WizMap <a href="https://poloclub.github.io/wizmap/"><img align="right" src="./src/imgs/icon-wizmap.svg" height="38"></img></a>

Exploring and interpreting large embeddings in your browser!

<table> <tr> <td colspan="3"><video width="100%" src='https://github.com/poloclub/wizmap/assets/15007159/26d26336-e611-4d00-b535-8837da59be82'></td> </tr> <tr></tr> <tr> <td><a href="https://poloclub.github.io/wizmap">🚀 Live Demo</a></td> <td><a href="https://youtu.be/8fJG87QVceQ">📺 Demo Video</a></td> <td><a href="https://arxiv.org/abs/2306.09328">📖 Research Paper</a></td> </tr> </table>

What is WizMap?

WizMap is a scalable interactive visualization tool to help you easily explore large machine learning embeddings. With a novel multi-resolution embedding summarization method and a familiar map-like interaction design, WizMap allows you to navigate and interpret embedding spaces with ease.

<table> <tr> <td>✅</td> <td>Scalable to <strong>millions</strong> of embedding point</td> </tr> <tr></tr> <tr> <td>✅</td> <td><strong>Multi-resolution</strong> embedding summaries</td> </tr> <tr></tr> <tr> <td>✅</td> <td><strong>Fast</strong> embedding search</td> </tr> <tr></tr> <tr> <td>✅</td> <td><strong>Multimodal</strong> data (text and image)</td> </tr> <tr></tr> <tr> <td>✅</td> <td>Animated <strong>embedding evolution</strong></td> </tr> <tr></tr> <tr> <td>✅</td> <td>Support <strong>computational notebooks</strong> (e.g., Jupyter, Colab, VS Code)</td> </tr> <tr></tr> <tr> <td>✅</td> <td><strong>Sharable URLs</strong></td> </tr> <tr></tr> </table>

WizMap Gallery

<table> <tr> <td><a href="https://poloclub.github.io/wizmap/?dataset=diffusiondb"><img src="https://i.imgur.com/hTcu3rJ.jpg" width="100%"></a></td> <td><a href="https://poloclub.github.io/wizmap/?dataset=diffusiondb"><img src="https://i.imgur.com/AYUEHWz.png" width="100%"></a></td> <td><a href="https://poloclub.github.io/wizmap/?dataset=diffusiondb"><img src="https://i.imgur.com/iyBzqHQ.png" width="100%"></a></td> </tr> <tr></tr> <tr> <td><a href="https://poloclub.github.io/wizmap/?dataset=diffusiondb">DiffusionDB Prompts + Images</a></td> <td><a href="https://poloclub.github.io/wizmap/?dataset=acl-abstracts">ACL Paper Abstracts</a></td> <td><a href="https://poloclub.github.io/wizmap/?dataset=imdb">IMDB Review Comments</a></td> </tr> <tr></tr> <tr> <td>1.8M text + 1.8M images</a></td> <td>63k text</td> <td>25k text</td> </tr> <tr></tr> <tr> <td><code>CLIP</code> Embedding</td> <td><code>all-MiniLM-L6-v2</code> Embedding</td> <td><code>all-MiniLM-L6-v2</code> Embedding</td> </tr> <tr></tr> </table>

Submit a PR to add your WizMap here! You can share your WizMap using a unique URL.

Web Demo

For a live web demo, visit: https://poloclub.github.io/wizmap.

Get Started

Clone or download this repository:

git clone git@github.com:poloclub/wizmap.git

Install the dependencies:

npm install

Then run WizMap:

npm run dev

Navigate to localhost:3000. You should see WizMap running in your browser :)

Use My Own Embeddings

To use WizMap with your embeddings, you first need to install the wizmap Python library.

pip install wizmap

Then take a look at this notebook for a detailed guide. If your dataset includes images, take a look at this notebook.

Spoiler alert: You'll be up and running with just two function calls from the wizmap library. These two JSON files contain pre-computed embedding summaries, distributions, and the original embedding data.

After generating the JSON files, you have two options to use WizMap.

Browser: You can click the folder icon on the bottom right of the WizMap demo and enter the URLs to the JSON files in the pop-up window.
Notebook: If you are familiar with computational notebooks (e.g., Jupyter Notebook, VSCode Notebook, Colab), you can directly use WizMap in your favorite notebook platform with wizmap.visualize().

Share My WizMap

You can easily share your embedding maps with collaborators by copying the unique URL of WizMap from your browser. For example, the URL below points to an IMDB embedding in WizMap.

https://poloclub.github.io/wizmap/?dataURL=https%3A%2F%2Fhuggingface.co%2Fdatasets%2Fxiaohk%2Fembeddings%2Fresolve%2Fmain%2Fimdb%2Fdata.ndjson&gridURL=https%3A%2F%2Fhuggingface.co%2Fdatasets%2Fxiaohk%2Fembeddings%2Fresolve%2Fmain%2Fimdb%2Fgrid.json

Credits

WizMap is created by <a href='https://zijie.wang/' target='_blank'>Jay Wang</a>, <a href='http://fredhohman.com/' target='_blank'>Fred Hohman</a>, and <a href='https://poloclub.github.io/polochau/' target='_blank'>Polo Chau</a>.

Citation

To learn more about WizMap, please read our research paper (published at ACL'23 System Demonstration).

@article{wangWizMapScalableInteractive2023,
  title = {{{WizMap}}: {{Scalable Interactive Visualization}} for {{Exploring Large Machine Learning Embeddings}}},
  shorttitle = {{{WizMap}}},
  author = {Wang, Zijie J. and Hohman, Fred and Chau, Duen Horng},
  year = {2023},
  url = {http://arxiv.org/abs/2306.09328},
  urldate = {2023-06-16},
  archiveprefix = {arxiv},
  journal = {arXiv 2306.09328}
}

License

The software is available under the MIT License.

Contact

If you have any questions, feel free to open an issue or contact Jay Wang.

Wizmap

Install / Use

README