When a Kindergartener Solves Calculus: Measuring Capability Leakage in Role-Prompted Reasoning Models
Injection is proposed, an inference-time intervention that combines explicit, role-specific capability guidelines with a guiding prefilled response prefix that improves role-capability alignment across models, reducing above-role accuracy by up to 0.562 while preserving in-role accuracy with a marginal drop of less tha...
Pakhapoom Sarapat, Saksorn Ruangtanusak, Kunat Pipatanakul et al.
· 0 citations