Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

Question Questions about reward in mlagents-learn console

Discussion in 'ML-Agents' started by Dream_Surpass, Mar 21, 2023.

  1. Dream_Surpass

    Dream_Surpass

    Joined:
    Dec 2, 2022
    Posts:
    18
    Does the Mean Reward log in console mean the sum of all the rewards(without discount) during one episode, or sum the reward contains discount(gamma)?

    upload_2023-3-21_21-43-25.png
     
  2. Joe-Soap

    Joe-Soap

    Joined:
    Oct 10, 2020
    Posts:
    2
    I stand to be corrected, but I've taken the Mean Reward to be the Mean reward achieved over all the previous episodes. So each time EndEpisode(...) is called, the total reward, for that episode, is the value that is used to make up the population of values for which the Mean is calculated.

    Anyone care to chime in if I'm off the mark?