6 papers
NEURON-Fabric: Architecture-Runtime Co-Design for Controlled Low-Bit Gradient Communication
Ziqiang Wang, Changcheng Huang, Chung-Horng Lung
Large-scale neural-network training repeatedly aggregates gradients across devices, making communication a central cost in distributed learning. Low-bit gradient aggregation can re…
NEURON-Fabric: CXL-Side Low-Bit Gradient Aggregation for Distributed Training
Ziqiang Wang, Changcheng Huang, Chung-Horng Lung
In large-model distributed training, especially large language model workloads, gradient All-Reduce increasingly stresses the memory and communication path. This paper asks whether…
FORGE: Self-Evolving Agent Memory With No Weight Updates via Population Broadcast
Igor Bogdanov, Chung-Horng Lung, Thomas Kunz +3
Can LLM agents improve decision-making through self-generated memory without gradient updates? We propose FORGE (Failure-Optimized Reflective Graduation and Evolution), a staged, p…
Context, Reasoning, and Hierarchy: A Cost-Performance Study of Compound LLM Agent Design in an Adversarial POMDP
Igor Bogdanov, Chung-Horng Lung, Thomas Kunz +3
Deploying compound LLM agents in adversarial, partially observable sequential environments requires navigating several design dimensions: (1) what the agent sees, (2) how it reason…
Green Traffic Engineering for Satellite Networks Using Segment Routing Flexible Algorithm
Jintao Liang, Pablo G. Madoery, Chung-Horng Lung +2
Large-scale low-Earth-orbit (LEO) constellations demand routing that simultaneously minimizes energy, guarantees delivery under congestion, and meets latency requirements for time-…
Green Satellite Networks Using Segment Routing and Software-Defined Networking
Jintao Liang, Pablo G. Madoery, Chung-Horng Lung +2
This paper presents a comprehensive evaluation of network performance in software defined networking (SDN)-based low Earth orbit (LEO) satellite networks, focusing on the Telesat L…