Re:DVC – Open Source Machine Learning Version Control System
by usasha on Monday February 11, 2019 @03:14AM
(#19132805)
Just ran thru official tutorial and I'm pretty impressed.
As I understood the idea is to:
- use git branch for each experiment (change of hyperparameters etc.)
- define pipeline stages (preprocessing, train/test split, model training, model validation)
- after this steps you can change any part of pipeline (say data preprocessing or model parameters) and run `dvc repro` to reproduce all stages for which dependancies changed and track metrics for all branches, which os pretty cool and reduce experiment logs in wiki
Re:DVC – Open Source Machine Learning Version Control System
by dmpetrov on Monday February 11, 2019 @06:21AM
(#19133634)
Exactly. And it is flexible as Git. You can define your own workflow. For example, some data scientists avoid using git-branches for experiments - they use directories.
As I understood the idea is to: - use git branch for each experiment (change of hyperparameters etc.) - define pipeline stages (preprocessing, train/test split, model training, model validation) - after this steps you can change any part of pipeline (say data preprocessing or model parameters) and run `dvc repro` to reproduce all stages for which dependancies changed and track metrics for all branches, which os pretty cool and reduce experiment logs in wiki