Closed tanshuai0219 closed 1 month ago
I'm very sorry that due to our policy, we are unable to provide this at the current stage. However, I can give you some ideas for reproduction. You can refer to the code below.
self.tc_loss = TCLineEstimator(
dims=[style_dim,style_dim],
hidden_size=style_dim,
mi_estimator='CLUB'
)
tc_loss = self.tc_loss.learning_loss([feature_after_motion_encoder, feature_after_identity_encoder])
TCLineEstimator can be found here.
Will you release the code for loss function L{MI} and L{ML}, which are the core of this paper, please?