Tell Me What To Learn: Generalizing Neural Memory to be Controllable in Natural Language

\Name Max S. Bennett \Email [email protected] \addr Columbia University \Name Thomas Zollo \Email [email protected] \addr Columbia University \Name Richard Zemel \Email [email protected] \addr Columbia University · 2026

Read Tell Me What To Learn: Generalizing Neural Memory to be Controllable in Natural Language by \Name Max S. Bennett \Email [email protected] \addr Columbia University \Name Thomas Zollo \Email [email protected] \addr Columbia University \Name Richard Zemel \Email [email protected] \addr Columbia University (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-23201.txt.

Description

Max Bennett and two co-authors propose a neural memory system users can steer with plain-language instructions. Modern models work in shifting settings where they must keep adapting to new tasks and new knowledge. Continual fine-tuning and in-context learning are costly and brittle, while neural memory methods promise light updates with little forgetting.

Existing memory models, however, assume one fixed goal and one uniform stream of information. They leave users no control over what the model keeps or drops over time. The new system takes learning instructions in natural language and updates selectively from varied sources. That control matters in settings such as health care and customer service, where fixed-goal memory updates fall short.

Catalog record

CREATOR\Name Max S. Bennett \Email [email protected] \addr Columbia University \Name Thomas Zollo \Email [email protected] \addr Columbia University \Name Richard Zemel \Email [email protected] \addr Columbia University
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE129 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-23201

Browse the full library · About the Tower of Babel