{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 380,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005263157894736842,
      "grad_norm": 12.642855011431948,
      "learning_rate": 5.2631578947368416e-08,
      "loss": 1.0803,
      "step": 1
    },
    {
      "epoch": 0.010526315789473684,
      "grad_norm": 12.34425414294974,
      "learning_rate": 1.0526315789473683e-07,
      "loss": 1.0462,
      "step": 2
    },
    {
      "epoch": 0.015789473684210527,
      "grad_norm": 11.451279866893907,
      "learning_rate": 1.5789473684210525e-07,
      "loss": 1.0479,
      "step": 3
    },
    {
      "epoch": 0.021052631578947368,
      "grad_norm": 12.195321064233118,
      "learning_rate": 2.1052631578947366e-07,
      "loss": 1.0667,
      "step": 4
    },
    {
      "epoch": 0.02631578947368421,
      "grad_norm": 11.732705675313184,
      "learning_rate": 2.631578947368421e-07,
      "loss": 1.0227,
      "step": 5
    },
    {
      "epoch": 0.031578947368421054,
      "grad_norm": 11.450018010840793,
      "learning_rate": 3.157894736842105e-07,
      "loss": 1.0279,
      "step": 6
    },
    {
      "epoch": 0.03684210526315789,
      "grad_norm": 11.469564671793638,
      "learning_rate": 3.684210526315789e-07,
      "loss": 1.0395,
      "step": 7
    },
    {
      "epoch": 0.042105263157894736,
      "grad_norm": 10.28912569661124,
      "learning_rate": 4.2105263157894733e-07,
      "loss": 1.0004,
      "step": 8
    },
    {
      "epoch": 0.04736842105263158,
      "grad_norm": 10.064429251308997,
      "learning_rate": 4.7368421052631574e-07,
      "loss": 1.0354,
      "step": 9
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": 11.11360392847431,
      "learning_rate": 5.263157894736842e-07,
      "loss": 1.0282,
      "step": 10
    },
    {
      "epoch": 0.05789473684210526,
      "grad_norm": 8.842870004697499,
      "learning_rate": 5.789473684210526e-07,
      "loss": 0.9573,
      "step": 11
    },
    {
      "epoch": 0.06315789473684211,
      "grad_norm": 9.419526776304732,
      "learning_rate": 6.31578947368421e-07,
      "loss": 1.0123,
      "step": 12
    },
    {
      "epoch": 0.06842105263157895,
      "grad_norm": 8.531235500445332,
      "learning_rate": 6.842105263157895e-07,
      "loss": 0.9723,
      "step": 13
    },
    {
      "epoch": 0.07368421052631578,
      "grad_norm": 7.612142885708635,
      "learning_rate": 7.368421052631578e-07,
      "loss": 0.9488,
      "step": 14
    },
    {
      "epoch": 0.07894736842105263,
      "grad_norm": 7.243400548958967,
      "learning_rate": 7.894736842105263e-07,
      "loss": 0.9566,
      "step": 15
    },
    {
      "epoch": 0.08421052631578947,
      "grad_norm": 7.715704166499068,
      "learning_rate": 8.421052631578947e-07,
      "loss": 0.9586,
      "step": 16
    },
    {
      "epoch": 0.08947368421052632,
      "grad_norm": 5.681343828224437,
      "learning_rate": 8.947368421052631e-07,
      "loss": 0.8972,
      "step": 17
    },
    {
      "epoch": 0.09473684210526316,
      "grad_norm": 5.909330321398334,
      "learning_rate": 9.473684210526315e-07,
      "loss": 0.9462,
      "step": 18
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.551955653512476,
      "learning_rate": 1e-06,
      "loss": 0.9716,
      "step": 19
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 3.9884175482678628,
      "learning_rate": 9.999810668616084e-07,
      "loss": 0.8733,
      "step": 20
    },
    {
      "epoch": 0.11052631578947368,
      "grad_norm": 4.3500944022312495,
      "learning_rate": 9.999242688802884e-07,
      "loss": 0.8819,
      "step": 21
    },
    {
      "epoch": 0.11578947368421053,
      "grad_norm": 4.612977019312943,
      "learning_rate": 9.998296103574966e-07,
      "loss": 0.8117,
      "step": 22
    },
    {
      "epoch": 0.12105263157894737,
      "grad_norm": 4.085876398056263,
      "learning_rate": 9.996970984619641e-07,
      "loss": 0.8479,
      "step": 23
    },
    {
      "epoch": 0.12631578947368421,
      "grad_norm": 3.8962745222123605,
      "learning_rate": 9.995267432291555e-07,
      "loss": 0.7967,
      "step": 24
    },
    {
      "epoch": 0.13157894736842105,
      "grad_norm": 4.906007211253714,
      "learning_rate": 9.993185575605073e-07,
      "loss": 0.846,
      "step": 25
    },
    {
      "epoch": 0.1368421052631579,
      "grad_norm": 4.076765235980574,
      "learning_rate": 9.99072557222452e-07,
      "loss": 0.8361,
      "step": 26
    },
    {
      "epoch": 0.14210526315789473,
      "grad_norm": 3.6469407214417875,
      "learning_rate": 9.987887608452234e-07,
      "loss": 0.8136,
      "step": 27
    },
    {
      "epoch": 0.14736842105263157,
      "grad_norm": 4.297874948498594,
      "learning_rate": 9.984671899214456e-07,
      "loss": 0.773,
      "step": 28
    },
    {
      "epoch": 0.15263157894736842,
      "grad_norm": 3.248821707316575,
      "learning_rate": 9.981078688045062e-07,
      "loss": 0.7295,
      "step": 29
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 3.2415876192930315,
      "learning_rate": 9.977108247067108e-07,
      "loss": 0.7677,
      "step": 30
    },
    {
      "epoch": 0.1631578947368421,
      "grad_norm": 3.63042098210306,
      "learning_rate": 9.972760876972224e-07,
      "loss": 0.7725,
      "step": 31
    },
    {
      "epoch": 0.16842105263157894,
      "grad_norm": 3.2114667781734085,
      "learning_rate": 9.968036906997853e-07,
      "loss": 0.738,
      "step": 32
    },
    {
      "epoch": 0.1736842105263158,
      "grad_norm": 3.022035623636562,
      "learning_rate": 9.962936694902306e-07,
      "loss": 0.7384,
      "step": 33
    },
    {
      "epoch": 0.17894736842105263,
      "grad_norm": 2.9523633250041854,
      "learning_rate": 9.957460626937662e-07,
      "loss": 0.7509,
      "step": 34
    },
    {
      "epoch": 0.18421052631578946,
      "grad_norm": 3.080112598494873,
      "learning_rate": 9.951609117820537e-07,
      "loss": 0.7605,
      "step": 35
    },
    {
      "epoch": 0.18947368421052632,
      "grad_norm": 4.039688332869512,
      "learning_rate": 9.945382610700657e-07,
      "loss": 0.7369,
      "step": 36
    },
    {
      "epoch": 0.19473684210526315,
      "grad_norm": 2.8312131444973034,
      "learning_rate": 9.938781577127306e-07,
      "loss": 0.7395,
      "step": 37
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.114482461656541,
      "learning_rate": 9.931806517013612e-07,
      "loss": 0.7159,
      "step": 38
    },
    {
      "epoch": 0.20526315789473684,
      "grad_norm": 2.7086708072786942,
      "learning_rate": 9.92445795859869e-07,
      "loss": 0.7222,
      "step": 39
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 3.7485449995739937,
      "learning_rate": 9.91673645840763e-07,
      "loss": 0.6949,
      "step": 40
    },
    {
      "epoch": 0.21578947368421053,
      "grad_norm": 2.8503331290525566,
      "learning_rate": 9.908642601209365e-07,
      "loss": 0.7092,
      "step": 41
    },
    {
      "epoch": 0.22105263157894736,
      "grad_norm": 3.079001602780246,
      "learning_rate": 9.900176999972364e-07,
      "loss": 0.7088,
      "step": 42
    },
    {
      "epoch": 0.22631578947368422,
      "grad_norm": 2.564364372698868,
      "learning_rate": 9.89134029581823e-07,
      "loss": 0.7025,
      "step": 43
    },
    {
      "epoch": 0.23157894736842105,
      "grad_norm": 3.1544438320150854,
      "learning_rate": 9.88213315797313e-07,
      "loss": 0.6982,
      "step": 44
    },
    {
      "epoch": 0.23684210526315788,
      "grad_norm": 3.3994379889515405,
      "learning_rate": 9.872556283717124e-07,
      "loss": 0.7051,
      "step": 45
    },
    {
      "epoch": 0.24210526315789474,
      "grad_norm": 2.7042300223916067,
      "learning_rate": 9.862610398331359e-07,
      "loss": 0.7234,
      "step": 46
    },
    {
      "epoch": 0.24736842105263157,
      "grad_norm": 2.6577071952641584,
      "learning_rate": 9.85229625504313e-07,
      "loss": 0.7011,
      "step": 47
    },
    {
      "epoch": 0.25263157894736843,
      "grad_norm": 3.050166736127533,
