10 papers
Bandwidth-Aware and Cost-Efficient Pipeline Parallel Scheduling in Geo-Distributed LLM Training
Han Zhang, Jianchun Liu, Hongli Xu
The rapid evolution of large language models (LLMs) has made geographically distributed training necessary due to GPU scarcity within a single cloud region. In such cross-region se…
Adaptive and Fine-grained Module-wise Expert Pruning for Efficient LoRA-MoE Fine-Tuning
Weihang Li, Jianchun Liu, Hongli Xu
LoRA-MoE has emerged as an effective paradigm for parameter-efficient fine-tuning, combining the low training cost of LoRA with the increased adaptation capacity of Mixture-of-Expe…
Improving LLM Reasoning via Dependency-Aware Query Decomposition and Logic-Parallel Content Expansion
Xianjun Gao, Jianchun Liu, Hongli Xu +1
The integration of Large Language Models (LLMs) into real-time Web applications, such as AI-powered search and conversational agents, presents a fundamental Web infrastructure chal…
SABlock: Semantic-Aware KV Cache Eviction with Adaptive Compression Block Size
Jinhan Chen, Jianchun Liu, Hongli Xu +2
The growing memory footprint of the Key-Value (KV) cache poses a severe scalability bottleneck for long-context Large Language Model (LLM) inference. While KV cache eviction has em…
Towards Communication-Efficient Decentralized Federated Graph Learning over Non-IID Data
Shilong Wang, Jianchun Liu, Hongli Xu +3
Decentralized Federated Graph Learning (DFGL) overcomes potential bottlenecks of the parameter server in FGL by establishing a peer-to-peer (P2P) communication network among worker…
Accelerating Mixture-of-Expert Inference with Adaptive Expert Split Mechanism
Jiaming Yan, Jianchun Liu, Hongli Xu +1
Mixture-of-Experts (MoE) has emerged as a promising architecture for modern large language models (LLMs). However, massive parameters impose heavy GPU memory (i.e., VRAM) demands,…