Feature Guides

Guides for runtime diagnostics, missingness, many-class, shift, noise, and benchmark guardrails.

Feature-specific guides for runtime diagnostics and data-shift controls.

Runtime observability metrics and diagnostic outputs.

Controlled injection of missing values into generated datasets.

Multi-class target generation with configurable class counts.

Distribution-shift controls for graph, mechanism, and noise.

Noise family selection, mixture modes, and per-dataset resolution.

Automated quality checks for benchmark suite runs.

Opt-in mechanism-family rollout workflows and diagnostics checks.