When to Adapt: Conditional Memory Adapters for Retention-Preserving Domain Specialization
Engram Adapter, a framework that repurposes pretraining-time conditional memory as a post-hoc adapter for frozen LLMs, improves in-domain accuracy while preserving 99.4%--100.1% of average OOD performance; on LegalBench it slightly exceeds the frozen base model on average, whereas comparable always-on baselines degrade...