Skip to content

Clarification and suggestions regarding trained_models_dir path #3

@vlyalcin

Description

@vlyalcin

Hi,

First of all, thank you for making this project open-source and sharing it with the community, it's much appreciated.

While exploring the code, I came across the following section:

module_path = os.path.abspath(__file__)
globals()['__file__'] = module_path
project_path = os.path.dirname(os.path.abspath(__file__))
trained_models_dir = os.path.join(project_path, 'trained_models')

self.models = []
model_ckpts = glob(os.path.join(trained_models_dir, '*.ckpt'))

I’d like to share a few thoughts and questions regarding this part:

Hardcoded path: Is there a reason why the trained_models directory path is hardcoded? Would it be more flexible to allow this to be passed as a configurable argument or environment variable?

Missing directory: I couldn’t find a trained_models/ directory or .ckpt files in the repository. Are these meant to be downloaded separately or generated?

Documentation: I didn’t see any mention of the trained_models_dir requirement in the README. Adding a short section about this, including how and where to obtain the necessary .ckpt files could help new users get started more easily.

Thanks again for your work!

Best regards,
Volkan

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