GGUF!

#1
by BahamutRU - opened

Let's go! =D q8 we waiting!

Owner

https://huggingface.co/Kquant03/CognitiveFusion-4x7B-GGUF

I made an even better model with about the same amount of params...working on 8x 7B right now. It might be a couple days before the next model this one is huge and merging them properly might become complicated

Thx! But small and fast MoE interesting as an experience. =)

Owner

It wouldn't be small or fast with 32x1.1b it's just crazy and huge and slow...that's why I did 4x7B with really specialized agents. It outperforms Mixtral Instruct in some cases.

Sign up or log in to comment