feat: add remote vertex model support (#237)

GarrettWu · web-flow · commit 0bfc4fb11768 · 2023-11-29T19:44:16.000Z
b/299356085
diff --git a/bigframes/ml/__init__.py b/bigframes/ml/__init__.py
@@ -26,4 +26,5 @@
     "llm",
     "forecasting",
     "imported",
+    "remote",
 ]
diff --git a/bigframes/ml/core.py b/bigframes/ml/core.py
@@ -294,13 +294,19 @@ def create_remote_model(
         self,
         session: bigframes.Session,
         connection_name: str,
+        input: Mapping[str, str] = {},
+        output: Mapping[str, str] = {},
         options: Mapping[str, Union[str, int, float, Iterable[str]]] = {},
     ) -> BqmlModel:
         """Create a session-temporary BQML remote model with the CREATE OR REPLACE MODEL statement
 
         Args:
             connection_name:
                 a BQ connection to talk with Vertex AI, of the format <PROJECT_NUMBER>.<REGION>.<CONNECTION_NAME>. https://cloud.google.com/bigquery/docs/create-cloud-resource-connection
+            input:
+                input schema for general remote models
+            output:
+                output schema for general remote models
             options:
                 a dict of options to configure the model. Generates a BQML OPTIONS clause
 
@@ -311,6 +317,8 @@ def create_remote_model(
         sql = self._model_creation_sql_generator.create_remote_model(
             connection_name=connection_name,
             model_ref=model_ref,
+            input=input,
+            output=output,
             options=options,
         )
 
diff --git a/bigframes/ml/remote.py b/bigframes/ml/remote.py
@@ -0,0 +1,157 @@
+# Copyright 2023 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""BigFrames general remote models."""
+
+from __future__ import annotations
+
+from typing import Mapping, Optional, Union
+import warnings
+
+import bigframes
+from bigframes import clients
+from bigframes.core import log_adapter
+from bigframes.ml import base, core, globals, utils
+import bigframes.pandas as bpd
+
+_SUPPORTED_DTYPES = (
+    "bool",
+    "string",
+    "int64",
+    "float64",
+    "array<bool>",
+    "array<string>",
+    "array<int64>",
+    "array<float64>",
+)
+
+_REMOTE_MODEL_STATUS = "remote_model_status"
+
+
+@log_adapter.class_logger
+class VertexAIModel(base.BaseEstimator):
+    """Remote model from a Vertex AI https endpoint. User must specify https endpoint, input schema and output schema.
+    How to deploy a model in Vertex AI https://cloud.google.com/bigquery/docs/bigquery-ml-remote-model-tutorial#Deploy-Model-on-Vertex-AI.
+
+    Args:
+        endpoint (str):
+            Vertex AI https endpoint.
+        input ({column_name: column_type}):
+            Input schema. Supported types are "bool", "string", "int64", "float64", "array<bool>", "array<string>", "array<int64>", "array<float64>".
+        output ({column_name: column_type}):
+            Output label schema. Supported the same types as the input.
+        session (bigframes.Session or None):
+            BQ session to create the model. If None, use the global default session.
+        connection_name (str or None):
+            Connection to connect with remote service. str of the format <PROJECT_NUMBER/PROJECT_ID>.<LOCATION>.<CONNECTION_ID>.
+            if None, use default connection in session context. BigQuery DataFrame will try to create the connection and attach
+            permission if the connection isn't fully setup.
+    """
+
+    def __init__(
+        self,
+        endpoint: str,
+        input: Mapping[str, str],
+        output: Mapping[str, str],
+        session: Optional[bigframes.Session] = None,
+        connection_name: Optional[str] = None,
+    ):
+        self.endpoint = endpoint
+        self.input = input
+        self.output = output
+        self.session = session or bpd.get_global_session()
+
+        self._bq_connection_manager = clients.BqConnectionManager(
+            self.session.bqconnectionclient, self.session.resourcemanagerclient
+        )
+        connection_name = connection_name or self.session._bq_connection
+        self.connection_name = self._bq_connection_manager.resolve_full_connection_name(
+            connection_name,
+            default_project=self.session._project,
+            default_location=self.session._location,
+        )
+
+        self._bqml_model_factory = globals.bqml_model_factory()
+        self._bqml_model: core.BqmlModel = self._create_bqml_model()
+
+    def _create_bqml_model(self):
+        # Parse and create connection if needed.
+        if not self.connection_name:
+            raise ValueError(
+                "Must provide connection_name, either in constructor or through session options."
+            )
+        connection_name_parts = self.connection_name.split(".")
+        if len(connection_name_parts) != 3:
+            raise ValueError(
+                f"connection_name must be of the format <PROJECT_NUMBER/PROJECT_ID>.<LOCATION>.<CONNECTION_ID>, got {self.connection_name}."
+            )
+        self._bq_connection_manager.create_bq_connection(
+            project_id=connection_name_parts[0],
+            location=connection_name_parts[1],
+            connection_id=connection_name_parts[2],
+            iam_role="aiplatform.user",
+        )
+
+        options = {
+            "endpoint": self.endpoint,
+        }
+
+        def standardize_type(v: str):
+            v = v.lower()
+            v = v.replace("boolean", "bool")
+
+            if v not in _SUPPORTED_DTYPES:
+                raise ValueError(
+                    f"Data type {v} is not supported. We only support {', '.join(_SUPPORTED_DTYPES)}."
+                )
+
+            return v
+
+        self.input = {k: standardize_type(v) for k, v in self.input.items()}
+        self.output = {k: standardize_type(v) for k, v in self.output.items()}
+
+        return self._bqml_model_factory.create_remote_model(
+            session=self.session,
+            connection_name=self.connection_name,
+            input=self.input,
+            output=self.output,
+            options=options,
+        )
+
+    def predict(
+        self,
+        X: Union[bpd.DataFrame, bpd.Series],
+    ) -> bpd.DataFrame:
+        """Predict the result from the input DataFrame.
+
+        Args:
+            X (bigframes.dataframe.DataFrame or bigframes.series.Series):
+                Input DataFrame or Series, which needs to comply with the input parameter of the model.
+
+        Returns:
+            bigframes.dataframe.DataFrame: DataFrame of shape (n_samples, n_input_columns + n_prediction_columns). Returns predicted values.
+        """
+
+        (X,) = utils.convert_to_dataframe(X)
+
+        df = self._bqml_model.predict(X)
+
+        # unlike LLM models, the general remote model status is null for successful runs.
+        if (df[_REMOTE_MODEL_STATUS].notna()).any():
+            warnings.warn(
+                f"Some predictions failed. Check column {_REMOTE_MODEL_STATUS} for detailed status. You may want to filter the failed rows and retry.",
+                RuntimeWarning,
+            )
+
+        return df
diff --git a/bigframes/ml/sql.py b/bigframes/ml/sql.py
@@ -57,6 +57,12 @@ def build_expressions(self, *expr_sqls: str) -> str:
         indent_str = "  "
         return "\n" + indent_str + f",\n{indent_str}".join(expr_sqls)
 
