quick_start/02_OpenAI_getting_started.ipynb

{
  "cells": [
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "# Azure OpenAI Quickstart"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "## Overview  \n",
        "> \"Large Language Model(LLM)은 주어진 텍스트 다음에 나올 텍스트를 토큰 단위로 생성하는 모델입니다. 입력 문자열이 주어지면 LLM은 다음에 나올 텍스트를 예측합니다\".  \n",
        "\n",
        "이 `Azure OpenAI Quickstarts` 노트북은 사용자에게 LLM 개념을 설명하고 LLM을 활용하기 위한 핵심 패키지 요구 사항, 프롬프트 디자인에 대한 간략한 소개 및 다양한 사용 사례에 대한 몇 가지 간단한 예를 소개합니다. 더 많은 빠른 시작 예제를 보려면 [공식 Azure Open AI 빠른 시작 설명서](https://learn.microsoft.com/en-us/azure/ai-services/openai/quickstart?pivots=programming-language-studio)를 참조하세요.  \n",
        "  \n",
        "해당 컨텐츠는 빠르게 실습을 진행할 수 있도록 컨테이너 기반으로 환경이 모두 설치가 된 상태로 시작할 수 있습니다. 따라서, [DevContainer](https://code.visualstudio.com/docs/devcontainers/containers) 기반으로 개발에 필요한 환경설정이 사전 정의되어 있습니다. GitHub Codespace를 활용하거나, 로컬에 Docker를 설치한 상태에서 Visual Stduio Code IDE에 해당 Repository를 다운로드 받을 경우, 자동으로 컨테이너에 개발환경(Python Runtime 3.11.4, Azure OpenAI 1.13.3)을 설치합니다. .env 파일에 필수 API 정보를 입력 하고 저장 후 사용하세요."
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "### Getting started with Azure OpenAI Service\n",
        "\n",
        "신규 고객은 Azure OpenAi 서비스에 [액세스 신청](https://aka.ms/oai/access)을 해야합니다.\n",
        "승인이 완료된 후 고객은 Azure Portal에 로그인하고 Azure OpenAI 서비스 리소스를 만들고 스튜디오를 통해 모델 실험을 시작할 수 있습니다.\n",
        "\n",
        "[Great resource for getting started quickly](https://techcommunity.microsoft.com/t5/educator-developer-blog/azure-openai-is-now-generally-available/ba-p/3719177 )\n"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "## Build your first prompt  \n",
        "이 짧은 연습은 간단한 작업 \"요약\"을 위해 OpenAI 모델에 프롬프트를 제출하기위한 기본 소개를 제공합니다.\n",
        "\n",
        "![](images/generative-AI-models-reduced.jpg)  \n",
        "\n",
        "\n",
        "**Steps**:  \n",
        "1. 파이썬 환경에 OpenAI 라이브러리를 설치\n",
        "2. 표준 헬퍼 라이브러리를 로드하고 OpenAI 보안 자격 증명을 설정\n",
        "3. OpenAI 작업에 적합한 모델을 선택\n",
        "4. 모델에 대한 간단한 프롬프트를 만듭니다.\n",
        "5. 모델 API에 요청을 제출하십시오!"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 1. 파이썬 환경에 OpenAI 라이브러리를 설치\n",
        "DevContainer 가 시작할 때, 자동으로 `requirements.txt` 에 기술한 라이브러리를 설치합니다.\n",
        "따라서 추가적으로 라이브러리 설치를 진행하지 않아도 즉시 실습할 수 있습니다.\n",
        "Python 버전은 **Python 3.11.4**, **Azure OpenAI 1.13.3** 를 사용합니다."
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "## 2. 표준 헬퍼 라이브러리를 로드하고 OpenAI 보안 자격 증명을 설정\n",
        "루트 디렉토리에 존재하는 `.evn.sample` 파일을 복사하여 `.env` 파일로 이름을 변경하고, Azure OpenAI Endpoint URL(AZURE_OPENAI_ENDPOINT) 및 API Key(AZURE_OPENAI_API_KEY)를 넣습니다.  \n",
        "만약 하단 코드가 정상적으로 수행되지 않을 경우, 해당 파일을 저장 후 파일을 닫은 후 새로 열어서 진행합니다. (커널 다시 시작을 해도 됩니다. 상단의 `재시작` 버튼 클릭)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "gather": {
          "logged": 1674829434433
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import json\n",
        "from openai import AzureOpenAI\n",
        "from dotenv import load_dotenv\n",
        "load_dotenv()\n",
        "\n",
        "client = AzureOpenAI(\n",
        "    azure_endpoint = os.getenv(\"AZURE_OPENAI_ENDPOINT\"),\n",
        "    api_key        = os.getenv(\"AZURE_OPENAI_API_KEY\"),\n",
        "    api_version    = os.getenv(\"OPENAI_API_VERSION\")\n",
        ")"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "## 3. 작업에 적합한 OpenAI 모델 선택\n",
        "2024년 7월 기준, 본 실습에서 다음의 모델을 사용합니다. 실습의 편의를 위해 모델을 배포할 때에는 아래 이름과 동일하게 배포하는 것을 추천 드립니다.\n",
        "- LLM 모델: `gpt-4o`\n",
        "- 임베딩 모델: `text-embedding-ada-002`\n",
        "\n",
        "`gpt-4o` 모델은 한글 토크나이저의 최적화로 기존 gpt-4 대비 더 빠르고 정확한 성능에 저렴한 비용을 제공합니다.  \n",
        "모델에 자세한 정보는 다음을 참고하세요: [Azure OpenAI Service models](https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models)  "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {},
      "outputs": [],
      "source": [
        "deployment_name = os.getenv(\"DEPLOYMENT_NAME\")\n",
        "deployment_embedding_name = os.getenv(\"DEPLOYMENT_EMBEDDING_NAME\")"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "## 4. Prompt Design  \n",
        "\n",
        "대형 언어 모델(LLM)의 마법은 방대한 양의 텍스트를 훈련함으로써, 다음에 나올 텍스트(토큰) 예측 오류를 최소화 하도록 만들어져 있습니다.\n",
        "예를 들어 다음과 같은 개념을 학습합니다:\n",
        "\n",
        "* 어떻게 쓰는지\n",
        "* 문법의 작동 방식\n",
        "* 의역하는 방법\n",
        "* 질문에 대답하는 방법\n",
        "* 대화를 이끄는 방법\n",
        "* 여러 언어로 작성하는 방법\n",
        "* 코딩하는 방법\n",
        "* 기타\n",
        "\n",
        "### 대형 언어 모델(LLM)을 제어하는 방법  \n",
        "LLM에 대한 모든 입력 중에서 가장 영향력 있는 것은 **텍스트 프롬프트** 입니다.\n",
        "\n",
        "대규모 언어 모델은 몇 가지 방법으로 출력을 생성하라는 메시지를 표시할 수 있습니다: \n",
        "\n",
        "* **지침(Instruction)**: 모델에게 원하는 지침을 설명하세요.  \n",
        "\n",
        "* **완료(Completion)**: 모델이 원하는 것의 시작을 끝낼수 있도록 유도하십시오.  \n",
        "\n",
        "* **시연(Demonstration)**: 다음 중 하나를 사용하여 원하는 것을 모델에 보여주십시오.  \n",
        "프롬프트 내에 몇 가지 예시를 주거나, 또는 수백 또는 수천 개의 예로 구성된 훈련 데이터 세트를 이용한 미세 조정을 할 수 있습니다.  \n",
        "\n",
        "\n",
        "### 프롬프트를 만들기 위한 세 가지 기본 지침\n",
        "\n",
        "* **Show and tell**: 명확하고 간결한 지침을 포함하고, 예시 또는 두 가지 조합을 통해 원하는 것을 명확히 기술합니다.  \n",
        "\n",
        "* **Provide quality data**: 양질의 예시를 제시함으로 원하는 결과물을 만들어 낼 수 있도록 유도합니다. 할루시네이션을 최소화 하는 가장 확실한 방법은 정확한 데이터를 전달하는 것으로 시작합니다. (RAG 패턴 활용)  \n",
        "\n",
        "* **Check your setting**: 온도(Temperature) 및 TOP_P 설정은 모델이 응답을 생성하는 데 결정적인 방법을 제어합니다. 정답이 하나만 있는 응답을 요청하는 경우 온도를 더 낮게(0) 설정하고 싶을 것입니다. 더 다양한 응답을 찾고 있다면 더 창의적인 응답을 원할 때에는 온도를 높게(1) 설정할 수 있습니다."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "gather": {
          "logged": 1674494935186
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [],
      "source": [
        "# Create your first prompt\n",
        "system_message = \"\"\"너는 긍정과 부정을 구분할 수 있는 에이전트야. 결과는 JSON 형식으로 공백 없이 반환해줘. 답변 예:{\"1\": \"긍정\", \"2\": \"부정\"}\"\"\"\n",
