Perceiver IO: A General Architecture for Structured Inputs & Outputs
Authors:
Andrew Jaegle,
Sebastian Borgeaud,
Jean-Baptiste Alayrac,
Carl Doersch,
Catalin Ionescu,
David Ding,
Skanda Koppula,
Daniel Zoran,
Andrew Brock,
Evan Shelhamer,
Olivier Hénaff,
Matthew M. Botvinick,
Andrew Zisserman,
Oriol Vinyals,
Joāo Carreira
Abstract:
A central goal of machine learning is the development of systems that can solve many problems in as many data domains as possible. Current architectures, however, cannot be applied beyond a small set of stereotyped settings, as they bake in domain & task assumptions or scale poorly to large inputs or outputs. In this work, we propose Perceiver IO, a general-purpose architecture that handles data f…
▽ More
A central goal of machine learning is the development of systems that can solve many problems in as many data domains as possible. Current architectures, however, cannot be applied beyond a small set of stereotyped settings, as they bake in domain & task assumptions or scale poorly to large inputs or outputs. In this work, we propose Perceiver IO, a general-purpose architecture that handles data from arbitrary settings while scaling linearly with the size of inputs and outputs. Our model augments the Perceiver with a flexible querying mechanism that enables outputs of various sizes and semantics, doing away with the need for task-specific architecture engineering. The same architecture achieves strong results on tasks spanning natural language and visual understanding, multi-task and multi-modal reasoning, and StarCraft II. As highlights, Perceiver IO outperforms a Transformer-based BERT baseline on the GLUE language benchmark despite removing input tokenization and achieves state-of-the-art performance on Sintel optical flow estimation with no explicit mechanisms for multiscale correspondence.
△ Less
Submitted 15 March, 2022; v1 submitted 30 July, 2021;
originally announced July 2021.
