Deep, multi-level explanations of every concept behind Recursive Reinforcement Language Models — definitions, the math explained, ELI5 → junior → PhD, a live concept graph, Q&A, and an Apify-powered paper→concept harvester.
0Concepts
0Learned
0Discovered
0Questions
📇 Cards
🕸 Concept Graph
💬 Ask
🔬 Discover papers
⚙ Settings
Pick a concept above — its related concepts appear as connected rectangles. Click any box to recenter.
Mode:
🔬 Paper search → concept harvesting
Searches papers via Apify on the backend, shows you the top results, you tick the ones you want, then Extract concepts turns them into cards via Claude Sonnet 4.6 (Bedrock). Tokens stay on the backend — set its URL in Settings.
📄 Top papers — pick the ones you want
🔐 Backend (Bedrock + Apify)
All tokens live on the backend, never in the browser. The backend calls Claude Sonnet 4.6 on AWS Bedrock and Apify for paper search. See backend/.
🧠 Answer mode
How Ask and concept-generation are powered.
Backend mode requires the backend running and reachable at the URL on the left.
Canonical data lives in concepts-data.json; the app seeds it into IndexedDB on first run and persists your learned status, questions, and discovered concepts there.