How to log MLFLow runs into Vectice

Vectice provides an easy way to retrieve already saved information from your MLflow runs and document them into Vectice.

To log model information from MLFlow use the following lines of code:

# Use the following model wrapper to retrieve your MLFlow run artifacts
model = Model.mlflow(run_id='your_run_id', client=mlflow, url='https://your_url.com')

# Log MLFlow artifacts in Vectice
iteration.log(model)
ParameterDescriptionDefault

run_id

A unique identifier for the MLflow experiment run.

required

client

The MLflow client object used for interacting with MLflow. This parameter is mandatory to ensure compatibility and avoid issues with specific configurations.

required

url

The URL to theMLFlow UI to access the run information.

optional

derived_from

List of datasets (or version ids) to link as lineage.

optional

You can access your MLFlow run artifacts in Vectice where you will see the model version metrics, properties, and a link to the MLFlow UI of your run.

Find this information by navigating to Models > Select model name > Select version.