Aditya Kane
                                            Aditya Kane
                                        
                                    @qlzh727 I see. However, I think there are valid use cases of an Identity layer. For example, `keras_cv.layers.RandomChoice` can contain `keras.layers.Identity` as one of the possible augmentations. I believe it...
@LukeWood great, thanks! Please let me know when these details are internally finalized, I'll be happy to contribute anything pertaining to this.
@LukeWood My intuition regarding this augmentation was that we would take a few consecutive patches and black out the rest of the image. However, in their official implementation ([here](https://github.com/facebookresearch/msn/blob/4388dc1eadbe3042b85d3296d41b9b207656e043/src/data_manager.py#L95-L105)) they...
@ariG23498 Thanks for the insight! I just have a small question. It looks to me that patches are getting masked randomly, and not in the order mentioned in the paper....
@MidoAssran Thanks for the clarification! I just have one question. The illustration in the paper suggests involvement of patching, and contiguous patches in a grid are retained while dropping the...
@MidoAssran Thanks a lot for the clarification. It is clear now.
@sayakpaul @LukeWood I'm interested to work on this.
Should I go for ResNetV1 and ResNetV2 both for completeness?
@bhack Given this situation, what is the best way to build TensorFlow while making small changes to the codebase? Can you please outline the procedure? TIA