Searcharxiv⌕ Search

arXiv subjects

Yuxuan Li

Publications and source records attributed to Yuxuan Li.

At least 19 recordsLinked to original sources
arXiv2026

ForgeTrain: Forging Production-Grade Training Frameworks via Harness-Driven AI Development

Training large models still relies on general-purpose frameworks such as Megatron-LM, whose generality tax constrains scenario-specific optimization and adds runtime overhead through accumulated abstraction. AI code generation reduces the cost of building a framework, and makes it affordable to forge one per scenario. We propose Forge Engineering: building a dedicated implementation from scratch for each scenario and iteratively optimizing it toward peak performance under correctness and usability constraints. Dedicated implementations inherit no abstraction boundaries, so they can integrate optimizations across the stack and reach a higher performance ceiling. We instantiate this paradigm for training frameworks as ForgeTrain, which holds a trusted framework as a golden reference and relaxes equivalence monotonically from Bit-for-Bit to Surpass. Experiments across multiple model--hardware configurations show that ForgeTrain consistently produces correct training engines and improves MFU over established training frameworks by 4.7--33.2%. To our knowledge this is the first production-grade training framework forged end-to-end by AI to match or surpass its human reference.

cs.SE↗
arXiv2026

ForgeMegakernel: A General Framework for Efficient Auto-Regressive Model Decode Megakernels

Auto-regressive model decode is bandwidth-bound, since every weight and key/value-cache byte crosses high-bandwidth memory once per token. A megakernel is an ideal solution, but existing automatic megakernel generation approaches cannot achieve both generalization across models and correctness guarantees. We present ForgeMegakernel, which generates a per-model high-performance decode megakernel using coding agents. ForgeMegakernel pairs a universal knowledge base of ten progressive milestones with an independent mid-state test oracle. The milestones provide the megakernel's structural properties: a fine-grained instruction stream for each SM, dependency counters replacing the global synchronization, and a shared-memory buffer pool for workload balance across SMs and greater parallelism. The test oracle derives the mid-states of the megakernel and checks the performance, error and precision during the generation process, guaranteeing a correct and trustworthy forged megakernel. We evaluated ForgeMegakernel on 14 representative decoding operations across eight model families spanning 0.6B-13B parameters. The generated megakernels achieved 50.5-85.9% MBU and geometric mean speedups of 1.21x over SGLang 0.5.18 and 1.54x over a megakernel compiler under identical configurations. Inside SGLang, evaluated on GSM8K with ragged prompts, all 14 megakernels decoded faster than the SGLang engine at comparable answer accuracy.

cs.DC↗
arXiv2026

The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement

Recursive self-improvement (RSI) enables AI systems to turn experience and feedback into persistent changes that improve both their capabilities and the process of future improvement. We first use the Headroom-Closed Index (HCI) to reveal the problems of existing LLMs, then introduce the RSI concept and its development roadmap: from improvement-execution autonomy, improvement-strategy autonomy, experience-acquisition autonomy, and environment-adaptation autonomy, to recursive meta-improvement. Next we examine RSI across scenarios (e.g., scientific discovery, embodied intelligence, software engineering), highlighting their distinct requirements and development speeds. Drawing on diverse industry practices and preliminary empirical evidence, we connect RSI research with practical systems and identify key challenges to achieving genuine RSI.

cs.LG↗
arXiv2026

Continual Policy Consolidation for Lifelong Robot Learning

Building a generalist robot policy requires continuously integrating new skills while preserving previously acquired behaviors. Directly optimizing a single policy over a growing task stream is difficult because robotic interaction is expensive, task distributions are heterogeneous, and sequential updates induce interference. To address these problems, we propose continual policy consolidation (CPC), a teacher--student framework that combines continual policy distillation with prioritized experience replay and expandable experts. This architecture separates skill acquisition from policy consolidation: teachers are trained independently through reinforcement learning, and their behaviors are continually distilled into a central generalist student. This decomposition retains the practical strength of reinforcement learning for task-specialized training while casting student-side consolidation as a supervised policy-learning problem. To balance stability and plasticity as the task stream grows, the student combines an expandable Transformer-based mixture-of-experts architecture with prioritized trajectory replay. Extensive experiments show that the student recovers a large proportion of teacher performance while achieving near-zero forgetting. These results demonstrate a scalable route for consolidating independently acquired robot skills into a continually growing generalist policy.

