arXiv subjects

Sen Yang

Publications and source records attributed to Sen Yang.

At least 37 records · Page 2Linked to original sources

arXiv2026

ProSPy: A Profiling-Driven SQL-Python Agentic Framework for Enterprise Text-to-SQL

Large language models have substantially advanced Text-to-SQL systems, yet applying them to enterprise-scale databases remains challenging. Real-world databases often contain large and heterogeneous schemas, incomplete metadata, dialect-specific SQL syntax, and complex analytical questions that are difficult to solve with a single SQL query. To address these challenges, we propose ProSPy, a Profiling-driven SQL--Python agentic framework for enterprise-scale Text-to-SQL. ProSPy structures the reasoning process into four stages: it first extracts fine-grained data evidence through automatic profiling, progressively prunes large schemas into task-relevant contexts, fetches intermediate views through a dialect-agnostic SQL interface, and finally performs flexible downstream analysis with Python. This design combines the efficiency of SQL over large databases with the flexibility of Python-based analysis, while reducing reliance on unreliable metadata and improving robustness across SQL dialects. Experiments on Spider 2.0-Lite and Spider 2.0-Snow show that ProSPy consistently outperforms strong baselines with both open-source and proprietary models, achieving execution accuracies of 60.15% and 60.51% with Claude-4.5-Opus, without majority voting. Further analysis shows that ProSPy is robust to SQL dialect variations and achieves a favorable trade-off between schema recall and precision.

cs.CL↗

arXiv2026

EviLink: Multi-Path Schema Linking with Uncertainty-Guided Evidence Acquisition for Large-Scale Text-to-SQL

Schema linking is a difficult and important step in large-scale Text-to-SQL, where systems must identify a compact yet sufficient schema context from large and ambiguous databases. Existing methods often treat schema linking as deterministic selection around a single SQL path, but complex questions may admit multiple valid realizations with different schema needs. We reframe schema linking as uncertainty-aware schema-need inference over multiple plausible SQL paths, where the system distinguishes required schema items from path-dependent uncertain ones and acquires evidence only where needed. We instantiate this reframing with EviLink, which combines multi-hypothesis schema grounding with uncertainty-guided evidence acquisition. Experiments on BIRD-Dev and Spider2-Snow show that this perspective improves the balance among schema completeness, schema relevance, and token cost. On Spider2-Snow, EviLink achieves 90.15% field-level strict recall rate, uses 123.30K average tokens, and improves downstream SQL generation under a fixed generator.

cs.CL↗

arXiv2026

ClawTrace: Cost-Aware Tracing for LLM Agent Skill Distillation

Skill-distillation pipelines learn reusable rules from LLM agent trajectories, but they lack a key signal: how much each step costs. Without per-step cost, a pipeline cannot distinguish adding a missing step to fix a bug from removing an expensive step that never affected the outcome. We use the cost-attribution gap to ask whether the rule types inside a distilled skill transfer the same way to new tasks. ClawTrace records cost-attributed agent traces and compiles each session into a TraceCard; CostCraft reads TraceCards and writes three kinds of skill patches: preserve, prune, and repair. We find a pattern aggregate metrics hide. On 30 held-out SpreadsheetBench tasks across two seeds, removing prune patches roughly tripled the quality-regression count without lowering median cost. Across the full 84-task SkillsBench transfer, CostCraft saves no aggregate cost. All three quality regressions trace to the preserve lane, and both quality wins trace to the prune lane: prune patches act as quality guardrails while preserve patches drive regressions. We argue that reusable agent skills should be evaluated at the rule-type level, not as monolithic instruction packages. To support this, we release ClawTrace, the TraceCard schema, and the full set of typed skills.

cs.AI↗

arXiv2026

Capability Conditioned Scaffolding for Professional Human LLM Collaboration

Large language model personalization typically adapts outputs to user preferences and style but does not account for differences in user evaluation capacity across domains of expertise. This limitation can encourage Professional Domain Drift, where users rely on AI generated reasoning in domains they cannot reliably evaluate. We introduce Capability Conditioned Scaffolding, a typed framework that partitions expertise into strong, mixed, and weak domains and conditions intervention behavior on structured capability profiles. A pilot evaluation across multiple MMLU subsets and four LLM substrates shows consistent profile conditioned intervention behavior, including categorical inversion under profile swapping and selective activation in mixed domain risk zones. These findings suggest that capability aware scaffolding can support more reliable professional human AI collaboration beyond stylistic personalization.

