Add missing linePragmas to ConsumeModel.tt (dotnet#4401)

* Add missing linePragmas to ConsumeModel.tt Fixes dotnet#4400 rebase commits
frank-dong-ms-zz · Nov 4, 2019 · 11cdcb7 · 11cdcb7
1 parent de6e5a2
commit 11cdcb7
Show file tree

Hide file tree

Showing 68 changed files with 660 additions and 242 deletions.
diff --git a/build/ci/job-template.yml b/build/ci/job-template.yml
@@ -10,7 +10,7 @@ parameters:
 jobs:
   - job: ${{ parameters.name }}
     ${{ if eq(parameters.codeCoverage, 'false') }}:
-      timeoutInMinutes: 60
+      timeoutInMinutes: 75
     ${{ if eq(parameters.codeCoverage, 'true') }}:
       timeoutInMinutes: 60
     strategy:

diff --git a/docs/samples/Microsoft.ML.AutoML.Samples/Microsoft.ML.AutoML.Samples.csproj b/docs/samples/Microsoft.ML.AutoML.Samples/Microsoft.ML.AutoML.Samples.csproj
@@ -10,5 +10,7 @@
     <NativeAssemblyReference Include="MatrixFactorizationNative" />
     <NativeAssemblyReference Include="FastTreeNative" />
   </ItemGroup>
-
+  <ItemGroup>
+    <PackageReference Include="SciSharp.TensorFlow.Redist" Version="$(TensorFlowVersion)" />
+  </ItemGroup>
 </Project>
diff --git a/pkg/Microsoft.ML.AutoML/Microsoft.ML.AutoML.nupkgproj b/pkg/Microsoft.ML.AutoML/Microsoft.ML.AutoML.nupkgproj
@@ -10,6 +10,8 @@
     <ProjectReference Include="../Microsoft.ML.LightGbm/Microsoft.ML.LightGbm.nupkgproj" />
     <ProjectReference Include="../Microsoft.ML.Mkl.Components/Microsoft.ML.Mkl.Components.nupkgproj" />
     <ProjectReference Include="../Microsoft.ML.Recommender/Microsoft.ML.Recommender.nupkgproj" />
+    <ProjectReference Include="../Microsoft.ML/Microsoft.ML.Dnn.nupkgproj" />
+    <ProjectReference Include="../Microsoft.ML/Microsoft.ML.ImageAnalytics.nupkgproj" />
   </ItemGroup>
 
 </Project>
diff --git a/src/Microsoft.ML.AutoML/API/ColumnInference.cs b/src/Microsoft.ML.AutoML/API/ColumnInference.cs
@@ -104,13 +104,20 @@ public sealed class ColumnInformation
         /// <value>The default value is a new, empty <see cref="Collection{String}"/>.</value>
         public ICollection<string> IgnoredColumnNames { get; }
 
+        /// <summary>
+        /// The dataset columns that are image paths.
+        /// </summary>
+        /// <value>The default value is a new, empty <see cref="Collection{String}"/>.</value>
+        public ICollection<string> ImagePathColumnNames { get; }
+
         public ColumnInformation()
         {
             LabelColumnName = DefaultColumnNames.Label;
             CategoricalColumnNames = new Collection<string>();
             NumericColumnNames = new Collection<string>();
             TextColumnNames = new Collection<string>();
             IgnoredColumnNames = new Collection<string>();
+            ImagePathColumnNames = new Collection<string>();
         }
     }
 }
diff --git a/src/Microsoft.ML.AutoML/API/ExperimentBase.cs b/src/Microsoft.ML.AutoML/API/ExperimentBase.cs
@@ -102,7 +102,6 @@ public ExperimentResult<TMetrics> Execute(IDataView trainData, ColumnInformation
             const int crossValRowCountThreshold = 15000;
 
