arXiv subjects

Zhuo Wang

Publications and source records attributed to Zhuo Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

FT-Dojo: Towards Autonomous LLM Fine-Tuning with Language Agents

Fine-tuning large language models for vertical domains remains labor-intensive, requiring practitioners to curate data, configure training, and iteratively diagnose model behavior. Despite growing interest in autonomous machine learning and language agents, end-to-end LLM fine-tuning has not been systematically studied as an interactive agent task. We introduce FT-Dojo, an interactive benchmark environment for autonomous LLM fine-tuning, comprising 13 tasks across 5 domains. Rather than a new collection of static datasets, FT-Dojo standardizes a task interface, shared raw-data repository, sandboxed execution environment, structured feedback protocol, and held-out evaluation procedure. We further develop FT-Agent, a fine-tuning-oriented autonomous framework that uses structured iteration planning, fail-fast validation, and multi-level feedback analysis to refine data and training strategies. Experiments show that FT-Agent provides a strong initial baseline, achieving the best performance on 10 out of 13 tasks, with additional controlled comparisons against frontier agents, open-source planning backbones, and multi-run statistics supporting the main findings. Case studies show that agents can recover from failures through cumulative learning, while still exposing limitations in causal diagnosis and long-horizon planning. The implementation is available at https://github.com/microsoft/rd-agent.

cs.AI↗

arXiv2026

Agent^2 RL-Bench: Can LLM Agents Engineer Agentic RL Post-Training?

We introduce Agent2 RL-Bench, a compact diagnostic benchmark for evaluating agentic RL post-training, which tests whether LLM agents can autonomously design, implement, debug, and execute post-training pipelines that improve foundation models. RL post-training increasingly drives model alignment and specialization, yet existing benchmarks are largely static, rewarding supervised fine-tuning or script generation without assessing an agent's ability to close an interactive RL loop. Agent2 RL-Bench provides a unified agent-facing interface: each run starts from an isolated workspace containing a base model, task data, instructions, and a grading API, and agents must iterate within a fixed budget by training models and submitting artifacts for evaluation. The benchmark spans six tasks across three levels, from static rule-based training to judge-based optimization and closed-loop online RL with trajectory collection. Two diagnostic skills, namely runtime recording and post-hoc summarization, enable structured analysis of agent behavior, facilitating smooth and effective iteration of the benchmark's evaluation framework. Across five agent systems and six driver LLMs, agents show intelligent behavior but clear limitations: one RL-oriented run improves ALFWorld from 4.85 to 93.28 via SFT warm-up and GRPO with online rollouts, yet DeepSearchQA remains difficult, most successful routes rely on supervised pipelines, and interactive outcomes show large single-run differences across agent stacks. Overall, Agent2 RL-Bench shows that current agents can sometimes engineer online RL, but stable agent-driven RL post-training remains rare under fixed budgets. It also demonstrates that our benchmark provides a strong and effective evaluation framework for future research in this direction. Code is available at https://github.com/microsoft/RD-Agent/blob/main/rdagent/scenarios/rl/autorl_bench/README.md

cs.AI↗

arXiv2026

Remember the Decision, Not the Description: A Rate-Distortion Framework for Agent Memory

Long-horizon language agents must operate under limited runtime memory, yet existing memory mechanisms often organize experience around descriptive criteria such as relevance, salience, or summary quality. For an agent, however, memory is valuable not because it faithfully describes the past, but because it preserves the distinctions between histories that must remain separated under a fixed budget to support good decisions. We cast this as a decision-centric rate-distortion problem, measuring memory quality by the loss in achievable decision quality induced by compression. This yields an exact forgetting boundary for what can be safely forgotten, and a memory-distortion frontier characterizing the optimal tradeoff between memory budget and decision quality. Motivated by this decision-centric view of memory, we propose DeMem, an online memory learner that refines its partition only when data certify that a shared state would induce decision conflict, and prove near-minimax regret guarantees. On both controlled synthetic diagnostics and long-horizon conversational benchmarks, DeMem yields consistent gains under the same runtime budget, supporting the principle that memory should preserve the distinctions that matter for decisions, not descriptions.