cs.CL↗

arXiv2026

AGN STORM 2. XII. Ground-Based Optical Photometry and Lag Measurements of Mrk 817

We present the ground-based imaging campaign and light curves of Markarian 817 as part of the multiwavelength monitoring program AGN STORM\,2. Observations were carried out over 1.4 years in \emph{uBgVriz} filters, with a median cadence of 0.4 days in \emph{g}. Reverberation lags are measured using three methods (ICCF, JAVELIN, and PyROA) with the Swift UVW2 band (1928 Å) as the reference light curve. The ICCF centroid lags range from $3.0\pm0.8$ days for the $u$ band up to $7.9\pm1.5$ days for $z$, and are consistent with a $τ\propto λ^{4/3}$ dependence, the relation expected for lamp-post reprocessing by a Shakura-Sunyaev disk. Lags measured with the other methods are systematically shorter, and deviate from a $λ^{4/3}$ power-law spectrum at long wavelengths. The lags exceed thin-disk reprocessing predictions by factors of $\sim$3-6, similar to the ``disk size discrepancy'' seen in other Seyfert galaxies. We divide the campaign into three epochs with different levels of mean luminosity and X-ray obscuring column density and find that the lags vary by as much as a factor of 2 between epochs. The intrinsic spectral energy distribution is bluer and brighter during the first third of the campaign, and the longest continuum reverberation lags are obtained during that period. These results suggest that changes in ionizing luminosity can produce large variations in continuum lags on short timescales by altering the diffuse continuum luminosity emitted by the broad-line region and/or obscuring outflow, although changes in obscuration between the central engine and broad-line region may also contribute to the lag variations.

astro-ph.GA↗

arXiv2026

Your Loss is My Gain: Low Stake Attacks on Liquid Staking Pools

Permissionless Proof-of-Stake (PoS) economic security is predicated on the high cost of violating consensus safety or liveness. We show that liquid staking introduces additional risks that are not captured by standard PoS economic security arguments. Through an empirical study of Ethereum data, we find that the operational performance of liquid staking pools is positively associated with subsequent normalized liquid staking token (LST) returns. Motivated by this, we present a cross-layer attack: a low-stake adversary can manipulate the consensus protocol to degrade a target pool's performance and take application-layer positions that profit if the market reprices the corresponding \gls{LST} in-line with the historically observed association. To make the consensus layer manipulation concrete, we develop a deep reinforcement learning (DRL) framework to automatically discover attack strategies. Our evaluation shows that the learned strategies can recover near-optimal theoretical attacks and uncover new manipulation behaviors that significantly degrade target pool performance. We further characterize feasible application-layer monetization channels and analyze leveraged shorting in detail using Monte Carlo simulations, showing that such attacks can be profitable with over one-half probability for LSTs of major staking pools. Our findings reveal a previously overlooked attack surface in PoS systems with liquid staking and expose a gap between consensus and economic security.

cs.GT↗

arXiv2026

StealthGraph: Exposing Domain-Specific Risks in LLMs through Knowledge-Graph-Guided Harmful Prompt Generation

Large language models (LLMs) are increasingly applied in specialized domains such as finance and healthcare, where they introduce unique safety risks. Domain-specific datasets of harmful prompts remain scarce and still largely rely on manual construction; public datasets mainly focus on explicit harmful prompts, which modern LLM defenses can often detect and refuse. In contrast, implicit harmful prompts-expressed through indirect domain knowledge-are harder to detect and better reflect real-world threats. We identify two challenges: transforming domain knowledge into actionable constraints and increasing the implicitness of generated harmful prompts. To address them, we propose an end-to-end framework that first performs knowledge-graph-guided harmful prompt generation to systematically produce domain-relevant prompts, and then applies two-strategy obfuscation rewriting to convert explicit harmful prompts into implicit variants via direct and context-enhanced rewriting. This framework yields high-quality datasets combining strong domain relevance with implicitness, enabling more realistic red-teaming and advancing LLM safety research. We release our code and datasets on GitHub.

cs.CL↗

arXiv2026

Scaffold-Conditioned Preference Triplets for Controllable Molecular Optimization with Large Language Models

