latent-label-data-augmentation

What problem does it solve?

Use this skill when a benchmark cannot support the desired latent labels cleanly and needs rewrites, matched pairs, counterbalancing, response generations, or synthetic augmentation. It covers benchmark repair for confounds, split construction, framing variants, and contrast-set design for mechanistic interpretability.

Core Features & Use Cases

• Guides structured augmentation moves (rewrite, pairing, counterbalancing) to reduce leakage and improve label validity.
• Enables generation of matched donor-target pairs and controls for causal questions, preserving core scenario semantics.
• Supports documentation and validation workflows to map repairs back to original gaps and to track phase status.

Quick Start

Provide an augmentation plan to rewrite, pair, or counterbalance latent labels for a benchmark repair.