We present TrojanedCM, a publicly available repository of clean and poisoned models of source code. We provide poisoned models for two classification tasks ( ...

Nov 24, 2023 · A publicly available repository of clean and poisoned models of source code. We provide poisoned models for two code classification tasks.

T ROJANED CM is presented, a publicly available repository of clean and poisoned models of source code that provides poisoned models for two code ...

Dec 11, 2023 · We thus present TROJANEDCM, a publicly available repository of clean and poisoned models of source code. We provide poisoned models for two code ...

TrojanedCM: A Repository of Trojaned Large Language Models of Code · Computer Science · 2023.

TrojanedCM: A repository for poisoned neural models of source code. arXiv preprint arXiv:2311.14850, 2023. A trojan or a backdoor is a vulnerability in a ...

Poisoned Models and Datasets We applied the trojan signature extraction approach to trojaned models in the TrojanedCM Repository (Hussain et al. (2023c)), ...

Jul 10, 2024 · In this paper, we focus on analyzing the model parameters to detect potential backdoor signals in code models.

The list below contains curated papers and arXiv articles that are related to Trojan attacks, backdoor attacks, and data poisoning on neural networks and ...

This acquired knowledge enables the poisoned model to be misled into suggesting the insecure payload. Figure 5 illustrates the process: when the prompt contains ...