There's no reason to gatekeep Reinforcement Learning...
RL is a powerful enough of a paradigm to admit a diversity of approaches. It is as much a problem statement as it is a solution approach. Problems arise when you elevate one over another.
Incredibly exciting work! https://t.co/aJEGtpF9eL. Cooperative, safe, driving can arise at scale from self-play training without human data, just as my group saw in Overcooked a few years ago. Caveat: in simulation. Bravo @EugeneVinitsky and coauthors!
Hello internet!
We’re still glowing in the joy that was being together at @RL_Conference earlier this month.
However, as promised, we’ve published the slides presented by each of our speakers on the workshop website
Had an amazing time presenting Fast TRAC at @RLFrameWorkshop and @ICBINB_RLC this weekend. Felt very lucky to meet a lot of the RL researchers I look up to!
This talk was fantastic among all the other fantastic talks we had @ICBINB_RLC! It’s safe to say that @audurand is a superstar. 🤩
We’ve been getting a lot of requests to have the speakers share their slides, we’re should have them up on our website soon. Stay tuned!
@audurand was one of my favourite talks at RLC because of her incredible commitment to get our methods to actually work! This is a huge undervalued problem in the community and everybody who pushes from abstract theory to concrete practice deserves a shoutout!
As always it was a delight to reconnect with friends and colleagues as well as meet new people (esp. those I’d only interacted with virtually!) with deep gratitude for my fellow organizers @RLFrameWorkshop and @ICBINB_RLC, I believe that we helped set the tone for @RL_Conference.
Thanks to everyone who came to our presentation and poster session at @ICBINB_RLC workshop! Had nice discussions about RL, Model Selection and hyperparameter tuning.
This talk facilitated a fantastic discussion that motivated going to the whiteboard (see the last picture) to drive some points home. 👏@wwdabney
You had me like this the whole time
We’re very much looking forward to seeing as many of you as possible during today’s workshops. We’ll be in Room 803 in Campus Center @UMassAmherst, it’s a cozy room which should promote some great discussion!