      "learning_rate": 9.841614634968843e-07,
      "loss": 0.7075,
      "step": 48
    },
    {
      "epoch": 0.2578947368421053,
      "grad_norm": 6.357403841745397,
      "learning_rate": 9.830566347054867e-07,
      "loss": 0.7151,
      "step": 49
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 2.545588371195535,
      "learning_rate": 9.819152228016257e-07,
      "loss": 0.6481,
      "step": 50
    },
    {
      "epoch": 0.26842105263157895,
      "grad_norm": 2.5786439940280035,
      "learning_rate": 9.807373142273394e-07,
      "loss": 0.6587,
      "step": 51
    },
    {
      "epoch": 0.2736842105263158,
      "grad_norm": 3.13815883329848,
      "learning_rate": 9.795229981886521e-07,
      "loss": 0.646,
      "step": 52
    },
    {
      "epoch": 0.2789473684210526,
      "grad_norm": 2.773920821177136,
      "learning_rate": 9.782723666488179e-07,
      "loss": 0.6955,
      "step": 53
    },
    {
      "epoch": 0.28421052631578947,
      "grad_norm": 2.617357302802378,
      "learning_rate": 9.769855143213574e-07,
      "loss": 0.6312,
      "step": 54
    },
    {
      "epoch": 0.2894736842105263,
      "grad_norm": 2.8145131931082252,
      "learning_rate": 9.756625386628832e-07,
      "loss": 0.6193,
      "step": 55
    },
    {
      "epoch": 0.29473684210526313,
      "grad_norm": 2.4833405259734307,
      "learning_rate": 9.7430353986572e-07,
      "loss": 0.6439,
      "step": 56
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5453834053616355,
      "learning_rate": 9.729086208503173e-07,
      "loss": 0.6436,
      "step": 57
    },
    {
      "epoch": 0.30526315789473685,
      "grad_norm": 2.753486017153059,
      "learning_rate": 9.71477887257454e-07,
      "loss": 0.6903,
      "step": 58
    },
    {
      "epoch": 0.3105263157894737,
      "grad_norm": 3.0871711482073767,
      "learning_rate": 9.700114474402388e-07,
      "loss": 0.6612,
      "step": 59
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 2.903286008491912,
      "learning_rate": 9.685094124559032e-07,
      "loss": 0.6685,
      "step": 60
    },
    {
      "epoch": 0.32105263157894737,
      "grad_norm": 2.6328385667684246,
      "learning_rate": 9.669718960573925e-07,
      "loss": 0.6562,
      "step": 61
    },
    {
      "epoch": 0.3263157894736842,
      "grad_norm": 2.4572855900723307,
      "learning_rate": 9.653990146847498e-07,
      "loss": 0.6747,
      "step": 62
    },
    {
      "epoch": 0.33157894736842103,
      "grad_norm": 3.2309917558351215,
      "learning_rate": 9.637908874562976e-07,
      "loss": 0.6727,
      "step": 63
    },
    {
      "epoch": 0.3368421052631579,
      "grad_norm": 2.6952262267407208,
      "learning_rate": 9.621476361596177e-07,
      "loss": 0.6816,
      "step": 64
    },
    {
      "epoch": 0.34210526315789475,
      "grad_norm": 2.4940546746298593,
      "learning_rate": 9.604693852423268e-07,
      "loss": 0.6452,
      "step": 65
    },
    {
      "epoch": 0.3473684210526316,
      "grad_norm": 3.7336842467494984,
      "learning_rate": 9.587562618026521e-07,
      "loss": 0.6358,
      "step": 66
    },
    {
      "epoch": 0.3526315789473684,
      "grad_norm": 2.4509664199477093,
      "learning_rate": 9.570083955798065e-07,
      "loss": 0.6626,
      "step": 67
    },
    {
      "epoch": 0.35789473684210527,
      "grad_norm": 2.91731520990893,
      "learning_rate": 9.552259189441624e-07,
      "loss": 0.6769,
      "step": 68
    },
    {
      "epoch": 0.3631578947368421,
      "grad_norm": 3.0789821120953365,
      "learning_rate": 9.534089668872273e-07,
      "loss": 0.6847,
      "step": 69
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 2.6786175238418726,
      "learning_rate": 9.515576770114197e-07,
      "loss": 0.655,
      "step": 70
    },
    {
      "epoch": 0.3736842105263158,
      "grad_norm": 2.634227056004132,
      "learning_rate": 9.496721895196495e-07,
      "loss": 0.7022,
      "step": 71
    },
    {
      "epoch": 0.37894736842105264,
      "grad_norm": 2.664437320388568,
      "learning_rate": 9.477526472046994e-07,
      "loss": 0.6619,
      "step": 72
    },
    {
      "epoch": 0.38421052631578945,
      "grad_norm": 4.512374130100286,
      "learning_rate": 9.457991954384103e-07,
      "loss": 0.667,
      "step": 73
    },
    {
      "epoch": 0.3894736842105263,
      "grad_norm": 2.8480285570523036,
      "learning_rate": 9.438119821606727e-07,
      "loss": 0.638,
      "step": 74
    },
    {
      "epoch": 0.39473684210526316,
      "grad_norm": 2.2934398392098574,
      "learning_rate": 9.417911578682228e-07,
      "loss": 0.6324,
      "step": 75
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2872153423968293,
      "learning_rate": 9.397368756032444e-07,
      "loss": 0.6718,
      "step": 76
    },
    {
      "epoch": 0.4052631578947368,
      "grad_norm": 2.6631337006486757,
      "learning_rate": 9.376492909417795e-07,
      "loss": 0.6606,
      "step": 77
    },
    {
      "epoch": 0.4105263157894737,
      "grad_norm": 2.5268933861039717,
      "learning_rate": 9.355285619819449e-07,
      "loss": 0.6477,
      "step": 78
    },
    {
      "epoch": 0.41578947368421054,
      "grad_norm": 2.492744860536843,
      "learning_rate": 9.333748493319602e-07,
      "loss": 0.6159,
      "step": 79
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 6.620410049291206,
      "learning_rate": 9.311883160979843e-07,
      "loss": 0.6229,
      "step": 80
    },
    {
      "epoch": 0.4263157894736842,
      "grad_norm": 2.8596307639279375,
      "learning_rate": 9.289691278717622e-07,
      "loss": 0.6608,
      "step": 81
    },
    {
      "epoch": 0.43157894736842106,
      "grad_norm": 2.2817426902775564,
      "learning_rate": 9.267174527180854e-07,
      "loss": 0.6233,
      "step": 82
    },
    {
      "epoch": 0.4368421052631579,
      "grad_norm": 2.338517201348536,
      "learning_rate": 9.244334611620627e-07,
      "loss": 0.6434,
      "step": 83
    },
    {
      "epoch": 0.4421052631578947,
      "grad_norm": 2.229568223588675,
      "learning_rate": 9.221173261762073e-07,
      "loss": 0.645,
      "step": 84
    },
    {
      "epoch": 0.4473684210526316,
      "grad_norm": 3.2473078589534654,
      "learning_rate": 9.197692231673359e-07,
      "loss": 0.6591,
      "step": 85
    },
    {
      "epoch": 0.45263157894736844,
      "grad_norm": 2.4771191451211605,
      "learning_rate": 9.173893299632855e-07,
      "loss": 0.623,
      "step": 86
    },
    {
      "epoch": 0.45789473684210524,
      "grad_norm": 2.7158915691477565,
      "learning_rate": 9.149778267994456e-07,
      "loss": 0.6423,
      "step": 87
    },
    {
      "epoch": 0.4631578947368421,
      "grad_norm": 2.6330534202405875,
      "learning_rate": 9.125348963051089e-07,
      "loss": 0.6598,
      "step": 88
    },
    {
      "epoch": 0.46842105263157896,
      "grad_norm": 2.524476758159517,
      "learning_rate": 9.100607234896396e-07,
      "loss": 0.6308,
      "step": 89
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 2.501226879638191,
      "learning_rate": 9.075554957284633e-07,
      "loss": 0.6393,
      "step": 90
    },
    {
      "epoch": 0.4789473684210526,
      "grad_norm": 3.933826469332694,
      "learning_rate": 9.050194027488754e-07,
      "loss": 0.643,
      "step": 91
    },
    {
      "epoch": 0.4842105263157895,
      "grad_norm": 2.494431369881789,
      "learning_rate": 9.024526366156732e-07,
      "loss": 0.6324,
      "step": 92
    },
    {
      "epoch": 0.48947368421052634,
      "grad_norm": 2.5647183175292976,
      "learning_rate": 8.998553917166108e-07,
      "loss": 0.6678,
      "step": 93
    },
    {
      "epoch": 0.49473684210526314,
      "grad_norm": 2.7418107425427403,
      "learning_rate": 8.972278647476763e-07,
      "loss": 0.6401,
      "step": 94
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.4908001630523504,
      "learning_rate": 8.945702546981968e-07,
      "loss": 0.6376,
      "step": 95
    },
    {
      "epoch": 0.5052631578947369,
      "grad_norm": 2.3820979238811097,
      "learning_rate": 8.918827628357677e-07,
