152334H

/

miqu-1-70b-sf

like 218

I read your comment on the exl2 issue about potential problems with permutations on the attention tensors. Would you be able to share what changes were made to the GGUF -> transformers script? In case there is a need to ever convert GGUF to transformers again in the future lol.

EDIT: Lmao alpin just asked the same thing.