Pachyderm

Model Store Overview

Vendor Name

Pachyderm

Stand-alone vs. Platform

Stand-alone tool

Delivery Model

Open-source.
Commercial and SaaS offerings come with additional enterprise features.

Clouds Supported

Any Kubernetes environment, cloud, local
SaaS offering

Pricing Model

Seat-based pricing for on-prem deployments.
Usage based pricing for hosted SaaS Service.
Expert services available

Service Level Guarantees

Pachyderm Enterprise: SLAs available but dependent on the plan.
Two support tiers available.

Support

Open source: Community
Pachyderm Enterprise: 24x7 available depending on plan

SSO, ACL

Pachyderm Enterprise provides full user and role-based access controls which allows for granular permission setting

Security and Compliance

Pachyderm Enterprise: advanced user management options

Model Store Capabalities

Setup

Installed via Helm
Data can be connected to training workflow by using S3 gateway or client SDKs
Data processing, transformations, and model training pipelines can be deployed at scale inside the product

Flexibility, Speed, and Accessibility

Pachyderm as a model store aims to be maximally flexible and scalable
Any type of file can be inserted into it, automatically generating associated metadata
Models and metadata can be accessed using the S3 gateway or one of the client SDKs

Model Versioning, Lineage, and Packaging

Traceability and compliance are primary features of the tool
All files are automatically versioned in Pachyderm
Any pipeline that is run in Pachyderm retains a full lineage for each version of your model (or any other artifact)

Log and Display of Metadata

Logs are automatically kept for any pipeline automatically and can be viewed in the Console.
For fine-grained experiment comparisons and monitoring, Pachyderm integrates with experimentation platforms

Comparing Experiments and Models

Users can create their own comparisons via pipelines.
The flexibility allows any type of comparison to be written and visualized.

Organizing and Searching Experiments and Models

Datasets, experiments, and models are can be organized into data repositories
These repositories can be searched in the Console view for files

Model Review, Collaboration, and Sharing

Any model (or any other artifact like a dataset), can be collaborated on and shared
All changes to the object are versioned, allowing the promotion and demotion of models/artifacts via branches or labels
All files in the system are immutable

CI/CD/CT Compatibility

Pipelines run automatically when artifacts change, functioning as CI/CD for artifacts
Pachyderm can also integrate with existing GitOps to incorporate code changes into pipelines

Integrations

Pachyderm integrates with a host of 3rd party ML tools. See the Pachyderm blog for a full list of supported integrations.

Reviews

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