Searcharxiv⌕ Search

arXiv subjects

Xiaowen Chu

Publications and source records attributed to Xiaowen Chu.

At least 37 records · Page 2Linked to original sources
arXiv2026

KernelFlume: Elastic Core-Attention Scaling for Agentic Long-Context Decoding

LLM serving is increasingly dominated by long and dynamic decode workloads from agents, reasoning models, and extended conversations. When bursty long-context demand exceeds deployed capacity, existing serving systems typically scale out by launching additional serving instances with model replicas. This instance-level elasticity increases KV capacity only by provisioning another full copy of the model, inheriting startup latency, memory overhead, and batch fragmentation. We present KernelFlume, a decode-centric architecture that disaggregates the stable projection/FFN path from core-attention computation: weight nodes execute dense projection/FFN kernels, while weightless attention nodes store token-range KV partitions and scale with request-state demand. To make this separation elastic, KernelFlume maintains a routing table that maps token ranges to attention-node endpoints. It updates routes at token boundaries and uses host-visible graph signals to drive pre-registered UCX endpoint communication outside the captured CUDA Graph. To preserve low per-token latency after disaggregation, KernelFlume combines query-first core-attention dispatch with inter-layer kernel pipelining, overlapping remote attention and communication with local projection/FFN work. On real GPU testbeds (intra-node A6000 and cross-node H100), under a dynamic long-context agentic workload serving Llama-3.1-8B, KernelFlume sustains flat p99 TPOTs of ~74 ms on A6000 and ~34 ms on H100, while lowering cost per million output tokens by up to 32% and 61%, respectively, relative to full-instance elastic scaling with ServerlessLLM, a state-of-the-art instance-startup method. Replaying the same trace at larger model scale in simulation projects a 56--66% cost reduction over ServerlessLLM, widening to 80--85% with cheaper heterogeneous attention-node hardware and persisting into the million-token context range.

cs.DC↗
arXiv2026

CoT-Core: Accelerating LLM Evaluation via CoT-Aware Coreset Selection

Evaluating Large Language Models (LLMs) incurs prohibitive computational overhead during continuous development processes. While coreset selection accelerates evaluation, existing methods either suffer from a severe ``cold start'' bottleneck requiring massive historical logs (e.g., Item Response Theory) or exhibit a surface lexical bias that misses the underlying reasoning manifold of tasks. We propose CoT-Core, a novel training-free core question selection framework. Recognizing that lexically disparate questions can share equivalent underlying logic, CoT-Core prompts LLMs to unroll zero-shot Chain-of-Thought (CoT) reasoning trajectories. Projecting these paths into a latent space effectively clusters questions by intrinsic logical equivalence rather than superficial text similarity. Extensive experiments on GSM8K, MMLU, MMLU-Pro, and GPQA demonstrate that CoT-Core drastically reduces evaluation costs while maintaining high-fidelity score estimation, and delineate the boundary conditions of reasoning-aware pruning, revealing that its efficacy is intrinsically gated by task complexity.

cs.AI↗
arXiv2026

Architecture-Aware Reinforcement Learning Makes Sliding-Window Attention Competitive in Math Reasoning

The rapid progress of reasoning and agentic large language models (LLMs) has increased the demand for long-context inference, but self-attention (SA) scales quadratically with context length. To address this, we study SWARR (Sliding-Window Attention with Reinforced Adaptation for Math Reasoning), a practical recipe for adapting SWA models to mathematical reasoning. SWARR has two stages: (1) efficient conversion from a pretrained SA model to SWA with supervised fine-tuning (SFT), which avoids pretraining a new base model, and (2) policy adaptation with reinforcement learning (RL). We find that SWA still underperforms SA after SFT, and we hypothesize that this gap is caused in part by a data-architecture mismatch: most SFT data are prepared for SA models and may contain long-range dependencies that are difficult for SWA to model. Because on-policy RL optimizes self-generated trajectories under the SWA constraint, it can adapt trajectories to better match SWA. Experiments on mathematical reasoning benchmarks show that this recipe substantially narrows the gap between SWA and SA, recovering much of the accuracy lost during SWA conversion while preserving the efficiency benefits of linear-complexity attention. Our central contribution is the empirical finding that RL changes the conclusion one would draw from conversion and SFT alone about SWA's viability for math reasoning.

