Last weekend, I decided to take a step to learn and build things, it is also a way to learn about LLM, in which I ended built InkChatGPT. A small and simple RAG

Last weekend, I decided to take a step to learn and build things, it is also a way to learn about LLM, in which I ended built InkChatGPT. A small and simple RAG LLM app that with tons of learning resources that I was afraid to start to learn with. I've been diving into the world of Large Language Models (LLMs), particularly the Retrieval Augmented Generation (RAG) approach. As a learning project, I built InkChatGPT, an AI agent that can assist with learning from multiple documents through a chat interface, inspired by projects like ChatPDF. Under the hood, I used LangChain as the LLM framework to simplify the backend, Streamlit for the front-end UI and deployment, and OpenAI's `gpt-3.5-turbo` model. For generating embeddings for document chunking, I leveraged HuggingFace's `all-MiniLM-L6-v2` model. Coming from a mobile development background, delving into ML, LLMs, prompt tuning, and various techniques has been an eye-opening experience. While the wealth of information can be overwhelming, building projects like InkChatGPT has been an incredible learning journey. I'd love to hear your thoughts and experiences with LLMs and RAG systems. Feel free to check out the InkChatGPT repo [1] and share your feedback. [1] https://github.com/vinhnx/InkChatGPT https://github.com/vinhnx/InkChatGPT