[GitHub] [incubator-mxnet] apeforest commented on issue #15499: Improve diagnose.py, adding build features info and binary library path.

[GitBox]

apeforest commented on issue #15499: Improve diagnose.py, adding build features 
info and binary library path.
URL: https://github.com/apache/incubator-mxnet/pull/15499#issuecomment-511091560
 
 
   One minor comment: the ✖ etc. symbols were not recognized when I import the 
txt file into MS office or Google Doc. Can we replace these unicode symbols by 
simple ascii, like Y/N?


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15499: Improve diagnose.py, adding build features info and binary library path.

[GitBox]

apeforest commented on a change in pull request #15499: Improve diagnose.py, 
adding build features info and binary library path.
URL: https://github.com/apache/incubator-mxnet/pull/15499#discussion_r303192850
 
 

 ##
 File path: tools/diagnose.py
 ##
 @@ -105,13 +110,20 @@ def check_mxnet():
 mx_dir = os.path.dirname(mxnet.__file__)
 print('Directory:', mx_dir)
 commit_hash = os.path.join(mx_dir, 'COMMIT_HASH')
-with open(commit_hash, 'r') as f:
-ch = f.read().strip()
-print('Commit Hash   :', ch)
+if os.path.exists(commit_hash):
+with open(commit_hash, 'r') as f:
+ch = f.read().strip()
+print('Commit Hash   :', ch)
+else:
+print('Commit hash file "{}" not found. Not installed from 
pre-built package or built from source.'.format(commit_hash))
+print('Library  :', mxnet.libinfo.find_lib_path())
+try:
+print('Build features:')
+print(get_build_features_str())
 
 Review comment:
   This will fail if user use this script to diagnose an mxnet script before 
1.5.0 release.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15499: Improve diagnose.py, adding build features info and binary library path.

[GitBox]

apeforest commented on a change in pull request #15499: Improve diagnose.py, 
adding build features info and binary library path.
URL: https://github.com/apache/incubator-mxnet/pull/15499#discussion_r303192850
 
 

 ##
 File path: tools/diagnose.py
 ##
 @@ -105,13 +110,20 @@ def check_mxnet():
 mx_dir = os.path.dirname(mxnet.__file__)
 print('Directory:', mx_dir)
 commit_hash = os.path.join(mx_dir, 'COMMIT_HASH')
-with open(commit_hash, 'r') as f:
-ch = f.read().strip()
-print('Commit Hash   :', ch)
+if os.path.exists(commit_hash):
+with open(commit_hash, 'r') as f:
+ch = f.read().strip()
+print('Commit Hash   :', ch)
+else:
+print('Commit hash file "{}" not found. Not installed from 
pre-built package or built from source.'.format(commit_hash))
+print('Library  :', mxnet.libinfo.find_lib_path())
+try:
+print('Build features:')
+print(get_build_features_str())
 
 Review comment:
   This will fail if user use this script to diagnose an mxnet script before 
1.5.0 release.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] intgogo opened a new issue #15529: MXNET_CUDNN_AUTOTUNE_DEFAULT problems

[GitBox]

intgogo opened a new issue #15529: MXNET_CUDNN_AUTOTUNE_DEFAULT problems
URL: https://github.com/apache/incubator-mxnet/issues/15529
 
 
   When I set MXNET_CUDNN_AUTOTUNE_DEFAULT =1, it won't stop.WTF~
   
   ```
   [12:39:44] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:03] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:18] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:23] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:33] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:39] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:45] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:52] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:40:59] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:05] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:14] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:19] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:26] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:31] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:38] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:45] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:53] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:41:59] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 
disable)
   [12:42:07] 
/root/lilw/mxnet/src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running 
performance tests to find the best convolution algorithm, this can take a 
while... (set the environment variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to 

[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #15529: MXNET_CUDNN_AUTOTUNE_DEFAULT problems

[GitBox]

mxnet-label-bot commented on issue #15529: MXNET_CUDNN_AUTOTUNE_DEFAULT problems
URL: 
https://github.com/apache/incubator-mxnet/issues/15529#issuecomment-511088398
 
 
   Hey, this is the MXNet Label Bot. 
Thank you for submitting the issue! I will try and suggest some labels so 
that the appropriate MXNet community members can help resolve it. 
Here are my recommended labels: Cuda, Bug


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303189996
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,763 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_to_file_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  void tensor_info_to_string(StreamType* os) {
+const int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303189612
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,763 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_to_file_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  void tensor_info_to_string(StreamType* os) {
+const int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] anirudh2290 opened a new pull request #15528: [WIP] Add AMP Conversion support for BucketingModule

[GitBox]

anirudh2290 opened a new pull request #15528: [WIP] Add AMP Conversion support 
for BucketingModule
URL: https://github.com/apache/incubator-mxnet/pull/15528
 
 
   ## Description ##
   - Add conversion APIs for bucketing module
   - Add support for saving and loading BucketingModule from a checkpoint
   - Add support for loading BucketingModule from a dict of bucket keys to 
symbols
   - Add tests
   
   ## Checklist ##
   ### Essentials ###
   Please feel free to remove inapplicable items for your PR.
   - [ ] The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to 
the relevant [JIRA issue](https://issues.apache.org/jira/projects/MXNET/issues) 
created (except PRs with tiny changes)
   - [ ] Changes are complete (i.e. I finished coding on this PR)
   - [ ] All changes have test coverage:
   - Unit tests are added for small changes to verify correctness (e.g. adding 
a new operator)
   - Nightly tests are added for complicated/long-running ones (e.g. changing 
distributed kvstore)
   - Build tests will be added for build configuration changes (e.g. adding a 
new build option with NCCL)
   - [ ] Code is well-documented: 
   - For user-facing API changes, API doc string has been updated. 
   - For new C++ functions in header files, their functionalities and arguments 
are documented. 
   - For new examples, README.md is added to explain the what the example does, 
the source of the dataset, expected performance on test set and reference to 
the original paper if applicable
   - Check the API doc at 
http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
   - [ ] To the my best knowledge, examples are either not affected by this 
change, or have been fixed to be compatible with this change
   
   ### Changes ###
   - [ ] Feature1, tests, (and when applicable, API doc)
   - [ ] Feature2, tests, (and when applicable, API doc)
   
   ## Comments ##
   - If this change is a backward incompatible change, why must this change be 
made.
   - Interesting edge cases to note here
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] anirudh2290 commented on issue #13335: MXPredCreate lost dtype

[GitBox]

anirudh2290 commented on issue #13335: MXPredCreate lost dtype
URL: 
https://github.com/apache/incubator-mxnet/issues/13335#issuecomment-511080169
 
 
   Resolved by #15245 


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] anirudh2290 closed issue #13335: MXPredCreate lost dtype

[GitBox]

anirudh2290 closed issue #13335: MXPredCreate lost dtype
URL: https://github.com/apache/incubator-mxnet/issues/13335
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet] branch master updated (cbb6f7f -> d677d1a)

[anirudh2290]

This is an automated email from the ASF dual-hosted git repository.

anirudh2290 pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.


from cbb6f7f  Docs: Fix misprints (#15505)
 add d677d1a  FP16 Support for C Predict API (#15245)

No new revisions were added by this update.

Summary of changes:
 amalgamation/python/mxnet_predict.py | 131 ++--
 include/mxnet/c_predict_api.h|  65 
 src/c_api/c_predict_api.cc   | 140 +--
 tests/python/gpu/test_predictor.py   | 128 
 4 files changed, 454 insertions(+), 10 deletions(-)
 create mode 100644 tests/python/gpu/test_predictor.py

[GitHub] [incubator-mxnet] anirudh2290 merged pull request #15245: FP16 Support for C Predict API

[GitBox]

anirudh2290 merged pull request #15245: FP16 Support for C Predict API
URL: https://github.com/apache/incubator-mxnet/pull/15245
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] anirudh2290 closed issue #14159: [Feature Request] Support fp16 for C Predict API

[GitBox]

anirudh2290 closed issue #14159: [Feature Request] Support fp16 for C Predict 
API
URL: https://github.com/apache/incubator-mxnet/issues/14159
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] braindotai removed a comment on issue #12185: from_logits definition seems different from what is expected?

[GitBox]

braindotai removed a comment on issue #12185: from_logits definition seems 
different from what is expected?
URL: 
https://github.com/apache/incubator-mxnet/issues/12185#issuecomment-510939653
 
 
   [Check this 
out](https://www.tensorflow.org/api_docs/python/tf/nn/softmax_cross_entropy_with_logits)
   Here it says "logits: Per-label activations, typically a linear output", 
which means `nd.dot(x, w) + b` in terms of MXNet. 
   There are actually two versions of logits, first is simply the linear 
layer(as mentions in the above link), and second is the unscaled log 
probabilities. That is the reason Tensorflow provided 2 versions of 
"softmax_cross_entropy_with_logits".
   
   In MXNet mx.gluon.loss.SoftmaxCrossEntropyLoss accepts the linear 
output(`nd.dot(x, w) + b`) as output in the argument. 
   You can check 
[here](https://gluon.mxnet.io/chapter02_supervised-learning/softmax-regression-gluon.html),
 the layer definition is `net = gluon.nn.Dense(num_outputs)`,
   defining the loss as `gluon.loss.SoftmaxCrossEntropyLoss()` and then 
calculating loss as
   ```python
   output = net(data)
   loss = softmax_cross_entropy(output, label)
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet] branch master updated: Docs: Fix misprints (#15505)

[wkcn]

This is an automated email from the ASF dual-hosted git repository.

wkcn pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git


The following commit(s) were added to refs/heads/master by this push:
 new cbb6f7f  Docs: Fix misprints (#15505)
cbb6f7f is described below

commit cbb6f7fd6e297c17fd267b29174a4ed29100c757
Author: Ruslan Baratov 
AuthorDate: Sat Jul 13 03:26:01 2019 +0300

Docs: Fix misprints (#15505)

* Docs: Fix 'bahavior' -> 'behavior'

