Replies: 1 comment
-
Thank you for your affirmation of OmniSafe. In fact, OmniSafe utilizes the Currently, OmniSafe does not support recording data for each individual step. This is because in the reinforcement learning paradigm, training often involves a large number of steps, reaching up to 1e6 steps or more, and recording data for each step becomes extremely challenging. If you could provide a detailed example regarding "tracking other parameter values my custom environment generates during each step," we can offer better guidance on how to adjust the logger to record such data. |
Beta Was this translation helpful? Give feedback.
-
Thank you for doing a great job on the Omnisafe framework! I'm using Omnisafe with a customized environment I created based on "omnisafe/tests/simple_env.py". I'm wondering if there's an easy way to track the reward throughout each episode to see how the learning is progressing. Additionally, I'd like to track other parameter values my custom environment generates during each step.
Beta Was this translation helpful? Give feedback.
All reactions