training_state.json

{
  "step": 20,
  "best_loss": 0.2503400317382964,
  "tokens_generated": 49152000,
  "config_snapshot": {
    "total_steps": 3050,
    "warmup_steps": 150,
    "stable_steps": 2400,
    "decay_steps": 500,
    "lr_peak": 0.0025,
    "lr_min": 0.00025,
    "weight_decay": 0.1,
    "beta1": 0.9,
    "beta2": 0.95
  }
}