DeepSeek-R1-Distill-Qwen-1.5B-GSPO-Basic/train_results.json

{
    "total_flos": 0.0,
    "train_loss": 0.013345611529686698,
    "train_runtime": 16054.0136,
    "train_samples": 3053,
    "train_samples_per_second": 0.951,
    "train_steps_per_second": 0.118
}