* Docs: Fix 'the the' -> 'the'

* retrigger CI

* retrigger CI
---
 NEWS.md   | 4 ++--
 R-package/R/viz.graph.R   | 4 ++--
 contrib/clojure-package/README.md | 2 +-
 docs/api/python/gluon/gluon.md| 2 +-
 docs/install/windows_setup.md | 2 +-
 docs/tutorials/mkldnn/MKLDNN_README.md| 2 +-
 example/gan/CGAN_mnist_R/README.md| 2 +-
 include/mxnet/ndarray.h   | 2 +-
 perl-package/AI-MXNet/lib/AI/MXNet/Gluon.pm   | 2 +-
 perl-package/AI-MXNet/lib/AI/MXNet/Module/Base.pm | 2 +-
 python/mxnet/contrib/onnx/onnx2mx/_op_translations.py | 2 +-
 python/mxnet/gluon/data/dataloader.py | 2 +-
 python/mxnet/module/base_module.py| 2 +-
 python/mxnet/module/python_module.py  | 2 +-
 .../core/src/main/scala/org/apache/mxnet/module/BaseModule.scala  | 2 +-
 .../org/apache/mxnetexamples/javaapi/infer/objectdetector/README.md   | 2 +-
 .../java/org/apache/mxnetexamples/javaapi/infer/predictor/README.md   | 2 +-
 .../scala/org/apache/mxnetexamples/infer/objectdetector/README.md | 2 +-
 src/operator/tensor/diag_op-inl.h | 2 +-
 src/operator/tensor/matrix_op.cc  | 2 +-
 tools/staticbuild/README.md   | 4 ++--
 21 files changed, 24 insertions(+), 24 deletions(-)

diff --git a/NEWS.md b/NEWS.md
index 59f8de8..ee8a73c 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -678,8 +678,8 @@ This fixes an buffer overflow detected by ASAN.
   This PR adds or updates the docs for the infer_range feature.
 
   Clarifies the param in the C op docs
-  Clarifies the param in the the Scala symbol docs
-  Adds the param for the the Scala ndarray docs
+  Clarifies the param in the Scala symbol docs
+  Adds the param for the Scala ndarray docs
   Adds the param for the Python symbol docs
   Adds the param for the Python ndarray docs
 
diff --git a/R-package/R/viz.graph.R b/R-package/R/viz.graph.R
index 5804372..ab876af 100644
--- a/R-package/R/viz.graph.R
+++ b/R-package/R/viz.graph.R
@@ -34,7 +34,7 @@
 #' @param symbol a \code{string} representing the symbol of a model.
 #' @param shape a \code{numeric} representing the input dimensions to the 
symbol.
 #' @param direction a \code{string} representing the direction of the graph, 
either TD or LR.
-#' @param type a \code{string} representing the rendering engine of the the 
graph, either graph or vis.
+#' @param type a \code{string} representing the rendering engine of the graph, 
either graph or vis.
 #' @param graph.width.px a \code{numeric} representing the size (width) of the 
graph. In pixels
 #' @param graph.height.px a \code{numeric} representing the size (height) of 
the graph. In pixels
 #'
@@ -169,4 +169,4 @@ graph.viz <- function(symbol, shape=NULL, direction="TD", 
type="graph", graph.wi
   return(graph_render)
 }
 
-globalVariables(c("color", "shape", "label", "id", ".", "op"))
\ No newline at end of file
+globalVariables(c("color", "shape", "label", "id", ".", "op"))
diff --git a/contrib/clojure-package/README.md 
b/contrib/clojure-package/README.md
index 7566ade..7bb417e 100644
--- a/contrib/clojure-package/README.md
+++ b/contrib/clojure-package/README.md
@@ -237,7 +237,7 @@ If you are having trouble getting started or have a 
question, feel free to reach
 There are quite a few examples in the examples directory. To use.
 
 `lein install` in the main project
-`cd` in the the example project of interest
+`cd` in the example project of interest
 
 There are README is every directory outlining instructions.
 
diff --git a/docs/api/python/gluon/gluon.md b/docs/api/python/gluon/gluon.md
index c063a71..19e462e 100644
--- a/docs/api/python/gluon/gluon.md
+++ b/docs/api/python/gluon/gluon.md
@@ -28,7 +28,7 @@
 
 The Gluon package is a high-level interface for MXNet designed to be easy to 
use, while keeping most of the flexibility of a low level API. Gluon supports 
both imperative and symbolic programming, making it easy to train complex 
models imperatively in Python and 

[GitHub] [incubator-mxnet] wkcn merged pull request #15505: Docs: Fix misprints

[GitBox]

wkcn merged pull request #15505: Docs: Fix misprints
URL: https://github.com/apache/incubator-mxnet/pull/15505
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] wkcn commented on issue #14894: Accelerate ROIPooling layer

[GitBox]

wkcn commented on issue #14894: Accelerate ROIPooling layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#issuecomment-511070139
 
 
   @larroy I think the main reason is that the indices of the maximum numbers 
are saved in this PR.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya commented on issue #15516: Fix memory leak reported by ASAN in NNVM to ONNX conversion

[GitBox]

ChaiBapchya commented on issue #15516: Fix memory leak reported by ASAN in NNVM 
to ONNX conversion
URL: https://github.com/apache/incubator-mxnet/pull/15516#issuecomment-511069536
 
 
   Gotcha! Makes sense!


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy edited a comment on issue #14894: Accelerate ROIPooling layer

[GitBox]

larroy edited a comment on issue #14894: Accelerate ROIPooling layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#issuecomment-511066311
 
 
   This is great, is the performance increase only due to type changes as in 
the description?


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy commented on issue #14894: Accelerate ROIPooling layer

[GitBox]

larroy commented on issue #14894: Accelerate ROIPooling layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#issuecomment-511066311
 
 
   This is great, is the performance increase only due to type changes as in 
the description? 


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] KellenSunderland opened a new pull request #15527: Small typo fixes in batch_norm-inl.h

[GitBox]

KellenSunderland opened a new pull request #15527: Small typo fixes in 
batch_norm-inl.h
URL: https://github.com/apache/incubator-mxnet/pull/15527
 
 
   ## Description ##
   Small typo fixes in batch_norm-inl.h


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet] branch KellenSunderland-patch-2 created (now 6c38f3e)

[kellen]

This is an automated email from the ASF dual-hosted git repository.

kellen pushed a change to branch KellenSunderland-patch-2
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.


  at 6c38f3e  Small typo fixes in batch_norm-inl.h

No new revisions were added by this update.

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

ChaiBapchya commented on a change in pull request #15490: Utility to help 
developers debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303175784
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,763 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_to_file_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  void tensor_info_to_string(StreamType* os) {
+const int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor 

[GitHub] [incubator-mxnet] zhreshold commented on issue #14894: Accelerate ROIPooling layer

[GitBox]

zhreshold commented on issue #14894: Accelerate ROIPooling layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#issuecomment-511060877
 
 
   Would be nice to apply to ROIAlign similarly since GluonCV have transitioned 
to use ROIAlign in recent networks.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet] branch master updated (2565fa2 -> 9c5acb4)

[sxjscience]

This is an automated email from the ASF dual-hosted git repository.

sxjscience pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.


from 2565fa2  fix nightly CI failure (#15452)
 add 9c5acb4  Accelerate ROIPooling layer (#14894)

No new revisions were added by this update.

Summary of changes:
 src/operator/roi_pooling-inl.h |  11 +++-
 src/operator/roi_pooling.cc| 128 ++---
 src/operator/roi_pooling.cu| 127 ++--
 3 files changed, 66 insertions(+), 200 deletions(-)

[GitHub] [incubator-mxnet] sxjscience merged pull request #14894: Accelerate ROIPooling layer

[GitBox]

sxjscience merged pull request #14894: Accelerate ROIPooling layer
URL: https://github.com/apache/incubator-mxnet/pull/14894
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector Tutorial

[GitBox]

ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector 
Tutorial
URL: https://github.com/apache/incubator-mxnet/pull/15517#discussion_r303174181
 
 

 ##
 File path: docs/faq/tensor_inspector_tutorial.md
 ##
 @@ -0,0 +1,164 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# Use TensorInspector to Help Debug Operators
+
+## Introduction
+
+When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, This utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
+
+
+## Usage 
+
+This utility locates in `src/common/tensor_inspector.h`. To use it in any 
operator code, just include `tensor_inspector`, construct an `TensorInspector` 
object, and call the APIs on that object. You can run any script that uses the 
operator you just modified then.
+
+The screenshot below shows a sample usage in 
`src/operator/nn/convolution-inl.h`.
+
+![Screen Shot 2019-07-08 at 5 03 46 
PM](https://user-images.githubusercontent.com/16669457/60850062-68690e00-a1a2-11e9-8268-033edde17aa4.png)
 
 Review comment:
   Would be great if you could give the image a recognizable name


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector Tutorial

[GitBox]

ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector 
Tutorial
URL: https://github.com/apache/incubator-mxnet/pull/15517#discussion_r303174854
 
 

