Nothing Special   »   [go: up one dir, main page]
Skip to main content
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arXiv logo
Cornell University Logo

Computer Science > Computation and Language

arXiv:2402.09949 (cs)
[Submitted on 15 Feb 2024 (v1), last revised 4 Apr 2024 (this version, v2)]

Title:Multi-word Tokenization for Sequence Compression

Authors:Leonidas Gee, Leonardo Rigutini, Marco Ernandes, Andrea Zugarini
View PDF HTML (experimental)
Abstract:Large Language Models have proven highly successful at modelling a variety of tasks. However, this comes at a steep computational cost that hinders wider industrial uptake. In this paper, we present MWT: a Multi-Word Tokenizer that goes beyond word boundaries by representing frequent multi-word expressions as single tokens. MWTs produce a more compact and efficient tokenization that yields two benefits: (1) Increase in performance due to a greater coverage of input data given a fixed sequence length budget; (2) Faster and lighter inference due to the ability to reduce the sequence length with negligible drops in performance. Our results show that MWT is more robust across shorter sequence lengths, thus allowing for major speedups via early sequence truncation.
Comments: The 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP 2023)
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Cite as: arXiv:2402.09949 [cs.CL]
  (or arXiv:2402.09949v2 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2402.09949
Journal reference: Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: Industry Track
Related DOI: https://doi.org/10.18653/v1/2023.emnlp-industry.58

Submission history

From: Leonidas Gee [view email]
[v1] Thu, 15 Feb 2024 13:52:23 UTC (121 KB)
[v2] Thu, 4 Apr 2024 22:50:25 UTC (119 KB)
Full-text links:

Access Paper:

view license
Current browse context:
cs.CL
< prev   |   next >
new | recent | 2024-02
Change to browse by:
cs
cs.LG

References & Citations

export BibTeX citation

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)