DVC

Model Store Overview

Vendor Name

DVC

Stand-alone vs. Platform

Stand-alone

Delivery Model

Open-source

Clouds Supported

Open-source

Pricing Model

Open-source

Service Level Guarantees

None

Support

Open-source: community (Discord server, GH issues, self-hosted forum, support email, social media)

SSO, ACL

N/A

Security and Compliance

None

Model Store Capabalities

Setup

Familiarity with Git workflows and CLI.

Flexibility, Speed, and Accessibility

Supports many cloud storage providers: S3, Azure, Google Cloud, Google Drive, Aliyun, SSH, HDFS, HTTP, WebDav (https://dvc.org/doc/command-reference/remote/modify#available-parameters-per-storage-type)

Model Versioning, Lineage, and Packaging

Uses Git to track metadata/hashes of data&models.
Uses YAML to store reproducible pipelines (data download sources, commands with data inputs & outputs).

Log and Display of Metadata

Supports parameters, metrics, plots, source control, and large artifacts like data and models across multiple commits & experiments.
Checkpoints for live per-epoch logging.

Comparing Experiments and Models

Tables & plots (see above) for parameters & metrics, including overlaid learning curves

Organizing and Searching Experiments and Models

Use Git-based experiments workflow to view and filter/sort by parameters, metrics, and commits.
Save results in standard Git workflow (branches, tags, commit messages, git-blame, git-grep, etc).

Model Review, Collaboration, and Sharing

N/A (user-provided cloud storage credentials can have limited permissions)

CI/CD/CT Compatibility

Cloud storage provider credentials may be provided as secret env vars in CI/CD

Integrations

Specifically designed to be language and user-workflow agnostic. Supports and integrates with anything.

Reviews

There are no reviews yet. Be the first to write one.