# TSMPD Search Demo Notebook

This directory contains a working Jupyter notebook that demonstrates semantic product search on the TSMPD-US-Public v1.0 dataset.

## What's Inside
- `tsmpd_search_demo.ipynb`: A step-by-step demo of embedding-based search using `faiss` and `sentence-transformers`.

## Features
- Load SBERT embeddings from parquet chunks
- Build and query a FAISS index for nearest neighbor search
- Explore how consumers might search across 355k+ small U.S. stores

## Setup
To run this notebook, install the following:
```bash
pip install faiss-cpu polars sentence-transformers
```

## Licensing
Code in this notebook is licensed under the MIT License.