Registry
W&B Registry is now in public preview. Visit this section to learn how to enable it for your deployment type.
W&B Registry is a curated central repository that stores and provides versioning, aliases, lineage tracking, and governance of models and datasets. Registry allows individuals and teams across the entire organization to share and collaboratively manage the lifecycle of all models, datasets and other artifacts. As the single source of truth for which models are in production, Registry provides the foundation for an effective CI/CD pipeline by identifying the right models to reproduce, retrain, evaluate, and deploy.
Use W&B Registry to:
- Bookmark your best artifacts for each machine learning task.
- Automate downstream processes and model CI/CD.
- Track an artifact’s lineage and audit the history of changes to production artifacts.
- Configure viewer, member, or admin access to a registry for all org users.
- Quickly find or reference important artifacts with a unique identifier known as aliases.
How it works
Track and publish your staged artifacts to W&B Registry in a few steps:
- Log an artifact version: In your training or experiment script, add a few lines of code to save the artifact to a W&B run.
- Link to registry: Bookmark the most relevant and valuable artifact version by linking it to a registry.
The following code snippet demonstrates how to log and link a model to the model registry inside W&B Registry:
import wandb
import random
# Start a new W&B run to track your experiment
run = wandb.init(project="registry_quickstart")
# Simulate logging model metrics
run.log({"acc": random.random()})
# Create a simulated model file
with open("my_model.txt", "w") as f:
f.write("Model: " + str(random.random()))
# log and link the model to the model registry inside W&B Registry
logged_artifact = run.log_artifact(artifact_or_path="./my_model.txt", name="gemma-finetuned-3twsov9e", type="model")
run.link_artifact(artifact=logged_artifact, target_path=f"<INSERT-ORG-NAME>/wandb-registry-model/registry-quickstart-collection"),
run.finish()
See learn more about linking to a registry, visit this guide.
Enable W&B Registry
Based on your deployment type, satisfy the following conditions to enable W&B Registry:
| Deployment type | How to enable |
|---|---|
| Multi-tenant Cloud | No action required. W&B Registry is available on the W&B App. |
| Dedicated Cloud | Contact your account team. The Solutions Architect (SA) Team will enable W&B Registry with your instance's operator console. Ensure your instance is on server release version 0.57.2 or newer. |
| Self-Managed | Enable the environment variable called ENABLE_REGISTRY_UI. To learn more about enabling environment variables in server, visit these docs. In self-managed instances, your infrastructure admin should enable this environment variable and set it to true. Ensure your instance is on server release version 0.57.2 or newer. |
Resources to get started
Depending on your use case, explore the following resources to get started with the W&B Registry:
Check out the tutorial video:
Take the W&B Model CI/CD course and learn how to:
- Use W&B Registry to manage and version your artifacts, track lineage, and promote models through different lifecycle stages.
- Automate your model management workflows using webhooks and launch jobs.
- See how Registry integrates with external ML systems and tools in your model development lifecycle for model evaluation, monitoring, and deployment.
Migrating from the legacy Model Registry to W&B Registry
The W&B Model Registry will be deprecated by the end of 2024. The contents in your Model Registry will be migrated to W&B Registry. Detailed information about the migration process from the legacy Model Registry to Registry will be posted soon.
The soon to be legacy W&B Model Registry App UI is still available until W&B Registry is made generally available. To view the legacy Model Registry: Navigate to the Model Registry from the homepage. A banner will appear to view the legacy Model Registry App UI.
Reach out to support@wandb.com with any questions or to speak to our product team about any concerns with the migration.