Amazon introduces two benchmark datasets for evaluating AI agents’ ability on code migration
DevOps & Developer Productivity Blog
This article introduces two benchmark datasets for evaluating AI agents' code migration capabilities across multiple programming languages and platforms.
- MigrationBench: 5,102 Java 8 repositories for migration to Java 17/21
- Poly-MigrationBench extends to .NET Framework, Node.js, and Python migrations
- Includes 100 .NET Framework, 74 Node.js, 83 Python repositories
- Provides standardized evaluation framework for reproducible research
- Addresses real-world migration challenges like AWS Lambda runtime upgrades
- Datasets and evaluation tools open-sourced on GitHub and Hugging Face
These benchmarks enable the research community to systematically evaluate and develop automated code migration solutions at repository scale.
The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.
Related articles
Dec 1
2025
2025
AWS expands AI Competency with new Agentic AI categories
Nov 19
2025
2025
How Amazon uses AI agents to support compliance screening of billions of transactions per day
Apr 23
2025
2025
Amazon introduces SWE-PolyBench, a multilingual benchmark for AI Coding Agents
Dec 1
2025
2025
AWS Transform adds new agentic AI capabilities for enterprise VMware migrations
The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.