cs.LG↗
arXiv2026

ForgeStencil: Automating Per-Case Stencil Specialization from Kernels to 100+ Real Applications

On modern GPUs the fastest stencil kernel depends on the stencil's shape, precision, and host application, and a kernel tuned for one case is rarely fastest for another. Stencil DSLs, code generators, and autotuners instead pursued generality: a single human-authored method reused across cases and validated mainly on microbenchmarks, because per-case specialization was too costly to scale. ForgeStencil starts from the opposite assumption. Code-synthesis agents have reduced that cost enough to build a fresh solution for each case and deploy it end-to-end in real software. A Kernel Agent synthesizes CUDA and forges a per-configuration matrix of specialized operators that matches or exceeds the strongest publicly available state-of-the-art (SOTA) baseline for each case. An App Agent extends the principle to whole applications: it locates hotspots, rewrites application structure, and validates and integrates each change across 100+ real industrial and scientific codes. Most of the measured speedup comes from structural and host-side rewrites, with pure stencil replacement in the minority; the gain also correlates negatively with how well the baseline was already tuned, consistent with gains coming from specialization rather than generic reuse. Every result is checked by a measurement-integrity harness that turns an overstated speedup into a system-level error. The forged kernels reach a same-precision f32 geometric mean of 2.35x against the per-case SOTA baselines (fp16 gains, 1.95x, disclosed separately; A100), and the end-to-end application median is 1.41x across 100 codes, all against same-architecture GPU baselines with program-provided validation and timing. Of 116 candidates, every one that failed the correctness, measurement, or speedup criteria was recorded as rejected or downgraded instead of written up as a speedup.

cs.DC↗
arXiv2026

Trust Stack for Mental Health AI: A Survey of Calibration across Human, Interaction, and AI Layers

Language-based AI is increasingly deployed for mental health support, yet trust is evaluated in interdisciplinary but operationally misaligned ways: NLP and AI work measures robustness, safety, privacy, and explanations, while psychotherapy, HCI, and regulatory work emphasize therapeutic fidelity, lived experience, empathy, and reliance. Empathetic chatbots can elicit strong user trust without commensurate safety, while safer systems are under-trusted when their boundaries are opaque, a calibration gap no single community owns. Through a structured scoping synthesis of 61 papers, we survey this landscape into a three-layer framework separating (L1) human-oriented trust, (L2) interaction-oriented trustworthiness, and (L3) AI-oriented trustworthiness, and map five stakeholder perspectives onto these layers. We outline a research agenda for building socio-technically aligned trustworthy AI for mental health support, highlighting that the central objective should shift from maximizing perceived trust to calibrating human trust to demonstrated interaction- and AI-level trustworthiness.

cs.CL↗
arXiv2026

PolyDebate: A Game-Orchestrated Multimodal System for Debate Skills Practice and Evaluation

Debate is a structured form of persuasive communication that trains argument construction, rebuttal, oral delivery, and audience awareness. These skills are valued in education, language learning, and professional communication. Recent AI debate systems and LLM-based judges have advanced argument generation and debate evaluation, but most remain text-centered and rarely support learners through a complete multimodal practice experience. We introduce PolyDebate, a game-orchestrated multimodal system for English debate practice and evaluation. PolyDebate guides learners through staged one-on-one (1v1) debates with an AI opponent, while skill cards, props, and coins make persuasive strategies explicit and turn practice into a game-like interaction. During each session, the system captures learner speech and visual delivery evidence, generates context-aware opponent responses, and produces rubric-informed stage-level and overall feedback. PolyDebate is available as both an immersive Unity 3D game version and a web platform version that share the same workflow and evaluation services. Four studies covering AI opponent quality, evaluation coverage, AI judge feedback, and user perception show that PolyDebate brings debate interaction, gamified scaffolding, multimodal assessment, and structured feedback together in a practical workflow for debate skills practice. The demonstration video is available at https://youtu.be/mHwBG1_8Ebk.

cs.HC↗
arXiv2026

Behaviorally Grounded User Profiles from the Wild for Personalized Alignment and Multi-Perspective Reasoning

