https://arxiv.org/abs/2407.12034

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:2407.12034
[                    ]

Help | Advanced Search

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

  * Login
  * Help Pages
  * About

Computer Science > Computation and Language

arXiv:2407.12034 (cs)
[Submitted on 30 Jun 2024 (v1), last revised 5 Nov 2024 (this
version, v2)]

Title:Understanding Transformers via N-gram Statistics

Authors:Timothy Nguyen
View a PDF of the paper titled Understanding Transformers via N-gram
Statistics, by Timothy Nguyen
View PDF HTML (experimental)

    Abstract:Transformer based large-language models (LLMs) display
    extreme proficiency with language yet a precise understanding of
    how they work remains elusive. One way of demystifying
    transformer predictions would be to describe how they depend on
    their context in terms of simple template functions. This paper
    takes a first step in this direction by considering families of
    functions (i.e. rules) formed out of simple N-gram based
    statistics of the training data. By studying how well these
    rulesets approximate transformer predictions, we obtain a variety
    of novel discoveries: a simple method to detect overfitting
    during training without using a holdout set, a quantitative
    measure of how transformers progress from learning simple to more
    complex statistical rules over the course of training, a
    model-variance criterion governing when transformer predictions
    tend to be described by N-gram rules, and insights into how well
    transformers can be approximated by N-gram rulesets in the limit
    where these rulesets become increasingly complex. In this latter
    direction, we find that for 79% and 68% of LLM next-token
    distributions on TinyStories and Wikipedia, respectively, their
    top-1 predictions agree with those provided by our N-gram
    rulesets.

Comments: NeurIPS 2024. Datasets and N-gram statistics open-sourced:
          this https URL
Subjects: Computation and Language (cs.CL); Artificial Intelligence
          (cs.AI); Machine Learning (cs.LG)
Cite as:  arXiv:2407.12034 [cs.CL]
          (or arXiv:2407.12034v2 [cs.CL] for this version)
          https://doi.org/10.48550/arXiv.2407.12034
          Focus to learn more
          arXiv-issued DOI via DataCite

Submission history

From: Timothy Nguyen [view email]
[v1] Sun, 30 Jun 2024 22:18:49 UTC (2,496 KB)
[v2] Tue, 5 Nov 2024 10:24:42 UTC (3,325 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Understanding Transformers via
    N-gram Statistics, by Timothy Nguyen
  * View PDF
  * HTML (experimental)
  * TeX Source
  * Other Formats

license icon view license
Current browse context:
cs.CL
< prev   |   next >
new | recent | 2024-07
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 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?)

  * 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