SFT-Compo-Distill-Qwen-1.5B / train_results.json
jyzhang1208's picture
Upload folder using huggingface_hub
7adfc15 verified
{
"epoch": 5.0,
"total_flos": 1.2037958677613773e+18,
"train_loss": 0.17663632503225785,
"train_runtime": 5647.7775,
"train_samples_per_second": 3.416,
"train_steps_per_second": 0.214
}