{
  "best_metric": 0.06687498837709427,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.922722029988466,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00461361014994233,
      "grad_norm": 15.70113754272461,
      "learning_rate": 1e-05,
      "loss": 4.4125,
      "step": 1
    },
    {
      "epoch": 0.00461361014994233,
      "eval_loss": 1.3732987642288208,
      "eval_runtime": 17.1142,
      "eval_samples_per_second": 21.327,
      "eval_steps_per_second": 5.376,
      "step": 1
    },
    {
      "epoch": 0.00922722029988466,
      "grad_norm": 20.68549346923828,
      "learning_rate": 2e-05,
      "loss": 4.6565,
      "step": 2
    },
    {
      "epoch": 0.01384083044982699,
      "grad_norm": 20.048234939575195,
      "learning_rate": 3e-05,
      "loss": 4.6348,
      "step": 3
    },
    {
      "epoch": 0.01845444059976932,
      "grad_norm": 14.199302673339844,
      "learning_rate": 4e-05,
      "loss": 4.4976,
      "step": 4
    },
    {
      "epoch": 0.02306805074971165,
      "grad_norm": 7.783128261566162,
      "learning_rate": 5e-05,
      "loss": 3.9834,
      "step": 5
    },
    {
      "epoch": 0.02768166089965398,
      "grad_norm": 6.201513767242432,
      "learning_rate": 6e-05,
      "loss": 3.7714,
      "step": 6
    },
    {
      "epoch": 0.03229527104959631,
      "grad_norm": 6.3549652099609375,
      "learning_rate": 7e-05,
      "loss": 3.5251,
      "step": 7
    },
    {
      "epoch": 0.03690888119953864,
      "grad_norm": 4.620194911956787,
      "learning_rate": 8e-05,
      "loss": 3.0183,
      "step": 8
    },
    {
      "epoch": 0.04152249134948097,
      "grad_norm": 5.089642524719238,
      "learning_rate": 9e-05,
      "loss": 2.8422,
      "step": 9
    },
    {
      "epoch": 0.0461361014994233,
      "grad_norm": 5.489530563354492,
      "learning_rate": 0.0001,
      "loss": 2.5396,
      "step": 10
    },
    {
      "epoch": 0.05074971164936563,
      "grad_norm": 4.536956787109375,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.4472,
      "step": 11
    },
    {
      "epoch": 0.05536332179930796,
      "grad_norm": 3.962470054626465,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.1924,
      "step": 12
    },
    {
      "epoch": 0.05997693194925029,
      "grad_norm": 3.0783560276031494,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.9028,
      "step": 13
    },
    {
      "epoch": 0.06459054209919261,
      "grad_norm": 3.398895263671875,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.995,
      "step": 14
    },
    {
      "epoch": 0.06920415224913495,
      "grad_norm": 2.8160057067871094,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.6265,
      "step": 15
    },
    {
      "epoch": 0.07381776239907728,
      "grad_norm": 2.524176597595215,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5802,
      "step": 16
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 2.023597002029419,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.3672,
      "step": 17
    },
    {
      "epoch": 0.08304498269896193,
      "grad_norm": 2.4030978679656982,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.4684,
      "step": 18
    },
    {
      "epoch": 0.08765859284890427,
      "grad_norm": 2.688441276550293,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.296,
      "step": 19
    },
    {
      "epoch": 0.0922722029988466,
      "grad_norm": 2.184069871902466,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3114,
      "step": 20
    },
    {
      "epoch": 0.09688581314878893,
      "grad_norm": 2.4564127922058105,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1625,
      "step": 21
    },
    {
      "epoch": 0.10149942329873125,
      "grad_norm": 2.848806619644165,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.3759,
      "step": 22
    },
    {
      "epoch": 0.1061130334486736,
      "grad_norm": 2.700578212738037,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1959,
      "step": 23
    },
    {
      "epoch": 0.11072664359861592,
      "grad_norm": 2.2797701358795166,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.0781,
      "step": 24
    },
    {
      "epoch": 0.11534025374855825,
      "grad_norm": 2.8311095237731934,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.1178,
      "step": 25
    },
    {
      "epoch": 0.11995386389850057,
      "grad_norm": 2.355239152908325,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.9605,
      "step": 26
    },
    {
      "epoch": 0.1245674740484429,
      "grad_norm": 2.595677137374878,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.0009,
      "step": 27
    },
    {
      "epoch": 0.12918108419838523,
      "grad_norm": 2.7560641765594482,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.9774,
      "step": 28
