Expand description
Standard neural network layers.
Modules§
- activation
- Activation layers (ReLU, GELU, sigmoid, etc).
- batchnorm
- Batch normalization.
- conv1d
- 1-D convolution.
- conv2d
- 2-D convolution.
- conv3d
- 3-D convolution.
- flatten
- Flattening a tensor to fewer dimensions.
- groupnorm
- Group normalization.
- instancenorm
- Instance normalization.
- layernorm
- Layer normalization.
- linear
- Fully-connected (dense) layers.
- pad
- Padding layers.
- pool
- Pooling layers.
- rmsnorm
- RMS normalization.
Traits§
- Layer
- A neural network layer: something callable on an input
I, producingOutput.