        "user_message = \"\"\"1. 모니터가 너무 뜨거워. 2. 모니터가 시장 반응이 너무 뜨거워.\"\"\""
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "## 5. Submit!"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "gather": {
          "logged": 1674494938225
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'{\"1\":\"부정\",\"2\":\"긍정\"}'"
            ]
          },
          "execution_count": 4,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# Simple API Call\n",
        "response = client.chat.completions.create(\n",
        "    model=deployment_name,\n",
        "    max_tokens=60,\n",
        "    messages=[\n",
        "        {\"role\": \"system\", \"content\": system_message},\n",
        "        {\"role\": \"user\", \"content\": user_message},\n",
        "    ]\n",
        ")\n",
        "\n",
        "response.choices[0].message.content"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "### 출력물 데이터 분석 \n",
        "해당 API의 결과 이외에 어떤 정보들이 포함되어져 있는지 확인해 봅시다.  "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{\n",
            "  \"id\": \"chatcmpl-9j3Ou791TfA0iOneBnlNLD57P01QZ\",\n",
            "  \"choices\": [\n",
            "    {\n",
            "      \"finish_reason\": \"stop\",\n",
            "      \"index\": 0,\n",
            "      \"logprobs\": null,\n",
            "      \"message\": {\n",
            "        \"content\": \"{\\\"1\\\":\\\"\\ubd80\\uc815\\\",\\\"2\\\":\\\"\\uae0d\\uc815\\\"}\",\n",
            "        \"role\": \"assistant\",\n",
            "        \"function_call\": null,\n",
            "        \"tool_calls\": null\n",
            "      }\n",
            "    }\n",
            "  ],\n",
            "  \"created\": 1720524860,\n",
            "  \"model\": \"gpt-4o-2024-05-13\",\n",
            "  \"object\": \"chat.completion\",\n",
            "  \"system_fingerprint\": \"fp_abc28019ad\",\n",
            "  \"usage\": {\n",
            "    \"completion_tokens\": 12,\n",
            "    \"prompt_tokens\": 91,\n",
            "    \"total_tokens\": 103\n",
            "  }\n",
            "}\n"
          ]
        }
      ],
      "source": [
        "print(json.dumps(response.model_dump(), indent=2))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "### Azure OpenAI 사용 비용 확인\n",
        "여기에는 gpt-4o (0513)을 사용했을 때의 [비용](https://azure.microsoft.com/ko-kr/pricing/details/cognitive-services/openai-service/#pricing)을 즉시 출력합니다.  \n",
        "\n",
        "| model | version | prompt cost(1k) | completion cost(1k) |\n",
        "|-----|-----|-----|-----|\n",
        "| gpt-4o | 0513| $0.005 | $0.015 |\n",
        "| gpt-35-turbo | 0125 | $0.0005 | $0.0015|\n",
        "\n",
        "gpt-35-turbo는 gpt-4o 대비 1/10 수준입니다.\n",
        "아래는 gpt-4o를 활용할 때 비용을 산출하는 예시입니다."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "프롬프트 토큰수: 91\n",
            "컴플리션 토큰수: 12\n",
            "모든 토큰수: 103\n",
            "gpt-4o 사용에 따른 비용:  0.63 원 +  0.25 원 =  0.88 원\n"
          ]
        }
      ],
      "source": [
        "# OpenAI API 사용 비용 계산 함수\n",
        "def openai_cost(model_name, response):\n",
        "    dollar_to_won = 1380\n",
        "    if model_name == \"gpt-4o\":\n",
        "        model_prompt_cost = 0.005\n",
        "        model_complet_cost = 0.015\n",
        "    elif model_name == \"gpt-35-turbo\":\n",
        "        model_prompt_cost = 0.0005\n",
        "        model_complet_cost = 0.0015\n",
        "    else:\n",
        "        print(\"선택된 모델이 없습니다.\")\n",
        "\n",
        "    print(\"프롬프트 토큰수:\", response.usage.prompt_tokens)\n",
        "    print(\"컴플리션 토큰수:\", response.usage.completion_tokens)\n",
        "    print(\"모든 토큰수:\", response.usage.total_tokens)\n",
        "\n",
        "    prompt_cost  = round(response.usage.prompt_tokens * model_prompt_cost / 1000 * dollar_to_won, 2)\n",
        "    complet_cost = round(response.usage.completion_tokens * model_complet_cost / 1000 * dollar_to_won, 2)\n",
        "\n",
        "    print(deployment_name, \"사용에 따른 비용: \", prompt_cost, \"원 + \", complet_cost, \"원 = \", prompt_cost + complet_cost , \"원\" )\n",
        "\n",
        "openai_cost(deployment_name, response)"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "### 동일한 호출을 반복합니다. 결과는 어떤가요?\n",
        "만약, gpt-35-turbo 모델을 미리 배포했을 경우, deployment_name을 `gpt-35-turbo` 로 변경해서 실행하여 비교해 보세요.  \n",
        "Azure OpenAI Stduio에서 `gpt-35-turbo` 모델이 배포되지 않았을 경우, 에러가 발생할 수 있습니다.  배포 후에 테스트 해 보세요."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "gather": {
          "logged": 1674494940872
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'{\"1\":\"부정\", \"2\":\"긍정\"}'"
            ]
          },
          "execution_count": 7,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# deployment_name = \"gpt-35-turbo\"\n",
        "\n",
        "# Simple API Call\n",
        "response = client.chat.completions.create(\n",
        "    model=deployment_name,\n",
        "    max_tokens=60,\n",
        "    messages=[\n",
        "        {\"role\": \"system\", \"content\": system_message},\n",
        "        {\"role\": \"user\", \"content\": user_message},\n",
        "    ]\n",
        ")\n",
        "\n",
        "response.choices[0].message.content"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "프롬프트 토큰수: 114\n",
            "컴플리션 토큰수: 13\n",
            "모든 토큰수: 127\n",
            "gpt-35-turbo 사용에 따른 비용:  0.08 원 +  0.03 원 =  0.11 원\n"
          ]
        }
      ],
      "source": [
        "# OpenAI API 사용 비용 계산\n",
        "openai_cost(deployment_name, response)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "gpt-4o는 gpt-35-turbo 대비 비용을 단순 비교하면 10배 더 비싸야 하지만, gpt-4o의 한글 토큰에 대한 최적화가 진행되어 그보다 낮은 8배 비용 차이가 나는 것을 알 수 있습니다."
