Utilizzare Agent Platform Vector Search con RAG Engine

Questa pagina mostra come connettere il motore RAG a Vector Search di Gemini Enterprise Agent Platform.

Puoi anche seguire la procedura utilizzando il notebook RAG Engine con Agent Platform Vector Search.

Configurazione di Agent Platform Vector Search

Agent Platform Vector Search si basa sulla tecnologia Vector Search sviluppata da Google Research. Con Vector Search puoi utilizzare la stessa infrastruttura alla base di prodotti Google come la Ricerca Google, YouTube e Google Play.

Per l'integrazione con RAG Engine, è necessario un indice Vector Search vuoto.

Configura l'SDK Vertex AI

Per configurare l'SDK Vertex AI, consulta Configurazione.

Crea indice Vector Search

Per creare un indice Vector Search compatibile con il tuo corpus RAG, l'indice deve soddisfare i seguenti criteri:

  1. IndexUpdateMethod deve essere STREAM_UPDATE, consulta Crea indice di flusso.

  2. Il tipo di misurazione della distanza deve essere impostato esplicitamente su uno dei seguenti valori:

    • DOT_PRODUCT_DISTANCE
    • COSINE_DISTANCE

  3. La dimensione del vettore deve essere coerente con il modello di embedding che prevedi di utilizzare nel corpus RAG. Altri parametri possono essere ottimizzati in base alle tue scelte, che determinano se i parametri aggiuntivi possono essere ottimizzati.

Python

Per scoprire come installare o aggiornare l'SDK Vertex AI Python, consulta Installa l'SDK Vertex AI Python. Per saperne di più, consulta la documentazione di riferimento dell'API Python. 

def vector_search_create_streaming_index(
    project: str, location: str, display_name: str, gcs_uri: Optional[str] = None
) -> aiplatform.MatchingEngineIndex:
    """Create a vector search index.

    Args:
        project (str): Required. Project ID
        location (str): Required. The region name
        display_name (str): Required. The index display name
        gcs_uri (str): Optional. The Google Cloud Storage uri for index content

    Returns:
        The created MatchingEngineIndex.
    """
    # Initialize the Vertex AI client
    aiplatform.init(project=project, location=location)

    # Create Index
    index = aiplatform.MatchingEngineIndex.create_tree_ah_index(
        display_name=display_name,
        contents_delta_uri=gcs_uri,
        description="Matching Engine Index",
        dimensions=100,
        approximate_neighbors_count=150,
        leaf_node_embedding_count=500,
        leaf_nodes_to_search_percent=7,
        index_update_method="STREAM_UPDATE",  # Options: STREAM_UPDATE, BATCH_UPDATE
        distance_measure_type=aiplatform.matching_engine.matching_engine_index_config.DistanceMeasureType.DOT_PRODUCT_DISTANCE,
    )

    return index

Crea un endpoint indice Vector Search

Gli endpoint pubblici sono supportati dal motore RAG.

Python

Per scoprire come installare o aggiornare l'SDK Vertex AI Python, consulta Installa l'SDK Vertex AI Python. Per saperne di più, consulta la documentazione di riferimento dell'API Python. 

def vector_search_create_index_endpoint(
    project: str, location: str, display_name: str
) -> None:
    """Create a vector search index endpoint.

    Args:
        project (str): Required. Project ID
        location (str): Required. The region name
        display_name (str): Required. The index endpoint display name
    """
    # Initialize the Vertex AI client
    aiplatform.init(project=project, location=location)

    # Create Index Endpoint
    index_endpoint = aiplatform.MatchingEngineIndexEndpoint.create(
        display_name=display_name,
        public_endpoint_enabled=True,
        description="Matching Engine Index Endpoint",
    )

    print(index_endpoint.name)

Esegui il deployment di un indice su un endpoint indice

Prima di eseguire la ricerca del vicino più prossimo, è necessario eseguire il deployment dell'indice in un endpoint indice.

Python

Per scoprire come installare o aggiornare l'SDK Vertex AI Python, consulta Installa l'SDK Vertex AI Python. Per saperne di più, consulta la documentazione di riferimento dell'API Python. 

def vector_search_deploy_index(
    project: str,
    location: str,
    index_name: str,
    index_endpoint_name: str,
    deployed_index_id: str,
) -> None:
    """Deploy a vector search index to a vector search index endpoint.

    Args:
        project (str): Required. Project ID
        location (str): Required. The region name
        index_name (str): Required. The index to update. A fully-qualified index
          resource name or a index ID.  Example:
          "projects/123/locations/us-central1/indexes/my_index_id" or
          "my_index_id".
        index_endpoint_name (str): Required. Index endpoint to deploy the index
          to.
        deployed_index_id (str): Required. The user specified ID of the
          DeployedIndex.
    """
    # Initialize the Vertex AI client
    aiplatform.init(project=project, location=location)

    # Create the index instance from an existing index
    index = aiplatform.MatchingEngineIndex(index_name=index_name)

    # Create the index endpoint instance from an existing endpoint.
    index_endpoint = aiplatform.MatchingEngineIndexEndpoint(
        index_endpoint_name=index_endpoint_name
    )

    # Deploy Index to Endpoint
    index_endpoint = index_endpoint.deploy_index(
        index=index, deployed_index_id=deployed_index_id
    )

    print(index_endpoint.deployed_indexes)

Se è la prima volta che esegui il deployment di un indice in un endpoint indice, la creazione e l'avvio automatico del backend richiedono circa 30 minuti prima che l'indice possa essere archiviato. Dopo il primo deployment, l'indice è pronto in pochi secondi. Per visualizzare lo stato del deployment dell'indice, apri la console di Vector Search, seleziona la scheda Endpoint indice e scegli l'endpoint indice.

Identifica il nome della risorsa dell'indice e dell'endpoint indice, che hanno i seguenti formati:

  • projects/${PROJECT_NUMBER}/locations/${LOCATION_ID}/indexes/${INDEX_ID}
  • projects/${PROJECT_NUMBER}/locations/${LOCATION_ID}/indexEndpoints/${INDEX_ENDPOINT_ID}.

Utilizza Agent Platform Vector Search in RAG Engine

Dopo aver configurato l'istanza di Vector Search, segui i passaggi descritti in questa sezione per impostarla come database vettoriale per l'applicazione RAG.

Imposta il database vettoriale per creare un corpus RAG

Quando crei il corpus RAG, specifica solo i valori completi di INDEX_ENDPOINT_NAME e INDEX_NAME. Assicurati di utilizzare l'ID numerico per i nomi delle risorse sia dell'indice che dell'endpoint indice. Il corpus RAG viene creato e associato automaticamente all'indice Vector Search. Le convalide vengono eseguite in base ai criteri. Se uno dei requisiti non viene soddisfatto, la richiesta viene rifiutata.

Python

Prima di provare questo esempio, segui le istruzioni di configurazione di Python nella guida rapida di Agent Platform per l'utilizzo delle librerie client.

Per eseguire l'autenticazione in Agent Platform, configura le Credenziali predefinite dell'applicazione. Per saperne di più, consulta Configura l'autenticazione per un ambiente di sviluppo locale. 


import agentplatform
from agentplatform import types

# TODO(developer): Update and un-comment below lines
# PROJECT_ID = "your-project-id"
# vector_search_index_name = "projects/{PROJECT_ID}/locations/{LOCATION}/indexes/{INDEX_ID}"
# vector_search_index_endpoint_name = "projects/{PROJECT_ID}/locations/{LOCATION}/indexEndpoints/{INDEX_ENDPOINT_ID}"
# display_name = "test_corpus"
# description = "Corpus Description"

# Initialize Agent Platform client once per session.
# See https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/rag-overview#supported-regions
client = agentplatform.Client(project=PROJECT_ID, location="us-east5")

# Configure embedding model (Optional)
embedding_model_config = types.RagEmbeddingModelConfig(
    vertex_prediction_endpoint=types.RagEmbeddingModelConfigVertexPredictionEndpoint(
        endpoint="publishers/google/models/text-embedding-005"
    )
)

# Configure Vector DB
vector_db = types.RagVectorDbConfigVertexVectorSearch(
    index=vector_search_index_name,
    index_endpoint=vector_search_index_endpoint_name
)

corpus = client.rag.create_corpus(
    rag_corpus=types.RagCorpus(
        display_name=display_name,
        description=description,
        rag_vector_db_config=types.RagVectorDbConfig(
            rag_embedding_model_config=embedding_model_config,
            vertex_vector_search=vector_db,
        ),
    )
)
print(corpus)
# Example response:
# RagCorpus(name='projects/1234567890/locations/us-central1/ragCorpora/1234567890',
# display_name='test_corpus', description='Corpus Description', embedding_model_config=...
# ...

