David Abel

dabelcs.bsky.social

05 March 2025 14:54

Barto and Sutton receive the Turing award! Amazing recognition for two kind, pioneering, and genuinely inspiring scientists. Congratulations!! awards.acm.org/about/2024-t...

Tom Schaul

schaul.bsky.social

24 February 2025 23:35

Some extra motivation for those of you in RLC deadline mode: our line-up of keynote speakers -- as all accepted papers get a talk, they may attend yours! @rl-conference.bsky.social

Eugene Vinitsky 🍒

eugenevinitsky.bsky.social

24 December 2024 10:24

Oh, sure. I don't have an exact proof that the state-of-the-art is suboptimal but warehousing companies such as Amazon still fund significant research on MAPF problems and co-optimization of task allocation and motion. That seems like a reason to believe existing solvers are suboptimal

Eugene Vinitsky 🍒

eugenevinitsky.bsky.social

22 December 2024 23:47

Into this regime and where we don't know until we try if these solutions will work! I think it's reasonable to conjecture but I'm not seeing, as an example, most control problems or combinatorial opt problems optimized via RL. These are problems that can be simulated fast with no sim-to-real gap

Csaba Szepesvari

skiandsolve.bsky.social

11 December 2024 22:48

Wow, o1 seems to have the biggest ego of any LLM I ever tested before. It gives me incorrect solutions to simple math problems and then insists on that I misunderstand things. And it cannot be swayed! Wow, just wow.