Searcharxiv⌕ Search

arXiv subjects

Yong Yu

Publications and source records attributed to Yong Yu.

At least 55 records · Page 3Linked to original sources
arXiv2026

A Survey of Process Reward Models: From Outcome Signals to Process Supervisions for Large Language Models

Although Large Language Models (LLMs) exhibit advanced reasoning ability, conventional alignment remains largely dominated by outcome reward models (ORMs) that judge only final answers. Process Reward Models(PRMs) address this gap by evaluating and guiding reasoning at the step or trajectory level. This survey provides a systematic overview of PRMs through the full loop: how to generate process data, build PRMs, and use PRMs for test-time scaling and reinforcement learning. We summarize applications across math, code, text, multimodal reasoning, robotics, and agents, and review emerging benchmarks. Our goal is to clarify design spaces, reveal open challenges, and guide future research toward fine-grained, robust reasoning alignment.

cs.CL↗
arXiv2026

Modular Representation Compression: Adapting LLMs for Efficient and Effective Recommendations

Recently, large language models (LLMs) have advanced recommendation systems (RSs), and recent works have begun to explore how to integrate LLMs into industrial RSs. While most approaches deploy LLMs offline to generate and pre-cache augmented representations for RSs, high-dimensional representations from LLMs introduce substantial storage and computational costs. Thus, it is crucial to compress LLM representations effectively. However, we identify a counterintuitive phenomenon during representation compression: Mid-layer Representation Advantage (MRA), where representations from middle layers of LLMs outperform those from final layers in recommendation tasks. This degraded final layer renders existing compression methods, which typically compress on the final layer, suboptimal. We interpret this based on modularity theory that LLMs develop spontaneous internal functional modularity and force the final layer to specialize in the proxy training task. Thus, we propose \underline{M}odul\underline{a}r \underline{R}epresentation \underline{C}ompression (MARC) to explicitly control the modularity of LLMs. First, Modular Adjustment explicitly introduces compression and task adaptation modules, enabling the LLM to operate strictly as a representation-learning module. Next, to ground each module to its specific task, Modular Task Decoupling uses information constraints and different network structures to decouple tasks. Extensive experiments validate that MARC addresses MRA and produces efficient representations. Notably, MARC achieved a 2.82% eCPM lift in an online A/B test within a large-scale commercial search advertising scenario.

cs.IR↗
arXiv2026

Enhancing astrometric registration of Chinese historical Astronomical Digital Plates with deep learning

China has systematically collected nighttime astronomical plates since 1900, creating a large historical dataset that has been digitized with optical scanners. For astrometric registration of these digitized plates, sources were first extracted using SExtractor, and then matched astrometrically with Astrometry.net and the Gaia catalog. However, suboptimal early storage conditions and subsequent environmental deterioration have impeded accurate source matching, resulting in processing failures for several thousand digitized plates. In this work, we introduce a Transformer-based classification model that takes cutouts of SExtractor-detected sources as input and leverages multi-scale feature fusion to identify trustworthy stellar sources on the plates. Trained on plates with successful astrometric calibration, our AI-based classifier was then applied to SExtractor detected sources of 1883 digitized plates, enabling us to complete the astrometric registration for 1353 of them. This AI-augmented pipeline streamlines the processing of historical plate archives and enhances their scientific value for long-term time-domain astronomical studies.

astro-ph.IM↗
arXiv2026

AdverMCTS: Combating Pseudo-Correctness in Code Generation via Adversarial Monte Carlo Tree Search

Recent advancements in Large Language Models (LLMs) have successfully employed search-based strategies to enhance code generation. However, existing methods typically rely on static, sparse public test cases for verification, leading to pseudo-correctness -- where solutions overfit the visible public tests but fail to generalize to hidden test cases. We argue that optimizing against a fixed, weak environment inherently limits robustness. To address this, we propose AdverMCTS, a novel adversarial Monte Carlo Tree Search framework that combats pseudo-correctness by coupling code search with active vulnerability discovery. AdverMCTS formulates generation as a minimax-style game between a Solver agent, which synthesizes code candidates, and an Attacker agent, which evolves to generate targeted corner test cases that exploit logical divergences in the current code pool. These discovered tests form a dynamic, progressively hostile filter that penalizes fragile reasoning. Extensive experiments demonstrate that AdverMCTS significantly outperforms state-of-the-art baselines, effectively reducing false positive rates and forcing the model to generalize beyond the initial constraints. The resources of this work are available at https://anonymous.4open.science/r/AdverMCTS_open-A255.

