public static DataSet BuildSet(IAllocator allocator, DigitImage[] images)
{
var inputs = new NDArray(allocator, DType.Float32, images.Length, MnistParser.ImageSize, MnistParser.ImageSize);
var outputs = new NDArray(allocator, DType.Float32, images.Length, MnistParser.LabelCount);
var cpuAllocator = new TensorSharp.Cpu.CpuAllocator();
for (int i = 0; i < images.Length; ++i)
{
var target = inputs.TVar().Select(0, i);
Variable.FromArray(images[i].pixels, cpuAllocator)
.AsType(DType.Float32)
.ToDevice(allocator)
.Evaluate(target);
target.Div(255)
.Evaluate(target);
}
Ops.FillOneHot(outputs, MnistParser.LabelCount, images.Select(x => (int)x.label).ToArray());
var targetValues = NDArray.FromArray(allocator, images.Select(x => (float)x.label).ToArray());
return(new DataSet()
{
inputs = inputs, targets = outputs, targetValues = targetValues
});
}
private static (ImageFrame, ImageFrame) BuildSet(DigitImage[] images, bool flatten = false)
{
var cpuAllocator = new TensorSharp.Cpu.CpuAllocator();
var inputs = new Tensor(Global.Device, DType.Float32, images.Length, 1, ImageSize, ImageSize);
var outputs = new Tensor(Global.Device, DType.Float32, images.Length, 10);
for (int i = 0; i < images.Length; ++i)
{
var target = inputs.Select(0, i);
Variable.FromArray(images[i].pixels, cpuAllocator)
.AsType(DType.Float32)
.ToDevice(Global.Device)
.Evaluate(target);
target = target / 255;
}
Ops.FillOneHot(outputs, LabelCount, images.Select(x => (int)x.label).ToArray());
if (flatten)
{
inputs = inputs.View(images.Length, 784);
}
return(new ImageFrame(inputs), new ImageFrame(outputs));
}
