prm_math_only_hf / train_results.json
DongfuJiang's picture
Duplicate from DongfuJiang/prm_version3_full_hf
600f251 verified
raw
history blame contribute delete
220 Bytes
{
"epoch": 0.9996064541519087,
"total_flos": 223330201436160.0,
"train_loss": 0.14531472616308316,
"train_runtime": 7100.3282,
"train_samples_per_second": 11.452,
"train_steps_per_second": 0.179
}