cs.SE↗
arXiv2026

UniCon: A Unified System for Efficient Robot Learning Transfers

Deploying learning-based controllers across heterogeneous robots is challenging due to platform differences, inconsistent interfaces, and inefficient middleware. To address these issues, we present UniCon, a lightweight framework that standardizes states, control flow, and instrumentation across platforms. It decomposes workflows into execution graphs with reusable components, separating system states from control logic to enable plug-and-play deployment across various robot morphologies. Unlike traditional middleware, it prioritizes efficiency through batched, vectorized data flow, minimizing communication overhead and improving inference latency. This modular, data-oriented approach enables seamless sim-to-real transfer with minimal re-engineering. We demonstrate that UniCon reduces code redundancy when transferring workflows and achieves higher inference efficiency compared to ROS-based systems. Deployed on over 12 robot models from 7 manufacturers, it has been successfully integrated into ongoing research projects, proving its effectiveness in real-world scenarios.

cs.RO↗
arXiv2026

Automatically Benchmarking LLM Code Agents through Agent-Driven Annotation and Evaluation

Recent advances in code agents have enabled automated software development at the project level, supported by large language models (LLMs). However, existing benchmarks for code agent evaluation face two major limitations. First, creating high-quality project-level evaluation datasets requires extensive domain expertise, leading to prohibitive annotation costs and limited diversity. Second, while recent Agent-as-a-Judge paradigms address the rigidity of traditional unit tests by enabling flexible metrics, their reliance on In-Context Learning (ICL) with general LLMs often results in inaccurate assessments that misalign with human standards. To address these challenges, we propose an agent-driven benchmark construction pipeline that leverages human supervision to efficiently generate diverse project-level tasks. Based on this, we introduce PRDBench, comprising 50 real-world Python projects across 20 domains, each with structured Product Requirement Documents (PRDs) and comprehensive criteria. Furthermore, to overcome the inaccuracy of general LLM judges, we propose a highly reliable evaluation framework powered by a specialized, fine-tuned model. Based on Qwen3-Coder-30B, our dedicated PRDJudge achieves over 90% human alignment in fixed-interface scenarios. Extensive experiments demonstrate that our suite provides a scalable, robust, and highly accurate framework for assessing state-of-the-art code agents.

cs.SE↗
arXiv2026

Low-noise Optomechanical Single Phonon-photon Conversion for Quantum Networks

Nano-structured optomechanical crystals (OMC) form an interface between mechanical modes with long coherence times and telecom optical photons, ideal for long-distance distribution of quantum information. However, the implementation of scalable quantum networks based on OMCs has been inhibited by thermal mechanical noise. Here, we overcome this limitation using a quasi-two-dimensional OMC and generate single photons via single phonon-photon conversion. In this work, we verify the low thermal noise and high purity of the generated single photons through a Hanbury Brown-Twiss experiment with $g^{(2)}(0)=0.35^{+0.10}_{-0.08}$. We perform Hong-Ou-Mandel interference of the emitted photons showcasing the indistinguishability and coherence with visibility $V=0.52 \pm 0.15$ after 1.43 km fiber delay. Lastly, we use two-photon interference to measure the temporal wavepackets of optomechanically generated single photons demonstrating narrow bandwidths as low as 10 MHz. Our results pave the way for multinode quantum networks of mechanical oscillators and hybrid entanglement generation between mechanical oscillators and telecom quantum emitters.