             var rowCount = DatasetDimensionsUtil.CountRows(trainData, crossValRowCountThreshold);
-
             if (rowCount < crossValRowCountThreshold)
             {
                 const int numCrossValFolds = 10;
@@ -160,7 +159,9 @@ public ExperimentResult<TMetrics> Execute(IDataView trainData, IDataView validat
         /// <remarks>
         /// Depending on the size of your data, the AutoML experiment could take a long time to execute.
         /// </remarks>
-        public ExperimentResult<TMetrics> Execute(IDataView trainData, IDataView validationData, ColumnInformation columnInformation, IEstimator<ITransformer> preFeaturizer = null, IProgress<RunDetail<TMetrics>> progressHandler = null)
+        public ExperimentResult<TMetrics> Execute(IDataView trainData, IDataView validationData,
+            ColumnInformation columnInformation, IEstimator<ITransformer> preFeaturizer = null,
+            IProgress<RunDetail<TMetrics>> progressHandler = null)
         {
             if (validationData == null)
             {
@@ -188,7 +189,9 @@ public ExperimentResult<TMetrics> Execute(IDataView trainData, IDataView validat
         /// <remarks>
         /// Depending on the size of your data, the AutoML experiment could take a long time to execute.
         /// </remarks>
-        public CrossValidationExperimentResult<TMetrics> Execute(IDataView trainData, uint numberOfCVFolds, ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizer = null, IProgress<CrossValidationRunDetail<TMetrics>> progressHandler = null)
+        public CrossValidationExperimentResult<TMetrics> Execute(IDataView trainData, uint numberOfCVFolds,
+            ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizer = null,
+            IProgress<CrossValidationRunDetail<TMetrics>> progressHandler = null)
         {
             UserInputValidationUtil.ValidateNumberOfCVFoldsArg(numberOfCVFolds);
             var splitResult = SplitUtil.CrossValSplit(Context, trainData, numberOfCVFolds, columnInformation?.SamplingKeyColumnName);

diff --git a/src/Microsoft.ML.AutoML/ColumnInference/ColumnInformationUtil.cs b/src/Microsoft.ML.AutoML/ColumnInference/ColumnInformationUtil.cs
@@ -56,6 +56,11 @@ internal static class ColumnInformationUtil
                 return ColumnPurpose.ItemId;
             }
 
+            if (columnInfo.ImagePathColumnNames.Contains(columnName))
+            {
+                return ColumnPurpose.ImagePath;
+            }
+
             return null;
         }
 
@@ -94,6 +99,9 @@ internal static ColumnInformation BuildColumnInfo(IEnumerable<(string name, Colu
                     case ColumnPurpose.TextFeature:
                         columnInfo.TextColumnNames.Add(column.name);
                         break;
+                    case ColumnPurpose.ImagePath:
+                        columnInfo.ImagePathColumnNames.Add(column.name);
+                        break;
                 }
             }
 
@@ -121,6 +129,7 @@ public static IEnumerable<string> GetColumnNames(ColumnInformation columnInforma
             AddStringsToListIfNotNull(columnNames, columnInformation.IgnoredColumnNames);
             AddStringsToListIfNotNull(columnNames, columnInformation.NumericColumnNames);
             AddStringsToListIfNotNull(columnNames, columnInformation.TextColumnNames);
+            AddStringsToListIfNotNull(columnNames, columnInformation.ImagePathColumnNames);
             return columnNames;
         }
 

diff --git a/src/Microsoft.ML.AutoML/ColumnInference/PurposeInference.cs b/src/Microsoft.ML.AutoML/ColumnInference/PurposeInference.cs
@@ -242,7 +242,6 @@ public static PurposeInference.Column[] InferPurposes(MLContext context, IDataVi
             ColumnInformation columnInfo)
         {
             data = context.Data.TakeRows(data, MaxRowsToRead);
-
             var allColumns = new List<IntermediateColumn>();
             var columnsToInfer = new List<IntermediateColumn>();
 

