Searcharxiv⌕ Search

arXiv subjects

Li Li

Publications and source records attributed to Li Li.

At least 91 records · Page 5Linked to original sources
arXiv2026

Capsule: Efficient Player Isolation for Datacenters

We introduce Capsule, a mechanism for seamlessly sharing datacenter resources across multiple players. It decouples player-local and global states to achieve isolation and to maximize cross-player sharing. Our evaluations show that Capsule increases datacenter resource utilization by accommodating up to 2.25x more players without degrading the user experience. This improvement stems from Capsule consuming up to 1.43x less GPU, 3.11x less VRAM, 3.7x less CPU, and 3.87x less RAM compared to the baseline. We evaluated Capsule across four applications and various hardware configurations, including three distinct servers and a multi-server cluster. These results demonstrate that the Capsule design is portable to other game engines.

cs.DC↗
arXiv2026

Thermodynamics and transport in holographic QCD with Gauss-Bonnet corrections

Thermodynamics and transport are investigated in a holographic QCD model that extends the Einstein--Maxwell--Dilaton framework by incorporating Gauss--Bonnet corrections. Model parameters are fixed using state-of-the-art lattice QCD thermodynamics. The analysis then examines the equation of state at zero and finite baryon chemical potential, the phase structure in the temperature and chemical potential plane, as well as the shear and bulk viscosity to entropy ratios, $η/s$ and $ζ/s$, via the corresponding fluctuation equations. For a constant Gauss--Bonnet coupling, the model preserves a reasonable description of the equation of state and generates a temperature-dependent $η/s$, although the resulting profile remains monotonic near the crossover region, which does not satisfy the phenomenological expectation. When the Gauss--Bonnet coupling is allowed to depend on the dilaton, a non-monotonic $η/s$ and a peaked $ζ/s$ are obtained while maintaining agreement with thermodynamic constraints. The resulting phase diagram contains a critical end point in a phenomenologically relevant region.

hep-th↗
arXiv2026

Critical phenomenon inside asymptotically flat black holes with spontaneous scalarization

We study the interior dynamics of spontaneously scalarized black holes in Einstein-Maxwell-Scalar theory with zero cosmological constant, revealing novel critical phenomena. We demonstrate that, for a wide range of scalar-electromagnetic couplings, scalarized black holes possess no smooth inner Cauchy horizon and instead evolve into a spacelike Kasner singularity. The scalar hair triggers a rapid collapse of the Einstein-Rosen bridge at the would-be Cauchy horizon. Near the critical point where scalarized black holes bifurcate from the Reissner-Nordstrom solution, we establish a robust scaling relation between the Kasner parameter and the charge-to-mass ratio of the hairy black hole, opening a new window into the remarkable simplicity underlying black hole interiors.

gr-qc↗
arXiv2026

Exploring Code Analysis: Zero-Shot Insights on Syntax and Semantics with LLMs

Code analysis is fundamental in Software Engineering, supporting debugging, optimization, and security assessment. Human developers approach it through syntax parsing, static semantics inference, and dynamic reasoning. Traditional tools are effective but limited by language specificity and weak cross-language generalization. Large language models (LLMs) are promising for code tasks, yet their capabilities for fundamental code analysis remain underexplored. We structure our study around three aspects aligned with human practices: syntax parsing, static semantics inference, and dynamic reasoning. We evaluate 21 state-of-the-art LLMs across nine tasks in four languages (C, Java, Python, Solidity), including AST generation, CFG construction, data dependency, taint analysis, and flaky test reasoning. We apply a three-layer evaluation protocol (automated metrics, expert adjudication, consistency validation) to 3,124 code samples, achieving high inter-rater reliability (Cohen's kappa = 0.844-0.936) and strong human-machine agreement (Gwet's AC1 = 0.500-0.727, F1 = 0.791-0.882). While the best LLMs excel in syntax parsing (AST 90%+, expression matching 84-100%) and show promise in static analysis, their dynamic reasoning remains limited (<70%) with high data-shift sensitivity (per-project F1 varying 0-1.0). This hierarchy holds across model families and scales, suggesting fundamental rather than transient limitations. These findings show how LLMs complement traditional analyzers: they offer cross-language generalization but non-deterministic outputs needing validation, while traditional tools give deterministic guarantees but need language-specific configuration. We contribute a validated evaluation framework with comparison against traditional analyzers (Tree-sitter, Soot, Joern) and task-specific applicability tiers. Benchmark: https://github.com/mathieu0905/llm_code_analysis.git

