I am assuming (perhaps incorrectly) that these benchmark results are obtained via online RL and not from a static dataset via offline RL. If so, how can we translate the x-axis of these plots to a more standard metric for online RL, such as "environment steps"?
Hi,
Thanks for providing the training code.
I've seen the results plots in the documentation (here: https://robosuite.ai/docs/algorithms/benchmarking.html) and read the whitepaper. However, the definition of "epoch" is unclear.
I am assuming (perhaps incorrectly) that these benchmark results are obtained via online RL and not from a static dataset via offline RL. If so, how can we translate the x-axis of these plots to a more standard metric for online RL, such as "environment steps"?