+    def build_schema(self, **kwargs: str) -> str:
+        """Encode a dict of values into a formatted schema type items for SQL"""
+        indent_str = "  "
+        param_strs = [f"{k} {v}" for k, v in kwargs.items()]
+        return "\n" + indent_str + f",\n{indent_str}".join(param_strs)
+
     def options(self, **kwargs: Union[str, int, float, Iterable[str]]) -> str:
         """Encode the OPTIONS clause for BQML"""
         return f"OPTIONS({self.build_parameters(**kwargs)})"
@@ -65,6 +71,14 @@ def struct_options(self, **kwargs: Union[int, float]) -> str:
         """Encode a BQ STRUCT as options."""
         return f"STRUCT({self.build_structs(**kwargs)})"
 
+    def input(self, **kwargs: str) -> str:
+        """Encode a BQML INPUT clause."""
+        return f"INPUT({self.build_schema(**kwargs)})"
+
+    def output(self, **kwargs: str) -> str:
+        """Encode a BQML OUTPUT clause."""
+        return f"OUTPUT({self.build_schema(**kwargs)})"
+
     # Connection
     def connection(self, conn_name: str) -> str:
         """Encode the REMOTE WITH CONNECTION clause for BQML. conn_name is of the format <PROJECT_NUMBER/PROJECT_ID>.<REGION>.<CONNECTION_NAME>."""
@@ -154,15 +168,19 @@ def create_remote_model(
         self,
         connection_name: str,
         model_ref: google.cloud.bigquery.ModelReference,
+        input: Mapping[str, str] = {},
+        output: Mapping[str, str] = {},
         options: Mapping[str, Union[str, int, float, Iterable[str]]] = {},
     ) -> str:
         """Encode the CREATE OR REPLACE MODEL statement for BQML remote model."""
-        options_sql = self.options(**options)
-
         parts = [f"CREATE OR REPLACE MODEL {self._model_id_sql(model_ref)}"]
+        if input:
+            parts.append(self.input(**input))
+        if output:
+            parts.append(self.output(**output))
         parts.append(self.connection(connection_name))
-        if options_sql:
-            parts.append(options_sql)
+        if options:
+            parts.append(self.options(**options))
         return "\n".join(parts)
 
     def create_imported_model(
diff --git a/docs/reference/bigframes.ml/index.rst b/docs/reference/bigframes.ml/index.rst
@@ -30,3 +30,5 @@ API Reference
     pipeline
 