cs.AI↗
arXiv2026

SDG-Track: A Heterogeneous Observer-Follower Framework for High-Resolution UAV Tracking on Embedded Platforms

Real-time tracking of small unmanned aerial vehicles (UAVs) on edge devices faces a fundamental resolution-speed conflict. Downsampling high-resolution imagery to standard detector input sizes causes small target features to collapse below detectable thresholds. Yet processing native 1080p frames on resource-constrained platforms yields insufficient throughput for smooth gimbal control. We propose SDG-Track, a Sparse Detection-Guided Tracker that adopts an Observer-Follower architecture to reconcile this conflict. The Observer stream runs a high-capacity detector at low frequency on the GPU to provide accurate position anchors from 1920x1080 frames. The Follower stream performs high-frequency trajectory interpolation via ROI-constrained sparse optical flow on the CPU. To handle tracking failures from occlusion or model drift caused by spectrally similar distractors, we introduce Dual-Space Recovery, a training-free re-acquisition mechanism combining color histogram matching with geometric consistency constraints. Experiments on a ground-to-air tracking station demonstrate that SDG-Track achieves 35.1 FPS system throughput while retaining 97.2\% of the frame-by-frame detection precision. The system successfully tracks agile FPV drones under real-world operational conditions on an NVIDIA Jetson Orin Nano. Our paper code is publicly available at https://github.com/Jeffry-wen/SDG-Track

cs.CV↗
arXiv2026

Semantic Integrity Matters: Benchmarking and Preserving High-Density Reasoning in KV Cache Compression

While Key-Value (KV) cache compression is essential for efficient LLM inference, current evaluations disproportionately focus on sparse retrieval tasks, potentially masking the degradation of High-Density Reasoning where Chain-of-Thought (CoT) coherence is critical. We introduce KVFundaBench to systematically evaluate this gap, revealing a sharp dichotomy: while retrieval tasks remain robust, reasoning tasks exhibit severe Task-Dependent Degradation under aggressive compression due to disrupted CoT links. Extending our analysis to the DeepSeek-R1 model, we uncover that its specialized attention patterns offer unique insights into the fragility of reasoning chains. Guided by these findings -- specifically the necessity of preserving few-shot examples as indivisible Semantic Units -- we propose ShotKV. This approach explicitly separates prefill and decoding phases to prioritize semantic integrity. Empirical results demonstrate that ShotKV achieves 9%-18% accuracy improvements on long-context generation tasks and effectively generalizes to document QA, all while delivering an 11% latency reduction compared to full cache inference.

cs.CL↗
arXiv2026

Position: LLM Inference Should Be Evaluated as Energy-to-Token Production

LLM inference is still evaluated mainly as a model or software problem: accuracy, latency, throughput, and hardware utilization. This is incomplete. At deployment scale, the relevant output is a quality-conditioned token produced under joint constraints from effective compute, delivered data-center power, cooling capacity, PUE, and utilization. We argue that the ML community should treat inference as \emph{energy-to-token production}. We formalize this view with a dimensionally consistent Token Production Function in which token rate is bounded by both compute-per-token and energy-per-token ceilings. Listed API prices vary by over an order of magnitude across providers, but we use price dispersion only as directional motivation, not as causal evidence of marginal cost. The core physical question is instead: under fixed quality and service targets, when does the binding constraint move from theoretical peak compute toward delivered power, cooling, and operational efficiency? Under this framing, system optimizations -- latent KV-cache compression, sparse or heavily compressed attention, quantization, routing, and difficulty-adaptive reasoning -- are not merely local engineering tricks. They are energy-to-token levers because they reduce FLOPs/token, joules/token, memory traffic, or utilization losses under fixed $(q^{*},s^{*})$. We therefore call for inference papers and benchmarks to report Joules/token, active binding constraint, PUE-adjusted delivered power, and utilization-adjusted token output alongside accuracy and latency.

