{
  "best_metric": 0.9330669508506887,
  "best_model_checkpoint": "rhp_WCETrain_tiny_BROLLLT_0001/run-fcec1_00000/checkpoint-1010",
  "epoch": 8.0,
  "eval_steps": 10,
  "global_step": 1616,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 43.166900634765625,
      "learning_rate": 1.3289785298903983e-05,
      "loss": 1.6043,
      "step": 1
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.821548461914062,
      "learning_rate": 1.328665165464581e-05,
      "loss": 0.4409,
      "step": 2
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.876423835754395,
      "learning_rate": 1.3283518010387641e-05,
      "loss": 0.2138,
      "step": 3
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.5617218017578125,
      "learning_rate": 1.328038436612947e-05,
      "loss": 0.1521,
      "step": 4
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.827425956726074,
      "learning_rate": 1.3277250721871299e-05,
      "loss": 0.1186,
      "step": 5
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5246403217315674,
      "learning_rate": 1.3274117077613126e-05,
      "loss": 0.0856,
      "step": 6
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.412988185882568,
      "learning_rate": 1.3270983433354957e-05,
      "loss": 0.0726,
      "step": 7
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9426465034484863,
      "learning_rate": 1.3267849789096786e-05,
      "loss": 0.0449,
      "step": 8
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9001221656799316,
      "learning_rate": 1.3264716144838613e-05,
      "loss": 0.0438,
      "step": 9
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.0339839458465576,
      "learning_rate": 1.3261582500580442e-05,
      "loss": 0.0564,
      "step": 10
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.596436023712158,
      "learning_rate": 1.3258448856322272e-05,
      "loss": 0.0655,
      "step": 11
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5015745162963867,
      "learning_rate": 1.32553152120641e-05,
      "loss": 0.0359,
      "step": 12
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3171091079711914,
      "learning_rate": 1.3252181567805929e-05,
      "loss": 0.0493,
      "step": 13
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.2199811935424805,
      "learning_rate": 1.3249047923547758e-05,
      "loss": 0.0465,
      "step": 14
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4519771337509155,
      "learning_rate": 1.3245914279289586e-05,
      "loss": 0.0303,
      "step": 15
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3096803426742554,
      "learning_rate": 1.3242780635031415e-05,
      "loss": 0.0306,
      "step": 16
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5370863676071167,
      "learning_rate": 1.3239646990773244e-05,
      "loss": 0.0295,
      "step": 17
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8910186290740967,
      "learning_rate": 1.3236513346515073e-05,
      "loss": 0.0453,
      "step": 18
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1542161703109741,
      "learning_rate": 1.3233379702256902e-05,
      "loss": 0.0218,
      "step": 19
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.2744879722595215,
      "learning_rate": 1.3230246057998731e-05,
      "loss": 0.0373,
      "step": 20
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1369712352752686,
      "learning_rate": 1.322711241374056e-05,
      "loss": 0.0387,
      "step": 21
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.7650647163391113,
      "learning_rate": 1.3223978769482387e-05,
      "loss": 0.0435,
      "step": 22
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.088339924812317,
      "learning_rate": 1.3220845125224218e-05,
      "loss": 0.0253,
      "step": 23
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8200860619544983,
      "learning_rate": 1.3217711480966047e-05,
      "loss": 0.0183,
      "step": 24
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1783356666564941,
      "learning_rate": 1.3214577836707874e-05,
      "loss": 0.0228,
      "step": 25
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5412431955337524,
      "learning_rate": 1.3211444192449703e-05,
      "loss": 0.0191,
      "step": 26
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3073039054870605,
      "learning_rate": 1.3208310548191532e-05,
      "loss": 0.0228,
      "step": 27
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2633461952209473,
      "learning_rate": 1.320517690393336e-05,
      "loss": 0.0306,
      "step": 28
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.998380184173584,
      "learning_rate": 1.320204325967519e-05,
      "loss": 0.0363,
      "step": 29
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8860704302787781,
      "learning_rate": 1.3198909615417019e-05,
      "loss": 0.0171,
      "step": 30
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.294479250907898,
      "learning_rate": 1.3195775971158848e-05,
      "loss": 0.017,
      "step": 31
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.840363502502441,
      "learning_rate": 1.3192642326900677e-05,
      "loss": 0.0211,
      "step": 32
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5014910697937012,
      "learning_rate": 1.3189508682642505e-05,
      "loss": 0.0651,
      "step": 33
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.305937767028809,
      "learning_rate": 1.3186375038384334e-05,
      "loss": 0.0229,
      "step": 34
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6240042448043823,
      "learning_rate": 1.3183241394126162e-05,
      "loss": 0.0278,
      "step": 35
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.8362228870391846,
      "learning_rate": 1.3180107749867992e-05,
      "loss": 0.028,
      "step": 36
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.944234609603882,
      "learning_rate": 1.3176974105609821e-05,
      "loss": 0.0357,
      "step": 37
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.427476167678833,
      "learning_rate": 1.317384046135165e-05,
      "loss": 0.029,
      "step": 38
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.966208815574646,
      "learning_rate": 1.3170706817093477e-05,
      "loss": 0.0139,
      "step": 39
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3142547607421875,
      "learning_rate": 1.3167573172835308e-05,
      "loss": 0.0278,
      "step": 40
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7708735466003418,
      "learning_rate": 1.3164439528577137e-05,
      "loss": 0.0231,
      "step": 41
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7919616103172302,
      "learning_rate": 1.3161305884318964e-05,
      "loss": 0.0191,
      "step": 42
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.11146354675293,
      "learning_rate": 1.3158172240060793e-05,
      "loss": 0.0261,
      "step": 43
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7685292959213257,
      "learning_rate": 1.3155038595802624e-05,
      "loss": 0.0141,
      "step": 44
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.605757474899292,
      "learning_rate": 1.315190495154445e-05,
      "loss": 0.0262,
      "step": 45
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0590410232543945,
      "learning_rate": 1.314877130728628e-05,
      "loss": 0.0294,
      "step": 46
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.779126763343811,
      "learning_rate": 1.3145637663028109e-05,
      "loss": 0.0172,
      "step": 47
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7185412645339966,
      "learning_rate": 1.3142504018769938e-05,
      "loss": 0.0334,
      "step": 48
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0430634021759033,
      "learning_rate": 1.3139370374511767e-05,
      "loss": 0.0172,
      "step": 49
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.957657814025879,
      "learning_rate": 1.3136236730253595e-05,
      "loss": 0.0492,
      "step": 50
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5715839862823486,
      "learning_rate": 1.3133103085995424e-05,
      "loss": 0.0135,
      "step": 51
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4538905620574951,
      "learning_rate": 1.3129969441737252e-05,
      "loss": 0.0327,
      "step": 52
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6182255744934082,
      "learning_rate": 1.3126835797479082e-05,
      "loss": 0.024,
      "step": 53
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.016416311264038,
      "learning_rate": 1.3123702153220911e-05,
      "loss": 0.0257,
      "step": 54
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7853610515594482,
      "learning_rate": 1.3120568508962738e-05,
      "loss": 0.0229,
      "step": 55
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7796930074691772,
      "learning_rate": 1.3117434864704567e-05,
      "loss": 0.0133,
      "step": 56
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3403170108795166,
      "learning_rate": 1.3114301220446398e-05,
      "loss": 0.0144,
      "step": 57
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8625386953353882,
      "learning_rate": 1.3111167576188225e-05,
      "loss": 0.0222,
      "step": 58
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6226298809051514,
      "learning_rate": 1.3108033931930054e-05,
      "loss": 0.0167,
      "step": 59
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6450958847999573,
      "learning_rate": 1.3104900287671883e-05,
      "loss": 0.0159,
      "step": 60
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.206281900405884,
      "learning_rate": 1.3101766643413712e-05,
      "loss": 0.0296,
      "step": 61
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6544156670570374,
      "learning_rate": 1.3098632999155541e-05,
      "loss": 0.016,
      "step": 62
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.539242744445801,
      "learning_rate": 1.309549935489737e-05,
      "loss": 0.0124,
      "step": 63
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.36065351963043213,
      "learning_rate": 1.3092365710639199e-05,
      "loss": 0.0128,
      "step": 64
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2435376644134521,
      "learning_rate": 1.3089232066381028e-05,
      "loss": 0.0287,
      "step": 65
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.375409722328186,
      "learning_rate": 1.3086098422122857e-05,
      "loss": 0.0193,
      "step": 66
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8153451681137085,
      "learning_rate": 1.3082964777864685e-05,
      "loss": 0.0343,
      "step": 67
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0442397594451904,
      "learning_rate": 1.3079831133606513e-05,
      "loss": 0.0253,
      "step": 68
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4716152846813202,
      "learning_rate": 1.3076697489348343e-05,
      "loss": 0.0108,
      "step": 69
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43147507309913635,
      "learning_rate": 1.3073563845090172e-05,
      "loss": 0.0126,
      "step": 70
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43620526790618896,
      "learning_rate": 1.3070430200832e-05,
      "loss": 0.0125,
      "step": 71
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.319814682006836,
      "learning_rate": 1.3067296556573828e-05,
      "loss": 0.0144,
      "step": 72
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4893972873687744,
      "learning_rate": 1.3064162912315659e-05,
      "loss": 0.0108,
      "step": 73
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6150720715522766,
      "learning_rate": 1.3061029268057488e-05,
      "loss": 0.0188,
      "step": 74
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5278173685073853,
      "learning_rate": 1.3057895623799315e-05,
      "loss": 0.0203,
      "step": 75
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.023555040359497,
      "learning_rate": 1.3054761979541144e-05,
      "loss": 0.012,
      "step": 76
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.470428705215454,
      "learning_rate": 1.3051628335282973e-05,
      "loss": 0.0232,
      "step": 77
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4155535697937012,
      "learning_rate": 1.3048494691024802e-05,
      "loss": 0.0273,
      "step": 78
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6539539694786072,
      "learning_rate": 1.3045361046766631e-05,
      "loss": 0.0231,
      "step": 79
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.05892276763916,
      "learning_rate": 1.304222740250846e-05,
      "loss": 0.015,
      "step": 80
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.7514607906341553,
      "learning_rate": 1.3039093758250289e-05,
      "loss": 0.0296,
      "step": 81
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1012790203094482,
      "learning_rate": 1.3035960113992118e-05,
      "loss": 0.0287,
      "step": 82
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6105172634124756,
      "learning_rate": 1.3032826469733947e-05,
      "loss": 0.0151,
      "step": 83
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6441468000411987,
      "learning_rate": 1.3029692825475776e-05,
      "loss": 0.011,
      "step": 84
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.22395525872707367,
      "learning_rate": 1.3026559181217603e-05,
      "loss": 0.0077,
      "step": 85
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0234441757202148,
      "learning_rate": 1.3023425536959433e-05,
      "loss": 0.0128,
      "step": 86
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6210427284240723,
      "learning_rate": 1.3020291892701262e-05,
      "loss": 0.0322,
      "step": 87
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.513728141784668,
      "learning_rate": 1.301715824844309e-05,
      "loss": 0.0225,
      "step": 88
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.410039186477661,
      "learning_rate": 1.3014024604184918e-05,
      "loss": 0.022,
      "step": 89
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5189119577407837,
      "learning_rate": 1.3010890959926749e-05,
      "loss": 0.0152,
      "step": 90
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.052048444747925,
      "learning_rate": 1.3007757315668576e-05,
      "loss": 0.0143,
      "step": 91
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2305142879486084,
      "learning_rate": 1.3004623671410405e-05,
      "loss": 0.0328,
      "step": 92
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.83760929107666,
      "learning_rate": 1.3001490027152234e-05,
      "loss": 0.0293,
      "step": 93
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.578233242034912,
      "learning_rate": 1.2998356382894063e-05,
      "loss": 0.0276,
      "step": 94
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5697505474090576,
      "learning_rate": 1.2995222738635892e-05,
      "loss": 0.0232,
      "step": 95
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2127329111099243,
      "learning_rate": 1.2992089094377721e-05,
      "loss": 0.0195,
      "step": 96
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7041822671890259,
      "learning_rate": 1.298895545011955e-05,
      "loss": 0.0372,
      "step": 97
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3832337260246277,
      "learning_rate": 1.2985821805861379e-05,
      "loss": 0.0109,
      "step": 98
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.687912702560425,
      "learning_rate": 1.2982688161603208e-05,
      "loss": 0.0237,
      "step": 99
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.15417742729187,
      "learning_rate": 1.2979554517345037e-05,
      "loss": 0.0414,
      "step": 100
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.773973822593689,
      "learning_rate": 1.2976420873086864e-05,
      "loss": 0.0076,
      "step": 101
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0944068431854248,
      "learning_rate": 1.2973287228828693e-05,
      "loss": 0.0135,
      "step": 102
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9463632702827454,
      "learning_rate": 1.2970153584570523e-05,
      "loss": 0.0184,
      "step": 103
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9553542137145996,
      "learning_rate": 1.296701994031235e-05,
      "loss": 0.0171,
      "step": 104
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7304505109786987,
      "learning_rate": 1.296388629605418e-05,
      "loss": 0.0292,
      "step": 105
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8401722311973572,
      "learning_rate": 1.2960752651796008e-05,
      "loss": 0.0176,
      "step": 106
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2365652322769165,
      "learning_rate": 1.2957619007537839e-05,
      "loss": 0.0225,
      "step": 107
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7007886171340942,
      "learning_rate": 1.2954485363279666e-05,
      "loss": 0.013,
      "step": 108
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.48461851477622986,
      "learning_rate": 1.2951351719021495e-05,
      "loss": 0.0099,
      "step": 109
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.42651697993278503,
      "learning_rate": 1.2948218074763324e-05,
      "loss": 0.0123,
      "step": 110
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.572167992591858,
      "learning_rate": 1.2945084430505153e-05,
      "loss": 0.0281,
      "step": 111
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.446479856967926,
      "learning_rate": 1.2941950786246982e-05,
      "loss": 0.008,
      "step": 112
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3084217309951782,
      "learning_rate": 1.2938817141988811e-05,
      "loss": 0.0246,
      "step": 113
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.232654571533203,
      "learning_rate": 1.293568349773064e-05,
      "loss": 0.0248,
      "step": 114
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.182873249053955,
      "learning_rate": 1.2932549853472469e-05,
      "loss": 0.0302,
      "step": 115
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.159117341041565,
      "learning_rate": 1.2929416209214298e-05,
      "loss": 0.0207,
      "step": 116
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2881197929382324,
      "learning_rate": 1.2926282564956127e-05,
      "loss": 0.0078,
      "step": 117
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5686463117599487,
      "learning_rate": 1.2923148920697954e-05,
      "loss": 0.0181,
      "step": 118
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.49851730465888977,
      "learning_rate": 1.2920015276439784e-05,
      "loss": 0.0109,
      "step": 119
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9558594822883606,
      "learning_rate": 1.2916881632181613e-05,
      "loss": 0.0196,
      "step": 120
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.739699125289917,
      "learning_rate": 1.291374798792344e-05,
      "loss": 0.0197,
      "step": 121
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5681390762329102,
      "learning_rate": 1.291061434366527e-05,
      "loss": 0.0134,
      "step": 122
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7622842788696289,
      "learning_rate": 1.29074806994071e-05,
      "loss": 0.013,
      "step": 123
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.403891086578369,
      "learning_rate": 1.2904347055148927e-05,
      "loss": 0.0108,
      "step": 124
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4364817142486572,
      "learning_rate": 1.2901213410890756e-05,
      "loss": 0.0117,
      "step": 125
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.48463937640190125,
      "learning_rate": 1.2898079766632585e-05,
      "loss": 0.0103,
      "step": 126
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8281981945037842,
      "learning_rate": 1.2894946122374414e-05,
      "loss": 0.0276,
      "step": 127
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32359665632247925,
      "learning_rate": 1.2891812478116243e-05,
      "loss": 0.0093,
      "step": 128
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23079122602939606,
      "learning_rate": 1.2888678833858072e-05,
      "loss": 0.006,
      "step": 129
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7822952270507812,
      "learning_rate": 1.2885545189599901e-05,
      "loss": 0.0379,
      "step": 130
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4681812524795532,
      "learning_rate": 1.2882411545341728e-05,
      "loss": 0.009,
      "step": 131
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6012412905693054,
      "learning_rate": 1.2879277901083559e-05,
      "loss": 0.0128,
      "step": 132
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.390124648809433,
      "learning_rate": 1.2876144256825388e-05,
      "loss": 0.0103,
      "step": 133
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8544343709945679,
      "learning_rate": 1.2873010612567215e-05,
      "loss": 0.0252,
      "step": 134
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.869128942489624,
      "learning_rate": 1.2869876968309044e-05,
      "loss": 0.0264,
      "step": 135
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36389434337615967,
      "learning_rate": 1.2866743324050875e-05,
      "loss": 0.0101,
      "step": 136
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8829776048660278,
      "learning_rate": 1.2863609679792702e-05,
      "loss": 0.0113,
      "step": 137
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.0819315910339355,
      "learning_rate": 1.286047603553453e-05,
      "loss": 0.0142,
      "step": 138
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8896161317825317,
      "learning_rate": 1.285734239127636e-05,
      "loss": 0.032,
      "step": 139
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.498758792877197,
      "learning_rate": 1.285420874701819e-05,
      "loss": 0.031,
      "step": 140
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8228998780250549,
      "learning_rate": 1.2851075102760017e-05,
      "loss": 0.0202,
      "step": 141
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9757546782493591,
      "learning_rate": 1.2847941458501846e-05,
      "loss": 0.0113,
      "step": 142
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9271151423454285,
      "learning_rate": 1.2844807814243675e-05,
      "loss": 0.0249,
      "step": 143
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37905001640319824,
      "learning_rate": 1.2841674169985504e-05,
      "loss": 0.0126,
      "step": 144
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2061004638671875,
      "learning_rate": 1.2838540525727333e-05,
      "loss": 0.0229,
      "step": 145
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2962275743484497,
      "learning_rate": 1.2835406881469162e-05,
      "loss": 0.0098,
      "step": 146
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0700948238372803,
      "learning_rate": 1.2832273237210991e-05,
      "loss": 0.015,
      "step": 147
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9125449657440186,
      "learning_rate": 1.282913959295282e-05,
      "loss": 0.0183,
      "step": 148
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9846622347831726,
      "learning_rate": 1.2826005948694649e-05,
      "loss": 0.0112,
      "step": 149
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.704027533531189,
      "learning_rate": 1.2822872304436478e-05,
      "loss": 0.0158,
      "step": 150
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9707388281822205,
      "learning_rate": 1.2819738660178305e-05,
      "loss": 0.0248,
      "step": 151
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4070046842098236,
      "learning_rate": 1.2816605015920134e-05,
      "loss": 0.0139,
      "step": 152
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5066566467285156,
      "learning_rate": 1.2813471371661965e-05,
      "loss": 0.0106,
      "step": 153
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3393709659576416,
      "learning_rate": 1.2810337727403792e-05,
      "loss": 0.0141,
      "step": 154
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.6112639904022217,
      "learning_rate": 1.280720408314562e-05,
      "loss": 0.0275,
      "step": 155
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.27600258588790894,
      "learning_rate": 1.280407043888745e-05,
      "loss": 0.0062,
      "step": 156
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0747978687286377,
      "learning_rate": 1.2800936794629279e-05,
      "loss": 0.0207,
      "step": 157
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.027299165725708,
      "learning_rate": 1.2797803150371107e-05,
      "loss": 0.0238,
      "step": 158
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7984327077865601,
      "learning_rate": 1.2794669506112936e-05,
      "loss": 0.012,
      "step": 159
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9884450435638428,
      "learning_rate": 1.2791535861854765e-05,
      "loss": 0.0222,
      "step": 160
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7388520836830139,
      "learning_rate": 1.2788402217596594e-05,
      "loss": 0.013,
      "step": 161
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4940875470638275,
      "learning_rate": 1.2785268573338423e-05,
      "loss": 0.0082,
      "step": 162
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9272128343582153,
      "learning_rate": 1.2782134929080252e-05,
      "loss": 0.0189,
      "step": 163
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0099061727523804,
      "learning_rate": 1.277900128482208e-05,
      "loss": 0.0174,
      "step": 164
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5828649997711182,
      "learning_rate": 1.277586764056391e-05,
      "loss": 0.0035,
      "step": 165
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9583665132522583,
      "learning_rate": 1.2772733996305739e-05,
      "loss": 0.0082,
      "step": 166
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4555276036262512,
      "learning_rate": 1.2769600352047566e-05,
      "loss": 0.0082,
      "step": 167
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6293783187866211,
      "learning_rate": 1.2766466707789395e-05,
      "loss": 0.0126,
      "step": 168
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6833214163780212,
      "learning_rate": 1.2763333063531226e-05,
      "loss": 0.0131,
      "step": 169
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7419341206550598,
      "learning_rate": 1.2760199419273053e-05,
      "loss": 0.014,
      "step": 170
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7213071584701538,
      "learning_rate": 1.2757065775014882e-05,
      "loss": 0.0124,
      "step": 171
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8481967449188232,
      "learning_rate": 1.275393213075671e-05,
      "loss": 0.0097,
      "step": 172
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0318468809127808,
      "learning_rate": 1.2750798486498541e-05,
      "loss": 0.0152,
      "step": 173
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.636877715587616,
      "learning_rate": 1.2747664842240369e-05,
      "loss": 0.0105,
      "step": 174
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.9102425575256348,
      "learning_rate": 1.2744531197982197e-05,
      "loss": 0.0142,
      "step": 175
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.412434101104736,
      "learning_rate": 1.2741397553724026e-05,
      "loss": 0.018,
      "step": 176
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2647795677185059,
      "learning_rate": 1.2738263909465854e-05,
      "loss": 0.0226,
      "step": 177
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5826900601387024,
      "learning_rate": 1.2735130265207684e-05,
      "loss": 0.0091,
      "step": 178
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3439165949821472,
      "learning_rate": 1.2731996620949513e-05,
      "loss": 0.0059,
      "step": 179
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7368501424789429,
      "learning_rate": 1.272886297669134e-05,
      "loss": 0.0157,
      "step": 180
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6436005234718323,
      "learning_rate": 1.272572933243317e-05,
      "loss": 0.0146,
      "step": 181
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2776414752006531,
      "learning_rate": 1.2722595688175e-05,
      "loss": 0.0079,
      "step": 182
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.715323805809021,
      "learning_rate": 1.2719462043916829e-05,
      "loss": 0.0223,
      "step": 183
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3590269088745117,
      "learning_rate": 1.2716328399658656e-05,
      "loss": 0.0193,
      "step": 184
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8406335115432739,
      "learning_rate": 1.2713194755400485e-05,
      "loss": 0.0194,
      "step": 185
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6337783932685852,
      "learning_rate": 1.2710061111142316e-05,
      "loss": 0.0107,
      "step": 186
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.198011875152588,
      "learning_rate": 1.2706927466884143e-05,
      "loss": 0.0176,
      "step": 187
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6882727146148682,
      "learning_rate": 1.2703793822625972e-05,
      "loss": 0.0177,
      "step": 188
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.989341139793396,
      "learning_rate": 1.27006601783678e-05,
      "loss": 0.0165,
      "step": 189
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5182610750198364,
      "learning_rate": 1.269752653410963e-05,
      "loss": 0.0139,
      "step": 190
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6215909719467163,
      "learning_rate": 1.2694392889851459e-05,
      "loss": 0.0092,
      "step": 191
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7435368299484253,
      "learning_rate": 1.2691259245593288e-05,
      "loss": 0.0163,
      "step": 192
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8726661801338196,
      "learning_rate": 1.2688125601335116e-05,
      "loss": 0.0117,
      "step": 193
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2530782222747803,
      "learning_rate": 1.2684991957076945e-05,
      "loss": 0.0151,
      "step": 194
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5423250794410706,
      "learning_rate": 1.2681858312818774e-05,
      "loss": 0.0055,
      "step": 195
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7455129027366638,
      "learning_rate": 1.2678724668560603e-05,
      "loss": 0.0126,
      "step": 196
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6774744987487793,
      "learning_rate": 1.267559102430243e-05,
      "loss": 0.0237,
      "step": 197
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3601767420768738,
      "learning_rate": 1.2672457380044261e-05,
      "loss": 0.0103,
      "step": 198
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9079448580741882,
      "learning_rate": 1.266932373578609e-05,
      "loss": 0.0117,
      "step": 199
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2807711660861969,
      "learning_rate": 1.2666190091527917e-05,
      "loss": 0.0098,
      "step": 200
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9219411611557007,
      "learning_rate": 1.2663056447269746e-05,
      "loss": 0.0128,
      "step": 201
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4324185252189636,
      "learning_rate": 1.2659922803011575e-05,
      "loss": 0.0072,
      "step": 202
    },
    {
      "epoch": 1.0,
      "eval_AVGf1": 0.9235035473191718,
      "eval_DIAG.f1": 0.9143646408839778,
      "eval_DIAG.precision": 0.927170868347339,
      "eval_DIAG.recall": 0.9019073569482289,
      "eval_MED.f1": 0.9382716049382717,
      "eval_MED.precision": 0.9661016949152542,
      "eval_MED.recall": 0.912,
      "eval_TREAT.f1": 0.9178743961352657,
      "eval_TREAT.precision": 0.9405940594059405,
      "eval_TREAT.recall": 0.8962264150943396,
      "eval_accuracy": 0.910707869205893,
      "eval_f1": 0.6958904109589041,
      "eval_loss": 0.012623640708625317,
      "eval_precision": 0.5664585191793042,
      "eval_recall": 0.9019886363636364,
      "eval_runtime": 943.4688,
      "eval_samples_per_second": 0.761,
      "eval_steps_per_second": 0.095,
      "step": 202
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2546284794807434,
      "learning_rate": 1.2656789158753404e-05,
      "loss": 0.0069,
      "step": 203
    },
    {
      "epoch": 1.01,
      "grad_norm": 2.1543943881988525,
      "learning_rate": 1.2653655514495233e-05,
      "loss": 0.0129,
