samples(generative-ai): Update Grounding Samples and remove Dimensions from MM Embeddings for Video (GoogleCloudPlatform#11858)

holtskinner · m-strzelczyk · gcf-owl-bot[bot] · web-flow · commit 4fe6277fb12e · 2024-06-19T18:35:22.000+02:00
* samples(aiplatform): Update Grounding with Google Search sample to use GA endpoint * Add variable for data_store_path * Change Gemini 1.5 Pro to 1.5 Flash * `line` -> `lines` * fix: Remove Dimensions from Video MM Embeddings - Add to Image Embeddings so one example exists * Change test text for test_gemini_single_turn_video_example * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md --------- Co-authored-by: Maciej Strzelczyk <strzelczyk@google.com> Co-authored-by: Owl Bot <gcf-owl-bot[bot]@users.noreply.github.com>
diff --git a/generative_ai/gemini_grounding_example.py b/generative_ai/gemini_grounding_example.py
@@ -13,22 +13,26 @@
 # limitations under the License.
 
 
-from vertexai.preview.generative_models import GenerationResponse
+from vertexai.generative_models import GenerationResponse
 
 
 def generate_text_with_grounding_web(project_id: str) -> GenerationResponse:
     # [START generativeaionvertexai_gemini_grounding_with_web]
     import vertexai
 
-    from vertexai.preview.generative_models import grounding
-    from vertexai.generative_models import GenerationConfig, GenerativeModel, Tool
+    from vertexai.generative_models import (
+        GenerationConfig,
+        GenerativeModel,
+        Tool,
+        grounding,
+    )
 
     # TODO(developer): Update and un-comment below line
     # project_id = "PROJECT_ID"
 
     vertexai.init(project=project_id, location="us-central1")
 
-    model = GenerativeModel(model_name="gemini-1.0-pro-002")
+    model = GenerativeModel(model_name="gemini-1.5-flash-001")
 
     # Use Google Search for grounding
     tool = Tool.from_google_search_retrieval(grounding.GoogleSearchRetrieval())
@@ -57,15 +61,14 @@ def generate_text_with_grounding_vertex_ai_search(
     from vertexai.preview.generative_models import grounding
     from vertexai.generative_models import GenerationConfig, GenerativeModel, Tool
 
-    # TODO(developer): Update and un-comment below line
+    # TODO(developer): Update and un-comment below lines
     # project_id = "PROJECT_ID"
+    # data_store_path = "projects/{project_id}/locations/{location}/collections/default_collection/dataStores/{data_store_id}"
 
     vertexai.init(project=project_id, location="us-central1")
 
-    model = GenerativeModel(model_name="gemini-1.0-pro-002")
+    model = GenerativeModel(model_name="gemini-1.5-flash-001")
 
-    # Use Vertex AI Search data store
-    # Format: projects/{project_id}/locations/{location}/collections/default_collection/dataStores/{data_store_id}
     tool = Tool.from_retrieval(
         grounding.Retrieval(grounding.VertexAISearch(datastore=data_store_path))
     )
diff --git a/generative_ai/multimodal_embedding_image.py b/generative_ai/multimodal_embedding_image.py
@@ -21,13 +21,16 @@ def get_image_embeddings(
     project_id: str,
     image_path: str,
     contextual_text: Optional[str] = None,
+    dimension: Optional[int] = 1408,
 ) -> MultiModalEmbeddingResponse:
     """Example of how to generate multimodal embeddings from image and text.
 
     Args:
         project_id: Google Cloud Project ID, used to initialize vertexai
         image_path: Path to image (local or Google Cloud Storage) to generate embeddings for.
         contextual_text: Text to generate embeddings for.
+        dimension: Dimension for the returned embeddings.
+            https://cloud.google.com/vertex-ai/docs/generative-ai/embeddings/get-multimodal-embeddings#low-dimension
     """
     # [START aiplatform_sdk_multimodal_embedding_image]
     import vertexai
@@ -42,6 +45,7 @@ def get_image_embeddings(
     embeddings = model.get_embeddings(
         image=image,
         contextual_text=contextual_text,
+        dimension=dimension,
     )
     print(f"Image Embedding: {embeddings.image_embedding}")
     print(f"Text Embedding: {embeddings.text_embedding}")
diff --git a/generative_ai/multimodal_embedding_image_video_text.py b/generative_ai/multimodal_embedding_image_video_text.py
@@ -22,7 +22,6 @@ def get_image_video_text_embeddings(
     image_path: str,
     video_path: str,
     contextual_text: Optional[str] = None,
-    dimension: Optional[int] = 1408,
     video_segment_config: Optional[VideoSegmentConfig] = None,
 ) -> MultiModalEmbeddingResponse:
     """Example of how to generate multimodal embeddings from image, video, and text.
@@ -33,8 +32,6 @@ def get_image_video_text_embeddings(
         image_path: Path to image (local or Google Cloud Storage) to generate embeddings for.
         video_path: Path to video (local or Google Cloud Storage) to generate embeddings for.
         contextual_text: Text to generate embeddings for.
-        dimension: Dimension for the returned embeddings.
-            https://cloud.google.com/vertex-ai/docs/generative-ai/embeddings/get-multimodal-embeddings#low-dimension
         video_segment_config: Define specific segments to generate embeddings for.
             https://cloud.google.com/vertex-ai/docs/generative-ai/embeddings/get-multimodal-embeddings#video-best-practices
     """
@@ -56,7 +53,6 @@ def get_image_video_text_embeddings(
         video=video,
         video_segment_config=video_segment_config,
         contextual_text=contextual_text,
-        dimension=dimension,
     )
 
     print(f"Image Embedding: {embeddings.image_embedding}")
diff --git a/generative_ai/multimodal_embedding_video.py b/generative_ai/multimodal_embedding_video.py
@@ -21,7 +21,6 @@ def get_video_embeddings(
     project_id: str,
     video_path: str,
     contextual_text: Optional[str] = None,
-    dimension: Optional[int] = 1408,
     video_segment_config: Optional[VideoSegmentConfig] = None,
 ) -> MultiModalEmbeddingResponse:
     """Example of how to generate multimodal embeddings from video and text.
@@ -31,8 +30,6 @@ def get_video_embeddings(
         location: Google Cloud Region, used to initialize vertexai
         video_path: Path to video (local or Google Cloud Storage) to generate embeddings for.
         contextual_text: Text to generate embeddings for.
-        dimension: Dimension for the returned embeddings.
-            https://cloud.google.com/vertex-ai/docs/generative-ai/embeddings/get-multimodal-embeddings#low-dimension
         video_segment_config: Define specific segments to generate embeddings for.
             https://cloud.google.com/vertex-ai/docs/generative-ai/embeddings/get-multimodal-embeddings#video-best-practices
     """
@@ -52,7 +49,6 @@ def get_video_embeddings(
         video=video,
         video_segment_config=video_segment_config,
         contextual_text=contextual_text,
-        dimension=dimension,
     )
 
     # Video Embeddings are segmented based on the video_segment_config.
diff --git a/generative_ai/test_gemini_examples.py b/generative_ai/test_gemini_examples.py
@@ -102,7 +102,9 @@ def test_gemini_single_turn_video_example() -> None:
     text = gemini_single_turn_video_example.generate_text(PROJECT_ID)
     text = text.lower()
     assert len(text) > 0
-    assert any([_ in text for _ in ("zoo", "tiger", "leaf", "water")])
+    assert any(
+        [_ in text for _ in ("zoo", "tiger", "leaf", "water", "animals", "photos")]
+    )
 
 
 @pytest.mark.skip(
