{
  "best_metric": 1.689497470855713,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "epoch": 1.408450704225352,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014084507042253521,
      "grad_norm": 1.6681244373321533,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 1.7632,
      "step": 1
    },
    {
      "epoch": 0.014084507042253521,
      "eval_loss": 2.132694959640503,
      "eval_runtime": 5.4514,
      "eval_samples_per_second": 22.013,
      "eval_steps_per_second": 2.752,
      "step": 1
    },
    {
      "epoch": 0.028169014084507043,
      "grad_norm": 1.6067739725112915,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 1.7556,
      "step": 2
    },
    {
      "epoch": 0.04225352112676056,
      "grad_norm": 1.749565601348877,
      "learning_rate": 5.000000000000001e-07,
      "loss": 1.9514,
      "step": 3
    },
    {
      "epoch": 0.056338028169014086,
      "grad_norm": 1.608572006225586,
      "learning_rate": 6.666666666666667e-07,
      "loss": 1.9827,
      "step": 4
    },
    {
      "epoch": 0.07042253521126761,
      "grad_norm": 1.913438081741333,
      "learning_rate": 8.333333333333333e-07,
      "loss": 2.0711,
      "step": 5
    },
    {
      "epoch": 0.08450704225352113,
      "grad_norm": 1.7730381488800049,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.9187,
      "step": 6
    },
    {
      "epoch": 0.09859154929577464,
      "grad_norm": 2.3010146617889404,
      "learning_rate": 1.1666666666666668e-06,
      "loss": 1.9674,
      "step": 7
    },
    {
      "epoch": 0.11267605633802817,
      "grad_norm": 1.8479571342468262,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 1.9355,
      "step": 8
    },
    {
      "epoch": 0.1267605633802817,
      "grad_norm": 2.0581886768341064,
      "learning_rate": 1.5e-06,
      "loss": 2.0487,
      "step": 9
    },
    {
      "epoch": 0.14084507042253522,
      "grad_norm": 2.3469154834747314,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 2.2273,
      "step": 10
    },
    {
      "epoch": 0.15492957746478872,
      "grad_norm": 2.158480405807495,
      "learning_rate": 1.8333333333333333e-06,
      "loss": 2.2333,
      "step": 11
    },
    {
      "epoch": 0.16901408450704225,
      "grad_norm": 2.586392641067505,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 2.1832,
      "step": 12
    },
    {
      "epoch": 0.18309859154929578,
      "grad_norm": 2.23938250541687,
      "learning_rate": 2.166666666666667e-06,
      "loss": 2.2508,
      "step": 13
    },
    {
      "epoch": 0.19718309859154928,
      "grad_norm": 2.1828818321228027,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 2.3221,
      "step": 14
    },
    {
      "epoch": 0.2112676056338028,
      "grad_norm": 2.2943806648254395,
      "learning_rate": 2.5e-06,
      "loss": 2.3427,
      "step": 15
    },
    {
      "epoch": 0.22535211267605634,
      "grad_norm": 3.1679396629333496,
      "learning_rate": 2.666666666666667e-06,
      "loss": 2.6016,
      "step": 16
    },
    {
      "epoch": 0.23943661971830985,
      "grad_norm": 3.9362666606903076,
      "learning_rate": 2.8333333333333335e-06,
      "loss": 2.7401,
      "step": 17
    },
    {
      "epoch": 0.2535211267605634,
      "grad_norm": 1.7012690305709839,
      "learning_rate": 3e-06,
      "loss": 1.7739,
      "step": 18
    },
    {
      "epoch": 0.2676056338028169,
      "grad_norm": 1.7209924459457397,
      "learning_rate": 3.1666666666666667e-06,
      "loss": 1.7283,
      "step": 19
    },
    {
      "epoch": 0.28169014084507044,
