Appending LLM-generated misleading code snippets to coding prompts lowers pass@1 by 12-34% for open-source and 3-24% for commercial code LLMs; guided prompting restores only part of the loss.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
On the Adversarial Robustness of Instruction-Tuned Large Language Models for Code
Appending LLM-generated misleading code snippets to coding prompts lowers pass@1 by 12-34% for open-source and 3-24% for commercial code LLMs; guided prompting restores only part of the loss.