      "step": 204
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1174274682998657,
      "learning_rate": 1.2650521870237062e-05,
      "loss": 0.0128,
      "step": 205
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.61497962474823,
      "learning_rate": 1.264738822597889e-05,
      "loss": 0.009,
      "step": 206
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5491525530815125,
      "learning_rate": 1.264425458172072e-05,
      "loss": 0.0095,
      "step": 207
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.27289947867393494,
      "learning_rate": 1.2641120937462549e-05,
      "loss": 0.0051,
      "step": 208
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7344390749931335,
      "learning_rate": 1.2637987293204378e-05,
      "loss": 0.0111,
      "step": 209
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.17898282408714294,
      "learning_rate": 1.2634853648946205e-05,
      "loss": 0.0041,
      "step": 210
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.37017083168029785,
      "learning_rate": 1.2631720004688035e-05,
      "loss": 0.009,
      "step": 211
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.4197113811969757,
      "learning_rate": 1.2628586360429864e-05,
      "loss": 0.0146,
      "step": 212
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6443505883216858,
      "learning_rate": 1.2625452716171692e-05,
      "loss": 0.0049,
      "step": 213
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.09277847409248352,
      "learning_rate": 1.262231907191352e-05,
      "loss": 0.0022,
      "step": 214
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.19542604684829712,
      "learning_rate": 1.2619185427655351e-05,
      "loss": 0.0041,
      "step": 215
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7286643385887146,
      "learning_rate": 1.261605178339718e-05,
      "loss": 0.0056,
      "step": 216
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.19170136749744415,
      "learning_rate": 1.2612918139139007e-05,
      "loss": 0.0035,
      "step": 217
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6621464490890503,
      "learning_rate": 1.2609784494880836e-05,
      "loss": 0.0047,
      "step": 218
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.38934943079948425,
      "learning_rate": 1.2606650850622667e-05,
      "loss": 0.0106,
      "step": 219
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.3695944547653198,
      "learning_rate": 1.2603517206364494e-05,
      "loss": 0.0083,
      "step": 220
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.20678924024105072,
      "learning_rate": 1.2600383562106323e-05,
      "loss": 0.0052,
      "step": 221
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.26029258966445923,
      "learning_rate": 1.2597249917848152e-05,
      "loss": 0.0057,
      "step": 222
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.3135945498943329,
      "learning_rate": 1.259411627358998e-05,
      "loss": 0.0047,
      "step": 223
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2602899372577667,
      "learning_rate": 1.259098262933181e-05,
      "loss": 0.0051,
      "step": 224
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.3139295279979706,
      "learning_rate": 1.2587848985073639e-05,
      "loss": 0.004,
      "step": 225
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.3058377504348755,
      "learning_rate": 1.2584715340815468e-05,
      "loss": 0.0027,
      "step": 226
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0445317029953003,
      "learning_rate": 1.2581581696557295e-05,
      "loss": 0.0067,
      "step": 227
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.8944097757339478,
      "learning_rate": 1.2578448052299125e-05,
      "loss": 0.0042,
      "step": 228
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.25380784273147583,
      "learning_rate": 1.2575314408040954e-05,
      "loss": 0.0046,
      "step": 229
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.22639180719852448,
      "learning_rate": 1.2572180763782782e-05,
      "loss": 0.0039,
      "step": 230
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7992675304412842,
      "learning_rate": 1.256904711952461e-05,
      "loss": 0.0112,
      "step": 231
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.23068396747112274,
      "learning_rate": 1.2565913475266441e-05,
      "loss": 0.0034,
      "step": 232
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.28125718235969543,
      "learning_rate": 1.2562779831008268e-05,
      "loss": 0.0036,
      "step": 233
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9341473579406738,
      "learning_rate": 1.2559646186750097e-05,
      "loss": 0.0094,
      "step": 234
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.3626331090927124,
      "learning_rate": 1.2556512542491926e-05,
      "loss": 0.0057,
      "step": 235
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9097028374671936,
      "learning_rate": 1.2553378898233755e-05,
      "loss": 0.013,
      "step": 236
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.5801799893379211,
      "learning_rate": 1.2550245253975584e-05,
      "loss": 0.0042,
      "step": 237
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.2265777587890625,
      "learning_rate": 1.2547111609717413e-05,
      "loss": 0.0334,
      "step": 238
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.21502335369586945,
      "learning_rate": 1.2543977965459242e-05,
      "loss": 0.0036,
      "step": 239
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.14358441531658173,
      "learning_rate": 1.254084432120107e-05,
      "loss": 0.0027,
      "step": 240
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.8060461282730103,
      "learning_rate": 1.25377106769429e-05,
      "loss": 0.0145,
      "step": 241
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.254424810409546,
      "learning_rate": 1.2534577032684729e-05,
      "loss": 0.0071,
      "step": 242
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.5932475328445435,
      "learning_rate": 1.2531443388426556e-05,
      "loss": 0.0062,
      "step": 243
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.220398411154747,
      "learning_rate": 1.2528309744168387e-05,
      "loss": 0.0041,
      "step": 244
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2945892810821533,
      "learning_rate": 1.2525176099910215e-05,
      "loss": 0.0043,
      "step": 245
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.3108975887298584,
      "learning_rate": 1.2522042455652043e-05,
      "loss": 0.0049,
      "step": 246
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.35047775506973267,
      "learning_rate": 1.2518908811393872e-05,
      "loss": 0.0053,
      "step": 247
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.33189657330513,
      "learning_rate": 1.2515775167135702e-05,
      "loss": 0.0108,
      "step": 248
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6183241009712219,
      "learning_rate": 1.2512641522877531e-05,
      "loss": 0.0108,
      "step": 249
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.21114368736743927,
      "learning_rate": 1.2509507878619358e-05,
      "loss": 0.0045,
      "step": 250
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.17985470592975616,
      "learning_rate": 1.2506374234361187e-05,
      "loss": 0.004,
      "step": 251
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.5435124635696411,
      "learning_rate": 1.2503240590103016e-05,
      "loss": 0.0073,
      "step": 252
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.5544082522392273,
      "learning_rate": 1.2500106945844845e-05,
      "loss": 0.0105,
      "step": 253
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.249291181564331,
      "learning_rate": 1.2496973301586674e-05,
      "loss": 0.009,
      "step": 254
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1447749137878418,
      "learning_rate": 1.2493839657328503e-05,
      "loss": 0.0141,
      "step": 255
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.27117419242858887,
      "learning_rate": 1.2490706013070332e-05,
      "loss": 0.0067,
      "step": 256
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.6128637194633484,
      "learning_rate": 1.248757236881216e-05,
      "loss": 0.0059,
      "step": 257
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2267618030309677,
      "learning_rate": 1.248443872455399e-05,
      "loss": 0.0052,
      "step": 258
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.16103290021419525,
      "learning_rate": 1.2481305080295819e-05,
      "loss": 0.0045,
      "step": 259
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.4214218556880951,
      "learning_rate": 1.2478171436037646e-05,
      "loss": 0.0064,
      "step": 260
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.2255360633134842,
      "learning_rate": 1.2475037791779477e-05,
      "loss": 0.0081,
      "step": 261
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.4919368326663971,
      "learning_rate": 1.2471904147521305e-05,
      "loss": 0.0132,
      "step": 262
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.15274657309055328,
      "learning_rate": 1.2468770503263133e-05,
      "loss": 0.004,
      "step": 263
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.2930839955806732,
      "learning_rate": 1.2465636859004962e-05,
      "loss": 0.0094,
      "step": 264
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.5679187178611755,
      "learning_rate": 1.2462503214746792e-05,
      "loss": 0.0097,
      "step": 265
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1493972539901733,
      "learning_rate": 1.245936957048862e-05,
      "loss": 0.009,
      "step": 266
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.2095317393541336,
      "learning_rate": 1.2456235926230448e-05,
      "loss": 0.0051,
      "step": 267
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.11597475409507751,
      "learning_rate": 1.2453102281972277e-05,
      "loss": 0.0027,
      "step": 268
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.3739665150642395,
      "learning_rate": 1.2449968637714106e-05,
      "loss": 0.0081,
      "step": 269
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8185667991638184,
      "learning_rate": 1.2446834993455935e-05,
      "loss": 0.0153,
      "step": 270
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.6023372411727905,
      "learning_rate": 1.2443701349197764e-05,
      "loss": 0.0078,
      "step": 271
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.19284436106681824,
      "learning_rate": 1.2440567704939593e-05,
      "loss": 0.0053,
      "step": 272
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2139939069747925,
      "learning_rate": 1.2437434060681422e-05,
      "loss": 0.0104,
      "step": 273
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7456642389297485,
      "learning_rate": 1.2434300416423251e-05,
      "loss": 0.0061,
      "step": 274
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.22498144209384918,
      "learning_rate": 1.243116677216508e-05,
      "loss": 0.0045,
      "step": 275
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.3246172070503235,
      "learning_rate": 1.2428033127906907e-05,
      "loss": 0.0104,
      "step": 276
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.33634820580482483,
      "learning_rate": 1.2424899483648736e-05,
      "loss": 0.0058,
      "step": 277
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.2140466868877411,
      "learning_rate": 1.2421765839390567e-05,
      "loss": 0.0045,
      "step": 278
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7434072494506836,
      "learning_rate": 1.2418632195132394e-05,
      "loss": 0.0068,
      "step": 279
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.998281478881836,
      "learning_rate": 1.2415498550874223e-05,
      "loss": 0.0112,
      "step": 280
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.35690030455589294,
      "learning_rate": 1.2412364906616052e-05,
      "loss": 0.0067,
      "step": 281
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0317761898040771,
      "learning_rate": 1.2409231262357882e-05,
      "loss": 0.0122,
      "step": 282
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.4638398587703705,
      "learning_rate": 1.240609761809971e-05,
      "loss": 0.0187,
      "step": 283
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8677943348884583,
      "learning_rate": 1.2402963973841538e-05,
      "loss": 0.0428,
      "step": 284
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.4307520389556885,
      "learning_rate": 1.2399830329583367e-05,
      "loss": 0.0097,
      "step": 285
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.3077194094657898,
      "learning_rate": 1.2396696685325196e-05,
      "loss": 0.0047,
      "step": 286
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.28216367959976196,
      "learning_rate": 1.2393563041067025e-05,
      "loss": 0.0081,
      "step": 287
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1216081380844116,
      "learning_rate": 1.2390429396808854e-05,
      "loss": 0.0087,
      "step": 288
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2751828134059906,
      "learning_rate": 1.2387295752550683e-05,
      "loss": 0.0039,
      "step": 289
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.18287332355976105,
      "learning_rate": 1.2384162108292512e-05,
      "loss": 0.003,
      "step": 290
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.24493369460105896,
      "learning_rate": 1.2381028464034341e-05,
      "loss": 0.0048,
      "step": 291
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.30328795313835144,
      "learning_rate": 1.237789481977617e-05,
      "loss": 0.009,
      "step": 292
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.19308710098266602,
      "learning_rate": 1.2374761175517997e-05,
      "loss": 0.0046,
      "step": 293
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.25658944249153137,
      "learning_rate": 1.2371627531259828e-05,
      "loss": 0.0053,
      "step": 294
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.6261961460113525,
      "learning_rate": 1.2368493887001657e-05,
      "loss": 0.0203,
      "step": 295
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2785145044326782,
      "learning_rate": 1.2365360242743484e-05,
      "loss": 0.0187,
      "step": 296
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.3342565894126892,
      "learning_rate": 1.2362226598485313e-05,
      "loss": 0.0053,
      "step": 297
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.2690996527671814,
      "learning_rate": 1.2359092954227143e-05,
      "loss": 0.0065,
      "step": 298
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.3764265179634094,
      "learning_rate": 1.235595930996897e-05,
      "loss": 0.0058,
      "step": 299
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4577929973602295,
      "learning_rate": 1.23528256657108e-05,
      "loss": 0.0162,
      "step": 300
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1272592544555664,
      "learning_rate": 1.2349692021452628e-05,
      "loss": 0.0091,
      "step": 301
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.3088337779045105,
      "learning_rate": 1.2346558377194457e-05,
      "loss": 0.0064,
      "step": 302
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2656728029251099,
      "learning_rate": 1.2343424732936286e-05,
      "loss": 0.0355,
      "step": 303
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6877207159996033,
      "learning_rate": 1.2340291088678115e-05,
      "loss": 0.0166,
      "step": 304
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.19311505556106567,
      "learning_rate": 1.2337157444419944e-05,
      "loss": 0.0049,
      "step": 305
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.26027628779411316,
      "learning_rate": 1.2334023800161771e-05,
      "loss": 0.0066,
      "step": 306
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.23496100306510925,
      "learning_rate": 1.2330890155903602e-05,
      "loss": 0.0049,
      "step": 307
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.23095273971557617,
      "learning_rate": 1.2327756511645431e-05,
      "loss": 0.0042,
      "step": 308
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7522357702255249,
      "learning_rate": 1.2324622867387258e-05,
      "loss": 0.0089,
      "step": 309
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.4904075562953949,
      "learning_rate": 1.2321489223129087e-05,
      "loss": 0.0056,
      "step": 310
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9362037181854248,
      "learning_rate": 1.2318355578870918e-05,
      "loss": 0.0175,
      "step": 311
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.005414605140686,
      "learning_rate": 1.2315221934612745e-05,
      "loss": 0.0111,
      "step": 312
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2100861817598343,
      "learning_rate": 1.2312088290354574e-05,
      "loss": 0.0043,
      "step": 313
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.20564229786396027,
      "learning_rate": 1.2308954646096403e-05,
      "loss": 0.005,
      "step": 314
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.3484073281288147,
      "learning_rate": 1.2305821001838233e-05,
      "loss": 0.0052,
      "step": 315
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.18312446773052216,
      "learning_rate": 1.230268735758006e-05,
      "loss": 0.0046,
      "step": 316
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.5120245814323425,
      "learning_rate": 1.229955371332189e-05,
      "loss": 0.0069,
      "step": 317
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.1598658263683319,
      "learning_rate": 1.2296420069063718e-05,
      "loss": 0.0022,
      "step": 318
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.5745701193809509,
      "learning_rate": 1.2293286424805547e-05,
      "loss": 0.0083,
      "step": 319
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.5370805263519287,
      "learning_rate": 1.2290152780547376e-05,
      "loss": 0.0128,
      "step": 320
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.09224429726600647,
      "learning_rate": 1.2287019136289205e-05,
      "loss": 0.0012,
      "step": 321
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.29811084270477295,
      "learning_rate": 1.2283885492031032e-05,
      "loss": 0.0079,
      "step": 322
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6683853268623352,
      "learning_rate": 1.2280751847772863e-05,
      "loss": 0.0092,
      "step": 323
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.3417896032333374,
      "learning_rate": 1.2277618203514692e-05,
      "loss": 0.0039,
      "step": 324
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.21969309449195862,
      "learning_rate": 1.2274484559256521e-05,
      "loss": 0.0055,
      "step": 325
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2806360125541687,
      "learning_rate": 1.2271350914998348e-05,
      "loss": 0.0038,
      "step": 326
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.6202746033668518,
      "learning_rate": 1.2268217270740177e-05,
      "loss": 0.0135,
      "step": 327
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8161211013793945,
      "learning_rate": 1.2265083626482008e-05,
      "loss": 0.0048,
      "step": 328
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.5728449821472168,
      "learning_rate": 1.2261949982223835e-05,
      "loss": 0.0104,
      "step": 329
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.39955753087997437,
      "learning_rate": 1.2258816337965664e-05,
      "loss": 0.0065,
      "step": 330
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.2015988677740097,
      "learning_rate": 1.2255682693707493e-05,
      "loss": 0.0054,
      "step": 331
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.20565283298492432,
      "learning_rate": 1.2252549049449322e-05,
      "loss": 0.0039,
      "step": 332
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.38143298029899597,
      "learning_rate": 1.224941540519115e-05,
      "loss": 0.0082,
      "step": 333
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2197386473417282,
      "learning_rate": 1.224628176093298e-05,
      "loss": 0.0034,
      "step": 334
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.39900702238082886,
      "learning_rate": 1.2243148116674808e-05,
      "loss": 0.0086,
      "step": 335
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.130195140838623,
      "learning_rate": 1.2240014472416637e-05,
      "loss": 0.0045,
      "step": 336
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.31819453835487366,
      "learning_rate": 1.2236880828158466e-05,
      "loss": 0.0061,
      "step": 337
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0156803131103516,
      "learning_rate": 1.2233747183900295e-05,
      "loss": 0.0104,
      "step": 338
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.5350903868675232,
      "learning_rate": 1.2230613539642122e-05,
      "loss": 0.0077,
      "step": 339
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.40196502208709717,
      "learning_rate": 1.2227479895383953e-05,
      "loss": 0.006,
      "step": 340
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.21756096184253693,
      "learning_rate": 1.2224346251125782e-05,
      "loss": 0.0055,
      "step": 341
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.543333113193512,
      "learning_rate": 1.222121260686761e-05,
      "loss": 0.0106,
      "step": 342
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.591719925403595,
      "learning_rate": 1.2218078962609438e-05,
      "loss": 0.0141,
      "step": 343
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.21601583063602448,
      "learning_rate": 1.2214945318351269e-05,
      "loss": 0.0065,
      "step": 344
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.4998062252998352,
      "learning_rate": 1.2211811674093096e-05,
      "loss": 0.0044,
      "step": 345
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2621588706970215,
      "learning_rate": 1.2208678029834925e-05,
      "loss": 0.0076,
      "step": 346
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2992645800113678,
      "learning_rate": 1.2205544385576754e-05,
      "loss": 0.0095,
      "step": 347
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.1572842001914978,
      "learning_rate": 1.2202410741318584e-05,
      "loss": 0.0039,
      "step": 348
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.561830461025238,
      "learning_rate": 1.2199277097060412e-05,
      "loss": 0.0094,
      "step": 349
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.15858852863311768,
      "learning_rate": 1.219614345280224e-05,
      "loss": 0.0035,
      "step": 350
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.49944812059402466,
      "learning_rate": 1.219300980854407e-05,
      "loss": 0.0063,
      "step": 351
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.2419816255569458,
      "learning_rate": 1.2189876164285897e-05,
      "loss": 0.0066,
      "step": 352
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.3870939314365387,
      "learning_rate": 1.2186742520027727e-05,
      "loss": 0.0035,
      "step": 353
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.2379673421382904,
      "learning_rate": 1.2183608875769556e-05,
      "loss": 0.0033,
      "step": 354
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.2873013913631439,
      "learning_rate": 1.2180475231511384e-05,
      "loss": 0.0045,
      "step": 355
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.5051774382591248,
      "learning_rate": 1.2177341587253213e-05,
      "loss": 0.0068,
      "step": 356
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2737874686717987,
      "learning_rate": 1.2174207942995043e-05,
      "loss": 0.0054,
      "step": 357
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6640222072601318,
      "learning_rate": 1.2171074298736872e-05,
      "loss": 0.0094,
      "step": 358
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.25164270401000977,
      "learning_rate": 1.21679406544787e-05,
      "loss": 0.0043,
      "step": 359
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.19681650400161743,
      "learning_rate": 1.2164807010220528e-05,
      "loss": 0.0046,
      "step": 360
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.20772317051887512,
      "learning_rate": 1.2161673365962359e-05,
      "loss": 0.0037,
      "step": 361
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.22849200665950775,
      "learning_rate": 1.2158539721704186e-05,
      "loss": 0.0047,
      "step": 362
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.535957396030426,
      "learning_rate": 1.2155406077446015e-05,
      "loss": 0.0096,
      "step": 363
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.689008891582489,
      "learning_rate": 1.2152272433187844e-05,
      "loss": 0.0109,
      "step": 364
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.148796081542969,
      "learning_rate": 1.2149138788929673e-05,
      "loss": 0.0128,
      "step": 365
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.19674113392829895,
      "learning_rate": 1.2146005144671502e-05,
      "loss": 0.0027,
      "step": 366
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.14471951127052307,
      "learning_rate": 1.214287150041333e-05,
      "loss": 0.0023,
      "step": 367
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.4351198971271515,
      "learning_rate": 1.213973785615516e-05,
      "loss": 0.0045,
      "step": 368
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.4286715090274811,
      "learning_rate": 1.2136604211896989e-05,
      "loss": 0.0057,
      "step": 369
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3822503089904785,
      "learning_rate": 1.2133470567638817e-05,
      "loss": 0.0093,
      "step": 370
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.5379936099052429,
      "learning_rate": 1.2130336923380646e-05,
      "loss": 0.0062,
      "step": 371
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.32353001832962036,
      "learning_rate": 1.2127203279122474e-05,
      "loss": 0.0047,
      "step": 372
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.44823935627937317,
      "learning_rate": 1.2124069634864304e-05,
      "loss": 0.0076,
      "step": 373
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.29466450214385986,
      "learning_rate": 1.2120935990606133e-05,
      "loss": 0.0076,
      "step": 374
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2160598784685135,
      "learning_rate": 1.211780234634796e-05,
      "loss": 0.0028,
      "step": 375
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.26719558238983154,
      "learning_rate": 1.211466870208979e-05,
      "loss": 0.0064,
      "step": 376
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.18223048746585846,
      "learning_rate": 1.2111535057831618e-05,
      "loss": 0.0025,
      "step": 377
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.6588431000709534,
      "learning_rate": 1.2108401413573447e-05,
      "loss": 0.0209,
      "step": 378
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.400428295135498,
      "learning_rate": 1.2105267769315276e-05,
      "loss": 0.0043,
      "step": 379
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7499300241470337,
      "learning_rate": 1.2102134125057105e-05,
      "loss": 0.0049,
      "step": 380
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8425832390785217,
      "learning_rate": 1.2099000480798934e-05,
      "loss": 0.0082,
      "step": 381
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4039020538330078,
      "learning_rate": 1.2095866836540763e-05,
      "loss": 0.0129,
      "step": 382
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.5410000085830688,
      "learning_rate": 1.2092733192282592e-05,
      "loss": 0.0066,
      "step": 383
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.3800564110279083,
      "learning_rate": 1.208959954802442e-05,
      "loss": 0.0062,
      "step": 384
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.3698054254055023,
      "learning_rate": 1.2086465903766248e-05,
      "loss": 0.0084,
      "step": 385
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.24255406856536865,
      "learning_rate": 1.2083332259508079e-05,
      "loss": 0.0047,
      "step": 386
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.5165615677833557,
      "learning_rate": 1.2080198615249907e-05,
      "loss": 0.0108,
      "step": 387
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.23940302431583405,
      "learning_rate": 1.2077064970991735e-05,
      "loss": 0.0056,
      "step": 388
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2536526322364807,
      "learning_rate": 1.2073931326733564e-05,
      "loss": 0.0042,
      "step": 389
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7617545127868652,
      "learning_rate": 1.2070797682475394e-05,
      "loss": 0.01,
      "step": 390
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.27217668294906616,
      "learning_rate": 1.2067664038217223e-05,
      "loss": 0.0068,
      "step": 391
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.18790052831172943,
      "learning_rate": 1.206453039395905e-05,
      "loss": 0.0044,
      "step": 392
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7538745403289795,
      "learning_rate": 1.206139674970088e-05,
      "loss": 0.0063,
      "step": 393
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8137382864952087,
      "learning_rate": 1.205826310544271e-05,
      "loss": 0.0059,
      "step": 394
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.606963634490967,
      "learning_rate": 1.2055129461184537e-05,
      "loss": 0.0073,
      "step": 395
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6240628361701965,
      "learning_rate": 1.2051995816926366e-05,
      "loss": 0.0132,
      "step": 396
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.1836826205253601,
      "learning_rate": 1.2048862172668195e-05,
      "loss": 0.0043,
      "step": 397
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2188057154417038,
      "learning_rate": 1.2045728528410024e-05,
      "loss": 0.0064,
      "step": 398
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.1544315665960312,
      "learning_rate": 1.2042594884151853e-05,
      "loss": 0.0033,
      "step": 399
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.46250271797180176,
      "learning_rate": 1.2039461239893682e-05,
      "loss": 0.0082,
      "step": 400
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8297383785247803,
      "learning_rate": 1.203632759563551e-05,
      "loss": 0.0154,
      "step": 401
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.5838727951049805,
      "learning_rate": 1.2033193951377338e-05,
      "loss": 0.0066,
      "step": 402
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.16540837287902832,
      "learning_rate": 1.2030060307119169e-05,
      "loss": 0.0041,
      "step": 403
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.4009976089000702,
      "learning_rate": 1.2026926662860997e-05,
      "loss": 0.0088,
      "step": 404
    },
    {
      "epoch": 2.0,
      "eval_AVGf1": 0.932645243203012,
      "eval_DIAG.f1": 0.9258241758241758,
      "eval_DIAG.precision": 0.9335180055401662,
      "eval_DIAG.recall": 0.9182561307901907,
      "eval_MED.f1": 0.9721115537848605,
      "eval_MED.precision": 0.9682539682539683,
      "eval_MED.recall": 0.976,
      "eval_TREAT.f1": 0.8999999999999999,
      "eval_TREAT.precision": 0.9086538461538461,
      "eval_TREAT.recall": 0.8915094339622641,
      "eval_accuracy": 0.9299317283507007,
      "eval_f1": 0.7431192660550459,
      "eval_loss": 0.010707378387451172,
      "eval_precision": 0.6230769230769231,
      "eval_recall": 0.9204545454545454,
      "eval_runtime": 927.947,
      "eval_samples_per_second": 0.774,
      "eval_steps_per_second": 0.097,
      "step": 404
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.186934232711792,
      "learning_rate": 1.2023793018602825e-05,
      "loss": 0.0133,
      "step": 405
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0642361640930176,
      "learning_rate": 1.2020659374344654e-05,
      "loss": 0.0072,
      "step": 406
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.16597823798656464,
      "learning_rate": 1.2017525730086484e-05,
      "loss": 0.0039,
      "step": 407