cs.CE↗
arXiv2026

LoRA-FA: Efficient and Effective Low Rank Representation Fine-tuning

Fine-tuning large language models (LLMs) is crucial for improving their performance on downstream tasks, but full-parameter fine-tuning (Full-FT) is computationally expensive and memory-intensive. Parameter-efficient fine-tuning (PEFT) methods, such as Low-Rank Adaptation (LoRA), address this by optimizing only a small subset of parameters. However, LoRA may underperform Full-FT in certain scenarios due to the intrinsic limitations of its low-rank gradients. In this work, we reveal an asymmetric, collapsible structure in LoRA's update: the low-rank modification to W can be reformulated as a single-layer linear regression, implying that one of the LoRA factors can be frozen without sacrificing expressivity. Leveraging this insight, we introduce LoRA-FA, which freezes the projection-down matrix A and trains only the projection-up matrix B. We further close the gap to Full-FT by deriving closed-form gradient corrections that minimize the discrepancy between the induced low-rank gradient and the full gradient. Through extensive experiments on diverse benchmarks, including GLUE, GSM8K, MT-Bench, and HumanEval, we demonstrate that LoRA-FA consistently achieves comparable performance to existing PEFT methods and Full-FT. Experiments on system efficiency show that LoRA-FA significantly reduces activation memory consumption and computational workload in fine-tuning. Our code is available at https://github.com/huggingface/peft.

cs.CL↗
arXiv2026

DiffAdapt: Difficulty-Adaptive Reasoning for Token-Efficient LLM Inference

Recent reasoning Large Language Models (LLMs) demonstrate remarkable problem-solving abilities but often generate long thinking traces whose utility is unclear. Our work aims to improve their efficiency, enabling them to reach high performance without overthinking. First, we analyze the entropy of token probabilities in reasoning traces. Across three models, we observe a consistent U-shaped entropy pattern: high entropy on easy problems despite high accuracy, low entropy on problems with medium difficulty, and high entropy on hard problems reflecting uncertainty. Specifically, we notice 22--25\% entropy reduction from easy to medium difficulty regions, suggesting an {overthinking} phenomenon on easy instances. Building on these insights, we introduce \textbf{DiffAdapt}, a lightweight framework that selects Easy/Normal/Hard inference strategies per question based on their difficulty and reasoning trace entropy. Each inference strategy consists of a fixed prompt, temperature and maximum token length. In contrast to existing efficiency optimization methods, our approach does not fine-tune base LLM but a small probe that classifies LLM's final hidden state, allowing inexpensive adaptation. We comprehensively evaluate our method on five models and eight benchmarks. Our method achieves comparable or improved accuracy while reducing token usage by up to 22.4\%, establishing a practical path toward compute-efficient reasoning.

cs.CL↗
arXiv2026

MDN: Parallelizing Stepwise Momentum for Delta Linear Attention

Linear Attention (LA) offers a promising paradigm for scaling large language models (LLMs) to long sequences by avoiding the quadratic complexity of self-attention. Recent LA models such as Mamba2 and GDN interpret linear recurrences as closed-form online stochastic gradient descent (SGD), but naive SGD updates suffer from rapid information decay and suboptimal convergence in optimization. While momentum-based optimizers provide a natural remedy, they pose challenges in simultaneously achieving training efficiency and effectiveness. To address this, we develop a chunkwise parallel algorithm for LA with a stepwise momentum rule by geometrically reordering the update coefficients. Further, from a dynamical systems perspective, we analyze the momentum-based recurrence as a second-order system that introduces complex conjugate eigenvalues. This analysis guides the design of stable gating constraints. The resulting model, Momentum DeltaNet (MDN), leverages Triton kernels to achieve comparable training throughput with competitive linear models such as Mamba2 and KDA. Extensive experiments on the 400M and 1.3B parameter models demonstrate consistent performance improvements over strong baselines, including Transformers, Mamba2 and GDN, across diverse downstream evaluation benchmarks. Code: https://github.com/HuuYuLong/MomentumDeltaNet .

