Skip to content

Merged Documentation

Documentation for MLFlow Methods

get_experiment_id_by_name

Functionality

This function takes an experiment name and returns its corresponding experiment ID using MLFlow. If no matching experiment is found, it returns None.

Parameters

  • experiment_name (str): The name of the experiment to retrieve the ID for.

Returns

  • str or None: The experiment ID if found, otherwise None.

Usage

  • Purpose: Retrieve the ID of a given experiment for tracking and logging purposes.

Example

experiment_id = get_experiment_id_by_name(
    "Example_Experiment"
)
print(experiment_id)

get_run_id_by_name

Functionality

Search for a run within a given experiment by matching the run name. It returns the run_id of the first matching run. If no run is found, the function returns None.

Parameters

  • experiment_id: MLFlow experiment identifier as a string.
  • run_name: MLFlow run name to filter the experiments.

Usage

  • Purpose: Retrieve the run_id for a specified run within an experiment.

Example

import mlflow

experiment_id = "123456"
run_name = "my_run"
run_id = get_run_id_by_name(experiment_id, run_name)
print(run_id)

get_mlflow_results_url

Functionality

Generates a URL for checking MLflow experiment results by setting the tracking URI and filtering experiments based on a naming convention.

Parameters

  • mlflow_url: MLflow tracking URI.
  • batch_id: ID representing the released batch.
  • model_id: ID of the model for constructing the experiment name.

Usage

  • Purpose: Retrieve the experiment ID of the first matching experiment, based on a naming convention derived from model and batch IDs.

Example

Assuming MLflow server at 'http://localhost', batch 'batch123', and model 'modelXYZ':

url = get_mlflow_results_url('http://localhost', 'batch123', 'modelXYZ')