[GitHub] ZiyueHuang commented on issue #8822: Can mx.nd.where(condition, x, y) supports if both x and y are None?

[GitBox]

ZiyueHuang commented on issue #8822: Can mx.nd.where(condition, x, y) supports 
if both x and y are None?
URL: 
https://github.com/apache/incubator-mxnet/issues/8822#issuecomment-353282731
 
 
   This requires dynamic memory allocation and infer shape at runtime when we 
get the data. So some operators like `tf.where`, `tf.unique`, etc. cannot be 
implemented in MXNet currently.
   
   Should we add support for dynamic shape?  @piiswrong @reminisce 
   
   Following is the way in tensorflow, (please correct me if I'm wrong since 
I'm not familiar with tensorflow)
   
   https://www.tensorflow.org/programmers_guide/faq
   
   > How can I determine the shape of a tensor in Python?
   > In TensorFlow, a tensor has both a static (inferred) shape and a dynamic 
(true) shape. The static shape can be read using the tf.Tensor.get_shape 
method: this shape is inferred from the operations that were used to create the 
tensor, and may be partially complete. If the static shape is not fully 
defined, the dynamic shape of a Tensor t can be determined by evaluating 
tf.shape(t).


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] marcoabreu commented on issue #9129: add tests for distribution generators

[GitBox]

marcoabreu commented on issue #9129: add tests for distribution generators
URL: https://github.com/apache/incubator-mxnet/pull/9129#issuecomment-353280274
 
 
   Please consider removing the print-statements as nose will handle outputting 
the names of all tests and various debug infos around it.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] marcoabreu commented on issue #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

marcoabreu commented on issue #9153: Mkl fix pr (fix an issue in prepare_mkl.sh 
and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#issuecomment-353279792
 
 
   LGTM


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] janelu9 opened a new issue #9161: how to improve the reading speed of a big data that con't be import to the memory one time

[GitBox]

janelu9 opened a new issue #9161: how  to improve the reading speed of a big 
data that con't be import to the memory one time
URL: https://github.com/apache/incubator-mxnet/issues/9161
 
 
   is there a cache like xgboost?
   http://xgboost.readthedocs.io/en/latest/how_to/external_memory.html


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] eric-haibin-lin commented on issue #9085: Add result sections in sparse tutorials

[GitBox]

eric-haibin-lin commented on issue #9085: Add result sections in sparse 
tutorials
URL: https://github.com/apache/incubator-mxnet/pull/9085#issuecomment-353273968
 
 
   @aaronmarkham 


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] eric-haibin-lin commented on a change in pull request #8302: Refactor operators & MKLDNN

[GitBox]

eric-haibin-lin commented on a change in pull request #8302: Refactor operators 
& MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r158207868
 
 

 ##
 File path: src/operator/nn/fully_connected.cc
 ##
 @@ -23,58 +23,165 @@
  * \brief fully connect operator
 */
 #include "./fully_connected-inl.h"
+#include "./mkldnn/mkldnn_ops-inl.h"
+#include "./mkldnn/mkldnn_base-inl.h"
 #if MXNET_USE_NNPACK == 1
 #include "./nnpack/nnpack_fully_connected-inl.h"
 #endif  // MXNET_USE_NNPACK
 
 namespace mxnet {
 namespace op {
-template<>
-Operator* CreateOp(FullyConnectedParam param, int dtype,
-std::vector *in_shape,
-std::vector *out_shape,
-Context ctx) {
-  Operator *op = NULL;
-#if MXNET_USE_NNPACK == 1
-  const size_t batch_size = (*in_shape)[0][0];
-  // nnp_fully_connected_inference will do optimization for batch-size = 1
-  // nnp_fully_connected_output will do optimization for batch-size > 1
-  switch (dtype) {
-  case mshadow::kFloat32:
-return new NNPACKFullyConnectedOp(param);
-  default:
-break;
+
+static bool FullyConnectedShape(const nnvm::NodeAttrs& attrs,
+std::vector *in_shape,
+std::vector *out_shape) {
+  const FullyConnectedParam& param = 
nnvm::get(attrs.parsed);
+  using namespace mshadow;
+  if (!param.no_bias) {
+CHECK_EQ(in_shape->size(), 3U) << "Input:[data, weight, bias]";
+  } else {
+CHECK_EQ(in_shape->size(), 2U) << "Input:[data, weight]";
+  }
+  CHECK_EQ(out_shape->size(), 1U);
+  TShape dshape = (*in_shape)[fullc::kData];
+  TShape oshape = (*out_shape)[0];
+  // require data to be known
+  if (dshape.ndim() ==  0) return false;
+
+  index_t num_input;
+  if (!param.flatten) {
+num_input = dshape[dshape.ndim()-1];
+  } else {
+num_input = dshape.ProdShape(1, dshape.ndim());
+  }
+  SHAPE_ASSIGN_CHECK(*in_shape, fullc::kWeight, Shape2(param.num_hidden, 
num_input));
+  if (!param.no_bias) {
+SHAPE_ASSIGN_CHECK(*in_shape, fullc::kBias, Shape1(param.num_hidden));
+  }
+
+  if (!param.flatten) {
+TShape result_shape(dshape);
+result_shape[dshape.ndim()-1] = param.num_hidden;
+SHAPE_ASSIGN_CHECK(*out_shape, 0, result_shape);
+  } else {
+SHAPE_ASSIGN_CHECK(*out_shape, 0, Shape2(dshape[0], param.num_hidden));
+  }
+  if (oshape.ndim() != 0) {
+dshape[0] = oshape[0];
+SHAPE_ASSIGN_CHECK(*in_shape, fullc::kData, dshape);
+  }
+  return true;
+}
+
+void FullyConnectedCompute_CPU(const nnvm::NodeAttrs& attrs, const OpContext 
&ctx,
+const std::vector &inputs, const std::vector &req,
+const std::vector &outputs) {
+#if MXNET_USE_MKLDNN == 1
+  if (SupportMKLDNN(inputs[0])) {
+MKLDNNFCForward(attrs, ctx, inputs, req, outputs);
+return;
+  }
+#endif
+  std::vector in_blobs(inputs.size());
+  for (size_t i = 0; i < in_blobs.size(); i++)
+in_blobs[i] = inputs[i].data();
+  std::vector out_blobs(outputs.size());
+  for (size_t i = 0; i < out_blobs.size(); i++)
+out_blobs[i] = outputs[i].data();
+  FullyConnectedCompute(attrs, ctx, in_blobs, req, out_blobs);
+}
+
+void FullyConnectedGradCompute_CPU(const nnvm::NodeAttrs& attrs,
+const OpContext &ctx, const std::vector &inputs,
+const std::vector &req, const std::vector &outputs) {
+#if MXNET_USE_MKLDNN == 1
+  if (SupportMKLDNN(inputs[0])) {
+MKLDNNFCBackward(attrs, ctx, inputs, req, outputs);
+return;
   }
 #endif
-  switch (dtype) {
-  case mshadow::kFloat32:
-op = new FullyConnectedOp(param);
-break;
-  case mshadow::kFloat64:
-op = new FullyConnectedOp(param);
-break;
-  case mshadow::kFloat16:
-LOG(FATAL) << "float16 fully connected layer is currently"
-  "only supported by CuDNN version.";
-break;
-  default:
-LOG(FATAL) << "Unsupported type " << dtype;
+  std::vector in_blobs(inputs.size());
+  for (size_t i = 0; i < in_blobs.size(); i++)
+in_blobs[i] = inputs[i].data();
+  std::vector out_blobs(outputs.size());
+  for (size_t i = 0; i < out_blobs.size(); i++)
+out_blobs[i] = outputs[i].data();
+  FullyConnectedGradCompute(attrs, ctx, in_blobs, req, out_blobs);
+}
+
+static bool FullyConnectedType(const nnvm::NodeAttrs& attrs,
+   std::vector *in_type, std::vector 
*out_type) {
+  CHECK_GE(in_type->size(), 1U);
+  return ElemwiseAttr(
+  attrs, in_type, out_type, -1);
+}
+
+struct FullyConnectedGrad {
+  const char *op_name;
+  std::vector operator()(const nnvm::NodePtr& n,
+  const std::vector& 
ograds) const {
+std::vector heads(ograds.begin(), ograds.end());
+heads.push_back(n->inputs[fullc::kData]);
+heads.push_back(n->inputs[fullc::kWeight]);
+return MakeGradNode(op_name, n, heads, n->attrs.dict);
   }
+};
 
