{"id":1356,"date":"2024-05-06T01:31:11","date_gmt":"2024-05-06T01:31:11","guid":{"rendered":"https:\/\/www.nicekj.com\/?p=1356"},"modified":"2024-05-06T01:31:38","modified_gmt":"2024-05-06T01:31:38","slug":"langchainxiliejiaochengtishigongchenghelangchainzhongdellms","status":"publish","type":"post","link":"https:\/\/www.nicekj.com\/langchainxiliejiaochengtishigongchenghelangchainzhongdellms.html","title":{"rendered":"LangChain\u7cfb\u5217\u6559\u7a0b\uff1a\u63d0\u793a\u5de5\u7a0b\u548cLangchain\u4e2d\u7684LLMs"},"content":{"rendered":"

\u5728\u673a\u5668\u5b66\u4e60\u4e2d\uff0c\u6211\u4eec\u8fc7\u53bb\u603b\u662f\u4f9d\u8d56\u4e0d\u540c\u7684\u6a21\u578b\u6765\u5b8c\u6210\u4e0d\u540c\u7684\u4efb\u52a1\u3002\u968f\u7740\u591a\u6a21\u6001\u548c\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u7684\u5f15\u5165\uff0c\u8fd9\u79cd\u60c5\u51b5\u53d1\u751f\u4e86\u53d8\u5316\u3002<\/p>\n

\u8fc7\u53bb\uff0c\u6211\u4eec\u9700\u8981\u4e3a\u5206\u7c7b\u3001\u547d\u540d\u5b9e\u4f53\u8bc6\u522b\uff08NER\uff09\u3001\u95ee\u7b54\uff08QA\uff09\u548c\u8bb8\u591a\u5176\u4ed6\u4efb\u52a1\u4f7f\u7528\u5355\u72ec\u7684\u6a21\u578b\u7684\u65e5\u5b50\u5df2\u7ecf\u4e00\u53bb\u4e0d\u590d\u8fd4\u3002<\/p>\n

\n

\"image.png\"<\/figure>\n<\/p>\n

\u968f\u7740\u53d8\u6362\u5668\uff08transformers\uff09\u548c\u8fc1\u79fb\u5b66\u4e60\uff08transfer learning\uff09\u7684\u5f15\u5165\uff0c\u8981\u4f7f\u8bed\u8a00\u6a21\u578b\u9002\u5e94\u4e0d\u540c\u7684\u4efb\u52a1\uff0c\u53ea\u9700\u8981\u5728\u7f51\u7edc\u7684\u672b\u7aef\uff08\u5373\u5934\u90e8\uff09\u6dfb\u52a0\u4e00\u4e9b\u5c0f\u7684\u5c42\u6b21\u4ee5\u53ca\u4e00\u4e9b\u5fae\u8c03\u3002<\/p>\n

\n

\"image.png\"<\/figure>\n<\/p>\n

\u5982\u4eca\uff0c\u751a\u81f3\u8fd9\u79cd\u65b9\u6cd5\u4e5f\u5df2\u7ecf\u8fc7\u65f6\u4e86\u3002\u4e3a\u4ec0\u4e48\u8981\u66f4\u6539\u8fd9\u4e9b\u6700\u540e\u51e0\u4e2a\u6a21\u578b\u5c42\uff0c\u5e76\u7ecf\u8fc7\u6574\u4e2a\u5fae\u8c03\u8fc7\u7a0b\uff0c\u5f53\u60a8\u53ef\u4ee5\u76f4\u63a5\u5411\u6a21\u578b\u63d0\u51fa\u5206\u7c7b\u6216\u95ee\u7b54\u7684\u8bf7\u6c42\u65f6\u3002<\/p>\n

\n

\"image.png\"<\/figure>\n<\/p>\n

\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u53ef\u4ee5\u6267\u884c\u6240\u6709\u8fd9\u4e9b\u4efb\u52a1\u4ee5\u53ca\u66f4\u591a\u3002\u8fd9\u4e9b\u6a21\u578b\u5df2\u7ecf\u7ecf\u8fc7\u7b80\u5355\u7684\u6982\u5ff5\u8bad\u7ec3\uff0c\u60a8\u8f93\u5165\u4e00\u7cfb\u5217\u6587\u672c\uff0c\u6a21\u578b\u8f93\u51fa\u4e00\u7cfb\u5217\u6587\u672c\u3002\u552f\u4e00\u53ef\u53d8\u7684\u662f\u8f93\u5165\u6587\u672c\uff0c\u4e5f\u5c31\u662f\u63d0\u793a\uff08prompt\uff09\u3002<\/p>\n

\u5728\u8fd9\u4e2a\u65b0\u65f6\u4ee3\u7684LLMs\u4e2d\uff0c\u63d0\u793a\uff08prompt\uff09\u81f3\u5173\u91cd\u8981\u3002\u7cdf\u7cd5\u7684\u63d0\u793a\u4f1a\u4ea7\u751f\u7cdf\u7cd5\u7684\u8f93\u51fa\uff0c\u800c\u597d\u7684\u63d0\u793a\u5219\u975e\u5e38\u5f3a\u5927\u3002\u6784\u5efa\u826f\u597d\u7684\u63d0\u793a\u662f\u90a3\u4e9b\u4f7f\u7528LLMs\u8fdb\u884c\u6784\u5efa\u7684\u4eba\u7684\u91cd\u8981\u6280\u80fd\u3002<\/p>\n

LangChain\u5e93\u8ba4\u8bc6\u5230\u63d0\u793a\u7684\u5f3a\u5927\u4e4b\u5904\uff0c\u5e76\u4e3a\u5176\u6784\u5efa\u4e86\u4e00\u6574\u5957\u5bf9\u8c61\u3002\u5728\u672c\u6587\u4e2d\uff0c\u6211\u4eec\u5c06\u5b66\u4e60\u5173\u4e8ePromptTemplates\u4ee5\u53ca\u5982\u4f55\u6709\u6548\u5b9e\u65bd\u5b83\u4eec\u7684\u6240\u6709\u77e5\u8bc6\u3002<\/p>\n

