You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Aug 28, 2021. It is now read-only.
I found we can't reproduce the experimental results after updating habitat-sim and habitat-api. I know it may be unrelated to this repo. But I am confused about the reason.
I hope that I can get some advice from you. It's useless even though I check their commit log.
Thanks
The text was updated successfully, but these errors were encountered:
I don't know if this is the cause, but the latest version of Habitat changed the default config so the episode order is now shuffled by default (which makes more sense, but might give different results). You can try with the old behavior by adding this to the config:
Sorry for replying late. I also find the pre-trained models perform worse after I use new habit-api and habitat-sim. One issue in habitat-api told that it happened because of different order of action space. However, this method includes three actions. I believe there is no influence for this repo. I will check again. Thanks a lot.
The problem is related to facebookresearch/habitat-lab#148. Following the instruction, I change the config with POINTGOAL_WITH_GPS_COMPASS_SENSOR . The results look good.
Hi,
I found we can't reproduce the experimental results after updating habitat-sim and habitat-api. I know it may be unrelated to this repo. But I am confused about the reason.
I hope that I can get some advice from you. It's useless even though I check their commit log.
Thanks
The text was updated successfully, but these errors were encountered: