rail-berkeley / rlkit

Collection of reinforcement learning algorithms
MIT License
2.43k stars 547 forks source link

evaluation/env_infos/failed Mean = 0.0 #133

Closed liuqi8827 closed 3 years ago

liuqi8827 commented 3 years ago

Thanks your great work. I'm facing a problem of getting the information of evaluation/env_infos/failed Mean. I found that the output of evaluation/env_infos/failed Mean always equal to 0.0. It never changed.

evaluation/env_infos/final/failed Mean    0.0
evaluation/env_infos/final/failed Std     0.0
evaluation/env_infos/final/failed Max     False
evaluation/env_infos/final/failed Min     False
evaluation/env_infos/initial/failed Mean  0.0
evaluation/env_infos/initial/failed Std   0.0
evaluation/env_infos/initial/failed Max   False
evaluation/env_infos/initial/failed Min   False
evaluation/env_infos/failed Mean          0.0
evaluation/env_infos/failed Std           0.0
evaluation/env_infos/failed Max           False
evaluation/env_infos/failed Min           False

Can you give some suggestions?

Thanks a lot!

vitchyr commented 3 years ago

Hi, are you sure that this is a logging issue? Could it be that failed is always equal to 0 in your environment?

liuqi8827 commented 3 years ago

@vitchyr Thanks for your reply.

It's my fault.