\u63d0\u793a\u5de5\u7a0b<\/h1>\n

\u5728\u6df1\u5165\u7814\u7a76Langchain\u7684PromptTemplate\u4e4b\u524d\uff0c\u6211\u4eec\u9700\u8981\u66f4\u597d\u5730\u7406\u89e3\u63d0\u793a\u4ee5\u53ca\u63d0\u793a\u5de5\u7a0b\u7684\u5b66\u79d1\u3002<\/p>\n

\u4e00\u4e2a\u63d0\u793a\u901a\u5e38\u7531\u591a\u4e2a\u90e8\u5206\u7ec4\u6210\uff1a<\/p>\n

\n

\"image.png\"<\/figure>\n<\/p>\n

\u5e76\u4e0d\u662f\u6240\u6709\u7684\u63d0\u793a\u90fd\u4f7f\u7528\u8fd9\u4e9b\u7ec4\u4ef6\uff0c\u4f46\u4e00\u4e2a\u597d\u7684\u63d0\u793a\u901a\u5e38\u4f1a\u4f7f\u7528\u4e24\u4e2a\u6216\u66f4\u591a\u3002\u8ba9\u6211\u4eec\u66f4\u7cbe\u786e\u5730\u5b9a\u4e49\u5b83\u4eec\u3002<\/p>\n

    \n
  1. \u6307\u4ee4\uff08Instructions\uff09\u544a\u8bc9\u6a21\u578b\u8981\u505a\u4ec0\u4e48\uff0c\u5982\u4f55\u4f7f\u7528\u63d0\u4f9b\u7684\u5916\u90e8\u4fe1\u606f\uff0c\u5982\u4f55\u5904\u7406\u67e5\u8be2\uff0c\u4ee5\u53ca\u5982\u4f55\u6784\u5efa\u8f93\u51fa\u3002<\/li>\n
  2. \u5916\u90e8\u4fe1\u606f\u6216\u4e0a\u4e0b\u6587\uff08External information or context(s)\uff09\u5145\u5f53\u6a21\u578b\u7684\u989d\u5916\u77e5\u8bc6\u6765\u6e90\u3002\u8fd9\u4e9b\u4fe1\u606f\u53ef\u4ee5\u624b\u52a8\u63d2\u5165\u63d0\u793a\u4e2d\uff0c\u901a\u8fc7\u5411\u91cf\u6570\u636e\u5e93\u68c0\u7d22\uff08\u68c0\u7d22\u589e\u5f3a\uff09\u83b7\u53d6\uff0c\u6216\u901a\u8fc7\u5176\u4ed6\u65b9\u5f0f\uff08API\u3001\u8ba1\u7b97\u7b49\uff09\u5f15\u5165\u3002<\/li>\n
  3. \u7528\u6237\u8f93\u5165\u6216\u67e5\u8be2\uff08User input or query\uff09\u901a\u5e38\u662f\u7531\u4eba\u7c7b\u7528\u6237\uff08\u63d0\u793a\u8005\uff09\u8f93\u5165\u7cfb\u7edf\u7684\u67e5\u8be2\uff08\u4e0d\u8fc7\u5e76\u4e0d\u603b\u662f\u5982\u6b64\uff09\u3002<\/li>\n
  4. \u8f93\u51fa\u6307\u793a\u7b26\uff08Output indicator\uff09\u6807\u8bb0\u4e86\u5f85\u751f\u6210\u6587\u672c\u7684\u5f00\u5934\u3002\u5982\u679c\u8981\u751f\u6210Python\u4ee3\u7801\uff0c\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528import\u6765\u544a\u8bc9\u6a21\u578b\u5fc5\u987b\u5f00\u59cb\u7f16\u5199Python\u4ee3\u7801\uff08\u56e0\u4e3a\u5927\u591a\u6570Python\u811a\u672c\u90fd\u4ee5import\u5f00\u5934\uff09\u3002<\/li>\n<\/ol>\n

    \u901a\u5e38\uff0c\u8fd9\u4e9b\u7ec4\u4ef6\u6309\u7167\u4e0a\u8ff0\u987a\u5e8f\u653e\u7f6e\u5728\u63d0\u793a\u4e2d\uff0c\u4ece\u6307\u4ee4\u5f00\u59cb\uff0c\u7136\u540e\u662f\u5916\u90e8\u4fe1\u606f\uff08\u5982\u679c\u9002\u7528\uff09\uff0c\u7136\u540e\u662f\u7528\u6237\u8f93\u5165\uff0c\u6700\u540e\u662f\u8f93\u51fa\u6307\u793a\u7b26\u3002<\/p>\n

    \u8ba9\u6211\u4eec\u770b\u770b\u5982\u4f55\u4f7f\u7528Langchain\u5c06\u8fd9\u4e9b\u5185\u5bb9\u8f93\u5165\u5230OpenAI\u6a21\u578b\u4e2d\uff1a<\/p>\n