Persona-driven techniques increasingly adapt large language models (LLMs) to diverse contexts. However, existing methods predominantly rely on rigid, synthetic personas that flatten individual variation, rely on stereotypes, and miss the nuanced signals driving actual human preferences. We introduce profile behavioral grounding, a framework for extracting open-ended, high-fidelity user profiles directly from authentic, anonymized social media posts. We evaluate these profiles across two paradigms: train-time personalization via supervised finetuning (SFT) and non-parametric test-time multi-perspective reasoning. Across complex recommendation and open-ended query benchmarks, behaviorally grounded profiles consistently improve base models and outperform synthetic profile baselines, driving stronger parametric alignment and enabling richer, multifaceted reasoning. Our findings establish open-ended, behavior-derived profiles as a highly diverse and effective foundation for the next generation of personalized language systems. Our code base is available at https://github.com/ServiceNow/behavior-grounding.

cs.CL↗
arXiv2026

VeinCast: Physics-Guided Dynamic Field Graphs with Graph-Conditioned Fusion for Global Medium-Range Weather Forecasting

Global medium-range weather forecasting requires modeling structured yet state-dependent interactions among heterogeneous atmospheric fields. Existing data-driven models largely learn these interactions implicitly, whereas equation-level physical constraints may inherit approximation and model-form biases. We present VeinCast, a physics-guided dynamic field graph and graph-conditioned fusion framework that jointly forecasts 69 surface and upper-air fields. Within each local window, its Physics-Guided Dynamic Field Graph combines predefined atmospheric relations with state-dependent Top-K residual edges and adapts Earth-window attention using the resulting graph context. Graph-Conditioned Latent Fusion further employs graph context and source-node centrality to guide field-to-latent aggregation, while bounded feedback preserves field-specific information. On the $1.5^\circ$ ERA5 benchmark, VeinCast demonstrates competitive forecasting performance across all 69 meteorological fields at lead times of up to 14 days, compared with representative global weather forecasting models including FuXi, Pangu-Weather, GraphCast, FengWu, and ARROW. Ablations confirm that the two modules provide complementary gains, demonstrating the effectiveness of relational-level physical guidance for data-driven weather forecasting.

cs.LG↗
arXiv2026

MemPrism: Task-Conditioned Relational Memory Views for Long-Horizon Agents

Long-horizon agents rely on memory to reuse experiences, yet existing memory systems often assume that evidence can be directly consumed through a fixed representation. This leads to representation mismatch, where relevant information is available but not organized for the current decision. To this end, we propose MemPrism, a task-conditioned relational memory framework that separates persistent experience storage from decision-time working memory. MemPrism records interactions as the event stream and dynamically constructs relational views according to the current task context. A lightweight view policy selects the relation structure, evidence range, outcome condition, and granularity, while a deterministic composer and render transform historical facts into a temporary optical working-memory view for a frozen task policy. Experiments on long-horizon embodied and web-agent benchmarks show that MemPrism consistently improves the task performance, especially as trajectories become longer, while reducing memory token consumption. Furthermore, the learned view policy transfers across different VLMs without additional adaptation, demonstrating the effectiveness of task-conditioned relational views as a general memory interface for agents.

cs.AI↗
arXiv2026

"Allow" to Achieve, Over-Privileged Inadvertently: The Unintended Cost of Task-Completion-Driven Pop-up Decisions in Mobile GUI Agents

Mobile GUI agents routinely encounter system permission dialogs during task execution, yet their ability to grant only permissions that are necessary for the delegated task remains largely unexamined. We present a systematic study of this capability, which we term Permission Literacy. We construct a four-level permission framework based on task relevance and privacy risk and validate the evaluated scenarios with three independent experts in GUI-agent safety. We inject Android-style permission popups into real GUI tasks and evaluate four frontier multimodal large language models using synchronized annotated screenshots and UI-tree hierarchies, making the requester, permission, justification, and available actions accessible to the agent. Beyond the main study, we conduct controlled interventions that separately vary task context and agent-visible requester identity. Under the same Calendar task, changing only the requester from Calendar to PiMusic reduces grants from 26/32 to 0/32, revealing a strong but task-conditioned App-Trust Bias. Holding a popup fixed while changing task context also substantially changes authorization decisions, revealing a systematic Task-Prior Override. Prompt interventions can reduce unnecessary grants, but their effectiveness is inconsistent across models and may come at the cost of suppressing legitimate grants. These results suggest that separating task execution from permission authorization is a promising design direction for future work.