      "loss": 0.6604,
      "step": 96
    },
    {
      "epoch": 0.5105263157894737,
      "grad_norm": 6.555925708049527,
      "learning_rate": 8.891655926910102e-07,
      "loss": 0.602,
      "step": 97
    },
    {
      "epoch": 0.5157894736842106,
      "grad_norm": 2.767476794957958,
      "learning_rate": 8.864189500421581e-07,
      "loss": 0.6351,
      "step": 98
    },
    {
      "epoch": 0.5210526315789473,
      "grad_norm": 2.5165154253276083,
      "learning_rate": 8.836430428994731e-07,
      "loss": 0.6385,
      "step": 99
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 2.3143831343939936,
      "learning_rate": 8.808380814894911e-07,
      "loss": 0.6369,
      "step": 100
    },
    {
      "epoch": 0.531578947368421,
      "grad_norm": 2.2866922921441972,
      "learning_rate": 8.780042782391027e-07,
      "loss": 0.6296,
      "step": 101
    },
    {
      "epoch": 0.5368421052631579,
      "grad_norm": 3.5036243819126316,
      "learning_rate": 8.751418477594643e-07,
      "loss": 0.6345,
      "step": 102
    },
    {
      "epoch": 0.5421052631578948,
      "grad_norm": 2.4647712993984623,
      "learning_rate": 8.722510068297453e-07,
      "loss": 0.6015,
      "step": 103
    },
    {
      "epoch": 0.5473684210526316,
      "grad_norm": 2.6893219919656612,
      "learning_rate": 8.693319743807115e-07,
      "loss": 0.6523,
      "step": 104
    },
    {
      "epoch": 0.5526315789473685,
      "grad_norm": 2.520295568093033,
      "learning_rate": 8.663849714781442e-07,
      "loss": 0.6275,
      "step": 105
    },
    {
      "epoch": 0.5578947368421052,
      "grad_norm": 2.5102646596872367,
      "learning_rate": 8.634102213060983e-07,
      "loss": 0.6235,
      "step": 106
    },
    {
      "epoch": 0.5631578947368421,
      "grad_norm": 2.420532153270969,
      "learning_rate": 8.604079491500009e-07,
      "loss": 0.6415,
      "step": 107
    },
    {
      "epoch": 0.5684210526315789,
      "grad_norm": 2.419586650925291,
      "learning_rate": 8.573783823795888e-07,
      "loss": 0.6083,
      "step": 108
    },
    {
      "epoch": 0.5736842105263158,
      "grad_norm": 2.4954004438078017,
      "learning_rate": 8.543217504316895e-07,
      "loss": 0.6043,
      "step": 109
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 2.631207821232483,
      "learning_rate": 8.512382847928461e-07,
      "loss": 0.5966,
      "step": 110
    },
    {
      "epoch": 0.5842105263157895,
      "grad_norm": 2.4487969119631607,
      "learning_rate": 8.48128218981785e-07,
      "loss": 0.6207,
      "step": 111
    },
    {
      "epoch": 0.5894736842105263,
      "grad_norm": 2.4819981729995795,
      "learning_rate": 8.449917885317319e-07,
      "loss": 0.6043,
      "step": 112
    },
    {
      "epoch": 0.5947368421052631,
      "grad_norm": 3.0806875014131254,
      "learning_rate": 8.418292309725738e-07,
      "loss": 0.6393,
      "step": 113
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.8047374095987743,
      "learning_rate": 8.386407858128706e-07,
      "loss": 0.6162,
      "step": 114
    },
    {
      "epoch": 0.6052631578947368,
      "grad_norm": 2.2653217647871426,
      "learning_rate": 8.354266945217159e-07,
      "loss": 0.6422,
      "step": 115
    },
    {
      "epoch": 0.6105263157894737,
      "grad_norm": 2.387247632448718,
      "learning_rate": 8.321872005104508e-07,
      "loss": 0.6256,
      "step": 116
    },
    {
      "epoch": 0.6157894736842106,
      "grad_norm": 2.70280163207189,
      "learning_rate": 8.289225491142291e-07,
      "loss": 0.6206,
      "step": 117
    },
    {
      "epoch": 0.6210526315789474,
      "grad_norm": 7.377993411312978,
      "learning_rate": 8.256329875734374e-07,
      "loss": 0.6425,
      "step": 118
    },
    {
      "epoch": 0.6263157894736842,
      "grad_norm": 2.3380220919877988,
      "learning_rate": 8.223187650149711e-07,
      "loss": 0.6116,
      "step": 119
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 3.323728626512576,
      "learning_rate": 8.18980132433368e-07,
      "loss": 0.608,
      "step": 120
    },
    {
      "epoch": 0.6368421052631579,
      "grad_norm": 3.1749966640550946,
      "learning_rate": 8.156173426717988e-07,
      "loss": 0.5815,
      "step": 121
    },
    {
      "epoch": 0.6421052631578947,
      "grad_norm": 7.107357659646092,
      "learning_rate": 8.122306504029193e-07,
      "loss": 0.597,
      "step": 122
    },
    {
      "epoch": 0.6473684210526316,
      "grad_norm": 4.479509224501224,
      "learning_rate": 8.088203121095829e-07,
      "loss": 0.5931,
      "step": 123
    },
    {
      "epoch": 0.6526315789473685,
      "grad_norm": 2.709896735413064,
      "learning_rate": 8.053865860654174e-07,
      "loss": 0.6132,
      "step": 124
    },
    {
      "epoch": 0.6578947368421053,
      "grad_norm": 3.1553051157828085,
      "learning_rate": 8.019297323152641e-07,
      "loss": 0.6462,
      "step": 125
    },
    {
      "epoch": 0.6631578947368421,
      "grad_norm": 2.6077332270949025,
      "learning_rate": 7.984500126554851e-07,
      "loss": 0.6412,
      "step": 126
    },
    {
      "epoch": 0.6684210526315789,
      "grad_norm": 2.687935639196366,
      "learning_rate": 7.949476906141359e-07,
      "loss": 0.6271,
      "step": 127
    },
    {
      "epoch": 0.6736842105263158,
      "grad_norm": 2.48251997551827,
      "learning_rate": 7.914230314310077e-07,
      "loss": 0.6027,
      "step": 128
    },
    {
      "epoch": 0.6789473684210526,
      "grad_norm": 2.9438977642779403,
      "learning_rate": 7.878763020375414e-07,
      "loss": 0.6412,
      "step": 129
    },
    {
      "epoch": 0.6842105263157895,
      "grad_norm": 4.032885299122092,
      "learning_rate": 7.843077710366104e-07,
      "loss": 0.6592,
      "step": 130
    },
    {
      "epoch": 0.6894736842105263,
      "grad_norm": 2.7903181652647406,
      "learning_rate": 7.807177086821801e-07,
      "loss": 0.6248,
      "step": 131
    },
    {
      "epoch": 0.6947368421052632,
      "grad_norm": 2.968921847649656,
      "learning_rate": 7.771063868588399e-07,
      "loss": 0.6231,
      "step": 132
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3977400854216486,
      "learning_rate": 7.734740790612136e-07,
      "loss": 0.6228,
      "step": 133
    },
    {
      "epoch": 0.7052631578947368,
      "grad_norm": 2.3140577613757425,
      "learning_rate": 7.698210603732454e-07,
      "loss": 0.5791,
      "step": 134
    },
    {
      "epoch": 0.7105263157894737,
      "grad_norm": 2.523785887846981,
      "learning_rate": 7.661476074473694e-07,
      "loss": 0.5997,
      "step": 135
    },
    {
      "epoch": 0.7157894736842105,
      "grad_norm": 2.7229089664651585,
      "learning_rate": 7.624539984835556e-07,
      "loss": 0.5994,
      "step": 136
    },
    {
      "epoch": 0.7210526315789474,
      "grad_norm": 2.4305539942752548,
      "learning_rate": 7.587405132082432e-07,
      "loss": 0.5941,
      "step": 137
    },
    {
      "epoch": 0.7263157894736842,
      "grad_norm": 2.253433002879369,
      "learning_rate": 7.550074328531544e-07,
      "loss": 0.6323,
      "step": 138
    },
    {
      "epoch": 0.7315789473684211,
      "grad_norm": 2.87605074258251,
      "learning_rate": 7.512550401339971e-07,
      "loss": 0.605,
      "step": 139
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 2.4820171983737964,
      "learning_rate": 7.47483619229054e-07,
      "loss": 0.6038,
      "step": 140
    },
    {
      "epoch": 0.7421052631578947,
      "grad_norm": 2.994909930351913,
      "learning_rate": 7.436934557576611e-07,
      "loss": 0.5898,
      "step": 141
    },
    {
      "epoch": 0.7473684210526316,
      "grad_norm": 2.4089107961391716,
      "learning_rate": 7.39884836758576e-07,
      "loss": 0.6191,
      "step": 142
    },
    {
      "epoch": 0.7526315789473684,
      "grad_norm": 2.484528784332179,
      "learning_rate": 7.360580506682413e-07,
      "loss": 0.593,
      "step": 143
    },
    {
      "epoch": 0.7578947368421053,
      "grad_norm": 2.526886015112745,
      "learning_rate": 7.322133872989398e-07,
