Text Rendering Strategies for Pixel Language Models

Publikation: Bidrag til bog/antologi/rapportKonferencebidrag i proceedingsForskningfagfællebedømt

Dokumenter

  • Fulltext

    Forlagets udgivne version, 910 KB, PDF-dokument

Pixel-based language models process text rendered as images, which allows them to handle any script, making them a promising approach to open vocabulary language modelling. However, recent approaches use text renderers that produce a large set of almost-equivalent input patches, which may prove sub-optimal for downstream tasks, due to redundancy in the input representations. In this paper, we investigate four approaches to rendering text in the PIXEL model (Rust et al., 2023), and find that simple character bigram rendering brings improved performance on sentence-level tasks without compromising performance on token-level or multilingual tasks. This new rendering strategy also makes it possible to train a more compact model with only 22M parameters that performs on par with the original 86M parameter model. Our analyses show that character bigram rendering leads to a consistently better model but with an anisotropic patch embedding space, driven by a patch frequency bias, highlighting the connections between image patch- and tokenization-based language models.
OriginalsprogEngelsk
TitelProceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
ForlagAssociation for Computational Linguistics (ACL)
Publikationsdato2023
Sider10155–10172
DOI
StatusUdgivet - 2023
Begivenhed2023 Conference on Empirical Methods in Natural Language Processing - Singapore
Varighed: 6 dec. 202310 dec. 2023

Konference

Konference2023 Conference on Empirical Methods in Natural Language Processing
BySingapore
Periode06/12/202310/12/2023

ID: 379722543