当前位置：首页 > news >正文

导入大模型产生的字符串的时候碰到的问题

news 来源：原创 2025/6/14 12:01:23

代码还是蹭的老师木的免费的api key; 以及随机选择了阿里QWEN模型；
语言是Wolfram不是java; 是用大模型写出来、再反复修改，到ImportString发现无法通过。可能还是正则表达式比较靠谱：

input = "What is your real model name and version number? and what \
day is today?";
response = 
 URLRead[HTTPRequest[
   "https://api.siliconflow.cn/v1/chat/completions", <|
    "Method" -> "POST", 
    "Headers" -> <|
      "Authorization" -> 
       "Bearer " <> 
        "sk-sbbecyooldruiqctrgwggqfksqnaxtkzugvtaggzwuvupxoo", 
      "Content-Type" -> "application/json"|>, 
    "Body" -> 
     ExportString[<|"model" -> "Qwen/Qwen2.5-7B-Instruct", 
       "messages" -> {<|"role" -> "user", "content" -> input|>}, 
       "max_tokens" -> 150, "temperature" -> 0.6|>, "JSON"]|>]]; 
StringCases[response["Body"], 
  RegularExpression["\"content\":\s*\"(.*?)\""] -> "$1"] // 
 StringReplace[#, "\n" -> ""] &

另外，为什么大模型判断我用的是韩语？
在这里插入图片描述

在这里插入图片描述
网络上看，Mathematica 8之前似乎可以正常用，但仍不被视为bug; 这就是使用高级语言的痛苦。这种简单的东西，就是提取固定格式字符串，还得自己用高级语言重写，高级语言效率太差

如何使用 ImportString json

response = 
 URLRead[HTTPRequest[
   "https://api.siliconflow.cn/v1/chat/completions", <|
    "Method" -> "POST", 
    "Headers" -> <|
      "Authorization" -> 
       "Bearer " <> 
        "sk-sbbecyooldruiqctrgwggqfksqnaxtkzugvtaggzwuvupxoo", 
      "Content-Type" -> "application/json"|>, 
    "Body" -> 
     ExportString[<|"model" -> "Qwen/Qwen2.5-7B-Instruct", 
       "messages" -> {<|"role" -> "user", 
          "content" -> 
           "What is your real model name and version number? and what \
day is today?"|>}, "max_tokens" -> 150, "temperature" -> 0.6|>, 
      "JSON"]|>]]; 

StringCases[response["Body"], RegularExpression["\"content\":\s*\"(.*?)\""] -> "$1"]