cristian-s.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

Few understand this jvns.ca/blog/2014/06...

submitted 40 days ago • 2 comments

"I think we have this idea that we're gonna make friends with bad guys and bend them to our will, and really we're just...friends with bad guys." --The Diplomat, S2 E4 (I've been binge-watching this, and it is excellent.)

submitted 40 days ago • 307 comments

But I heard someone say LLMs are world models?

submitted 41 days ago • 8 comments

Tree at night by firelight. Backyard in Natick, MA. 01-02-2025. #DansDayOutdoors

submitted 55 days ago • 0 comments

Migration Series. www.labelsonhumans.com #scape #photography #colourphotography

submitted 41 days ago • 10 comments

Brandon is a wonderful research colleague and I could not endorse enough trying to work with Brandon

submitted 51 days ago • 1 comment

I have a draft of my introduction to cooperative multi-agent reinforcement learning on arxiv. Check it out and let me know any feedback you have. The plan is to polish and extend the material into a more comprehensive text with Frans Oliehoek. arxiv.org/abs/2405.06161

submitted 50 days ago • 3 comments

Instead of listing my publications, as the year draws to an end, I want to shine the spotlight on the commonplace assumption that productivity must always increase. Good research is disruptive and thinking time is central to high quality scholarship and necessary for disruptive research.

submitted 69 days ago • 21 comments

It’s scathing critique season, this time for reinforcement learning. We need this, the science cannot get better without it. Usual suspects: training brittleness (over reliance on hyperparameter tuning), bad & slow sims, overemphasis on generality, LLMs dominating discourse, tabula rasa RL is hard

submitted 63 days ago • 1 comment

Chomsky, Varofakis & Greenwald issued joint declaration calling for Panama, Greenland & Canada to negotiate, trade land for peace and not provoke nuclear armed madman... Lol

submitted 65 days ago • 59 comments

the feds: stop trying to turn this guy into some cool antihero with a badass public image also the feds: treat him like they’ve captured the joker

submitted 69 days ago • 1385 comments

Catch my poster tomorrow at the NeurIPS MLSB Workshop! We present a simple (yet effective 😁) multimodal Transformer for molecules, supporting multiple 3D conformations & showing promise for transfer learning. Interested in molecular representation learning? Let’s chat 👋!

submitted 74 days ago • 0 comments

Fresh off the presses: In "Learning on compressed molecular representations" Jan Weinreich and I looked into whether GZIP performed better than Neural Networks in chemical machine learning tasks. Yes, you've read that right. TL;DR: Yes, GZIP can perform better than baseline GNNs and MLPs. It can ..

submitted 98 days ago • 4 comments

submitted 73 days ago • 226 comments

Ilya Sutskever calls data the "fossil fuel of AI" – the finite power source that kickstarted the rapid initial rise, but that is now running out and needs to be replaced with more sophisticated and sustainable methods. x.com/_jasonwei/st...

submitted 75 days ago • 1 comment

Most in the industry have been talking abt this for upwards of a year. Current form of pre training has been dead. We're at a token wall. But scale is here to stay. I'm excited by: -pretraining with human (?) prefs, merging pre and post training -online exploration for (synthetic) data collection

submitted 75 days ago • 2 comments

Is this really the correct citation for the Gaussian distribution? Why is the citation count so low?

submitted 75 days ago • 4 comments

We will run out of data for pretraining and see diminishing returns. In many application domains such as in the sciences we also have to be very careful on what data we pretrain to be effective. It is important to adaptively generate new data from physical simulators. Excited about the work below

submitted 75 days ago • 0 comments

I think math books should use more descriptive titles: - Precalculus: a modern approach with a lot of calculus - An introduction to probability with only a gentle amount of gaslighting - Category Theory

submitted 75 days ago • 0 comments

I guess we have Boltz-1 to thank for pushing all these other AF3 clones to adopt fully open-source licenses and models.

submitted 75 days ago • 1 comment

Transatlantic flights suck but at least I finally got a chance to read this book by @neurograce.bsky.social

submitted 76 days ago • 1 comment

Lol

submitted 76 days ago • 2 comments

Apparently the recipient of the Best Paper award at NeurIPS 2024 did not play fairly: var-integrity-report.github.io

submitted 77 days ago • 4 comments

if one more person says that AI uses up fresh water i am going to have a fucking stroke

submitted 77 days ago • 25 comments

Sometimes you're the heron, sometimes you're the frog. We live in a society.

