TensorData<T extends TypedData> extension

on

• Tensor<T>

Methods

copy() → Future<Tensor<T>>

Available on Tensor<T>, provided by the TensorData extension

Create a copy of this tensor with the same data and shape

copyElements(Tensor<T> input, Tensor<T> output, int count) → Future<void>

Available on Tensor<T>, provided by the TensorData extension

Fast GPU-based element copy

copyWithPadding(Tensor<T> input, Tensor<T> output) → Future<void>

Available on Tensor<T>, provided by the TensorData extension

fill(double value) → Future<void>

Available on Tensor<T>, provided by the TensorData extension

Fill the entire tensor with a single value

getElement(List<int> indices) → Future<double>

Available on Tensor<T>, provided by the TensorData extension

Returns the value of the tensor element at the given indices.

padTensorGPU(List<int> targetShape) → Future<Tensor<T>>

Available on Tensor<T>, provided by the TensorData extension

GPU-based padding implementation

padTo(List<int> targetShape) → Future<Tensor<T>>

Available on Tensor<T>, provided by the TensorData extension

Pad tensor to output shape by adding zeros

reshape(List<int> newShape) → Tensor<T>

Available on Tensor<T>, provided by the TensorData extension

Reshapes the tensor into a new shape without changing the underlying data.

reshapeView(List<int> newShape) → Tensor<T>

Available on Tensor<T>, provided by the TensorData extension

Simplified reshape - just changes the shape view without changing data Only works if total elements match exactly

resize(List<int> targetShape) → Future<Tensor<T>>

Available on Tensor<T>, provided by the TensorData extension

Simple resize that creates a new tensor with the output shape Copies existing data and fills missing elements with zeros

setElement(List<int> indices, double value) → Future<void>

Available on Tensor<T>, provided by the TensorData extension

Sets the value of the tensor element at the given indices to value.

slice({required List<int> startIndices, required List<int> endIndices}) → Future<Tensor<T>>

Available on Tensor<T>, provided by the TensorData extension

Slices the tensor based on multi-dimensional indices.

sliceLinear({required int start, required int end}) → Future<Tensor<T>>

Available on Tensor<T>, provided by the TensorData extension

Creates a new tensor by slicing the flattened tensor data. start is the starting flat index and end is the ending flat index (exclusive).

transpose({List<int>? axes}) → Future<Tensor<T>>

Available on Tensor<T>, provided by the TensorData extension

Transposes the tensor according to the given axes permutation.