This is an automated email from the ASF dual-hosted git repository. reminisce pushed a change to branch numpy in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.
discard 29281cd [Numpy] fix test_operator_gpu.test_upsampling_bilinear_with_type (#14557) discard cc078c2 [Numpy] Misc fix (#14612) discard 5ac0b4a Fix pooling_v1 and deformable_convolution param initialization (#14577) discard 2b1599f Fix cpp package build after using new shape definition (#14554) discard c6bab57 fix R-package (#14536) discard 12018d2 fix concat and slice (#14549) discard db37dd9 [numpy] Fix numpy import in python2 (#14537) discard fcefc5a [numpy] Fix test_dynamic_shape.test_dynamic_shape (#14538) discard 38f2e06 Fix a bug to pass the test in test_contrib_rnn (#14520) discard 64c61b9 [numpy] Fix unit tests after introducing numpy compatible shapes (#14487) discard 47e2348 [WIP] Use new shape definition (#14453) discard 0731af7 [Numpy] Change semantics of ndim for operators in `src/operator/contrib` (#14409) discard b4497e7 [numpy] Shape support scalar tensor (#14315) new ce99e49 Cudnn conv dgrad algo filtering (#14310) new 4432af1 [MXNET-1226] add Docs update for MXNet Java (#14395) new ae55b75 fix Makefile (#14424) new 9fd3153 [MXNET-1291] solve pylint errors in examples with issue no.12205 (#13938) new 88b3741 Disables flaky TestStochasticTiming_2D test (#14412) new b077965 Add dtype visualization to plot_network (#14066) new 74c2274 Support multi-threading for Custom Operator (#14363) new d1fcda9 Fix entropy for uint8 (#14150) new d001eaf what's new - add 1.4.0 release (#14435) new 43173f5 moveaxis operator now accepts negative indices and sequence of ints as well. (#14321) new 226212b Add repr for SymbolBlock (#14423) new a091d36 temporarily disable integ tests with a dependency on origami repo (#14448) new f602b0d fix OOM error during resource allocation (#14444) new 63ed258 Correct update count with Gluon trainer and update_on_kvstore=False (#14377) new c2f939f Update MKL-DNN to v0.18 release (was: fix the Dense layer issue) (#13668) new 020e832 Speedup _contrib_index_copy (#14359) new ab5b44c Fix crashes on visualization (#14425) new ed77d6d add contributors from intel (#14455) new d671528 begin=end not a valid input (#14403) new 3ab1dec Fix memory leak for size-zero ndarray (#14365) new c56c146 [Doc] Start the tutorials for MKL-DNN backend (#14202) new c31bb7e Enforce determinism for backwards compatibility checker (#14463) new f838f67 [MKL-DNN] Enable s8 support for inner product and 3d input with flatten=false (#14466) new 56b7b67 Fixes the test_sgld (#14473) new f98820c Revert "Fix memory leak for size-zero ndarray (#14365)" (#14477) new 4b1811c fix custom operation in fork (#14451) new 3b28e62 Change Straight Dope to Dive into Deep Learning (#14465) new 95d4680 Added link to landing page for Java examples (#14481) new a01bdee Fixes test_operator_gpu.test_multinomial_generator (#14475) new a88c562 [MXNET-949] Module API to Gluon API tutorial (#12542) new 29e13b4 Fixed tutorial warnings (#14472) new 056fce4 Add examples of running MXNet with Horovod (#14286) new a9458ca Fixes for CI downloads (#14504) new f8a0dbc Enhance PartitionGraph (#14277) new 092af36 [MXNET-1285] Draw bounding box with Scala/Java Image API (#14474) new 651a6c0 reenable the test (#14483) new c4cd49c Fix script retrieval (#14519) new 3d20f2a add filter to warnings (#14532) new 67c10f9 Adds context parameter to check_rnn_layer_forward calls in test_lstmp (#14529) new 5d2a451 Performance improving for MKL-DNN Quantized FullyConnected (#14528) new 09daf22 speedup SequenceMask on GPU (#14445) new 645c778 Tidy up storage allocation and deallocation (#14480) new 102b46f Memory fixes. Resolves #10867, and resolves #14080 (#14372) new 84c2ae1 Remove unnecessary "also" in README.md (#14543) new b20f08b [clojure]: add comp-metric based on CompositeEvalMetric (#14553) new 9f5dfbf Chouffe/clojure fix tests (#14531) new 4d04238 [clojure][image] add draw-bounding-box interop (#14533) new 5f19362 fix tests (#14565) new 8c2a25f Enhance subgraph API (#14113) new 4075212 Do not touch GPU 0 during ReleaseAll (#14550) new b6eac1d Change CUB submodule to track Nvidia CUB project. (#13322) new 09ba8be Fixes static build script for cub directory rename (#14578) new 33b6543 example/ssd/evaluate/eval_metric.py (#14561) new e2f5b47 Support SyncBatchNorm5D (#14542) new 6392666 Disable Flaky Test test_poisson_generator (#14540) new 9e4ee99 [MXNET-1357] Fix the cpp-examples to add exception handling (#14441) new dde77d4 Updates gpu tests to use CUDNN_VERSION supplied by the environment but default to 7.0.3 if not set (#14595) new 214ce63 fix build cpp examples option (#14562) new a6fd0e9 Fix flaky test poisson generator & test_negative_binomial_generator (#14571) new 6478691 Fixing unintentional variable overloading (#14438) new b482a44 [MXNET-1379] update reshape operator (#14600) new 28fe552 Updated documentation about nightly tests (#14493) new 53743d6 set _scale in Trainer using optimizer rescale_grad (#14593) new 5241c1b Add Gluon Transformer Crop (#14259) new 7983ef1 fix quantize graph pass (#14605) new b3ab101 GELU (#14449) new a80177c [numpy] Shape support scalar tensor (#14315) new 6340f6d [Numpy] Change semantics of ndim for operators in `src/operator/contrib` (#14409) new 9aeac83 [WIP] Use new shape definition (#14453) new 42cf92c [numpy] Fix unit tests after introducing numpy compatible shapes (#14487) new f10b1ab Fix a bug to pass the test in test_contrib_rnn (#14520) new 77a508b [numpy] Fix test_dynamic_shape.test_dynamic_shape (#14538) new 808d036 [numpy] Fix numpy import in python2 (#14537) new 50de181 fix concat and slice (#14549) new 387141e fix R-package (#14536) new 38e1699 Fix cpp package build after using new shape definition (#14554) new 6292add Fix pooling_v1 and deformable_convolution param initialization (#14577) new cc6a63b [Numpy] Misc fix (#14612) new 921045f [Numpy] fix test_operator_gpu.test_upsampling_bilinear_with_type (#14557) This update added new revisions after undoing existing revisions. That is to say, some revisions that were in the old version of the branch are not in the new version. This situation occurs when a user --force pushes a change and generates a repository containing something like this: * -- * -- B -- O -- O -- O (29281cd) \ N -- N -- N refs/heads/numpy (921045f) You should already have received notification emails for all of the O revisions, and so the following emails describe only the N revisions from the common base, B. Any revisions marked "omit" are not gone; other references still refer to them. Any revisions marked "discard" are gone forever. The 9467 revisions listed above as "new" are entirely new to this repository and will be described in separate emails. The revisions listed as "add" were already present in the repository and have only been added to this reference. Summary of changes: .gitmodules | 6 +- 3rdparty/cub | 1 - 3rdparty/mkldnn | 2 +- 3rdparty/nvidia_cub | 1 + CMakeLists.txt | 5 +- CONTRIBUTORS.md | 18 + LICENSE | 4 +- MKLDNN_README.md | 314 +------------- Makefile | 6 +- NEWS.md | 2 +- R-package/tests/testthat/get_data.R | 11 +- R-package/vignettes/CatsDogsFinetune.Rmd | 4 +- .../classifyRealImageWithPretrainedModel.Rmd | 2 +- README.md | 7 +- ci/docker/install/ubuntu_mklml.sh | 2 +- ci/docker/runtime_functions.sh | 12 +- cmake/DownloadMKLML.cmake | 14 +- contrib/clojure-package/integration-tests.sh | 2 +- .../src/org/apache/clojure_mxnet/eval_metric.clj | 18 +- .../src/org/apache/clojure_mxnet/image.clj | 66 +++ .../src/org/apache/clojure_mxnet/util.clj | 17 +- .../clojure-package/test/dev/generator_test.clj | 87 ++-- .../org/apache/clojure_mxnet/eval_metric_test.clj | 9 + .../org/apache/clojure_mxnet/executor_test.clj | 8 +- .../clojure_mxnet/infer/imageclassifier_test.clj | 12 +- .../test/org/apache/clojure_mxnet/module_test.clj | 38 +- .../test/org/apache/clojure_mxnet/ndarray_test.clj | 12 +- .../org/apache/clojure_mxnet/operator_test.clj | 12 +- .../test/org/apache/clojure_mxnet/symbol_test.clj | 10 +- .../test/org/apache/clojure_mxnet/util_test.clj | 4 +- cpp-package/CMakeLists.txt | 2 +- cpp-package/example/alexnet.cpp | 2 + cpp-package/example/charRNN.cpp | 3 + cpp-package/example/feature_extract/run.sh | 7 +- cpp-package/example/get_data.sh | 61 ++- cpp-package/example/googlenet.cpp | 2 + cpp-package/example/inception_bn.cpp | 2 + .../inference/unit_test_inception_inference.sh | 4 +- cpp-package/example/lenet.cpp | 3 + cpp-package/example/lenet_with_mxdataiter.cpp | 2 + cpp-package/example/mlp.cpp | 3 + cpp-package/example/mlp_cpu.cpp | 2 + cpp-package/example/mlp_csv.cpp | 2 + cpp-package/example/mlp_gpu.cpp | 2 + cpp-package/example/resnet.cpp | 2 + cpp-package/example/test_score.cpp | 4 +- cpp-package/example/utils.h | 9 + docs/_static/mxnet-theme/navbar.html | 4 +- docs/api/perl/index.md | 2 +- docs/community/ecosystem.md | 2 +- docs/faq/env_var.md | 3 + docs/faq/perf.md | 2 +- docs/gluon/index.md | 14 +- docs/install/index.md | 6 +- docs/install/java_setup.md | 32 +- docs/install/ubuntu_setup.md | 2 +- docs/install/windows_setup.md | 4 +- docs/tutorials/c++/subgraphAPI.md | 40 +- docs/tutorials/gluon/customop.md | 20 + .../gluon/gluon_from_experiment_to_deployment.md | 4 +- docs/tutorials/gluon/hybrid.md | 8 +- docs/tutorials/gluon/save_load_params.md | 5 +- docs/tutorials/index.md | 9 +- docs/tutorials/java/mxnet_java_on_intellij.md | 27 +- .../tutorials/mkldnn/MKLDNN_README.md | 14 +- docs/tutorials/{vision => mkldnn}/index.md | 0 docs/tutorials/onnx/fine_tuning_gluon.md | 5 +- docs/tutorials/onnx/inference_on_onnx_model.md | 6 +- docs/tutorials/python/matrix_factorization.md | 289 ------------- docs/tutorials/python/module_to_gluon.md | 365 +++++++++++++++++ example/README.md | 1 + example/bayesian-methods/bdk_demo.py | 2 +- example/capsnet/capsulenet.py | 7 +- example/ctc/multiproc_data.py | 1 - example/distributed_training-horovod/README.md | 201 +++++++++ .../distributed_training-horovod/gluon_mnist.py | 186 +++++++++ .../distributed_training-horovod/module_mnist.py | 162 ++++++++ .../resnet50_imagenet.py | 453 +++++++++++++++++++++ example/gluon/dc_gan/dcgan.py | 67 +-- example/gluon/lstm_crf/lstm_crf.py | 62 +-- example/quantization/imagenet_gen_qsym_mkldnn.py | 2 - example/recommenders/matrix_fact.py | 8 +- example/ssd/evaluate/eval_metric.py | 2 +- include/mxnet/c_api_test.h | 2 +- include/mxnet/ndarray.h | 8 +- julia/models/Inception/get.sh | 2 +- matlab/get_inception_model.sh | 2 +- mkldnn.mk | 5 + perl-package/AI-MXNet/lib/AI/MXNet.pm | 12 +- python/mxnet/contrib/quantization.py | 12 +- python/mxnet/gluon/block.py | 8 + python/mxnet/gluon/data/vision/transforms.py | 61 +++ python/mxnet/gluon/nn/activations.py | 21 +- python/mxnet/gluon/trainer.py | 8 +- python/mxnet/image/image.py | 2 +- python/mxnet/ndarray/ndarray.py | 33 +- python/mxnet/optimizer/optimizer.py | 17 +- python/mxnet/test_utils.py | 2 +- python/mxnet/visualization.py | 71 +++- scala-package/.gitignore | 1 + .../src/main/scala/org/apache/mxnet/Executor.scala | 47 ++- .../src/main/scala/org/apache/mxnet/Image.scala | 54 +++ .../src/main/scala/org/apache/mxnet/LibInfo.scala | 5 +- .../src/main/scala/org/apache/mxnet/NDArray.scala | 13 +- .../main/scala/org/apache/mxnet/Optimizer.scala | 20 +- .../scala/org/apache/mxnet/ResourceScope.scala | 21 +- .../src/main/scala/org/apache/mxnet/Symbol.scala | 21 +- .../scala/org/apache/mxnet/javaapi/Image.scala | 46 ++- .../mxnet/module/DataParallelExecutorGroup.scala | 11 +- .../scala/org/apache/mxnet/optimizer/Adam.scala | 101 +++-- .../java/org/apache/mxnet/javaapi/ImageTest.java | 20 +- .../test/scala/org/apache/mxnet/ImageSuite.scala | 21 + .../test/scala/org/apache/mxnet/NDArraySuite.scala | 8 +- .../org/apache/mxnet/ResourceScopeSuite.scala | 33 ++ .../javaapi/infer/objectdetector/README.md | 2 +- .../infer/objectdetector/SSDClassifierExample.java | 41 +- .../org/apache/mxnet/infer/ObjectDetector.scala | 2 +- .../mxnet/infer/javaapi/ObjectDetectorOutput.scala | 4 +- .../infer/javaapi/ObjectDetectorOutputTest.java | 4 +- .../main/native/org_apache_mxnet_native_c_api.cc | 13 +- .../main/native/org_apache_mxnet_native_c_api.h | 8 +- src/c_api/c_api.cc | 1 + src/c_api/c_api_symbolic.cc | 17 +- src/c_api/c_api_test.cc | 21 +- src/executor/graph_executor.cc | 347 ++++++++++------ src/initialize.cc | 5 + src/ndarray/ndarray.cc | 8 +- src/operator/contrib/index_copy-inl.h | 93 +---- src/operator/contrib/index_copy.cc | 120 ++++++ src/operator/contrib/index_copy.cu | 110 ++++- src/operator/contrib/sync_batch_norm-inl.h | 31 +- src/operator/custom/custom-inl.h | 72 ++-- src/operator/custom/custom.cc | 5 - src/operator/image/crop-inl.h | 190 +++++++++ src/operator/image/{resize.cc => crop.cc} | 76 ++-- src/operator/leaky_relu-inl.h | 21 +- src/operator/mshadow_op.h | 19 + src/operator/nn/batch_norm.cu | 4 +- src/operator/nn/cudnn/cudnn_batch_norm-inl.h | 14 +- src/operator/nn/cudnn/cudnn_convolution-inl.h | 25 +- src/operator/nn/cudnn/cudnn_deconvolution-inl.h | 24 +- src/operator/nn/fully_connected-inl.h | 6 + src/operator/nn/fully_connected.cc | 10 +- src/operator/nn/mkldnn/mkldnn_base.cc | 2 + src/operator/operator_tune.cc | 2 + .../mkldnn/mkldnn_quantized_fully_connected.cc | 32 +- src/operator/quantization/quantize_graph_pass.cc | 12 +- .../quantization/quantized_fully_connected.cc | 55 ++- .../quantization/quantized_fully_connected.cu | 5 + src/operator/sequence_mask-inl.h | 79 +--- src/operator/sequence_mask.cc | 64 +++ src/operator/sequence_mask.cu | 59 +++ .../{partition_graph.cc => build_subgraph.cc} | 387 ++++++++---------- ...property.cc => default_subgraph_property_v2.cc} | 25 +- ...rty.cc => mkldnn_conv_post_quantize_property.h} | 35 +- ...dnn_conv_property.cc => mkldnn_conv_property.h} | 22 +- src/operator/subgraph/mkldnn/mkldnn_fc.cc | 56 +-- ...perty.cc => mkldnn_fc_post_quantize_property.h} | 28 +- ...{mkldnn_fc_property.cc => mkldnn_fc_property.h} | 39 +- .../mkldnn_post_quantize_align_scale_property.h | 164 ++++++++ ...ldnn_conv-inl.h => mkldnn_subgraph_property.cc} | 25 +- src/operator/subgraph/subgraph_property.h | 218 +++++++++- src/operator/tensor/matrix_op-inl.h | 67 ++- src/resource.cc | 30 +- src/storage/cpu_device_storage.h | 25 +- src/storage/cpu_shared_storage_manager.h | 7 +- src/storage/gpu_device_storage.h | 19 +- src/storage/naive_storage_manager.h | 2 +- src/storage/pinned_memory_storage.h | 21 +- src/storage/pooled_storage_manager.h | 36 +- src/storage/storage.cc | 12 +- src/storage/storage_manager.h | 11 +- tests/cpp/include/test_util.h | 8 +- tests/cpp/operator/batchnorm_test.cc | 4 +- tests/cpp/operator/mkldnn_test.cc | 2 +- tests/cpp/storage/storage_test.cc | 15 + tests/nightly/README.md | 10 +- .../model_backward_compat_checker.sh | 3 +- .../train_mxnet_legacy_models.sh | 1 + tests/python/gpu/test_gluon_gpu.py | 225 ++++------ tests/python/gpu/test_kvstore_gpu.py | 7 +- tests/python/gpu/test_operator_gpu.py | 26 ++ tests/python/mkl/test_subgraph.py | 114 +++++- tests/python/quantization/test_quantization.py | 123 +++++- tests/python/unittest/test_gluon.py | 143 ++++++- tests/python/unittest/test_gluon_data_vision.py | 75 +++- tests/python/unittest/test_gluon_trainer.py | 21 +- tests/python/unittest/test_ndarray.py | 72 +++- tests/python/unittest/test_operator.py | 203 +++++++++ tests/python/unittest/test_optimizer.py | 5 +- tests/python/unittest/test_random.py | 10 +- tests/python/unittest/test_subgraph_op.py | 89 ++-- tests/python/unittest/test_viz.py | 2 + tests/tutorials/test_sanity_tutorials.py | 2 + tests/tutorials/test_tutorials.py | 7 +- tools/dependencies/README.md | 2 +- tools/dependencies/curl.sh | 4 +- tools/dependencies/eigen.sh | 4 +- tools/dependencies/libpng.sh | 4 +- tools/dependencies/libtiff.sh | 4 +- tools/dependencies/libturbojpeg.sh | 4 +- tools/dependencies/libz.sh | 4 +- tools/dependencies/lz4.sh | 4 +- tools/dependencies/make_shared_dependencies.sh | 28 +- tools/dependencies/openblas.sh | 4 +- tools/dependencies/opencv.sh | 4 +- tools/dependencies/openssl.sh | 4 +- tools/dependencies/protobuf.sh | 4 +- tools/dependencies/zmq.sh | 4 +- tools/staticbuild/build_lib.sh | 2 +- 210 files changed, 5357 insertions(+), 2245 deletions(-) delete mode 160000 3rdparty/cub create mode 160000 3rdparty/nvidia_cub copy MKLDNN_README.md => docs/tutorials/mkldnn/MKLDNN_README.md (92%) copy docs/tutorials/{vision => mkldnn}/index.md (100%) delete mode 100644 docs/tutorials/python/matrix_factorization.md create mode 100644 docs/tutorials/python/module_to_gluon.md create mode 100644 example/distributed_training-horovod/README.md create mode 100644 example/distributed_training-horovod/gluon_mnist.py create mode 100644 example/distributed_training-horovod/module_mnist.py create mode 100644 example/distributed_training-horovod/resnet50_imagenet.py create mode 100644 src/operator/image/crop-inl.h copy src/operator/image/{resize.cc => crop.cc} (51%) rename src/operator/subgraph/{partition_graph.cc => build_subgraph.cc} (69%) copy src/operator/subgraph/{default_subgraph_property.cc => default_subgraph_property_v2.cc} (73%) rename src/operator/subgraph/mkldnn/{mkldnn_conv_post_quantize_property.cc => mkldnn_conv_post_quantize_property.h} (84%) rename src/operator/subgraph/mkldnn/{mkldnn_conv_property.cc => mkldnn_conv_property.h} (93%) rename src/operator/subgraph/mkldnn/{mkldnn_fc_post_quantize_property.cc => mkldnn_fc_post_quantize_property.h} (91%) rename src/operator/subgraph/mkldnn/{mkldnn_fc_property.cc => mkldnn_fc_property.h} (85%) create mode 100644 src/operator/subgraph/mkldnn/mkldnn_post_quantize_align_scale_property.h copy src/operator/subgraph/mkldnn/{mkldnn_conv-inl.h => mkldnn_subgraph_property.cc} (60%)