Skip to content

iflytek/spark-ai-go

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

讯飞星火大模型接入库 (spark-ai-go)

本Go SDK库帮助用户更快体验讯飞星火大模型

项目地址

前言

感谢开源的力量,希望讯飞开源越做越好,星火大模型效果越来越好!。

!img

本logo出自星火大模型

感谢社区(LangchainGo项目以及SparkLLM部分committer)[项目正在开发中]

近期规划新特性[待演进]

  • 极简的接入,快速调用讯飞星火大模型
  • LLM类统一接口,快速切换业界大模型
  • Python版本SDK进行中

如何使用

示例代码

  • 前置条件 需要在 xfyun.cn 申请有权限的

    • app_id
    • api_key
    • api_secret
  • URL/Domain配置请查看doc

  • 运行测试脚本需要提前将 .env.example 拷贝为 .env并配置其中变量

内如如下:

# spark 授权信息
SPARKAI_APP_ID=
SPARKAI_API_KEY=
SPARKAI_API_SECRET=
SPARKAI_DOMAIN=
SPARKAI_URL=wss:https://spark-api.xf-yun.com/v3.5/chat

一次性返回结果(非流式)

func TestLLMClientStream(t *testing.T) {
	SPARK_API_KEY := os.Getenv(ApiKeyEnvVarName)
	SPARK_API_SECRET := os.Getenv(ApiSecretEnvVarName)
	SPARK_API_BASE := os.Getenv(BaseURLEnvVarName)
	SPARK_APP_ID := os.Getenv(AppIdEnvVarName)
	SPARK_DOMAIN = os.Getenv(SparkDomainEnvVarName)
	_, client, err := spark.NewClient(spark.WithBaseURL(SPARK_API_BASE), spark.WithApiKey(SPARK_API_KEY), spark.WithApiSecret(SPARK_API_SECRET), spark.WithAppId(SPARK_APP_ID), spark.WithAPIDomain(SPARK_DOMAIN))
	if err != nil {
		fmt.Println(err.Error())
		return
	}
	ctx := context.Background()
	r := &sparkclient.ChatRequest{
		Domain: &SPARK_DOMAIN,
		Messages: []messages.ChatMessage{
			&messages.GenericChatMessage{
				Role:    "user",
				Content: "帮我润色并简化这句话: 国内科技大厂,创业公司都在演进大模型AI Agent且AI Agent框架仍然处于发展初期,采用开源化的AIAgent演进路线有助于快速构建影响力\n",
			},
		},
	}
    res, err := client.CreateChat(ctx, r)
    if err != nil {
        fmt.Print(err.Error())
        return
    }
    fmt.Print(res.GetContent())

}

流式返回结果

func TestLLMClientStream(t *testing.T) {
	SPARK_API_KEY := os.Getenv(ApiKeyEnvVarName)
	SPARK_API_SECRET := os.Getenv(ApiSecretEnvVarName)
	SPARK_API_BASE := os.Getenv(BaseURLEnvVarName)
	SPARK_APP_ID := os.Getenv(AppIdEnvVarName)
	SPARK_DOMAIN = os.Getenv(SparkDomainEnvVarName)
	_, client, err := spark.NewClient(spark.WithBaseURL(SPARK_API_BASE), spark.WithApiKey(SPARK_API_KEY), spark.WithApiSecret(SPARK_API_SECRET), spark.WithAppId(SPARK_APP_ID), spark.WithAPIDomain(SPARK_DOMAIN))
	if err != nil {
		fmt.Println(err.Error())
		return
	}
	ctx := context.Background()
	r := &sparkclient.ChatRequest{
		Domain: &SPARK_DOMAIN,
		Messages: []messages.ChatMessage{
			&messages.GenericChatMessage{
				Role:    "user",
				Content: "帮我润色并简化这句话: 国内科技大厂,创业公司都在演进大模型AI Agent且AI Agent框架仍然处于发展初期,采用开源化的AIAgent演进路线有助于快速构建影响力\n",
			},
		},
	}
	_, err = client.CreateChatWithCallBack(ctx, r, func(msg messages.ChatMessage) error {
		fmt.Print(msg.GetContent())
		return nil
	})
	if err != nil {
		fmt.Print(err.Error())
		return
	}

}

其中可以在CreateWithCallBack传入一个func用于回调处理流式数据

FunctionCall功能

func TestLLMFunctionCN(t *testing.T) {
	SPARK_API_KEY := os.Getenv(ApiKeyEnvVarName)
	SPARK_API_SECRET := os.Getenv(ApiSecretEnvVarName)
	SPARK_API_BASE := os.Getenv(BaseURLEnvVarName)
	SPARK_APP_ID := os.Getenv(AppIdEnvVarName)
	SPARK_DOMAIN = os.Getenv(SparkDomainEnvVarName)
	_, client, _ := NewClient(WithBaseURL(SPARK_API_BASE), WithApiKey(SPARK_API_KEY), WithApiSecret(SPARK_API_SECRET), WithAppId(SPARK_APP_ID), WithAPIDomain(SPARK_DOMAIN))
	ctx := context.Background()
	r := &sparkclient.ChatRequest{
		Domain: &SPARK_DOMAIN,
		Messages: []messages.ChatMessage{
			&messages.GenericChatMessage{
				Role:    "user",
				Content: " 帮我生成一份2023ATP报表放到 /workspace目录下 ",
			},
		},
		Functions: []messages.FunctionDefinition{
			{
				Name:        "generate_biz_report",
				Description: "生成运营分析报表工具",
				Parameters: map[string]any{
					"type": "object",
					"properties": map[string]any{
						"biz": map[string]any{
							"type":        "string",
							"description": "业务类型,取值范围为 ase, aipaas, atp",
						},
						"output_dir": map[string]any{
							"type":        "string",
							"description": "生成存放位置",
						},
						"type": map[string]any{
							"type":        "string",
							"description": "报表类型,取值范围是 txt, excel, markdown之一,默认为excel",
						},
						"start": map[string]any{
							"type":        "string",
							"description": "起始时间年份",
						},
						"end": map[string]any{
							"type":        "string",
							"description": "结束时间年份",
						},
					},
					"required": []string{
						"output_dir", "type", "start", "end",
					},
				},
			},
			{
				Name:        "translate_transcript",
				Description: "using translate_text function to translate the script",
				Parameters: map[string]any{
					"type": "object",
					"properties": map[string]any{
						"source_language": map[string]any{
							"type":        "string",
							"description": "source language",
						},
						"target_language": map[string]any{
							"type":        "string",
							"description": "target language",
						},
					},
					"required": []string{
						"source_language",
						"source_language",
					},
				},
			},
		},
	}
	_, err := client.CreateChatWithCallBack(ctx, r, func(msg messages.ChatMessage) error {
		fmt.Print(msg.GetContent())
		return nil
	})
	if err != nil {
		return
	}

}

欢迎贡献

扫码加入交流群

img

已知问题

  • 项目目前开发阶段,有一些冗余代码,人力有限,部分思想借鉴开源实现

URL

  • wss:https://spark-api.xf-yun.com/v3.5/chat

致谢

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published