      "grad_norm": 1.7125264406204224,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.92,
      "step": 20
    },
    {
      "epoch": 0.29577464788732394,
      "grad_norm": 1.5941029787063599,
      "learning_rate": 3.5e-06,
      "loss": 1.9873,
      "step": 21
    },
    {
      "epoch": 0.30985915492957744,
      "grad_norm": 1.7670009136199951,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 2.0733,
      "step": 22
    },
    {
      "epoch": 0.323943661971831,
      "grad_norm": 1.5573713779449463,
      "learning_rate": 3.833333333333334e-06,
      "loss": 2.0259,
      "step": 23
    },
    {
      "epoch": 0.3380281690140845,
      "grad_norm": 1.9398852586746216,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.8444,
      "step": 24
    },
    {
      "epoch": 0.352112676056338,
      "grad_norm": 2.0217692852020264,
      "learning_rate": 4.166666666666667e-06,
      "loss": 1.9226,
      "step": 25
    },
    {
      "epoch": 0.36619718309859156,
      "grad_norm": 1.7835967540740967,
      "learning_rate": 4.333333333333334e-06,
      "loss": 1.9009,
      "step": 26
    },
    {
      "epoch": 0.38028169014084506,
      "grad_norm": 1.8909916877746582,
      "learning_rate": 4.5e-06,
      "loss": 2.0627,
      "step": 27
    },
    {
      "epoch": 0.39436619718309857,
      "grad_norm": 1.7021043300628662,
      "learning_rate": 4.666666666666667e-06,
      "loss": 2.1645,
      "step": 28
    },
    {
      "epoch": 0.4084507042253521,
      "grad_norm": 2.2201666831970215,
      "learning_rate": 4.833333333333333e-06,
      "loss": 2.1412,
      "step": 29
    },
    {
      "epoch": 0.4225352112676056,
      "grad_norm": 1.8209506273269653,
      "learning_rate": 5e-06,
      "loss": 2.1778,
      "step": 30
    },
    {
      "epoch": 0.43661971830985913,
      "grad_norm": 1.9495118856430054,
      "learning_rate": 4.997482666353287e-06,
      "loss": 2.1262,
      "step": 31
    },
    {
      "epoch": 0.4507042253521127,
      "grad_norm": 2.1168172359466553,
      "learning_rate": 4.989935734988098e-06,
      "loss": 2.2203,
      "step": 32
    },
    {
      "epoch": 0.4647887323943662,
      "grad_norm": 2.568599224090576,
      "learning_rate": 4.977374404419838e-06,
      "loss": 2.3985,
      "step": 33
    },
    {
      "epoch": 0.4788732394366197,
      "grad_norm": 3.04622220993042,
      "learning_rate": 4.959823971496575e-06,
      "loss": 2.5179,
      "step": 34
    },
    {
      "epoch": 0.49295774647887325,
      "grad_norm": 1.2564030885696411,
      "learning_rate": 4.937319780454559e-06,
      "loss": 1.6984,
      "step": 35
    },
    {
      "epoch": 0.5070422535211268,
      "grad_norm": 1.4525283575057983,
      "learning_rate": 4.909907151739634e-06,
      "loss": 1.716,
      "step": 36
    },
    {
      "epoch": 0.5211267605633803,
      "grad_norm": 1.3202933073043823,
      "learning_rate": 4.8776412907378845e-06,
      "loss": 1.8016,
      "step": 37
    },
    {
      "epoch": 0.5352112676056338,
      "grad_norm": 1.3897333145141602,
      "learning_rate": 4.8405871765993435e-06,
      "loss": 1.8423,
      "step": 38
    },
    {
      "epoch": 0.5492957746478874,
      "grad_norm": 1.9675092697143555,
      "learning_rate": 4.7988194313786275e-06,
      "loss": 1.8888,
      "step": 39
    },
    {
      "epoch": 0.5633802816901409,
      "grad_norm": 1.5221786499023438,
      "learning_rate": 4.752422169756048e-06,
      "loss": 1.9842,
      "step": 40
    },
    {
      "epoch": 0.5774647887323944,
      "grad_norm": 1.920662760734558,
