Searcharxiv⌕ Search

arXiv subjects

Haojie Wang

Publications and source records attributed to Haojie Wang.

At least 19 recordsLinked to original sources
arXiv2026

Decoupling Readiness from Release for Tail-Aware Scheduling of Agentic LLM Workflows

Agentic LLM workflows consist of sequences of model turns interleaved with tool interactions, so their end-to-end completion time depends not only on inference speed but also on when ready turns are released. Most runtimes release each turn immediately upon readiness. Under contention, this eager release policy can accumulate released but unfinished work; once submitted, those turns can no longer be reordered by the workflow-level policy, increasing tail latency. We present a tail-risk-aware turn release scheduling method that jointly decides which ready turn to release next and how much released but unfinished work to maintain. The method uses a mean--Conditional Value-at-Risk (CVaR) objective to capture the evolving tail risk of unfinished workflows, incorporates online estimates of turn work when prioritizing ready turns, and adapts the released work budget to observed queue pressure. We evaluate the method using real agent execution traces from software engineering tasks across multiple LLMs and workflow arrival rates. The method performs comparably to eager release under light load and substantially reduces the P95 of workflow flow time under contention, achieving up to a \(3.50\times\) speedup.

cs.AI↗
arXiv2026

Epoch: Compiling Diffusion Blocks for Sparse MoE Serving

Diffusion language models generate text by refining a fixed-size block of token positions through many forward passes, a loop that does not match the per-forward execution unit used by most LLM serving systems. A dense MoE runtime binds all work to the refinement-iteration clock: it rebuilds similar routing structure on every forward, recomputes expert outputs for positions whose logits are already dead, and sends those positions through dense expert-parallel collectives. This paper presents \sys{}, a serving system that treats the diffusion block as a compilation unit. \sys{} compiles a small \emph{block plan} for the block-clock structure of one diffusion block and refreshes every value that can affect a live decode decision on the iteration clock. \sys{} realizes this plan along three dense axes of an MoE forward: \atlas{} compiles a coverage-driven active expert support per layer while recomputing gate logits every iteration; \lsp{} keeps full sequence shards as model state but routes only live, newly decoded, and refresh-required positions through fresh routed-expert computation; \freshlane{} carries this fresh token--expert worklist through expert-parallel dispatch, kernels, and combine, then restores the dense logical shard at the layer boundary. We implement \sys{} on 8 NVIDIA H100 GPUs and evaluate it on three open-weight block-diffusion MoE models (LLaDA-MoE, LLaDA2.0-mini, and LLaDA2.0-Flash, spanning 7B to 100B total parameters) across GSM8K, HumanEval, MGSM, and MT-Bench. \sys{} improves end-to-end execution time by up to 2.7$\times$ over the strongest surviving baseline under the same 8-GPU placement and remains feasible at the largest batch sizes where multiple baselines run out of memory, while preserving task quality relative to the dense reference.

cs.DC↗
arXiv2026

Entanglement certification via causal-order interferometry in a quantum switch

Entanglement certification is often performed on states that have already undergone noisy transmission or processing. Noise can reduce the surviving entanglement and can also cause a given criterion to fail even when entanglement remains. In this context, the quantum switch, a paradigmatic realization of indefinite causal order (ICO), coherently controls the orders in which two channels act and has been shown to offer advantages across a range of quantum information-processing tasks. Here we ask whether this coherent control enlarges the noise-parameter region in which entanglement remains certifiable. We regard the two order branches as the arms of a causal-order interferometer and insert a local unitary between the channel uses to tune their interference. For stochastic Pauli noise, a postselected ICO output can exhibit greater entanglement negativity than any classical mixture of the two definite orders; in particular, we identify regimes where its negativity remains nonzero while that of every classical mixture vanishes. A suitable local Pauli unitary substantially enlarges this ICO-only region, while an input-dependent path-difference indicator qualitatively links operator noncommutativity to the postselected negativity gain. Numerical examples extend the advantage to local amplitude-damping noise and two-qutrit Weyl noise. At a representative Weyl-noise point for the $3\times 3$ positive-partial-transpose (PPT) Tiles bound-entangled state, a nondecomposable witness detects the postselected ICO output, whereas an analytic bound excludes detection of the definite-order outputs and their mixtures by the entire locally rotated witness family. These results identify causal-order interferometry as a strategy for enhancing entanglement certification across distinct noise models and dimensions.

