jie.hang
jie.hang
AttributeError: module 'byteps.torch' has no attribute 'broadcast' AttributeError: module 'byteps.torch' has no attribute 'is_initialized' which is not compatible with horovord in pytorch.
Unfortunately, this NGT package does not seem to support installation on the loongson CPU(mips) self-designed by China. 
The following error is reported when compiling: `$make time_two_op` `nvcc -std=c++11 -c -o tensorflow_time_two/python/ops/_time_two_ops.cu.o tensorflow_time_two/cc/kernels/time_two_kernels.cu.cc -I/usr/local/lib/python3.6/dist-packages/tensorflow_core/include -D_GLIBCXX_USE_CXX11_ABI=0 -fPIC -O2 -std=c++11 -L/usr/local/lib/python3.6/dist-packages/tensorflow_core -l:libtensorflow_framework.so.1 -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC -DNDEBUG --expt-relaxed-constexpr...
when I compiled your code tf_blazeface.cpp under MNN source code, error occured. ``` CMakeFiles/blazeface.out.dir/tf_blazeface.cpp.o: In function `std::unique_ptr::~unique_ptr()': tf_blazeface.cpp:(.text._ZNSt10unique_ptrI6RevertSt14default_deleteIS0_EED2Ev[_ZNSt10unique_ptrI6RevertSt14default_deleteIS0_EED5Ev]+0x10): undefined reference to `Revert::~Revert()' CMakeFiles/blazeface.out.dir/tf_blazeface.cpp.o: In function `main': tf_blazeface.cpp:(.text.startup.main+0x13c): undefined reference to...
When I execute the following python code, I don’t know why a kernel like CUDAFunctor_add appears. `import torch from apex.normalization import MixedFusedRMSNorm datatype = torch.bfloat16 input = torch.randn([1024, 8192], dtype=datatype).cuda()...
Excuse me again, I have 2 questions? (1) I have delete my project and corresponding file, but when I use pip to install tflex or tensorflow-tflex, system still Using cached...
I try to compile apex from source based on official pytorch1.8.1 whl: torch-1.8.1-cp36-cp36m-manylinux1_x86_64.whl, ``` python setup.py build --cpp_ext --cuda_ext python setup.py bdist_wheel --cpp_ext --cuda_ext -d build_pip pip3 install build_pip/apex-0.1-cp36-cp36m-linux_x86_64.whl ```...