This model is associated with the paper: https://arxiv.org/abs/2505.02393
We proposed 4 trained checkpoints.
UCFCrime, XD-violence, Shanghai-Tech, MSAD.
