New RAG feature for ComfyUI IF AI tools
Вставка
- Опубліковано 3 жов 2024
- NEW RAG system for ComfyUI IF_AI tools
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
VIDEO LINKS📄🖍️o(≧o≦)o🔥
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
IF_AI tools with RAG
github.com/if-...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
SOCIAL MEDIA LINKS!
✨ Support my (*・‿・)ノ⌒*:･ﾟ✧
character available at ko-fi.com/impa...
SD related civitai.com/us...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
A1111 forge sdnext
github.com/if-...
ComfyUI
github.com/if-...
github.com/if-...
github.com/if-...
github.com/if-...
github.com/if-...
ollama.com/dow...
ollama.com/dow...
ollama.com/dow...
My model for Ollama
ollama run impactframes/llama3_ifai_sd_prompt_mkr_q4km
ollama run impactframes/ifai_promptmkr_dolphin_phi3:latest
My model for other serving tools
huggingface.co...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
🔥NOTES
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Windows version coming soon
ComfyUi is getting deeper and deeper 😮
Yes and this is just the beginning 💪
Excellent. Thx!
Thanks I still have to put a lot of work into tweaking it but hey I got the RAG feature working
Awesome tool man! Great work!
Thank you, it's not perfect but already can do some things that might be useful for a new type of workflows
great update my friend 😉👍
Thank you 🙂 I will work on refining it. Hopefully people can make even better workflows with it.
Wow! Comfy is really becoming my favourite thing on the computer. Can we use this workflow to create descriptions from an image too?
Yes it can you can even save the description and make a Llava dataset
Would you be able to add batch for image captioning? That would be amazing for datasets, especially with the tag functionality to have the model edit the entire dataset caption files once output with consistent keywords across the dataset
Yes I am going to add it
@@impactframesthank you
Hey, how are you? Any news on this update?
@@___x__x_r___xa__x_____f______ Yes I am still working on it the way python and comfy works makes it very difficult. It kind of works but is too much for users to do so I can't release it like this.
@@impactframesno worries I appreciate your efforts. I realise it has to be manageable and functional cheers for the update
Hello there, i was wondering if this RAG system could work for captioning dataset images for training with a personal caption style (specific terms I will later use in prompting). I can see your tool allows the use of a txt file to guide inference). Bot sure how to implement batching for all the images in the dataset to be processed
Yes it can buy you need to make your own workflow to do that a powerful model or even gpt4 o mini can be awesome for this I think the new LLaMa3.1 are also multy model so this can be a banger
Hace you thought About rag for images?
hello again, i dont see a field for passing documents in the IF Chat Prompt. The extension is up to date
@@___x__x_r___xa__x_____f______ yes that's because the rag features only work on the Linux version, I made a Linux branch but you are probably using main. I am working on Windows version it should be ready in a week or two
@@impactframes that would be super