0

I am working on an MLOps project on Azure Databricks and after training the model using the 'SparkXGBRegressorModel' library I want to log it to Feature Store. The problem is that using the fs.log_model() function it raises an error because of the model data type which is 'SparkXGBRegressorModel'.

Does anyone know how can I log this model to Feature Store?

I tried to register it to Feature Store using this code I found on the Databricks documentation:

fs.log_model(
            model=xgb_model,
            artifact_path=f"{model_name}_artifact_path",
            flavor=mlflow.xgboost,
            training_set = training_set ,
            registered_model_name = model_name
        )

But as I said it raised the following error:

AttributeError: 'SparkXGBRegressorModel' object has no attribute 'save_model'

Φιλιππος Πριοβολος
  • 1
  • 1

0 Answers0