Request to Share Stage 1 Metrics for Reproducibility on nuScenes

[wangziyi2016]

Hi, thank you for your excellent work on DL4AGX and for open-sourcing this repository!

I'm currently working on reproducing the UniAD model on the nuScenes dataset. I was wondering if you could share the Stage 1 results, such as Tracking AMOTA and Mapping IoU-lane, in addition to the final Stage 2 results published in this document.

Results from the PyTorch version alone would be sufficient—we’re mainly looking to validate that Stage 1 training is proceeding correctly before moving to the next stage.

Thanks again for your contributions, and looking forward to your response!

[Jian-danai]

Hi, thanks for your interest, our UniAD project mainly focuses on planning performance and TensorRT enablement. we don't have plan to share stage 1 detailed metrics. Let us know if you are unable to reproduce UniAD-tiny stage2 planning metrics.