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "## 다양한 케이스에 대한 활용법\n",
        "1. 요약\n",
        "2. 분류\n",
        "3. 제품 이름 생성\n",
        "4. 임베딩"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "### 1. 요약\n",
        "LLM은 다양한 케이스에서 사용이 가능합니다. 요약하기 위한 방법은 다음과 같습니다.\n",
        "텍스트에 대한 토큰 길이를 확인할 수 있는 방법:\n",
        "1. 토크나이저: https://platform.openai.com/tokenizer "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "gather": {
          "logged": 1674495198534
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [],
      "source": [
        "original_text = \"Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches.\\n\\nTl;dr\"\n",
        "system_message = \"Summarize the text below as a bullet point list of the most important points using Korean.\"\n",
        "user_message = \"Text: ```\" + original_text + \"```\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "gather": {
          "logged": 1674495201868
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "- 최근 연구에서는 대규모 텍스트 코퍼스에서 사전 학습 후 특정 작업에 대해 미세 조정(fine-tuning)하는 방식이 NLP 작업 및 벤치마크에서 상당한 성과를 보였음.\n",
            "- 일반적으로 작업에 무관한 아키텍처를 사용하지만, 여전히 수천 또는 수만 개의 예제가 포함된 작업별 미세 조정 데이터셋이 필요함.\n",
            "- 반면에 인간은 대개 몇 가지 예제나 간단한 지시만으로 새로운 언어 작업을 수행할 수 있음.\n",
            "- 현재의 NLP 시스템은 이러한 능력을 대부분 갖추지 못하고 있음.\n",
            "- 본 연구에서는 언어 모델의 규모를 확장함으로써 작업에 무관한 소수 샷(few-shot) 성능이 크게 향상됨을 보여줌.\n",
            "- 때로는 이전 최첨단 미세 조정 접근 방식과 경쟁할 정도의 성과를 보이기도 함.\n"
          ]
        }
      ],
      "source": [
        "# deployment_name = os.getenv(\"DEPLOYMENT_NAME\")\n",
        "\n",
        "#Setting a few additional, typical parameters during API Call\n",
        "response = client.chat.completions.create(\n",
        "  model=deployment_name,\n",
        "  messages=[\n",
        "    {\"role\": \"system\", \"content\": system_message},\n",
        "    {\"role\": \"user\", \"content\": user_message},\n",
        "  ],\n",
        "  temperature=0.7,\n",
        "  max_tokens=300\n",
        ")\n",
        "\n",
        "print(response.choices[0].message.content)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "#### 참고하는 Text 문서가 없을 때\n",
        "아래와 같이 참고하는 Text 문서가 없을 때에 어떻게 결과가 출력하는지 살펴 봅니다. RAG에서 다음과 같이 모델 자체 지식이 아닌 전달 받은 데이터 기반으로 출력하도록 구성할 수 있습니다."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Please provide the text you would like summarized, and I'll be happy to help!\n"
          ]
        }
      ],
      "source": [
        "original_text = \"\"\n",
        "system_message = \"Summarize the text below as a bullet point list of the most important points using Korean.\"\n",
        "user_message = \"Text: ```\" + original_text + \"```\"\n",
        "\n",
        "#Setting a few additional, typical parameters during API Call\n",
        "response = client.chat.completions.create(\n",
        "  model=deployment_name,\n",
        "  messages=[\n",
        "    {\"role\": \"system\", \"content\": system_message},\n",
        "    {\"role\": \"user\", \"content\": user_message},\n",
        "  ],\n",
        "  temperature=0.7,\n",
        "  max_tokens=300\n",
        ")\n",
        "\n",
        "print(response.choices[0].message.content)"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "### 2. 분류\n",
        "\n",
        "제공된 범주에 맞게 적절한 항목 분류를 추론합니다. 다음 예에서는 프롬프트에서 분류 할 범주와 텍스트를 모두 제공합니다.\n",
        "\n",
        "고객 문의 : 안녕하세요, 최근 노트북 키보드의 열쇠 중 하나가 최근에 파산되었으며 교체가 필요합니다.\n",
        "\n",
        "분류 카테고리 :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "gather": {
          "logged": 1674499424645
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [],
      "source": [
        "system_message = \"\"\"Classify the following inquiry into one of the following:\n",
        "categories: [Pricing, Hardware Support, Software Support]\n",
        "\"\"\"\n",
        "user_message = \"\"\"inquiry: Hello, one of the keys on my laptop keyboard broke recently and I'll need a replacement.\n",
        "Classified category:\n",
        "\"\"\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "gather": {
          "logged": 1674499378518
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Classified category: Hardware Support\n"
          ]
        }
      ],
      "source": [
        "response = client.chat.completions.create(\n",
        "  model=deployment_name,\n",
        "  messages=[\n",
        "    {\"role\": \"system\", \"content\": system_message},\n",
        "    {\"role\": \"user\", \"content\": user_message},\n",
        "  ],\n",
        "  temperature=0,\n",
        "  max_tokens=60\n",
        ")\n",
        "\n",
        "print(response.choices[0].message.content)"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "### 3. 제품 이름 생성\n",
        "시드 단어를 기반으로 제품 이름을 만듭니다. 여기에는 이름을 생성할 제품에 대한 프롬프트 정보가 포함되어 있습니다. 또한 생성 패턴을 보여주는 비슷한 예를 제공합니다. 보다 창의적인 반응성을 높이기 위해 온도 값을 높게 설정했습니다.\n",
        "\n",
        "제품 설명 : 발 크기에 맞는 신발 한 쌍.  \n",
        "종자 단어 : 적응성, 적합, 옴니 피트.  \n",
        "\n",
        "> temperature 값이 0에서 1 사이로 변경하여 반복 실행시 어떤 변화가 있는지 확인 합니다."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "gather": {
          "logged": 1674257087279
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [],
      "source": [
        "system_message = \"\"\"Come up with five product names that fit the given product description and seed words.\"\"\"\n",
        "user_message = \"\"\"Product description: A pair of shoes that can fit any foot size.\n",
        "Seed words: adaptable, fit, omni-fit.\n",
        "Product names:\"\"\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "1. AdaptFit Shoes\n",
            "2. OmniStride Footwear\n",
            "3. FlexiFit Sneakers\n",
            "4. UniversalStep Shoes\n",
            "5. AllSize Comfort Shoes\n"
          ]
        }
      ],
      "source": [
        "response = client.chat.completions.create(\n",
        "  model=deployment_name,\n",
        "  messages=[\n",
        "    {\"role\": \"system\", \"content\": system_message},\n",
        "    {\"role\": \"user\", \"content\": user_message},\n",
        "  ],\n",
        "  temperature=0.8,\n",
        "  # temperature=0,\n",
        "  max_tokens=60\n",
        ")\n",
        "\n",
        "print(response.choices[0].message.content)"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "### 4. 임베딩\n",
        "이 섹션에서는 임베딩을 검색하고 단어, 문장 및 문서 사이의 유사성을 찾는 방법을 보여줍니다."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {},
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "\n",
        "def cosine_similarity(a, b):\n",
        "    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "gather": {
          "logged": 1674829424097
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [],
      "source": [
        "text = 'the quick brown fox jumped over the lazy dog'"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "gather": {
          "logged": 1674829446467
        },
        "jupyter": {
