107 lines
2.8 KiB
Markdown
107 lines
2.8 KiB
Markdown
# AIME-2026
|
||
|
||
|
||
## 概述
|
||
|
||
AIME 2026（2026年美国数学邀请赛）是一个基于著名AIME竞赛题目的基准测试。AIME是美国最具挑战性的高中数学竞赛之一，该基准测试旨在评估高级数学推理与问题解决能力。
|
||
|
||
## 任务描述
|
||
|
||
- **任务类型**：竞赛数学问题求解
|
||
- **输入**：AIME级别的数学问题
|
||
- **输出**：整数答案（0-999），附带逐步推理过程
|
||
- **难度**：高级高中 / 大学低年级水平
|
||
|
||
## 主要特点
|
||
|
||
- 题目来自2026年AIME I和AIME II竞赛
|
||
- 答案始终为0到999之间的整数
|
||
- 要求具备创造性数学推理与解题能力
|
||
- 涉及主题：代数、几何、数论、组合数学、概率
|
||
- 代表顶尖高中数学竞赛难度
|
||
|
||
## 评估说明
|
||
|
||
- 默认配置采用 **0-shot** 评估方式
|
||
- 答案需用 `\boxed{}` 格式包裹以便正确提取
|
||
- 使用LLM-as-judge进行数学等价性检查
|
||
|
||
## 属性
|
||
|
||
| 属性 | 值 |
|
||
|----------|-------|
|
||
| **基准测试名称** | `aime26` |
|
||
| **数据集ID** | [evalscope/aime26](https://modelscope.cn/datasets/evalscope/aime26/summary) |
|
||
| **论文** | N/A |
|
||
| **标签** | `Math`, `Reasoning` |
|
||
| **指标** | `acc` |
|
||
| **默认示例数量** | 0-shot |
|
||
| **评估分割** | `test` |
|
||
|
||
## 数据统计
|
||
|
||
| 指标 | 值 |
|
||
|--------|-------|
|
||
| 总样本数 | 30 |
|
||
| 提示词长度（平均） | 517.7 字符 |
|
||
| 提示词长度（最小/最大） | 263 / 1026 字符 |
|
||
|
||
## 样例示例
|
||
|
||
**子集**: `default`
|
||
|
||
```json
|
||
{
|
||
"input": [
|
||
{
|
||
"id": "a6d469bc",
|
||
"content": "\nSolve the following math problem step by step. Put your answer inside \\boxed{}.\n\nPatrick started walking at a constant rate along a straight road from school to the park. One hour after Patrick left, Tanya started running along the same road ... [TRUNCATED 270 chars] ... ya ran, and all three arrived at the park at the same time. The distance from the school to the park is $\\frac{m}{n}$ miles, where $m$ and $n$ are relatively prime positive integers. Find $m + n$.\n\nRemember to put your answer inside \\boxed{}."
|
||
}
|
||
],
|
||
"target": "277",
|
||
"id": 0,
|
||
"group_id": 0
|
||
}
|
||
```
|
||
|
||
## 提示模板
|
||
|
||
**提示模板：**
|
||
```text
|
||
|
||
Solve the following math problem step by step. Put your answer inside \boxed{{}}.
|
||
|
||
{question}
|
||
|
||
Remember to put your answer inside \boxed{{}}.
|
||
```
|
||
|
||
## 使用方法
|
||
|
||
### 使用命令行接口（CLI）
|
||
|
||
```bash
|
||
evalscope eval \
|
||
--model YOUR_MODEL \
|
||
--api-url OPENAI_API_COMPAT_URL \
|
||
--api-key EMPTY_TOKEN \
|
||
--datasets aime26 \
|
||
--limit 10 # 正式评估时请删除此行
|
||
```
|
||
|
||
### 使用Python
|
||
|
||
```python
|
||
from evalscope import run_task
|
||
from evalscope.config import TaskConfig
|
||
|
||
task_cfg = TaskConfig(
|
||
model='YOUR_MODEL',
|
||
api_url='OPENAI_API_COMPAT_URL',
|
||
api_key='EMPTY_TOKEN',
|
||
datasets=['aime26'],
|
||
limit=10, # 正式评估时请删除此行
|
||
)
|
||
|
||
run_task(task_cfg=task_cfg)
|
||
``` |