Search Unity

Trainning is more and more slow

Discussion in 'ML-Agents' started by jokerHHH, Jul 19, 2021.

  1. jokerHHH

    jokerHHH

    Joined:
    May 11, 2021
    Posts:
    9
    When I train the agent to learn in my own environment, in the beginning, it works well, trainning is fast. But after a few steps the trainng is becoming more and more slow. It took me more than 1000s to summary 10000 steps. Why would this happened?
    upload_2021-7-19_10-58-34.png
     
    jokerHHH, Jul 19, 2021
    #1

  2. ervteng_unity

    ervteng_unity

    Unity Technologies

    Joined:
    Dec 6, 2018
    Posts:
    150
    You may have a memory leak - can you check the memory usage of both Python and Unity over time?
     
    ervteng_unity, Jul 21, 2021
    #2