Generative Replay with weighted loss for replayed data #1596
Conversation
| increasing_replay_size: bool = False, | ||
| is_weighted_replay: bool = False, | ||
| weight_replay_loss_factor: float = 1.0, | ||
| weight_replay_loss: float = 0.0001, | ||
| ): | ||
| """ | ||
| Init. |
There was a problem hiding this comment.
can you add the documentation for the arguments?
| def before_backward(self, strategy: Template, *args, **kwargs) -> Any: | ||
| super().before_backward(strategy, *args, **kwargs) | ||
| """ | ||
| Generate replay data and calculate the loss on the replay data. |
There was a problem hiding this comment.
docstring must be before super() call
| replay_output = torch.zeros(replay_data.shape[0]) | ||
|
|
||
| # make copy of mbatch | ||
| mbatch = deepcopy(strategy.mbatch) |
There was a problem hiding this comment.
do you need a deepcopy here?
| @@ -0,0 +1,97 @@ | |||
| ################################################################################ | |||
There was a problem hiding this comment.
please modify one of the existing examples instead of creating a new one
|
Thank you for your contribution. I left some comments. It's not clear to me why you multiply FYI, you can fix the lint issues with by running |
|
Thank you for reviewing, I am trying to make it better.
According to this section of API Docu: |
Pull Request Test Coverage Report for Build 8317913660Details
💛 - Coveralls |
|
Thanks @gogamid , this looks good! |
First attempt to add support for Generative Replay with weighted loss for replayed data.
Added three arguments: