Decision transformer

tags

Transformers, GPT, Reinforcement learning

paper

(Chen et al. 2021)

Architecture

This is a decoder model that uses a GPT-like model to encode and predict trajectories for Reinforcement learning tasks. It has essentially the same characteristics as GPT.

Bibliography

1. Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch. June 24, 2021. "Decision Transformer: Reinforcement Learning via Sequence Modeling". arXiv. DOI.