trainer_state.json

{
  "best_global_step": 700,
  "best_metric": 5.510611534118652,
  "best_model_checkpoint": "./qwen3moe_tinystories_sft/checkpoint-700",
  "epoch": 0.9996631862579993,
  "eval_steps": 100,
  "global_step": 742,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.13472549680026946,
      "grad_norm": 58221.64453125,
      "learning_rate": 3.3221476510067115e-05,
      "loss": 5.6966,
      "step": 100
    },
    {
      "epoch": 0.13472549680026946,
      "eval_loss": 10.694306373596191,
      "eval_runtime": 119.9061,
      "eval_samples_per_second": 41.699,
      "eval_steps_per_second": 2.61,
      "step": 100
    },
    {
      "epoch": 0.2694509936005389,
      "grad_norm": 53346.890625,
      "learning_rate": 4.5784148397976396e-05,
      "loss": 4.7677,
      "step": 200
    },
    {
      "epoch": 0.2694509936005389,
      "eval_loss": 8.538475036621094,
      "eval_runtime": 118.4911,
      "eval_samples_per_second": 42.197,
      "eval_steps_per_second": 2.642,
      "step": 200
    },
    {
      "epoch": 0.40417649040080833,
      "grad_norm": 60593.984375,
      "learning_rate": 3.735244519392918e-05,
      "loss": 3.8182,
      "step": 300
    },
    {
      "epoch": 0.40417649040080833,
      "eval_loss": 6.992630481719971,
      "eval_runtime": 122.3318,
      "eval_samples_per_second": 40.872,
      "eval_steps_per_second": 2.559,
      "step": 300
    },
    {
      "epoch": 0.5389019872010778,
      "grad_norm": 45884.55078125,
      "learning_rate": 2.8920741989881955e-05,
      "loss": 3.268,
      "step": 400
    },
    {
      "epoch": 0.5389019872010778,
      "eval_loss": 6.24953031539917,
      "eval_runtime": 122.2793,
      "eval_samples_per_second": 40.89,
      "eval_steps_per_second": 2.56,
      "step": 400
    },
    {
      "epoch": 0.6736274840013473,
      "grad_norm": 37802.12890625,
      "learning_rate": 2.048903878583474e-05,
      "loss": 2.9965,
      "step": 500
    },
    {
      "epoch": 0.6736274840013473,
      "eval_loss": 5.843188762664795,
      "eval_runtime": 121.956,
      "eval_samples_per_second": 40.998,
      "eval_steps_per_second": 2.567,
      "step": 500
    },
    {
      "epoch": 0.8083529808016167,
      "grad_norm": 34398.84375,
      "learning_rate": 1.205733558178752e-05,
      "loss": 2.8499,
      "step": 600
    },
    {
      "epoch": 0.8083529808016167,
      "eval_loss": 5.611515045166016,
      "eval_runtime": 121.7403,
      "eval_samples_per_second": 41.071,
      "eval_steps_per_second": 2.571,
      "step": 600
    },
    {
      "epoch": 0.9430784776018861,
      "grad_norm": 29294.10546875,
      "learning_rate": 3.625632377740304e-06,
      "loss": 2.7606,
      "step": 700
    },
    {
      "epoch": 0.9430784776018861,
      "eval_loss": 5.510611534118652,
      "eval_runtime": 121.6665,
      "eval_samples_per_second": 41.096,
      "eval_steps_per_second": 2.573,
      "step": 700
    }
  ],
  "logging_steps": 100,
  "max_steps": 742,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0700859494762496e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}