checkpoint-4/trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8421052631578947,
  "eval_steps": 6,
  "global_step": 4,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": false,
  "log_history": [
    {
      "clip_ratio": 0.0,
      "completion_length": 10352.974609375,
      "epoch": 0.21052631578947367,
      "grad_norm": 0.13259537518024445,
      "kl": 0.0,
      "learning_rate": 1.6666666666666667e-05,
      "loss": -0.11016345024108887,
      "memory(GiB)": 186.69,
      "response_clip_ratio": 0.11328125,
      "reward": -0.002658387296833098,
      "reward_std": 0.06134121119976044,
      "rewards/CosineReward": -0.0026579967816360295,
      "rewards/RepetitionPenalty": -3.8975886695880035e-07,
      "step": 1,
      "train_speed(iter/s)": 0.000241
    },
    {
      "clip_ratio": 0.0,
      "epoch": 0.42105263157894735,
      "grad_norm": 0.1320001482963562,
      "kl": 0.0,
      "learning_rate": 3.3333333333333335e-05,
      "loss": -0.11016345024108887,
      "memory(GiB)": 186.69,
      "step": 2,
      "train_speed(iter/s)": 0.000466
    },
    {
      "clip_ratio": 1.3441811461234465e-05,
      "completion_length": 10439.369140625,
      "epoch": 0.631578947368421,
      "grad_norm": 0.08990391343832016,
      "kl": 9.50181856751442e-07,
      "learning_rate": 5e-05,
      "loss": -0.06604708731174469,
      "memory(GiB)": 186.69,
      "response_clip_ratio": 0.13671875,
      "reward": 0.0006296975770965219,
      "reward_std": 0.07172460854053497,
      "rewards/CosineReward": 0.0006298604130279273,
      "rewards/RepetitionPenalty": -1.6200439745261974e-07,
      "step": 3,
      "train_speed(iter/s)": 0.00035
    },
    {
      "clip_ratio": 1.70210253145342e-05,
      "epoch": 0.8421052631578947,
      "grad_norm": 0.0967094898223877,
      "kl": 1.1101365089416504e-05,
      "learning_rate": 6.666666666666667e-05,
      "loss": -0.06727766245603561,
      "memory(GiB)": 186.69,
      "step": 4,
      "train_speed(iter/s)": 0.000458
    }
  ],
  "logging_steps": 1,
  "max_steps": 60,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 2,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}