reasoning-degeneration-dev/sdc-scores-medium-v1

---
license: mit
tags:
- semantic-distance-coding
- scores
- medium
---

# sdc-scores-medium-v1

Aggregated pass@1 scores — Medium tier, mainstream languages only. Esoteric baselines removed (paper transcriptions, not per-tier measurements).

## Dataset Info

- **Rows**: 8
- **Columns**: 10

## Columns

| Column | Type | Description |
|--------|------|-------------|
| language | Value('string') | Programming language name |
| tiobe_rank | Value('int64') | TIOBE index rank (1=Python, 47=OCaml) |
| tiobe_pct | Value('float64') | TIOBE index percentage share |
| condition | Value('string') | zero-shot |
| pass_at_1 | Value('float64') | % of 20 problems solved, averaged over 3 runs |
| pass_at_1_std | Value('float64') | Standard deviation of pass@1 across 3 runs |
| compile_rate | Value('float64') | % that compiled successfully |
| num_problems | Value('int64') | Number of problems evaluated |
| num_runs | Value('int64') | Number of independent runs |
| per_problem | List({'pass_rate': Value('float64'), 'problem_id': Value('string')}) | List of per-problem pass rates across runs |

## Generation Parameters

```json
{
"script_name": "run_medium_zero_shot.py",
"model": "gpt-5-2",
"description": "Aggregated pass@1 scores \u2014 Medium tier, mainstream languages only. Esoteric baselines removed (paper transcriptions, not per-tier measurements).",
"tier": "medium",
"hyperparameters": {},
"input_datasets": []
}
```

## Experiment Documentation

For complete experiment details, see [https://github.com/Zayne-sprague/SC-Research-Notes/tree/main/experiments/semantic-distance-coding](https://github.com/Zayne-sprague/SC-Research-Notes/tree/main/experiments/semantic-distance-coding)

## Usage

```python
from datasets import load_dataset

dataset = load_dataset("reasoning-degeneration-dev/sdc-scores-medium-v1", split="train")
print(f"Loaded {len(dataset)} rows")
```

---

*This dataset is tracked in [reasoning-degeneration-dev/PROJECT-MANIFEST](https://huggingface.co/datasets/reasoning-degeneration-dev/PROJECT-MANIFEST)*

---
许可证：MIT许可证
标签：
- 语义距离编码（semantic-distance-coding）
- 分数
- 中等层级
---

# sdc-scores-medium-v1

聚合后的pass@1分数——仅针对中等层级的主流编程语言，已移除小众基准测试（仅保留论文转录内容，而非分层测量结果）。

## 数据集信息

- **行数**：8
- **列数**：10

## 列信息

| 列名 | 数据类型 | 描述 |
|--------|------|-------------|
| language | Value('string') | 编程语言名称 |
| tiobe_rank | Value('int64') | TIOBE指数排名（1代表Python，47代表OCaml） |
| tiobe_pct | Value('float64') | TIOBE指数市场份额占比 |
| condition | Value('string') | 零样本（zero-shot） |
| pass_at_1 | Value('float64') | 20道题目中被正确解答的比例，取3次运行的平均值 |
| pass_at_1_std | Value('float64') | 3次运行中pass@1指标的标准差 |
| compile_rate | Value('float64') | 代码编译成功的比例 |
| num_problems | Value('int64') | 参与评估的题目数量 |
| num_runs | Value('int64') | 独立运行的次数 |
| per_problem | List({'pass_rate': Value('float64'), 'problem_id': Value('string')}) | 各题目单次运行通过率的列表 |

## 生成参数

json
{
"脚本名称": "run_medium_zero_shot.py",
"模型": "gpt-5-2",
"描述": "聚合后的pass@1分数——仅针对中等层级的主流编程语言，已移除小众基准测试（仅保留论文转录内容，而非分层测量结果）",
"层级": "中等",
"超参数": {},
"输入数据集": []
}

## 实验文档

完整实验细节请参阅：[https://github.com/Zayne-sprague/SC-Research-Notes/tree/main/experiments/semantic-distance-coding](https://github.com/Zayne-sprague/SC-Research-Notes/tree/main/experiments/semantic-distance-coding)

## 使用方法

python
from datasets import load_dataset

dataset = load_dataset("reasoning-degeneration-dev/sdc-scores-medium-v1", split="train")
print(f"已加载 {len(dataset)} 行数据")

*本数据集已在 [reasoning-degeneration-dev/PROJECT-MANIFEST](https://huggingface.co/datasets/reasoning-degeneration-dev/PROJECT-MANIFEST) 中进行追踪*