Compiling Deterministic Structure into SLM Harnesses

arXiv:2604.17450v2 Announce Type: replace Abstract: Enterprise SLM deployment faces epistemic asymmetry: small models cannot self-correct reasoning errors, while frontier LLMs incur prohibitive costs and data sovereignty risks at scale. We propose Semantic Gradient Descent (SGDe), a teacher-student...