 ##
 File path: docs/faq/tensor_inspector_tutorial.md
 ##
 @@ -0,0 +1,164 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# Use TensorInspector to Help Debug Operators
+
+## Introduction
+
+When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, This utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
+
+
+## Usage 
+
+This utility locates in `src/common/tensor_inspector.h`. To use it in any 
operator code, just include `tensor_inspector`, construct an `TensorInspector` 
object, and call the APIs on that object. You can run any script that uses the 
operator you just modified then.
+
+The screenshot below shows a sample usage in 
`src/operator/nn/convolution-inl.h`.
+
+![Screen Shot 2019-07-08 at 5 03 46 
PM](https://user-images.githubusercontent.com/16669457/60850062-68690e00-a1a2-11e9-8268-033edde17aa4.png)
+
+
+## Functionalities/APIs
+
+### Create a TensorInspector Object from Tensor, TBlob, and NDArray Objects
+
+You can create a `TensorInspector` object by passing in two things: 1) an 
object of type `Tensor`, `Tbob`, or `NDArray`, and 2) an `RunContext` object.
+
+Essentially, `TensorInspector` can be understood as a wrapper class around 
`TBlob`. Internally, the `Tensor`, `Tbob`, or `NDArray` object that you passed 
in will all be converted to a `TBlob` object. The `RunContext` object is used 
when the the tensor is a GPU tensor; in such case, we need to use the context 
information to copy the data from GPU memory to CPU/main memory.
+
+Below are the three constructors:
+
+```c++
+// Construct from Tensor object
+template
+TensorInspector(const  mshadow::Tensor& ts, const 
RunContext& ctx);
+
+// Construct from TBlob object
+TensorInspector(const TBlob& tb, const RunContext& ctx);
+
+// Construct from NDArray object
+TensorInspector(const NDArray& arr, const RunContext& ctx):
+```
+
+### Print Tensor Value (Static) 
+
+To print out the tensor value in a nicely structured way,  you can use this 
API:
+
+```c++
+void print_string();
+```
+
+This API will print the entire tensor to `std::cout` and preserve the shape 
(it supports all dimensions from 1 and up). You can copy the output and 
interpret it with any `JSON` loader. Also, on the last line of the output you 
can find some useful information about the tensor. Refer to the case below, we 
are able to know that this is a float-typed tensor with shape 20x1x5x5.
+
+![Screen Shot 2019-07-08 at 4 07 16 
PM](https://user-images.githubusercontent.com/16669457/60848554-d8c06100-a19b-11e9-9fe0-23e79a7a371a.png)
+
+If instead of printing the tensor to `std::cout`, you just need a `string`, 
you can use this API:
+```c++
+std::string void to_string();
+```
+
+### Interactively Print Tensor Value (Dynamic) 
+
+When debugging, situations might occur that at compilation time, you do not 
know which part of a tensor to inspect. Also, sometimes, it would be nice to 
pause the operator control flow to “zoom into” a specific, erroneous part of a 
tensor multiple times until you are satisfied. In this regard, you can use this 
API to interactively inspect the tensor:
+
+```c++
+void  interactive_print(std::string tag =  "") {
+```
+
+This API will set a "break point" in your code, so that you will enter a loop 
that will keep asking you for further command. In the API call, `tag` is an 
optional parameter to give the call a name, so that you can identify it when 
you have multiple `interactive_print()` calls in different parts of your code. 
A visit count will tell you for how many times have you stepped into this 
particular "break point", should this operator be called more than once. Note 
that all `interactive_print()` calls are properly locked, so you can use it in 
many different places without issues.
+
+![Screen Shot 2019-07-10 at 5 29 07 
PM](https://user-images.githubusercontent.com/16669457/61013632-5325e800-a338-11e9-90e6-607f17d81495.png)
+
+Refer the screenshot above, there are many useful commands available: you can 
type "e" to print out the entire tensor, ''d" to dump the tensor to file (see 
below), "b" to break from this command loop, and "s" to skip all future 
`interactive_print()`. Most importantly, in this screen, you can specify a part 
of the tensor that you are particularly interested in and want to print out. 
For example, for this 20x1x5x5 tensor, you can type in "0, 0" and presss enter 
to check the sub-tensor with shape 5x5 at coordinate (0, 0). 
+
+### Check Tensor Value
+
+Sometimes, developers might want to check if the tensor contains unexpected 
values which could be negative 

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector Tutorial

[GitBox]

ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector 
Tutorial
URL: https://github.com/apache/incubator-mxnet/pull/15517#discussion_r303174924
 
 

 ##
 File path: docs/faq/tensor_inspector_tutorial.md
 ##
 @@ -0,0 +1,164 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# Use TensorInspector to Help Debug Operators
+
+## Introduction
+
+When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, This utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
+
+
+## Usage 
+
+This utility locates in `src/common/tensor_inspector.h`. To use it in any 
operator code, just include `tensor_inspector`, construct an `TensorInspector` 
object, and call the APIs on that object. You can run any script that uses the 
operator you just modified then.
+
+The screenshot below shows a sample usage in 
`src/operator/nn/convolution-inl.h`.
+
+![Screen Shot 2019-07-08 at 5 03 46 
PM](https://user-images.githubusercontent.com/16669457/60850062-68690e00-a1a2-11e9-8268-033edde17aa4.png)
+
+
+## Functionalities/APIs
+
+### Create a TensorInspector Object from Tensor, TBlob, and NDArray Objects
+
+You can create a `TensorInspector` object by passing in two things: 1) an 
object of type `Tensor`, `Tbob`, or `NDArray`, and 2) an `RunContext` object.
+
+Essentially, `TensorInspector` can be understood as a wrapper class around 
`TBlob`. Internally, the `Tensor`, `Tbob`, or `NDArray` object that you passed 
in will all be converted to a `TBlob` object. The `RunContext` object is used 
when the the tensor is a GPU tensor; in such case, we need to use the context 
information to copy the data from GPU memory to CPU/main memory.
+
+Below are the three constructors:
+
+```c++
+// Construct from Tensor object
+template
+TensorInspector(const  mshadow::Tensor& ts, const 
RunContext& ctx);
+
+// Construct from TBlob object
+TensorInspector(const TBlob& tb, const RunContext& ctx);
+
+// Construct from NDArray object
+TensorInspector(const NDArray& arr, const RunContext& ctx):
+```
+
+### Print Tensor Value (Static) 
+
+To print out the tensor value in a nicely structured way,  you can use this 
API:
+
+```c++
+void print_string();
+```
+
+This API will print the entire tensor to `std::cout` and preserve the shape 
(it supports all dimensions from 1 and up). You can copy the output and 
interpret it with any `JSON` loader. Also, on the last line of the output you 
can find some useful information about the tensor. Refer to the case below, we 
are able to know that this is a float-typed tensor with shape 20x1x5x5.
+
+![Screen Shot 2019-07-08 at 4 07 16 
PM](https://user-images.githubusercontent.com/16669457/60848554-d8c06100-a19b-11e9-9fe0-23e79a7a371a.png)
+
+If instead of printing the tensor to `std::cout`, you just need a `string`, 
you can use this API:
+```c++
+std::string void to_string();
+```
+
+### Interactively Print Tensor Value (Dynamic) 
+
+When debugging, situations might occur that at compilation time, you do not 
know which part of a tensor to inspect. Also, sometimes, it would be nice to 
pause the operator control flow to “zoom into” a specific, erroneous part of a 
tensor multiple times until you are satisfied. In this regard, you can use this 
API to interactively inspect the tensor:
+
+```c++
+void  interactive_print(std::string tag =  "") {
+```
+
+This API will set a "break point" in your code, so that you will enter a loop 
that will keep asking you for further command. In the API call, `tag` is an 
optional parameter to give the call a name, so that you can identify it when 
you have multiple `interactive_print()` calls in different parts of your code. 
A visit count will tell you for how many times have you stepped into this 
particular "break point", should this operator be called more than once. Note 
that all `interactive_print()` calls are properly locked, so you can use it in 
many different places without issues.
+
+![Screen Shot 2019-07-10 at 5 29 07 
PM](https://user-images.githubusercontent.com/16669457/61013632-5325e800-a338-11e9-90e6-607f17d81495.png)
+
+Refer the screenshot above, there are many useful commands available: you can 
type "e" to print out the entire tensor, ''d" to dump the tensor to file (see 
below), "b" to break from this command loop, and "s" to skip all future 
`interactive_print()`. Most importantly, in this screen, you can specify a part 
of the tensor that you are particularly interested in and want to print out. 
For example, for this 20x1x5x5 tensor, you can type in "0, 0" and presss enter 
to check the sub-tensor with shape 5x5 at coordinate (0, 0). 
+
+### Check Tensor Value
+
+Sometimes, developers might want to check if the tensor contains unexpected 
values which could be negative 

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector Tutorial

[GitBox]

ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector 
Tutorial
URL: https://github.com/apache/incubator-mxnet/pull/15517#discussion_r303174371
 
 

 ##
 File path: docs/faq/tensor_inspector_tutorial.md
 ##
 @@ -0,0 +1,164 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# Use TensorInspector to Help Debug Operators
+
+## Introduction
+
+When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, This utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
+
+
+## Usage 
+
+This utility locates in `src/common/tensor_inspector.h`. To use it in any 
operator code, just include `tensor_inspector`, construct an `TensorInspector` 
object, and call the APIs on that object. You can run any script that uses the 
operator you just modified then.
+
+The screenshot below shows a sample usage in 
`src/operator/nn/convolution-inl.h`.
+
+![Screen Shot 2019-07-08 at 5 03 46 
PM](https://user-images.githubusercontent.com/16669457/60850062-68690e00-a1a2-11e9-8268-033edde17aa4.png)
+
+
+## Functionalities/APIs
+
+### Create a TensorInspector Object from Tensor, TBlob, and NDArray Objects
+
+You can create a `TensorInspector` object by passing in two things: 1) an 
object of type `Tensor`, `Tbob`, or `NDArray`, and 2) an `RunContext` object.
+
+Essentially, `TensorInspector` can be understood as a wrapper class around 
`TBlob`. Internally, the `Tensor`, `Tbob`, or `NDArray` object that you passed 
in will all be converted to a `TBlob` object. The `RunContext` object is used 
when the the tensor is a GPU tensor; in such case, we need to use the context 
information to copy the data from GPU memory to CPU/main memory.
 
