README.md

metadata

license: cc-by-sa-4.0
task_categories:
  - text-generation
language:
  - en
tags:
  - code
pretty_name: Doocs LeetCode Solutions
size_categories:
  - 10K<n<100K

Doocs LeetCode Solutions

A comprehensive dataset of LeetCode problems and solutions created from the Doocs LeetCode repository. This dataset is designed for fine-tuning large language models to understand programming problems and generate code solutions.

Description

• Repository: Doocs LeetCode Solutions
• Total Problems: 3500+
• Total Solutions: 15,000+ (across multiple languages)
• Size: ~60 MB (Parquet format)
• Languages:
  • C
  • Cangjie
  • C++
  • C#
  • Dart
  • Go
  • Java
  • JavaScript
  • Kotlin
  • Nim
  • PHP
  • Python
  • Ruby
  • Rust
  • Scala
  • Bash
  • SQL
  • Swift
  • TypeScript

Structure

Data Fields

Field	Type	Description	Example
id	string	Problem ID	"0001"
title	string	Problem title (slugified)	"two-sum"
difficulty	string	Problem difficulty level	"Easy", "Medium", "Hard"
description	string	Problem description in Markdown	"Given an array of integers..."
tags	string	Problem category tags	"Array; Hash Table"
language	string	Programming language of solution	"Python", "Java", "C++"
solution	string	Complete solution code	"class Solution:\n def..."

Data Splits

The dataset contains a single comprehensive split containing all problems and solutions:

Split	Problems	Solutions
train	3500+	15,000+

How to Use

Using Hugging Face datasets Library

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("olegshulyakov/doocs-leetcode-solutions")

# Access a sample
sample = dataset['train'][0]
print(f"Problem {sample['id']}: {sample['title']}")
print(f"Difficulty: {sample['difficulty']}")
print(f"Tags: {sample['tags']}")
print(f"Language: {sample['language']}")
print(f"Solution:\n{sample['solution']}")

Dataset Creation

Source Data

• Collected from Doocs LeetCode repository
• Solutions cover 14+ programming languages
• Includes problems from LeetCode's entire problem set

Preprocessing

1. Repository cloned and processed using our dataset generator tool.
2. Metadata extracted from README_EN.md files.
3. Solution files parsed and mapped to programming languages.
4. Special characters and encoding issues resolved.

Intended Use

Primary Use

• Fine-tuning code generation models
• Training programming problem-solving AI
• Educational purposes for learning algorithms

Possible Uses

• Benchmarking code generation models
• Studying programming patterns across languages
• Analyzing problem difficulty characteristics
• Creating programming tutorials and examples

Limitations

• Solutions may not be optimal (community solutions)
• Some edge cases might not be covered
• Problem descriptions may contain markdown/html formatting
• Limited to problems available in the Doocs repository

License

This dataset is licensed under the CC-BY-SA-4.0 License - see LICENSE file for details.

Copyright

The copyright of this project belongs to Doocs community.

Contact

For questions or issues regarding the dataset, please open an issue on GitHub.