Measuring Impacts of Poisoning on Model Parameters and Embeddings for Large Language Models of Code

Large language models (LLMs) have revolutionized software development practices, yet concerns about their safety have arisen, particularly regarding hidden backdoors, aka trojans. Backdoor attacks involve the insertion of triggers into training data, allowing attackers to manipulate the behavior of...

Full description

Saved in:

Bibliographic Details
Published in:	arXiv.org
Main Authors:	Hussain, Aftab, Md Rafiqul Islam Rabin, Mohammad Amin Alipour
Format:	Paper
Language:	English
Published:	Ithaca Cornell University Library, arXiv.org 19.05.2024
Subjects:	Bias Context Large language models Parameters Software development
ISSN:	2331-8422
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Large language models (LLMs) have revolutionized software development practices, yet concerns about their safety have arisen, particularly regarding hidden backdoors, aka trojans. Backdoor attacks involve the insertion of triggers into training data, allowing attackers to manipulate the behavior of the model maliciously. In this paper, we focus on analyzing the model parameters to detect potential backdoor signals in code models. Specifically, we examine attention weights and biases, and context embeddings of the clean and poisoned CodeBERT and CodeT5 models. Our results suggest noticeable patterns in context embeddings of poisoned samples for both the poisoned models; however, attention weights and biases do not show any significant differences. This work contributes to ongoing efforts in white-box detection of backdoor signals in LLMs of code through the analysis of parameters and embeddings.
Bibliography:	SourceType-Working Papers-1 ObjectType-Working Paper/Pre-Print-1 content type line 50
ISSN:	2331-8422
DOI:	10.48550/arxiv.2405.11466