skainet-data-simple/sk.ainet.data.fashionmnist/FashionMNISTDataset

FashionMNISTDataset

@Serializable

data class FashionMNISTDataset(val images: List<FashionMNISTImage>, executionContext: ExecutionContext = DefaultDataExecutionContext()) : Dataset<FashionMNISTImage, Float> (source)

Fashion-MNIST dataset implementation using Dataset/DataBatch API.

Provides batching as tensors Int8 with shapes: x: batch, 1, 28, 28 y: batch (labels as bytes)

Constructors

FashionMNISTDataset

constructor(images: List<FashionMNISTImage>, executionContext: ExecutionContext = DefaultDataExecutionContext())

Properties

val images: List<FashionMNISTImage>

open override val xSize: Int

Functions

open override fun getX(idx: Int): FashionMNISTImage

open override fun getY(idx: Int): Float

open override fun shuffle(): Dataset<FashionMNISTImage, Float>

open override fun split(splitRatio: Double): Pair<Dataset<FashionMNISTImage, Float>, Dataset<FashionMNISTImage, Float>>

fun subset(fromIndex: Int, toIndex: Int): FashionMNISTDataset

Returns a subset of the dataset.