NLP-progress
A community-maintained list of NLP tasks, datasets, and current best published results.
- Stars
- 23k
- Forks
- 3.6k
- Licence
- MIT
- Language
- Python
- Last update
- 2.3 years ago
- Type
- Curated list
No commits in over a year — check for forks or alternatives before adopting.
What it is
NLP-progress is a collection of Markdown pages, one per natural language processing task, that lists benchmark datasets and the current state-of-the-art results for each task. It is meant as a quick reference for someone who wants to see what the best published systems achieve on a problem such as named entity recognition or machine translation. Contributors add results and datasets through pull requests.
- Browse the table of contents to find datasets for a task such as sentiment analysis or summarization
- Compare reported scores for a benchmark before choosing a model to study
- Follow links to public leaderboards that the README points to for certain tasks
- Use the structured JSON export described in the README to process the task data programmatically
Great for
- Researchers looking for benchmark datasets for a specific NLP task
- Students who want an overview of which NLP tasks exist and how they are evaluated
- Practitioners checking which published methods currently score highest on a task
- Anyone who wants a starting point for finding public leaderboards
Not ideal for
- People who need a software library or tool to run NLP code, since the README describes a reference of results rather than an installable package
- Readers who need guaranteed up-to-date results, because the README relies on community contributions and the listed last push date is 2024-07-28
How to get it
- 1
Read it online
Clone the repository to read the files locally
git clone https://github.com/sebastianruder/NLP-progress - 2
Run it
- Open https://nlpprogress.com/ in a browser to read the site
- Pick a task from the table of contents, such as named entity recognition
- Read the dataset description and the results table, sorted with the best result on top
- Follow the Code column links to find implementations where listed
- 3
Get the most out of it
- Use the structured JSON export described in structured/README.md to work with the data in code
- Check the Code column for official implementations before reproducing a result
- Follow the Chinese NLP link in the README for more Chinese-language tasks and datasets
- Use the wish list in the README to see which tasks and datasets contributors have not yet added
Downloads
Strengths
- Covers many core and newer NLP tasks in one place
- Pages link to papers and leaderboards so results can be checked at the source
- Contributions follow a documented process with guidelines on what counts as a valid result
Trade-offs
- It is a list of results rather than working code, so it does not run models itself
- The README notes that results are preferred from published papers, so very recent work may be missing
- Coverage of non-English languages is uneven, with some languages listing only one or two tasks
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| NLP-progress (this) | 23k | Curated list | beginner | 2.3 years ago | Researchers looking for benchmark datasets for a specific NLP task |
| Made-With-ML | 49.9k | Learning resource | intermediate | 8 months ago | Software engineers who want to learn how ML models get deployed and maintained |
| stanford-tensorflow-tutorials | 10.4k | Learning resource | intermediate | 5.9 years ago | Students following the original CS 20 course materials |
| spaCy | 34k | Library | intermediate | 10 days ago | Python developers building text-processing features into an application |
| DocsGPT | 18.4k | App | intermediate | 4 days ago | Teams that want an internal assistant over their own documents and wikis |
| ml-visuals | 17.4k | Other | beginner | 3.7 years ago | Machine learning students preparing presentations or thesis figures |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.