      "loss": 0.6036,
      "step": 144
    },
    {
      "epoch": 0.7631578947368421,
      "grad_norm": 3.1068042639656226,
      "learning_rate": 7.283511378168457e-07,
      "loss": 0.5986,
      "step": 145
    },
    {
      "epoch": 0.7684210526315789,
      "grad_norm": 2.426127834823539,
      "learning_rate": 7.244715947199749e-07,
      "loss": 0.6097,
      "step": 146
    },
    {
      "epoch": 0.7736842105263158,
      "grad_norm": 2.3123421734560115,
      "learning_rate": 7.20575051816033e-07,
      "loss": 0.5765,
      "step": 147
    },
    {
      "epoch": 0.7789473684210526,
      "grad_norm": 3.0322275767396927,
      "learning_rate": 7.166618042001639e-07,
      "loss": 0.6209,
      "step": 148
    },
    {
      "epoch": 0.7842105263157895,
      "grad_norm": 2.0885878316769113,
      "learning_rate": 7.127321482326026e-07,
      "loss": 0.6038,
      "step": 149
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 2.6655995523304443,
      "learning_rate": 7.087863815162299e-07,
      "loss": 0.6039,
      "step": 150
    },
    {
      "epoch": 0.7947368421052632,
      "grad_norm": 3.5462821286242603,
      "learning_rate": 7.048248028740349e-07,
      "loss": 0.5951,
      "step": 151
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.471112364580461,
      "learning_rate": 7.008477123264847e-07,
      "loss": 0.5965,
      "step": 152
    },
    {
      "epoch": 0.8052631578947368,
      "grad_norm": 2.5061966397777162,
      "learning_rate": 6.968554110688019e-07,
      "loss": 0.5871,
      "step": 153
    },
    {
      "epoch": 0.8105263157894737,
      "grad_norm": 2.406700451452237,
      "learning_rate": 6.928482014481558e-07,
      "loss": 0.6105,
      "step": 154
    },
    {
      "epoch": 0.8157894736842105,
      "grad_norm": 3.1910017405823705,
      "learning_rate": 6.888263869407632e-07,
      "loss": 0.587,
      "step": 155
    },
    {
      "epoch": 0.8210526315789474,
      "grad_norm": 2.467282283935378,
      "learning_rate": 6.847902721289067e-07,
      "loss": 0.6227,
      "step": 156
    },
    {
      "epoch": 0.8263157894736842,
      "grad_norm": 2.295127372658213,
      "learning_rate": 6.807401626778679e-07,
      "loss": 0.5899,
      "step": 157
    },
    {
      "epoch": 0.8315789473684211,
      "grad_norm": 2.6433009830471406,
      "learning_rate": 6.766763653127772e-07,
      "loss": 0.5892,
      "step": 158
    },
    {
      "epoch": 0.8368421052631579,
      "grad_norm": 3.0749394453309526,
      "learning_rate": 6.725991877953867e-07,
      "loss": 0.6041,
      "step": 159
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 2.345915225733439,
      "learning_rate": 6.68508938900761e-07,
      "loss": 0.586,
      "step": 160
    },
    {
      "epoch": 0.8473684210526315,
      "grad_norm": 3.0520257716599097,
      "learning_rate": 6.644059283938937e-07,
      "loss": 0.6134,
      "step": 161
    },
    {
      "epoch": 0.8526315789473684,
      "grad_norm": 2.747377304181717,
      "learning_rate": 6.602904670062475e-07,
      "loss": 0.5633,
      "step": 162
    },
    {
      "epoch": 0.8578947368421053,
      "grad_norm": 2.179146773849276,
      "learning_rate": 6.561628664122226e-07,
      "loss": 0.5706,
      "step": 163
    },
    {
      "epoch": 0.8631578947368421,
      "grad_norm": 2.3135324153171495,
      "learning_rate": 6.520234392055521e-07,
      "loss": 0.6167,
      "step": 164
    },
    {
      "epoch": 0.868421052631579,
      "grad_norm": 9.71533571058622,
      "learning_rate": 6.478724988756284e-07,
      "loss": 0.6069,
      "step": 165
    },
    {
      "epoch": 0.8736842105263158,
      "grad_norm": 2.4920221204476554,
      "learning_rate": 6.437103597837629e-07,
      "loss": 0.5963,
      "step": 166
    },
    {
      "epoch": 0.8789473684210526,
      "grad_norm": 3.0329738688798145,
      "learning_rate": 6.395373371393769e-07,
      "loss": 0.5822,
      "step": 167
    },
    {
      "epoch": 0.8842105263157894,
      "grad_norm": 2.6956141149357973,
      "learning_rate": 6.353537469761315e-07,
      "loss": 0.5937,
      "step": 168
    },
    {
      "epoch": 0.8894736842105263,
      "grad_norm": 2.5218470185601998,
      "learning_rate": 6.311599061279931e-07,
      "loss": 0.6025,
      "step": 169
    },
    {
      "epoch": 0.8947368421052632,
      "grad_norm": 2.384640301104755,
      "learning_rate": 6.269561322052377e-07,
      "loss": 0.6249,
      "step": 170
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.5020602687904634,
      "learning_rate": 6.227427435703995e-07,
      "loss": 0.5954,
      "step": 171
    },
    {
      "epoch": 0.9052631578947369,
      "grad_norm": 2.5824028414049014,
      "learning_rate": 6.185200593141591e-07,
      "loss": 0.6262,
      "step": 172
    },
    {
      "epoch": 0.9105263157894737,
      "grad_norm": 3.3671575875344177,
      "learning_rate": 6.14288399231178e-07,
      "loss": 0.5806,
      "step": 173
    },
    {
      "epoch": 0.9157894736842105,
      "grad_norm": 2.8730810418787165,
      "learning_rate": 6.100480837958801e-07,
      "loss": 0.5756,
      "step": 174
    },
    {
      "epoch": 0.9210526315789473,
      "grad_norm": 2.4545013421309414,
      "learning_rate": 6.057994341381812e-07,
      "loss": 0.5878,
      "step": 175
    },
    {
      "epoch": 0.9263157894736842,
      "grad_norm": 2.3488962229180923,
      "learning_rate": 6.015427720191692e-07,
      "loss": 0.5887,
      "step": 176
    },
    {
      "epoch": 0.9315789473684211,
      "grad_norm": 2.4164866608841695,
      "learning_rate": 5.97278419806736e-07,
      "loss": 0.5847,
      "step": 177
    },
    {
      "epoch": 0.9368421052631579,
      "grad_norm": 2.536723291986412,
      "learning_rate": 5.93006700451164e-07,
      "loss": 0.6001,
      "step": 178
    },
    {
      "epoch": 0.9421052631578948,
      "grad_norm": 2.578895932915379,
      "learning_rate": 5.887279374606679e-07,
      "loss": 0.5695,
      "step": 179
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 2.2778696227839865,
      "learning_rate": 5.844424548768951e-07,
      "loss": 0.6,
      "step": 180
    },
    {
      "epoch": 0.9526315789473684,
      "grad_norm": 2.0983119878164005,
      "learning_rate": 5.801505772503853e-07,
      "loss": 0.574,
      "step": 181
    },
    {
      "epoch": 0.9578947368421052,
      "grad_norm": 2.276622095316042,
      "learning_rate": 5.758526296159905e-07,
      "loss": 0.5896,
      "step": 182
    },
    {
      "epoch": 0.9631578947368421,
      "grad_norm": 2.3702885416300936,
      "learning_rate": 5.7154893746826e-07,
      "loss": 0.6187,
      "step": 183
    },
    {
      "epoch": 0.968421052631579,
      "grad_norm": 2.6682239348053076,
      "learning_rate": 5.672398267367901e-07,
      "loss": 0.5975,
      "step": 184
    },
    {
      "epoch": 0.9736842105263158,
      "grad_norm": 2.4953199970370754,
      "learning_rate": 5.629256237615402e-07,
      "loss": 0.5833,
      "step": 185
    },
    {
      "epoch": 0.9789473684210527,
      "grad_norm": 2.3148878554267633,
      "learning_rate": 5.586066552681179e-07,
      "loss": 0.5656,
      "step": 186
    },
    {
      "epoch": 0.9842105263157894,
      "grad_norm": 2.5271611837279138,
      "learning_rate": 5.542832483430363e-07,
      "loss": 0.5645,
      "step": 187
    },
    {
      "epoch": 0.9894736842105263,
      "grad_norm": 2.494396905400437,
      "learning_rate": 5.499557304089418e-07,
      "loss": 0.6034,
      "step": 188
    },
    {
      "epoch": 0.9947368421052631,
      "grad_norm": 2.5191956286524677,
      "learning_rate": 5.456244291998182e-07,
      "loss": 0.6025,
      "step": 189
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3574219439115813,
      "learning_rate": 5.412896727361662e-07,
      "loss": 0.5931,
      "step": 190
    },
    {
      "epoch": 1.0052631578947369,
      "grad_norm": 2.3448910845211666,
      "learning_rate": 5.369517893001619e-07,
      "loss": 0.5247,
      "step": 191
    },
    {
      "epoch": 1.0105263157894737,
      "grad_norm": 3.1508229753956085,
