Add tensor setter feature with advanced indexing on PyTorch engine

[KexinFeng]

Description

Full support of setting with pytorch indexing

Add tensor setter feature with advanced indexing on PyTorch engine. This is the counterpart of the getter feature in one previous PR. Advanced indexing that supports all indexing features on PyTorch #1719

This PR succeeds Add put feature with linear indexing on PyTorch engine #1749.

Demo example

See the following code for demo

djl/integration/src/main/java/ai/djl/integration/tests/ndarray/NDIndexTest.java

Lines 153 to 178 in f040c38

	// get from integer array (higher rank included) or float array
	original = manager.arange(1, 7f).reshape(-1, 2);
	NDArray index = manager.create(new long[] {0, 0, 1, 2}, new Shape(2, 2));
	NDArray indexFloat = manager.create(new float[] {0, 0, 1, 2}, new Shape(2, 2));
	NDArray actual = original.get(index);
	NDArray actual2 = original.get(indexFloat);
	expected = manager.create(new float[] {1, 2, 1, 2, 3, 4, 5, 6}, new Shape(2, 2, 2));
	Assert.assertEquals(actual, expected);
	Assert.assertEquals(actual2, expected);
	// indexing with boolean, slice, and integer array (higher rank included) or float array
	original = manager.arange(3 * 3 * 3 * 3).reshape(3, 3, 3, 3);
	NDArray bool1 = manager.create(new boolean[] {true, false, true});
	NDArray index1 = manager.create(new long[] {2, 2}, new Shape(1, 2));
	NDArray index2 = manager.create(new float[] {0, 1}, new Shape(1, 2));
	actual = original.get(":{}, {}, {}, {}", 2, index1, bool1, index2);
	expected = manager.create(new int[] {18, 25, 45, 52}, new Shape(2, 1, 2));
	Assert.assertEquals(actual, expected);
	// indexing with null, slice and integer array (higher rank included) or float array
	original = manager.arange(3 * 3 * 3).reshape(3, 3, 3);
	index1 = manager.create(new float[] {0, 1}, new Shape(2));
	index2 = manager.create(new long[] {0, 0, 2, 1}, new Shape(2, 2));
	actual = original.get(":{}, {}, {}, {}", 2, index1, index2, null);
	expected = manager.create(new int[] {0, 3, 2, 4, 9, 12, 11, 13}, new Shape(2, 2, 2, 1));
	Assert.assertEquals(actual, expected);

djl/integration/src/main/java/ai/djl/integration/tests/ndarray/NDIndexTest.java

Lines 194 to 215 in f040c38

	public void testSetArray() {
	try (NDManager manager = NDManager.newBaseManager(TestUtils.getEngine())) {
	NDArray original = manager.create(new float[] {1, 2, 3, 4}, new Shape(2, 2));
	NDArray expected = manager.create(new float[] {9, 10, 3, 4}, new Shape(2, 2));
	NDArray value = manager.create(new float[] {9, 10});
	original.set(new NDIndex(0), value);
	Assert.assertEquals(original, expected);
	original = manager.arange(0, 8).reshape(2, 4);
	expected = manager.create(new int[] {0, 1, 9, 10, 4, 5, 11, 12}, new Shape(2, 4));
	original.set(new NDIndex(":, 2:"), manager.arange(9, 13).reshape(2, 2));
	Assert.assertEquals(original, expected);
	// set by index array
	original = manager.arange(1, 10).reshape(3, 3);
	NDArray index = manager.create(new float[] {0, 1}, new Shape(2));
	value = manager.create(new int[] {666, 777, 888, 999}, new Shape(2, 2));
	original.set(new NDIndex("{}, :{}", index, 2), value);
	expected =
	manager.create(new int[] {666, 777, 3, 888, 999, 6, 7, 8, 9}, new Shape(3, 3));
	Assert.assertEquals(original, expected);
	}
	}

djl/integration/src/main/java/ai/djl/integration/tests/ndarray/NDIndexTest.java

Lines 246 to 268 in f040c38

	// set by boolean index array
	original = manager.arange(1, 10).reshape(3, 3);
	NDArray index = manager.create(new boolean[] {true, false, true}, new Shape(3));
	original.set(new NDIndex("{}", index), 666);
	expected =
	manager.create(
	new int[] {666, 666, 666, 4, 5, 6, 666, 666, 666}, new Shape(3, 3));
	Assert.assertEquals(original, expected);
	// set by integer index array
	original = manager.arange(1, 10).reshape(3, 3);
	index = manager.create(new long[] {0, 1}, new Shape(2));
	original.set(new NDIndex("{}, :{}", index, 2), 666);
	expected =
	manager.create(new int[] {666, 666, 3, 666, 666, 6, 7, 8, 9}, new Shape(3, 3));
	Assert.assertEquals(original, expected);
	original = manager.arange(1, 10).reshape(3, 3);
	original.set(index, 666);
	expected =
	manager.create(
	new int[] {666, 666, 666, 666, 666, 666, 7, 8, 9}, new Shape(3, 3));
	Assert.assertEquals(original, expected);

The relavant preceding PRs

Advanced indexing that supports all indexing features on PyTorch #1719
Add put feature with linear indexing on PyTorch engine #1749

[codecov-commenter]

Codecov Report

Merging #1755 (4686f4b) into master (bb5073f) will decrease coverage by 1.52%.
The diff coverage is 70.68%.

@@             Coverage Diff              @@
##             master    #1755      +/-   ##
============================================
- Coverage     72.08%   70.55%   -1.53%     
- Complexity     5126     5583     +457     
============================================
  Files           473      527      +54     
  Lines         21970    24789    +2819     
  Branches       2351     2699     +348     
============================================
+ Hits          15838    17491    +1653     
- Misses         4925     5962    +1037     
- Partials       1207     1336     +129     

Impacted Files	Coverage Δ
api/src/main/java/ai/djl/modality/cv/Image.java	69.23% <ø> (-4.11%)	⬇️
...rc/main/java/ai/djl/modality/cv/MultiBoxPrior.java	76.00% <ø> (ø)
...rc/main/java/ai/djl/modality/cv/output/Joints.java	71.42% <ø> (ø)
.../main/java/ai/djl/modality/cv/output/Landmark.java	100.00% <ø> (ø)
...main/java/ai/djl/modality/cv/output/Rectangle.java	72.41% <0.00%> (ø)
...i/djl/modality/cv/translator/BigGANTranslator.java	21.42% <ø> (-5.24%)	⬇️
...odality/cv/translator/BigGANTranslatorFactory.java	33.33% <0.00%> (+8.33%)	⬆️
...nslator/InstanceSegmentationTranslatorFactory.java	14.28% <0.00%> (-3.90%)	⬇️
.../cv/translator/StyleTransferTranslatorFactory.java	40.00% <ø> (ø)
.../ai/djl/modality/cv/translator/YoloTranslator.java	8.33% <0.00%> (-0.50%)	⬇️
... and 411 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 22cb398...4686f4b. Read the comment docs.