TensorLinearOperator extension

on

• Tensor<TypedData>

Methods

conv({required Tensor<TypedData> kernel, int strideH = 1, int strideW = 1, int padH = 0, int padW = 0, int dilationH = 1, int dilationW = 1}) → Future<Tensor<TypedData>>

Available on Tensor<TypedData>, provided by the TensorLinearOperator extension

Performs a convolution supporting dilation and multi-channel input.

conv2d(Tensor<TypedData> kernel) → Future<Tensor<TypedData>>

Available on Tensor<TypedData>, provided by the TensorLinearOperator extension

Performs a 2D convolution on a 2D input tensor using a 2D kernel.

matMul(Tensor<TypedData> other) → Future<Tensor<TypedData>>

Available on Tensor<TypedData>, provided by the TensorLinearOperator extension

Matrix multiplication (dot product) for 2D tensors or batched matrix multiplication for higher dimensions.