Hello, today I'm releasing a big update!
It doesn't add any content but greatly improve the performances of the game.
Because it brings a lot of breaking changes, it will remain in experimental for some time.
To use this new version, enable it in the beta tab on Steam.

Limitations
- The models KoboldAI/GPT-J-6B-Adventure and KoboldAI/GPT-J-6B-Skein aren't supporting the new version yet.
- You have to redownload all the models.
Client
- Made compatible the client with the new server's version.
- The AI's files are now all stored in the same folder in `models`
- Added many new options to optimize the performances.
- Added an option to allow the server to download the AI by itself (It will be useful for custom models, later).
Server
- The server can now run the AI on both the GPU and CPU at the same time.
- No more memory peak when loading the AI
- Added an option to limit the memory used (for both the GPU and CPU)
- Added an option to write data in a folder instead of using the memory
- The config file has now much more options.
