Who we are
Perplexify · Group P3
A lightweight playground to compare base vs fine-tuned language models using perplexity visualizations. Built for fast iteration, demos, and sharing experiments with your team.
What it does
Drop in text, generate embeddings, and see PCA projections with perplexity heatmaps for base and fine-tuned models side-by-side.
Why it helps
Quickly spot divergence between training runs, validate fine-tune quality, and build lightweight regression suites without leaving the browser.
How to use it
Start on the home page, add your sample text, or run a saved test suite. The “About” button in the header always brings you back here if you need a refresher.