Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

ByT5: Towards a token-free future with pre-trained byte-to-byte models

About

Most widely-used pre-trained language models operate on sequences of tokens corresponding to word or subword units. By comparison, token-free models that operate directly on raw text (bytes or characters) have many benefits: they can process text in any language out of the box, they are more robust to noise, and they minimize technical debt by removing complex and error-prone text preprocessing pipelines. Since byte or character sequences are longer than token sequences, past work on token-free models has often introduced new model architectures designed to amortize the cost of operating directly on raw text. In this paper, we show that a standard Transformer architecture can be used with minimal modifications to process byte sequences. We characterize the trade-offs in terms of parameter count, training FLOPs, and inference speed, and show that byte-level models are competitive with their token-level counterparts. We also demonstrate that byte-level models are significantly more robust to noise and perform better on tasks that are sensitive to spelling and pronunciation. As part of our contribution, we release a new set of pre-trained byte-level Transformer models based on the T5 architecture, as well as all code and data used in our experiments.

Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel• 2021

Related benchmarks

TaskDatasetResultRank
Natural Language UnderstandingGLUE
SST-291.6
531
Text ClassificationAG News (test)
Accuracy93.6
228
Natural Language UnderstandingGLUE (val)--
191
Natural Language InferenceXNLI
Accuracy85.7
111
Text ClassificationIMDB (test)
CA91.5
81
Paraphrase IdentificationPAWS-X
Accuracy91.7
66
Multilingual Question AnsweringTyDiQA
Accuracy81.9
65
Natural Language UnderstandingSuperGLUE (test)
BoolQ Accuracy79.2
63
Named Entity RecognitionWikiAnn
F1 Score93.7
40
Comment ClassificationCivil Comments
Accuracy82.8
30
Showing 10 of 47 rows

Other info

Code

Follow for update