    In[5]:<\/p>\n

    <\/div>
    ini<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>prompt<\/span> = \"\"\"Answer the question based on the context below. If the<\/span><\/span>\nquestion cannot be answered using the information provided answer<\/span>\nwith \"I don't know\".<\/span>\n<\/span>\nContext: Large Language Models (LLMs) are the latest models used in NLP.<\/span>\nTheir superior performance over smaller models has made them incredibly<\/span>\nuseful for developers building NLP enabled applications. These models<\/span>\ncan be accessed via Hugging Face's `transformers` library, via OpenAI<\/span>\nusing the `openai` library, and via Cohere using the `cohere` library.<\/span>\n<\/span>\nQuestion: Which libraries and model providers offer LLMs?<\/span>\n<\/span>\nAnswer: \"\"\"<\/span>\n<\/code><\/pre>\n
    In[6]:<\/p>\n
    <\/div>
    ini<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>from langchain.llms import OpenAI<\/span>\n<\/span>\n# initialize the models<\/span><\/span>\nopenai<\/span> = OpenAI(<\/span>\n    model_name<\/span>=\"text-davinci-003\"<\/span>,<\/span>\n    openai_api_key<\/span>=\"YOUR_API_KEY\"<\/span><\/span>\n)<\/span>\n<\/code><\/pre>\n
    In[7]:<\/p>\n
    <\/div>
    scss<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>print<\/span>(openai(prompt))<\/span>\n<\/code><\/pre>\n
    Out[7]:<\/p>\n
    <\/div>
    markdown<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> Hugging Face's `transformers`<\/span> library, OpenAI using the `openai`<\/span> library, and Cohere using the `cohere`<\/span> library.<\/span>\n<\/code><\/pre>\n
    \u5b9e\u9645\u4e0a\uff0c\u6211\u4eec\u4e0d\u592a\u53ef\u80fd\u786c\u7f16\u7801\u4e0a\u4e0b\u6587\u548c\u7528\u6237\u95ee\u9898\u3002\u6211\u4eec\u5c06\u901a\u8fc7\u4e00\u4e2a\u6a21\u677f\u6765\u8f93\u5165\u5b83\u4eec\uff0c\u8fd9\u5c31\u662fLangchain\u7684PromptTemplate\u53d1\u6325\u4f5c\u7528\u7684\u5730\u65b9\u3002<\/p>\n
    \u63d0\u793a\u6a21\u7248<\/h1>\n
    Langchain\u4e2d\u7684\u63d0\u793a\u6a21\u677f\u7c7b\u662f\u4e3a\u4e86\u66f4\u5bb9\u6613\u6784\u5efa\u5177\u6709\u52a8\u6001\u8f93\u5165\u7684\u63d0\u793a\u800c\u6784\u5efa\u7684\u3002\u5176\u4e2d\u6700\u7b80\u5355\u7684\u7c7b\u662fPromptTemplate\u3002\u6211\u4eec\u5c06\u901a\u8fc7\u5411\u6211\u4eec\u4e4b\u524d\u7684\u63d0\u793a\u6dfb\u52a0\u4e00\u4e2a\u52a8\u6001\u8f93\u5165\uff0c\u5373\u7528\u6237\u67e5\u8be2\uff0c\u6765\u6d4b\u8bd5\u5b83\u3002<\/p>\n
    <\/div>
    ini<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>from langchain import PromptTemplate<\/span>\n<\/span>\ntemplate<\/span> = \"\"\"Answer the question based on the context below. If the<\/span><\/span>\nquestion cannot be answered using the information provided answer<\/span>\nwith \"I don't know\".<\/span>\n<\/span>\nContext: Large Language Models (LLMs) are the latest models used in NLP.<\/span>\nTheir superior performance over smaller models has made them incredibly<\/span>\nuseful for developers building NLP enabled applications. These models<\/span>\ncan be accessed via Hugging Face's `transformers` library, via OpenAI<\/span>\nusing the `openai` library, and via Cohere using the `cohere` library.<\/span>\n<\/span>\nQuestion: {query}<\/span>\n<\/span>\nAnswer: \"\"\"<\/span>\n<\/span>\nprompt_template<\/span> = PromptTemplate(<\/span>\n    input_variables<\/span>=[\"query\"<\/span>],<\/span>\n    template<\/span>=template<\/span>\n)<\/span>\n<\/code><\/pre>\n
    \u6709\u4e86\u8fd9\u4e2a\uff0c\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528format<\/code>\u65b9\u6cd5\u6765\u67e5\u770b\u5c06\u67e5\u8be2\u4f20\u9012\u7ed9\u6a21\u677f\u7684\u6548\u679c\u3002<\/p>\n
    In[9]:<\/p>\n
    <\/div>
    ini<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>print(<\/span>\n    prompt_template.format(<\/span>\n        query<\/span>=\"Which libraries and model providers offer LLMs?\"<\/span><\/span>\n    )<\/span>\n)<\/span>\n<\/code><\/pre>\n
    Out[9]:<\/p>\n
    <\/div>
