search

google-deepmind / open_x_embodiment

Apache License 2.0
623 stars 41 forks source link

sub-optimal #18

Open rainbow979 opened 8 months ago

rainbow979 commented 8 months ago

There are some datasets that have sub-optimal trajectories (e.g. maniskill, bridge). However, I found the total reward of all trajectories are same (i.e. 1). Is there anything wrong?

And is there any way to split the sub-optimal trajectories?

quanvuong commented 7 months ago

Thanks for letting us know and we'll investigate this issue