REST

  # TODO(developer): Update and un-comment the following lines:
  # CORPUS_DISPLAY_NAME = "YOUR_CORPUS_DISPLAY_NAME"
  # Full index/indexEndpoint resource name
  # Index: projects/${PROJECT_NUMBER}/locations/${LOCATION_ID}/indexes/${INDEX_ID}
  # IndexEndpoint: projects/${PROJECT_NUMBER}/locations/${LOCATION_ID}/indexEndpoints/${INDEX_ENDPOINT_ID}
  # INDEX_RESOURCE_NAME = "YOUR_INDEX_ENDPOINT_RESOURCE_NAME"
  # INDEX_NAME = "YOUR_INDEX_RESOURCE_NAME"
  # Call CreateRagCorpus API to create a new RagCorpus
  curl -X POST -H "Authorization: Bearer $(gcloud auth print-access-token)" -H "Content-Type: application/json" https://${LOCATION_ID}-aiplatform.googleapis.com/v1beta1/projects/${PROJECT_NUMBER}/locations/${LOCATION_ID}/ragCorpora -d '{
        "display_name" : '\""${CORPUS_DISPLAY_NAME}"\"',
        "rag_vector_db_config" : {
                "vertex_vector_search": {
                  "index":'\""${INDEX_NAME}"\"'
              "index_endpoint":'\""${INDEX_ENDPOINT_NAME}"\"'
                }
          }
    }'

  # Call ListRagCorpora API to verify the RagCorpus is created successfully
  curl -sS -X GET \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $(gcloud auth print-access-token)" \
  "https://${LOCATION_ID}-aiplatform.googleapis.com/v1beta1/projects/${PROJECT_NUMBER}/locations/${LOCATION_ID}/ragCorpora"

Importa file utilizzando l'API RAG

Utilizza il metodo API ragFiles.import per importare i file da Cloud Storage o Google Drive nell'indice di Vector Search. I file vengono incorporati e archiviati nell'indice di Vector Search.

REST

# TODO(developer): Update and uncomment the following lines:
# RAG_CORPUS_ID = "your-rag-corpus-id"
#
# Google Cloud Storage bucket/file location.
# For example, "gs://rag-fos-test/"
# GCS_URIS= "your-gcs-uris"

# Call ImportRagFiles API to embed files and store in the BigQuery table
curl -X POST \
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
-H "Content-Type: application/json" \
https://us-central1-aiplatform.googleapis.com/v1beta1/projects/${PROJECT_NUMBER}/locations/us-central1/ragCorpora/${RAG_CORPUS_ID}/ragFiles:import \
-d '{
  "import_rag_files_config": {
    "gcs_source": {
      "uris": '\""${GCS_URIS}"\"'
    },
    "rag_file_chunking_config": {
      "chunk_size": 512
    }
  }
}'

# Call ListRagFiles API to verify the files are imported successfully
curl -X GET \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
https://us-central1-aiplatform.googleapis.com/v1beta1/projects/${PROJECT_NUMBER}/locations/us-central1/ragCorpora/${RAG_CORPUS_ID}/ragFiles

Python

Per scoprire come installare o aggiornare l'SDK Vertex AI Python, consulta Installa l'SDK Vertex AI Python. Per saperne di più, consulta la documentazione di riferimento dell'API Python. 


import agentplatform
from agentplatform import types

from google.genai import types as genai_types

# TODO(developer): Update and un-comment below lines
# PROJECT_ID = "your-project-id"
# corpus_name = "projects/{PROJECT_ID}/locations/us-east5/ragCorpora/{rag_corpus_id}"
# paths = ["gs://my_bucket/my_files_dir/*"]

# Initialize Agent Platform client once per session.
# See https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/rag-overview#supported-regions
client = agentplatform.Client(project=PROJECT_ID, location="us-east5")

response = client.rag.import_files(
    name=corpus_name,
    import_config=types.ImportRagFilesConfig(
        gcs_source=genai_types.GcsSource(uris=paths),
        rag_file_transformation_config=types.RagFileTransformationConfig(
            rag_file_chunking_config=types.RagFileChunkingConfig(
                chunk_size=512,
                chunk_overlap=100,
            )
        ),  # optional
        max_embedding_requests_per_min=900,  # optional
    ),
)

print(f"Imported {response.imported_rag_files_count} files.")
# Example response:
# Imported 2 files.

Recupera i contesti pertinenti utilizzando l'API RAG

Al termine dell'importazione dei file, il contesto pertinente può essere recuperato dall'indice di Vector Search utilizzando l'API RetrieveContexts.

