This paper proposes Supervised Memory Training (SMT), a method for training nonlinear RNNs that sidesteps recurrent credit propagation entirely.

This paper introduces KGERMAR, a framework that constructs dynamic, context-specific knowledge graphs during inference for long-context language modeling, achieving lower perplexity and better memory…

MemGuard introduces a type-aware memory framework to prevent heterogeneous memory contamination in long-term memory-augmented LLMs, significantly improving memory reliability and efficiency.

The paper introduces Reasoning in Memory (RiM), a latent reasoning method that replaces autoregressive token generation with fixed memory blocks to enable compute-efficient internal working memory for…

ElasticMem introduces a novel framework that treats memory as an elastic latent resource, allowing LLM agents to adaptively manage and inject variable-budget memories for improved performance in long-…

The paper quantifies the exact parametric memory capacity of LLMs using LoRA and proposes a new optimization strategy, MemFT, to enhance memory fidelity.

The paper introduces VikingMem, a novel Memory Base Management System that effectively manages the persistent state of long-term LLM interactions by selectively extracting, evolving, and compressing m…

The paper introduces and evaluates five parameter alignment strategies that significantly mitigate catastrophic forgetting when continually pretraining multilingual expert language models across multi…

MemGraphRAG introduces a novel memory-based multi-agent system to construct globally consistent and structurally sound knowledge graphs, significantly improving retrieval-augmented generation for comp…

The paper introduces RHELM, a new benchmark designed to test LLMs' long-term memory by simulating realistic, complex, and evolving dialogues that integrate multiple heterogeneous data sources.

The paper introduces RefMem-Bench, a new benchmark for measuring reflective memory in long-horizon dialogue, and proposes REMIND, a framework that significantly improves models' ability to synthesize…

This study systematically evaluates a wide range of chunking methods for Retrieval-Augmented Generation (RAG) to assess their effectiveness and highlight the overlooked challenges associated with chun…

This paper proposes a multi-turn retrieval-augmented generation pipeline for conversational systems across four domains.

The paper proposes LaSR, a context-aware training paradigm that uses latent reasoning to significantly improve speech recognition, especially for specialized terminology, without adding latency.

PromptEmbedder introduces a dual-LLM framework that efficiently and transferably adapts text embeddings by decoupling task-specific knowledge from the backbone model, significantly reducing computatio…

This paper introduces a new scaling law for sparse language models trained with limited data, demonstrating that sparsity can significantly improve performance and delay data saturation during multi-e…

The paper reframes context distillation as a latent memory management problem, proposing a modular framework using LoRA adapters and a Self-Gating mechanism for efficient, selective memory retrieval a…

GRKV introduces a training-free KV-cache merging method that uses global regression to distribute information from evicted tokens, solving the over-merging problem inherent in span-based retention.

The paper introduces MemCog, a Memory-as-Cognition system that integrates memory access directly into the reasoning process, significantly improving agent performance, especially in proactive memory r…

This paper demonstrates that targeted adaptation of the small language model Phi Silica, using dataset curation and fine-tuning, significantly improves its performance in short-form text rewriting, na…