a1_math_deepmath / all_results.json
neginr's picture
End of training
1b16787 verified
{
"epoch": 5.0,
"total_flos": 2.0745947931067023e+18,
"train_loss": 0.28116225621960905,
"train_runtime": 21824.7333,
"train_samples_per_second": 7.239,
"train_steps_per_second": 0.057
}