cs.AI↗

arXiv2026

RobustSora: De-Watermarked Benchmark for Robust AI-Generated Video Detection

The proliferation of AI-generated video models poses new challenges to information integrity and digital trust. A key confound, however, remains unaddressed: commercial generators embed visible overlay watermarks for provenance tracking, yet no existing benchmark controls for this variable, leaving open whether detectors learn genuine generation artefacts or merely associate watermark patterns with AI-generated labels. We present RobustSora, a benchmark of 6,500 manually verified videos in four categories: Authentic-Clean (A-C), Generated-Watermarked (G-W), Generated-DeWatermarked (G-DeW), and Authentic-Spoofed (A-S), sourced from Vript, DVF, and UltraVideo (authentic) and from Sora, Sora 2, Pika, Open-Sora 2, and KLing (generated). Two evaluation tasks isolate watermark effects: Task-I (Watermark Erasure Robustness) tests detection on watermark-removed AI videos; Task-II (Watermark Spoofing Robustness) measures false-alarm rates on authentic videos injected with fake watermarks. Across ten models spanning specialized detectors, transformer classifiers, and MLLMs, watermark manipulation induces accuracy changes of $-9.4$ to $+1.6$ pp (mean 6.6 pp; $p{<}0.01$ for 7/10 models on each task). A placebo control bounds inpainting-artefact confounds at $\le$2 pp, and a watermark-aware training augmentation recovers 3-4 pp on both tasks, together providing causal evidence that detectors actively rely on watermark cues. Per-generator breakdown shows that Sora 2 induces drops of $-11$ to $-14$ pp versus $-3$ to $-6$ pp for Pika and Open-Sora 2, indicating that watermark prominence, rather than detector architecture, is the principal driver of dependency. These results argue for watermark-aware evaluation and training in AIGC video detection. Dataset, evaluation code, and pretrained checkpoints will be released.

cs.CV↗

arXiv2026

CoTEvol: Self-Evolving Chain-of-Thoughts for Data Synthesis in Mathematical Reasoning

Large Language Models (LLMs) exhibit strong mathematical reasoning when trained on high-quality Chain-of-Thought (CoT) that articulates intermediate steps, yet costly CoT curation hinders further progress. While existing remedies such as distillation from stronger LLMs and self-synthesis based on test-time search alleviate this issue, they often suffer from diminishing returns or high computing overhead.In this work, we propose CoTEvol, a genetic evolutionary framework that casts CoT generation as a population-based search over reasoning trajectories.Candidate trajectories are iteratively evolved through reflective global crossover at the trajectory level and local mutation guided by uncertainty at the step level, enabling holistic recombination and fine-grained refinement. Lightweight, task-aware fitness functions are designed to guide the evolutionary process toward accurate and diverse reasoning. Empirically, CoTEvol improves correct-CoT synthesis success by over 30% and enhances structural diversity, with markedly improved efficiency. LLMs trained on these evolutionary CoT data achieve an average gain of 6.6% across eight math benchmarks, outperforming previous distillation and self-synthesis approaches. These results underscore the promise of evolutionary CoT synthesis as a scalable and effective method for mathematical reasoning tasks.

cs.AI↗

arXiv2026

A Data-Driven Regional Model for Skillful Medium-Range Typhoon Prediction

Accurate prediction of tropical cyclones remains a major challenge for both numerical weather prediction and emerging artificial intelligence weather prediction systems. While recent global AI models have demonstrated strong skill in large-scale circulation prediction, they often struggle to represent the mesoscale structures critical for tropical cyclone intensity and precipitation. Here we develop the Hybrid Intelligent Typhoon System (HITS), a regional AI forecasting framework for medium-range typhoon prediction over the Asia-Pacific region, trained on a newly constructed 9-km high-resolution typhoon reanalysis dataset. The model combines regional autoregressive prediction with large scale dynamical constraints from the state-of-the-art ECMWF Artificial Intelligence Forecasting System (AIFS), allowing it to remain dynamically consistent with the evolving large-scale circulation while resolving mesoscale structures. HITS is further extended with a structure-aware perceptual training strategy (HITS-LPIPS) that improves the representation of convective and typhoon rainband structures. Experiments show that the hybrid framework substantially improves precipitation structure and typhoon intensity forecasts compared with both purely autoregressive regional AI models and standalone AI downscaling approaches. In particular, HITS-LPIPS reduces intensity errors by up to 47.8% relative to AIFS at a 72 hour lead time and produces a near-unbiased wind-pressure relationship for simulated typhoons. These results demonstrate that dynamically constrained regional AI systems provide a promising pathway for improving medium-range typhoon prediction.

