What's the difference between this and https://huggingface.co/bartowski/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview-GGUF

#1
by ksze - opened

Could you enlighten me as to why there are two versions?

Both are derived from the original https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview.
This one using a newer version of llama.cpp (b4546) vs the other one using an older version (b4514). So what are the practical implications?

https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview/discussions/1#67947fb94e725a560dbe7594

They updated the weights in place, in order to preserve the original I uploaded this one as v0.1

I could probably be a bit more clear on the model card!

@bartowski So it means the one without the v0.1 in the name has updated weights, which fixes the problem where it "struggles with long-chain reasoning and tends to provide immediate answers directly", correct?

no sorry, v0.1 is the updated one, I meant to update the model card 🤦 i'll do it right now haha

Sign up or log in to comment