tensorflow gpu version not enabled in okwrtdsh/anaconda3:keras-9.2-cudnn7

[alidadsetan]

Running the container: sudo nvidia-docker run -it okwrtdsh/anaconda3:keras-9.2-cudnn7 bash

"attaching" to container: sudo docker exec -it container_name bash

Running python repl and importing tensorflow to check whether gpu is enabled in tensorflow:

from tensorflow.python.client import device_lib
print(device_lib.list_local_devices())

which prints:

2018-07-05 14:56:47.988072: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
[name: "/device:CPU:0"
device_type: "CPU"
memory_limit: 268435456
locality {
}
incarnation: 3815053792300585653
]

That shows device_type as cpu. Can this be a bug?

[okwrtdsh]

@CowboyBebug Please, check your CUDA Toolkit and Compatible Driver Versions. keras-9.2-cudnn7 requires 396.26 or later (Linux x86_64). You can check the driver version with nvidia-smi command.