physics.ao-ph↗

arXiv2026

Insights from Ex-Typhoon Halong (2025) -- An Arctic Cyclone of Tropical Origin

An Arctic cyclone, Ex-Typhoon Halong, produced strong winds and devastating flooding in southwestern Alaska during 11-12 October 2025. This study examines the evolution of Halong after its transition into an extratropical cyclone through the analysis of ERA5 reanalysis and WRF model simulations. It is found that warm sea surface temperature (SST) anomalies over the western North Pacific preconditioned ex-Halong for intensification by increasing water-vapor content and reducing static stability. Quasi-geostrophic lifting associated with a subsequent interaction with another extratropical cyclone led to the rapid deepening of ex-Halong. This case demonstrates that tropical cyclones can transition into extratropical systems that are intensified by anomalously warm ocean waters, exacerbating impacts in high latitudes. Further analyses indicate that an increasing fraction of Alaskan cyclones has originated in tropical latitudes (south of 30°N) in recent decades. In particular, the frequency of Arctic cyclones of tropical origin increased by a factor of four in August and by a factor of three in September during 1980-2025 compared with 1940-1979.

physics.ao-ph↗

arXiv2026

Guideline Forest: Retrieval-Augmented Reasoning with Branching Experience-Induced Guidelines

Retrieval-augmented generation (RAG) has been widely adopted to ground large language models (LLMs) in external knowledge, yet it remains largely underexplored for improving reasoning. Existing methods either rely on online exploration during inference or heuristic supervision over reasoning trajectories, but they fail to effectively accumulate and reuse past reasoning experience. We propose Guideline Forest, a retrieval-augmented reasoning framework that explicitly leverages experience to guide multi-step reasoning. The framework stores high-quality, label-consistent reasoning traces as reusable memory, retrieves relevant experiences for new problems, and induces them into structured guidelines that steer reasoning and enable controlled branching and aggregation. Experiments on mathematical (GSM8K, MATH-500) and programming (MBPP, HumanEval) benchmarks demonstrate consistent improvements over strong reasoning baselines, including CoT, ReAct, ToT, FoT, and AFlow. Further analyses show that experience retrieval, guideline-induced diversity, and stepwise aggregation are key to the framework's effectiveness. Beyond single-model reasoning, Guideline Forest generalizes to enhance diverse reasoning paradigms and supports multi-model collaboration, highlighting its flexibility and scalability.

cs.AI↗

arXiv2026

Millimeter-scale rigid diamond probe for high sensitivity endoscopic-magnetometry applications

Magnetometry based on diamond nitrogen-vacancy (NV) centers has been extensively studied for applications requiring diverse capabilities, spanning from nanometer spatial resolution to subpicotesla sensitivity. Among various applications, diamond magnetometers can demonstrate high sensitivity magnetic sensing within millimeter-scale size for endoscopic applications. However, the trade-off between sensitivity and spatial resolution of diamond magnetometry makes it difficult to achieve such a probe. In this study, we present a millimeter-scale rigid diamond magnetometer probe with enhanced sensitivity via optimizing the optical design. By coupling the frustum diamond with the miniaturized compound parabolic concentrator (CPC) lens, we enhance the fluorescence collection efficiency by 37% within 4 mm diameter, and the achieved sensitivity is 200 pT/Hz1/2 based on the sample with the resonance linewidth of ~8 MHz. With this verified structure, endoscopes with mm-size probe and picotesla sensitivity can be projected for surgical and industrial applications in the future.

physics.ins-det↗

arXiv2026

High-Resolution Underwater Camouflaged Object Detection: GBU-UCOD Dataset and Topology-Aware and Frequency-Decoupled Networks

Underwater Camouflaged Object Detection (UCOD) is a challenging task due to the extreme visual similarity between targets and backgrounds across varying marine depths. Existing methods often struggle with topological fragmentation of slender creatures in the deep sea and the subtle feature extraction of transparent organisms. In this paper, we propose DeepTopo-Net, a novel framework that integrates topology-aware modeling with frequency-decoupled perception. To address physical degradation, we design the Water-Conditioned Adaptive Perceptor (WCAP), which employs Riemannian metric tensors to dynamically deform convolutional sampling fields. Furthermore, the Abyssal-Topology Refinement Module (ATRM) is developed to maintain the structural connectivity of spindly targets through skeletal priors. Specifically, we first introduce GBU-UCOD, the first high-resolution (2K) benchmark tailored for marine vertical zonation, filling the data gap for hadal and abyssal zones. Extensive experiments on MAS3K, RMAS, and our proposed GBU-UCOD datasets demonstrate that DeepTopo-Net achieves state-of-the-art performance, particularly in preserving the morphological integrity of complex underwater patterns. The datasets and codes will be released at https://github.com/Wuwenji18/GBU-UCOD.

cs.CV↗

arXiv2025

LoongFlow: Directed Evolutionary Search via a Cognitive Plan-Execute-Summarize Paradigm

