Issue while implementing Reinforcement algorithm with tensorflow

Hello Everyone,
I was trying to rewrite the reinforcement learning algorithm from scratch.
but it’s not learning anything literally. I repeatedly compared the code, and it seemed the same

that a link of the jupyter.
the issue is the total points are going up and down crazily with each episode.

When you set up your own environment, you have to be extremely careful that get get all of the same versions of the tools and packages that are used in Coursera Labs. Incompatible versions are extremely common.