MLFlow
This extension adds operators to use MLFlow for experiment tracking
and model registration to the platfrom.
Broadly speaking the
extension covers two use cases:
Track your experiments.
While building models you try a lot of
combinations of models and preprocessing. Often you want to track them
to see whats good. The extension provides operators to use a MLFlow
server to do this. Namely:
- Create Experiment
- Create
Run
- Delete Experiment
- Delete Run
- Get
Runs
- Get Run
- Get Experiment
- Log Run
The
central operator is Log Run.
Track your deployments.
The second use case is to maintain
deployed models. MLFlow offers a Model Registry. This extension allows
you to use models from a worklflow within the model registry.
The operators are:
- Create Model
- Create Model
Version
- Retrieve Model
- Transition State
- Get
Model Information
For questions please consult community.rapidminer.com or our professional support.
--- Version 1.2.0 ---
- Added an option to give a run a custom name in Create Run
- Added a parameter to Log Run to allow the user to log custom parameters
- Log Run is now always logging the operator name as well as a parameter, if a parameter list is connected
--- Version 1.1.0 ---
- Added an option to get more logging for all operators. You can activate it in the settings.
- Added a global time_out to the settings for all the HTTP requests. Set this to 60sec as default.
--- Version 1.0.5 ---
- Minor fix which fixes a bug that performances not coming from a validation operator are not loggable.
Product Details
Version | 1.2.0 |
File size | 13 MB |
Downloads | 1998 (27 Today) |
Vendor | RapidMiner Labs |
Category | Operators |
Released | 9/6/24 |
Last Update | 9/6/24 11:51 AM |
License | AGPL |
Product web site | rapidminer.com |
Rating | (0)
|
Comments
Log in to post comments.