As much as I dislike all the baggage that comes with Elon owning Twitter, it is the only place I know with a critical mass of AI talent sharing papers and thoughts. Even in read-only mode, it's incredibly useful as a way to stay up to date with interesting AI papers and discussions.
The website was launched in October. It is already used by researchers in most of the leading universities and tech companies and is growing fast. Let me know what you think.. cheers!
Thanks! That's something we are considering, given how many people mentioned it and offered help. Right now, the help we need is writing a paper detailing all aspects of the system and comparing it with similar systems. We have a group in Slack: if you want to contribute, that would be great!
Love the site, but the one thing I am missing is instantly seeing what papers are trending right now, instead of having to manually select a timeframe. Think hacker news frontpage. Would be interesting to add a "hot" filter, or similar. Average weighting by exponential decay over time? Not sure how you usually do that with pagerank.
Thanks! Yes, that's in the roadmap; I believe it will be complete in 3-4 cycles. Right now, we are finishing the topics module. It will provide us with more granular filtering tools, and enable us to also discover trending topics, etc. It will be eye-opening... but it's taking longer than expected because of the dataset size.
Discord is where our conversations go to die, yes; but (non-private) Github repos are clonable and very searchable - so I'm not sure why you wrote that.
I use https://huggingface.co/papers and frankly more than enough, usually anything significant enough has a paper there or is mentioned in the papers unless it's completely proprietary stuff (Midjourney, NovelAI etc) so they're not very interesting from a research point of view.
how does this even merit inclusion on arxiv? Does it even count as science? We're talking meta-computer-science. Meanwhile, It's hard enough for me to get the endorsement to publish my actual math paper on there, which has afik new results. The computer science section of arxiv feels like a giant citation ring of mediocre, math-light papers.
It’s basically academia washing for the new tech bubble.
They saw how closely and uncritically people followed papers about COVID and realized if they released “papers” about whatever llm product they were working on it would lend some kind of credibility that they never got with cryptocurrency or Web 3.0 or the metaverse.
So glad the AK account exists. As a researcher, I've always wanted some guy with an econ degree and a year of ML eng to recommend me papers after glancing at them for maybe 30 seconds.
I am genuinely baffled that researchers in the field think there is any value in the service AK provides. I'd wager I could create an equally effective bot with the following process:
1) Create a historical dataset of publications and their citation counts
2) For each publication extract the following features:
- H-index of first author
- Maximum H-index of all authors
- Number of author affiliations in {top-10 school, deepmind, meta, openai, nvidia}
- Number of times the phrase "state-of-the-art" appears
- Which latex template is used (NeurIPS, ICML, etc.)
- Number of images in the paper
- Whether there is an image on the first page
- Whether "all you need" appears in the title
- Whether the publication has a linked project page
3) Train a shallow decision tree with citation counts as the regression target
"Do I recognize the latex template" is my number one filter when clicking through the new arxiv papers each day, so I definitely buy that that would work.
A friend of mine created a bot to do basically this, except it also looks at the current page rank associated with researchers recommending that paper. I've seen a lot of good looking papers (decent school/group/conference submission/etc.) that don't end up contributing to the field. Top researchers and Professors tend to have a better intuition of importance by reading the abstract and a quick skim.
There are have been many, many services that have tried to automate paper selection based on these heuristics. None of them have had the staying power of AK's account. As someone with a PhD in machine learning from Stanford, I can attest AK's taste is quite good.
You should be ashamed of yourself and apologize. What's more likely? There's paid shills lying that he provides non-negative value, or you're missing something when asserting for every ML PhD on earth, they must think his extremely popular work has negative value?
Apparently my innuendo has not been taken well, so I'll clarify in more straightforward terms: @abidlabs is the employer of the individual running the aforementioned twitter account.
If this is true then not disclosing that is extremely unethical of @abidlabs, damn near intellectual malfeasance, and reflects very poorly on the rest of their work.
