Hi, thank you for providing this model.I am trying to use it in vllm. It seems that the speculative decoding does not work at all. I am curious if the drafter model is fine tuned as well? Thanks!
· Sign up or log in to comment
