.3393,
      "step": 132
    },
    {
      "epoch": 0.6314540059347181,
      "grad_norm": 0.3257688581943512,
      "learning_rate": 1.9278052513379256e-05,
      "loss": 0.331,
      "step": 133
    },
    {
      "epoch": 0.6362017804154303,
      "grad_norm": 0.33901330828666687,
      "learning_rate": 1.9257239692688907e-05,
      "loss": 0.342,
      "step": 134
    },
    {
      "epoch": 0.6409495548961425,
      "grad_norm": 0.3109529912471771,
      "learning_rate": 1.9236142678573983e-05,
      "loss": 0.3151,
      "step": 135
    },
    {
      "epoch": 0.6456973293768546,
      "grad_norm": 0.34090420603752136,
      "learning_rate": 1.921476211870408e-05,
      "loss": 0.3293,
      "step": 136
    },
    {
      "epoch": 0.6504451038575668,
      "grad_norm": 0.3669479787349701,
      "learning_rate": 1.9193098669453532e-05,
      "loss": 0.3574,
      "step": 137
    },
    {
      "epoch": 0.6551928783382789,
      "grad_norm": 0.3427859842777252,
      "learning_rate": 1.9171152995881257e-05,
      "loss": 0.3352,
      "step": 138
    },
    {
      "epoch": 0.6599406528189911,
      "grad_norm": 0.3781029284000397,
      "learning_rate": 1.9148925771710347e-05,
      "loss": 0.3384,
      "step": 139
    },
    {
      "epoch": 0.6646884272997032,
      "grad_norm": 0.4394756257534027,
      "learning_rate": 1.912641767930738e-05,
      "loss": 0.3234,
      "step": 140
    },
    {
      "epoch": 0.6694362017804154,
      "grad_norm": 0.3640480637550354,
      "learning_rate": 1.9103629409661468e-05,
      "loss": 0.3266,
      "step": 141
    },
    {
      "epoch": 0.6741839762611276,
      "grad_norm": 0.32769304513931274,
      "learning_rate": 1.908056166236305e-05,
      "loss": 0.3255,
      "step": 142
    },
    {
      "epoch": 0.6789317507418398,
      "grad_norm": 0.3409084379673004,
      "learning_rate": 1.9057215145582418e-05,
      "loss": 0.3285,
      "step": 143
    },
    {
      "epoch": 0.6836795252225519,
      "grad_norm": 0.3296876549720764,
      "learning_rate": 1.9033590576047967e-05,
      "loss": 0.3197,
      "step": 144
    },
    {
      "epoch": 0.6884272997032641,
      "grad_norm": 0.3788384199142456,
      "learning_rate": 1.900968867902419e-05,
      "loss": 0.33,
      "step": 145
    },
    {
      "epoch": 0.6931750741839763,
      "grad_norm": 0.31414154171943665,
      "learning_rate": 1.898551018828944e-05,
      "loss": 0.313,
      "step": 146
    },
    {
      "epoch": 0.6979228486646885,
      "grad_norm": 0.4816839396953583,
      "learning_rate": 1.8961055846113358e-05,
      "loss": 0.3407,
      "step": 147
    },
    {
      "epoch": 0.7026706231454006,
      "grad_norm": 0.3418005406856537,
      "learning_rate": 1.8936326403234125e-05,
      "loss": 0.3311,
      "step": 148
    },
    {
      "epoch": 0.7074183976261128,
      "grad_norm": 0.31916749477386475,
      "learning_rate": 1.8911322618835393e-05,
      "loss": 0.3042,
      "step": 149
    },
    {
      "epoch": 0.712166172106825,
      "grad_norm": 1.1523184776306152,
      "learning_rate": 1.888604526052299e-05,
      "loss": 0.3218,
      "step": 150
    },
    {
      "epoch": 0.716913946587537,
      "grad_norm": 0.2973329424858093,
      "learning_rate": 1.8860495104301346e-05,
      "loss": 0.3061,
      "step": 151
    },
    {
      "epoch": 0.7216617210682492,
      "grad_norm": 0.3566143214702606,
      "learning_rate": 1.8834672934549677e-05,
      "loss": 0.3287,
      "step": 152
    },
    {
      "epoch": 0.7264094955489614,
