An interactive web app for exploring student evaluations of AI models across homework assignments. Just drag and drop special_participation_site and you're good to go.
| View | Description |
|---|---|
| Matrix View | Contribution matrix showing which students evaluated which AI models per homework, with transpose toggle |
| Threads View | Browse all posts with metadata, badges, and expandable details |
| Homework Insights | AI-generated summaries analyzing model performance per assignment |
| Model Insights | AI-generated summaries of each model's overall strengths and weaknesses |
| Performance Analyzer | Score AI model performance (1–5 scale) using an on-device LLM |
- Advanced search with prefixes:
author:John,model:Claude,hw:HW5 - Filter by Contributor, Homework, and Model
- Sort by newest/oldest, homework, model (A–Z), or contributor (A–Z)
- Collapsible filter panel
- PDF/Text attachment support with direct download links
- Renders Markdown and LaTeX in chat responses
- AI Chat Assistant — Floating chat widget running LLMs entirely in-browser (Llama 3.2, SmolLM2, Qwen 2.5)
- RAG — Toggle semantic search over threads for context-aware responses
- Per-thread Q&A — Ask questions about individual threads in the post modal
- Performance Analyzer — On-device LLM scores how well AI models performed based on student reports
- Visualizations — Charts by homework or model, including holistic views
- Theme toggle (Auto / Light / Dark) with persistence
- Static site — no build toolchain required; works with any HTTP server
process_data.pyprocesses raw Ed thread data- Tag corrections via
tag_fix/corrections.json
- Version History — Commit ids of our initial red team submission and final blue team submission