trainer_state.json

{
  "best_metric": 0.7974429202242631,
  "best_model_checkpoint": "00-allnli-p9-allnli-p9-allnli-p9-allnli-old-best/checkpoint-2600",
  "epoch": 1.0586319218241043,
  "global_step": 2600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08,
      "learning_rate": 3e-05,
      "loss": 0.129,
      "step": 200
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.75e-05,
      "loss": 0.0665,
      "step": 400
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.5e-05,
      "loss": 0.0754,
      "step": 600
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.25e-05,
      "loss": 0.0806,
      "step": 800
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.0895,
      "step": 1000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.0959,
      "step": 1200
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.5e-05,
      "loss": 0.1116,
      "step": 1400
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.25e-05,
      "loss": 0.1274,
      "step": 1600
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.999999999999999e-06,
      "loss": 0.1436,
      "step": 1800
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.5e-06,
      "loss": 0.1621,
      "step": 2000
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.9999999999999996e-06,
      "loss": 0.1891,
      "step": 2200
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.4999999999999998e-06,
      "loss": 0.2067,
      "step": 2400
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0,
      "loss": 0.1854,
      "step": 2600
    },
    {
      "epoch": 1.06,
      "step": 2600,
      "total_flos": 172275249862656.0,
      "train_loss": 0.12790855224315936,
      "train_runtime": 1337.6408,
      "train_samples_per_second": 248.796,
      "train_steps_per_second": 1.944
    }
  ],
  "max_steps": 2600,
  "num_train_epochs": 2,
  "total_flos": 172275249862656.0,
  "trial_name": null,
  "trial_params": null
}