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.22433412075042725,
      "learning_rate": 1.2014392085828311e-05,
      "loss": 0.0036,
      "step": 408
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.2633878290653229,
      "learning_rate": 1.201125844157014e-05,
      "loss": 0.0054,
      "step": 409
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.16141216456890106,
      "learning_rate": 1.200812479731197e-05,
      "loss": 0.0031,
      "step": 410
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.17187462747097015,
      "learning_rate": 1.2004991153053798e-05,
      "loss": 0.0045,
      "step": 411
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.22352737188339233,
      "learning_rate": 1.2001857508795627e-05,
      "loss": 0.0035,
      "step": 412
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.26726987957954407,
      "learning_rate": 1.1998723864537456e-05,
      "loss": 0.0048,
      "step": 413
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.14169646799564362,
      "learning_rate": 1.1995590220279285e-05,
      "loss": 0.0031,
      "step": 414
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.3768385648727417,
      "learning_rate": 1.1992456576021114e-05,
      "loss": 0.0068,
      "step": 415
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.07901391386985779,
      "learning_rate": 1.1989322931762943e-05,
      "loss": 0.0014,
      "step": 416
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.2737289369106293,
      "learning_rate": 1.1986189287504772e-05,
      "loss": 0.0033,
      "step": 417
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.3107191026210785,
      "learning_rate": 1.1983055643246599e-05,
      "loss": 0.0033,
      "step": 418
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.10891890525817871,
      "learning_rate": 1.197992199898843e-05,
      "loss": 0.0021,
      "step": 419
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.06144319474697113,
      "learning_rate": 1.1976788354730259e-05,
      "loss": 0.001,
      "step": 420
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7917850017547607,
      "learning_rate": 1.1973654710472086e-05,
      "loss": 0.0041,
      "step": 421
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.45074304938316345,
      "learning_rate": 1.1970521066213915e-05,
      "loss": 0.0039,
      "step": 422
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.3067933917045593,
      "learning_rate": 1.1967387421955745e-05,
      "loss": 0.0028,
      "step": 423
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.2231537401676178,
      "learning_rate": 1.1964253777697574e-05,
      "loss": 0.0023,
      "step": 424
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.1787988841533661,
      "learning_rate": 1.1961120133439402e-05,
      "loss": 0.0031,
      "step": 425
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.2547398507595062,
      "learning_rate": 1.195798648918123e-05,
      "loss": 0.0043,
      "step": 426
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.5918833017349243,
      "learning_rate": 1.195485284492306e-05,
      "loss": 0.0064,
      "step": 427
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.47568148374557495,
      "learning_rate": 1.1951719200664888e-05,
      "loss": 0.0079,
      "step": 428
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.629791259765625,
      "learning_rate": 1.1948585556406717e-05,
      "loss": 0.0059,
      "step": 429
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.4274146854877472,
      "learning_rate": 1.1945451912148546e-05,
      "loss": 0.0085,
      "step": 430
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.7731022834777832,
      "learning_rate": 1.1942318267890373e-05,
      "loss": 0.0116,
      "step": 431
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.1572871059179306,
      "learning_rate": 1.1939184623632204e-05,
      "loss": 0.0015,
      "step": 432
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.6938714981079102,
      "learning_rate": 1.1936050979374033e-05,
      "loss": 0.0027,
      "step": 433
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0601812601089478,
      "learning_rate": 1.1932917335115862e-05,
      "loss": 0.0093,
      "step": 434
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.09081868082284927,
      "learning_rate": 1.1929783690857689e-05,
      "loss": 0.0015,
      "step": 435
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.5807023644447327,
      "learning_rate": 1.192665004659952e-05,
      "loss": 0.0084,
      "step": 436
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.3588840663433075,
      "learning_rate": 1.1923516402341349e-05,
      "loss": 0.0022,
      "step": 437
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.1769259124994278,
      "learning_rate": 1.1920382758083176e-05,
      "loss": 0.0035,
      "step": 438
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.4556799829006195,
      "learning_rate": 1.1917249113825005e-05,
      "loss": 0.0075,
      "step": 439
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.26164647936820984,
      "learning_rate": 1.1914115469566835e-05,
      "loss": 0.0027,
      "step": 440
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.24021293222904205,
      "learning_rate": 1.1910981825308663e-05,
      "loss": 0.0044,
      "step": 441
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.7946174740791321,
      "learning_rate": 1.1907848181050492e-05,
      "loss": 0.0085,
      "step": 442
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.507349967956543,
      "learning_rate": 1.190471453679232e-05,
      "loss": 0.0083,
      "step": 443
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.18560951948165894,
      "learning_rate": 1.190158089253415e-05,
      "loss": 0.003,
      "step": 444
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.2595173120498657,
      "learning_rate": 1.1898447248275978e-05,
      "loss": 0.002,
      "step": 445
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.5530809164047241,
      "learning_rate": 1.1895313604017807e-05,
      "loss": 0.0037,
      "step": 446
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.27164676785469055,
      "learning_rate": 1.1892179959759636e-05,
      "loss": 0.0024,
      "step": 447
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.0849897861480713,
      "learning_rate": 1.1889046315501465e-05,
      "loss": 0.0221,
      "step": 448
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.355968952178955,
      "learning_rate": 1.1885912671243294e-05,
      "loss": 0.0069,
      "step": 449
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.1489480435848236,
      "learning_rate": 1.1882779026985123e-05,
      "loss": 0.0022,
      "step": 450
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.42592746019363403,
      "learning_rate": 1.187964538272695e-05,
      "loss": 0.0046,
      "step": 451
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3183438777923584,
      "learning_rate": 1.1876511738468779e-05,
      "loss": 0.0104,
      "step": 452
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.20666322112083435,
      "learning_rate": 1.187337809421061e-05,
      "loss": 0.004,
      "step": 453
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.2114042192697525,
      "learning_rate": 1.1870244449952437e-05,
      "loss": 0.0037,
      "step": 454
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.2127334624528885,
      "learning_rate": 1.1867110805694266e-05,
      "loss": 0.0035,
      "step": 455
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5146394968032837,
      "learning_rate": 1.1863977161436095e-05,
      "loss": 0.0128,
      "step": 456
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.28809142112731934,
      "learning_rate": 1.1860843517177925e-05,
      "loss": 0.0038,
      "step": 457
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.34094706177711487,
      "learning_rate": 1.1857709872919753e-05,
      "loss": 0.0043,
      "step": 458
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.17759913206100464,
      "learning_rate": 1.1854576228661582e-05,
      "loss": 0.0018,
      "step": 459
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.32923200726509094,
      "learning_rate": 1.185144258440341e-05,
      "loss": 0.0035,
      "step": 460
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.12852923572063446,
      "learning_rate": 1.184830894014524e-05,
      "loss": 0.0024,
      "step": 461
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.2824360728263855,
      "learning_rate": 1.1845175295887068e-05,
      "loss": 0.0024,
      "step": 462
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.4875888526439667,
      "learning_rate": 1.1842041651628897e-05,
      "loss": 0.0059,
      "step": 463
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.6058448553085327,
      "learning_rate": 1.1838908007370724e-05,
      "loss": 0.0052,
      "step": 464
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1942172050476074,
      "learning_rate": 1.1835774363112555e-05,
      "loss": 0.009,
      "step": 465
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.5531330108642578,
      "learning_rate": 1.1832640718854384e-05,
      "loss": 0.0053,
      "step": 466
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.5312083959579468,
      "learning_rate": 1.1829507074596213e-05,
      "loss": 0.0036,
      "step": 467
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9267328381538391,
      "learning_rate": 1.182637343033804e-05,
      "loss": 0.011,
      "step": 468
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.12396935373544693,
      "learning_rate": 1.182323978607987e-05,
      "loss": 0.0021,
      "step": 469
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.7546250224113464,
      "learning_rate": 1.18201061418217e-05,
      "loss": 0.0049,
      "step": 470
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.2428586781024933,
      "learning_rate": 1.1816972497563527e-05,
      "loss": 0.0038,
      "step": 471
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.21396037936210632,
      "learning_rate": 1.1813838853305356e-05,
      "loss": 0.003,
      "step": 472
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.2456817477941513,
      "learning_rate": 1.1810705209047187e-05,
      "loss": 0.005,
      "step": 473
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.38682255148887634,
      "learning_rate": 1.1807571564789014e-05,
      "loss": 0.0075,
      "step": 474
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.11746039986610413,
      "learning_rate": 1.1804437920530843e-05,
      "loss": 0.0017,
      "step": 475
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.16832947731018066,
      "learning_rate": 1.1801304276272672e-05,
      "loss": 0.0035,
      "step": 476
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.13616953790187836,
      "learning_rate": 1.17981706320145e-05,
      "loss": 0.002,
      "step": 477
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.1772916316986084,
      "learning_rate": 1.179503698775633e-05,
      "loss": 0.0024,
      "step": 478
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.19214826822280884,
      "learning_rate": 1.1791903343498158e-05,
      "loss": 0.0023,
      "step": 479
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4829951524734497,
      "learning_rate": 1.1788769699239987e-05,
      "loss": 0.0027,
      "step": 480
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.5726773738861084,
      "learning_rate": 1.1785636054981815e-05,
      "loss": 0.0032,
      "step": 481
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.10879381746053696,
      "learning_rate": 1.1782502410723645e-05,
      "loss": 0.0013,
      "step": 482
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.4862184226512909,
      "learning_rate": 1.1779368766465474e-05,
      "loss": 0.0061,
      "step": 483
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.1533650159835815,
      "learning_rate": 1.1776235122207301e-05,
      "loss": 0.007,
      "step": 484
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.39571061730384827,
      "learning_rate": 1.177310147794913e-05,
      "loss": 0.0099,
      "step": 485
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.21099461615085602,
      "learning_rate": 1.176996783369096e-05,
      "loss": 0.0036,
      "step": 486
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.21915002167224884,
      "learning_rate": 1.1766834189432788e-05,
      "loss": 0.002,
      "step": 487
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.1712527722120285,
      "learning_rate": 1.1763700545174617e-05,
      "loss": 0.0039,
      "step": 488
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.1915467530488968,
      "learning_rate": 1.1760566900916446e-05,
      "loss": 0.0034,
      "step": 489
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.32483038306236267,
      "learning_rate": 1.1757433256658277e-05,
      "loss": 0.0059,
      "step": 490
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.2606416940689087,
      "learning_rate": 1.1754299612400104e-05,
      "loss": 0.0046,
      "step": 491
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.15190906822681427,
      "learning_rate": 1.1751165968141933e-05,
      "loss": 0.0012,
      "step": 492
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.08609986305236816,
      "learning_rate": 1.1748032323883762e-05,
      "loss": 0.0011,
      "step": 493
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.6363046765327454,
      "learning_rate": 1.174489867962559e-05,
      "loss": 0.0043,
      "step": 494
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.32831451296806335,
      "learning_rate": 1.174176503536742e-05,
      "loss": 0.0023,
      "step": 495
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.359994411468506,
      "learning_rate": 1.1738631391109248e-05,
      "loss": 0.006,
      "step": 496
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.0524345636367798,
      "learning_rate": 1.1735497746851076e-05,
      "loss": 0.0052,
      "step": 497
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.4743441939353943,
      "learning_rate": 1.1732364102592906e-05,
      "loss": 0.003,
      "step": 498
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.18251554667949677,
      "learning_rate": 1.1729230458334735e-05,
      "loss": 0.002,
      "step": 499
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0514787435531616,
      "learning_rate": 1.1726096814076564e-05,
      "loss": 0.0193,
      "step": 500
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.1741778552532196,
      "learning_rate": 1.1722963169818391e-05,
      "loss": 0.0028,
      "step": 501
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.5104990005493164,
      "learning_rate": 1.171982952556022e-05,
      "loss": 0.0073,
      "step": 502
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.1928783506155014,
      "learning_rate": 1.1716695881302051e-05,
      "loss": 0.0021,
      "step": 503
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.6509090065956116,
      "learning_rate": 1.1713562237043878e-05,
      "loss": 0.0051,
      "step": 504
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.18126338720321655,
      "learning_rate": 1.1710428592785707e-05,
      "loss": 0.0028,
      "step": 505
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0450201034545898,
      "learning_rate": 1.1707294948527536e-05,
      "loss": 0.0079,
      "step": 506
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.17173361778259277,
      "learning_rate": 1.1704161304269365e-05,
      "loss": 0.0019,
      "step": 507
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.44778165221214294,
      "learning_rate": 1.1701027660011194e-05,
      "loss": 0.0077,
      "step": 508
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.40937212109565735,
      "learning_rate": 1.1697894015753023e-05,
      "loss": 0.0015,
      "step": 509
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.5868229866027832,
      "learning_rate": 1.1694760371494852e-05,
      "loss": 0.0037,
      "step": 510
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8154226541519165,
      "learning_rate": 1.169162672723668e-05,
      "loss": 0.0034,
      "step": 511
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9846629500389099,
      "learning_rate": 1.168849308297851e-05,
      "loss": 0.0103,
      "step": 512
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.09263516962528229,
      "learning_rate": 1.1685359438720338e-05,
      "loss": 0.0007,
      "step": 513
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7513229846954346,
      "learning_rate": 1.1682225794462166e-05,
      "loss": 0.0058,
      "step": 514
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.6275089383125305,
      "learning_rate": 1.1679092150203996e-05,
      "loss": 0.0092,
      "step": 515
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.1911066174507141,
      "learning_rate": 1.1675958505945825e-05,
      "loss": 0.0025,
      "step": 516
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.10452564805746078,
      "learning_rate": 1.1672824861687652e-05,
      "loss": 0.002,
      "step": 517
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9370364546775818,
      "learning_rate": 1.1669691217429481e-05,
      "loss": 0.0181,
      "step": 518
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.21984761953353882,
      "learning_rate": 1.1666557573171312e-05,
      "loss": 0.005,
      "step": 519
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.503592610359192,
      "learning_rate": 1.166342392891314e-05,
      "loss": 0.0043,
      "step": 520
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.7691473364830017,
      "learning_rate": 1.1660290284654968e-05,
      "loss": 0.0065,
      "step": 521
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.15278437733650208,
      "learning_rate": 1.1657156640396797e-05,
      "loss": 0.0028,
      "step": 522
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.19572284817695618,
      "learning_rate": 1.1654022996138628e-05,
      "loss": 0.0015,
      "step": 523
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.22262245416641235,
      "learning_rate": 1.1650889351880455e-05,
      "loss": 0.0039,
      "step": 524
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6352301836013794,
      "learning_rate": 1.1647755707622284e-05,
      "loss": 0.0114,
      "step": 525
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.10295292735099792,
      "learning_rate": 1.1644622063364113e-05,
      "loss": 0.0014,
      "step": 526
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.26818275451660156,
      "learning_rate": 1.164148841910594e-05,
      "loss": 0.0049,
      "step": 527
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.14778026938438416,
      "learning_rate": 1.163835477484777e-05,
      "loss": 0.0029,
      "step": 528
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.34743914008140564,
      "learning_rate": 1.16352211305896e-05,
      "loss": 0.0028,
      "step": 529
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.16241776943206787,
      "learning_rate": 1.1632087486331427e-05,
      "loss": 0.0052,
      "step": 530
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.5407586693763733,
      "learning_rate": 1.1628953842073256e-05,
      "loss": 0.0049,
      "step": 531
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.7160522937774658,
      "learning_rate": 1.1625820197815086e-05,
      "loss": 0.0071,
      "step": 532
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.17300841212272644,
      "learning_rate": 1.1622686553556915e-05,
      "loss": 0.0049,
      "step": 533
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.3690812587738037,
      "learning_rate": 1.1619552909298742e-05,
      "loss": 0.0201,
      "step": 534
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9223749041557312,
      "learning_rate": 1.1616419265040571e-05,
      "loss": 0.0077,
      "step": 535
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8624359965324402,
      "learning_rate": 1.1613285620782402e-05,
      "loss": 0.0139,
      "step": 536
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.3416508436203003,
      "learning_rate": 1.161015197652423e-05,
      "loss": 0.0032,
      "step": 537
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.29625454545021057,
      "learning_rate": 1.1607018332266058e-05,
      "loss": 0.0035,
      "step": 538
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.13982036709785461,
      "learning_rate": 1.1603884688007887e-05,
      "loss": 0.0036,
      "step": 539
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.16581672430038452,
      "learning_rate": 1.1600751043749716e-05,
      "loss": 0.0038,
      "step": 540
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.255969524383545,
      "learning_rate": 1.1597617399491545e-05,
      "loss": 0.025,
      "step": 541
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.22581994533538818,
      "learning_rate": 1.1594483755233374e-05,
      "loss": 0.0041,
      "step": 542
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.22700616717338562,
      "learning_rate": 1.1591350110975203e-05,
      "loss": 0.0034,
      "step": 543
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.2995148301124573,
      "learning_rate": 1.1588216466717032e-05,
      "loss": 0.0047,
      "step": 544
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.890937864780426,
      "learning_rate": 1.158508282245886e-05,
      "loss": 0.0077,
      "step": 545
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.85826575756073,
      "learning_rate": 1.158194917820069e-05,
      "loss": 0.0133,
      "step": 546
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.1715591996908188,
      "learning_rate": 1.1578815533942517e-05,
      "loss": 0.0051,
      "step": 547
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.1344786435365677,
      "learning_rate": 1.1575681889684347e-05,
      "loss": 0.0033,
      "step": 548
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.12326278537511826,
      "learning_rate": 1.1572548245426176e-05,
      "loss": 0.0032,
      "step": 549
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.21264688670635223,
      "learning_rate": 1.1569414601168004e-05,
      "loss": 0.004,
      "step": 550
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.33767762780189514,
      "learning_rate": 1.1566280956909832e-05,
      "loss": 0.0044,
      "step": 551
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.2840878963470459,
      "learning_rate": 1.1563147312651661e-05,
      "loss": 0.0035,
      "step": 552
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.39995479583740234,
      "learning_rate": 1.156001366839349e-05,
      "loss": 0.0045,
      "step": 553
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.22125373780727386,
      "learning_rate": 1.155688002413532e-05,
      "loss": 0.0043,
      "step": 554
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9674707055091858,
      "learning_rate": 1.1553746379877148e-05,
      "loss": 0.0059,
      "step": 555
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.23246456682682037,
      "learning_rate": 1.1550612735618977e-05,
      "loss": 0.004,
      "step": 556
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.13465391099452972,
      "learning_rate": 1.1547479091360806e-05,
      "loss": 0.0027,
      "step": 557
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.14751525223255157,
      "learning_rate": 1.1544345447102635e-05,
      "loss": 0.0021,
      "step": 558
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.024561882019043,
      "learning_rate": 1.1541211802844464e-05,
      "loss": 0.0108,
      "step": 559
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.10713528096675873,
      "learning_rate": 1.1538078158586291e-05,
      "loss": 0.0015,
      "step": 560
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.6990461945533752,
      "learning_rate": 1.1534944514328122e-05,
      "loss": 0.0052,
      "step": 561
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.2596141993999481,
      "learning_rate": 1.153181087006995e-05,
      "loss": 0.0053,
      "step": 562
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.11822185665369034,
      "learning_rate": 1.1528677225811778e-05,
      "loss": 0.0021,
      "step": 563
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.3999606668949127,
      "learning_rate": 1.1525543581553607e-05,
      "loss": 0.0055,
      "step": 564
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.2715095579624176,
      "learning_rate": 1.1522409937295437e-05,
      "loss": 0.0041,
      "step": 565
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8980044722557068,
      "learning_rate": 1.1519276293037266e-05,
      "loss": 0.0063,
      "step": 566
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.6569909453392029,
      "learning_rate": 1.1516142648779094e-05,
      "loss": 0.006,
      "step": 567
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.7011352181434631,
      "learning_rate": 1.1513009004520922e-05,
      "loss": 0.0066,
      "step": 568
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.15736383199691772,
      "learning_rate": 1.1509875360262753e-05,
      "loss": 0.004,
      "step": 569
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.10100263357162476,
      "learning_rate": 1.150674171600458e-05,
      "loss": 0.0016,
      "step": 570
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.2181003987789154,
      "learning_rate": 1.150360807174641e-05,
      "loss": 0.0045,
      "step": 571
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.23993754386901855,
      "learning_rate": 1.1500474427488238e-05,
      "loss": 0.0038,
      "step": 572
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.3511224091053009,
      "learning_rate": 1.1497340783230067e-05,
      "loss": 0.0044,
      "step": 573
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.15094974637031555,
      "learning_rate": 1.1494207138971896e-05,
      "loss": 0.002,
      "step": 574
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5774611234664917,
      "learning_rate": 1.1491073494713725e-05,
      "loss": 0.0092,
      "step": 575
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.4431806802749634,
      "learning_rate": 1.1487939850455554e-05,
      "loss": 0.0032,
      "step": 576
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.3538781702518463,
      "learning_rate": 1.1484806206197381e-05,
      "loss": 0.0027,
      "step": 577
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9895443916320801,
      "learning_rate": 1.1481672561939212e-05,
      "loss": 0.0197,
      "step": 578
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.13569994270801544,
      "learning_rate": 1.147853891768104e-05,
      "loss": 0.002,
      "step": 579
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.15131524205207825,
      "learning_rate": 1.1475405273422868e-05,
      "loss": 0.0034,
      "step": 580
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.24596454203128815,
      "learning_rate": 1.1472271629164697e-05,
      "loss": 0.0038,
      "step": 581
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.1823067665100098,
      "learning_rate": 1.1469137984906527e-05,
      "loss": 0.0041,
      "step": 582
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.10222851485013962,
      "learning_rate": 1.1466004340648355e-05,
      "loss": 0.0012,
      "step": 583
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.7349010109901428,
      "learning_rate": 1.1462870696390184e-05,
      "loss": 0.0045,
      "step": 584
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8193291425704956,
      "learning_rate": 1.1459737052132013e-05,
      "loss": 0.0063,
      "step": 585
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.4586693048477173,
      "learning_rate": 1.1456603407873841e-05,
      "loss": 0.0066,
      "step": 586
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.14474442601203918,
      "learning_rate": 1.145346976361567e-05,
      "loss": 0.0032,
      "step": 587
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.6175042390823364,
      "learning_rate": 1.14503361193575e-05,
      "loss": 0.008,
      "step": 588
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.3044759929180145,
      "learning_rate": 1.1447202475099328e-05,
      "loss": 0.0072,
      "step": 589
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.48347410559654236,
      "learning_rate": 1.1444068830841157e-05,
      "loss": 0.0079,
      "step": 590
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.1298187971115112,
      "learning_rate": 1.1440935186582986e-05,
      "loss": 0.0232,
      "step": 591
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.3358183205127716,
      "learning_rate": 1.1437801542324815e-05,
      "loss": 0.0037,
      "step": 592
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.8314073085784912,
      "learning_rate": 1.1434667898066642e-05,
      "loss": 0.0075,
      "step": 593
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.134234428405762,
      "learning_rate": 1.1431534253808473e-05,
      "loss": 0.0062,
      "step": 594
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.5944139361381531,
      "learning_rate": 1.1428400609550302e-05,
      "loss": 0.0056,
      "step": 595
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.1745123714208603,
      "learning_rate": 1.1425266965292129e-05,
      "loss": 0.0032,
      "step": 596
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.575960099697113,
      "learning_rate": 1.1422133321033958e-05,
      "loss": 0.0067,
      "step": 597
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9467850923538208,
      "learning_rate": 1.1418999676775789e-05,
      "loss": 0.0092,
      "step": 598
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.2177359014749527,
      "learning_rate": 1.1415866032517617e-05,
      "loss": 0.003,
      "step": 599
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.4297947883605957,
      "learning_rate": 1.1412732388259445e-05,
      "loss": 0.005,
      "step": 600
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.7549340724945068,
      "learning_rate": 1.1409598744001274e-05,
      "loss": 0.0066,
      "step": 601
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.7796575427055359,
      "learning_rate": 1.1406465099743103e-05,
      "loss": 0.0057,
      "step": 602
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.15445953607559204,
      "learning_rate": 1.1403331455484931e-05,
      "loss": 0.003,
      "step": 603
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.22926171123981476,
      "learning_rate": 1.140019781122676e-05,
      "loss": 0.0049,
      "step": 604
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.26526951789855957,
      "learning_rate": 1.139706416696859e-05,
      "loss": 0.0042,
      "step": 605
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.2891643047332764,
      "learning_rate": 1.1393930522710417e-05,
      "loss": 0.0056,
      "step": 606
    },
    {
      "epoch": 3.0,
      "eval_AVGf1": 0.9195884504490279,
      "eval_DIAG.f1": 0.8802228412256266,
      "eval_DIAG.precision": 0.9002849002849003,
      "eval_DIAG.recall": 0.8610354223433242,
      "eval_MED.f1": 0.9554655870445343,
      "eval_MED.precision": 0.9672131147540983,
      "eval_MED.recall": 0.944,
      "eval_TREAT.f1": 0.923076923076923,
      "eval_TREAT.precision": 0.9411764705882353,
      "eval_TREAT.recall": 0.9056603773584906,
      "eval_accuracy": 0.9390945023356091,
      "eval_f1": 0.7638804148871263,
      "eval_loss": 0.015417341142892838,
      "eval_precision": 0.6695187165775401,
      "eval_recall": 0.8892045454545454,
      "eval_runtime": 950.0777,
      "eval_samples_per_second": 0.756,
      "eval_steps_per_second": 0.095,
      "step": 606
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.20340119302272797,
      "learning_rate": 1.1390796878452247e-05,
      "loss": 0.0042,
      "step": 607
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.3265801668167114,
      "learning_rate": 1.1387663234194076e-05,
      "loss": 0.0039,
      "step": 608
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.16601444780826569,
      "learning_rate": 1.1384529589935905e-05,
      "loss": 0.0021,
      "step": 609
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.2744130492210388,
      "learning_rate": 1.1381395945677732e-05,
      "loss": 0.0028,
      "step": 610
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.286461740732193,
