arXiv subjects

Zhengchao Wang

Publications and source records attributed to Zhengchao Wang.

4 recordsLinked to original sources

arXiv2026

OLED-MoE: Accelerating MoE-Based dLLM Inference via Inter-Iteration Locality-Aware Expert Offloading

Semi-autoregressive diffusion large language models (dLLMs) improve decoding parallelism through iterative block-wise denoising, but scaling them with mixture-of-experts (MoE) layers introduces a large expert parameter footprint that exceeds memory-constrained GPU capacity. Expert offloading is a natural remedy, yet existing MoE serving systems target autoregressive decoding and rely on intra-iteration layer-wise prefetching: while computing one layer, they predict and load experts for subsequent layers. Under dLLM inference, block-wise routing expands the active expert working set within each iteration, making such prefetches difficult to complete in time and costly when mispredicted. Consequently, existing prefetch-based solutions often degenerate into on-demand expert loading with high decoding latency. We propose OLED-MoE, an expert offloading system that shifts the optimization target from intra-iteration prefetching to inter-iteration expert retention. Its key insight is that adjacent denoising iterations exhibit strong expert routing overlap, and token confidence indicates which experts are likely to be reused. OLED-MoE uses confidence-guided inter-iteration prediction to retain high-value experts in GPU memory without introducing extra prefetch traffic. It further compensates unavoidable cache misses through CPU-GPU cooperative execution, jointly considering dynamic expert computation load and predicted future reuse. Across diverse dLLM workloads, OLED-MoE reduces time per output token (TPOT) by 1.23x-7.93x and improves expert cache utilization by 1.44x-4.23x over state-of-the-art offloading systems. Notably, OLED-MoE approaches full-residency performance while using only 40% of the expert GPU memory, incurring merely 23% higher TPOT despite a 60% reduction in expert memory footprint. OLED-MoE's source code is publicly available at https://github.com/flashserve/OLED-MoE.

cs.DC↗

arXiv2026

Multi-Armed Bernoulli Bandits via Minimax Single-Arm Stopping

We develop an index policy for finite-horizon Bernoulli multi-armed bandits from minimax solutions to single-arm bandit (SAB) problems. Each SAB problem involves choosing between an unknown Bernoulli arm and a known reward. We show that minimizing worst-case regret of SAB problems over all non-anticipative policies admits an exact semi-infinite linear programming formulation. The resulting stopping policies offer a natural way to compare arms: the higher the known reward against which a policy continues sampling, the more promising the unknown arm. We turn this intuition into indices based on cumulative continuation probabilities, with a monotone adjustment and a reward-shortfall cap. By relating index errors to the regret of single-arm stopping policies, we establish a distribution-free regret bound of $4.45\sqrt{KT}+10.75K$ for $K$ arms and horizon $T$. This bound matches the minimax-optimal regret order established in the literature. The guarantee extends to rewards supported on $[0,1]$ through Bernoulli randomization. We also provide a finite-grid implementation with quantified approximation loss. In numerical experiments, the SAB-based index policy achieves lower worst-case regret than every tested benchmark policy across all evaluated numbers of arms and horizons, while closely matching the grid-based MAB minimax policy in the two-arm setting.

cs.LG↗

arXiv2026

RAC: Reference-Aware Activation Compression for Communication-Efficient Split LLM Inference

Large language model (LLM) agents repeatedly process long, privacy-sensitive contexts, while cloud-only deployment exposes user data beyond the trusted endpoint and fully local deployment often requires costly hardware. Split inference offers a middle ground by executing the model head, tail, and tools locally and the middle layers in the cloud, but its local-cloud-local path transfers boundary hidden states at every invocation and creates a critical communication bottleneck. We present \system, a reference-aware codec that retrieves exact-token historical spans for prefill uplinks, reuses the reconstructed uplink state for same-round prefill downlinks, and generates boundary-specific decode references with lightweight causal predictors. RAC applies grouped affine alignment and calibrated residual quantization with optional prefill outliers, while sender-side wire-format reconstruction synchronizes subsequent references and offline calibration accounts for quality and packed representation costs. Across three models and nine evaluated model-link pairs, Raw-to-RAC mean time to first token (TTFT) and time per output token (TPOT) ratios are 1.24-2.72$\times$ and 1.01-2.79$\times$, while the 12 non-perplexity task-score changes range from $-0.40$ to $+2.50$ points.

cs.DC↗

arXiv2025

RAGPulse: An Open-Source RAG Workload Trace to Optimize RAG Serving Systems

Retrieval-Augmented Generation (RAG) is a critical paradigm for building reliable, knowledge-intensive Large Language Model (LLM) applications. However, the multi-stage pipeline (retrieve, generate) and unique workload characteristics (e.g., knowledge dependency) of RAG systems pose significant challenges for serving performance optimization. Existing generic LLM inference traces fail to capture these RAG-specific dynamics, creating a significant performance gap between academic research and real-world deployment. To bridge this gap, this paper introduces RAGPulse, an open-source RAG workload trace dataset. This dataset was collected from an university-wide Q&A system serving that has served more than 40,000 students and faculties since April 2024. We detail RAGPulse's system architecture, its privacy-preserving hash-based data format, and provide an in-depth statistical analysis. Our analysis reveals that real-world RAG workloads exhibit significant temporal locality and a highly skewed hot document access pattern. RAGPulse provides a high-fidelity foundation for researchers to develop and validate novel optimization strategies for RAG systems, such as content-aware batching and retrieval caching, ultimately enhancing the efficiency and reliability of RAG services. The code is available at https://github.com/flashserve/RAGPulse.

cs.LG↗