chore: increase max_new_tokens parameter to 4096 across chat model and inference functions for improved response generation

Implement initial project structure with Dockerfile, requirements, and Streamlit app. Added model loading and inference utilities, along with chat management features. Updated entry point and added new dependencies.