共用方式為


資料來源 - Azure AI 搜尋服務

在您的資料上使用 Azure OpenAI 時,Azure AI 搜尋服務的可設定選項。 API 第 2024-02-01 版支援此資料來源。

名稱 類型​​ 必要 描述
parameters 參數 True 設定 Azure 搜尋服務時要使用的參數。
type 字串 True 必須是 azure_search

參數

姓名 類型​​ 必要 描述
endpoint string True Azure 搜尋服務資源所要使用的絕對端點路徑。
index_name 字串 True 要用於參考 Azure 搜尋服務資源中的索引名稱。
authentication 其中一個 ApiKeyAuthenticationOptionsSystemAssignedManagedIdentityAuthenticationOptionsUserAssignedManagedIdentityAuthenticationOptionsonYourDataAccessTokenAuthenticationOptions True 存取已定義資料來源時要使用的驗證方法。
embedding_dependency 其中一個 DeploymentNameVectorizationSourceEndpointVectorizationSource False 向量搜尋的內嵌相依性。 當 query_typevectorvector_simple_hybridvector_semantic_hybrid 時為必填。
fields_mapping FieldsMappingOptions False 與搜尋索引互動時要使用的自訂欄位對應行為。
filter 字串 False 搜尋篩選。
in_scope boolean False 查詢是否應限制為使用已編製索引的資料。 預設值為 True
query_type QueryType False 要與 Azure 搜尋服務搭配使用的查詢類型。 預設為 simple
role_information 字串 False 提供模型關於其行為方式的指示,以及產生回應時應該參考的任何內容。 您可以描述助理的特質,並對其告知如何格式化回應。
semantic_configuration 字串 False 查詢的語意設定。 當 query_typesemanticvector_semantic_hybrid 時為必填。
strictness 整數 False 已設定的搜尋相關性篩選嚴格度。 嚴格度越高,精確度就越高,但答案的召回率越低。 預設值為 3
top_n_documents 整數 False 要為設定的查詢顯示的已設定文件數目上限。 預設值為 5
max_search_queries 整數 False 針對一個使用者訊息應傳送給搜尋提供者的重寫查詢數目上限。 若未指定,系統會決定要傳送的查詢數目。
allow_partial_result 整數 False 如果指定為 true,系統將允許使用部分搜尋結果,要求將在所有查詢都失敗時失敗。 若未指定或指定為 false,要求將隨著任何搜尋查詢失敗而失敗。
include_contexts 陣列 False 輸出內容包含的屬性。 若未指定,則預設值為 citationsintent。 值可以是 citationsintentall_retrieved_documents

API 金鑰驗證選項

Azure OpenAI On Your Data 在使用 API 金鑰時的驗證選項。

名稱 類型​​ 必要 描述
key string True 要用於驗證的 API 金鑰。
type 字串 True 必須是 api_key

系統指派的受控身分識別驗證選項

Azure OpenAI On Your Data 在使用系統指派的受控識別時的驗證選項。

名稱 類型​​ 必要 描述
type string True 必須是 system_assigned_managed_identity

使用者指派的受控身分識別驗證選項

Azure OpenAI On Your Data 在使用使用者指派的受控識別時的驗證選項。

名稱 類型​​ 必要 描述
managed_identity_resource_id string True 要用於驗證的使用者指派受控識別的資源識別碼。
type 字串 True 必須是 user_assigned_managed_identity

存取權杖驗證選項

Azure OpenAI On Your Data 在使用存取權杖時的驗證選項。

名稱 類型​​ 必要 描述
access_token string True 要用於驗證的存取權杖。
type 字串 True 必須是 access_token

部署名稱向量化來源

套用向量搜尋時,資料上的 Azure OpenAI 所使用的向量化來源詳細資料。 此向量化來源是以相同 Azure OpenAI 資源中的內部內嵌模型部署名稱為基礎。 此向量化來源可讓您在沒有 Azure OpenAI API 金鑰的情況下使用向量搜尋,而不需要 Azure OpenAI 公用網路存取。