cs.CR↗
arXiv2026

SieveIVF: Threshold-Aware IVF Execution for Large-Scale Training Data Deduplication

Embedding-based training data deduplication retrieves candidate duplicate edges above an application similarity threshold, but fixed-probe inverted-file (IVF) search ignores this predicate when giving every query the same partition budget. Across four Hunyuan workloads, qualifying neighbors appear early despite sharply varying search depths. We present SieveIVF, a threshold-aware IVF executor that stops after $W$ consecutive searches find no qualifying candidate. The systems challenge is to preserve partition-major batching when each query's remaining work depends on prior results. Continuous batching groups ready queries by partition. A lookahead scheduler layers on top, exposing only work committed by the stopping rule to increase concurrency without changing stopping decisions or returned results. We implement SieveIVF in Lance. At $W=8$, SieveIVF is $4.1$--$7.6\times$ faster than fixed-probe IVF on four 10M Hunyuan workloads and $6.1$--$8.4\times$ faster on two public 100M workloads under the same index and search parameters, with pooled filtered top-10 recall losses of $0.03$--$1.13$ percentage points on Hunyuan and $1.43$--$2.29$ percentage points on the public workloads. These results show how an application predicate can guide IVF work allocation without changing the index or bounded top-$k$ interface.

cs.DB↗
arXiv2026

A Survey of Large Audio Language Models: Generalization, Trustworthiness, and Outlook

Advances in Large Language Models (LLMs) have paved the way for Multimodal Large Language Models (MLLMs). Among these, Large Audio Language Models (LALMs) are essential for realizing universal auditory intelligence. Despite their remarkable performance, the escalation of LALMs' capabilities has significantly outpaced the development of systemic frameworks to ensure their trustworthiness. This survey provides a comprehensive investigation into the endogenous mechanisms of LALMs, detailing the architectural innovations and alignment algorithms that facilitate emergent reasoning. Specifically, we analyze how the transition to unified end-to-end frameworks and the integration of continuous acoustic signals expand the attack surface. To rigorously evaluate the risks within these paradigms, we establish a comprehensive taxonomy of trustworthiness, categorizing critical vulnerabilities such as cross-modal jailbreaking, latent acoustic backdoors, and biometric privacy leakage. We review the state-of-the-art LALMs through six analytical pillars: hallucination, robustness, safety, privacy, fairness, and authentication. The pronounced imbalance between a mature offensive landscape and underdeveloped defenses highlights persistent trustworthiness gaps and multidimensional risks in audio-centric intelligence. Finally, we propose a roadmap advocating for ``Defense-in-Depth'' architectures, causal auditory world modeling, and intrinsic representation engineering to support the development of more reliable and trustworthy audio intelligence. Our project has been uploaded to GitHub https://github.com/Kwwwww74/Awesome-Trustworthy-AudioLLMs.

cs.SD↗
arXiv2026

Intraday Gas Fee Heterogeneity on Ethereum: Evidence from Operational Firms

Ethereum's EIP-1559 fee mechanism was designed under the assumption of homogeneous, myopic agents responding to a single congestion signal. We examine how this assumption interacts with the heterogeneous demand structure of real-world Ethereum users. Analyzing 62,142 confirmed transactions from seven operational firms across seven industries (January--March 2026), we document significant intraday gas-fee variation: fees peak at hour~12 UTC (7\,AM ET, $\hatβ_{12}=\$0.054$ above the U.S.\ evening baseline, $p<0.001$) and are associated with periods of elevated speculative-arbitrage activity. Operational firms exhibit heterogeneous scheduling responses moderated by transaction deferrability and gas intensity. Residual cost floors, i.e. the gap between observed expenditure and the counterfactual under perfect off-peak scheduling, range from 40.7\% to 92.5\% of actual expenditure, and persist even during the lowest-cost hours ($h\in\{20,21,22,23\}$ UTC, 3--6\,PM ET). We introduce an On-Chain Scheduling Matrix that maps firms to four scheduling regimes as a practical framework for managing gas-fee exposure under the current mechanism.

