Irfan Essa

VER: Scaling On-Policy RL Leads to the Emergence of Navigation in Embodied Rearrangement

2022 · arXiv (Cornell University)

We present Variable Experience Rollout (VER), a technique for efficiently scaling batched on-policy reinforcement learning in heterogenous environments (where different environments take vastly different times to generate rollouts) to many GPUs residing on, potentially, many …