cs.SE↗
arXiv2026

RAP: Runtime Adaptive Pruning for LLM Inference

Large language models (LLMs) excel at language understanding and generation, but their enormous computational and memory requirements hinder deployment. Compression offers a potential solution to mitigate these constraints. However, most existing methods rely on fixed heuristics and thus fail to adapt to runtime memory variations or heterogeneous KV-cache demands arising from diverse user requests. To address these limitations, we propose RAP, an elastic pruning framework driven by reinforcement learning (RL) that dynamically adjusts compression strategies in a runtime-aware manner. Specifically, RAP dynamically tracks the evolving ratio between model parameters and KV-cache across practical execution. Recognizing that FFNs house most parameters, whereas parameter -light attention layers dominate KV-cache formation, the RL agent retains only those components that maximize utility within the current memory budget, conditioned on instantaneous workload and device state. Extensive experiments results demonstrate that RAP outperforms state-of-the-art baselines, marking the first time to jointly consider model weights and KV-cache on the fly.

cs.LG↗
arXiv2026

Scenario Generation in Roundabouts with Adjustable Interaction Intensity

Roundabouts, characterized by frequent merging and yielding interactions, remain a safety-critical corner case for the development and testing of intelligent driving functions. However, extracting sufficient near-critical scenarios from naturalistic data is inefficient. Most existing scenario generation methods provide limited controllability over interaction intensity and criticality, making systematic safety testing and detailed analysis difficult. This paper presents an interaction-aware roundabout scenario generator with continuously adjustable interaction intensity. Geometric routes and temporal progress profiles are first decoupled and mapped to latent codes using pretrained autoencoders. Conditional latent generation is then performed with Wasserstein Generative Adversarial Networks (WGAN) to generate scenarios. Yielding is modeled as a controllable timing intervention via a compact yield code during the approach-to-entry segment, where interaction intensity is modulated by scaling the code with a factor $λ$. Results demonstrate enhanced timing-latent fidelity and plausible interaction responses compared to a baseline model. Under criticality-calibrated scaling, increasing $λ$ expands the safety margin, providing a scalable and controlled testing mechanism.

cs.RO↗
arXiv2026

"Someone Hid It": Query-Agnostic Black-Box Attacks on LLM-Based Retrieval

Large language models (LLMs) have been serving as effective backbones for retrieval systems, including Retrieval-Augmentation-Generation (RAG), Dense Information Retriever (IR), and Agent Memory Retrieval. Recent studies have demonstrated that such LLM-based Retrieval (LLMR) is vulnerable to adversarial attacks, which manipulates documents by token-level injections and enables adversaries to either boost or diminish these documents in retrieval tasks. However, existing attack studies mainly (1) presume a known query is given to the attacker, and (2) highly rely on access to the victim model's parameters or interactions, which are hardly accessible in real-world scenarios, leading to limited validity. To further explore the secure risks of LLMR, we propose a practical black-box attack method that generates transferable injection tokens based on zero-shot surrogate LLMs without need of victim queries or victim models knowledge. The effectiveness of our attack raises such a robustness issue that similar effects may arise from benign or unintended document edits in the real world. To achieve our attack, we first establish a theoretical framework of LLMR and empirically verify it. Under the framework, we simulate the transferable attack as a min-max problem, and propose an adversarial learning mechanism that finds optimal adversarial tokens with learnable query samples. Our attack is validated to be effective on benchmark datasets across popular LLM retrievers.

