deepseek-8b-orpo-lora / train_results.json
zfz1's picture
Model save
3cbea15 verified
raw
history blame
231 Bytes
{
"epoch": 1.9936102236421727,
"total_flos": 0.0,
"train_loss": 1.3676127867820935,
"train_runtime": 5524.9246,
"train_samples": 20000,
"train_samples_per_second": 7.24,
"train_steps_per_second": 0.056
}