an ooba booga (or any llm runner and hoster thing) for discord.py
Do you want free llm on your discord account (or bot)? Well then this repo is for you!
- A PC running oobabooga text generation webui (anything simillar will work but you have to tinker with the get_assistant_response function)
- Another (or the same PC) running this Note: if you're running the py on another PC Please change the
127.0.0.1part in line 150 to your other pc's IP Address - discord.py==1.7.2 (anything newer or older won't work)
- A LLM model (could be anything you want but I recommend Dolphin LLaMA3)
- Make sure you have ooba booga or anything simillar as described above, now run it
- Go to your Web UI Panel, then session, in extensions click openai, press
Save UI defaults to settings.yamland press apply flags/extensions and restart - Go to model, download a LLM Model (if you haven't already) and load it
- Run the .py, type the Discord Authentication Token and enjoy. (you only need to type it once as it'll save to a file)
Selfbotting is against Discord Terms of Service, use the selfbot function at your own risk. Otherwise, use the bot version instead. This obviously doesn't come with a LLM by itself, you must host it on your own.
Username for each message so the LLM knows who sent that message (message context mode only)Added ✅- Actual working 2k characters seperator Workingonit 🤔
More support for llm hosters by using OpenAI module instead of sending a requestAdded ✅ (openai-module-ver branch)Working message context (Can probably be fixed by using a better model)should be added- Working replying context (as of right now it will only work if you reply to it directly, if you reply to another persons message and ping the bot then it wont work; though that is currently being implemented)