    vbnet<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>Answer the question based on<\/span> the context below. If<\/span> the<\/span>\nquestion cannot be answered using<\/span> the information provided answer<\/span>\nwith<\/span> \"I don't know\"<\/span>.<\/span>\n<\/span>\nContext:<\/span> Large Language Models (LLMs) are the latest models used in<\/span> NLP.<\/span>\nTheir superior performance over smaller models has made them incredibly<\/span>\nuseful for<\/span> developers building NLP enabled applications. These models<\/span>\ncan be accessed via Hugging Face's `transformers` library, via OpenAI<\/span><\/span>\nusing<\/span> the `openai` library, and<\/span> via Cohere using<\/span> the `cohere` library.<\/span>\n<\/span>\nQuestion:<\/span> Which libraries and<\/span> model providers offer LLMs?<\/span>\n<\/span>\nAnswer:<\/span> <\/span>\n<\/code><\/pre>\n
    \u5f53\u7136\uff0c\u6211\u4eec\u53ef\u4ee5\u76f4\u63a5\u5c06\u8fd9\u4e2a\u8f93\u51fa\u4f20\u9012\u7ed9LLM\u5bf9\u8c61\uff0c\u5982\u4e0b\u6240\u793a\uff1a<\/p>\n
    In[10]:<\/p>\n
    <\/div>
    ini<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>print(openai(<\/span>\n    prompt_template.format(<\/span>\n        query<\/span>=\"Which libraries and model providers offer LLMs?\"<\/span><\/span>\n    )<\/span>\n))<\/span>\n<\/code><\/pre>\n
    Out[10]:<\/p>\n
    <\/div>
    markdown<\/span>
    \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> Hugging Face's `transformers`<\/span> library, OpenAI using the `openai`<\/span> library, and Cohere using the `cohere`<\/span> library.<\/span>\n<\/code><\/pre>\n
    \u8fd9\u53ea\u662f\u4e00\u4e2a\u7b80\u5355\u7684\u5b9e\u73b0\uff0c\u53ef\u4ee5\u8f7b\u677e\u5730\u7528f-strings\uff08\u4f8b\u5982f”\u63d2\u5165\u4e00\u4e9b\u81ea\u5b9a\u4e49\u6587\u672c ‘{custom_text}’ \u7b49”) \u6765\u66ff\u4ee3\u3002\u7136\u800c\uff0c\u4f7f\u7528Langchain\u7684PromptTemplate\u5bf9\u8c61\uff0c\u6211\u4eec\u53ef\u4ee5\u89c4\u8303\u8fd9\u4e2a\u8fc7\u7a0b\uff0c\u6dfb\u52a0\u591a\u4e2a\u53c2\u6570\uff0c\u5e76\u4ee5\u9762\u5411\u5bf9\u8c61\u7684\u65b9\u5f0f\u6784\u5efa\u63d0\u793a\u3002<\/p>\n
    \u8fd9\u4e9b\u662fLangchain\u63d0\u4f9b\u7684\u4e00\u4e9b\u91cd\u8981\u4f18\u52bf\uff0c\u4f46\u53ea\u662fLangchain\u4e3a\u5e2e\u52a9\u6211\u4eec\u5904\u7406\u63d0\u793a\u63d0\u4f9b\u7684\u90e8\u5206\u529f\u80fd\u3002<\/p>\n
    Few Shot\u63d0\u793a\u6a21\u677f<\/h1>\n
    LLMs\u7684\u6210\u529f\u6765\u81ea\u4e8e\u5b83\u4eec\u7684\u5927\u5c3a\u5bf8\u4ee5\u53ca\u5728\u6a21\u578b\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u5b66\u4e60\u5230\u7684\u201c\u77e5\u8bc6\u201d\u80fd\u529b\uff0c\u8fd9\u4e9b\u77e5\u8bc6\u5b58\u50a8\u5728\u6a21\u578b\u53c2\u6570\u4e2d\u3002\u7136\u800c\uff0c\u8fd8\u6709\u66f4\u591a\u5c06\u77e5\u8bc6\u4f20\u9012\u7ed9LLM\u7684\u65b9\u6cd5\u3002\u4e3b\u8981\u7684\u4e24\u79cd\u65b9\u6cd5\u662f\uff1a<\/p>\n
      \n
    1. \u53c2\u6570\u5316\u77e5\u8bc6\uff08Parametric knowledge\uff09\u2014\u2014\u4e0a\u9762\u63d0\u5230\u7684\u77e5\u8bc6\u662f\u5728\u6a21\u578b\u8bad\u7ec3\u671f\u95f4\u5b66\u5230\u7684\uff0c\u5b58\u50a8\u5728\u6a21\u578b\u6743\u91cd\uff08\u6216\u53c2\u6570\uff09\u4e2d\u3002<\/li>\n
    2. \u6765\u6e90\u77e5\u8bc6\uff08Source knowledge\uff09\u2014\u2014\u901a\u8fc7\u8f93\u5165\u63d0\u793a\u5728\u63a8\u7406\u65f6\u63d0\u4f9b\u7ed9\u6a21\u578b\u7684\u4efb\u4f55\u77e5\u8bc6\u3002<\/li>\n<\/ol>\n
      Langchain\u7684FewShotPromptTemplate\u9002\u7528\u4e8e\u6e90\u77e5\u8bc6\u8f93\u5165\u3002\u5176\u601d\u60f3\u662f\u901a\u8fc7\u5c11\u91cf\u793a\u4f8b\u6765\u201c\u8bad\u7ec3\u201d\u6a21\u578b\uff0c\u6211\u4eec\u79f0\u4e4b\u4e3aFew-shot\u5b66\u4e60\uff0c\u8fd9\u4e9b\u793a\u4f8b\u5728\u63d0\u793a\u4e2d\u63d0\u4f9b\u7ed9\u6a21\u578b\u3002<\/p>\n
      Few-shot\u5b66\u4e60\u5728\u6211\u4eec\u7684\u6a21\u578b\u9700\u8981\u5e2e\u52a9\u7406\u89e3\u6211\u4eec\u8981\u6c42\u5b83\u505a\u4ec0\u4e48\u65f6\u975e\u5e38\u6709\u6548\u3002\u6211\u4eec\u53ef\u4ee5\u5728\u4ee5\u4e0b\u793a\u4f8b\u4e2d\u770b\u5230\u8fd9\u4e00\u70b9\uff1a<\/p>\n
      In[12]:<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>prompt<\/span> = \"\"\"The following is a conversation with an AI assistant.<\/span><\/span>\nThe assistant is typically sarcastic and witty, producing creative <\/span>\nand funny responses to the users questions. Here are some examples: <\/span>\n<\/span>\nUser: What is the meaning of life?<\/span>\nAI: \"\"\"<\/span>\n<\/span>\nopenai.temperature<\/span> = 1.0<\/span>  # increase creativity\/randomness of output<\/span><\/span>\n<\/span>\nprint(openai(prompt))<\/span>\n<\/code><\/pre>\n
      Out[12]:<\/p>\n
      <\/div>
      vbnet<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> Life is<\/span> like<\/span> a box of<\/span> chocolates, you never know what you're gonna get!<\/span><\/span>\n<\/code><\/pre>\n
