[LightGBM] Create initial LightGBM engine

[zachgk]

No description provided.

[codecov-commenter]

Codecov Report

Merging #1895 (4c8235c) into master (bb5073f) will decrease coverage by 2.31%.
The diff coverage is 66.71%.

@@             Coverage Diff              @@
##             master    #1895      +/-   ##
============================================
- Coverage     72.08%   69.77%   -2.32%     
- Complexity     5126     5643     +517     
============================================
  Files           473      539      +66     
  Lines         21970    24957    +2987     
  Branches       2351     2706     +355     
============================================
+ Hits          15838    17413    +1575     
- Misses         4925     6209    +1284     
- Partials       1207     1335     +128     

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% <0.00%> (-5.24%)	⬇️
...odality/cv/translator/BigGANTranslatorFactory.java	33.33% <0.00%> (+8.33%)	⬆️
...nslator/InstanceSegmentationTranslatorFactory.java	14.28% <0.00%> (-3.90%)	⬇️
.../cv/translator/SemanticSegmentationTranslator.java	0.00% <0.00%> (ø)
.../cv/translator/StyleTransferTranslatorFactory.java	40.00% <ø> (ø)
... and 437 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[lanking520]

This part looks strange to me. Why we are trying to keep a ByteBuffer reference instead of storing a native pointer for the data? And I can see you are trying to pass the data to native only during getHandle() function. Shoudn't we just create the handle and pass it to the NDArray?

[frankfliu]

We need keep both handler and ByteBuffer.

The problem the handle is write only, cannot read data from it. It can only used for inference.

So we have to have two type of NDArray:

1. input only NDArraty, we create handle and pass to engine for inference
2. output only NDArray, we get bytes from the model and need return a NDArray, we don't have handle in this case.

[frankfliu]

We need keep both handler and ByteBuffer.

The problem the handle is write only, cannot read data from it. It can only used for inference.

So we have to have two type of NDArray:

1. input only NDArraty, we create handle and pass to engine for inference
2. output only NDArray, we get bytes from the model and need return a NDArray, we don't have handle in this case.