Skip to content

Merge master into release/preview for 0.7#1469

Merged
shauheen merged 161 commits intodotnet:release/previewfrom
shauheen:rc107
Oct 31, 2018
Merged

Merge master into release/preview for 0.7#1469
shauheen merged 161 commits intodotnet:release/previewfrom
shauheen:rc107

Conversation

@shauheen
Copy link
Contributor

@shauheen shauheen commented Oct 31, 2018

This PR merges master into release branch for 0.7

shauheen and others added 30 commits September 26, 2018 13:13
* Remove the error tracing when assembly loading fails for Maml.

Also adding our native assemblies to the list to skip, so they aren't attempted to be loaded.

Fix #1034
Helps the user to relate to the macOS version faster.
* Multiclass logistic Regression tests enabled

* threshold providing in tests

* defining tolerance as a constant in baseTestBaseline Class

* upper case camel for constant and _ for large decimal numbers
…800)

* clarifying roadmap to mention current text/NLP features

* updated the documentation to fix review comments
* updating documentation of TensorFlowTransform

* making some some changes in the description
* (Fix issue #1050) Renamed namespace to Microsoft.ML.Transforms to conform with TF transform.

* (Fix issue #1051) Added public create methods

* (Fixed issue 1053) IDV names don't have to be the same as OnnxModel node names.

* Added license header
* move Harness-related code to Harness folder

* make sure that we always use recommended config

* mention the external dependencies in the README

* add nuget.config file so BDN can restore all packages

* add comments to the config so I am not the only person who understands it

* don't enable MemoryDiagnoser by default, it requires one extra iteration which is expensive for long running benchmarks

* don't add nuget.config file, generate it on the fly when needed by BDN

* generate a .csproj file that will handle both native dependencies and nuget.config file issue

* describe authoring new benchmarks in the docs

* add some integration tests that make sure that the benchmarks are not broken

* register the right assemblies after recent change of assembly loading, makes all benchmark work again ;)

* make Ranking benchmarks work

* code review: split Helpers.cs into multiple files, cleanup the code, don't hardcode the dependencies
* Change all uses of System.Linq.Append to System.Linq.Concat.

* Address code review comments
* Kmeans to estimator.
* Adding the Clustering training context, the clustering Evaluate method, the KMeans extension on the clustering context and a tet for it.
… API. (#1082)

This way, all components are registered before the Experiment tries to instantiate them.

Fix #1042
* update ml.scoring to stable version

* small change to kick off mac build

* small change to kick off build

* Kick off build

* small change to kick off build
* simple fixe for warning issue in keytovalue transform

* changed test file outputs so that they don't include the warnings
* AP xtensions
* lbfgs derived classes take more arguments in their public ctors
* adding pigstensions for lr, multilr, possion
* Ogd static xtensions.
* namespace change for pigstensions
Fixes #987 
Adds a document describing high-level API concepts, as well as the 'cookbook' with a variety of samples.
* Adding the Ranker TrainContext, the Ranker TrainerEstimatorReconcilier, and an Evaluate method + metrics class to the existing RankerEvaluator.
* Adding the FastTree ranking xtension and test.
* Grouping the xtensions in classes with more meaningful names, since the docs site displays the methods per class, not file.
Adds project references for OnnxTransform, TensorFlowTransorm, and a NativeAssemblyReference for SymSGD
* Add a workaround for the tests hanging while loading MKL.

The workaround is to ensure the MKL library is loaded very early in the test process, so it doesn't cause the deadlock.

Workaround #1073

Another deadlock also occurs when running TestAutoInference and TestPipelineSweeper in parallel. Marking these tests to not run in parallel anymore.

Workaround #1095

Moving back to the Azure Hosted VS2017 pool to run the tests now that we've narrowed the deadlocks down.
…1032)

* Add instructions for building for .NET Core 3.0, and make them work.

Fix #1011

* Add config specific properties for the Intrinsics configs.

* Allow tests to be run against .NET Core 3.0
shauheen and others added 20 commits October 29, 2018 15:39
* Enable statically-typed matrix factorization

* Address comments
1. Add copyright
2. Try fix mac tolerance
3. Use MLContext with static pipeline

* Add another example for in-memory matrix factorization
…1328)

* Renaming the namespaces where the transforms live from Runtime.Data.Transforms to ML.Transforms.
Addressing PR comments.
* Change TryParse* methods to return false instead of throw.