quant-ph↗
arXiv2026

LayerScope: Predictive Cross-Layer Scheduling for Efficient Multi-Batch MoE Inference on Legacy Servers

Mixture-of-Experts (MoE) models face memory and PCIe latency bottlenecks when deployed on commodity hardware. Offloading expert weights to CPU memory results in PCIe transfer latency that exceeds GPU computation by several folds. We present PreScope, a prediction-driven expert scheduling system that addresses three key challenges: inaccurate activation prediction, PCIe bandwidth competition, and cross-device scheduling complexity. Our solution includes: 1) Learnable Layer-Aware Predictor (LLaPor) that captures layer-specific expert activation patterns; 2) Prefetch-Aware Cross-Layer Scheduling (PreSched) that generates globally optimal plans balancing prefetching costs and loading overhead; 3) Asynchronous I/O Optimizer (AsyncIO) that decouples I/O from computation, eliminating waiting bubbles. PreScope achieves 141% higher throughput and 74.6% lower latency than state-of-the-art solutions.

cs.LG↗
arXiv2026

PLANET v2.0: A comprehensive Protein-Ligand Affinity Prediction Model Based on Mixture Density Network

Drug discovery represents a time-consuming and financially intensive process, and virtual screening can accelerate it. Scoring functions, as one of the tools guiding virtual screening, have their precision closely tied to screening efficiency. In our previous study, we developed a graph neural network model called PLANET (Protein-Ligand Affinity prediction NETwork), but it suffers from the defect in representing protein-ligand contact maps. Incorrect binding modes inevitably lead to poor affinity predictions, so accurate prediction of the protein-ligand contact map is desired to improve PLANET. In this study, we have proposed PLANET v2.0 as an upgraded version. The model is trained via multi-objective training strategy and incorporates the Mixture Density Network to predict binding modes. Except for the probability density distributions of non-covalent interactions, we innovatively employ another Gaussian mixture model to describe the relationship between distance and energy of each interaction pair and predict protein-ligand affinity like calculating the mathematical expectation. As on the CASF-2016 benchmark, PLANET v2.0 demonstrates excellent scoring power, ranking power, and docking power. The screening power of PLANET v2.0 gets notably improved compared to PLANET and Glide SP and it demonstrates robust validation on a commercial ultra-large-scale dataset. Given its efficiency and accuracy, PLANET v2.0 can hopefully become one of the practical tools for virtual screening workflows. PLANET v2.0 is freely available at https://www.pdbbind-plus.org.cn/planetv2.

cs.LG↗
arXiv2026

Non-commutativity as a Universal Characterization for Enhanced Quantum Metrology

A central challenge in quantum metrology is to effectively harness quantum resources to surpass classical precision bounds. Although recent studies suggest that the indefinite causal order may enable sensitivities to attain the super-Heisenberg scaling, the physical origins of such enhancements remain elusive. Here, we introduce the nilpotency index $\mathcal{K}$, which quantifies the depth of non-commutativity between operators during the encoding process, can act as a fundamental parameter governing quantum-enhanced sensing. We show that a finite $\mathcal{K}$ yields an enhanced scaling of root-mean-square error as $N^{-(1+\mathcal{K})}$. Meanwhile, the requirement for indefinite causal order arises only when the nested commutators become constant. Remarkably, in the limit $\mathcal{K} \to \infty$, exponential precision scaling $N^{-1}e^{-N}$ is achievable. We propose experimentally feasible protocols implementing these mechanisms, providing a systematic pathway towards practical quantum-enhanced metrology.

quant-ph↗
arXiv2025

RailS: Load Balancing for All-to-All Communication in Distributed Mixture-of-Experts Training