          "outputs_hidden": true,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "[-0.0046152775175869465, 0.009807861410081387, -0.015035336837172508, -0.00635036826133728, -0.011387523263692856, 0.015568234957754612, -0.024145355448126793, -0.0158600602298975, -0.015923500061035156, -0.029512399807572365, 0.02156968228518963, 0.02121441625058651, 0.018930567428469658, 0.004018939100205898, -0.0006974461721256375, -0.007644549012184143, 0.024906639009714127, -0.004215604159981012, 0.011089353822171688, -0.008577120490372181, -0.008983138017356396, 0.02122710458934307, -0.006413808558136225, -0.008532712236046791, 0.006416980642825365, 0.013208257965743542, 0.007524012587964535, -0.0032576564699411392, -0.009040234610438347, 0.0014020293019711971, 0.00668660132214427, 0.003920606803148985, -0.039409078657627106, -0.0026660761795938015, -0.012751488946378231, -0.021645810455083847, -0.0037651783786714077, -0.010486671701073647, 0.026238882914185524, -0.04544859007000923, 0.00947162788361311, 0.01566973887383938, -0.0224832221865654, -0.01119720283895731, -0.002677178243175149, 0.011945797130465508, 0.009953773580491543, -0.012269342318177223, -0.02277504652738571, 0.010309039615094662, -0.0005515335942618549, 0.006572409067302942, -0.009547756053507328, -0.011565156280994415, -0.004415440838783979, -0.007200467400252819, -0.0015931430971249938, -0.006223487667739391, 0.009173458442091942, -0.005582741461694241, 0.01110204216092825, 0.013588899746537209, -0.005414624698460102, -0.005833330098539591, 0.0007747639319859445, -0.009351091459393501, 0.01105763390660286, -0.010759465396404266, -0.013094065710902214, -0.005138659384101629, 0.029588527977466583, -0.0017398486379534006, 8.143149898387492e-05, 0.005896770395338535, 0.03146636113524437, -0.008856257423758507, -0.027456937357783318, 0.008957762271165848, 0.01437555905431509, 0.011463651433587074, 0.02974078431725502, -0.04222582280635834, -0.006128327455371618, 0.01427405420690775, 0.017611011862754822, 0.002988035324960947, -0.025718674063682556, 0.03770887851715088, -0.01678628847002983, 0.011305050924420357, 0.0055129569955170155, 0.030578196048736572, 0.008285295218229294, -0.0055954293347895145, -0.016925856471061707, 0.006769073661416769, -0.0007160817040130496, 0.018917879089713097, -0.01020753476768732, -0.025769425556063652, 0.006864234339445829, -0.011546123772859573, -0.02702554315328598, -0.002705726306885481, 0.006578753236681223, -0.02712704800069332, -0.0052591958083212376, 0.02212795615196228, 0.01796627603471279, -0.014134486205875874, 0.016215326264500618, 0.023625146597623825, 0.017801331356167793, -0.039941977709531784, -0.004015767481178045, -0.011469995602965355, 0.009839581325650215, -0.00965560507029295, -0.009135394357144833, -0.009033890441060066, 0.009414532221853733, 0.017839396372437477, 0.02212795615196228, -0.002654974116012454, -0.014210614375770092, -0.010372479446232319, -0.004006251227110624, 0.007384444121271372, -0.010905377566814423, 0.003470181254670024, 0.032608285546302795, 0.023802777752280235, 0.03309043124318123, -0.012396222911775112, -0.03679534047842026, 0.02339676022529602, 0.0050593591295182705, -0.011089353822171688, -0.03055281937122345, -0.022533973678946495, -0.0015709389699622989, 0.016367582604289055, -0.006673913449048996, 0.008723033592104912, 0.022470533847808838, 0.023713963106274605, 0.02140473760664463, 0.008583464659750462, -0.002150624291971326, -0.0065153129398822784, 0.018296165391802788, -0.002030087634921074, 0.02984228916466236, 0.00042425660649314523, 0.009630228392779827, 0.015657052397727966, 0.003758834209293127, 0.014400934800505638, -0.028091339394450188, -0.021836131811141968, 0.024145355448126793, 0.021785378456115723, -0.00025118369376286864, -0.008266263641417027, 0.0020364318042993546, 0.03146636113524437, 0.03298892453312874, 0.003092711791396141, 0.008196479640901089, -0.010328071191906929, 0.004881726577877998, 0.03281129151582718, -0.011501715518534184, 0.03760737553238869, 0.002126834122464061, 0.0067056333646178246, 0.02158237062394619, -0.0074859485030174255, -0.03230376914143562, 0.0016764083411544561, 0.0039618429727852345, 0.01414717361330986, 0.038850802928209305, 0.02386621944606304, 0.00505618704482913, 0.02221677266061306, 0.005547848995774984, -0.027507688850164413, -0.007479604333639145, -0.0009944257326424122, 0.009211523458361626, 0.001749364659190178, -0.018575303256511688, -0.013525459915399551, -0.6857636570930481, -0.029893042519688606, 0.003863510675728321, -0.010562799870967865, 0.021836131811141968, 0.034054722636938095, -0.012117085978388786, -0.010638928040862083, -0.00045795924961566925, 0.035780295729637146, -0.030324434861540794, 0.002107802079990506, 0.004700921941548586, -0.0030197554733604193, -0.009712700732052326, -0.0034130848944187164, 0.02730467915534973, -0.02274966984987259, -0.008215511217713356, 0.020833775401115417, 0.004545493517071009, 0.0069276741705834866, -0.010721400380134583, 0.00197140546515584, -0.00509742321446538, 0.011476339772343636, -0.00220930646173656, 0.005519301164895296, -0.0013972712913528085, 0.017281122505664825, -0.002001539571210742, 0.018714871257543564, -0.006566064897924662, 0.0053955926559865475, 0.04283485189080238, 0.004523289389908314, -0.00600779103115201, 0.020618079230189323, 0.040043480694293976, 0.05369582027196884, -0.035678792744874954, -0.013436643406748772, 0.014616631902754307, 0.008856257423758507, -0.011431931518018246, 0.0012188456021249294, 0.02531265653669834, 0.017103489488363266, -0.005392420571297407, -0.015809308737516403, 0.014071045443415642, 0.017801331356167793, 0.016101133078336716, 0.005008607171475887, -0.018346916884183884, 0.02684791013598442, 0.01084193680435419, -0.0002571312361396849, -0.0016795804258435965, -0.002469411352649331, -0.010962474159896374, 0.001914309337735176, -0.010315382853150368, 0.008818193338811398, -0.022648166865110397, 0.003879370866343379, -0.0071687474846839905, 0.010023558512330055, 0.02212795615196228, -0.0144136231392622, 0.02257203869521618, 0.01985679566860199, -0.010480327531695366, 0.010042590089142323, 0.014565879479050636, 0.010740432888269424, 0.008983138017356396, -0.006410636473447084, -0.01169838011264801, 0.006464560516178608, -0.012282030656933784, -0.002735860412940383, -0.03945982828736305, -0.0013512771110981703, 0.03603405877947807, 0.01712886430323124, -0.0351712703704834, -0.006312304176390171, -0.00509742321446538, -0.024259548634290695, 0.016900479793548584, 0.02439911663532257, -0.031796250492334366, -0.0008374112076126039, -0.012459663674235344, 0.008221855387091637, -0.005858706310391426, 0.038774676620960236, 0.04113465175032616, -0.032608285546302795, 0.012142462655901909, -0.0016653063939884305, -0.0035336215514689684, -0.004095067735761404, 0.014121797867119312, 0.030628949403762817, 0.016824351623654366, 0.0016415162244811654, 0.01740800216794014, -0.010911721736192703, -0.00870400108397007, -0.017534881830215454, 0.01721768081188202, 0.00514500355347991, -0.006572409067302942, -0.024906639009714127, 0.015631675720214844, -0.0048373183235526085, 0.015517483465373516, -0.019793355837464333, 0.009357435628771782, -0.011685692705214024, 0.008780129253864288, -0.000971428700722754, 0.0076128290966153145, 0.0019095512107014656, -0.02674640528857708, -0.029537776485085487, -0.01624070107936859, -0.004913446493446827, -0.007105307187885046, 0.020351629704236984, 0.03552653640508652, -0.002068151952698827, -0.003240210236981511, -0.00446302117779851, 0.014083733782172203, 0.0059697264805436134, 0.018626054748892784, -0.01603769324719906, -0.031491734087467194, 0.009249587543308735, -0.014362870715558529, -0.01360158808529377, 0.007682613097131252, -0.019996363669633865, -0.02584555372595787, 0.004634309560060501, 0.006604128982871771, 0.03001992218196392, -0.012186869978904724, -0.023422136902809143, -0.012434286996722221, -0.015377914533019066, 0.023066870868206024, -0.005160863511264324, -0.03237989917397499, -0.004428128711879253, -0.004504257347434759, -0.02837047539651394, 0.027685321867465973, 0.014172550290822983, -0.006242519710212946, -0.004263184033334255, 0.010575488209724426, -0.03506976366043091, 0.003165668109431863, 0.023054184392094612, -0.024919327348470688, -0.02712704800069332, -0.004913446493446827, -0.02267354167997837, -0.006359884049743414, -0.0013457260793074965, -0.017268434166908264, 0.011704724282026291, -0.027228550985455513, -0.024462556466460228, -0.002049119910225272, -0.010543768294155598, -0.004539149347692728, 