{
  "epoch": 2,
  "base_model": "Qwen/Qwen2.5-14B-Instruct",
  "metrics": {
    "sft/loss_mean": 0.5524452924728394,
    "sft/loss_final": 0.5536863803863525,
    "sft/loss_std": 0.052169021074377195,
    "sft/grad_norm_mean": 1.3986235870004038,
    "sft/grad_norm_max": 2.046875,
    "sft/tokens_trained": 3829137,
    "sft/samples_trained": 79264,
    "sft/optimizer_steps": 2477,
    "sft/loss_per_token": 0.0003566250023621232,
    "global_step": 4954,
    "epoch": 2
  }
}