cs.CR↗
arXiv2026

TVRN: Invertible Neural Networks for Compression-Aware Temporal Video Rescaling

To fit diverse display and bandwidth constraints, high-frame-rate videos are temporally downscaled to low-frame-rate (LFR) and later upscaled, requiring joint optimization for effective frame-rate rescaling. However, existing methods typically link the two operations via training objectives, without fully exploiting their reciprocal nature, which may cause high-frequency information loss. Moreover, they overlook the impact of lossy codecs on LFR videos, limiting real-world applicability. In this work, we propose an end-to-end framework for compression-aware frame-rate rescaling, named TVRN. To regularize high-frequency information lost during frame-rate downscaling, TVRN adopts an invertible architecture that combines a Multi-Input Multi-Output Temporal Wavelet Transform with a high-frequency reconstruction module. To enable end-to-end training through non-differentiable lossy codecs, we design a surrogate network that approximates their gradients. Finally, to improve robustness under various compression levels, we extend TVRN to an asymmetric architecture by incorporating compression-aware features learned via a learning-to-rank strategy. Extensive experiments show that TVRN outperforms existing methods in reconstruction quality under industrial video compression settings. Source code is publicly available at https://github.com/fengxinmin/TVRN_public.

eess.IV↗
arXiv2026

EP-GRPO: Entropy-Progress Aligned Group Relative Policy Optimization with Implicit Process Guidance

Reinforcement learning with verifiable rewards (RLVR), particularly Group Relative Policy Optimization (GRPO), has advanced LLM reasoning. However, GRPO suffers from three credit assignment failures: uniform token-level granularity that ignores heterogeneous informational value, uniform polarity that penalizes correct steps and rewards incorrect ones, and zero-variance collapse that erases outcome-driven gradients. We systematically quantify these failures, revealing highly non-uniform token informativeness, widespread step-level polarity misalignment, and substantial training waste. To address these limitations, we propose Entropy-Progress Aligned GRPO (EP-GRPO), a framework that mines the model's intrinsic information flow for dense, self-supervised guidance. EP-GRPO integrates entropy-gated modulation to prioritize high entropy decision pivots, implicit process signals from policy divergence anchored to outcome advantages for directional token-level feedback without external reward models, and cumulative entropy mapping that enables progress-aligned advantage normalization, naturally maintaining gradient flow under zero reward variance. Extensive experiments on mathematical reasoning benchmarks demonstrate that EP-GRPO achieves superior accuracy and efficiency compared to GRPO and its variants. The code will be available.

cs.LG↗
arXiv2026

Towards Agentic Runtime Healing

Self-healing systems have long been a focus of research, aiming to enable software to recover from unexpected runtime errors without human intervention. Traditional approaches rely on predefined heuristic rules, such as reusing error handlers or rolling back to checkpoints, but these methods struggle to adapt to the diverse range of runtime errors. The emergence of Large Language Models offers a new opportunity to address this challenge. Leveraging their ability to understand and generate code and natural language, we propose using LLMs to dynamically generate error-handling strategies in real time, tailored to specific runtime contexts such as error messages and program states. We demonstrate the feasibility of this approach by designing such a framework, Healer, and empirically showing that it can handle runtime errors with a high success rate. When an unanticipated runtime error occurs, Healer leverages its internal LLM to generate bespoke error-handling code. The generated healing code is then executed to produce a corrected program state, allowing the program to continue execution with minimal disruption. We evaluate Healer across four code datasets and three state-of-the-art LLMs (GPT-3.5, GPT-4, and CodeQwen-7B), where GPT-4 can successfully recover from 72.8% of runtime errors, underscoring the promise of LLMs in this domain. Despite these promising results, challenges remain, particularly regarding the trustworthiness of LLM-generated code and its integration into existing systems. We mention potential solutions, such as safety checks and Healer-aware programming, to mitigate risks and ensure reliable operation. This work represents the first step toward agentic runtime healing, paving the way for more adaptive, resilient, and self-healing software systems.

cs.SE↗
arXiv2026

R-CoT: A Reasoning-Layer Watermark via Redundant Chain-of-Thought in Large Language Models

Large language models (LLMs) are widely deployed in multiple scenarios due to reasoning capabilities. In order to prevent the models from being misused, watermarking is generally employed to ensure ownership. However, most existing watermarking methods rely on superficial modifications to the model's output distribution, rendering the watermark vulnerable to perturbation and removal. To overcome this challenge, this paper introduces a reasoning-layer framework termed Redundant Chain-of-Thought (R-CoT), which embeds watermarks into the reasoning path. A dual-trajectory optimization mechanism based on GRPO enables the native and the watermark reasoning path to coexist within a shared parameter space, internalizing the watermark as a distinct reasoning policy. Therefore, the watermark is embedded into the model's stable reasoning path, avoiding the watermark failure caused by output-level perturbations. Experimental results show that, compared with existing methods, R-CoT achieves high watermark effectiveness and strong robustness. Under fine-tuning and other post-training operations, the true positive rate (TPR) consistently remains above 95%, exhibiting only marginal degradation.

