Do you plan to create GGUF versions of different model sizes for the NSFW AIO based on Qwen 2511?
#28
by daijava - opened
We would appreciate it if you could create GGUF models from Q2 to Q8.
Try this
https://huggingface.co/Arunk25/Qwen-Image-Edit-Rapid-AIO-GGUF/tree/main
We would appreciate it if you could create GGUF models from Q2 to Q8.
Q2 and Q3 are completely unreliable for image models.
The bare minimum is Q4 (and even that one is doing weird, because the latest qwen image edit models are thinking models, and thinking models usually require Q6 or higher -> optionally you can also opt for mxfp4, which has q8 for the attention heads (the thinking part))