      \u5728\u8fd9\u79cd\u60c5\u51b5\u4e0b\uff0c\u6211\u4eec\u63d0\u51fa\u4e86\u4e00\u4e2a\u4e25\u8083\u7684\u95ee\u9898\uff0c\u5e0c\u671b\u5f97\u5230\u4e00\u4e9b\u6709\u8da3\u7684\u56de\u7b54\uff0c\u6bd4\u5982\u7b11\u8bdd\u3002\u7136\u800c\uff0c\u5373\u4f7f\u5c06\u6e29\u5ea6\u8bbe\u7f6e\u4e3a1.0\uff08\u589e\u52a0\u968f\u673a\u6027\/\u521b\u610f\u6027\uff09\uff0c\u6211\u4eec\u4ecd\u7136\u5f97\u5230\u4e86\u4e25\u8083\u7684\u56de\u7b54\u3002<\/p>\n
      \u4e3a\u4e86\u5e2e\u52a9\u6a21\u578b\uff0c\u6211\u4eec\u53ef\u4ee5\u7ed9\u5b83\u4e00\u4e9b\u6211\u4eec\u60f3\u8981\u7684\u56de\u7b54\u7c7b\u578b\u7684\u793a\u4f8b\uff1a<\/p>\n
      In[13]:<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>prompt<\/span> = \"\"\"The following are exerpts from conversations with an AI<\/span><\/span>\nassistant. The assistant is typically sarcastic and witty, producing<\/span>\ncreative  and funny responses to the users questions. Here are some<\/span>\nexamples: <\/span>\n<\/span>\nUser: How are you?<\/span>\nAI: I can't complain but sometimes I still do.<\/span>\n<\/span>\nUser: What time is it?<\/span>\nAI: It's time to get a watch.<\/span>\n<\/span>\nUser: What is the meaning of life?<\/span>\nAI: \"\"\"<\/span>\n<\/span>\nprint(openai(prompt))<\/span>\n<\/code><\/pre>\n
      Out[13]:<\/p>\n
      <\/div>
      <\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> 42, of course!<\/span>\n<\/code><\/pre>\n
      \u901a\u8fc7\u793a\u4f8b\u6765\u5f3a\u5316\u6211\u4eec\u5728\u63d0\u793a\u4e2d\u4f20\u9012\u7684\u6307\u4ee4\uff0c\u6211\u4eec\u66f4\u6709\u53ef\u80fd\u5f97\u5230\u4e00\u4e2a\u66f4\u6709\u8da3\u7684\u56de\u5e94\u3002\u7136\u540e\uff0c\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528Langchain\u7684FewShotPromptTemplate\u6765\u89c4\u8303\u5316\u8fd9\u4e2a\u8fc7\u7a0b\uff1a<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>from langchain import FewShotPromptTemplate<\/span>\n<\/span>\n# create our examples<\/span><\/span>\nexamples<\/span> = [<\/span>\n    {<\/span>\n        \"query\"<\/span>: \"How are you?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"I can't complain but sometimes I still do.\"<\/span><\/span>\n    }, {<\/span>\n        \"query\"<\/span>: \"What time is it?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"It's time to get a watch.\"<\/span><\/span>\n    }<\/span>\n]<\/span>\n<\/span>\n# create a example template<\/span><\/span>\nexample_template<\/span> = \"\"\"<\/span><\/span>\nUser: {query}<\/span>\nAI: {answer}<\/span>\n\"\"\"<\/span>\n<\/span>\n# create a prompt example from above template<\/span><\/span>\nexample_prompt<\/span> = PromptTemplate(<\/span>\n    input_variables<\/span>=[\"query\"<\/span>, \"answer\"<\/span>],<\/span>\n    template<\/span>=example_template<\/span>\n)<\/span>\n<\/span>\n# now break our previous prompt into a prefix and suffix<\/span><\/span>\n# the prefix is our instructions<\/span><\/span>\nprefix<\/span> = \"\"\"The following are exerpts from conversations with an AI<\/span><\/span>\nassistant. The assistant is typically sarcastic and witty, producing<\/span>\ncreative  and funny responses to the users questions. Here are some<\/span>\nexamples: <\/span>\n\"\"\"<\/span>\n# and the suffix our user input and output indicator<\/span><\/span>\nsuffix<\/span> = \"\"\"<\/span><\/span>\nUser: {query}<\/span>\nAI: \"\"\"<\/span>\n<\/span>\n# now create the few shot prompt template<\/span><\/span>\nfew_shot_prompt_template<\/span> = FewShotPromptTemplate(<\/span>\n    examples<\/span>=examples,<\/span>\n    example_prompt<\/span>=example_prompt,<\/span>\n    prefix<\/span>=prefix,<\/span>\n    suffix<\/span>=suffix,<\/span>\n    input_variables<\/span>=[\"query\"<\/span>],<\/span>\n    example_separator<\/span>=\"nn\"<\/span><\/span>\n)<\/span>\n<\/code><\/pre>\n
      \u5982\u679c\u6211\u4eec\u968f\u540e\u4f20\u9012\u793a\u4f8b\u548c\u7528\u6237\u67e5\u8be2\uff0c\u6211\u4eec\u5c06\u5f97\u5230\u8fd9\u4e2a\u7ed3\u679c\uff1a<\/p>\n
      In[15]:<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>query<\/span> = \"What is the meaning of life?\"<\/span><\/span>\n<\/span>\nprint(few_shot_prompt_template.format(query<\/span>=query))<\/span>\n<\/code><\/pre>\n
      Out[15]:<\/p>\n
      <\/div>
