How to combine OC22 and OC20 dataset for model training using fairchem?

[zouzihan123]

What would you like to report?

Dear fairchem staff, I want to make a dataset by mixing OC20 and OC22. But, in the Open Catalysis project, I found that the data formats provided for OC20 and OC22 are not the same. Where, OC22 is provided as .lmdb, but OC20 (training set) is provided as .extxyz.xz. I would like to ask if fairchem project has code that reads and combines the two formats of the dataset together, I have only found code for loading .lmdb before.
Sincerely