Molecular property optimization is central to drug discovery, yet many deep learning methods rely on black-box scoring and offer limited control over scaffold preservation, often producing unstable or biologically implausible edits. While large language models (LLMs) are promising molecular generators, optimization remains constrained by the lack of chemistry-grounded preference supervision and principled data curation. We introduce \textbf{Scaffold-Conditioned Preference Triplets (SCPT)}, a pipeline that constructs similarity-constrained triplets $\langle\text{scaffold}, \text{better}, \text{worse}\rangle$ via scaffold alignment and chemistry-driven filters for validity, synthesizability, and meaningful property gains. Using these preferences, we align a pretrained molecular LLM as a conditional editor, enabling property-improving edits that retain the scaffold. Across single- and multi-objective benchmarks, SCPT improves optimization success and property gains while maintaining higher scaffold similarity than competitive baselines. Compared with representative non-LLM molecular optimization methods, SCPT-trained LLMs are better suited to scaffold-constrained and multi-objective optimization. In addition, models trained on single-property and two-property supervision generalize effectively to three-property tasks, indicating promising extrapolative generalization under limited higher-order supervision. SCPT also provides controllable data-construction knobs that yield a predictable similarity-gain frontier, enabling systematic adaptation to diverse optimization regimes.

cs.LG↗

arXiv2026

CycleChart: A Unified Consistency-Based Learning Framework for Bidirectional Chart Understanding and Generation

Current chart-related tasks, such as chart generation (NL2Chart), chart schema parsing, chart data parsing, and chart question answering (ChartQA), are typically studied in isolation, preventing models from learning the shared semantics that link chart creation and interpretation. We introduce CycleChart, a consistency-based learning framework for bidirectional chart understanding and generation. Unlike conventional multi-task approaches that draw training samples independently across tasks, CycleChart organizes all tasks around each single data instance. From a source table and natural-language query, the model generates a chart specification, renders and executes it, then learns to recover the schema and underlying data from the resulting chart image. This per-instance lifecycle design lets the model capture the full chain of transformations, from raw data through visual encoding to structured recovery, and a generate--parse consistency objective enforces semantic alignment between the forward generation and reverse parsing directions. To support this framework, we construct CycleChart-Bench, a lifecycle-aligned benchmark where every chart sample carries aligned annotations for generation, schema parsing, data parsing, and question answering. CycleChart achieves strong results across all four tasks and transfers effectively to unseen external benchmarks, demonstrating improved cross-task generalization and marking a step toward more general chart understanding models.

cs.CL↗

arXiv2026

Diffusion-based Evolutionary Optimization for 3D Multi-Objective Molecular Generation

Optimizing conflicting molecular properties while strictly adhering to complex 3D structural constraints constitutes a challenging Constrained Multi-Objective Optimization Problem (CMOP). Traditional Evolutionary Algorithms (EAs) destroy chemical valency in 3D space, whereas 3D diffusion models act as rigid generators requiring costly retraining for novel objectives. To bridge this gap, we propose a progressive algorithmic suite. First, we introduce the Evolutionary-Guided Diffusion (EGD) operator, which executes crossover and mutation at an optimally calibrated noise level, leveraging a pre-trained denoising network to project chimeric states back onto the valid chemical manifold. Second, to combat the severe loss of molecular structural diversity inherent in traditional EMO frameworks, we design a Structure-Aware Environmental Selection (SAES) mechanism that explicitly enforces structural distinctiveness. Finally, synergizing EGD and SAES, we develop the Diffusion-based Evolutionary Molecular Optimization (DEMO) framework for CMOPs. To safely navigate disjoint feasible regions, DEMO employs a tri-population architecture with distinct goals: exploring novel chemical scaffolds, refining partially assembled intermediates, and fine-tuning perfectly feasible elite molecules. Extensive experiments across single-property targeting, unconstrained MOPs, multi-fragment CMOPs, and 3D protein-ligand docking demonstrate that our method comprehensively outperforms state-of-the-art baselines and traditional EMO frameworks. Operating entirely zero-shot, this suite consistently discovers highly diverse, chemically valid Pareto frontiers.

cs.NE↗

arXiv2026

Geographical Centralization Resilience in Ethereum's Block-Building Paradigms

Decentralization has an important geographic dimension that conventional metrics, such as stake distribution, often overlook. Validator location affects resilience to regional shocks (e.g., outages, natural disasters, or government intervention) as well as fairness in reward access. Yet major blockchain protocols do not encode geographical location in their rules; instead, validator locations emerge from a combination of economic incentives, regulatory constraints, infrastructure availability, and validator deployment choices. When some locations offer systematic advantages, validators may strategically co-locate to increase expected rewards, as in Ethereum, where validators cluster along the Atlantic corridor, which exhibits favorable latency. In this paper, we develop a formal model of validators' geographical positioning incentives under Ethereum's protocol design, capturing the interaction between its two block-building paradigms, local and external block building, and the distribution of validators and information sources. We analyze the model under a mean-field approximation and complement it with agent-based simulations calibrated with real-world latency data to quantify how these incentives translate into geographical concentration under heterogeneous geographic and infrastructural conditions. Our results show that Ethereum's block-building architecture is not geographically neutral. Both paradigms create location-dependent payoffs and incentives to move closer to payoff-relevant parties to reduce propagation delays, though through different mechanisms. Asymmetric access to information sources further increases geographical centralization. We also show that consensus parameters, including attestation thresholds and slot times, affect latency sensitivity and can strengthen these effects. Finally, we discuss implications for protocol design and possible mitigation directions.

