Nvidia’s new technique cuts LLM reasoning costs by 8x without losing accuracy

Nvidia researchers developed dynamic memory sparsification (DMS), a technique that compresses the KV cache in large language models by up to 8x while maintaining reasoning accuracy — and it can be ...

'Observational memory' cuts AI agent costs 10x and outscores RAG on long-context benchmarks

As AI agents move into production, teams are rethinking memory. Mastra’s open-source observational memory shows how stable ...
Science Daily

Dynamic DNA structures and the formation of memory

An international collaborative research team has discovered that G-quadraplex DNA (G4-DNA) accumulates in neurons and dynamically controls the activation and repression of genes underlying long-term ...
Nature

Dynamic Random Access Memory Technologies

Dynamic random access memory (DRAM) remains a cornerstone of modern electronic systems, enabling rapid data storage and retrieval. Recent developments have focused on capacitorless designs – notably ...
Semiconductor Engineering

Dynamic KV Cache Scheduling in Heterogeneous Memory Systems for LLM Inference (Rensselaer Polytechnic Institute, IBM)

A new technical paper titled “Accelerating LLM Inference via Dynamic KV Cache Placement in Heterogeneous Memory System” was published by researchers at Rensselaer Polytechnic Institute and IBM. “Large ...
Frontiers

Computational Models of Memory and Attention: Advancing Mechanistic Understanding of Brain with Artificial Intelligence and Beyond

Human memory and attention are core cognitive functions that shape perception, learning, and decision-making. And whilst decades of research have provided ...
Nature

Dynamic Random Access Memory Security and Efficiency

Dynamic Random Access Memory (DRAM) remains a central element in computing architectures, but its intrinsic vulnerabilities and power demands have spurred a wealth of research focused on enhancing ...