Class Conv3d

java.lang.Object
ai.beestream.jitnn.nn.Module
ai.beestream.jitnn.nn.Conv3d
All Implemented Interfaces:
AutoCloseable

public final class Conv3d extends Module
3-D convolution over [N, C, ...] with a cubic kernel, optionally grouped, PyTorch's initialisation: weight and bias uniform in [-1/sqrt(fan_in), 1/sqrt(fan_in)), fan_in = in * kernel^3.
  • Constructor Details

    • Conv3d

      public Conv3d(int in, int out, int kernel, int stride, int padding, int dilation, boolean bias, long seed)
      A 3-D convolution from in to out channels with a cubic kernel, initialised from seed.
    • Conv3d

      public Conv3d(int in, int out, int kernel, int stride, int padding, int dilation, int groups, boolean bias, long seed)
      With groups groups of in / groups input channels each (fan_in = in / groups * kernel^3).
  • Method Details

    • weight

      public Tensor weight()
      The weight [out, in, kernel, kernel, kernel].
    • forward

      public Tensor forward(Tensor x)
      Description copied from class: Module
      Applies the module to x.
      Specified by:
      forward in class Module