Codestin Search App

lanking520 · 2018-06-14T21:39:19Z

Description

This PR contains examples written from the new API.
@nswamy @yzhliu @andrewfayres
All examples pushed here are tested locally on my Mac. It lives in the new package called org.apache.mxnet.examples
Be aware: CNN Example is specifically designed to run on GPU. Disable the CPU

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant JIRA issue created (except PRs with tiny changes)
Changes are complete (i.e. I finished coding on this PR)
All changes have test coverage:
Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
Code is well-documented:
For user-facing API changes, API doc string has been updated.
For new C++ functions in header files, their functionalities and arguments are documented.
For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$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

lanking520 · 2018-06-15T23:04:14Z

Currently facing the CI problem as shown below:

- Example CI - CNN Example *** FAILED ***

  java.nio.charset.MalformedInputException: Input length = 1

  at java.nio.charset.CoderResult.throwException(CoderResult.java:281)

  at sun.nio.cs.StreamDecoder.implRead(StreamDecoder.java:339)

  at sun.nio.cs.StreamDecoder.read(StreamDecoder.java:178)

  at java.io.InputStreamReader.read(InputStreamReader.java:184)

  at java.io.BufferedReader.read1(BufferedReader.java:210)

  at java.io.BufferedReader.read(BufferedReader.java:286)

  at java.io.Reader.read(Reader.java:140)

  at scala.io.BufferedSource.mkString(BufferedSource.scala:96)

  at org.apache.mxnet.examples.cnntextclassification.DataHelper$.loadMRDataAndLabels(DataHelper.scala:53)

  at org.apache.mxnet.examples.cnntextclassification.DataHelper$.loadMSDataWithWord2vec(DataHelper.scala:166)

  ...

I add a fix on codec to see if we can solve that... Here is the link I found helpful. Cannot reproduce using a ubuntu machine

lanking520 · 2018-06-18T01:41:50Z

The previous problem solved, currently facing the memory leak issues:

- Example CI - CNN Example *** FAILED ***

  org.apache.mxnet.MXNetError: [23:56:58] src/storage/./pooled_storage_manager.h:118: cudaMalloc failed: out of memory



Stack trace returned 10 entries:

[bt] (0) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(dmlc::StackTrace[abi:cxx11]()+0x1bc) [0x7f74407f0f1c]

[bt] (1) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(dmlc::LogMessageFatal::~LogMessageFatal()+0x28) [0x7f74407f2138]

[bt] (2) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(mxnet::storage::GPUPooledStorageManager::Alloc(mxnet::Storage::Handle*)+0x159) [0x7f744342e7c9]

[bt] (3) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(mxnet::StorageImpl::Alloc(mxnet::Storage::Handle*)+0x5d) [0x7f744343084d]

[bt] (4) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(mxnet::NDArray::CheckAndAlloc() const+0x238) [0x7f74409a9048]

[bt] (5) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(+0x33c19f0) [0x7f7442f429f0]

[bt] (6) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(mxnet::imperative::PushFCompute(std::function<void (nnvm::NodeAttrs const&, mxnet::OpContext const&, std::vector<mxnet::TBlob, std::allocator<mxnet::TBlob> > const&, std::vector<mxnet::OpReqType, std::allocator<mxnet::OpReqType> > const&, std::vector<mxnet::TBlob, std::allocator<mxnet::TBlob> > const&)> const&, nnvm::Op const*, nnvm::NodeAttrs const&, mxnet::Context const&, std::vector<mxnet::engine::Var*, std::allocator<mxnet::engine::Var*> > const&, std::vector<mxnet::engine::Var*, std::allocator<mxnet::engine::Var*> > const&, std::vector<mxnet::Resource, std::allocator<mxnet::Resource> > const&, std::vector<mxnet::NDArray*, std::allocator<mxnet::NDArray*> > const&, std::vector<mxnet::NDArray*, std::allocator<mxnet::NDArray*> > const&, std::vector<unsigned int, std::allocator<unsigned int> > const&, std::vector<mxnet::OpReqType, std::allocator<mxnet::OpReqType> > const&)::{lambda(mxnet::RunContext)#1}::operator()(mxnet::RunContext) const+0x1d5) [0x7f7442f6a0c5]

[bt] (7) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(+0x3896c7b) [0x7f7443417c7b]

[bt] (8) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(mxnet::engine::ThreadedEngine::ExecuteOprBlock(mxnet::RunContext, mxnet::engine::OprBlock*)+0x8e5) [0x7f7443411575]

[bt] (9) /work/mxnet/scala-package/native/linux-x86_64-gpu/target/libmxnet-scala-linux-x86_64-gpu.so(void mxnet::engine::ThreadedEnginePerDevice::GPUWorker<(dmlc::ConcurrentQueueType)0>(mxnet::Context, bool, mxnet::engine::ThreadedEnginePerDevice::ThreadWorkerBlock<(dmlc::ConcurrentQueueType)0>*, std::shared_ptr<dmlc::ManualEvent> const&)+0xeb) [0x7f744342869b]

  at org.apache.mxnet.Base$.checkCall(Base.scala:131)

  at org.apache.mxnet.NDArray.internal(NDArray.scala:942)

  at org.apache.mxnet.NDArray.toArray(NDArray.scala:935)

  at org.apache.mxnet.NDArrayFuncReturn.toArray(NDArray.scala:1152)

  at org.apache.mxnet.examples.cnntextclassification.CNNTextClassification$$anonfun$trainCNN$1$$anonfun$apply$mcVI$sp$1.apply$mcVI$sp(CNNTextClassification.scala:150)

  at scala.collection.immutable.Range.foreach$mVc$sp(Range.scala:160)

  at org.apache.mxnet.examples.cnntextclassification.CNNTextClassification$$anonfun$trainCNN$1.apply$mcVI$sp(CNNTextClassification.scala:127)

  at scala.collection.immutable.Range.foreach$mVc$sp(Range.scala:160)

  at org.apache.mxnet.examples.cnntextclassification.CNNTextClassification$.trainCNN(CNNTextClassification.scala:121)

  at org.apache.mxnet.examples.cnntextclassification.CNNTextClassification$.test(CNNTextClassification.scala:248)

  ...