0.01905744895339012, 0.006442356389015913, -0.022635478526353836, 0.015479418449103832, -0.005179895553737879, 0.024525998160243034, 0.01264998409897089, -0.02395503595471382, -0.010264631360769272, 0.02120172791182995, 0.027228550985455513, 0.012954497709870338, -0.01351277157664299, -0.005570053122937679, -0.020199373364448547, 0.005598601419478655, 0.03012142702937126, 0.01492114458233118, 0.009306683205068111, 0.01632951758801937, -0.007701645139604807, 0.007517668418586254, -0.026949414983391762, 0.021912259981036186, -0.03184700012207031, -0.0017366765532642603, -0.0038539946544915438, 0.006207627709954977, 2.276166560477577e-05, -0.020009052008390427, 1.185167820949573e-05, 0.019628411158919334, -0.038952309638261795, 0.017204994335770607, 0.02395503595471382, -0.012307406403124332, 0.022445157170295715, 0.008773785084486008, -0.01278955303132534, 0.002838950837031007, -0.02068151906132698, 0.015631675720214844, -0.00523382006213069, 0.0222040843218565, 0.02467825450003147, 0.00875475350767374, 0.0156951155513525, 0.0016351721715182066, -0.004123615566641092, -0.023409448564052582, 0.0033496448304504156, 0.017154240980744362, 0.01658327877521515, 0.0085707763209939, 0.0006907056085765362, 0.007003802806138992, -0.010880001820623875, 0.014451687224209309, -0.0013147989520803094, 0.017890147864818573, -0.0071814353577792645, 0.00784755777567625, -0.014248678460717201, 0.011692036874592304, 0.011095697991549969, 0.02946164831519127, -0.005681073758751154, -0.01724305748939514, 0.0062964437529444695, -0.0020427757408469915, -0.0035209334455430508, -0.024881262332201004, 0.027913706377148628, 0.007428852375596762, -0.014946521259844303, 0.021074848249554634, 0.004961026832461357, 0.02920788712799549, 0.02522384002804756, 0.0041711959056556225, 0.0030895397067070007, 0.008976793847978115, 0.010270975530147552, 0.01885443925857544, -0.0005618426366709173, -0.006369400303810835, 0.013030625879764557, 0.013309762813150883, -0.020161308348178864, -0.02566792257130146, -0.005351184401661158, 0.0017065424472093582, -0.017839396372437477, 0.02050388604402542, 0.013119441457092762, -0.035501159727573395, 0.013994917273521423, 0.01877831108868122, 0.02030087634921074, -0.021924948319792747, -0.027279304340481758, -0.006185423582792282, 0.004234636202454567, 0.00398721918463707, 0.005319464020431042, -0.028065962716937065, -0.0006573994760401547, -0.0034511492121964693, 0.022533973678946495, -0.0001861574564827606, -0.004961026832461357, -0.00013233238132670522, -0.009706356562674046, -0.008691312745213509, -0.013588899746537209, 0.020009052008390427, -0.011235266923904419, 0.0005757202161476016, -0.028522731736302376, 0.008380455896258354, -0.0007061691721901298, -0.007219499442726374, -0.007365412078797817, 0.025883618742227554, -0.00035427408874966204, -0.013119441457092762, -0.0015772830229252577, -0.001289422856643796, 0.019260456785559654, -0.020161308348178864, -0.014438998885452747, -0.01750950701534748, 0.013677716255187988, 0.0023123968858271837, -0.00884357001632452, -0.020186685025691986, -0.011381179094314575, 0.02829434722661972, 0.010055278427898884, -0.013030625879764557, -0.015898125246167183, -0.006458216346800327, 0.022457845509052277, 0.08049298077821732, 0.020998720079660416, -0.018461110070347786, 0.013170193880796432, 0.01894325576722622, -0.02730467915534973, -0.022889239713549614, -0.006369400303810835, 0.028142090886831284, -0.018105844035744667, -0.0039269509725272655, -0.008989482186734676, 0.005516129080206156, -0.008006158284842968, 0.0239296592772007, -0.007581108715385199, 0.003619265742599964, -0.008596152998507023, 0.001638344256207347, 0.0018048748606815934, -0.0049324785359203815, -0.0036605019122362137, -9.461517038289458e-05, 0.012446975335478783, -0.014451687224209309, 0.0036827060393989086, -0.0025899477768689394, 0.016887791454792023, 0.02837047539651394, -0.029588527977466583, -0.0034321171697229147, 0.00798078253865242, 0.015644364058971405, 0.006680257618427277, 0.02120172791182995, 0.020719582214951515, -0.005084735341370106, 0.014667383395135403, 0.03281129151582718, -0.03154248744249344, 0.011730100959539413, 0.007543044630438089, 0.004409096669405699, 0.004605761729180813, 0.024284925311803818, 0.0002622857573442161, -0.01794089935719967, 0.016532527282834053, 0.0060395109467208385, -0.0296646561473608, 0.03636394813656807, -0.005287744104862213, -0.01028366293758154, 0.002195032313466072, 0.0060712313279509544, 0.025896307080984116, -0.01950152963399887, -0.004244151990860701, -0.013208257965743542, 0.0014710206305608153, -0.01033441536128521, -0.020085180178284645, -0.002930939197540283, -0.009452596306800842, -0.0015376328956335783, -0.025553729385137558, -0.011673004366457462, -0.008805505931377411, -0.0064835925586521626, -0.014933832921087742, -0.021455489099025726, -0.018524549901485443, -0.01877831108868122, -0.010010870173573494, 0.015238345600664616, 0.030222931876778603, 0.0103788236156106, -0.0014020293019711971, 0.004475709050893784, 0.010702368803322315, -0.01731918565928936, -0.016278766095638275, -0.0010079068597406149, -0.02611200325191021, 0.002415487077087164, 0.00970001332461834, 0.013398579321801662, -0.006455044727772474, 0.011850637383759022, 0.012066333554685116, 0.010803872719407082, 0.008545400574803352, -0.010810216888785362, 0.005846018437296152, 0.02267354167997837, -0.01114645041525364, 0.01242794282734394, 0.03456224501132965, 0.0191462654620409, -0.005585913080722094, 0.03291279822587967, -0.01704004965722561, 0.001068175071850419, 0.000574530684389174, 0.014642007648944855, -0.007606484927237034, 0.029867665842175484, 0.005030811298638582, -0.020173996686935425, -0.03293817490339279, 0.031694743782281876, -0.00552881695330143, 0.014565879479050636, 0.012459663674235344, 0.01092440914362669, 0.014984585344791412, 0.01578393206000328, 0.011374834924936295, -0.00015156270819716156, -0.009763453155755997, -0.00856443215161562, -0.021341297775506973, -0.001030903891660273, 0.02946164831519127, -0.025274591520428658, 0.021633122116327286, -0.004041143227368593, -0.015352538786828518, -0.0030308575369417667, 0.003375020809471607, 0.011806229129433632, 0.007067243102937937, 0.0015542858745902777, -0.009801517240703106, -0.04199744015932083, -0.01084193680435419, -0.0036795339547097683, 0.012383535504341125, -0.01351277157664299, 0.0012093295808881521, -0.031237974762916565, -0.008849913254380226, 0.012434286996722221, -0.04270797222852707, 0.01986948400735855, -0.025515664368867874, -0.0018508690409362316, -0.014426310546696186, -0.0023679069709032774, 0.013994917273521423, -0.01568242721259594, -0.010740432888269424, -0.024246860295534134, -0.016126509755849838, -0.002261644694954157, -0.03618631511926651, 0.007803149521350861, -0.0156951155513525, 0.04017036035656929, 0.017636386677622795, 0.030045298859477043, -0.0034130848944187164, 0.0048151141963899136, 0.024322988465428352, -0.02674640528857708, -0.003698565997183323, -0.0016573762986809015, -0.003179942024871707, -0.016189949586987495, -0.007035522721707821, 0.029588527977466583, 0.0058174701407551765, 0.026594148948788643, 0.018803687766194344, -0.017788643017411232, 0.011311395093798637, -0.0036953941453248262, 0.009433563798666, -0.012129774317145348, -0.016253389418125153, 0.020427757874131203, -0.0054907528683543205, 0.016304142773151398, -0.0072639076970517635, -0.008602497167885303, -0.004085551481693983, 0.010435919277369976, 0.03293817490339279, 0.03735361248254776, 0.0065153129398822784, 0.02202645130455494, 0.0008723033242858946, -0.004995918832719326, 0.0047199539840221405, -0.003314752597361803, -0.027177799493074417, -0.0018524550832808018, 0.010987849906086922, 0.006461388431489468, 