 Review comment:
   ```suggestion
   Essentially, `TensorInspector` can be understood as a wrapper class around 
`TBlob`. Internally, the `Tensor`, `Tbob`, or `NDArray` object that you passed 
in will be converted to a `TBlob` object. The `RunContext` object is used when 
the the tensor is a GPU tensor; in such a case, we need to use the context 
information to copy the data from GPU memory to CPU/main memory.
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector Tutorial

[GitBox]

ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector 
Tutorial
URL: https://github.com/apache/incubator-mxnet/pull/15517#discussion_r303173725
 
 

 ##
 File path: docs/faq/tensor_inspector_tutorial.md
 ##
 @@ -0,0 +1,164 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# Use TensorInspector to Help Debug Operators
+
+## Introduction
+
+When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, This utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
+
+
+## Usage 
+
+This utility locates in `src/common/tensor_inspector.h`. To use it in any 
operator code, just include `tensor_inspector`, construct an `TensorInspector` 
object, and call the APIs on that object. You can run any script that uses the 
operator you just modified then.
 
 Review comment:
   ```suggestion
   This utility is located in `src/common/tensor_inspector.h`. To use it in any 
operator code, just include `tensor_inspector`, construct an `TensorInspector` 
object, and call the APIs on that object. You can run any script that uses the 
operator you just modified then.
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector Tutorial

[GitBox]

ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector 
Tutorial
URL: https://github.com/apache/incubator-mxnet/pull/15517#discussion_r303173914
 
 

 ##
 File path: docs/faq/tensor_inspector_tutorial.md
 ##
 @@ -0,0 +1,164 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# Use TensorInspector to Help Debug Operators
+
+## Introduction
+
+When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, This utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
 
 Review comment:
   
