duke comments

Repositories
Issues
Comments

Results 2 comments of


                                            duke

[Bug]: MTP Implementation Inconsistency Between DeepSeek Paper and vllm

this picture is for training. in inference, MTP draft worker takes same input and previous hidden state, then outputs draft token

[Bug]: MTP Implementation Inconsistency Between DeepSeek Paper and vllm

> > this picture is for training. in inference, MTP draft worker takes same input and previous hidden state, then outputs draft token > > Thanks for the reply, that...