Microsoft MVP성태의 닷넷 이야기
.NET Framework: 2124. C# - Semantic Kernel의 Planner 사용 예제 [링크 복사], [링크+제목 복사],
조회: 11230
글쓴 사람
정성태 (techsharer at outlook.com)
홈페이지
첨부 파일

(시리즈 글이 8개 있습니다.)
.NET Framework: 2117. C# - (OpenAI 기반의) Microsoft Semantic Kernel을 이용한 자연어 처리
; https://www.sysnet.pe.kr/2/0/13345

.NET Framework: 2118. C# - Semantic Kernel의 Prompt chaining 예제
; https://www.sysnet.pe.kr/2/0/13347

.NET Framework: 2119. C# - Semantic Kernel의 "Basic Loading of the Kernel" 예제
; https://www.sysnet.pe.kr/2/0/13348

.NET Framework: 2120. C# - Semantic Kernel의 Skill과 Function 사용 예제
; https://www.sysnet.pe.kr/2/0/13349

.NET Framework: 2121. C# - Semantic Kernel의 대화 문맥 유지
; https://www.sysnet.pe.kr/2/0/13352

.NET Framework: 2123. C# - Semantic Kernel의 ChatGPT 대화 구현
; https://www.sysnet.pe.kr/2/0/13355

.NET Framework: 2124. C# - Semantic Kernel의 Planner 사용 예제
; https://www.sysnet.pe.kr/2/0/13357

.NET Framework: 2125. C# - Semantic Kernel의 Semantic Memory 사용 예제
; https://www.sysnet.pe.kr/2/0/13358




C# - Semantic Kernel의 Planner 사용 예제

Semantic Kernel의 구성 요소는 다음의 그림에서 잘 확인할 수 있습니다.

[출처: https://devblogs.microsoft.com/semantic-kernel/hello-world/]
sk_planner_1.png

ASK는 Kernel에 보내는 요청이고, Kernel은 이미 그동안 우리가 만든 코드의 시작에 해당하는 타입이었습니다.

var kernel = Kernel.Builder.Build(); // C# - Semantic Kernel의 "Basic Loading of the Kernel" 예제

그림에서 S는 Skill인데 이것도 "C# - Semantic Kernel의 Skill과 Function 사용 예제" 글을 통해 이미 다뤘습니다.

나머지는 M(Memory), C(Connector)와 오늘의 주제인 Planner인데요, 이에 대한 예제를 다음의 글에서 다루고 있습니다.

Introduction to the Planner
; https://github.com/microsoft/semantic-kernel/blob/main/samples/notebooks/dotnet/05-using-the-planner.ipynb

코드는 기존 예제와 동일하게 Kernel을 준비하는 것으로 시작하고,

KernelConfig kernelConfig = new KernelConfig();
kernelConfig.AddOpenAITextCompletionService("default", "text-davinci-003", apiKey);

var kernel = Kernel.Builder
    .WithConfiguration(kernelConfig)
    .Build();

그다음 planner를 만들고 Kernel에 Skill/Function을 담고 있습니다.

var planner = new SequentialPlanner(kernel);

var skillsDirectory = Path.Combine(Directory.GetCurrentDirectory(), "skills");
kernel.ImportSemanticSkillFromDirectory(skillsDirectory, "SummarizeSkill");
kernel.ImportSemanticSkillFromDirectory(skillsDirectory, "WriterSkill");

/*
SummarizeSkill
    Function: MakeAbstractReadable, Notegen, Summarize, Topics

WriterSkill
    Function: Acronym, AcronymGenerator, AcronymReverse, Brainstorm, EmailGen, EmailTo, 
              EnglishImprover, NovelChapter, NovelChapterWithNotes, NovelOutline, Rewrite, 
              ShortPoem, StoryGen, TellMeMore, Translate, TwoSentenceSummary
*/

해당 스킬들은 github에 예제로 제공되고 있는 것들인데, clone 시킨 디렉터리에서 복사할 수 있습니다. 프로젝트에 파일로 추가하는 경우 각각 CopyToOutputDirectory 설정을 해야 하는데, 일일이 파일 하나씩 하는 것은 번잡하므로 와일드카드 문자열을 사용해 "skills\**\*"로 지정하면 skills 디렉터리 하위에 있는 모든 디렉터리 및 파일들을 복사하게 됩니다. 아래는 csproj에 그 설정을 추가한 것입니다.

