no_pipeline_math_0.3k / train_results.json
neginr's picture
End of training
748d467 verified
{
"epoch": 13.0,
"total_flos": 2.50173978574848e+16,
"train_loss": 0.20831933366134764,
"train_runtime": 2041.6117,
"train_samples_per_second": 2.012,
"train_steps_per_second": 0.064
}