Background:
Purpose is to develop a machine learning algorithm involving tensorflow.
Problem:
Importing tensorflow results in the error: "invalid proto descriptor for file "tensorboard/compat/proto/resource_handle.proto"" (see error log after my code at the end.)
Action taken:
Tried running pip uninstall protobuf. Next, pip install --no-binary protobuf protobuf.
But, that creates a second error that there is "no google protobuf. If I install protobufs I run into the first error again.
Platform:
Ubuntu 18.10, v 64 bit forGPU. Python 3.6.8.
My Code:
import tensorflow-nightly-gpu #(regardless of version of tf I get the error)
import pandas as pd
...
Error Log:
Couldn't build proto file into descriptor pool!
Invalid proto descriptor for file
"tensorboard/compat/proto/resource_handle.proto":
tensorboard.ResourceHandleProto.device: "tensorboard.ResourceHandleProto.device" is already defined in file "tensorboard/src/resource_handle.proto".
tensorboard.ResourceHandleProto.container: "tensorboard.ResourceHandleProto.container" is already defined in file "tensorboard/src/resource_handle.proto".
tensorboard.ResourceHandleProto.name: "tensorboard.ResourceHandleProto.name" is already defined in file "tensorboard/src/resource_handle.proto".
tensorboard.ResourceHandleProto.hash_code: "tensorboard.ResourceHandleProto.hash_code" is already defined in file "tensorboard/src/resource_handle.proto".
tensorboard.ResourceHandleProto.maybe_type_name: "tensorboard.ResourceHandleProto.maybe_type_name" is already defined in file "tensorboard/src/resource_handle.proto".
Solved my own problem with these steps:
1.Remove all existing GPU, CUDA, CUDNN drivers and remove all tensorflow from my computer
Install drivers for GPU, CUDA, CUDNN with instructions here, https://www.tensorflow.org/install/gpu
Install tensorflow from here, phttps://www.tensorflow.org/install/pip
Related
I am trying to use cupy to speed up some background numpy operations in my python code but when attempting to import cupy I am told that the DLL load failed, importError: DLL load failed.
I am fairly inexperienced when it comes to handling things like this so any help would be greatly appreciated. The entire error message and trace is copied below:
when I tried pip install cupy-cuda112, cupy import is working but pytorch import is not working.
Error loading "C:\Users\CT1\miniconda3\envs\DentiumRIP\lib\site-packages\torch\lib\cublas64_11.dll" or one of its dependencies.
And, when I do conda install pytorch torchvision torchaudio cudatoolkit=11.2 -c pytorch -c conda-forge, the cupy import doesn't work.
please tell me how to fix it..
Failed to import CuPy.
If you installed CuPy via wheels (cupy-cudaXXX or cupy-rocm-X-X), make sure that the package matches with the version of CUDA or ROCm installed.
On Linux, you may need to set LD_LIBRARY_PATH environment variable depending on how you installed CUDA/ROCm.
On Windows, try setting CUDA_PATH environment variable.
Check the Installation Guide for details:
https://docs.cupy.dev/en/latest/install.html
Original error:
ImportError: DLL load failed: 지정된 프로시저를 찾을 수 없습니다.
I am using Windows 10 OS, Python Version - 3.8
Installed Method - conda install pytorch torchvision cudatoolkit=10.2 -c pytorch
It was successfully installed, when trying to run torch.cuda.is_available() throws up an error as -
AttributeError: partially initialized module 'torch' has no attribute 'cuda' (most likely due to a circular import)
I faced this error other importing such as -torch.nn as nn-
I checked the package list (!pip list ) it is present
Why I can't use pytoch on spyder ? How to tackle this problem ?
I think you have a file named torch.py in your working directory
Please rename it to something else
I am able to successfully install Dlib with CUDA support in Windows 10 but getting an error during "import dlib" in my python code of computer vision project.
Environment: Windows 10, Python 3.7.6 (Anaconda), CUDA 11, CuDNN 10.2
Error Message:
>>> import dlib
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "C:\ProgramData\Anaconda3\lib\site-packages\dlib-19.20.99-py3.7-win-amd64.egg\dlib\__init__.py", line 12, in <module>
from _dlib_pybind11 import *
ImportError: DLL load failed: The specified module could not be found.
This can be solved by copying the cudnn64_7.dll (available here: https://developer.nvidia.com/cudnn)
into the %CUDA_PATH%/bin directory (probably something like this: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin)
I am able to find and fix this issue. CUDA 11 installation wasn't able to add few of the directories into the PATH environment variable (Windows 10). It was truncated due to the max length of 2048 characters. I have removed a few of the unused software paths from PATH value and after reinstallation, dlib 19.20 is working with CUDA 11 now.
I created an issue on DLIB Github under the following link which has more information regarding error logs and snapshots for this issue.
https://github.com/davisking/dlib/issues/2097
In my environment, the problem was due to an error somewhere in the build process that resulted code to load CuDNN dynamic libraries not being included in the generated file dlib/__init__.py despite having no build error. In my case the file always included this strange block of code:
if 'OFF' == 'ON':
add_lib_to_dll_path('cudnn-NOTFOUND')
add_lib_to_dll_path('C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.5/lib/x64/cudart.lib')
The second line's cudnn-NOTFOUND gave a clue of what happened with my build.
