A community-driven repository tracking datasets and state-of-the-art results for common NLP tasks across multiple languages.
NLP-progress is a community-maintained repository that systematically tracks the state-of-the-art in Natural Language Processing. It aggregates benchmark datasets and model performance across dozens of NLP tasks, providing a quick reference for researchers to understand current capabilities and gaps. The project serves as a living document of the field's evolution.
NLP researchers, machine learning engineers, and students who need to benchmark models, identify research opportunities, or survey current SOTA results for specific tasks.
It offers a uniquely comprehensive and structured overview of NLP progress across many tasks and languages, all in one open, community-updated location. Unlike scattered papers or leaderboards, it provides a centralized, easily navigable resource.
Repository to track the progress in Natural Language Processing (NLP), including the datasets and the current state-of-the-art for the most common NLP tasks.
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Tracks over 40 core NLP tasks from machine translation to sentiment analysis, providing a one-stop reference for diverse research needs, as listed in the detailed table of contents.
Includes progress for languages beyond English like Chinese, French, and Vietnamese, aiding global NLP initiatives and reducing English-centric bias in benchmarking.
Built on GitHub with clear contribution guidelines, allowing the community to submit updates via pull requests, ensuring the resource evolves with the field.
Provides machine-readable JSON exports of all tasks and SOTA tables, enabling easy data extraction for meta-analyses or integration into research workflows.
Prefers results from published papers with exceptions for influential preprints, enhancing credibility and making it a reliable academic reference.
Relies on manual pull requests for updates, causing delays in reflecting the latest SOTA or fast-moving research breakthroughs, unlike automated leaderboards.
Coverage varies by language and task; for example, English has extensive entries while some languages like Nepali have only one task, limiting utility for low-resource NLP.
No built-in verification process for submitted results, relying on community oversight which can lead to errors or biased data persisting without correction.
Markdown-based tables lack interactive features like filtering or visualization, making it less engaging and harder to navigate than dynamic platforms like Papers with Code.