Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Issues loading pretrained model #21

Open
kvantricht opened this issue Oct 18, 2023 · 3 comments
Open

Issues loading pretrained model #21

kvantricht opened this issue Oct 18, 2023 · 3 comments

Comments

@kvantricht
Copy link
Contributor

kvantricht commented Oct 18, 2023

When I'm trying to install Presto as a package not in editable mode (e.g. to pack it to run on our cluster), I'm experiencing a couple of issues:

  • the current pretrained model loading method does not work as the model weights are not considered as package data (probably separate issue).
  • An alternative would be to accept a non-default model weights location in the load_pretrained method here but at the moment this is not possible
  • As a last resort, I was trying to copy the data folder manually and using update_data_dir from here, but while the data_dir variable is successfully updated, default_model_path from here is not, so it has actually no effect for loading the pretrained model
@kvantricht kvantricht changed the title Non-default data_dir Issues loading pretrained model Oct 18, 2023
@gabrieltseng
Copy link
Collaborator

In addition to the fix in #22, I think it might make sense to move the weights into the presto folder - what do you think?

@kvantricht
Copy link
Contributor Author

Yes I think that makes sense, although it needs to be tested if that would resolve the issue. Pathlike reference to package files doesn't always work on all systems.

@gabrieltseng
Copy link
Collaborator

Okay - I'll make this change (with some tests) in addition to the notes you left in #20 to a new PR

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants