Full Pipeline — chaining stages

What it does

Every stage can run standalone — the Pipeline page chains them into one reproducible computational graph: data synthesis → evaluation → prediction → optimization, with TEA, synthetic-data nodes and more as intermediate steps.

The graph editor

Build the pipeline visually: add nodes, connect outputs to inputs, configure each stage, then execute the whole workflow from one screen. A pipeline run is recorded as a job like any other.

Automatic handoffs

Each stage's output becomes the next stage's input — no manual export/import between stages. Synthetic data is natively available as an intermediate node.

Reproducibility

Every pipeline run keeps its full configuration, so you can revisit, rerun and share it with collaborators — the entire experimental workflow is one artifact.