v3.1.0

• Started using pipenv to (hopefully) make dependency errors can go away
• Fixed the huggingface llm download error in setup.py among some other minor setup.py improvements

v3.0.1

Fixed an issue with setup.py

v3.0.0

Finally got WhisperX and ExLlamaV2 working! That means free, local, real-time responses. Unfortunately, even though we all love Megumin, I decided to move more to the M.I.T.S.U.H.A. character. However, if you truly want to walk the path of explosions, you can just edit the .env file lore and change all instances of M.I.T.S.U.H.A. to Megumin in the code and conversation.jsonl.

v2.1.1

Added automatic GPU support in setup.py

v2.1.0

Added time and date awareness to the AI.

v2.0.2

Some more minor setup bug fixes

v2.0.1

Small fixes in requirements, OneReality.bat, and setup.py

v2.0.0

Basically v2.0.0-beta but with automatic easy installation by @jaxfry!

v2.0.0-beta

Core of the AI is finally complete! This means all the essential features are done for software. All the updates from now on will be new cool features and hardware, but the base is done! The reason this is beta is because @jaxfry is going to help me get easy install set up soon.

• Got realtime easy no wsl vits inference working
• Korean is now supported

v2.0.0-alpha

New version is almost ready for release, only issue is that inference is pretty slow, 10-15s. Other than that, new features are looking great including:

• Multilingual nearly perfect megumin voice with English, Japanese, and Chinese
• Everything local (except whisper by default as whisper.cpp doesnt work good at all for me at least, however i will leave it as an option)
• Virtually infinite long term memory using vector databases
• Alexa like smart home control using Tuya API
• Open apps without specifying path, only name of app
• Easier setup with dotenv