      "learning_rate": 4.701488829641845e-06,
      "loss": 1.7825,
      "step": 41
    },
    {
      "epoch": 0.5915492957746479,
      "grad_norm": 1.7375679016113281,
      "learning_rate": 4.646121984004666e-06,
      "loss": 1.7749,
      "step": 42
    },
    {
      "epoch": 0.6056338028169014,
      "grad_norm": 1.4709947109222412,
      "learning_rate": 4.586433134303257e-06,
      "loss": 1.7423,
      "step": 43
    },
    {
      "epoch": 0.6197183098591549,
      "grad_norm": 1.4375509023666382,
      "learning_rate": 4.522542485937369e-06,
      "loss": 1.9147,
      "step": 44
    },
    {
      "epoch": 0.6338028169014085,
      "grad_norm": 1.6048940420150757,
      "learning_rate": 4.454578706170075e-06,
      "loss": 1.9161,
      "step": 45
    },
    {
      "epoch": 0.647887323943662,
      "grad_norm": 1.590658187866211,
      "learning_rate": 4.382678665009028e-06,
      "loss": 1.9271,
      "step": 46
    },
    {
      "epoch": 0.6619718309859155,
      "grad_norm": 1.5676381587982178,
      "learning_rate": 4.3069871595684795e-06,
      "loss": 1.9127,
      "step": 47
    },
    {
      "epoch": 0.676056338028169,
      "grad_norm": 1.5853455066680908,
      "learning_rate": 4.227656622467162e-06,
      "loss": 1.899,
      "step": 48
    },
    {
      "epoch": 0.6901408450704225,
      "grad_norm": 1.6616050004959106,
      "learning_rate": 4.144846814849282e-06,
      "loss": 1.9497,
      "step": 49
    },
    {
      "epoch": 0.704225352112676,
      "grad_norm": 1.7431186437606812,
      "learning_rate": 4.058724504646834e-06,
      "loss": 2.0654,
      "step": 50
    },
    {
      "epoch": 0.704225352112676,
      "eval_loss": 1.8435356616973877,
      "eval_runtime": 5.5505,
      "eval_samples_per_second": 21.62,
      "eval_steps_per_second": 2.702,
      "step": 50
    },
    {
      "epoch": 0.7183098591549296,
      "grad_norm": 2.7816450595855713,
      "learning_rate": 3.969463130731183e-06,
      "loss": 2.3131,
      "step": 51
    },
    {
      "epoch": 0.7323943661971831,
      "grad_norm": 1.5190588235855103,
      "learning_rate": 3.8772424536302565e-06,
      "loss": 1.6635,
      "step": 52
    },
    {
      "epoch": 0.7464788732394366,
      "grad_norm": 1.3857049942016602,
      "learning_rate": 3.782248193514766e-06,
      "loss": 1.5654,
      "step": 53
    },
    {
      "epoch": 0.7605633802816901,
      "grad_norm": 1.2096006870269775,
      "learning_rate": 3.684671656182497e-06,
      "loss": 1.7222,
      "step": 54
    },
    {
      "epoch": 0.7746478873239436,
      "grad_norm": 1.2725486755371094,
      "learning_rate": 3.5847093477938955e-06,
      "loss": 1.7869,
      "step": 55
    },
    {
      "epoch": 0.7887323943661971,
      "grad_norm": 1.3458586931228638,
      "learning_rate": 3.4825625791348093e-06,
      "loss": 1.8483,
      "step": 56
    },
    {
      "epoch": 0.8028169014084507,
      "grad_norm": 1.5015977621078491,
      "learning_rate": 3.3784370602033572e-06,
      "loss": 1.5314,
      "step": 57
    },
    {
      "epoch": 0.8169014084507042,
      "grad_norm": 1.5502326488494873,
      "learning_rate": 3.272542485937369e-06,
      "loss": 1.63,
      "step": 58
    },
    {
      "epoch": 0.8309859154929577,
      "grad_norm": 1.6416538953781128,
      "learning_rate": 3.165092113916688e-06,
      "loss": 1.5956,
      "step": 59
    },
    {
      "epoch": 0.8450704225352113,
      "grad_norm": 1.6144381761550903,
      "learning_rate": 3.056302334890786e-06,
      "loss": 1.6909,
      "step": 60
    },
    {
      "epoch": 0.8591549295774648,
      "grad_norm": 1.3809669017791748,
      "learning_rate": 2.946392236996592e-06,
      "loss": 1.801,
      "step": 61
    },
    {
      "epoch": 0.8732394366197183,
      "grad_norm": 1.4957369565963745,
      "learning_rate": 2.835583164544139e-06,
      "loss": 1.8188,
      "step": 62
    },
    {
      "epoch": 0.8873239436619719,
      "grad_norm": 1.5907005071640015,
      "learning_rate": 2.724098272258584e-06,
      "loss": 1.7431,
      "step": 63
    },
    {
      "epoch": 0.9014084507042254,
      "grad_norm": 1.412441372871399,
      "learning_rate": 2.6121620758762877e-06,
      "loss": 1.8455,
      "step": 64
    },
    {
      "epoch": 0.9154929577464789,
      "grad_norm": 1.4778364896774292,
      "learning_rate": 2.5e-06,
      "loss": 1.824,
      "step": 65
    },
    {
      "epoch": 0.9295774647887324,
      "grad_norm": 1.4273475408554077,
      "learning_rate": 2.3878379241237136e-06,
      "loss": 1.8471,
      "step": 66
    },
    {
      "epoch": 0.9436619718309859,
      "grad_norm": 2.055863380432129,
      "learning_rate": 2.2759017277414165e-06,
      "loss": 2.0734,
      "step": 67
    },
    {
      "epoch": 0.9577464788732394,
      "grad_norm": 2.6091489791870117,
      "learning_rate": 2.1644168354558623e-06,
      "loss": 2.1574,
      "step": 68
    },
    {
      "epoch": 0.971830985915493,
      "grad_norm": 0.8804201483726501,
      "learning_rate": 2.053607763003409e-06,
      "loss": 1.6816,
      "step": 69
    },
    {
      "epoch": 0.9859154929577465,
      "grad_norm": 1.1886156797409058,
      "learning_rate": 1.9436976651092143e-06,
      "loss": 1.587,
      "step": 70
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3310304880142212,
      "learning_rate": 1.8349078860833125e-06,
      "loss": 1.8521,
      "step": 71
    },
    {
      "epoch": 1.0140845070422535,
      "grad_norm": 1.2212010622024536,
      "learning_rate": 1.7274575140626318e-06,
      "loss": 1.5759,
      "step": 72
    },
    {
      "epoch": 1.028169014084507,
      "grad_norm": 1.4360296726226807,
      "learning_rate": 1.6215629397966432e-06,
      "loss": 1.558,
      "step": 73
    },
    {
      "epoch": 1.0422535211267605,
      "grad_norm": 1.3856598138809204,
      "learning_rate": 1.5174374208651913e-06,
      "loss": 1.6417,
      "step": 74
    },
    {
      "epoch": 1.056338028169014,
      "grad_norm": 1.0938535928726196,
      "learning_rate": 1.415290652206105e-06,
      "loss": 1.6678,
      "step": 75
    },
    {
      "epoch": 1.0704225352112675,
      "grad_norm": 1.2646450996398926,
      "learning_rate": 1.3153283438175036e-06,
      "loss": 1.7071,
      "step": 76
    },
    {
      "epoch": 1.084507042253521,
      "grad_norm": 1.1934901475906372,
      "learning_rate": 1.217751806485235e-06,
      "loss": 1.7286,
      "step": 77
    },
    {
      "epoch": 1.0985915492957747,
      "grad_norm": 1.6086108684539795,
      "learning_rate": 1.122757546369744e-06,
      "loss": 1.5267,
      "step": 78
    },
    {
      "epoch": 1.1126760563380282,
      "grad_norm": 1.856623649597168,
      "learning_rate": 1.0305368692688175e-06,
      "loss": 1.5899,
      "step": 79
    },
    {
      "epoch": 1.1267605633802817,
      "grad_norm": 1.3756409883499146,
      "learning_rate": 9.412754953531664e-07,
      "loss": 1.5355,
      "step": 80
    },
    {
      "epoch": 1.1408450704225352,
      "grad_norm": 1.3555829524993896,
      "learning_rate": 8.551531851507186e-07,
      "loss": 1.6508,
      "step": 81
    },
    {
      "epoch": 1.1549295774647887,
