docs(plans): add Spec-3 question-gen-v2 implementation plan
9 tasks: type extension → postprocess → families → sampler_v2 → generator_v2 → gates → run_store → pipeline_v2 → CLI integration. Includes structured-logging schemas/metrics and Codex review revisions.
This commit is contained in:
@@ -0,0 +1,33 @@
|
||||
---
|
||||
type: metric
|
||||
node_id: metric:qgen-answer-uniformity
|
||||
title: 答案位置分布均匀性
|
||||
date: 2026-07-12
|
||||
---
|
||||
|
||||
# 答案位置分布均匀性
|
||||
|
||||
## 定义
|
||||
|
||||
最终接受题目中,正确答案在 A/B/C/D 四个位置的分布是否均匀。
|
||||
|
||||
## 判定方式
|
||||
|
||||
- **类型**: 硬性(卡方检验)
|
||||
- **阈值**: χ² 检验 p > 0.05(不拒绝均匀分布假设)
|
||||
- **计算**: 对 accepted 题目统计 correct_answer 的 A/B/C/D 频次,做 χ²(df=3) 检验
|
||||
- **数据源**: 题目 JSON 文件(correct_answer 字段)
|
||||
|
||||
## 基线
|
||||
|
||||
设计 §7.3 要求 shuffle 后必须均匀;不依赖历史基线。
|
||||
|
||||
## 验证方式
|
||||
|
||||
```python
|
||||
from scipy.stats import chisquare
|
||||
observed = [count_A, count_B, count_C, count_D]
|
||||
expected = [total/4] * 4
|
||||
stat, p = chisquare(observed, expected)
|
||||
assert p > 0.05, f"答案位置分布不均匀: p={p:.4f}"
|
||||
```
|
||||
Reference in New Issue
Block a user