mix different datasets in training

[Zhao-Qihao]

What is the feature?

I want to train a centerpoint model with two different datasets: nuscenes and lyft dataset ,only use their common classes for trainning

Any other context?

No response