Peng Chen

@sh1r0 Hi I just install the cmake 3.5.2 cmake --version: cmake version 3.5.2 When I do not set ANDROID_ABI and run ./scripts/build_gflags.sh, it is OK. However when ANDROID_ABI is set...

Besides, when I check the caffe-android-lib/gflags-2.1.2/build/CMakeFiles/CMakeError.log file, I found `/workspace/soft/android-ndk-r10d/platforms/android-21/arch-arm64/usr/lib/libc.so: undefined reference to`dlclose@LIBC' 13 /workspace/soft/android-ndk-r10d/platforms/android-21/arch-arm64/usr/lib/libc.so: undefined reference to `dlopen@LIBC' 14 /workspace/soft/android-ndk-r10d/platforms/android-21/arch-arm64/usr/lib/libc.so: undefined reference to`dlerror@LIBC' 15 /workspace/soft/android-ndk-r10d/platforms/android-21/arch-arm64/usr/lib/libc.so: undefined reference to `dladdr@LIBC'...

@sh1r0 Hi, I verified all the path manually by dumping compile and link flags (also the ANDROID_ABI variable) in the android-cmake/android.toolchain.cmake. And now the error was gone. (^_^)

I implemented a pytorch version of LQ-Net at https://github.com/aim-uofa/model-quantization The repo also contains other quantization methods, such as Dorefa-Net, Xnor-net, Xnor-net++, PACT, LSQ, LSQ+,FATNN.

@Desmond-97 你这边尝试的结论是否是： 对于QAT量化，以RepVGG方式训练的全精度模型初始化，fintune融合后的单分支模型，效果非常差。但是对于这个融合后的单分支模型，直接用PTQ算法，量化精度几乎没损失？ 也就是说RepVGG训练出来的模型不适合QAT的量化，只要用PTQ就行了～～

I also prefer the feature however have little knowledge how to implement it. I guess by changing the iteration folder might help (in _layerout/reveal.html): " {% for post in site.posts...

@ddemirdjian Is there any reason or benefit to add negative examples?

Thanks for reporting the requirement. I will pay some effort on the converting.

Hi, @shuangyichen For the onnx converting script, one might consider the following aspects: - What version of onnx op set is going to support? As far as I know some...

just update some progress. - As Group Norm is not well supported in onnx, I'am training the Full-BN version of ABC-net. - In the meanwhile, I am developing the onnx...