[Request] GGUF for the Text Encoder
Can you please provide quants for the pruned 32B text encoder? The only other quants for it aren't compatible with your fork.
Of course. I'll ping you when done
Of course. I'll ping you when done
Thank you, because i tried other quants and they were giving tensor out of shape errors with your dynamic vram clip loader.
I thought I'd rather make sure to support existing GGUFs and keep things interoperable rather than making similar, but incompatible quants, so if you update my fork of the loader nodes () then it will load https://huggingface.co/nif0/Qwen3-VL-32B-Instruct-MiniMax-H3-GGUF well (Do not forget to download the mmproj too - only one is needed)
https://github.com/molbal/ComfyUI-GGUF/releases/tag/v26-08-14 (or from here: https://registry.comfy.org/publishers/molbal/nodes/comfyui-gguf-reboot)
I thought I'd rather make sure to support existing GGUFs and keep things interoperable rather than making similar, but incompatible quants, so if you update my fork of the loader nodes () then it will load https://huggingface.co/nif0/Qwen3-VL-32B-Instruct-MiniMax-H3-GGUF well (Do not forget to download the mmproj too - only one is needed)
https://github.com/molbal/ComfyUI-GGUF/releases/tag/v26-08-14 (or from here: https://registry.comfy.org/publishers/molbal/nodes/comfyui-gguf-reboot)
Will it also work with the dynamic vram gguf clip loader?
Yes, I tested it before I uploaded the nodes