1 points | by imranq 4 hours ago
1 comments
But the unique feature: AI-generated "explainers" that break down complex papers with interactive visualizations. Example: https://neurips2025.pages.dev/explainers/linear_attention/
It explains why attention is hard to optimize, shows the math with interactive demos, and includes critical analysis of limitations.
The explainers are generated using Gemini3 to parse papers and create:
- Interactive visualizations
- Step-by-step mathematical walkthroughs
- Critical analysis sections
- "What would convince me?" sections
Tech stack: OpenReview API, Gemini API for explainer generation, static hosting on Cloudflare Pages for speed.
I'm planning to generate explainers for more papers based on what people find interesting, so any feedback would be amazing
But the unique feature: AI-generated "explainers" that break down complex papers with interactive visualizations. Example: https://neurips2025.pages.dev/explainers/linear_attention/
It explains why attention is hard to optimize, shows the math with interactive demos, and includes critical analysis of limitations.
The explainers are generated using Gemini3 to parse papers and create:
- Interactive visualizations
- Step-by-step mathematical walkthroughs
- Critical analysis sections
- "What would convince me?" sections
Tech stack: OpenReview API, Gemini API for explainer generation, static hosting on Cloudflare Pages for speed.
I'm planning to generate explainers for more papers based on what people find interesting, so any feedback would be amazing