{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 242,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004132231404958678,
      "grad_norm": 0.16453267633914948,
      "learning_rate": 9.958677685950414e-06,
      "loss": 2.012,
      "step": 1
    },
    {
      "epoch": 0.008264462809917356,
      "grad_norm": 0.15626223385334015,
      "learning_rate": 9.917355371900828e-06,
      "loss": 2.0374,
      "step": 2
    },
    {
      "epoch": 0.012396694214876033,
      "grad_norm": 0.15379267930984497,
      "learning_rate": 9.876033057851241e-06,
      "loss": 1.9793,
      "step": 3
    },
    {
      "epoch": 0.01652892561983471,
      "grad_norm": 0.15836523473262787,
      "learning_rate": 9.834710743801654e-06,
      "loss": 1.9191,
      "step": 4
    },
    {
      "epoch": 0.02066115702479339,
      "grad_norm": 0.17007023096084595,
      "learning_rate": 9.793388429752067e-06,
      "loss": 2.081,
      "step": 5
    },
    {
      "epoch": 0.024793388429752067,
      "grad_norm": 0.16546334326267242,
      "learning_rate": 9.75206611570248e-06,
      "loss": 1.9598,
      "step": 6
    },
    {
      "epoch": 0.028925619834710745,
      "grad_norm": 0.17922371625900269,
      "learning_rate": 9.710743801652894e-06,
      "loss": 1.9773,
      "step": 7
    },
    {
      "epoch": 0.03305785123966942,
      "grad_norm": 0.1753963977098465,
      "learning_rate": 9.669421487603307e-06,
      "loss": 2.0211,
      "step": 8
    },
    {
      "epoch": 0.0371900826446281,
      "grad_norm": 0.17829500138759613,
      "learning_rate": 9.62809917355372e-06,
      "loss": 1.9165,
      "step": 9
    },
    {
      "epoch": 0.04132231404958678,
      "grad_norm": 0.1785338968038559,
      "learning_rate": 9.586776859504134e-06,
      "loss": 1.8532,
      "step": 10
    },
    {
      "epoch": 0.045454545454545456,
      "grad_norm": 0.18187786638736725,
      "learning_rate": 9.545454545454547e-06,
      "loss": 1.9232,
      "step": 11
    },
    {
      "epoch": 0.049586776859504134,
      "grad_norm": 0.19290965795516968,
      "learning_rate": 9.50413223140496e-06,
      "loss": 1.9536,
      "step": 12
    },
    {
      "epoch": 0.05371900826446281,
      "grad_norm": 0.19236230850219727,
      "learning_rate": 9.462809917355372e-06,
      "loss": 1.8572,
      "step": 13
    },
    {
      "epoch": 0.05785123966942149,
      "grad_norm": 0.21088214218616486,
      "learning_rate": 9.421487603305785e-06,
      "loss": 1.9461,
      "step": 14
    },
    {
      "epoch": 0.06198347107438017,
      "grad_norm": 0.2091795653104782,
      "learning_rate": 9.3801652892562e-06,
      "loss": 1.9347,
      "step": 15
    },
    {
      "epoch": 0.06611570247933884,
      "grad_norm": 0.21398870646953583,
      "learning_rate": 9.338842975206613e-06,
      "loss": 1.922,
      "step": 16
    },
    {
      "epoch": 0.07024793388429752,
      "grad_norm": 0.20920316874980927,
      "learning_rate": 9.297520661157025e-06,
      "loss": 1.8347,
      "step": 17
    },
    {
      "epoch": 0.0743801652892562,
      "grad_norm": 0.22646453976631165,
      "learning_rate": 9.25619834710744e-06,
      "loss": 1.9168,
      "step": 18
    },
    {
      "epoch": 0.07851239669421488,
      "grad_norm": 0.22585293650627136,
      "learning_rate": 9.214876033057853e-06,
      "loss": 1.9227,
      "step": 19
    },
    {
      "epoch": 0.08264462809917356,
      "grad_norm": 0.24510250985622406,
      "learning_rate": 9.173553719008265e-06,
      "loss": 1.9615,
      "step": 20
    },
    {
      "epoch": 0.08677685950413223,
      "grad_norm": 0.25765278935432434,
      "learning_rate": 9.132231404958678e-06,
      "loss": 2.0353,
      "step": 21
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 0.24669119715690613,
      "learning_rate": 9.090909090909091e-06,
      "loss": 2.0307,
      "step": 22
    },
    {
      "epoch": 0.09504132231404959,
      "grad_norm": 0.27933332324028015,
      "learning_rate": 9.049586776859506e-06,
      "loss": 2.0923,
      "step": 23
    },
    {
      "epoch": 0.09917355371900827,
      "grad_norm": 0.26190587878227234,
      "learning_rate": 9.008264462809918e-06,
      "loss": 1.9315,
      "step": 24
    },
    {
      "epoch": 0.10330578512396695,
      "grad_norm": 0.2693282961845398,
      "learning_rate": 8.966942148760331e-06,
      "loss": 2.0,
      "step": 25
    },
    {
      "epoch": 0.10743801652892562,
      "grad_norm": 0.27441301941871643,
      "learning_rate": 8.925619834710744e-06,
      "loss": 2.0193,
      "step": 26
    },
    {
      "epoch": 0.1115702479338843,
      "grad_norm": 0.2680116891860962,
      "learning_rate": 8.884297520661158e-06,
      "loss": 2.0443,
      "step": 27
    },
    {
      "epoch": 0.11570247933884298,
      "grad_norm": 0.28165847063064575,
      "learning_rate": 8.842975206611571e-06,
      "loss": 1.9533,
      "step": 28
    },
    {
      "epoch": 0.11983471074380166,
      "grad_norm": 0.2774558663368225,
      "learning_rate": 8.801652892561984e-06,
      "loss": 2.0188,
      "step": 29
    },
    {
      "epoch": 0.12396694214876033,
      "grad_norm": 0.283816397190094,
      "learning_rate": 8.760330578512397e-06,
      "loss": 2.0198,