      "learning_rate": 1.1378262301419563e-05,
      "loss": 0.003,
      "step": 611
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.12513302266597748,
      "learning_rate": 1.1375128657161392e-05,
      "loss": 0.0015,
      "step": 612
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.12108122557401657,
      "learning_rate": 1.1371995012903219e-05,
      "loss": 0.0012,
      "step": 613
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.1972023993730545,
      "learning_rate": 1.1368861368645048e-05,
      "loss": 0.0013,
      "step": 614
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.14991775155067444,
      "learning_rate": 1.1365727724386879e-05,
      "loss": 0.0016,
      "step": 615
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.18378549814224243,
      "learning_rate": 1.1362594080128706e-05,
      "loss": 0.0032,
      "step": 616
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.4889930486679077,
      "learning_rate": 1.1359460435870535e-05,
      "loss": 0.0032,
      "step": 617
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.29404857754707336,
      "learning_rate": 1.1356326791612364e-05,
      "loss": 0.0038,
      "step": 618
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9558826684951782,
      "learning_rate": 1.1353193147354193e-05,
      "loss": 0.0043,
      "step": 619
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.09746597707271576,
      "learning_rate": 1.1350059503096021e-05,
      "loss": 0.0012,
      "step": 620
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.2898968458175659,
      "learning_rate": 1.134692585883785e-05,
      "loss": 0.0052,
      "step": 621
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0910760164260864,
      "learning_rate": 1.134379221457968e-05,
      "loss": 0.0171,
      "step": 622
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.12335193902254105,
      "learning_rate": 1.1340658570321508e-05,
      "loss": 0.0017,
      "step": 623
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.780931293964386,
      "learning_rate": 1.1337524926063337e-05,
      "loss": 0.0097,
      "step": 624
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.14997728168964386,
      "learning_rate": 1.1334391281805166e-05,
      "loss": 0.002,
      "step": 625
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.08798888325691223,
      "learning_rate": 1.1331257637546993e-05,
      "loss": 0.0014,
      "step": 626
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.18135200440883636,
      "learning_rate": 1.1328123993288822e-05,
      "loss": 0.0024,
      "step": 627
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.14975248277187347,
      "learning_rate": 1.1324990349030653e-05,
      "loss": 0.0016,
      "step": 628
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.04505351558327675,
      "learning_rate": 1.132185670477248e-05,
      "loss": 0.0004,
      "step": 629
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.3864043653011322,
      "learning_rate": 1.1318723060514309e-05,
      "loss": 0.003,
      "step": 630
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.14274364709854126,
      "learning_rate": 1.1315589416256138e-05,
      "loss": 0.0011,
      "step": 631
    },
    {
      "epoch": 3.13,
      "grad_norm": 6.088502883911133,
      "learning_rate": 1.1312455771997969e-05,
      "loss": 0.0089,
      "step": 632
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.4871140122413635,
      "learning_rate": 1.1309322127739796e-05,
      "loss": 0.0026,
      "step": 633
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.0961884260177612,
      "learning_rate": 1.1306188483481625e-05,
      "loss": 0.0017,
      "step": 634
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.12078428268432617,
      "learning_rate": 1.1303054839223454e-05,
      "loss": 0.0019,
      "step": 635
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.11315777152776718,
      "learning_rate": 1.1299921194965283e-05,
      "loss": 0.0016,
      "step": 636
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.13208875060081482,
      "learning_rate": 1.1296787550707112e-05,
      "loss": 0.0013,
      "step": 637
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.1346849501132965,
      "learning_rate": 1.129365390644894e-05,
      "loss": 0.0024,
      "step": 638
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.14288070797920227,
      "learning_rate": 1.1290520262190768e-05,
      "loss": 0.0013,
      "step": 639
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.15530893206596375,
      "learning_rate": 1.1287386617932598e-05,
      "loss": 0.0026,
      "step": 640
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.6599082350730896,
      "learning_rate": 1.1284252973674427e-05,
      "loss": 0.0043,
      "step": 641
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.9387555122375488,
      "learning_rate": 1.1281119329416256e-05,
      "loss": 0.0054,
      "step": 642
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.13693058490753174,
      "learning_rate": 1.1277985685158083e-05,
      "loss": 0.0019,
      "step": 643
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.15290617942810059,
      "learning_rate": 1.1274852040899914e-05,
      "loss": 0.0024,
      "step": 644
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.3050028085708618,
      "learning_rate": 1.1271718396641743e-05,
      "loss": 0.0012,
      "step": 645
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.09243263304233551,
      "learning_rate": 1.126858475238357e-05,
      "loss": 0.0012,
      "step": 646
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.248701810836792,
      "learning_rate": 1.1265451108125399e-05,
      "loss": 0.0038,
      "step": 647
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.2949126958847046,
      "learning_rate": 1.126231746386723e-05,
      "loss": 0.0039,
      "step": 648
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.14467991888523102,
      "learning_rate": 1.1259183819609057e-05,
      "loss": 0.0018,
      "step": 649
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.1594603806734085,
      "learning_rate": 1.1256050175350886e-05,
      "loss": 0.0026,
      "step": 650
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.2044917643070221,
      "learning_rate": 1.1252916531092715e-05,
      "loss": 0.0034,
      "step": 651
    },
    {
      "epoch": 3.23,
      "grad_norm": 3.5424976348876953,
      "learning_rate": 1.1249782886834544e-05,
      "loss": 0.0185,
      "step": 652
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.2858648896217346,
      "learning_rate": 1.1246649242576373e-05,
      "loss": 0.0019,
      "step": 653
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.7578615546226501,
      "learning_rate": 1.1243515598318202e-05,
      "loss": 0.0045,
      "step": 654
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.3441081643104553,
      "learning_rate": 1.124038195406003e-05,
      "loss": 0.0031,
      "step": 655
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6860391497612,
      "learning_rate": 1.1237248309801858e-05,
      "loss": 0.0052,
      "step": 656
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.8070695996284485,
      "learning_rate": 1.1234114665543688e-05,
      "loss": 0.006,
      "step": 657
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.943175196647644,
      "learning_rate": 1.1230981021285517e-05,
      "loss": 0.0126,
      "step": 658
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8102656006813049,
      "learning_rate": 1.1227847377027344e-05,
      "loss": 0.0054,
      "step": 659
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.16605380177497864,
      "learning_rate": 1.1224713732769173e-05,
      "loss": 0.0022,
      "step": 660
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.13527774810791016,
      "learning_rate": 1.1221580088511004e-05,
      "loss": 0.0027,
      "step": 661
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.13172458112239838,
      "learning_rate": 1.1218446444252831e-05,
      "loss": 0.001,
      "step": 662
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.5032154321670532,
      "learning_rate": 1.121531279999466e-05,
      "loss": 0.0069,
      "step": 663
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.3353384733200073,
      "learning_rate": 1.1212179155736489e-05,
      "loss": 0.0037,
      "step": 664
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.10639052093029022,
      "learning_rate": 1.120904551147832e-05,
      "loss": 0.0026,
      "step": 665
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.35866278409957886,
      "learning_rate": 1.1205911867220147e-05,
      "loss": 0.0045,
      "step": 666
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.7908288240432739,
      "learning_rate": 1.1202778222961976e-05,
      "loss": 0.0072,
      "step": 667
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.28295791149139404,
      "learning_rate": 1.1199644578703805e-05,
      "loss": 0.0028,
      "step": 668
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.4234902262687683,
      "learning_rate": 1.1196510934445634e-05,
      "loss": 0.0029,
      "step": 669
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.31143835186958313,
      "learning_rate": 1.1193377290187463e-05,
      "loss": 0.0076,
      "step": 670
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.08443977683782578,
      "learning_rate": 1.1190243645929292e-05,
      "loss": 0.001,
      "step": 671
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.1352209746837616,
      "learning_rate": 1.1187110001671119e-05,
      "loss": 0.0015,
      "step": 672
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.11498111486434937,
      "learning_rate": 1.118397635741295e-05,
      "loss": 0.0018,
      "step": 673
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.1816299855709076,
      "learning_rate": 1.1180842713154778e-05,
      "loss": 0.0026,
      "step": 674
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.1718875765800476,
      "learning_rate": 1.1177709068896607e-05,
      "loss": 0.0032,
      "step": 675
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.34595584869384766,
      "learning_rate": 1.1174575424638434e-05,
      "loss": 0.0023,
      "step": 676
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.12826266884803772,
      "learning_rate": 1.1171441780380263e-05,
      "loss": 0.0021,
      "step": 677
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.08362850546836853,
      "learning_rate": 1.1168308136122094e-05,
      "loss": 0.0006,
      "step": 678
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.16141323745250702,
      "learning_rate": 1.1165174491863921e-05,
      "loss": 0.002,
      "step": 679
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.29531630873680115,
      "learning_rate": 1.116204084760575e-05,
      "loss": 0.0033,
      "step": 680
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.13579827547073364,
      "learning_rate": 1.1158907203347579e-05,
      "loss": 0.002,
      "step": 681
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.4698866307735443,
      "learning_rate": 1.1155773559089408e-05,
      "loss": 0.0069,
      "step": 682
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.33812540769577026,
      "learning_rate": 1.1152639914831237e-05,
      "loss": 0.0074,
      "step": 683
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.1119675412774086,
      "learning_rate": 1.1149506270573066e-05,
      "loss": 0.0014,
      "step": 684
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.15889808535575867,
      "learning_rate": 1.1146372626314895e-05,
      "loss": 0.0022,
      "step": 685
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.7536487579345703,
      "learning_rate": 1.1143238982056724e-05,
      "loss": 0.0045,
      "step": 686
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.06529586017131805,
      "learning_rate": 1.1140105337798553e-05,
      "loss": 0.001,
      "step": 687
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.11882104724645615,
      "learning_rate": 1.1136971693540382e-05,
      "loss": 0.0014,
      "step": 688
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1629424095153809,
      "learning_rate": 1.1133838049282209e-05,
      "loss": 0.0052,
      "step": 689
    },
    {
      "epoch": 3.42,
      "grad_norm": 3.0102016925811768,
      "learning_rate": 1.113070440502404e-05,
      "loss": 0.0139,
      "step": 690
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.11936141550540924,
      "learning_rate": 1.1127570760765868e-05,
      "loss": 0.0008,
      "step": 691
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.07433454692363739,
      "learning_rate": 1.1124437116507696e-05,
      "loss": 0.0013,
      "step": 692
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.13326257467269897,
      "learning_rate": 1.1121303472249525e-05,
      "loss": 0.0019,
      "step": 693
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.4734951853752136,
      "learning_rate": 1.1118169827991355e-05,
      "loss": 0.0037,
      "step": 694
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.14695709943771362,
      "learning_rate": 1.1115036183733182e-05,
      "loss": 0.0014,
      "step": 695
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.15714237093925476,
      "learning_rate": 1.1111902539475011e-05,
      "loss": 0.0016,
      "step": 696
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6829473376274109,
      "learning_rate": 1.110876889521684e-05,
      "loss": 0.0029,
      "step": 697
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.11529865860939026,
      "learning_rate": 1.1105635250958669e-05,
      "loss": 0.0014,
      "step": 698
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.23391352593898773,
      "learning_rate": 1.1102501606700498e-05,
      "loss": 0.0042,
      "step": 699
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.1852315217256546,
      "learning_rate": 1.1099367962442327e-05,
      "loss": 0.0031,
      "step": 700
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.17035040259361267,
      "learning_rate": 1.1096234318184156e-05,
      "loss": 0.0014,
      "step": 701
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.15209169685840607,
      "learning_rate": 1.1093100673925983e-05,
      "loss": 0.0023,
      "step": 702
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.8433559536933899,
      "learning_rate": 1.1089967029667814e-05,
      "loss": 0.0085,
      "step": 703
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.5330628156661987,
      "learning_rate": 1.1086833385409643e-05,
      "loss": 0.0068,
      "step": 704
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.6550683975219727,
      "learning_rate": 1.108369974115147e-05,
      "loss": 0.003,
      "step": 705
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.13696719706058502,
      "learning_rate": 1.1080566096893299e-05,
      "loss": 0.0027,
      "step": 706
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.2149571031332016,
      "learning_rate": 1.107743245263513e-05,
      "loss": 0.0029,
      "step": 707
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.2994069457054138,
      "learning_rate": 1.1074298808376958e-05,
      "loss": 0.0033,
      "step": 708
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.4017512798309326,
      "learning_rate": 1.1071165164118786e-05,
      "loss": 0.0036,
      "step": 709
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.2800564467906952,
      "learning_rate": 1.1068031519860615e-05,
      "loss": 0.0038,
      "step": 710
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.2627919018268585,
      "learning_rate": 1.1064897875602445e-05,
      "loss": 0.0036,
      "step": 711
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.869141697883606,
      "learning_rate": 1.1061764231344272e-05,
      "loss": 0.0128,
      "step": 712
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.20604509115219116,
      "learning_rate": 1.1058630587086101e-05,
      "loss": 0.0015,
      "step": 713
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.16295376420021057,
      "learning_rate": 1.105549694282793e-05,
      "loss": 0.0018,
      "step": 714
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.07981301099061966,
      "learning_rate": 1.1052363298569759e-05,
      "loss": 0.0007,
      "step": 715
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.10447100549936295,
      "learning_rate": 1.1049229654311588e-05,
      "loss": 0.0026,
      "step": 716
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.5674294233322144,
      "learning_rate": 1.1046096010053417e-05,
      "loss": 0.0046,
      "step": 717
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.2773357033729553,
      "learning_rate": 1.1042962365795246e-05,
      "loss": 0.0038,
      "step": 718
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.15468092262744904,
      "learning_rate": 1.1039828721537075e-05,
      "loss": 0.0028,
      "step": 719
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.2658460736274719,
      "learning_rate": 1.1036695077278904e-05,
      "loss": 0.0025,
      "step": 720
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.12997522950172424,
      "learning_rate": 1.1033561433020733e-05,
      "loss": 0.0018,
      "step": 721
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.12985467910766602,
      "learning_rate": 1.103042778876256e-05,
      "loss": 0.0016,
      "step": 722
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.22345779836177826,
      "learning_rate": 1.102729414450439e-05,
      "loss": 0.0031,
      "step": 723
    },
    {
      "epoch": 3.58,
      "grad_norm": 3.8086893558502197,
      "learning_rate": 1.102416050024622e-05,
      "loss": 0.0101,
      "step": 724
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.475942462682724,
      "learning_rate": 1.1021026855988047e-05,
      "loss": 0.0029,
      "step": 725
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.6605396866798401,
      "learning_rate": 1.1017893211729876e-05,
      "loss": 0.0029,
      "step": 726
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.18381348252296448,
      "learning_rate": 1.1014759567471705e-05,
      "loss": 0.0024,
      "step": 727
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.16097219288349152,
      "learning_rate": 1.1011625923213533e-05,
      "loss": 0.0053,
      "step": 728
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.6381688117980957,
      "learning_rate": 1.1008492278955362e-05,
      "loss": 0.0084,
      "step": 729
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.19089730083942413,
      "learning_rate": 1.1005358634697191e-05,
      "loss": 0.0035,
      "step": 730
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.4287155568599701,
      "learning_rate": 1.100222499043902e-05,
      "loss": 0.0041,
      "step": 731
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.34275293350219727,
      "learning_rate": 1.099909134618085e-05,
      "loss": 0.0019,
      "step": 732
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.08921957015991211,
      "learning_rate": 1.0995957701922678e-05,
      "loss": 0.0012,
      "step": 733
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.3156415522098541,
      "learning_rate": 1.0992824057664507e-05,
      "loss": 0.0036,
      "step": 734
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.20399633049964905,
      "learning_rate": 1.0989690413406334e-05,
      "loss": 0.0014,
      "step": 735
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.3836483955383301,
      "learning_rate": 1.0986556769148165e-05,
      "loss": 0.002,
      "step": 736
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.17072108387947083,
      "learning_rate": 1.0983423124889994e-05,
      "loss": 0.0021,
      "step": 737
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.18817174434661865,
      "learning_rate": 1.0980289480631821e-05,
      "loss": 0.0045,
      "step": 738
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.10868512839078903,
      "learning_rate": 1.097715583637365e-05,
      "loss": 0.0012,
      "step": 739
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.14934095740318298,
      "learning_rate": 1.097402219211548e-05,
      "loss": 0.0019,
      "step": 740
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.4736967086791992,
      "learning_rate": 1.097088854785731e-05,
      "loss": 0.0039,
      "step": 741
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.966816782951355,
      "learning_rate": 1.0967754903599137e-05,
      "loss": 0.0094,
      "step": 742
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.10213013738393784,
      "learning_rate": 1.0964621259340966e-05,
      "loss": 0.0018,
      "step": 743
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.12555892765522003,
      "learning_rate": 1.0961487615082796e-05,
      "loss": 0.0016,
      "step": 744
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.14296932518482208,
      "learning_rate": 1.0958353970824624e-05,
      "loss": 0.0018,
      "step": 745
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.3138982951641083,
      "learning_rate": 1.0955220326566452e-05,
      "loss": 0.0031,
      "step": 746
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.2067403644323349,
      "learning_rate": 1.0952086682308281e-05,
      "loss": 0.0027,
      "step": 747
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.29844823479652405,
      "learning_rate": 1.094895303805011e-05,
      "loss": 0.0023,
      "step": 748
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7445510029792786,
      "learning_rate": 1.094581939379194e-05,
      "loss": 0.0262,
      "step": 749
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.9664465188980103,
      "learning_rate": 1.0942685749533768e-05,
      "loss": 0.0049,
      "step": 750
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.141004279255867,
      "learning_rate": 1.0939552105275597e-05,
      "loss": 0.0012,
      "step": 751
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.1615574061870575,
      "learning_rate": 1.0936418461017424e-05,
      "loss": 0.0039,
      "step": 752
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.0874723568558693,
      "learning_rate": 1.0933284816759255e-05,
      "loss": 0.0015,
      "step": 753
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.5668485760688782,
      "learning_rate": 1.0930151172501084e-05,
      "loss": 0.0044,
      "step": 754
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.3322218954563141,
      "learning_rate": 1.0927017528242911e-05,
      "loss": 0.0026,
      "step": 755
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.8034695386886597,
      "learning_rate": 1.092388388398474e-05,
      "loss": 0.0039,
      "step": 756
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.038783073425293,
      "learning_rate": 1.092075023972657e-05,
      "loss": 0.0051,
      "step": 757
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.24261260032653809,
      "learning_rate": 1.0917616595468398e-05,
      "loss": 0.0028,
      "step": 758
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.8278446793556213,
      "learning_rate": 1.0914482951210227e-05,
      "loss": 0.0052,
      "step": 759
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.3076862096786499,
      "learning_rate": 1.0911349306952056e-05,
      "loss": 0.0025,
      "step": 760
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.5605326890945435,
      "learning_rate": 1.0908215662693885e-05,
      "loss": 0.0042,
      "step": 761
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.1564527451992035,
      "learning_rate": 1.0905082018435714e-05,
      "loss": 0.0022,
      "step": 762
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.2191602885723114,
      "learning_rate": 1.0901948374177542e-05,
      "loss": 0.0028,
      "step": 763
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.12013883143663406,
      "learning_rate": 1.0898814729919371e-05,
      "loss": 0.003,
      "step": 764
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.27376076579093933,
      "learning_rate": 1.08956810856612e-05,
      "loss": 0.0006,
      "step": 765
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.37962606549263,
      "learning_rate": 1.089254744140303e-05,
      "loss": 0.0028,
      "step": 766
    },
    {
      "epoch": 3.8,
      "grad_norm": 2.2985994815826416,
      "learning_rate": 1.0889413797144858e-05,
      "loss": 0.0037,
      "step": 767
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.1201179027557373,
      "learning_rate": 1.0886280152886685e-05,
      "loss": 0.0021,
      "step": 768
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.10593807697296143,
      "learning_rate": 1.0883146508628516e-05,
      "loss": 0.0009,
      "step": 769
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.23085354268550873,
      "learning_rate": 1.0880012864370345e-05,
      "loss": 0.0029,
      "step": 770
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.39966341853141785,
      "learning_rate": 1.0876879220112172e-05,
      "loss": 0.0043,
      "step": 771
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7965002059936523,
      "learning_rate": 1.0873745575854001e-05,
      "loss": 0.0131,
      "step": 772
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.1829119324684143,
      "learning_rate": 1.0870611931595832e-05,
      "loss": 0.0026,
      "step": 773
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.09523776173591614,
      "learning_rate": 1.086747828733766e-05,
      "loss": 0.0009,
      "step": 774
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.12286712229251862,
      "learning_rate": 1.0864344643079488e-05,
      "loss": 0.0021,
      "step": 775
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.12610690295696259,
      "learning_rate": 1.0861210998821317e-05,
      "loss": 0.0019,
      "step": 776
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.33074650168418884,
      "learning_rate": 1.0858077354563146e-05,
      "loss": 0.0026,
      "step": 777
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.38875433802604675,
      "learning_rate": 1.0854943710304975e-05,
      "loss": 0.017,
      "step": 778
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.1672883778810501,
      "learning_rate": 1.0851810066046804e-05,
      "loss": 0.0019,
      "step": 779
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.18982449173927307,
      "learning_rate": 1.0848676421788632e-05,
      "loss": 0.0019,
      "step": 780
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.20724084973335266,
      "learning_rate": 1.084554277753046e-05,
      "loss": 0.0016,
      "step": 781
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.41346630454063416,
      "learning_rate": 1.084240913327229e-05,
      "loss": 0.0036,
      "step": 782
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.2011241912841797,
      "learning_rate": 1.083927548901412e-05,
      "loss": 0.0039,
      "step": 783
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.1061830148100853,
      "learning_rate": 1.0836141844755948e-05,
      "loss": 0.0011,
      "step": 784
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.2667681574821472,
      "learning_rate": 1.0833008200497775e-05,
      "loss": 0.002,
      "step": 785
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.12072428315877914,
      "learning_rate": 1.0829874556239606e-05,
      "loss": 0.0018,
      "step": 786
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.10592997819185257,
      "learning_rate": 1.0826740911981435e-05,
      "loss": 0.0013,
      "step": 787
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.2694378197193146,
      "learning_rate": 1.0823607267723262e-05,
      "loss": 0.0046,
      "step": 788
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.15724769234657288,
      "learning_rate": 1.0820473623465091e-05,
      "loss": 0.0023,
      "step": 789
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.24358107149600983,
      "learning_rate": 1.0817339979206922e-05,
      "loss": 0.0031,
      "step": 790
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.63752281665802,
      "learning_rate": 1.0814206334948749e-05,
      "loss": 0.004,
      "step": 791
    },
    {
      "epoch": 3.92,
      "grad_norm": 3.1439943313598633,
      "learning_rate": 1.0811072690690578e-05,
      "loss": 0.0107,
      "step": 792
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.19486427307128906,
      "learning_rate": 1.0807939046432407e-05,
      "loss": 0.0032,
      "step": 793
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.13371670246124268,
      "learning_rate": 1.0804805402174236e-05,
      "loss": 0.0034,
      "step": 794
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.2311642169952393,
      "learning_rate": 1.0801671757916065e-05,
      "loss": 0.0064,
      "step": 795
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.15413545072078705,
      "learning_rate": 1.0798538113657894e-05,
      "loss": 0.0026,
      "step": 796
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.0836484357714653,
      "learning_rate": 1.0795404469399722e-05,
      "loss": 0.0008,
      "step": 797
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.48144063353538513,
      "learning_rate": 1.0792270825141551e-05,
      "loss": 0.004,
      "step": 798
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.1532559096813202,
      "learning_rate": 1.078913718088338e-05,
      "loss": 0.0016,
      "step": 799
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.08767285197973251,
      "learning_rate": 1.078600353662521e-05,
      "loss": 0.0024,
      "step": 800
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.1613958328962326,
      "learning_rate": 1.0782869892367036e-05,
      "loss": 0.0039,
      "step": 801
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.4709697365760803,
      "learning_rate": 1.0779736248108865e-05,
      "loss": 0.0039,
      "step": 802
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.762413501739502,
      "learning_rate": 1.0776602603850696e-05,
      "loss": 0.0152,
      "step": 803
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.15357325971126556,
      "learning_rate": 1.0773468959592523e-05,
      "loss": 0.0015,
      "step": 804
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.17114724218845367,
      "learning_rate": 1.0770335315334352e-05,
      "loss": 0.0017,
      "step": 805
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.1969446986913681,
      "learning_rate": 1.0767201671076181e-05,
      "loss": 0.0031,
      "step": 806
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.15421070158481598,
      "learning_rate": 1.076406802681801e-05,
      "loss": 0.0035,
      "step": 807
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.05711280182003975,
      "learning_rate": 1.0760934382559839e-05,
      "loss": 0.0003,
      "step": 808
    },
    {
      "epoch": 4.0,
      "eval_AVGf1": 0.927578093306288,
      "eval_DIAG.f1": 0.8937931034482758,
      "eval_DIAG.precision": 0.9050279329608939,
      "eval_DIAG.recall": 0.8828337874659401,
      "eval_MED.f1": 0.976,
      "eval_MED.precision": 0.976,
      "eval_MED.recall": 0.976,
      "eval_TREAT.f1": 0.9129411764705883,
      "eval_TREAT.precision": 0.9107981220657277,
      "eval_TREAT.recall": 0.9150943396226415,
      "eval_accuracy": 0.9462809917355371,
      "eval_f1": 0.785758133824432,
      "eval_loss": 0.015093267895281315,
      "eval_precision": 0.6918918918918919,
      "eval_recall": 0.9090909090909091,
      "eval_runtime": 963.4496,
      "eval_samples_per_second": 0.745,
      "eval_steps_per_second": 0.093,
      "step": 808
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.1919289231300354,
      "learning_rate": 1.0757800738301668e-05,
      "loss": 0.0013,
      "step": 809
    },
    {
      "epoch": 4.01,
      "grad_norm": 0.36885520815849304,
      "learning_rate": 1.0754667094043497e-05,
      "loss": 0.001,
      "step": 810
    },
    {
      "epoch": 4.01,
      "grad_norm": 0.14645346999168396,
      "learning_rate": 1.0751533449785326e-05,
      "loss": 0.0012,
      "step": 811
    },
    {
      "epoch": 4.02,
      "grad_norm": 0.11528532207012177,
      "learning_rate": 1.0748399805527155e-05,
      "loss": 0.0012,
      "step": 812
    },
    {
      "epoch": 4.02,
      "grad_norm": 0.29233884811401367,
      "learning_rate": 1.0745266161268984e-05,
      "loss": 0.0022,
      "step": 813
    },
    {
      "epoch": 4.03,