<Project Sdk="Microsoft.NET.Sdk">

  <PropertyGroup>
    <OutputType>Exe</OutputType>
    <TargetFramework>net7.0</TargetFramework>
    <ImplicitUsings>enable</ImplicitUsings>
    <Nullable>enable</Nullable>
  </PropertyGroup>

  <ItemGroup>
    <PackageReference Include="Microsoft.SemanticKernel" Version="0.13.277.1-preview" />
  </ItemGroup>

  <ItemGroup>
    <None Update="skills\**\*">
      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
    </None>
  </ItemGroup>

</Project>

그다음 코드에서는 planner를 이용해 사용자의 요청에 기반한 Planner를 생성하는데요,

var ask = "Tomorrow is Valentine's day. I need to come up with a few date ideas and e-mail them to my significant other.";
var originalPlan = await planner.CreatePlanAsync(ask);

Console.WriteLine("Original plan:\n");
Console.WriteLine(JsonSerializer.Serialize(originalPlan, new JsonSerializerOptions { WriteIndented = true }));

재미있는 건, 어떤 스킬을 사용하라고 지정하지도 않았다는 점입니다. 즉, 단순히 현재 Kernel에 포함된 Skill/Function을 이용해 사용자의 ASK를 만족하는 Planner를 구성하는데요, 아래는 originalPlan의 내부 구조를 JSON 포맷으로 보여줍니다.

Original plan:

{
  "state": [
    {
      "Key": "INPUT",
      "Value": ""
    }
  ],
  "steps": [
    {
      "state": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "steps": [],
      "named_parameters": [
        {
          "Key": "INPUT",
          "Value": "Valentine\u0027s day date ideas"
        }
      ],
      "named_outputs": [
        {
          "Key": "DATE_IDEAS",
          "Value": ""
        },
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "next_step_index": 0,
      "name": "Brainstorm",
      "skill_name": "WriterSkill",
      "description": "Given a goal or topic description generate a list of ideas"
    },
    {
      "state": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "steps": [],
      "named_parameters": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "named_outputs": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "next_step_index": 0,
      "name": "",
      "skill_name": "Microsoft.SemanticKernel.Orchestration.Plan",
      "description": ""
    },
    {
      "state": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "steps": [],
      "named_parameters": [
        {
          "Key": "sender",
          "Value": "Me"
        },
        {
          "Key": "to",
          "Value": "My Significant Other"
        },
        {
          "Key": "INPUT",
          "Value": "$DATE_IDEA_1;$DATE_IDEA_2;$DATE_IDEA_3"
        }
      ],
      "named_outputs": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "next_step_index": 0,
      "name": "EmailTo",
      "skill_name": "WriterSkill",
      "description": "Turn bullet points into an email to someone, using a polite tone"
    }
  ],
  "named_parameters": [
    {
      "Key": "INPUT",
      "Value": ""
    }
  ],
  "named_outputs": [
    {
      "Key": "INPUT",
      "Value": ""
    }
  ],
  "next_step_index": 0,
  "name": "Tomorrow is Valentine\u0027s day. I need to come up with a few date ideas and e-mail them to my significant other.",
  "skill_name": "Microsoft.SemanticKernel.Orchestration.Plan",
  "description": "Tomorrow is Valentine\u0027s day. I need to come up with a few date ideas and e-mail them to my significant other."
}

어찌 보면, 저것 하나가 새롭게 "Skill"이 된 것입니다. 결국 이러한 Planner대로 실행을 하면,

var originalPlanResult = await kernel.RunAsync(originalPlan);

Console.WriteLine("Original Plan results:\n");
Console.WriteLine(Utils.WordWrap(originalPlanResult.Result, 80));

다음과 같은 출력을 볼 수 있습니다.

Original Plan results:

Dear My Significant Other,

I wanted to take a moment to thank you for being my
significant other. You have been a source of joy and comfort in my life, and I
am so grateful for your presence. I am so lucky to have you in my life.

Thanks,
Me

그러니까, 사용자의 "Tomorrow is Valentine's day. I need to come up with a few date ideas and e-mail them to my significant other." 입력을 바탕으로, 내부에 구성한 Skill/Function을 활용해 위의 출력을 만들어 낸 것입니다.