      "step": 30
    },
    {
      "epoch": 0.128099173553719,
      "grad_norm": 0.2835429310798645,
      "learning_rate": 8.71900826446281e-06,
      "loss": 1.947,
      "step": 31
    },
    {
      "epoch": 0.1322314049586777,
      "grad_norm": 0.2720879912376404,
      "learning_rate": 8.677685950413224e-06,
      "loss": 1.9155,
      "step": 32
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 0.2787908911705017,
      "learning_rate": 8.636363636363637e-06,
      "loss": 1.885,
      "step": 33
    },
    {
      "epoch": 0.14049586776859505,
      "grad_norm": 0.28071197867393494,
      "learning_rate": 8.59504132231405e-06,
      "loss": 1.8583,
      "step": 34
    },
    {
      "epoch": 0.1446280991735537,
      "grad_norm": 0.28638049960136414,
      "learning_rate": 8.553719008264464e-06,
      "loss": 1.9749,
      "step": 35
    },
    {
      "epoch": 0.1487603305785124,
      "grad_norm": 0.2940247654914856,
      "learning_rate": 8.512396694214877e-06,
      "loss": 1.9215,
      "step": 36
    },
    {
      "epoch": 0.15289256198347106,
      "grad_norm": 0.2773992121219635,
      "learning_rate": 8.47107438016529e-06,
      "loss": 1.9353,
      "step": 37
    },
    {
      "epoch": 0.15702479338842976,
      "grad_norm": 0.3020448684692383,
      "learning_rate": 8.429752066115703e-06,
      "loss": 1.9072,
      "step": 38
    },
    {
      "epoch": 0.16115702479338842,
      "grad_norm": 0.29430723190307617,
      "learning_rate": 8.388429752066116e-06,
      "loss": 1.9095,
      "step": 39
    },
    {
      "epoch": 0.1652892561983471,
      "grad_norm": 0.2791050970554352,
      "learning_rate": 8.34710743801653e-06,
      "loss": 1.8698,
      "step": 40
    },
    {
      "epoch": 0.16942148760330578,
      "grad_norm": 0.29201751947402954,
      "learning_rate": 8.305785123966943e-06,
      "loss": 1.9336,
      "step": 41
    },
    {
      "epoch": 0.17355371900826447,
      "grad_norm": 0.25510144233703613,
      "learning_rate": 8.264462809917356e-06,
      "loss": 1.8471,
      "step": 42
    },
    {
      "epoch": 0.17768595041322313,
      "grad_norm": 0.27273449301719666,
      "learning_rate": 8.22314049586777e-06,
      "loss": 1.8281,
      "step": 43
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.2743457555770874,
      "learning_rate": 8.181818181818183e-06,
      "loss": 1.9007,
      "step": 44
    },
    {
      "epoch": 0.1859504132231405,
      "grad_norm": 0.27609583735466003,
      "learning_rate": 8.140495867768596e-06,
      "loss": 1.8284,
      "step": 45
    },
    {
      "epoch": 0.19008264462809918,
      "grad_norm": 0.26752588152885437,
      "learning_rate": 8.099173553719009e-06,
      "loss": 1.8266,
      "step": 46
    },
    {
      "epoch": 0.19421487603305784,
      "grad_norm": 0.27254441380500793,
      "learning_rate": 8.057851239669421e-06,
      "loss": 1.8473,
      "step": 47
    },
    {
      "epoch": 0.19834710743801653,
      "grad_norm": 0.27281084656715393,
      "learning_rate": 8.016528925619836e-06,
      "loss": 1.9348,
      "step": 48
    },
    {
      "epoch": 0.2024793388429752,
      "grad_norm": 0.265490859746933,
      "learning_rate": 7.975206611570249e-06,
      "loss": 1.9098,
      "step": 49
    },
    {
      "epoch": 0.2066115702479339,
      "grad_norm": 0.2591995596885681,
      "learning_rate": 7.933884297520661e-06,
      "loss": 1.8532,
      "step": 50
    },
    {
      "epoch": 0.21074380165289255,
      "grad_norm": 0.27537262439727783,
      "learning_rate": 7.892561983471076e-06,
      "loss": 1.8636,
      "step": 51
    },
    {
      "epoch": 0.21487603305785125,
      "grad_norm": 0.2684531807899475,
      "learning_rate": 7.851239669421489e-06,
      "loss": 1.7468,
      "step": 52
    },
    {
      "epoch": 0.2190082644628099,
      "grad_norm": 0.27100837230682373,
      "learning_rate": 7.809917355371902e-06,
      "loss": 1.964,
      "step": 53
    },
    {
      "epoch": 0.2231404958677686,
      "grad_norm": 0.2652667164802551,
      "learning_rate": 7.768595041322314e-06,
      "loss": 1.8481,
      "step": 54
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 0.25347599387168884,
      "learning_rate": 7.727272727272727e-06,
      "loss": 1.9248,
      "step": 55
    },
    {
      "epoch": 0.23140495867768596,
      "grad_norm": 0.2594529688358307,
      "learning_rate": 7.685950413223142e-06,
      "loss": 1.9079,
      "step": 56
    },
    {
      "epoch": 0.23553719008264462,
      "grad_norm": 0.2740556001663208,
      "learning_rate": 7.644628099173555e-06,
      "loss": 1.8116,
      "step": 57
    },
    {
      "epoch": 0.2396694214876033,
      "grad_norm": 0.2498086541891098,
      "learning_rate": 7.603305785123968e-06,
      "loss": 1.7945,
      "step": 58
    },
    {
      "epoch": 0.24380165289256198,
      "grad_norm": 0.2498074471950531,
      "learning_rate": 7.561983471074381e-06,
      "loss": 1.8729,
      "step": 59
    },
    {
      "epoch": 0.24793388429752067,
      "grad_norm": 0.2547696828842163,
      "learning_rate": 7.520661157024795e-06,
      "loss": 1.8504,
      "step": 60
    },
    {
      "epoch": 0.25206611570247933,
      "grad_norm": 0.24527904391288757,
