DVC (Data Version Control)
About this tool
A CLI for data version control and ML experiment management. Git stores version metadata for large datasets and models, while files live in cloud or other remote storage. Define pipelines and rerun only steps affected by changed dependencies. dvc exp runs and compares experiments locally, helping groups reproduce data-intensive research.
- Feature 1: Version data and models in Git metadata, with files on S3, Azure, Google Cloud, SSH or other storage.
- Feature 2: Declare pipeline dependencies and outputs; rerun affected steps only.
- Feature 3: Run, compare and apply experiments with dvc exp.
- Feature 4: A VS Code extension supports experiments and data management.
- Feature 5: The CLI can be wrapped as a skill for agents.
Resource information
- Maintainer
- Treeverse (lakeFS); took over from Iterative in November 2025.
- Type
- Tool / library
- Category
- Experiments & data analysis
- Runs on
- Tool or library
- Who it's for
- ML, computer-vision, bioinformatics and other data-intensive teams fixing dataset and model versions for reproducible experiments and collaboration.
- License at time of listing
- Apache-2.0
- Source
- GitHub
