OpenR1-Qwen-3B-SFT-Instruct/all_results.json

{
    "total_flos": 2439615715737600.0,
    "train_loss": 0.4280901599204403,
    "train_runtime": 26178.4362,
    "train_samples": 93733,
    "train_samples_per_second": 1.968,
    "train_steps_per_second": 0.123
}