when i fine tuned model with lora, and when i inference it , i get different results each run.
1 Like
There are several possible causes, but if do_sample=False is not specified, the output will usually be different each time.