    },
    {
      "epoch": 0.13379469434832755,
      "grad_norm": 2.496370792388916,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0094,
      "step": 29
    },
    {
      "epoch": 0.1384083044982699,
      "grad_norm": 2.242828845977783,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9715,
      "step": 30
    },
    {
      "epoch": 0.14302191464821223,
      "grad_norm": 2.752673864364624,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.9229,
      "step": 31
    },
    {
      "epoch": 0.14763552479815456,
      "grad_norm": 2.201143503189087,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.8835,
      "step": 32
    },
    {
      "epoch": 0.1522491349480969,
      "grad_norm": 1.9557386636734009,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.7726,
      "step": 33
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 2.219505786895752,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.8736,
      "step": 34
    },
    {
      "epoch": 0.16147635524798154,
      "grad_norm": 2.194694757461548,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.7686,
      "step": 35
    },
    {
      "epoch": 0.16608996539792387,
      "grad_norm": 2.1537692546844482,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.7776,
      "step": 36
    },
    {
      "epoch": 0.1707035755478662,
      "grad_norm": 2.1905946731567383,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.8058,
      "step": 37
    },
    {
      "epoch": 0.17531718569780855,
      "grad_norm": 2.362812042236328,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.8182,
      "step": 38
    },
    {
      "epoch": 0.17993079584775087,
      "grad_norm": 2.339319944381714,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.7534,
      "step": 39
    },
    {
      "epoch": 0.1845444059976932,
      "grad_norm": 1.571523904800415,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.6239,
      "step": 40
    },
    {
      "epoch": 0.18915801614763553,
      "grad_norm": 1.7810354232788086,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.6016,
      "step": 41
    },
    {
      "epoch": 0.19377162629757785,
      "grad_norm": 2.2040226459503174,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.6992,
      "step": 42
    },
    {
      "epoch": 0.19838523644752018,
      "grad_norm": 1.8784388303756714,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.5785,
      "step": 43
    },
    {
      "epoch": 0.2029988465974625,
      "grad_norm": 2.0472021102905273,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.5436,
      "step": 44
    },
    {
      "epoch": 0.20761245674740483,
      "grad_norm": 1.978790283203125,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.5644,
      "step": 45
    },
    {
      "epoch": 0.2122260668973472,
      "grad_norm": 2.206702947616577,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.5628,
      "step": 46
    },
    {
      "epoch": 0.21683967704728951,
      "grad_norm": 2.2076473236083984,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.6329,
      "step": 47
    },
    {
      "epoch": 0.22145328719723184,
      "grad_norm": 2.2554423809051514,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.5721,
      "step": 48
    },
    {
      "epoch": 0.22606689734717417,
      "grad_norm": 2.839228868484497,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6017,
      "step": 49
    },
    {
      "epoch": 0.2306805074971165,
      "grad_norm": 3.1892142295837402,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.8681,
      "step": 50
    },
    {
      "epoch": 0.2306805074971165,
      "eval_loss": 0.1744069755077362,
      "eval_runtime": 17.4323,
      "eval_samples_per_second": 20.938,
      "eval_steps_per_second": 5.278,
      "step": 50
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 1.7297769784927368,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.5304,
      "step": 51
    },
    {
      "epoch": 0.23990772779700115,
      "grad_norm": 1.6736814975738525,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.5341,
      "step": 52
    },
    {
      "epoch": 0.24452133794694347,
      "grad_norm": 2.0269315242767334,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6766,
      "step": 53
    },
    {
      "epoch": 0.2491349480968858,
      "grad_norm": 2.130960464477539,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.6106,
      "step": 54
    },
    {
      "epoch": 0.2537485582468281,
      "grad_norm": 1.8690334558486938,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.7397,
      "step": 55
    },
    {
      "epoch": 0.25836216839677045,
      "grad_norm": 1.8371063470840454,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.5668,
      "step": 56
    },
    {
      "epoch": 0.2629757785467128,
