Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
IzzyViz is a Python library that visualizes attention scores in transformer models, supporting architectures like BERT, GPT-2, and T5. It provides flexible heatmap functions for comparing attention patterns, tracking stability, and highlighting key regions, outputting static PDFs suitable for research.
Parse Score
Sources
reddit.com shapes more of what AI says about IzzyViz than any other source, at 100% of its citations.