xlinx
commented
3 days ago let me try explanin. before and after action mean before LLM and after LLM.
- before example -- call get prompt from outside txt file. -- https://github.com/xlinx/sd-webui-decadetw-auto-prompt-llm/tree/main/before-after-actions -- in bat batch file return some text will send to ur-prompt-LLM -- other word in this case mean, u can get the dynamic prompt from web or any outside resource.
- after example -- in this case we call a command -- curl http://localhost:11434/api/generate -d '{"keep_alive": 0}' -- that because some one has few GPU RAM(8G), he want run LLM use GPU 7G, but after LLM finished, SDWEB ui need gram too, so after LLM-action call the special command for release LLM-VRAM everytime.
above are example u can run any command u want. ex: send LLM-result using LINE IM app. https://engineering.linecorp.com/en/blog/using-line-notify-to-send-messages-to-line-from-the-command-line
LadyFlames
i dont know how the use Scripts etc actually works with this LLM Prompt if someone could explain it here´s what i suggest about the thing that could be added to this it could it be made so that lets say if someone uses regional Prompter or autochar or ADetailer or Controlnets or LoRa´s or Embedding´s or Textual inversions for instance it could be done like a Drop Down List to select the Script or embedding or LoRa or textual inversion or ControlNet itself but would it use them automatically or do they need to be manually configured in their actual sections in the webui itself etc or it could be done like if someone puts any of the things to the system prompt itself it would search civitai automatically and put the proper LoRa´s etc to the prompt itself but it might be a bit too hard to pull off and it might be also a good idea to allow the StoryBoard/History to be cleared without restarting the WebUI everyTime