Firstly, thanks for the contribution of your work.
I would like to ask have you ever submitted your evaluation results on Argoverse 1 Leaderboard?
I followed your data extractor and data-module setting, and trained the DeMo on AV1. I can get similar evaluation results on AV1 VAL set as listed in your paper. However, when i submitted the evaluation results on test set(to be precisely, test_obs set in AV1), the results is bad. It seems overfitting happens.
Can you help look into this issue? thanks.
Firstly, thanks for the contribution of your work.
I would like to ask have you ever submitted your evaluation results on Argoverse 1 Leaderboard?
I followed your data extractor and data-module setting, and trained the DeMo on AV1. I can get similar evaluation results on AV1 VAL set as listed in your paper. However, when i submitted the evaluation results on test set(to be precisely, test_obs set in AV1), the results is bad. It seems overfitting happens.
Can you help look into this issue? thanks.