[2210.14215] In-context Reinforcement Learning with Algorithm Distillation

arxiv.org · 7,878 words · saved by 1 readers

We propose Algorithm Distillation (AD), a method for distilling reinforcement learning (RL) algorithms into neural networks by modeling their training histories with a causal sequence model. Algorithm Distillation treats learning to reinforcement learn as an across-episode sequential prediction problem. A dataset of learning histories is generated by a source RL algorithm, and then a causal transformer is trained by autoregressively predicting actions given their preceding learning histories as context. Unlike sequential policy prediction architectures that distill post-learning or expert sequences, AD is able to improve its policy entirely in-context without updating its network parameters. We demonstrate that AD can reinforcement learn in-context in a variety of environments with sparse rewards, combinatorial task structure, and pixel-based observations, and find that AD learns a more data-efficient RL algorithm than the one that generated the source data.

Preprint I N - CONTEXT R EINFORCEMENT L EARNING WITH A LGORITHM D ISTILLATION Michael Laskin∗ Luyu Wang∗ Junhyuk Oh Emilio Parisotto Stephen Spencer Richie Steigerwald DJ Strouse Steven Hansen Angelos Filos Ethan Brooks Maxime Gazeau Himanshu Sahni Satinder Singh Volodymyr…

[2210.14215] In-context Reinforcement Learning with Algorithm Distillation

related reading