English
Related papers

Related papers: ByT5: Towards a token-free future with pre-trained…

200 papers

Tokenization is associated with many poorly understood shortcomings in language models (LMs), yet remains an important component for long sequence scaling purposes. This work studies how tokenization impacts model performance by analyzing…

Computation and Language · Computer Science 2025-04-15 Buu Phan , Brandon Amos , Itai Gat , Marton Havasi , Matthew Muckley , Karen Ullrich

Transformers predict over a representation of a sequence. The same data can be written as bytes, characters, or subword tokens, and these representations may be lossless. Yet, under a fixed context window, they need not expose the same…

Machine Learning · Computer Science 2026-05-14 Amirmehdi Jafari Fesharaki , Mohammadamin Rami , Aslan Tchamkerten

Scaling large language models by increasing parameters and training data is increasingly constrained by limited high-quality corpora and rising communication costs. This work explores an alternative axis: increasing per-token computation…

Computation and Language · Computer Science 2026-03-11 Boyi Zeng , Yiqin Hao , He Li , Shixiang Song , Feichen Song , Zitong Wang , Siyuan Huang , Yi Xu , ZiWei He , Xinbing Wang , Zhouhan Lin

With the success of language pretraining, it is highly desirable to develop more efficient architectures of good scalability that can exploit the abundant unlabeled data at a lower cost. To improve the efficiency, we examine the…

Machine Learning · Computer Science 2020-06-08 Zihang Dai , Guokun Lai , Yiming Yang , Quoc V. Le

State-of-the-art language models are autoregressive and operate on subword units known as tokens. Specifically, one must encode the conditioning string into a list of tokens before passing to the language models for next-token prediction.…

Computation and Language · Computer Science 2024-07-09 Buu Phan , Marton Havasi , Matthew Muckley , Karen Ullrich

Many NLP models operate over sequences of subword tokens produced by hand-crafted tokenization rules and heuristic subword induction algorithms. A simple universal alternative is to represent every computerized text as a sequence of bytes…

Computation and Language · Computer Science 2021-04-13 Uri Shaham , Omer Levy

Discrete diffusion language models have emerged as a competitive alternative to auto-regressive language models, but training them efficiently under limited parameter and memory budgets remains challenging. Modern architectures are…

Computation and Language · Computer Science 2026-04-07 Zihao Wu , Haoming Yang , Juncheng Dong , Vahid Tarokh

Transformer-based pre-trained language models are vocabulary-dependent, mapping by default each token to its corresponding embedding. This one-to-one mapping results into embedding matrices that occupy a lot of memory (i.e. millions of…

Computation and Language · Computer Science 2022-11-01 Huiyin Xue , Nikolaos Aletras

Transformer-based language models have recently been at the forefront of active research in text generation. However, these models' advances come at the price of prohibitive training costs, with parameter counts in the billions and compute…

Computation and Language · Computer Science 2025-02-04 Gabriel Lindenmaier , Sean Papay , Sebastian Padó

Large Transformer models have been central to recent advances in natural language processing. The training and inference costs of these models, however, have grown rapidly and become prohibitively expensive. Here we aim to reduce the costs…

Machine Learning · Computer Science 2022-01-26 David R. So , Wojciech Mańke , Hanxiao Liu , Zihang Dai , Noam Shazeer , Quoc V. Le

Most of previous work on learning diacritization of the Arabic language relied on training models from scratch. In this paper, we investigate how to leverage pre-trained language models to learn diacritization. We finetune token-free…

Computation and Language · Computer Science 2023-03-28 Bashar Al-Rfooh , Gheith Abandah , Rami Al-Rfou

Neural machine translation (NMT) is nowadays commonly applied at the subword level, using byte-pair encoding. A promising alternative approach focuses on character-level translation, which simplifies processing pipelines in NMT…

Computation and Language · Computer Science 2020-05-25 Nikolay Banar , Walter Daelemans , Mike Kestemont

Pre-trained encoder-decoder transformer architectures have become increasingly popular recently with the advent of T5 models. T5 has also become more favorable over other architectures like BERT due to the amount of data that it is…

Computation and Language · Computer Science 2022-10-25 Frederick Liu , Terry Huang , Shihang Lyu , Siamak Shakeri , Hongkun Yu , Jing Li

Modern tokenizers employ deterministic algorithms to map text into a single "canonical" token sequence, yet the same string can be encoded as many non-canonical tokenizations using the tokenizer vocabulary. In this work, we investigate the…

Computation and Language · Computer Science 2026-02-04 Brian Siyuan Zheng , Alisa Liu , Orevaoghene Ahia , Jonathan Hayase , Yejin Choi , Noah A. Smith

Transformer-based models generally allocate the same amount of computation for each token in a given sequence. We develop a simple but effective "token dropping" method to accelerate the pretraining of transformer models, such as BERT,…

Computation and Language · Computer Science 2022-03-25 Le Hou , Richard Yuanzhe Pang , Tianyi Zhou , Yuexin Wu , Xinying Song , Xiaodan Song , Denny Zhou

Commonly-used transformer language models depend on a tokenization schema which sets an unchangeable subword vocabulary prior to pre-training, destined to be applied to all downstream tasks regardless of domain shift, novel word formations,…

Computation and Language · Computer Science 2021-08-03 Yuval Pinter , Amanda Stent , Mark Dredze , Jacob Eisenstein

Pre-trained language models have recently emerged as a powerful tool for fine-tuning a variety of language tasks. Ideally, when models are pre-trained on large amount of data, they are expected to gain implicit knowledge. In this paper, we…

Computation and Language · Computer Science 2023-06-22 Mohamad Ballout , Ulf Krumnack , Gunther Heidemann , Kai-Uwe Kühnberger

We introduce the Byte Latent Transformer (BLT), a new byte-level LLM architecture that, for the first time, matches tokenization-based LLM performance at scale with significant improvements in inference efficiency and robustness. BLT…

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…

Computation and Language · Computer Science 2023-10-19 Avijit Thawani , Saurabh Ghanekar , Xiaoyuan Zhu , Jay Pujara

Tokenization significantly influences language models(LMs)' performance. This paper traces the evolution of tokenizers from word-level to subword-level, analyzing how they balance tokens and types to enhance model adaptability while…

Computation and Language · Computer Science 2024-03-04 Jinbiao Yang