cs.CR↗
arXiv2026

A Survey on LLM-based Conversational User Simulation

User simulation has long played a vital role in computer science due to its potential to support a wide range of applications. Language, as the primary medium of human communication, forms the foundation of social interaction and behavior. Consequently, simulating conversational behavior has become a key area of study. Recent advancements in large language models (LLMs) have significantly catalyzed progress in this domain by enabling high-fidelity generation of synthetic user conversation. In this paper, we survey recent advancements in LLM-based conversational user simulation. We introduce a novel taxonomy covering user granularity and simulation objectives. Additionally, we systematically analyze core techniques and evaluation methodologies. We aim to keep the research community informed of the latest advancements in conversational user simulation and to further facilitate future research by identifying open challenges and organizing existing work under a unified framework.

cs.CL↗
arXiv2026

DM-ASR: Diarization-aware Multi-speaker ASR with Large Language Models

Multi-speaker automatic speech recognition (ASR) aims to transcribe conversational speech involving multiple speakers, requiring the model to capture not only what was said, but also who said it and sometimes when it was spoken. Recent Speech-LLM approaches have shown the potential of unified modeling for this task, but jointly learning speaker attribution, temporal structure, and lexical recognition remains difficult and data-intensive. At the current stage, leveraging reliable speaker diarization as an explicit structural prior provides a practical and efficient way to simplify this task. To effectively exploit such priors, we propose DM-ASR, a diarization-aware multi-speaker ASR framework that reformulates the task as a multi-turn dialogue generation process. Given an audio chunk and diarization results, DM-ASR decomposes transcription into a sequence of speaker- and time-conditioned queries, each corresponding to one speaker in one time segment. This formulation converts multi-speaker recognition into a series of structured sub-tasks, explicitly decoupling speaker-temporal structure from linguistic content and enabling effective integration of diarization cues with the reasoning capability of large language models. We further introduce an optional word-level timestamp prediction mechanism that interleaves word and timestamp tokens, yielding richer structured outputs and better transcription quality. Our analysis shows that diarization systems provide more reliable speaker identities and segment-level boundaries, while LLMs excel at modeling linguistic content and long-range dependencies, demonstrating their complementary strengths. Experiments on Mandarin and English benchmarks show that the proposed approach achieves strong performance with relatively small models and training data, while remaining competitive with or outperforming existing unified approaches.

eess.AS↗
arXiv2026

Bubble dynamics and vortex formation in holographic first-order superfluid phase transitions

We investigate bubble dynamics in a holographic superfluid undergoing a first-order phase transition with spontaneous $U(1)$ symmetry breaking. Near the nucleation threshold, the system exhibits universal critical behavior governed by a single unstable mode, leading to logarithmic scaling of the time spent near the critical solution. The terminal bubble wall velocity increases with charge density but remains small due to strong dissipation. In multi-bubble collisions, vortex formation depends sensitively on the initial phases and deviates significantly from the geodesic rule. Notably, we identify a regime where three-bubble collisions produce a vortex-antivortex pair that subsequently annihilates, a phenomenon not predicted by the geodesic rule. The lifetime of this pair scales logarithmically with the distance to the critical collision radius. Our results underscore the crucial role of non-equilibrium dynamics in strongly coupled superfluids and provide new insights into topological defect formation during first-order phase transitions.

hep-th↗
arXiv2026

Optical Appearance of Scalarized Kerr-Newman Black Holes with Multiple Light Rings

