c1_math_10d_4s / all_results.json
neginr's picture
End of training
d660ef3 verified
{
"epoch": 4.982278481012658,
"total_flos": 5.900557800872346e+18,
"train_loss": 0.36615794705666177,
"train_runtime": 54580.8511,
"train_samples_per_second": 2.895,
"train_steps_per_second": 0.023
}