Dvc

DVC, short for Data Version Control, is an open source tool that brings Git style version control to data, models and machine learning experiments. It works as an extension of Git: large datasets and model files are tracked with small metadata files committed alongside the code, while the actual content lives in remote storage, so a team can reproduce any past state of a project and share data without committing it to Git. It is aimed at individual data scientists and small to mid sized ML projects. Core capabilities include data and model versioning, pipelines defined in dvc.yaml with a run cache, metrics, parameters and plots, and experiment tracking and comparison, including Hydra composition. Remote storage backends include Amazon S3, Azure Blob Storage, Google Cloud Storage, Google Drive, Aliyun OSS, SSH and SFTP, HDFS, HTTP and WebDAV, and the documentation covers CI/CD for machine learning, data registries and integrations with Amazon SageMaker, Databricks, Evidently and Hugging Face. A VS Code extension provides a visual interface for experiments and plots. DVC is free and open source, and it has no pricing page. The site steers larger organizations toward lakeFS, a separate data version control system for petabyte scale object stores and data lakes, offered as a free open source edition and as lakeFS Enterprise, whose pricing is available through a demo request. DVC now sits alongside lakeFS on its website, which positions DVC for small data science projects and lakeFS for enterprise data infrastructure.