This study investigates the optical appearance of rotating scalarized Kerr-Newman black holes in the Einstein-Maxwell-scalar theory with exponential coupling. By analyzing equatorial null geodesics, these black holes are classified into six types according to the number and properties of their light rings. Combining slow-rotation analysis with full numerical ray tracing, we investigate images of black holes illuminated by a geometrically and optically thin accretion disk. Unlike the Kerr case, where the image is typically governed by a single outer photon shell and a single critical curve, scalarized Kerr-Newman black holes can develop an additional inner photon shell outside the event horizon. This extra shell gives rise to an inner critical curve inside the usual outer one, which may be absent, partial, or closed depending on the black hole parameters and the observer's inclination. Moreover, it generates new higher-order images in the region between the two critical curves, some of which exhibit crescent-like morphologies distinct from the nearly circular higher-order images familiar from Kerr black holes. These features enrich the optical appearance of scalarized black holes and could serve as distinctive observational signatures in future high-resolution observations.

gr-qc↗
arXiv2026

Spotlight and Shadow: Attention-Guided Dual-Anchor Introspective Decoding for MLLM Hallucination Mitigation

Multimodal Large Language Models (MLLMs) have demonstrated remarkable reasoning capabilities yet continue to suffer from hallucination, where generated text contradicts visual content. In this paper, we introduce Dual-Anchor Introspective Decoding (DaID), a novel contrastive decoding framework that dynamically calibrates each token generation by mining the model's internal perceptual discrepancies. Specifically, DaID identifies a Spotlight layer to amplify visual factual signals and a Shadow layer to suppress textual inertia. By leveraging visual attention distributions to guide this dual-anchor selection process, our method ensures precise, token-specific adaptation. Experimental results across multiple benchmarks and MLLMs demonstrate that DaID significantly mitigates hallucination while enhancing general reasoning capabilities.

cs.CV↗
arXiv2026

High-Mobility Indium Native Oxide Transistors via Liquid-Metal Printing in Air

Oxide semiconductors have emerged as common channel materials in transistors and hold promise for next-generation electronics, yet achieving high mobility typically requires costly vacuum-based techniques. Here, ultrathin (5-nm) indium native oxide (InOx) prepared by ambient-air liquid-metal printing (LMP) at low temperature (250 °C), is applied as semiconducting channel in field-effect transistor (FET). The resulting InOx is found to be polycrystalline with large lateral grains that extend vertically throughout the film thickness. InOx FETs in a transfer length method (TLM) configuration demonstrate a high conductivity mobility (uCON) of 125 cm2 V-1 s-1, with systematic analysis of contact resistance confirming potential for channel length scaling. Integration with atomic-layer-deposited (ALD) gate dielectrics further reveals excellent compatibility, for instance, InOx FET integrated with HfO2 exhibits a high field-effect mobility (uFE) of 107 cm2 V-1 s-1, an on/off current ratio (ION/IOFF) of >107, a subthreshold swing (SS) of 204 mV dec-1, a gate leakage of <10-6 A cm-2, while maintaining stable performance over 104 endurance cycles without degradation. Post-fabrication oxygen-plasma treatment is applied to achieve enhancement-mode operation and a depletion-load inverter is demonstrated, exhibiting a voltage gain of 69.8 V/V. These results demonstrate the great potential of LMP InOx as semiconducting channel in high-performance and power-efficient transistors for next-generation oxide electronics.

cond-mat.mtrl-sci↗
arXiv2026

Beyond End-to-End: Dynamic Chain Optimization for Private LLM Adaptation on the Edge

Federated fine-tuning enables privacy-preserving LLM adaptation but faces a critical bottleneck: the disparity between LLMs' high memory demands and edge devices' limited capacity. To break the memory barrier, we propose Chain Federated Fine-Tuning (ChainFed), an innovative paradigm that forgoes end-to-end updates in favor of a sequential, layer-by-layer manner. It first trains the initial adapter to convergence, freezes its weights, and then proceeds to the next. This iterative train-and-freeze process forms an optimization chain, gradually enhancing the model's task-specific proficiency. ChainFed further integrates three core techniques: 1) Dynamic Layer Co-Tuning to bridge semantic gaps between sequentially tuned layers and facilitate information flow; 2) Globally Perceptive Optimization to endow each adapter with foresight beyond its local objective; 3) Function-Oriented Adaptive Tuning to automatically identify the optimal fine-tuning starting point. Extensive experiments on multiple benchmarks demonstrate the superiority of ChainFed over existing methods, boosting average accuracy by up to 46.46\%.

cs.DC↗