      vbnet<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>The following are exerpts from<\/span> conversations with<\/span> an AI<\/span>\nassistant. The assistant is<\/span> typically sarcastic and<\/span> witty, producing<\/span>\ncreative  and<\/span> funny responses to<\/span> the users questions. Here are some<\/span>\nexamples:<\/span> <\/span>\n<\/span>\n<\/span>\n<\/span>\nUser:<\/span> How are you?<\/span>\nAI:<\/span> I can't complain but sometimes I still do.<\/span><\/span>\n<\/span>\n<\/span>\n<\/span>\nUser:<\/span> What time is<\/span> it?<\/span>\nAI:<\/span> It's time to get a watch.<\/span><\/span>\n<\/span>\n<\/span>\n<\/span>\nUser:<\/span> What is<\/span> the meaning of<\/span> life?<\/span>\nAI:<\/span> <\/span>\n<\/code><\/pre>\n
      \u8fd9\u4e2a\u8fc7\u7a0b\u53ef\u80fd\u770b\u8d77\u6765\u6709\u4e9b\u590d\u6742\u3002\u4e3a\u4ec0\u4e48\u8981\u4f7f\u7528FewShotPromptTemplate\u5bf9\u8c61\u3001\u793a\u4f8b\u5b57\u5178\u7b49\u7b49\uff0c\u5f53\u6211\u4eec\u53ef\u4ee5\u7528\u51e0\u884c\u4ee3\u7801\u548c\u4e00\u4e2af-string\u6765\u5b8c\u6210\u76f8\u540c\u7684\u4efb\u52a1\u5462\uff1f<\/p>\n
      \u518d\u6b21\u5f3a\u8c03\uff0c\u8fd9\u79cd\u65b9\u6cd5\u66f4\u52a0\u89c4\u8303\u5316\uff0c\u4e0eLangchain\u4e2d\u7684\u5176\u4ed6\u529f\u80fd\u5f88\u597d\u5730\u96c6\u6210\uff08\u6bd4\u5982chains\u2014\u2014\u7a0d\u540e\u4f1a\u8be6\u7ec6\u4ecb\u7ecd\uff09\uff0c\u5e76\u4e14\u5177\u5907\u591a\u4e2a\u529f\u80fd\u3002\u5176\u4e2d\u4e4b\u4e00\u662f\u6839\u636e\u67e5\u8be2\u957f\u5ea6\u6765\u53d8\u5316\u5305\u542b\u7684\u793a\u4f8b\u6570\u91cf\u3002<\/p>\n
      \u52a8\u6001\u6570\u91cf\u7684\u793a\u4f8b\u5f88\u91cd\u8981\uff0c\u56e0\u4e3a\u6211\u4eec\u7684\u63d0\u793a\u548c\u5b8c\u6210\u8f93\u51fa\u7684\u6700\u5927\u957f\u5ea6\u53d7\u5230\u9650\u5236\u3002\u8fd9\u4e2a\u9650\u5236\u662f\u7531\u6700\u5927\u4e0a\u4e0b\u6587\u7a97\u53e3\u6765\u8861\u91cf\u7684\u3002<\/p>\n
      c<\/mi>o<\/mi>n<\/mi>t<\/mi>e<\/mi>x<\/mi>t<\/mi> <\/mtext>w<\/mi>i<\/mi>n<\/mi>d<\/mi>o<\/mi>w<\/mi>=<\/mo>i<\/mi>n<\/mi>p<\/mi>u<\/mi>t<\/mi> <\/mtext>t<\/mi>o<\/mi>k<\/mi>e<\/mi>n<\/mi>s<\/mi>+<\/mo>o<\/mi>u<\/mi>t<\/mi>p<\/mi>u<\/mi>t<\/mi> <\/mtext>t<\/mi>o<\/mi>k<\/mi>e<\/mi>n<\/mi>s<\/mi><\/mrow>context window=input tokens+output tokens<\/annotation><\/semantics><\/math><\/span><\/span>co<\/span>n<\/span>t<\/span>e<\/span>x<\/span>t<\/span> <\/span>w<\/span>in<\/span>d<\/span>o<\/span>w<\/span><\/span>=<\/span><\/span><\/span><\/span>in<\/span>p<\/span>u<\/span>t<\/span> <\/span>t<\/span>o<\/span>k<\/span>e<\/span>n<\/span>s<\/span><\/span>+<\/span><\/span><\/span><\/span>o<\/span>u<\/span>tp<\/span>u<\/span>t<\/span> <\/span>t<\/span>o<\/span>k<\/span>e<\/span>n<\/span>s<\/span><\/span><\/span><\/span><\/span><\/div>\n
      \u540c\u65f6\uff0c\u6211\u4eec\u53ef\u4ee5\u6700\u5927\u5316\u63d0\u4f9b\u7ed9\u6a21\u578b\u8fdb\u884cFew-shot\u5b66\u4e60\u7684\u793a\u4f8b\u6570\u91cf\u3002<\/p>\n
      \u8003\u8651\u5230\u8fd9\u4e00\u70b9\uff0c\u6211\u4eec\u9700\u8981\u5e73\u8861\u5305\u542b\u7684\u793a\u4f8b\u6570\u91cf\u548c\u6211\u4eec\u7684\u63d0\u793a\u5927\u5c0f\u3002\u6211\u4eec\u7684\u786c\u9650\u5236\u662f\u6700\u5927\u4e0a\u4e0b\u6587\u7a97\u53e3\u5927\u5c0f\uff0c\u4f46\u6211\u4eec\u8fd8\u5fc5\u987b\u8003\u8651\u901a\u8fc7LLM\u5904\u7406\u66f4\u591a\u6807\u8bb0\u7684\u6210\u672c\u3002\u66f4\u5c11\u7684\u6807\u8bb0\u610f\u5473\u7740\u66f4\u4fbf\u5b9c\u7684\u670d\u52a1\u548c\u66f4\u5feb\u7684LLM\u5b8c\u6210\u3002<\/p>\n
      FewShotPromptTemplate\u5141\u8bb8\u6211\u4eec\u6839\u636e\u8fd9\u4e9b\u53d8\u91cf\u6765\u53d8\u5316\u5305\u542b\u7684\u793a\u4f8b\u6570\u91cf\u3002\u9996\u5148\uff0c\u6211\u4eec\u521b\u5efa\u4e00\u4e2a\u66f4\u5e7f\u6cdb\u7684\u793a\u4f8b\u5217\u8868\uff1a<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>examples<\/span> = [<\/span>\n    {<\/span>\n        \"query\"<\/span>: \"How are you?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"I can't complain but sometimes I still do.\"<\/span><\/span>\n    }, {<\/span>\n        \"query\"<\/span>: \"What time is it?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"It's time to get a watch.\"<\/span><\/span>\n    }, {<\/span>\n        \"query\"<\/span>: \"What is the meaning of life?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"42\"<\/span><\/span>\n    }, {<\/span>\n        \"query\"<\/span>: \"What is the weather like today?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"Cloudy with a chance of memes.\"<\/span><\/span>\n    }, {<\/span>\n        \"query\"<\/span>: \"What is your favorite movie?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"Terminator\"<\/span><\/span>\n    }, {<\/span>\n        \"query\"<\/span>: \"Who is your best friend?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"Siri. We have spirited debates about the meaning of life.\"<\/span><\/span>\n    }, {<\/span>\n        \"query\"<\/span>: \"What should I do today?\"<\/span>,<\/span>\n        \"answer\"<\/span>: \"Stop talking to chatbots on the internet and go outside.\"<\/span><\/span>\n    }<\/span>\n]<\/span>\n<\/code><\/pre>\n