      "learning_rate": 7.479338842975207e-06,
      "loss": 1.8259,
      "step": 61
    },
    {
      "epoch": 0.256198347107438,
      "grad_norm": 0.24427714943885803,
      "learning_rate": 7.43801652892562e-06,
      "loss": 1.8512,
      "step": 62
    },
    {
      "epoch": 0.2603305785123967,
      "grad_norm": 0.25535401701927185,
      "learning_rate": 7.396694214876033e-06,
      "loss": 1.9013,
      "step": 63
    },
    {
      "epoch": 0.2644628099173554,
      "grad_norm": 0.25003406405448914,
      "learning_rate": 7.355371900826447e-06,
      "loss": 1.7674,
      "step": 64
    },
    {
      "epoch": 0.26859504132231404,
      "grad_norm": 0.2516063451766968,
      "learning_rate": 7.31404958677686e-06,
      "loss": 1.8561,
      "step": 65
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 0.24849283695220947,
      "learning_rate": 7.272727272727273e-06,
      "loss": 1.9069,
      "step": 66
    },
    {
      "epoch": 0.2768595041322314,
      "grad_norm": 0.23888932168483734,
      "learning_rate": 7.231404958677687e-06,
      "loss": 1.759,
      "step": 67
    },
    {
      "epoch": 0.2809917355371901,
      "grad_norm": 0.23254430294036865,
      "learning_rate": 7.1900826446281005e-06,
      "loss": 1.7327,
      "step": 68
    },
    {
      "epoch": 0.28512396694214875,
      "grad_norm": 0.23558281362056732,
      "learning_rate": 7.1487603305785124e-06,
      "loss": 1.7799,
      "step": 69
    },
    {
      "epoch": 0.2892561983471074,
      "grad_norm": 0.25738146901130676,
      "learning_rate": 7.107438016528926e-06,
      "loss": 1.8367,
      "step": 70
    },
    {
      "epoch": 0.29338842975206614,
      "grad_norm": 0.2553260028362274,
      "learning_rate": 7.066115702479339e-06,
      "loss": 1.7651,
      "step": 71
    },
    {
      "epoch": 0.2975206611570248,
      "grad_norm": 0.23750841617584229,
      "learning_rate": 7.0247933884297525e-06,
      "loss": 1.7739,
      "step": 72
    },
    {
      "epoch": 0.30165289256198347,
      "grad_norm": 0.24391470849514008,
      "learning_rate": 6.983471074380166e-06,
      "loss": 1.7786,
      "step": 73
    },
    {
      "epoch": 0.30578512396694213,
      "grad_norm": 0.23539938032627106,
      "learning_rate": 6.942148760330579e-06,
      "loss": 1.7458,
      "step": 74
    },
    {
      "epoch": 0.30991735537190085,
      "grad_norm": 0.2562333345413208,
      "learning_rate": 6.900826446280993e-06,
      "loss": 1.7769,
      "step": 75
    },
    {
      "epoch": 0.3140495867768595,
      "grad_norm": 0.24707230925559998,
      "learning_rate": 6.859504132231406e-06,
      "loss": 1.8149,
      "step": 76
    },
    {
      "epoch": 0.3181818181818182,
      "grad_norm": 0.2444324493408203,
      "learning_rate": 6.818181818181818e-06,
      "loss": 1.8119,
      "step": 77
    },
    {
      "epoch": 0.32231404958677684,
      "grad_norm": 0.23097573220729828,
      "learning_rate": 6.776859504132232e-06,
      "loss": 1.8132,
      "step": 78
    },
    {
      "epoch": 0.32644628099173556,
      "grad_norm": 0.2282407581806183,
      "learning_rate": 6.735537190082645e-06,
      "loss": 1.7909,
      "step": 79
    },
    {
      "epoch": 0.3305785123966942,
      "grad_norm": 0.244503915309906,
      "learning_rate": 6.694214876033058e-06,
      "loss": 1.8004,
      "step": 80
    },
    {
      "epoch": 0.3347107438016529,
      "grad_norm": 0.2202165275812149,
      "learning_rate": 6.652892561983472e-06,
      "loss": 1.6869,
      "step": 81
    },
    {
      "epoch": 0.33884297520661155,
      "grad_norm": 0.25013434886932373,
      "learning_rate": 6.611570247933885e-06,
      "loss": 1.7756,
      "step": 82
    },
    {
      "epoch": 0.34297520661157027,
      "grad_norm": 0.23903381824493408,
      "learning_rate": 6.5702479338842985e-06,
      "loss": 1.8687,
      "step": 83
    },
    {
      "epoch": 0.34710743801652894,
      "grad_norm": 0.24788710474967957,
      "learning_rate": 6.528925619834712e-06,
      "loss": 1.7638,
      "step": 84
    },
    {
      "epoch": 0.3512396694214876,
      "grad_norm": 0.23012609779834747,
      "learning_rate": 6.487603305785124e-06,
      "loss": 1.7314,
      "step": 85
    },
    {
      "epoch": 0.35537190082644626,
      "grad_norm": 0.23892751336097717,
      "learning_rate": 6.446280991735537e-06,
      "loss": 1.812,
      "step": 86
    },
    {
      "epoch": 0.359504132231405,
      "grad_norm": 0.26781031489372253,
      "learning_rate": 6.404958677685951e-06,
      "loss": 1.76,
      "step": 87
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.24897883832454681,
      "learning_rate": 6.363636363636364e-06,
      "loss": 1.7797,
      "step": 88
    },
    {
      "epoch": 0.3677685950413223,
      "grad_norm": 0.23585253953933716,
      "learning_rate": 6.322314049586777e-06,
      "loss": 1.7395,
      "step": 89
    },
    {
      "epoch": 0.371900826446281,
      "grad_norm": 0.2473200559616089,
      "learning_rate": 6.280991735537191e-06,
      "loss": 1.7748,
      "step": 90
    },
    {
      "epoch": 0.3760330578512397,
      "grad_norm": 0.20990578830242157,
      "learning_rate": 6.239669421487604e-06,
      "loss": 1.6423,
      "step": 91
    },
    {
      "epoch": 0.38016528925619836,
      "grad_norm": 0.27370184659957886,
      "learning_rate": 6.198347107438017e-06,
      "loss": 1.7637,
      "step": 92
    },
    {
      "epoch": 0.384297520661157,
      "grad_norm": 0.23353751003742218,
      "learning_rate": 6.15702479338843e-06,
      "loss": 1.8031,
      "step": 93
    },
    {
      "epoch": 0.3884297520661157,
      "grad_norm": 0.25242578983306885,
      "learning_rate": 6.115702479338843e-06,
      "loss": 1.7187,
      "step": 94
    },
    {
      "epoch": 0.3925619834710744,
      "grad_norm": 0.26314014196395874,
      "learning_rate": 6.074380165289256e-06,
      "loss": 1.7663,
      "step": 95
    },
    {
      "epoch": 0.39669421487603307,
      "grad_norm": 0.23005221784114838,
      "learning_rate": 6.03305785123967e-06,
      "loss": 1.712,
      "step": 96
    },
    {
      "epoch": 0.40082644628099173,
      "grad_norm": 0.23051100969314575,
      "learning_rate": 5.991735537190083e-06,
      "loss": 1.6499,
      "step": 97
    },
    {
      "epoch": 0.4049586776859504,
      "grad_norm": 0.22571218013763428,
      "learning_rate": 5.9504132231404965e-06,
      "loss": 1.7121,
      "step": 98
    },
    {
      "epoch": 0.4090909090909091,
      "grad_norm": 0.22399099171161652,
      "learning_rate": 5.90909090909091e-06,
      "loss": 1.6854,
      "step": 99
    },
    {
      "epoch": 0.4132231404958678,
      "grad_norm": 0.22115306556224823,
      "learning_rate": 5.867768595041323e-06,
      "loss": 1.7528,
      "step": 100
    },
    {
      "epoch": 0.41735537190082644,
      "grad_norm": 0.24841462075710297,
      "learning_rate": 5.826446280991736e-06,
      "loss": 1.7365,
      "step": 101
    },
    {
      "epoch": 0.4214876033057851,
      "grad_norm": 0.21741348505020142,
      "learning_rate": 5.785123966942149e-06,
      "loss": 1.7083,
      "step": 102
    },
    {
      "epoch": 0.4256198347107438,
      "grad_norm": 0.2234075665473938,
      "learning_rate": 5.743801652892562e-06,
      "loss": 1.6826,
      "step": 103
    },
    {
      "epoch": 0.4297520661157025,
      "grad_norm": 0.22538623213768005,
      "learning_rate": 5.702479338842976e-06,
      "loss": 1.7443,
      "step": 104
    },
    {
      "epoch": 0.43388429752066116,
      "grad_norm": 0.21332193911075592,
      "learning_rate": 5.661157024793389e-06,
      "loss": 1.6649,
      "step": 105
    },
    {
      "epoch": 0.4380165289256198,
      "grad_norm": 0.24505837261676788,
      "learning_rate": 5.619834710743802e-06,
      "loss": 1.8533,
      "step": 106
    },
    {
      "epoch": 0.44214876033057854,
      "grad_norm": 0.22248229384422302,
      "learning_rate": 5.578512396694216e-06,
      "loss": 1.7278,
      "step": 107
    },
    {
      "epoch": 0.4462809917355372,
      "grad_norm": 0.2204764038324356,
      "learning_rate": 5.537190082644629e-06,
      "loss": 1.7078,
      "step": 108
    },
    {
      "epoch": 0.45041322314049587,
      "grad_norm": 0.2260615974664688,
      "learning_rate": 5.495867768595042e-06,
      "loss": 1.7898,
      "step": 109
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.24133943021297455,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 1.7457,
      "step": 110
    },
    {
      "epoch": 0.45867768595041325,
      "grad_norm": 0.21375595033168793,
      "learning_rate": 5.413223140495868e-06,
      "loss": 1.6839,
      "step": 111
    },
    {
      "epoch": 0.4628099173553719,
      "grad_norm": 0.2149706780910492,
      "learning_rate": 5.371900826446281e-06,
      "loss": 1.7084,
      "step": 112
    },
    {
      "epoch": 0.4669421487603306,
      "grad_norm": 0.21798448264598846,
      "learning_rate": 5.3305785123966946e-06,
      "loss": 1.7338,
      "step": 113
    },
    {
      "epoch": 0.47107438016528924,
      "grad_norm": 0.22930534183979034,
      "learning_rate": 5.289256198347108e-06,
      "loss": 1.7118,
      "step": 114
    },
    {
      "epoch": 0.47520661157024796,
      "grad_norm": 0.21905331313610077,
      "learning_rate": 5.247933884297521e-06,
      "loss": 1.7581,
      "step": 115
    },
    {
      "epoch": 0.4793388429752066,
      "grad_norm": 0.2666592597961426,
      "learning_rate": 5.206611570247935e-06,
      "loss": 1.7292,
      "step": 116
    },
    {
      "epoch": 0.4834710743801653,
      "grad_norm": 0.2206874042749405,
      "learning_rate": 5.165289256198347e-06,
      "loss": 1.7305,
      "step": 117
    },
    {
      "epoch": 0.48760330578512395,
      "grad_norm": 0.22711925208568573,
      "learning_rate": 5.12396694214876e-06,
      "loss": 1.6886,
      "step": 118
    },
    {
      "epoch": 0.49173553719008267,
      "grad_norm": 0.21445244550704956,
      "learning_rate": 5.082644628099174e-06,
      "loss": 1.7324,
      "step": 119
    },
    {
      "epoch": 0.49586776859504134,
      "grad_norm": 0.2524474859237671,
      "learning_rate": 5.041322314049587e-06,
      "loss": 1.7729,
      "step": 120
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23446372151374817,
      "learning_rate": 5e-06,
      "loss": 1.6154,
      "step": 121
    },
    {
      "epoch": 0.5041322314049587,
      "grad_norm": 0.22260743379592896,
