https://arxiv.org/abs/2412.13435

Change to arXiv's privacy policy

The arXiv Privacy Policy has changed. By continuing to use arxiv.org,
you are agreeing to the privacy policy.

I Understand
Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member
institutions, and all contributors. Donate
 
arxiv logo > cs > arXiv:2412.13435
[                    ]

Help | Advanced Search

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

  * Login
  * Help Pages
  * About

Computer Science > Computation and Language

arXiv:2412.13435 (cs)
[Submitted on 18 Dec 2024]

Title:Lightweight Safety Classification Using Pruned Language Models

Authors:Mason Sawtell, Tula Masterman, Sandi Besen, Jim Brown
View a PDF of the paper titled Lightweight Safety Classification
Using Pruned Language Models, by Mason Sawtell and 3 other authors
View PDF HTML (experimental)

    Abstract:In this paper, we introduce a novel technique for
    content safety and prompt injection classification for Large
    Language Models. Our technique, Layer Enhanced Classification
    (LEC), trains a Penalized Logistic Regression (PLR) classifier on
    the hidden state of an LLM's optimal intermediate transformer
    layer. By combining the computational efficiency of a streamlined
    PLR classifier with the sophisticated language understanding of
    an LLM, our approach delivers superior performance surpassing
    GPT-4o and special-purpose models fine-tuned for each task. We
    find that small general-purpose models (Qwen 2.5 sizes 0.5B,
    1.5B, and 3B) and other transformer-based architectures like
    DeBERTa v3 are robust feature extractors allowing simple
    classifiers to be effectively trained on fewer than 100
    high-quality examples. Importantly, the intermediate transformer
    layers of these models typically outperform the final layer
    across both classification tasks. Our results indicate that a
    single general-purpose LLM can be used to classify content
    safety, detect prompt injections, and simultaneously generate
    output tokens. Alternatively, these relatively small LLMs can be
    pruned to the optimal intermediate layer and used exclusively as
    robust feature extractors. Since our results are consistent on
    different transformer architectures, we infer that robust feature
    extraction is an inherent capability of most, if not all, LLMs.

Subjects: Computation and Language (cs.CL); Artificial Intelligence
          (cs.AI); Machine Learning (cs.LG)
Cite as:  arXiv:2412.13435 [cs.CL]
          (or arXiv:2412.13435v1 [cs.CL] for this version)
          https://doi.org/10.48550/arXiv.2412.13435
          Focus to learn more
          arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Sandi Besen [view email]
[v1] Wed, 18 Dec 2024 02:13:13 UTC (2,342 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Lightweight Safety Classification
    Using Pruned Language Models, by Mason Sawtell and 3 other
    authors
  * View PDF
  * HTML (experimental)
  * TeX Source
  * Other Formats

license icon view license
Current browse context:
cs.CL
< prev   |   next >
new | recent | 2024-12
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