{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.962025316455696,
  "eval_steps": 500,
  "global_step": 78,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0379746835443038,
      "grad_norm": 8.919463626846706,
      "learning_rate": 1.25e-06,
      "loss": 1.0163,
      "step": 1
    },
    {
      "epoch": 0.0759493670886076,
      "grad_norm": 9.60137056223418,
      "learning_rate": 2.5e-06,
      "loss": 1.022,
      "step": 2
    },
    {
      "epoch": 0.11392405063291139,
      "grad_norm": 9.018461084100906,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 1.0331,
      "step": 3
    },
    {
      "epoch": 0.1518987341772152,
      "grad_norm": 7.682785361281325,
      "learning_rate": 5e-06,
      "loss": 0.974,
      "step": 4
    },
    {
      "epoch": 0.189873417721519,
      "grad_norm": 5.058597946896709,
      "learning_rate": 6.25e-06,
      "loss": 0.9246,
      "step": 5
    },
    {
      "epoch": 0.22784810126582278,
      "grad_norm": 4.0276911689980635,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.8771,
      "step": 6
    },
    {
      "epoch": 0.26582278481012656,
      "grad_norm": 3.1073568178930335,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.7337,
      "step": 7
    },
    {
      "epoch": 0.3037974683544304,
      "grad_norm": 2.6475486861587942,
      "learning_rate": 1e-05,
      "loss": 0.7302,
      "step": 8
    },
    {
      "epoch": 0.34177215189873417,
      "grad_norm": 2.348615751641208,
      "learning_rate": 9.994965332706574e-06,
      "loss": 0.6831,
      "step": 9
    },
    {
      "epoch": 0.379746835443038,
      "grad_norm": 2.8449595570405855,
      "learning_rate": 9.979871469976197e-06,
      "loss": 0.7247,
      "step": 10
    },
    {
      "epoch": 0.4177215189873418,
      "grad_norm": 2.6270454490314115,
      "learning_rate": 9.954748808839675e-06,
      "loss": 0.6686,
      "step": 11
    },
    {
      "epoch": 0.45569620253164556,
      "grad_norm": 1.6838959248688399,
      "learning_rate": 9.91964794299315e-06,
      "loss": 0.6473,
      "step": 12
    },
    {
      "epoch": 0.4936708860759494,
      "grad_norm": 1.7629961618421153,
      "learning_rate": 9.874639560909118e-06,
      "loss": 0.6759,
      "step": 13
    },
    {
      "epoch": 0.5316455696202531,
      "grad_norm": 1.5563884198675755,
      "learning_rate": 9.819814303479268e-06,
      "loss": 0.6204,
      "step": 14
    },
    {
      "epoch": 0.569620253164557,
      "grad_norm": 1.4501226522021222,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.6304,
      "step": 15
    },
    {
      "epoch": 0.6075949367088608,
      "grad_norm": 1.413552009755346,
      "learning_rate": 9.681174353198687e-06,
      "loss": 0.6135,
      "step": 16
    },
    {
      "epoch": 0.6455696202531646,
      "grad_norm": 1.3630362406113175,
      "learning_rate": 9.597638862757255e-06,
      "loss": 0.6414,
      "step": 17
    },
    {
      "epoch": 0.6835443037974683,
      "grad_norm": 1.4535958511961327,
      "learning_rate": 9.504844339512096e-06,
      "loss": 0.6024,
      "step": 18
    },
    {
      "epoch": 0.7215189873417721,
      "grad_norm": 1.3773142640612739,
      "learning_rate": 9.40297765928369e-06,
      "loss": 0.6314,
      "step": 19
    },
    {
      "epoch": 0.759493670886076,
      "grad_norm": 1.2296088752044154,
      "learning_rate": 9.292243968009332e-06,
      "loss": 0.5745,
