iiinova — where your digital self keeps evolving

Reposted by Natasha Jaques

Allen School @uwcse.bsky.social · 23/04/2026

#HuskyGivingDay is here! Gifts of any amount help unlock up to $10K more for #UWAllen priorities like undergraduate scholarships, which ensures that an Allen School education remains within reach of students regardless of their means. Let's go! bit.ly/hgd-allen-sc... #PoweredByYou #HuskyExperience

Portrait of Benjamin Moskalensky (B.S., ‘25) with quote: “Being able to complete this experience of going to college, getting a high-level education, providing back to the research community, getting involved in my own startup — all these things were made possible.”

273

Natasha Jaques @natashajaques.bsky.social · 03/10/2025

Instead of behavior cloning, what if you asked an LLM to write code to describe how an agent was acting, and used this to predict their future behavior? Our new paper "Modeling Others' Minds as Code" shows this outperforms BC by 2x, and reaches human-level performance in predicting human behavior.

1132

Natasha Jaques @natashajaques.bsky.social · 01/08/2025

My husband presenting his work on caregiving 😍

0120

Natasha Jaques @natashajaques.bsky.social · 09/07/2025

By optimizing for intrinsic curiosity, the LLM learns how to ask a series of questions over the course of the conversation to improve the accuracy of its user model. This generates conversations which reveal significantly more information about the user.

030

Natasha Jaques @natashajaques.bsky.social · 09/07/2025

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.

2123

Natasha Jaques @natashajaques.bsky.social · 01/07/2025

This work shows the benefit of RL training for improving reasoning skills when there is no possibility for data leakage. AND how continuously evolving multi-agent competition leads to the development of emergent skills that generalize to novel tasks.

090

Natasha Jaques @natashajaques.bsky.social · 01/07/2025

We analyze the results and find that LLMs learn emergent reasoning patterns like case-by-case analysis and expected value calculation that transfer to improve performance on math questions.

170

Natasha Jaques @natashajaques.bsky.social · 01/07/2025

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?

2597

Reposted by Natasha Jaques

Natasha Jaques @natashajaques.bsky.social · 12/06/2025

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.

1163

Natasha Jaques @natashajaques.bsky.social · 12/06/2025

1163

Reposted by Natasha Jaques

Kunal Jha @kjha02.bsky.social · 09/06/2025

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

0102

Reposted by Natasha Jaques

Christian Guckelsberger @creativeendvs.bsky.social · 03/06/2025

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!

bit.ly

Doctoral Researcher and Postdoc positions to work on Computational Social Intrinsic Motivation (SIM) | Aalto University

The Autotelic Interaction Research (AIR) group at the Dept. of Computer Science, Aalto University, Finland is looking for 1 Doctoral Researcher (2+2 years) and 1 Postdoc (2 years) to work on Computational Social Intrinsic Motivation (SIM)

194

Natasha Jaques @natashajaques.bsky.social · 19/04/2025

Way to go KJ for producing such an insightful paper in the first few months of your PhD!

010

Natasha Jaques @natashajaques.bsky.social · 19/04/2025

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

1164

Reposted by Natasha Jaques

Kunal Jha @kjha02.bsky.social · 19/04/2025

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%...

1267

Natasha Jaques @natashajaques.bsky.social · 12/04/2025

Got a weird combination of mail today.

080

Natasha Jaques @natashajaques.bsky.social · 28/03/2025

I had a ton of fun using this as a kid. I actually made my high school English class project a giant hypercard-based video game where I drew each frame in Paint and hid buttons behind the hand-drawn elements that let you navigate the world. That was so fun...😍

020

Natasha Jaques @natashajaques.bsky.social · 27/03/2025

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

youtube.com

Reinforcement Learning (RL) for LLMs

YouTube video by Natasha Jaques

55110

Reposted by Natasha Jaques

Sharon 🪳🌹 @sharonk.bsky.social · 12/03/2025

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

606165392173

Natasha Jaques @natashajaques.bsky.social · 16/02/2025

Yes! Or you could focus on developing better MARL algorithms for the corporations that let them cooperate to solve the social dilemma more effectively. Similar to MARL benchmarks like Melting Pot but for a more impactful domain

010

Reposted by Natasha Jaques

xiaoxuanh.bsky.social @xiaoxuanh.bsky.social · 13/02/2025

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)

121

Natasha Jaques @natashajaques.bsky.social · 13/02/2025

In contrast, MARL enables testing new policies with many more agents over a long time horizon. We hope this benchmark will enable researchers in the RL and MARL communities to develop sophisticated cooperation algorithms in the context of a societally impactful problem!

040

Natasha Jaques @natashajaques.bsky.social · 13/02/2025

I’m really excited about this, as I think MARL provides a new tool in the toolbox for investigating this problem. Existing work on ESG disclosures focuses on empirical studies (can’t test counterfactual policies), or analytical economics models (limited to 2 players or short time intervals)

141

Natasha Jaques @natashajaques.bsky.social · 13/02/2025

...providing corporations with more reliable information about climate risks — and we show that this significantly improves corporations’ ability to mitigate climate change, even without the influence of investors!

130

Natasha Jaques @natashajaques.bsky.social · 13/02/2025

The parameters of the environment are carefully benchmarked to real-world data (e.g. from IPCC reports), and many of our experiments reveal findings consistent with empirical work on climate change and ESG disclosures. But we can also test new policy interventions, such as..

