My husband presenting his work on caregiving 😍

Excited to release our latest paper on a new multi-turn RL objective for training LLMs to *learn how to learn* to adapt to the user. This enables it to adapt and personalize to novel users, whereas the multi-turn RLHF baseline fails to generalize effectively to new users.

In our latest paper, we discovered a surprising result: training LLMs with self-play reinforcement learning on zero-sum games (like poker) significantly improves performance on math and reasoning benchmarks, zero-shot. Whaaat? How does this work?

Just posted a talk I gave about this work! youtu.be/mxWJ9k2XKbk

RLHF is the main technique for ensuring LLM safety, but it provides no guarantees that they won’t say something harmful. Instead, we use online adversarial training to achieve theoretical safety guarantees and substantial empirical safety improvements over RLHF, without sacrificing capabilities.

RLHF is the main technique for ensuring LLM safety, but it provides no guarantees that they won’t say something harmful. Instead, we use online adversarial training to achieve theoretical safety guarantees and substantial empirical safety improvements over RLHF, without sacrificing capabilities.

Oral @icmlconf.bsky.social !!! Can't wait to share our work and hear the community's thoughts on it, should be a fun talk! Can't thank my collaborators enough: @cogscikid.bsky.social y.social @liangyanchenggg @simon-du.bsky.social @maxkw.bsky.social @natashajaques.bsky.social

At @rldmdublin2025.bsky.social this week? Check out our social learning workshop from @amritalamba.bsky.social and I tomorrow! Inc. talks from @natashajaques.bsky.social, @nitalon.bsky.social, @carocharp.bsky.social, @kartikchandra.bsky.social & more! Full schedule: sites.google.com/view/rldm202...

1/4 Join us and the Autotelic Interaction Research (AIR) group @aalto.fi / Finland to work on Computational Social Intrinsic Motivation (SIM) as PhD (4y) or postdoc (2y). Job ad w project description and application instructions: bit.ly/4jyNLGv. We're looking forward to learning about you!

Human-AI cooperation is important, but existing work trains on the same 5 Overcooked layouts, creating brittle strategies. Instead, we find that training on billions of procedurally generated tasks trains agents to learn general cooperative norms that transfer to humans... like avoiding collision

Our new paper (first one of my PhD!) on cooperative AI reveals a surprising insight: Environment Diversity > Partner Diversity. Agents trained in self-play across many environments learn cooperative norms that transfer to humans on novel tasks. shorturl.at/fqsNN%F0%9F%...

Got a weird combination of mail today.

Recorded a recent "talk" / rant about RL fine-tuning of LLMs for a guest lecture in Stanford CSE234: youtube.com/watch?v=NTSY.... Covers some of my lab's recent work on personalized RLHF, as well as some mild Schmidhubering about my own early contributions to this space

next Canadian government should think of boosting research funding up here and trying to grab as many American postdocs and researchers as possible

AI has shown great potential in boosting efficiency. But can it help human society make better decisions as a whole? 🤔 In this project, using MARL, we explore this by studying the impact of an ESG disclosure mandate—a highly controversial policy. (1/6)

Our latest work uses multi-agent reinforcement learning to model corporate investment in climate change mitigation as a social dilemma. We create a new benchmark, and show that corporations are greedily motivated to pollute without mitigating their emissions, but if all companies defect...

The paper I spoke about at the #NeurIPS2024 ML for Systems workshop is now on arxiv arxiv.org/abs/2412.15573! We use multi-agent RL to solve a classic combinatorial optimization problem (the Sequential Assignment Problem), by combining MARL with a classical polynomial time assignment algorithm. 1/3

Repost for updating my favourite deep learning meme of all time

Astonishing how many RL bottlenecks are resolved simply by “make simulator go fast”. What if we had prioritized engineering over algorithms years ago?

University of Washington researchers craft method of fine-tuning AI chatbots for individual taste www.geekwire.com/2024/univers...

UW News put out a Q&A about our recent work on Variational Preference Learning, a technique for personalizing Reinforcement Learning from Human Feedback (RLHF) washington.edu/news/2024/12...

Really excited to present my work this Sunday @NeurIPS on how we might approach training a generalist agent capable of cooperation at scale: coordinating with many novel partners on many novel tasks has never been easier! Come by the IMOL workshop to check it out and chat more!

Even though the Social RL lab only got started ~1 year ago, I’m super excited to announce that we have 10 people from the lab presenting their work at #NeurIPS2024. Delighted to officially introduce our lab: socialrl.cs.washington.edu! Thread with all our NeurIPS work below 👇

Let's cycle through the memes for this one until it stops... 😇😅🙏

Is it Bad to leave Twitter? No. Here are 7+ years of insights from my lab’s research that explain why. Featuring work w/ @williambrady.bsky.social @killianmcloughlin.bsky.social 🧵

Better VQ-VAEs with this one weird rotation trick! I missed this when it came out, but I love papers like this: a simple change to an already powerful technique, that significantly improves results without introducing complexity or hyperparameters.

Bayesian Fashion

The Reality for AI Startups

New relationship sin just dropped: weaponized nonchalance. When someone insists they just don't care about something, forcing you to handle it. "I guess the mess just doesn't bother me!"

Poem by Joseph Fasano

Before the election, twitter turned into a non-stop propaganda machine. Now that the election is over, it has reverted back to recommending ML to me mostly. I hated that experience and never want to be at the whims of someone like that again

NeurIPS Test of Time Awards: Generative Adversarial Nets Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, Yoshua Bengio Sequence to Sequence Learning with Neural Networks Ilya Sutskever, Oriol Vinyals, Quoc V. Le

Amazing blog post on flow matching, stunning visuals! It also makes the connection with normalising flows crystal clear. Incredible effort!

Alibaba has their own version on GPT-o1. This might be the best description of “o1-type”systems so far arxiv.org/abs/2411.14405

Beyond excited to introduce AlphaQubit, now published in Nature! AlphaQubit is a neural network for quantum error correction and achieves state-of-the-art accuracy on simulated and real-world data.

So I tried to post on X about how I'm moving to BlueSky and why. The tweet got less than 10% of the views of my 5 most recent tweets. Think the algorithm is suppressing this information? Once again, so much for a "free and open town square".

Now that @jeffclune.bsky.social and @joelbot3000.bsky.social are here, time for an Open-Endedness starter pack. go.bsky.app/MdVxrtD

Very expensive and mostly confirms established theories? /jk jk

Re: the scale is dead debate. Isn't it pretty obvious that just scaling is never going to work if your method breaks down on OOD inputs? The world is non-stationary, so it's constantly presenting new OOD inputs.

This chart is pretty insane. More of an overnight social/political-type shift than typical tech adoption imo: techcrunch.com/2024/11/19/b...

One of my first posts on twitter was "fuck twitter". I'd just like to reiterate that sentiment today, as I join bluesky

Thanks for the invite to get on here @eugenevinitsky.bsky.social and @sharky6000.bsky.social. Excited to not see a post from Elon as every other item in my feed. Maybe I'll actually get bold and start posting more of my real thoughts, since only the cool people seem to be on here so far :)

Lets get the multi-agent learning community started up here: go.bsky.app/9gsefkW