A mini Retrieval-Augmented Generation (RAG) project using local Ollama models.
- Loads memories from a text file
- Converts memories into embeddings
- Converts user question into an embedding
- Finds relevant memories using cosine similarity
- Sends retrieved context to a local LLM
- Generates an answer based on retrieved context
- Python
- Ollama
- nomic-embed-text
- qwen2.5-coder
- NumPy
python3 main.py