Skip to content

Conversation

jeejeelee
Copy link
Collaborator

    1. Baichuan-13B's hidden_size is 5120, and the merged qkv size is 15360. To enable Baichuan-13B to support LoRA, it should incorporate an additional size of 15360 into Punica.
    1. fix lora/text_baichuan.py function name error, by the way

@jeejeelee jeejeelee changed the title Kernel] Add punica dimension for Baichuan-13B [Kernel] Add punica dimension for Baichuan-13B Apr 13, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants