d1_math_all_10k / all_results.json
ryanmarten's picture
End of training
1a83fce verified
{
"epoch": 4.992,
"total_flos": 2.06491114012672e+18,
"train_loss": 0.368717604302443,
"train_runtime": 61246.8225,
"train_samples_per_second": 0.816,
"train_steps_per_second": 0.006
}