ScaleDiff-7B / all_results.json
QizhiPei's picture
Upload initial model
bee8598 verified
raw
history blame
209 Bytes
{
"epoch": 3.0,
"total_flos": 2.437260619231934e+21,
"train_loss": 0.21978757351587389,
"train_runtime": 735453.379,
"train_samples_per_second": 2.384,
"train_steps_per_second": 0.074
}