   ```suggestion
   When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, this utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector Tutorial

[GitBox]

ChaiBapchya commented on a change in pull request #15517: [WIP]Tensor Inspector 
Tutorial
URL: https://github.com/apache/incubator-mxnet/pull/15517#discussion_r303174576
 
 

 ##
 File path: docs/faq/tensor_inspector_tutorial.md
 ##
 @@ -0,0 +1,164 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# Use TensorInspector to Help Debug Operators
+
+## Introduction
+
+When developing new operators, developers need to deal with tensor objects 
extensively. This new utility, Tensor Inspector, mainly aims to help developers 
debug by providing unified interfaces to print, check, and dump the tensor 
value. To developers' convenience, This utility works for all the three data 
types: Tensors, TBlobs, and NDArrays. Also, it supports both CPU and GPU 
tensors.
+
+
+## Usage 
+
+This utility locates in `src/common/tensor_inspector.h`. To use it in any 
operator code, just include `tensor_inspector`, construct an `TensorInspector` 
object, and call the APIs on that object. You can run any script that uses the 
operator you just modified then.
+
+The screenshot below shows a sample usage in 
`src/operator/nn/convolution-inl.h`.
+
+![Screen Shot 2019-07-08 at 5 03 46 
PM](https://user-images.githubusercontent.com/16669457/60850062-68690e00-a1a2-11e9-8268-033edde17aa4.png)
+
+
+## Functionalities/APIs
+
+### Create a TensorInspector Object from Tensor, TBlob, and NDArray Objects
+
+You can create a `TensorInspector` object by passing in two things: 1) an 
object of type `Tensor`, `Tbob`, or `NDArray`, and 2) an `RunContext` object.
+
+Essentially, `TensorInspector` can be understood as a wrapper class around 
`TBlob`. Internally, the `Tensor`, `Tbob`, or `NDArray` object that you passed 
in will all be converted to a `TBlob` object. The `RunContext` object is used 
when the the tensor is a GPU tensor; in such case, we need to use the context 
information to copy the data from GPU memory to CPU/main memory.
+
+Below are the three constructors:
+
+```c++
+// Construct from Tensor object
+template
+TensorInspector(const  mshadow::Tensor& ts, const 
RunContext& ctx);
+
+// Construct from TBlob object
+TensorInspector(const TBlob& tb, const RunContext& ctx);
+
+// Construct from NDArray object
+TensorInspector(const NDArray& arr, const RunContext& ctx):
+```
+
+### Print Tensor Value (Static) 
+
+To print out the tensor value in a nicely structured way,  you can use this 
API:
+
+```c++
+void print_string();
+```
+
+This API will print the entire tensor to `std::cout` and preserve the shape 
(it supports all dimensions from 1 and up). You can copy the output and 
interpret it with any `JSON` loader. Also, on the last line of the output you 
can find some useful information about the tensor. Refer to the case below, we 
are able to know that this is a float-typed tensor with shape 20x1x5x5.
+
+![Screen Shot 2019-07-08 at 4 07 16 
PM](https://user-images.githubusercontent.com/16669457/60848554-d8c06100-a19b-11e9-9fe0-23e79a7a371a.png)
 
 Review comment:
   likewise


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] sxjscience commented on issue #14894: Accelerate ROIPooling layer

[GitBox]

sxjscience commented on issue #14894: Accelerate ROIPooling layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#issuecomment-511059573
 
 
   I think it should be good. One concern is it may be backward incompatible 
due to the `atomicAdd`. Although I feel it's reasonable to use the faster 
version, I need to confirm with the GluonCV team whether it will break the 
current training scripts. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] sxjscience commented on a change in pull request #14894: Accelerate ROIPooling layer

[GitBox]

sxjscience commented on a change in pull request #14894: Accelerate ROIPooling 
layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#discussion_r303173596
 
 

 ##
 File path: src/operator/roi_pooling.cc
 ##
 @@ -137,91 +140,18 @@ template
 inline void ROIPoolBackwardAcc(const Tensor _grad,
const Tensor _grad,
const Tensor ,
-   const Tensor _idx,
+   const Tensor _idx,
const float spatial_scale_) {
   const Dtype *top_diff = out_grad.dptr_;
-  const Dtype *bottom_rois = bbox.dptr_;
   Dtype *bottom_diff = in_grad.dptr_;
-  Dtype *argmax_data = max_idx.dptr_;
-
-  const int batch_size_ = in_grad.size(0);
-  const int channels_ = in_grad.size(1);
-  const int height_ = in_grad.size(2);
-  const int width_ = in_grad.size(3);
-  const int pooled_height_ = out_grad.size(2);
-  const int pooled_width_ = out_grad.size(3);
-
-  const int num_rois = bbox.size(0);
-
-  for (int b = 0; b < batch_size_; ++b) {
-for (int c = 0; c < channels_; ++c) {
-  for (int h = 0; h < height_; ++h) {
-for (int w = 0; w < width_; ++w) {
-  int offset_bottom_diff = (b * channels_ + c) * height_ * width_;
-  offset_bottom_diff += h * width_ + w;
-
-  Dtype gradient = 0;
-  // Accumulate gradient over all ROIs that pooled this element
-  for (int roi_n = 0; roi_n < num_rois; ++roi_n) {
-const Dtype* offset_bottom_rois = bottom_rois + roi_n * 5;
-int roi_batch_ind = offset_bottom_rois[0];
-assert(roi_batch_ind >= 0);
-assert(roi_batch_ind < batch_size_);
-if (b != roi_batch_ind) {
-  continue;
-}
+  index_t *argmax_data = max_idx.dptr_;
 
-int roi_start_w = std::round(offset_bottom_rois[1] * 
spatial_scale_);
-int roi_start_h = std::round(offset_bottom_rois[2] * 
spatial_scale_);
-int roi_end_w = std::round(offset_bottom_rois[3] * spatial_scale_);
-int roi_end_h = std::round(offset_bottom_rois[4] * spatial_scale_);
+  const index_t count = out_grad.shape_.Size();
 
-bool in_roi = (w >= roi_start_w && w <= roi_end_w &&
-   h >= roi_start_h && h <= roi_end_h);
-if (!in_roi) {
-  continue;
-}
-
-// force malformed ROIs to be 1 * 1
-int roi_height = max(roi_end_h - roi_start_h + 1, 1);
-int roi_width = max(roi_end_w - roi_start_w + 1, 1);
-const Dtype bin_size_h = static_cast(roi_height)
- / static_cast(pooled_height_);
-const Dtype bin_size_w = static_cast(roi_width)
- / static_cast(pooled_width_);
-
-// compute pooled regions correspond to original (h, w) point
-int phstart = static_cast(floor(static_cast(h - 
roi_start_h)
- / bin_size_h));
-int pwstart = static_cast(floor(static_cast(w - 
roi_start_w)
- / bin_size_w));
-int phend = static_cast(ceil(static_cast(h - 
roi_start_h + 1)
-  / bin_size_h));
-int pwend = static_cast(ceil(static_cast(w - 
roi_start_w + 1)
-  / bin_size_w));
-
-// clip to boundaries of pooled region
-phstart = min(max(phstart, 0), pooled_height_);
-phend = min(max(phend, 0), pooled_height_);
-pwstart = min(max(pwstart, 0), pooled_width_);
-pwend = min(max(pwend, 0), pooled_width_);
-
-// accumulate over gradients in pooled regions
-int offset = (roi_n * channels_ + c) * pooled_height_ * 
pooled_width_;
-const Dtype* offset_top_diff = top_diff + offset;
-const Dtype* offset_argmax_data = argmax_data + offset;
-for (int ph = phstart; ph < phend; ++ph) {
-  for (int pw = pwstart; pw < pwend; ++pw) {
-const int pooled_index = ph * pooled_width_ + pw;
-if (static_cast(offset_argmax_data[pooled_index]) == h * 
width_ + w) {
-  gradient += offset_top_diff[pooled_index];
-}
-  }
-}
-  }
-  bottom_diff[offset_bottom_diff] += gradient;
-}
-  }
+  for (int index = 0; index < count; ++index) {
+index_t max_idx = argmax_data[index];
+if (max_idx >= 0) {
+  bottom_diff[max_idx] += top_diff[index];
 
 Review comment:
   It's correct, sorry for the misreading.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the 

[GitHub] [incubator-mxnet] sxjscience commented on a change in pull request #14894: Accelerate ROIPooling layer

[GitBox]

sxjscience commented on a change in pull request #14894: Accelerate ROIPooling 
layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#discussion_r303173424
 
 

 ##
 File path: src/operator/roi_pooling.cc
 ##
 @@ -137,91 +140,18 @@ template
 inline void ROIPoolBackwardAcc(const Tensor _grad,
const Tensor _grad,
const Tensor ,
-   const Tensor _idx,
+   const Tensor _idx,
const float spatial_scale_) {
   const Dtype *top_diff = out_grad.dptr_;
-  const Dtype *bottom_rois = bbox.dptr_;
   Dtype *bottom_diff = in_grad.dptr_;
-  Dtype *argmax_data = max_idx.dptr_;
-
-  const int batch_size_ = in_grad.size(0);
-  const int channels_ = in_grad.size(1);
-  const int height_ = in_grad.size(2);
-  const int width_ = in_grad.size(3);
-  const int pooled_height_ = out_grad.size(2);
-  const int pooled_width_ = out_grad.size(3);
-
-  const int num_rois = bbox.size(0);
-
-  for (int b = 0; b < batch_size_; ++b) {
-for (int c = 0; c < channels_; ++c) {
-  for (int h = 0; h < height_; ++h) {
-for (int w = 0; w < width_; ++w) {
-  int offset_bottom_diff = (b * channels_ + c) * height_ * width_;
-  offset_bottom_diff += h * width_ + w;
-
-  Dtype gradient = 0;
-  // Accumulate gradient over all ROIs that pooled this element
-  for (int roi_n = 0; roi_n < num_rois; ++roi_n) {
-const Dtype* offset_bottom_rois = bottom_rois + roi_n * 5;
-int roi_batch_ind = offset_bottom_rois[0];
-assert(roi_batch_ind >= 0);
-assert(roi_batch_ind < batch_size_);
-if (b != roi_batch_ind) {
-  continue;
-}
+  index_t *argmax_data = max_idx.dptr_;
 
-int roi_start_w = std::round(offset_bottom_rois[1] * 
spatial_scale_);
-int roi_start_h = std::round(offset_bottom_rois[2] * 
spatial_scale_);
-int roi_end_w = std::round(offset_bottom_rois[3] * spatial_scale_);
-int roi_end_h = std::round(offset_bottom_rois[4] * spatial_scale_);
+  const index_t count = out_grad.shape_.Size();
 
-bool in_roi = (w >= roi_start_w && w <= roi_end_w &&
-   h >= roi_start_h && h <= roi_end_h);
-if (!in_roi) {
-  continue;
-}
-
-// force malformed ROIs to be 1 * 1
-int roi_height = max(roi_end_h - roi_start_h + 1, 1);
-int roi_width = max(roi_end_w - roi_start_w + 1, 1);
-const Dtype bin_size_h = static_cast(roi_height)
- / static_cast(pooled_height_);
-const Dtype bin_size_w = static_cast(roi_width)
- / static_cast(pooled_width_);
-
-// compute pooled regions correspond to original (h, w) point
-int phstart = static_cast(floor(static_cast(h - 
roi_start_h)
- / bin_size_h));
-int pwstart = static_cast(floor(static_cast(w - 
roi_start_w)
- / bin_size_w));
-int phend = static_cast(ceil(static_cast(h - 
roi_start_h + 1)
-  / bin_size_h));
-int pwend = static_cast(ceil(static_cast(w - 
roi_start_w + 1)
-  / bin_size_w));
-
-// clip to boundaries of pooled region
-phstart = min(max(phstart, 0), pooled_height_);
-phend = min(max(phend, 0), pooled_height_);
-pwstart = min(max(pwstart, 0), pooled_width_);
-pwend = min(max(pwend, 0), pooled_width_);
-
-// accumulate over gradients in pooled regions
-int offset = (roi_n * channels_ + c) * pooled_height_ * 
pooled_width_;
-const Dtype* offset_top_diff = top_diff + offset;
-const Dtype* offset_argmax_data = argmax_data + offset;
-for (int ph = phstart; ph < phend; ++ph) {
-  for (int pw = pwstart; pw < pwend; ++pw) {
-const int pooled_index = ph * pooled_width_ + pw;
-if (static_cast(offset_argmax_data[pooled_index]) == h * 
width_ + w) {
-  gradient += offset_top_diff[pooled_index];
-}
-  }
-}
-  }
-  bottom_diff[offset_bottom_diff] += gradient;
-}
-  }
+  for (int index = 0; index < count; ++index) {
+index_t max_idx = argmax_data[index];
+if (max_idx >= 0) {
+  bottom_diff[max_idx] += top_diff[index];
 
 Review comment:
   ~~Is it possible that two threads are writing to the same max_idx?~~


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the

[GitHub] [incubator-mxnet] sxjscience commented on a change in pull request #14894: Accelerate ROIPooling layer

[GitBox]

sxjscience commented on a change in pull request #14894: Accelerate ROIPooling 
layer
URL: https://github.com/apache/incubator-mxnet/pull/14894#discussion_r303173424
 
 

 ##
 File path: src/operator/roi_pooling.cc
 ##
 @@ -137,91 +140,18 @@ template
 inline void ROIPoolBackwardAcc(const Tensor _grad,
const Tensor _grad,
const Tensor ,
-   const Tensor _idx,
+   const Tensor _idx,
const float spatial_scale_) {
   const Dtype *top_diff = out_grad.dptr_;
-  const Dtype *bottom_rois = bbox.dptr_;
   Dtype *bottom_diff = in_grad.dptr_;
-  Dtype *argmax_data = max_idx.dptr_;
-
-  const int batch_size_ = in_grad.size(0);
-  const int channels_ = in_grad.size(1);
-  const int height_ = in_grad.size(2);
-  const int width_ = in_grad.size(3);
