I have been working with my project using PPO to train my agents, but now, I am trying to train them using SAC to compare the algorithms and every time I run the training the Unity environment get stuck. However, I the training is still doing its job asa I get the summaries, but i can not see what are the agents doing in the environment. Any idea?
Do the agents move at all, or is the environment completely stuck? Sometimes the training process takes up CPU and Unity will slow down the rendering - but in the backend the agents are still moving and it’s still training properly.
Yeah, they are still training, but I can not see how they are doing wich is kind of important for me. I have to see the transtion of colors that they take.
Try running with a lower timescale using --time-scale=X when running mlagents-learn (default is 20).
1 Like