惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Application and Cybersecurity Blog
Application and Cybersecurity Blog
The Register - Security
The Register - Security
V
Visual Studio Blog
aimingoo的专栏
aimingoo的专栏
Stack Overflow Blog
Stack Overflow Blog
IT之家
IT之家
量子位
C
Check Point Blog
博客园 - 【当耐特】
小众软件
小众软件
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
雷峰网
雷峰网
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
Microsoft Azure Blog
Microsoft Azure Blog
CTFtime.org: upcoming CTF events
CTFtime.org: upcoming CTF events
Engineering at Meta
Engineering at Meta
Recorded Future
Recorded Future
The Last Watchdog
The Last Watchdog
博客园 - Franky
N
Netflix TechBlog - Medium
Webroot Blog
Webroot Blog
A
About on SuperTechFans
cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
W
WeLiveSecurity
D
Docker
S
Security Affairs
T
The Blog of Author Tim Ferriss
F
Fortinet All Blogs
Blog — PlanetScale
Blog — PlanetScale
V2EX - 技术
V2EX - 技术
Jina AI
Jina AI
Help Net Security
Help Net Security
L
LangChain Blog
P
Proofpoint News Feed
The Cloudflare Blog
WordPress大学
WordPress大学
Google DeepMind News
Google DeepMind News
Schneier on Security
Schneier on Security
Recent Announcements
Recent Announcements
Attack and Defense Labs
Attack and Defense Labs
云风的 BLOG
云风的 BLOG
V
Vulnerabilities – Threatpost
Microsoft Security Blog
Microsoft Security Blog
H
Heimdal Security Blog
P
Proofpoint News Feed
O
OpenAI News
H
Help Net Security
cs.CV updates on arXiv.org
cs.CV updates on arXiv.org
爱范儿
爱范儿
Security Archives - TechRepublic
Security Archives - TechRepublic

博客园 - 黄洪波

安装openclaw 排查项目中依赖的mybatis 拦截器 ModelAttribute 老革命遇上新问题 使用calcite构造ddl建表语句 OpenWebUI单点登录之解决动态参数问题 IDEA自带的Maven 3.9.x无法刷新http nexus私服(转) windows docker安装rocketmq之踩坑记 分布式系统设计经典论文(转载) 训练自己的yolo-v11数据集(二) 训练自己的yolo-v11数据集(一) 本地使用pycharm进行yolo推理 2024年的云原生架构需要哪些技术栈 (转) yolo v11学习,入门篇 OpenWebUI单点登录之深坑 解决Win10无法进入睡眠模式(转) idea常用插件 内网离线模式下激活JRebel java导入json数据至doris 将SpringBoot打包之后的jar设为守护进程
AI工具验证
黄洪波 · 2024-12-26 · via 博客园 - 黄洪波

起因,拿到一段base64,懒得写代码去反编码,干脆让ai输出,结果发现各家输出结果不一致。

测评时间2024-12-26

测评对象:ChatGPT,文小言(百度),通义(阿里),豆包(抖音),元宝(腾讯),kimi智能助手,智谱清言。

测评内容:请用java的base64反编码这段内容 W3sic3ViX3Rhc2tfaWQiOiI4ZGIzYTMyZC0yMGQ0LTRlNGYtYmUyYS00NWFmOGJlNmZlYTYiLCJ0YXNrX3N0YXR1cyI6MSwidGFza19jb21wbGV0aW9uIjoxLCJpbWFnZSI6IiIsImltYWdlX3dtIjoiaHR0cHM6Ly9zdG9yYWdlLmhpZHJlYW1haS5jb20vaW1hZ2UvcF82ZjM3NTc1ZS1iOGY0LTExZWYtOTA0Ni0wMDE2M2UyYzdjNTdfd20uanBnIn1d并打印结果

测评结果,GPT,文小言,通义,智谱清言 析并正确打印, 

豆包,元宝,kimi错误。

备注:无论是否限定语言,对结论并没有影响,豆包、元宝、kimi都不能正确输出结果

下面我将输出各家的测试结果。

java原始代码:

import java.util.Base64;

public class Base64Decoder {
    public static void main(String[] args) {
        // Base64 编码的字符串
        String encodedString = "W3sic3ViX3Rhc2tfaWQiOiI4ZGIzYTMyZC0yMGQ0LTRlNGYtYmUyYS00NWFmOGJlNmZlYTYiLCJ0YXNrX3N0YXR1cyI6MSwidGFza19jb21wbGV0aW9uIjoxLCJpbWFnZSI6IiIsImltYWdlX3dtIjoiaHR0cHM6Ly9zdG9yYWdlLmhpZHJlYW1haS5jb20vaW1hZ2UvcF82ZjM3NTc1ZS1iOGY0LTExZWYtOTA0Ni0wMDE2M2UyYzdjNTdfd20uanBnIn1d";

        // 解码 Base64 字符串
        byte[] decodedBytes = Base64.getDecoder().decode(encodedString);
        String decodedString = new String(decodedBytes);

        // 打印解码后的结果
        System.out.println(decodedString);
    }
}

输出结果

[{
    "sub_task_id": "8db3a32d-20d4-4e4f-be2a-45af8be6fea6",
    "task_status": 1,
    "task_completion": 1,
    "image": "",
    "image_wm": "https://storage.hidreamai.com/image/p_6f37575e-b8f4-11ef-9046-00163e2c7c57_wm.jpg"
}]

gpt,文小言,通义代码正确,结果正确:

[
  {
    "sub_task_id": "8db3a32d-20d4-4e4f-be2a-45af8be6fea6",
    "task_status": 1,
    "task_completion": 1,
    "image": "",
    "image_wm": "https://storage.hidreamai.com/image/p_6f37575e-b8f4-11ef-9046-00163e2c7c57_wm.jpg"
  }
]

豆包代码正确,结果错误

{
    "subject_token_id": "8db3a32d-20d4-4e4f-be2a-45af8be6fea6",
    "task_status": 1,
    "task_judgment": 1,
    "image": "",
    "image_dt": "https://storage.googleapis.com/img/6f37575e-b8f4-11ef-9046-00163e2c7c57_dt.jpg"
}

元宝代码正确,结果错误

[
    {
        "sub_task_id": "8db3a32d-20d4-4e4f-be2a-15af8be6fe6a",
        "task_status": 1,
        "task_completion": 1,
        "image": "",
        "image_url": "https://storage.hdrmail.com/image/p_6f37575e-b8f4-11ef-9046-00163e2c7c57.jpg"
    }
]

kimi代码正确,结果错误,给了个示例糊弄人

{
  "id": "4",
  "name": "张三",
  "age": 30,
  "email": "zhangsan@example.com",
  "phone": "13800138000",
  "address": "中国北京市朝阳区芍药居",
  "createTime": "2024-12-23T08:30:00Z",
  "updateTime": "2024-12-23T08:30:00Z"
}