      "grad_norm": 0.3308376967906952,
      "learning_rate": 1.8808579543997892e-05,
      "loss": 0.3034,
      "step": 153
    },
    {
      "epoch": 0.7311572700296736,
      "grad_norm": 0.33520743250846863,
      "learning_rate": 1.8782215733702286e-05,
      "loss": 0.3296,
      "step": 154
    },
    {
      "epoch": 0.7359050445103857,
      "grad_norm": 0.3926364481449127,
      "learning_rate": 1.8755582313020912e-05,
      "loss": 0.3135,
      "step": 155
    },
    {
      "epoch": 0.7406528189910979,
      "grad_norm": 0.37003567814826965,
      "learning_rate": 1.8728680099588748e-05,
      "loss": 0.3455,
      "step": 156
    },
    {
      "epoch": 0.7454005934718101,
      "grad_norm": 0.31512436270713806,
      "learning_rate": 1.870150991929261e-05,
      "loss": 0.3146,
      "step": 157
    },
    {
      "epoch": 0.7501483679525223,
      "grad_norm": 0.35992729663848877,
      "learning_rate": 1.867407260624578e-05,
      "loss": 0.3138,
      "step": 158
    },
    {
      "epoch": 0.7548961424332344,
      "grad_norm": 0.33734387159347534,
      "learning_rate": 1.864636900276241e-05,
      "loss": 0.3453,
      "step": 159
    },
    {
      "epoch": 0.7596439169139466,
      "grad_norm": 0.403726726770401,
      "learning_rate": 1.8618399959331642e-05,
      "loss": 0.3176,
      "step": 160
    },
    {
      "epoch": 0.7643916913946588,
      "grad_norm": 0.33617129921913147,
      "learning_rate": 1.8590166334591533e-05,
      "loss": 0.3235,
      "step": 161
    },
    {
      "epoch": 0.769139465875371,
      "grad_norm": 0.3692370355129242,
      "learning_rate": 1.8561668995302668e-05,
      "loss": 0.3167,
      "step": 162
    },
    {
      "epoch": 0.7738872403560831,
      "grad_norm": 0.31387221813201904,
      "learning_rate": 1.8532908816321557e-05,
      "loss": 0.2978,
      "step": 163
    },
    {
      "epoch": 0.7786350148367952,
      "grad_norm": 0.39496850967407227,
      "learning_rate": 1.850388668057379e-05,
      "loss": 0.3167,
      "step": 164
    },
    {
      "epoch": 0.7833827893175074,
      "grad_norm": 0.29015660285949707,
      "learning_rate": 1.8474603479026912e-05,
      "loss": 0.3111,
      "step": 165
    },
    {
      "epoch": 0.7881305637982196,
      "grad_norm": 0.4949202835559845,
      "learning_rate": 1.844506011066308e-05,
      "loss": 0.3231,
      "step": 166
    },
    {
      "epoch": 0.7928783382789317,
      "grad_norm": 0.3161013126373291,
      "learning_rate": 1.841525748245147e-05,
      "loss": 0.3235,
      "step": 167
    },
    {
      "epoch": 0.7976261127596439,
      "grad_norm": 0.3269205093383789,
      "learning_rate": 1.8385196509320424e-05,
      "loss": 0.3302,
      "step": 168
    },
    {
      "epoch": 0.8023738872403561,
      "grad_norm": 0.2890471816062927,
      "learning_rate": 1.8354878114129368e-05,
      "loss": 0.297,
      "step": 169
    },
    {
      "epoch": 0.8071216617210683,
      "grad_norm": 0.3379496932029724,
      "learning_rate": 1.8324303227640472e-05,
      "loss": 0.3032,
      "step": 170
    },
    {
      "epoch": 0.8118694362017804,
      "grad_norm": 0.3012542128562927,
      "learning_rate": 1.8293472788490096e-05,
      "loss": 0.3119,
      "step": 171
    },
    {
      "epoch": 0.8166172106824926,
      "grad_norm": 0.31910914182662964,
      "learning_rate": 1.826238774315995e-05,
      "loss": 0.3249,
      "step": 172
    },
    {
      "epoch": 0.8213649851632048,
      "grad_norm": 0.2994840741157532,
      "learning_rate": 1.8231049045948054e-05,