As I followed instructions on this page, copying all the binaries and include files to the right places within the CUDA directory, I only needed to modify the code to (similar to what Epic Chen's answer suggests but I got rid of the if clause and the bad code line):
add_lib_to_dll_path('C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.5/lib/x64/cudart.lib')
That workaround fixed the problem for me.
My configuration is CUDA 11.5, CuDNN 8.3.1.22, dlib compiled using Visual Studio 2019. The environment variable CMAKE_PREFIX_PATH to the CuDNN directory to get the compiler to find the include files and libraries.
If you are using Anaconda, uninstall dlib and reinstall dlib.
In anaconda command prompt, type
pip uninstall dlib
After successfully uninstalling, type
pip install dlib
It helped me fix the problem.
Try to check the __init__.py file which the error message indicate as below.
Your path is not the same as me.
In the __init__.py file, the if statement should be 'ON' == 'ON'
Besides, the following library paths must be correct. Your version may not be the same as me.
I am trying to train InceptionV3 on a dataset that I have.I am using python3.5 on spyder with keras and tensorflow as backend.
I am getting the following error:
could not initialize a memory descriptor, in file tensorflow/core/kernels/mkl_maxpooling_op.cc:578
I have tried installing tensor flow with conda install -c anaconda tensorflow but that didn't removed the error.
I am trying to install nltk_data package to my environment natlang using conda by giving the following command:
(natlang) C:\Users\asus>conda install -c conda-forge nltk_data
I receive the following errors:
Verifying transaction: failed
CondaVerificationError: The package for nltk_data located at
C:\Users\asus\Anaconda3\pkgs\nltk_data-2017.10.22-py_0
appears to be corrupted. The path
'lib/nltk_data/corpora/propbank/frames/con.xml'
specified in the package manifest cannot be found.
ClobberError: This transaction has incompatible packages due to a shared
path.
packages: conda-forge::nltk_data-2017.10.22-py_0, conda-forge::nltk_data-
2017.10.22-py_0
path: 'lib/nltk_data/corpora/nombank.1.0/readme'
ClobberError: This transaction has incompatible packages due to a shared
path.
packages: conda-forge::nltk_data-2017.10.22-py_0, conda-forge::nltk_data-
2017.10.22-py_0
path: 'lib/nltk_data/corpora/nombank.1.0/readme-dictionaries'
ClobberError: This transaction has incompatible packages due to a shared
path.
packages: conda-forge::nltk_data-2017.10.22-py_0, conda-forge::nltk_data-
2017.10.22-py_0
path: 'lib/nltk_data/corpora/nombank.1.0/readme-nombank-proposition-
structure'
I am working on Anaconda 3, python version 3.6.5, windows 10 enterprise.
Can someone please tell me why this error is occurring and how can I fix it.
Background: I originally wanted to use punkt in one of my programs using the code lines:
import nltk_data
nltk.download()
This would open the nltk downloader and after installing all the packages including punkt, on further running the program I would still encounter the following error:
LookupError:
Resource [93mpunkt[0m not found.
Please use the NLTK Downloader to obtain the resource:
[31m>>> import nltk
>>> nltk.download('punkt')
I tried rerunning the nltk.donwload() and nltk.download('punkt') a couple of times with no change. So then I decided to simply install the nltk_data package to my environment based on the assumption that if I install the package to the env itself, I won't have to use the nltk.download function to use punkt.
Summarizing, I have the following two questions:
If I install the nltk_data package to my evn, do I still need to use the nltk.download function in my code? If yes, how do I resolve the lookup error?
If installing to the evn is enough, then how do I resolve the clobber error?
(ps: I apologize if this sounds stupid, I am very new to machine learning and working with python in general.)
The nltk_data repository is a collection of zipfiles and xml meta data. Usually, it is not installation through packaging tools such as conda or pip.
But there is this utility from conda-forge that tries to install the nltk_data, https://github.com/conda-forge/nltk_data-feedstock
To use it, on the terminal/command prompt/console, first add the conda-forge channel:
conda config --add channels conda-forge
Then you shouldn't need the -c option, and just use:
conda install nltk_data
Please try the above and see whether that get rids of the ClobberError.
This error is requesting you to download a specific nltk dataset call punkt:
Please use the NLTK Downloader to obtain the resource:
>>> import nltk
>>> nltk.download('punkt')
Running nltk.download() without specifying which specific dataset you want to download will call up a tkinter GUI which normally wouldn't be possible if you are accessing your machine remotely without a GUI.
If you're unsure of which resource you need, I would suggest using the popular collection.
import nltk
nltk.download('popular')
Answering 2 que first- there have been similar issues all across windows machines. Its better to use the ntlk.download() function if you want to use punkt or a similar module.
1) The lookup error can easily be resolved. It was because of a typo. Instead of
import nltk_data
it should be
import nltk.data