IBM/Project_CodeNet resource

Health v2 · maintenance only

Adoption not part of the score

1689 stars · 203 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Project CodeNet is a large-scale curated dataset of ~14 million code submissions across 50+ programming languages, built to support AI-for-Code research such as code search, summarization, completion, and code-to-code translation. This repository hosts tools and utilities in Python for accessing, selecting, and pre-processing the dataset, which is hosted on DAX.

Use cases

• train machine learning models on source code
• build code-to-code translation models
• research code summarization and completion
• benchmark AI models for code understanding
• filter and preprocess large code corpora by language or problem
• study competitive programming submissions for AI research

When to choose

• you need a large, diverse, curated dataset of code with metadata for AI-for-Code research
• you want benchmark problems with solutions in many programming languages
• you need Python tooling to access and preprocess code datasets

When to avoid

• you need a production code analysis tool rather than a research dataset
• you want a small, clean dataset for quick experiments
• you need real-world production code rather than competitive programming submissions

Facets

dataset · maturity stable

machine-learning nlp data-science parser artificial-intelligence machine-learning programming-languages education developer-tools python cross-platform ai-for-code code-dataset code-generation code-translation benchmark competitive-programming

Member repositories

For agents

markdown · JSON · MCP: product_card(name="IBM/Project_CodeNet")