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

[Feature request] CLI for generating templates #6

Closed
tachukao opened this issue Feb 4, 2023 · 1 comment
Closed

[Feature request] CLI for generating templates #6

tachukao opened this issue Feb 4, 2023 · 1 comment

Comments

@tachukao
Copy link

tachukao commented Feb 4, 2023

Description

  • Please correct me if I'm off-base here, but my current understanding of the intended usage of this software is for users to:
    1. git clone this repo
    2. modify the relevant template yaml files/scripts (e.g., run_pbt.py) for their problem/dataset
    3. deploy ray or kubeflow according to instructions in the documentation.
  • I believe these templates files are going to make it much easier for neuroscientists to apply AutoLFADS at scale to their own datasets
  • However, the current usage would also mean that users would have to either
    • work entirely within this repo with multiple versions of essentially the same files, making it challenging to use git to version control their own projects, or
    • manually copy these template files into their own projects

Proposal

  • I think a potential solution to make this process more streamlined/modular is to provide a (pip installable) command line interface that automatically generates the relevant templates/scripts in a user-specified target directory.

  • For example, for folks wanting to deploy autolfads with ray, they could do something like:

    autolfads-deploy generate --strategy ray --target [target_dir]

    This would generate everything under the ray folder in target_dir. Folks can then modify those files inside target_dir.

  • I think this would provide a clear API for how to use this software package, as it wasn't entirely clear to me before. Automatic tests can also be written to test this generation process. Otherwise, it seems like automatic testing is missing at the moment?

ref: Joss review

@a9p
Copy link
Member

a9p commented Mar 9, 2023

Thank you for your feedback! We overhauled the examples and documentation per your noted concerns. We additionally added some clarification around experiment management as an alternative to the above proposal that should provide reduced user burden/barrier to using this repository. And an end-to-end test via ci was introduced in #17. Feel free to reopen this issue if we missed addressing any of your concerns!

@a9p a9p closed this as completed Mar 9, 2023
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