@inproceedings{carroll2022uni,
author = {Carroll, Micah and Paradise, Orr and Lin, Jessy and Stevenson, Raluca and Sun, Mingfei and Bignell, Dave and Milani, Stephanie and Hofmann, Katja and Hausknecht, Matthew and Dragan, Anca and Devlin, Sam},
title = {Uni[MASK]: Unified Inference in Sequential Decision Problems},
booktitle = {2022 Neural Information Processing Systems},
year = {2022},
month = {November},
abstract = {Randomly masking and predicting word tokens has been a successful approach in pre-training language models for a variety of downstream tasks. In this work, we observe that the same idea also applies naturally to sequential decision-making, where many well-studied tasks like behavior cloning, offline reinforcement learning, inverse dynamics, and waypoint conditioning correspond to different sequence maskings over a sequence of states, actions, and returns. We introduce the UniMASK framework, which provides a unified way to specify models which can be trained on many different sequential decision-making tasks. We show that a single UniMASK model is often capable of carrying out many tasks with performance similar to or better than single-task models. Additionally, after fine-tuning, our UniMASK models consistently outperform comparable single-task models.},
url = {http://approjects.co.za/?big=en-us/research/publication/unimask-unified-inference-in-sequential-decision-problems/},
}