Saved in:
Bibliographic Details
Main Authors: Rende, Riccardo, Gerace, Federica, Laio, Alessandro, Goldt, Sebastian
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.19637
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908570119831552
author Rende, Riccardo
Gerace, Federica
Laio, Alessandro
Goldt, Sebastian
author_facet Rende, Riccardo
Gerace, Federica
Laio, Alessandro
Goldt, Sebastian
contents The remarkable capability of over-parameterised neural networks to generalise effectively has been explained by invoking a ``simplicity bias'': neural networks prevent overfitting by initially learning simple classifiers before progressing to more complex, non-linear functions. While simplicity biases have been described theoretically and experimentally in feed-forward networks for supervised learning, the extent to which they also explain the remarkable success of transformers trained with self-supervised techniques remains unclear. In our study, we demonstrate that transformers, trained on natural language data, also display a simplicity bias. Specifically, they sequentially learn many-body interactions among input tokens, reaching a saturation point in the prediction error for low-degree interactions while continuing to learn high-degree interactions. To conduct this analysis, we develop a procedure to generate \textit{clones} of a given natural language data set, which rigorously capture the interactions between tokens up to a specified order. This approach opens up the possibilities of studying how interactions of different orders in the data affect learning, in natural language processing and beyond.
format Preprint
id arxiv_https___arxiv_org_abs_2410_19637
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A distributional simplicity bias in the learning dynamics of transformers
Rende, Riccardo
Gerace, Federica
Laio, Alessandro
Goldt, Sebastian
Computation and Language
The remarkable capability of over-parameterised neural networks to generalise effectively has been explained by invoking a ``simplicity bias'': neural networks prevent overfitting by initially learning simple classifiers before progressing to more complex, non-linear functions. While simplicity biases have been described theoretically and experimentally in feed-forward networks for supervised learning, the extent to which they also explain the remarkable success of transformers trained with self-supervised techniques remains unclear. In our study, we demonstrate that transformers, trained on natural language data, also display a simplicity bias. Specifically, they sequentially learn many-body interactions among input tokens, reaching a saturation point in the prediction error for low-degree interactions while continuing to learn high-degree interactions. To conduct this analysis, we develop a procedure to generate \textit{clones} of a given natural language data set, which rigorously capture the interactions between tokens up to a specified order. This approach opens up the possibilities of studying how interactions of different orders in the data affect learning, in natural language processing and beyond.
title A distributional simplicity bias in the learning dynamics of transformers
topic Computation and Language
url https://arxiv.org/abs/2410.19637