DVC
Data & ML🦉 Data Versioning and ML Experiments
Release History
3.66.1This patch release includes minor non‑functional changes: a typo fix in the flatten‑dict dependency and added logging of paths during dry‑run garbage collection.
3.66.0Breaking1 featureThis release restricts the `pathspec` dependency to versions below 1, introduces the `--collapse-foreach-matrix` CLI option, and updates the `flufl-lock` version bounds.
3.65.0Breaking1 featureVersion 3.65.0 adds bulk remote entry checks and drops the pytest-test-utils dependency, requiring test suite updates if that utility was used.
3.64.21 featureThe release introduces a new `bearer_token_command` feature to the WebDAV backend, enabling dynamic token acquisition.
3.64.11 featureThe release adds support for the "--no-hydra" flag to the exp command.
3.64.03 fixes4 featuresThis release adds Python 3.14 support, enhances the move command, and introduces config variable completion, while fixing several bugs.
3.63.0Breaking4 fixes7 featuresThis release introduces several new CLI options and enhancements, adds granular change reporting, improves detection of file moves, and includes multiple bug fixes, with a breaking change to `dvc status --cloud` target handling.
3.62.06 fixes3 featuresVersion 3.62.0 introduces new flags and hints for experiments, improves hashing security, and includes several bug fixes across data status, analytics, CLI, and experiment execution.
3.61.02 fixes3 featuresThis release adds target‑limited data status, ignores non‑remote files when push is disabled, and introduces a wait_for_lock option, along with bug fixes for data status in new Git repos and cache directory handling.
3.60.13 fixesThis patch release (3.60.1) fixes cache directory handling and updates the dvc-s3 dependency.
3.60.01 fix2 featuresVersion 3.60.0 introduces a renamed argument in dvcfs, adds anonymous login for the gs filesystem, fixes S3 configpath handling, and bumps the minimum dvc-gs version to 3.0.2.
3.59.22 fixes1 featureThis release adds a default display for "dvc remote list", forwards arguments to _get_remote_config respecting core/no_scm, and improves repository opening robustness, along with minor maintenance updates.
3.59.12 fixesPatch release 3.59.1 resolves a broken progress bar during folder imports and prevents showing zero or negative sizes for directory entries in ls/ls-url.
3.59.03 featuresThis release introduces tree and level options for the `ls-url` command and adds official Python 3.13 support, along with an updated Studio URL.
Common Errors
ModuleNotFoundError1 reportThe "ModuleNotFoundError" in dvc, like the "no module gpg" example, indicates a missing Python package dependency. To fix this, install the necessary package using pip. For example, run `pip install python-gnupg` to resolve the `no module gpg` error; ensure the correct package name is used.
GitAuthError1 reportGitAuthError in DVC usually indicates authentication problems when DVC tries to interact with a remote Git repository, often because of SSH key issues or proxy configurations. Ensure your SSH key is properly configured and added to your SSH agent, and that your `dvc remote` and Git configurations have the correct URL using SSH instead of HTTPS. If using a proxy, properly set up your `http_proxy` and `https_proxy` environment variables or Git's proxy settings to allow DVC and Git to communicate through the proxy server.
ArtifactNotFoundError1 reportArtifactNotFoundError in DVC usually means a DVC-tracked file or directory is missing from your local workspace or remote storage. To fix this, first ensure the missing files are committed to DVC and the corresponding data cache exists; then, run `dvc pull` to download the missing artifacts from your remote storage into your local workspace. Finally, verify that the submodule data is also pulled.
FileNotFoundError1 reportFileNotFoundError in dvc often arises when dvc cannot locate a file or directory referenced in your dvc.yaml, stage definitions, or tracked data. Double-check all file paths in your dvc.yaml, stage definitions (dependencies/outputs), and experiments to ensure they exist and are correctly specified relative to your project's root. Specifically, verify capitalization and spelling, and use absolute paths if necessary to avoid ambiguity.
Related Data & ML Packages
An Open Source Machine Learning Framework for Everyone
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
Tensors and Dynamic neural networks in Python with strong GPU acceleration
scikit-learn: machine learning in Python
Flexible and powerful data analysis / manipulation library for Python, providing labeled data structures similar to R data.frame objects, statistical functions, and much more
Streamlit — A faster way to build and share data apps.
Subscribe to Updates
Get notified when new versions are released