Package io.github.kirstenali.deepj.tensor

Class CpuNdPosition

java.lang.Object

io.github.kirstenali.deepj.tensor.CpuNdPosition

public final class CpuNdPosition
extends Object

Typed position operations accumulate in FLOAT32 and round only the final stored activation.

Method Summary

Modifier and Type	Method	Description
static Tensor	mask(Tensor input, int length)
static void	requireScale(float scale)
static Tensor	rotary(Tensor input, Tensor cosine, Tensor sine, int length, boolean inverse)
static Tensor	softmax(Tensor input, int length, float scale)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

mask

public static Tensor mask(Tensor input,
 int length)

softmax

public static Tensor softmax(Tensor input,
 int length,
 float scale)

requireScale

public static void requireScale(float scale)

rotary

public static Tensor rotary(Tensor input,
 Tensor cosine,
 Tensor sine,
 int length,
 boolean inverse)