iamalbert / pytorch-wordemb Goto Github PK

I downloaded and installed torchwordemb-0.0.9 via python setup.py install.and the i got an error related to return {vocab, dest};,I change the original code to return VocabAndTensor(vocab, dest);, and run python setup.py install again.then i install torchwordemb-0.0.9 successful；
but when i import torchwordemb，i meet an error ：

 import torchwordemb
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: /usr/local/lib/python3.6/dist-packages/torchwordemb-0.0.9-py3.6-linux-x86_64.egg/torchwordemb.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _ZN6caffe26detail37_typeMetaDataInstance_preallocated_29E

i have found some solutions but it didn't work for me, anyone meet this problem?

  Downloading torchwordemb-0.0.8.tar.gz
    Complete output from command python setup.py egg_info:
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/private/var/folders/wm/rqr_vbc104g66ffx__hwnw3dkl2z1b/T/pip-build-3KaXlf/torchwordemb/setup.py", line 17, in <module>
        import build
      File "build.py", line 2, in <module>
        from torch.utils.ffi import create_extension
      File "/Users/dblythe/.environments/torch/lib/python2.7/site-packages/torch/utils/ffi/__init__.py", line 14, in <module>
        raise ImportError("torch.utils.ffi requires the cffi package")
    ImportError: torch.utils.ffi requires the cffi package

This is fixed by pip install cffi

pytorch-wordemb/src/loadwordemb.cpp:101:41: error: non-constant-expression cannot be narrowed from type 'size_t' (aka 'unsigned long') to 'long long' in initializer list [-Wc++11-narrowing]
dest.resize_(torch::IntArrayRef{n_word, dim});

-Wc++11-narrowing
prevents compilation due to incompatible types Typedef Documentation using c10::IntArrayRef = ArrayRef<int64_t>
and
dest.resize_(torch::IntArrayRef{n_word, dim});

correct to:
dest.resize_(torch::IntArrayRef{(int64_t)n_word, (int64_t)dim});

or propose better / correct handling of types...

On installation with gcc version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.11),
I faced the error below when I run pip install torchwordemb inside of the docker container of the image nvidia/cuda:10.0-cudnn7-devel-ubuntu16.04

    src/loadwordemb.cpp: In function ‘VocabAndTensor load_word2vec_bin(const char*)’:
    src/loadwordemb.cpp:139:52: warning: narrowing conversion of ‘n_word’ from ‘size_t {aka long unsigned int}’ to ‘long int’ inside { } [-Wnarrowing]
             dest.resize_(torch::IntArrayRef{n_word, dim});
                                                        ^
    src/loadwordemb.cpp:139:52: warning: narrowing conversion of ‘n_word’ from ‘size_t {aka long unsigned int}’ to ‘long int’ inside { } [-Wnarrowing]
    src/loadwordemb.cpp:139:52: warning: narrowing conversion of ‘dim’ from ‘size_t {aka long unsigned int}’ to ‘long int’ inside { } [-Wnarrowing]
    src/loadwordemb.cpp:139:52: warning: narrowing conversion of ‘dim’ from ‘size_t {aka long unsigned int}’ to ‘long int’ inside { } [-Wnarrowing]
    src/loadwordemb.cpp:157:24: error: converting to ‘VocabAndTensor {aka std::tuple<pybind11::dict, at::Tensor>}’ from initializer list would use explicit constructor ‘constexpr std::tuple<_T1, _T2>::tuple(_U1&&, _U2&&) [with _U1 = pybind11::dict&; _U2 = at::Tensor&; <template-parameter-2-3> = void; _T1 = pybind11::dict; _T2 = at::Tensor]’
         return {vocab, dest};
                            ^
    error: command 'gcc' failed with exit status 1
    ----------------------------------------

but I succeeded to install with gcc version 6.3.0 20170516 (Debian 6.3.0-18+deb9u1).

Is this error due to the conflict of gcc versions?

$ pip install --user torchwordemb
Collecting torchwordemb
  Using cached https://files.pythonhosted.org/packages/ab/26/a811077fa6971b2d010cb7c10bbed18cb5182219706b71d7d151f9e6a838/torchwordemb-0.0.8.tar.gz
    Complete output from command python setup.py egg_info:
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/tmp/pip-install-avn748hy/torchwordemb/setup.py", line 17, in <module>
        import build
      File "/tmp/pip-install-avn748hy/torchwordemb/build.py", line 2, in <module>
        from torch.utils.ffi import create_extension
      File "/opt/python/lib/python3.6/site-packages/torch/utils/ffi/__init__.py", line 1, in <module>
        raise ImportError("torch.utils.ffi is deprecated. Please use cpp extensions instead.")
    ImportError: torch.utils.ffi is deprecated. Please use cpp extensions instead.
    
    ----------------------------------------
Command "python setup.py egg_info" failed with error code 1 in /tmp/pip-install-avn748hy/torchwordemb/

I have PyTorch 1.0. When I install PyTorch 0.4.0 on a virtual environment and then install torchwordemb on top of that, I don't get this error.

I loaded the word2vec embeddings trained by gensim, and I encountered the Segmentation fault (core dumped) error.

Error during installation:
src/loadwordemb.cpp:157:24: error: converting to ‘VocabAndTensor {aka std::tuple<pybind11::dict, at::Tensor>}’ from initializer list would use explicit constructor ‘constexpr std::tuple<_T1, _T2>::tuple(_U1&&, _U2&&) [with _U1 = pybind11::dict&; _U2 = at::Tensor&; = void; _T1 = pybind11::dict; _T2 = at::Tensor]’
return {vocab, dest};

This problem has been bothering me for a long time. Is it a compatibility issue with the Pytorch version? Or other questions, thank you.