feat: LLM - Support streaming prediction for code chat models

PiperOrigin-RevId: 558364254

Author: Ark-kun

tests/system/aiplatform/test_language_models.py

@@ -260,8 +260,19 @@ def test_code_generation_streaming(self):
 
         for response in model.predict_streaming(
             prefix="def reverse_string(s):",
-            suffix="    return s",
+            # code-bison does not support suffix
+            # suffix="    return s",
             max_output_tokens=128,
             temperature=0,
         ):
             assert response.text
+
+    def test_code_chat_model_send_message_streaming(self):
+        aiplatform.init(project=e2e_base._PROJECT, location=e2e_base._LOCATION)
+
+        chat_model = language_models.ChatModel.from_pretrained("codeodechat-bison@001")
+        chat = chat_model.start_chat()
+
+        message1 = "Please help write a function to calculate the max of two numbers"
+        for response in chat.send_message_streaming(message1):
+            assert response.text

tests/unit/aiplatform/test_language_models.py

@@ -1938,6 +1938,51 @@ def test_code_chat(self):
             assert prediction_parameters["temperature"] == message_temperature
             assert prediction_parameters["maxDecodeSteps"] == message_max_output_tokens
 
+    def test_code_chat_model_send_message_streaming(self):
+        """Tests the chat generation model."""
+        aiplatform.init(
+            project=_TEST_PROJECT,
+            location=_TEST_LOCATION,
+        )
+        with mock.patch.object(
+            target=model_garden_service_client.ModelGardenServiceClient,
+            attribute="get_publisher_model",
+            return_value=gca_publisher_model.PublisherModel(
+                _CODECHAT_BISON_PUBLISHER_MODEL_DICT
+            ),
+        ):
+            model = language_models.CodeChatModel.from_pretrained("codechat-bison@001")
+
+        chat = model.start_chat(temperature=0.0)
+
+        # Using list instead of a generator so that it can be reused.
+        response_generator = [
+            gca_prediction_service.StreamingPredictResponse(
+                outputs=[_streaming_prediction.value_to_tensor(response_dict)]
+            )
+            for response_dict in _TEST_CHAT_PREDICTION_STREAMING
+        ]
+
+        with mock.patch.object(
+            target=prediction_service_client.PredictionServiceClient,
+            attribute="server_streaming_predict",
+            return_value=response_generator,
+        ):
+            message_text1 = (
+                "Please help write a function to calculate the max of two numbers"
+            )
+            # New messages are not added until the response is fully read
+            assert not chat.message_history
+            for response in chat.send_message_streaming(message_text1):
+                assert len(response.text) > 10
+            # New messages are only added after the response is fully read
+            assert chat.message_history
+
+        assert len(chat.message_history) == 2
+        assert chat.message_history[0].author == chat.USER_AUTHOR
+        assert chat.message_history[0].content == message_text1
+        assert chat.message_history[1].author == chat.MODEL_AUTHOR
+
     def test_code_generation(self):
         """Tests code generation with the code generation model."""
         aiplatform.init(

vertexai/language_models/_language_models.py

@@ -59,13 +59,6 @@ def _get_model_id_from_tuning_model_id(tuning_model_id: str) -> str:
     return f"publishers/google/models/{model_name}@{version}"
 
 
-@dataclasses.dataclass
-class _PredictionRequest:
-    """A single-instance prediction request."""
-    instance: Dict[str, Any]
-    parameters: Optional[Dict[str, Any]] = None
-
-
 class _LanguageModel(_model_garden_models._ModelGardenModel):
     """_LanguageModel is a base class for all language models."""
 
@@ -1234,6 +1227,34 @@ def send_message(
             temperature=temperature,
         )
 
+    def send_message_streaming(
+        self,
+        message: str,
+        *,
+        max_output_tokens: Optional[int] = None,
+        temperature: Optional[float] = None,
+    ) -> Iterator[TextGenerationResponse]:
+        """Sends message to the language model and gets a streamed response.
+
+        The response is only added to the history once it's fully read.
+
+        Args:
+            message: Message to send to the model
+            max_output_tokens: Max length of the output text in tokens. Range: [1, 1024].
+                Uses the value specified when calling `ChatModel.start_chat` by default.
+            temperature: Controls the randomness of predictions. Range: [0, 1]. Default: 0.
+                Uses the value specified when calling `ChatModel.start_chat` by default.
+
+        Returns:
+            A stream of `TextGenerationResponse` objects that contain partial
+            responses produced by the model.
+        """
+        return super().send_message_streaming(
+            message=message,
+            max_output_tokens=max_output_tokens,
+            temperature=temperature,
+        )
+
 
 class CodeGenerationModel(_LanguageModel):
     """A language model that generates code.