pyflink.table.table_environment.TableEnvironment.create_model

TableEnvironment.create_model(model_path: str, model_descriptor: ModelDescriptor, ignore_if_exists: bool | None = False)

Registers the given ModelDescriptor as a catalog model similar to SQL models.

The ModelDescriptor is converted into a CatalogModel and stored in the catalog.

If the model should not be permanently stored in a catalog, use create_temporary_model() instead.

Examples:

>>> table_env.create_model("MyModel", ModelDescriptor.for_provider("OPENAI")
...     .input_schema(Schema.new_builder()
...                   .column("f0", DataTypes.STRING())
...                   .build())
...     .output_schema(Schema.new_builder()
...                   .column("label", DataTypes.STRING())
...                   .build())
...     .option("task", "regression")
...     .option("type", "remote")
...     .
...     .
...     .build(),
...  True)

Parameters:

• model_path – The path under which the model will be registered.

• model_descriptor – Template for creating a CatalogModel instance.

• ignore_if_exists – If a model exists under the given path and this flag is set, no operation is executed. An exception is thrown otherwise.

Added in version 2.1.0.