From c49f9f9793c42ec8b6ed627672b11f5127734fbc Mon Sep 17 00:00:00 2001 From: Ross Smith <rosmith@microsoft.com> Date: Fri, 15 Jun 2018 17:32:55 +0100 Subject: [PATCH 1/2] Initial Propose Sweeps --- Microsoft.ML.sln | 7 ++++ src/Microsoft.ML.Sweeper/Algorithms/Grid.cs | 2 +- .../Microsoft.ML.Sweeper.Tests.csproj | 21 ++++++++++++ .../Microsoft.ML.Sweeper.Tests/SweeperTest.cs | 32 +++++++++++++++++++ 4 files changed, 61 insertions(+), 1 deletion(-) create mode 100644 test/Microsoft.ML.Sweeper.Tests/Microsoft.ML.Sweeper.Tests.csproj create mode 100644 test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs diff --git a/Microsoft.ML.sln b/Microsoft.ML.sln index 3529c0e5b76..436aedcb0b7 100644 --- a/Microsoft.ML.sln +++ b/Microsoft.ML.sln @@ -106,6 +106,8 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Microsoft.ML.Maml", "src\Mi EndProject Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Microsoft.ML.Console", "src\Microsoft.ML.Console\Microsoft.ML.Console.csproj", "{362A98CF-FBF7-4EBB-A11B-990BBF845B15}" EndProject +Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Microsoft.ML.Sweeper.Tests", "test\Microsoft.ML.Sweeper.Tests\Microsoft.ML.Sweeper.Tests.csproj", "{3DEB504D-7A07-48CE-91A2-8047461CB3D4}" +EndProject Global GlobalSection(SolutionConfigurationPlatforms) = preSolution Debug|Any CPU = Debug|Any CPU @@ -208,6 +210,10 @@ Global {362A98CF-FBF7-4EBB-A11B-990BBF845B15}.Debug|Any CPU.Build.0 = Debug|Any CPU {362A98CF-FBF7-4EBB-A11B-990BBF845B15}.Release|Any CPU.ActiveCfg = Release|Any CPU {362A98CF-FBF7-4EBB-A11B-990BBF845B15}.Release|Any CPU.Build.0 = Release|Any CPU + {3DEB504D-7A07-48CE-91A2-8047461CB3D4}.Debug|Any CPU.ActiveCfg = Debug|Any CPU + {3DEB504D-7A07-48CE-91A2-8047461CB3D4}.Debug|Any CPU.Build.0 = Debug|Any CPU + {3DEB504D-7A07-48CE-91A2-8047461CB3D4}.Release|Any CPU.ActiveCfg = Release|Any CPU + {3DEB504D-7A07-48CE-91A2-8047461CB3D4}.Release|Any CPU.Build.0 = Release|Any CPU EndGlobalSection GlobalSection(SolutionProperties) = preSolution HideSolutionNode = FALSE @@ -243,6 +249,7 @@ Global {7A9DB75F-2CA5-4184-9EF5-1F17EB39483F} = {AED9C836-31E3-4F3F-8ABC-929555D3F3C4} {64F40A0D-D4C2-4AA7-8470-E9CC437827E4} = {09EADF06-BE25-4228-AB53-95AE3E15B530} {362A98CF-FBF7-4EBB-A11B-990BBF845B15} = {09EADF06-BE25-4228-AB53-95AE3E15B530} + {3DEB504D-7A07-48CE-91A2-8047461CB3D4} = {AED9C836-31E3-4F3F-8ABC-929555D3F3C4} EndGlobalSection GlobalSection(ExtensibilityGlobals) = postSolution SolutionGuid = {41165AF1-35BB-4832-A189-73060F82B01D} diff --git a/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs b/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs index 734cf93e301..aeb66137fb8 100644 --- a/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs +++ b/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs @@ -64,7 +64,7 @@ protected SweeperBase(ArgumentsBase args, IHostEnvironment env, IValueGenerator[ SweepParameters = sweepParameters; } - public virtual ParameterSet[] ProposeSweeps(int maxSweeps, IEnumerable<IRunResult> previousRuns) + public virtual ParameterSet[] ProposeSweeps(int maxSweeps, IEnumerable<IRunResult> previousRuns = null) { var prevParamSets = previousRuns?.Select(r => r.ParameterSet).ToList() ?? new List<ParameterSet>(); var result = new List<ParameterSet>(); diff --git