submitted 77 days ago • 7 comments

Every year at NeurIPS, I get a sense of where the community is headed. I'm so happy that the era of larger language models on larger datasets is coming to an end.

submitted 77 days ago • 4 comments

Baby blep! Postcard from my collection - copyright 1900, mailed 1907.

submitted 78 days ago • 209 comments

Hi NeurIPS! Explore ~4,500 NeurIPS papers in this interactive visualization: jalammar.github.io/assets/neuri... (Click on a point to see the paper on the website) Uses @cohere.com models and @lelandmcinnes.bsky.social's datamapplot/umap to help make sense of the overwhelming scale of NeurIPS.

submitted 78 days ago • 1 comment

Omg! Science Twitter is truly dead. The mass migration of all the good folks to here is something to behold—like the great wildebeest migration in the Serengeti!

submitted 78 days ago • 9 comments

Expanding on this, it's worth noting that LLMs and generative models working as well as they did absolutely caught OpenAI by surprise. Before (google's) Lamda's success they were plugging away at deep reinforcement learning. They are entirely freestyling with no idea what happens next.

submitted 78 days ago • 3 comments

I want to dispel a very commonly-held assumption: none of these companies had a plan with generative AI. They threw cash at it because it looked like it might be profitable, and have thrown more cash as it became less profitable because none of them have any other ideas, or plan of any kind.

submitted 78 days ago • 36 comments

The ML for Molecules and Materials workshop on Dec 6, 2024, brought over 700 attendees together, online and in Berlin, to explore innovations in drug discovery, protein folding, and material design through expert talks and posters. Get more details here: bit.ly/49rIey9 #ELLISPrograms

submitted 78 days ago • 0 comments

A Nebula that extends its hand into space

submitted 78 days ago • 30 comments

My cat has two and only two settings and they are: 1) awww look at the sleepy little baby - or - 2) I swear to god if this motherf...

submitted 78 days ago • 297 comments

The founder of Insitro, Daphne Koller, is at NVIDIA HQ today to present on ML in drug R&D. With 4 forces driving the inflection - 1) rise of transformer architectures 2) high volume bio & clinical data 3) advances in compute and GPUs 4) the growing community of experts blending ML and bio

submitted 79 days ago • 1 comment

submitted 79 days ago • 186 comments

Just saying.

submitted 79 days ago • 13 comments

After two years, our paper on generative models for structure-based drug design is finally out in @natcomputsci.bsky.social www.nature.com/articles/s43...

submitted 79 days ago • 2 comments

People say large language models are a watershed because, for the first time in human history, we've invented a technology we cannot control - but we haven't been able to control printers for 40 years.

submitted 81 days ago • 10 comments

Very happy to share our new paper that will appear in @unireps.bsky.social We show improvements of zero-shot model stitching through invariance to symmetries in parameter space, and topological regularization of the latent spaces more info below 👇

submitted 81 days ago • 1 comment

Fascinating paper on where 6000 global elites went to college. Billionaires, CEOs, heads of state, central bankers, etc. In a word: Harvard. Fully 10% of global elites went to Harvard. Elite US schools are over-represented (23% IvyPlus), but nobody comes close to Harvard. 🧵

submitted 82 days ago • 36 comments

Curious about all this inference-time scaling hype? Attend our NeurIPS tutorial: Beyond Decoding: Meta-Generation Algorithms for LLMs (Tue. 1:30)! We have a top-notch panelist lineup. Our website: cmu-l3.github.io/neurips2024-...

submitted 82 days ago • 1 comment

extraordinary news from Romania where its constitutional court has cancelled the deciding round of the country’s presidential election scheduled for Sunday after allegations that Russia used TikTok to promote the leading candidate www.ft.com/content/2248...

submitted 82 days ago • 21 comments

A thread of curious animals interrupting wildlife photographers 😍 1.Thread 🧵

submitted 82 days ago • 372 comments

Liking this interaction with @mmbronstein.bsky.social and Denis Danilov so much I'm reposting it here

submitted 82 days ago • 3 comments