TT_L0.2_H0.2_grpo/train_results.json

{
    "total_flos": 0.0,
    "train_loss": 3.2957177609205244e-09,
    "train_runtime": 10011.2078,
    "train_samples": 7000,
    "train_samples_per_second": 1.279,
    "train_steps_per_second": 0.02
}