Bug fix with assert() during training

build/darknet/darknet.vcxproj

@@ -133,13 +133,12 @@
       <IntrinsicFunctions>true</IntrinsicFunctions>
       <SDLCheck>true</SDLCheck>
       <AdditionalIncludeDirectories>$(OPENCV_DIR)\include;C:\opencv_3.0\opencv\build\include;..\..\include;..\..\3rdparty\stb\include;..\..\3rdparty\pthreads\include;%(AdditionalIncludeDirectories);$(CudaToolkitIncludeDir);$(CUDNN)\include;$(cudnn)\include</AdditionalIncludeDirectories>
-      <PreprocessorDefinitions>OPENCV;CUDNN_HALF;CUDNN;_TIMESPEC_DEFINED;_SCL_SECURE_NO_WARNINGS;_CRT_SECURE_NO_WARNINGS;_CRT_RAND_S;GPU;WIN32;NDEBUG;_CONSOLE;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>OPENCV;CUDNN_HALF;CUDNN;_TIMESPEC_DEFINED;_SCL_SECURE_NO_WARNINGS;_CRT_SECURE_NO_WARNINGS;_CRT_RAND_S;GPU;WIN32;_CONSOLE;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <CLanguageStandard>c11</CLanguageStandard>
       <CppLanguageStandard>c++1y</CppLanguageStandard>
       <PrecompiledHeaderCompileAs>CompileAsCpp</PrecompiledHeaderCompileAs>
       <CompileAs>Default</CompileAs>
-      <UndefinePreprocessorDefinitions>
-      </UndefinePreprocessorDefinitions>
+      <UndefinePreprocessorDefinitions>NDEBUG</UndefinePreprocessorDefinitions>
       <MultiProcessorCompilation>true</MultiProcessorCompilation>
     </ClCompile>
     <Link>

src/network_kernels.cu

@@ -161,11 +161,12 @@ void forward_backward_network_gpu(network net, float *x, float *y)
     for (i = 0; i < net.n; ++i) {
         layer l = net.layers[i];
         if (net.cudnn_half){
-            assert((l.c*l.n*l.size*l.size) > 0);
             if (l.type == CONVOLUTIONAL && l.weights_gpu && l.weights_gpu16) {
+                assert((l.c*l.n*l.size*l.size) > 0);
                 cuda_convert_f32_to_f16(l.weights_gpu, l.c*l.n*l.size*l.size, l.weights_gpu16);
             }
             else if (l.type == CRNN && l.input_layer->weights_gpu && l.input_layer->weights_gpu16) {
+                assert((l.input_layer->c*l.input_layer->n*l.input_layer->size*l.input_layer->size) > 0);
                 cuda_convert_f32_to_f16(l.input_layer->weights_gpu, l.input_layer->nweights, l.input_layer->weights_gpu16);
                 cuda_convert_f32_to_f16(l.self_layer->weights_gpu, l.self_layer->nweights, l.self_layer->weights_gpu16);
                 cuda_convert_f32_to_f16(l.output_layer->weights_gpu, l.output_layer->nweights, l.output_layer->weights_gpu16);