Discover & Fork
Browse thousands of public graphs from the community, fork what works, and build on research done by other people.
The Community tab
Open the Graphs Hub and click Community. You'll see a catalog of public graphs published by other users. Each card shows:
- The graph name and a one-line description.
- A miniature preview of the canvas.
- The author, license, and last update date.
- Tags (e.g.,
computer-vision,fine-tuning,dataset).
Searching and filtering
Use the search bar at the top to query by keyword. Filters let you narrow by:
- Tag — Domain or technique.
- Node types used — e.g., graphs that use a specific Custom node you also use.
- Stars — Sort by popularity.
- Recency — Newest first, or recently updated.
Previewing without forking
Click any card to open the graph in View Mode. View Mode is fully read-only:
- You can pan, zoom, and click nodes to inspect their config.
- You can trace paths and see run data if the author published runs.
- You cannot edit anything.
View Mode is the right place to evaluate whether a graph fits your needs before forking.
Forking
Click Fork in the top-right of View Mode. Connectify creates a new project in your workspace that's a complete copy of the source graph:
- All nodes, connections, subgraphs, and variants.
- Config values on every node.
- A reference back to the source for attribution.
Forks are independent — edits in your fork don't propagate back, and updates to the source don't override your fork.
Forks remember their origin
The forked project has a small "Forked from …" chip in the header. Click it to jump to the source — useful when you want to check what the original author changed.
Publishing your own graph
When you're ready to share, open the project, click Publish, and fill in:
- Title and description.
- Tags (be specific —
image-classificationbeatsml). - License (MIT, Apache 2.0, CC-BY, etc.).
- Visibility (Public, Unlisted, or Private).
Published graphs appear in the Community tab within a few minutes. You can unpublish or update at any time.
Attribution
When you fork a graph, the source is recorded automatically. If you build on someone's work and publish your version, please leave the attribution chip in place — it's how the community finds related work and how authors get credit.