그럼, 이 상태에서 Shakespeare 화법으로 말하라는 Skill/Function도 (위에서 만든 Kernel에) 추가해 볼까요? ^^

string skPrompt = """
{{$input}}

Rewrite the above in the style of Shakespeare.
""";
var shakespeareFunction = kernel.CreateSemanticFunction(skPrompt, "shakespeare", "ShakespeareSkill", maxTokens: 2000, temperature: 0.2, topP: 0.5);

이번 Skill/Function은 ImportSemanticSkillFromDirectory가 아닌 동적으로 CreateSemanticFunction을 이용해 추가했습니다. 이러한 변화를 반영해 Planner를 다시 생성하고,

var ask = @"Tomorrow is Valentine's day. I need to come up with a few date ideas.
She likes Shakespeare so write using his style. E-mail these ideas to my significant other";
var newPlan = await planner.CreatePlanAsync(ask);

Planner의 내부 변화를 확인합니다.

Console.WriteLine("Updated plan:\n");
Console.WriteLine(JsonSerializer.Serialize(newPlan, new JsonSerializerOptions { WriteIndented = true }));

Updated plan:

{
  "state": [
    {
      "Key": "INPUT",
      "Value": ""
    }
  ],
  "steps": [
    {
      "state": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "steps": [],
      "named_parameters": [
        {
          "Key": "INPUT",
          "Value": "Date ideas for Valentine\u0027s day"
        }
      ],
      "named_outputs": [
        {
          "Key": "INPUT",
          "Value": ""
        },
        {
          "Key": "DATE_IDEAS",
          "Value": ""
        }
      ],
      "next_step_index": 0,
      "name": "Brainstorm",
      "skill_name": "WriterSkill",
      "description": "Given a goal or topic description generate a list of ideas"
    },
    {
      "state": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "steps": [],
      "named_parameters": [
        {
          "Key": "INPUT",
          "Value": "$DATE_IDEAS"
        }
      ],
      "named_outputs": [
        {
          "Key": "INPUT",
          "Value": ""
        },
        {
          "Key": "SHAKESPEARE_IDEAS",
          "Value": ""
        }
      ],
      "next_step_index": 0,
      "name": "shakespeare",
      "skill_name": "ShakespeareSkill",
      "description": "Generic function, unknown purpose"
    },
    {
      "state": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "steps": [],
      "named_parameters": [
        {
          "Key": "sender",
          "Value": "Myself"
        },
        {
          "Key": "INPUT",
          "Value": "$SHAKESPEARE_IDEAS"
        },
        {
          "Key": "to",
          "Value": "My Significant Other"
        }
      ],
      "named_outputs": [
        {
          "Key": "INPUT",
          "Value": ""
        }
      ],
      "next_step_index": 0,
      "name": "EmailTo",
      "skill_name": "WriterSkill",
      "description": "Turn bullet points into an email to someone, using a polite tone"
    }
  ],
  "named_parameters": [
    {
      "Key": "INPUT",
      "Value": ""
    }
  ],
  "named_outputs": [
    {
      "Key": "INPUT",
      "Value": ""
    }
  ],
  "next_step_index": 0,
  "name": "Tomorrow is Valentine\u0027s day. I need to come up with a few date ideas.\r\nShe likes Shakespeare so write using his style. E-mail these ideas to my significant other",
  "skill_name": "Microsoft.SemanticKernel.Orchestration.Plan",
  "description": "Tomorrow is Valentine\u0027s day. I need to come up with a few date ideas.\r\nShe likes Shakespeare so write using his style. E-mail these ideas to my significant other"
}

이에 따라 텍스트를 생성하면,

var newPlanResult = await kernel.RunAsync(newPlan);
Console.WriteLine("New Plan results:\n");
Console.WriteLine(newPlanResult.Result);

이런 출력을 확인할 수 있습니다.

New Plan results:

Dear My Significant Other,

I hope you're doing well. I wanted to suggest some activities that we could do together.

We could make a delicious feast together. We could take a romantic walk. We could watch a romantic movie. We could attend a local event. We could take a dancing class. We could visit a museum. We could go to a spa. We could go stargazing. We could go to a concert. We could have a picnic.

I look forward to hearing your thoughts.

Thanks,
Myself

음... 잘하면 시라노 같은 친구도 만들 수 있겠군요. ^^

(첨부 파일은 이 글의 예제 코드를 포함합니다.)




[이 글에 대해서 여러분들과 의견을 공유하고 싶습니다. 틀리거나 미흡한 부분 또는 의문 사항이 있으시면 언제든 댓글 남겨주십시오.]







