no_pipeline_math_3k / train_results.json
neginr's picture
End of training
35f2819 verified
{
"epoch": 7.0,
"total_flos": 2.2511058919253606e+17,
"train_loss": 0.25922639719593576,
"train_runtime": 5246.8102,
"train_samples_per_second": 4.216,
"train_steps_per_second": 0.044
}