      "grad_norm": 1.3804004192352295,
      "learning_rate": 7.723433775328385e-07,
      "loss": 1.6912,
      "step": 82
    },
    {
      "epoch": 1.1690140845070423,
      "grad_norm": 1.4337102174758911,
      "learning_rate": 6.930128404315214e-07,
      "loss": 1.7094,
      "step": 83
    },
    {
      "epoch": 1.1830985915492958,
      "grad_norm": 1.4616302251815796,
      "learning_rate": 6.17321334990973e-07,
      "loss": 1.7183,
      "step": 84
    },
    {
      "epoch": 1.1971830985915493,
      "grad_norm": 1.3623261451721191,
      "learning_rate": 5.454212938299256e-07,
      "loss": 1.6798,
      "step": 85
    },
    {
      "epoch": 1.2112676056338028,
      "grad_norm": 1.4235385656356812,
      "learning_rate": 4.774575140626317e-07,
      "loss": 1.7621,
      "step": 86
    },
    {
      "epoch": 1.2253521126760563,
      "grad_norm": 1.6084195375442505,
      "learning_rate": 4.1356686569674344e-07,
      "loss": 1.8549,
      "step": 87
    },
    {
      "epoch": 1.2394366197183098,
      "grad_norm": 2.6670219898223877,
      "learning_rate": 3.538780159953348e-07,
      "loss": 2.1107,
      "step": 88
    },
    {
      "epoch": 1.2535211267605635,
      "grad_norm": 1.3784013986587524,
      "learning_rate": 2.98511170358155e-07,
      "loss": 1.5806,
      "step": 89
    },
    {
      "epoch": 1.267605633802817,
      "grad_norm": 1.3833378553390503,
      "learning_rate": 2.4757783024395244e-07,
      "loss": 1.5403,
      "step": 90
    },
    {
      "epoch": 1.2816901408450705,
      "grad_norm": 1.3122845888137817,
      "learning_rate": 2.0118056862137358e-07,
      "loss": 1.5813,
      "step": 91
    },
    {
      "epoch": 1.295774647887324,
      "grad_norm": 1.1090465784072876,
      "learning_rate": 1.59412823400657e-07,
      "loss": 1.6799,
      "step": 92
    },
    {
      "epoch": 1.3098591549295775,
      "grad_norm": 1.3463518619537354,
      "learning_rate": 1.223587092621162e-07,
      "loss": 1.7081,
      "step": 93
    },
    {
      "epoch": 1.323943661971831,
      "grad_norm": 1.2838189601898193,
      "learning_rate": 9.00928482603669e-08,
      "loss": 1.7572,
      "step": 94
    },
    {
      "epoch": 1.3380281690140845,
      "grad_norm": 1.5930005311965942,
      "learning_rate": 6.268021954544095e-08,
      "loss": 1.4535,
      "step": 95
    },
    {
      "epoch": 1.352112676056338,
      "grad_norm": 1.4882570505142212,
      "learning_rate": 4.017602850342584e-08,
      "loss": 1.5066,
      "step": 96
    },
    {
      "epoch": 1.3661971830985915,
      "grad_norm": 1.4822651147842407,
      "learning_rate": 2.262559558016325e-08,
      "loss": 1.4942,
      "step": 97
    },
    {
      "epoch": 1.380281690140845,
      "grad_norm": 1.6636682748794556,
      "learning_rate": 1.006426501190233e-08,
      "loss": 1.6294,
      "step": 98
    },
    {
      "epoch": 1.3943661971830985,
      "grad_norm": 1.2824631929397583,
      "learning_rate": 2.5173336467135266e-09,
      "loss": 1.6908,
      "step": 99
    },
    {
      "epoch": 1.408450704225352,
      "grad_norm": 1.4730621576309204,
      "learning_rate": 0.0,
      "loss": 1.6844,
      "step": 100
    },
    {
      "epoch": 1.408450704225352,
      "eval_loss": 1.689497470855713,
      "eval_runtime": 5.5773,
      "eval_samples_per_second": 21.516,
      "eval_steps_per_second": 2.689,
      "step": 100
    }
  ],
  "logging_steps": 1,
  "max_steps": 100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.719288523312333e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}