Frameworks

Introduction

Diffgram has two frameworks for developing.

Workflows (Backend)
Scripts (Frontend)

Both frameworks unlock a growing ecosystem of ML programs, provide opinionated development options, and standard ways to install other tools.

Examples of areas include:

Interaction with an Ecosystem of tech
MLOps
ML Modeling
ML Programs (Such as sampling, filtering)
Training Data Orchestration
Human task workflow
Surfacing your Training Data Processes
QA Assurance programs
ML Maintenance

Both frameworks are in "Preview" on our Quality Bar Status.
We encourage you to try them out and share your insights with the community.

Workflow Backend

Workflows

Interactive Scripts Frontend

Scripts

SDK, CLI, Standard Ecosystem

Ecosystem
Please note the SDK, CLI, etc. are considered part of the standard ecosystem and interact with frameworks.

Baseline, Frameworks and Components

The baseline configuration supports anything that can be reasonably implemented or configured directly.

Frameworks are more open ended, in that you can integrate or install 3rd party programs, write your own programs that integrate, or extend functions etc. Frameworks are opinionated.

Components are open further, allowing you to develop custom software that uses existing Diffgram technical components, such as JS classes, Vue components, etc.

Direction

Over time we expect these frameworks will further integrate with each other, and other diffgram development options.