增强检索知识库系统4

第9节: OpenAI模型对接

以下是openAI的以下帖子: openAI中文网 , openAI官网 ,

获取openAI key的方法:

【OpenAI】获取OpenAI API Key的多种方式全攻略：从入门到精通，再到详解教程！_openai api key获取-优快云博客

• 引入maven依赖

<dependency>
    <groupId>org.springframework.ai</groupId>
    <artifactId>spring-ai-openai-spring-boot-starter</artifactId>
</dependency>

• 配置yml文件

spring:
  ai:   
     openai:
      base-url: https://pro-share-aws-api.zcyai.com/
      api-key: *****
      embedding-model: text-embedding-ada-002
    rag:
      embed: nomic-embed-text #nomic-embed-text、text-embedding-ada-002

• 配置openAI接口

    Flux<ChatResponse> generateStreamRag(String model, String ragTag, String message);

• 配置openAI配置类

 @Bean
    public OpenAiApi openAiApi(@Value("${spring.ai.openai.base-url}") String baseUrl, @Value("${spring.ai.openai.api-key}") String apikey) {
        return new OpenAiApi(baseUrl, apikey);
    }

@Bean
    public SimpleVectorStore vectorStore(@Value("${spring.ai.rag.embed}") String model, OllamaApi ollamaApi, OpenAiApi openAiApi) {
        if ("nomic-embed-text".equalsIgnoreCase(model)) {
            OllamaEmbeddingClient embeddingClient = new OllamaEmbeddingClient(ollamaApi);
            embeddingClient.withDefaultOptions(OllamaOptions.create().withModel("nomic-embed-text"));
            return new SimpleVectorStore(embeddingClient);
        } else {
            OpenAiEmbeddingClient embeddingClient = new OpenAiEmbeddingClient(openAiApi);
            return new SimpleVectorStore(embeddingClient);
        }
    }

    @Bean
    public PgVectorStore pgVectorStore(@Value("${spring.ai.rag.embed}") String model, OllamaApi ollamaApi, OpenAiApi openAiApi, JdbcTemplate jdbcTemplate) {
        if ("nomic-embed-text".equalsIgnoreCase(model)) {
            OllamaEmbeddingClient embeddingClient = new OllamaEmbeddingClient(ollamaApi);
            embeddingClient.withDefaultOptions(OllamaOptions.create().withModel("nomic-embed-text"));
            return new PgVectorStore(jdbcTemplate, embeddingClient);
        } else {
            OpenAiEmbeddingClient embeddingClient = new OpenAiEmbeddingClient(openAiApi);
            return new PgVectorStore(jdbcTemplate, embeddingClient);
        }
    }

• UI对接

对话框

上传文件

上传知识Git

云服务器部署上线

• 我使用的openAI https://ai.nengyongai.cn/v1

这两个名字要相同

• DockerFile文件

# 基础镜像
FROM openjdk:17-jdk-slim

# 作者
MAINTAINER guslegend

# 配置
ENV PARAMS=""

# 时区
ENV TZ=PRC
RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone

# 添加应用
ADD target/ai-rag-app.jar /ai-rag-app.jar

ENTRYPOINT ["sh","-c","java -jar $JAVA_OPTS /ai-rag-app.jar $PARAMS"]

• build.sh

# 普通镜像构建，随系统版本构建 amd/arm
docker build -t guslegend/ai-rag-app:1.0 -f ./Dockerfile .

# 兼容 amd、arm 构建镜像
# docker buildx build --load --platform liunx/amd64,linux/arm64 -t guslegend/ai-rag-app:1.2 -f ./Dockerfile . --push

构建你的镜像 https://hub.docker.com/repository/create