Papers are scored in real-time based on how reproducible they are (based on their Github repos) and how interesting they are (based on Twitter).

Stanford, CA
Joined July 2018
Tracking 50,000+ papers. We're probably the fastest growing meta journal of all time.
Replying to @karpathy
Might I suggest checking out assert.pub. Many of the same features as arxiv-sanity. Rates papers based on Twitter and Github repos. Crawls all categories of arxiv (not just ML ones), biorxiv, and many others. See the most interesting papers each day.
1
We now automatically extract out the sample size (N=181) from papers. This makes it easy to see the papers that are more likely to be verifiable. Enjoy!
Interesting or reproducible? See the papers that you want to see.
Replying to @sirajraval
@sirajraval Huge fan! You always talk about the best papers. I wanted to make it easier for you to find the most talked about papers on Arxiv. Check it out assert.pub. If you think other wizards would find it helpful, would love a RT! #100DaysOfMLCode
7
30
Replying to @sirajraval
@sirajraval HUGE FAN! I watch your videos religiously. I built a website to highlight new Arxiv papers that put their code on GitHub. I wanted to share it with you. link -> assert.pub. If you think other wizards would find it interesting would love a shout out.
Drew this picture to explain how assert.pub works.
1
8
13
Replying to @GizemKaraali_
We use a combo of twitter and github. Drew a graphic to help explain it a bit more. Trying to democratize the review process.
1
1