Hyperparameter Management¶

Hyperparameter management allows runtime-driven control of experiment settings.

Supported patterns¶

Register a Python dictionary.
Register a YAML file path and let Weightslab watch for updates.

Minimal examples¶

import weightslab as wl

# Dict-based registration
wl.watch_or_edit(
    {
        "experiment_name": "seg_exp",
        "training_steps_to_do": 1000,
        "optimizer": {"lr": 1e-3},
    },
    flag="hyperparameters",
    name="seg_exp",
)

# File-based registration and polling
wl.watch_or_edit(
    "./config.yaml",
    flag="hyperparameters",
    defaults={"optimizer": {"lr": 1e-3}},
    poll_interval=1.0,
)

Typical controlled values¶

Learning rate and optimizer settings
Batch size and dataloader settings
Logging paths and experiment metadata
Service toggles (CLI/gRPC)

Tips¶

Set root_log_dir early to keep all artifacts under one experiment folder.
Keep defaults in code and environment-specific overrides in YAML.
Version-control your baseline YAML templates.