Allow selecting a different CLIP for Qwen models #939
Description
Feature Idea
Currently, SwarmUI automatically downloads and uses qwen_2.5_vl_7b_fp8_scaled.safetensors when using Qwen image generation. I would rather use a GGUF version of this model (and perhaps experiment with various finetunes); for example, the Q_6 of the LLM would save some VRAM while providing about the same (or better) quality as the FP8 version (qwen_2.5_vl_7b_fp8_scaled.safetensors).
Is it possible to add the choice of the Qwen encoder for those models that utilize it? (I think Lumina also uses Qwen LLM as encoder) under Advanced Model Addons?
Other
No response