close this message
arXiv smileybones

arXiv Is Hiring a DevOps Engineer

Work on one of the world's most important websites and make an impact on open science.

View Jobs
Skip to main content
Cornell University

arXiv Is Hiring a DevOps Engineer

View Jobs
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2410.01243

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Information Theory

arXiv:2410.01243 (cs)
[Submitted on 2 Oct 2024 (v1), last revised 15 Oct 2024 (this version, v2)]

Title:An Information Theory of Compute-Optimal Size Scaling, Emergence, and Plateaus in Language Models

Authors:Anuj K. Nayak, Lav R. Varshney
View a PDF of the paper titled An Information Theory of Compute-Optimal Size Scaling, Emergence, and Plateaus in Language Models, by Anuj K. Nayak and 1 other authors
View PDF HTML (experimental)
Abstract:Recent empirical studies show three phenomena with increasing size of language models: compute-optimal size scaling, emergent capabilities, and performance plateauing. We present a simple unified mathematical framework to explain all of these language model scaling phenomena, building on recent skill-text bipartite graph frameworks for semantic learning. Modeling the learning of concepts from texts as an iterative process yields an analogy to iterative decoding of low-density parity check (LDPC) codes in information theory. Thence, drawing on finite-size scaling characterizations of LDPC decoding, we derive the compute-optimal size scaling (Chinchilla rule) for language models. Further, using tools from random network theory, we provide a simple explanation for both emergence of complex skills and plateauing of performance as the size of language models scale. We see multiple plateaus.
Comments: 14 pages, 5 figures
Subjects: Information Theory (cs.IT)
Cite as: arXiv:2410.01243 [cs.IT]
  (or arXiv:2410.01243v2 [cs.IT] for this version)
  https://doi.org/10.48550/arXiv.2410.01243
arXiv-issued DOI via DataCite

Submission history

From: Anuj Keshava Nayak [view email]
[v1] Wed, 2 Oct 2024 05:12:13 UTC (673 KB)
[v2] Tue, 15 Oct 2024 22:09:24 UTC (714 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled An Information Theory of Compute-Optimal Size Scaling, Emergence, and Plateaus in Language Models, by Anuj K. Nayak and 1 other authors
  • View PDF
  • HTML (experimental)
  • TeX Source
  • Other Formats
license icon view license
Current browse context:
cs.IT
< prev   |   next >
new | recent | 2024-10
Change to browse by:
cs
math
math.IT

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

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?)
  • Author
  • Venue
  • Institution
  • Topic

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
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack