🤖 check out @chongyiz1 ‘s new work on a new unsupervised pre-training method for RL!
Representation learning is all about capturing the right prior. What is the right prior for *reinforcement learning*?View quoted post on X
We propose a new unsupervised pre-training method for RL: chongyi-zheng.github.io/onestep-fb.
🧵⬇️