SOURCE-LINKED INTELLIGENCE

How Output Format Confounds Data Quality and Capability in Instruction Tuning

arXiv · AI, language, vision and robotics · article · Sep 2, 2026 · UTC

Instruction-tuning data are judged by quality metrics, and tuned models are judged by benchmarks, but both judgments pass through an output interface: the surface format in which an answer is written. Using gradient signatures across 12 tasks, four semantically equivalent interfaces, three model families, and controlled corruptions, we show that this interface confounds both measurements. Spectral statistics such as effective rank are provably invariant to interface rotation and empirically blind to semantic corruption, while the direction of the update carries the quality signal. The interfac

Read original source ↗ Open in workspace

recordType

paper

region

Global