Will try to look into the code to see if we are not disposing some ndarrays

lanking520 · 2018-06-18T16:20:17Z

Currently, I have switched from epoch=200 into 30 and set dropout to 0.0 to match Python configuration. It seemed the training result doesn't looks good

nswamy · 2018-06-29T11:30:50Z

@lanking520
Though not ideal this example is not tuned to achieve great accuracy. I think we should call this out explicitly in the README that the example is for illustration purposes only at this time.

nswamy · 2018-06-29T11:31:32Z

@@ -0,0 +1,23 @@
+# CNN Text Classification Example for Scala
+This is the example using Scala type-safe api doing CNN text classification. 
+Currently, I cannot reproduce the same result in the python example here. 


This example is only for Illustration and not modeled to achieve the best accuracy.

nswamy · 2018-07-03T21:56:32Z

+This is the example using Scala type-safe api doing CNN text classification. 
+This example is only for Illustration and not modeled to achieve the best accuracy.
+
+Please contribute to improve the dev accuracy of the model.


nswamy · 2018-07-03T23:47:00Z

- */
+  * An Implementation of the paper
+  * Convolutional Neural Networks for Sentence Classification
+  */


nswamy · 2018-07-04T00:16:56Z

-    val sm = Symbol.SoftmaxOutput()()(Map("data" -> fc, "label" -> inputY))
+    val fc = Symbol.api.FullyConnected(data = Some(hDrop), num_hidden = numLabel)
+    val sm = Symbol.api.SoftmaxOutput(data = Some(fc), label = Some(inputY))
+    fc.dispose()


have you checked that this does not crash. Isn't SoftmaxOutput using FC during training

Yes, I have. Not crashing

yzhliu · 2018-07-11T18:29:13Z

Can we merge?

* init commit for CNN * Add changes to pass CPU and GPU test * Add codec to solve the mkstring issue * Change dropout and epoch number * change epoch to 10 and Java download * add README.md * adding dispose method to avoid memory leaks * dispose unused Symbols

lanking520 requested a review from yzhliu as a code owner June 14, 2018 21:39

lanking520 changed the title ~~[MXNET-531] CNN Examples for Scala new APi~~ [MXNET-531] CNN Examples for Scala new API Jun 25, 2018

lanking520 force-pushed the example-cnn branch 2 times, most recently from acddbe1 to 147972c Compare June 26, 2018 20:12

nswamy reviewed Jun 29, 2018

View reviewed changes

lanking520 force-pushed the example-cnn branch 2 times, most recently from e4729aa to 539c091 Compare July 3, 2018 22:09

nswamy reviewed Jul 4, 2018

View reviewed changes

lanking520 added 11 commits July 11, 2018 13:16

init commit for CNN

030bfb8

Add changes to pass CPU and GPU test

b40f55d

Add codec to solve the mkstring issue

8006d8e

Change dropout and epoch number

4a0a839

change epoch to 10 and Java download

921bfc0

add README.md

7ecc1d4

adding dispose method to avoid memory leaks

61766c6

dispose unused Symbols

997f7ec

Remove the author tag and improve the documentation

0a552bc

move the files

bcf6039

remove white space

3730150

lanking520 force-pushed the example-cnn branch from 539c091 to 3730150 Compare July 11, 2018 20:16

nswamy merged commit 5495e2a into apache:master Jul 16, 2018

lanking520 deleted the example-cnn branch July 17, 2018 21:57

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[MXNET-531] CNN Examples for Scala new API#11292

[MXNET-531] CNN Examples for Scala new API#11292
nswamy merged 11 commits into
apache:masterfrom
lanking520:example-cnn

lanking520 commented Jun 14, 2018 •

edited

Loading

Uh oh!

lanking520 commented Jun 15, 2018 •

edited

Loading

Uh oh!

lanking520 commented Jun 18, 2018

Uh oh!

lanking520 commented Jun 18, 2018

Uh oh!

nswamy commented Jun 29, 2018

Uh oh!

nswamy Jun 29, 2018

Uh oh!

nswamy Jul 3, 2018

Uh oh!

nswamy Jul 3, 2018

Uh oh!

nswamy Jul 4, 2018 •

edited

Loading

Uh oh!

lanking520 Jul 4, 2018

Uh oh!

yzhliu commented Jul 11, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

lanking520 commented Jun 14, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Checklist

Essentials

Uh oh!

lanking520 commented Jun 15, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lanking520 commented Jun 18, 2018

Uh oh!

lanking520 commented Jun 18, 2018

Uh oh!

nswamy commented Jun 29, 2018

Uh oh!

nswamy Jun 29, 2018

Choose a reason for hiding this comment

Uh oh!

nswamy Jul 3, 2018

Choose a reason for hiding this comment

Uh oh!

nswamy Jul 3, 2018

Choose a reason for hiding this comment

Uh oh!

nswamy Jul 4, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lanking520 Jul 4, 2018

Choose a reason for hiding this comment

Uh oh!

yzhliu commented Jul 11, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

lanking520 commented Jun 14, 2018 •

edited

Loading

lanking520 commented Jun 15, 2018 •

edited

Loading

nswamy Jul 4, 2018 •

edited

Loading