Got through my first chess lesson in many years using chess dot com

Spontaneously, I would love to thank Dr. Lanctot for making me believe in multi-agnency as something more than just a straightforward combinatorial combination of the learning agents. I still really think about it as a good research direction. However, not many people think the same, unfortunately.

spreading for my 179 followers maybe one of you can fly to Copenhagen and enjoy the European scientific discussion there

Good!

Google added torch frontend for jax. Hopefully it'll ease simultaneous support of both frameworks google.github.io/torchax/

Collaboration with Dynamic Open Ad Hoc Team via Team State Modelling Jing Sun, Cong Zhang, Zhiguang Cao Action editor: Quanming Yao https://openreview.net/forum?id=BukMU42P3G #teammates #teammate #teamwork

picture

Interesting idea to exponentially scale the fog-of-war chess to multi-player setting by Alexandra Botez store.steampowered.com/app/3202290/... Kaggle Arena competitor?

Interesting thing I have known since this morning. I can't create individual plans or work for almost any applications with marl because of sanctions. I see. Good to know, at least. And it's not stated almost anywhere.

at least I am trying

Rapid RL experimentation is great. But how do you catch silent errors before they slip by? In this post, I share tools and habits that help me move quickly from idea to result without sacrificing reliability.

but isn't it just like 200 euro

I have recently self-reflected on my huge library in YouTube (google) music that had been up since 2013-14. What might be motivation/inspiration/hesitation to music taste of yours? Mine was radio.

underrated research field right there: counter-strike has one of the biggest gaming communities on the planet with plenty of data stored a recorded since 2007 as david also said, it has the own dlc (callouts) -- it's a great sandbox for imitation of human behaviour and its skillwise diversity

I guess, I should challenge myself a bit neuroevolutionbook.com/code-exercises

opponent shaping in LLMs by @mircomusolesi.bsky.social 's people arxiv.org/abs/2510.08255

FDG'26 call for papers! We will be hosting the ACM conference Foundations of Digital Games 2026 conference here at the Royal Danish Academy in Copenhagen August 10-13 2026. The call for papers is out now: fdg2026.org/call-for-pap... We hope to see you (and your paper) at FDG’26!

Mjlab Isaac Lab API, powered by MuJoCo-Warp, for RL and robotics research. github.com/mujocolab/mj...

I really struggle with writing papers. Working alone, it's very hard to fully outline the expectations and not anxiously doubt myself, jumping from the ideas draft to the code and back. Constantly failing my own approaching deadlines is disappointing, especially compared to the other students.

Samples of generated code from the appendix look so curious not to check the paper out

Today, I have read today and remember what it's like to feel like a secondhand person

an approach worth looking at github.com/McGill-NLP/t...

π-thon

gasp

For all the details, please give the paper a read! Paper: arxiv.org/abs/2507.06466 Infinite thanks to @jeffclune.com and @cong-ml.bsky.social for all their guidance!

A nice idea by Valve corp. to use play testing as a method to improve design (content generation) of their games. It matches very well with the modern quality-diversity, reinforcement learning and learning to vote (get quantitive feedback) youtu.be/9Yomqk0C6kE?...

Evolution of trust. Using game theory in multi agent systems by Elias Fernández

At #EWRL, we presented 4 papers, which we summarize below. - A Theoretical Justification for AsymAC Algorithms. - Informed AsymAC: Theoretical Insights and Open Questions. - Behind the Myth of Exploration in Policy Gradients. - Off-Policy MaxEntRL with Future State-Action Visitation Measures.

wow, really cool

Are there any #ELLISPhD scholars who do their research in the direction of multi-agent systems? If so, what kind of research are you fond of doing? Asking for a friend.

New paper challenges how we think about Theory of Mind. What if we model others as executing simple behavioral scripts rather than reasoning about complex mental states? Our algorithm, ROTE (Representing Others' Trajectories as Executables), treats behavior prediction as program synthesis.

startup culture may actually kill science or elevate the progress to the different level 2020s can be a breaking point for our society

In machine learning, do you need to know any optimization algorithm other than stochastic gradient descent? A reluctant but best-faith argument for no.

I need my own Indian-peeled AI collaboration bot

Sam Ganzfried Quadratic Programming Approach for Nash Equilibrium Computation in Multiplayer Imperfect-Information Games https://arxiv.org/abs/2509.25618

Being honest, social networks bring me emotional swings. I dont really like talking with people irl, and I have recently noticed that using social media affects my mental health in a negative way.

good that there's a public release, maybe'll try with chess models this weekend github.com/google/tunix...

A cool approach for the preference learning in the rlhf based on Nash solving: arxiv.org/abs/2509.23102

Probabilistic models (diffusion, normalised flows) are a way to train models of the transition dynamics. Would be interesting to understand how value models can(not) be described with these models. This can help to uncover how to use language models for planning. Maybe some toy experiments'll do.

I usually repost to bookmark but now do it to endorse. The multiagent group at dm is really responsible for the most crucial advancements in multi-agent learning for the last decade, I would say.

www.julian.ac/blog/2025/09... I hope, we will achieve agi in 2026, and it'll just kill us all as irrational beings are these people contractually obligated to publish this sort of sloppy and slippery posts

New #Reproducibility Certification: [Re] Cooperate or Collapse: Emergence of Sustainable Cooperation in a Society of LLM Agents Oliver van Erven, Konstantinos Zafeirakis, Jacobus Smit, Julio Smidi, Luc Buijs https://openreview.net/forum?id=EWWxSkUchO #cooperation #agents #abstraction

Cool job posting! These people do great work and maybe even single-handedly drag the whole field. Apply !!!

libtorch compilation experience has been ... awful

Watch the stories of the eternal @michaelhbowling.bsky.social about the atari learning environment www.youtube.com/watch?v=fP_B...