cs.LG↗
arXiv2026

SpatialGrammar: A Domain-Specific Language for LLM-Based 3D Indoor Scene Generation

Automatically generating interactive 3D indoor scenes from natural language is crucial for virtual reality, gaming, and embodied AI. However, existing LLM-based approaches often suffer from spatial errors and collisions, in part because common scene representations-raw coordinates or verbose code-are difficult for models to reason about 3D spatial relationships and physical constraints. We propose SpatialGrammar, a domain-specific language that represents gravity-aligned indoor layouts as BEV grid placements with deterministic compilation to valid 3D geometry, enabling verifiable constraint checking. Building on this representation, we develop (1) SG-Agent, a closed-loop system that uses compiler feedback to iteratively refine scenes and enforce collision constraints, and (2) SG-Mini, a 104M-parameter model trained entirely on compiler-validated synthetic data. Across 159 test scenes spanning five scenarios of different complexity, SG-Agent improves spatial fidelity and physical plausibility over prior methods, while SG-Mini performs competitively against larger LLM-based baselines on single-shot generation scenarios.

cs.AI↗
arXiv2026

ZipCCL: Efficient Lossless Data Compression of Communication Collectives for Accelerating LLM Training

Communication has emerged as a critical bottleneck in the distributed training of large language models (LLMs). While numerous approaches have been proposed to reduce communication overhead, the potential of lossless compression has remained largely underexplored since compression and decompression typically consume larger overheads than the benefits of reduced communication traffic. We observe that the communication data, including activations, gradients and parameters, during training often follows a near-Gaussian distribution, which is a key feature for data compression. Thus, we introduce ZipCCL, a lossless compressed communication library of collectives for LLM training. ZipCCL is equipped with our novel techniques: (1) theoretically grounded exponent coding that exploits the Gaussian distribution of LLM tensors to accelerate compression without expensive online statistics, (2) GPU-optimized compression and decompression kernels that carefully design memory access patterns and pipeline using communication-aware data layout, and (3) adaptive communication strategies that dynamically switch collective operations based on workload patterns and system characteristics. Evaluated on a 64-GPU cluster using both mixture-of-experts and dense transformer models, ZipCCL reduces communication time by up to 1.35$\times$ and achieves end-to-end training speedups of up to 1.18$\times$ without any impact on model quality.

cs.DC↗
arXiv2026

ExpertFlow: Efficient Mixture-of-Experts Inference via Predictive Expert Caching and Token Scheduling

Sparse Mixture-of-Experts (MoE) models can outperform dense large language models at similar computation by activating only a small set of experts per token. However, stacking many expert modules introduces substantial parameter memory, which makes MoE models difficult to deploy in memory-constrained environments such as single-GPU devices. Offloading alleviates this issue by storing inactive experts in CPU memory and loading them on demand, but existing methods remain limited: static caches disregard input-dependent routing, and methods that train separate models to predict expert usage ahead of time are often inaccurate or require significant training cost. We propose ExpertFlow, a lightweight MoE inference system that addresses this routing dependency through three coordinated components: 1) a transformer-based routing path predictor that estimates expert usage across all MoE layers in a single forward pass, 2) a token scheduler that groups tokens with similar predicted routes to improve expert utilization, and 3) a predictive expert cache that loads only the required experts while correcting mispredictions at runtime. Together, these components enable efficient expert loading and execution, reducing GPU memory usage by up to 93.72% and improving inference throughput by up to 10x over strong offloading baselines on a single GPU.

