Skip to content

tech and home datasets as pre-training datasets #3

@Jrui66

Description

@Jrui66

Thank you for sharing this valuable work! I'm currently trying to use tech and home datasets as pre-training datasets, but encountered an implementation issue.

After examining the error trace and reviewing the source code, I noticed that in the load_node_dataset function within dataset_benchmark.py, there doesn't appear to be handling logic for the tech and home dataset identifiers specifically.

Could you kindly clarify:

  1. Is this an intentional design choice (i.e., these datasets aren't supported for pre-training)?
  2. If they should be supported, is there a specific format or preprocessing step I might be missing?
  3. Would you recommend any workaround to include these datasets?

Appreciate your guidance!

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