no_pipeline_math_10k / all_results.json
neginr's picture
End of training
fe60081 verified
{
"epoch": 4.992,
"total_flos": 5.358630760129495e+17,
"train_loss": 0.2751285242537657,
"train_runtime": 10768.1535,
"train_samples_per_second": 4.643,
"train_steps_per_second": 0.036
}