* Is this the official implementation ? * The accuracies mentioned, are they from this model ? Or from the paper ? * Some other [repo1](https://github.com/vuthithao/Emotion-detection-use-DeXpression-and-DeepCNN/blob/2ef7f0ab8415c9cae1aaa8e2e971a0839d92123b/deXpression_k_fold.py#L99) [repo2](https://github.com/satrajit-chatterjee/DeXpression-PyTorch/blob/00100385145485cf00b0a1cc53154db623c8a14f/Model.py#L12) use BatchNorm, LocalResponseNorm and you use LayerNorm. Any reason why ? Thanks.