-  const int pooled_height_ = out_grad.size(2);
-  const int pooled_width_ = out_grad.size(3);
-
-  const int num_rois = bbox.size(0);
-
-  for (int b = 0; b < batch_size_; ++b) {
-for (int c = 0; c < channels_; ++c) {
-  for (int h = 0; h < height_; ++h) {
-for (int w = 0; w < width_; ++w) {
-  int offset_bottom_diff = (b * channels_ + c) * height_ * width_;
-  offset_bottom_diff += h * width_ + w;
-
-  Dtype gradient = 0;
-  // Accumulate gradient over all ROIs that pooled this element
-  for (int roi_n = 0; roi_n < num_rois; ++roi_n) {
-const Dtype* offset_bottom_rois = bottom_rois + roi_n * 5;
-int roi_batch_ind = offset_bottom_rois[0];
-assert(roi_batch_ind >= 0);
-assert(roi_batch_ind < batch_size_);
-if (b != roi_batch_ind) {
-  continue;
-}
+  index_t *argmax_data = max_idx.dptr_;
 
-int roi_start_w = std::round(offset_bottom_rois[1] * 
spatial_scale_);
-int roi_start_h = std::round(offset_bottom_rois[2] * 
spatial_scale_);
-int roi_end_w = std::round(offset_bottom_rois[3] * spatial_scale_);
-int roi_end_h = std::round(offset_bottom_rois[4] * spatial_scale_);
+  const index_t count = out_grad.shape_.Size();
 
-bool in_roi = (w >= roi_start_w && w <= roi_end_w &&
-   h >= roi_start_h && h <= roi_end_h);
-if (!in_roi) {
-  continue;
-}
-
-// force malformed ROIs to be 1 * 1
-int roi_height = max(roi_end_h - roi_start_h + 1, 1);
-int roi_width = max(roi_end_w - roi_start_w + 1, 1);
-const Dtype bin_size_h = static_cast(roi_height)
- / static_cast(pooled_height_);
-const Dtype bin_size_w = static_cast(roi_width)
- / static_cast(pooled_width_);
-
-// compute pooled regions correspond to original (h, w) point
-int phstart = static_cast(floor(static_cast(h - 
roi_start_h)
- / bin_size_h));
-int pwstart = static_cast(floor(static_cast(w - 
roi_start_w)
- / bin_size_w));
-int phend = static_cast(ceil(static_cast(h - 
roi_start_h + 1)
-  / bin_size_h));
-int pwend = static_cast(ceil(static_cast(w - 
roi_start_w + 1)
-  / bin_size_w));
-
-// clip to boundaries of pooled region
-phstart = min(max(phstart, 0), pooled_height_);
-phend = min(max(phend, 0), pooled_height_);
-pwstart = min(max(pwstart, 0), pooled_width_);
-pwend = min(max(pwend, 0), pooled_width_);
-
-// accumulate over gradients in pooled regions
-int offset = (roi_n * channels_ + c) * pooled_height_ * 
pooled_width_;
-const Dtype* offset_top_diff = top_diff + offset;
-const Dtype* offset_argmax_data = argmax_data + offset;
-for (int ph = phstart; ph < phend; ++ph) {
-  for (int pw = pwstart; pw < pwend; ++pw) {
-const int pooled_index = ph * pooled_width_ + pw;
-if (static_cast(offset_argmax_data[pooled_index]) == h * 
width_ + w) {
-  gradient += offset_top_diff[pooled_index];
-}
-  }
-}
-  }
-  bottom_diff[offset_bottom_diff] += gradient;
-}
-  }
+  for (int index = 0; index < count; ++index) {
+index_t max_idx = argmax_data[index];
+if (max_idx >= 0) {
+  bottom_diff[max_idx] += top_diff[index];
 
 Review comment:
   Is it possible that two threads are writing to the same max_idx?


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL 

[GitHub] [incubator-mxnet] anirudh2290 opened a new pull request #15526: Fix AMP Tutorial warnings and failures

[GitBox]

anirudh2290 opened a new pull request #15526: Fix AMP Tutorial warnings and 
failures
URL: https://github.com/apache/incubator-mxnet/pull/15526
 
 
   ## Description ##
   - Ignored warnings
   - removed usage of __file__.
   
   ## Checklist ##
   ### Essentials ###
   Please feel free to remove inapplicable items for your PR.
   - [ ] The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to 
the relevant [JIRA issue](https://issues.apache.org/jira/projects/MXNET/issues) 
created (except PRs with tiny changes)
   - [ ] Changes are complete (i.e. I finished coding on this PR)
   - [ ] All changes have test coverage:
   - Unit tests are added for small changes to verify correctness (e.g. adding 
a new operator)
   - Nightly tests are added for complicated/long-running ones (e.g. changing 
distributed kvstore)
   - Build tests will be added for build configuration changes (e.g. adding a 
new build option with NCCL)
   - [ ] Code is well-documented: 
   - For user-facing API changes, API doc string has been updated. 
   - For new C++ functions in header files, their functionalities and arguments 
are documented. 
   - For new examples, README.md is added to explain the what the example does, 
the source of the dataset, expected performance on test set and reference to 
the original paper if applicable
   - Check the API doc at 
http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
   - [ ] To the my best knowledge, examples are either not affected by this 
change, or have been fixed to be compatible with this change
   
   ### Changes ###
   - [ ] Feature1, tests, (and when applicable, API doc)
   - [ ] Feature2, tests, (and when applicable, API doc)
   
   ## Comments ##
   - If this change is a backward incompatible change, why must this change be 
made.
   - Interesting edge cases to note here
   
   @roywei 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] aaronmarkham commented on issue #15524: refine Nano setup directions

[GitBox]

aaronmarkham commented on issue #15524: refine Nano setup directions
URL: https://github.com/apache/incubator-mxnet/pull/15524#issuecomment-511056119
 
 
   Nudging CI again...


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] aaronmarkham commented on issue #15523: add julia env settings

[GitBox]

aaronmarkham commented on issue #15523: add julia env settings
URL: https://github.com/apache/incubator-mxnet/pull/15523#issuecomment-511055829
 
 
   Nudging CI since so many tests failed for no apparent reason.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya commented on issue #15522: [Opperf] Add optimizer update operator benchmarks to opperf

[GitBox]

ChaiBapchya commented on issue #15522: [Opperf] Add  optimizer update operator 
benchmarks to opperf
URL: https://github.com/apache/incubator-mxnet/pull/15522#issuecomment-511055731
 
 
   @mxnet-label-bot add [Benchmark, Operator]


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy edited a comment on issue #15516: Fix memory leak reported by ASAN in NNVM to ONNX conversion

[GitBox]

larroy edited a comment on issue #15516: Fix memory leak reported by ASAN in 
NNVM to ONNX conversion
URL: https://github.com/apache/incubator-mxnet/pull/15516#issuecomment-511052670
 
 
   Not really, in those cases you want a semantic name, even if you are 
wrapping your Cat in a pointer you want to name it:
   
   `unique_ptr cat = make_unique() `
   
   And not 
   
   `unique_ptr unique_ptr = make_unique()`
   
   Semantic naming is important, we have already type information from the 
compiler and we are in a statically typed language, no need to name a pointer 
_ptr. And definitely not "data" for the name of a variable, this is a code 
smell.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy commented on issue #15516: Fix memory leak reported by ASAN in NNVM to ONNX conversion

[GitBox]

larroy commented on issue #15516: Fix memory leak reported by ASAN in NNVM to 
ONNX conversion
URL: https://github.com/apache/incubator-mxnet/pull/15516#issuecomment-511052670
 
 
   Not really, in those cases you want a semantic name, even if you are 
wrapping your Cat in a pointer you want to name it:
   
   unique_ptr cat = make_unique() 
   
   And not 
   
   unique_ptr unique_ptr = make_unique().
   
   Semantic naming is important, we have already type information from the 
compiler and we are in a statically typed language, no need to name a pointer 
_ptr. And definitely not "data" for the name of a variable, this is a code 
smell.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303166837
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303166837
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303166316
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] KellenSunderland commented on a change in pull request #15399: Add unit tests for TensorRT integration and fix some bugs

[GitBox]

KellenSunderland commented on a change in pull request #15399: Add unit tests 
for TensorRT integration and fix some bugs
URL: https://github.com/apache/incubator-mxnet/pull/15399#discussion_r302338157
 
 

 ##
 File path: src/operator/subgraph/tensorrt/tensorrt-inl.h
 ##
 @@ -180,6 +253,17 @@ class TensorrtProperty : public SubgraphProperty {
 n->attrs.name = "TensorRT" + std::to_string(subgraph_id);
 n->attrs.op = Op::Get("_TensorRT");
 CHECK(n->attrs.op);
+// prevent using Gamma value if using fix_gamma on BatchNorm
+DFSVisit(new_sym.outputs, [](const nnvm::NodePtr& node) {
+  if (node->op() == Op::Get("BatchNorm")) {
 
 Review comment:
   Why not just check for FixGamma = true during nnvm -> Onnx conversion and 
set gamma to 1 if it's true?


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303153431
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303150705
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303139896
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303139943
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[incubator-mxnet] branch master updated (9ff6c46 -> 2565fa2)

[anirudh2290]

This is an automated email from the ASF dual-hosted git repository.

anirudh2290 pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.


from 9ff6c46  Add -R option to ci/build.py to avoid rebuilding containers 
(#15426)
 add 2565fa2  fix nightly CI failure (#15452)

No new revisions were added by this update.

Summary of changes:
 ci/docker/runtime_functions.sh|  6 +--
 docs/tutorials/amp/amp_tutorial.md|  1 +
 tests/nightly/estimator/test_estimator_cnn.py | 32 -
 tests/nightly/estimator/test_sentiment_rnn.py | 69 ---
 4 files changed, 55 insertions(+), 53 deletions(-)

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303133599
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303132894
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303129096
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303130822
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303130854
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303130688
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303130767
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303130652
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303130733
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
 
 Review comment:
   added


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303129096
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303129096
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] Zha0q1 commented on issue #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on issue #15490: Utility to help developers debug operators: 
Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#issuecomment-511013592
 
 
   > Nice work, just a few comments, looks good.
   
   Thank you for the detailed advices!


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303118403
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303113524
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
 
 Review comment:
   const?


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303113874
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303118876
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303118036
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303113816
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303115908
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303119170
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303117010
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303119928
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

larroy commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303119597
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; ++i) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] larroy commented on issue #15465: [RFC] Integrate TVM into Apache MXNet

[GitBox]

larroy commented on issue #15465: [RFC] Integrate TVM into Apache MXNet
URL: 
https://github.com/apache/incubator-mxnet/issues/15465#issuecomment-510999368
 
 
   Could we add more information on how will it impact the build systems and 
dependencies? While integrating TVM is exciting, I think is fair to gauge the 
increase of complexity, build and release implications. Would it be possible to 