      "step": 20
    },
    {
      "epoch": 0.7974683544303798,
      "grad_norm": 1.2150770234431787,
      "learning_rate": 9.172866268606514e-06,
      "loss": 0.5841,
      "step": 21
    },
    {
      "epoch": 0.8354430379746836,
      "grad_norm": 1.2037317303070096,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.5992,
      "step": 22
    },
    {
      "epoch": 0.8734177215189873,
      "grad_norm": 1.1076898621884657,
      "learning_rate": 8.90915741234015e-06,
      "loss": 0.5866,
      "step": 23
    },
    {
      "epoch": 0.9113924050632911,
      "grad_norm": 1.2237515758915307,
      "learning_rate": 8.765357330018056e-06,
      "loss": 0.6294,
      "step": 24
    },
    {
      "epoch": 0.9493670886075949,
      "grad_norm": 1.070772970037307,
      "learning_rate": 8.613974319136959e-06,
      "loss": 0.5817,
      "step": 25
    },
    {
      "epoch": 0.9873417721518988,
      "grad_norm": 1.005637029521485,
      "learning_rate": 8.455313244934324e-06,
      "loss": 0.5935,
      "step": 26
    },
    {
      "epoch": 1.0253164556962024,
      "grad_norm": 1.949924086297769,
      "learning_rate": 8.289693629698564e-06,
      "loss": 0.8746,
      "step": 27
    },
    {
      "epoch": 1.0632911392405062,
      "grad_norm": 1.1271189769276084,
      "learning_rate": 8.117449009293668e-06,
      "loss": 0.5111,
      "step": 28
    },
    {
      "epoch": 1.1012658227848102,
      "grad_norm": 1.0289556715560984,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.4902,
      "step": 29
    },
    {
      "epoch": 1.139240506329114,
      "grad_norm": 0.9399304419510998,
      "learning_rate": 7.754484907260513e-06,
      "loss": 0.4532,
      "step": 30
    },
    {
      "epoch": 1.1772151898734178,
      "grad_norm": 1.0345517850809745,
      "learning_rate": 7.564496387029532e-06,
      "loss": 0.4962,
      "step": 31
    },
    {
      "epoch": 1.2151898734177216,
      "grad_norm": 1.0309938475305984,
      "learning_rate": 7.369343312364994e-06,
      "loss": 0.4964,
      "step": 32
    },
    {
      "epoch": 1.2531645569620253,
      "grad_norm": 1.1294663507088059,
      "learning_rate": 7.169418695587791e-06,
      "loss": 0.4966,
      "step": 33
    },
    {
      "epoch": 1.2911392405063291,
      "grad_norm": 1.093354343130434,
      "learning_rate": 6.965125158269619e-06,
      "loss": 0.4917,
      "step": 34
    },
    {
      "epoch": 1.3291139240506329,
      "grad_norm": 1.047840967438168,
      "learning_rate": 6.7568741204067145e-06,
      "loss": 0.4539,
      "step": 35
    },
    {
      "epoch": 1.3670886075949367,
      "grad_norm": 1.079414799158266,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.5063,
      "step": 36
    },
    {
      "epoch": 1.4050632911392404,
      "grad_norm": 1.102805791853958,
      "learning_rate": 6.330184227833376e-06,
      "loss": 0.4752,
      "step": 37
    },
    {
      "epoch": 1.4430379746835442,
      "grad_norm": 1.1567331870502953,
      "learning_rate": 6.112604669781572e-06,
      "loss": 0.4989,
      "step": 38
    },
    {
      "epoch": 1.481012658227848,
      "grad_norm": 1.0359359264373975,
      "learning_rate": 5.892784473993184e-06,
      "loss": 0.4452,
      "step": 39
    },
    {
      "epoch": 1.518987341772152,
      "grad_norm": 1.1090872515281387,
      "learning_rate": 5.671166329088278e-06,
      "loss": 0.5109,
      "step": 40
    },
    {
