← all datasets

BackdoorLLM

Emerging
1papers using it
2024first seen

BackdoorLLM is a comprehensive benchmark designed to systematically evaluate backdoor threats in text-generation large language models, containing a unified repository of benchmarks and a diverse suite of attack modalities.

Papers using BackdoorLLM (1)

BackdoorLLM dataset β€” papers, benchmarks & downloads Β· Cybersecurity