cs.AI↗
arXiv2026

Hybrid Quantum-Classical Spatiotemporal Forecasting for 3D Cloud Fields

Accurate forecasting of three-dimensional (3D) cloud fields is important for atmospheric analysis and short-range numerical weather prediction, yet it remains challenging because cloud evolution involves cross-layer interactions, nonlocal dependencies, and multiscale spatiotemporal dynamics. Existing spatiotemporal prediction models based on convolutions, recurrence, or attention often rely on locality-biased representations and therefore struggle to preserve fine cloud structures in volumetric forecasting tasks. To address this issue, we propose QENO, a hybrid quantum-inspired spatiotemporal forecasting framework for 3D cloud fields. The proposed architecture consists of four components: a classical spatiotemporal encoder for compact latent representation, a topology-aware quantum enhancement block for modeling nonlocal couplings in latent space, a dynamic fusion temporal unit for integrating measurement-derived quantum features with recurrent memory, and a decoder for reconstructing future cloud volumes. Experiments on CMA-MESO 3D cloud fields show that QENO consistently outperforms representative baselines, including ConvLSTM, PredRNN++, Earthformer, TAU, and SimVP variants, in terms of MSE, MAE, RMSE, SSIM, and threshold-based detection metrics. In particular, QENO achieves an MSE of 0.2038, an RMSE of 0.4514, and an SSIM of 0.6291, while also maintaining a compact parameter budget. These results indicate that topology-aware hybrid quantum-classical feature modeling is a promising direction for 3D cloud structure forecasting and atmospheric Earth observation data analysis.

cs.LG↗
arXiv2026

Is Your LLM-as-a-Recommender Agent Trustable? LLMs' Recommendation is Easily Hacked by Biases (Preferences)

Current Large Language Models (LLMs) are gradually exploited in practically valuable agentic workflows such as Deep Research, E-commerce recommendation, and job recruitment. In these applications, LLMs need to select some optimal solutions from massive candidates, which we term as \textit{LLM-as-a-Recommender} paradigm. However, the reliability of using LLM agents for recommendations is underexplored. In this work, we introduce a \textbf{Bias} \textbf{Rec}ommendation \textbf{Bench}mark (\textbf{BiasRecBench}) to highlight the critical vulnerability of such agents to biases in high-value real-world tasks. The benchmark includes three practical domains: paper review, e-commerce, and job recruitment. We construct a \textsc{Bias Synthesis Pipeline with Calibrated Quality Margins} that 1) synthesizes evaluation data by controlling the quality gap between optimal and sub-optimal options to provide a calibrated testbed to elicit the vulnerability to biases; 2) injects contextual biases that are logical and suitable for option contexts. Extensive experiments on both SOTA (Gemini-{2.5,3}-pro, GPT-4o, DeepSeek-R1) and small-scale LLMs reveal that agents frequently succumb to injected biases despite having sufficient reasoning capabilities to identify the ground truth. These findings expose a significant reliability bottleneck in current agentic workflows, calling for specialized alignment strategies for LLM-as-a-Recommender. The complete code and evaluation datasets will be made publicly available shortly.

cs.CY↗
arXiv2026

ZipServ: Fast and Memory-Efficient LLM Inference with Hardware-Aware Lossless Compression

Lossless model compression holds tremendous promise for alleviating the memory and bandwidth bottlenecks in bit-exact Large Language Model (LLM) serving. However, existing approaches often result in substantial inference slowdowns due to fundamental design mismatches with GPU architectures: at the kernel level, variable-length bitstreams produced by traditional entropy codecs break SIMT parallelism; at the system level, decoupled pipelines lead to redundant memory traffic. We present ZipServ, a lossless compression framework co-designed for efficient LLM inference. ZipServ introduces Tensor-Core-Aware Triple Bitmap Encoding (TCA-TBE), a novel fixed-length format that enables constant-time, parallel decoding, together with a fused decompression-GEMM (ZipGEMM) kernel that decompresses weights on-the-fly directly into Tensor Core registers. This "load-compressed, compute-decompressed" design eliminates intermediate buffers and maximizes compute intensity. Experiments show that ZipServ reduces the model size by up to 30%, achieves up to 2.21x kernel-level speedup over NVIDIA's cuBLAS, and expedites end-to-end inference by an average of 1.22x over vLLM. ZipServ is the first lossless compression system that provides both storage savings and substantial acceleration for LLM inference on GPUs.

