Get Started

代码沙盒服务主要提供两个功能：运行代码和判断题目对错。

支持的编程语言：

Python (python, pytest)
C++
C#
Go (go, go test)
Java (javac, junit)
NodeJS
Typescript (tsx, jest)
Lean

Scala
Kotlin
PHP
Rust
Bash
Lua
R
Perl

D
Ruby
Julia
Verilog
CUDA (GPU)
Python (GPU)

已经实现的开源数据集：

本地部署

docker run -it -p 8080:8080 volcengine/sandbox-fusion:server-20250609

国内用户可以使用以下镜像源：

docker run -it -p 8080:8080 vemlp-cn-beijing.cr.volces.com/preset-images/code-sandbox:server-20250609

使用

代码沙盒

提示

http://localhost:8080/SandboxFusion/playground/sandbox 提供了一个简单的测试页面

在shell中执行下面的指令来请求沙盒执行一段python代码：

curl 'http://localhost:8080/run_code' \
  -H 'Content-Type: application/json' \
  --data-raw '{"code": "print(\"Hello, world!\")", "language": "python"}'

样例输出：

{
  "status": "Success",
  "message": "",
  "compile_result": null,
  "run_result": {
    "status": "Finished",
    "execution_time": 0.016735315322875977,
    "return_code": 0,
    "stdout": "Hello, world!\n",
    "stderr": ""
  },
  "executor_pod_name": null,
  "files": {}
}

也可以用Python脚本完成类似的请求，一个运行C++代码的例子：

import requests
import json

response = requests.post('http://localhost:8080/run_code', json={
    'code': '''
#include <iostream>

int main() {
    std::cout << "Hello, world!" << std::endl;
    return 0;
}
''',
    'language': 'cpp',
})

print(json.dumps(response.json(), indent=2))

样例输出：

{
  "status": "Success",
  "message": "",
  "compile_result": {
    "status": "Finished",
    "execution_time": 0.45870447158813477,
    "return_code": 0,
    "stdout": "",
    "stderr": ""
  },
  "run_result": {
    "status": "Finished",
    "execution_time": 0.002761363983154297,
    "return_code": 0,
    "stdout": "Hello, world!\n",
    "stderr": ""
  },
  "executor_pod_name": null,
  "files": {}
}

数据集

提示

http://localhost:8080/SandboxFusion/playground/datasets 提供了一个简单的测试页面

Datasets模块实现了各类不同Code数据集的判断逻辑，并提供了统一的接口。以MBPP数据集为例，下面是通过沙盒服务给模型输出做评估的流程：

获取MBPP所有题目的prompt：

curl 'http://localhost:8080/get_prompts' \
  -H 'Content-Type: application/json' \
  --data-raw '{"dataset":"mbpp","config":{}}'

输出：

[
  {
    "id": 11,
    "prompt": "Write a python function to remove first and last occurrence of a given character from the string. Your code should satisfy these tests:\n\nassert remove_Occ(\"hello\",\"l\") == \"heo\"",
    "labels": {
      "challenge_test_list": [
        "assert remove_Occ(\"hellolloll\",\"l\") == \"helollol\"",
        "assert remove_Occ(\"\",\"l\") == \"\""
      ],
      "test_setup_code": ""
    }
  },
  {
    "id": 12,
    "prompt": "Write a function to sort a given matrix in ascending order according to the sum of its rows. Your code should satisfy these tests:\n\nassert sort_matrix([[1, 2, 3], [2, 4, 5], [1, 1, 1]])==[[1, 1, 1], [1, 2, 3], [2, 4, 5]]",
    "labels": {
      "challenge_test_list": [],
      "test_setup_code": ""
    }
  },
  ...
]

提交模型的输出，获得题目对错的结果：

curl 'http://localhost:8080/submit' \
  -H 'Content-Type: application/json' \
  --data-raw '{"dataset":"mbpp","id":"11","completion":"Here is a Python function that removes the first and last occurrence of a given character from a string:\n\n```python\ndef remove_Occ(s, char):\n    first_occ = s.find(char)\n    last_occ = s.rfind(char)\n    \n    if first_occ == -1 or first_occ == last_occ:\n        return s\n    \n    # Remove the first occurrence\n    s = s[:first_occ] + s[first_occ + 1:]\n    \n    # Adjust the index for the last occurrence since the string is now one character shorter\n    last_occ -= 1\n    \n    # Remove the last occurrence\n    s = s[:last_occ] + s[last_occ + 1:]\n    \n    return s\n\n# Test the function\nassert remove_Occ(\"hello\", \"l\") == \"heo\"\n```\n\nThis function works as follows:\n1. It finds the index of the first occurrence of the given character.\n2. It finds the index of the last occurrence of the given character.\n3. If the character does not exist in the string or only occurs once, it simply returns the original string.\n4. Otherwise, it constructs a new string by removing the first occurrence and then adjusts the index for the last occurrence before removing it.\n\nYou can run the provided test to ensure the function works as expected.","config":{}}'

输出：

{
  "id": "11",
  "accepted": true,
  "extracted_code": "def remove_Occ(s, char):\n    first_occ = s.find(char)\n    last_occ = s.rfind(char)\n    \n    if first_occ == -1 or first_occ == last_occ:\n        return s\n    \n    # Remove the first occurrence\n    s = s[:first_occ] + s[first_occ + 1:]\n    \n    # Adjust the index for the last occurrence since the string is now one character shorter\n    last_occ -= 1\n    \n    # Remove the last occurrence\n    s = s[:last_occ] + s[last_occ + 1:]\n    \n    return s\n\n# Test the function\nassert remove_Occ(\"hello\", \"l\") == \"heo\"",
  "full_code": "def remove_Occ(s, char):\n    first_occ = s.find(char)\n    last_occ = s.rfind(char)\n    \n    if first_occ == -1 or first_occ == last_occ:\n        return s\n    \n    # Remove the first occurrence\n    s = s[:first_occ] + s[first_occ + 1:]\n    \n    # Adjust the index for the last occurrence since the string is now one character shorter\n    last_occ -= 1\n    \n    # Remove the last occurrence\n    s = s[:last_occ] + s[last_occ + 1:]\n    \n    return s\n\n# Test the function\nassert remove_Occ(\"hello\", \"l\") == \"heo\"\n\nassert remove_Occ(\"hello\",\"l\") == \"heo\"\nassert remove_Occ(\"abcda\",\"a\") == \"bcd\"\nassert remove_Occ(\"PHP\",\"P\") == \"H\"",
  "test_code": null,
  "tests": [
    {
      "passed": true,
      "exec_info": {
        "status": "Success",
        "message": "",
        "compile_result": null,
        "run_result": {
          "status": "Finished",
          "execution_time": 0.017310619354248047,
          "return_code": 0,
          "stdout": "",
          "stderr": ""
        },
        "executor_pod_name": null,
        "files": {}
      },
      "test_info": null
    }
  ],
  "extracted_type": null,
  "extra": null
}

本地部署​

使用​

代码沙盒​

数据集​

本地部署

使用

代码沙盒

数据集