cli for linear classifier

[edyoshikun]

Probably best to checkout the README.md on how to use this.

• Add viscy-dynaclr CLI entry point with two new commands for linear classifier workflows:
  • train-linear-classifier — train a logistic regression classifier on DynaCLR cell embeddings with W&B tracking
  • apply-linear-classifier — apply a trained classifier to new embeddings and save predictions
• Add core implementation for LinearClassifierPipeline (classifier + optional StandardScaler/PCA) with full W&B artifact management
• Add Pydantic-based configuration with validation for both training and inference configs
• Add example YAML configs and documentation (annotations schema, naming conventions, README)

[Soorya19Pradeep]

@edyoshikun, I tried out the cli. I think I have followed the instructions for setting up my wandb settings, including pip installations and logging in using API key and setting base url to https://czi.wandb.io. The training seems to be working, but it is not getting stored. I am getting the following error at the end:

============================================================
SAVING MODEL AND LOGGING TO WANDB

wandb: Loading settings from /home/soorya.pradeep/.config/wandb/settings
wandb: [wandb.login()] Loaded credentials for https://czi.wandb.io from /home/soorya.pradeep/.netrc.
wandb: W&B API key is configured. Use wandb login --relogin to force relogin

❌ Training failed: Error uploading run: returned error 401: {"data":{"upsertBucket":null},"errors":[{"message":"user is not logged in","path":["upsertBucket"],"extensions":{"code":"PERMISSION_ERROR"}}]}
Aborted!
Do you have any suggestions? Thanks!

[Soorya19Pradeep]

The permission issue is resolved now. Thank you!
The trained classifier is now stored in my personal profile. Do you know which setting I need to change to get them to store in the computational imaging profile?