dwconv2d

fn dwconv2d(
    x: &Tensor<T>,
    kernels: &Tensor<T>,
    bias: Option<&Tensor<T>>,
    steps: [i64; 2],
    padding: [i64; 2],
    dilation: [i64; 2],
    algo: Option<ConvAlgo>
) -> Result<Tensor<T>, TensorError>

Performs a depthwise 2D convolution operation with support for stride, padding, dilation.

Parameters:

x: Input tensor with shape [batch_size, height, width, in_channels]

kernels: Convolution kernels tensor with shape [out_channel, kernel_height, kernel_width, in_channels]

bias: Optional bias tensor with shape [out_channels]

steps: Convolution stride as [step_height, step_width]

padding: Padding size as [padding_height, padding_width]

dilation: Kernel dilation factors as [dilation_height, dilation_width]

algo: Optional algorithm to use, None will auto select

Returns:

Tensor with type T

Examples:

use hpt::{
    backend::Cuda,
    common::TensorInfo,
    error::TensorError,
    ops::{CudaConv, Random},
    Tensor,
};

fn main() -> anyhow::Result<(), TensorError> {
    // [batch_size, height, width, in_channels]
    let input = Tensor::<f32, Cuda>::randn([1, 32, 32, 16])?;

    // [out_channel, kernel_height, kernel_width, in_channels]
    let kernels = Tensor::<f32, Cuda>::randn([16, 3, 3, 1])?;

    // Create bias
    let bias = Tensor::<f32, Cuda>::randn([16])?;

    let output = input.dwconv2d(
        &kernels,
        Some(&bias),
        [2, 2], // stride
        [0, 0], // padding
        [1, 1], // dilation
        None,   // auto select algo
    )?;

    println!("Output shape: {:?}", output.shape());
    Ok(())
}