zettabrain-rag 0.1.0

Local private RAG pipeline — LangChain + Ollama + ChromaDB + NFS support

ZettaBrain RAG

Local private RAG pipeline — your documents, your hardware, zero cloud.

---

Install

pip install zettabrain-rag

Requires Python 3.10+ and Ollama.

---

One-time setup

1. Install Ollama and pull models

curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3.1:8b
ollama pull nomic-embed-text

2. Mount NFS share and build vector store

sudo zettabrain-setup

Prompts for NFS server IP and export path, mounts at /mnt/Rag-data, then builds the vector store automatically.

3. Start chatting

zettabrain-chat

---

Commands

Command	What it does
sudo zettabrain-setup	NFS mount wizard + auto vector store build
zettabrain-chat	Start interactive RAG chat
zettabrain-chat --rebuild	Rebuild vector store then chat
zettabrain-chat --debug	Show retrieved chunks on every query
zettabrain-ingest	Ingest documents without starting chat
zettabrain-ingest --file /path/to/file.pdf	Ingest a single file
zettabrain-ingest --stats	Show what's in the vector store
zettabrain-ingest --clear	Wipe the vector store
zettabrain-status	Show install paths and store statistics

---

Directory structure after install

/zettabrain/
├── nfs_setup.sh              ← NFS mount wizard (run via sudo zettabrain-setup)
└── src/
    ├── 03_langchain_rag.py   ← main RAG pipeline
    ├── 05_ingest_documents.py← ingestion utility
    ├── 01_chromadb_setup.py  ← diagnostic: verify ChromaDB
    ├── 02_embeddings_test.py ← diagnostic: verify embeddings
    └── zettabrain_vectorstore/  ← auto-generated, do not commit

---

Configuration via environment variables

export ZETTABRAIN_DOCS=/mnt/Rag-data          # documents folder (NFS mount)
export ZETTABRAIN_CHROMA=./zettabrain_vectorstore
export ZETTABRAIN_LLM_MODEL=llama3.1:8b
export ZETTABRAIN_EMBED_MODEL=nomic-embed-text
export ZETTABRAIN_CHUNK_SIZE=1500
export OLLAMA_HOST=http://localhost:11434

---

Supported document formats

.pdf .txt .md .docx

---

Hardware guide

RAM	Model	Notes
8GB	llama3.2:3b	Basic
16GB	llama3.1:8b	Recommended
32GB	mistral-nemo:12b	Better reasoning
Apple M3/M4	llama3.1:70b-q4	Excellent

---

License

MIT — © ZettaBrain