Qwen3-8B-SFT-envbench_qwen-green-yellow/result_model/train_results.json

{
    "epoch": 5.0,
    "num_input_tokens_seen": 2242920,
    "total_flos": 1.0185196765918003e+17,
    "train_loss": 0.19328440070152283,
    "train_runtime": 354.8577,
    "train_samples_per_second": 1.071,
    "train_steps_per_second": 0.07
}