      "grad_norm": 1.5481998920440674,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.6382,
      "step": 57
    },
    {
      "epoch": 0.2675893886966551,
      "grad_norm": 1.6908756494522095,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.6662,
      "step": 58
    },
    {
      "epoch": 0.2722029988465975,
      "grad_norm": 1.72432279586792,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.6328,
      "step": 59
    },
    {
      "epoch": 0.2768166089965398,
      "grad_norm": 1.4916110038757324,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.6373,
      "step": 60
    },
    {
      "epoch": 0.28143021914648214,
      "grad_norm": 1.9055370092391968,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.4967,
      "step": 61
    },
    {
      "epoch": 0.28604382929642447,
      "grad_norm": 1.4707900285720825,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.4727,
      "step": 62
    },
    {
      "epoch": 0.2906574394463668,
      "grad_norm": 1.6812090873718262,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.538,
      "step": 63
    },
    {
      "epoch": 0.2952710495963091,
      "grad_norm": 1.600934624671936,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.5162,
      "step": 64
    },
    {
      "epoch": 0.29988465974625145,
      "grad_norm": 1.7742993831634521,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.5814,
      "step": 65
    },
    {
      "epoch": 0.3044982698961938,
      "grad_norm": 1.5534722805023193,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.5618,
      "step": 66
    },
    {
      "epoch": 0.3091118800461361,
      "grad_norm": 1.628890872001648,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.5436,
      "step": 67
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 1.6108001470565796,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.5618,
      "step": 68
    },
    {
      "epoch": 0.31833910034602075,
      "grad_norm": 1.3222441673278809,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.5345,
      "step": 69
    },
    {
      "epoch": 0.3229527104959631,
      "grad_norm": 1.8731218576431274,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.5441,
      "step": 70
    },
    {
      "epoch": 0.3275663206459054,
      "grad_norm": 1.59487783908844,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.4835,
      "step": 71
    },
    {
      "epoch": 0.33217993079584773,
      "grad_norm": 1.8421398401260376,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5247,
      "step": 72
    },
    {
      "epoch": 0.33679354094579006,
      "grad_norm": 1.5689805746078491,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.493,
      "step": 73
    },
    {
      "epoch": 0.3414071510957324,
      "grad_norm": 1.6961421966552734,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.6119,
      "step": 74
    },
    {
      "epoch": 0.3460207612456747,
      "grad_norm": 1.6617010831832886,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.4983,
      "step": 75
    },
    {
      "epoch": 0.3506343713956171,
      "grad_norm": 1.5688445568084717,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.5141,
      "step": 76
    },
    {
      "epoch": 0.3552479815455594,
      "grad_norm": 1.977229118347168,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.4587,
      "step": 77
    },
    {
      "epoch": 0.35986159169550175,
      "grad_norm": 1.6480942964553833,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4615,
      "step": 78
    },
    {
      "epoch": 0.3644752018454441,
      "grad_norm": 1.800323486328125,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.5088,
      "step": 79
    },
    {
      "epoch": 0.3690888119953864,
      "grad_norm": 1.5188491344451904,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.4654,
      "step": 80
    },
    {
      "epoch": 0.3737024221453287,
      "grad_norm": 2.1865861415863037,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.5313,
      "step": 81
    },
    {
      "epoch": 0.37831603229527105,
      "grad_norm": 1.6185697317123413,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.5178,
      "step": 82
    },
    {
      "epoch": 0.3829296424452134,
      "grad_norm": 1.5496866703033447,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.4488,
      "step": 83
    },
    {
      "epoch": 0.3875432525951557,
      "grad_norm": 1.704131007194519,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.4499,
      "step": 84
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 1.5845049619674683,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.4855,
      "step": 85
    },
    {
      "epoch": 0.39677047289504036,
      "grad_norm": 1.68927800655365,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.4365,
      "step": 86
    },
