Minor bug that removes best performing trajectory in gym experiments
Closed this issue · 1 comments
micahcarroll commented
I believe that this line should have a <=
rather than a <
in order for the code to not cut out the best performing trajectory even when using pct_traj = 1.
decision-transformer/gym/experiment.py
Line 109 in f04280e
To replicate, use a dataset with 2 trajectories and use pct_traj = 1.
and the resulting num_trajectories
will just be 1 rather than 2.
kzl commented
Thanks for catching this! I think you're right.