113 lines
2.7 KiB
Markdown
113 lines
2.7 KiB
Markdown
# MRI-MCQA
|
||
|
||
## 概述
|
||
|
||
MRI-MCQA 是一个专门针对磁共振成像（MRI）的多项选择题基准测试，用于评估 AI 模型对 MRI 物理原理、扫描协议、图像采集及临床应用的理解能力。
|
||
|
||
## 任务描述
|
||
|
||
- **任务类型**：医学影像知识多项选择问答
|
||
- **输入**：与 MRI 相关的问题及其多个选项
|
||
- **输出**：正确答案的字母
|
||
- **领域**：医学影像、MRI 物理、放射学
|
||
|
||
## 主要特点
|
||
|
||
- 专注于 MRI 技术及其应用
|
||
- 考察对 MRI 物理原理和扫描协议的理解
|
||
- 涵盖临床 MRI 应用和序列
|
||
- 专为评估医学影像 AI 系统而设计
|
||
- 采用多项选择格式以实现标准化评估
|
||
|
||
## 评估说明
|
||
|
||
- 默认配置使用 **0-shot** 评估
|
||
- 在测试集（test split）上进行评估
|
||
- 使用简单准确率（accuracy）作为评估指标
|
||
- 无训练集可用
|
||
|
||
## 属性
|
||
|
||
| 属性 | 值 |
|
||
|----------|-------|
|
||
| **基准测试名称** | `mri_mcqa` |
|
||
| **数据集 ID** | [extraordinarylab/mri-mcqa](https://modelscope.cn/datasets/extraordinarylab/mri-mcqa/summary) |
|
||
| **论文** | N/A |
|
||
| **标签** | `Knowledge`, `MCQ`, `Medical` |
|
||
| **指标** | `acc` |
|
||
| **默认示例数量** | 0-shot |
|
||
| **评估划分** | `test` |
|
||
|
||
## 数据统计
|
||
|
||
| 指标 | 值 |
|
||
|--------|-------|
|
||
| 总样本数 | 563 |
|
||
| 提示词长度（平均） | 457.23 字符 |
|
||
| 提示词长度（最小/最大） | 259 / 888 字符 |
|
||
|
||
## 样例示例
|
||
|
||
**子集**: `default`
|
||
|
||
```json
|
||
{
|
||
"input": [
|
||
{
|
||
"id": "84f179d7",
|
||
"content": "Answer the following multiple choice question. The entire content of your response should be of the following format: 'ANSWER: [LETTER]' (without quotes) where [LETTER] is one of A,B,C,D.\n\nWhich cardiac chambers are typically imaged on the short-axis view?\n\nA) RA and RV\nB) RA and LA\nC) LA and LV\nD) RV and LV"
|
||
}
|
||
],
|
||
"choices": [
|
||
"RA and RV",
|
||
"RA and LA",
|
||
"LA and LV",
|
||
"RV and LV"
|
||
],
|
||
"target": "D",
|
||
"id": 0,
|
||
"group_id": 0,
|
||
"metadata": {}
|
||
}
|
||
```
|
||
|
||
## 提示模板
|
||
|
||
**提示模板：**
|
||
```text
|
||
Answer the following multiple choice question. The entire content of your response should be of the following format: 'ANSWER: [LETTER]' (without quotes) where [LETTER] is one of {letters}.
|
||
|
||
{question}
|
||
|
||
{choices}
|
||
```
|
||
|
||
## 使用方法
|
||
|
||
### 使用 CLI
|
||
|
||
```bash
|
||
evalscope eval \
|
||
--model YOUR_MODEL \
|
||
--api-url OPENAI_API_COMPAT_URL \
|
||
--api-key EMPTY_TOKEN \
|
||
--datasets mri_mcqa \
|
||
--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=['mri_mcqa'],
|
||
limit=10, # 正式评估时请删除此行
|
||
)
|
||
|
||
run_task(task_cfg=task_cfg)
|
||
``` |