[최초 등록일: ]
[최종 수정일: 5/16/2023]

Creative Commons License
이 저작물은 크리에이티브 커먼즈 코리아 저작자표시-비영리-변경금지 2.0 대한민국 라이센스에 따라 이용하실 수 있습니다.
by SeongTae Jeong, mailto:techsharer at outlook.com

비밀번호

댓글 작성자
 




1  [2]  3  4  5  6  7  8  9  10  11  12  13  14  15  ...
NoWriterDateCnt.TitleFile(s)
13893정성태2/27/20252225Linux: 115. eBPF (bpf2go) - ARRAY / HASH map 기본 사용법
13892정성태2/24/20252981닷넷: 2325. C# - PowerShell과 연동하는 방법파일 다운로드1
13891정성태2/23/20252500닷넷: 2324. C# - 프로세스의 성능 카운터용 인스턴스 이름을 구하는 방법파일 다운로드1
13890정성태2/21/20252320닷넷: 2323. C# - 프로세스 메모리 중 Private Working Set 크기를 구하는 방법(Win32 API)파일 다운로드1
13889정성태2/20/20253050닷넷: 2322. C# - 프로세스 메모리 중 Private Working Set 크기를 구하는 방법(성능 카운터, WMI) [1]파일 다운로드1
13888정성태2/17/20252498닷넷: 2321. Blazor에서 발생할 수 있는 async void 메서드의 부작용
13887정성태2/17/20253070닷넷: 2320. Blazor의 razor 페이지에서 code-behind 파일로 코드를 분리 및 DI 사용법
13886정성태2/15/20252572VS.NET IDE: 196. Visual Studio - Code-behind처럼 cs 파일을 그룹핑하는 방법
13885정성태2/14/20253234닷넷: 2319. ASP.NET Core Web API / Razor 페이지에서 발생할 수 있는 async void 메서드의 부작용
13884정성태2/13/20253521닷넷: 2318. C# - (async Task가 아닌) async void 사용 시의 부작용파일 다운로드1
13883정성태2/12/20253260닷넷: 2317. C# - Memory Mapped I/O를 이용한 PCI Configuration Space 정보 열람파일 다운로드1
13882정성태2/10/20252577스크립트: 70. 파이썬 - oracledb 패키지 연동 시 Thin / Thick 모드
13881정성태2/7/20252832닷넷: 2316. C# - Port I/O를 이용한 PCI Configuration Space 정보 열람파일 다운로드1
13880정성태2/5/20253167오류 유형: 947. sshd - Failed to start OpenSSH server daemon.
13879정성태2/5/20253406오류 유형: 946. Ubuntu - N: Updating from such a repository can't be done securely, and is therefore disabled by default.
13878정성태2/3/20253197오류 유형: 945. Windows - 최대 절전 모드 시 DRIVER_POWER_STATE_FAILURE 발생 (pacer.sys)
13877정성태1/25/20253249닷넷: 2315. C# - PCI 장치 열거 (레지스트리, SetupAPI)파일 다운로드1
13876정성태1/25/20253706닷넷: 2314. C# - ProcessStartInfo 타입의 Arguments와 ArgumentList파일 다운로드1
13875정성태1/24/20253133스크립트: 69. 파이썬 - multiprocessing 패키지의 spawn 모드로 동작하는 uvicorn의 workers
13874정성태1/24/20253555스크립트: 68. 파이썬 - multiprocessing Pool의 기본 프로세스 시작 모드(spawn, fork)
13873정성태1/23/20252983디버깅 기술: 217. WinDbg - PCI 장치 열거파일 다운로드1
13872정성태1/23/20252883오류 유형: 944. WinDbg - 원격 커널 디버깅이 연결은 되지만 Break (Ctrl + Break) 키를 눌러도 멈추지 않는 현상
13871정성태1/22/20253292Windows: 278. Windows - 윈도우를 다른 모니터 화면으로 이동시키는 단축키 (Window + Shift + 화살표)
13870정성태1/18/20253731개발 환경 구성: 741. WinDbg - 네트워크 커널 디버깅이 가능한 NIC 카드 지원 확대
13869정성태1/18/20253456개발 환경 구성: 740. WinDbg - _NT_SYMBOL_PATH 환경 변수에 설정한 경로로 심벌 파일을 다운로드하지 않는 경우
13868정성태1/17/20253109Windows: 277. Hyper-V - Windows 11 VM의 Enhanced Session 모드로 로그인을 할 수 없는 문제
1  [2]  3  4  5  6  7  8  9  10  11  12  13  14  15  ...