quant-ph↗
arXiv2026

Turing Test on Screen: A Benchmark for Mobile GUI Agent Humanization

The rise of autonomous GUI agents has triggered adversarial countermeasures from digital platforms, yet existing research prioritizes utility and robustness over the critical dimension of anti-detection. We argue that for agents to survive in human-centric ecosystems, they must evolve Humanization capabilities. We introduce the ``Turing Test on Screen,'' formally modeling the interaction as a MinMax optimization problem between a detector and an agent aiming to minimize behavioral divergence. We then collect a new high-fidelity dataset of mobile touch dynamics, and conduct our analysis that vanilla LMM-based agents are easily detectable due to unnatural kinematics. Consequently, we establish the Agent Humanization Benchmark (AHB) and detection metrics to quantify the trade-off between imitability and utility. Finally, we propose methods ranging from heuristic noise to data-driven behavioral matching, demonstrating that agents can achieve high imitability theoretically and empirically without sacrificing performance. This work shifts the paradigm from whether an agent can perform a task to how it performs it within a human-centric ecosystem, laying the groundwork for seamless coexistence in adversarial digital environments.

cs.AI↗
arXiv2026

LogitsCoder: Towards Efficient Chain-of-Thought Path Search via Logits Preference Decoding for Code Generation

Code generation remains a challenging task that requires precise and structured reasoning. Existing Test Time Scaling (TTS) methods, including structured tree search, have made progress in exploring reasoning paths but still face two major challenges: (1) underthinking, where reasoning chains tend to be shallow and fail to capture the full complexity of problems; and (2) overthinking, where overly verbose reasoning leads to inefficiency and increased computational costs. To address these issues, we propose LogitsCoder, a novel framework that enhances chain-of-thought reasoning through lightweight, logit-level control mechanisms for code generation. LogitsCoder iteratively generates and refines reasoning steps by first steering token selection toward statistically preferred patterns via Logits Preference Decoding, then selecting and aggregating diverse reasoning paths using Logits Rank Based Path Selection and Thoughts Aggregation. This results in coherent and effective reasoning chains that balance depth and efficiency. Extensive experiments demonstrate that LogitsCoder produces more efficient and higher-quality reasoning chains, leading to superior code generation performance compared to baseline methods.

cs.CL↗
arXiv2026

Plan-MCTS: Plan Exploration for Action Exploitation in Web Navigation

Large Language Models (LLMs) have empowered autonomous agents to handle complex web navigation tasks. While recent studies integrate tree search to enhance long-horizon reasoning, applying these algorithms in web navigation faces two critical challenges: sparse valid paths that lead to inefficient exploration, and a noisy context that dilutes accurate state perception. To address this, we introduce Plan-MCTS, a framework that reformulates web navigation by shifting exploration to a semantic Plan Space. By decoupling strategic planning from execution grounding, it transforms sparse action space into a Dense Plan Tree for efficient exploration, and distills noisy contexts into an Abstracted Semantic History for precise state awareness. To ensure efficiency and robustness, Plan-MCTS incorporates a Dual-Gating Reward to strictly validate both physical executability and strategic alignment and Structural Refinement for on-policy repair of failed subplans. Extensive experiments on WebArena demonstrate that Plan-MCTS achieves state-of-the-art performance, surpassing current approaches with higher task effectiveness and search efficiency.

cs.AI↗
arXiv2026

Adaptive Milestone Reward for GUI Agents

Reinforcement Learning (RL) has emerged as a mainstream paradigm for training Mobile GUI Agents, yet it struggles with the temporal credit assignment problem inherent in long-horizon tasks. A primary challenge lies in the trade-off between reward fidelity and density: outcome reward offers high fidelity but suffers from signal sparsity, while process reward provides dense supervision but remains prone to bias and reward hacking. To resolve this conflict, we propose the Adaptive Milestone Reward (ADMIRE) mechanism. ADMIRE constructs a verifiable, adaptive reward system by anchoring trajectory to milestones, which are dynamically distilled from successful explorations. Crucially, ADMIRE integrates an asymmetric credit assignment strategy that denoises successful trajectories and scaffolds failed trajectories. Extensive experiments demonstrate that ADMIRE consistently yields over 10% absolute improvement in success rate across different base models on AndroidWorld. Moreover, the method exhibits robust generalizability, achieving strong performance across diverse RL algorithms and heterogeneous environments such as web navigation and embodied tasks.

