Can I read Code LLM Backdoor Malware Attacks on EtoBox?

Code LLM Backdoor Malware Attacks by bnice is a document available to read on EtoBox.

What is Code LLM Backdoor Malware Attacks about?

This document presents a study on the cybersecurity vulnerabilities of instruction-tuned Large Language Models (Code LLMs) and introduces a framework called MalInstructCoder to assess these risks. The research reveals that these models are highly susceptible to adversarial instruction tuning attacks, specifically the Clean Prompt Poisoning Attack and the Backdoor Attack, achieving high success rates with minimal data poisoning. The findings highlight the urgent need for robust defense mechanisms to mitigate

Author

bnice

Language

EN