      "learning_rate": 5.326111074107951e-07,
      "loss": 0.5441,
      "step": 192
    },
    {
      "epoch": 1.0157894736842106,
      "grad_norm": 2.4870945738772305,
      "learning_rate": 5.282679557989896e-07,
      "loss": 0.5467,
      "step": 193
    },
    {
      "epoch": 1.0210526315789474,
      "grad_norm": 2.462541528431831,
      "learning_rate": 5.239226633827073e-07,
      "loss": 0.5653,
      "step": 194
    },
    {
      "epoch": 1.0263157894736843,
      "grad_norm": 2.58644706797543,
      "learning_rate": 5.195755592420386e-07,
      "loss": 0.5446,
      "step": 195
    },
    {
      "epoch": 1.0315789473684212,
      "grad_norm": 2.335054761677809,
      "learning_rate": 5.152269725942813e-07,
      "loss": 0.5615,
      "step": 196
    },
    {
      "epoch": 1.0368421052631578,
      "grad_norm": 2.624795958426164,
      "learning_rate": 5.108772327690064e-07,
      "loss": 0.5462,
      "step": 197
    },
    {
      "epoch": 1.0421052631578946,
      "grad_norm": 2.260065923566593,
      "learning_rate": 5.06526669183118e-07,
      "loss": 0.532,
      "step": 198
    },
    {
      "epoch": 1.0473684210526315,
      "grad_norm": 2.2958442656392495,
      "learning_rate": 5.021756113159061e-07,
      "loss": 0.5585,
      "step": 199
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 2.426007034037249,
      "learning_rate": 4.978243886840939e-07,
      "loss": 0.5175,
      "step": 200
    },
    {
      "epoch": 1.0578947368421052,
      "grad_norm": 2.1330228014416734,
      "learning_rate": 4.93473330816882e-07,
      "loss": 0.5483,
      "step": 201
    },
    {
      "epoch": 1.063157894736842,
      "grad_norm": 2.6733289129426225,
      "learning_rate": 4.891227672309935e-07,
      "loss": 0.5351,
      "step": 202
    },
    {
      "epoch": 1.068421052631579,
      "grad_norm": 2.225783335907971,
      "learning_rate": 4.847730274057186e-07,
      "loss": 0.5231,
      "step": 203
    },
    {
      "epoch": 1.0736842105263158,
      "grad_norm": 2.5020405988561065,
      "learning_rate": 4.804244407579613e-07,
      "loss": 0.5527,
      "step": 204
    },
    {
      "epoch": 1.0789473684210527,
      "grad_norm": 2.2536599203492664,
      "learning_rate": 4.7607733661729287e-07,
      "loss": 0.5461,
      "step": 205
    },
    {
      "epoch": 1.0842105263157895,
      "grad_norm": 2.183669632744124,
      "learning_rate": 4.717320442010104e-07,
      "loss": 0.5506,
      "step": 206
    },
    {
      "epoch": 1.0894736842105264,
      "grad_norm": 2.3465016065867492,
      "learning_rate": 4.6738889258920476e-07,
      "loss": 0.5701,
      "step": 207
    },
    {
      "epoch": 1.0947368421052632,
      "grad_norm": 2.496841967109924,
      "learning_rate": 4.630482106998381e-07,
      "loss": 0.559,
      "step": 208
    },
    {
      "epoch": 1.1,
      "grad_norm": 2.238933045944526,
      "learning_rate": 4.5871032726383385e-07,
      "loss": 0.5678,
      "step": 209
    },
    {
      "epoch": 1.1052631578947367,
      "grad_norm": 3.3765686416844725,
      "learning_rate": 4.5437557080018175e-07,
      "loss": 0.575,
      "step": 210
    },
    {
      "epoch": 1.1105263157894736,
      "grad_norm": 2.6001233660513687,
      "learning_rate": 4.500442695910581e-07,
      "loss": 0.5444,
      "step": 211
    },
    {
      "epoch": 1.1157894736842104,
      "grad_norm": 2.4980230677779156,
      "learning_rate": 4.4571675165696364e-07,
      "loss": 0.532,
      "step": 212
    },
    {
      "epoch": 1.1210526315789473,
      "grad_norm": 2.38332156122142,
      "learning_rate": 4.4139334473188206e-07,
      "loss": 0.5275,
      "step": 213
    },
    {
      "epoch": 1.1263157894736842,
      "grad_norm": 2.7714314545935115,
      "learning_rate": 4.3707437623845987e-07,
      "loss": 0.5623,
      "step": 214
    },
    {
      "epoch": 1.131578947368421,
      "grad_norm": 2.314265141849706,
      "learning_rate": 4.327601732632098e-07,
      "loss": 0.5526,
      "step": 215
    },
    {
      "epoch": 1.1368421052631579,
      "grad_norm": 2.345637212860385,
      "learning_rate": 4.2845106253173996e-07,
      "loss": 0.5475,
      "step": 216
    },
    {
      "epoch": 1.1421052631578947,
      "grad_norm": 2.404757712189546,
      "learning_rate": 4.241473703840096e-07,
      "loss": 0.5478,
      "step": 217
    },
    {
      "epoch": 1.1473684210526316,
      "grad_norm": 2.7886042663656676,
      "learning_rate": 4.198494227496147e-07,
      "loss": 0.5449,
      "step": 218
    },
    {
      "epoch": 1.1526315789473685,
      "grad_norm": 2.240431787069853,
      "learning_rate": 4.1555754512310474e-07,
      "loss": 0.5305,
      "step": 219
    },
    {
      "epoch": 1.1578947368421053,
      "grad_norm": 2.275044580303918,
      "learning_rate": 4.112720625393321e-07,
      "loss": 0.555,
      "step": 220
    },
    {
      "epoch": 1.1631578947368422,
      "grad_norm": 2.176470249987148,
      "learning_rate": 4.069932995488361e-07,
      "loss": 0.5342,
      "step": 221
    },
    {
      "epoch": 1.168421052631579,
      "grad_norm": 2.3180546727082874,
      "learning_rate": 4.027215801932641e-07,
      "loss": 0.5545,
      "step": 222
    },
    {
      "epoch": 1.1736842105263159,
      "grad_norm": 2.500478770242888,
      "learning_rate": 3.984572279808306e-07,
      "loss": 0.541,
      "step": 223
    },
    {
      "epoch": 1.1789473684210527,
      "grad_norm": 3.2129334508357736,
      "learning_rate": 3.9420056586181876e-07,
      "loss": 0.531,
      "step": 224
    },
    {
      "epoch": 1.1842105263157894,
      "grad_norm": 3.190794082513424,
      "learning_rate": 3.8995191620411994e-07,
      "loss": 0.5538,
      "step": 225
    },
    {
      "epoch": 1.1894736842105262,
      "grad_norm": 2.2027187652727087,
      "learning_rate": 3.85711600768822e-07,
      "loss": 0.5553,
      "step": 226
    },
    {
      "epoch": 1.194736842105263,
      "grad_norm": 9.415768580485024,
      "learning_rate": 3.814799406858408e-07,
      "loss": 0.5352,
      "step": 227
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.5331004049850994,
      "learning_rate": 3.772572564296004e-07,
      "loss": 0.552,
      "step": 228
    },
    {
      "epoch": 1.2052631578947368,
      "grad_norm": 2.5568166677231936,
      "learning_rate": 3.730438677947624e-07,
      "loss": 0.524,
      "step": 229
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 2.430967299620147,
      "learning_rate": 3.688400938720071e-07,
      "loss": 0.5596,
      "step": 230
    },
    {
      "epoch": 1.2157894736842105,
      "grad_norm": 2.2841927182707686,
      "learning_rate": 3.646462530238683e-07,
      "loss": 0.5504,
      "step": 231
    },
    {
      "epoch": 1.2210526315789474,
      "grad_norm": 2.709784732818941,
      "learning_rate": 3.60462662860623e-07,
      "loss": 0.5485,
      "step": 232
    },
    {
      "epoch": 1.2263157894736842,
      "grad_norm": 2.35859083822343,
      "learning_rate": 3.5628964021623696e-07,
      "loss": 0.556,
      "step": 233
    },
    {
      "epoch": 1.231578947368421,
      "grad_norm": 2.4468595923428214,
      "learning_rate": 3.521275011243715e-07,
      "loss": 0.5603,
      "step": 234
    },
    {
      "epoch": 1.236842105263158,
      "grad_norm": 2.4049891202411886,
      "learning_rate": 3.47976560794448e-07,
      "loss": 0.5585,
      "step": 235
    },
    {
      "epoch": 1.2421052631578948,
      "grad_norm": 2.292589627214253,
      "learning_rate": 3.4383713358777735e-07,
      "loss": 0.5383,
      "step": 236
    },
    {
      "epoch": 1.2473684210526317,
      "grad_norm": 2.3752679374582426,
      "learning_rate": 3.3970953299375257e-07,
      "loss": 0.5516,
      "step": 237
    },
    {
      "epoch": 1.2526315789473683,
      "grad_norm": 2.4876254806908116,
      "learning_rate": 3.3559407160610644e-07,
      "loss": 0.5484,
      "step": 238
    },
    {
      "epoch": 1.2578947368421054,
      "grad_norm": 2.762247825036067,
      "learning_rate": 3.3149106109923896e-07,
      "loss": 0.5537,
      "step": 239
    },
    {
      "epoch": 1.263157894736842,