      "epoch": 1.5569620253164556,
      "grad_norm": 0.9826252180561131,
      "learning_rate": 5.448196544517168e-06,
      "loss": 0.4698,
      "step": 41
    },
    {
      "epoch": 1.5949367088607596,
      "grad_norm": 1.0004703109965065,
      "learning_rate": 5.224324151752575e-06,
      "loss": 0.4639,
      "step": 42
    },
    {
      "epoch": 1.6329113924050633,
      "grad_norm": 1.1270253939624921,
      "learning_rate": 5e-06,
      "loss": 0.5,
      "step": 43
    },
    {
      "epoch": 1.6708860759493671,
      "grad_norm": 1.0376132729963399,
      "learning_rate": 4.775675848247427e-06,
      "loss": 0.4597,
      "step": 44
    },
    {
      "epoch": 1.7088607594936709,
      "grad_norm": 1.017016374658824,
      "learning_rate": 4.551803455482833e-06,
      "loss": 0.4446,
      "step": 45
    },
    {
      "epoch": 1.7468354430379747,
      "grad_norm": 1.0515748720432072,
      "learning_rate": 4.3288336709117246e-06,
      "loss": 0.4769,
      "step": 46
    },
    {
      "epoch": 1.7848101265822784,
      "grad_norm": 0.9888312668680488,
      "learning_rate": 4.107215526006818e-06,
      "loss": 0.4451,
      "step": 47
    },
    {
      "epoch": 1.8227848101265822,
      "grad_norm": 1.0280345882946786,
      "learning_rate": 3.887395330218429e-06,
      "loss": 0.5149,
      "step": 48
    },
    {
      "epoch": 1.8607594936708862,
      "grad_norm": 1.0750601897232568,
      "learning_rate": 3.669815772166625e-06,
      "loss": 0.493,
      "step": 49
    },
    {
      "epoch": 1.8987341772151898,
      "grad_norm": 1.0111703724765564,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.4458,
      "step": 50
    },
    {
      "epoch": 1.9367088607594938,
      "grad_norm": 1.0498789075006694,
      "learning_rate": 3.2431258795932863e-06,
      "loss": 0.4528,
      "step": 51
    },
    {
      "epoch": 1.9746835443037973,
      "grad_norm": 0.9516004076755519,
      "learning_rate": 3.0348748417303826e-06,
      "loss": 0.4331,
      "step": 52
    },
    {
      "epoch": 2.0126582278481013,
      "grad_norm": 2.006683544107566,
      "learning_rate": 2.83058130441221e-06,
      "loss": 0.6866,
      "step": 53
    },
    {
      "epoch": 2.050632911392405,
      "grad_norm": 1.046494807487131,
      "learning_rate": 2.6306566876350072e-06,
      "loss": 0.3663,
      "step": 54
    },
    {
      "epoch": 2.088607594936709,
      "grad_norm": 0.9808436555671575,
      "learning_rate": 2.43550361297047e-06,
      "loss": 0.392,
      "step": 55
    },
    {
      "epoch": 2.1265822784810124,
      "grad_norm": 0.9540646631903068,
      "learning_rate": 2.245515092739488e-06,
      "loss": 0.3795,
      "step": 56
    },
    {
      "epoch": 2.1645569620253164,
      "grad_norm": 0.9184797146617196,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.3679,
      "step": 57
    },
    {
      "epoch": 2.2025316455696204,
      "grad_norm": 0.8996090227370815,
      "learning_rate": 1.8825509907063328e-06,
      "loss": 0.3464,
      "step": 58
    },
    {
      "epoch": 2.240506329113924,
      "grad_norm": 0.9622263940838064,
      "learning_rate": 1.7103063703014372e-06,
      "loss": 0.3853,
      "step": 59
    },
    {
      "epoch": 2.278481012658228,
      "grad_norm": 0.9779626691809532,
      "learning_rate": 1.544686755065677e-06,
      "loss": 0.3464,
      "step": 60
    },
    {
      "epoch": 2.3164556962025316,
