I mentioned earlier that I was working on implementing permanent memory using embeddings, and that work is done and ready for alpha!
Some new cool stuff (you can see the main list of features on the README or in my original post):
- Permanent memory using embeddings, you can chat more accurately, for longer, and preserve tokens better with this
- AI-based Automatic server moderation was also added
- Group chats with GPT3
- Better image prompt optimizer, now works exceptionally well for Midjourney and Stable Diffusion too!
- Entirely revamped for discord slash commands, reliability, and performance in high-throughput environments