      "grad_norm": 2.162751322141153,
      "learning_rate": 3.274008122046132e-07,
      "loss": 0.566,
      "step": 240
    },
    {
      "epoch": 1.268421052631579,
      "grad_norm": 3.581918002736885,
      "learning_rate": 3.2332363468722267e-07,
      "loss": 0.5474,
      "step": 241
    },
    {
      "epoch": 1.2736842105263158,
      "grad_norm": 2.284588189418711,
      "learning_rate": 3.192598373221322e-07,
      "loss": 0.5479,
      "step": 242
    },
    {
      "epoch": 1.2789473684210526,
      "grad_norm": 2.268653101760438,
      "learning_rate": 3.152097278710933e-07,
      "loss": 0.5047,
      "step": 243
    },
    {
      "epoch": 1.2842105263157895,
      "grad_norm": 2.2436341824829715,
      "learning_rate": 3.1117361305923686e-07,
      "loss": 0.5442,
      "step": 244
    },
    {
      "epoch": 1.2894736842105263,
      "grad_norm": 3.526093157139422,
      "learning_rate": 3.071517985518442e-07,
      "loss": 0.5301,
      "step": 245
    },
    {
      "epoch": 1.2947368421052632,
      "grad_norm": 2.4393150883883377,
      "learning_rate": 3.03144588931198e-07,
      "loss": 0.552,
      "step": 246
    },
    {
      "epoch": 1.3,
      "grad_norm": 2.25707824609034,
      "learning_rate": 2.9915228767351535e-07,
      "loss": 0.5216,
      "step": 247
    },
    {
      "epoch": 1.305263157894737,
      "grad_norm": 2.4659870700865554,
      "learning_rate": 2.9517519712596494e-07,
      "loss": 0.5531,
      "step": 248
    },
    {
      "epoch": 1.3105263157894738,
      "grad_norm": 2.211365210217085,
      "learning_rate": 2.9121361848377013e-07,
      "loss": 0.5464,
      "step": 249
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 2.4990811569238147,
      "learning_rate": 2.872678517673975e-07,
      "loss": 0.551,
      "step": 250
    },
    {
      "epoch": 1.3210526315789473,
      "grad_norm": 2.256640053743778,
      "learning_rate": 2.833381957998362e-07,
      "loss": 0.537,
      "step": 251
    },
    {
      "epoch": 1.3263157894736843,
      "grad_norm": 2.439046289136518,
      "learning_rate": 2.7942494818396687e-07,
      "loss": 0.5446,
      "step": 252
    },
    {
      "epoch": 1.331578947368421,
      "grad_norm": 2.4445909166764994,
      "learning_rate": 2.75528405280025e-07,
      "loss": 0.5522,
      "step": 253
    },
    {
      "epoch": 1.3368421052631578,
      "grad_norm": 2.503169992163442,
      "learning_rate": 2.7164886218315444e-07,
      "loss": 0.5315,
      "step": 254
    },
    {
      "epoch": 1.3421052631578947,
      "grad_norm": 3.1819620554512564,
      "learning_rate": 2.6778661270106025e-07,
      "loss": 0.5657,
      "step": 255
    },
    {
      "epoch": 1.3473684210526315,
      "grad_norm": 2.4239145590188675,
      "learning_rate": 2.639419493317587e-07,
      "loss": 0.5467,
      "step": 256
    },
    {
      "epoch": 1.3526315789473684,
      "grad_norm": 2.325463024514789,
      "learning_rate": 2.601151632414241e-07,
      "loss": 0.5445,
      "step": 257
    },
    {
      "epoch": 1.3578947368421053,
      "grad_norm": 2.6223884759828704,
      "learning_rate": 2.56306544242339e-07,
      "loss": 0.5484,
      "step": 258
    },
    {
      "epoch": 1.3631578947368421,
      "grad_norm": 2.8460302822874852,
      "learning_rate": 2.5251638077094603e-07,
      "loss": 0.5417,
      "step": 259
    },
    {
      "epoch": 1.368421052631579,
      "grad_norm": 2.24730489630462,
      "learning_rate": 2.487449598660029e-07,
      "loss": 0.5285,
      "step": 260
    },
    {
      "epoch": 1.3736842105263158,
      "grad_norm": 2.3421953136620406,
      "learning_rate": 2.449925671468456e-07,
      "loss": 0.5272,
      "step": 261
    },
    {
      "epoch": 1.3789473684210527,
      "grad_norm": 2.4559845843844523,
      "learning_rate": 2.412594867917568e-07,
      "loss": 0.5421,
      "step": 262
    },
    {
      "epoch": 1.3842105263157896,
      "grad_norm": 3.115271141122453,
      "learning_rate": 2.3754600151644444e-07,
      "loss": 0.5516,
      "step": 263
    },
    {
      "epoch": 1.3894736842105262,
      "grad_norm": 2.593266549185136,
      "learning_rate": 2.3385239255263073e-07,
      "loss": 0.5425,
      "step": 264
    },
    {
      "epoch": 1.3947368421052633,
      "grad_norm": 2.2082484863636194,
      "learning_rate": 2.3017893962675454e-07,
      "loss": 0.542,
      "step": 265
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.283360151388717,
      "learning_rate": 2.2652592093878665e-07,
      "loss": 0.5554,
      "step": 266
    },
    {
      "epoch": 1.4052631578947368,
      "grad_norm": 2.3911086301133446,
      "learning_rate": 2.2289361314116006e-07,
      "loss": 0.5194,
      "step": 267
    },
    {
      "epoch": 1.4105263157894736,
      "grad_norm": 2.593370895355469,
      "learning_rate": 2.1928229131782006e-07,
      "loss": 0.5504,
      "step": 268
    },
    {
      "epoch": 1.4157894736842105,
      "grad_norm": 2.393818171006261,
      "learning_rate": 2.1569222896338963e-07,
      "loss": 0.5452,
      "step": 269
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 2.4163961537782717,
      "learning_rate": 2.121236979624586e-07,
      "loss": 0.5244,
      "step": 270
    },
    {
      "epoch": 1.4263157894736842,
      "grad_norm": 2.3320647289960883,
      "learning_rate": 2.085769685689923e-07,
      "loss": 0.5484,
      "step": 271
    },
    {
      "epoch": 1.431578947368421,
      "grad_norm": 2.343996890572737,
      "learning_rate": 2.0505230938586415e-07,
      "loss": 0.5619,
      "step": 272
    },
    {
      "epoch": 1.436842105263158,
      "grad_norm": 6.066669575842648,
      "learning_rate": 2.0154998734451472e-07,
      "loss": 0.5276,
      "step": 273
    },
    {
      "epoch": 1.4421052631578948,
      "grad_norm": 2.2887239747665045,
      "learning_rate": 1.9807026768473579e-07,
      "loss": 0.5576,
      "step": 274
    },
    {
      "epoch": 1.4473684210526316,
      "grad_norm": 2.1868450818312644,
      "learning_rate": 1.9461341393458252e-07,
      "loss": 0.5505,
      "step": 275
    },
    {
      "epoch": 1.4526315789473685,
      "grad_norm": 5.547035869334998,
      "learning_rate": 1.911796878904171e-07,
      "loss": 0.5304,
      "step": 276
    },
    {
      "epoch": 1.4578947368421051,
      "grad_norm": 2.356826285784176,
      "learning_rate": 1.877693495970809e-07,
      "loss": 0.5072,
      "step": 277
    },
    {
      "epoch": 1.4631578947368422,
      "grad_norm": 2.1453867969594818,
      "learning_rate": 1.8438265732820125e-07,
      "loss": 0.5542,
      "step": 278
    },
    {
      "epoch": 1.4684210526315788,
      "grad_norm": 2.487577099034787,
      "learning_rate": 1.8101986756663196e-07,
      "loss": 0.5136,
      "step": 279
    },
    {
      "epoch": 1.4736842105263157,
      "grad_norm": 2.1415604665665064,
      "learning_rate": 1.776812349850289e-07,
      "loss": 0.5188,
      "step": 280
    },
    {
      "epoch": 1.4789473684210526,
      "grad_norm": 2.2729801926850026,
      "learning_rate": 1.743670124265627e-07,
      "loss": 0.5354,
      "step": 281
    },
    {
      "epoch": 1.4842105263157894,
      "grad_norm": 2.962685274358996,
      "learning_rate": 1.7107745088577087e-07,
      "loss": 0.5444,
      "step": 282
    },
    {
      "epoch": 1.4894736842105263,
      "grad_norm": 2.493997771189138,
      "learning_rate": 1.6781279948954918e-07,
      "loss": 0.5124,
      "step": 283
    },
    {
      "epoch": 1.4947368421052631,
      "grad_norm": 2.174484329414595,
      "learning_rate": 1.6457330547828403e-07,
      "loss": 0.5186,
      "step": 284
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.294851477833156,
      "learning_rate": 1.6135921418712955e-07,
      "loss": 0.5459,
      "step": 285
    },
    {
      "epoch": 1.5052631578947369,
      "grad_norm": 2.431178684998425,
      "learning_rate": 1.5817076902742622e-07,
      "loss": 0.5133,
      "step": 286
    },
    {
      "epoch": 1.5105263157894737,
      "grad_norm": 2.591260995377334,
      "learning_rate": 1.5500821146826804e-07,
      "loss": 0.5263,