      "loss": 0.3106,
      "step": 173
    },
    {
      "epoch": 0.826112759643917,
      "grad_norm": 0.31799837946891785,
      "learning_rate": 1.8199457658939425e-05,
      "loss": 0.3109,
      "step": 174
    },
    {
      "epoch": 0.8308605341246291,
      "grad_norm": 0.3020702302455902,
      "learning_rate": 1.816761455197657e-05,
      "loss": 0.3016,
      "step": 175
    },
    {
      "epoch": 0.8356083086053413,
      "grad_norm": 0.33007511496543884,
      "learning_rate": 1.8135520702629677e-05,
      "loss": 0.3326,
      "step": 176
    },
    {
      "epoch": 0.8403560830860534,
      "grad_norm": 0.34143343567848206,
      "learning_rate": 1.8103177096166632e-05,
      "loss": 0.3245,
      "step": 177
    },
    {
      "epoch": 0.8451038575667655,
      "grad_norm": 0.325300008058548,
      "learning_rate": 1.8070584725522763e-05,
      "loss": 0.32,
      "step": 178
    },
    {
      "epoch": 0.8498516320474777,
      "grad_norm": 0.38607487082481384,
      "learning_rate": 1.803774459127034e-05,
      "loss": 0.289,
      "step": 179
    },
    {
      "epoch": 0.8545994065281899,
      "grad_norm": 0.31634339690208435,
      "learning_rate": 1.8004657701587893e-05,
      "loss": 0.3185,
      "step": 180
    },
    {
      "epoch": 0.8593471810089021,
      "grad_norm": 0.32501840591430664,
      "learning_rate": 1.7971325072229227e-05,
      "loss": 0.3291,
      "step": 181
    },
    {
      "epoch": 0.8640949554896142,
      "grad_norm": 0.3348029553890228,
      "learning_rate": 1.7937747726492256e-05,
      "loss": 0.324,
      "step": 182
    },
    {
      "epoch": 0.8688427299703264,
      "grad_norm": 0.34350311756134033,
      "learning_rate": 1.7903926695187595e-05,
      "loss": 0.3351,
      "step": 183
    },
    {
      "epoch": 0.8735905044510386,
      "grad_norm": 0.34770363569259644,
      "learning_rate": 1.7869863016606893e-05,
      "loss": 0.3368,
      "step": 184
    },
    {
      "epoch": 0.8783382789317508,
      "grad_norm": 0.40538617968559265,
      "learning_rate": 1.783555773649097e-05,
      "loss": 0.3027,
      "step": 185
    },
    {
      "epoch": 0.8830860534124629,
      "grad_norm": 0.37002649903297424,
      "learning_rate": 1.7801011907997728e-05,
      "loss": 0.3296,
      "step": 186
    },
    {
      "epoch": 0.8878338278931751,
      "grad_norm": 0.3094744384288788,
      "learning_rate": 1.7766226591669787e-05,
      "loss": 0.3211,
      "step": 187
    },
    {
      "epoch": 0.8925816023738873,
      "grad_norm": 0.3551942706108093,
      "learning_rate": 1.773120285540195e-05,
      "loss": 0.315,
      "step": 188
    },
    {
      "epoch": 0.8973293768545995,
      "grad_norm": 0.3153332471847534,
      "learning_rate": 1.7695941774408424e-05,
      "loss": 0.312,
      "step": 189
    },
    {
      "epoch": 0.9020771513353115,
      "grad_norm": 0.30479609966278076,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.3176,
      "step": 190
    },
    {
      "epoch": 0.9068249258160237,
      "grad_norm": 0.3118721842765808,
      "learning_rate": 1.7624711915499767e-05,
      "loss": 0.3366,
      "step": 191
    },
    {
      "epoch": 0.9115727002967359,
      "grad_norm": 0.3304102122783661,
      "learning_rate": 1.75887453243118e-05,
      "loss": 0.3321,
      "step": 192
    },
    {
      "epoch": 0.916320474777448,
      "grad_norm": 0.29333484172821045,
      "learning_rate": 1.755254576178535e-05,
      "loss": 0.306,
      "step": 193
    },
    {
      "epoch": 0.9210682492581602,
