Long Context Modeling with Ranked Memory-Augmented Retrieval

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Alselwi, Ghadir, Xue, Hao, Jameel, Shoaib, Suleiman, Basem, Salim, Flora D., Razzak, Imran
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917506754543616
author Alselwi, Ghadir
Xue, Hao
Jameel, Shoaib
Suleiman, Basem
Salim, Flora D.
Razzak, Imran
author_facet Alselwi, Ghadir
Xue, Hao
Jameel, Shoaib
Suleiman, Basem
Salim, Flora D.
Razzak, Imran
contents Effective long-term memory management is crucial for language models handling extended contexts. We introduce the Enhanced Ranked Memory Augmented Retrieval (ERMAR) framework, which dynamically ranks memory entries based on relevance. Unlike prior models, ERMAR employs a novel relevance scoring mechanism and a pointwise re-ranking model for key-value embeddings, inspired by learning-to-rank techniques in information retrieval. By integrating historical usage patterns and adaptive retrieval, ERMAR achieves state-of-the-art results on standard benchmarks, demonstrating superior scalability and performance in long-context tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2503_14800
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Long Context Modeling with Ranked Memory-Augmented Retrieval
Alselwi, Ghadir
Xue, Hao
Jameel, Shoaib
Suleiman, Basem
Salim, Flora D.
Razzak, Imran
Information Retrieval
Artificial Intelligence
Machine Learning
Effective long-term memory management is crucial for language models handling extended contexts. We introduce the Enhanced Ranked Memory Augmented Retrieval (ERMAR) framework, which dynamically ranks memory entries based on relevance. Unlike prior models, ERMAR employs a novel relevance scoring mechanism and a pointwise re-ranking model for key-value embeddings, inspired by learning-to-rank techniques in information retrieval. By integrating historical usage patterns and adaptive retrieval, ERMAR achieves state-of-the-art results on standard benchmarks, demonstrating superior scalability and performance in long-context tasks.
title Long Context Modeling with Ranked Memory-Augmented Retrieval
topic Information Retrieval
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2503.14800