      "step": 287
    },
    {
      "epoch": 1.5157894736842106,
      "grad_norm": 2.296988414628801,
      "learning_rate": 1.5187178101821502e-07,
      "loss": 0.5406,
      "step": 288
    },
    {
      "epoch": 1.5210526315789474,
      "grad_norm": 2.518137143341776,
      "learning_rate": 1.4876171520715397e-07,
      "loss": 0.5537,
      "step": 289
    },
    {
      "epoch": 1.526315789473684,
      "grad_norm": 4.157747210914889,
      "learning_rate": 1.4567824956831042e-07,
      "loss": 0.5321,
      "step": 290
    },
    {
      "epoch": 1.5315789473684212,
      "grad_norm": 2.304265529094299,
      "learning_rate": 1.426216176204112e-07,
      "loss": 0.5453,
      "step": 291
    },
    {
      "epoch": 1.5368421052631578,
      "grad_norm": 2.2909234105971894,
      "learning_rate": 1.395920508499991e-07,
      "loss": 0.5329,
      "step": 292
    },
    {
      "epoch": 1.5421052631578949,
      "grad_norm": 2.642177484706335,
      "learning_rate": 1.3658977869390164e-07,
      "loss": 0.5309,
      "step": 293
    },
    {
      "epoch": 1.5473684210526315,
      "grad_norm": 2.4224952394007966,
      "learning_rate": 1.336150285218558e-07,
      "loss": 0.5151,
      "step": 294
    },
    {
      "epoch": 1.5526315789473686,
      "grad_norm": 2.303479541727698,
      "learning_rate": 1.3066802561928853e-07,
      "loss": 0.5529,
      "step": 295
    },
    {
      "epoch": 1.5578947368421052,
      "grad_norm": 3.187263476988102,
      "learning_rate": 1.2774899317025467e-07,
      "loss": 0.5496,
      "step": 296
    },
    {
      "epoch": 1.563157894736842,
      "grad_norm": 2.1628631688530007,
      "learning_rate": 1.248581522405358e-07,
      "loss": 0.5525,
      "step": 297
    },
    {
      "epoch": 1.568421052631579,
      "grad_norm": 2.406176175843737,
      "learning_rate": 1.219957217608974e-07,
      "loss": 0.5429,
      "step": 298
    },
    {
      "epoch": 1.5736842105263158,
      "grad_norm": 2.262578641804591,
      "learning_rate": 1.1916191851050871e-07,
      "loss": 0.5066,
      "step": 299
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 2.4369446864539395,
      "learning_rate": 1.1635695710052689e-07,
      "loss": 0.5285,
      "step": 300
    },
    {
      "epoch": 1.5842105263157895,
      "grad_norm": 2.2109343115341993,
      "learning_rate": 1.1358104995784185e-07,
      "loss": 0.5569,
      "step": 301
    },
    {
      "epoch": 1.5894736842105264,
      "grad_norm": 2.330458766354278,
      "learning_rate": 1.1083440730898974e-07,
      "loss": 0.5665,
      "step": 302
    },
    {
      "epoch": 1.594736842105263,
      "grad_norm": 2.305815278527985,
      "learning_rate": 1.0811723716423232e-07,
      "loss": 0.516,
      "step": 303
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.3079199650608766,
      "learning_rate": 1.0542974530180327e-07,
      "loss": 0.5285,
      "step": 304
    },
    {
      "epoch": 1.6052631578947367,
      "grad_norm": 2.513610078891517,
      "learning_rate": 1.027721352523237e-07,
      "loss": 0.544,
      "step": 305
    },
    {
      "epoch": 1.6105263157894738,
      "grad_norm": 2.5767331454181255,
      "learning_rate": 1.0014460828338928e-07,
      "loss": 0.5293,
      "step": 306
    },
    {
      "epoch": 1.6157894736842104,
      "grad_norm": 2.199657550666274,
      "learning_rate": 9.754736338432679e-08,
      "loss": 0.5391,
      "step": 307
    },
    {
      "epoch": 1.6210526315789475,
      "grad_norm": 2.754603403945606,
      "learning_rate": 9.498059725112468e-08,
      "loss": 0.536,
      "step": 308
    },
    {
      "epoch": 1.6263157894736842,
      "grad_norm": 2.536336143602128,
      "learning_rate": 9.244450427153683e-08,
      "loss": 0.5468,
      "step": 309
    },
    {
      "epoch": 1.631578947368421,
      "grad_norm": 2.3411920268532587,
      "learning_rate": 8.993927651036049e-08,
      "loss": 0.5151,
      "step": 310
    },
    {
      "epoch": 1.6368421052631579,
      "grad_norm": 2.1051916003559255,
      "learning_rate": 8.746510369489102e-08,
      "loss": 0.5238,
      "step": 311
    },
    {
      "epoch": 1.6421052631578947,
      "grad_norm": 3.544977524584269,
      "learning_rate": 8.502217320055426e-08,
      "loss": 0.559,
      "step": 312
    },
    {
      "epoch": 1.6473684210526316,
      "grad_norm": 2.058244791333844,
      "learning_rate": 8.261067003671446e-08,
      "loss": 0.5206,
      "step": 313
    },
    {
      "epoch": 1.6526315789473685,
      "grad_norm": 2.446605948134395,
      "learning_rate": 8.023077683266399e-08,
      "loss": 0.5369,
      "step": 314
    },
    {
      "epoch": 1.6578947368421053,
      "grad_norm": 2.4153323922926,
      "learning_rate": 7.78826738237926e-08,
      "loss": 0.5458,
      "step": 315
    },
    {
      "epoch": 1.663157894736842,
      "grad_norm": 2.6104861029979016,
      "learning_rate": 7.556653883793724e-08,
      "loss": 0.5451,
      "step": 316
    },
    {
      "epoch": 1.668421052631579,
      "grad_norm": 2.3100120056396296,
      "learning_rate": 7.328254728191463e-08,
      "loss": 0.5363,
      "step": 317
    },
    {
      "epoch": 1.6736842105263157,
      "grad_norm": 2.5621407650897936,
      "learning_rate": 7.103087212823778e-08,
      "loss": 0.5099,
      "step": 318
    },
    {
      "epoch": 1.6789473684210527,
      "grad_norm": 2.118171619643243,
      "learning_rate": 6.881168390201581e-08,
      "loss": 0.5322,
      "step": 319
    },
    {
      "epoch": 1.6842105263157894,
      "grad_norm": 2.907352165494029,
      "learning_rate": 6.66251506680397e-08,
      "loss": 0.555,
      "step": 320
    },
    {
      "epoch": 1.6894736842105265,
      "grad_norm": 2.2807676681250446,
      "learning_rate": 6.447143801805515e-08,
      "loss": 0.5408,
      "step": 321
    },
    {
      "epoch": 1.694736842105263,
      "grad_norm": 2.297176719362461,
      "learning_rate": 6.23507090582206e-08,
      "loss": 0.5226,
      "step": 322
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.287950688640522,
      "learning_rate": 6.026312439675551e-08,
      "loss": 0.5096,
      "step": 323
    },
    {
      "epoch": 1.7052631578947368,
      "grad_norm": 3.6162969976516024,
      "learning_rate": 5.820884213177712e-08,
      "loss": 0.5365,
      "step": 324
    },
    {
      "epoch": 1.7105263157894737,
      "grad_norm": 2.2306965460746473,
      "learning_rate": 5.618801783932725e-08,
      "loss": 0.5287,
      "step": 325
    },
    {
      "epoch": 1.7157894736842105,
      "grad_norm": 2.2028587942070588,
      "learning_rate": 5.420080456158971e-08,
      "loss": 0.5566,
      "step": 326
    },
    {
      "epoch": 1.7210526315789474,
      "grad_norm": 2.18142780543031,
      "learning_rate": 5.2247352795300626e-08,
      "loss": 0.5375,
      "step": 327
    },
    {
      "epoch": 1.7263157894736842,
      "grad_norm": 2.197343563558157,
      "learning_rate": 5.0327810480350344e-08,
      "loss": 0.5487,
      "step": 328
    },
    {
      "epoch": 1.731578947368421,
      "grad_norm": 2.316942598992064,
      "learning_rate": 4.84423229885802e-08,
      "loss": 0.5432,
      "step": 329
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 2.3842561102459356,
      "learning_rate": 4.659103311277274e-08,
      "loss": 0.5237,
      "step": 330
    },
    {
      "epoch": 1.7421052631578946,
      "grad_norm": 2.177074041941887,
      "learning_rate": 4.477408105583741e-08,
      "loss": 0.5203,
      "step": 331
    },
    {
      "epoch": 1.7473684210526317,
      "grad_norm": 2.312588944046282,
      "learning_rate": 4.2991604420193396e-08,
      "loss": 0.5548,
      "step": 332
    },
    {
      "epoch": 1.7526315789473683,
      "grad_norm": 3.476255191649119,
      "learning_rate": 4.124373819734794e-08,
      "loss": 0.5258,
      "step": 333
    },
    {
      "epoch": 1.7578947368421054,
      "grad_norm": 2.14006154277014,
      "learning_rate": 3.953061475767339e-08,
      "loss": 0.5094,
      "step": 334
    },
    {
      "epoch": 1.763157894736842,
      "grad_norm": 2.152901623992758,
      "learning_rate": 3.7852363840382316e-08,
