Papers with Code auto-links methods and benchmarks across 121k papers for just $5
NielsRogge · x · 2026-10-23
NielsRogge announced a new Papers with Code feature powered by Decision Models: methods and benchmarks mentioned in paper abstracts are now highlighted, and hovering over them links to the corresponding pages. The pipeline is built with Jev from @typesafeai.
How it works:
- A rule-based algorithm first identifies candidate keywords in each abstract.
- Jev then classifies, for each keyword, whether it's genuinely relevant and should be linked.
The striking part: running this over 121,000 papers cost only $5, a nice showcase of a rules-plus-small-model pipeline for cheap large-scale annotation.
More from Apps
- ChatGPT web chat has a built-in cloud browser, and users are just finding out — bdsqlsz · 2026-10-23
- SEO Expert Warns: LLM Discovery Ends in Google AI Overview Linking to Competitors — lilyraynyc · 2026-10-23
- Tiny Computers App Turns iPhone Into 10 Classic Machines From Macintosh to Windows XP — xiaohu · 2026-10-23
- Open-source MCP server lets AI agents control a real iPhone via Safari screen sharing — semstr · 2026-10-23
- He Turned His Apple Watch into a Gang of AI Agents, Built Entirely with Claude Code — mikemockup · 2026-10-23
- Paul Hudson's iPhone Duo app test: hits like Things, misses like YouTube and Gmail — rudrank · 2026-10-23