README.md

Retrieval Augmented Generation with local LLMs

Advanced RAG, 'locally' on Google Colab via HuggingFace 🤗

Goal: Complex PDF question answering, many pages including figures.

• LlamaIndex pipelines, semantic partitioning, re-ranking, response synthesis
• Efficient LLM : HuggingFace/TinyLlama
• End to end Evaluation: DeepEval
• Next experiment: Efficient Multi-modal RAG for more complex visual QA HuggingFace/VILA

Vanilla RAG, locally on M1 macbook.

Goal: Privacy-preserving 🤫 sensistive document question answering 📄.

• Ollama + Mistral LLM, Chroma DB 🍭, Nomic Embeddings 🍪
• Langchain 🦜🔗 pipeline

---

"People who buy things are suckers." - Ron Swanson