Microsoft MVP성태의 닷넷 이야기
.NET Framework: 2119. C# - Semantic Kernel의 "Basic Loading of the Kernel" 예제 [링크 복사], [링크+제목 복사],
조회: 14269
글쓴 사람
정성태 (techsharer at outlook.com)
홈페이지
첨부 파일
 
(연관된 글이 1개 있습니다.)
(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 "Basic Loading of the Kernel" 예제

아래의 예제를 보면,

Breadcrumbssemantic-kernel/samples/notebooks/dotnet/01-basic-loading-the-kernel.ipynb
; https://github.com/microsoft/semantic-kernel/blob/main/samples/notebooks/dotnet/01-basic-loading-the-kernel.ipynb

Kernel 개체를 Generic Host처럼 초기화하는 것을 보여줍니다.

해당 코드에서는 NullLogger를 사용해서,

ILogger myLogger = NullLogger.Instance;
IKernel kernel_2 = Kernel.Builder
    .WithLogger(myLogger)
    .Build();

아무런 출력이 없지만 이것을 우리가 임의로 만든 ConsoleLogger로,

using Microsoft.Extensions.Logging;

namespace ConsoleApp1;

// https://learn.microsoft.com/en-us/dotnet/core/extensions/custom-logging-provider
public class ConsoleLogger : ILogger
{
    public IDisposable? BeginScope<TState>(TState state) where TState : notnull => default!;

    public bool IsEnabled(LogLevel logLevel)
    {
        return true;
    }

    public void Log<TState>(LogLevel logLevel, EventId eventId, TState state, Exception? exception, Func<TState, Exception?, string> formatter)
    {
        Console.WriteLine(formatter(state, exception));
    }
}

지난번 글의 summarize 함수와 연결해 열역학 법칙에 대해서만 테스트해 보면,

private static async Task<int> Main(string[] args)
{
    // ...[생략]...

    ILogger myLogger = new ConsoleLogger();
    var kernel = Kernel.Builder
        .WithLogger(myLogger)
        .Build();

    // ...[생략]...

    var prompt = @"{{$input}}

One line TLDR with the fewest words.";

    var summarize = kernel.CreateSemanticFunction(prompt);

    string text1 = @"
1st Law of Thermodynamics - Energy cannot be created or destroyed.
2nd Law of Thermodynamics - For a spontaneous process, the entropy of the universe increases.
3rd Law of Thermodynamics - A perfect crystal at zero Kelvin has zero entropy.";

    Console.WriteLine(await summarize.InvokeAsync(text1));

    return 0;
}

이런 결과를 얻게 됩니다.

Extracting blocks from template: {{$input}}

One line TLDR with the fewest words.
Rendering string template: {{$input}}

One line TLDR with the fewest words.
Extracting blocks from template: {{$input}}

One line TLDR with the fewest words.
Rendering list of 2 blocks
Rendered prompt:
1st Law of Thermodynamics - Energy cannot be created or destroyed.
2nd Law of Thermodynamics - For a spontaneous process, the entropy of the universe increases.
3rd Law of Thermodynamics - A perfect crystal at zero Kelvin has zero entropy.

One line TLDR with the fewest words.

Energy conserved, entropy increases, zero entropy at 0K.




자, 그다음 KernelConfig도 이렇게 바꿔서 초기화할 수 있습니다.

KernelConfig kernelConfig = new KernelConfig();
kernelConfig.AddOpenAITextCompletionService(
    "davinci-openai",
    "text-davinci-003",               // OpenAI Model name
    apiKey       // OpenAI API Key
);

ILogger myLogger = new ConsoleLogger();
var kernel = Kernel.Builder
    .WithLogger(myLogger)
    .WithConfiguration(kernelConfig)
    .Build();

kernelConfig에는 1개 이상의 서비스를 연결해 둘 수 있는데요, 예를 들어 아래는 OpenAI와 Azure OpenAI 서비스를 함께 지정하고 있습니다. (혹은 동일한 OpenAI여도 Model 이름이 다른 서비스를 추가할 수도 있습니다.)

kernelConfig.AddAzureTextCompletionService(
    "Azure_curie",                          // alias used in the prompt templates' config.json
    "my-finetuned-Curie",                   // Azure OpenAI *Deployment ID*
    "https://contoso.openai.azure.com/",    // Azure OpenAI *Endpoint*
    "...your Azure OpenAI Key..."           // Azure OpenAI *Key*
);

kernelConfig.AddOpenAITextCompletionService(
    "davinci-openai",
    "text-davinci-003",               // OpenAI Model name
    apiKey       // OpenAI API Key
);

