d1_science_long_paragraphs_3k / train_results.json
ryanmarten's picture
End of training
9a8e45c verified
{
"epoch": 6.850632911392405,
"total_flos": 3.6328761054776525e+17,
"train_loss": 0.4001694982871413,
"train_runtime": 15845.3324,
"train_samples_per_second": 1.396,
"train_steps_per_second": 0.014
}