b2_math_random / all_results.json
neginr's picture
End of training
a16287e verified
{
"epoch": 5.0,
"total_flos": 1.8600293763309896e+18,
"train_loss": 0.2672070719030222,
"train_runtime": 23060.9507,
"train_samples_per_second": 6.851,
"train_steps_per_second": 0.054
}