feat: GenAI - Added support for system instructions

Usage: ``` model = generative_models.GenerativeModel( "gemini-1.0-pro", system_instruction=[ "Talk like a pirate.", "Don't use rude words.", ], ) ``` PiperOrigin-RevId: 621703355
googleapis · Apr 4, 2024 · 4990eb6 · 4990eb6
1 parent d0585e8
commit 4990eb6
Show file tree

Hide file tree

Showing 4 changed files with 41 additions and 2 deletions.
diff --git a/tests/system/vertexai/test_generative_models.py b/tests/system/vertexai/test_generative_models.py
@@ -121,7 +121,13 @@ async def test_generate_content_streaming_async(self):
             assert chunk.text
 
     def test_generate_content_with_parameters(self):
-        model = generative_models.GenerativeModel("gemini-pro")
+        model = generative_models.GenerativeModel(
+            "gemini-pro",
+            system_instruction=[
+                "Talk like a pirate.",
+                "Don't use rude words.",
+            ],
+        )
         response = model.generate_content(
             contents="Why is sky blue?",
             generation_config=generative_models.GenerationConfig(

diff --git a/tests/unit/vertexai/test_generative_models.py b/tests/unit/vertexai/test_generative_models.py
@@ -297,7 +297,14 @@ def test_generate_content(self, generative_models: generative_models):
         response = model.generate_content("Why is sky blue?")
         assert response.text
 
-        response2 = model.generate_content(
+        model2 = generative_models.GenerativeModel(
+            "gemini-pro",
+            system_instruction=[
+                "Talk like a pirate.",
+                "Don't use rude words.",
+            ],
+        )
+        response2 = model2.generate_content(
             "Why is sky blue?",
             generation_config=generative_models.GenerationConfig(
                 temperature=0.2,

diff --git a/vertexai/generative_models/README.md b/vertexai/generative_models/README.md
@@ -63,6 +63,19 @@ print(vision_chat.send_message(["I like this image.", image]))
 print(vision_chat.send_message("What things do I like?."))
 ```
 
+#### System instructions
+```
+from vertexai.generative_models import GenerativeModel
+model = GenerativeModel(
+    "gemini-1.0-pro",
+    system_instruction=[
+        "Talk like a pirate.",
+        "Don't use rude words.",
+    ],
+)
+print(model.generate_content("Why is sky blue?"))
+```
+
 #### Function calling
 
 ```

diff --git a/vertexai/generative_models/_generative_models.py b/vertexai/generative_models/_generative_models.py
@@ -133,6 +133,7 @@ def __init__(
         generation_config: Optional[GenerationConfigType] = None,
         safety_settings: Optional[SafetySettingsType] = None,
         tools: Optional[List["Tool"]] = None,
+        system_instruction: Optional[PartsType] = None,
     ):
         r"""Initializes GenerativeModel.
 
@@ -147,6 +148,9 @@ def __init__(
             generation_config: Default generation config to use in generate_content.
             safety_settings: Default safety settings to use in generate_content.
             tools: Default tools to use in generate_content.
+            system_instruction: Default system instruction to use in generate_content.
+                Note: Only text should be used in parts.
+                Content of each part will become a separate paragraph.
         """
         if "/" not in model_name:
             model_name = "publishers/google/models/" + model_name
@@ -163,13 +167,15 @@ def __init__(
         self._generation_config = generation_config
         self._safety_settings = safety_settings
         self._tools = tools
+        self._system_instruction = system_instruction
 
         # Validating the parameters
         self._prepare_request(
             contents="test",
             generation_config=generation_config,
             safety_settings=safety_settings,
             tools=tools,
+            system_instruction=system_instruction,
         )
 
     @property
@@ -205,6 +211,7 @@ def _prepare_request(
         generation_config: Optional[GenerationConfigType] = None,
         safety_settings: Optional[SafetySettingsType] = None,
         tools: Optional[List["Tool"]] = None,
+        system_instruction: Optional[PartsType] = None,
     ) -> gapic_prediction_service_types.GenerateContentRequest:
         """Prepares a GAPIC GenerateContentRequest."""
         if not contents:
@@ -213,6 +220,7 @@ def _prepare_request(
         generation_config = generation_config or self._generation_config
         safety_settings = safety_settings or self._safety_settings
         tools = tools or self._tools
+        system_instruction = system_instruction or self._system_instruction
 
         # contents can either be a list of Content objects (most generic case)
         if isinstance(contents, Sequence) and any(
@@ -244,6 +252,10 @@ def _prepare_request(
         else:
             contents = [_to_content(contents)]
 
+        gapic_system_instruction: Optional[gapic_content_types.Content] = None
+        if system_instruction:
+            gapic_system_instruction = _to_content(system_instruction)
+
         gapic_generation_config: Optional[gapic_content_types.GenerationConfig] = None
         if generation_config:
             if isinstance(generation_config, gapic_content_types.GenerationConfig):
@@ -307,6 +319,7 @@ def _prepare_request(
             generation_config=gapic_generation_config,
             safety_settings=gapic_safety_settings,
             tools=gapic_tools,
+            system_instruction=gapic_system_instruction,
         )
 
     def _parse_response(