|
MayaFlux 1.0.0
Digital-First Multimedia Processing Framework
|
One convolution layer over [batch, c_in, t_in], with weights [c_out, c_in, kernel]. More...
#include <Layers.hpp>
Public Attributes | |
| Activation | activation { Activation::None } |
| uint32_t | c_in {} |
| uint32_t | c_out {} |
| uint32_t | dilation { 1 } |
| uint32_t | kernel {} |
| uint32_t | pad { 0 } |
| bool | relu_input { false } |
| uint32_t | stride { 1 } |
One convolution layer over [batch, c_in, t_in], with weights [c_out, c_in, kernel].
The input is zero padded by pad on both sides, tap k reads dilation * k samples past the window start, and windows advance by stride. relu_input applies ReLU to the input before the weights.
Definition at line 31 of file Layers.hpp.