@sharky6000.bsky.social why github is kind of up and down, you might be interested to take a look

I am already thinking about awesome meta-game solving papers repository to track the progress.....

Sure, but while IL can work with IID data, RL -- I am not sure about that. That's why we need special assumptions of the data, which you highlighted in your post. It satisfies my initial question, xD

what about github repo/blog where we read some papers, by hand, and try to understand their role in context of modern science?

for imitation learning (same as behavioural clonning), I guess, data has to be as less interdependent as possible (the Markov assumption isn't really the requirement). More or less, the same with other Monte-Carlo methods :) A question is why temporal dependence might be pivotal to exploit the data

something about Markovian property of the RL data

I have updated my tutorial on making Vision Language Action models. This tutorial starts with a basic Transformer and walks people through the steps to transform it into a full VLA that uses PaliGemma as the pretrained VLM. Links below.

what about temporal dependenices in data

aamas papers started to flood arxiv, good.

Sample-Efficient Policy Space Response Oracles with Joint Experience Best Response Multi-agent reinforcement learning (MARL) offers a scalable alternative to exact game-theoretic analysis but suffers from non-stationarity and the need to maintain diverse populations of strategies th...

PSRO with JOINT (!!!) experience best response

arxiv.org/abs/2602.06599

interesting that you paste equation to "latex" it instead of just doing it in any markdown environment

that's why ram costs 1337$ per stick, I guess

I wrote about how I don’t know math but still am somehow a successful computer scientist. I have strong feelings about this. But I also want to understand.
togelius.blogspot.com/2026/02/math...

GitHub - ash80/diffusion-gpt: From baby GPT to diffusion GPT: An annotated implementation of a character-level discrete diffusion model (adapted from Karpathy’s baby GPT). From baby GPT to diffusion GPT: An annotated implementation of a character-level discrete diffusion model (adapted from Karpathy’s baby GPT). - ash80/diffusion-gpt

github.com/ash80/diffus...

e

The most important finding from this analysis! See the post for more details

insulting AI?

I enjoyed the book after I took decision theory during my bachelor's degree and took this coursera course: www.coursera.org/learn/narrat...

I am bad at economics but least I've got a high level overview of what Dr. Lancot is describing in the thread, partially motivating me to model interactions...

TABX: A High-Throughput Sandbox Battle Simulator for Multi-Agent Reinforcement Learning The design of environments plays a critical role in shaping the development and evaluation of cooperative multi-agent reinforcement learning (MARL) algorithms. While existing benchmarks highlight crit...

Looks goofy but arxiv.org/abs/2602.01665

#tutorial

I found myself not having time and — mainly — desire to read. Quite sad and don't know what to do.

In-context learning is the most affordable to the general public way to do and understand meta-learning and why formalisation of it was genius.

What a great week at the @aliceworkshop.org (Artificial Life, Intelligence, Complexity & Evolution) in Copenhagen.

Our multidisciplinary group worked intensely for the whole week and we got the 2nd prize!!

Thanks to the amazing organizers from the REAL lab and the jury.

Balatro - Wikipedia

@sharky6000.bsky.social might be interested!

balatro is like poker but rogue-like

en.wikipedia.org/wiki/Balatro

A cool benchmark
balatrobench.com

Talks from the World Models Workshop, happening at MILA in Montreal!

@sharky6000.bsky.social can one submit a game to play with LLMs, one like it should be an org with the results and resources?

I have still been teased with civ/wow/colonisation like things...

Symphony-Coord: Emergent Coordination in Decentralized Agent Systems Multi-agent large language model systems can tackle complex multi-step tasks by decomposing work and coordinating specialized behaviors. However, current coordination mechanisms typically rely on stat...

I am not sure in UCB, in particular, but I like the perspective of this paper arxiv.org/abs/2602.00966

I've the same thing on my mind, maybe we can unify it.

The 2026 IFAAMAS Influential Paper Award Committee has selected two winners for this year’s award.

🔹Book Award
Rules of Encounter by Jeffrey S. Rosenschein & Gilad Zlotkin

🔹Collection of Papers Award
Influential works by Amy Greenwald, Keith Hall, @Junling Hu, Michael Wellman, and Amir Jafari.

Decoding Life on Earth | Google and the Earth Biogenome Project Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

It once took 13 years and $3 billion to sequence the human genome.

Now, we're using Google’s AI tools to sequence animal genomes in just days to help save endangered species 🧬 (1/4) ↓

goo.gle/4kgx18P

during this crazy session of the alyssa workshop, we found out that RL is not that deep...

he definitely enjoys the process

Paper page - TTCS: Test-Time Curriculum Synthesis for Self-Evolving Join the discussion on this paper page

I like huggingface.co/papers/2601....