Skip to content

Training settings about SURREAL #9

Description

@zhangyujie9999

Now I get the training data of 3D-coded, and make a training experiment with 1000 data. But I got a bad result. Could you tell me how to set the parameters related to SURREAL?

In the process of training, I also found that in dataset.py, the values of source_normals and target_normals were set to none, which led to the two lines of code could not run.

data["source_normals"] = torch.cat([_source_normals, _target_normals, _source_normals], dim=0).contiguous()
data["target_normals"] = torch.cat([_target_normals, _source_normals, _source_normals], dim=0).contiguous()

I think if the value passed in is none, then these two lines of code will not work, so I annotated them. What negative impact will this have on the training process?

Looking forward to your help, thx!

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions