Skip to content

Question about the code used for Nuscenes training having a negative loss when trained on modelnet40 #14

@CorgiBoyG

Description

@CorgiBoyG

Thanks for doing such a great job and for providing the network code for Nuscenes training. I noticed that the projection parameters in the code for loading the dataset for Nuscenes are self-generated. I then referenced this for the modelnet40 dataset, where the overlap handling is referenced to the RPMNet protocol. Regarding the projection parameters, I kept them the same. The result is that after epoch about 2~4, the loss is negative. What should I do and could you provide some good suggestions to help me please. Looking forward to your reply and thanks again for doing such a good job.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions