{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 7299,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.20550760378133992,
      "grad_norm": 0.9523736834526062,
      "learning_rate": 1.8629949308124403e-05,
      "loss": 0.1346,
      "step": 500
    },
    {
      "epoch": 0.20550760378133992,
      "eval_accuracy": 0.9909558067831449,
      "eval_loss": 0.03258030116558075,
      "eval_runtime": 45.1976,
      "eval_samples_per_second": 107.638,
      "eval_steps_per_second": 13.474,
      "step": 500
    },
    {
      "epoch": 0.41101520756267984,
      "grad_norm": 0.014003735035657883,
      "learning_rate": 1.7259898616248804e-05,
      "loss": 0.047,
      "step": 1000
    },
    {
      "epoch": 0.41101520756267984,
      "eval_accuracy": 0.9872559095580679,
      "eval_loss": 0.07240545004606247,
      "eval_runtime": 44.1247,
      "eval_samples_per_second": 110.256,
      "eval_steps_per_second": 13.802,
      "step": 1000
    },
    {
      "epoch": 0.6165228113440198,
      "grad_norm": 0.00866780523210764,
      "learning_rate": 1.58898479243732e-05,
      "loss": 0.0342,
      "step": 1500
    },
    {
      "epoch": 0.6165228113440198,
      "eval_accuracy": 0.9901336073997945,
      "eval_loss": 0.055349577218294144,
      "eval_runtime": 42.3578,
      "eval_samples_per_second": 114.855,
      "eval_steps_per_second": 14.378,
      "step": 1500
    },
    {
      "epoch": 0.8220304151253597,
      "grad_norm": 0.028850242495536804,
      "learning_rate": 1.4519797232497603e-05,
      "loss": 0.0284,
      "step": 2000
    },
    {
      "epoch": 0.8220304151253597,
      "eval_accuracy": 0.9790339157245632,
      "eval_loss": 0.12142039090394974,
      "eval_runtime": 42.7356,
      "eval_samples_per_second": 113.84,
      "eval_steps_per_second": 14.25,
      "step": 2000
    },
    {
      "epoch": 1.0275380189066996,
      "grad_norm": 0.0015924024628475308,
      "learning_rate": 1.3149746540622004e-05,
      "loss": 0.0231,
      "step": 2500
    },
    {
      "epoch": 1.0275380189066996,
      "eval_accuracy": 0.9843782117163412,
      "eval_loss": 0.0819912925362587,
      "eval_runtime": 38.0093,
      "eval_samples_per_second": 127.995,
      "eval_steps_per_second": 16.022,
      "step": 2500
    },
    {
      "epoch": 1.2330456226880395,
      "grad_norm": 0.007981816306710243,
      "learning_rate": 1.1779695848746405e-05,
      "loss": 0.0223,
      "step": 3000
    },
    {
      "epoch": 1.2330456226880395,
      "eval_accuracy": 0.9856115107913669,
      "eval_loss": 0.06681085377931595,
      "eval_runtime": 45.9508,
      "eval_samples_per_second": 105.874,
      "eval_steps_per_second": 13.253,
      "step": 3000
    },
    {
      "epoch": 1.4385532264693794,
      "grad_norm": 0.0008939368417486548,
      "learning_rate": 1.0409645156870804e-05,
      "loss": 0.0102,
      "step": 3500
    },
    {
      "epoch": 1.4385532264693794,
      "eval_accuracy": 0.9716341212744091,
      "eval_loss": 0.17282135784626007,
      "eval_runtime": 44.1285,
      "eval_samples_per_second": 110.246,
      "eval_steps_per_second": 13.801,
      "step": 3500
    },
    {
      "epoch": 1.6440608302507194,
      "grad_norm": 0.0006680354126729071,
      "learning_rate": 9.039594464995205e-06,
      "loss": 0.0092,
      "step": 4000
    },
    {
      "epoch": 1.6440608302507194,
      "eval_accuracy": 0.9895169578622816,
      "eval_loss": 0.07168136537075043,
      "eval_runtime": 45.5874,
      "eval_samples_per_second": 106.718,
      "eval_steps_per_second": 13.359,
      "step": 4000
    },
    {
      "epoch": 1.8495684340320593,
      "grad_norm": 0.000525481125805527,
      "learning_rate": 7.669543773119606e-06,
      "loss": 0.0102,
      "step": 4500
    },
    {
      "epoch": 1.8495684340320593,
      "eval_accuracy": 0.9882836587872559,
      "eval_loss": 0.06586801260709763,
      "eval_runtime": 44.823,
      "eval_samples_per_second": 108.538,
      "eval_steps_per_second": 13.587,
      "step": 4500
    },
    {
      "epoch": 2.055076037813399,
      "grad_norm": 0.0005221966421231627,
      "learning_rate": 6.299493081244007e-06,
      "loss": 0.0067,
      "step": 5000
    },
    {
      "epoch": 2.055076037813399,
      "eval_accuracy": 0.9872559095580679,
      "eval_loss": 0.07754110544919968,
      "eval_runtime": 44.8882,
      "eval_samples_per_second": 108.38,
      "eval_steps_per_second": 13.567,
      "step": 5000
    },
    {
      "epoch": 2.260583641594739,
      "grad_norm": 0.00018965282652061433,
      "learning_rate": 4.929442389368407e-06,
      "loss": 0.0028,
      "step": 5500
    },
    {
      "epoch": 2.260583641594739,
      "eval_accuracy": 0.9827338129496402,
      "eval_loss": 0.10580458492040634,
      "eval_runtime": 44.8351,
      "eval_samples_per_second": 108.509,
      "eval_steps_per_second": 13.583,
      "step": 5500
    },
    {
      "epoch": 2.466091245376079,
      "grad_norm": 0.0001756457204464823,
      "learning_rate": 3.5593916974928076e-06,
      "loss": 0.0022,
      "step": 6000
    },
    {
      "epoch": 2.466091245376079,
      "eval_accuracy": 0.9905447070914697,
      "eval_loss": 0.058701325207948685,
      "eval_runtime": 45.273,
      "eval_samples_per_second": 107.459,
      "eval_steps_per_second": 13.452,
      "step": 6000
    },
    {
      "epoch": 2.671598849157419,
      "grad_norm": 0.00028377334820106626,
      "learning_rate": 2.189341005617208e-06,
      "loss": 0.0,
      "step": 6500
    },
    {
      "epoch": 2.671598849157419,
      "eval_accuracy": 0.9835560123329907,
      "eval_loss": 0.11242231726646423,
      "eval_runtime": 41.9421,
      "eval_samples_per_second": 115.993,
      "eval_steps_per_second": 14.52,
      "step": 6500
    },
    {
      "epoch": 2.877106452938759,
      "grad_norm": 0.0001175385041278787,
      "learning_rate": 8.192903137416085e-07,
      "loss": 0.0008,
      "step": 7000
    },
    {
      "epoch": 2.877106452938759,
      "eval_accuracy": 0.9864337101747174,
      "eval_loss": 0.08880383521318436,
      "eval_runtime": 39.0528,
      "eval_samples_per_second": 124.575,
      "eval_steps_per_second": 15.594,
      "step": 7000
    }
  ],
  "logging_steps": 500,
  "max_steps": 7299,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.08775763062404e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}