e1_math_all_qwq / train_results.json
neginr's picture
End of training
068bfb2 verified
{
"epoch": 5.0,
"total_flos": 1.6351601386649027e+18,
"train_loss": 0.5656346910636917,
"train_runtime": 7027.9189,
"train_samples_per_second": 22.482,
"train_steps_per_second": 0.176
}