Best practices to organize hyperparameter searches

When you run many hyperparameter searches, tag your runs so you can group and find related experiments later.

Tag runs in your code

Pass unique tags when you initialize each run:

import wandb

with wandb.init(tags=["sweep-3", "lr-0.01"]) as run:
    # Your training code here
    pass

For more information about wandb.init(), see the wandb.init() reference and Add labels to runs with tags.

Open your project and go to the Runs tab.
Click Filter above the runs table.
Choose Tags, select an operator such as is or in, and enter the tag you set in code.

See Filter runs for more filter options.

⌘I