b1_math_top_1_10k / train_results.json
ryanmarten's picture
End of training
bb73bf0 verified
{
"epoch": 4.992,
"total_flos": 7.594055696090399e+17,
"train_loss": 0.2745184522790787,
"train_runtime": 30427.5029,
"train_samples_per_second": 1.643,
"train_steps_per_second": 0.013
}