cs.LG↗
arXiv2026

Phase Transition With Rapini-Papoular Surface Anchoring

We analyze the dynamical (in)stability of nematic liquid crystals in the presence of external magnetic fields and Rapini-Papoular surface potential. The P-HAN transition is investigated using a simplified 3D Ericksen-Leslie system. We find the thickness threshold of the P-HAN transition. If the thickness of the nematic layer exceeds this threshold, there is a global-in-time suitable weak solution converging exponentially to a nontrivial equilibrium state as time tends to infinity. If the thickness is no more than the threshold, the global-in-time suitable weak solution has a trivial long-time asymptotic limit. Our results rigorously justify the P-HAN transition discussed in the physics literature.

math.AP↗
arXiv2026

Position: The Real Barrier to LLM Agent Usability is Agentic ROI

Large Language Model (LLM) agents represent a promising shift in human-AI interaction, moving beyond passive prompt-response systems to autonomous agents capable of reasoning, planning, and goal-directed action. While LLM agents are technically capable of performing a broad range of tasks, not all of these capabilities translate into meaningful usability. This position paper argues that the central question for LLM agent usability is no longer whether a task can be automated, but whether it delivers sufficient Agentic Return on Investment (Agentic ROI). Agentic ROI reframes evaluation from raw performance to a holistic, utility-driven perspective, guiding when, where, and for whom LLM agents should be deployed. Despite widespread application in high-ROI tasks like coding and scientific research, we identify a critical usability gap in mass-market, everyday applications. To address this, we propose a zigzag developmental trajectory: first scaling up to improve information gain and time savings, then scaling down to reduce cost. We present a strategic roadmap across these phases to make LLM agents truly usable, accessible, and scalable in real-world applications.

cs.CL↗
arXiv2026

ReMiT: RL-Guided Mid-Training for Iterative LLM Evolution

Standard training pipelines for large language models (LLMs) are typically unidirectional, progressing from pre-training to post-training. However, the potential for a bidirectional process--where insights from post-training retroactively improve the pre-trained foundation--remains unexplored. We aim to establish a self-reinforcing flywheel: a cycle in which reinforcement learning (RL)-tuned model strengthens the base model, which in turn enhances subsequent post-training performance, requiring no specially trained teacher or reference model. To realize this, we analyze training dynamics and identify the mid-training (annealing) phase as a critical turning point for model capabilities. This phase typically occurs at the end of pre-training, utilizing high-quality corpora under a rapidly decaying learning rate. Building upon this insight, we introduce ReMiT (Reinforcement Learning-Guided Mid-Training). Specifically, ReMiT leverages the reasoning priors of RL-tuned models to dynamically reweight tokens during the mid-training phase, prioritizing those pivotal for reasoning. Empirically, ReMiT achieves an average improvement of 3\% on 10 pre-training benchmarks, spanning math, code, and general reasoning, and sustains these gains by over 2\% throughout the post-training pipeline. These results validate an iterative feedback loop, enabling continuous and self-reinforcing evolution of LLMs.

cs.CL↗
arXiv2026

CATArena: Evaluating Evolutionary Capabilities of Code Agents via Iterative Tournaments

Current evaluation for Large Language Model (LLM) code agents predominantly focus on generating functional code in single-turn scenarios, which fails to evaluate the agent's capability for continuous code optimization and multi-turn iterative development. To bridge this gap, we introduce CATArena, a framework designed to evaluate the evolutionary capabilities of code agents via iterative tournaments. Agents engage in multi-turn tournaments and continuously refine their code through self-reflection and peer-learning based on comprehensive execution feedback. For evaluation, we propose a dual-metric system to decouple static generation proficiency from evolutionary potential. Extensive experiments reveal that an agent's evolutionary potential is not strictly correlated with its initial proficiency. Our analysis further reveals that current agents struggle to concurrently leverage both peer-learning and self-reflection for effective performance gains. Furthermore, the results validate CATArena's high extensibility and resistance to variance tasks, establishing it as a continuous and reliable standard for assessing the evolutionary capability of LLM code agents.