-  return op;
+inline static bool FCStorageType(const nnvm::NodeAttrs& attrs,
+ const int dev_mask,

[GitHub] RogerChern commented on issue #8822: Can mx.nd.where(condition, x, y) supports if both x and y are None?

[GitBox]

RogerChern commented on issue #8822: Can mx.nd.where(condition, x, y) supports 
if both x and y are None?
URL: 
https://github.com/apache/incubator-mxnet/issues/8822#issuecomment-353272179
 
 
   Now we have mx.sym.scatter and mx.sym.gather in v1.0.0 which demand a fully 
implemented mx.sym.where. Please take this into consideration.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] ChidanandKumarKS opened a new issue #9160: How to convert data in range -m to +n to data in range 0 to1? Is there are any python API

[GitBox]

ChidanandKumarKS opened a new issue #9160: How to convert data in range -m to 
+n to data in range 0 to1? Is there are any python API
URL: https://github.com/apache/incubator-mxnet/issues/9160
 
 
   I want to covert floating point data in  the range -m to  pythom APIs ava+n 
to the range o to 1.
   Is there any API in mxnet? kindly do the needfull


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] Adesun commented on issue #9102: Cmake option USE_OPERATOR_TUNING on Windows BUG

[GitBox]

Adesun commented on issue #9102: Cmake option USE_OPERATOR_TUNING on Windows BUG
URL: 
https://github.com/apache/incubator-mxnet/issues/9102#issuecomment-353251204
 
 
   same problem. mxnet can be built successfully when enable 
USE_OPERATOR_TUNING,
   but It just hung up  when running script , never execute it.
   turn it off and rebuild it, mxnet works normally.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] Adesun commented on issue #9102: Cmake option USE_OPERATOR_TUNING on Windows BUG

[GitBox]

Adesun commented on issue #9102: Cmake option USE_OPERATOR_TUNING on Windows BUG
URL: 
https://github.com/apache/incubator-mxnet/issues/9102#issuecomment-353251204
 
 
   same problem. mxnet can be built successfully when enable 
USE_OPERATOR_TUNING,
   but It just hung up  when running script , never execute it..


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] chowkamlee81 closed issue #9155: mx.nd.tanh and mx.symbol.tanh yields different results.

[GitBox]

chowkamlee81 closed issue #9155: mx.nd.tanh and mx.symbol.tanh yields different 
results.
URL: https://github.com/apache/incubator-mxnet/issues/9155
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] chowkamlee81 commented on issue #9155: mx.nd.tanh and mx.symbol.tanh yields different results.

[GitBox]

chowkamlee81 commented on issue #9155: mx.nd.tanh and mx.symbol.tanh yields 
different results.
URL: 
https://github.com/apache/incubator-mxnet/issues/9155#issuecomment-353250337
 
 
   @sxjscience , closing the comment as you suggested earlier PR


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] chowkamlee81 commented on issue #9143: maximum value of activation function using tanh goes beyoynd maximum value...

[GitBox]

chowkamlee81 commented on issue #9143: maximum value of activation function 
using tanh goes beyoynd maximum value...
URL: 
https://github.com/apache/incubator-mxnet/issues/9143#issuecomment-353250239
 
 
   @sxjscience , yeah you are correct.
   I did a mistake as the overflow was coming thru because of upsampling API 
which is of default by nature.
   mx.sym.Activation(a, act_type='tanh') rocks as defined but it was my 
mistake. Sorry for trouble
   Thanks for your kind suggestions. Closing the PR.
   @szha thanks for your kind reply. Closing the PR.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] chowkamlee81 closed issue #9143: maximum value of activation function using tanh goes beyoynd maximum value...

[GitBox]

chowkamlee81 closed issue #9143: maximum value of activation function using 
tanh goes beyoynd maximum value...
URL: https://github.com/apache/incubator-mxnet/issues/9143
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] jinhuang415 commented on issue #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

jinhuang415 commented on issue #9153: Mkl fix pr (fix an issue in 
prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#issuecomment-353066075
 
 
   I discussed internally within team and we would like to add this into CI 
unittest, since it only applies to MKLML, so we will create a separate 
directory (tests/python/cpu) to hold the test case and only runs for MKLML 
sanity build. I will update the Jenkinsfile to add this test and update the PR 
later.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] zhaoningning commented on issue #9156: float64 data backward error using gluon

[GitBox]

zhaoningning commented on issue #9156: float64 data backward error using  gluon
URL: 
https://github.com/apache/incubator-mxnet/issues/9156#issuecomment-353238428
 
 
   @sxjscience  But I use float64 data and float64 parameters,  still need to 
cast the loss to float32 ?  I have to use float64 data type   because it may 
generate very small values in the forward process.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] szha closed pull request #9090: Add Wikitext for gluon/rnnlm

[GitBox]

szha closed pull request #9090: Add Wikitext for gluon/rnnlm
URL: https://github.com/apache/incubator-mxnet/pull/9090
 
 
   

This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:

As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):

diff --git a/example/gluon/word_language_model/README.md 
b/example/gluon/word_language_model/README.md
index f200c164a7..ff8ea56b20 100644
--- a/example/gluon/word_language_model/README.md
+++ b/example/gluon/word_language_model/README.md
@@ -3,6 +3,7 @@
 This example trains a multi-layer RNN (Elman, GRU, or LSTM) on Penn Treebank 
(PTB) language modeling benchmark.
 
 The model obtains the state-of-the-art result on PTB using LSTM, getting a 
test perplexity of ~72.
+And ~97 ppl in WikiText-2, outperform than basic LSTM(99.3) and reach 
Variational LSTM(96.3).
 