    {
      "epoch": 0.4013840830449827,
      "grad_norm": 1.558206558227539,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.4406,
      "step": 87
    },
    {
      "epoch": 0.405997693194925,
      "grad_norm": 1.6405448913574219,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.4586,
      "step": 88
    },
    {
      "epoch": 0.41061130334486734,
      "grad_norm": 1.4575345516204834,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.4403,
      "step": 89
    },
    {
      "epoch": 0.41522491349480967,
      "grad_norm": 1.6011104583740234,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4758,
      "step": 90
    },
    {
      "epoch": 0.419838523644752,
      "grad_norm": 1.3472517728805542,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.2868,
      "step": 91
    },
    {
      "epoch": 0.4244521337946944,
      "grad_norm": 1.6547609567642212,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.4138,
      "step": 92
    },
    {
      "epoch": 0.4290657439446367,
      "grad_norm": 1.7545020580291748,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.4362,
      "step": 93
    },
    {
      "epoch": 0.43367935409457903,
      "grad_norm": 1.4638218879699707,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3809,
      "step": 94
    },
    {
      "epoch": 0.43829296424452135,
      "grad_norm": 1.8458184003829956,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.4869,
      "step": 95
    },
    {
      "epoch": 0.4429065743944637,
      "grad_norm": 1.4978622198104858,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3235,
      "step": 96
    },
    {
      "epoch": 0.447520184544406,
      "grad_norm": 1.1788842678070068,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.2182,
      "step": 97
    },
    {
      "epoch": 0.45213379469434833,
      "grad_norm": 2.009841203689575,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3887,
      "step": 98
    },
    {
      "epoch": 0.45674740484429066,
      "grad_norm": 2.058389902114868,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.4368,
      "step": 99
    },
    {
      "epoch": 0.461361014994233,
      "grad_norm": 2.85821533203125,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.4363,
      "step": 100
    },
    {
      "epoch": 0.461361014994233,
      "eval_loss": 0.0986216738820076,
      "eval_runtime": 17.4315,
      "eval_samples_per_second": 20.939,
      "eval_steps_per_second": 5.278,
      "step": 100
    },
    {
      "epoch": 0.4659746251441753,
      "grad_norm": 1.2732893228530884,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3516,
      "step": 101
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 1.0777982473373413,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.2961,
      "step": 102
    },
    {
      "epoch": 0.47520184544405997,
      "grad_norm": 1.1578322649002075,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3652,
      "step": 103
    },
    {
      "epoch": 0.4798154555940023,
      "grad_norm": 1.4592502117156982,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.4224,
      "step": 104
    },
    {
      "epoch": 0.4844290657439446,
      "grad_norm": 1.6004899740219116,
      "learning_rate": 5e-05,
      "loss": 0.3997,
      "step": 105
    },
    {
      "epoch": 0.48904267589388695,
      "grad_norm": 1.5356582403182983,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3947,
      "step": 106
    },
    {
      "epoch": 0.4936562860438293,
      "grad_norm": 1.501625418663025,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.4262,
      "step": 107
    },
    {
      "epoch": 0.4982698961937716,
      "grad_norm": 1.3154499530792236,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.3248,
      "step": 108
    },
    {
      "epoch": 0.5028835063437139,
      "grad_norm": 1.2077476978302002,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.2978,
      "step": 109
    },
    {
      "epoch": 0.5074971164936563,
      "grad_norm": 1.2540231943130493,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3185,
      "step": 110
    },
    {
      "epoch": 0.5121107266435986,
      "grad_norm": 1.1510992050170898,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.3075,
      "step": 111
    },
    {
      "epoch": 0.5167243367935409,
      "grad_norm": 1.567811131477356,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.4239,
      "step": 112
    },
    {
      "epoch": 0.5213379469434832,
      "grad_norm": 1.3366222381591797,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.336,
      "step": 113
    },
    {
      "epoch": 0.5259515570934256,
      "grad_norm": 1.5086486339569092,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3836,
      "step": 114
    },
    {
      "epoch": 0.5305651672433679,
