Appearance
Simulation Workflow Builder
Assemble a ColliderML simulation pipeline stage by stage — geometry → generation → simulation → digitization → reconstruction — then validate it and submit it to the SaaS backend. Add or remove stages to build exactly the workflow you want.
The Copilot (bottom-right, on every page) can read and edit this pipeline for you: ask it to "build a ttbar pu200 workflow and validate it" and watch the cards appear. On other pages the same copilot answers documentation questions.
No stages yet. Add one below, or ask the copilot to build a pipeline for you.
HuggingFace token (for live submit)
Verified by the backend against huggingface.co; needed to authorize /v1/simulate.
How submission works
Submit sends {channel, events, pileup, seed} to POST /v1/simulate on the backend (COLLIDERML_BACKEND, default http://localhost:8000). The endpoint verifies your HuggingFace token, then dispatches to Perlmutter via the Superfacility API — or completes in mock mode (~2s) when SFAPI credentials aren't set. Results land as a HuggingFace dataset you can pull with colliderml.load(...).