plan/sft-qwen2.5-7b-instruct-graph-planning-bs128/all_results.json

{
    "epoch": 2.997242140099283,
    "num_input_tokens_seen": 69953200,
    "total_flos": 2.976146663409713e+18,
    "train_loss": 0.004280612113766934,
    "train_runtime": 8852.4475,
    "train_samples_per_second": 4.914,
    "train_steps_per_second": 0.038
}