Merge pull request kherud#58 from kherud/refactoring

Update to llama.cpp b702

Update release workflow

Version 2.3.5

- Fixed handling of special tokens in tokenizer

Version 2.3.4

Library loading from APK on Android

Version 2.3.3

- Fixed some JNI references leading to segmentation faults
- Fixed JNI wrong data type method usage

Version 2.3.2

- Add android support

Version 2.3.1

- Fixed macos aarch64 build
- Defaulting models.home variable

Version 2.3.0

- Support for latest llama.cpp version
- Support for mixtral
- Extended pre-built libraries

Version 2.2.1

- added more information to llm outputs (token/probabilities)
- some github workflow bug fixes

Version 2.2.0

- multiple fixes and update to latest llama.cpp
- add infilling