Options
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
03-31-2025 08:58 PM
I had this same problem, and followed the same steps, but it did not work until I also explicitly set the experiment_id of the children.
EXPERIMENT_NAME = '/Users/my_user_name/my_experiment'
experiment_id = mlflow.get_experiment_by_name(EXPERIMENT_NAME)
if experiment_id is None:
experiment_id = mlflow.create_experiment(EXPERIMENT_NAME)
else:
experiment_id = experiment_id.experiment_id
mlflow.set_experiment(experiment_id=experiment_id)
#
#
#
def train_model(params):
with mlflow.start_run(nested=True, parent_run_id=params['parent_run_id'], experiment_id=params['experiment_id']):
# training and logging here
#
#
#
search_space = #dict of parameters
with mlflow.start_run(run_name=my_run_name) as parent_run:
run_id_value = parent_run.info.run_id
search_space['parent_run_id'] = run_id_value
search_space['experiment_id'] = experiment_id
best_params = fmin(fn=train_model,
space=search_space,
#etc
)