저렇게 2개의 서비스를 추가한 경우, 기본적으로 첫 번째 서비스가 "default"로 처리된다고 합니다. 만약, 기본 서비스를 다른 걸로 지정하고 싶다면 SetDefaultTextCompletionService 메서드를 호출하면 되는데요, 아래의 코드는 위의 예제에서 두 번째 OpenAI 서비스를 default로 지정하고 있습니다.

kernelConfig.SetDefaultTextCompletionService("davinci-openai");

아하... 여기서 수수께끼가 하나 풀리는군요. ^^ 지난 글에서 AddOpenAITextCompletionService의 첫 번째 인자가 "Service ID"라고만 설명돼 있어 잘 모르겠다고 했는데 실은 임의의 문자열로 kernelConfig에 추가되는 서비스를 구분하기 위한 식별자로서만 의미가 있었던 것입니다.

따라서, 그냥 아무 문자열이나 고유하게만 붙여주면 되는 것입니다. ^^




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

[연관 글]






[최초 등록일: ]
[최종 수정일: 5/10/2023]

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)
13804정성태11/7/20247067Linux: 101. eBPF 함수의 인자를 다루는 방법
13803정성태11/7/20246645닷넷: 2309. C# - .NET Core에서 바뀐 DateTime.Ticks의 정밀도
13802정성태11/6/20247049Windows: 269. GetSystemTimeAsFileTime과 GetSystemTimePreciseAsFileTime의 차이점파일 다운로드1
13801정성태11/5/20246870Linux: 100. eBPF의 2가지 방식 - libbcc와 libbpf(CO-RE)
13800정성태11/3/20247843닷넷: 2308. C# - ICU 라이브러리를 활용한 문자열의 대소문자 변환 [2]파일 다운로드1
13799정성태11/2/20245616개발 환경 구성: 732. 모바일 웹 브라우저에서 유니코드 문자가 표시되지 않는 경우
13798정성태11/2/20247276개발 환경 구성: 731. 유니코드 - 출력 예시 및 폰트 찾기
13797정성태11/1/20247260C/C++: 185. C++ - 문자열의 대소문자를 변환하는 transform + std::tolower/toupper 방식의 문제점파일 다운로드1
13796정성태10/31/20246773C/C++: 184. C++ - ICU dll을 이용하는 예제 코드 (Windows)파일 다운로드1
13795정성태10/31/20245966Windows: 268. Windows - 리눅스 환경처럼 공백으로 끝나는 프롬프트 만들기
13794정성태10/30/20246048닷넷: 2307. C# - 윈도우에서 한글(및 유니코드)을 포함한 콘솔 프로그램을 컴파일 및 실행하는 방법
13793정성태10/28/20246013C/C++: 183. C++ - 윈도우에서 한글(및 유니코드)을 포함한 콘솔 프로그램을 컴파일 및 실행하는 방법
13792정성태10/27/20245401Linux: 99. Linux - 프로세스의 실행 파일 경로 확인
13791정성태10/27/20245860Windows: 267. Win32 API의 A(ANSI) 버전은 DBCS를 사용할까요?파일 다운로드1
13790정성태10/27/20245364Linux: 98. Ubuntu 22.04 - 리눅스 커널 빌드 및 업그레이드
13789정성태10/27/20246331Linux: 97. menuconfig에 CONFIG_DEBUG_INFO_BTF, CONFIG_DEBUG_INFO_BTF_MODULES 옵션이 없는 경우
13788정성태10/26/20245124Linux: 96. eBPF (bpf2go) - fentry, fexit를 이용한 트레이스
13787정성태10/26/20246488개발 환경 구성: 730. github - Linux 커널 repo를 윈도우 환경에서 git clone하는 방법 [1]
13786정성태10/26/20246266Windows: 266. Windows - 대소문자 구분이 가능한 파일 시스템
13785정성태10/23/20245605C/C++: 182. 윈도우가 운영하는 2개의 Code Page파일 다운로드1
13784정성태10/23/20246213Linux: 95. eBPF - kprobe를 이용한 트레이스
13783정성태10/23/20245725Linux: 94. eBPF - vmlinux.h 헤더 포함하는 방법 (bpf2go에서 사용)
13782정성태10/23/20245421Linux: 93. Ubuntu 22.04 - 커널 이미지로부터 커널 함수 역어셈블
13781정성태10/22/20245976오류 유형: 930. WSL + eBPF: modprobe: FATAL: Module kheaders not found in directory
13780정성태10/22/20247036Linux: 92. WSL 2 - 커널 이미지로부터 커널 함수 역어셈블
13779정성태10/22/20245607개발 환경 구성: 729. WSL 2 - Mariner VM 커널 이미지 업데이트 방법
1  2  3  4  5  6  [7]  8  9  10  11  12  13  14  15  ...