conglu1997 / v-d4rl

Challenges and Opportunities in Offline Reinforcement Learning from Visual Observations
MIT License
94 stars 9 forks source link

Inconsistency between 64px and 84 datasets #10

Closed yihaosun1124 closed 9 months ago

yihaosun1124 commented 10 months ago

Thanks for the lovely open-source benchmark. However, I noticed inconsistencies between the 64px and 84px datasets. Specifically, I observed that the number of transitions in the 64px dataset is less than in the 84px one. For instance, the cheetah_run_medium_replay dataset in 84px has 200k transitions (which aligns with the description in your paper), but the 64px version only has 50k. I'm wondering if you might have missed uploading some of the data for the 64px version.

conglu1997 commented 10 months ago

Hey! How are you computing this?

I just checked the 64px version https://drive.google.com/drive/folders/1xAcRdQHgZlcs2tTcLQNOT_wMO-bm3w44?usp=drive_link and there are 400 files each with 1 episode each (500 steps in each), which makes 200K.