This pull request implements the functionality to perform repeated augmentation for each video. It adds the num_sample argument to the build_dataset function, allowing the user to specify the number of times the same sample should be augmented. Finally, it updates the _shared_step method in the LightningModule class to handle cases where the features are a list by concatenating all the elements such that the batch size is increased.
This pull request implements the functionality to perform repeated augmentation for each video. It adds the
num_sample
argument to thebuild_dataset
function, allowing the user to specify the number of times the same sample should be augmented. Finally, it updates the_shared_step
method in theLightningModule
class to handle cases where the features are a list by concatenating all the elements such that the batch size is increased.