[Feature] Support Medusa decode (Medusa: Simple Framework for Accelerating LLM Generation with Multiple Decoding Heads)

[yimuu]

Motivation
Medusa: Simple Framework for Accelerating LLM Generation with Multiple Decoding Heads (https://github.com/FasterDecoding/Medusa) The proposed method can greatly improve the inference speed

Related resources
https://github.com/FasterDecoding/Medusa

[pujiang2018]

Yes, we are aware of the Medusa method. Are you currently using it, or do you plan to use it in the near future?

[yimuu]

Yes, we are aware of the Medusa method. Are you currently using it, or do you plan to use it in the near future?

We are going to use it in the near future. Hope to provide support