{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 15000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "grad_norm": 22.972631454467773,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.2552,
      "step": 500
    },
    {
      "epoch": 0.2,
      "grad_norm": 46.487945556640625,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.4654,
      "step": 1000
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.95026683807373,
      "learning_rate": 5e-05,
      "loss": 0.4089,
      "step": 1500
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.9401936531066895,
      "learning_rate": 4.983095894354858e-05,
      "loss": 0.367,
      "step": 2000
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.8206381797790527,
      "learning_rate": 4.9326121764495596e-05,
      "loss": 0.3275,
      "step": 2500
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.384533882141113,
      "learning_rate": 4.849231551964771e-05,
      "loss": 0.3179,
      "step": 3000
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.695955276489258,
      "learning_rate": 4.734081600808531e-05,
      "loss": 0.2881,
      "step": 3500
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.10525286942720413,
      "learning_rate": 4.588719528532342e-05,
      "loss": 0.2975,
      "step": 4000
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.812150001525879,
      "learning_rate": 4.415111107797445e-05,
      "loss": 0.2413,
      "step": 4500
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.31458476185798645,
      "learning_rate": 4.215604094671835e-05,
      "loss": 0.2518,
      "step": 5000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.935,
      "eval_loss": 0.29133525490760803,
      "eval_runtime": 226.6205,
      "eval_samples_per_second": 17.651,
      "eval_steps_per_second": 2.206,
      "step": 5000
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.748766899108887,
      "learning_rate": 3.9928964792569655e-05,
      "loss": 0.2025,
      "step": 5500
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6717832684516907,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.1751,
      "step": 6000
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.013602585531771183,
      "learning_rate": 3.490199415097892e-05,
      "loss": 0.1917,
      "step": 6500
    },
    {
      "epoch": 1.4,
      "grad_norm": 38.89192199707031,
      "learning_rate": 3.217008081777726e-05,
      "loss": 0.1644,
      "step": 7000
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1758809089660645,
      "learning_rate": 2.9341204441673266e-05,
      "loss": 0.1846,
      "step": 7500
    },
    {
      "epoch": 1.6,
      "grad_norm": 9.477453204453923e-06,
      "learning_rate": 2.6453620722761896e-05,
      "loss": 0.1546,
      "step": 8000
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.899080753326416,
      "learning_rate": 2.3546379277238107e-05,
      "loss": 0.1478,
      "step": 8500
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.797894689545501e-06,
      "learning_rate": 2.0658795558326743e-05,
      "loss": 0.1483,
      "step": 9000
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.011263616383075714,
      "learning_rate": 1.7829919182222752e-05,
      "loss": 0.165,
      "step": 9500
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.0016277525573968887,
      "learning_rate": 1.509800584902108e-05,
      "loss": 0.11,
      "step": 10000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.95825,
      "eval_loss": 0.1764126569032669,
      "eval_runtime": 226.3715,
      "eval_samples_per_second": 17.67,
      "eval_steps_per_second": 2.209,
      "step": 10000
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.00017825818213168532,
      "learning_rate": 1.2500000000000006e-05,
      "loss": 0.0418,
      "step": 10500
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.011653249152004719,
      "learning_rate": 1.0071035207430352e-05,
      "loss": 0.0573,
      "step": 11000
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.5024826300068526e-06,
      "learning_rate": 7.843959053281663e-06,
      "loss": 0.038,
      "step": 11500
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.023587645962834358,
      "learning_rate": 5.848888922025553e-06,
      "loss": 0.0434,
      "step": 12000
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.05811120569705963,
      "learning_rate": 4.112804714676594e-06,
      "loss": 0.0395,
      "step": 12500
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.06877656280994415,
      "learning_rate": 2.659183991914696e-06,
      "loss": 0.0387,
      "step": 13000
    },
    {
      "epoch": 2.7,
      "grad_norm": 9.670445251686033e-06,
      "learning_rate": 1.5076844803522922e-06,
      "loss": 0.037,
      "step": 13500
    },
    {
      "epoch": 2.8,
      "grad_norm": 61.91555404663086,
      "learning_rate": 6.738782355044049e-07,
      "loss": 0.0528,
      "step": 14000
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9232193153584376e-05,
      "learning_rate": 1.6904105645142444e-07,
      "loss": 0.0432,
      "step": 14500
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5172378908800965e-09,
      "learning_rate": 0.0,
      "loss": 0.0381,
      "step": 15000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.959,
      "eval_loss": 0.24073761701583862,
      "eval_runtime": 226.1826,
      "eval_samples_per_second": 17.685,
      "eval_steps_per_second": 2.211,
      "step": 15000
    }
  ],
  "logging_steps": 500,
  "max_steps": 15000,
  "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": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}