     preprocessing
+
+    remote
diff --git a/docs/reference/bigframes.ml/remote.rst b/docs/reference/bigframes.ml/remote.rst
@@ -0,0 +1,7 @@
+bigframes.ml.remote
+===================
+
+.. automodule:: bigframes.ml.remote
+    :members:
+    :inherited-members:
+    :undoc-members:
diff --git a/docs/templates/toc.yml b/docs/templates/toc.yml
@@ -108,6 +108,12 @@
       - name: PaLM2TextEmbeddingGenerator
         uid: bigframes.ml.llm.PaLM2TextEmbeddingGenerator
       name: llm
+    - items:
+      - name: Overview
+        uid: bigframes.ml.remote
+      - name: VertexAIModel
+        uid: bigframes.ml.remote.VertexAIModel
+      name: remote
     - items:
       - name: metrics
         uid: bigframes.ml.metrics
diff --git a/tests/system/small/ml/conftest.py b/tests/system/small/ml/conftest.py
@@ -29,6 +29,7 @@
     imported,
     linear_model,
     llm,
+    remote,
 )
 
 
@@ -247,6 +248,46 @@ def palm2_embedding_generator_multilingual_model(
     )
 
 
+@pytest.fixture(scope="session")
+def linear_remote_model_params() -> dict:
+    # Pre-deployed endpoint of linear reg model in Vertex.
+    # bigframes-test-linreg2 -> bigframes-test-linreg-endpoint2
+    return {
+        "input": {"culmen_length_mm": "float64"},
+        "output": {"predicted_body_mass_g": "array<float64>"},
+        "endpoint": "https://us-central1-aiplatform.googleapis.com/v1/projects/1084210331973/locations/us-central1/endpoints/3193318217619603456",
+    }
+
+
+@pytest.fixture(scope="session")
+def bqml_linear_remote_model(
+    session, bq_connection, linear_remote_model_params
+) -> core.BqmlModel:
+    options = {
+        "endpoint": linear_remote_model_params["endpoint"],
+    }
+    return globals.bqml_model_factory().create_remote_model(
+        session=session,
+        input=linear_remote_model_params["input"],
+        output=linear_remote_model_params["output"],
+        connection_name=bq_connection,
+        options=options,
+    )
+
+
+@pytest.fixture(scope="session")
+def linear_remote_vertex_model(
+    session, bq_connection, linear_remote_model_params
+) -> remote.VertexAIModel:
+    return remote.VertexAIModel(
+        endpoint=linear_remote_model_params["endpoint"],
+        input=linear_remote_model_params["input"],
+        output=linear_remote_model_params["output"],
+        session=session,
+        connection_name=bq_connection,
+    )
+
+
 @pytest.fixture(scope="session")
 def time_series_bqml_arima_plus_model(
     session, time_series_arima_plus_model_name
diff --git a/tests/system/small/ml/test_core.py b/tests/system/small/ml/test_core.py
@@ -289,6 +289,22 @@ def test_model_predict_with_unnamed_index(
     )
 
 
+def test_remote_model_predict(
+    bqml_linear_remote_model: core.BqmlModel, new_penguins_df
+):
+    predictions = bqml_linear_remote_model.predict(new_penguins_df).to_pandas()
+    expected = pd.DataFrame(
+        {"predicted_body_mass_g": [[3739.54], [3675.79], [3619.54]]},
+        index=pd.Index([1633, 1672, 1690], name="tag_number", dtype="Int64"),
+    )
+    pd.testing.assert_frame_equal(
+        predictions[["predicted_body_mass_g"]].sort_index(),
+        expected,
+        check_exact=False,
+        rtol=0.1,
+    )
+
+
 @pytest.mark.flaky(retries=2, delay=120)
 def test_model_generate_text(
     bqml_palm2_text_generator_model: core.BqmlModel, llm_text_df
diff --git a/tests/system/small/ml/test_remote.py b/tests/system/small/ml/test_remote.py
@@ -0,0 +1,33 @@
+# Copyright 2023 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import pandas as pd
+
+from bigframes.ml import remote
+
+
+def test_remote_linear_vertex_model_predict(
+    linear_remote_vertex_model: remote.VertexAIModel, new_penguins_df
+):
+    predictions = linear_remote_vertex_model.predict(new_penguins_df).to_pandas()
+    expected = pd.DataFrame(
+        {"predicted_body_mass_g": [[3739.54], [3675.79], [3619.54]]},
+        index=pd.Index([1633, 1672, 1690], name="tag_number", dtype="Int64"),
+    )
+    pd.testing.assert_frame_equal(
+        predictions[["predicted_body_mass_g"]].sort_index(),
+        expected,
+        check_exact=False,
+        rtol=0.1,
+    )
diff --git a/tests/unit/ml/test_sql.py b/tests/unit/ml/test_sql.py

Original file line number	Diff line number	Diff line change
`@@ -26,4 +26,5 @@`
`26`	`26`	`"llm",`
`27`	`27`	`"forecasting",`
`28`	`28`	`"imported",`
	`29`	`+ "remote",`
`29`	`30`	`]`