      "grad_norm": 0.29916074872016907,
      "learning_rate": 1.074213251701081e-05,
      "loss": 0.0017,
      "step": 814
    },
    {
      "epoch": 4.03,
      "grad_norm": 0.020575886592268944,
      "learning_rate": 1.0738998872752641e-05,
      "loss": 0.0002,
      "step": 815
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.500468909740448,
      "learning_rate": 1.073586522849447e-05,
      "loss": 0.0026,
      "step": 816
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.08279713243246078,
      "learning_rate": 1.07327315842363e-05,
      "loss": 0.002,
      "step": 817
    },
    {
      "epoch": 4.05,
      "grad_norm": 0.07128813862800598,
      "learning_rate": 1.0729597939978127e-05,
      "loss": 0.0016,
      "step": 818
    },
    {
      "epoch": 4.05,
      "grad_norm": 0.26631614565849304,
      "learning_rate": 1.0726464295719957e-05,
      "loss": 0.0011,
      "step": 819
    },
    {
      "epoch": 4.06,
      "grad_norm": 0.1340247094631195,
      "learning_rate": 1.0723330651461786e-05,
      "loss": 0.0021,
      "step": 820
    },
    {
      "epoch": 4.06,
      "grad_norm": 0.12595820426940918,
      "learning_rate": 1.0720197007203613e-05,
      "loss": 0.0014,
      "step": 821
    },
    {
      "epoch": 4.07,
      "grad_norm": 0.115996815264225,
      "learning_rate": 1.0717063362945442e-05,
      "loss": 0.0021,
      "step": 822
    },
    {
      "epoch": 4.07,
      "grad_norm": 0.12639397382736206,
      "learning_rate": 1.0713929718687273e-05,
      "loss": 0.0036,
      "step": 823
    },
    {
      "epoch": 4.08,
      "grad_norm": 0.05830858647823334,
      "learning_rate": 1.07107960744291e-05,
      "loss": 0.001,
      "step": 824
    },
    {
      "epoch": 4.08,
      "grad_norm": 0.036114998161792755,
      "learning_rate": 1.0707662430170929e-05,
      "loss": 0.0003,
      "step": 825
    },
    {
      "epoch": 4.09,
      "grad_norm": 0.08563323318958282,
      "learning_rate": 1.0704528785912758e-05,
      "loss": 0.0019,
      "step": 826
    },
    {
      "epoch": 4.09,
      "grad_norm": 0.1124243438243866,
      "learning_rate": 1.0701395141654587e-05,
      "loss": 0.0017,
      "step": 827
    },
    {
      "epoch": 4.1,
      "grad_norm": 0.04704216495156288,
      "learning_rate": 1.0698261497396416e-05,
      "loss": 0.0007,
      "step": 828
    },
    {
      "epoch": 4.1,
      "grad_norm": 0.2557905614376068,
      "learning_rate": 1.0695127853138245e-05,
      "loss": 0.0023,
      "step": 829
    },
    {
      "epoch": 4.11,
      "grad_norm": 0.6211841702461243,
      "learning_rate": 1.0691994208880074e-05,
      "loss": 0.0029,
      "step": 830
    },
    {
      "epoch": 4.11,
      "grad_norm": 0.08716987073421478,
      "learning_rate": 1.0688860564621901e-05,
      "loss": 0.0007,
      "step": 831
    },
    {
      "epoch": 4.12,
      "grad_norm": 0.087484210729599,
      "learning_rate": 1.0685726920363731e-05,
      "loss": 0.0003,
      "step": 832
    },
    {
      "epoch": 4.12,
      "grad_norm": 0.07249535620212555,
      "learning_rate": 1.068259327610556e-05,
      "loss": 0.0002,
      "step": 833
    },
    {
      "epoch": 4.13,
      "grad_norm": 0.08692945539951324,
      "learning_rate": 1.0679459631847388e-05,
      "loss": 0.0015,
      "step": 834
    },
    {
      "epoch": 4.13,
      "grad_norm": 0.24230727553367615,
      "learning_rate": 1.0676325987589217e-05,
      "loss": 0.0064,
      "step": 835
    },
    {
      "epoch": 4.14,
      "grad_norm": 0.3160689175128937,
      "learning_rate": 1.0673192343331047e-05,
      "loss": 0.0028,
      "step": 836
    },
    {
      "epoch": 4.14,
      "grad_norm": 0.4063141942024231,
      "learning_rate": 1.0670058699072874e-05,
      "loss": 0.0021,
      "step": 837
    },
    {
      "epoch": 4.15,
      "grad_norm": 0.07834106683731079,
      "learning_rate": 1.0666925054814703e-05,
      "loss": 0.0008,
      "step": 838
    },
    {
      "epoch": 4.15,
      "grad_norm": 0.11609429121017456,
      "learning_rate": 1.0663791410556532e-05,
      "loss": 0.0028,
      "step": 839
    },
    {
      "epoch": 4.16,
      "grad_norm": 0.26402267813682556,
      "learning_rate": 1.0660657766298361e-05,
      "loss": 0.0019,
      "step": 840
    },
    {
      "epoch": 4.16,
      "grad_norm": 0.10171741247177124,
      "learning_rate": 1.065752412204019e-05,
      "loss": 0.0008,
      "step": 841
    },
    {
      "epoch": 4.17,
      "grad_norm": 0.33649611473083496,
      "learning_rate": 1.0654390477782019e-05,
      "loss": 0.0027,
      "step": 842
    },
    {
      "epoch": 4.17,
      "grad_norm": 0.08745944499969482,
      "learning_rate": 1.0651256833523848e-05,
      "loss": 0.0008,
      "step": 843
    },
    {
      "epoch": 4.18,
      "grad_norm": 0.2022976577281952,
      "learning_rate": 1.0648123189265677e-05,
      "loss": 0.0009,
      "step": 844
    },
    {
      "epoch": 4.18,
      "grad_norm": 0.15411056578159332,
      "learning_rate": 1.0644989545007506e-05,
      "loss": 0.0019,
      "step": 845
    },
    {
      "epoch": 4.19,
      "grad_norm": 0.06805182993412018,
      "learning_rate": 1.0641855900749335e-05,
      "loss": 0.0008,
      "step": 846
    },
    {
      "epoch": 4.19,
      "grad_norm": 2.399332284927368,
      "learning_rate": 1.0638722256491162e-05,
      "loss": 0.0134,
      "step": 847
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.1378047615289688,
      "learning_rate": 1.0635588612232993e-05,
      "loss": 0.0015,
      "step": 848
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.23969024419784546,
      "learning_rate": 1.0632454967974821e-05,
      "loss": 0.0043,
      "step": 849
    },
    {
      "epoch": 4.21,
      "grad_norm": 0.1560378074645996,
      "learning_rate": 1.062932132371665e-05,
      "loss": 0.0032,
      "step": 850
    },
    {
      "epoch": 4.21,
      "grad_norm": 0.19125576317310333,
      "learning_rate": 1.0626187679458478e-05,
      "loss": 0.0021,
      "step": 851
    },
    {
      "epoch": 4.22,
      "grad_norm": 0.12119518965482712,
      "learning_rate": 1.0623054035200307e-05,
      "loss": 0.0015,
      "step": 852
    },
    {
      "epoch": 4.22,
      "grad_norm": 0.09724880009889603,
      "learning_rate": 1.0619920390942137e-05,
      "loss": 0.0004,
      "step": 853
    },
    {
      "epoch": 4.23,
      "grad_norm": 0.08055749535560608,
      "learning_rate": 1.0616786746683964e-05,
      "loss": 0.0013,
      "step": 854
    },
    {
      "epoch": 4.23,
      "grad_norm": 0.06786857545375824,
      "learning_rate": 1.0613653102425793e-05,
      "loss": 0.0007,
      "step": 855
    },
    {
      "epoch": 4.24,
      "grad_norm": 0.06449595093727112,
      "learning_rate": 1.0610519458167622e-05,
      "loss": 0.0006,
      "step": 856
    },
    {
      "epoch": 4.24,
      "grad_norm": 0.128890722990036,
      "learning_rate": 1.0607385813909451e-05,
      "loss": 0.0028,
      "step": 857
    },
    {
      "epoch": 4.25,
      "grad_norm": 5.841109752655029,
      "learning_rate": 1.060425216965128e-05,
      "loss": 0.0174,
      "step": 858
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.09092911332845688,
      "learning_rate": 1.0601118525393109e-05,
      "loss": 0.0017,
      "step": 859
    },
    {
      "epoch": 4.26,
      "grad_norm": 0.22573551535606384,
      "learning_rate": 1.0597984881134938e-05,
      "loss": 0.0008,
      "step": 860
    },
    {
      "epoch": 4.26,
      "grad_norm": 0.14501002430915833,
      "learning_rate": 1.0594851236876767e-05,
      "loss": 0.0014,
      "step": 861
    },
    {
      "epoch": 4.27,
      "grad_norm": 0.26200586557388306,
      "learning_rate": 1.0591717592618596e-05,
      "loss": 0.0028,
      "step": 862
    },
    {
      "epoch": 4.27,
      "grad_norm": 0.0637667328119278,
      "learning_rate": 1.0588583948360425e-05,
      "loss": 0.0003,
      "step": 863
    },
    {
      "epoch": 4.28,
      "grad_norm": 0.48093774914741516,
      "learning_rate": 1.0585450304102252e-05,
      "loss": 0.002,
      "step": 864
    },
    {
      "epoch": 4.28,
      "grad_norm": 0.2839108407497406,
      "learning_rate": 1.0582316659844083e-05,
      "loss": 0.0013,
      "step": 865
    },
    {
      "epoch": 4.29,
      "grad_norm": 0.4579899311065674,
      "learning_rate": 1.0579183015585912e-05,
      "loss": 0.0016,
      "step": 866
    },
    {
      "epoch": 4.29,
      "grad_norm": 0.4219379127025604,
      "learning_rate": 1.0576049371327739e-05,
      "loss": 0.0035,
      "step": 867
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.14920270442962646,
      "learning_rate": 1.0572915727069568e-05,
      "loss": 0.0007,
      "step": 868
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.07550948858261108,
      "learning_rate": 1.0569782082811398e-05,
      "loss": 0.0006,
      "step": 869
    },
    {
      "epoch": 4.31,
      "grad_norm": 0.153974711894989,
      "learning_rate": 1.0566648438553226e-05,
      "loss": 0.0031,
      "step": 870
    },
    {
      "epoch": 4.31,
      "grad_norm": 0.13354970514774323,
      "learning_rate": 1.0563514794295054e-05,
      "loss": 0.0021,
      "step": 871
    },
    {
      "epoch": 4.32,
      "grad_norm": 0.04836566001176834,
      "learning_rate": 1.0560381150036883e-05,
      "loss": 0.0003,
      "step": 872
    },
    {
      "epoch": 4.32,
      "grad_norm": 0.13605570793151855,
      "learning_rate": 1.0557247505778712e-05,
      "loss": 0.0012,
      "step": 873
    },
    {
      "epoch": 4.33,
      "grad_norm": 0.11106490343809128,
      "learning_rate": 1.0554113861520541e-05,
      "loss": 0.0015,
      "step": 874
    },
    {
      "epoch": 4.33,
      "grad_norm": 0.13186205923557281,
      "learning_rate": 1.055098021726237e-05,
      "loss": 0.0018,
      "step": 875
    },
    {
      "epoch": 4.34,
      "grad_norm": 0.3905848562717438,
      "learning_rate": 1.0547846573004199e-05,
      "loss": 0.0014,
      "step": 876
    },
    {
      "epoch": 4.34,
      "grad_norm": 0.13211289048194885,
      "learning_rate": 1.0544712928746026e-05,
      "loss": 0.0014,
      "step": 877
    },
    {
      "epoch": 4.35,
      "grad_norm": 0.8153514266014099,
      "learning_rate": 1.0541579284487857e-05,
      "loss": 0.0066,
      "step": 878
    },
    {
      "epoch": 4.35,
      "grad_norm": 0.0976765975356102,
      "learning_rate": 1.0538445640229686e-05,
      "loss": 0.001,
      "step": 879
    },
    {
      "epoch": 4.36,
      "grad_norm": 0.06659042835235596,
      "learning_rate": 1.0535311995971513e-05,
      "loss": 0.001,
      "step": 880
    },
    {
      "epoch": 4.36,
      "grad_norm": 0.15370504558086395,
      "learning_rate": 1.0532178351713342e-05,
      "loss": 0.002,
      "step": 881
    },
    {
      "epoch": 4.37,
      "grad_norm": 0.11330357193946838,
      "learning_rate": 1.0529044707455173e-05,
      "loss": 0.0024,
      "step": 882
    },
    {
      "epoch": 4.37,
      "grad_norm": 0.14851288497447968,
      "learning_rate": 1.0525911063197002e-05,
      "loss": 0.0017,
      "step": 883
    },
    {
      "epoch": 4.38,
      "grad_norm": 0.1697237342596054,
      "learning_rate": 1.0522777418938829e-05,
      "loss": 0.0029,
      "step": 884
    },
    {
      "epoch": 4.38,
      "grad_norm": 0.39677751064300537,
      "learning_rate": 1.0519643774680658e-05,
      "loss": 0.0021,
      "step": 885
    },
    {
      "epoch": 4.39,
      "grad_norm": 0.5811119675636292,
      "learning_rate": 1.0516510130422488e-05,
      "loss": 0.0111,
      "step": 886
    },
    {
      "epoch": 4.39,
      "grad_norm": 0.07648241519927979,
      "learning_rate": 1.0513376486164316e-05,
      "loss": 0.0008,
      "step": 887
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.09037857502698898,
      "learning_rate": 1.0510242841906144e-05,
      "loss": 0.0031,
      "step": 888
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.06013543903827667,
      "learning_rate": 1.0507109197647973e-05,
      "loss": 0.0022,
      "step": 889
    },
    {
      "epoch": 4.41,
      "grad_norm": 0.12190402299165726,
      "learning_rate": 1.0503975553389802e-05,
      "loss": 0.0015,
      "step": 890
    },
    {
      "epoch": 4.41,
      "grad_norm": 0.06382696330547333,
      "learning_rate": 1.0500841909131631e-05,
      "loss": 0.0018,
      "step": 891
    },
    {
      "epoch": 4.42,
      "grad_norm": 0.11207643896341324,
      "learning_rate": 1.049770826487346e-05,
      "loss": 0.0016,
      "step": 892
    },
    {
      "epoch": 4.42,
      "grad_norm": 0.15361686050891876,
      "learning_rate": 1.0494574620615289e-05,
      "loss": 0.0012,
      "step": 893
    },
    {
      "epoch": 4.43,
      "grad_norm": 0.21753041446208954,
      "learning_rate": 1.0491440976357118e-05,
      "loss": 0.0021,
      "step": 894
    },
    {
      "epoch": 4.43,
      "grad_norm": 0.2366906851530075,
      "learning_rate": 1.0488307332098947e-05,
      "loss": 0.0017,
      "step": 895
    },
    {
      "epoch": 4.44,
      "grad_norm": 0.9420850872993469,
      "learning_rate": 1.0485173687840776e-05,
      "loss": 0.0104,
      "step": 896
    },
    {
      "epoch": 4.44,
      "grad_norm": 0.05524289608001709,
      "learning_rate": 1.0482040043582603e-05,
      "loss": 0.0009,
      "step": 897
    },
    {
      "epoch": 4.45,
      "grad_norm": 0.09353532642126083,
      "learning_rate": 1.0478906399324434e-05,
      "loss": 0.0007,
      "step": 898
    },
    {
      "epoch": 4.45,
      "grad_norm": 0.23325134813785553,
      "learning_rate": 1.0475772755066263e-05,
      "loss": 0.0013,
      "step": 899
    },
    {
      "epoch": 4.46,
      "grad_norm": 0.07453510165214539,
      "learning_rate": 1.047263911080809e-05,
      "loss": 0.0006,
      "step": 900
    },
    {
      "epoch": 4.46,
      "grad_norm": 0.27760934829711914,
      "learning_rate": 1.0469505466549919e-05,
      "loss": 0.0021,
      "step": 901
    },
    {
      "epoch": 4.47,
      "grad_norm": 0.3367156982421875,
      "learning_rate": 1.0466371822291748e-05,
      "loss": 0.0037,
      "step": 902
    },
    {
      "epoch": 4.47,
      "grad_norm": 0.14733706414699554,
      "learning_rate": 1.0463238178033577e-05,
      "loss": 0.004,
      "step": 903
    },
    {
      "epoch": 4.48,
      "grad_norm": 0.14544622600078583,
      "learning_rate": 1.0460104533775406e-05,
      "loss": 0.002,
      "step": 904
    },
    {
      "epoch": 4.48,
      "grad_norm": 0.15930630266666412,
      "learning_rate": 1.0456970889517234e-05,
      "loss": 0.0018,
      "step": 905
    },
    {
      "epoch": 4.49,
      "grad_norm": 0.14259351789951324,
      "learning_rate": 1.0453837245259063e-05,
      "loss": 0.0018,
      "step": 906
    },
    {
      "epoch": 4.49,
      "grad_norm": 0.20260144770145416,
      "learning_rate": 1.0450703601000892e-05,
      "loss": 0.0028,
      "step": 907
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.10381250083446503,
      "learning_rate": 1.0447569956742721e-05,
      "loss": 0.0006,
      "step": 908
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.08324310928583145,
      "learning_rate": 1.044443631248455e-05,
      "loss": 0.0016,
      "step": 909
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.2987682819366455,
      "learning_rate": 1.0441302668226377e-05,
      "loss": 0.0029,
      "step": 910
    },
    {
      "epoch": 4.51,
      "grad_norm": 0.3024025857448578,
      "learning_rate": 1.0438169023968208e-05,
      "loss": 0.0018,
      "step": 911
    },
    {
      "epoch": 4.51,
      "grad_norm": 0.14482058584690094,
      "learning_rate": 1.0435035379710037e-05,
      "loss": 0.0022,
      "step": 912
    },
    {
      "epoch": 4.52,
      "grad_norm": 0.0930890291929245,
      "learning_rate": 1.0431901735451864e-05,
      "loss": 0.0023,
      "step": 913
    },
    {
      "epoch": 4.52,
      "grad_norm": 0.07957684993743896,
      "learning_rate": 1.0428768091193693e-05,
      "loss": 0.0005,
      "step": 914
    },
    {
      "epoch": 4.53,
      "grad_norm": 0.14982189238071442,
      "learning_rate": 1.0425634446935524e-05,
      "loss": 0.0009,
      "step": 915
    },
    {
      "epoch": 4.53,
      "grad_norm": 0.21297410130500793,
      "learning_rate": 1.0422500802677351e-05,
      "loss": 0.001,
      "step": 916
    },
    {
      "epoch": 4.54,
      "grad_norm": 0.5605773329734802,
      "learning_rate": 1.041936715841918e-05,
      "loss": 0.0073,
      "step": 917
    },
    {
      "epoch": 4.54,
      "grad_norm": 0.03863908350467682,
      "learning_rate": 1.0416233514161009e-05,
      "loss": 0.0003,
      "step": 918
    },
    {
      "epoch": 4.55,
      "grad_norm": 0.04396303370594978,
      "learning_rate": 1.041309986990284e-05,
      "loss": 0.0003,
      "step": 919
    },
    {
      "epoch": 4.55,
      "grad_norm": 0.6286675333976746,
      "learning_rate": 1.0409966225644667e-05,
      "loss": 0.0048,
      "step": 920
    },
    {
      "epoch": 4.56,
      "grad_norm": 0.1118682473897934,
      "learning_rate": 1.0406832581386496e-05,
      "loss": 0.0015,
      "step": 921
    },
    {
      "epoch": 4.56,
      "grad_norm": 0.0693434625864029,
      "learning_rate": 1.0403698937128325e-05,
      "loss": 0.0006,
      "step": 922
    },
    {
      "epoch": 4.57,
      "grad_norm": 0.03157630190253258,
      "learning_rate": 1.0400565292870153e-05,
      "loss": 0.0003,
      "step": 923
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.9115592241287231,
      "learning_rate": 1.0397431648611982e-05,
      "loss": 0.0023,
      "step": 924
    },
    {
      "epoch": 4.58,
      "grad_norm": 0.11953920125961304,
      "learning_rate": 1.0394298004353811e-05,
      "loss": 0.001,
      "step": 925
    },
    {
      "epoch": 4.58,
      "grad_norm": 0.12054043263196945,
      "learning_rate": 1.039116436009564e-05,
      "loss": 0.0018,
      "step": 926
    },
    {
      "epoch": 4.59,
      "grad_norm": 0.16741815209388733,
      "learning_rate": 1.0388030715837467e-05,
      "loss": 0.0006,
      "step": 927
    },
    {
      "epoch": 4.59,
      "grad_norm": 0.07631529122591019,
      "learning_rate": 1.0384897071579298e-05,
      "loss": 0.0004,
      "step": 928
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.06158828362822533,
      "learning_rate": 1.0381763427321127e-05,
      "loss": 0.0008,
      "step": 929
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.11724244058132172,
      "learning_rate": 1.0378629783062954e-05,
      "loss": 0.001,
      "step": 930
    },
    {
      "epoch": 4.61,
      "grad_norm": 0.11568804830312729,
      "learning_rate": 1.0375496138804783e-05,
      "loss": 0.0034,
      "step": 931
    },
    {
      "epoch": 4.61,
      "grad_norm": 0.18305525183677673,
      "learning_rate": 1.0372362494546614e-05,
      "loss": 0.0012,
      "step": 932
    },
    {
      "epoch": 4.62,
      "grad_norm": 0.10010791569948196,
      "learning_rate": 1.0369228850288441e-05,
      "loss": 0.0009,
      "step": 933
    },
    {
      "epoch": 4.62,
      "grad_norm": 0.16863921284675598,
      "learning_rate": 1.036609520603027e-05,
      "loss": 0.0017,
      "step": 934
    },
    {
      "epoch": 4.63,
      "grad_norm": 0.0711769312620163,
      "learning_rate": 1.0362961561772099e-05,
      "loss": 0.0014,
      "step": 935
    },
    {
      "epoch": 4.63,
      "grad_norm": 0.1584193855524063,
      "learning_rate": 1.0359827917513928e-05,
      "loss": 0.001,
      "step": 936
    },
    {
      "epoch": 4.64,
      "grad_norm": 0.11551675945520401,
      "learning_rate": 1.0356694273255757e-05,
      "loss": 0.0022,
      "step": 937
    },
    {
      "epoch": 4.64,
      "grad_norm": 0.15103916823863983,
      "learning_rate": 1.0353560628997586e-05,
      "loss": 0.002,
      "step": 938
    },
    {
      "epoch": 4.65,
      "grad_norm": 0.9653202891349792,
      "learning_rate": 1.0350426984739415e-05,
      "loss": 0.008,
      "step": 939
    },
    {
      "epoch": 4.65,
      "grad_norm": 0.1497216522693634,
      "learning_rate": 1.0347293340481243e-05,
      "loss": 0.0012,
      "step": 940
    },
    {
      "epoch": 4.66,
      "grad_norm": 0.16273315250873566,
      "learning_rate": 1.0344159696223072e-05,
      "loss": 0.0021,
      "step": 941
    },
    {
      "epoch": 4.66,
      "grad_norm": 0.1253347098827362,
      "learning_rate": 1.0341026051964901e-05,
      "loss": 0.0012,
      "step": 942
    },
    {
      "epoch": 4.67,
      "grad_norm": 0.08757926523685455,
      "learning_rate": 1.0337892407706729e-05,
      "loss": 0.0014,
      "step": 943
    },
    {
      "epoch": 4.67,
      "grad_norm": 0.12076185643672943,
      "learning_rate": 1.033475876344856e-05,
      "loss": 0.0013,
      "step": 944
    },
    {
      "epoch": 4.68,
      "grad_norm": 0.484214186668396,
      "learning_rate": 1.0331625119190388e-05,
      "loss": 0.0015,
      "step": 945
    },
    {
      "epoch": 4.68,
      "grad_norm": 0.04215659201145172,
      "learning_rate": 1.0328491474932215e-05,
      "loss": 0.0005,
      "step": 946
    },
    {
      "epoch": 4.69,
      "grad_norm": 0.05397907271981239,
      "learning_rate": 1.0325357830674044e-05,
      "loss": 0.0007,
      "step": 947
    },
    {
      "epoch": 4.69,
      "grad_norm": 0.10055296868085861,
      "learning_rate": 1.0322224186415875e-05,
      "loss": 0.0021,
      "step": 948
    },
    {
      "epoch": 4.7,
      "grad_norm": 0.341290682554245,
      "learning_rate": 1.0319090542157702e-05,
      "loss": 0.0026,
      "step": 949
    },
    {
      "epoch": 4.7,
      "grad_norm": 0.17511200904846191,
      "learning_rate": 1.0315956897899531e-05,
      "loss": 0.0008,
      "step": 950
    },
    {
      "epoch": 4.71,
      "grad_norm": 0.5351629853248596,
      "learning_rate": 1.031282325364136e-05,
      "loss": 0.0035,
      "step": 951
    },
    {
      "epoch": 4.71,
      "grad_norm": 0.13497723639011383,
      "learning_rate": 1.0309689609383189e-05,
      "loss": 0.0031,
      "step": 952
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.0873773843050003,
      "learning_rate": 1.0306555965125018e-05,
      "loss": 0.0007,
      "step": 953
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.07625476270914078,
      "learning_rate": 1.0303422320866847e-05,
      "loss": 0.0007,
      "step": 954
    },
    {
      "epoch": 4.73,
      "grad_norm": 0.07660971581935883,
      "learning_rate": 1.0300288676608676e-05,
      "loss": 0.0007,
      "step": 955
    },
    {
      "epoch": 4.73,
      "grad_norm": 0.10834752023220062,
      "learning_rate": 1.0297155032350503e-05,
      "loss": 0.0006,
      "step": 956
    },
    {
      "epoch": 4.74,
      "grad_norm": 0.06748098134994507,
      "learning_rate": 1.0294021388092333e-05,
      "loss": 0.0014,
      "step": 957
    },
    {
      "epoch": 4.74,
      "grad_norm": 0.35559016466140747,
      "learning_rate": 1.0290887743834162e-05,
      "loss": 0.0022,
      "step": 958
    },
    {
      "epoch": 4.75,
      "grad_norm": 0.23248696327209473,
      "learning_rate": 1.0287754099575991e-05,
      "loss": 0.0015,
      "step": 959
    },
    {
      "epoch": 4.75,
      "grad_norm": 0.061237312853336334,
      "learning_rate": 1.0284620455317819e-05,
      "loss": 0.0014,
      "step": 960
    },
    {
      "epoch": 4.76,
      "grad_norm": 0.0791211649775505,
      "learning_rate": 1.028148681105965e-05,
      "loss": 0.0016,
      "step": 961
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.5654101371765137,
      "learning_rate": 1.0278353166801478e-05,
      "loss": 0.0159,
      "step": 962
    },
    {
      "epoch": 4.77,
      "grad_norm": 0.11773218214511871,
      "learning_rate": 1.0275219522543305e-05,
      "loss": 0.0024,
      "step": 963
    },
    {
      "epoch": 4.77,
      "grad_norm": 0.5579394102096558,
      "learning_rate": 1.0272085878285134e-05,
      "loss": 0.0026,
      "step": 964
    },
    {
      "epoch": 4.78,
      "grad_norm": 0.05599897727370262,
      "learning_rate": 1.0268952234026965e-05,
      "loss": 0.0008,
      "step": 965
    },
    {
      "epoch": 4.78,
      "grad_norm": 0.09302478283643723,
      "learning_rate": 1.0265818589768792e-05,
      "loss": 0.0006,
      "step": 966
    },
    {
      "epoch": 4.79,
      "grad_norm": 0.59910649061203,
      "learning_rate": 1.0262684945510621e-05,
      "loss": 0.0079,
      "step": 967
    },
    {
      "epoch": 4.79,
      "grad_norm": 0.09593573957681656,
      "learning_rate": 1.025955130125245e-05,
      "loss": 0.0024,
      "step": 968
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.09404628723859787,
      "learning_rate": 1.0256417656994279e-05,
      "loss": 0.0015,
      "step": 969
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.14857031404972076,
      "learning_rate": 1.0253284012736108e-05,
      "loss": 0.0025,
      "step": 970
    },
    {
      "epoch": 4.81,
      "grad_norm": 0.3911975026130676,
      "learning_rate": 1.0250150368477937e-05,
      "loss": 0.0034,
      "step": 971
    },
    {
      "epoch": 4.81,
      "grad_norm": 0.03253992274403572,
      "learning_rate": 1.0247016724219766e-05,
      "loss": 0.0003,
      "step": 972
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.0538798570632935,
      "learning_rate": 1.0243883079961595e-05,
      "loss": 0.006,
      "step": 973
    },
    {
      "epoch": 4.82,
      "grad_norm": 0.06660764664411545,
      "learning_rate": 1.0240749435703424e-05,
      "loss": 0.0008,
      "step": 974
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.2144451141357422,
      "learning_rate": 1.0237615791445252e-05,
      "loss": 0.0066,
      "step": 975
    },
    {
      "epoch": 4.83,
      "grad_norm": 0.3344007432460785,
      "learning_rate": 1.023448214718708e-05,
      "loss": 0.002,
      "step": 976
    },
    {
      "epoch": 4.84,
      "grad_norm": 0.0840497687458992,
      "learning_rate": 1.0231348502928909e-05,
      "loss": 0.0016,
      "step": 977
    },
    {
      "epoch": 4.84,
      "grad_norm": 0.34377655386924744,
      "learning_rate": 1.022821485867074e-05,
      "loss": 0.001,
      "step": 978
    },
    {
      "epoch": 4.85,
      "grad_norm": 0.07830354571342468,
      "learning_rate": 1.0225081214412566e-05,
      "loss": 0.0009,
      "step": 979
    },
    {
      "epoch": 4.85,
      "grad_norm": 0.1842389702796936,
      "learning_rate": 1.0221947570154395e-05,
      "loss": 0.0015,
      "step": 980
    },
    {
      "epoch": 4.86,
      "grad_norm": 0.15005484223365784,
      "learning_rate": 1.0218813925896224e-05,
      "loss": 0.0016,
      "step": 981
    },
    {
      "epoch": 4.86,
      "grad_norm": 0.21731935441493988,
      "learning_rate": 1.0215680281638053e-05,
      "loss": 0.0008,
      "step": 982
    },
    {
      "epoch": 4.87,
      "grad_norm": 0.1569393426179886,
      "learning_rate": 1.0212546637379882e-05,
      "loss": 0.0022,
      "step": 983
    },
    {
      "epoch": 4.87,
      "grad_norm": 0.0299218837171793,
      "learning_rate": 1.0209412993121711e-05,
      "loss": 0.0002,
      "step": 984
    },
    {
      "epoch": 4.88,
      "grad_norm": 0.06104075908660889,
      "learning_rate": 1.020627934886354e-05,
      "loss": 0.0016,
      "step": 985
    },
    {
      "epoch": 4.88,
      "grad_norm": 0.4784625172615051,
      "learning_rate": 1.0203145704605369e-05,
      "loss": 0.0017,
      "step": 986
    },
    {
      "epoch": 4.89,
      "grad_norm": 0.14487329125404358,
      "learning_rate": 1.0200012060347198e-05,
      "loss": 0.0024,
      "step": 987
    },
    {
      "epoch": 4.89,
      "grad_norm": 0.18043585121631622,
      "learning_rate": 1.0196878416089027e-05,
      "loss": 0.0008,
      "step": 988
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.4947347342967987,
      "learning_rate": 1.0193744771830854e-05,
      "loss": 0.0041,
      "step": 989
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.14530658721923828,
      "learning_rate": 1.0190611127572685e-05,
      "loss": 0.0018,
      "step": 990
    },
    {
      "epoch": 4.91,
      "grad_norm": 0.3404431939125061,
      "learning_rate": 1.0187477483314514e-05,
      "loss": 0.006,
      "step": 991
    },
    {
      "epoch": 4.91,
      "grad_norm": 0.06990846991539001,
      "learning_rate": 1.0184343839056342e-05,
      "loss": 0.0012,
      "step": 992
    },
    {
      "epoch": 4.92,
      "grad_norm": 0.052425261586904526,
      "learning_rate": 1.018121019479817e-05,
      "loss": 0.0003,
      "step": 993
    },
    {
      "epoch": 4.92,
      "grad_norm": 0.18092483282089233,
      "learning_rate": 1.017807655054e-05,
      "loss": 0.0035,
      "step": 994
    },
    {
      "epoch": 4.93,
      "grad_norm": 0.4231566786766052,
      "learning_rate": 1.017494290628183e-05,
      "loss": 0.007,
      "step": 995
    },
    {
      "epoch": 4.93,
      "grad_norm": 0.0965544804930687,
      "learning_rate": 1.0171809262023656e-05,
      "loss": 0.0009,
      "step": 996
    },
    {
      "epoch": 4.94,
      "grad_norm": 0.1900458186864853,
      "learning_rate": 1.0168675617765485e-05,
      "loss": 0.0027,
      "step": 997
    },
    {
      "epoch": 4.94,
      "grad_norm": 0.02809087373316288,
      "learning_rate": 1.0165541973507316e-05,
      "loss": 0.0002,
      "step": 998
    },
    {
      "epoch": 4.95,
      "grad_norm": 0.15995682775974274,
      "learning_rate": 1.0162408329249143e-05,
      "loss": 0.002,
      "step": 999
    },
    {
      "epoch": 4.95,
      "grad_norm": 0.10238388925790787,
      "learning_rate": 1.0159274684990972e-05,
      "loss": 0.0024,
      "step": 1000
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.08678873628377914,
      "learning_rate": 1.0156141040732801e-05,
      "loss": 0.0013,
      "step": 1001
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.0725141391158104,
      "learning_rate": 1.015300739647463e-05,
      "loss": 0.0006,
      "step": 1002
    },
    {
      "epoch": 4.97,
      "grad_norm": 0.03946508467197418,
      "learning_rate": 1.0149873752216459e-05,
      "loss": 0.0004,
      "step": 1003
    },
    {
      "epoch": 4.97,
      "grad_norm": 0.1928601711988449,
      "learning_rate": 1.0146740107958288e-05,
      "loss": 0.0016,
      "step": 1004
    },
    {
      "epoch": 4.98,
      "grad_norm": 0.070815309882164,
      "learning_rate": 1.0143606463700117e-05,
      "loss": 0.0007,
      "step": 1005
    },
    {
      "epoch": 4.98,
      "grad_norm": 0.1003846675157547,
      "learning_rate": 1.0140472819441944e-05,
      "loss": 0.0011,
      "step": 1006
    },
    {
      "epoch": 4.99,
      "grad_norm": 0.07330068200826645,
      "learning_rate": 1.0137339175183775e-05,
      "loss": 0.001,
      "step": 1007
    },
    {
      "epoch": 4.99,
      "grad_norm": 0.0775037333369255,
      "learning_rate": 1.0134205530925604e-05,
      "loss": 0.0004,
      "step": 1008
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.08864942193031311,
      "learning_rate": 1.013107188666743e-05,
      "loss": 0.0017,
      "step": 1009
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.2823582887649536,
      "learning_rate": 1.012793824240926e-05,
      "loss": 0.0019,
      "step": 1010
    },
    {
      "epoch": 5.0,
      "eval_AVGf1": 0.9330669508506887,
      "eval_DIAG.f1": 0.912621359223301,
      "eval_DIAG.precision": 0.9293785310734464,
      "eval_DIAG.recall": 0.896457765667575,
      "eval_MED.f1": 0.9718875502008033,
      "eval_MED.precision": 0.9758064516129032,
      "eval_MED.recall": 0.968,
      "eval_TREAT.f1": 0.9146919431279621,
      "eval_TREAT.precision": 0.919047619047619,
      "eval_TREAT.recall": 0.910377358490566,
      "eval_accuracy": 0.9522098454904779,
      "eval_f1": 0.8067754077791719,
      "eval_loss": 0.017719808965921402,
      "eval_precision": 0.7224719101123596,
      "eval_recall": 0.9133522727272727,
      "eval_runtime": 944.8436,
      "eval_samples_per_second": 0.76,
      "eval_steps_per_second": 0.095,
      "step": 1010
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.05410182848572731,
      "learning_rate": 1.012480459815109e-05,
      "loss": 0.0008,
      "step": 1011
    },
    {
      "epoch": 5.01,
      "grad_norm": 0.1399615854024887,
      "learning_rate": 1.0121670953892918e-05,
      "loss": 0.0015,
      "step": 1012
    },
    {
      "epoch": 5.01,
      "grad_norm": 0.07447698712348938,
      "learning_rate": 1.0118537309634746e-05,
      "loss": 0.0008,
      "step": 1013
    },
    {
      "epoch": 5.02,
      "grad_norm": 0.04005727544426918,
      "learning_rate": 1.0115403665376575e-05,
      "loss": 0.0004,
      "step": 1014
    },
    {
      "epoch": 5.02,
      "grad_norm": 0.0774712935090065,
