d1_math_all_large / train_results.json
neginr's picture
End of training
4a910c0 verified
{
"epoch": 4.986425339366516,
"total_flos": 1.3210208775492862e+19,
"train_loss": 0.37584147680889474,
"train_runtime": 27424.4359,
"train_samples_per_second": 10.307,
"train_steps_per_second": 0.02
}