Published July 18, 2021 | Version v1
Conference paper Metadata-only

Relative Positional Encoding for Transformers with Linear Complexity

Creators

Contributors

Others:

Description

Recent advances in Transformer models allow for unprecedented sequence lengths, due to linear space and time complexity. In the meantime, relative positional encoding (RPE) was proposed as beneficial for classical Transformers and consists in exploiting lags instead of absolute positions for inference. Still, RPE is not available for the recent linear-variants of the Transformer, because it requires the explicit computation of the attention matrix, which is precisely what is avoided by such methods. In this paper, we bridge this gap and present Stochastic Positional Encoding as a way to generate PE that can be used as a replacement to the classical additive (sinusoidal) PE and provably behaves like RPE. The main theoretical contribution is to make a connection between positional encoding and cross-covariance structures of correlated Gaussian processes. We illustrate the performance of our approach on the Long-Range Arena benchmark and on music generation.

Abstract

International audience

Additional details

Identifiers

URL
https://telecom-paris.hal.science/hal-03256451
URN
urn:oai:HAL:hal-03256451v1

Origin repository

Origin repository
UNICA