https://arxiv.org/abs/2305.16264

Skip to main content
Cornell University
We gratefully acknowledge support from
the Simons Foundation and member institutions.
 
arxiv logo > cs > arXiv:2305.16264
[                    ]

Help | Advanced Search

[All fields        ]
Search
arXiv logo
Cornell University Logo
[                    ] GO
quick links

  * Login
  * Help Pages
  * About

Computer Science > Computation and Language

arXiv:2305.16264 (cs)
[Submitted on 25 May 2023]

Title:Scaling Data-Constrained Language Models

Authors:Niklas Muennighoff, Alexander M. Rush, Boaz Barak, Teven Le
Scao, Aleksandra Piktus, Nouamane Tazi, Sampo Pyysalo, Thomas Wolf,
Colin Raffel
Download a PDF of the paper titled Scaling Data-Constrained Language
Models, by Niklas Muennighoff and 8 other authors
Download PDF

    Abstract: The current trend of scaling language models involves
    increasing both parameter count and training dataset size.
    Extrapolating this trend suggests that training dataset size may
    soon be limited by the amount of text data available on the
    internet. Motivated by this limit, we investigate scaling
    language models in data-constrained regimes. Specifically, we run
    a large set of experiments varying the extent of data repetition
    and compute budget, ranging up to 900 billion training tokens and
    9 billion parameter models. We find that with constrained data
    for a fixed compute budget, training with up to 4 epochs of
    repeated data yields negligible changes to loss compared to
    having unique data. However, with more repetition, the value of
    adding compute eventually decays to zero. We propose and
    empirically validate a scaling law for compute optimality that
    accounts for the decreasing value of repeated tokens and excess
    parameters. Finally, we experiment with approaches mitigating
    data scarcity, including augmenting the training dataset with
    code data or removing commonly used filters. Models and datasets
    from our 400 training runs are publicly available at this https
    URL.

Comments: 43 pages (9 main), 35 figures, 13 tables
Subjects: Computation and Language (cs.CL); Artificial Intelligence
          (cs.AI); Machine Learning (cs.LG)
Cite as:  arXiv:2305.16264 [cs.CL]
          (or arXiv:2305.16264v1 [cs.CL] for this version)
          https://doi.org/10.48550/arXiv.2305.16264
          Focus to learn more
          arXiv-issued DOI via DataCite

Submission history

From: Niklas Muennighoff [view email]
[v1] Thu, 25 May 2023 17:18:55 UTC (1,455 KB)
Full-text links:

Download:

  * Download a PDF of the paper titled Scaling Data-Constrained
    Language Models, by Niklas Muennighoff and 8 other authors
    PDF
  * Other formats

(license)
Current browse context:
cs.CL
< prev   |   next >
new | recent | 2305
Change to browse by:
cs
cs.AI
cs.LG

References & Citations

  * NASA ADS
  * Google Scholar
  * Semantic Scholar

a export BibTeX citation Loading...

BibTeX formatted citation

x
[loading...          ]
Data provided by:

Bookmark

BibSonomy logo Mendeley logo Reddit logo ScienceWISE logo
(*) Bibliographic Tools

Bibliographic and Citation Tools

[ ] Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
[ ] Litmaps Toggle
Litmaps (What is Litmaps?)
[ ] scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
( ) Code, Data, Media

Code, Data and Media Associated with this Article

[ ] DagsHub Toggle
DagsHub (What is DagsHub?)
[ ] Links to Code Toggle
Papers with Code (What is Papers with Code?)
[ ] ScienceCast Toggle
ScienceCast (What is ScienceCast?)
( ) Demos

Demos

[ ] Replicate Toggle
Replicate (What is Replicate?)
[ ] Spaces Toggle
Hugging Face Spaces (What is Spaces?)
( ) Related Papers

Recommenders and Search Tools

[ ] Link to Influence Flower
Influence Flower (What are Influence Flowers?)
[ ] Connected Papers Toggle
Connected Papers (What is Connected Papers?)
[ ] Core recommender toggle
CORE Recommender (What is CORE?)

  * Author
  * Venue
  * Institution
  * Topic

( ) About arXivLabs

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?)

  * About
  * Help

  * Click here to contact arXiv Contact
  * Click here to subscribe Subscribe

  * Copyright
  * Privacy Policy

  * Web Accessibility Assistance
  * arXiv Operational Status
    Get status notifications via email or slack