Effective long-term memory management is crucial for language models handling extended contexts. We introduce a novel framework that dynamically ranks memory entries based on relevance. Unlike previous works, our model introduces a novel relevance scoring and a pointwise re-ranking model for key-value embeddings, inspired by learning-to-rank techniques in information retrieval. Enhanced Ranked Memory Augmented Retrieval ERMAR achieves state-of-the-art results on standard benchmarks.
View on arXiv@article{alselwi2025_2503.14800,
title={ Long Context Modeling with Ranked Memory-Augmented Retrieval },
author={ Ghadir Alselwi and Hao Xue and Shoaib Jameel and Basem Suleiman and Flora D. Salim and Imran Razzak },
journal={arXiv preprint arXiv:2503.14800},
year={ 2025 }
}