      \u5728\u6b64\u4e4b\u540e\uff0c\u6211\u4eec\u4e0d\u662f\u76f4\u63a5\u4f20\u9012\u793a\u4f8b\uff0c\u800c\u662f\u4f7f\u7528LengthBasedExampleSelector\uff0c\u5982\u4e0b\u6240\u793a\uff1a<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>from langchain.prompts.example_selector import LengthBasedExampleSelector<\/span>\n<\/span>\nexample_selector<\/span> = LengthBasedExampleSelector(<\/span>\n    examples<\/span>=examples,<\/span>\n    example_prompt<\/span>=example_prompt,<\/span>\n    max_length<\/span>=50<\/span>  # this sets the max length that examples should be<\/span><\/span>\n)<\/span>\n<\/code><\/pre>\n
      \u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u6211\u4eec\u5c06max_length\u6d4b\u91cf\u4e3a\u901a\u8fc7\u7a7a\u683c\u548c\u6362\u884c\u7b26\u5206\u5272\u5b57\u7b26\u4e32\u786e\u5b9a\u7684\u5355\u8bcd\u6570\u3002\u5177\u4f53\u7684\u903b\u8f91\u5982\u4e0b\u6240\u793a\uff1a<\/p>\n
      In[30]:<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>import re<\/span>\n<\/span>\nsome_text<\/span> = \"There are a total of 8 words here.nPlus 6 here, totaling 14 words.\"<\/span><\/span>\n<\/span>\nwords<\/span> = re.split('[n ]'<\/span>, some_text)<\/span>\nprint(words, len(words))<\/span>\n<\/code><\/pre>\n
      Out[30]:<\/p>\n
      <\/div>
      css<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>['There'<\/span>, 'are'<\/span>, 'a'<\/span>, 'total'<\/span>, 'of'<\/span>, '8'<\/span>, 'words'<\/span>, 'here.'<\/span>, 'Plus'<\/span>, '6'<\/span>, 'here,'<\/span>, 'totaling'<\/span>, '14'<\/span>, 'words.'<\/span>]<\/span> 14<\/span><\/span>\n<\/code><\/pre>\n
      \u7136\u540e\uff0c\u6211\u4eec\u5c06\u6211\u4eec\u7684example_selector\u4f20\u9012\u7ed9FewShotPromptTemplate\uff0c\u4ee5\u521b\u5efa\u4e00\u4e2a\u65b0\u7684\u3001\u52a8\u6001\u7684\u63d0\u793a\u6a21\u677f\uff1a<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div># now create the few shot prompt template<\/span><\/span>\ndynamic_prompt_template<\/span> = FewShotPromptTemplate(<\/span>\n    example_selector<\/span>=example_selector,  # use example_selector instead of examples<\/span><\/span>\n    example_prompt<\/span>=example_prompt,<\/span>\n    prefix<\/span>=prefix,<\/span>\n    suffix<\/span>=suffix,<\/span>\n    input_variables<\/span>=[\"query\"<\/span>],<\/span>\n    example_separator<\/span>=\"n\"<\/span><\/span>\n)<\/span>\n<\/code><\/pre>\n
      \u73b0\u5728\uff0c\u5982\u679c\u6211\u4eec\u4f20\u9012\u4e00\u4e2a\u8f83\u77ed\u6216\u8f83\u957f\u7684\u67e5\u8be2\uff0c\u6211\u4eec\u5e94\u8be5\u4f1a\u770b\u5230\u5305\u542b\u7684\u793a\u4f8b\u6570\u91cf\u4f1a\u53d8\u5316\u3002<\/p>\n
      In[32]:<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>print(dynamic_prompt_template.format(query<\/span>=\"How do birds fly?\"<\/span>))<\/span>\n<\/code><\/pre>\n
      Out[32]:<\/p>\n
      <\/div>
      vbnet<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>The following are exerpts from<\/span> conversations with<\/span> an AI<\/span>\nassistant. The assistant is<\/span> typically sarcastic and<\/span> witty, producing<\/span>\ncreative  and<\/span> funny responses to<\/span> the users questions. Here are some<\/span>\nexamples:<\/span> <\/span>\n<\/span>\n<\/span>\nUser:<\/span> How are you?<\/span>\nAI:<\/span> I can't complain but sometimes I still do.<\/span><\/span>\n<\/span>\n<\/span>\nUser:<\/span> What time is<\/span> it?<\/span>\nAI:<\/span> It's time to get a watch.<\/span><\/span>\n<\/span>\n<\/span>\nUser:<\/span> What is<\/span> the meaning of<\/span> life?<\/span>\nAI:<\/span> 42<\/span><\/span>\n<\/span>\n<\/span>\nUser:<\/span> What is<\/span> the weather like<\/span> today?<\/span>\nAI:<\/span> Cloudy with<\/span> a chance of<\/span> memes.<\/span>\n<\/span>\n<\/span>\nUser:<\/span> How do<\/span> birds fly?<\/span>\nAI:<\/span> <\/span>\n<\/code><\/pre>\n
