Skip to content

Notebooks

Run the RAG Lab examples directly in your browser — no local setup required — or pull them down and run them on your own machine for full control.

What you'll learn

  • How to open and run any notebook in Google Colab with one click
  • How to run notebooks locally with Jupyter or VS Code
  • Which notebook matches which tutorial in the site

Cloud vs local: which should I use?

Google Colab Local runtime
Setup Zero — just a Google account Python + Jupyter or VS Code
GPU Free tier available (~12 h cap, not guaranteed) Your hardware
Ollama Not supported on the default cloud runtime Fully supported
Persistence Resets on session end Your disk

Quick start recommendation

If you just want to read along and run cells without installing anything, use Colab. If you are working through the Ollama + Chroma tutorial or need a persistent environment, run locally.

Ollama steps require a local runtime

Notebooks that use Ollama (e.g. loading llama3 or nomic-embed-text) cannot connect to a locally running Ollama server from the Colab cloud runtime. Either skip those cells in Colab or switch to a local runtime (Runtime → Change runtime type → Local runtime in Colab). Notebooks that only use sentence-transformers or the OpenAI-compatible API work fine on the default Colab cloud.

Available notebooks

01 — Minimal RAG

Open In Colab

Build a retrieval-augmented generation pipeline from scratch in under 50 lines of Python, using sentence-transformers for embeddings and a simple in-memory vector store.

Colab compatible: Yes — uses sentence-transformers (pip-installable), no Ollama required.

Matches tutorial: Minimal RAG

02 — RAG Evaluation

Open In Colab

Measure the quality of your retrieval pipeline and generated answers with standard metrics including precision@k, recall@k, MRR, and faithfulness scoring.

Colab compatible: Yes — evaluation helpers are pure Python or lightweight pip packages.

Matches tutorial: Evaluation

Running notebooks locally

Option 1 — Jupyter Lab / Notebook

# Install Jupyter if you haven't already
pip install jupyterlab

# Clone the repo
git clone https://github.com/sitharaj88/rag-lab.git
cd rag-lab

# Install project dependencies
pip install -r requirements.txt

# Launch Jupyter Lab
jupyter lab notebooks/

Navigate to any .ipynb file in the file browser on the left and click to open it.

Option 2 — VS Code

# Install the Jupyter extension from the Extensions panel (Ctrl+Shift+X)
# Search: "Jupyter" by Microsoft

# Open the repo folder
code /path/to/rag-lab

# Open any .ipynb file directly — VS Code renders it natively

Kernel selection

When VS Code asks you to select a kernel, choose the Python environment where you installed requirements.txt (usually the virtual-env you created during environment setup).

Keeping outputs clean for git

# Strip outputs before committing so diffs stay readable
pip install nbstripout
nbstripout --install          # installs a git pre-commit hook

Where to go next