REST

# TODO(developer): Update and uncomment the following lines:
# RETRIEVAL_QUERY="your-retrieval-query"
#
# Full RAG corpus resource name
# Format:
# "projects/${PROJECT_NUMBER}/locations/us-central1/ragCorpora/${RAG_CORPUS_ID}"
# RAG_CORPUS_RESOURCE="your-rag-corpus-resource"

# Call RetrieveContexts API to retrieve relevant contexts
curl -X POST \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $(gcloud auth print-access-token)" \
https://us-central1-aiplatform.googleapis.com/v1beta1/projects/${PROJECT_NUMBER}/locations/us-central1:retrieveContexts \
  -d '{
    "vertex_rag_store": {
      "rag_resources": {
          "rag_corpus": '\""${RAG_CORPUS_RESOURCE}"\"',
        },
    },
    "query": {
      "text": '\""${RETRIEVAL_QUERY}"\"',
      "similarity_top_k": 10
    }
  }'

Python

Per scoprire come installare o aggiornare l'SDK Vertex AI Python, consulta Installa l'SDK Vertex AI Python. Per saperne di più, consulta la documentazione di riferimento dell'API Python. 


import agentplatform

from agentplatform import types
from google.genai import types as genai_types

# TODO(developer): Update and un-comment below lines
# PROJECT_ID = "your-project-id"
# corpus_name = "projects/[PROJECT_ID]/locations/us-east5/ragCorpora/[rag_corpus_id]"

# Initialize Agent Platform client once per session.
# See https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/rag-overview#supported-regions
client = agentplatform.Client(project=PROJECT_ID, location="us-east5")

response = client.rag.retrieve_contexts(
    vertex_rag_store=genai_types.VertexRagStore(
        rag_resources=[
            genai_types.VertexRagStoreRagResource(
                rag_corpus=corpus_name,
                # Optional: supply IDs from `rag.list_files()`.
                # rag_file_ids=["rag-file-1", "rag-file-2", ...],
            )
        ],
    ),
    query=types.RagQuery(
        text="Hello World!",
        rag_retrieval_config=genai_types.RagRetrievalConfig(
            top_k=10,
            filter=genai_types.RagRetrievalConfigFilter(
                vector_distance_threshold=0.5
            ),
        ),
    )
)
print(response)
# Example response:
# contexts {
#   contexts {
#     source_uri: "gs://your-bucket-name/file.txt"
#     text: "....
#   ....

Genera contenuti utilizzando l'API Gemini di Agent Platform

Per generare contenuti utilizzando i modelli Gemini, effettua una chiamata all'API GenerateContent di Agent Platform. Specificando RAG_CORPUS_RESOURCE nella richiesta, l'API recupera automaticamente i dati dall'indice Vector Search.

REST

# TODO(developer): Update and uncomment the following lines:
# MODEL_ID=gemini-2.5-flash
# GENERATE_CONTENT_PROMPT="your-generate-content-prompt"

# GenerateContent with contexts retrieved from the FeatureStoreOnline index
curl -X POST \
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
-H "Content-Type: application/json"  https://us-central1-aiplatform.googleapis.com/v1beta1/projects/${PROJECT_NUMBER}/locations/us-central1/publishers/google/models/${MODEL_ID}:generateContent \
-d '{
  "contents": {
    "role": "user",
    "parts": {
      "text": '\""${GENERATE_CONTENT_PROMPT}"\"'
    }
  },
  "tools": {
    "retrieval": {
      "vertex_rag_store": {
        "rag_resources": {
            "rag_corpus": '\""${RAG_CORPUS_RESOURCE}"\"',
          },
        "similarity_top_k": 8,
        "vector_distance_threshold": 0.32
      }
    }
  }
}'

Python

Per scoprire come installare o aggiornare l'SDK Vertex AI Python, consulta Installa l'SDK Vertex AI Python. Per saperne di più, consulta la documentazione di riferimento dell'API Python. 


from google import genai
from google.genai import types as genai_types

# TODO(developer): Update and un-comment below lines
# PROJECT_ID = "your-project-id"
# corpus_name = "projects/{PROJECT_ID}/locations/us-east5/ragCorpora/{rag_corpus_id}"

rag_retrieval_tool = genai_types.Tool(
    retrieval=genai_types.Retrieval(
        vertex_rag_store=genai_types.VertexRagStore(
            rag_resources=[
                genai_types.VertexRagStoreRagResource(
                    rag_corpus=corpus_name
                )
            ],
            rag_retrieval_config=genai_types.RagRetrievalConfig(
                top_k=10,
                filter=genai_types.RagRetrievalConfigFilter(
                    vector_distance_threshold=0.5
                ),
            ),
        ),
    )
)

# Create a GenAI SDK client to make a generate_content request
genai_client = genai.Client(vertexai=True, project=PROJECT_ID, location="global")

response = genai_client.models.generate_content(
    model="gemini-3.8-flash",
    contents="Why is the sky blue?",
    config=genai_types.GenerateContentConfig(
        tools=[rag_retrieval_tool]
    ),
)
print(response.text)
# Example response:
#   The sky appears blue due to a phenomenon called Rayleigh scattering.
#   Sunlight, which contains all colors of the rainbow, is scattered
#   by the tiny particles in the Earth's atmosphere....
#   ...

Passaggi successivi