      "grad_norm": 0.34330320358276367,
      "learning_rate": 1.7516114339231984e-05,
      "loss": 0.3173,
      "step": 194
    },
    {
      "epoch": 0.9258160237388724,
      "grad_norm": 0.28864824771881104,
      "learning_rate": 1.747945217508129e-05,
      "loss": 0.2955,
      "step": 195
    },
    {
      "epoch": 0.9305637982195846,
      "grad_norm": 0.32548072934150696,
      "learning_rate": 1.7442560394846518e-05,
      "loss": 0.3284,
      "step": 196
    },
    {
      "epoch": 0.9353115727002967,
      "grad_norm": 0.3424006402492523,
      "learning_rate": 1.740544013109005e-05,
      "loss": 0.315,
      "step": 197
    },
    {
      "epoch": 0.9400593471810089,
      "grad_norm": 0.31073182821273804,
      "learning_rate": 1.73680925233886e-05,
      "loss": 0.3186,
      "step": 198
    },
    {
      "epoch": 0.9448071216617211,
      "grad_norm": 0.37291258573532104,
      "learning_rate": 1.7330518718298263e-05,
      "loss": 0.2982,
      "step": 199
    },
    {
      "epoch": 0.9495548961424333,
      "grad_norm": 0.2813059091567993,
      "learning_rate": 1.7292719869319295e-05,
      "loss": 0.3189,
      "step": 200
    },
    {
      "epoch": 0.9543026706231454,
      "grad_norm": 0.297116219997406,
      "learning_rate": 1.72546971368607e-05,
      "loss": 0.2888,
      "step": 201
    },
    {
      "epoch": 0.9590504451038576,
      "grad_norm": 0.3253256380558014,
      "learning_rate": 1.7216451688204623e-05,
      "loss": 0.327,
      "step": 202
    },
    {
      "epoch": 0.9637982195845697,
      "grad_norm": 0.3983827233314514,
      "learning_rate": 1.717798469747049e-05,
      "loss": 0.309,
      "step": 203
    },
    {
      "epoch": 0.9685459940652819,
      "grad_norm": 0.30336683988571167,
      "learning_rate": 1.7139297345578992e-05,
      "loss": 0.3222,
      "step": 204
    },
    {
      "epoch": 0.973293768545994,
      "grad_norm": 0.3853195309638977,
      "learning_rate": 1.7100390820215805e-05,
      "loss": 0.3182,
      "step": 205
    },
    {
      "epoch": 0.9780415430267062,
      "grad_norm": 0.3312664031982422,
      "learning_rate": 1.7061266315795146e-05,
      "loss": 0.2972,
      "step": 206
    },
    {
      "epoch": 0.9827893175074184,
      "grad_norm": 0.3694183826446533,
      "learning_rate": 1.7021925033423096e-05,
      "loss": 0.3177,
      "step": 207
    },
    {
      "epoch": 0.9875370919881306,
      "grad_norm": 0.3218733072280884,
      "learning_rate": 1.698236818086073e-05,
      "loss": 0.304,
      "step": 208
    },
    {
      "epoch": 0.9922848664688427,
      "grad_norm": 0.33860743045806885,
      "learning_rate": 1.694259697248704e-05,
      "loss": 0.3031,
      "step": 209
    },
    {
      "epoch": 0.9970326409495549,
      "grad_norm": 0.42556554079055786,
      "learning_rate": 1.690261262926165e-05,
      "loss": 0.3196,
      "step": 210
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38205158710479736,
      "learning_rate": 1.686241637868734e-05,
      "loss": 0.2881,
      "step": 211
    },
    {
      "epoch": 1.004747774480712,
      "grad_norm": 0.36370348930358887,
      "learning_rate": 1.682200945477235e-05,
      "loss": 0.3082,
      "step": 212
    },
    {
      "epoch": 1.0094955489614243,
      "grad_norm": 0.3192894160747528,
      "learning_rate": 1.6781393097992512e-05,
      "loss": 0.3233,
      "step": 213
    },
    {
      "epoch": 1.0142433234421364,
      "grad_norm": 0.34595000743865967,
      "learning_rate": 1.6740568555253153e-05,
