checkpoint-3546/trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 1000,
  "global_step": 3546,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008463817181548879,
      "grad_norm": 5.875010013580322,
      "learning_rate": 0.0,
      "loss": 3.4842,
      "step": 1
    },
    {
      "epoch": 0.8463817181548878,
      "grad_norm": 3.085141897201538,
      "learning_rate": 2.3945471638984644e-05,
      "loss": 2.2583,
      "step": 1000
    },
    {
      "epoch": 0.8463817181548878,
      "eval_cosine_accuracy": 0.9730781316757202,
      "eval_loss": 1.2993038892745972,
      "eval_runtime": 22.9248,
      "eval_samples_per_second": 414.79,
      "eval_steps_per_second": 1.658,
      "step": 1000
    },
    {
      "epoch": 1.6923402454506982,
      "grad_norm": 2.947434425354004,
      "learning_rate": 1.4544030084612974e-05,
      "loss": 1.6255,
      "step": 2000
    },
    {
      "epoch": 1.6923402454506982,
      "eval_cosine_accuracy": 0.9754968881607056,
      "eval_loss": 1.2366809844970703,
      "eval_runtime": 22.438,
      "eval_samples_per_second": 423.789,
      "eval_steps_per_second": 1.694,
      "step": 2000
    },
    {
      "epoch": 2.5382987727465087,
      "grad_norm": 3.0690841674804688,
      "learning_rate": 5.142588530241304e-06,
      "loss": 1.4763,
      "step": 3000
    },
    {
      "epoch": 2.5382987727465087,
      "eval_cosine_accuracy": 0.9767588376998901,
      "eval_loss": 1.2182397842407227,
      "eval_runtime": 22.8781,
      "eval_samples_per_second": 415.638,
      "eval_steps_per_second": 1.661,
      "step": 3000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 3546,
  "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": 256,
  "trial_name": null,
  "trial_params": null
}