d1_science_long_paragraphs / all_results.json
neginr's picture
End of training
d3321b4 verified
{
"epoch": 5.0,
"total_flos": 1.9732071672467948e+18,
"train_loss": 0.3764596720576769,
"train_runtime": 22536.0676,
"train_samples_per_second": 7.011,
"train_steps_per_second": 0.055
}