Vector Embeddings Workshop – University of Vavuniya

This open-source project introduces students to vector embeddings, vector stores, and modern semantic search techniques. It was developed and delivered as a workshop for undergraduate students at the University of Vavuniya.

🔍 Overview

• Basics of vector representations
• Introduction to semantic similarity
• PCA-based dimensionality reduction
• 3D interactive visualization of embeddings
• Color mapping based on 3D vectors

📁 Project Structure

.
├── demo/                     # Source code and visual outputs
│   ├── Color vectors for demo.py
│   ├── colorful_vectors_3d.html
├── data/                     # Input embeddings (GloVe)
│   └── glove.6B.100d.txt
├── docs/                     # Presentation slides
│   └── Vector-Stores-and-Embeddings-The-Future-of-Search.pdf
├── notebooks/                # Jupyter demonstration notebook
│   └── embedding_demo.ipynb
├── requirements.txt
├── LICENSE
└── README.md

🚀 Getting Started

# Clone the repo
https://github.com/your-username/vector-embeddings-workshop-vavuniya.git

# Install dependencies
pip install -r requirements.txt

# Run the demo
python "demo/Color vectors for demo.py"

# Open the visualization
open "demo/colorful_vectors_3d.html"

📦 Requirements

• Python 3.7+
• numpy
• plotly
• scikit-learn
• matplotlib

🧠 Learnings

• Dimensionality reduction helps visualize high-dimensional embeddings
• Color mapping gives intuition about vector closeness
• Word embeddings (e.g. GloVe) can be visually explored in 3D

🌐 Live Interactive Demo

Explore the 3D visualization of word embeddings here:
👉 Interactive Plotly Demo

Rotate and hover to understand how words cluster in high-dimensional space!

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgements

Thanks to students and faculty of the University of Vavuniya. Workshop and code developed by Nisaharan Genhatharan.