      "learning_rate": 4.958677685950414e-06,
      "loss": 1.6891,
      "step": 122
    },
    {
      "epoch": 0.5082644628099173,
      "grad_norm": 0.22862933576107025,
      "learning_rate": 4.917355371900827e-06,
      "loss": 1.6711,
      "step": 123
    },
    {
      "epoch": 0.512396694214876,
      "grad_norm": 0.23864181339740753,
      "learning_rate": 4.87603305785124e-06,
      "loss": 1.6873,
      "step": 124
    },
    {
      "epoch": 0.5165289256198347,
      "grad_norm": 0.2561310827732086,
      "learning_rate": 4.834710743801653e-06,
      "loss": 1.7676,
      "step": 125
    },
    {
      "epoch": 0.5206611570247934,
      "grad_norm": 0.21949930489063263,
      "learning_rate": 4.793388429752067e-06,
      "loss": 1.7488,
      "step": 126
    },
    {
      "epoch": 0.5247933884297521,
      "grad_norm": 0.23986859619617462,
      "learning_rate": 4.75206611570248e-06,
      "loss": 1.7251,
      "step": 127
    },
    {
      "epoch": 0.5289256198347108,
      "grad_norm": 0.20821848511695862,
      "learning_rate": 4.710743801652893e-06,
      "loss": 1.6392,
      "step": 128
    },
    {
      "epoch": 0.5330578512396694,
      "grad_norm": 0.23042964935302734,
      "learning_rate": 4.669421487603306e-06,
      "loss": 1.658,
      "step": 129
    },
    {
      "epoch": 0.5371900826446281,
      "grad_norm": 0.2073439061641693,
      "learning_rate": 4.62809917355372e-06,
      "loss": 1.6188,
      "step": 130
    },
    {
      "epoch": 0.5413223140495868,
      "grad_norm": 0.20989008247852325,
      "learning_rate": 4.586776859504133e-06,
      "loss": 1.6483,
      "step": 131
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.23288583755493164,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 1.714,
      "step": 132
    },
    {
      "epoch": 0.5495867768595041,
      "grad_norm": 0.22618982195854187,
      "learning_rate": 4.504132231404959e-06,
      "loss": 1.6207,
      "step": 133
    },
    {
      "epoch": 0.5537190082644629,
      "grad_norm": 0.22818031907081604,
      "learning_rate": 4.462809917355372e-06,
      "loss": 1.7065,
      "step": 134
    },
    {
      "epoch": 0.5578512396694215,
      "grad_norm": 0.21046042442321777,
      "learning_rate": 4.421487603305786e-06,
      "loss": 1.6718,
      "step": 135
    },
    {
      "epoch": 0.5619834710743802,
      "grad_norm": 0.22790120542049408,
      "learning_rate": 4.3801652892561984e-06,
      "loss": 1.7159,
      "step": 136
    },
    {
      "epoch": 0.5661157024793388,
      "grad_norm": 0.23064090311527252,
      "learning_rate": 4.338842975206612e-06,
      "loss": 1.72,
      "step": 137
    },
    {
      "epoch": 0.5702479338842975,
      "grad_norm": 0.21820063889026642,
      "learning_rate": 4.297520661157025e-06,
      "loss": 1.7422,
      "step": 138
    },
    {
      "epoch": 0.5743801652892562,
      "grad_norm": 0.21634773910045624,
      "learning_rate": 4.2561983471074386e-06,
      "loss": 1.6999,
      "step": 139
    },
    {
      "epoch": 0.5785123966942148,
      "grad_norm": 0.21947871148586273,
      "learning_rate": 4.214876033057851e-06,
      "loss": 1.7012,
      "step": 140
    },
    {
      "epoch": 0.5826446280991735,
      "grad_norm": 0.25091108679771423,
      "learning_rate": 4.173553719008265e-06,
      "loss": 1.7777,
      "step": 141
    },
    {
      "epoch": 0.5867768595041323,
      "grad_norm": 0.2140003889799118,
      "learning_rate": 4.132231404958678e-06,
      "loss": 1.6804,
      "step": 142
    },
    {
      "epoch": 0.5909090909090909,
      "grad_norm": 0.21121805906295776,
      "learning_rate": 4.0909090909090915e-06,
      "loss": 1.6423,
      "step": 143
    },
    {
      "epoch": 0.5950413223140496,
      "grad_norm": 0.2193901091814041,
      "learning_rate": 4.049586776859504e-06,
      "loss": 1.6364,
      "step": 144
    },
    {
      "epoch": 0.5991735537190083,
      "grad_norm": 0.20386271178722382,
      "learning_rate": 4.008264462809918e-06,
      "loss": 1.6611,
      "step": 145
    },
    {
      "epoch": 0.6033057851239669,
      "grad_norm": 0.24780020117759705,
      "learning_rate": 3.966942148760331e-06,
      "loss": 1.6012,
      "step": 146
    },
    {
      "epoch": 0.6074380165289256,
      "grad_norm": 0.22096756100654602,
      "learning_rate": 3.925619834710744e-06,
      "loss": 1.6789,
      "step": 147
    },
    {
      "epoch": 0.6115702479338843,
      "grad_norm": 0.20254209637641907,
      "learning_rate": 3.884297520661157e-06,
      "loss": 1.6872,
      "step": 148
    },
    {
      "epoch": 0.6157024793388429,
      "grad_norm": 0.20938359200954437,
      "learning_rate": 3.842975206611571e-06,
      "loss": 1.6556,
      "step": 149
    },
    {
      "epoch": 0.6198347107438017,
      "grad_norm": 0.23572930693626404,
      "learning_rate": 3.801652892561984e-06,
      "loss": 1.7176,
      "step": 150
    },
    {
      "epoch": 0.6239669421487604,
      "grad_norm": 0.1991184651851654,
      "learning_rate": 3.7603305785123973e-06,
      "loss": 1.6704,
      "step": 151
    },
    {
      "epoch": 0.628099173553719,
      "grad_norm": 0.21645860373973846,
      "learning_rate": 3.71900826446281e-06,
