Python 在模型记录 mlflow 期间设置标签

当使用mlflow.sklearn.log_model-you work with the experiment registry which is run-focused时，只能描述和标记实验和运行.

如果你想在模型上设置标签，你可以 Select need to work with the model registry.

我推荐的解决方案是设置为register the model when logging using registered_model_name(也有更细粒度的方法)，并使用MLFlowClientAPI设置已注册模型的自定义属性(如标记).

以下是一个有效的示例:

import mlflow
from mlflow.client import MlflowClient

mlflow.set_tracking_uri('http://0.0.0.0:5000')

experiment_name = 'test_mlflow'
try:
    experiment_id = mlflow.create_experiment(experiment_name)
except:
    experiment_id = mlflow.get_experiment_by_name(experiment_name).experiment_id

from sklearn.linear_model import LogisticRegression
from sklearn.datasets import load_iris
from sklearn.metrics import accuracy_score

with mlflow.start_run(experiment_id = experiment_id):
    # log performance and register the model
    X, y = load_iris(return_X_y=True)
    params = {"C": 0.1, "random_state": 42}
    mlflow.log_params(params)
    lr = LogisticRegression(**params).fit(X, y)
    y_pred = lr.predict(X)
    mlflow.log_metric("accuracy", accuracy_score(y, y_pred))
    mlflow.sklearn.log_model(lr, 
        artifact_path="models", 
        registered_model_name='test-model'
    )
    # set extra tags on the model
    client = MlflowClient(mlflow.get_tracking_uri())
    model_info = client.get_latest_versions('test-model')[0]
    client.set_model_version_tag(
        name='test-model',
        version=model_info.version,
        key='task',
        value='regression'
    )

下面是插图

另请参阅这篇关于MLFlow Client的优秀文档.