      "loss": 0.2932,
      "step": 214
    },
    {
      "epoch": 1.0189910979228487,
      "grad_norm": 0.32576438784599304,
      "learning_rate": 1.669953707985084e-05,
      "loss": 0.2944,
      "step": 215
    },
    {
      "epoch": 1.0237388724035608,
      "grad_norm": 0.3424221873283386,
      "learning_rate": 1.6658299931434857e-05,
      "loss": 0.305,
      "step": 216
    },
    {
      "epoch": 1.028486646884273,
      "grad_norm": 0.32766616344451904,
      "learning_rate": 1.6616858375968596e-05,
      "loss": 0.3266,
      "step": 217
    },
    {
      "epoch": 1.0332344213649851,
      "grad_norm": 0.34646332263946533,
      "learning_rate": 1.657521368569064e-05,
      "loss": 0.316,
      "step": 218
    },
    {
      "epoch": 1.0379821958456974,
      "grad_norm": 0.30172574520111084,
      "learning_rate": 1.6533367139075732e-05,
      "loss": 0.2973,
      "step": 219
    },
    {
      "epoch": 1.0427299703264095,
      "grad_norm": 0.33457738161087036,
      "learning_rate": 1.649132002079552e-05,
      "loss": 0.3015,
      "step": 220
    },
    {
      "epoch": 1.0474777448071217,
      "grad_norm": 0.32709312438964844,
      "learning_rate": 1.6449073621679128e-05,
      "loss": 0.2967,
      "step": 221
    },
    {
      "epoch": 1.0522255192878338,
      "grad_norm": 0.32702985405921936,
      "learning_rate": 1.6406629238673507e-05,
      "loss": 0.3375,
      "step": 222
    },
    {
      "epoch": 1.056973293768546,
      "grad_norm": 0.3030819296836853,
      "learning_rate": 1.6363988174803638e-05,
      "loss": 0.3171,
      "step": 223
    },
    {
      "epoch": 1.0617210682492582,
      "grad_norm": 0.3053749203681946,
      "learning_rate": 1.632115173913252e-05,
      "loss": 0.3188,
      "step": 224
    },
    {
      "epoch": 1.0664688427299702,
      "grad_norm": 0.33132317662239075,
      "learning_rate": 1.627812124672099e-05,
      "loss": 0.325,
      "step": 225
    },
    {
      "epoch": 1.0712166172106825,
      "grad_norm": 0.32221683859825134,
      "learning_rate": 1.6234898018587336e-05,
      "loss": 0.3249,
      "step": 226
    },
    {
      "epoch": 1.0759643916913946,
      "grad_norm": 0.32629886269569397,
      "learning_rate": 1.619148338166677e-05,
      "loss": 0.308,
      "step": 227
    },
    {
      "epoch": 1.0807121661721069,
      "grad_norm": 0.31524035334587097,
      "learning_rate": 1.614787866877066e-05,
      "loss": 0.304,
      "step": 228
    },
    {
      "epoch": 1.085459940652819,
      "grad_norm": 0.33473050594329834,
      "learning_rate": 1.6104085218545633e-05,
      "loss": 0.3222,
      "step": 229
    },
    {
      "epoch": 1.0902077151335312,
      "grad_norm": 0.28698474168777466,
      "learning_rate": 1.6060104375432476e-05,
      "loss": 0.3057,
      "step": 230
    },
    {
      "epoch": 1.0949554896142433,
      "grad_norm": 0.2982139587402344,
      "learning_rate": 1.601593748962485e-05,
      "loss": 0.3074,
      "step": 231
    },
    {
      "epoch": 1.0997032640949556,
      "grad_norm": 0.29852068424224854,
      "learning_rate": 1.5971585917027864e-05,
      "loss": 0.3152,
      "step": 232
    },
    {
      "epoch": 1.1044510385756676,
      "grad_norm": 0.28795862197875977,
      "learning_rate": 1.5927051019216428e-05,
      "loss": 0.3084,
      "step": 233
    },
    {
      "epoch": 1.10919881305638,
      "grad_norm": 0.27636635303497314,
      "learning_rate": 1.588233416339345e-05,
      "loss": 0.3079,
      "step": 234
    },
    {
      "epoch": 1.113946587537092,
