qwen-7b-lora-math / train_results.json
Wuhuwill's picture
Upload LoRA weights and configs
fc4562b verified
raw
history blame contribute delete
220 Bytes
{
"epoch": 2.9955555555555557,
"total_flos": 5.872683475324109e+17,
"train_loss": 0.9816320471350691,
"train_runtime": 11250.2702,
"train_samples_per_second": 1.2,
"train_steps_per_second": 0.15
}