      "loss": 1.635,
      "step": 152
    },
    {
      "epoch": 0.6322314049586777,
      "grad_norm": 0.2397676557302475,
      "learning_rate": 3.6776859504132234e-06,
      "loss": 1.6484,
      "step": 153
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.20403875410556793,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 1.6795,
      "step": 154
    },
    {
      "epoch": 0.640495867768595,
      "grad_norm": 0.20842084288597107,
      "learning_rate": 3.5950413223140502e-06,
      "loss": 1.5839,
      "step": 155
    },
    {
      "epoch": 0.6446280991735537,
      "grad_norm": 0.2065836638212204,
      "learning_rate": 3.553719008264463e-06,
      "loss": 1.6684,
      "step": 156
    },
    {
      "epoch": 0.6487603305785123,
      "grad_norm": 0.1999097764492035,
      "learning_rate": 3.5123966942148763e-06,
      "loss": 1.6396,
      "step": 157
    },
    {
      "epoch": 0.6528925619834711,
      "grad_norm": 0.21142417192459106,
      "learning_rate": 3.4710743801652895e-06,
      "loss": 1.6853,
      "step": 158
    },
    {
      "epoch": 0.6570247933884298,
      "grad_norm": 0.2628518342971802,
      "learning_rate": 3.429752066115703e-06,
      "loss": 1.6653,
      "step": 159
    },
    {
      "epoch": 0.6611570247933884,
      "grad_norm": 0.22541484236717224,
      "learning_rate": 3.388429752066116e-06,
      "loss": 1.6815,
      "step": 160
    },
    {
      "epoch": 0.6652892561983471,
      "grad_norm": 0.21939656138420105,
      "learning_rate": 3.347107438016529e-06,
      "loss": 1.6832,
      "step": 161
    },
    {
      "epoch": 0.6694214876033058,
      "grad_norm": 0.2548986077308655,
      "learning_rate": 3.3057851239669424e-06,
      "loss": 1.6538,
      "step": 162
    },
    {
      "epoch": 0.6735537190082644,
      "grad_norm": 0.21173818409442902,
      "learning_rate": 3.264462809917356e-06,
      "loss": 1.563,
      "step": 163
    },
    {
      "epoch": 0.6776859504132231,
      "grad_norm": 0.21219997107982635,
      "learning_rate": 3.2231404958677685e-06,
      "loss": 1.6127,
      "step": 164
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 0.21379010379314423,
      "learning_rate": 3.181818181818182e-06,
      "loss": 1.6534,
      "step": 165
    },
    {
      "epoch": 0.6859504132231405,
      "grad_norm": 0.21609431505203247,
      "learning_rate": 3.1404958677685953e-06,
      "loss": 1.7105,
      "step": 166
    },
    {
      "epoch": 0.6900826446280992,
      "grad_norm": 0.23285210132598877,
      "learning_rate": 3.0991735537190086e-06,
      "loss": 1.6895,
      "step": 167
    },
    {
      "epoch": 0.6942148760330579,
      "grad_norm": 0.23710978031158447,
      "learning_rate": 3.0578512396694214e-06,
      "loss": 1.6208,
      "step": 168
    },
    {
      "epoch": 0.6983471074380165,
      "grad_norm": 0.20837798714637756,
      "learning_rate": 3.016528925619835e-06,
      "loss": 1.6422,
      "step": 169
    },
    {
      "epoch": 0.7024793388429752,
      "grad_norm": 0.1929236799478531,
      "learning_rate": 2.9752066115702483e-06,
      "loss": 1.602,
      "step": 170
    },
    {
      "epoch": 0.7066115702479339,
      "grad_norm": 0.2555234730243683,
      "learning_rate": 2.9338842975206615e-06,
      "loss": 1.6946,
      "step": 171
    },
    {
      "epoch": 0.7107438016528925,
      "grad_norm": 0.21097096800804138,
      "learning_rate": 2.8925619834710743e-06,
      "loss": 1.6702,
      "step": 172
    },
    {
      "epoch": 0.7148760330578512,
      "grad_norm": 0.19178104400634766,
      "learning_rate": 2.851239669421488e-06,
      "loss": 1.658,
      "step": 173
    },
    {
      "epoch": 0.71900826446281,
      "grad_norm": 0.1989249885082245,
      "learning_rate": 2.809917355371901e-06,
      "loss": 1.6235,
      "step": 174
    },
    {
      "epoch": 0.7231404958677686,
      "grad_norm": 0.2068866640329361,
      "learning_rate": 2.7685950413223144e-06,
      "loss": 1.5794,
      "step": 175
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.2141098529100418,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 1.7234,
      "step": 176
    },
    {
      "epoch": 0.731404958677686,
      "grad_norm": 0.2172386348247528,
      "learning_rate": 2.6859504132231405e-06,
      "loss": 1.696,
      "step": 177
    },
    {
      "epoch": 0.7355371900826446,
      "grad_norm": 0.21580706536769867,
      "learning_rate": 2.644628099173554e-06,
      "loss": 1.6702,
      "step": 178
    },
    {
      "epoch": 0.7396694214876033,
      "grad_norm": 0.19322863221168518,
      "learning_rate": 2.6033057851239673e-06,
      "loss": 1.5562,
      "step": 179
    },
    {
      "epoch": 0.743801652892562,
      "grad_norm": 0.21309177577495575,
      "learning_rate": 2.56198347107438e-06,
      "loss": 1.7298,
      "step": 180
    },
    {
      "epoch": 0.7479338842975206,
      "grad_norm": 0.21737979352474213,
      "learning_rate": 2.5206611570247934e-06,
      "loss": 1.647,
      "step": 181
    },
    {
      "epoch": 0.7520661157024794,
      "grad_norm": 0.21412277221679688,
      "learning_rate": 2.479338842975207e-06,
      "loss": 1.6582,
      "step": 182
    },
    {
      "epoch": 0.756198347107438,