0.026974789798259735, -0.0011665074853226542, 0.01624070107936859, -0.013144818134605885, 0.024145355448126793, -0.0038444786332547665, 0.005376560613512993, -0.011406555771827698, -0.012136118486523628, -0.0014337494503706694, -0.025870930403470993, -0.010740432888269424, -0.03481600433588028, -0.011520748026669025, -0.016849728301167488, 0.006077575031667948, 0.01966647431254387, -0.011799884960055351, 0.010822905227541924, -0.014464374631643295, -0.03755662217736244, -0.004291732329875231, 0.0014274053974077106, 0.045296333730220795, -0.010315382853150368, 0.00802519079297781, 0.0117427883669734, -0.00466920156031847, 0.0015178078319877386, 0.01615188457071781, 0.012630952522158623, 0.00514183146879077, -0.00048531783977523446, 0.01830885373055935, -0.008900665678083897, -0.004830974154174328, -0.019983675330877304, -0.008989482186734676, -0.024449868127703667, -0.010588176548480988, 0.027989834547042847, 0.0008524782606400549, -0.00017882218526210636, -0.008735720999538898, -0.017534881830215454, -0.0004873003636021167, 0.011032258160412312, -0.015162217430770397, -0.00600461894646287, -0.02420879527926445, -0.00018011081556323916, -0.042022816836833954, 0.009357435628771782, -0.01894325576722622, 0.0035399654880166054, -0.03075582906603813, -0.012726112268865108, 0.0072068115696311, 0.015555547550320625, -0.02755844034254551, -0.006267895922064781, -0.018994009122252464, 0.01264363992959261, -0.025071583688259125, -0.002328256843611598, 0.013322451151907444, 0.0026295979041606188, -0.022242149338126183, -0.02021206170320511, -0.0059348344802856445, 0.025261905044317245, -0.01427405420690775, -0.01667209528386593, -0.006026823073625565, 0.012529447674751282, 0.01712886430323124, -0.019793355837464333, 0.0005384490359574556, -0.028954125940799713, -0.00467237364500761, 0.015720492228865623, 0.02112559974193573, -0.0011228922521695495, -0.0041838837787508965, 0.007251219358295202, -0.009503347799181938, 0.002283848589286208, 0.0040347995236516, -0.014223301783204079, 0.009795173071324825, -0.019603034481406212, -0.0006090262904763222, -0.017027361318469048, -0.0015233587473630905, 0.0030054813250899315, 0.02077033556997776, 0.0002999533899128437, -0.006965738721191883, 0.02439911663532257, -0.028142090886831284, 0.017649075016379356, -0.03083195723593235, 0.006125155370682478, -0.024970078840851784, 0.007784117478877306, 0.01020753476768732, 0.00635036826133728, 0.0144136231392622, -0.020021740347146988, 0.0032798603642731905, -0.02503351867198944, -0.005351184401661158, -0.003619265742599964, 0.02283848635852337, -0.024373739957809448, 0.01105763390660286, 0.00034356856485828757, 0.01110838633030653, -0.031872376799583435, -0.007879277691245079, 0.03311580419540405, 0.0012362917186692357, 0.024805134162306786, 0.03583104908466339, -0.02458943799138069, 0.006978426594287157, 0.005636665504425764, 0.009649260900914669, 0.012262999080121517, -0.02239440567791462, -0.007790461648255587, -0.010321727022528648, 0.02857348509132862, 0.0037778664845973253, -0.01092440914362669, -0.011469995602965355, -0.003381364978849888, -0.01658327877521515, 0.015745867043733597, -0.023701274767518044, -0.003048303537070751, 0.018194660544395447, 0.0021157320588827133, -0.02702554315328598, 0.01107032224535942, -0.013436643406748772, -0.011996549554169178, -0.013576211407780647, -0.005084735341370106, -0.030502067878842354, -0.01966647431254387, -0.007790461648255587, 0.002493201522156596, 0.013829972594976425, -0.011666660197079182, -0.02538878470659256, -0.008945073932409286, -0.006661225575953722, -0.010137750767171383, 0.026594148948788643, 0.003847650717943907, 0.02121441625058651, -0.02276235818862915, 0.011298706755042076, 0.014350182376801968, -0.013309762813150883, 0.030146801844239235, -0.012973529286682606, -0.009725389070808887, -0.0066104731522500515, -0.002271160716190934, -0.0022299245465546846, -0.000904023414477706, -0.004811942111700773, -0.0103788236156106, 0.0048246304504573345, 0.021988388150930405, 0.010734088718891144, 0.01101956982165575, -0.008253575302660465, 6.408455374184996e-05, -0.0036668458487838507, -0.019260456785559654, 0.007530356757342815, -0.008520024828612804, 0.011330426670610905, -0.026822533458471298, -0.0112416110932827, 0.006692945491522551, -0.012117085978388786, -0.02628963626921177, -0.005750857759267092, 0.017750579863786697, -0.0058713941834867, 0.00957313273102045, -0.02211526781320572, -0.004995918832719326, -0.01088634505867958, -0.029766160994768143, 0.013791908510029316, -0.012903745286166668, 0.030070673674345016, 0.024970078840851784, -0.011806229129433632, -0.01024559885263443, 0.0009341575787402689, 0.0012545307399705052, -0.013779220171272755, -0.021734626963734627, 0.00711799506098032, -0.017369939014315605, -0.0014369214186444879, -0.029690032824873924, 0.0018746590940281749, -0.022876551374793053, 0.006826169788837433, -0.022001076489686966, -0.015504795126616955, -0.017204994335770607, 0.024779757484793663, -0.008640561252832413, -0.021150976419448853, -0.0023092248011380434, -0.022242149338126183, 0.00394281093031168, -0.0009872887749224901, -0.010397855192422867, 0.006921330466866493, 0.003955499269068241, -0.01823272556066513, -0.0068325139582157135, 4.004665242973715e-05, -0.023358697071671486, -0.013017937541007996, 0.015314474701881409, -0.006997458636760712, 0.021975699812173843, 0.2084900140762329, -0.0017826707335188985, 0.0050688753835856915, 0.03687146678566933, -0.010550112463533878, 0.030349811539053917, 0.021709250286221504, -0.0035494815092533827, -0.006718321703374386, 0.02575673721730709, -0.002483685500919819, 0.012967185117304325, -0.02150624245405197, 0.004618449602276087, 0.014603943563997746, 0.008735720999538898, -0.005179895553737879, -0.0025566418189555407, -0.014337494969367981, -0.03009605035185814, -0.014591255225241184, -0.00600779103115201, -0.026797158643603325, -0.004443989135324955, 0.010689680464565754, 0.01912088878452778, -0.007898310199379921, -0.0009555686265230179, 0.032785918563604355, 0.01849917508661747, -0.017369939014315605, -0.0026216679252684116, -0.011140106245875359, 0.02258472517132759, -0.014781576581299305, 0.00793637428432703, 0.016469087451696396, 0.01848648674786091, 0.012592887505888939, -0.003911091014742851, 0.020998720079660416, 0.0015186007367447019, -0.007308315951377153, -0.014730824157595634, 0.0021236620377749205, 0.025439536198973656, 0.017369939014315605, 0.0011141692521050572, -0.012237622402608395, 0.027812201529741287, -0.03352182358503342, 0.0009436735999770463, 0.011140106245875359, 0.030425939708948135, -0.0028563968371599913, 0.014819640666246414, 0.02320644073188305, 0.005646181292831898, 0.013132129795849323, 0.011653972789645195, 0.0058523621410131454, 0.035044390708208084, -0.02811671420931816, 0.00240438524633646, -0.008989482186734676, 0.010575488209724426, -0.010810216888785362, 0.0035463094245642424, -0.008177447132766247, 0.005325808189809322, -0.009541411884129047, -0.005693761631846428, -0.0013528631534427404, 0.005179895553737879, -0.003581201657652855, -0.03245602920651436, 0.012459663674235344, 0.01196482963860035, 0.006423324346542358, 0.018651431426405907, -0.01984410732984543, 0.015999628230929375, -0.008228199556469917, -0.021150976419448853, -0.014845016412436962, -0.016012316569685936, 0.009858613833785057, -0.004999090917408466, -0.031212598085403442, 0.007225843612104654, -8.400875958614051e-05, -0.032608285546302795, 0.004548665136098862, -0.0030134115368127823, -0.00019051897106692195, 0.001505912747234106, 0.0112416110932827, 0.028700364753603935, -0.0022426124196499586, -0.009186146780848503, -0.031415607780218124, -0.05394957959651947, 0.008602497167885303, 0.003984047099947929, 0.011603220365941525, -0.009903022088110447, 