Quick Question on cudNN error

when i run the code on the GPU, soon after i refresh the kernel, I get the following error:
RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED

But when I click the run option again (without refreshing the kernel), the error vanishes and the code executes. Does this mean that the code is not properly implemented on the GPU?

What is your CUDA version?

CUDA version is 9.0
CudNN is 7005
torch version is 1.0.1