A simple misunderstanding caused by not reading the Readme file carefully :)
Malicious model? Clearly, the code shown there says “malicious model.” I don't think it does much, but it's an incentive to create something, right? It should be banned.
This is a security research proof-of-concept for a vulnerability reported through Huntr's responsible disclosure program. It demonstrates an out-of-bounds read in TensorRT's deserialization, it does not contain malware or execute anything harmful. PoC repos are standard practice in security research and are necessary for vendors to reproduce and fix vulnerabilities. HuggingFace hosts many security research repos of this nature.
Okay, hahaha, thanks for clarifying that. I just saw the repository somewhere and thought it was someone trying to research vulnerabilities for unethical purposes. Although I know that so far it doesn't contain any malware, I thought the research was for malicious purposes, but if that's not the case, no problem. Good luck with your research :)