a/test/Microsoft.ML.Sweeper.Tests/Microsoft.ML.Sweeper.Tests.csproj b/test/Microsoft.ML.Sweeper.Tests/Microsoft.ML.Sweeper.Tests.csproj new file mode 100644 index 00000000000..59a9a3221e4 --- /dev/null +++ b/test/Microsoft.ML.Sweeper.Tests/Microsoft.ML.Sweeper.Tests.csproj @@ -0,0 +1,21 @@ +<Project Sdk="Microsoft.NET.Sdk"> + + <PropertyGroup> + <TargetFramework>netcoreapp2.0</TargetFramework> + + <IsPackable>false</IsPackable> + </PropertyGroup> + + <ItemGroup> + <PackageReference Include="Microsoft.NET.Test.Sdk" Version="15.7.0" /> + <PackageReference Include="xunit" Version="2.3.1" /> + <PackageReference Include="xunit.runner.visualstudio" Version="2.3.1" /> + <DotNetCliToolReference Include="dotnet-xunit" Version="2.3.1" /> + </ItemGroup> + + <ItemGroup> + <ProjectReference Include="..\..\src\Microsoft.ML.Sweeper\Microsoft.ML.Sweeper.csproj" /> + <ProjectReference Include="..\Microsoft.ML.TestFramework\Microsoft.ML.TestFramework.csproj" /> + </ItemGroup> + +</Project> diff --git a/test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs b/test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs new file mode 100644 index 00000000000..a1eeae70de1 --- /dev/null +++ b/test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs @@ -0,0 +1,32 @@ +using Microsoft.ML.Runtime; +using Microsoft.ML.Runtime.CommandLine; +using Microsoft.ML.Runtime.Data; +using Microsoft.ML.Runtime.RunTests; +using Microsoft.ML.Runtime.Sweeper; +using System; +using System.IO; +using Xunit; + +namespace Microsoft.ML.Sweeper.Tests +{ + public class SweeperTest + { + [Fact] + public void SweeperReturnsDistinctValues() + { + var args = new DiscreteParamArguments(); + args.Name = "Amazing"; + args.Values = new string[] { "one" }; + var valueGenerator = new DiscreteValueGenerator(args); + using (var writer = new StreamWriter(new MemoryStream())) + using (var env = new TlcEnvironment(42, outWriter: writer, errWriter: writer)) + { + var sweeper = new UniformRandomSweeper(env, new SweeperBase.ArgumentsBase(), new[] { valueGenerator }); + var results = sweeper.ProposeSweeps(2); + Assert.NotNull(results); + int length = results.Length; + Assert.Equal(1, length); + } + } + } +} From 4d88ca8ab9f95067291a1bb6d9f41b5cf69faa70 Mon Sep 17 00:00:00 2001 From: Ross Smith <rosmith@microsoft.com> Date: Mon, 18 Jun 2018 13:29:49 +0100 Subject: [PATCH 2/2] Changed List to HashSet to ensure that there are no duplicates --- src/Microsoft.ML.Core/Prediction/ISweeper.cs | 5 ++ src/Microsoft.ML.Sweeper/Algorithms/Grid.cs | 6 +-- .../Microsoft.ML.Sweeper.Tests/SweeperTest.cs | 49 ++++++++++++++++--- 3 files changed, 49 insertions(+), 11 deletions(-) diff --git a/src/Microsoft.ML.Core/Prediction/ISweeper.cs b/src/Microsoft.ML.Core/Prediction/ISweeper.cs index a0a1850be0d..b3dd0dc3daa 100644 --- a/src/Microsoft.ML.Core/Prediction/ISweeper.cs +++ b/src/Microsoft.ML.Core/Prediction/ISweeper.cs @@ -174,6 +174,11 @@ public override string ToString() { return string.Join(" ", _parameterValues.Select(kvp => string.Format("{0}={1}", kvp.Value.Name, kvp.Value.ValueText)).ToArray()); } + + public override int GetHashCode() + { + return _hash; + } } /// <summary> diff --git a/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs b/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs index aeb66137fb8..53e7046ebd6 100644 --- a/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs +++ b/src/Microsoft.ML.Sweeper/Algorithms/Grid.cs @@ -67,7 +67,7 @@ protected SweeperBase(ArgumentsBase args, IHostEnvironment env, IValueGenerator[ public virtual ParameterSet[] ProposeSweeps(int maxSweeps, IEnumerable<IRunResult> previousRuns = null) { var prevParamSets = previousRuns?.Select(r => r.ParameterSet).ToList() ?? new List<ParameterSet>(); - var result = new List<ParameterSet>(); + var result = new HashSet<ParameterSet>(); for (int i = 0; i < maxSweeps; i++) { ParameterSet paramSet; @@ -150,12 +150,12 @@ public RandomGridSweeper(IHostEnvironment env, Arguments args, IValueGenerator[] } } - public override ParameterSet[] ProposeSweeps(int maxSweeps, IEnumerable<IRunResult> previousRuns) + public override ParameterSet[] ProposeSweeps(int maxSweeps, IEnumerable<IRunResult> previousRuns = null) { if (_nGridPoints == 0) return base.ProposeSweeps(maxSweeps, previousRuns); - var result = new List<ParameterSet>(); + var result = new HashSet<ParameterSet>(); var prevParamSets = (previousRuns != null) ? previousRuns.Select(r => r.ParameterSet).ToList() : new List<ParameterSet>(); diff --git a/test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs b/test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs index a1eeae70de1..9a03d9ec302 100644 --- a/test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs +++ b/test/Microsoft.ML.Sweeper.Tests/SweeperTest.cs @@ -12,21 +12,54 @@ namespace Microsoft.ML.Sweeper.Tests public class SweeperTest { [Fact] - public void SweeperReturnsDistinctValues() + public void UniformRandomSweeperReturnsDistinctValuesWhenProposeSweep() { - var args = new DiscreteParamArguments(); - args.Name = "Amazing"; - args.Values = new string[] { "one" }; - var valueGenerator = new DiscreteValueGenerator(args); + DiscreteValueGenerator valueGenerator = CreateDiscreteValueGenerator(); + using (var writer = new StreamWriter(new MemoryStream())) using (var env = new TlcEnvironment(42, outWriter: writer, errWriter: writer)) { - var sweeper = new UniformRandomSweeper(env, new SweeperBase.ArgumentsBase(), new[] { valueGenerator }); - var results = sweeper.ProposeSweeps(2); + var sweeper = new UniformRandomSweeper(env, + new SweeperBase.ArgumentsBase(), + new[] { valueGenerator }); + + var results = sweeper.ProposeSweeps(5000); Assert.NotNull(results); + int length = results.Length; - Assert.Equal(1, length); + Assert.Equal(2, length); } } + + [Fact] + public void RandomGridSweeperReturnsDistinctValuesWhenProposeSweep() + { + DiscreteValueGenerator valueGenerator = CreateDiscreteValueGenerator(); + + using (var writer = new StreamWriter(new MemoryStream())) + using (var env = new TlcEnvironment(42, outWriter: writer, errWriter: writer)) + { + var sweeper = new RandomGridSweeper(env, + new RandomGridSweeper.Arguments(), + new[] { valueGenerator }); + + var results = sweeper.ProposeSweeps(5000); + Assert.NotNull(results); + + int length = results.Length; + Assert.Equal(2, length); + } + } + + private static DiscreteValueGenerator CreateDiscreteValueGenerator() + { + var args = new DiscreteParamArguments() + { + Name = "TestParam", + Values = new string[] { "one", "two" } + }; + + return new DiscreteValueGenerator(args); + } } }