 The following techniques have been adopted for SOTA results: 
 - [LSTM for LM](https://arxiv.org/pdf/1409.2329.pdf)
@@ -10,20 +11,37 @@ The following techniques have been adopted for SOTA results:
 
 ## Data
 
+### PTB
+
 The PTB data is the processed version from [(Mikolov et al, 
2010)](http://www.fit.vutbr.cz/research/groups/speech/publi/2010/mikolov_interspeech2010_IS100722.pdf):
 
 ```bash
+bash get_ptb_data.sh
 python data.py
 ```
 
+### Wiki Text
+
+The wikitext-2 data is downloaded from [(The wikitext long term dependency 
language modeling 
dataset)](https://www.salesforce.com/products/einstein/ai-research/the-wikitext-dependency-language-modeling-dataset/):
+
+```bash
+bash get_wikitext2_data.sh
+```
+
+
 ## Usage
 
 Example runs and the results:
 
 ```
-python train.py --cuda --tied --nhid 650 --emsize 650 --dropout 0.5# 
Test ppl of 75.3
-python train.py --cuda --tied --nhid 1500 --emsize 1500 --dropout 0.65  # 
Test ppl of 72.0
+python train.py -data ./data/ptb. --cuda --tied --nhid 650 --emsize 650 
--dropout 0.5# Test ppl of 75.3 in ptb
+python train.py -data ./data/ptb. --cuda --tied --nhid 1500 --emsize 1500 
--dropout 0.65  # Test ppl of 72.0 in ptb
+```
+
 ```
+python train.py -data ./data/wikitext-2/wiki. --cuda --tied --nhid 256 
--emsize 256  # Test ppl of 97.07 in wikitext-2 
+```
+
 
 
 
diff --git a/example/gluon/word_language_model/get_wikitext2_data.sh 
b/example/gluon/word_language_model/get_wikitext2_data.sh
new file mode 100755
index 00..e9b8461c40
--- /dev/null
+++ b/example/gluon/word_language_model/get_wikitext2_data.sh
@@ -0,0 +1,36 @@
+#!/usr/bin/env bash
+
+# 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.
+
+
+RNN_DIR=$(cd `dirname $0`; pwd)
+DATA_DIR="${RNN_DIR}/data/"
+
+if [[ ! -d "${DATA_DIR}" ]]; then
+  echo "${DATA_DIR} doesn't exist, will create one";
+  mkdir -p ${DATA_DIR}
+fi
+
+wget -P ${DATA_DIR} 
https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-v1.zip
+cd ${DATA_DIR}
+unzip wikitext-2-v1.zip
+
+# rename
+mv ${DATA_DIR}/wikitext-2/wiki.test.tokens ${DATA_DIR}/wikitext-2/wiki.test.txt
+mv ${DATA_DIR}/wikitext-2/wiki.valid.tokens 
${DATA_DIR}/wikitext-2/wiki.valid.txt
+mv ${DATA_DIR}/wikitext-2/wiki.train.tokens 
${DATA_DIR}/wikitext-2/wiki.train.txt
diff --git a/example/gluon/word_language_model/train.py 
b/example/gluon/word_language_model/train.py
index b419277dcf..eb584b822a 100644
--- a/example/gluon/word_language_model/train.py
+++ b/example/gluon/word_language_model/train.py
@@ -24,7 +24,7 @@
 import data
 
 parser = argparse.ArgumentParser(description='MXNet Autograd PennTreeBank 
RNN/LSTM Language Model')
-parser.add_argument('--data', type=str, default='./data/ptb.',
+parser.add_argument('--data', type=str, default='./data/wikitext-2/wiki.',
 help='location of the data corpus')
 parser.add_argument('--model', type=str, default='lstm',
 help='type of recurrent net (rnn_tanh, rnn_relu, lstm, 
gru)')


 


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

[incubator-mxnet] branch master updated: Add Wikitext for gluon/rnnlm (#9090)

[zhasheng]

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

zhasheng 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 5c3acff  Add Wikitext for gluon/rnnlm (#9090)
5c3acff is described below

commit 5c3acff3b7bdb177a4731094faa724e31387715d
Author: Zihao Zheng 
AuthorDate: Thu Dec 21 09:09:02 2017 +0800

Add Wikitext for gluon/rnnlm (#9090)

* Add wikitext-2 data for rnnlm example in gluon.

* Add Wikitext2 for rnnlm.

* Add performance data in WikiText-2.
---
 example/gluon/word_language_model/README.md| 22 +++--
 .../word_language_model/get_wikitext2_data.sh  | 36 ++
 example/gluon/word_language_model/train.py |  2 +-
 3 files changed, 57 insertions(+), 3 deletions(-)

diff --git a/example/gluon/word_language_model/README.md 
b/example/gluon/word_language_model/README.md
index f200c16..ff8ea56 100644
--- a/example/gluon/word_language_model/README.md
+++ b/example/gluon/word_language_model/README.md
@@ -3,6 +3,7 @@
 This example trains a multi-layer RNN (Elman, GRU, or LSTM) on Penn Treebank 
(PTB) language modeling benchmark.
 
 The model obtains the state-of-the-art result on PTB using LSTM, getting a 
test perplexity of ~72.
+And ~97 ppl in WikiText-2, outperform than basic LSTM(99.3) and reach 
Variational LSTM(96.3).
 
 The following techniques have been adopted for SOTA results: 
 - [LSTM for LM](https://arxiv.org/pdf/1409.2329.pdf)
@@ -10,20 +11,37 @@ The following techniques have been adopted for SOTA results:
 
 ## Data
 
+### PTB
+
 The PTB data is the processed version from [(Mikolov et al, 
2010)](http://www.fit.vutbr.cz/research/groups/speech/publi/2010/mikolov_interspeech2010_IS100722.pdf):
 
 ```bash
+bash get_ptb_data.sh
 python data.py
 ```
 
+### Wiki Text
+
+The wikitext-2 data is downloaded from [(The wikitext long term dependency 
language modeling 
dataset)](https://www.salesforce.com/products/einstein/ai-research/the-wikitext-dependency-language-modeling-dataset/):
+
+```bash
+bash get_wikitext2_data.sh
+```
+
+
 ## Usage
 
 Example runs and the results:
 
 ```
-python train.py --cuda --tied --nhid 650 --emsize 650 --dropout 0.5# 
Test ppl of 75.3
-python train.py --cuda --tied --nhid 1500 --emsize 1500 --dropout 0.65  # 
Test ppl of 72.0
+python train.py -data ./data/ptb. --cuda --tied --nhid 650 --emsize 650 
--dropout 0.5# Test ppl of 75.3 in ptb
+python train.py -data ./data/ptb. --cuda --tied --nhid 1500 --emsize 1500 
--dropout 0.65  # Test ppl of 72.0 in ptb
+```
+
 ```
+python train.py -data ./data/wikitext-2/wiki. --cuda --tied --nhid 256 
--emsize 256  # Test ppl of 97.07 in wikitext-2 
+```
+
 
 
 
diff --git a/example/gluon/word_language_model/get_wikitext2_data.sh 
b/example/gluon/word_language_model/get_wikitext2_data.sh
new file mode 100755
index 000..e9b8461
--- /dev/null
+++ b/example/gluon/word_language_model/get_wikitext2_data.sh
@@ -0,0 +1,36 @@
+#!/usr/bin/env bash
+
+# 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.
+
+
+RNN_DIR=$(cd `dirname $0`; pwd)
+DATA_DIR="${RNN_DIR}/data/"
+
+if [[ ! -d "${DATA_DIR}" ]]; then
+  echo "${DATA_DIR} doesn't exist, will create one";
+  mkdir -p ${DATA_DIR}
+fi
+
+wget -P ${DATA_DIR} 
https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-v1.zip
+cd ${DATA_DIR}
+unzip wikitext-2-v1.zip
+
+# rename
+mv ${DATA_DIR}/wikitext-2/wiki.test.tokens ${DATA_DIR}/wikitext-2/wiki.test.txt
+mv ${DATA_DIR}/wikitext-2/wiki.valid.tokens 
${DATA_DIR}/wikitext-2/wiki.valid.txt
+mv ${DATA_DIR}/wikitext-2/wiki.train.tokens 
${DATA_DIR}/wikitext-2/wiki.train.txt
diff --git a/example/gluon/word_language_model/train.py 
b/example/gluon/word_language_model/train.py
index b419277..eb584b8 100644
--- a/example/gluon/word_language_model/train.py
+++ b/example/gluon/word_language_model/train.py
@@ -24,7 +24,7 @@ import model
 import data
 
 parser = argparse.ArgumentParser(description='MXNet Autograd PennTreeBank 
RNN/LSTM Language Model')
-parser.add_argument('--data', type=str, default='./data/ptb.',
+parser.add_argument('--data', type=str, default='./da