      "grad_norm": 0.3473345637321472,
      "learning_rate": 1.5837436722347902e-05,
      "loss": 0.2836,
      "step": 235
    },
    {
      "epoch": 1.1186943620178043,
      "grad_norm": 0.3350068926811218,
      "learning_rate": 1.5792360074412612e-05,
      "loss": 0.3066,
      "step": 236
    },
    {
      "epoch": 1.1234421364985163,
      "grad_norm": 0.33414971828460693,
      "learning_rate": 1.5747105603422013e-05,
      "loss": 0.3082,
      "step": 237
    },
    {
      "epoch": 1.1281899109792284,
      "grad_norm": 0.32059019804000854,
      "learning_rate": 1.570167469866962e-05,
      "loss": 0.3103,
      "step": 238
    },
    {
      "epoch": 1.1329376854599407,
      "grad_norm": 0.32460328936576843,
      "learning_rate": 1.5656068754865388e-05,
      "loss": 0.3151,
      "step": 239
    },
    {
      "epoch": 1.1376854599406527,
      "grad_norm": 0.3252885341644287,
      "learning_rate": 1.561028917209291e-05,
      "loss": 0.3248,
      "step": 240
    },
    {
      "epoch": 1.142433234421365,
      "grad_norm": 0.3324680030345917,
      "learning_rate": 1.5564337355766412e-05,
      "loss": 0.3119,
      "step": 241
    },
    {
      "epoch": 1.147181008902077,
      "grad_norm": 0.3315082788467407,
      "learning_rate": 1.551821471658763e-05,
      "loss": 0.3094,
      "step": 242
    },
    {
      "epoch": 1.1519287833827894,
      "grad_norm": 0.3459841310977936,
      "learning_rate": 1.5471922670502472e-05,
      "loss": 0.2982,
      "step": 243
    },
    {
      "epoch": 1.1566765578635014,
      "grad_norm": 0.30599355697631836,
      "learning_rate": 1.5425462638657597e-05,
      "loss": 0.2802,
      "step": 244
    },
    {
      "epoch": 1.1614243323442137,
      "grad_norm": 0.3451800048351288,
      "learning_rate": 1.5378836047356725e-05,
      "loss": 0.3065,
      "step": 245
    },
    {
      "epoch": 1.1661721068249258,
      "grad_norm": 0.28921040892601013,
      "learning_rate": 1.5332044328016916e-05,
      "loss": 0.2935,
      "step": 246
    },
    {
      "epoch": 1.170919881305638,
      "grad_norm": 0.3126339912414551,
      "learning_rate": 1.5285088917124555e-05,
      "loss": 0.2794,
      "step": 247
    },
    {
      "epoch": 1.1756676557863501,
      "grad_norm": 0.33224689960479736,
      "learning_rate": 1.5237971256191325e-05,
      "loss": 0.3068,
      "step": 248
    },
    {
      "epoch": 1.1804154302670624,
      "grad_norm": 0.33340543508529663,
      "learning_rate": 1.5190692791709891e-05,
      "loss": 0.2895,
      "step": 249
    },
    {
      "epoch": 1.1851632047477745,
      "grad_norm": 0.3247314989566803,
      "learning_rate": 1.5143254975109538e-05,
      "loss": 0.3005,
      "step": 250
    },
    {
      "epoch": 1.1899109792284865,
      "grad_norm": 0.3434915542602539,
      "learning_rate": 1.5095659262711588e-05,
      "loss": 0.2991,
      "step": 251
    },
    {
      "epoch": 1.1946587537091988,
      "grad_norm": 0.28737980127334595,
      "learning_rate": 1.5047907115684695e-05,
      "loss": 0.2603,
      "step": 252
    },
    {
      "epoch": 1.1946587537091988,
      "eval_loss": 0.15092316269874573,
      "eval_runtime": 136.3285,
      "eval_samples_per_second": 13.174,
      "eval_steps_per_second": 0.829,
      "step": 252
    }
  ],
  "logging_steps": 1,
  "max_steps": 630,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 126,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.752167478826369e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}