Create Experiment
(MLFlow)
Synopsis
This operator creates a new run on a mlflow serverDescription
This operator creates a new run into an existing experiment. The experiment needs to be defined using the experiment_id parameter.
A A run is simply put a single execution of your training process and can contain performances, metrics and other parameters
This operator acts similary to mlflow.client.create_run() in python.
Input
con (Connection)Connection to a MLFlow Server
through (Data table)Through port to send data through
Output
- exa (Data table)
Table with the information on the generated run
- run
A RunIOObject, which holds all the information on the given run and can be used to log things into the run.
- through (Data table)
Through port to send data through
Parameters
- experiment name Name of the experiment to create the run in
- define run name If set to true the user has to provide a name for the created run. If set to false a random name is used.
- run name The name of the run. Only available if define run name is set to true.