About preprocess logic or code for raw dataset

[Feiyu-Yao]

Hi Xueyi!

Thanks for the great work! Would you mind share the code or logic for preprocessing the raw datasets? Now only the processed data i savailable in the project and it's hard to figure out what clip_clean['f1']-clip_clean['f13'] (line 262 in data_loaders/humanml/data/dataset_ours_single_seq.py) means from preprocessed data.

Looking forward to your reply,

Best,
Feiyu

[Xueyi Liu]

Hi Feiyu,

Sorry for the late response. I've been overwhelmed with final projects and exams recently. The data preprocessing for GRAB is mainly adapted from that used in TOCH. I vaguely remember making some small changes to this script before, but I can't quite recall the specifics. Additionally, I'm unable to quickly locate the preprocessing code I used at the time, as it might be hidden in the part of my code that I haven't organized yet. The preprocessing code for other datasets is also not organized yet. The process is a little bit complicated, and I'm afraid it's difficult to describe clearly in words. After my final term, I will allocate some time to work on this. I kindly ask for your continued patience. Thank you!

Best,
Xueyi