R2EGym-Qwen3-8B-Agent-Coder-Instruct1-merged_bucketab_4sources_11/train_results.json

{
    "epoch": 6.0,
    "total_flos": 665846364340224.0,
    "train_loss": 0.11275661619404306,
    "train_runtime": 28497.6157,
    "train_samples_per_second": 0.777,
    "train_steps_per_second": 0.097
}