* Make TextLoader throw on bad values

* Make TextLoader throw on bad values, and fix unit tests.

* Update Release baseline

* Fix one more unit test
* Adding the catalog extensions for  Concat, CopyColumns, Hash, KeyToVal
* adding airquality, infert datasets. Added samples for Term and KeyToVal estimators.

* Adding tests for the NormalizerCatalog and the TextTransformCatalog.
Adding samples for the ConcatEstimator, and KeyToValue, Term that will need to get referenced from the respective catalogs when they happen.

re-organized the samples based on static-dynamic. Renamed.
…olumnsTransform (#1371)

* Removes ChooseColumnsTransform and DropColumnsTransform classes
replacing them with SelectColumnsTransform. These changes include:
* Updates to SelectColumnsTransform to respect ordering when keeping
columns. For example, if the input is ABC and CB is selected, the output
will be CB.
* Updates to code that used Choose or Drop columns, replacing with
SelectColumns.
* Updates to baseline output for tests to pass
* Re-enabled the SavePipeline tests

This fixes #1342
These changes are also related to #754
* Enhancements to online linear trainers to make them stateless.

* Factor stateful logic into a separate internal object.
* Remove direct usage of Console.Writeline
* Opportunistic fixes of minor issues.
* Nuke failing Mac test temporarily
* more reorg and namespace move

* renaming the HashEstimator and taking care of consequences
Added a custom mapping transformer/estimator
* adding all extensions for the Text related transformation

* adding keyToVector extensions and renaming the estimatorto conform to 1318

* Adding SelectColumns xtensions. Renaming SelectEstimator and Textnormalizer

* KeyToBinary extensions

* Adding extensions for the Image transforms. Some renaming
Fix an unassigned field for matrix factorization.
# Conflicts:
#	build/Dependencies.props
#	src/Microsoft.ML.Data/Evaluators/ClusteringEvaluator.cs
#	src/Microsoft.ML.Data/Scorers/PredictionTransformer.cs
#	src/Microsoft.ML.KMeansClustering/KMeansPlusPlusTrainer.cs
#	src/Microsoft.ML.KMeansClustering/KMeansStatic.cs
#	src/Microsoft.ML.Legacy/AssemblyRegistration.cs
#	src/Microsoft.ML.PCA/WrappedPcaTransform.cs
#	src/Microsoft.ML.StandardLearners/Standard/LogisticRegression/LbfgsPredictorBase.cs
#	src/Microsoft.ML.StandardLearners/Standard/LogisticRegression/MulticlassLogisticRegression.cs
#	src/Microsoft.ML.StandardLearners/Standard/Online/AveragedLinear.cs
#	src/Microsoft.ML.StandardLearners/Standard/Online/AveragedPerceptron.cs
#	src/Microsoft.ML.StandardLearners/Standard/Online/LinearSvm.cs
#	src/Microsoft.ML.StandardLearners/Standard/Online/OnlineGradientDescent.cs
#	src/Microsoft.ML.StandardLearners/Standard/Online/OnlineLearnerStatic.cs
#	src/Microsoft.ML.StandardLearners/Standard/Online/OnlineLinear.cs
#	src/Microsoft.ML.TensorFlow/doc.xml
#	src/Microsoft.ML.Transforms/NAReplaceTransform.cs
#	src/Microsoft.ML.Transforms/Text/TextTransform.cs
#	test/Microsoft.ML.StaticPipelineTesting/Training.cs
#	test/Microsoft.ML.Tests/TrainerEstimators/LbfgsTests.cs
#	test/Microsoft.ML.Tests/TrainerEstimators/TrainerEstimators.cs
@shauheen shauheen requested review from TomFinley and sfilipi October 31, 2018 20:06
@shauheen shauheen merged commit c5cef31 into dotnet:release/preview Oct 31, 2018
@shauheen shauheen deleted the rc107 branch October 31, 2018 20:22
@ghost ghost locked as resolved and limited conversation to collaborators Mar 27, 2022
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.