Reviving PapersWithCode (by Hugging Face) [P]

2026-05-18 · Source: Machine Learning · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Research Methodology & Innovation, Robotics & Autonomous Systems · Depth: Expert, short

Summary

Hugging Face has launched a revived version of PapersWithCode, a platform for tracking research papers and their associated code, which had been unmaintained after its acquisition by Meta. The new platform, available at paperswithcode.co, uses AI agents to parse papers at scale and automatically generate leaderboards, with human verification for results. It currently features trending papers based on GitHub star velocity, categorization by domain (e.g., OCR), methods (e.g., RLVR), evaluation results for high-impact papers like Qwen 3.5, and leaderboards for various domains such as MMTEB and COCO val 2017. The site also supports citation counts, automated linking of GitHub repositories and project pages, external papers beyond Arxiv, and Harness reports for coding agent benchmarks like Terminal Bench. User authentication is handled via "Sign in with HF," and data backups are stored using Storage Buckets.

Key takeaway

For ML researchers and AI engineers tracking the latest advancements, the revived PapersWithCode offers a centralized, updated resource for discovering state-of-the-art models and their implementations. You should explore the new platform at paperswithcode.co to find trending papers, benchmark results for models like Qwen 3.5, and specific methods. Consider contributing feedback or offering help to shape its future development, especially regarding reproducibility and consistent baselines.

Key insights

Hugging Face has revived PapersWithCode, using AI agents and human verification to track research papers and code.

Principles

• Automate data extraction with AI agents.
• Prioritize human verification for accuracy.
• Integrate community feedback for feature development.

Method

The platform parses high-impact papers using AI agents to generate leaderboards, which are then human-verified. It categorizes content by domain, tracks methods, and links to code repositories and evaluation results.

In practice

• Explore SOTA models on leaderboards like Terminal Bench 2.0.
• Filter papers by citation count or domain.
• Submit feedback and feature requests to Hugging Face.

Topics

• PapersWithCode Platform
• AI-powered Paper Parsing
• State-of-the-Art Leaderboards
• ML Research Discovery
• ML Benchmarking

Best for: AI Engineer, NLP Engineer, Computer Vision Engineer, AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• How to Actually Get Started with HuggingFace 🤗 — Hugging Face provides an open-source ecosystem for ML/NLP, simplifying model access, development, and deployment.
• Would a 2000-2021 ML paper even get accepted today? [D] — ML paper acceptance standards have shifted, with higher expectations for traditional research but a potentially lower bar for LLM-centric engineering.
• Arena Leaderboard Dataset — Arena released its historical AI capability leaderboards as a public dataset for community analysis and open science.
• Import AI 461: “Alignment is not on track”; FrontierCode; and synthetic research interns — AI progress demands new benchmarks for safety, cultural reasoning, coding quality, and research assistance, alongside faster, more aligned models.
• Automating LLM Post-Training with Hugging Face’s ml-intern — Hugging Face's ml-intern models ML research as a multi-step workflow, not just a tool-calling chat assistant.
• The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers — AI research documentation practices improved significantly from 2014-2024, driven by a broader open science movement.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.