load operators compiled with TVM dynamically? This would open a path to 
modularize MXNet and reduce the library size / target inference only version in 
a more streamlined way than with amalgamation hacks.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] marcoabreu commented on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

marcoabreu commented on issue #15525: Nightly tests / integration tests in PRs 
are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510999095
 
 
   I'm certainly in favour of a context sensitive bot that gives certain 
comments if certain types of changes are detected. But enforcing directly isn't 
possible besides adding actual checks.
   
   For example, we asked people in the past to run make lint locally. I could 
give you statistics about how often linting fails in ci, but trust me that 
something that simple isn't even followed most of the time. So basically we 
can't request anything from people unless we enforce it, which I'm opposed to 
in this particular case


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy commented on issue #15270: Fix warnings in CLang.

[GitBox]

larroy commented on issue #15270: Fix warnings in CLang.
URL: https://github.com/apache/incubator-mxnet/pull/15270#issuecomment-510997782
 
 
   Done, could we get those small PRs that fix warnings merged?


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy edited a comment on issue #15426: Add -R option to ci/build.py to avoid rebuilding containers

[GitBox]

larroy edited a comment on issue #15426: Add -R option to ci/build.py to avoid 
rebuilding containers
URL: https://github.com/apache/incubator-mxnet/pull/15426#issuecomment-510997277
 
 
   Danke für die merge.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] larroy commented on issue #15426: Add -R option to ci/build.py to avoid rebuilding containers

[GitBox]

larroy commented on issue #15426: Add -R option to ci/build.py to avoid 
rebuilding containers
URL: https://github.com/apache/incubator-mxnet/pull/15426#issuecomment-510997277
 
 
   Danke fuer die merge.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] anirudh2290 commented on a change in pull request #15245: FP16 Support for C Predict API

[GitBox]

anirudh2290 commented on a change in pull request #15245: FP16 Support for C 
Predict API
URL: https://github.com/apache/incubator-mxnet/pull/15245#discussion_r303110259
 
 

 ##
 File path: amalgamation/python/mxnet_predict.py
 ##
 @@ -160,10 +249,18 @@ def forward(self, **kwargs):
 >>> predictor.forward(data=mydata)
 >>> out = predictor.get_output(0)
 """
+if self.type_dict and len(self.type_dict) != len(kwargs.items()):
+raise ValueError("number of kwargs should be same as len of 
type_dict" \
+ "Please check your forward pass inputs" \
+ "or type_dict passed to Predictor instantiation")
+
 for k, v in kwargs.items():
 if not isinstance(v, np.ndarray):
 raise ValueError("Expect numpy ndarray as input")
-v = np.asarray(v, dtype=np.float32, order='C')
+if self.type_dict and k in self.type_dict:
+v = np.asarray(v, dtype=self.type_dict[k], order='C')
+else:
+v = np.asarray(v, dtype=np.float32, order='C')
 
 Review comment:
   we use the lookup table to map the dtype int, which comes from mxnet backend 
to numpy dtype. For cases, we are not determining it from backend dtype I think 
it is better to use np.float32 rather than _DTYPE_MX_TO_NP[0].


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] roywei edited a comment on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

roywei edited a comment on issue #15525: Nightly tests / integration tests in 
PRs are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510993342
 
 
   - Nightly tests are supposed to fail and we can revert whichever PR caused 
it, other wise it's too costly to run even selected jobs only for PR 
validation, imagine each commit will trigger a big job.
   - Most of the time, when contributors are changing nightly test 
configurations and adding nightly tests, they test it anyway, it's not likely 
to fail.
   - Test tuorials is a special case as many people are contributing to 
tutorials. But most of them don't know they have to add any dependency package 
used in tutorial in CI docker file and warnings generated are considered as 
failure. That's most of the reasons for CI failure.
   
   To start with, without any new features been added to CI, maybe some way to 
enforce contributors to test locally before merging PR.  (Having label bot to 
leave a message if nightly test change detected?)


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] roywei edited a comment on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

roywei edited a comment on issue #15525: Nightly tests / integration tests in 
PRs are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510993342
 
 
   - Nightly tests are supposed to fail and we can revert whichever PR caused 
it, other wise it's too costly to run even selected jobs only for PR 
validation, imagine each commit will trigger a big job.
   - Most of the time, when contributors are changing nightly test 
configurations and adding nightly tests, they test it anyway, it's not likely 
to fail.
   - Test tuorials is a special case as many people are contributing to 
tutorials. But most of them don't know they have to add any dependency package 
used in tutorial in CI docker file and warnings generated are considered as 
failure. That's most of the reasons for CI failure.
   
   To start with, without any new features been added, maybe some way to 
enforce contributors to test locally before merging PR.  (Having label bot to 
leave a message if nightly test change detected?)


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] roywei edited a comment on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

roywei edited a comment on issue #15525: Nightly tests / integration tests in 
PRs are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510993342
 
 
   - Nightly tests are supposed to fail and we can revert whichever PR caused 
it, other wise it's too costly to run even selected jobs only for PR 
validation, imagine each commit will trigger a big job.
   - Most of the time, when contributors are changing nightly test 
configurations and adding nightly tests, they test it anyway, it's not likely 
to fail.
   - Test tuorials is a special case as many people are contributing to 
tutorials. But most of them don't know they have to add any dependency package 
used in tutorial in CI docker file and warnings generated are considered as 
failure. That's most of the reasons for CI failure.
   
   To start with, without any new features been added, maybe some way to 
enforce contributors to test locally before merging PR.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] roywei edited a comment on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

roywei edited a comment on issue #15525: Nightly tests / integration tests in 
PRs are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510993342
 
 
   - Nightly tests are supposed to fail and we can revert whichever PR caused 
it, other wise it's too costly to run even selected jobs only for PR 
validation, imagine each commit will trigger a big job.
   - Most of the time, when contributors are changing nightly test 
configurations and adding nightly tests, they test it anyway, it's not likely 
to fail.
   - Test tuorials is a special case as many people are contributing to 
tutorials. But most of them don't know they have to add any dependency package 
used in tutorial in CI docker file and warnings are considered as failure. 
That's most of the reasons for CI failure.
   
   To start with, without any new features been added, maybe some way to 
enforce contributors to test tutorials locally before merging PR.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] roywei edited a comment on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

roywei edited a comment on issue #15525: Nightly tests / integration tests in 
PRs are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510993342
 
 
   - Nightly tests are supposed to fail and we can revert whichever PR caused 
it, other wise it's too costly to run even selected jobs only for PR 
validation, imagine each commit will trigger a big job.
   - Most of the time, when contributors are changing nightly test 
configurations and adding nightly tests, they test it anyway, it's not likely 
to fail.
   - Test tuorials is a special case as many people are contributing to 
tutorials. But most of them don't know they have to add any dependency package 
used in tutorial in CI docker file and warnings generated are considered as 
failure. That's most of the reasons for CI failure.
   
   To start with, without any new features been added, maybe some way to 
enforce contributors to test tutorials locally before merging PR.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] roywei edited a comment on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

roywei edited a comment on issue #15525: Nightly tests / integration tests in 
PRs are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510993342
 
 
   - Nightly tests are supposed to fail and we can revert whichever PR caused 
it, other wise it's too costly to run even selected jobs only for PR 
validation, imagine each commit will trigger a big job.
   - Most of the time, when contributors are changing nightly test 
configurations and adding nightly tests, they test it anyway, it's not likely 
to fail.
   - Test tuorials is a special case as many people are contributing to 
tutorials. But most of them don't know they have to add any dependency in CI 
and warnings are considered as failure. That's most of the reasons for CI 
failure.
   
   To start with, without any new features been added, maybe some way to 
enforce contributors to test tutorials locally before merging PR.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] roywei commented on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

roywei commented on issue #15525: Nightly tests / integration tests in PRs are 
not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510993342
 
 
   - Nightly tests are supposed to fail and we can revert whichever PR caused 
it, other wise it's too costly to run even for PR validation, imagine each 
commit will trigger a big job.
   - Most of the time, when contributors are changing nightly test 
configurations and adding nightly tests, they test it anyway, it's not likely 
to fail.
   - Test tuorials is a special case as many people are contributing to 
tutorials. But most of them don't know they have to add any dependency in CI 
and warnings are considered as failure. That's most of the reasons for CI 
failure.
   
   To start with, without any new features been added, maybe some way to 
enforce contributors to test tutorials locally before merging PR.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] marcoabreu commented on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

marcoabreu commented on issue #15525: Nightly tests / integration tests in PRs 
are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510985789
 
 
   I'd love to have context sensitive validations. Unfortunately, this feature 
doesn't exist out of the box.
   
   In general I think that we consider master bleeding edge. Thus, it's okay to 
have less coverage for the sake of speed and costs. If a PR breaks something, 
we usually revert it and get back to the original author.
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #15490: Utility to help developers debug operators: Tensor Inspector

[GitBox]

Zha0q1 commented on a change in pull request #15490: Utility to help developers 
debug operators: Tensor Inspector
URL: https://github.com/apache/incubator-mxnet/pull/15490#discussion_r303081013
 
 

 ##
 File path: src/common/tensor_inspector.h
 ##
 @@ -0,0 +1,733 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * Copyright (c) 2019 by Contributors
+ * \file tensor_inspector.h
+ * \brief utility to inspect tensor objects
+ * \author Zhaoqi Zhu
+ */
+
+#ifndef MXNET_COMMON_TENSOR_INSPECTOR_H_
+#define MXNET_COMMON_TENSOR_INSPECTOR_H_
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include "../../3rdparty/mshadow/mshadow/base.h"
+#include "../../tests/cpp/include/test_util.h"
+
+namespace mxnet {
+
+/*!
+ * \brief this singleton struct mediates individual TensorInspector objects
