unsloth

/

Qwen3-235B-A22B-GGUF

like 54

Follow

Unsloth AI 6.3k

Maybe with UD 🙏

I also waiting UD version IQ2_XXS for 235B-A22B

Hey guys thanks for all the hype! We'll need to work with llama.cpp to get down to the issues with quantization of larger Qwen models

Sadly I don't think low bit I quants are possible - I tried and continuously tried debugging, but imatrix has nans and zeros - I'm going to have to work with the llama.cpp and Hugging Face folks and Qwen folks to get to the bottom of things!

For now, the quants we provide are the only "stable" ones! Apologies!