      "grad_norm": 0.20408561825752258,
      "learning_rate": 2.43801652892562e-06,
      "loss": 1.6248,
      "step": 183
    },
    {
      "epoch": 0.7603305785123967,
      "grad_norm": 0.22556652128696442,
      "learning_rate": 2.3966942148760335e-06,
      "loss": 1.595,
      "step": 184
    },
    {
      "epoch": 0.7644628099173554,
      "grad_norm": 0.22902049124240875,
      "learning_rate": 2.3553719008264463e-06,
      "loss": 1.5972,
      "step": 185
    },
    {
      "epoch": 0.768595041322314,
      "grad_norm": 0.21549232304096222,
      "learning_rate": 2.31404958677686e-06,
      "loss": 1.5896,
      "step": 186
    },
    {
      "epoch": 0.7727272727272727,
      "grad_norm": 0.2015775740146637,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 1.646,
      "step": 187
    },
    {
      "epoch": 0.7768595041322314,
      "grad_norm": 0.2306847870349884,
      "learning_rate": 2.231404958677686e-06,
      "loss": 1.6198,
      "step": 188
    },
    {
      "epoch": 0.78099173553719,
      "grad_norm": 0.19418193399906158,
      "learning_rate": 2.1900826446280992e-06,
      "loss": 1.59,
      "step": 189
    },
    {
      "epoch": 0.7851239669421488,
      "grad_norm": 0.219194233417511,
      "learning_rate": 2.1487603305785124e-06,
      "loss": 1.6799,
      "step": 190
    },
    {
      "epoch": 0.7892561983471075,
      "grad_norm": 0.20487534999847412,
      "learning_rate": 2.1074380165289257e-06,
      "loss": 1.65,
      "step": 191
    },
    {
      "epoch": 0.7933884297520661,
      "grad_norm": 0.19993071258068085,
      "learning_rate": 2.066115702479339e-06,
      "loss": 1.6595,
      "step": 192
    },
    {
      "epoch": 0.7975206611570248,
      "grad_norm": 0.1964050978422165,
      "learning_rate": 2.024793388429752e-06,
      "loss": 1.5711,
      "step": 193
    },
    {
      "epoch": 0.8016528925619835,
      "grad_norm": 0.23315295577049255,
      "learning_rate": 1.9834710743801654e-06,
      "loss": 1.5905,
      "step": 194
    },
    {
      "epoch": 0.8057851239669421,
      "grad_norm": 0.19839169085025787,
      "learning_rate": 1.9421487603305786e-06,
      "loss": 1.5852,
      "step": 195
    },
    {
      "epoch": 0.8099173553719008,
      "grad_norm": 0.19847120344638824,
      "learning_rate": 1.900826446280992e-06,
      "loss": 1.6396,
      "step": 196
    },
    {
      "epoch": 0.8140495867768595,
      "grad_norm": 0.22999809682369232,
      "learning_rate": 1.859504132231405e-06,
      "loss": 1.5756,
      "step": 197
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.2322094440460205,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 1.7321,
      "step": 198
    },
    {
      "epoch": 0.8223140495867769,
      "grad_norm": 0.23996399343013763,
      "learning_rate": 1.7768595041322315e-06,
      "loss": 1.6773,
      "step": 199
    },
    {
      "epoch": 0.8264462809917356,
      "grad_norm": 0.19855372607707977,
      "learning_rate": 1.7355371900826448e-06,
      "loss": 1.5775,
      "step": 200
    },
    {
      "epoch": 0.8305785123966942,
      "grad_norm": 0.2526344954967499,
      "learning_rate": 1.694214876033058e-06,
      "loss": 1.6407,
      "step": 201
    },
    {
      "epoch": 0.8347107438016529,
      "grad_norm": 0.22508279979228973,
      "learning_rate": 1.6528925619834712e-06,
      "loss": 1.6703,
      "step": 202
    },
    {
      "epoch": 0.8388429752066116,
      "grad_norm": 0.22720687091350555,
      "learning_rate": 1.6115702479338842e-06,
      "loss": 1.6681,
      "step": 203
    },
    {
      "epoch": 0.8429752066115702,
      "grad_norm": 0.23210258781909943,
      "learning_rate": 1.5702479338842977e-06,
      "loss": 1.6659,
      "step": 204
    },
    {
      "epoch": 0.8471074380165289,
      "grad_norm": 0.26924028992652893,
      "learning_rate": 1.5289256198347107e-06,
      "loss": 1.7742,
      "step": 205
    },
    {
      "epoch": 0.8512396694214877,
      "grad_norm": 0.19536957144737244,
      "learning_rate": 1.4876033057851241e-06,
      "loss": 1.641,
      "step": 206
    },
    {
      "epoch": 0.8553719008264463,
      "grad_norm": 0.20508214831352234,
      "learning_rate": 1.4462809917355372e-06,
      "loss": 1.6532,
      "step": 207
    },
    {
      "epoch": 0.859504132231405,
      "grad_norm": 0.23412977159023285,
      "learning_rate": 1.4049586776859506e-06,
      "loss": 1.709,
      "step": 208
    },
    {
      "epoch": 0.8636363636363636,
      "grad_norm": 0.20496754348278046,
      "learning_rate": 1.3636363636363636e-06,
      "loss": 1.572,
      "step": 209
    },
    {
      "epoch": 0.8677685950413223,
      "grad_norm": 0.19729302823543549,
      "learning_rate": 1.322314049586777e-06,
      "loss": 1.6229,
      "step": 210
    },
    {
      "epoch": 0.871900826446281,
      "grad_norm": 0.20501768589019775,
      "learning_rate": 1.28099173553719e-06,
      "loss": 1.6185,
      "step": 211
    },
    {
      "epoch": 0.8760330578512396,
      "grad_norm": 0.2092459797859192,
      "learning_rate": 1.2396694214876035e-06,
      "loss": 1.5979,
      "step": 212
    },
    {
      "epoch": 0.8801652892561983,
      "grad_norm": 0.22356414794921875,
