multiple_samples_none_numina_aime / train_results.json
sedrickkeh's picture
End of training
02681e3 verified
raw
history blame
218 Bytes
{
"epoch": 2.9725490196078432,
"total_flos": 121055548211200.0,
"train_loss": 0.7961941848671625,
"train_runtime": 4090.1751,
"train_samples_per_second": 2.985,
"train_steps_per_second": 0.031
}