warmspringwinds / pytorch-cpp Goto Github PK

View Code? Open in Web Editor NEW

370.0 370.0 68.0 8.95 MB

Pytorch C++ Library

CMake 3.58% C++ 93.39% Jupyter Notebook 3.03%

pytorch-cpp's People

Contributors

Stargazers

Watchers

pytorch-cpp's Issues

ATen is out of date

I am getting:

4 errors generated.
make[2]: *** [src/ATen/CMakeFiles/ATen.dir/ATen/CPUByteType.cpp.o] Error 1
make[2]: *** Waiting for unfinished jobs....
/Volumes/Hodge/dev/pytorch-cpp/ATen/build/src/ATen/ATen/CPUCharType.cpp:1913:12: error: conversion from
'long' to 'at::Scalar' is ambiguous
return THCharTensor_sumall(self_->tensor);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
AT_FORALL_SCALAR_TYPES(DEFINE_IMPLICIT_CTOR)
^
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/build/src/ATen/ATen/CPUCharType.cpp:1947:12: error: conversion from
'long' to 'at::Scalar' is ambiguous
return THCharTensor_prodall(self_->tensor);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
AT_FORALL_SCALAR_TYPES(DEFINE_IMPLICIT_CTOR)
^
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/build/src/ATen/ATen/CPUCharType.cpp:2002:12: error: conversion from
'long' to 'at::Scalar' is ambiguous
return THCharTensor_trace(self_->tensor);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
AT_FORALL_SCALAR_TYPES(DEFINE_IMPLICIT_CTOR)
^
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/build/src/ATen/ATen/CPUCharType.cpp:2534:12: error: conversion from
'long' to 'at::Scalar' is ambiguous
return THCharTensor_dot(self_->tensor, tensor_->tensor);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
AT_FORALL_SCALAR_TYPES(DEFINE_IMPLICIT_CTOR)
^
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
/Volumes/Hodge/dev/pytorch-cpp/ATen/src/ATen/../ATen/Scalar.h:25:26: note: candidate constructor
4 errors generated.

I will test to see if that is fixed on the ATen master.

Conversion from float std::vector to Aten Tensor and vice versa?

pretrained pytorch model

Hi,

I've a pretrained model which is saved as a pth.tar file, so is it possible to use pytorch-cpp to inference?

Errors in loading state_dict

When I run the convert_weights.ipynb with my own model which is a pretrained Dual Path Network(DPN)：

A mistake happend ：

Could you please tell me why this would happen? (When training the DPN model everything is fine)

About deploy

Hi, thank you first, I want to use C++ api to call the deep learning model of pytorch. but I not sure the project is easy to deploy in centos 6.5 or 7.0. I try tensorflow cpp api before but it's too difficult to deploy the environment , so do you have some advice ?

Project Progress

Dear @warmspringwinds,
Your project is fantastic. Do you have nay plan for continuing it? For example by implementing other PyTorch's layers (e.g., nn.Dropout2d, nn.Conv3d)?

preprocess_batch might cause transfer to CPU, possible bug

Inspect the preprocess_batch -- it might cause the transfer
to CPU in case we are working with GPU tensors.

Having fixed that will improve the performance.

https://github.com/warmspringwinds/pytorch-cpp/blob/master/src/pytorch.cpp#L1280

Support of customized models

Hi! Does pytorch-cpp support any conversion from customized models built in pytorch, or does it only support standard models (e.g. vgg, res-net) and your own segmentation models?
Big thanks!

ATen CUDA error

Hi @warmspringwinds ,

After I switched to CUDA 8.0, I still can't compile ATen cloned from your repo while I can do this from the original master repo.

After compiled it, I got two .so file which is cuda and cpu version. Then I did a very simple test:

cout << ones(CUDA(kFloat), {3,4}) << "\n";

and I got below error message:

cannot initialize CUDA without ATen_cuda library (initCUDA at /home/pharrell/codebase/github/pytorch-cpp/ATen/aten/src/ATen/detail/CUDAHooksInterface.h:42)
frame #0: at::Context::lazyInitCUDA()::{lambda()#1}::operator()() const + 0x32 (0x409a56 in ./test_aten)
frame #1: void std::_Bind_simple<at::Context::lazyInitCUDA()::{lambda()#1} ()>::_M_invoke<>(std::_Index_tuple<>) + 0x28 (0x40afb0 in ./test_aten)
frame #2: std::_Bind_simple<at::Context::lazyInitCUDA()::{lambda()#1} ()>::operator()() + 0x2c (0x40ace0 in ./test_aten)
frame #3: void std::__once_call_impl<std::_Bind_simple<at::Context::lazyInitCUDA()::{lambda()#1} ()> >() + 0x17 (0x40a7a4 in ./test_aten)
frame #4: + 0xea99 (0x7feec5bdda99 in /lib/x86_64-linux-gnu/libpthread.so.0)
frame #5: ./test_aten() [0x408ccd]
frame #6: void std::call_onceat::Context::lazyInitCUDA()::{lambda()#1}(std::once_flag&, at::Context::lazyInitCUDA()::{lambda()#1}&&) + 0x77 (0x40a0f7 in ./test_aten)
frame #7: at::Context::lazyInitCUDA() + 0x3d (0x409b37 in ./test_aten)
frame #8: at::Context::initCUDAIfNeeded(at::Backend) + 0x21 (0x409b81 in ./test_aten)
frame #9: at::Context::getTypeOpt(at::Backend, at::ScalarType) + 0x23 (0x40984b in ./test_aten)
frame #10: at::Context::getType(at::Backend, at::ScalarType) + 0x4a (0x40992a in ./test_aten)
frame #11: ./test_aten() [0x408da9]
frame #12: ./test_aten() [0x408de1]
frame #13: main + 0x65a (0x409467 in ./test_aten)
frame #14: __libc_start_main + 0xf0 (0x7feec3e2e830 in /lib/x86_64-linux-gnu/libc.so.6)
frame #15: _start + 0x29 (0x408bb9 in ./test_aten)

Aborted (core dumped)

I've already added those .so in the CMakeList which is TARGET_LINK_LIBRARIES(test_aten ${CUDA_LIBRARIES} ${ATen_BINARY_DIR}/src/ATen/libATen_cuda.so ${ATen_BINARY_DIR}/src/ATen/libATen_cpu.so).

The cpu part is working fine.

Do you have any thoughts on this problem? Thanks!

test speed

Have you tested the speed? I get a lower speed(30ms/img) with resnet18 224*224 bachsize1

How to replace get_hdf5_file_keys()

Where Are you get this function ? What is version of hdf5 version? I use 1.10.1 And I can't find get_hdf5_file_keys() fucntion

how to turn on intel MKL for BLAS?

In CPU mode, what is the BLAS as default? how to support intel MKL?

Thanks,
Adam Wang

Does Pytorch-cpp support Pytorch 0.3.1?

How to find the operation of C++ functions respect to python?

Hi, your work is great, and recently, I need to deploy the unet in C++, and I want to use pytorch model directly in C++. So ,I need write a C++ operation of ConvTranspose2d. I noticed that you write the operation of Conv2d, use the functions SpatialDilatedConvolution_updateOutput. so ,if I want to use ConvTranspose2d, which functions can I use, and how can i found them or are there any docs?
Thx very much!!

How to convert Type of Aten? from kByte to kFloat?

Make error

Hi, ATen can not be made successfully. Here is the error message I got. Any thoughts on this? I spent almost two days on this and can not make any progress. My system is Ubuntu 16.04 with CUDA 9.0 and CUDNN 7.0. Thanks!

Scanning dependencies of target TH
[ 0%] Building C object lib/TH/CMakeFiles/TH.dir/THAllocator.c.o
[ 1%] Building C object lib/TH/CMakeFiles/TH.dir/THSize.c.o
[ 2%] Building C object lib/TH/CMakeFiles/TH.dir/THGeneral.c.o
[ 3%] Building C object lib/TH/CMakeFiles/TH.dir/THBlas.c.o
[ 3%] Building C object lib/TH/CMakeFiles/TH.dir/THLapack.c.o
[ 3%] Building C object lib/TH/CMakeFiles/TH.dir/THHalf.c.o
[ 3%] Building C object lib/TH/CMakeFiles/TH.dir/THTensor.c.o
[ 3%] Building C object lib/TH/CMakeFiles/TH.dir/THStorage.c.o
[ 3%] Building C object lib/TH/CMakeFiles/TH.dir/THLogAdd.c.o
[ 4%] Building C object lib/TH/CMakeFiles/TH.dir/THRandom.c.o
[ 4%] Building C object lib/TH/CMakeFiles/TH.dir/THFile.c.o
[ 5%] Building C object lib/TH/CMakeFiles/TH.dir/THDiskFile.c.o
[ 5%] Building C object lib/TH/CMakeFiles/TH.dir/THMemoryFile.c.o
[ 5%] Building C object lib/TH/CMakeFiles/TH.dir/THAtomic.c.o
[ 6%] Building C object lib/TH/CMakeFiles/TH.dir/THVector.c.o
[ 6%] Building C object lib/TH/CMakeFiles/TH.dir/generic/simd/convolve.c.o
[ 6%] Building C object lib/TH/CMakeFiles/TH.dir/generic/simd/convolve5x5_sse.c.o
[ 7%] Building C object lib/TH/CMakeFiles/TH.dir/vector/AVX.c.o
[ 7%] Building C object lib/TH/CMakeFiles/TH.dir/generic/simd/convolve5x5_avx.c.o
[ 7%] Building C object lib/TH/CMakeFiles/TH.dir/vector/AVX2.c.o
[ 8%] Linking C shared library libTH.so
[ 8%] Built target TH
[ 8%] Building NVCC (Device) object lib/THC/CMakeFiles/THC.dir/THC_generated_THCReduceApplyUtils.cu.o
Scanning dependencies of target THS
Scanning dependencies of target THNN
[ 8%] Building NVCC (Device) object lib/THC/CMakeFiles/THC.dir/THC_generated_THCStorage.cu.o
[ 8%] Building NVCC (Device) object lib/THC/CMakeFiles/THC.dir/THC_generated_THCBlas.cu.o
[ 9%] Building NVCC (Device) object lib/THC/CMakeFiles/THC.dir/THC_generated_THCHalf.cu.o
[ 10%] Building NVCC (Device) object lib/THC/CMakeFiles/THC.dir/THC_generated_THCStorageCopy.cu.o
[ 10%] Building NVCC (Device) object lib/THC/CMakeFiles/THC.dir/THC_generated_THCSleep.cu.o
[ 10%] Building C object lib/THS/CMakeFiles/THS.dir/THSTensor.c.o
[ 10%] Building C object lib/THNN/CMakeFiles/THNN.dir/init.c.o
CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THStorage.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THStorageCopy.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THCStorage.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THCStorageCopy.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THTensor.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THTensorCopy.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THTensorRandom.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THTensorMath.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THTensorConv.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THTensorLapack.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THCTensor.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THCTensorCopy.h

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THCStorageCopy.cu

CMake Warning at /usr/share/cmake-3.5/Modules/FindCUDA/make2cmake.cmake:65 (message):
Removing non-existent dependency file: generic/THStorage.h