0.0053955926559865475, -0.002764408476650715, 0.001549527863971889, -0.010860969312489033, -0.0041711959056556225, 0.005449516698718071, 0.0036065776366740465, 0.006692945491522551, 0.011311395093798637, -0.007441540248692036, -0.0029245950281620026, -0.019603034481406212, -0.00961754098534584, -0.004155335947871208, 0.0072639076970517635, -0.0254902895539999, -0.023282568901777267, 0.011533435434103012, -0.0033242686185985804, -0.0204785093665123, 0.00441861292347312, 0.011901388876140118, -0.01197117380797863, -0.036490827798843384, 0.01658327877521515, -0.018968632444739342, 0.011863324791193008, -0.0007216327358037233, -0.006451872643083334, 0.005078391171991825, -0.010892689228057861, -0.014007605612277985, 0.007213155273348093, 0.018346916884183884, 0.010093342512845993, 0.0067373537458479404, -0.010600863955914974, 0.016392959281802177, 0.010619896464049816, -0.03202463313937187, 0.023155687376856804, -0.008678625337779522, -0.02357439324259758, -0.017534881830215454, -0.021455489099025726, 0.004313936457037926, -0.0053733885288238525, -0.0010182159021496773, 0.035957928746938705, -0.03737898916006088, -0.02293999120593071, -0.02811671420931816, 0.007993470877408981, 0.01346201915293932, -0.0400688573718071, -0.00801884662359953, 0.004009423311799765, -0.007790461648255587, -0.013994917273521423, -0.019564971327781677, -0.1614934802055359, -0.005249680019915104, 0.013944164849817753, 0.010442263446748257, -0.005554193165153265, -0.011990205384790897, 0.04859522357583046, -0.016519838944077492, -0.026518020778894424, -0.006315476261079311, 0.009839581325650215, 0.0017826707335188985, -0.04308861121535301, 0.00127514882478863, -0.01733187399804592, 0.003828618675470352, -0.030324434861540794, 0.013842660933732986, 0.013956853188574314, 0.0018445249879732728, 0.012396222911775112, -0.007517668418586254, 0.02277504652738571, -0.005509784910827875, 0.003470181254670024, 0.022533973678946495, -0.00879916176199913, 0.01020119059830904, -0.005106939468532801, -0.018613366410136223, -0.025528352707624435, -0.012770520523190498, -0.003454321064054966, -0.013893412426114082, 0.0036509858909994364, 0.006217143498361111, -0.008723033592104912, 0.00030788342701271176, -0.01922239363193512, 0.03009605035185814, 0.02401847578585148, 0.0455247163772583, 0.029893042519688606, 0.009966461919248104, -0.004919790662825108, 0.008678625337779522, -0.0012854578671976924, 0.017116177827119827, 0.0036827060393989086, -0.0012101226020604372, -0.016938544809818268, -0.012104397639632225, 0.022698918357491493, -0.00673100957646966, 0.006236175540834665, 0.0012045715702697635, -0.0055129569955170155, 0.015263722278177738, -0.010594520717859268, 0.008488303981721401, -0.01941271312534809, -0.02773607335984707, 0.0052813999354839325, 0.0036319538485258818, -0.038317907601594925, -0.013766532763838768, -0.023358697071671486, 0.026340387761592865, -0.03425772860646248, -0.011419243179261684, -0.0061790794134140015, -0.007086275145411491, 0.007714333478361368, 0.008285295218229294, -0.005538333207368851, 0.010594520717859268, -0.017788643017411232, 0.0023108108434826136, 0.001172058517113328, 0.002060221740975976, -0.005506612826138735, 0.01437555905431509, 0.00036160938907414675, -0.0034828693605959415, 0.006572409067302942, -0.00399356335401535, 0.0014860876835882664, -0.00012697960482910275, 0.001632000203244388, 0.008171102963387966, -0.005455860868096352, -0.050295423716306686, -0.0030197554733604193, -0.017065424472093582, 0.004821458365768194, 0.02503351867198944, 0.012091710232198238, 0.005696933716535568, 0.0023028808645904064, 0.00784755777567625, -0.023688586428761482, -0.01437555905431509, -0.024830510839819908, 0.0023536330554634333, 0.008323359303176403, 0.010397855192422867, 0.019526906311511993, 0.012878368608653545, 0.024056538939476013, -0.000257527717621997, -0.008259919472038746, 0.025376096367836, 0.008742065168917179, 0.028801869601011276, -0.0032735164277255535, -0.0006189388805069029, 0.007517668418586254, -0.003505073254927993, -0.0001923032250488177, -0.0005130729405209422, 0.06750041991472244, 0.006252035964280367, -0.025096960365772247, -0.00437420466914773, -0.0067373537458479404, 0.02603587508201599, -0.10647810250520706, -0.031212598085403442, 0.010543768294155598, 0.007175091188400984, -0.011489028111100197, 0.021150976419448853, -0.0049515110440552235, 0.0021775863133370876, 0.003930123057216406, 0.028421228751540184, 0.011311395093798637, 0.0004908688715659082, 0.005541505292057991, -0.026974789798259735, 0.03329343721270561, 0.006233003921806812, 0.027989834547042847, -0.006102951243519783, -0.01950152963399887, 0.019273145124316216, -0.006055370904505253, 0.0032449683640152216, 0.0012767348671332002, -0.022610101848840714, -0.023054184392094612, -0.0052687120623886585, -0.028268972411751747, 0.01015043817460537, -0.0015526999486610293, -0.004063347354531288, 0.014007605612277985, -0.012047301977872849, 0.0032798603642731905, -0.0025534697342664003, 0.01360158808529377, 0.016354894265532494, 0.00011092129716416821, -0.0010324899340048432, 0.01333513855934143, -0.013157506473362446, 0.0036382977850735188, 0.0067373537458479404, 0.008310671895742416, -0.016621343791484833, 0.028345100581645966, -0.0066104731522500515, 0.016431022435426712, 0.01487039215862751, -0.004247324075549841, -0.003955499269068241, -0.02367589809000492, 0.001571731991134584, -0.022267524152994156, -0.016519838944077492, 0.039840470999479294, 0.00630278792232275, 0.01192676555365324, 0.0239296592772007, 0.009027546271681786, 0.025414161384105682, 0.02022474817931652, -0.013728468678891659, -0.0037049101665616035, 0.03400396928191185, 0.028903374448418617, -0.0026977963279932737, -0.023739337921142578, -0.0041616796515882015, 0.004548665136098862, -0.0034352890215814114, -0.023701274767518044, 0.03263365849852562, -0.014312118291854858, 0.0223182775080204, -0.010810216888785362, -0.0014131313655525446, -0.0068198260851204395, -0.04143916442990303, 0.015200281515717506, -0.011774509213864803, -0.02292730286717415, -0.02211526781320572, -0.0010507289553061128, -0.02447524480521679, 0.03156786412000656, 0.00702917855232954, -0.008342391811311245, 0.01975529082119465, -0.004656513687223196, 0.00621397141367197, 0.007790461648255587, 0.009313027374446392, 0.015479418449103832, -0.04816383123397827, -0.024449868127703667, 0.013969541527330875, 0.0028643268160521984, -0.00630596000701189, -0.0017033704789355397, 0.021163664758205414, -0.014642007648944855, -0.021024096757173538, -0.021442800760269165, 0.016164572909474373, -0.022686230018734932, -0.013944164849817753, 0.016976607963442802, 0.022077204659581184, -0.016532527282834053, -0.022369029000401497, -0.013347826898097992, -0.008088630624115467, -0.015466731041669846, 0.00938281137496233, -0.02077033556997776, 0.006020478904247284, 0.008323359303176403, -0.023980410769581795, -0.003831790527328849, -0.009141738526523113, 0.002329842885956168, 0.007035522721707821, 0.01986948400735855, 0.001134787336923182, 0.02574405074119568, -0.0008175860857591033, -0.011165482923388481, 0.002055463846772909, -0.0059221466071903706, 0.014109109528362751, -0.0302736833691597, -0.012459663674235344, -0.0049293069168925285, -0.035957928746938705, 0.004501085262745619, 0.01712886430323124, 0.0034796972759068012, -0.015479418449103832, 0.004726298153400421, -0.00544317252933979, 0.002729516476392746, 0.01959034614264965, -0.006420152261853218, -0.024602126330137253, 0.005459032952785492, -0.03491750732064247, -0.031872376799583435, 0.0035399654880166054, -0.002927767112851143, 0.00600461894646287, 0.024272236973047256, 6.998251774348319e-05, 0.01978066749870777, 0.01622801460325718, -0.010626240633428097, -0.015644364058971405, -0.0030562335159629583, -0.033318813890218735, 