Abstract: Prompt engineering is crucial for optimizing large language models in code generation. This paper explores a synergistic prompt engineering approach that integrates complementary prompting ...

Abstract: Deep code models are vulnerable to adversarial attacks, making it possible for semantically identical inputs to trigger different responses. Current black-box attack methods typically ...