diff --git a/src/Microsoft.ML.AutoML/EstimatorExtensions/EstimatorExtensionCatalog.cs b/src/Microsoft.ML.AutoML/EstimatorExtensions/EstimatorExtensionCatalog.cs
@@ -19,7 +19,8 @@ internal enum EstimatorName
         OneHotHashEncoding,
         TextFeaturizing,
         TypeConverting,
-        ValueToKeyMapping
+        ValueToKeyMapping,
+        ImageLoading
     }
 
     internal class EstimatorExtensionCatalog
@@ -38,6 +39,7 @@ internal class EstimatorExtensionCatalog
             { EstimatorName.TextFeaturizing, typeof(TextFeaturizingExtension) },
             { EstimatorName.TypeConverting, typeof(TypeConvertingExtension) },
             { EstimatorName.ValueToKeyMapping, typeof(ValueToKeyMappingExtension) },
+            { EstimatorName.ImageLoading, typeof(ImageLoadingExtension) }
         };
 
         public static IEstimatorExtension GetExtension(EstimatorName estimatorName)

diff --git a/src/Microsoft.ML.AutoML/EstimatorExtensions/EstimatorExtensions.cs b/src/Microsoft.ML.AutoML/EstimatorExtensions/EstimatorExtensions.cs
@@ -2,6 +2,7 @@
 // The .NET Foundation licenses this file to you under the MIT license.
 // See the LICENSE file in the project root for more information.
 
+using System.Collections.Generic;
 using Microsoft.ML.Data;
 using Microsoft.ML.Transforms;
 
@@ -260,6 +261,7 @@ public static SuggestedTransform CreateSuggestedTransform(MLContext context, str
         {
             var pipelineNode = new PipelineNode(EstimatorName.ValueToKeyMapping.ToString(),
                 PipelineNodeType.Transform, inColumn, outColumn);
+
             var estimator = CreateInstance(context, inColumn, outColumn);
             return new SuggestedTransform(pipelineNode, estimator);
         }
@@ -269,4 +271,29 @@ private static IEstimator<ITransformer> CreateInstance(MLContext context, string
             return context.Transforms.Conversion.MapValueToKey(outColumn, inColumn);
         }
     }
+
+    internal class ImageLoadingExtension : IEstimatorExtension
+    {
+        public static string ImageFolder { private get; set; }
+        public IEstimator<ITransformer> CreateInstance(MLContext context, PipelineNode pipelineNode)
+        {
+            return CreateInstance(context, pipelineNode.InColumns[0], pipelineNode.OutColumns[0]);
+        }
+
+        public static SuggestedTransform CreateSuggestedTransform(MLContext context, string inColumn, string outColumn)
+        {
+            var pipelineNodeProperty = new Dictionary<string, object>()
+            {
+                { "imageFolder", ImageFolder },
+            };
+            var pipelineNode = new PipelineNode(EstimatorName.ImageLoading.ToString(), PipelineNodeType.Transform, inColumn, outColumn, pipelineNodeProperty);
+            var estimator = CreateInstance(context, inColumn, outColumn);
+            return new SuggestedTransform(pipelineNode, estimator);
+        }
+
+        private static IEstimator<ITransformer> CreateInstance(MLContext context, string inColumn, string outColumn)
+        {
+            return context.Transforms.LoadRawImageBytes(outColumn, ImageFolder, inColumn);
+        }
+    }
 }
diff --git a/src/Microsoft.ML.AutoML/Experiment/Experiment.cs b/src/Microsoft.ML.AutoML/Experiment/Experiment.cs
@@ -62,7 +62,9 @@ public IList<TRunDetail> Execute()
 
                 // get next pipeline
                 var getPiplelineStopwatch = Stopwatch.StartNew();
-                var pipeline = PipelineSuggester.GetNextInferredPipeline(_context, _history, _datasetColumnInfo, _task, _optimizingMetricInfo.IsMaximizing, _experimentSettings.CacheBeforeTrainer, _trainerWhitelist);
+                var pipeline = PipelineSuggester.GetNextInferredPipeline(_context, _history, _datasetColumnInfo, _task,
+                    _optimizingMetricInfo.IsMaximizing, _experimentSettings.CacheBeforeTrainer, _trainerWhitelist);
+
                 var pipelineInferenceTimeInSeconds = getPiplelineStopwatch.Elapsed.TotalSeconds;
 