[GitHub] sandeep-krishnamurthy opened a new issue #9159: Symbol.bind() do not warn or error on dtype mismatch between Symbol and binding Data

[GitBox]

sandeep-krishnamurthy opened a new issue #9159: Symbol.bind() do not warn or 
error on dtype mismatch between Symbol and binding Data
URL: https://github.com/apache/incubator-mxnet/issues/9159
 
 
   ## Description
   Symbol bind() API do not warn or error out when user creates a Symbol with 
dtype="int32" and then passes "float32" data. Final result is "float32".
   
   Users who create a symbolic graph with "int32" and by mistake pass on 
float32 data are never warned. It silently converts everything to "float32".
   
   ## Environment info (Required)
   
   Latest MXNet pip install on Mac.
   
   ```
   In [16]: array1 = mx.nd.array([[1,2,3],[1,2,3]])
   
   In [17]: array2 = mx.nd.array([[1,1,1],[1,1,1]])
   
   In [18]: a = mx.sym.Variable('a', dtype='int32')
   
   In [19]: b = mx.sym.Variable('b', dtype='int32')
   
   In [20]: res1 = mx.sym.broadcast_plus(a, b)
   
   In [22]: array1.dtype
   Out[22]: numpy.float32
   
   In [23]: array2.dtype
   Out[23]: numpy.float32
   
   
   In [24]: executor2 = res1.bind(mx.cpu(), args={'a':array1, 'b':array2})
   
   In [25]: output2 = executor2.forward()
   
   In [26]: output2[0]
   Out[26]: 
   
   [[ 2.  3.  4.]
[ 2.  3.  4.]]
   
   
   In [27]: output2[0].dtype
   Out[27]: numpy.float32
   
   
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] szha commented on issue #9143: maximum value of activation function using tanh goes beyoynd maximum value...

[GitBox]

szha commented on issue #9143: maximum value of activation function using tanh 
goes beyoynd maximum value...
URL: 
https://github.com/apache/incubator-mxnet/issues/9143#issuecomment-353224528
 
 
   @chowkamlee81 did you turn on MKL options when getting the incorrect results?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience commented on issue #9143: maximum value of activation function using tanh goes beyoynd maximum value...

[GitBox]

sxjscience commented on issue #9143: maximum value of activation function using 
tanh goes beyoynd maximum value...
URL: 
https://github.com/apache/incubator-mxnet/issues/9143#issuecomment-353218641
 
 
   @chowkamlee81 It's really strange as I can obtain the correct result of 
mx.sym.Activation using the following script.
   
   ```python
   import mxnet as mx
   import numpy as np
   a = mx.sym.var('a')
   b = mx.sym.Activation(a, act_type='tanh')
   exe = b.simple_bind(ctx=mx.gpu(), a=(100,100))
   a_np = np.random.normal(0, 1, (100,100))
   res = exe.forward(a=mx.nd.array(a_np), is_train=True)[0].asnumpy()
   print(res.max(), res.min())
   ```
   Result:
   ```
   0.999537 -0.998478
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] javelinjs commented on issue #9119: fix random generator: do not gen seed each time

[GitBox]

javelinjs commented on issue #9119: fix random generator: do not gen seed each 
time
URL: https://github.com/apache/incubator-mxnet/pull/9119#issuecomment-353217391
 
 
   I have refactor `RandomGenerator` for readability, and add openmp for CPU.  
ping @piiswrong @asmushetzel 
   I also merge @sxjscience 's PR 
https://github.com/apache/incubator-mxnet/pull/9129 in.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] javelinjs commented on issue #9119: fix random generator: do not gen seed each time

[GitBox]

javelinjs commented on issue #9119: fix random generator: do not gen seed each 
time
URL: https://github.com/apache/incubator-mxnet/pull/9119#issuecomment-353217021
 
 
   @piiswrong By using `LaunchRNG` I want to hide the underlying implementation 
to users. Otherwise,
   * Users need to understand gpu state is not thread-safe, and pick a curand 
state in `Op::Map`, then loop the array carefully.
   * The implementation of `Launch` implies two adjacent array entries are 
accessed by two adjacent threads in one block. But in my understanding, we 
should generate successive numbers from one state, i.e., one thread, as much as 
we can. With `Launch`, this will make users suffering calculating the array 
element index, which can easily make mistake and the codes will be weird.
   * The curand states are allocated in global memory. For efficiency, we copy 
it to local memory when launching a kernel (and copy back to global at the end, 
as suggested in NV's doc). Users probably do not want to do it in every 
`Op::Map` function.
   
   I think `LaunchRNG` is a convenient helper function to use. Users can still 
use `Launch` if they want to control everything.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] chowkamlee81 commented on issue #9143: maximum value of activation function using tanh goes beyoynd maximum value...

[GitBox]

chowkamlee81 commented on issue #9143: maximum value of activation function 
using tanh goes beyoynd maximum value...
URL: 
https://github.com/apache/incubator-mxnet/issues/9143#issuecomment-353215545
 
 
   I removed the above 2 lines(2 times upsampling symbol)  as suggested and 
directly binded the symbol conv_scale0_relu. Now the max value of 1.72 which is 
still greater than 1.
   
   I had done one more experiment removed **mx.symbol.activation(tanh)** and 
used ndarray func **mx.nd.tanh** after the **model.forward** function 
   I removed **tanh activation** function and used only previous line
   conv4_scale0 = mx.symbol.Convolution(name='conv4_scale0', 
data=conv3_scale0_relu, num_filter=1, pad=(1, 1),kernel=(3, 3), stride=(1, 1), 
no_bias=True)
   Binded the symbol using model 
conv4_scale0.bind(ctx=mx.gpu(0), args=arg_arrays, args_grad=grad_arrays, 
grad_req=reqs)
   model.forward(is_train=True)
pred= model.outputs[0]
   pred_tanh = mx.nd.tanh(pred,name='tanh_func')
   Now the maximum value of tanh is strictly 1 which is between 1 and -1.
   
   The question is why mx.symbol.activation(tanh)/mx.symbol.tanh results are 
different from mx.nd.tanh?
   Kindly suggest as iam unable to move forward, seeking ur help
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] eric-haibin-lin commented on a change in pull request #8972: Profiling enhancements, python API, vtune and chrome tracing objects, etc.

