Skip to content
AI’s

evaluation

Dimensionality Reduction

Techniques for reducing the number of features in a dataset while preserving important information. Dimensionality reduction helps with visualization, noise removal, and improving model efficiency on high-dimensional data.

In practice

t-SNE reduces 768-dimensional embeddings to 2D for visualization, showing how similar documents cluster together.