I am using TensorFlow with GPU support on my local machine.
The README states:
(Optional) If you plan to install TensorFlow with GPU support on your local machine, follow the guide to install the necessary NVIDIA software on your system.
The official Tensorflow install page always refers to the latest Tensoflow install and this project has a locked Tensorflow version.
cudnn fails when attempting to call 'dog_detector(img_path)'
F tensorflow/stream_executor/cuda/cuda_dnn.cc:222] Check failed: s.ok() could not find cudnnCreate in cudnn DSO; dlerror: /home/jason/anaconda3/envs/dog-project/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow.so: undefined symbol: cudnnCreate
For me the fix was to open the 'dog-project' env: source activate dog-project
Use conda to install the toolkit: conda install -c cudatoolkit
Install a DOWNGRADED cudnn: conda install -c marta-sd cudnn
I am using TensorFlow with GPU support on my local machine. The README states:
The official Tensorflow install page always refers to the latest Tensoflow install and this project has a locked Tensorflow version.
cudnn fails when attempting to call 'dog_detector(img_path)'
For me the fix was to open the 'dog-project' env: source activate dog-project Use conda to install the toolkit: conda install -c cudatoolkit Install a DOWNGRADED cudnn: conda install -c marta-sd cudnn