名稱 類型​​ 必要 描述
deployment_name string True 相同 Azure OpenAI 資源內的內嵌模型部署名稱。
type 字串 True 必須是 deployment_name
dimensions 整數 False 內嵌應有的維度數目。 僅支援 text-embedding-3 和更新模型。

端點向量化來源

套用向量搜尋時,資料上的 Azure OpenAI 所使用的向量化來源詳細資料。 此向量化來源是以 Azure OpenAI 內嵌 API 端點為基礎。

名稱 類型​​ 必要 描述
endpoint string True 指定應從中擷取內嵌的資源端點 URL。 其格式應為 https://{YOUR_RESOURCE_NAME}.openai.azure.com/openai/deployments/YOUR_DEPLOYMENT_NAME/embeddings。 不允許 api-version 查詢參數。
authentication ApiKeyAuthenticationOptions True 指定從指定端點擷取內嵌時要使用的驗證選項。
type 字串 True 必須是 endpoint
dimensions 整數 False 內嵌應有的維度數目。 僅支援 text-embedding-3 和更新模型。

欄位對應選項

選擇性設定,可控制使用已設定的 Azure 搜尋服務資源時處理欄位的方式。

名稱 類型​​ 必要 描述
content_fields string[] False 應視為內容的索引欄位名稱。
vector_fields string[] False 代表向量資料的欄位名稱。
content_fields_separator 字串 False 內容欄位應使用的分隔符號模式。 預設值為 \n
filepath_field 字串 False 要用作檔案路徑的索引欄位名稱。
title_field 字串 False 要用作標題的索引欄位名稱。
url_field 字串 False 要用作 URL 的索引欄位名稱。

查詢類型

Azure 搜尋服務擷取查詢在用作資料上的 Azure OpenAI 時所應執行的類型。

列舉值 描述
simple 代表預設的簡單查詢剖析器。
semantic 代表用於進階語意建模的語意查詢剖析器。
vector 代表對計算資料的向量搜尋。
vector_simple_hybrid 代表簡單查詢策略與向量資料的組合。
vector_semantic_hybrid 代表語意搜尋與向量資料查詢的組合。

範例

先決條件:

  • 設定從 Azure OpenAI 系統指派的受控識別給 Azure 搜尋服務的角色指派。 必要角色:Search Index Data ReaderSearch Service Contributor
  • 設定從使用者到 Azure OpenAI 資源的角色指派。 必要角色:Cognitive Services OpenAI User
  • 安裝 Az CLI,然後執行 az login
  • 定義下列環境變數:AzureOpenAIEndpointChatCompletionsDeploymentNameSearchEndpointSearchIndex
export AzureOpenAIEndpoint=https://example.openai.azure.com/
export ChatCompletionsDeploymentName=turbo
export SearchEndpoint=https://example.search.windows.net
export SearchIndex=example-index

安裝最新的 pip 套件 openaiazure-identity

import os
from openai import AzureOpenAI
from azure.identity import DefaultAzureCredential, get_bearer_token_provider

endpoint = os.environ.get("AzureOpenAIEndpoint")
deployment = os.environ.get("ChatCompletionsDeploymentName")
search_endpoint = os.environ.get("SearchEndpoint")
search_index = os.environ.get("SearchIndex")

token_provider = get_bearer_token_provider(DefaultAzureCredential(), "https://cognitiveservices.azure.com/.default")

client = AzureOpenAI(
    azure_endpoint=endpoint,
    azure_ad_token_provider=token_provider,
    api_version="2024-02-01",
)

completion = client.chat.completions.create(
    model=deployment,
    messages=[
        {
            "role": "user",
            "content": "Who is DRI?",
        },
    ],
    extra_body={
        "data_sources": [
            {
                "type": "azure_search",
                "parameters": {
                    "endpoint": search_endpoint,
                    "index_name": search_index,
                    "authentication": {
                        "type": "system_assigned_managed_identity"
                    }
                }
            }
        ]
    }
)

print(completion.model_dump_json(indent=2))