Created a ComfyUI node for local prompt enhancing and inferencing based on your cloud solution

#37
by Nojah - opened

Hi, great work all you fantastic devs at THUDM.

I got inspired by your prompt enhancer in your space and found an cloud solution in convert_demo.py.
Therefore I build a custom prompt enhancer node which is using local instances of current GLM-4 model of choise.
Since glm-4v-9b is a vision model, I added the possibility combine image and text to be enhanced.

Also added node for inference with custom system prompts.

Beyond that I added the option to unload model for those with limited VRAM.

Feel free to test it and let me know if I can improve it.

Link to repo: Nojahhh/ComfyUI_GLM4_Wrapper

Nojah changed discussion title from Created a ComfyUI node for prompt enhancing and inferencing based on your cloud solution to Created a ComfyUI node for local prompt enhancing and inferencing based on your cloud solution

Sign up or log in to comment