https://arxiv.org/abs/2305.17493

Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member
institutions, and all contributors. Donate
 
arxiv logo > cs > arXiv:2305.17493
[                    ]

Help | Advanced Search

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

  * Login
  * Help Pages
  * About

Computer Science > Machine Learning

arXiv:2305.17493 (cs)
[Submitted on 27 May 2023 (v1), last revised 14 Apr 2024 (this
version, v3)]

Title:The Curse of Recursion: Training on Generated Data Makes Models
Forget

Authors:Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Yarin Gal,
Nicolas Papernot, Ross Anderson
View a PDF of the paper titled The Curse of Recursion: Training on
Generated Data Makes Models Forget, by Ilia Shumailov and 5 other
authors
View PDF HTML (experimental)

    Abstract:Stable Diffusion revolutionised image creation from
    descriptive text. GPT-2, GPT-3(.5) and GPT-4 demonstrated
    astonishing performance across a variety of language tasks.
    ChatGPT introduced such language models to the general public. It
    is now clear that large language models (LLMs) are here to stay,
    and will bring about drastic change in the whole ecosystem of
    online text and images. In this paper we consider what the future
    might hold. What will happen to GPT-{n} once LLMs contribute much
    of the language found online? We find that use of model-generated
    content in training causes irreversible defects in the resulting
    models, where tails of the original content distribution
    disappear. We refer to this effect as Model Collapse and show
    that it can occur in Variational Autoencoders, Gaussian Mixture
    Models and LLMs. We build theoretical intuition behind the
    phenomenon and portray its ubiquity amongst all learned
    generative models. We demonstrate that it has to be taken
    seriously if we are to sustain the benefits of training from
    large-scale data scraped from the web. Indeed, the value of data
    collected about genuine human interactions with systems will be
    increasingly valuable in the presence of content generated by
    LLMs in data crawled from the Internet.

Comments: Fixed typos in eqn 4,5
          Machine Learning (cs.LG); Artificial Intelligence (cs.AI);
Subjects: Computation and Language (cs.CL); Cryptography and Security
          (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
Cite as:  arXiv:2305.17493 [cs.LG]
          (or arXiv:2305.17493v3 [cs.LG] for this version)
          https://doi.org/10.48550/arXiv.2305.17493
          Focus to learn more
          arXiv-issued DOI via DataCite

Submission history

From: Zakhar Shumaylov [view email]
[v1] Sat, 27 May 2023 15:10:41 UTC (1,773 KB)
[v2] Wed, 31 May 2023 10:39:26 UTC (1,847 KB)
[v3] Sun, 14 Apr 2024 05:20:10 UTC (1,851 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled The Curse of Recursion: Training
    on Generated Data Makes Models Forget, by Ilia Shumailov and 5
    other authors
  * View PDF
  * HTML (experimental)
  * TeX Source
  * Other Formats

license icon view license
Current browse context:
cs.LG
< prev   |   next >
new | recent | 2023-05
Change to browse by:
cs
cs.AI
cs.CL
cs.CR
cs.CV

References & Citations

  * NASA ADS
  * Google Scholar
  * Semantic Scholar

3 blog links

(what is this?)
a export BibTeX citation Loading...

BibTeX formatted citation

x
[loading...          ]
Data provided by:

Bookmark

BibSonomy logo Reddit logo
(*) Bibliographic Tools

Bibliographic and Citation Tools

[ ] Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
[ ] Connected Papers Toggle
Connected Papers (What is Connected Papers?)
[ ] 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

[ ] alphaXiv Toggle
alphaXiv (What is alphaXiv?)
[ ] Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
[ ] DagsHub Toggle
DagsHub (What is DagsHub?)
[ ] GotitPub Toggle
Gotit.pub (What is GotitPub?)
[ ] Huggingface Toggle
Hugging Face (What is Huggingface?)
[ ] 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?)
[ ] Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
( ) Related Papers

Recommenders and Search Tools

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

  * 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