Codeforces Competitive Programming Submissions

Millions of real user code submissions from the Codeforces competitive programming platform. The dataset contains human solutions to challenging algorithmic optimization problems, curated by open-r1 and last updated in May 2025.

Use Cases

• Train a code generation model on millions of human-written algorithmic solutions to improve code reasoning capabilities.
• Analyze patterns in user submissions for specific algorithmic problem categories to identify common solution strategies.
• Benchmark the performance of large language models on competitive programming tasks using real-world human code as a reference.
• Study the evolution of coding styles and solution efficiency across different contest difficulty levels.

Strengths

• Contains millions of real-world human code submissions.
• Focuses on challenging algorithmic optimization problems from a popular competitive programming platform.
• Data is sourced from Codeforces, a major hub for competitive programmers.

Limitations

• Specific details on rows, columns, file formats, and data structure are unknown.
• Potential for incomplete metadata or missing context about individual contests and problems.
• License information is not provided, which may restrict certain uses.

Provenance

Source

Codeforces website

Collection Method

Collection of real user code submissions from the platform.

Time Range

null

Freshness

Last updated 2025-05-14.

Geography

null