0.009255930781364441, -0.005551021080464125, -0.0006074403063394129, -0.03319193422794342, 0.01278955303132534, -0.003003895515576005, 0.007650893181562424, 0.02022474817931652, 0.015809308737516403, -0.003099055727943778, 0.00149005267303437, -0.02547760121524334, 0.013500083237886429, -0.02628963626921177, -0.015022649429738522, 0.020529262721538544, -0.003137120045721531, 0.0032798603642731905, 0.00510376738384366, 0.002497959416359663, -0.004037971142679453, 0.0021093881223350763, -0.01985679566860199, 0.006661225575953722, 0.02456406131386757, -0.011939452961087227, -0.006991114467382431, 0.019361961632966995, 0.016545215621590614, 0.006188595667481422, -0.020871838554739952, -0.017484130337834358, -0.0007902274955995381, 0.023079559206962585, -0.008145727217197418, -0.005363872274756432, -0.006528000812977552, 0.0034067409578710794, -0.012047301977872849, 0.001993609592318535, -0.024817822501063347, 0.01187601312994957, -0.0041838837787508965, 0.014020293019711971, 0.00332109653390944, 0.00031601169030182064, -0.01084193680435419, -0.030984213575720787, 0.0004611312469933182, 0.004777050111442804, -0.007225843612104654, -0.02239440567791462, -0.0018286649137735367, 0.0005130729405209422, -0.0005364665412344038, -0.016824351623654366, -0.010017214342951775, 0.013170193880796432, -0.004843662492930889, 0.031060341745615005, 0.0036573298275470734, -0.014045669697225094, -0.031491734087467194, 0.03344569355249405, 0.019564971327781677, -0.005493924953043461, 0.020529262721538544, -0.006410636473447084, 0.017166929319500923, 0.005275056231766939, 0.011806229129433632, -0.031415607780218124, -0.02963928133249283, 0.008107663132250309, -0.003857166739180684, -0.0013901343336328864, 0.000964291684795171, 0.003612921806052327, -0.009985494427382946, -0.025426847860217094, 0.011679348535835743, 0.03816564753651619, -0.0079871267080307, 0.056639447808265686, 0.030527444556355476, -0.016380270943045616, -0.003011825494468212, 0.003381364978849888, 0.013500083237886429, 0.012922776862978935, 0.016177261248230934, -0.03697297349572182, -0.01509877759963274, 0.009452596306800842, -0.007282939739525318, 0.019070137292146683, -0.03073045238852501, -0.017750579863786697, -0.027431560680270195, -0.031390231102705, -0.00960485264658928, -0.007701645139604807, -0.020009052008390427, 0.025046207010746002, 0.0015590438852086663, 0.005823814310133457, -0.004872210323810577, -0.01830885373055935, -0.024437181651592255, 0.03623706474900246, 0.0014662626199424267, -0.002385352971032262, -0.020706893876194954, -0.00399039126932621, -0.0020871839951723814, -0.020275501534342766, -0.014109109528362751, 0.029436271637678146, -0.0010665890295058489, 0.016570590436458588, 0.013144818134605885, 0.03207538649439812, -0.00523382006213069, -0.016570590436458588, 0.011520748026669025, -0.013880725018680096, -0.008532712236046791, 0.02593437023460865, 0.03933294862508774, -0.029131758958101273, 0.018879815936088562, -0.04113465175032616]\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "1536"
            ]
          },
          "execution_count": 18,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "vector = client.embeddings.create(input = [text], model=deployment_embedding_name).data[0].embedding\n",
        "print(vector)\n",
        "len(vector)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "gather": {
          "logged": 1674829555255
        },
        "jupyter": {
          "outputs_hidden": false,
          "source_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "The new movie is awesome\tThe dog plays in the garden\tScore: 0.7480\n",
            "The new movie is awesome\tThis recent movie is so good\tScore: 0.9191\n",
            "The new movie is awesome\tThe new movie is awesome\tScore: 1.0000\n"
          ]
        }
      ],
      "source": [
        "sentences1 = ['The new movie is awesome',  \n",
        "              'The new movie is awesome',  \n",
        "              'The new movie is awesome']  \n",
        "  \n",
        "sentences2 = ['The dog plays in the garden',  \n",
        "              'This recent movie is so good',  \n",
        "              'The new movie is awesome']  \n",
        "\n",
        "embeddings1 = [client.embeddings.create(input = s, model=deployment_embedding_name).data[0].embedding for s in sentences1]  \n",
        "embeddings2 = [client.embeddings.create(input = s, model=deployment_embedding_name).data[0].embedding for s in sentences2]  \n",
        "  \n",
        "for i in range(len(sentences1)):  \n",
        "    print(\"{}\\t{}\\tScore: {:.4f}\".format(sentences1[i], sentences2[i], cosine_similarity(embeddings1[i], embeddings2[i])))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "다음은 한글 문장을 임베딩 API를 이용하여 벡터 값으로 바꾼 후 비교한 결과입니다.  \n",
        "유사도 점수가 어떤 차이가 날 수 있는지를 확인 합니다."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t바나나 우유는 목욕 후에 마시면 더 맛있습니다.\tScore: 1.0000\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t바나나 우유는 후에 목욕 마시면 더 맛있습니다.\tScore: 0.9920\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t목욕 후에 바나나 우유를 마시면 더 맛있습니다.\tScore: 0.9726\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t맛있습니다 더 마시면 후에 목욕 우유는 바나나\tScore: 0.9393\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t딸기 우유는 목욕 전에 마시면 더 맛있습니다.\tScore: 0.9259\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t우유는 시리얼이랑 먹으면 더 고소한 맛이 납니다.\tScore: 0.8716\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t어제는 비랑 눈이 와서 날씨가 더 추웠습니다.\tScore: 0.7890\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\t다가오는 2024 OpenAI DevDay는 직관할 수 있을까?\tScore: 0.7240\n",
            "바나나 우유는 목욕 후에 마시면 더 맛있습니다.\tABCDE FGHI JKL MNO PQR STU VWXYZ ABCDEFG\tScore: 0.6949\n"
          ]
        }
      ],
      "source": [
        "sentences1 = [\n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "            ]\n",
        "  \n",
        "sentences2 = [\n",
        "              '바나나 우유는 목욕 후에 마시면 더 맛있습니다.',  \n",
        "              '바나나 우유는 후에 목욕 마시면 더 맛있습니다.',  \n",
        "              '목욕 후에 바나나 우유를 마시면 더 맛있습니다.',\n",
        "              '맛있습니다 더 마시면 후에 목욕 우유는 바나나',  \n",
        "              '딸기 우유는 목욕 전에 마시면 더 맛있습니다.',  \n",
        "              '우유는 시리얼이랑 먹으면 더 고소한 맛이 납니다.',\n",
        "              '어제는 비랑 눈이 와서 날씨가 더 추웠습니다.',\n",
        "              '다가오는 2024 OpenAI DevDay는 직관할 수 있을까?',\n",
        "              'ABCDE FGHI JKL MNO PQR STU VWXYZ ABCDEFG',\n",
        "            ]\n",
        "\n",
        "embeddings1 = [client.embeddings.create(input = s, model=deployment_embedding_name).data[0].embedding for s in sentences1]  \n",
        "embeddings2 = [client.embeddings.create(input = s, model=deployment_embedding_name).data[0].embedding for s in sentences2]  \n",
        "  \n",
        "for i in range(len(sentences1)):  \n",
        "    print(\"{}\\t{}\\tScore: {:.4f}\".format(sentences1[i], sentences2[i], cosine_similarity(embeddings1[i], embeddings2[i])))"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      },
      "source": [
        "# 궁금증 있을 때 연락 주세요.  \n",
        "Prompt Engineering과 관련 궁금증이 있을 경우 아래 이메일로 요청 주세요.  \n",
        "MS Korea 김현수 이메일: [<hyounsookim@microsoft.com>](hyounsookim@microsoft.com)"
      ]
    }
  ],
  "metadata": {
    "kernel_info": {
      "name": "python310-sdkv2"
    },
    "kernelspec": {
      "display_name": "base",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.11.4"
    },
    "microsoft": {
      "host": {
        "AzureML": {
          "notebookHasBeenCompleted": true
        }
      }
    },
    "nteract": {
      "version": "nteract-front-end@1.0.0"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 2
}