training_state.json

{
  "step": 2008000,
  "epoch": 53,
  "vocab_size": 30574,
  "model_vocab_size": 30592,
  "config": {
    "optimizer_type": "adamw",
    "lr": 0.0001,
    "weight_decay": 0.01,
    "warmup_steps": 6000,
    "scheduler_type": "cosine",
    "scheduler_params": {
      "eta_min": 1e-07
    }
  }
}