CNTK Eval Examples

Eval samples in CNTK binary download package

The CNTK binary download package on the CNTK Releases page includes samples for using the eval library in C++, C#/.NET and Python.

CNTK Library Eval C++/C# Examples

The CNTKLibraryEvalExamples contains code samples demonstrating how to use the CNTK Library Eval API in C++ and C#.

• CNTKLibraryCSEvalCPUOnlyExamples uses the CNTK Library CPU-Only Nuget package to evaluate models on CPU-only devices in C#.
• CNTKLibraryCSEvalGPUExamples uses the CNTK Library GPU Nuget package to evaluate models on GPU devices in C#.
• CNTKLibraryCPPEvalCPUOnlyExamples uses the CNTK Library C++ API to evaluate models on CPU-only devices. It uses the CNTK Library CPU-Only Nuget package.
• CNTKLibraryCPPEvalGPUExamples uses the CNTK Library C++ API to evaluate models on GPU devices. It uses the CNTK Library GPU Nuget package.

On Windows,

• You need Visual Studio 2015 update 3 for using these samples.
• The samples should be built for the 64-bit target platform. Otherwise some issues arise when calling the library. Please also refer to the Troubleshoot CNTK page for more information.
• After a successful build, the executable is saved under the $(SolutionDir)..\..$(Platform)$(ProjectName).$(Configuration)\ folder, e.g. ..\..\X64\CNTKLibraryCSEvalCPUOnlyExamples.Release\CNTKLibraryCSEvalCPUOnlyExamples.exe.

On Linux, only C++ is supported. Please refer to Makefile for building samples. The target name CNTKLIBRARY_CPP_EVAL_EXAMPLES is used to build CNTKLibraryCPPEvalExamples.

CNTK Library Eval Java Examples

• This API is still experimental and subject to change.

• If you are using the bindings in your java project, add the cntk.jar file to the classpath. If you are working with an IDE you should add this as an unmanaged jar. The cntk.jar file can be found in the distributed CNTK binaries. You can also build cntk.jar from code base by yourself. In this case, it can be found in the "java" folder of CNTK's output directory. Please refer to the Linux and Windows instructions for more details.

• The JavaEvalTest contains an example Java project that evaluates a CNN using the Java API.

Examples for evaluating multiple requests in parallel

The CNTKLibraryEvalExamples includes sample code to evaluate multiple evaluation requests in parallel. The sample code also shows how to share model parameters among multiple instances of the same model to reduce memory usage.

• EvaluateMultipleImagesInParallel() in C# demonstrates how to evaluate concurrent requests using CNTK C#/.NET Managed API.
• The MultiThreadsEvaluationWithLoadModel() is the C++ example showing concurrent evaluation using CNTK C++ Library API.

EvalDll Eval C++/C# Examples

The EvalClients.sln contains the following projects demonstrating how to use the EvalDll library in C++ and C#.

• CPPEvalClient: this sample uses the C++ EvalDll.
• CPPEvalExtendedClient: this sample uses the C++ extended Eval interface in EvalDll to evaluate a RNN model.
• CSEvalClient: this sample uses the C# EvalDll (only for Windows). It uses the CNTK EvalDll Nuget Package.

On Windows, The solution file EvalClients.sln is used to build and run samples. Please note

• You need Visual Studio 2015 update 3 for using these samples.
• The samples should be built for the 64-bit target platform. Otherwise some issues arise when calling the library. Please also refer to the Troubleshoot CNTK page for more information.
• After a successful build, the executable is saved under the $(SolutionDir)..\..$(Platform)$(ProjectName).$(Configuration)\ folder, e.g. ..\..\X64\CPPEvalClient.Release\CppEvalClient.exe.

On Linux, please refer to Makefile for building samples. The target name EVAL_CLIENT, and EVAL_EXTENDED_CLIENT are used to build these projects.

Python Eval Examples

You can also use Python to evaluate a pre-trained model as described here.