d1_math_all / train_results.json
sedrickkeh's picture
End of training
8700249 verified
{
"epoch": 5.0,
"total_flos": 5.052740113642029e+18,
"train_loss": 0.34700557460427767,
"train_runtime": 58005.5556,
"train_samples_per_second": 2.724,
"train_steps_per_second": 0.021
}