Jacob Beck and Risto Vuorio
TalkRL: The Reinforcement Learning Podcast
00:00
The Barriers to Realistic High-Dimensional Task Distribution
Three questions from zoha rimon, a researcher at technion. What do you think are the barriers we'll need to tackle to make meta rl work on realistic high-dimensional task distributions? We don't quite have like a good training task distribution where we would train the meta rl agent that would then generalize to the other tasks so there's been efforts in this direction right and now deepmind has their uh xland i think it's called "Xland"
Transcript
Play full episode