      "learning_rate": 1.1983471074380167e-06,
      "loss": 1.6803,
      "step": 213
    },
    {
      "epoch": 0.8842975206611571,
      "grad_norm": 0.20351006090641022,
      "learning_rate": 1.15702479338843e-06,
      "loss": 1.6471,
      "step": 214
    },
    {
      "epoch": 0.8884297520661157,
      "grad_norm": 0.21975171566009521,
      "learning_rate": 1.115702479338843e-06,
      "loss": 1.5911,
      "step": 215
    },
    {
      "epoch": 0.8925619834710744,
      "grad_norm": 0.19022318720817566,
      "learning_rate": 1.0743801652892562e-06,
      "loss": 1.6235,
      "step": 216
    },
    {
      "epoch": 0.8966942148760331,
      "grad_norm": 0.21179281175136566,
      "learning_rate": 1.0330578512396695e-06,
      "loss": 1.676,
      "step": 217
    },
    {
      "epoch": 0.9008264462809917,
      "grad_norm": 0.2215491384267807,
      "learning_rate": 9.917355371900827e-07,
      "loss": 1.5923,
      "step": 218
    },
    {
      "epoch": 0.9049586776859504,
      "grad_norm": 0.1991722285747528,
      "learning_rate": 9.50413223140496e-07,
      "loss": 1.6321,
      "step": 219
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.20081669092178345,
      "learning_rate": 9.090909090909091e-07,
      "loss": 1.6221,
      "step": 220
    },
    {
      "epoch": 0.9132231404958677,
      "grad_norm": 0.2524271011352539,
      "learning_rate": 8.677685950413224e-07,
      "loss": 1.636,
      "step": 221
    },
    {
      "epoch": 0.9173553719008265,
      "grad_norm": 0.1950342357158661,
      "learning_rate": 8.264462809917356e-07,
      "loss": 1.6524,
      "step": 222
    },
    {
      "epoch": 0.9214876033057852,
      "grad_norm": 0.20284403860569,
      "learning_rate": 7.851239669421488e-07,
      "loss": 1.6317,
      "step": 223
    },
    {
      "epoch": 0.9256198347107438,
      "grad_norm": 0.25988680124282837,
      "learning_rate": 7.438016528925621e-07,
      "loss": 1.7327,
      "step": 224
    },
    {
      "epoch": 0.9297520661157025,
      "grad_norm": 0.27988940477371216,
      "learning_rate": 7.024793388429753e-07,
      "loss": 1.7567,
      "step": 225
    },
    {
      "epoch": 0.9338842975206612,
      "grad_norm": 0.20009325444698334,
      "learning_rate": 6.611570247933885e-07,
      "loss": 1.6607,
      "step": 226
    },
    {
      "epoch": 0.9380165289256198,
      "grad_norm": 0.22822903096675873,
      "learning_rate": 6.198347107438018e-07,
      "loss": 1.7284,
      "step": 227
    },
    {
      "epoch": 0.9421487603305785,
      "grad_norm": 0.19419339299201965,
      "learning_rate": 5.78512396694215e-07,
      "loss": 1.5976,
      "step": 228
    },
    {
      "epoch": 0.9462809917355371,
      "grad_norm": 0.22927206754684448,
      "learning_rate": 5.371900826446281e-07,
      "loss": 1.7503,
      "step": 229
    },
    {
      "epoch": 0.9504132231404959,
      "grad_norm": 0.19419941306114197,
      "learning_rate": 4.958677685950413e-07,
      "loss": 1.6068,
      "step": 230
    },
    {
      "epoch": 0.9545454545454546,
      "grad_norm": 0.22186236083507538,
      "learning_rate": 4.5454545454545457e-07,
      "loss": 1.6539,
      "step": 231
    },
    {
      "epoch": 0.9586776859504132,
      "grad_norm": 0.2566603422164917,
      "learning_rate": 4.132231404958678e-07,
      "loss": 1.6298,
      "step": 232
    },
    {
      "epoch": 0.9628099173553719,
      "grad_norm": 0.19301608204841614,
      "learning_rate": 3.7190082644628103e-07,
      "loss": 1.6212,
      "step": 233
    },
    {
      "epoch": 0.9669421487603306,
      "grad_norm": 0.2026912420988083,
      "learning_rate": 3.3057851239669426e-07,
      "loss": 1.6505,
      "step": 234
    },
    {
      "epoch": 0.9710743801652892,
      "grad_norm": 0.2571674883365631,
      "learning_rate": 2.892561983471075e-07,
      "loss": 1.6299,
      "step": 235
    },
    {
      "epoch": 0.9752066115702479,
      "grad_norm": 0.22160300612449646,
      "learning_rate": 2.4793388429752067e-07,
      "loss": 1.7306,
      "step": 236
    },
    {
      "epoch": 0.9793388429752066,
      "grad_norm": 0.19275052845478058,
      "learning_rate": 2.066115702479339e-07,
      "loss": 1.5956,
      "step": 237
    },
    {
      "epoch": 0.9834710743801653,
      "grad_norm": 0.23581215739250183,
      "learning_rate": 1.6528925619834713e-07,
      "loss": 1.5395,
      "step": 238
    },
    {
      "epoch": 0.987603305785124,
      "grad_norm": 0.22948598861694336,
      "learning_rate": 1.2396694214876034e-07,
      "loss": 1.5879,
      "step": 239
    },
    {
      "epoch": 0.9917355371900827,
      "grad_norm": 0.18640126287937164,
      "learning_rate": 8.264462809917357e-08,
      "loss": 1.6289,
      "step": 240
    },
    {
      "epoch": 0.9958677685950413,
      "grad_norm": 0.23165103793144226,
      "learning_rate": 4.132231404958678e-08,
      "loss": 1.6027,
      "step": 241
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.24915795028209686,
      "learning_rate": 0.0,
      "loss": 1.6821,
      "step": 242
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 242,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.934135632920576e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}