Unable to run antmaze_sac.py

Farama-Foundation / D4RL

A collection of reference environments for offline reinforcement learning

Apache License 2.0

1.32k stars 279 forks source link

Unable to run antmaze_sac.py #21

Closed s-gupta closed 4 years ago

s-gupta commented 4 years ago

Similar to #9, I believe appropriate version of rlkit is also needed to run antmaze_sac.py. Any estimate when the appropriate version of rlkit can be made available?

In particular, current code crashes at MdpPathCollector with the error init() got an unexpected keyword argument 'sparse_reward'

aviralkumar2907 commented 4 years ago

Thank you for pointing this out. We have added a script compatible with rlkit: https://github.com/rail-berkeley/d4rl/blob/master/scripts/antmaze_sac_rlkit.py

This should work with rlkit. Version differences in gym may give rise to an assertion error, which is fixable using instructions at the top of the launcher script.