v0.0.2

• Clear previous chat messages when LLMInference::load_model is called
• Allow rendering non-ASCII characters on the chat interface generated by the LLMs/SLMs
• Show token generation speed (in tokens/second) for the latest message in the chat interface (#1)