      "loss": 0.5359,
      "step": 335
    },
    {
      "epoch": 1.768421052631579,
      "grad_norm": 2.2182151450123437,
      "learning_rate": 3.6209112543702236e-08,
      "loss": 0.5026,
      "step": 336
    },
    {
      "epoch": 1.7736842105263158,
      "grad_norm": 2.2236206549096713,
      "learning_rate": 3.460098531525018e-08,
      "loss": 0.5636,
      "step": 337
    },
    {
      "epoch": 1.7789473684210526,
      "grad_norm": 2.384680110369391,
      "learning_rate": 3.3028103942607353e-08,
      "loss": 0.5464,
      "step": 338
    },
    {
      "epoch": 1.7842105263157895,
      "grad_norm": 2.9760197665338937,
      "learning_rate": 3.149058754409678e-08,
      "loss": 0.5301,
      "step": 339
    },
    {
      "epoch": 1.7894736842105263,
      "grad_norm": 2.3297740433944254,
      "learning_rate": 2.9988552559761294e-08,
      "loss": 0.536,
      "step": 340
    },
    {
      "epoch": 1.7947368421052632,
      "grad_norm": 3.9190533049512073,
      "learning_rate": 2.85221127425459e-08,
      "loss": 0.5358,
      "step": 341
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.4877390644915085,
      "learning_rate": 2.7091379149682682e-08,
      "loss": 0.5317,
      "step": 342
    },
    {
      "epoch": 1.805263157894737,
      "grad_norm": 2.250535562910496,
      "learning_rate": 2.5696460134279953e-08,
      "loss": 0.5289,
      "step": 343
    },
    {
      "epoch": 1.8105263157894735,
      "grad_norm": 2.2819477227854446,
      "learning_rate": 2.4337461337116892e-08,
      "loss": 0.5534,
      "step": 344
    },
    {
      "epoch": 1.8157894736842106,
      "grad_norm": 2.488467928210033,
      "learning_rate": 2.301448567864256e-08,
      "loss": 0.5169,
      "step": 345
    },
    {
      "epoch": 1.8210526315789473,
      "grad_norm": 2.5228811063557366,
      "learning_rate": 2.1727633351181994e-08,
      "loss": 0.5516,
      "step": 346
    },
    {
      "epoch": 1.8263157894736843,
      "grad_norm": 2.1521113439043438,
      "learning_rate": 2.0477001811347984e-08,
      "loss": 0.5327,
      "step": 347
    },
    {
      "epoch": 1.831578947368421,
      "grad_norm": 2.2553781295744786,
      "learning_rate": 1.9262685772660603e-08,
      "loss": 0.5468,
      "step": 348
    },
    {
      "epoch": 1.836842105263158,
      "grad_norm": 2.3149958407144946,
      "learning_rate": 1.8084777198374313e-08,
      "loss": 0.5553,
      "step": 349
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 2.2277223935869235,
      "learning_rate": 1.6943365294513235e-08,
      "loss": 0.5614,
      "step": 350
    },
    {
      "epoch": 1.8473684210526315,
      "grad_norm": 2.3882894364212,
      "learning_rate": 1.5838536503115675e-08,
      "loss": 0.5449,
      "step": 351
    },
    {
      "epoch": 1.8526315789473684,
      "grad_norm": 2.2020675552755,
      "learning_rate": 1.4770374495687133e-08,
      "loss": 0.5334,
      "step": 352
    },
    {
      "epoch": 1.8578947368421053,
      "grad_norm": 2.488077093971711,
      "learning_rate": 1.37389601668641e-08,
      "loss": 0.5346,
      "step": 353
    },
    {
      "epoch": 1.8631578947368421,
      "grad_norm": 2.2318977307050356,
      "learning_rate": 1.274437162828751e-08,
      "loss": 0.5377,
      "step": 354
    },
    {
      "epoch": 1.868421052631579,
      "grad_norm": 2.68631686113767,
      "learning_rate": 1.1786684202687025e-08,
      "loss": 0.5704,
      "step": 355
    },
    {
      "epoch": 1.8736842105263158,
      "grad_norm": 2.5296212288222724,
      "learning_rate": 1.086597041817705e-08,
      "loss": 0.5546,
      "step": 356
    },
    {
      "epoch": 1.8789473684210525,
      "grad_norm": 2.2211846561034654,
      "learning_rate": 9.98230000276351e-09,
      "loss": 0.5302,
      "step": 357
    },
    {
      "epoch": 1.8842105263157896,
      "grad_norm": 2.6225570074321554,
      "learning_rate": 9.135739879063464e-09,
      "loss": 0.5302,
      "step": 358
    },
    {
      "epoch": 1.8894736842105262,
      "grad_norm": 2.1746290538615027,
      "learning_rate": 8.326354159236882e-09,
      "loss": 0.5465,
      "step": 359
    },
    {
      "epoch": 1.8947368421052633,
      "grad_norm": 2.4217463634793037,
      "learning_rate": 7.554204140131137e-09,
      "loss": 0.5293,
      "step": 360
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.608159649984883,
      "learning_rate": 6.819348298638839e-09,
      "loss": 0.5158,
      "step": 361
    },
    {
      "epoch": 1.905263157894737,
      "grad_norm": 2.286787198440065,
      "learning_rate": 6.1218422872694186e-09,
      "loss": 0.5566,
      "step": 362
    },
    {
      "epoch": 1.9105263157894736,
      "grad_norm": 2.3164313990764187,
      "learning_rate": 5.46173892993429e-09,
      "loss": 0.5344,
      "step": 363
    },
    {
      "epoch": 1.9157894736842105,
      "grad_norm": 2.271875111172559,
      "learning_rate": 4.839088217946208e-09,
      "loss": 0.5479,
      "step": 364
    },
    {
      "epoch": 1.9210526315789473,
      "grad_norm": 2.3900875539689257,
      "learning_rate": 4.2539373062336904e-09,
      "loss": 0.5368,
      "step": 365
    },
    {
      "epoch": 1.9263157894736842,
      "grad_norm": 2.280769261985202,
      "learning_rate": 3.7063305097694287e-09,
      "loss": 0.5503,
      "step": 366
    },
    {
      "epoch": 1.931578947368421,
      "grad_norm": 2.591641861271708,
      "learning_rate": 3.196309300214528e-09,
      "loss": 0.5516,
      "step": 367
    },
    {
      "epoch": 1.936842105263158,
      "grad_norm": 6.988790160030566,
      "learning_rate": 2.72391230277752e-09,
      "loss": 0.5146,
      "step": 368
    },
    {
      "epoch": 1.9421052631578948,
      "grad_norm": 2.3712111188938714,
      "learning_rate": 2.289175293289314e-09,
      "loss": 0.5327,
      "step": 369
    },
    {
      "epoch": 1.9473684210526314,
      "grad_norm": 2.2537886816054127,
      "learning_rate": 1.8921311954937514e-09,
      "loss": 0.539,
      "step": 370
    },
    {
      "epoch": 1.9526315789473685,
      "grad_norm": 2.5549021483750742,
      "learning_rate": 1.5328100785542696e-09,
      "loss": 0.5858,
      "step": 371
    },
    {
      "epoch": 1.9578947368421051,
      "grad_norm": 2.260161761471616,
      "learning_rate": 1.2112391547766109e-09,
      "loss": 0.5325,
      "step": 372
    },
    {
      "epoch": 1.9631578947368422,
      "grad_norm": 2.0728662282209296,
      "learning_rate": 9.27442777547971e-10,
      "loss": 0.5319,
      "step": 373
    },
    {
      "epoch": 1.9684210526315788,
      "grad_norm": 2.235237067721937,
      "learning_rate": 6.814424394926965e-10,
      "loss": 0.5291,
      "step": 374
    },
    {
      "epoch": 1.973684210526316,
      "grad_norm": 2.448448284365754,
      "learning_rate": 4.732567708445879e-10,
      "loss": 0.5517,
      "step": 375
    },
    {
      "epoch": 1.9789473684210526,
      "grad_norm": 2.2224536930373695,
      "learning_rate": 3.0290153803591567e-10,
      "loss": 0.5473,
      "step": 376
    },
    {
      "epoch": 1.9842105263157894,
      "grad_norm": 2.27924977744772,
      "learning_rate": 1.7038964250343236e-10,
      "loss": 0.541,
      "step": 377
    },
    {
      "epoch": 1.9894736842105263,
      "grad_norm": 2.329045504499136,
      "learning_rate": 7.573111971148627e-11,
      "loss": 0.5445,
      "step": 378
    },
    {
      "epoch": 1.9947368421052631,
      "grad_norm": 2.3718632081482105,
      "learning_rate": 1.893313839157473e-11,
      "loss": 0.5472,
      "step": 379
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.301821768415375,
      "learning_rate": 0.0,
      "loss": 0.5179,
      "step": 380
    },
    {
      "epoch": 2.0,
      "step": 380,
      "total_flos": 1205553369972736.0,
      "train_loss": 0.6109166437073758,
      "train_runtime": 4792.8804,
      "train_samples_per_second": 5.067,
      "train_steps_per_second": 0.079
    }
  ],
  "logging_steps": 1,
  "max_steps": 380,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1205553369972736.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}