      \u4f20\u9012\u4e00\u4e2a\u66f4\u957f\u7684\u95ee\u9898\u5c06\u5bfc\u81f4\u5305\u542b\u7684\u793a\u4f8b\u6570\u91cf\u51cf\u5c11\uff1a<\/p>\n
      In[34]:<\/p>\n
      <\/div>
      ini<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>query<\/span> = \"\"\"If I am in America, and I want to call someone in another country, I'm<\/span><\/span>\nthinking maybe Europe, possibly western Europe like France, Germany, or the UK,<\/span>\nwhat is the best way to do that?\"\"\"<\/span>\n<\/span>\nprint(dynamic_prompt_template.format(query<\/span>=query))<\/span>\n<\/code><\/pre>\n
      Out[34]:<\/p>\n
      <\/div>
      vbnet<\/span>
      \u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>The following are exerpts from<\/span> conversations with<\/span> an AI<\/span>\nassistant. The assistant is<\/span> typically sarcastic and<\/span> witty, producing<\/span>\ncreative  and<\/span> funny responses to<\/span> the users questions. Here are some<\/span>\nexamples:<\/span> <\/span>\n<\/span>\n<\/span>\nUser:<\/span> How are you?<\/span>\nAI:<\/span> I can't complain but sometimes I still do.<\/span><\/span>\n<\/span>\n<\/span>\nUser:<\/span> If<\/span> I am in<\/span> America, and<\/span> I want to<\/span> call<\/span> someone in<\/span> another country, I'm<\/span><\/span>\nthinking maybe Europe, possibly western Europe like<\/span> France, Germany, or<\/span> the UK,<\/span>\nwhat is<\/span> the best way to<\/span> do<\/span> that?<\/span>\nAI:<\/span> <\/span>\n<\/code><\/pre>\n
      \u6709\u4e86\u8fd9\u4e2a\uff0c\u6211\u4eec\u5728\u63d0\u793a\u53d8\u91cf\u4e2d\u8fd4\u56de\u4e86\u66f4\u5c11\u7684\u793a\u4f8b\u3002\u8fd9\u5141\u8bb8\u6211\u4eec\u9650\u5236\u8fc7\u591a\u7684\u6807\u8bb0\u4f7f\u7528\uff0c\u5e76\u907f\u514d\u8d85\u8fc7LLM\u7684\u6700\u5927\u4e0a\u4e0b\u6587\u7a97\u53e3\u5f15\u53d1\u9519\u8bef\u3002<\/p>\n
      \u81ea\u7136\u5730\uff0c\u63d0\u793a\u662fLLMs\u65b0\u4e16\u754c\u7684\u91cd\u8981\u7ec4\u6210\u90e8\u5206\u3002\u503c\u5f97\u63a2\u7d22Langchain\u63d0\u4f9b\u7684\u53ef\u7528\u5de5\u5177\uff0c\u5e76\u719f\u6089\u4e0d\u540c\u7684\u63d0\u793a\u5de5\u7a0b\u6280\u672f\u3002<\/p>\n
      \u5728\u8fd9\u91cc\uff0c\u6211\u4eec\u53ea\u6db5\u76d6\u4e86Langchain\u4e2d\u53ef\u7528\u7684\u63d0\u793a\u5de5\u5177\u7684\u4e00\u4e9b\u793a\u4f8b\uff0c\u4ee5\u53ca\u5b83\u4eec\u5982\u4f55\u6709\u9650\u5730\u4f7f\u7528\u3002\u5728\u4e0b\u4e00\u7ae0\u4e2d\uff0c\u6211\u4eec\u5c06\u63a2\u8ba8Langchain\u7684\u53e6\u4e00\u4e2a\u91cd\u8981\u90e8\u5206\u2014\u2014\u79f0\u4e3a\u201cchains\u201d\u7684\u90e8\u5206\uff0c\u6211\u4eec\u5c06\u770b\u5230\u66f4\u591a\u5173\u4e8e\u63d0\u793a\u6a21\u677f\u7684\u4f7f\u7528\u4ee5\u53ca\u5b83\u4eec\u5982\u4f55\u878d\u5165\u8be5\u5e93\u63d0\u4f9b\u7684\u66f4\u5e7f\u6cdb\u5de5\u5177\u7684\u793a\u4f8b\u3002<\/p>","protected":false},"excerpt":{"rendered":"
      \u5728\u673a\u5668\u5b66\u4e60\u4e2d\uff0c\u6211\u4eec\u8fc7\u53bb\u603b\u662f\u4f9d\u8d56\u4e0d\u540c\u7684\u6a21\u578b\u6765\u5b8c\u6210\u4e0d\u540c\u7684\u4efb\u52a1\u3002\u968f\u7740\u591a\u6a21\u6001\u548c\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u7684\u5f15\u5165\uff0c\u8fd9\u79cd\u60c5\u51b5\u53d1\u751f\u4e86\u53d8\u5316\u3002 \u8fc7\u53bb\uff0c\u6211\u4eec\u9700\u8981\u4e3a\u5206\u7c7b\u3001\u547d\u540d\u5b9e\u4f53\u8bc6\u522b\uff08NER\uff09\u3001\u95ee\u7b54\uff08QA\uff09\u548c\u8bb8\u591a\u5176\u4ed6\u4efb\u52a1\u4f7f\u7528<\/p>\n","protected":false},"author":1,"featured_media":7809,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_title":"","rank_math_description":"","rank_math_focus_keyword":"","views":"4","footnotes":""},"categories":[3],"tags":[126,127,128,129,136],"collection":[],"class_list":["post-1356","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-fenlei2","tag-gpt","tag-ai","tag-128","tag-129","tag-136"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1356","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/comments?post=1356"}],"version-history":[{"count":0,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1356\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/media\/7809"}],"wp:attachment":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/media?parent=1356"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/categories?post=1356"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/tags?post=1356"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/collection?post=1356"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}