[GitBox]

eric-haibin-lin commented on a change in pull request #8972: Profiling 
enhancements, python API, vtune and chrome tracing objects, etc.
URL: https://github.com/apache/incubator-mxnet/pull/8972#discussion_r158164957
 
 

 ##
 File path: include/mxnet/c_api.h
 ##
 @@ -227,10 +228,131 @@ MXNET_DLL int MXSetProfilerConfig(int mode, const char* 
filename);
  */
 MXNET_DLL int MXSetProfilerState(int state);
 
-/*! \brief Save profile and stop profiler */
+/*!
+ * \brief Save profile and stop profiler
+ * \param append true if appending to current profile file, false for truncate
+ * \return
+ */
 MXNET_DLL int MXDumpProfile();
 
-/*! \brief Set the number of OMP threads to use */
+/*!
+ * \brief Set whether to continuously write the profiling data to a file
+ * \param continuous_dump true to continuously write profiling data to a file
+ * \param delay_in_seconds Number of seconds (or fraction of seconds) to delay 
between writes
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXSetContinuousProfileDump(int continuous_dump, float 
delay_in_seconds);
+
+/*!
+ * \brief Pause profiler tuning collection
+ * \param paused If nonzero, profiling pauses. Otherwise, profiling 
resumes/continues
+ * \return 0 when success, -1 when failure happens.
+ * \note pausing and resuming is global and not recursive
+ */
+MXNET_DLL int MXProfilePause(int paused);
+
+/*!
+ * \brief Create profiling domain
+ * \param domain String representing the domain name to create
+ * \param out Return domain object
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXProfileCreateDomain(const char *domain, ProfileHandle *out);
+
+/*!
+ * \brief Create profile task
+ * \param name Name of the task
+ * \param domain Domain of the task
+ * \param out Output handle
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXProfileCreateTask(ProfileHandle domain,
+  const char *task_name,
+  ProfileHandle *out);
+
+/*!
+ * \brief Create profile frame
+ * \param name Name of the frame
+ * \param domain Domain of the frame
+ * \param out Output handle
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXProfileCreateFrame(ProfileHandle domain,
+   const char *frame_name,
+   ProfileHandle *out);
+
+/*!
+ * \brief Create profile event
+ * \param name Name of the event
+ * \param out Output handle
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXProfileCreateEvent(const char *event_name, ProfileHandle *out);
+
+/*!
+ * \brief Create profile counter
+ * \param name Name of the counter
+ * \param domain Domain of the counter
+ * \param out Output handle
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXProfileCreateCounter(ProfileHandle domain,
+ const char *counter_name,
+ ProfileHandle *out);
+
+/*!
+ * \brief Destroy a frame
+ * \param frame_handle Handle to frame to destroy
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXProfileDestroyHandle(ProfileHandle frame_handle);
+
+/*!
+ * \brief Start timing the duration of a profile duration object such as an 
event, task or frame
+ * \param duration_handle handle to the duration object
+ * \return 0 when success, -1 when failure happens.
+ */
+MXNET_DLL int MXProfileDurationStart(ProfileHandle duration_handle);
+
+/*!
+ * \brief Stoptiming the duration of a profile duration object such as an 
event, task or frame
 
 Review comment:
   nit: `Stoptiming` -> `Stop timing `


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] eric-haibin-lin commented on issue #9136: Autograd in R-package

[GitBox]

eric-haibin-lin commented on issue #9136: Autograd in R-package
URL: 
https://github.com/apache/incubator-mxnet/issues/9136#issuecomment-353210647
 
 
   @thirdwing any plan to support autograd in R? 


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience closed issue #9158: Should we move the model_zoo.custom_layers to nn.basic_layers?

[GitBox]

sxjscience closed issue #9158: Should we move the model_zoo.custom_layers to 
nn.basic_layers?
URL: https://github.com/apache/incubator-mxnet/issues/9158
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] aaronmarkham commented on a change in pull request #9123: Update module.md to include results of the code snippet

[GitBox]

aaronmarkham commented on a change in pull request #9123: Update module.md to 
include results of the code snippet
URL: https://github.com/apache/incubator-mxnet/pull/9123#discussion_r158144249
 
 

 ##
 File path: docs/tutorials/basic/module.md
 ##
 @@ -62,6 +66,13 @@ net = mx.sym.SoftmaxOutput(net, name='softmax')
 mx.viz.plot_network(net)
 ```
 
+
+
+
+![svg](https://raw.githubusercontent.com/eric-haibin-lin/web-data/eric-haibin-lin-patch-1/mxnet/doc/tutorials/basic/module/output_3_0.svg?sanitize=true)
 
 Review comment:
   Seems like this might be a transient storage point. What about putting on S3?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] aaronmarkham commented on a change in pull request #9123: Update module.md to include results of the code snippet

[GitBox]

aaronmarkham commented on a change in pull request #9123: Update module.md to 
include results of the code snippet
URL: https://github.com/apache/incubator-mxnet/pull/9123#discussion_r158144249
 
 

 ##
 File path: docs/tutorials/basic/module.md
 ##
 @@ -62,6 +66,13 @@ net = mx.sym.SoftmaxOutput(net, name='softmax')
 mx.viz.plot_network(net)
 ```
 
+
+
+
+![svg](https://raw.githubusercontent.com/eric-haibin-lin/web-data/eric-haibin-lin-patch-1/mxnet/doc/tutorials/basic/module/output_3_0.svg?sanitize=true)
 
 Review comment:
   Seems like this might be a transient storage point. What about putting the 
image on S3?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience opened a new issue #9158: Should we move the custom layers to the nn folder?

[GitBox]

sxjscience opened a new issue #9158: Should we move the custom layers to the nn 
folder?
URL: https://github.com/apache/incubator-mxnet/issues/9158
 
 
   Should we move the layers in 
https://github.com/apache/incubator-mxnet/blob/master/python/mxnet/gluon/model_zoo/custom_layers.py
 to `nn.basic_layers`? This file adds the `HybridConcurrent` layer and the 
`Identity` layer.
   
   @szha @piiswrong 


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] anirudh2290 commented on issue #9131: random_uniform causes VM to crash

[GitBox]

anirudh2290 commented on issue #9131: random_uniform causes VM to crash
URL: 
https://github.com/apache/incubator-mxnet/issues/9131#issuecomment-353184879
 
 
   #7335 


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] bhavinthaker commented on issue #9004: test_nccl.py script causes a core dump on P2.16xlarge instance when ran against NCCL enabled MXNet build.

[GitBox]

bhavinthaker commented on issue #9004: test_nccl.py script causes a core dump 
on P2.16xlarge instance when ran against NCCL enabled MXNet build.
URL: 
https://github.com/apache/incubator-mxnet/issues/9004#issuecomment-353178351
 
 
   Update from Nvidia: "Issue has been fixed, and will be part of the next 
release 2.2.1 or later".  
   
   This issue can be kept open for verification in Nvidia NCCL 2.2.1.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] leleamol commented on a change in pull request #9124: Removing the PYTHONPATH inserts from the examples to avoid portabiliity issues.

