snorkel-team / snorkel

A system for quickly generating training data with weak supervision
https://snorkel.org
Apache License 2.0
5.81k stars 857 forks source link

Show a progress bar on LabelModel.fit #1624

Closed nbeuchat closed 3 years ago

nbeuchat commented 3 years ago

Is your feature request related to a problem? Please describe.

When fitting a LabelModel with hundreds of labeling functions and thousands of datapoints, it can take several minutes to complete. When calling fit() on my LabelModel, I currently have no clue how many epochs were already processed nor an estimated ETA (even if verbose is set to True).

Describe the solution you'd like

When calling LabelModel().fit(), a tqdm progress bar would be displayed and updated after each epoch. This would at least be shown if verbose=True

Describe alternatives you've considered

time library and print statements

Additional context

Snorkel version: snorkel==0.9.6 Run in Jupyter lab

rjurney commented 3 years ago

Is this possible? I mean, does the matrix optimization give a percentage completeness?