Output of code with a seed differs each run

I have been getting into reinforcement learning lately and wanted to see how the Augmented Random Search would work. I used a code found on github: https://gist.github.com/ZSalloum/61828d5f2b3805b675fb5b34dbf6b73d#file-ars-py

The code works fine but the output differs each run. Even though there is a random seed set at line 143.

I feel like the only line where randomness is used is in line 67 where random matrices are drawn using numpy. When I used that method by itself, the random seed worked just fine.

Is there something I am missing here?

Thanks in advance