130

Natasha Jaques @natashajaques.bsky.social · 13/02/2025

overall climate risks increase, leading to lower market wealth. By introducing climate-conscious investor agents to simulate the possible effects of ESG disclosures, we show that investors can actually incentivize profit-motivated companies to invest in climate mitigation.

141

Natasha Jaques @natashajaques.bsky.social · 13/02/2025

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...

2355

Natasha Jaques @natashajaques.bsky.social · 02/01/2025

Impressive results that strongly outperform the best classical techniques. Several MARL methods work well, showing the promise of MARL for combinatorial optimization and systems problems, as in our earlier work on MARL for Microprocessor Design arxiv.org/abs/2211.16385 3/3

0120

Natasha Jaques @natashajaques.bsky.social · 02/01/2025

The classical solver can give an optimal, conflict-free assignment for a single timestep if it's given known utility values, but can't plan assignments over time. So, we learn the future estimated payoff for each single-timestep assignment using RL! 2/3

1121

Natasha Jaques @natashajaques.bsky.social · 02/01/2025

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

19617

Natasha Jaques @natashajaques.bsky.social · 21/12/2024

Repost for updating my favourite deep learning meme of all time

1131

Reposted by Natasha Jaques

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 21/12/2024

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

8826

Natasha Jaques @natashajaques.bsky.social · 18/12/2024

This work was also recently awarded the #3 Best Paper award in the Pluralistic Alignment workshop at NeurIPS! Congrats @sriyash.bsky.social @yanmingwan.bsky.social @hamishivi.bsky.social @abhishekunique7.bsky.social

060

Reposted by Natasha Jaques

Kunal Jha @kjha02.bsky.social · 12/12/2024

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!

0113

Natasha Jaques @natashajaques.bsky.social · 11/12/2024

On Sunday in the Open World Agents workshop, we will have Yanming Wan @yanmingwan.bsky.social presenting his work on following ambiguous natural language instructions with FISER sites.google.com/view/fiser-h..., and Eric Ye presenting his paper on multi-agent CraftAx (East Building MTG 1-3 + S.FOY)

sites.google.com

FISER

Human's natural language instruction is inherently ambiguous. Standard language grounding and planning methods fail to resolve ambiguity. We propose FISER, which explicitly reasons about human's inter...

130

Natasha Jaques @natashajaques.bsky.social · 11/12/2024

Also on Sunday, Kunal Jha @kjha02.bsky.social l will be presenting his recent work InfiniteKitchen: Cross-environment Cooperation for Zero-shot Multi-agent Coordination at the Intrinsically Motivated Open-ended Learning workshop imol-workshop.github.io

imol-workshop.github.io

IMOL@NeurIPS 2024

Intrinsically Motivated Open-ended Learning NeurIPS 2024 in-person Workshop, December 15, Vancouver. imol.workshop@gmail.com. Description How do humans develop broad and flexible repertoires of knowle...

161

Natasha Jaques @natashajaques.bsky.social · 11/12/2024

On Sunday, I will be giving a talk about our recent work and participating in the panel at the Open World Agents workshop sites.google.com/view/open-wo.... I will also be speaking and on the panel at the ML for Systems workshop mlforsystems.org!

sites.google.com

Open-World Agents

News 11/28/1014 - We've moved! The new location of OWA-2024 is East Building - MTG 1-3 + S.FOY (the city side of the 2nd floor of the East Building). Here is a map. 10/29/2024 - We've updated the work...

130

Natasha Jaques @natashajaques.bsky.social · 11/12/2024

On Saturday we have Michael Li presenting his work Genetic Curriculum Learning for Distribution Generalization on the Travelling Salesman Problem at the Math.AI workshop mathai2024.github.io

mathai2024.github.io

MATH-AI

The 4th Workshop on Mathematical Reasoning and AI

120

Natasha Jaques @natashajaques.bsky.social · 11/12/2024

On Thursday we have Sriyash Poddar @sriyash.bsky.social and Yanming Wan @yanmingwan.bsky.social presenting their spotlight talk on VPL nips.cc/virtual/2024..., a personalized RLHF method for pluralistic alignment (West Ballroom A-D #7102, 4:30-7:30pm). More info here: weirdlabuw.github.io/vpl/

nips.cc

NeurIPS Poster Personalizing Reinforcement Learning from Human Feedback with Variational Preference LearningNeurIPS 2024

120

Natasha Jaques @natashajaques.bsky.social · 11/12/2024

Today (Wednesday) we have Yancheng Liang and Daphne Chen presenting GAMMA nips.cc/virtual/2024..., a generative model that improves zero-shot human-AI cooperation (West Ballroom A-D #6510, 4:30-7:30pm). You can play with their models with the live demo here: sites.google.com/view/human-a...

nips.cc

NeurIPS Poster Learning to Cooperate with Humans using Generative AgentsNeurIPS 2024

130

Natasha Jaques @natashajaques.bsky.social · 11/12/2024

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 👇

socialrl.cs.washington.edu

SocialRL Lab

We are the Social Reinforcement Learning Lab at the University of Washington.

15210

Reposted by Natasha Jaques

Marc Lanctot @sharky6000.bsky.social · 02/12/2024

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

1386

Reposted by Natasha Jaques

M.J. Crockett @mjcrockett.bsky.social · 01/12/2024

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 🧵

691435540

Reposted by Natasha Jaques

Sander Dieleman @sedielem.bsky.social · 02/12/2024

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.

18613