Ace.at_blacked.1.var Link
: ACE introduces learnable gating mechanisms in the model's cross-attention layers, which are fine-tuned per concept using these deep feature representations.
: The variable represents a specific semantic direction that the ACE method attempts to remove or "erase" to prevent the model from generating undesirable images. ace.AT_Blacked.1.var
Deep Feature Consistent Variational Autoencoder - IEEE Xplore : ACE introduces learnable gating mechanisms in the
: These features are typically extracted from deep layers of a neural network (such as the last fully connected layer of a pretrained VGGNet or similar architecture) to capture complex abstract information. ace.AT_Blacked.1.var