[GitBox]

leleamol commented on a change in pull request #9124: Removing the PYTHONPATH 
inserts from the examples to avoid portabiliity issues.
URL: https://github.com/apache/incubator-mxnet/pull/9124#discussion_r158113471
 
 

 ##
 File path: example/ctc/ocr_predict.py
 ##
 @@ -21,8 +21,6 @@
 from __future__ import print_function
 import sys, os
 curr_path = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
-sys.path.append("../../amalgamation/python/")
-sys.path.append("../../python/")
 
 from mxnet_predict import Predictor
 
 Review comment:
   I am not very experienced with this code. It would take me some time to 
understand and implement your suggestion.
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] szha commented on a change in pull request #9148: raise warning when we detect Block inside nested list/dict

[GitBox]

szha commented on a change in pull request #9148: raise warning when we detect 
Block inside nested list/dict
URL: https://github.com/apache/incubator-mxnet/pull/9148#discussion_r158112091
 
 

 ##
 File path: python/mxnet/gluon/block.py
 ##
 @@ -230,6 +256,7 @@ def params(self):
 def collect_params(self):
 """Returns a :py:class:`ParameterDict` containing this 
:py:class:`Block` and all of its
 children's Parameters."""
+self._check_container_with_block()
 
 Review comment:
   I agree that collect_params is probably a good place to put this check, 
because that's where the problem of not properly registering the blocks would 
surface.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] marfago commented on issue #9154: random_normal with wrong parameters crashes

[GitBox]

marfago commented on issue #9154: random_normal with wrong parameters crashes
URL: 
https://github.com/apache/incubator-mxnet/issues/9154#issuecomment-353143907
 
 
   Probably would be better to create a new issue with a list where to report 
all the issues like this and #9131 


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience closed issue #9154: random_normal with wrong parameters crashes

[GitBox]

sxjscience closed issue #9154: random_normal with wrong parameters crashes
URL: https://github.com/apache/incubator-mxnet/issues/9154
 
 
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience commented on issue #9154: random_normal with wrong parameters crashes

[GitBox]

sxjscience commented on issue #9154: random_normal with wrong parameters crashes
URL: 
https://github.com/apache/incubator-mxnet/issues/9154#issuecomment-353143032
 
 
   I see. I think we can merge the discussion into the other issue 
https://github.com/apache/incubator-mxnet/issues/9131


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] marfago commented on issue #9154: random_normal with wrong parameters crashes

[GitBox]

marfago commented on issue #9154: random_normal with wrong parameters crashes
URL: 
https://github.com/apache/incubator-mxnet/issues/9154#issuecomment-353142812
 
 
   @sxjscience I agree if the error is wrapped in a python exception. In this 
case the VM exits due to a c++ assert.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience commented on issue #9143: maximum value of activation function using tanh goes beyoynd maximum value...

[GitBox]

sxjscience commented on issue #9143: maximum value of activation function using 
tanh goes beyoynd maximum value...
URL: 
https://github.com/apache/incubator-mxnet/issues/9143#issuecomment-353142661
 