      "learning_rate": 1.0112270021118404e-05,
      "loss": 0.001,
      "step": 1015
    },
    {
      "epoch": 5.03,
      "grad_norm": 0.01586790755391121,
      "learning_rate": 1.0109136376860233e-05,
      "loss": 0.0001,
      "step": 1016
    },
    {
      "epoch": 5.03,
      "grad_norm": 0.10197090357542038,
      "learning_rate": 1.0106002732602062e-05,
      "loss": 0.001,
      "step": 1017
    },
    {
      "epoch": 5.04,
      "grad_norm": 0.15525425970554352,
      "learning_rate": 1.0102869088343891e-05,
      "loss": 0.0019,
      "step": 1018
    },
    {
      "epoch": 5.04,
      "grad_norm": 0.09304305166006088,
      "learning_rate": 1.009973544408572e-05,
      "loss": 0.0016,
      "step": 1019
    },
    {
      "epoch": 5.05,
      "grad_norm": 6.917536735534668,
      "learning_rate": 1.0096601799827549e-05,
      "loss": 0.0094,
      "step": 1020
    },
    {
      "epoch": 5.05,
      "grad_norm": 0.6672495007514954,
      "learning_rate": 1.0093468155569378e-05,
      "loss": 0.0016,
      "step": 1021
    },
    {
      "epoch": 5.06,
      "grad_norm": 0.0702059268951416,
      "learning_rate": 1.0090334511311205e-05,
      "loss": 0.0008,
      "step": 1022
    },
    {
      "epoch": 5.06,
      "grad_norm": 0.08075249940156937,
      "learning_rate": 1.0087200867053036e-05,
      "loss": 0.0006,
      "step": 1023
    },
    {
      "epoch": 5.07,
      "grad_norm": 0.08803071826696396,
      "learning_rate": 1.0084067222794865e-05,
      "loss": 0.0006,
      "step": 1024
    },
    {
      "epoch": 5.07,
      "grad_norm": 0.08024802803993225,
      "learning_rate": 1.0080933578536694e-05,
      "loss": 0.0009,
      "step": 1025
    },
    {
      "epoch": 5.08,
      "grad_norm": 0.013280569575726986,
      "learning_rate": 1.007779993427852e-05,
      "loss": 0.0001,
      "step": 1026
    },
    {
      "epoch": 5.08,
      "grad_norm": 0.011612316593527794,
      "learning_rate": 1.007466629002035e-05,
      "loss": 0.0001,
      "step": 1027
    },
    {
      "epoch": 5.09,
      "grad_norm": 0.35925835371017456,
      "learning_rate": 1.007153264576218e-05,
      "loss": 0.0016,
      "step": 1028
    },
    {
      "epoch": 5.09,
      "grad_norm": 0.9871984124183655,
      "learning_rate": 1.0068399001504008e-05,
      "loss": 0.0033,
      "step": 1029
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.06335406750440598,
      "learning_rate": 1.0065265357245837e-05,
      "loss": 0.0007,
      "step": 1030
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.1352735459804535,
      "learning_rate": 1.0062131712987665e-05,
      "loss": 0.0017,
      "step": 1031
    },
    {
      "epoch": 5.11,
      "grad_norm": 0.043752092868089676,
      "learning_rate": 1.0058998068729494e-05,
      "loss": 0.0008,
      "step": 1032
    },
    {
      "epoch": 5.11,
      "grad_norm": 0.647517740726471,
      "learning_rate": 1.0055864424471323e-05,
      "loss": 0.0016,
      "step": 1033
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.10731631517410278,
      "learning_rate": 1.0052730780213152e-05,
      "loss": 0.0011,
      "step": 1034
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.13756334781646729,
      "learning_rate": 1.0049597135954981e-05,
      "loss": 0.001,
      "step": 1035
    },
    {
      "epoch": 5.13,
      "grad_norm": 0.11022597551345825,
      "learning_rate": 1.004646349169681e-05,
      "loss": 0.0009,
      "step": 1036
    },
    {
      "epoch": 5.13,
      "grad_norm": 0.12070055305957794,
      "learning_rate": 1.0043329847438639e-05,
      "loss": 0.001,
      "step": 1037
    },
    {
      "epoch": 5.14,
      "grad_norm": 0.037480127066373825,
      "learning_rate": 1.0040196203180468e-05,
      "loss": 0.0002,
      "step": 1038
    },
    {
      "epoch": 5.14,
      "grad_norm": 2.8844544887542725,
      "learning_rate": 1.0037062558922295e-05,
      "loss": 0.003,
      "step": 1039
    },
    {
      "epoch": 5.15,
      "grad_norm": 0.09766790270805359,
      "learning_rate": 1.0033928914664126e-05,
      "loss": 0.001,
      "step": 1040
    },
    {
      "epoch": 5.15,
      "grad_norm": 0.20324283838272095,
      "learning_rate": 1.0030795270405955e-05,
      "loss": 0.0038,
      "step": 1041
    },
    {
      "epoch": 5.16,
      "grad_norm": 0.1715298891067505,
      "learning_rate": 1.0027661626147782e-05,
      "loss": 0.0026,
      "step": 1042
    },
    {
      "epoch": 5.16,
      "grad_norm": 0.13229705393314362,
      "learning_rate": 1.002452798188961e-05,
      "loss": 0.0022,
      "step": 1043
    },
    {
      "epoch": 5.17,
      "grad_norm": 0.09102973341941833,
      "learning_rate": 1.0021394337631441e-05,
      "loss": 0.0009,
      "step": 1044
    },
    {
      "epoch": 5.17,
      "grad_norm": 0.06405163556337357,
      "learning_rate": 1.0018260693373269e-05,
      "loss": 0.0009,
      "step": 1045
    },
    {
      "epoch": 5.18,
      "grad_norm": 0.06822778284549713,
      "learning_rate": 1.0015127049115098e-05,
      "loss": 0.0006,
      "step": 1046
    },
    {
      "epoch": 5.18,
      "grad_norm": 0.1688670665025711,
      "learning_rate": 1.0011993404856927e-05,
      "loss": 0.0011,
      "step": 1047
    },
    {
      "epoch": 5.19,
      "grad_norm": 0.1795819252729416,
      "learning_rate": 1.0008859760598755e-05,
      "loss": 0.0011,
      "step": 1048
    },
    {
      "epoch": 5.19,
      "grad_norm": 0.5702873468399048,
      "learning_rate": 1.0005726116340584e-05,
      "loss": 0.0014,
      "step": 1049
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.15224650502204895,
      "learning_rate": 1.0002592472082413e-05,
      "loss": 0.0013,
      "step": 1050
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.23804889619350433,
      "learning_rate": 9.999458827824242e-06,
      "loss": 0.0006,
      "step": 1051
    },
    {
      "epoch": 5.21,
      "grad_norm": 0.5042240619659424,
      "learning_rate": 9.99632518356607e-06,
      "loss": 0.0029,
      "step": 1052
    },
    {
      "epoch": 5.21,
      "grad_norm": 0.019530117511749268,
      "learning_rate": 9.9931915393079e-06,
      "loss": 0.0001,
      "step": 1053
    },
    {
      "epoch": 5.22,
      "grad_norm": 0.06926476210355759,
      "learning_rate": 9.990057895049729e-06,
      "loss": 0.0012,
      "step": 1054
    },
    {
      "epoch": 5.22,
      "grad_norm": 0.08432062715291977,
      "learning_rate": 9.986924250791556e-06,
      "loss": 0.0006,
      "step": 1055
    },
    {
      "epoch": 5.23,
      "grad_norm": 0.6996254920959473,
      "learning_rate": 9.983790606533385e-06,
      "loss": 0.0008,
      "step": 1056
    },
    {
      "epoch": 5.23,
      "grad_norm": 0.3922226130962372,
      "learning_rate": 9.980656962275216e-06,
      "loss": 0.003,
      "step": 1057
    },
    {
      "epoch": 5.24,
      "grad_norm": 0.2075994312763214,
      "learning_rate": 9.977523318017043e-06,
      "loss": 0.0009,
      "step": 1058
    },
    {
      "epoch": 5.24,
      "grad_norm": 0.09908422082662582,
      "learning_rate": 9.974389673758872e-06,
      "loss": 0.0021,
      "step": 1059
    },
    {
      "epoch": 5.25,
      "grad_norm": 0.05718173086643219,
      "learning_rate": 9.971256029500701e-06,
      "loss": 0.0011,
      "step": 1060
    },
    {
      "epoch": 5.25,
      "grad_norm": 0.46550452709198,
      "learning_rate": 9.968122385242531e-06,
      "loss": 0.0077,
      "step": 1061
    },
    {
      "epoch": 5.26,
      "grad_norm": 0.3536960780620575,
      "learning_rate": 9.964988740984359e-06,
      "loss": 0.0007,
      "step": 1062
    },
    {
      "epoch": 5.26,
      "grad_norm": 0.3480602204799652,
      "learning_rate": 9.961855096726188e-06,
      "loss": 0.0028,
      "step": 1063
    },
    {
      "epoch": 5.27,
      "grad_norm": 0.5524848699569702,
      "learning_rate": 9.958721452468017e-06,
      "loss": 0.0021,
      "step": 1064
    },
    {
      "epoch": 5.27,
      "grad_norm": 1.776950478553772,
      "learning_rate": 9.955587808209845e-06,
      "loss": 0.0151,
      "step": 1065
    },
    {
      "epoch": 5.28,
      "grad_norm": 0.1563672423362732,
      "learning_rate": 9.952454163951674e-06,
      "loss": 0.0024,
      "step": 1066
    },
    {
      "epoch": 5.28,
      "grad_norm": 0.07350972294807434,
      "learning_rate": 9.949320519693503e-06,
      "loss": 0.0002,
      "step": 1067
    },
    {
      "epoch": 5.29,
      "grad_norm": 0.04931764677166939,
      "learning_rate": 9.946186875435332e-06,
      "loss": 0.0006,
      "step": 1068
    },
    {
      "epoch": 5.29,
      "grad_norm": 0.10255689173936844,
      "learning_rate": 9.943053231177161e-06,
      "loss": 0.0022,
      "step": 1069
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.07300114631652832,
      "learning_rate": 9.93991958691899e-06,
      "loss": 0.0004,
      "step": 1070
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.10363590717315674,
      "learning_rate": 9.936785942660819e-06,
      "loss": 0.0012,
      "step": 1071
    },
    {
      "epoch": 5.31,
      "grad_norm": 0.0520465150475502,
      "learning_rate": 9.933652298402646e-06,
      "loss": 0.0004,
      "step": 1072
    },
    {
      "epoch": 5.31,
      "grad_norm": 0.1241404339671135,
      "learning_rate": 9.930518654144475e-06,
      "loss": 0.0012,
      "step": 1073
    },
    {
      "epoch": 5.32,
      "grad_norm": 0.06866321712732315,
      "learning_rate": 9.927385009886306e-06,
      "loss": 0.0003,
      "step": 1074
    },
    {
      "epoch": 5.32,
      "grad_norm": 0.09852733463048935,
      "learning_rate": 9.924251365628133e-06,
      "loss": 0.0003,
      "step": 1075
    },
    {
      "epoch": 5.33,
      "grad_norm": 0.4700741171836853,
      "learning_rate": 9.921117721369962e-06,
      "loss": 0.0021,
      "step": 1076
    },
    {
      "epoch": 5.33,
      "grad_norm": 0.18251173198223114,
      "learning_rate": 9.917984077111791e-06,
      "loss": 0.0009,
      "step": 1077
    },
    {
      "epoch": 5.34,
      "grad_norm": 0.05008579418063164,
      "learning_rate": 9.91485043285362e-06,
      "loss": 0.0005,
      "step": 1078
    },
    {
      "epoch": 5.34,
      "grad_norm": 0.06560137867927551,
      "learning_rate": 9.911716788595449e-06,
      "loss": 0.0007,
      "step": 1079
    },
    {
      "epoch": 5.35,
      "grad_norm": 0.06501755863428116,
      "learning_rate": 9.908583144337278e-06,
      "loss": 0.0004,
      "step": 1080
    },
    {
      "epoch": 5.35,
      "grad_norm": 0.04674607515335083,
      "learning_rate": 9.905449500079107e-06,
      "loss": 0.0003,
      "step": 1081
    },
    {
      "epoch": 5.36,
      "grad_norm": 0.5025933384895325,
      "learning_rate": 9.902315855820936e-06,
      "loss": 0.0008,
      "step": 1082
    },
    {
      "epoch": 5.36,
      "grad_norm": 0.06976941227912903,
      "learning_rate": 9.899182211562764e-06,
      "loss": 0.0007,
      "step": 1083
    },
    {
      "epoch": 5.37,
      "grad_norm": 0.09702540934085846,
      "learning_rate": 9.896048567304593e-06,
      "loss": 0.0012,
      "step": 1084
    },
    {
      "epoch": 5.37,
      "grad_norm": 0.4600895941257477,
      "learning_rate": 9.89291492304642e-06,
      "loss": 0.0035,
      "step": 1085
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.10747282952070236,
      "learning_rate": 9.889781278788251e-06,
      "loss": 0.0007,
      "step": 1086
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.17369237542152405,
      "learning_rate": 9.88664763453008e-06,
      "loss": 0.0009,
      "step": 1087
    },
    {
      "epoch": 5.39,
      "grad_norm": 0.06625343859195709,
      "learning_rate": 9.883513990271907e-06,
      "loss": 0.0005,
      "step": 1088
    },
    {
      "epoch": 5.39,
      "grad_norm": 0.1844579577445984,
      "learning_rate": 9.880380346013736e-06,
      "loss": 0.0016,
      "step": 1089
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.07982013374567032,
      "learning_rate": 9.877246701755567e-06,
      "loss": 0.0017,
      "step": 1090
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.06205949932336807,
      "learning_rate": 9.874113057497394e-06,
      "loss": 0.0007,
      "step": 1091
    },
    {
      "epoch": 5.41,
      "grad_norm": 0.5191577076911926,
      "learning_rate": 9.870979413239223e-06,
      "loss": 0.0011,
      "step": 1092
    },
    {
      "epoch": 5.41,
      "grad_norm": 0.11758501082658768,
      "learning_rate": 9.867845768981052e-06,
      "loss": 0.0009,
      "step": 1093
    },
    {
      "epoch": 5.42,
      "grad_norm": 0.048592131584882736,
      "learning_rate": 9.864712124722883e-06,
      "loss": 0.0013,
      "step": 1094
    },
    {
      "epoch": 5.42,
      "grad_norm": 1.0238181352615356,
      "learning_rate": 9.86157848046471e-06,
      "loss": 0.0033,
      "step": 1095
    },
    {
      "epoch": 5.43,
      "grad_norm": 0.32875964045524597,
      "learning_rate": 9.858444836206539e-06,
      "loss": 0.0006,
      "step": 1096
    },
    {
      "epoch": 5.43,
      "grad_norm": 0.4964665174484253,
      "learning_rate": 9.855311191948368e-06,
      "loss": 0.0029,
      "step": 1097
    },
    {
      "epoch": 5.44,
      "grad_norm": 0.12145067751407623,
      "learning_rate": 9.852177547690195e-06,
      "loss": 0.0009,
      "step": 1098
    },
    {
      "epoch": 5.44,
      "grad_norm": 0.2871690094470978,
      "learning_rate": 9.849043903432026e-06,
      "loss": 0.0017,
      "step": 1099
    },
    {
      "epoch": 5.45,
      "grad_norm": 0.1191554144024849,
      "learning_rate": 9.845910259173854e-06,
      "loss": 0.0026,
      "step": 1100
    },
    {
      "epoch": 5.45,
      "grad_norm": 0.06786814332008362,
      "learning_rate": 9.842776614915683e-06,
      "loss": 0.0009,
      "step": 1101
    },
    {
      "epoch": 5.46,
      "grad_norm": 0.07721118628978729,
      "learning_rate": 9.83964297065751e-06,
      "loss": 0.0007,
      "step": 1102
    },
    {
      "epoch": 5.46,
      "grad_norm": 0.02987983636558056,
      "learning_rate": 9.836509326399341e-06,
      "loss": 0.0002,
      "step": 1103
    },
    {
      "epoch": 5.47,
      "grad_norm": 0.16932423412799835,
      "learning_rate": 9.83337568214117e-06,
      "loss": 0.0012,
      "step": 1104
    },
    {
      "epoch": 5.47,
      "grad_norm": 0.06452585011720657,
      "learning_rate": 9.830242037882997e-06,
      "loss": 0.0006,
      "step": 1105
    },
    {
      "epoch": 5.48,
      "grad_norm": 0.053415339440107346,
      "learning_rate": 9.827108393624826e-06,
      "loss": 0.0004,
      "step": 1106
    },
    {
      "epoch": 5.48,
      "grad_norm": 0.10779280960559845,
      "learning_rate": 9.823974749366657e-06,
      "loss": 0.0008,
      "step": 1107
    },
    {
      "epoch": 5.49,
      "grad_norm": 0.07948265969753265,
      "learning_rate": 9.820841105108484e-06,
      "loss": 0.0005,
      "step": 1108
    },
    {
      "epoch": 5.49,
      "grad_norm": 0.07489346712827682,
      "learning_rate": 9.817707460850313e-06,
      "loss": 0.0002,
      "step": 1109
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.11462302505970001,
      "learning_rate": 9.814573816592142e-06,
      "loss": 0.0025,
      "step": 1110
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.3671611547470093,
      "learning_rate": 9.811440172333971e-06,
      "loss": 0.0021,
      "step": 1111
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.08901914954185486,
      "learning_rate": 9.8083065280758e-06,
      "loss": 0.0003,
      "step": 1112
    },
    {
      "epoch": 5.51,
      "grad_norm": 0.07918983697891235,
      "learning_rate": 9.805172883817629e-06,
      "loss": 0.0005,
      "step": 1113
    },
    {
      "epoch": 5.51,
      "grad_norm": 0.14054358005523682,
      "learning_rate": 9.802039239559458e-06,
      "loss": 0.0012,
      "step": 1114
    },
    {
      "epoch": 5.52,
      "grad_norm": 0.13522937893867493,
      "learning_rate": 9.798905595301287e-06,
      "loss": 0.0024,
      "step": 1115
    },
    {
      "epoch": 5.52,
      "grad_norm": 0.2978847622871399,
      "learning_rate": 9.795771951043116e-06,
      "loss": 0.0013,
      "step": 1116
    },
    {
      "epoch": 5.53,
      "grad_norm": 0.03400041162967682,
      "learning_rate": 9.792638306784944e-06,
      "loss": 0.001,
      "step": 1117
    },
    {
      "epoch": 5.53,
      "grad_norm": 0.1356572061777115,
      "learning_rate": 9.789504662526772e-06,
      "loss": 0.0008,
      "step": 1118
    },
    {
      "epoch": 5.54,
      "grad_norm": 4.149552822113037,
      "learning_rate": 9.786371018268602e-06,
      "loss": 0.0068,
      "step": 1119
    },
    {
      "epoch": 5.54,
      "grad_norm": 0.44581738114356995,
      "learning_rate": 9.783237374010431e-06,
      "loss": 0.0006,
      "step": 1120
    },
    {
      "epoch": 5.55,
      "grad_norm": 0.1605798602104187,
      "learning_rate": 9.780103729752258e-06,
      "loss": 0.0007,
      "step": 1121
    },
    {
      "epoch": 5.55,
      "grad_norm": 0.08159391582012177,
      "learning_rate": 9.776970085494087e-06,
      "loss": 0.0008,
      "step": 1122
    },
    {
      "epoch": 5.56,
      "grad_norm": 0.06133635342121124,
      "learning_rate": 9.773836441235916e-06,
      "loss": 0.0009,
      "step": 1123
    },
    {
      "epoch": 5.56,
      "grad_norm": 0.07942347973585129,
      "learning_rate": 9.770702796977745e-06,
      "loss": 0.0008,
      "step": 1124
    },
    {
      "epoch": 5.57,
      "grad_norm": 0.23408243060112,
      "learning_rate": 9.767569152719574e-06,
      "loss": 0.0022,
      "step": 1125
    },
    {
      "epoch": 5.57,
      "grad_norm": 0.1679341048002243,
      "learning_rate": 9.764435508461403e-06,
      "loss": 0.0008,
      "step": 1126
    },
    {
      "epoch": 5.58,
      "grad_norm": 0.10121217370033264,
      "learning_rate": 9.761301864203232e-06,
      "loss": 0.0005,
      "step": 1127
    },
    {
      "epoch": 5.58,
      "grad_norm": 0.06341489404439926,
      "learning_rate": 9.758168219945061e-06,
      "loss": 0.0003,
      "step": 1128
    },
    {
      "epoch": 5.59,
      "grad_norm": 2.859222650527954,
      "learning_rate": 9.75503457568689e-06,
      "loss": 0.0051,
      "step": 1129
    },
    {
      "epoch": 5.59,
      "grad_norm": 0.13927514851093292,
      "learning_rate": 9.751900931428719e-06,
      "loss": 0.0014,
      "step": 1130
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.022672904655337334,
      "learning_rate": 9.748767287170546e-06,
      "loss": 0.0002,
      "step": 1131
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.10486946254968643,
      "learning_rate": 9.745633642912377e-06,
      "loss": 0.0016,
      "step": 1132
    },
    {
      "epoch": 5.61,
      "grad_norm": 0.10309460014104843,
      "learning_rate": 9.742499998654206e-06,
      "loss": 0.0007,
      "step": 1133
    },
    {
      "epoch": 5.61,
      "grad_norm": 0.1305740624666214,
      "learning_rate": 9.739366354396034e-06,
      "loss": 0.0021,
      "step": 1134
    },
    {
      "epoch": 5.62,
      "grad_norm": 0.8879368901252747,
      "learning_rate": 9.736232710137862e-06,
      "loss": 0.0025,
      "step": 1135
    },
    {
      "epoch": 5.62,
      "grad_norm": 0.18100330233573914,
      "learning_rate": 9.733099065879692e-06,
      "loss": 0.0013,
      "step": 1136
    },
    {
      "epoch": 5.63,
      "grad_norm": 0.21542318165302277,
      "learning_rate": 9.729965421621521e-06,
      "loss": 0.0014,
      "step": 1137
    },
    {
      "epoch": 5.63,
      "grad_norm": 0.05076437443494797,
      "learning_rate": 9.726831777363349e-06,
      "loss": 0.0002,
      "step": 1138
    },
    {
      "epoch": 5.64,
      "grad_norm": 0.07321081310510635,
      "learning_rate": 9.723698133105177e-06,
      "loss": 0.0018,
      "step": 1139
    },
    {
      "epoch": 5.64,
      "grad_norm": 0.014493919909000397,
      "learning_rate": 9.720564488847008e-06,
      "loss": 0.0001,
      "step": 1140
    },
    {
      "epoch": 5.65,
      "grad_norm": 0.221954345703125,
      "learning_rate": 9.717430844588835e-06,
      "loss": 0.003,
      "step": 1141
    },
    {
      "epoch": 5.65,
      "grad_norm": 0.19959115982055664,
      "learning_rate": 9.714297200330664e-06,
      "loss": 0.0012,
      "step": 1142
    },
    {
      "epoch": 5.66,
      "grad_norm": 1.541774868965149,
      "learning_rate": 9.711163556072493e-06,
      "loss": 0.0102,
      "step": 1143
    },
    {
      "epoch": 5.66,
      "grad_norm": 0.13119812309741974,
      "learning_rate": 9.708029911814322e-06,
      "loss": 0.0013,
      "step": 1144
    },
    {
      "epoch": 5.67,
      "grad_norm": 0.05773743987083435,
      "learning_rate": 9.704896267556151e-06,
      "loss": 0.0005,
      "step": 1145
    },
    {
      "epoch": 5.67,
      "grad_norm": 0.10364692658185959,
      "learning_rate": 9.70176262329798e-06,
      "loss": 0.0014,
      "step": 1146
    },
    {
      "epoch": 5.68,
      "grad_norm": 0.06675044447183609,
      "learning_rate": 9.698628979039809e-06,
      "loss": 0.001,
      "step": 1147
    },
    {
      "epoch": 5.68,
      "grad_norm": 0.06865566223859787,
      "learning_rate": 9.695495334781636e-06,
      "loss": 0.0004,
      "step": 1148
    },
    {
      "epoch": 5.69,
      "grad_norm": 0.1334363967180252,
      "learning_rate": 9.692361690523467e-06,
      "loss": 0.0007,
      "step": 1149
    },
    {
      "epoch": 5.69,
      "grad_norm": 0.11968280375003815,
      "learning_rate": 9.689228046265296e-06,
      "loss": 0.0012,
      "step": 1150
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.06916464865207672,
      "learning_rate": 9.686094402007123e-06,
      "loss": 0.0003,
      "step": 1151
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.36292409896850586,
      "learning_rate": 9.682960757748952e-06,
      "loss": 0.0024,
      "step": 1152
    },
    {
      "epoch": 5.71,
      "grad_norm": 0.08964971452951431,
      "learning_rate": 9.679827113490782e-06,
      "loss": 0.0018,
      "step": 1153
    },
    {
      "epoch": 5.71,
      "grad_norm": 0.38522523641586304,
      "learning_rate": 9.67669346923261e-06,
      "loss": 0.0027,
      "step": 1154
    },
    {
      "epoch": 5.72,
      "grad_norm": 0.13483880460262299,
      "learning_rate": 9.673559824974439e-06,
      "loss": 0.0023,
      "step": 1155
    },
    {
      "epoch": 5.72,
      "grad_norm": 0.12277685850858688,
      "learning_rate": 9.670426180716267e-06,
      "loss": 0.0013,
      "step": 1156
    },
    {
      "epoch": 5.73,
      "grad_norm": 0.5885747075080872,
      "learning_rate": 9.667292536458096e-06,
      "loss": 0.003,
      "step": 1157
    },
    {
      "epoch": 5.73,
      "grad_norm": 0.10298590362071991,
      "learning_rate": 9.664158892199925e-06,
      "loss": 0.0013,
      "step": 1158
    },
    {
      "epoch": 5.74,
      "grad_norm": 0.4284578561782837,
      "learning_rate": 9.661025247941754e-06,
      "loss": 0.0026,
      "step": 1159
    },
    {
      "epoch": 5.74,
      "grad_norm": 0.14644823968410492,
      "learning_rate": 9.657891603683583e-06,
      "loss": 0.0024,
      "step": 1160
    },
    {
      "epoch": 5.75,
      "grad_norm": 0.13728320598602295,
      "learning_rate": 9.654757959425412e-06,
      "loss": 0.0018,
      "step": 1161
    },
    {
      "epoch": 5.75,
      "grad_norm": 0.08643432706594467,
      "learning_rate": 9.651624315167241e-06,
      "loss": 0.0033,
      "step": 1162
    },
    {
      "epoch": 5.76,
      "grad_norm": 0.03757728636264801,
      "learning_rate": 9.64849067090907e-06,
      "loss": 0.0003,
      "step": 1163
    },
    {
      "epoch": 5.76,
      "grad_norm": 0.10620026290416718,
      "learning_rate": 9.645357026650897e-06,
      "loss": 0.0021,
      "step": 1164
    },
    {
      "epoch": 5.77,
      "grad_norm": 0.19958724081516266,
      "learning_rate": 9.642223382392728e-06,
      "loss": 0.0018,
      "step": 1165
    },
    {
      "epoch": 5.77,
      "grad_norm": 0.12463889271020889,
      "learning_rate": 9.639089738134557e-06,
      "loss": 0.0013,
      "step": 1166
    },
    {
      "epoch": 5.78,
      "grad_norm": 0.10599002987146378,
      "learning_rate": 9.635956093876384e-06,
      "loss": 0.002,
      "step": 1167
    },
    {
      "epoch": 5.78,
      "grad_norm": 0.783067524433136,
      "learning_rate": 9.632822449618213e-06,
      "loss": 0.0062,
      "step": 1168
    },
    {
      "epoch": 5.79,
      "grad_norm": 0.15511901676654816,
      "learning_rate": 9.629688805360043e-06,
      "loss": 0.0015,
      "step": 1169
    },
    {
      "epoch": 5.79,
      "grad_norm": 0.09003478288650513,
      "learning_rate": 9.626555161101872e-06,
      "loss": 0.0015,
      "step": 1170
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.15353353321552277,
      "learning_rate": 9.6234215168437e-06,
      "loss": 0.0012,
      "step": 1171
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.1537737250328064,
      "learning_rate": 9.620287872585529e-06,
      "loss": 0.0019,
      "step": 1172
    },
    {
      "epoch": 5.81,
      "grad_norm": 0.22379910945892334,
      "learning_rate": 9.617154228327357e-06,
      "loss": 0.0019,
      "step": 1173
    },
    {
      "epoch": 5.81,
      "grad_norm": 1.4833083152770996,
      "learning_rate": 9.614020584069186e-06,
      "loss": 0.0028,
      "step": 1174
    },
    {
      "epoch": 5.82,
      "grad_norm": 0.04819599166512489,
      "learning_rate": 9.610886939811015e-06,
      "loss": 0.0002,
      "step": 1175
    },
    {
      "epoch": 5.82,
      "grad_norm": 0.09544014930725098,
      "learning_rate": 9.607753295552844e-06,
      "loss": 0.0016,
      "step": 1176
    },
    {
      "epoch": 5.83,
      "grad_norm": 0.09164503961801529,
      "learning_rate": 9.604619651294673e-06,
      "loss": 0.0018,
      "step": 1177
    },
    {
      "epoch": 5.83,
      "grad_norm": 0.1084987223148346,
      "learning_rate": 9.601486007036502e-06,
      "loss": 0.0018,
      "step": 1178
    },
    {
      "epoch": 5.84,
      "grad_norm": 0.20399239659309387,
      "learning_rate": 9.598352362778331e-06,
      "loss": 0.0014,
      "step": 1179
    },
    {
      "epoch": 5.84,
      "grad_norm": 0.05658655986189842,
      "learning_rate": 9.59521871852016e-06,
      "loss": 0.0003,
      "step": 1180
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.6490488052368164,
      "learning_rate": 9.592085074261987e-06,
      "loss": 0.0036,
      "step": 1181
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.07432391494512558,
      "learning_rate": 9.588951430003818e-06,
      "loss": 0.0004,
      "step": 1182
    },
    {
      "epoch": 5.86,
      "grad_norm": 0.17615073919296265,
      "learning_rate": 9.585817785745647e-06,
      "loss": 0.0021,
      "step": 1183
    },
    {
      "epoch": 5.86,
      "grad_norm": 0.14254455268383026,
      "learning_rate": 9.582684141487474e-06,
      "loss": 0.0022,
      "step": 1184
    },
    {
      "epoch": 5.87,
      "grad_norm": 0.15713784098625183,
      "learning_rate": 9.579550497229303e-06,
      "loss": 0.0012,
      "step": 1185
    },
    {
      "epoch": 5.87,
      "grad_norm": 0.08358027786016464,
      "learning_rate": 9.576416852971133e-06,
      "loss": 0.0004,
      "step": 1186
    },
    {
      "epoch": 5.88,
      "grad_norm": 0.07834039628505707,
      "learning_rate": 9.57328320871296e-06,
      "loss": 0.0007,
      "step": 1187
    },
    {
      "epoch": 5.88,
      "grad_norm": 0.11377963423728943,
      "learning_rate": 9.57014956445479e-06,
      "loss": 0.0019,
      "step": 1188
    },
    {
      "epoch": 5.89,
      "grad_norm": 0.10509398579597473,
      "learning_rate": 9.567015920196619e-06,
      "loss": 0.001,
      "step": 1189
    },
    {
      "epoch": 5.89,
      "grad_norm": 0.280445396900177,
      "learning_rate": 9.563882275938447e-06,
      "loss": 0.0022,
      "step": 1190
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.677524745464325,
      "learning_rate": 9.560748631680276e-06,
      "loss": 0.0033,
      "step": 1191
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.3794487416744232,
      "learning_rate": 9.557614987422105e-06,
      "loss": 0.0027,
      "step": 1192
    },
    {
      "epoch": 5.91,
      "grad_norm": 0.10558668524026871,
      "learning_rate": 9.554481343163934e-06,
      "loss": 0.0021,
      "step": 1193
    },
    {
      "epoch": 5.91,
      "grad_norm": 0.10567694157361984,
      "learning_rate": 9.551347698905763e-06,
      "loss": 0.0018,
      "step": 1194
    },
    {
      "epoch": 5.92,
      "grad_norm": 0.299349844455719,
      "learning_rate": 9.548214054647592e-06,
      "loss": 0.0015,
      "step": 1195
    },
    {
      "epoch": 5.92,
      "grad_norm": 0.6275417804718018,
      "learning_rate": 9.545080410389421e-06,
      "loss": 0.0066,
      "step": 1196
    },
    {
      "epoch": 5.93,
      "grad_norm": 0.11963768303394318,
      "learning_rate": 9.541946766131248e-06,
      "loss": 0.0015,
      "step": 1197
    },
    {
      "epoch": 5.93,
      "grad_norm": 0.04511168971657753,
      "learning_rate": 9.538813121873077e-06,
      "loss": 0.0004,
      "step": 1198
    },
    {
      "epoch": 5.94,
      "grad_norm": 0.22727777063846588,
      "learning_rate": 9.535679477614908e-06,
      "loss": 0.0019,
      "step": 1199
    },
    {
      "epoch": 5.94,
      "grad_norm": 0.18388329446315765,
      "learning_rate": 9.532545833356735e-06,
      "loss": 0.0034,
      "step": 1200
    },
    {
      "epoch": 5.95,
      "grad_norm": 0.12182167917490005,
      "learning_rate": 9.529412189098564e-06,
      "loss": 0.001,
      "step": 1201
    },
    {
      "epoch": 5.95,
      "grad_norm": 0.7836220264434814,
      "learning_rate": 9.526278544840393e-06,
      "loss": 0.0034,
      "step": 1202
    },
    {
      "epoch": 5.96,
      "grad_norm": 0.0982346311211586,
      "learning_rate": 9.523144900582224e-06,
      "loss": 0.0014,
      "step": 1203
    },
    {
      "epoch": 5.96,
      "grad_norm": 0.13183607161045074,
      "learning_rate": 9.52001125632405e-06,
      "loss": 0.0008,
      "step": 1204
    },
    {
      "epoch": 5.97,
      "grad_norm": 0.1630563586950302,
      "learning_rate": 9.51687761206588e-06,
      "loss": 0.0022,
      "step": 1205
    },
    {
      "epoch": 5.97,
      "grad_norm": 0.276093065738678,
      "learning_rate": 9.513743967807709e-06,
      "loss": 0.0035,
      "step": 1206
    },
    {
      "epoch": 5.98,
      "grad_norm": 0.13917453587055206,
      "learning_rate": 9.510610323549538e-06,
      "loss": 0.0009,
      "step": 1207
    },
    {
      "epoch": 5.98,
      "grad_norm": 2.921734571456909,
      "learning_rate": 9.507476679291366e-06,
      "loss": 0.0027,
      "step": 1208
    },
    {
      "epoch": 5.99,
      "grad_norm": 0.10660935193300247,
      "learning_rate": 9.504343035033195e-06,
      "loss": 0.0012,
      "step": 1209
    },
    {
      "epoch": 5.99,
      "grad_norm": 0.3314594626426697,
      "learning_rate": 9.501209390775024e-06,
      "loss": 0.0015,
      "step": 1210
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.030933953821659088,
      "learning_rate": 9.498075746516853e-06,
      "loss": 0.0001,
      "step": 1211
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.44004684686660767,
      "learning_rate": 9.494942102258682e-06,
      "loss": 0.0027,
      "step": 1212
    },
    {
      "epoch": 6.0,
      "eval_AVGf1": 0.9270583819922669,
      "eval_DIAG.f1": 0.8922651933701659,
      "eval_DIAG.precision": 0.9047619047619048,
      "eval_DIAG.recall": 0.8801089918256131,
      "eval_MED.f1": 0.96,
      "eval_MED.precision": 0.96,
      "eval_MED.recall": 0.96,
      "eval_TREAT.f1": 0.9289099526066351,
      "eval_TREAT.precision": 0.9333333333333333,
      "eval_TREAT.recall": 0.9245283018867925,
      "eval_accuracy": 0.9543657923104564,
      "eval_f1": 0.8073278584965257,
      "eval_loss": 0.01961512863636017,
      "eval_precision": 0.726962457337884,
      "eval_recall": 0.9076704545454546,
      "eval_runtime": 964.899,
      "eval_samples_per_second": 0.744,
      "eval_steps_per_second": 0.093,
      "step": 1212
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.20426569879055023,
      "learning_rate": 9.491808458000511e-06,
      "loss": 0.0019,
      "step": 1213
    },
    {
      "epoch": 6.01,
      "grad_norm": 0.11194341629743576,
      "learning_rate": 9.488674813742338e-06,
      "loss": 0.0004,
      "step": 1214
    },
    {
      "epoch": 6.01,
      "grad_norm": 0.21565930545330048,
      "learning_rate": 9.485541169484169e-06,
      "loss": 0.0005,
      "step": 1215
    },
    {
      "epoch": 6.02,
      "grad_norm": 0.05033720284700394,
      "learning_rate": 9.482407525225998e-06,
      "loss": 0.0007,
