search

ZikangZhou / HiVT

[CVPR 2022] HiVT: Hierarchical Vector Transformer for Multi-Agent Motion Prediction
https://openaccess.thecvf.com/content/CVPR2022/papers/Zhou_HiVT_Hierarchical_Vector_Transformer_for_Multi-Agent_Motion_Prediction_CVPR_2022_paper.pdf
Apache License 2.0
601 stars 118 forks source link

Performance of GRUDecoder #17

Closed Sungmin-Woo closed 1 year ago

Sungmin-Woo commented 1 year ago

Hello, @ZikangZhou. Thanks for your code! How much is the performance gap between the models using GRUDecoder and MLPDecoder? Do you have any idea why GRUDecoder leads to lower performance? I am looking forward to your reply!

ZikangZhou commented 1 year ago

According to my experiments, GRUDecoder is actually a better choice than MLPDecoder 😂😂 There is no obvious performance gap in terms of ADE, but using GRU can improve FDE marginally. RNN has never died!

Sungmin-Woo commented 1 year ago

Thank you for the quick reply. I'm glad to hear that! My issue is solved now :)