      "grad_norm": 1.1578775644302368,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2972,
      "step": 115
    },
    {
      "epoch": 0.5351787773933102,
      "grad_norm": 1.1253985166549683,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3638,
      "step": 116
    },
    {
      "epoch": 0.5397923875432526,
      "grad_norm": 1.3894917964935303,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3488,
      "step": 117
    },
    {
      "epoch": 0.544405997693195,
      "grad_norm": 1.7476497888565063,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3868,
      "step": 118
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 1.524665355682373,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3953,
      "step": 119
    },
    {
      "epoch": 0.5536332179930796,
      "grad_norm": 1.092214584350586,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2821,
      "step": 120
    },
    {
      "epoch": 0.558246828143022,
      "grad_norm": 1.6726285219192505,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.3561,
      "step": 121
    },
    {
      "epoch": 0.5628604382929643,
      "grad_norm": 1.4599559307098389,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3482,
      "step": 122
    },
    {
      "epoch": 0.5674740484429066,
      "grad_norm": 1.2500851154327393,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3491,
      "step": 123
    },
    {
      "epoch": 0.5720876585928489,
      "grad_norm": 1.643224835395813,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3094,
      "step": 124
    },
    {
      "epoch": 0.5767012687427913,
      "grad_norm": 1.27620530128479,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3124,
      "step": 125
    },
    {
      "epoch": 0.5813148788927336,
      "grad_norm": 1.1907225847244263,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.2576,
      "step": 126
    },
    {
      "epoch": 0.5859284890426759,
      "grad_norm": 1.194791316986084,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3291,
      "step": 127
    },
    {
      "epoch": 0.5905420991926182,
      "grad_norm": 1.4398959875106812,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3788,
      "step": 128
    },
    {
      "epoch": 0.5951557093425606,
      "grad_norm": 1.4079384803771973,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.4317,
      "step": 129
    },
    {
      "epoch": 0.5997693194925029,
      "grad_norm": 1.3933075666427612,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3008,
      "step": 130
    },
    {
      "epoch": 0.6043829296424452,
      "grad_norm": 1.7076566219329834,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.4375,
      "step": 131
    },
    {
      "epoch": 0.6089965397923875,
      "grad_norm": 1.4131172895431519,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.409,
      "step": 132
    },
    {
      "epoch": 0.6136101499423299,
      "grad_norm": 1.5254442691802979,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.413,
      "step": 133
    },
    {
      "epoch": 0.6182237600922722,
      "grad_norm": 1.3601816892623901,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.374,
      "step": 134
    },
    {
      "epoch": 0.6228373702422145,
      "grad_norm": 1.8306043148040771,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3003,
      "step": 135
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 1.5551977157592773,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.41,
      "step": 136
    },
    {
      "epoch": 0.6320645905420992,
      "grad_norm": 1.0856026411056519,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.2385,
      "step": 137
    },
    {
      "epoch": 0.6366782006920415,
      "grad_norm": 0.9590676426887512,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.2239,
      "step": 138
    },
    {
      "epoch": 0.6412918108419838,
      "grad_norm": 1.2713561058044434,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3078,
      "step": 139
    },
    {
      "epoch": 0.6459054209919262,
      "grad_norm": 1.2558138370513916,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.265,
      "step": 140
    },
    {
      "epoch": 0.6505190311418685,
      "grad_norm": 1.2806661128997803,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.2942,
      "step": 141
    },
    {
      "epoch": 0.6551326412918108,
      "grad_norm": 1.2513595819473267,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2516,
      "step": 142
    },
    {
      "epoch": 0.6597462514417531,
      "grad_norm": 1.692915916442871,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3054,
      "step": 143
    },
    {
      "epoch": 0.6643598615916955,
      "grad_norm": 1.1031126976013184,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.1817,
      "step": 144
    },
    {
      "epoch": 0.6689734717416378,
      "grad_norm": 1.4994665384292603,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3032,
      "step": 145
    },
    {
      "epoch": 0.6735870818915801,
      "grad_norm": 1.6487501859664917,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.3576,
      "step": 146
    },
    {
      "epoch": 0.6782006920415224,
      "grad_norm": 1.571150779724121,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2995,
      "step": 147
    },
    {
      "epoch": 0.6828143021914648,
      "grad_norm": 1.8149317502975464,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2893,
      "step": 148
    },
    {
      "epoch": 0.6874279123414071,
      "grad_norm": 2.061163902282715,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3869,
      "step": 149
    },
    {
      "epoch": 0.6920415224913494,
      "grad_norm": 2.8944008350372314,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.4262,
      "step": 150
    },
    {
      "epoch": 0.6920415224913494,
      "eval_loss": 0.07600851356983185,
      "eval_runtime": 17.4466,
      "eval_samples_per_second": 20.921,
      "eval_steps_per_second": 5.273,
      "step": 150
    },
    {
      "epoch": 0.6966551326412919,
      "grad_norm": 1.4078782796859741,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2749,
      "step": 151
    },
    {
      "epoch": 0.7012687427912342,
      "grad_norm": 1.229753017425537,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.258,
      "step": 152
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 1.5120552778244019,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3187,
      "step": 153
    },
    {
      "epoch": 0.7104959630911188,
      "grad_norm": 1.2669744491577148,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.3104,
      "step": 154
    },
    {
      "epoch": 0.7151095732410612,
      "grad_norm": 1.3387261629104614,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.3011,
      "step": 155
    },
    {
      "epoch": 0.7197231833910035,
      "grad_norm": 1.2108829021453857,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.2581,
      "step": 156
    },
    {
      "epoch": 0.7243367935409458,
      "grad_norm": 1.1918522119522095,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3597,
      "step": 157
    },
    {
      "epoch": 0.7289504036908881,
      "grad_norm": 1.235791802406311,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2529,
      "step": 158
    },
    {
      "epoch": 0.7335640138408305,
      "grad_norm": 1.2199387550354004,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.2697,
      "step": 159
    },
    {
      "epoch": 0.7381776239907728,
      "grad_norm": 1.063336968421936,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.2394,
      "step": 160
    },
    {
      "epoch": 0.7427912341407151,
      "grad_norm": 1.2216490507125854,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.2775,
      "step": 161
    },
    {
      "epoch": 0.7474048442906575,
      "grad_norm": 1.233976125717163,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2557,
      "step": 162
    },
    {
      "epoch": 0.7520184544405998,
      "grad_norm": 0.9891316890716553,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.2407,
      "step": 163
    },
    {
      "epoch": 0.7566320645905421,
      "grad_norm": 0.9953312277793884,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2158,
      "step": 164
    },
    {
      "epoch": 0.7612456747404844,
      "grad_norm": 1.2919557094573975,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2765,
      "step": 165
    },
    {
      "epoch": 0.7658592848904268,
      "grad_norm": 1.3856021165847778,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.2925,
      "step": 166
    },
    {
      "epoch": 0.7704728950403691,
      "grad_norm": 1.2595807313919067,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3496,
      "step": 167
    },
    {
      "epoch": 0.7750865051903114,
      "grad_norm": 1.1151498556137085,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2587,
      "step": 168
    },
    {
      "epoch": 0.7797001153402537,
      "grad_norm": 1.088828206062317,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2545,
      "step": 169
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 1.1316912174224854,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.2654,
      "step": 170
    },
    {
      "epoch": 0.7889273356401384,
      "grad_norm": 1.545020580291748,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3321,
      "step": 171
    },
    {
      "epoch": 0.7935409457900807,
      "grad_norm": 1.372032642364502,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.308,
      "step": 172
    },
    {
      "epoch": 0.798154555940023,
