d1_science_all / train_results.json
neginr's picture
End of training
8431ad9 verified
{
"epoch": 4.982278481012658,
"total_flos": 2.519968592317874e+18,
"train_loss": 0.3888714426780135,
"train_runtime": 39739.1661,
"train_samples_per_second": 3.976,
"train_steps_per_second": 0.031
}