> As a researcher, I've always wanted some guy with an econ degree and a year of ML eng to recommend me papers after glancing at them for maybe 30 seconds.
This kind of elitism is baffling given the quality of work from independent researchers recently.
And no, I have no I have no “vested interest” (whatever the hell thats supposed to mean) in someone’s Twitter.
If someone has done more of the quantitative side of econ, they are well positioned to pick up ML real fast. And the average AI/ML paper simply isn't very difficult to understand. I was a comp sci undergrad and some of the econometrics focused folks were much closer to ML work than anyone doing comp sci (this was quite some time ago though).
Despite the paper's claims to the contrary, they most likely have the causality absolutely backward. The two accounts they analyse, AK (@_akhaliq) and Aran Komatsuzaki (@arankomatsuzaki), have much better taste than anyone or anything else I've come across that are as prolific as they are. There's a reason lots of us ML/AI researchers follow them closely, which is that over an extended period of time we've noticed that they keep surfacing interesting and important work.
They do a much better job of it than some random reviewer, so you would not expect some other papers with the same review scores, venues, and embeddings to have the same citation counts. But this is the premise that this paper entirely relies on as the basis of the analysis, and as a result the paper fails to make a strong case.
The idea that thousands of researchers would decide to follow two particular Twitter accounts for no good reason, and would like and retweet posts from those accounts just because they're "influencers", makes no sense at all. Accounts become influencers for a reason. In this case, the accounts are not from people who are otherwise famous for some external reason -- literally the only thing they're known for is posting papers. If they weren't good at finding papers that researchers found interesting, they wouldn't get followed.
BTW, some years ago, Miles Brundage's Twitter account was the main one that researchers followed for this purpose. Nowadays he's the head of policy at OpenAI and no longer does this. But back when he did, the '@brundagebot' bot was created that tried to replicate his curation choices using an algorithm.
50 comments
[ 2.5 ms ] story [ 123 ms ] threadDoes a good alternative to this exist?
I know Musk would be upset though.
We created this tool to help:
https://trendingpapers.com/
The website was launched in October. It is already used by researchers in most of the leading universities and tech companies and is growing fast. Let me know what you think.. cheers!
...Yes, basically siloed off and unsearchable.
> Sign in to search code on GitHub - Before you can access our code search functionality please sign in or create a free account.
how does this even merit inclusion on arxiv? Does it even count as science? We're talking meta-computer-science. Meanwhile, It's hard enough for me to get the endorsement to publish my actual math paper on there, which has afik new results. The computer science section of arxiv feels like a giant citation ring of mediocre, math-light papers.
They saw how closely and uncritically people followed papers about COVID and realized if they released “papers” about whatever llm product they were working on it would lend some kind of credibility that they never got with cryptocurrency or Web 3.0 or the metaverse.
I am genuinely baffled that researchers in the field think there is any value in the service AK provides. I'd wager I could create an equally effective bot with the following process:
This kind of elitism is baffling given the quality of work from independent researchers recently.
And no, I have no I have no “vested interest” (whatever the hell thats supposed to mean) in someone’s Twitter.
They do a much better job of it than some random reviewer, so you would not expect some other papers with the same review scores, venues, and embeddings to have the same citation counts. But this is the premise that this paper entirely relies on as the basis of the analysis, and as a result the paper fails to make a strong case.
The idea that thousands of researchers would decide to follow two particular Twitter accounts for no good reason, and would like and retweet posts from those accounts just because they're "influencers", makes no sense at all. Accounts become influencers for a reason. In this case, the accounts are not from people who are otherwise famous for some external reason -- literally the only thing they're known for is posting papers. If they weren't good at finding papers that researchers found interesting, they wouldn't get followed.
BTW, some years ago, Miles Brundage's Twitter account was the main one that researchers followed for this purpose. Nowadays he's the head of policy at OpenAI and no longer does this. But back when he did, the '@brundagebot' bot was created that tried to replicate his curation choices using an algorithm.
[1]: https://news.ycombinator.com/item?id=39124309