d1_math_gpt_10k / train_results.json
ryanmarten's picture
End of training
f321403 verified
raw
history blame contribute delete
212 Bytes
{
"epoch": 4.992,
"total_flos": 1.9006487815992115e+18,
"train_loss": 0.34742337488211117,
"train_runtime": 58720.8205,
"train_samples_per_second": 0.851,
"train_steps_per_second": 0.007
}