Python Forum

Full Version: Can't find some libraries while importing TensorFlow
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Hello,

While trying to import TensorFlow, I see that some libraries are missed.
>>> import tensorflow as tf
2020-03-16 15:55:06.178671: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer.so.6'; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.2/lib64
2020-03-16 15:55:06.178790: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer_plugin.so.6'; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.2/lib64
2020-03-16 15:55:06.178805: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
>>> 
What should I check in my setup ?
Thanks.
The last line suggests that if you are not using NVidia GPU you can ignore the error. Have you tried proceeding?
(Mar-16-2020, 04:18 PM)jefsummers Wrote: [ -> ]The last line suggests that if you are not using NVidia GPU you can ignore the error. Have you tried proceeding?

I've GPU on my laptop.
I installed TF using this instruction:
pip3 install --upgrade tensorflow-gpu
I installed Tensorflor for GPU last week, and i had to download Nvidia Cuda and downloaded som DLL files too. There is guides on how to do it on youtube and google.
(Mar-18-2020, 04:29 PM)Finpyth Wrote: [ -> ]I installed Tensorflor for GPU last week, and i had to download Nvidia Cuda and downloaded som DLL files too. There is guides on how to do it on youtube and google.
I work on Ubuntu, so no DLLs.
I've also followed some manual that was outdated (unfortunately I realized it later).
Some components that I installed (NVIDIA drivers, CUDA, cuDNN, TF) are unfortunately not compatible.
Thanks. Finally I succeeded to remove whole NVIDIA staff