                 // break if no candidates returned, means no valid pipeline available
@@ -75,6 +77,7 @@ public IList<TRunDetail> Execute()
                 _logger.Trace($"Evaluating pipeline {pipeline.ToString()}");
                 (SuggestedPipelineRunDetail suggestedPipelineRunDetail, TRunDetail runDetail)
                     = _runner.Run(pipeline, _modelDirectory, _history.Count + 1);
+
                 _history.Add(suggestedPipelineRunDetail);
                 WriteIterationLog(pipeline, suggestedPipelineRunDetail, iterationStopwatch);
 

diff --git a/src/Microsoft.ML.AutoML/Experiment/RecipeInference.cs b/src/Microsoft.ML.AutoML/Experiment/RecipeInference.cs
@@ -15,7 +15,7 @@ internal static class RecipeInference
         public static IEnumerable<SuggestedTrainer> AllowedTrainers(MLContext mlContext, TaskKind task,
             ColumnInformation columnInfo, IEnumerable<TrainerName> trainerWhitelist)
         {
-            var trainerExtensions = TrainerExtensionCatalog.GetTrainers(task, trainerWhitelist);
+            var trainerExtensions = TrainerExtensionCatalog.GetTrainers(task, trainerWhitelist, columnInfo);
 
             var trainers = new List<SuggestedTrainer>();
             foreach (var trainerExtension in trainerExtensions)

diff --git a/src/Microsoft.ML.AutoML/Experiment/Runners/CrossValRunner.cs b/src/Microsoft.ML.AutoML/Experiment/Runners/CrossValRunner.cs
@@ -2,6 +2,7 @@
 // The .NET Foundation licenses this file to you under the MIT license.
 // See the LICENSE file in the project root for more information.
 
+using System;
 using System.Collections.Generic;
 using System.IO;
 using System.Linq;

diff --git a/src/Microsoft.ML.AutoML/Experiment/Runners/RunnerUtil.cs b/src/Microsoft.ML.AutoML/Experiment/Runners/RunnerUtil.cs
@@ -24,7 +24,7 @@ public static (ModelContainer model, TMetrics metrics, Exception exception, doub
         {
             try
             {
-                var estimator = pipeline.ToEstimator();
+                var estimator = pipeline.ToEstimator(trainData, validData);
                 var model = estimator.Fit(trainData);
 
                 var scoredData = model.Transform(validData);

diff --git a/src/Microsoft.ML.AutoML/Experiment/SuggestedPipeline.cs b/src/Microsoft.ML.AutoML/Experiment/SuggestedPipeline.cs
@@ -40,7 +40,7 @@ public SuggestedPipeline(IEnumerable<SuggestedTransform> transforms,
         public override bool Equals(object obj)
         {
             var pipeline = obj as SuggestedPipeline;
-            if(pipeline == null)
+            if (pipeline == null)
             {
                 return false;
             }
@@ -55,7 +55,7 @@ public override int GetHashCode()
         public Pipeline ToPipeline()
         {
             var pipelineElements = new List<PipelineNode>();
-            foreach(var transform in Transforms)
+            foreach (var transform in Transforms)
             {
                 pipelineElements.Add(transform.PipelineNode);
             }
@@ -74,9 +74,9 @@ public static SuggestedPipeline FromPipeline(MLContext context, Pipeline pipelin
             SuggestedTrainer trainer = null;
 
             var trainerEncountered = false;
-            foreach(var pipelineNode in pipeline.Nodes)
+            foreach (var pipelineNode in pipeline.Nodes)
             {
-                if(pipelineNode.NodeType == PipelineNodeType.Trainer)
+                if (pipelineNode.NodeType == PipelineNodeType.Trainer)
                 {
                     var trainerName = (TrainerName)Enum.Parse(typeof(TrainerName), pipelineNode.Name);
                     var trainerExtension = TrainerExtensionCatalog.GetTrainerExtension(trainerName);
@@ -105,7 +105,8 @@ public static SuggestedPipeline FromPipeline(MLContext context, Pipeline pipelin
             return new SuggestedPipeline(transforms, transformsPostTrainer, trainer, context, pipeline.CacheBeforeTrainer);
         }
 