The transition from static Large Language Models (LLMs) to self-improving agents is hindered by the lack of structured reasoning in traditional evolutionary approaches. Existing methods often struggle with premature convergence and inefficient exploration in high-dimensional code spaces. To address these challenges, we introduce LoongFlow, a self-evolving agent framework that achieves state-of-the-art solution quality with significantly reduced computational costs. Unlike "blind" mutation operators, LoongFlow integrates LLMs into a cognitive "Plan-Execute-Summarize" (PES) paradigm, effectively mapping the evolutionary search to a reasoning-heavy process. To sustain long-term architectural coherence, we incorporate a hybrid evolutionary memory system. By synergizing Multi-Island models with MAP-Elites and adaptive Boltzmann selection, this system theoretically balances the exploration-exploitation trade-off, maintaining diverse behavioral niches to prevent optimization stagnation. We instantiate LoongFlow with a General Agent for algorithmic discovery and an ML Agent for pipeline optimization. Extensive evaluations on the AlphaEvolve benchmark and Kaggle competitions demonstrate that LoongFlow outperforms leading baselines (e.g., OpenEvolve, ShinkaEvolve) by up to 60% in evolutionary efficiency while discovering superior solutions. LoongFlow marks a substantial step forward in autonomous scientific discovery, enabling the generation of expert-level solutions with reduced computational overhead.

cs.AI↗

arXiv2025

Toward Patch Robustness Certification and Detection for Deep Learning Systems Beyond Consistent Samples

Patch robustness certification is an emerging kind of provable defense technique against adversarial patch attacks for deep learning systems. Certified detection ensures the detection of all patched harmful versions of certified samples, which mitigates the failures of empirical defense techniques that could (easily) be compromised. However, existing certified detection methods are ineffective in certifying samples that are misclassified or whose mutants are inconsistently pre icted to different labels. This paper proposes HiCert, a novel masking-based certified detection technique. By focusing on the problem of mutants predicted with a label different from the true label with our formal analysis, HiCert formulates a novel formal relation between harmful samples generated by identified loopholes and their benign counterparts. By checking the bound of the maximum confidence among these potentially harmful (i.e., inconsistent) mutants of each benign sample, HiCert ensures that each harmful sample either has the minimum confidence among mutants that are predicted the same as the harmful sample itself below this bound, or has at least one mutant predicted with a label different from the harmful sample itself, formulated after two novel insights. As such, HiCert systematically certifies those inconsistent samples and consistent samples to a large extent. To our knowledge, HiCert is the first work capable of providing such a comprehensive patch robustness certification for certified detection. Our experiments show the high effectiveness of HiCert with a new state-of the-art performance: It certifies significantly more benign samples, including those inconsistent and consistent, and achieves significantly higher accuracy on those samples without warnings and a significantly lower false silent ratio.

cs.SE↗

arXiv2025

Kondo-coupled van der Waals antiferromagnet with high-mobility quasiparticles

Two-dimensional van der Waals (vdW) materials exhibit high carrier mobility and tunability, making them suitable for low-power, high-performance electronic and spintronic applications. Incorporating narrow-band electronic correlation effects could further promote tunability, though mass renormalization may impact carrier mobility. It is therefore challenging to identify a vdW material with both high mobility and strong correlation. Herein, by a combination of optical spectroscopy and high-field quantum-oscillation measurements, we observe significant effective-mass enhancement in CeTe$_3$ at low temperature, arising from not only the band-structure modulation by antiferromagnetic ordering but also the narrow-band correlation effect. Despite the mass enhancement, the quantum mobility surprisingly \textit{increases} and reaches $\sim$2403 cm$^2$/Vs, likely benefiting from topological protection. Remarkably, these unique properties are maintained in atomically thin nanoflakes with quantum mobility enhanced to $\sim$3158 cm$^2$/Vs. Thus, CeTe$_3$ emerges as a promising Kondo-coupled vdW antiferromagnetic metal with high-mobility quasiparticles, potentially unlocking new device concepts.

cond-mat.str-el↗

arXiv2025

AnyEnhance: A Unified Generative Model with Prompt-Guidance and Self-Critic for Voice Enhancement

We introduce AnyEnhance, a unified generative model for voice enhancement that processes both speech and singing voices. Based on a masked generative model, AnyEnhance is capable of handling both speech and singing voices, supporting a wide range of enhancement tasks including denoising, dereverberation, declipping, super-resolution, and target speaker extraction, all simultaneously and without fine-tuning. AnyEnhance introduces a prompt-guidance mechanism for in-context learning, which allows the model to natively accept a reference speaker's timbre. In this way, it could boost enhancement performance when a reference audio is available and enable the target speaker extraction task without altering the underlying architecture. Moreover, we also introduce a self-critic mechanism into the generative process for masked generative models, yielding higher-quality outputs through iterative self-assessment and refinement. Extensive experiments on various enhancement tasks demonstrate AnyEnhance outperforms existing methods in terms of both objective metrics and subjective listening tests. Demo audios are publicly available at https://amphionspace.github.io/anyenhance. An open-source implementation is provided at https://github.com/viewfinder-annn/anyenhance-v1-ccf-aatc.