Training Mixture-of-Experts (MoE) models introduces sparse and highly imbalanced all-to-all communication that dominates iteration time. Conventional load-balancing methods fail to exploit the deterministic topology of Rail architectures, leaving multi-NIC bandwidth underutilized. We present RailS, a distributed load-balancing framework that minimizes all-to-all completion time in MoE training. RailS leverages the Rail topology's symmetry to prove that uniform sending ensures uniform receiving, transforming global coordination into local scheduling. Each node independently executes a Longest Processing Time First (LPT) spraying scheduler to proactively balance traffic using local information. RailS activates N parallel rails for fine-grained, topology-aware multipath transmission. Across synthetic and real-world MoE workloads, RailS improves bus bandwidth by 20%--78% and reduces completion time by 17%--78%. For Mixtral workloads, it shortens iteration time by 18%--40% and achieves near-optimal load balance, fully exploiting architectural parallelism in distributed training.

cs.DC↗
arXiv2025

Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning

The rapid growth of large language model (LLM) services imposes increasing demands on distributed GPU inference infrastructure. Most existing scheduling systems follow a reactive paradigm, relying solely on the current system state to make decisions, without considering how task demand and resource availability evolve over time. This lack of temporal awareness in reactive approaches leads to inefficient GPU utilization, high task migration overhead, and poor system responsiveness under dynamic workloads. In this work, we identify the fundamental limitations of these instantaneous-state-only scheduling approaches and propose Temporal Optimal Resource scheduling via Two-layer Architecture (TORTA). TORTA introduces a spatiotemporal scheduling framework that captures both long-term workload patterns and short-term execution constraints. It adopts a two-layer design: a macro-level scheduler leverages reinforcement learning and optimal transport to coordinate inter-region task distribution, while a micro-level allocator refines task-to-server assignments within each region to reduce latency and switching costs. Experimental results across multiple network topologies show that TORTA reduces average inference response time by up to 15\%, improves load balance by approximately 4-5\%, and cuts total operational cost by 10-20\% compared to state-of-the-art baseline methods.

cs.DC↗
arXiv2025

Multi-Static Target Position Estimation and System Optimization for Cell-Free mMIMO-OTFS ISAC

This paper investigates multi-static position estimation in cell-free massive multiple-input multiple-output (CF mMIMO) architectures, where orthogonal time frequency space (OTFS) is used as an integrated sensing and communication (ISAC) signal. A maximum likelihood position estimation scheme is proposed, where the required search space is reduced by employing a common reference system. Closed-form expressions for the Cramér-Rao lower bound and the position error bound (PEB) in multi-static position estimation are derived, providing quantitative evaluations of sensing performance. These theoretical bounds are further generalized into a universal structure to support other ISAC signals. To enhance overall system performance and adapt to dynamic network requirements, a joint AP operation mode selection and power allocation algorithm is developed to maximize the minimum user communication spectral efficiency (SE) while ensuring a specified sensing PEB requirement. Moreover, a decomposition method is introduced to achieve a better tradeoff between complexity and ISAC performance. The results verify the effectiveness of the proposed algorithms, demonstrating the superiority of the OTFS signal through a nearly twofold SE gain over the orthogonal frequency division multiplexing (OFDM) signal. These findings highlight promising advantages of the CF-ISAC systems from a novel parameter estimation perspective, particularly in high-mobility vehicle-to-everything applications.

cs.IT↗
arXiv2025

NineToothed: A Triton-Based High-Level Domain-Specific Language for Machine Learning

The emergence of deep learning domain-specific languages (DSLs) has substantially reduced the obstacles in developing high-performance, cross-platform compute kernels. However, current DSLs, such as Triton, still demand that developers possess expertise in parallel programming and expose them to many low-level details. This requirement complicates the development process and adds to the difficulty of maintaining compute kernels. Consequently, developing a new programming model that supports serial programming for deep learning workloads is crucial. This paper introduces NineToothed, a domain-specific language that offers serial semantics for machine learning programming. Through the automatic transformation of serial code into parallel code, NineToothed significantly streamlines the development process while causing minimal performance degradation. NineToothed encompasses (1) a language with tensor-oriented metaprogramming (TOM) that adopts the arrange-and-apply paradigm, enabling the expression of tiled computations without the need to manage low-level details and (2) a code generator for generating high-performance parallel code. Our evaluation results indicate that NineToothed can greatly simplify compute kernel development while maintaining performance comparable to that of Triton.

