Please add gguf‑quantized support for text encoder model
The text encoder used by this model is the standard Qwen 3 VL 8B Instruct model. The repository uses a different name, but the underlying weights and tensor shapes are identical to the original.
You can safely use any working GGUF version of the Qwen 3 VL 8B Instruct text encoder. Just make sure to also download the mmproj file, as it is provided separately in GGUF format.
The text encoder used by this model is the standard Qwen 3 VL 8B Instruct model. The repository uses a different name, but the underlying weights and tensor shapes are identical to the original.
You can safely use any working GGUF version of the Qwen 3 VL 8B Instruct text encoder. Just make sure to also download the mmproj file, as it is provided separately in GGUF format.
https://huggingface.co/jdopensource/JoyAI-Image-Edit-ComfyUI/discussions/2#6a589f4e73e92916e06378c7
The official staff for this model responded that the text encoder has been fine‑tuned.
The official staff for this model responded that the text encoder has been fine‑tuned.
i've used qwen3vl_8b_abliterated_fp8_scaled.safetensors with joyai_image_edit_plus-Q6_K.gguf model from this repo (thanks @vantagewithai so much for it) and it worked well.
so, it should work also well with GGUF versions