I created this small repo to make it easier to stay up-to-date with research in ML. It uses the paper recommendations posted (almost) daily on huggingface.co/papers (Thanks a lot to AK) and stores all of them in papers.json. A simple web interface allows searching all recommended papers, sorting them by upvotes (from huggingface) or date. You can also select papers by arxiv tag by clicking on the tags below the author list.

Try it out on https://mlpapers.netlify.app/.

• Web interface for keyword search within the paper titles, abstracts, tags and authors.
• Sort papers by publication date or number of upvotes.
• Define a date range with start and end date
• Select arxiv tags or authors by clicking on them
• Combine search queries with a semicolon
• The paper list gets updated daily and upvotes get updated every sunday (see branch "uptodate")

Clone the repository and install the requirements:

git clone https://github.com/mwoedlinger/HuggingfacePapersViewer.git
cd HuggingfacePapersViewer
pip install -r requirements.txt

Search papers with the web interface:

python -m http.server

Open a browser and go to http://localhost:8000 to see the web interface.

You can manually update the papers.json file with update_papers.py

python update_papers.py --start_date YYYY-MM-DD

which updates the papers.json file with all papers posted since YYYY-MM-DD (you can also specify an end date with --end_date if you are only interested in papers between two specific dates). This might take a moment (took me around 5 seconds per day).

Updating the upvotes alone can be done with update_votes.py.