Ishan Durugkar

Tulips in my garden are past their prime now 😞
Can totally see why people wanted to invest in these drops of sunshine though

Very creative!

Nuance?! Oh no!

Hey Pablo! Is this role for the Summer? I can point some folks to this great opportunity

Top notch. "discourse about large models as intelligent agents is fundamentally misconceived. ... Large Models should not be viewed primarily as intelligent agents, but as a new kind of cultural and social technology, allowing humans to take advantage of information other humans have accumulated."

Oof! I can imagine. Eternally grateful that I'm working from home right now and can just take a break to go hang with the baby whenever

Very interesting reframing of the perspective through which we should look at the impacts of LLMs and other large models. To me, it is intuitively more satisfying than the debate around whether AGI is around the corner.

This thread presents a very different viewpoint compared to my intuitions and the debate I usually see online. Looking forward to diving in!

Very cool application of MARL! I'm excited to see more systems like this in the wild.

*Please repost* @sjgreenwood.bsky.social and I just launched a new personalized feed (*please pin*) that we hope will become a "must use" for #academicsky. The feed shows posts about papers filtered by *your* follower network. It's become my default Bluesky experience bsky.app/profile/pape...

Oof, yes. Agree with most of it too. I worry we're staring at another AI winter 😞

Andrew Barto and Richard Sutton are the recipients of the 2024 ACM A.M. Turing Award for developing the conceptual and algorithmic foundations of reinforcement learning. Andrew Barto and Richard Sutton are the recipients of the 2024 ACM A.M. Turing Award for developing the conceptual and algorithmic foundations of reinforcement learning. In a series of papers beginnin...

I got hooked on RL at UMass, where Andy and Rich started their collaboration. Really happy to hear that they are this year's Turing Award winners!

www.acm.org/media-center...

Good to know!

Whoa! I should try!

The first part of my suggestion was kinda serious

Maybe we should add some sort of comment section to ArXiv

Then we can have some experienced researchers who are not affiliated with the authors decide which papers have crossed the threshold to be classified as finalized papers.

Many people have asked me about the France Action Summit.

I think a summit is typically most valuable as a catalyst, not as a solution in itself.

But, will share some observations.

Now we need to train a Minecraft agent that can play Atari games in there

🚨🚨 RLC deadline has been extended by a week! Abstract deadline is Feb. 21 with a paper deadline of Feb. 28 🚨🚨. Please spread the word!

Oh no

Haha it's so hard to keep these exciting results to yourself until you can talk about them properly! Looks like it was totally worth it though

Whoa! Very cool stuff!

grad student descent is passe. Time for LLM descent!

Welcome to the other side of ICML dear bleary-eyed RL researchers. Time to get ready for the RLC deadline, Feb. 14 AOE! So excited for your papers.
See the call for more info: rl-conference.cc/callforpaper...

The LLM is training us to give it a dense signal

It seems to make the problem of exploration more tractable. The agent doesn't have to explore everything, it just needs to explore efficiently in the neighborhood of the base policy.

So maybe it's actually removing some of the hardest challenges from general RL?

Hello world!