basic-rag

Basic rag implementation

A basic RAG implementation with docker.

Requirements

• install ollama
• run

ollama pull mistral (or other models)

• run

OLLAMA_HOST=0.0.0.0 ollama serve

Docker stuff

• build the image

docker build -t allamak:0.1 .

• run

docker run -it --rm -p 8501:8501 -v /tmp/tt:/data -v /tmp:/xx -e REQUIRED_EXTS='.pdf' -e PERSIST_DIR='/xx/storage' -e LLM_SERVER_URL="your.ollama.host:port"  allamak:0.1

• access localhost:8501
• ask away once the index is done loading

fine-tuning

• embeddings, indexing, querying in llamaindex would help

refs:

• llamaindex
• ollama

PR sought XD