Greetings from Rio, killing some time before #ICLR2026 @iclr-conf.bsky.social

Getting ready to fly to Rio. 😎🇧🇷

Why is the default option in the rebuttal acknowledgements at ICML to accept the paper? I'm very confused on how to use these buttons. Can someone explain them to me?

That is even if the baseline reported 80% on a similar task

Neural networks are highly non-convex, so approximate error minimizers need not look anything like each other in parameter space. But we show that nevertheless (for many model sizes) approximate error minimizers must closely agree in function/prediction space despite this!

It's this time of the year again: your baselines cannot be PPO and SAC.

How AI is changing the nature of mathematical research What machine learning theorists learned using AI agents to generate proofs — and what comes next.

Michael @mkearnsphilly.bsky.social ) and I wrote a blog post about our experiences using AI for research, and our thoughts on what these developments will mean for research, publication, and education: www.amazon.science/blog/how-ai-...

I'm so glad that so many research problems are finally being treated as first class citizens rather than afterthoughts. 🤔

Too many papers sound like this

Hierarchical Context-Aware Diffusion-Transformer Meta-World-Model Reinforcement Learning with Causally Disentangled Preference-Aligned Self-Supervised Compositional Multi-Scale Latent Skill Priors for Long-Horizon Generalist Decision Making

One reason I work on replicable and consistent RL is because it is has always been at the top of the list of criteria for reliability.

Excuse me? Surely telling me that didn't require much thinking.

I have seen multiple times now that a reviewer said sth like: the proofs are simple -> reject the paper. That is completely counter-productive. A theorem needs to generate new insights. If we learn something new from something simple that should be preferred. Don't believe me? Ask someone famous:

Why are you doing this to me

I think it's relatively simple. One side already has the job they want and the other side needs citations to get that job. And everyone tells me advertising work is how you get citations. I have been tempted to go back because on bsky, interactions have become fewer and fewer. Not going to though...

Yet somehow every now and then a paper becomes very popular even though its findings are similar to those of many others. This paper gets cited while the others don't. Was it just luck?

While I agree with the sentiment let me play devil's advocate. You only get invited to give talks if your work is already well known. Conferences have become too large to even find relevant people. And social media posts are only marginally relevant if you don't already have a large following.

I understand that that is okay but at some point it honestly becomes disheartening if you constantly have to reach out to people. There are others who don't seem to have to; what are they doing differently?

I’m trying to understand whether this is mostly about keyword mismatch, venue visibility, social media, etc.
For example, when I search terms like “high update ratio RL” on Scholar, our papers show up near the top.

scholar.google.com/scholar?hl=e...

Where are things going wrong?

I’ve been thinking about a practical question and would love some opinions:

How do your papers actually get discovered/cited?

I was searching for recent work on high update ratio RL and found several very closely related papers tackling the same failure modes we study. None cited our earlier work.

🚀 Excited to share REPPO, a new on-policy RL agent!

TL;DR: Replace PPO with REPPO for fewer hyperparameter headaches and more robust training.

REPPO, led by @cvoelcker.bsky.social, will be presented at ICLR 2026. How does it work? 🧵👇

Scaling Laws in Particle Physics Data! This is a result I've been itching to share and it's finally out. One of the big open questions is how much better AI-based methods at particle colliders can still become. 1/4

Not a particle physics person but extremely curious, can you elaborate what we might hope to learn from these models in the future? What physics might we discover using them?

"Scientific reviewers should have experience publishing scientific work in related areas" is really not that hot of a take.

Clicking like on any relevant ICLR paper. Encourage people to post their work here more!

How do I see this?

The other paper accepted to @iclr-conf.bsky.social 2026 🇧🇷. Our work on replicable RL sheds some light on how to consistently make decisions in RL.

@ericeaton.bsky.social @mkearnsphilly.bsky.social @aaroth.bsky.social @sikatasengupta.bsky.social @optimistsinc.bsky.social

Two papers accepted to @iclr-conf.bsky.social 2026! One of the is REPPO, see below! I think it deserves a lot more recognition. Let's chat about it in Rio! 🇧🇷

Quite disheartening that there isn't a single workshop at ICLR to present my RL work but there several topics that are listed 5 or 6 times just named differently.

That's correct, we did make it bold

Our number went down by 0.01 but it's very expensive to run so we can't have error bars. Our algorithm is so much better than the rest, new SOTA!