README.md

metadata

license: apache-2.0
dataset_info:
  features:
    - name: repo
      dtype: string
    - name: base_commit
      dtype: string
    - name: license
      dtype: string
  splits:
    - name: test
      num_examples: 3
programming_languages:
  - java
tags:
  - coding
  - code-migration
  - java
  - amazon
  - amazon-science
  - aws

MIGRATION-BENCH

1. 📖 Overview

🤗 MIGRATION-BENCH is a large-scale code migration benchmark dataset at the repository level, across multiple programming languages.

• Current and initial release includes java 8 repositories with the maven build system, as of May 2025.

2. Datasets

There are three datasets in 🤗 MIGRATION-BENCH:

• All repositories included in the datasets are available on GitHub, under the MIT or Apache-2.0 license.

Index	Dataset	Size	Notes
1	🤗 AmazonScience/migration-bench-java-full	5,102	Each repo has a test directory or at least one test case
2	🤗 AmazonScience/migration-bench-java-selected	300	A subset of 🤗 migration-bench-java-full
3	🤗 AmazonScience/migration-bench-java-utg	4,814	The unit test generation (utg) dataset, disjoint with 🤗 migration-bench-java-full

3. Metadata

Metadata is provided in the csv file for each dataset:

1. repo (str): The original repo URL without the https://github.com/ prefix
2. base_commit (str): Base commit id
   • At this commit with java 8 and maven, the repository is able to (1) compile and (2) pass existing unit tests and integration tests if any
   • It is the starting point for code migration from java 8
3. license (str): The license of the repository, either MIT or Apache2.0 for the whole dataset

4. 📚 Citation