econ.EM↗
arXiv2026

Visual prompt engineering for video models

In the age of foundation models, a model is only as good as its prompt. For this reason, prompt engineering has become an essential technique for improving language model performance. Since video models are currently becoming foundation models for visual tasks (e.g., visual reasoning), we here ask whether they similarly benefit from visual prompt engineering: automatically modifying the task image to improve model performance. For example, for a visual physics reasoning task ("Where does the ball land, after passing a set of obstacles?"), an abstract sketch-like scene can be turned into a photorealistic version with a simple call to an image editing model. We find that visual prompt engineering, or VIPE for short, improves video reasoning performance across tasks. In fact, for video models, visual prompt engineering can be even more effective than classic text-based prompt engineering or test-time scaling. Ultimately, just as text-based prompt engineering systematically improves language model performance, visual prompt engineering can serve as a simple, compute-efficient approach to elicit better visual reasoning performance from video models. Example videos on our project page at https://visual-prompt-engineering.github.io/.

cs.CV↗
arXiv2026

FUSAR-R1: A Large-Scale Reasoning Model for Intelligent Interpretation of SAR Images

In recent years, large-scale vision-language models have been driving a paradigm shift in intelligent remote sensing image interpretation. By incorporating textual semantic information, the cognitive expression, semantic understanding, and human-computer interaction capabilities of interpretation models have been significantly improved, achieving initial progress in the field of Synthetic Aperture Radar (SAR) image interpretation. However, SAR images are affected by factors such as coherent imaging mechanisms, complex scattering characteristics, speckle noise interference, and target-background coupling, resulting in complex and variable image features with significant uncertainties and specializations. Existing SAR vision-language models do not yet possess the step-by-step analysis, logical judgment, and self-correction capabilities of human experts, making it difficult to support reliable intelligent interpretation in complex scenarios. To address this issue, this paper proposes a large-scale reasoning model, FUSAR-R1, for intelligent interpretation of SAR images. The model first constructs explicit chain-of-thought reasoning data by simulating the interpretation process of human experts and uses this data to guide instruction learning, thereby endowing the model with basic reasoning capabilities. Subsequently, a reinforcement learning strategy is introduced to optimize the model's outputs based on inference results, enabling self-correction and more reliable reasoning. Experimental results demonstrate that FUSAR-R1 consistently outperforms existing multimodal large-scale models across various SAR interpretation tasks, including target detection, target counting and classification, and land-cover category recognition.

cs.AI↗
arXiv2026

Simon-SR: Spatially Adaptive Modulation and Visual Prompt Adaptation for Text-Reinforced Super-Resolution

Single Image Super-Resolution (SISR) reconstructs high-quality images from low-resolution inputs. While recent multi-modal methods improve perceptual quality, they remain sensitive to erroneous priors and require expensive annotations. To address these issues, we propose Simon-SR, a multi-modal SISR framework leveraging learnable prompts for efficient semantic mining and robust text-image fusion. Our approach combines Contrastive Prompt Learning with Prompt-Guided Spatially Adaptive Refinement to enhance multi-modal alignment. Experiments demonstrate that Simon-SR surpasses state-of-the-art methods, achieving maximum improvements of 0.50 dB in PSNR, 0.0133 in SSIM, and 0.0695 in LPIPS. Code will be released.

cs.CV↗
arXiv2026

Reasoning LLM Improves Speaker Recognition in Long-form TV Dramas

Long-form TV dramas present a formidable challenge for comprehensive video understanding, where deciphering complex storyline often relies on \textbf{speaker recognition}, the task of accurately attributing each spoken utterance to its respective character. In this paper, we advance this field through two primary contributions. (1) We introduce \textbf{DramaSR-532K}, a large-scale benchmark comprising 532K annotated dialogue lines across more than 900 unique characters, necessitating the integration of auditory, linguistic, and visual cues for speaker recognition. (2) We propose \textbf{DramaSR-LRM}, a robust approach built upon a large reasoning model (LRM). DramaSR-LRM is designed to autonomously aggregate contextual evidence via multimodal tool-use, synthesizing diverse inputs to achieve high-fidelity attribution. Experimental results demonstrate that DramaSR-LRM significantly outperforms existing baselines, particularly on short utterances where acoustic biometrics are inherently unreliable. \textit{All the data and code will be made publicly available at the project page: https://www.github.com/198808xc/DramaSR-LRM.}

cs.CL↗