deepseek-math-7B-finetuned/train_results.json

{
    "epoch": 1.0,
    "total_flos": 4644350000627712.0,
    "train_loss": 0.08661497140495507,
    "train_runtime": 70638.1483,
    "train_samples_per_second": 2.677,
    "train_steps_per_second": 0.084
}