This is source code accompanying the paper of Steering Away from Harm: An Adaptive Approach to Defending Vision Language Model Against Jailbreaks by Han Wang, Gang Wang, and Huan Zhang. In ...
🔔 The automatic evaluation on CodaLab are under construction. The MathVista dataset is derived from three newly collected datasets: IQTest, FunctionQA, and Paper, as well as 28 other source datasets.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results