Memory Dial: A Training Framework for Controllable Memorization in Language Models

ArXi:2604.05074v1 Announce Type: new Memorization in language models is widely studied but remains difficult to isolate and control. Understanding when and what models memorize is essential for explaining their predictions, yet existing approaches are post-hoc: they can detect memorization in trained models, but cannot disentangle its effects from architecture, data, or optimization. We