'Stable Baselines 3 DDPG and TD3 Agents both stop at a specific step and compute forever
I'm testing different DRL Agents on a Custom Environment using Stable Baselines 3 and Gym. I tested A2C and PPO Agents on the environment and everything works fine but when I test an Agent with DDPG or TD3 both stop a the same time step 974.759. The program is not stopped, you can see that they are still running and computing but they basically wont progress.
I implemented some print statements to see on which step they are and which actions they took but at the named step they just dont take any more actions and get in the next step but just compute. I restarted the code several times and also waited couple of hours but nothing happenes.
Does someone know what I'm doing wrong or what I could do to solve the problem? I appreciate your support and would be happy to hear your suggestions!
Edit: The action space is continuous as well as the observation space
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|