what is the mean of train_episodes and valid_episodes?

Question

what is the mean of train_episodes and valid_episodes?

GeorgeDUT opened this issue 4 years ago · 4 comments

the test.py writes a file about "task", "train_episodes", "valid_episodes". "train_episodes", "valid_episodes" are the total rewards of an episode?

Answer 1 · 2020-10-29T13:59:56.000Z

Yes train_returns and valid_returns are the (undiscounted) cumulated rewards for each task and each episode.

Answer 2 · 2020-11-01T15:30:15.000Z

what is the difference between "train_return" and "valid_return"?

Answer 3 · 2020-11-01T18:45:22.000Z

train_return correspond to the returns for the trajectories sampled with the initial policy (before adaptation), and valid_return correspond to the returns for the trajectories sampled with the adapted policy. Taking the notations of the paper (Algorithm 3), train_returns correspond to the returns on D and valid_returns correspond to the returns on D'.

Answer 4 · 2020-11-02T12:45:55.000Z

thanks so much, I get it