Hi, I am working on a few experiments, exploring deep reinforcement learning using ml agents. one is a simple card game where the opponent is controlled by an intelligent agent, the other is a driving game where the agents control cars. Both work very well with PPO. However, I am facing a problem with SAC. While training with SAC, Unity constantly freezes after a few seconds of training for a minute and longer, then continues the training for some seconds, freezes again, and so on... I tried all kinds of different hyperparameter, but the problem remains. Training therefore takes a really long time (for example a few hours for "only" 100K steps). Also, the agents don't seem to improve. RAM and CPU seem fine. I have no idea where the problem lies, or is this normal behavior for SAC? I am using Unity 2022.3 and Python 3.9.13 and ml agents 2.0.1. I would really appreciate if someone could help me with this problem
#ML-Agents, SAC freezing.
crude bane