Show HN: Papers with Code – trending ML research with available code (opens in new tab)

(paperswithcode.com)

9 pointsrstoj7y ago4 comments

4 comments

rstojOP7y ago

Hi all,

Sharing a project I’ve been working on in spare time: Papers with Code.

Papers with Code links the latest machine learning papers on ArXiv with code on GitHub. It allows you to see “trending research” allowing you to keep up-to-date with what’s popular in the ML community, as well as the most popular historical papers. Additionally it shows the framework the implementation was coded in: TensorFlow/PyTorch/MXNet/etc. Lastly, you can search for papers and choose a code implementation of your choice.

Links are automatically scraped from arxiv papers and github repos, so in some cases e.g. the SNIPER repo there are three entries because this library implements three papers.

Let me know your thoughts, and if there’s any other feature ideas you have for the site,

Thanks! Robert

ivansavz7y ago

I like that it shows the DL tool used. Might be good for people learning to see some state-of-the-art stuff before choosing a framework.

See also http://www.gitxiv.com/ which seems similar.

stealthcat7y ago

Really awesome piece of work!

Might need tags by datasets used, i.e. mnist is image classification, COCO is object recognition, etc.

oliverx07y ago

This is amazing thanks so much! I think it is a great way to learn deep learning.

j / k navigate · click thread line to collapse