      "step": 1216
    },
    {
      "epoch": 6.02,
      "grad_norm": 0.1762872040271759,
      "learning_rate": 9.479273880967825e-06,
      "loss": 0.0008,
      "step": 1217
    },
    {
      "epoch": 6.03,
      "grad_norm": 0.049365490674972534,
      "learning_rate": 9.476140236709654e-06,
      "loss": 0.0007,
      "step": 1218
    },
    {
      "epoch": 6.03,
      "grad_norm": 0.10195271670818329,
      "learning_rate": 9.473006592451485e-06,
      "loss": 0.0004,
      "step": 1219
    },
    {
      "epoch": 6.04,
      "grad_norm": 0.26775848865509033,
      "learning_rate": 9.469872948193312e-06,
      "loss": 0.007,
      "step": 1220
    },
    {
      "epoch": 6.04,
      "grad_norm": 0.04559099301695824,
      "learning_rate": 9.46673930393514e-06,
      "loss": 0.0002,
      "step": 1221
    },
    {
      "epoch": 6.05,
      "grad_norm": 0.007405612152069807,
      "learning_rate": 9.46360565967697e-06,
      "loss": 0.0001,
      "step": 1222
    },
    {
      "epoch": 6.05,
      "grad_norm": 0.07509858906269073,
      "learning_rate": 9.460472015418799e-06,
      "loss": 0.0003,
      "step": 1223
    },
    {
      "epoch": 6.06,
      "grad_norm": 0.04566282778978348,
      "learning_rate": 9.457338371160628e-06,
      "loss": 0.0003,
      "step": 1224
    },
    {
      "epoch": 6.06,
      "grad_norm": 0.06073032319545746,
      "learning_rate": 9.454204726902456e-06,
      "loss": 0.0015,
      "step": 1225
    },
    {
      "epoch": 6.07,
      "grad_norm": 0.2727704346179962,
      "learning_rate": 9.451071082644285e-06,
      "loss": 0.0008,
      "step": 1226
    },
    {
      "epoch": 6.07,
      "grad_norm": 0.053360383957624435,
      "learning_rate": 9.447937438386113e-06,
      "loss": 0.0002,
      "step": 1227
    },
    {
      "epoch": 6.08,
      "grad_norm": 0.056087981909513474,
      "learning_rate": 9.444803794127943e-06,
      "loss": 0.0003,
      "step": 1228
    },
    {
      "epoch": 6.08,
      "grad_norm": 0.11166010051965714,
      "learning_rate": 9.441670149869772e-06,
      "loss": 0.0003,
      "step": 1229
    },
    {
      "epoch": 6.09,
      "grad_norm": 0.2141570895910263,
      "learning_rate": 9.4385365056116e-06,
      "loss": 0.0006,
      "step": 1230
    },
    {
      "epoch": 6.09,
      "grad_norm": 0.035683523863554,
      "learning_rate": 9.435402861353428e-06,
      "loss": 0.0011,
      "step": 1231
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.20922675728797913,
      "learning_rate": 9.432269217095259e-06,
      "loss": 0.0004,
      "step": 1232
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.06270837038755417,
      "learning_rate": 9.429135572837086e-06,
      "loss": 0.0014,
      "step": 1233
    },
    {
      "epoch": 6.11,
      "grad_norm": 0.08441109210252762,
      "learning_rate": 9.426001928578915e-06,
      "loss": 0.0018,
      "step": 1234
    },
    {
      "epoch": 6.11,
      "grad_norm": 0.1049419641494751,
      "learning_rate": 9.422868284320744e-06,
      "loss": 0.0009,
      "step": 1235
    },
    {
      "epoch": 6.12,
      "grad_norm": 0.16950374841690063,
      "learning_rate": 9.419734640062575e-06,
      "loss": 0.0024,
      "step": 1236
    },
    {
      "epoch": 6.12,
      "grad_norm": 0.008717745542526245,
      "learning_rate": 9.416600995804402e-06,
      "loss": 0.0001,
      "step": 1237
    },
    {
      "epoch": 6.13,
      "grad_norm": 0.10020992159843445,
      "learning_rate": 9.41346735154623e-06,
      "loss": 0.0007,
      "step": 1238
    },
    {
      "epoch": 6.13,
      "grad_norm": 0.04810984805226326,
      "learning_rate": 9.41033370728806e-06,
      "loss": 0.0001,
      "step": 1239
    },
    {
      "epoch": 6.14,
      "grad_norm": 0.38938790559768677,
      "learning_rate": 9.407200063029889e-06,
      "loss": 0.0022,
      "step": 1240
    },
    {
      "epoch": 6.14,
      "grad_norm": 0.09656047821044922,
      "learning_rate": 9.404066418771718e-06,
      "loss": 0.0006,
      "step": 1241
    },
    {
      "epoch": 6.15,
      "grad_norm": 0.22869130969047546,
      "learning_rate": 9.400932774513546e-06,
      "loss": 0.0015,
      "step": 1242
    },
    {
      "epoch": 6.15,
      "grad_norm": 0.17612148821353912,
      "learning_rate": 9.397799130255375e-06,
      "loss": 0.0016,
      "step": 1243
    },
    {
      "epoch": 6.16,
      "grad_norm": 0.05888248607516289,
      "learning_rate": 9.394665485997204e-06,
      "loss": 0.0007,
      "step": 1244
    },
    {
      "epoch": 6.16,
      "grad_norm": 0.16711938381195068,
      "learning_rate": 9.391531841739033e-06,
      "loss": 0.0012,
      "step": 1245
    },
    {
      "epoch": 6.17,
      "grad_norm": 0.047442857176065445,
      "learning_rate": 9.388398197480862e-06,
      "loss": 0.0003,
      "step": 1246
    },
    {
      "epoch": 6.17,
      "grad_norm": 0.04538818076252937,
      "learning_rate": 9.38526455322269e-06,
      "loss": 0.0009,
      "step": 1247
    },
    {
      "epoch": 6.18,
      "grad_norm": 0.02311217226088047,
      "learning_rate": 9.382130908964518e-06,
      "loss": 0.0001,
      "step": 1248
    },
    {
      "epoch": 6.18,
      "grad_norm": 0.09773791581392288,
      "learning_rate": 9.378997264706349e-06,
      "loss": 0.0002,
      "step": 1249
    },
    {
      "epoch": 6.19,
      "grad_norm": 0.07052435725927353,
      "learning_rate": 9.375863620448176e-06,
      "loss": 0.001,
      "step": 1250
    },
    {
      "epoch": 6.19,
      "grad_norm": 0.16447028517723083,
      "learning_rate": 9.372729976190005e-06,
      "loss": 0.0003,
      "step": 1251
    },
    {
      "epoch": 6.2,
      "grad_norm": 0.05764441192150116,
      "learning_rate": 9.369596331931834e-06,
      "loss": 0.0001,
      "step": 1252
    },
    {
      "epoch": 6.2,
      "grad_norm": 0.0552435964345932,
      "learning_rate": 9.366462687673663e-06,
      "loss": 0.0001,
      "step": 1253
    },
    {
      "epoch": 6.21,
      "grad_norm": 0.10032617300748825,
      "learning_rate": 9.363329043415492e-06,
      "loss": 0.0004,
      "step": 1254
    },
    {
      "epoch": 6.21,
      "grad_norm": 0.008596989326179028,
      "learning_rate": 9.36019539915732e-06,
      "loss": 0.0003,
      "step": 1255
    },
    {
      "epoch": 6.22,
      "grad_norm": 0.06413231790065765,
      "learning_rate": 9.35706175489915e-06,
      "loss": 0.0007,
      "step": 1256
    },
    {
      "epoch": 6.22,
      "grad_norm": 0.010608188807964325,
      "learning_rate": 9.353928110640979e-06,
      "loss": 0.0,
      "step": 1257
    },
    {
      "epoch": 6.23,
      "grad_norm": 0.02950921840965748,
      "learning_rate": 9.350794466382808e-06,
      "loss": 0.0002,
      "step": 1258
    },
    {
      "epoch": 6.23,
      "grad_norm": 0.012921888381242752,
      "learning_rate": 9.347660822124637e-06,
      "loss": 0.0001,
      "step": 1259
    },
    {
      "epoch": 6.24,
      "grad_norm": 0.1503705233335495,
      "learning_rate": 9.344527177866464e-06,
      "loss": 0.0002,
      "step": 1260
    },
    {
      "epoch": 6.24,
      "grad_norm": 0.239127516746521,
      "learning_rate": 9.341393533608294e-06,
      "loss": 0.0014,
      "step": 1261
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.3748297393321991,
      "learning_rate": 9.338259889350123e-06,
      "loss": 0.0014,
      "step": 1262
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.03920583054423332,
      "learning_rate": 9.33512624509195e-06,
      "loss": 0.0007,
      "step": 1263
    },
    {
      "epoch": 6.26,
      "grad_norm": 0.13970781862735748,
      "learning_rate": 9.33199260083378e-06,
      "loss": 0.0007,
      "step": 1264
    },
    {
      "epoch": 6.26,
      "grad_norm": 0.04359162598848343,
      "learning_rate": 9.32885895657561e-06,
      "loss": 0.0001,
      "step": 1265
    },
    {
      "epoch": 6.27,
      "grad_norm": 0.08924025297164917,
      "learning_rate": 9.325725312317437e-06,
      "loss": 0.0005,
      "step": 1266
    },
    {
      "epoch": 6.27,
      "grad_norm": 0.01958087645471096,
      "learning_rate": 9.322591668059266e-06,
      "loss": 0.0001,
      "step": 1267
    },
    {
      "epoch": 6.28,
      "grad_norm": 0.14290888607501984,
      "learning_rate": 9.319458023801095e-06,
      "loss": 0.0006,
      "step": 1268
    },
    {
      "epoch": 6.28,
      "grad_norm": 0.04646167904138565,
      "learning_rate": 9.316324379542926e-06,
      "loss": 0.0001,
      "step": 1269
    },
    {
      "epoch": 6.29,
      "grad_norm": 0.010225554928183556,
      "learning_rate": 9.313190735284753e-06,
      "loss": 0.0001,
      "step": 1270
    },
    {
      "epoch": 6.29,
      "grad_norm": 0.04966873303055763,
      "learning_rate": 9.310057091026582e-06,
      "loss": 0.0009,
      "step": 1271
    },
    {
      "epoch": 6.3,
      "grad_norm": 3.1506245136260986,
      "learning_rate": 9.30692344676841e-06,
      "loss": 0.0075,
      "step": 1272
    },
    {
      "epoch": 6.3,
      "grad_norm": 0.3142995834350586,
      "learning_rate": 9.303789802510238e-06,
      "loss": 0.002,
      "step": 1273
    },
    {
      "epoch": 6.31,
      "grad_norm": 0.20508092641830444,
      "learning_rate": 9.300656158252069e-06,
      "loss": 0.0019,
      "step": 1274
    },
    {
      "epoch": 6.31,
      "grad_norm": 0.22843855619430542,
      "learning_rate": 9.297522513993898e-06,
      "loss": 0.0012,
      "step": 1275
    },
    {
      "epoch": 6.32,
      "grad_norm": 0.053507156670093536,
      "learning_rate": 9.294388869735725e-06,
      "loss": 0.0007,
      "step": 1276
    },
    {
      "epoch": 6.32,
      "grad_norm": 0.1270291656255722,
      "learning_rate": 9.291255225477554e-06,
      "loss": 0.0003,
      "step": 1277
    },
    {
      "epoch": 6.33,
      "grad_norm": 0.10943729430437088,
      "learning_rate": 9.288121581219384e-06,
      "loss": 0.0004,
      "step": 1278
    },
    {
      "epoch": 6.33,
      "grad_norm": 0.08934703469276428,
      "learning_rate": 9.284987936961213e-06,
      "loss": 0.0018,
      "step": 1279
    },
    {
      "epoch": 6.34,
      "grad_norm": 0.009777539409697056,
      "learning_rate": 9.28185429270304e-06,
      "loss": 0.0001,
      "step": 1280
    },
    {
      "epoch": 6.34,
      "grad_norm": 0.055330950766801834,
      "learning_rate": 9.27872064844487e-06,
      "loss": 0.0002,
      "step": 1281
    },
    {
      "epoch": 6.35,
      "grad_norm": 0.16996553540229797,
      "learning_rate": 9.2755870041867e-06,
      "loss": 0.0003,
      "step": 1282
    },
    {
      "epoch": 6.35,
      "grad_norm": 0.18581070005893707,
      "learning_rate": 9.272453359928527e-06,
      "loss": 0.0009,
      "step": 1283
    },
    {
      "epoch": 6.36,
      "grad_norm": 0.12458633631467819,
      "learning_rate": 9.269319715670356e-06,
      "loss": 0.0003,
      "step": 1284
    },
    {
      "epoch": 6.36,
      "grad_norm": 0.03636595979332924,
      "learning_rate": 9.266186071412185e-06,
      "loss": 0.0009,
      "step": 1285
    },
    {
      "epoch": 6.37,
      "grad_norm": 0.0804811641573906,
      "learning_rate": 9.263052427154014e-06,
      "loss": 0.0005,
      "step": 1286
    },
    {
      "epoch": 6.37,
      "grad_norm": 0.6475043892860413,
      "learning_rate": 9.259918782895843e-06,
      "loss": 0.0021,
      "step": 1287
    },
    {
      "epoch": 6.38,
      "grad_norm": 0.12106894701719284,
      "learning_rate": 9.256785138637672e-06,
      "loss": 0.0008,
      "step": 1288
    },
    {
      "epoch": 6.38,
      "grad_norm": 0.032399192452430725,
      "learning_rate": 9.253651494379501e-06,
      "loss": 0.0002,
      "step": 1289
    },
    {
      "epoch": 6.39,
      "grad_norm": 0.05032498016953468,
      "learning_rate": 9.25051785012133e-06,
      "loss": 0.0004,
      "step": 1290
    },
    {
      "epoch": 6.39,
      "grad_norm": 0.06361556798219681,
      "learning_rate": 9.247384205863159e-06,
      "loss": 0.0004,
      "step": 1291
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.06766597181558609,
      "learning_rate": 9.244250561604988e-06,
      "loss": 0.0007,
      "step": 1292
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.15950827300548553,
      "learning_rate": 9.241116917346815e-06,
      "loss": 0.0007,
      "step": 1293
    },
    {
      "epoch": 6.41,
      "grad_norm": 0.6865561604499817,
      "learning_rate": 9.237983273088645e-06,
      "loss": 0.0022,
      "step": 1294
    },
    {
      "epoch": 6.41,
      "grad_norm": 0.03201720863580704,
      "learning_rate": 9.234849628830474e-06,
      "loss": 0.0006,
      "step": 1295
    },
    {
      "epoch": 6.42,
      "grad_norm": 0.16338810324668884,
      "learning_rate": 9.231715984572302e-06,
      "loss": 0.0007,
      "step": 1296
    },
    {
      "epoch": 6.42,
      "grad_norm": 3.605842113494873,
      "learning_rate": 9.22858234031413e-06,
      "loss": 0.0055,
      "step": 1297
    },
    {
      "epoch": 6.43,
      "grad_norm": 0.09851399809122086,
      "learning_rate": 9.22544869605596e-06,
      "loss": 0.0005,
      "step": 1298
    },
    {
      "epoch": 6.43,
      "grad_norm": 0.15105152130126953,
      "learning_rate": 9.222315051797788e-06,
      "loss": 0.0003,
      "step": 1299
    },
    {
      "epoch": 6.44,
      "grad_norm": 0.12006644159555435,
      "learning_rate": 9.219181407539617e-06,
      "loss": 0.0009,
      "step": 1300
    },
    {
      "epoch": 6.44,
      "grad_norm": 0.11241912096738815,
      "learning_rate": 9.216047763281446e-06,
      "loss": 0.0007,
      "step": 1301
    },
    {
      "epoch": 6.45,
      "grad_norm": 0.7943034768104553,
      "learning_rate": 9.212914119023275e-06,
      "loss": 0.0022,
      "step": 1302
    },
    {
      "epoch": 6.45,
      "grad_norm": 0.16033807396888733,
      "learning_rate": 9.209780474765104e-06,
      "loss": 0.0013,
      "step": 1303
    },
    {
      "epoch": 6.46,
      "grad_norm": 0.15609796345233917,
      "learning_rate": 9.206646830506933e-06,
      "loss": 0.0024,
      "step": 1304
    },
    {
      "epoch": 6.46,
      "grad_norm": 0.14939875900745392,
      "learning_rate": 9.203513186248762e-06,
      "loss": 0.0019,
      "step": 1305
    },
    {
      "epoch": 6.47,
      "grad_norm": 0.0932713970541954,
      "learning_rate": 9.20037954199059e-06,
      "loss": 0.0013,
      "step": 1306
    },
    {
      "epoch": 6.47,
      "grad_norm": 0.03884293884038925,
      "learning_rate": 9.19724589773242e-06,
      "loss": 0.0008,
      "step": 1307
    },
    {
      "epoch": 6.48,
      "grad_norm": 0.10668661445379257,
      "learning_rate": 9.194112253474249e-06,
      "loss": 0.0006,
      "step": 1308
    },
    {
      "epoch": 6.48,
      "grad_norm": 0.22162935137748718,
      "learning_rate": 9.190978609216076e-06,
      "loss": 0.0021,
      "step": 1309
    },
    {
      "epoch": 6.49,
      "grad_norm": 0.19467799365520477,
      "learning_rate": 9.187844964957905e-06,
      "loss": 0.0016,
      "step": 1310
    },
    {
      "epoch": 6.49,
      "grad_norm": 0.1241166964173317,
      "learning_rate": 9.184711320699736e-06,
      "loss": 0.0012,
      "step": 1311
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.48941531777381897,
      "learning_rate": 9.181577676441564e-06,
      "loss": 0.0015,
      "step": 1312
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.08845467120409012,
      "learning_rate": 9.178444032183392e-06,
      "loss": 0.0005,
      "step": 1313
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.1015854924917221,
      "learning_rate": 9.17531038792522e-06,
      "loss": 0.0009,
      "step": 1314
    },
    {
      "epoch": 6.51,
      "grad_norm": 4.338428974151611,
      "learning_rate": 9.172176743667051e-06,
      "loss": 0.0067,
      "step": 1315
    },
    {
      "epoch": 6.51,
      "grad_norm": 0.05533049255609512,
      "learning_rate": 9.169043099408878e-06,
      "loss": 0.0008,
      "step": 1316
    },
    {
      "epoch": 6.52,
      "grad_norm": 0.05854766070842743,
      "learning_rate": 9.165909455150707e-06,
      "loss": 0.0005,
      "step": 1317
    },
    {
      "epoch": 6.52,
      "grad_norm": 0.026727799326181412,
      "learning_rate": 9.162775810892536e-06,
      "loss": 0.0001,
      "step": 1318
    },
    {
      "epoch": 6.53,
      "grad_norm": 0.4555393159389496,
      "learning_rate": 9.159642166634365e-06,
      "loss": 0.002,
      "step": 1319
    },
    {
      "epoch": 6.53,
      "grad_norm": 0.21162594854831696,
      "learning_rate": 9.156508522376194e-06,
      "loss": 0.0002,
      "step": 1320
    },
    {
      "epoch": 6.54,
      "grad_norm": 0.3529793918132782,
      "learning_rate": 9.153374878118023e-06,
      "loss": 0.0014,
      "step": 1321
    },
    {
      "epoch": 6.54,
      "grad_norm": 1.8511605262756348,
      "learning_rate": 9.150241233859852e-06,
      "loss": 0.0113,
      "step": 1322
    },
    {
      "epoch": 6.55,
      "grad_norm": 0.4039936661720276,
      "learning_rate": 9.14710758960168e-06,
      "loss": 0.0035,
      "step": 1323
    },
    {
      "epoch": 6.55,
      "grad_norm": 0.05341874063014984,
      "learning_rate": 9.14397394534351e-06,
      "loss": 0.0003,
      "step": 1324
    },
    {
      "epoch": 6.56,
      "grad_norm": 0.26365119218826294,
      "learning_rate": 9.140840301085339e-06,
      "loss": 0.0016,
      "step": 1325
    },
    {
      "epoch": 6.56,
      "grad_norm": 0.13120470941066742,
      "learning_rate": 9.137706656827166e-06,
      "loss": 0.0004,
      "step": 1326
    },
    {
      "epoch": 6.57,
      "grad_norm": 0.051032353192567825,
      "learning_rate": 9.134573012568995e-06,
      "loss": 0.0003,
      "step": 1327
    },
    {
      "epoch": 6.57,
      "grad_norm": 0.101485975086689,
      "learning_rate": 9.131439368310826e-06,
      "loss": 0.0017,
      "step": 1328
    },
    {
      "epoch": 6.58,
      "grad_norm": 0.12698890268802643,
      "learning_rate": 9.128305724052653e-06,
      "loss": 0.0013,
      "step": 1329
    },
    {
      "epoch": 6.58,
      "grad_norm": 0.2696094214916229,
      "learning_rate": 9.125172079794482e-06,
      "loss": 0.0005,
      "step": 1330
    },
    {
      "epoch": 6.59,
      "grad_norm": 0.16069243848323822,
      "learning_rate": 9.12203843553631e-06,
      "loss": 0.0029,
      "step": 1331
    },
    {
      "epoch": 6.59,
      "grad_norm": 0.14140133559703827,
      "learning_rate": 9.11890479127814e-06,
      "loss": 0.0006,
      "step": 1332
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.08995630592107773,
      "learning_rate": 9.115771147019968e-06,
      "loss": 0.0013,
      "step": 1333
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.2836974859237671,
      "learning_rate": 9.112637502761797e-06,
      "loss": 0.0023,
      "step": 1334
    },
    {
      "epoch": 6.61,
      "grad_norm": 0.09461076557636261,
      "learning_rate": 9.109503858503626e-06,
      "loss": 0.0003,
      "step": 1335
    },
    {
      "epoch": 6.61,
      "grad_norm": 0.22359655797481537,
      "learning_rate": 9.106370214245455e-06,
      "loss": 0.0015,
      "step": 1336
    },
    {
      "epoch": 6.62,
      "grad_norm": 0.13185620307922363,
      "learning_rate": 9.103236569987284e-06,
      "loss": 0.0004,
      "step": 1337
    },
    {
      "epoch": 6.62,
      "grad_norm": 0.8337807059288025,
      "learning_rate": 9.100102925729113e-06,
      "loss": 0.0042,
      "step": 1338
    },
    {
      "epoch": 6.63,
      "grad_norm": 0.12954169511795044,
      "learning_rate": 9.09696928147094e-06,
      "loss": 0.0012,
      "step": 1339
    },
    {
      "epoch": 6.63,
      "grad_norm": 0.0349908284842968,
      "learning_rate": 9.093835637212771e-06,
      "loss": 0.0005,
      "step": 1340
    },
    {
      "epoch": 6.64,
      "grad_norm": 0.024683836847543716,
      "learning_rate": 9.0907019929546e-06,
      "loss": 0.0001,
      "step": 1341
    },
    {
      "epoch": 6.64,
      "grad_norm": 0.16871754825115204,
      "learning_rate": 9.087568348696427e-06,
      "loss": 0.0023,
      "step": 1342
    },
    {
      "epoch": 6.65,
      "grad_norm": 0.4865623712539673,
      "learning_rate": 9.084434704438256e-06,
      "loss": 0.0016,
      "step": 1343
    },
    {
      "epoch": 6.65,
      "grad_norm": 0.029191847890615463,
      "learning_rate": 9.081301060180087e-06,
      "loss": 0.0002,
      "step": 1344
    },
    {
      "epoch": 6.66,
      "grad_norm": 0.17333358526229858,
      "learning_rate": 9.078167415921916e-06,
      "loss": 0.0012,
      "step": 1345
    },
    {
      "epoch": 6.66,
      "grad_norm": 0.3575497567653656,
      "learning_rate": 9.075033771663743e-06,
      "loss": 0.0021,
      "step": 1346
    },
    {
      "epoch": 6.67,
      "grad_norm": 0.0754239484667778,
      "learning_rate": 9.071900127405572e-06,
      "loss": 0.001,
      "step": 1347
    },
    {
      "epoch": 6.67,
      "grad_norm": 0.5000989437103271,
      "learning_rate": 9.0687664831474e-06,
      "loss": 0.0019,
      "step": 1348
    },
    {
      "epoch": 6.68,
      "grad_norm": 0.2830958664417267,
      "learning_rate": 9.06563283888923e-06,
      "loss": 0.0024,
      "step": 1349
    },
    {
      "epoch": 6.68,
      "grad_norm": 0.13291119039058685,
      "learning_rate": 9.062499194631058e-06,
      "loss": 0.0007,
      "step": 1350
    },
    {
      "epoch": 6.69,
      "grad_norm": 0.08536079525947571,
      "learning_rate": 9.059365550372887e-06,
      "loss": 0.0013,
      "step": 1351
    },
    {
      "epoch": 6.69,
      "grad_norm": 0.057019829750061035,
      "learning_rate": 9.056231906114716e-06,
      "loss": 0.0014,
      "step": 1352
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.31738245487213135,
      "learning_rate": 9.053098261856545e-06,
      "loss": 0.0005,
      "step": 1353
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.028770040720701218,
      "learning_rate": 9.049964617598374e-06,
      "loss": 0.0002,
      "step": 1354
    },
    {
      "epoch": 6.71,
      "grad_norm": 3.084144353866577,
      "learning_rate": 9.046830973340203e-06,
      "loss": 0.0083,
      "step": 1355
    },
    {
      "epoch": 6.71,
      "grad_norm": 0.10060601681470871,
      "learning_rate": 9.04369732908203e-06,
      "loss": 0.0011,
      "step": 1356
    },
    {
      "epoch": 6.72,
      "grad_norm": 0.2072688788175583,
      "learning_rate": 9.040563684823861e-06,
      "loss": 0.0027,
      "step": 1357
    },
    {
      "epoch": 6.72,
      "grad_norm": 0.049768704921007156,
      "learning_rate": 9.03743004056569e-06,
      "loss": 0.0001,
      "step": 1358
    },
    {
      "epoch": 6.73,
      "grad_norm": 0.12242576479911804,
      "learning_rate": 9.034296396307517e-06,
      "loss": 0.0013,
      "step": 1359
    },
    {
      "epoch": 6.73,
      "grad_norm": 0.09496069699525833,
      "learning_rate": 9.031162752049346e-06,
      "loss": 0.0003,
      "step": 1360
    },
    {
      "epoch": 6.74,
      "grad_norm": 0.6738590598106384,
      "learning_rate": 9.028029107791177e-06,
      "loss": 0.0032,
      "step": 1361
    },
    {
      "epoch": 6.74,
      "grad_norm": 0.09428558498620987,
      "learning_rate": 9.024895463533004e-06,
      "loss": 0.0008,
      "step": 1362
    },
    {
      "epoch": 6.75,
      "grad_norm": 0.10895764827728271,
      "learning_rate": 9.021761819274833e-06,
      "loss": 0.0003,
      "step": 1363
    },
    {
      "epoch": 6.75,
      "grad_norm": 0.9728694558143616,
      "learning_rate": 9.018628175016662e-06,
      "loss": 0.0139,
      "step": 1364
    },
    {
      "epoch": 6.76,
      "grad_norm": 0.11927837878465652,
      "learning_rate": 9.01549453075849e-06,
      "loss": 0.0005,
      "step": 1365
    },
    {
      "epoch": 6.76,
      "grad_norm": 0.10513833910226822,
      "learning_rate": 9.01236088650032e-06,
      "loss": 0.0008,
      "step": 1366
    },
    {
      "epoch": 6.77,
      "grad_norm": 0.20020245015621185,
      "learning_rate": 9.009227242242149e-06,
      "loss": 0.0018,
      "step": 1367
    },
    {
      "epoch": 6.77,
      "grad_norm": 0.10981647670269012,
      "learning_rate": 9.006093597983977e-06,
      "loss": 0.0003,
      "step": 1368
    },
    {
      "epoch": 6.78,
      "grad_norm": 0.16399769484996796,
      "learning_rate": 9.002959953725806e-06,
      "loss": 0.0007,
      "step": 1369
    },
    {
      "epoch": 6.78,
      "grad_norm": 0.08855575323104858,
      "learning_rate": 8.999826309467635e-06,
      "loss": 0.0016,
      "step": 1370
    },
    {
      "epoch": 6.79,
      "grad_norm": 0.10650718957185745,
      "learning_rate": 8.996692665209464e-06,
      "loss": 0.0007,
      "step": 1371
    },
    {
      "epoch": 6.79,
      "grad_norm": 0.19293570518493652,
      "learning_rate": 8.993559020951291e-06,
      "loss": 0.0012,
      "step": 1372
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.09654835611581802,
      "learning_rate": 8.99042537669312e-06,
      "loss": 0.0013,
      "step": 1373
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.08592084795236588,
      "learning_rate": 8.987291732434951e-06,
      "loss": 0.001,
      "step": 1374
    },
    {
      "epoch": 6.81,
      "grad_norm": 0.3163571059703827,
      "learning_rate": 8.984158088176778e-06,
      "loss": 0.0007,
      "step": 1375
    },
    {
      "epoch": 6.81,
      "grad_norm": 0.09828496724367142,
      "learning_rate": 8.981024443918607e-06,
      "loss": 0.0008,
      "step": 1376
    },
    {
      "epoch": 6.82,
      "grad_norm": 0.15653057396411896,
      "learning_rate": 8.977890799660436e-06,
      "loss": 0.0013,
      "step": 1377
    },
    {
      "epoch": 6.82,
      "grad_norm": 0.11857650429010391,
      "learning_rate": 8.974757155402267e-06,
      "loss": 0.0015,
      "step": 1378
    },
    {
      "epoch": 6.83,
      "grad_norm": 0.05723285302519798,
      "learning_rate": 8.971623511144094e-06,
      "loss": 0.0004,
      "step": 1379
    },
    {
      "epoch": 6.83,
      "grad_norm": 0.05242699384689331,
      "learning_rate": 8.968489866885923e-06,
      "loss": 0.0004,
      "step": 1380
    },
    {
      "epoch": 6.84,
      "grad_norm": 0.11757563799619675,
      "learning_rate": 8.965356222627752e-06,
      "loss": 0.001,
      "step": 1381
    },
    {
      "epoch": 6.84,
      "grad_norm": 0.16820941865444183,
      "learning_rate": 8.96222257836958e-06,
      "loss": 0.0024,
      "step": 1382
    },
    {
      "epoch": 6.85,
      "grad_norm": 0.057220879942178726,
      "learning_rate": 8.95908893411141e-06,
      "loss": 0.0016,
      "step": 1383
    },
    {
      "epoch": 6.85,
      "grad_norm": 0.03631538525223732,
      "learning_rate": 8.955955289853239e-06,
      "loss": 0.0002,
      "step": 1384
    },
    {
      "epoch": 6.86,
      "grad_norm": 0.5945540070533752,
      "learning_rate": 8.952821645595066e-06,
      "loss": 0.0027,
      "step": 1385
    },
    {
      "epoch": 6.86,
      "grad_norm": 2.137747049331665,
      "learning_rate": 8.949688001336896e-06,
      "loss": 0.0015,
      "step": 1386
    },
    {
      "epoch": 6.87,
      "grad_norm": 0.06284510344266891,
      "learning_rate": 8.946554357078725e-06,
      "loss": 0.0008,
      "step": 1387
    },
    {
      "epoch": 6.87,
      "grad_norm": 0.315872460603714,
      "learning_rate": 8.943420712820554e-06,
      "loss": 0.001,
      "step": 1388
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.03504334017634392,
      "learning_rate": 8.940287068562381e-06,
      "loss": 0.0003,
      "step": 1389
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.09268735349178314,
      "learning_rate": 8.937153424304212e-06,
      "loss": 0.0015,
      "step": 1390
    },
    {
      "epoch": 6.89,
      "grad_norm": 0.028745530173182487,
      "learning_rate": 8.934019780046041e-06,
      "loss": 0.0002,
      "step": 1391
    },
    {
      "epoch": 6.89,
      "grad_norm": 0.19896036386489868,
      "learning_rate": 8.930886135787868e-06,
      "loss": 0.0018,
      "step": 1392
    },
    {
      "epoch": 6.9,
      "grad_norm": 8.520423889160156,
      "learning_rate": 8.927752491529697e-06,
      "loss": 0.0055,
      "step": 1393
    },
    {
      "epoch": 6.9,
      "grad_norm": 0.03246511146426201,
      "learning_rate": 8.924618847271528e-06,
      "loss": 0.0003,
      "step": 1394
    },
    {
      "epoch": 6.91,
      "grad_norm": 0.05635882541537285,
      "learning_rate": 8.921485203013355e-06,
      "loss": 0.0002,
      "step": 1395
    },
    {
      "epoch": 6.91,
      "grad_norm": 0.27429547905921936,
      "learning_rate": 8.918351558755184e-06,
      "loss": 0.0019,
      "step": 1396
    },
    {
      "epoch": 6.92,
      "grad_norm": 0.16052931547164917,
      "learning_rate": 8.915217914497013e-06,
      "loss": 0.0014,
      "step": 1397
    },
    {
      "epoch": 6.92,
      "grad_norm": 0.1595965027809143,
      "learning_rate": 8.912084270238842e-06,
      "loss": 0.0008,
      "step": 1398
    },
    {
      "epoch": 6.93,
      "grad_norm": 0.12766516208648682,
      "learning_rate": 8.90895062598067e-06,
      "loss": 0.0013,
      "step": 1399
    },
    {
      "epoch": 6.93,
      "grad_norm": 0.5140051245689392,
      "learning_rate": 8.9058169817225e-06,
      "loss": 0.0007,
      "step": 1400
    },
    {
      "epoch": 6.94,
      "grad_norm": 0.16029460728168488,
      "learning_rate": 8.902683337464329e-06,
      "loss": 0.0008,
      "step": 1401
    },
    {
      "epoch": 6.94,
      "grad_norm": 0.1636640727519989,
      "learning_rate": 8.899549693206156e-06,
      "loss": 0.0003,
      "step": 1402
    },
    {
      "epoch": 6.95,
      "grad_norm": 0.1541077047586441,
      "learning_rate": 8.896416048947986e-06,
      "loss": 0.0003,
      "step": 1403
    },
    {
      "epoch": 6.95,
      "grad_norm": 0.06340734660625458,
      "learning_rate": 8.893282404689815e-06,
      "loss": 0.0003,
      "step": 1404
    },
    {
      "epoch": 6.96,
      "grad_norm": 0.0907476469874382,
      "learning_rate": 8.890148760431643e-06,
      "loss": 0.0027,
      "step": 1405
    },
    {
      "epoch": 6.96,
      "grad_norm": 0.13232512772083282,
      "learning_rate": 8.887015116173471e-06,
      "loss": 0.0003,
      "step": 1406
    },
    {
      "epoch": 6.97,
      "grad_norm": 0.1368613839149475,
      "learning_rate": 8.883881471915302e-06,
      "loss": 0.0014,
      "step": 1407
    },
    {
      "epoch": 6.97,
      "grad_norm": 0.6859922409057617,
      "learning_rate": 8.88074782765713e-06,
      "loss": 0.0045,
      "step": 1408
    },
    {
      "epoch": 6.98,
      "grad_norm": 0.08766862004995346,
      "learning_rate": 8.877614183398958e-06,
      "loss": 0.0005,
      "step": 1409
    },
    {
      "epoch": 6.98,
      "grad_norm": 0.19953495264053345,
      "learning_rate": 8.874480539140787e-06,
      "loss": 0.004,
      "step": 1410
    },
    {
      "epoch": 6.99,
      "grad_norm": 0.15451595187187195,
      "learning_rate": 8.871346894882618e-06,
      "loss": 0.0032,
      "step": 1411
    },
    {
      "epoch": 6.99,
      "grad_norm": 0.1213899701833725,
      "learning_rate": 8.868213250624445e-06,
      "loss": 0.0016,
      "step": 1412
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.015314236283302307,
      "learning_rate": 8.865079606366274e-06,
      "loss": 0.0001,
      "step": 1413
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.05983460694551468,
      "learning_rate": 8.861945962108103e-06,
      "loss": 0.0002,
      "step": 1414
    },
    {
      "epoch": 7.0,
      "eval_AVGf1": 0.9226002425253532,
      "eval_DIAG.f1": 0.8737864077669902,
      "eval_DIAG.precision": 0.8898305084745762,
      "eval_DIAG.recall": 0.8583106267029973,
      "eval_MED.f1": 0.968,
      "eval_MED.precision": 0.968,
      "eval_MED.recall": 0.968,
      "eval_TREAT.f1": 0.9260143198090692,
      "eval_TREAT.precision": 0.9371980676328503,
      "eval_TREAT.recall": 0.9150943396226415,
      "eval_accuracy": 0.9473589651455264,
      "eval_f1": 0.7909604519774011,
      "eval_loss": 0.022287117317318916,
      "eval_precision": 0.7086614173228346,
      "eval_recall": 0.8948863636363636,
      "eval_runtime": 951.4215,
      "eval_samples_per_second": 0.755,
      "eval_steps_per_second": 0.095,
      "step": 1414
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.2583630383014679,
      "learning_rate": 8.858812317849932e-06,
      "loss": 0.0009,
      "step": 1415
    },
    {
      "epoch": 7.01,
      "grad_norm": 0.032550957053899765,
      "learning_rate": 8.85567867359176e-06,
      "loss": 0.0005,
      "step": 1416
    },
    {
      "epoch": 7.01,
      "grad_norm": 0.054444119334220886,
      "learning_rate": 8.85254502933359e-06,
      "loss": 0.0003,
      "step": 1417
    },
    {
      "epoch": 7.02,
      "grad_norm": 2.381608009338379,
      "learning_rate": 8.849411385075417e-06,
      "loss": 0.009,
      "step": 1418
    },
    {
      "epoch": 7.02,
      "grad_norm": 1.0916979312896729,
      "learning_rate": 8.846277740817248e-06,
      "loss": 0.0064,
      "step": 1419
    },
    {
      "epoch": 7.03,
      "grad_norm": 0.11403799802064896,
      "learning_rate": 8.843144096559076e-06,
      "loss": 0.0005,
      "step": 1420
    },
    {
      "epoch": 7.03,
      "grad_norm": 0.029737714678049088,
      "learning_rate": 8.840010452300905e-06,
      "loss": 0.0004,
      "step": 1421
    },
    {
      "epoch": 7.04,
      "grad_norm": 0.24564360082149506,
      "learning_rate": 8.836876808042733e-06,
      "loss": 0.0019,
      "step": 1422
    },
    {
      "epoch": 7.04,
      "grad_norm": 0.036392781883478165,
      "learning_rate": 8.833743163784562e-06,
      "loss": 0.0003,
      "step": 1423
    },
    {
      "epoch": 7.05,
      "grad_norm": 0.10857929289340973,
      "learning_rate": 8.830609519526392e-06,
      "loss": 0.0015,
      "step": 1424
    },
    {
      "epoch": 7.05,
      "grad_norm": 0.013592868112027645,
      "learning_rate": 8.82747587526822e-06,
      "loss": 0.0002,
      "step": 1425
    },
    {
      "epoch": 7.06,
      "grad_norm": 0.07900795340538025,
      "learning_rate": 8.824342231010048e-06,
      "loss": 0.0004,
      "step": 1426
    },
    {
      "epoch": 7.06,
      "grad_norm": 0.09350398927927017,
      "learning_rate": 8.821208586751877e-06,
      "loss": 0.0005,
      "step": 1427
    },
    {
      "epoch": 7.07,
      "grad_norm": 0.09924134612083435,
      "learning_rate": 8.818074942493706e-06,
      "loss": 0.002,
      "step": 1428
    },
    {
      "epoch": 7.07,
      "grad_norm": 0.101247139275074,
      "learning_rate": 8.814941298235535e-06,
      "loss": 0.0011,
      "step": 1429
    },
    {
      "epoch": 7.08,
      "grad_norm": 0.07916533201932907,
      "learning_rate": 8.811807653977364e-06,
      "loss": 0.001,
      "step": 1430
    },
    {
      "epoch": 7.08,
      "grad_norm": 0.0360180102288723,
      "learning_rate": 8.808674009719193e-06,
      "loss": 0.0002,
      "step": 1431
    },
    {
      "epoch": 7.09,
      "grad_norm": 0.06990933418273926,
      "learning_rate": 8.805540365461022e-06,
      "loss": 0.0003,
      "step": 1432
    },
    {
      "epoch": 7.09,
      "grad_norm": 0.10635387152433395,
      "learning_rate": 8.80240672120285e-06,
      "loss": 0.0007,
      "step": 1433
    },
    {
      "epoch": 7.1,
      "grad_norm": 0.11420906335115433,
      "learning_rate": 8.79927307694468e-06,
      "loss": 0.0004,
      "step": 1434
    },
    {
      "epoch": 7.1,
      "grad_norm": 0.3094460964202881,
      "learning_rate": 8.796139432686507e-06,
      "loss": 0.0013,
      "step": 1435
    },
    {
      "epoch": 7.11,
      "grad_norm": 0.06705201417207718,
      "learning_rate": 8.793005788428338e-06,
      "loss": 0.0002,
      "step": 1436
    },
    {
      "epoch": 7.11,
      "grad_norm": 0.21862205862998962,
      "learning_rate": 8.789872144170166e-06,
      "loss": 0.0015,
      "step": 1437
    },
    {
      "epoch": 7.12,
      "grad_norm": 0.040000639855861664,
      "learning_rate": 8.786738499911994e-06,
      "loss": 0.0002,
      "step": 1438
    },
    {
      "epoch": 7.12,
      "grad_norm": 0.09396147727966309,
      "learning_rate": 8.783604855653823e-06,
      "loss": 0.0009,
      "step": 1439
    },
    {
      "epoch": 7.13,
      "grad_norm": 0.03484032675623894,
      "learning_rate": 8.780471211395653e-06,
      "loss": 0.0002,
      "step": 1440
    },
    {
      "epoch": 7.13,
      "grad_norm": 0.049223948270082474,
      "learning_rate": 8.77733756713748e-06,
      "loss": 0.0004,
      "step": 1441
    },
    {
      "epoch": 7.14,
      "grad_norm": 0.14196506142616272,
      "learning_rate": 8.77420392287931e-06,
      "loss": 0.0013,
      "step": 1442
    },
    {
      "epoch": 7.14,
      "grad_norm": 0.01425865013152361,
      "learning_rate": 8.771070278621138e-06,
      "loss": 0.0001,
      "step": 1443
    },
    {
      "epoch": 7.15,
      "grad_norm": 0.050805121660232544,
      "learning_rate": 8.767936634362969e-06,
      "loss": 0.0004,
      "step": 1444
    },
    {
      "epoch": 7.15,
      "grad_norm": 0.08674886077642441,
      "learning_rate": 8.764802990104796e-06,
      "loss": 0.0006,
      "step": 1445
    },
    {
      "epoch": 7.16,
      "grad_norm": 0.012550003826618195,
      "learning_rate": 8.761669345846625e-06,
      "loss": 0.0001,
      "step": 1446
    },
    {
      "epoch": 7.16,
      "grad_norm": 0.12520700693130493,
      "learning_rate": 8.758535701588454e-06,
      "loss": 0.0012,
      "step": 1447
    },
    {
      "epoch": 7.17,
      "grad_norm": 0.03863045945763588,
      "learning_rate": 8.755402057330281e-06,
      "loss": 0.0005,
      "step": 1448
    },
    {
      "epoch": 7.17,
      "grad_norm": 0.1170501708984375,
      "learning_rate": 8.752268413072112e-06,
      "loss": 0.0004,
      "step": 1449
    },
    {
      "epoch": 7.18,
      "grad_norm": 0.40645456314086914,
      "learning_rate": 8.74913476881394e-06,
      "loss": 0.0008,
      "step": 1450
    },
    {
      "epoch": 7.18,
      "grad_norm": 0.04544227942824364,
      "learning_rate": 8.746001124555768e-06,
      "loss": 0.0004,
      "step": 1451
    },
    {
      "epoch": 7.19,
      "grad_norm": 0.037970904260873795,
      "learning_rate": 8.742867480297597e-06,
      "loss": 0.0003,
      "step": 1452
    },
    {
      "epoch": 7.19,
      "grad_norm": 0.04321594908833504,
      "learning_rate": 8.739733836039428e-06,
      "loss": 0.0006,
      "step": 1453
    },
    {
      "epoch": 7.2,
      "grad_norm": 0.04812179133296013,
      "learning_rate": 8.736600191781256e-06,
      "loss": 0.0005,
      "step": 1454
    },
    {
      "epoch": 7.2,
      "grad_norm": 0.03246890753507614,
      "learning_rate": 8.733466547523084e-06,
      "loss": 0.0002,
      "step": 1455
    },
    {
      "epoch": 7.21,
      "grad_norm": 0.07406389713287354,
      "learning_rate": 8.730332903264913e-06,
      "loss": 0.0003,
      "step": 1456
    },
    {
      "epoch": 7.21,
      "grad_norm": 0.24286481738090515,
      "learning_rate": 8.727199259006743e-06,
      "loss": 0.0013,
      "step": 1457
    },
    {
      "epoch": 7.22,
      "grad_norm": 0.04155365377664566,
      "learning_rate": 8.72406561474857e-06,
      "loss": 0.0001,
      "step": 1458
    },
    {
      "epoch": 7.22,
      "grad_norm": 3.967560052871704,
      "learning_rate": 8.7209319704904e-06,
      "loss": 0.0043,
      "step": 1459
    },
    {
      "epoch": 7.23,
      "grad_norm": 0.0018141287146136165,
      "learning_rate": 8.717798326232228e-06,
      "loss": 0.0,
      "step": 1460
    },
    {
      "epoch": 7.23,
      "grad_norm": 0.04049459099769592,
      "learning_rate": 8.714664681974057e-06,
      "loss": 0.0002,
      "step": 1461
    },
    {
      "epoch": 7.24,
      "grad_norm": 0.06598058342933655,
      "learning_rate": 8.711531037715886e-06,
      "loss": 0.0011,
      "step": 1462
    },
    {
      "epoch": 7.24,
      "grad_norm": 0.02757769264280796,
      "learning_rate": 8.708397393457715e-06,
      "loss": 0.0002,
      "step": 1463
    },
    {
      "epoch": 7.25,
      "grad_norm": 1.0547513961791992,
      "learning_rate": 8.705263749199544e-06,
      "loss": 0.0014,
      "step": 1464
    },
    {
      "epoch": 7.25,
      "grad_norm": 0.12713591754436493,
      "learning_rate": 8.702130104941373e-06,
      "loss": 0.0014,
      "step": 1465
    },
    {
      "epoch": 7.26,
      "grad_norm": 0.09068585932254791,
      "learning_rate": 8.698996460683202e-06,
      "loss": 0.0005,
      "step": 1466
    },
    {
      "epoch": 7.26,
      "grad_norm": 0.0511365681886673,
      "learning_rate": 8.69586281642503e-06,
      "loss": 0.0003,
      "step": 1467
    },
    {
      "epoch": 7.27,
      "grad_norm": 0.04100288078188896,
      "learning_rate": 8.692729172166858e-06,
      "loss": 0.0001,
      "step": 1468
    },
    {
      "epoch": 7.27,
      "grad_norm": 0.16439126431941986,
      "learning_rate": 8.689595527908689e-06,
      "loss": 0.0012,
      "step": 1469
    },
    {
      "epoch": 7.28,
      "grad_norm": 0.020900987088680267,
      "learning_rate": 8.686461883650518e-06,
      "loss": 0.0001,
      "step": 1470
    },
    {
      "epoch": 7.28,
      "grad_norm": 1.8823357820510864,
      "learning_rate": 8.683328239392345e-06,
      "loss": 0.0014,
      "step": 1471
    },
    {
      "epoch": 7.29,
      "grad_norm": 0.15858346223831177,
      "learning_rate": 8.680194595134174e-06,
      "loss": 0.0002,
      "step": 1472
    },
    {
      "epoch": 7.29,
      "grad_norm": 2.235989570617676,
      "learning_rate": 8.677060950876003e-06,
      "loss": 0.0035,
      "step": 1473
    },
    {
      "epoch": 7.3,
      "grad_norm": 0.05070332810282707,
      "learning_rate": 8.673927306617832e-06,
      "loss": 0.0004,
      "step": 1474
    },
    {
      "epoch": 7.3,
      "grad_norm": 0.03802021965384483,
      "learning_rate": 8.67079366235966e-06,
      "loss": 0.0001,
      "step": 1475
    },
    {
      "epoch": 7.31,
      "grad_norm": 0.06680741906166077,
      "learning_rate": 8.66766001810149e-06,
      "loss": 0.0003,
      "step": 1476
    },
    {
      "epoch": 7.31,
      "grad_norm": 0.1655208021402359,
      "learning_rate": 8.664526373843318e-06,
      "loss": 0.0007,
      "step": 1477
    },
    {
      "epoch": 7.32,
      "grad_norm": 0.07759912312030792,
      "learning_rate": 8.661392729585147e-06,
      "loss": 0.0004,
      "step": 1478
    },
    {
      "epoch": 7.32,
      "grad_norm": 0.24478192627429962,
      "learning_rate": 8.658259085326976e-06,
      "loss": 0.0008,
      "step": 1479
    },
    {
      "epoch": 7.33,
      "grad_norm": 0.026276985183358192,
      "learning_rate": 8.655125441068805e-06,
      "loss": 0.0003,
      "step": 1480
    },
    {
      "epoch": 7.33,
      "grad_norm": 0.05617845430970192,
      "learning_rate": 8.651991796810632e-06,
      "loss": 0.0001,
      "step": 1481
    },
    {
      "epoch": 7.34,
      "grad_norm": 0.2544419765472412,
      "learning_rate": 8.648858152552463e-06,
      "loss": 0.0013,
      "step": 1482
    },
    {
      "epoch": 7.34,
      "grad_norm": 0.11766302585601807,
      "learning_rate": 8.645724508294292e-06,
      "loss": 0.0013,
      "step": 1483
    },
    {
      "epoch": 7.35,
      "grad_norm": 0.14714832603931427,
      "learning_rate": 8.642590864036119e-06,
      "loss": 0.0013,
      "step": 1484
    },
    {
      "epoch": 7.35,
      "grad_norm": 0.1815035194158554,
      "learning_rate": 8.639457219777948e-06,
      "loss": 0.0014,
      "step": 1485
    },
    {
      "epoch": 7.36,
      "grad_norm": 0.12049174308776855,
      "learning_rate": 8.636323575519779e-06,
      "loss": 0.0009,
      "step": 1486
    },
    {
      "epoch": 7.36,
      "grad_norm": 0.04539098963141441,
      "learning_rate": 8.633189931261608e-06,
      "loss": 0.0004,
      "step": 1487
    },
    {
      "epoch": 7.37,
      "grad_norm": 0.04131612926721573,
      "learning_rate": 8.630056287003435e-06,
      "loss": 0.0005,
      "step": 1488
    },
    {
      "epoch": 7.37,
      "grad_norm": 0.10081498324871063,
      "learning_rate": 8.626922642745264e-06,
      "loss": 0.0002,
      "step": 1489
    },
    {
      "epoch": 7.38,
      "grad_norm": 0.8296879529953003,
      "learning_rate": 8.623788998487094e-06,
      "loss": 0.0014,
      "step": 1490
    },
    {
      "epoch": 7.38,
      "grad_norm": 0.8818601965904236,
      "learning_rate": 8.620655354228922e-06,
      "loss": 0.0009,
      "step": 1491
    },
    {
      "epoch": 7.39,
      "grad_norm": 0.2535895109176636,
      "learning_rate": 8.61752170997075e-06,
      "loss": 0.0014,
      "step": 1492
    },
    {
      "epoch": 7.39,
      "grad_norm": 0.1295892894268036,
      "learning_rate": 8.61438806571258e-06,
      "loss": 0.0009,
      "step": 1493
    },
    {
      "epoch": 7.4,
      "grad_norm": 0.1679765284061432,
      "learning_rate": 8.611254421454408e-06,
      "loss": 0.0003,
      "step": 1494
    },
    {
      "epoch": 7.4,
      "grad_norm": 0.5332298278808594,
      "learning_rate": 8.608120777196237e-06,
      "loss": 0.0019,
      "step": 1495
    },
    {
      "epoch": 7.41,
      "grad_norm": 0.10120099037885666,
      "learning_rate": 8.604987132938066e-06,
      "loss": 0.0003,
      "step": 1496
    },
    {
      "epoch": 7.41,
      "grad_norm": 0.3983917832374573,
      "learning_rate": 8.601853488679895e-06,
      "loss": 0.0015,
      "step": 1497
    },
    {
      "epoch": 7.42,
      "grad_norm": 0.3076419532299042,
      "learning_rate": 8.598719844421722e-06,
      "loss": 0.0008,
      "step": 1498
    },
    {
      "epoch": 7.42,
      "grad_norm": 0.09644710272550583,
      "learning_rate": 8.595586200163553e-06,
      "loss": 0.0015,
      "step": 1499
    },
    {
      "epoch": 7.43,
      "grad_norm": 0.2960636019706726,
      "learning_rate": 8.592452555905382e-06,
      "loss": 0.0008,
      "step": 1500
    },
    {
      "epoch": 7.43,
      "grad_norm": 0.056584540754556656,
      "learning_rate": 8.58931891164721e-06,
      "loss": 0.0004,
      "step": 1501
    },
    {
      "epoch": 7.44,
      "grad_norm": 0.04823360964655876,
      "learning_rate": 8.586185267389038e-06,
      "loss": 0.0005,
      "step": 1502
    },
    {
      "epoch": 7.44,
      "grad_norm": 0.19233112037181854,
      "learning_rate": 8.583051623130869e-06,
      "loss": 0.0016,
      "step": 1503
    },
    {
      "epoch": 7.45,
      "grad_norm": 1.1256970167160034,
      "learning_rate": 8.579917978872696e-06,
      "loss": 0.009,
      "step": 1504
    },
    {
      "epoch": 7.45,
      "grad_norm": 0.24481019377708435,
      "learning_rate": 8.576784334614525e-06,
      "loss": 0.0021,
      "step": 1505
    },
    {
      "epoch": 7.46,
      "grad_norm": 0.07226970791816711,
      "learning_rate": 8.573650690356354e-06,
      "loss": 0.0002,
      "step": 1506
    },
    {
      "epoch": 7.46,
      "grad_norm": 0.11479971557855606,
      "learning_rate": 8.570517046098183e-06,
      "loss": 0.001,
      "step": 1507
    },
    {
      "epoch": 7.47,
      "grad_norm": 2.8813960552215576,
      "learning_rate": 8.567383401840012e-06,
      "loss": 0.0072,
      "step": 1508
    },
    {
      "epoch": 7.47,
      "grad_norm": 0.6767382621765137,
      "learning_rate": 8.56424975758184e-06,
      "loss": 0.0064,
      "step": 1509
    },
    {
      "epoch": 7.48,
      "grad_norm": 0.42827898263931274,
      "learning_rate": 8.56111611332367e-06,
      "loss": 0.0016,
      "step": 1510
    },
    {
      "epoch": 7.48,
      "grad_norm": 0.19683079421520233,
      "learning_rate": 8.557982469065498e-06,
      "loss": 0.0011,
      "step": 1511
    },
    {
      "epoch": 7.49,
      "grad_norm": 0.8824372887611389,
      "learning_rate": 8.554848824807327e-06,
      "loss": 0.0036,
      "step": 1512
    },
    {
      "epoch": 7.49,
      "grad_norm": 0.22761614620685577,
      "learning_rate": 8.551715180549156e-06,
      "loss": 0.002,
      "step": 1513
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.15321722626686096,
      "learning_rate": 8.548581536290983e-06,
      "loss": 0.0008,
      "step": 1514
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.07099834084510803,
      "learning_rate": 8.545447892032814e-06,
      "loss": 0.0013,
      "step": 1515
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.04534110426902771,
      "learning_rate": 8.542314247774643e-06,
      "loss": 0.0006,
      "step": 1516
    },
    {
      "epoch": 7.51,
      "grad_norm": 0.055894553661346436,
      "learning_rate": 8.53918060351647e-06,
      "loss": 0.0004,
      "step": 1517
    },
    {
      "epoch": 7.51,
      "grad_norm": 0.06372444331645966,
      "learning_rate": 8.5360469592583e-06,
      "loss": 0.0008,
      "step": 1518
    },
    {
      "epoch": 7.52,
      "grad_norm": 0.07277344167232513,
      "learning_rate": 8.53291331500013e-06,
      "loss": 0.0005,
      "step": 1519
    },
    {
      "epoch": 7.52,
      "grad_norm": 0.03549089655280113,
      "learning_rate": 8.529779670741959e-06,
      "loss": 0.0004,
      "step": 1520
    },
    {
      "epoch": 7.53,
      "grad_norm": 0.09486376494169235,
      "learning_rate": 8.526646026483786e-06,
      "loss": 0.0008,
      "step": 1521
    },
    {
      "epoch": 7.53,
      "grad_norm": 0.17272357642650604,
      "learning_rate": 8.523512382225615e-06,
      "loss": 0.0018,
      "step": 1522
    },
    {
      "epoch": 7.54,
      "grad_norm": 0.7529717683792114,
      "learning_rate": 8.520378737967444e-06,
      "loss": 0.005,
      "step": 1523
    },
    {
      "epoch": 7.54,
      "grad_norm": 0.06549325585365295,
      "learning_rate": 8.517245093709273e-06,
      "loss": 0.0005,
      "step": 1524
    },
    {
      "epoch": 7.55,
      "grad_norm": 0.0821935161948204,
      "learning_rate": 8.514111449451102e-06,
      "loss": 0.0011,
      "step": 1525
    },
    {
      "epoch": 7.55,
      "grad_norm": 1.24702787399292,
      "learning_rate": 8.51097780519293e-06,
      "loss": 0.0022,
      "step": 1526
    },
    {
      "epoch": 7.56,
      "grad_norm": 2.245776414871216,
      "learning_rate": 8.507844160934758e-06,
      "loss": 0.0072,
      "step": 1527
    },
    {
      "epoch": 7.56,
      "grad_norm": 0.06965713948011398,
      "learning_rate": 8.504710516676588e-06,
      "loss": 0.0006,
      "step": 1528
    },
    {
      "epoch": 7.57,
      "grad_norm": 0.23954026401042938,
      "learning_rate": 8.501576872418417e-06,
      "loss": 0.0005,
      "step": 1529
    },
    {
      "epoch": 7.57,
      "grad_norm": 0.03892820328474045,
      "learning_rate": 8.498443228160246e-06,
      "loss": 0.0002,
      "step": 1530
    },
    {
      "epoch": 7.58,
      "grad_norm": 0.12306425720453262,
      "learning_rate": 8.495309583902074e-06,
      "loss": 0.0006,
      "step": 1531
    },
    {
      "epoch": 7.58,
      "grad_norm": 1.9655427932739258,
      "learning_rate": 8.492175939643904e-06,
      "loss": 0.008,
      "step": 1532
    },
    {
      "epoch": 7.59,
      "grad_norm": 0.03611445054411888,
      "learning_rate": 8.489042295385733e-06,
      "loss": 0.0002,
      "step": 1533
    },
    {
      "epoch": 7.59,
      "grad_norm": 0.31665894389152527,
      "learning_rate": 8.48590865112756e-06,
      "loss": 0.0007,
      "step": 1534
    },
    {
      "epoch": 7.6,
      "grad_norm": 0.015489790588617325,
      "learning_rate": 8.48277500686939e-06,
      "loss": 0.0001,
      "step": 1535
    },
    {
      "epoch": 7.6,
      "grad_norm": 0.18178120255470276,
      "learning_rate": 8.47964136261122e-06,
      "loss": 0.0022,
      "step": 1536
    },
    {
      "epoch": 7.61,
      "grad_norm": 0.12825702130794525,
      "learning_rate": 8.476507718353047e-06,
      "loss": 0.0006,
      "step": 1537
    },
    {
      "epoch": 7.61,
      "grad_norm": 0.19960515201091766,
      "learning_rate": 8.473374074094876e-06,
      "loss": 0.0009,
      "step": 1538
    },
    {
      "epoch": 7.62,
      "grad_norm": 0.4772973358631134,
      "learning_rate": 8.470240429836705e-06,
      "loss": 0.0028,
      "step": 1539
    },
    {
      "epoch": 7.62,
      "grad_norm": 0.10756939649581909,
      "learning_rate": 8.467106785578534e-06,
      "loss": 0.0006,
      "step": 1540
    },
    {
      "epoch": 7.63,
      "grad_norm": 0.08755452930927277,
      "learning_rate": 8.463973141320363e-06,
      "loss": 0.0008,
      "step": 1541
    },
    {
      "epoch": 7.63,
      "grad_norm": 0.0944322720170021,
      "learning_rate": 8.460839497062192e-06,
      "loss": 0.0011,
      "step": 1542
    },
    {
      "epoch": 7.64,
      "grad_norm": 0.6957164406776428,
      "learning_rate": 8.45770585280402e-06,
      "loss": 0.0181,
      "step": 1543
    },
    {
      "epoch": 7.64,
      "grad_norm": 0.009558900259435177,
      "learning_rate": 8.45457220854585e-06,
      "loss": 0.0,
      "step": 1544
    },
    {
      "epoch": 7.65,
      "grad_norm": 0.017615167424082756,
      "learning_rate": 8.451438564287678e-06,
      "loss": 0.0002,
      "step": 1545
    },
    {
      "epoch": 7.65,
      "grad_norm": 0.1276034116744995,
      "learning_rate": 8.448304920029507e-06,
      "loss": 0.0011,
      "step": 1546
    },
    {
      "epoch": 7.66,
      "grad_norm": 0.1463834047317505,
      "learning_rate": 8.445171275771335e-06,
      "loss": 0.0013,
      "step": 1547
    },
    {
      "epoch": 7.66,
      "grad_norm": 0.18083390593528748,
      "learning_rate": 8.442037631513164e-06,
      "loss": 0.0007,
      "step": 1548
    },
    {
      "epoch": 7.67,
      "grad_norm": 1.9444671869277954,
      "learning_rate": 8.438903987254994e-06,
      "loss": 0.0056,
      "step": 1549
    },
    {
      "epoch": 7.67,
      "grad_norm": 0.11981119960546494,
      "learning_rate": 8.435770342996821e-06,
      "loss": 0.0011,
      "step": 1550
    },
    {
      "epoch": 7.68,
      "grad_norm": 0.060943931341171265,
      "learning_rate": 8.43263669873865e-06,
      "loss": 0.0003,
      "step": 1551
    },
    {
      "epoch": 7.68,
      "grad_norm": 0.09760569781064987,
      "learning_rate": 8.42950305448048e-06,
      "loss": 0.001,
      "step": 1552
    },
    {
      "epoch": 7.69,
      "grad_norm": 0.25665757060050964,
      "learning_rate": 8.42636941022231e-06,
      "loss": 0.001,
      "step": 1553
    },
    {
      "epoch": 7.69,
      "grad_norm": 0.2564047574996948,
      "learning_rate": 8.423235765964137e-06,
      "loss": 0.0012,
      "step": 1554
    },
    {
      "epoch": 7.7,
      "grad_norm": 0.25705191493034363,
      "learning_rate": 8.420102121705966e-06,
      "loss": 0.0017,
      "step": 1555
    },
    {
      "epoch": 7.7,
      "grad_norm": 0.021398914977908134,
      "learning_rate": 8.416968477447795e-06,
      "loss": 0.0002,
      "step": 1556
    },
    {
      "epoch": 7.71,
      "grad_norm": 0.07756762206554413,
      "learning_rate": 8.413834833189624e-06,
      "loss": 0.0004,
      "step": 1557
    },
    {
      "epoch": 7.71,
      "grad_norm": 0.015044590458273888,
      "learning_rate": 8.410701188931453e-06,
      "loss": 0.0002,
      "step": 1558
    },
    {
      "epoch": 7.72,
      "grad_norm": 0.16661861538887024,
      "learning_rate": 8.407567544673282e-06,
      "loss": 0.0003,
      "step": 1559
    },
    {
      "epoch": 7.72,
      "grad_norm": 0.08453238755464554,
      "learning_rate": 8.404433900415109e-06,
      "loss": 0.0009,
      "step": 1560
    },
    {
      "epoch": 7.73,
      "grad_norm": 0.016007263213396072,
      "learning_rate": 8.40130025615694e-06,
      "loss": 0.0001,
      "step": 1561
    },
    {
      "epoch": 7.73,
      "grad_norm": 0.5927816033363342,
      "learning_rate": 8.398166611898768e-06,
      "loss": 0.0015,
      "step": 1562
    },
    {
      "epoch": 7.74,
      "grad_norm": 0.04829038679599762,
      "learning_rate": 8.395032967640597e-06,
      "loss": 0.0018,
      "step": 1563
    },
    {
      "epoch": 7.74,
      "grad_norm": 0.117977574467659,
      "learning_rate": 8.391899323382425e-06,
      "loss": 0.0016,
      "step": 1564
    },
    {
      "epoch": 7.75,
      "grad_norm": 0.015656594187021255,
      "learning_rate": 8.388765679124255e-06,
      "loss": 0.0001,
      "step": 1565
    },
    {
      "epoch": 7.75,
      "grad_norm": 0.0762481614947319,
      "learning_rate": 8.385632034866084e-06,
      "loss": 0.0003,
      "step": 1566
    },
    {
      "epoch": 7.76,
      "grad_norm": 0.10039056837558746,
      "learning_rate": 8.382498390607911e-06,
      "loss": 0.0005,
      "step": 1567
    },
    {
      "epoch": 7.76,
      "grad_norm": 0.15250973403453827,
      "learning_rate": 8.37936474634974e-06,
      "loss": 0.0011,
      "step": 1568
    },
    {
      "epoch": 7.77,
      "grad_norm": 0.24381236732006073,
      "learning_rate": 8.376231102091571e-06,
      "loss": 0.0021,
      "step": 1569
    },
    {
      "epoch": 7.77,
      "grad_norm": 0.2731243073940277,
      "learning_rate": 8.373097457833398e-06,
      "loss": 0.0016,
      "step": 1570
    },
    {
      "epoch": 7.78,
      "grad_norm": 0.6716006398200989,
      "learning_rate": 8.369963813575227e-06,
      "loss": 0.003,
      "step": 1571
    },
    {
      "epoch": 7.78,
      "grad_norm": 0.11152507364749908,
      "learning_rate": 8.366830169317056e-06,
      "loss": 0.0003,
      "step": 1572
    },
    {
      "epoch": 7.79,
      "grad_norm": 0.1737097203731537,
      "learning_rate": 8.363696525058885e-06,
      "loss": 0.0004,
      "step": 1573
    },
    {
      "epoch": 7.79,
      "grad_norm": 0.03141438215970993,
      "learning_rate": 8.360562880800714e-06,
      "loss": 0.001,
      "step": 1574
    },
    {
      "epoch": 7.8,
      "grad_norm": 0.10333050787448883,
      "learning_rate": 8.357429236542543e-06,
      "loss": 0.0007,
      "step": 1575
    },
    {
      "epoch": 7.8,
      "grad_norm": 0.020588140934705734,
      "learning_rate": 8.354295592284372e-06,
      "loss": 0.0003,
      "step": 1576
    },
    {
      "epoch": 7.81,
      "grad_norm": 0.10353291779756546,
      "learning_rate": 8.351161948026199e-06,
      "loss": 0.0013,
      "step": 1577
    },
    {
      "epoch": 7.81,
      "grad_norm": 0.10130604356527328,
      "learning_rate": 8.34802830376803e-06,
      "loss": 0.002,
      "step": 1578
    },
    {
      "epoch": 7.82,
      "grad_norm": 0.1906624734401703,
      "learning_rate": 8.344894659509858e-06,
      "loss": 0.0013,
      "step": 1579
    },
    {
      "epoch": 7.82,
      "grad_norm": 0.1221800222992897,
      "learning_rate": 8.341761015251686e-06,
      "loss": 0.001,
      "step": 1580
    },
    {
      "epoch": 7.83,
      "grad_norm": 0.19712169468402863,
      "learning_rate": 8.338627370993515e-06,
      "loss": 0.0005,
      "step": 1581
    },
    {
      "epoch": 7.83,
      "grad_norm": 0.06970426440238953,
      "learning_rate": 8.335493726735345e-06,
      "loss": 0.0007,
      "step": 1582
    },
    {
      "epoch": 7.84,
      "grad_norm": 0.044812049716711044,
      "learning_rate": 8.332360082477172e-06,
      "loss": 0.0001,
      "step": 1583
    },
    {
      "epoch": 7.84,
      "grad_norm": 0.20344525575637817,
      "learning_rate": 8.329226438219001e-06,
      "loss": 0.001,
      "step": 1584
    },
    {
      "epoch": 7.85,
      "grad_norm": 0.4748174548149109,
      "learning_rate": 8.32609279396083e-06,
      "loss": 0.0018,
      "step": 1585
    },
    {
      "epoch": 7.85,
      "grad_norm": 0.05396106094121933,
      "learning_rate": 8.322959149702661e-06,
      "loss": 0.001,
      "step": 1586
    },
    {
      "epoch": 7.86,
      "grad_norm": 0.19149939715862274,
      "learning_rate": 8.319825505444488e-06,
      "loss": 0.0007,
      "step": 1587
    },
    {
      "epoch": 7.86,
      "grad_norm": 0.031779736280441284,
      "learning_rate": 8.316691861186317e-06,
      "loss": 0.0002,
      "step": 1588
    },
    {
      "epoch": 7.87,
      "grad_norm": 0.14006322622299194,
      "learning_rate": 8.313558216928146e-06,
      "loss": 0.0014,
      "step": 1589
    },
    {
      "epoch": 7.87,
      "grad_norm": 0.6385043859481812,
      "learning_rate": 8.310424572669975e-06,
      "loss": 0.0015,
      "step": 1590
    },
    {
      "epoch": 7.88,
      "grad_norm": 0.13299055397510529,
      "learning_rate": 8.307290928411804e-06,
      "loss": 0.001,
      "step": 1591
    },
    {
      "epoch": 7.88,
      "grad_norm": 0.03988081216812134,
      "learning_rate": 8.304157284153633e-06,
      "loss": 0.0002,
      "step": 1592
    },
    {
      "epoch": 7.89,
      "grad_norm": 0.29338309168815613,
      "learning_rate": 8.30102363989546e-06,
      "loss": 0.0026,
      "step": 1593
    },
    {
      "epoch": 7.89,
      "grad_norm": 0.08290848135948181,
      "learning_rate": 8.29788999563729e-06,
      "loss": 0.0003,
      "step": 1594
    },
    {
      "epoch": 7.9,
      "grad_norm": 0.030147776007652283,
      "learning_rate": 8.29475635137912e-06,
      "loss": 0.0002,
      "step": 1595
    },
    {
      "epoch": 7.9,
      "grad_norm": 0.13874447345733643,
      "learning_rate": 8.291622707120949e-06,
      "loss": 0.0024,
      "step": 1596
    },
    {
      "epoch": 7.91,
      "grad_norm": 0.11522466689348221,
      "learning_rate": 8.288489062862776e-06,
      "loss": 0.0013,
      "step": 1597
    },
    {
      "epoch": 7.91,
      "grad_norm": 0.09395524114370346,
      "learning_rate": 8.285355418604605e-06,
      "loss": 0.0007,
      "step": 1598
    },
    {
      "epoch": 7.92,
      "grad_norm": 0.5775045156478882,
      "learning_rate": 8.282221774346435e-06,
      "loss": 0.0014,
      "step": 1599
    },
    {
      "epoch": 7.92,
      "grad_norm": 0.07949704676866531,
      "learning_rate": 8.279088130088263e-06,
      "loss": 0.0005,
      "step": 1600
    },
    {
      "epoch": 7.93,
      "grad_norm": 0.1521735042333603,
      "learning_rate": 8.275954485830091e-06,
      "loss": 0.001,
      "step": 1601
    },
    {
      "epoch": 7.93,
      "grad_norm": 0.184163436293602,
      "learning_rate": 8.27282084157192e-06,
      "loss": 0.0006,
      "step": 1602
    },
    {
      "epoch": 7.94,
      "grad_norm": 0.05156800150871277,
      "learning_rate": 8.26968719731375e-06,
      "loss": 0.0002,
      "step": 1603
    },
    {
      "epoch": 7.94,
      "grad_norm": 0.03423110768198967,
      "learning_rate": 8.266553553055578e-06,
      "loss": 0.0001,
      "step": 1604
    },
    {
      "epoch": 7.95,
      "grad_norm": 0.05142718181014061,
      "learning_rate": 8.263419908797407e-06,
      "loss": 0.0003,
      "step": 1605
    },
    {
      "epoch": 7.95,
      "grad_norm": 0.13468870520591736,
      "learning_rate": 8.260286264539236e-06,
      "loss": 0.0005,
      "step": 1606
    },
    {
      "epoch": 7.96,
      "grad_norm": 13.702183723449707,
      "learning_rate": 8.257152620281065e-06,
      "loss": 0.0033,
      "step": 1607
    },
    {
      "epoch": 7.96,
      "grad_norm": 0.10961530357599258,
      "learning_rate": 8.254018976022894e-06,
      "loss": 0.0015,
      "step": 1608
    },
    {
      "epoch": 7.97,
      "grad_norm": 0.10425278544425964,
      "learning_rate": 8.250885331764723e-06,
      "loss": 0.0009,
      "step": 1609
    },
    {
      "epoch": 7.97,
      "grad_norm": 0.5799276232719421,
      "learning_rate": 8.24775168750655e-06,
      "loss": 0.0028,
      "step": 1610
    },
    {
      "epoch": 7.98,
      "grad_norm": 0.40733763575553894,
      "learning_rate": 8.24461804324838e-06,
      "loss": 0.0015,
      "step": 1611
    },
    {
      "epoch": 7.98,
      "grad_norm": 0.1184542179107666,
      "learning_rate": 8.24148439899021e-06,
      "loss": 0.0008,
      "step": 1612
    },
    {
      "epoch": 7.99,
      "grad_norm": 0.05381670594215393,
      "learning_rate": 8.238350754732037e-06,
      "loss": 0.0007,
      "step": 1613
    },
    {
      "epoch": 7.99,
      "grad_norm": 0.11424726992845535,
      "learning_rate": 8.235217110473866e-06,
      "loss": 0.0013,
      "step": 1614
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.020441917702555656,
      "learning_rate": 8.232083466215696e-06,
      "loss": 0.0001,
      "step": 1615
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.10560420155525208,
      "learning_rate": 8.228949821957524e-06,
      "loss": 0.0009,
      "step": 1616
    },
    {
      "epoch": 8.0,
      "eval_AVGf1": 0.9292189709109185,
      "eval_DIAG.f1": 0.8901098901098902,
      "eval_DIAG.precision": 0.8975069252077562,
      "eval_DIAG.recall": 0.8828337874659401,
      "eval_MED.f1": 0.9718875502008033,
      "eval_MED.precision": 0.9758064516129032,
      "eval_MED.recall": 0.968,
      "eval_TREAT.f1": 0.9256594724220623,
      "eval_TREAT.precision": 0.9414634146341463,
      "eval_TREAT.recall": 0.910377358490566,
      "eval_accuracy": 0.9588573481854115,
      "eval_f1": 0.8195247270391779,
      "eval_loss": 0.023975053802132607,
      "eval_precision": 0.7479484173505275,
      "eval_recall": 0.90625,
      "eval_runtime": 959.9692,
      "eval_samples_per_second": 0.748,
      "eval_steps_per_second": 0.094,
      "step": 1616
    }
  ],
  "logging_steps": 1,
  "max_steps": 4242,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 21,
  "save_steps": 500,
  "total_flos": 6.276937408121242e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": {
    "adam_beta1": 0.92,
    "adam_beta2": 0.995,
    "learning_rate": 1.3292918943162154e-05,
    "lr_scheduler_type": "linear",
    "num_train_epochs": 21
  }
}