qwen2-5_code_ablate_duplications_3/train_results.json

{
    "epoch": 2.994773519163763,
    "total_flos": 520358168002560.0,
    "train_loss": 0.6214590458553707,
    "train_runtime": 8980.1589,
    "train_samples_per_second": 6.133,
    "train_steps_per_second": 0.064
}