      "grad_norm": 1.253602147102356,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.3051,
      "step": 173
    },
    {
      "epoch": 0.8027681660899654,
      "grad_norm": 1.0796576738357544,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2455,
      "step": 174
    },
    {
      "epoch": 0.8073817762399077,
      "grad_norm": 1.2191009521484375,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3088,
      "step": 175
    },
    {
      "epoch": 0.81199538638985,
      "grad_norm": 1.156672477722168,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.2994,
      "step": 176
    },
    {
      "epoch": 0.8166089965397924,
      "grad_norm": 1.047947645187378,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.2264,
      "step": 177
    },
    {
      "epoch": 0.8212226066897347,
      "grad_norm": 1.2244176864624023,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.2601,
      "step": 178
    },
    {
      "epoch": 0.825836216839677,
      "grad_norm": 1.3187599182128906,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3073,
      "step": 179
    },
    {
      "epoch": 0.8304498269896193,
      "grad_norm": 1.4453531503677368,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.3635,
      "step": 180
    },
    {
      "epoch": 0.8350634371395617,
      "grad_norm": 1.3755645751953125,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.3503,
      "step": 181
    },
    {
      "epoch": 0.839677047289504,
      "grad_norm": 1.1200528144836426,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.2619,
      "step": 182
    },
    {
      "epoch": 0.8442906574394463,
      "grad_norm": 1.4638115167617798,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2792,
      "step": 183
    },
    {
      "epoch": 0.8489042675893888,
      "grad_norm": 1.595523476600647,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3399,
      "step": 184
    },
    {
      "epoch": 0.8535178777393311,
      "grad_norm": 1.3847228288650513,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.2623,
      "step": 185
    },
    {
      "epoch": 0.8581314878892734,
      "grad_norm": 1.5647356510162354,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.2925,
      "step": 186
    },
    {
      "epoch": 0.8627450980392157,
      "grad_norm": 1.3908151388168335,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2971,
      "step": 187
    },
    {
      "epoch": 0.8673587081891581,
      "grad_norm": 1.0677061080932617,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2059,
      "step": 188
    },
    {
      "epoch": 0.8719723183391004,
      "grad_norm": 1.219547152519226,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.2267,
      "step": 189
    },
    {
      "epoch": 0.8765859284890427,
      "grad_norm": 1.4710617065429688,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.2931,
      "step": 190
    },
    {
      "epoch": 0.881199538638985,
      "grad_norm": 1.2974404096603394,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2169,
      "step": 191
    },
    {
      "epoch": 0.8858131487889274,
      "grad_norm": 1.2070374488830566,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.2571,
      "step": 192
    },
    {
      "epoch": 0.8904267589388697,
      "grad_norm": 1.254785180091858,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2548,
      "step": 193
    },
    {
      "epoch": 0.895040369088812,
      "grad_norm": 1.3049482107162476,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.2574,
      "step": 194
    },
    {
      "epoch": 0.8996539792387543,
      "grad_norm": 1.2218841314315796,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.215,
      "step": 195
    },
    {
      "epoch": 0.9042675893886967,
      "grad_norm": 1.7114369869232178,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.3408,
      "step": 196
    },
    {
      "epoch": 0.908881199538639,
      "grad_norm": 1.6129153966903687,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.2528,
      "step": 197
    },
    {
      "epoch": 0.9134948096885813,
      "grad_norm": 1.6780078411102295,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.2238,
      "step": 198
    },
    {
      "epoch": 0.9181084198385236,
      "grad_norm": 1.9415208101272583,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2899,
      "step": 199
    },
    {
      "epoch": 0.922722029988466,
      "grad_norm": 1.605064034461975,
      "learning_rate": 0.0,
      "loss": 0.2231,
      "step": 200
    },
    {
      "epoch": 0.922722029988466,
      "eval_loss": 0.06687498837709427,
      "eval_runtime": 17.4347,
      "eval_samples_per_second": 20.935,
      "eval_steps_per_second": 5.277,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.50336002064384e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}