cs.CR↗

arXiv2026

The Eleventh NTIRE 2026 Efficient Super-Resolution Challenge Report

This paper reviews the NTIRE 2026 challenge on efficient single-image super-resolution with a focus on the proposed solutions and results. The aim of this challenge is to devise a network that reduces one or several aspects, such as runtime, parameters, and FLOPs, while maintaining PSNR of around 26.90 dB on the DIV2K_LSDIR_valid dataset, and 26.99 dB on the DIV2K_LSDIR_test dataset. The challenge had 95 registered participants, and 15 teams made valid submissions. They gauge the state-of-the-art results for efficient single-image super-resolution.

cs.CV↗

arXiv2026

Computational Pathology in the Era of Emerging Foundation and Agentic AI -- International Expert Perspectives on Clinical Integration and Translational Readiness

Recent breakthroughs in artificial intelligence through foundation models and agents have accelerated the evolution of computational pathology. Demonstrated performance gains reported across academia in benchmarking datasets in predictive tasks such as diagnosis, prognosis, and treatment response have ignited substantial enthusiasm for clinical application. Despite this development momentum, real world adoption has lagged, as implementation faces economic, technical, and administrative challenges. Beyond existing discussions of technical architectures and comparative performance, this review considers how these emerging AI systems can be responsibly integrated into medical practice by connecting deployable clinical relevance with downstream analytical capabilities and their technical maturity, operational readiness, and economic and regulatory context. Drawing on perspectives from an international group, we provide a practical assessment of current capabilities and barriers to adoption in patient care settings.

cs.CE↗

arXiv2026

Testing Gravitational-Wave Signal From Verification Binaries with Space-Based Gravitational-Wave Detectors

Space-based gravitational wave (GW) detectors will open the millihertz band to survey ultra-compact binaries (UCBs). \textit{Verification binaries} (VBs) is a key to verifying the performance of space-based GW detectors because its parameters are known from electromagnetic observations and it is expected to be a detectable source of GW. We evaluated 73 VBs, computing their detection prospects and parameter estimation precision for individual GW detectors and networks. Among single detectors, DECIGO shows the highest sensitivity, detecting 71 sources at signal-to-noise ratio $ρ$ $\geq$ 5, compared to 42 for LISA, 32 for Taiji, and 27 for TianQin, while the full TianQin + LISA + Taiji + DECIGO network improves this to 73 detectable sources. For parameter estimation, individual detectors achieve median precisions on the order of $\sim 10^{-2}-10^{-1} \, \text{M}_{\odot}$ for chirp mass, $\sim 1\,\text{kpc}$ for distance, $\sim 1-17\,\text{deg}$ for inclination and $\sim 10^{-4}-10^{-2}\,\text{deg}^2$ for sky localization. The complete TianQin + LISA + Taiji + DECIGO network enhances these constraints substantially, reducing the median uncertainties to approximately $\sim 10^{-2} \, \text{M}_{\odot}$ in chirp mass, $\sim 10^{-2}\,\text{kpc}$ in distance, $\sim 1\,\text{deg}$ in inclination and $\sim 10^{-4}\,\text{deg}^2$ in sky localization. The upcoming space-based GW detectors, especially their networks, have outstanding observational capabilities for UCB, which will advance our research on multi-messenger astronomy and deepen our understanding of UCB in the Milky Way.

