StarVector

StarVector is an open-source, browser-based tool for visualizing and exploring high-dimensional data through interactive 2D and 3D projections. It is designed to help users understand complex embeddings, such as word vectors, sentence embeddings, image features, or model representations, by mapping them into an interpretable visual space. Users can upload their own vector datasets, often in common formats like CSV or JSON, and view them as scatter plots with customizable colors, labels, and groupings.

The tool supports common dimensionality reduction techniques, such as PCA and t-SNE, to project high-dimensional vectors into lower-dimensional spaces while preserving meaningful structure. Users can zoom, pan, and hover to inspect individual points, examine neighborhood relationships, and identify clusters or outliers. StarVector is particularly useful for researchers and practitioners working in natural language processing, machine learning, and data science who need to debug models, compare embeddings, or present results.