Datasetpublic

happylife365/code-quality-poor

Low Quality Code Dataset Overview This dataset contains 444 Python code samples with poor quality scores (62-74 out of 100). These samples can be used for: Training models to recognize bad code patterns Contrastive learning (good vs bad code) Code quality classification tasks Statistics Metric Value Total samples 444 Quality range 62-74 Average quality ~70 Source Distribution Source Count BigOBench 425… See the full description on the dataset page: https://huggingface.co/datasets/happylife365/code-quality-poor.

sourceHugging Faceupdated 7mo agoView on Hugging Face

0likes3downloads

Dataset Card

Low Quality Code Dataset

Overview

This dataset contains 444 Python code samples with poor quality scores (62-74 out of 100).

These samples can be used for:

—Training models to recognize bad code patterns
—Contrastive learning (good vs bad code)
—Code quality classification tasks

Statistics

Metric	Value
Total samples	444
Quality range	62-74
Average quality	~70

Source Distribution

Source	Count
BigOBench	425
OpenCodeReasoning-0	9
rStar-Coder-seed	6
rStar-Coder-synthetic	3
OpenCodeReasoning-1	1

Common Issues in Poor Quality Code

—Long methods (>50 lines)
—Deep nesting (5+ levels)
—Low/no comments
—High cyclomatic complexity
—PEP8 violations

Fields

Field	Description
`code`	The Python code snippet
`source_dataset`	Original dataset
`quality_score`	Overall quality (62-74)
`quality_tier`	"poor"
`loc`	Lines of code
`pylint_score`	Pylint score (0-10)
`nesting_depth_max`	Maximum nesting

Usage

python

from datasets import load_dataset

dataset = load_dataset("happylife365/code-quality-poor")

Related Dataset

—High quality code: happylife365/code-quality-annotated

Generated: 2026-02-04