 
   @chowkamlee81 Could you remove these two lines and use `conv4_scale0_relu` 
directly to bind the symbol?
   ```
   conv4_scale0_relu = mx.symbol.UpSampling(conv4_scale0_relu, num_filter=1, 
scale=2, sample_type='bilinear',num_args=2, name="upsampling0")
   
   heat_map = mx.symbol.UpSampling(conv4_scale0_relu, num_filter=1, scale=2, 
sample_type='bilinear', num_args=2,name="upsampling1")
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience commented on issue #9154: random_normal with wrong parameters crashes

[GitBox]

sxjscience commented on issue #9154: random_normal with wrong parameters crashes
URL: 
https://github.com/apache/incubator-mxnet/issues/9154#issuecomment-353142091
 
 
   @marfago Since the standard deviation can never be negative, it's reasonable 
to throw an error.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience commented on issue #9156: float64 data backward error using gluon

[GitBox]

sxjscience commented on issue #9156: float64 data backward error using  gluon
URL: 
https://github.com/apache/incubator-mxnet/issues/9156#issuecomment-353141853
 
 
   @zhaoningning You need to cast the type to float32 explicitly. Use 
`arr.astype(np.float32)` to cast the data type.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] sxjscience commented on issue #9155: mx.nd.tanh and mx.symbol.tanh yields different results.

[GitBox]

sxjscience commented on issue #9155: mx.nd.tanh and mx.symbol.tanh yields 
different results.
URL: 
https://github.com/apache/incubator-mxnet/issues/9155#issuecomment-353141341
 
 
   @chowkamlee81 Would you provide a reproducible example?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] piiswrong commented on issue #9119: fix random generator: do not gen seed each time

[GitBox]

piiswrong commented on issue #9119: fix random generator: do not gen seed each 
time
URL: https://github.com/apache/incubator-mxnet/pull/9119#issuecomment-353140613
 
 
   I don't think we need LaunchRNG. Why not Launch with N = rnd->size() ?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] piiswrong closed pull request #9149: Fix float16 min and max

[GitBox]

piiswrong closed pull request #9149: Fix float16 min and max
URL: https://github.com/apache/incubator-mxnet/pull/9149
 
 
   

This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:

As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):

diff --git a/mshadow b/mshadow
index 984a3a7c25..3d87ed2a4b 16
--- a/mshadow
+++ b/mshadow
@@ -1 +1 @@
-Subproject commit 984a3a7c253a9b590c17206f8d926bdaafdea997
+Subproject commit 3d87ed2a4b47ef749c616f208cee45d920fb6e6e
diff --git a/tests/python/unittest/test_operator.py 
b/tests/python/unittest/test_operator.py
index 19c4e65d3d..ba1b99183f 100644
--- a/tests/python/unittest/test_operator.py
+++ b/tests/python/unittest/test_operator.py
@@ -4668,6 +4668,14 @@ def test_slice_forward_backward(a, index):
 check_numeric_gradient(slice_sym, [in_data])
 
 
+def test_float16_min_max():
+"""Test for issue: https://github.com/apache/incubator-mxnet/issues/9007""";
+a = mx.nd.array([np.finfo('float16').min, np.finfo('float16').max], 
dtype='float16')
+assert a.dtype == np.float16
+assert np.finfo('float16').min == mx.nd.min(a).asscalar()
+assert np.finfo('float16').max == mx.nd.max(a).asscalar()
+
+
 if __name__ == '__main__':
 import nose
 nose.runmodule()


 


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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: Fix float16 min and max (#9149)

[jxie]

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

jxie 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 080d29c  Fix float16 min and max (#9149)
080d29c is described below

commit 080d29cb1a2daa903f99a6fa7c3f5f6dba2771dd
Author: reminisce 
AuthorDate: Wed Dec 20 10:06:29 2017 -0800

Fix float16 min and max (#9149)

* Add unittest for float16 min and max

* Add mshadow fix
---
 mshadow| 2 +-
 tests/python/unittest/test_operator.py | 8 
 2 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/mshadow b/mshadow
index 984a3a7..3d87ed2 16
--- a/mshadow
+++ b/mshadow
@@ -1 +1 @@
-Subproject commit 984a3a7c253a9b590c17206f8d926bdaafdea997
+Subproject commit 3d87ed2a4b47ef749c616f208cee45d920fb6e6e
diff --git a/tests/python/unittest/test_operator.py 
b/tests/python/unittest/test_operator.py
index 19c4e65..ba1b991 100644
--- a/tests/python/unittest/test_operator.py
+++ b/tests/python/unittest/test_operator.py
@@ -4668,6 +4668,14 @@ def test_slice():
 check_numeric_gradient(slice_sym, [in_data])
 
 
+def test_float16_min_max():
+"""Test for issue: https://github.com/apache/incubator-mxnet/issues/9007""";
+a = mx.nd.array([np.finfo('float16').min, np.finfo('float16').max], 
dtype='float16')
+assert a.dtype == np.float16
+assert np.finfo('float16').min == mx.nd.min(a).asscalar()
+assert np.finfo('float16').max == mx.nd.max(a).asscalar()
+
+
 if __name__ == '__main__':
 import nose
 nose.runmodule()

-- 
To stop receiving notification emails like this one, please contact
['"comm...@mxnet.apache.org" '].

[GitHub] sxjscience commented on a change in pull request #9148: raise warning when we detect Block inside nested list/dict

[GitBox]

sxjscience commented on a change in pull request #9148: raise warning when we 
detect Block inside nested list/dict
URL: https://github.com/apache/incubator-mxnet/pull/9148#discussion_r158091622
 
 

 ##
 File path: python/mxnet/gluon/block.py
 ##
 @@ -230,6 +256,7 @@ def params(self):
 def collect_params(self):
 """Returns a :py:class:`ParameterDict` containing this 
:py:class:`Block` and all of its
 children's Parameters."""
+self._check_container_with_block()
 
 Review comment:
   @piiswrong Would it be okay to accept the current approach? The 
collect_params will always be called if the users want to initialize the 
network and they may notice the warning.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] bhavinthaker commented on issue #9004: test_nccl.py script causes a core dump on P2.16xlarge instance when ran against NCCL enabled MXNet build.

[GitBox]

bhavinthaker commented on issue #9004: test_nccl.py script causes a core dump 
on P2.16xlarge instance when ran against NCCL enabled MXNet build.
URL: 
https://github.com/apache/incubator-mxnet/issues/9004#issuecomment-353109885
 
 
   Any update on the two requests above?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] giantpoplar commented on issue #1623: Distributed training over Infiniband

[GitBox]

giantpoplar commented on issue #1623: Distributed training over Infiniband
URL: 
https://github.com/apache/incubator-mxnet/issues/1623#issuecomment-353085091
 
 
   Hi, I met the same issue.Did you succeed to train over ib? @vcodreanu 
   I  replaced the two tcp: by sdp: in  ps-lite/src/van.cc. But It didn't work 
on my environment.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] jinhuang415 commented on issue #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

jinhuang415 commented on issue #9153: Mkl fix pr (fix an issue in 
prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#issuecomment-353066075
 
 
   My previous thought is to run this test by hand when we want to verify if 
MKLML is installed correctly, and this test case only applies to Linux (not 
applied to windows/MAC and other OS) so if needed I can add check for Linux OS 
and skip check for other OS. This test case will pass only when installed with 
MKLML, so if using openblas or GPU it will fail. So if we want to add it into 
the Jenkins we may only test with MKLML scenario. I checked Jenkinsfile, in 
order to add this into CI build, looks like we need to add new def like 
python2[3]_mklml_ut() to only invoke my newly added test case 
(tests/python/cpu/test_mklml.py) and invoked by MKLML build only, doing this 
can make sure all MKLML build linked with correct library for sanity check, but 
it will also bring the Jenkins logic a little more complicated.
   Please let me know your suggestions on it.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] jinhuang415 commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

jinhuang415 commented on a change in pull request #9153: Mkl fix pr (fix an 
issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158021822
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
 
 Review comment:
   Thanks for the suggestion, will update the docstring accordingly.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix 
an issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158004212
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
+when compiling with Intel MKLML library, the method is try
+to import mxnet module and see if correct mklml library is
+mapped to this process's address space
 
 Review comment:
   Copy suggestion: This test will verify that MXNet is built/installed 
correctly when compiled with Intel MKLML library.  The method will try to 
import the mxnet module and see if the mklml library is mapped to this 
process's address space.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix 
an issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158007200
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
 
 Review comment:
   Copy suggestion: This test will verify that MXNet is built/installed 
correctly when compiled with Intel MKLML library.  The method will try to 
import the mxnet module and see if the mklml library is mapped to this 
process's address space.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix 
an issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158003130
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
+when compiling with Intel MKLML library, the method is try
+to import mxnet module and see if correct mklml library is
+mapped to this process's address space
+"""
+logging.basicConfig(level=logging.INFO)
+try:
+#pylint: disable=unused-variable
+import mxnet as mx
+except (ImportError, OSError) as e:
+assert 0, "Import mxnet error: %s. Please double check your build/" \
+   "install steps or environment variable settings" % str(e)
+
+pid = os.getpid()
+rc = os.system("cat /proc/" + str(pid) + \
 
 Review comment:
   Does this work on windows?  If not, would it be possible to detect if you're 
running in Windows and skip the test?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix 
an issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158002621
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
+when compiling with Intel MKLML library, the method is try
+to import mxnet module and see if correct mklml library is
+mapped to this process's address space
+"""
+logging.basicConfig(level=logging.INFO)
 
 Review comment:
   Will this change the global logging level?  Would this be a problem if say a 
user wanted to run the entire test suit in warning only mode?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix 
an issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158004212
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
+when compiling with Intel MKLML library, the method is try
+to import mxnet module and see if correct mklml library is
+mapped to this process's address space
 
 Review comment:
   Copy suggestion: This test will verify that MXNet is built/installed 
correctly when compiled with Intel MKLML library.  The method will try to 
import the mxnet module and see if the mklml library is mapped to this 
process's address space.


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix 
an issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158003130
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
+when compiling with Intel MKLML library, the method is try
+to import mxnet module and see if correct mklml library is
+mapped to this process's address space
+"""
+logging.basicConfig(level=logging.INFO)
+try:
+#pylint: disable=unused-variable
+import mxnet as mx
+except (ImportError, OSError) as e:
+assert 0, "Import mxnet error: %s. Please double check your build/" \
+   "install steps or environment variable settings" % str(e)
+
+pid = os.getpid()
+rc = os.system("cat /proc/" + str(pid) + \
 
 Review comment:
   Does this work on windows?  If not, would it be possible to detect if you're 
running in Windows and skip the test?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix an issue in prepare_mkl.sh and add test case)

[GitBox]

KellenSunderland commented on a change in pull request #9153: Mkl fix pr (fix 
an issue in prepare_mkl.sh and add test case)
URL: https://github.com/apache/incubator-mxnet/pull/9153#discussion_r158002621
 
 

 ##
 File path: tests/python/cpu/test_mklml.py
 ##
 @@ -0,0 +1,52 @@
+# 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.
+
+"""
+MKLML related test cases
+"""
+
+import logging
+import os
+
+def test_mklml_install():
+"""
+This function will check if MXNet is built/installed correctly
+when compiling with Intel MKLML library, the method is try
+to import mxnet module and see if correct mklml library is
+mapped to this process's address space
+"""
+logging.basicConfig(level=logging.INFO)
 
 Review comment:
   Will this change the global logging level?  Would this be a problem if say a 
user wanted to run the entire test suit in warning only mode?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] schlichtanders opened a new issue #9157: YARN support data locality

[GitBox]

schlichtanders opened a new issue #9157: YARN support data locality
URL: https://github.com/apache/incubator-mxnet/issues/9157
 
 
   Hadoop / YARN is famous for bringing computation to the data and not vice 
versa. Does MXNet's YARN support also supports this kind of data locality?
   If not, is it easy to add to the YARN client?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] schlichtanders commented on issue #533: [YARN] MXNet on YARN Example

[GitBox]

schlichtanders commented on issue #533: [YARN] MXNet on YARN Example
URL: https://github.com/apache/incubator-mxnet/issues/533#issuecomment-353041987
 
 
   why was this closed if it is not resolved?


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] asmushetzel commented on a change in pull request #9119: fix random generator: do not gen seed each time

[GitBox]

asmushetzel commented on a change in pull request #9119: fix random generator: 
do not gen seed each time
URL: https://github.com/apache/incubator-mxnet/pull/9119#discussion_r158001140
 
 

 ##
 File path: src/operator/mxnet_op.h
 ##
 @@ -448,6 +449,23 @@ struct Kernel {
 #endif
   }
 
+  /*!
+   * \brief Launch a generic CPU kernel with native random generator.
+   * \tparam rnd CPU random generator
+   * \tparam Args Varargs type to eventually pass to the OP::Map() functoion
+   * \param N Number of iterations
+   * \param args Varargs to eventually pass to the OP::Map() functoion
+   */
+  template
+  inline static void LaunchNativeRandomGenerator(mshadow::Stream *,
+ 
common::random::RandGenerator *rnd,
+ const int N, Args... args) {
+// do not use openmp since it does not guarantee the output order.
 
 Review comment:
   Adopting the same design would solve this ordering issue. And we don't need 
to pre-allocate thousands of CPU-samplers, 256 would certainly be enough. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] zhaoningning opened a new issue #9156: float64 data backward error using gluon

[GitBox]

zhaoningning opened a new issue #9156: float64 data backward error using  gluon
URL: https://github.com/apache/incubator-mxnet/issues/9156
 
 
   I write a custom loss in gluon,and when  using  float32 data type, 
everything is ok, but  whien I changed to  use float64 data type,there is a  
error says: 
   ?include/mxnet/././tensor_blob.h:217: Check failed: 
mshadow::DataType::kFlag == type_flag_ TBlob.get_with_shape: data type 
do not match specified type.Expected: 0 v.s. given 1?? 
   this happend   after  loss is calculated ,when  loss.backward ()  is 
executed.
mxnet version is 1.0.0, ubuntu 14.04,python2.7


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] feevos commented on issue #9138: I cannot make mxnet.ndarray.UpSampling work with bilinear interpolation

[GitBox]

feevos commented on issue #9138: I cannot make mxnet.ndarray.UpSampling work 
with bilinear interpolation
URL: 
https://github.com/apache/incubator-mxnet/issues/9138#issuecomment-353008453
 
 
   @chowkamlee81  
   
   I tried the code you've sent me (exact copy/paste), I again obtain the same 
error - can this be a version related error? In addition, from the 
documentation of 
[ndarray.UpSampling](https://mxnet.incubator.apache.org/api/python/ndarray/ndarray.html#mxnet.ndarray.UpSampling)
 the ```num_filter``` variable is not related (at least to my understanding) to 
the number of channels of the input, e.g. for input ```shape = 
[2,16,256,256]``` the variable ```num_filter``` is not related to the number 
```shape[1]==16```, it is somehow related to the bilinear interpolation. By the 
way, the argument ```num_args``` doesn't appear in the official documentation 
either.  
   