-        public IEstimator<ITransformer> ToEstimator()
+        public IEstimator<ITransformer> ToEstimator(IDataView trainset = null,
+            IDataView validationSet = null)
         {
             IEstimator<ITransformer> pipeline = new EstimatorChain<ITransformer>();
 
@@ -118,8 +119,13 @@ public IEstimator<ITransformer> ToEstimator()
                 }
             }
 
+            if (validationSet != null)
+            {
+                validationSet = pipeline.Fit(validationSet).Transform(validationSet);
+            }
+
             // Get learner
-            var learner = Trainer.BuildTrainer();
+            var learner = Trainer.BuildTrainer(validationSet);
 
             if (_cacheBeforeTrainer)
             {

diff --git a/src/Microsoft.ML.AutoML/Experiment/SuggestedTrainer.cs b/src/Microsoft.ML.AutoML/Experiment/SuggestedTrainer.cs
@@ -2,6 +2,7 @@
 // The .NET Foundation licenses this file to you under the MIT license.
 // See the LICENSE file in the project root for more information.
 
+using System;
 using System.Collections.Generic;
 using System.Linq;
 using Microsoft.ML.Trainers;
@@ -41,14 +42,14 @@ public SuggestedTrainer Clone()
             return new SuggestedTrainer(_mlContext, _trainerExtension, _columnInfo, HyperParamSet?.Clone());
         }
 
-        public ITrainerEstimator<IPredictionTransformer<object>, object> BuildTrainer()
+        public ITrainerEstimator<IPredictionTransformer<object>, object> BuildTrainer(IDataView validationSet = null)
         {
             IEnumerable<SweepableParam> sweepParams = null;
             if (HyperParamSet != null)
             {
                 sweepParams = SweepParams;
             }
-            return _trainerExtension.CreateInstance(_mlContext, sweepParams, _columnInfo);
+            return _trainerExtension.CreateInstance(_mlContext, sweepParams, _columnInfo, validationSet);
         }
 
         public override string ToString()
@@ -63,7 +64,7 @@ public override string ToString()
 
         public PipelineNode ToPipelineNode()
         {
-            var sweepParams = SweepParams.Where(p => p.RawValue != null);
+            var sweepParams = SweepParams?.Where(p => p.RawValue != null);
             return _trainerExtension.CreatePipelineNode(sweepParams, _columnInfo);
         }
 

diff --git a/src/Microsoft.ML.AutoML/Microsoft.ML.AutoML.csproj b/src/Microsoft.ML.AutoML/Microsoft.ML.AutoML.csproj
@@ -8,6 +8,8 @@
   <ItemGroup>
     <ProjectReference Include="..\Microsoft.ML.Core\Microsoft.ML.Core.csproj" />
     <ProjectReference Include="..\Microsoft.ML.CpuMath\Microsoft.ML.CpuMath.csproj" />
+    <ProjectReference Include="..\Microsoft.ML.Dnn\Microsoft.ML.Dnn.csproj" />
+    <ProjectReference Include="..\Microsoft.ML.ImageAnalytics\Microsoft.ML.ImageAnalytics.csproj" />
     <ProjectReference Include="..\Microsoft.ML.LightGbm\Microsoft.ML.LightGbm.csproj" />
     <ProjectReference Include="..\Microsoft.ML.Mkl.Components\Microsoft.ML.Mkl.Components.csproj" />
     <ProjectReference Include="..\Microsoft.ML.Recommender\Microsoft.ML.Recommender.csproj" />