Research & Papers

Hugging Face revives PapersWithCode with CVPR 2026 conference support

Browse 1,000+ CVPR 2026 papers with arXiv IDs, GitHub links, and evaluations.

Deep Dive

Niels, part of the open-source team at Hugging Face, announced a major update to paperswithcode.co, a revival of the original PapersWithCode website that launched two weeks ago. The platform lets users track state-of-the-art (SOTA) results across diverse AI domains—from agents and computer vision to time-series forecasting. The latest feature adds dedicated conference support, enabling easy browsing of papers from top-tier AI conferences like NeurIPS, CVPR, and ICML. For CVPR 2026, which kicks off next week in Denver, all papers have been indexed with corresponding arXiv IDs, organized by task, and enriched with linked GitHub repositories, project pages, Hugging Face artifacts (e.g., models, datasets), and evaluation benchmarks. Users can also filter for Oral and Spotlight papers.

The move addresses a long-standing community need: the original PapersWithCode site was sunset after acquisition, leaving a gap in centralized SOTA tracking. By rebuilding it as an open-source project under Hugging Face, Niels aims to provide a seamless, up-to-date resource for researchers and practitioners. The conference page (paperswithcode.co/conferences) currently highlights CVPR 2026 but promises ongoing coverage for NeurIPS, ICML, and others. The platform is free to use, with feedback encouraged via the Hugging Face community. This revival not only restores a beloved tool but also integrates tightly with the Hugging Face ecosystem, making it easier to go from paper to model to deployment.

Key Points
  • paperswithcode.co indexes all CVPR 2026 papers with arXiv IDs, grouped by task.
  • Papers include linked GitHub repos, project pages, Hugging Face artifacts, and evaluations.
  • Conference support extends to NeurIPS, ICML, and future events; Oral/Spotlight filters available.

Why It Matters

Revives a critical open-source SOTA tracker with Hugging Face integration, easing paper-to-model workflows for AI researchers.