gr-qc↗

arXiv2026

Reconstructing KV Caches with Cross-layer Fusion For Enhanced Transformers

Transformer decoders have achieved strong results across tasks, but the memory required for the KV cache becomes prohibitive at long sequence lengths. Although Cross-layer KV Cache sharing (e.g., YOCO, CLA) offers a path to mitigate KV Cache bottleneck, it typically underperforms within-layer methods like GQA. To understand the root cause, we investigate the information flow of keys and values of the top-layers. Our preliminary reveals a clear distribution: values are predominantly derived from the bottom layer, while keys draw more information from both bottom and middle layers. Building upon this, we propose FusedKV, whose top-layer KV caches are a learnable fusion of the most informative ones from the bottom and middle layers. This fusion operates directly on post-RoPE keys, preserving relative positional information without the computational cost of re-applying rotary embeddings. To further improve efficiency, we propose FusedKV-Lite, an cross-layer sharing approach, where top-layer KV caches are directly derived from the bottom-layer values and the middle-layer keys. Compared to FusedKV, FusedKV-Lite reduces I/O overhead at the cost of a slight increase in perplexity. In experiments on LLMs ranging from 332M to 4B parameters, our proposed method reduce 50\% cache memory while achieving lower validation perplexity than the standard Transformer decoder, establishing it as a memory-efficient, high-performance architectural alternative.

cs.CL↗

arXiv2026

Boost+: Equitable, Incentive-Compatible Block Building

Block space on the blockchain is scarce and must be allocated efficiently through block building. However, Ethereum's current block-building ecosystem, MEV-Boost, has become highly centralized due to integration, which distorts competition, reduces blockspace efficiency, and obscures MEV flow transparency. To guarantee equitability and economic efficiency in block building, we propose $\mathrm{Boost+}$, a system that decouples the process into collecting and ordering transactions, and ensures equal access to all collected transactions. The core of $\mathrm{Boost+}$ is the mechanism $\mathit{M}_{\mathrm{Boost+}}$, built around a default algorithm. $\mathit{M}_{\mathrm{Boost+}}$ aligns incentives for both searchers (intermediaries that generate or route transactions) and builders: Truthful bidding is a dominant strategy for all builders. For searchers, truthful reporting is dominant whenever the default algorithm dominates competing builders, and it remains dominant for all conflict-free transactions, even when builders may win. We further show that even if a searcher can technically integrate with a builder, non-integration combined with truthful bidding still dominates any deviation for conflict-free transactions. We also implement a concrete default algorithm informed by empirical analysis of real-world transactions and evaluate its efficacy using historical transaction data.

cs.CR↗

arXiv2026

PLawBench: A Rubric-Based Benchmark for Evaluating LLMs in Real-World Legal Practice

As large language models (LLMs) are increasingly applied to legal domain-specific tasks, evaluating their ability to perform legal work in real-world settings has become essential. However, existing legal benchmarks rely on simplified and highly standardized tasks, failing to capture the ambiguity, complexity, and reasoning demands of real legal practice. Moreover, prior evaluations often adopt coarse, single-dimensional metrics and do not explicitly assess fine-grained legal reasoning. To address these limitations, we introduce PLawBench, a Practical Law Benchmark designed to evaluate LLMs in realistic legal practice scenarios. Grounded in real-world legal workflows, PLawBench models the core processes of legal practitioners through three task categories: public legal consultation, practical case analysis, and legal document generation. These tasks assess a model's ability to identify legal issues and key facts, perform structured legal reasoning, and generate legally coherent documents. PLawBench comprises 850 questions across 13 practical legal scenarios, with each question accompanied by expert-designed evaluation rubrics, resulting in approximately 12,500 rubric items for fine-grained assessment. Using an LLM-based evaluator aligned with human expert judgments, we evaluate 10 state-of-the-art LLMs. Experimental results show that none achieves strong performance on PLawBench, revealing substantial limitations in the fine-grained legal reasoning capabilities of current LLMs and highlighting important directions for future evaluation and development of legal LLMs. Data is available at: https://github.com/skylenage/PLawbench.

cs.CL↗

arXiv2026

Deformations of Chow groups via cyclic homology

Let $X$ be a smooth projective variety over an arbitrary field $k$ of characteristic zero. We explore infinitesimal deformations of the Chow group $CH^{p}(X)$ via its formal completion $\widehat{CH}^{p}$, a functor defined on the category of local augmented Artinian $k$-algebras. Under a natural vanishing condition on Hodge cohomology groups, for certain augmented graded Artinian $k$-algebras $A$ with the maximal ideal $m_{A}$, we prove that \[ \widehat{CH}^{p}(A) \cong H^{p}(X, Ω^{p-1}_{X/ k})\otimes_{k}m_{A}. \]This extends earlier results of Bloch and others from the case where $k$ is algebraic over $\mathbb{Q}$ to arbitrary fields of characteristic zero,and gives a partial affirmative answer to a general question linking the pro-representability of Chow groups to a specific set of Hodge-theoretic vanishing conditions.

math.AG↗