LIBERO-90 dataset preprocessing #17
Description
Thank you very much for your excellent work.
I am currently working with the LIBERO-90 dataset and trying to reproduce the data processing steps mentioned in your documentation.
The documentation states that:
"steps 1 and 2 follow exactly similar to CALVIN."
However, I encountered an issue when running the statistics calculation script (python scripts/dataset/calculate_<meanstd|minmax>_stats.py) specifically on the LIBERO-90 data. I pin the error related to:
"the varying privilege state vector sizes (scene_obs equivalent in CALVIN) for different LIBERO-90 scenes and layouts."
Given this potential incompatibility with the CALVIN-style state calculation due to variable state sizes, I would be very grateful if you could clarify what is the specific and correctprocedure you followed for data processing for the LIBERO-90 dataset?
Greatly appreciated for your help!