      "grad_norm": 0.9378190673008402,
      "learning_rate": 1.3860256808630429e-06,
      "loss": 0.3371,
      "step": 61
    },
    {
      "epoch": 2.3544303797468356,
      "grad_norm": 0.8972139661296104,
      "learning_rate": 1.234642669981946e-06,
      "loss": 0.3443,
      "step": 62
    },
    {
      "epoch": 2.392405063291139,
      "grad_norm": 0.9360265191219411,
      "learning_rate": 1.0908425876598512e-06,
      "loss": 0.376,
      "step": 63
    },
    {
      "epoch": 2.430379746835443,
      "grad_norm": 0.9947565995212894,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.3583,
      "step": 64
    },
    {
      "epoch": 2.4683544303797467,
      "grad_norm": 1.0027810733182683,
      "learning_rate": 8.271337313934869e-07,
      "loss": 0.3662,
      "step": 65
    },
    {
      "epoch": 2.5063291139240507,
      "grad_norm": 0.9888266043757836,
      "learning_rate": 7.077560319906696e-07,
      "loss": 0.3847,
      "step": 66
    },
    {
      "epoch": 2.5443037974683547,
      "grad_norm": 0.9648071863163608,
      "learning_rate": 5.9702234071631e-07,
      "loss": 0.3656,
      "step": 67
    },
    {
      "epoch": 2.5822784810126582,
      "grad_norm": 0.9133407869676421,
      "learning_rate": 4.951556604879049e-07,
      "loss": 0.325,
      "step": 68
    },
    {
      "epoch": 2.620253164556962,
      "grad_norm": 0.9550194021350166,
      "learning_rate": 4.0236113724274716e-07,
      "loss": 0.3666,
      "step": 69
    },
    {
      "epoch": 2.6582278481012658,
      "grad_norm": 0.9733249521797224,
      "learning_rate": 3.18825646801314e-07,
      "loss": 0.3282,
      "step": 70
    },
    {
      "epoch": 2.6962025316455698,
      "grad_norm": 0.9563717338588541,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.391,
      "step": 71
    },
    {
      "epoch": 2.7341772151898733,
      "grad_norm": 0.8720648332000553,
      "learning_rate": 1.801856965207338e-07,
      "loss": 0.3197,
      "step": 72
    },
    {
      "epoch": 2.7721518987341773,
      "grad_norm": 0.9592437361357863,
      "learning_rate": 1.253604390908819e-07,
      "loss": 0.3426,
      "step": 73
    },
    {
      "epoch": 2.810126582278481,
      "grad_norm": 0.9383478522787649,
      "learning_rate": 8.035205700685167e-08,
      "loss": 0.3334,
      "step": 74
    },
    {
      "epoch": 2.848101265822785,
      "grad_norm": 0.9203775930113144,
      "learning_rate": 4.52511911603265e-08,
      "loss": 0.3562,
      "step": 75
    },
    {
      "epoch": 2.8860759493670884,
      "grad_norm": 0.907311538257859,
      "learning_rate": 2.012853002380466e-08,
      "loss": 0.334,
      "step": 76
    },
    {
      "epoch": 2.9240506329113924,
      "grad_norm": 1.0275019574304114,
      "learning_rate": 5.034667293427053e-09,
      "loss": 0.3905,
      "step": 77
    },
    {
      "epoch": 2.962025316455696,
      "grad_norm": 0.9131149619494413,
      "learning_rate": 0.0,
      "loss": 0.3404,
      "step": 78
    },
    {
      "epoch": 2.962025316455696,
      "step": 78,
      "total_flos": 17187987922944.0,
      "train_loss": 0.5260231238909256,
      "train_runtime": 493.6478,
      "train_samples_per_second": 15.193,
      "train_steps_per_second": 0.158
    }
  ],
  "logging_steps": 1,
  "max_steps": 78,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 17187987922944.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}