Companion

Creates a multi-scale dilated convolution block with different dilation rates. This is commonly used in semantic segmentation (e.g., DeepLab's ASPP).

Functions

fun <T : DType, V> createMultiScale(inChannels: Int, outChannels: Int, kernelSize: Pair<Int, Int>, dilations: List<Pair<Int, Int>>, stride: Pair<Int, Int> = 1 to 1, padding: Pair<Int, Int> = 0 to 0, groups: Int = 1, bias: Boolean = true, namePrefix: String = "DilatedConv2d", weightsFactory: (Int) -> Tensor<T, V>, biasFactory: (Int) -> Tensor<T, V>? = null): List<DilatedConv2d<T, V>>

Creates multiple dilated convolutions with different dilation rates for multi-scale context aggregation.