Hello, so I created a new Jupyter Notebook using a Tensorflow 1.5 + GPU + py3 machine. When trying to test the sample notebooks (Hello Tensorflow, Getting Started…) they output an error “libcuda.so.1: cannot open shared object file: No such file or directory”. I suppose that the nvidia drivers are not installed or not up-to-date? Do I need to take any steps to set this up? Thanks.
@adifatol We’re not able to reproduce this issue. Is it possible you selected the GPU container and ran it on a CPU instance (C2 or C7)? This would throw a CUDA error because CUDA is not available on the CPU instances. If you’re running on a CPU instance, select the TensorFlow 1.5.0 CPU Py3 container.
Hello, you are right! I was using a C2 instance, sorry I didn’t see that!
Thanks for your help!