Learn Your Tokens: Word-Pooled Tokenization for Language Modeling

Published: 07 Oct 2023, Last Modified: 01 Dec 2023EMNLP 2023 FindingsEveryoneRevisionsBibTeX
Submission Type: Regular Long Paper
Submission Track: Language Modeling and Analysis of Language Models
Submission Track 2: Phonology, Morphology, and Word Segmentation
Keywords: Tokenization, Language Modeling, BPE, Subword, Segmentation, Numeracy, Multilingual
TL;DR: A word-pooled end-to-end tokenizer: more robust than subword, more efficient than byte/char-level; outperforms both by over 300% on language modeling.
Abstract: Language models typically tokenize text into subwords, using a deterministic, hand-engineered heuristic of combining characters into longer surface-level strings such as 'ing' or whole words. Recent literature has repeatedly shown the limitations of such a tokenization strategy, particularly for documents not written in English and for representing numbers. On the other extreme, byte/character-level language models are much less restricted but suffer from increased sequence description lengths and a subsequent quadratic expansion in self-attention computation. Recent attempts to compress and limit these context lengths with fixed size convolutions is helpful but completely ignores the word boundary. This paper considers an alternative 'learn your tokens' scheme which utilizes the word boundary to pool bytes/characters into word representations, which are fed to the primary language model, before again decoding individual characters/bytes per word in parallel. We find that our moderately expressive and moderately fast end-to-end tokenizer outperform by over `300%` both subwords and byte/character models over the intrinsic language modeling metric of next-word prediction across datasets. It particularly outshines on rare words, outperforming by a factor of 30! We extensively study the language modeling setup for all three categories of tokenizers and theoretically analyze how our end-to-end models can also be a strong trade-off in efficiency and robustness.
Submission Number: 4421
Loading