Qwen2.5-7B_ultrafeedback_chosen/train_results.json

{
    "total_flos": 1.1343544468855849e+18,
    "train_loss": 1.2924116627842772,
    "train_runtime": 849.2265,
    "train_samples_per_second": 61.495,
    "train_steps_per_second": 0.961
}