cs.DC↗
arXiv2025

A Homogeneous Graph Neural Network for Precoding and Power Allocation in Scalable Wireless Networks

Deep learning is widely used in wireless communications but struggles with fixed neural network sizes, which limit their adaptability in environments where the number of users and antennas varies. To overcome this, this paper introduced a generalization strategy for precoding and power allocation in scalable wireless networks. Initially, we employ an innovative approach to abstract the wireless network into a homogeneous graph. This primarily focuses on bypassing the heterogeneous features between transmitter (TX) and user entities to construct a virtual homogeneous graph serving optimization objectives, thereby enabling all nodes in the virtual graph to share the same neural network. This ``TX entity'' is known as a base station (BS) in cellular networks and an access point (AP) in cell-free networks. Subsequently, we design a universal graph neural network, termed the information carrying graph neural network (ICGNN), to capture and integrate information from this graph, maintaining permutation invariance. Lastly, using ICGNN as the core algorithm, we tailor the neural network's input and output for specific problem requirements and validate its performance in two scenarios: 1) in cellular networks, we develop a matrix-inverse-free multi-user multi-input multi-output (MU-MIMO) precoding scheme using the conjugate gradient (CG) method, adaptable to varying user and antenna numbers; 2) in a cell-free network, facing dynamic variations in the number of users served by APs, the number of APs serving each user, and the number of antennas per AP, we propose a universal power allocation scheme. Simulations demonstrate that the proposed approach not only significantly reduces computational complexity but also achieves, and potentially exceeds, the spectral efficiency (SE) of conventional algorithms.

eess.SP↗
arXiv2025

LLMs for Supply Chain Management

The development of large language models (LLMs) has provided new tools for research in supply chain management (SCM). In this paper, we introduce a retrieval-augmented generation (RAG) framework that dynamically integrates external knowledge into the inference process, and develop a domain-specialized SCM LLM, which demonstrates expert-level competence by passing standardized SCM examinations and beer game tests. We further employ the use of LLMs to conduct horizontal and vertical supply chain games, in order to analyze competition and cooperation within supply chains. Our experiments show that RAG significantly improves performance on SCM tasks. Moreover, game-theoretic analysis reveals that the LLM can reproduce insights from the classical SCM literature, while also uncovering novel behaviors and offering fresh perspectives on phenomena such as the bullwhip effect. This paper opens the door for exploring cooperation and competition for complex supply chain network through the lens of LLMs.

cs.AI↗
arXiv2025

SpecRouter: Adaptive Routing for Multi-Level Speculative Decoding in Large Language Models

Large Language Models (LLMs) present a critical trade-off between inference quality and computational cost: larger models offer superior capabilities but incur significant latency, while smaller models are faster but less powerful. Existing serving strategies often employ fixed model scales or static two-stage speculative decoding, failing to dynamically adapt to the varying complexities of user requests or fluctuations in system performance. This paper introduces \systemname{}, a novel framework that reimagines LLM inference as an adaptive routing problem solved through multi-level speculative decoding. \systemname{} dynamically constructs and optimizes inference "paths" (chains of models) based on real-time feedback, addressing the limitations of static approaches. Our contributions are threefold: (1) An \textbf{adaptive model chain scheduling} mechanism that leverages performance profiling (execution times) and predictive similarity metrics (derived from token distribution divergence) to continuously select the optimal sequence of draft and verifier models, minimizing predicted latency per generated token. (2) A \textbf{multi-level collaborative verification} framework where intermediate models within the selected chain can validate speculative tokens, reducing the verification burden on the final, most powerful target model. (3) A \textbf{synchronized state management} system providing efficient, consistent KV cache handling across heterogeneous models in the chain, including precise, low-overhead rollbacks tailored for asynchronous batch processing inherent in multi-level speculation. Preliminary experiments demonstrate the validity of our method.

cs.LG↗
arXiv2025

Multi-Target Position Error Bound and Power Allocation Scheme for Cell-Free mMIMO-OTFS ISAC Systems

