How did you split the train/val/test dataset for objaverse?

modelscope / richdreamer

Live Demo：https://modelscope.cn/studios/Damo_XR_Lab/3D_AIGC

https://aigc3d.github.io/richdreamer/

Apache License 2.0

362 stars 13 forks source link

Closed thucz closed 5 months ago

thucz commented 5 months ago

Could you tell me how you split the objaverse dataset?

lingtengqiu commented 5 months ago

we random sample 99% Objaverse for training and 1% for validation.

thucz commented 5 months ago

Thanks!