cs.DC↗
arXiv2026

Are Dilemmas and Conflicts in LLM Alignment Solvable? A View from Priority Graph

As Large Language Models (LLMs) become more powerful and autonomous, they increasingly face conflicts and dilemmas in many scenarios. We first summarize and taxonomize these diverse conflicts. Then, we model the LLM's preferences to make different choices as a priority graph, where instructions and values are nodes, and the edges represent context-specific priorities determined by the model's output distribution. This graph reveals that a unified stable LLM alignment is very challenging, because the graph is neither static nor necessarily consistent in different contexts. Besides, it also reveals a potential vulnerability: priority hacking, where adversaries can craft deceptive contexts to manipulate the graph and bypass safety alignments. To counter this, we propose a runtime verification mechanism, enabling LLMs to query external sources to ground their context and resist manipulation. While this approach enhances robustness, we also acknowledge that many ethical and value dilemmas are philosophically irreducible, posing a long-term, open challenge for the future of AI alignment.

cs.AI↗
arXiv2026

Rethinking Deep Research from the Perspective of Web Content Distribution Matching

Despite the integration of search tools, Deep Search Agents often suffer from a misalignment between reasoning-driven queries and the underlying web indexing structures. Existing frameworks treat the search engine as a static utility, leading to queries that are either too coarse or too granular to retrieve precise evidence. We propose WeDas, a Web Content Distribution Aware framework that incorporates search-space structural characteristics into the agent's observation space. Central to our method is the Query-Result Alignment Score, a metric quantifying the compatibility between agent intent and retrieval outcomes. To overcome the intractability of indexing the dynamic web, we introduce a few-shot probing mechanism that iteratively estimates this score via limited query accesses, allowing the agent to dynamically recalibrate sub-goals based on the local content landscape. As a plug-and-play module, WeDas consistently improves sub-goal completion and accuracy across four benchmarks, effectively bridging the gap between high-level reasoning and low-level retrieval.

cs.LG↗
arXiv2026

Janus-Q: End-to-End Event-Driven Trading via Hierarchical-Gated Reward Modeling

Financial market movements are often driven by discrete financial events conveyed through news, whose impacts are heterogeneous, abrupt, and difficult to capture under purely numerical prediction objectives. These limitations have motivated growing interest in using textual information as the primary source of trading signals in learning-based systems. Two key challenges hinder existing approaches: (1) the absence of large-scale, event-centric datasets that jointly model news semantics and statistically grounded market reactions, and (2) the misalignment between language model reasoning and financially valid trading behavior under dynamic market conditions. To address these challenges, we propose Janus-Q, an end-to-end event-driven trading framework that elevates financial news events from auxiliary signals to primary decision units. Janus-Q unifies event-centric data construction and model optimization under a two-stage paradigm. Stage I focuses on event-centric data construction, building a large-scale financial news event dataset comprising 62,400 articles annotated with 10 fine-grained event types, associated stocks, sentiment labels, and event-driven cumulative abnormal return (CAR). Stage II performs decision-oriented fine-tuning, combining supervised learning with reinforcement learning guided by a Hierarchical Gated Reward Model (HGRM), which explicitly captures trade-offs among multiple trading objectives. Extensive experiments demonstrate that Janus-Q achieves more consistent, interpretable, and profitable trading decisions than market indices and LLM baselines, improving the Sharpe Ratio by up to 102.0% while increasing direction accuracy by over 17.5% compared to the strongest competing strategies.

cs.CL↗