Inquiry Regarding Argoverse2 Dataset Configuration and load_interval Settings

[SPA-junghokim]

Dear Author,

I would like to express my profound gratitude for your exceptional research, which has significantly contributed to the advancement of our work.
Your willingness to share the setup details for training on the nuscenes and argoverse2 datasets has been immensely helpful, accelerating our research progress.

I have a query regarding the configuration settings for the argoverse2 dataset.
I noticed that the load_interval is set to 4 in the config.
I am curious if this setting was also used in the experiments described in the maptrv2 paper.
Given that the argoverse2 dataset comprises 110k training samples and the nuscenes dataset contains around 28k, it seems that adjusting the load_interval to 1 and running for 6 epochs would align the number of training iterations between the two datasets more closely.
Could you please confirm if this is the case?

Your insights on this matter would greatly enhance our understanding and assist in optimizing our training procedure. Thank you once again for your invaluable support and for sharing your knowledge with us.

Looking forward to your response.

Best regards,
Jungho Kim.