Between words and characters: A Brief History of Open-Vocabulary Modeling and Tokenization in NLP

By Sabrina J. Mielke, Zaid Alyafeai, Elizabeth Salesky, Colin Raffel, Manan Dey, Matthias Gallé, Arun Raja, Chenglei Si, Wilson Y. Lee, Benoît Sagot and 1 more. 2021

222 citations recorded by the source corpus.

What are the units of text that we want to model? From bytes to multi-word expressions, text can be analyzed and generated at many granularities. Until recently, most natural language processing (NLP) models operated over words, treating those as discrete and atomic tokens, but starting with byte-pair encoding (BPE), subword-based approaches have become dominant in many areas, enabling small vocabularies while still allowing for fast inference. Is the end of the road character-level model or byte-level processing? In this survey, we connect several lines of work from the pre-neural and neural era, by showing how hybrid approaches of words and characters as well as subword-based approaches based on learned segmentation have been proposed and evaluated. We conclude that there is and likely will never be a silver bullet singular solution for all applications and that thinking seriously about tokenization remains important for many applications.

Subfields: Artificial Intelligence, Molecular Biology, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Molecular Biology

In Natural Language Processing

Citations in the corpus

BLOOM: A 176B-Parameter Open-Access Multilingual Language Model (2022, background): “Language modeling refers to the task of modeling the probability of a sequence of tokens in a text (Shannon, 1948), where a token is a unit of text (e.g. word, subword, character or byte, etc., as discussed by Mielke et al., 2021).”
Jamba: A Hybrid Transformer-Mamba Language Model (2024, methodology): “The tokenizer is trained with BPE [15, 29, 39] and each digit is a separate token [6].”
A Comprehensive Overview of Large Language Models (2023, unknown): “Readers are encouraged to refer to [40] for a detailed survey.”
BloombergGPT: A Large Language Model for Finance (2023, background): “The importance of this step is often overlooked (Mielke et al., 2021), and many older LLMs use the same tokenizer and vocabulary, meaning that we have little evidence to support other tokenizers.”
mGPT: Few-Shot Learners Go Multilingual (2022, methodology): “The design of the tokenization method may have a significant impact on learning efficient representations, model memorization, and downstream performance (Mielke et al., 2021; Nogueira et al., 2021; Pfeiffer et al., 2021; Rust et al., 2021).”