   Thanks!
   
   ## Code:
   
   ```Python
   import mxnet as mx
   import mxnet.ndarray as nd
   
   xx = nd.random_normal(shape=[1,1,256,256],ctx=mx.cpu()) ###(NCHW)
   temp = nd.UpSampling(xx, num_filter=1, scale=2, 
sample_type='bilinear',num_args=2,name="upsampling1")
   ```
   
   
   ## Error
   
   ```Python
   ---
   MXNetErrorTraceback (most recent call last)
in ()
 3 
 4 xx = nd.random_normal(shape=[1,1,256,256],ctx=mx.cpu()) ###(NCHW)
   > 5 temp = nd.UpSampling(xx, num_filter=1, scale=2, 
sample_type='bilinear', num_args=2,name="upsampling1")
   
   
/home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/ndarray/register.pyc 
in UpSampling(*data, **kwargs)
   
   /home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/_ctypes/ndarray.pyc 
in _imperative_invoke(handle, ndargs, keys, vals, out)
90 c_str_array(keys),
91 c_str_array([str(s) for s in vals]),
   ---> 92 ctypes.byref(out_stypes)))
93 
94 if original_output is not None:
   
   /home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/base.pyc in 
check_call(ret)
   144 """
   145 if ret != 0:
   --> 146 raise MXNetError(py_str(_LIB.MXGetLastError()))
   147 
   148 
   
   MXNetError: [17:05:49] src/c_api/../imperative/imperative_utils.h:303: Check 
failed: num_inputs == infered_num_inputs (1 vs. 2) Operator UpSampling expects 
2 inputs, but got 1 instead.
   
   Stack trace returned 10 entries:
   [bt] (0) 
/home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/libmxnet.so(+0x289a1c) 
[0x7f6e699d6a1c]
   [bt] (1) 
/home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/libmxnet.so(+0x240538f)
 [0x7f6e6bb5238f]
   [bt] (2) 
/home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/libmxnet.so(+0x24029a2)
 [0x7f6e6bb4f9a2]
   [bt] (3) 
/home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/libmxnet.so(MXImperativeInvokeEx+0x63)
 [0x7f6e6bb4ffb3]
   [bt] (4) 
/home/dia021/anaconda2/lib/python2.7/lib-dynload/_ctypes.so(ffi_call_unix64+0x4c)
 [0x7f6ea919357c]
   [bt] (5) 
/home/dia021/anaconda2/lib/python2.7/lib-dynload/_ctypes.so(ffi_call+0x1f5) 
[0x7f6ea9192cd5]
   [bt] (6) 
/home/dia021/anaconda2/lib/python2.7/lib-dynload/_ctypes.so(_ctypes_callproc+0x3e6)
 [0x7f6ea918a376]
   [bt] (7) 
/home/dia021/anaconda2/lib/python2.7/lib-dynload/_ctypes.so(+0x9db3) 
[0x7f6ea9181db3]
   [bt] (8) 
/home/dia021/anaconda2/bin/../lib/libpython2.7.so.1.0(PyObject_Call+0x53) 
[0x7f6eae213e93]
   [bt] (9) 
/home/dia021/anaconda2/bin/../lib/libpython2.7.so.1.0(PyEval_EvalFrameEx+0x715d)
 [0x7f6eae2c680d]
   ```


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] chowkamlee81 commented on issue #9138: I cannot make mxnet.ndarray.UpSampling work with bilinear interpolation

[GitBox]

chowkamlee81 commented on issue #9138: I cannot make mxnet.ndarray.UpSampling 
work with bilinear interpolation
URL: 
https://github.com/apache/incubator-mxnet/issues/9138#issuecomment-353004436
 
 
   Your shape is shape =[2,16,256,256]
   But number of filter =2 in your code
   Also in your code batch size=2
   
   check below code :
   xx = nd.random_normal(shape=[1,1,256,256],ctx=mx.cpu()) ###(NCHW)
   temp = nd.UpSampling(xx, num_filter=1, scale=2, sample_type='bilinear', 
num_args=2,name="upsampling1")
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on 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] Jing-Luo commented on issue #8603: Contrib operators for object-detection bounding box related stuffs

[GitBox]

Jing-Luo commented on issue #8603: Contrib operators for object-detection 
bounding box related stuffs
URL: https://github.com/apache/incubator-mxnet/pull/8603#issuecomment-352992699
 
 
   @piiswrong @cjolivier01 why this pr hasn't been merged yet?


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