Question about the code on bench2drive benchmark

[bugmakerwww]

Thanks for the great work.
The paper has demonstrate the comparable result on both navsim and bench2drive benchmark. However, I can only get the navsim training and evaluation code in this repo.
Do you use different code for training, feature builder and closeloop evaluation?

[owl-10]

We use the same model files but rewrote the training and inference code. I’ll release the results and code when I have time soon.