chore: rename project to TokenTrace and set default language to Chinese

feat: add Tiny-NLA activation explanation with trained model weights

init: TokenTrace - LLM interpretability toolbox