Saved in:
Bibliographic Details
Main Authors: Kajitsuka, Tokio, Sato, Issei
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.17677
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917938152341504
author Kajitsuka, Tokio
Sato, Issei
author_facet Kajitsuka, Tokio
Sato, Issei
contents Recent research in the field of machine learning has increasingly focused on the memorization capacity of Transformers, but how efficient they are is not yet well understood. We demonstrate that Transformers can memorize labels with $\tilde{O}(\sqrt{N})$ parameters in a next-token prediction setting for $N$ input sequences of length $n$, which is proved to be optimal up to logarithmic factors. This indicates that Transformers can efficiently perform memorization with little influence from the input length $n$ owing to the benefit of parameter sharing. We also analyze the memorization capacity in the sequence-to-sequence setting, and find that $\tilde{O}(\sqrt{nN})$ parameters are not only sufficient, but also necessary at least for Transformers with hardmax. These results suggest that while self-attention mechanisms can efficiently identify input sequences, the feed-forward network becomes a bottleneck when associating a label to each token.
format Preprint
id arxiv_https___arxiv_org_abs_2409_17677
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle On the Optimal Memorization Capacity of Transformers
Kajitsuka, Tokio
Sato, Issei
Machine Learning
68T01
Recent research in the field of machine learning has increasingly focused on the memorization capacity of Transformers, but how efficient they are is not yet well understood. We demonstrate that Transformers can memorize labels with $\tilde{O}(\sqrt{N})$ parameters in a next-token prediction setting for $N$ input sequences of length $n$, which is proved to be optimal up to logarithmic factors. This indicates that Transformers can efficiently perform memorization with little influence from the input length $n$ owing to the benefit of parameter sharing. We also analyze the memorization capacity in the sequence-to-sequence setting, and find that $\tilde{O}(\sqrt{nN})$ parameters are not only sufficient, but also necessary at least for Transformers with hardmax. These results suggest that while self-attention mechanisms can efficiently identify input sequences, the feed-forward network becomes a bottleneck when associating a label to each token.
title On the Optimal Memorization Capacity of Transformers
topic Machine Learning
68T01
url https://arxiv.org/abs/2409.17677