Hi,
I am trying to understand what timesteps_per_iteration is. One iteration is one .train() call of the algorithm which will update the estimation. What does timesteps_per_iteration define and how is it related to episodes? I believe it is not horizon also.