qwen32B_models/qwen_32B_d1_iter2_model/train_results.json

{
    "epoch": 1.974025974025974,
    "total_flos": 34479036694528.0,
    "train_loss": 0.6159840069319072,
    "train_runtime": 412.9286,
    "train_samples_per_second": 2.95,
    "train_steps_per_second": 0.184
}