cuDNN compatibility with TensorFlow

[DiegoZoracKy]

I'm unable to run some notebooks due to the following issue:

Loaded runtime CuDNN library: 7104 (compatibility version 7100) but source was compiled with 7005 (compatibility version 7000)

As described here: keras-team/keras#9567 it seems that downgrading the cuDNN version to 7.0.* make it work.

I fixed the problem locally by running the following inside the container:

apt-get update && apt-get install -y --allow-downgrades --no-install-recommends \
    libcudnn7=7.0.4.31-1+cuda9.0 \
    libcudnn7-dev=7.0.4.31-1+cuda9.0 

Would you consider updating your image to use this version of cuDNN?

Full log:

2018-07-20 00:43:05.491803: E tensorflow/stream_executor/cuda/cuda_dnn.cc:396] Loaded runtime CuDNN library: 7104 (compatibility version 7100) but source was compiled with 7005 (compatibility version 7000).  If using a binary install, upgrade your CuDNN library to match.  If building from sources, make sure the library loaded at runtime matches a compatible version specified during compile configuration.
2018-07-20 00:43:05.492799: F tensorflow/core/kernels/conv_ops.cc:712] Check failed: stream->parent()->GetConvolveAlgorithms( conv_parameters.ShouldIncludeWinogradNonfusedAlgo<T>(), &algorithms) 
[I 00:43:08.479 NotebookApp] KernelRestarter: restarting kernel (1/5), keep random ports