Releases: lalaland-ai/lala-companion
v0.0.12
v0.0.11
- new VRMCompanion component. updated from latest @lalaland-ai RnD.
- mouth lip sync based on volume of audio playing.
- better lighting for overlay 3d scene
- prompts optimized, 25% less tokens.
- setup for self operating emotions
v0.0.10
- fix lala not moving, animations url 404 (oops)
- add a few more diverse animations
- bump package.json versions to latest everything
v0.0.9
Basic self operating AI computer demo. Shout out to: @OthersideAI https://github.com/OthersideAI/self-operating-computer for the prompts and ideas. But we just had to make it Typescript ;)
It deff does fail some X/Y cords sometimes, depends what you request.
Some things it did correctly though:
- write me a tweet in timeline on webpage, it found text box, clicks in and wrote tweet good.
- write me a Haiku poem in my Notepad app.
- find this file: "Documents/github/lala-companion/README.md" - it was able to drill down through file explorer and even took a novel approach of just using the search bar on it's own and typed in the path.
This is preliminary. We can improve many ways, script functions and just executing child process commands may be what's needed for better stablity on current GPT4-V... I'm sure GPT5-V will greatly improve X/Y... prepare for the AGI.
v0.0.8
-
refactor Vite -> Webpack because nutjs was broken on Vite and also for some reason it was taking 35 minutes for windows builds after just adding MUI packages. Now its 3 minute builds like normal and everything works. sry Vite.
-
installer bundle sizes are cut in half, 50% from Vite -> Webpack.
-
added screenshot ability button to test for future self operating computer functions, also moves your mouse as test for now.
v0.0.7
- request darwin mic permissions
v0.0.6
- add MUI, cleanup design, theme
- toggle for always on user microphone
- prompt text box for regular chats to AI
v0.0.5
- realtime whisper user microphone to AI prompt magic
v0.0.4
First user "ready" version of basic demo. Early access.
- gpt4 text streaming
- AI voice audio
- render 3D VRM Avatar in overlay
- initial greeting to user onload