d1_math_gpt_3k / all_results.json
ryanmarten's picture
End of training
f160e88 verified
{
"epoch": 6.850632911392405,
"total_flos": 8.149472647147684e+17,
"train_loss": 0.3728780325369111,
"train_runtime": 25402.5047,
"train_samples_per_second": 0.871,
"train_steps_per_second": 0.009
}