Save episode rewards in ray.tune

Question

I am training several agents with PPO algorithms in a multi-agent environment using rllib/ray. I am using the ray.tune() command to train the agents and then lo