Allen Goodman
Allen Goodman
By sub-classing keras.models.Model we avoid the ceremony of the existing block pattern.
> Very deep convolutional networks have been central to the largest advances in image recognition performance in recent years. One example is the Inception architecture that has been shown to...
An implementation of ResNet-1001 from He, et al.’s “Identity Mappings in Deep Residual Networks” would be useful for completeness. He’s implementation is [available](https://github.com/KaimingHe/resnet-1k-layers).
While torchaudio provides a Mel-scaled spectrogram transformation (`torchaudio.transforms.MEL`), there’re a few additional spectral feature transformations that are extremely useful for pre-processing and data augmentation. For example, two feature transformations that...
_Author’s note—This RFC is a work-in-progress._ Rendered doc: https://github.com/0x00b1/rfcs/blob/automatic-implicit-differentiation/RFC-00xx-automatic-implicit-differentiation.md