+ * so that we can control the global behavior from each of them
+ */
+struct InspectorManager {
+  static InspectorManager* get() {
+static std::mutex mtx;
+static std::unique_ptr im = nullptr;
+if (!im) {
+  std::unique_lock lk(mtx);
+  if (!im)
+im = std::make_unique();
+}
+return im.get();
+  }
+  /* !\brief mutex used to lock interactive_print() and check_value() */
+  std::mutex mutex_;
+  /* !\brief skip all interactive prints */
+  bool interactive_print_skip_all_ = false;
+  /* !\brief skip all value checks */
+  bool check_value_skip_all_ = false;
+  /* !\brief visit count for interactive print tags */
+  std::unordered_map interactive_print_tag_counter_;
+  /* !\brief visit count for check value tags */
+  std::unordered_map check_value_tag_counter_;
+  /* !\brief visit count for dump value tags */
+  std::unordered_map dump_value_tag_counter_;
+};
+
+/*!
+ * \brief Enum for building value checkers for TensorInspector::check_value()
+ */
+enum CheckerType {
+  NegativeChecker,  // check if is negative
+  PositiveChecker,  // check if is positive
+  ZeroChecker,  // check if is zero
+  NaNChecker,  // check if is NaN, will always return false if DType is not a 
float type
+  InfChecker,  // check if is infinity, will always return false if DType is 
not a float type
+  PositiveInfChecker,  // check if is positive infinity,
+   // will always return false if DType is not a float type
+  NegativeInfChecker,  // check if is nagative infinity,
+   // will always return false if DType is not a float type
+  FiniteChecker,  // check if is finite, will always return false if DType is 
not a float type
+  NormalChecker,  // check if is neither infinity nor NaN
+  AbnormalChecker,  // chekck if is infinity or nan
+};
+
+/**
+ *  ___  _   _ 
+ * |__   __||_   _| | |
+ *| | ___ _ __  ___  ___  _ __| |  _ __  ___ _ __   ___  ___| |_ ___  _ __ 
+ *| |/ _ \ '_ \/ __|/ _ \| '__| | | '_ \/ __| '_ \ / _ \/ __| __/ _ \| '__|
+ *| |  __/ | | \__ \ (_) | | _| |_| | | \__ \ |_) |  __/ (__| || (_) | |   
+ *|_|\___|_| |_|___/\___/|_||_|_| |_|___/ .__/ \___|\___|\__\___/|_|   
+ *  | |
+ *  |_|   
+ */
+
+/*!
+ * \brief This class provides a unified interface to inspect the value of all 
data types
+ * including Tensor, TBlob, and NDArray. If the tensor resides on GPU, then it 
will be
+ * copied from GPU memory back to CPU memory to be operated on. Internally, 
all data types
+ * are stored as a TBlob object tb_.
+ */
+class TensorInspector {
+ private:
+  /*!
+   * \brief generate the tensor info, including data type and shape 
+   * \tparam DType the data type
+   * \tparam StreamType the type of the stream object
+   * \param os stream object to output to
+   */
+  template
+  inline void tensor_info_to_string(StreamType* os) {
+int dimension = tb_.ndim();
+*os << "<" << typeid(tb_.dptr()[0]).name() << " Tensor ";
+*os << tb_.shape_[0];
+for (int i = 1; i < dimension; i++) {
+  *os << 'x' << tb_.shape_[i];
+}
+*os << ">" << std::endl;
+  }
+
+  /*!
+   * \brief output the tensor info, 

[GitHub] [incubator-mxnet] sandeep-krishnamurthy opened a new issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

sandeep-krishnamurthy opened a new issue #15525: Nightly tests / integration 
tests in PRs are not tested before merging
URL: https://github.com/apache/incubator-mxnet/issues/15525
 
 
   We do not test nightly/integration tests added in a PR at PR CI stage. This 
leads to a PR with failing integration test to get merged and all our nightly 
CI jobs starts failing.
   
   We need to make sure everything in the PR is tested before merge. For this 
particular problem, one suggested solution could be - If there are nightly 
tests in the PR, it needs to be executed as part of PR validation.
   
   Suggestions? 
   @roywei @marcoabreu @lanking520 


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #15525: Nightly tests / integration tests in PRs are not tested before merging

[GitBox]

mxnet-label-bot commented on issue #15525: Nightly tests / integration tests in 
PRs are not tested before merging
URL: 
https://github.com/apache/incubator-mxnet/issues/15525#issuecomment-510983068
 
 
   Hey, this is the MXNet Label Bot. 
Thank you for submitting the issue! I will try and suggest some labels so 
that the appropriate MXNet community members can help resolve it. 
Here are my recommended labels: Test


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] haojin2 commented on a change in pull request #15349: Numpy Tensordot Operator

[GitBox]

haojin2 commented on a change in pull request #15349: Numpy Tensordot Operator 
URL: https://github.com/apache/incubator-mxnet/pull/15349#discussion_r303094273
 
 

 ##
 File path: src/operator/numpy/np_tensordot_op.cc
 ##
 @@ -0,0 +1,222 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * \file np_tensordot_op.cc
+ * \brief CPU Implementation of numpy-compatible tensordot
+ */
+
+#include 
+#include "np_tensordot_op-inl.h"
+
+namespace mxnet {
+namespace op {
+
+bool TensordotOpShape(const nnvm::NodeAttrs& attrs,
+  mxnet::ShapeVector *in_attrs,
+  mxnet::ShapeVector *out_attrs) {
+  CHECK_EQ(in_attrs->size(), 2U);
+  CHECK_EQ(out_attrs->size(), 1U);
+
+  const mxnet::TShape& a_shape = in_attrs->at(0);
+  const mxnet::TShape& b_shape = in_attrs->at(1);
+
+  if (!ndim_is_known(a_shape) || !ndim_is_known(b_shape)) {
+return false;
+  }
+
+  if ((a_shape.ndim() < 1) || (b_shape.ndim() < 1)) {
+return false;
+  }
+
+  const TensordotParam& param = nnvm::get(attrs.parsed);
+  const Tuple& a_axes_summed = param.a_axes_summed;
+  const Tuple& b_axes_summed = param.b_axes_summed;
+
+  Tuple a_axes_remained;
+  Tuple b_axes_remained;
+  Tuple a_axes;
+  Tuple b_axes;
+  GetReorderedAxes(a_axes_summed, _axes_remained, _axes, b_axes_summed, 
_axes_remained,
+   _axes, a_shape, b_shape);
+
+  CHECK_EQ(a_axes_summed.ndim(), b_axes_summed.ndim());
+
+  mxnet::TShape out_shape(a_axes_remained.ndim() + b_axes_remained.ndim(), -1);
+  for (int i = 0; i < a_axes_remained.ndim(); i++) {
+out_shape[i] = a_shape[a_axes_remained[i]];
+  }
+  for (int i = 0; i < b_axes_remained.ndim(); i++) {
+out_shape[a_axes_remained.ndim() + i] = b_shape[b_axes_remained[i]];
+  }
+  SHAPE_ASSIGN_CHECK(*out_attrs, 0, out_shape);
+
+  mxnet::TShape tem_shape1(a_axes.ndim(), -1);
+  for (int i = 0; i < a_axes_remained.ndim(); i++) {
+tem_shape1[a_axes_remained[i]] = out_shape[i];
+  }
+  for (int i = 0; i < a_axes_summed.ndim(); i++) {
+tem_shape1[a_axes_summed[i]] = b_shape[b_axes_summed[i]];
+  }
+  SHAPE_ASSIGN_CHECK(*in_attrs, 0, tem_shape1);
+
+  mxnet::TShape tem_shape2(b_axes.ndim(), -1);
+  for (int i = 0; i < b_axes_remained.ndim(); i++) {
+tem_shape2[b_axes_remained[i]] = out_shape[a_axes_remained.ndim() + i];
+  }
+  for (int i = 0; i < b_axes_summed.ndim(); i++) {
+tem_shape2[b_axes_summed[i]] = a_shape[a_axes_summed[i]];
+  }
+  SHAPE_ASSIGN_CHECK(*in_attrs, 1, tem_shape2);
+
+  return shape_is_known(*in_attrs) && shape_is_known(*out_attrs);
+}
+
+DMLC_REGISTER_PARAMETER(TensordotParam);
+
+NNVM_REGISTER_OP(_npi_tensordot)
+.set_attr_parser(mxnet::op::ParamParser)
+.set_num_inputs(2)
+.set_num_outputs(1)
+.set_attr("FListInputNames",
+  [](const NodeAttrs& attrs) {
+return std::vector{"a", "b"};
+  })
+.set_attr("FInferShape", TensordotOpShape)
+.set_attr("FInferType", mxnet::op::ElemwiseType<2, 1>)
+.set_attr("FResourceRequest",
+  [](const NodeAttrs& attrs) {
+return std::vector{ResourceRequest::kTempSpace};
+  })
+.set_attr("FCompute", TensordotOpForward)
+.set_attr("FGradient", 
mxnet::op::ElemwiseGradUseIn{"_backward_npi_tensordot"})
+.add_argument("a", "NDArray-or-Symbol", "First input")
+.add_argument("b", "NDArray-or-Symbol", "Second input")
+.add_arguments(TensordotParam::__FIELDS__());
+
+NNVM_REGISTER_OP(_backward_npi_tensordot)
+.set_attr_parser(mxnet::op::ParamParser)
+.set_num_inputs(3)
+.set_num_outputs(2)
+.set_attr("TIsBackward", true)
+.set_attr("FResourceRequest",
+  [](const NodeAttrs& attrs) {
+return std::vector{ResourceRequest::kTempSpace};
+  })
+.set_attr("FCompute", TensordotOpBackward);
+
+bool TensordotIntAxesOpShape(const nnvm::NodeAttrs& attrs,
+ mxnet::ShapeVector *in_attrs,
+ mxnet::ShapeVector *out_attrs) {
+  CHECK_EQ(in_attrs->size(), 2U);
+  CHECK_EQ(out_attrs->size(), 1U);
+
+  const mxnet::TShape& a_shape = in_attrs->at(0);
+  const mxnet::TShape& b_shape = in_attrs->at(1);
+
+  if (!ndim_is_known(a_shape) || !ndim_is_known(b_shape)) {
+return false;
+  }
+
+  if ((a_shape.ndim() < 1) || (b_shape.ndim() < 1)) {
+return false;
+  }
+
+  const 

[GitHub] [incubator-mxnet] haojin2 commented on a change in pull request #15349: Numpy Tensordot Operator

[GitBox]

haojin2 commented on a change in pull request #15349: Numpy Tensordot Operator 
URL: https://github.com/apache/incubator-mxnet/pull/15349#discussion_r303094358
 
 

 ##
 File path: tests/python/unittest/test_numpy_op.py
 ##
 @@ -26,6 +26,151 @@
 from mxnet.test_utils import check_numeric_gradient
 from common import assertRaises, with_seed
 import random
+import collections
+
 
 Review comment:
   one more blank line.


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] aaronmarkham opened a new pull request #15524: refine Nano setup directions

[GitBox]

aaronmarkham opened a new pull request #15524: refine Nano setup directions
URL: https://github.com/apache/incubator-mxnet/pull/15524
 
 
   ## Description ##
   After another run-through on a fresh Nano, plus trying out these new pip 
wheels, we found some refinements that should make the setup go a little more 
cleanly.
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet] branch master updated (6a564be -> 9ff6c46)

[marcoabreu]

This is an automated email from the ASF dual-hosted git repository.

marcoabreu pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.


from 6a564be  cuda/cuDNN lib version checking.  Force cuDNN v7 usage. 
(#15449)
 add 9ff6c46  Add -R option to ci/build.py to avoid rebuilding containers 
(#15426)

No new revisions were added by this update.

Summary of changes:
 ci/build.py | 10 +-
 1 file changed, 9 insertions(+), 1 deletion(-)

[GitHub] [incubator-mxnet] marcoabreu merged pull request #15426: Add -R option to ci/build.py to avoid rebuilding containers

[GitBox]

marcoabreu merged pull request #15426: Add -R option to ci/build.py to avoid 
rebuilding containers
URL: https://github.com/apache/incubator-mxnet/pull/15426
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] marcoabreu commented on issue #15426: Add -R option to ci/build.py to avoid rebuilding containers

[GitBox]

marcoabreu commented on issue #15426: Add -R option to ci/build.py to avoid 
rebuilding containers
URL: https://github.com/apache/incubator-mxnet/pull/15426#issuecomment-510979685
 
 
   Ye it's probably user id - for whatever reason docker cares about that...


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] aaronmarkham opened a new pull request #15523: add julia env settings

[GitBox]

aaronmarkham opened a new pull request #15523: add julia env settings
URL: https://github.com/apache/incubator-mxnet/pull/15523
 
 
   ## Description ##
   I missed this env var requirement in the Julia docs PR, #15454. 
   
   ## Comment
   We should probably add some CI test for the website publish routine so this 
kind of thing doesn't get overlooked until after a merge.
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] [incubator-mxnet] ChaiBapchya opened a new pull request #15522: Add optimizer update operator benchmarks to opperf

[GitBox]

ChaiBapchya opened a new pull request #15522: Add  optimizer update operator 
benchmarks to opperf
URL: https://github.com/apache/incubator-mxnet/pull/15522
 
 
   ## Description ##
   Added neural network optimizer (Adam, sgd, etc) update operators to the 
existing set of opperf benchmarks
   
   ## Checklist ##
   ### Essentials ###
   Please feel free to remove inapplicable items for your PR.
   - [X] Changes are complete (i.e. I finished coding on this PR)
   - [X] All changes have test coverage:
   - [X] Code is well-documented: 
   - [X] To the my best knowledge, examples are either not affected by this 
change, or have been fixed to be compatible with this change
   
   ### Changes ###
   - [ ] Added optimizer update ops to opperf
   
   @sandeep-krishnamurthy @apeforest 


This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services