cs.SD↗

arXiv2025

FinHEAR: Human Expertise and Adaptive Risk-Aware Temporal Reasoning for Financial Decision-Making

Financial decision-making presents unique challenges for language models, demanding temporal reasoning, adaptive risk assessment, and responsiveness to dynamic events. While large language models (LLMs) show strong general reasoning capabilities, they often fail to capture behavioral patterns central to human financial decisions-such as expert reliance under information asymmetry, loss-averse sensitivity, and feedback-driven temporal adjustment. We propose FinHEAR, a multi-agent framework for Human Expertise and Adaptive Risk-aware reasoning. FinHEAR orchestrates specialized LLM-based agents to analyze historical trends, interpret current events, and retrieve expert-informed precedents within an event-centric pipeline. Grounded in behavioral economics, it incorporates expert-guided retrieval, confidence-adjusted position sizing, and outcome-based refinement to enhance interpretability and robustness. Empirical results on curated financial datasets show that FinHEAR consistently outperforms strong baselines across trend prediction and trading tasks, achieving higher accuracy and better risk-adjusted returns.

cs.LG↗

arXiv2025

Rich unconventional Hall effects in a single quasi-kagome Kondo Weyl semimetal candidate Ce$_3$TiSb$_5$

It is generally believed that electronic correlation, geometric frustration, and topology, \textit{individually}, can facilitate the emergence of various intriguing properties that have attracted a broad audience for both fundamental research and potential applications. Here, we report a series of unconventional Hall effects observed in a \textit{single} compound - quasi-kagome Kondo Weyl semimetal candidate Ce$_3$TiSb$_5$. In the paramagnetic phase, signature of dynamic $c$-$f$ hybridization is revealed by a reduction of anomalous Hall effect and is connected to frustration-promoted incoherent Kondo scattering. A large topological Hall effect exceeding 0.2 $μΩ\cdot$cm is found at low temperatures, which should be ascribed to the non-collinear magnetic texture. In addition, a peculiar loop-shaped Hall effect with switching chirality is also seen, which is inferred to be associated with magnetic domain walls that pin history-dependent spin chirality and / or Fermi-arc surface states projected from the in-gap Weyl nodes. These exotic results place Ce$_3$TiSb$_5$ in a regime of highly-frustrated antiferromagnetic dense Kondo lattice with a nontrivial topology on an ``extended" global phase diagram, and highlight the interplay among electronic correlation, geometric frustration and topology.

cond-mat.str-el↗

arXiv2025

From Implicit Exploration to Structured Reasoning: Leveraging Guideline and Refinement for LLMs

Large language models (LLMs) have advanced general-purpose reasoning, showing strong performance across diverse tasks. However, existing methods often rely on implicit exploration, where the model follows stochastic and unguided reasoning paths-like walking without a map. This leads to unstable reasoning paths, lack of error correction, and limited learning from past experience. To address these issues, we propose a framework that shifts from implicit exploration to structured reasoning through guideline and refinement. First, we extract structured reasoning patterns from successful trajectories and reflective signals from failures. During inference, the model follows these guidelines step-by-step, with refinement applied after each step to correct errors and stabilize the reasoning process. Experiments on BBH and four additional benchmarks (GSM8K, MATH-500, MBPP, HumanEval) show that our method consistently outperforms strong baselines across diverse reasoning tasks. Structured reasoning with stepwise execution and refinement improves stability and generalization, while guidelines transfer well across domains and flexibly support cross-model collaboration, matching or surpassing supervised fine-tuning in effectiveness and scalability.

cs.AI↗

arXiv2025

Iterative Distortion Cancellation Algorithms for Single-Sideband Systems

We propose an iterative distortion cancellation algorithm to digitally mitigate the impact of double-sideband dither signal amplitude from the automatic bias control module on Kramers-Kronig receivers without modifying physical layer structures. The algorithm utilizes the KK relation for initial signal decisions and reconstructs the distortion caused by dither signals. Experimental tests in back-to-back showed it improved tolerance to dither amplitudes up to 10% Vπ. For 80-km fiber transmission, the algorithm increased the receiver sensitivity by more than 1 dB, confirming the effectiveness of the proposed distortion cancellation method.

eess.SP↗