AdapterSwap: Continuous Training of LLMs with Data Removal and Access-Control Guarantees

Large language models (LLMs) are increasingly capable of completing knowledge intensive tasks by recalling information from a static pretraining corpus. Here we are concerned with LLMs in the context of evolving data requirements. For instance: batches of new data that are introduced periodically; s...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Fleshman, William, Khan, Aleem, Marone, Marc, Benjamin Van Durme
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 12.04.2024
Subjects
Access control
Data collection
Inference
Large language models
Online AccessGet full text

Cover

More Information
Summary:Large language models (LLMs) are increasingly capable of completing knowledge intensive tasks by recalling information from a static pretraining corpus. Here we are concerned with LLMs in the context of evolving data requirements. For instance: batches of new data that are introduced periodically; subsets of data with user-based access controls; or requirements on dynamic removal of documents with guarantees that associated knowledge cannot be recalled. We wish to satisfy these requirements while at the same time ensuring a model does not forget old information when new data becomes available. To address these issues, we introduce AdapterSwap, a training and inference scheme that organizes knowledge from a data collection into a set of low-rank adapters, which are dynamically composed during inference. Our experiments demonstrate AdapterSwap's ability to support efficient continual learning, while also enabling organizations to have fine-grained control over data access and deletion.
ISSN:2331-8422