This paper investigates multi-target position estimation in cell-free massive multiple-input multiple-output (CF mMIMO) architectures, where orthogonal time frequency and space (OTFS) is used as an integrated sensing and communication (ISAC) signal. Closed-form expressions for the Cramér-Rao lower bound and the positioning error bound (PEB) in multi-target position estimation are derived, providing quantitative evaluations of sensing performance. To enhance the overall performance of the ISAC system, a power allocation algorithm is developed to maximize the minimum user communication signal-to-interference-plus-noise ratio while ensuring a specified sensing PEB requirement. The results validate the proposed PEB expression and its approximation, clearly illustrating the coordination gain enabled by ISAC. Further, the superiority of using the multi-static CF mMIMO architecture over traditional cellular ISAC is demonstrated, and the advantages of OTFS signals in high-mobility scenarios are highlighted.

cs.IT↗
arXiv2025

FastCache: Optimizing Multimodal LLM Serving through Lightweight KV-Cache Compression Framework

Multi-modal Large Language Models (MLLMs) serving systems commonly employ KV-cache compression to reduce memory footprint. However, existing compression methods introduce significant processing overhead and queuing delays, particularly in concurrent serving scenarios. We present \texttt{FastCache}, a novel serving framework that effectively addresses these challenges through two key innovations: (1) a dynamic batching strategy that optimizes request scheduling across prefill, compression, and decode stages, and (2) an efficient KV-cache memory pool mechanism that eliminates memory fragmentation while maintaining high GPU utilization. Our comprehensive experiments on the GQA and MileBench datasets demonstrate that \texttt{FastCache} achieves up to 19.3$\times$ reduction in Time-To-First-Token (TTFT) and 12.1$\times$ improvement in throughput compared to state-of-the-art baselines. The system maintains stable performance under high-concurrency scenarios (up to 40 req/s) while reducing average memory consumption by 20\%. These results establish \texttt{FastCache} as an efficient solution for real-world LLM serving systems with KV-cache compression.

cs.MM↗
arXiv2025

ARIC: An Activity Recognition Dataset in Classroom Surveillance Images

The application of activity recognition in the ``AI + Education" field is gaining increasing attention. However, current work mainly focuses on the recognition of activities in manually captured videos and a limited number of activity types, with little attention given to recognizing activities in surveillance images from real classrooms. Activity recognition in classroom surveillance images faces multiple challenges, such as class imbalance and high activity similarity. To address this gap, we constructed a novel multimodal dataset focused on classroom surveillance image activity recognition called ARIC (Activity Recognition In Classroom). The ARIC dataset has advantages of multiple perspectives, 32 activity categories, three modalities, and real-world classroom scenarios. In addition to the general activity recognition tasks, we also provide settings for continual learning and few-shot continual learning. We hope that the ARIC dataset can act as a facilitator for future analysis and research for open teaching scenarios. You can download preliminary data from https://ivipclab.github.io/publication_ARIC/ARIC.

cs.CV↗
arXiv2025

TritonBench: Benchmarking Large Language Model Capabilities for Generating Triton Operators

Triton, a high-level Python-like language designed for building efficient GPU kernels, is widely adopted in deep learning frameworks due to its portability, flexibility, and accessibility. However, programming and parallel optimization still require considerable trial and error from Triton developers. Despite advances in large language models (LLMs) for conventional code generation, these models struggle to generate accurate, performance-optimized Triton code, as they lack awareness of its specifications and the complexities of GPU programming. More critically, there is an urgent need for systematic evaluations tailored to Triton. In this work, we introduce TritonBench, the first comprehensive benchmark for Triton operator generation. TritonBench features two evaluation channels: a curated set of 184 real-world operators from GitHub and a collection of operators aligned with PyTorch interfaces. Unlike conventional code benchmarks prioritizing functional correctness, TritonBench also profiles efficiency performance on widely deployed GPUs aligned with industry applications. Our study reveals that current state-of-the-art code LLMs struggle to generate efficient Triton operators, highlighting a significant gap in high-performance code generation. TritonBench will be available at https://github.com/thunlp/TritonBench.

cs.CL↗