https://huggingface.co/mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated

#170
by StatusQuo209 - opened

3.1 8B abliterated just dropped. You will have to get the latest llama.cpp for the ROPE fixes.

URL: https://huggingface.co/mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated

Thank you for what you do man. Really.

StatusQuo209 changed discussion title from mlabonne / Meta-Llama-3.1-8B-abliterated to https://huggingface.co/mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated

I don't think there are any rope fixes yet (at least I don't see them in the changelog). I'm holding the 3.1 models back at the moment, but I can quantize this and then requant it again once the situazion has cleared up.

It'd being redone as we speak - static quants are done, it's now waiting for the imatrix to be generated. Will probably be done within the next 24 hours or even earlier.

mradermacher changed discussion status to closed

Sign up or log in to comment