6 papers · ranked by Valyu relevance
Arsham Mikaeili Namini, Ali Saberi, Hamed S Najafabadi, Peter Robinson
Runtime benchmarking demonstrated substantial improvements from both architectural redesign and parallelization ([btag334-F1]). Even without parallelism, GEDI 2.0 ran up to $∼$3 $\times$ faster than the legacy implementation, confirming substantial performance gains from improved algorithms and memory access patterns…
Ragnar Bjornsson
We introduce ASH, a multi-scale, multi-theory modeling program for quantum mechanics (QM), molecular mechanics (MM), and hybrid calculations, written in the Python programming language. ASH is written in response to the increasingly diverse computational chemistry software landscape that features more QM and MM…
Sangjin Lee, Sunggon Kim, Yongseok Son, Agbotiname Lucky Imoize
We propose ScaleDefrag, a parallel and asynchronous defragmentation tool that reduces defragmentation time by up to 3.8× compared to e4defrag, while improving scalability on multi-core systems. Flash-based solid-state drives (SSDs) have been widely adopted in various large-scale storage systems including cloud and HPC…
Teh Noranis Mohd Aris, Ningning Chen, Norwati Mustapha, Maslina Zolkepli + 1 more
To address the inefficiencies in sample utilization and policy instability in asynchronous distributed reinforcement learning, we propose TPDEB-a dual experience replay framework that integrates prioritized sampling and temporal diversity. While recent distributed RL systems have scaled well, they often suffer from…
Annika Österdiekhoff, Nils Wendel Heinrich, Nele Russwinkel, Stefan Kopp + 1 more
Having to control multiple tasks in parallel poses challenges for humans and artificial agents alike. In artificial intelligence, specific forms of reinforcement learning (RL), most notably hierarchical and model-based RL, have shown promising results in scenarios where tasks or skills need to be switched adaptively.…
Kai Xu, Diming Zhang, Xuguo Wang, Alessandra Rizzardi
To address the bottlenecks of missing decision-making closed loop, insufficient experience reuse, and decoupled resource scheduling in industrial LLM deployment, this paper proposes LLM-Conductor, a three-layer collaborative architecture that enables monitoring-feedback autonomous decision-making, structured policy…