Loading page…
The paper keeps the Memorizer fixed during Researcher training and does not study Memorizer fine-tuning or joint Memorizer–Researcher optimization. The backbone comparison isolates model choice with a fixed trained Researcher. · CiteArk