cs.AI↗
arXiv2026

Metasurface-assisted balanced-injection synchronization for turbulence-resilient long-haul chaotic free-space link

Optical chaotic synchronization between coupled nonlinear lasers underpins most chaos-based applications, including complex laser network dynamics, secure communication, key distribution, and reinforcement learning. In free-space links, however, chaotic synchronization is highly vulnerable to stochastic fluctuation induced by atmospheric turbulence, which results in temporal injection imbalances at symmetric receivers and triggers intermittent desynchronization. Here, we introduce a full Poincaré vector beam-enabled balanced-injection synchronization (BIS) mechanism, which passively mitigates coupling fluctuations and preserves injection symmetry through a complementary metasurface pair, without requiring any channel estimation or active control. Over a 3.2 km urban link under moderately strong turbulence, BIS suppresses coupling power fluctuations by a factor of 4.6 (from 0.4511 to 0.0975). It eliminates desynchronization events and increases the high-quality synchronization probability from 58.6% to 91.0%. This enables a record-high bit rate-distance product of 720 Gbps \cdot km, reducing communication interruption probability by up to 77% compared to Gaussian beam transmission. Our innovative strategy bridges the gap between nanophotonics and engineering optics, offering a new insight into advancing next-generation LiDAR, secure communication, and integrated sensing and communication systems in turbulent environments.

physics.optics↗
arXiv2026

2MASS Re-processing I: The Search for Faint Objects

We present an automated, DAOFind-based pipeline developed to reprocess J-band Atlas All Sky Release Survey Images from the Two Micron All Sky Survey (2MASS). By optimizing the detection parameters and implementing a screening procedure that jointly evaluates the signal-to-noise ratio and central sharpness, the pipeline effectively identifies faint point sources that were previously undetected. Applying this method to eight representative sky regions improves the 2MASS detection limit from 16.20 to 16.60 mag and increases the number of detected point sources by approximately 21.36% relative to the 2MASS Point Source Catalog, with a false-positive rate of only 4.80%. These results demonstrate that the proposed reprocessing pipeline can substantially enhance the scientific yield of archival 2MASS data, providing valuable faint-source supplements for studies of time-domain variability, Galactic structure, and cold, low-luminosity objects.

astro-ph.IM↗
arXiv2026

CoreCodeBench: Decoupling Code Intelligence via Fine-Grained Repository-Level Tasks

The evaluation of Large Language Models (LLMs) for software engineering has shifted towards complex, repository-level tasks. However, existing benchmarks predominantly rely on coarse-grained pass rates that treat programming proficiency as a monolithic capability, obscuring specific cognitive bottlenecks. Furthermore, the static nature of these benchmarks renders them vulnerable to data contamination and performance saturation. To address these limitations, we introduce CoreCodeBench, a configurable repository-level benchmark designed to dissect coding capabilities through atomized tasks. Leveraging our automated framework, CorePipe, we extract and transform Python repositories into a comprehensive suite of tasks that isolate distinct cognitive demands within identical code contexts. Unlike static evaluations, CoreCodeBench supports controllable difficulty scaling to prevent saturation and ensures superior data quality. It achieves a 78.55% validity yield, significantly surpassing the 31.7% retention rate of SWE-bench-Verified. Extensive experiments with state-of-the-art LLMs reveal a significant capability misalignment, evidenced by distinct ranking shifts across cognitive dimensions. This indicates that coding proficiency is non-monolithic, as strength in one aspect does not necessarily translate to others. These findings underscore the necessity of our fine-grained taxonomy in diagnosing model deficiencies and offer a sustainable, rigorous framework for evolving code intelligence. The code for CorePipe is available at https://github.com/AGI-Eval-Official/CoreCodeBench, and the data for CoreCodeBench can be accessed at https://huggingface.co/collections/tubehhh/corecodebench-68256d2faabf4b1610a08caa.

cs.SE↗