how to sweep hyperparameters

[Jasha10]

This discussion is based on the comment by @LukeLIN-web on issue #2555.

Did you know how to sweep the hyperparameters? I write config.yaml

defaults:
  - model: ???
  - _self_
  
root: '/tmp/datasets'

hydra:
  sweeper:
    params:
      model: sage
      dataset: reddit
      model.num_train_worker: range(2, 7)

But it shows:

You must specify 'model', e.g, model=<OPTION>
Available options:
        sage

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.
See https://hydra.cc/docs/1.2/upgrades/1.1_to_1.2/changes_to_job_working_dir/ for more information.
  ret = run_job(
Error executing job with overrides: []
Traceback (most recent call last):
  File "microbatch_reddit_quiver_queue.py", line 152, in main
    model_name, dataset_name = conf.model.name, conf.dataset.name
omegaconf.errors.ConfigAttributeError: Key 'model' is not in struct
    full_key: model
    object_type=dict

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.

[Jasha10]

You need to pass the --multirun flag:

python app.py --multirun