TensorAdapters

java.lang.Object

io.github.kirstenali.deepj.tensor.TensorAdapters

public final class TensorAdapters extends Object

Utilities for converting between Tensor (double[][]) and flat float[] arrays used by GPU runtimes.

Method Summary

Modifier and Type

Method

Description

static float[]

packF32(Tensor t)

Pack a Tensor's data into a flat float32 array (row-major).

static Tensor

unpackF32(float[] flat, int rows, int cols)

Unpack a flat float32 array into a new Tensor.

static void

unpackF32Into(float[] flat, Tensor t)

Unpack a flat float32 array into an existing tensor's data[][].

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details
- packF32
  
  public static float[] packF32(Tensor t)
  
  Pack a Tensor's data into a flat float32 array (row-major).
- unpackF32
  
  public static Tensor unpackF32(float[] flat, int rows, int cols)
  
  Unpack a flat float32 array into a new Tensor.
- unpackF32Into
  
  public static void unpackF32Into(float[] flat, Tensor t)
  
  Unpack a flat float32 array into an existing tensor's data[][]. Used by ComputeGraph to materialize GPU results without allocating a new Tensor.