arXiv subjects

Jiang Zhou

Publications and source records attributed to Jiang Zhou.

At least 19 recordsLinked to original sources

arXiv2026

Environment Evolution for Terminal Agents

Scaling interactive and verifiable environments is critical for training terminal agents. As frontier models become more capable, environments synthesized from scratch become less challenging and thus provide limited learning signals. Recent co-evolution methods iteratively synthesize environments near the model's learnable frontier based on weaknesses exposed during rollouts. However, their dependence on on-policy rollouts limits generalization and the continuous provision of learning signals as the model becomes stronger. In this paper, we propose environment evolution, which incrementally increases environment difficulty off-policy and schedules the evolved environments generation by generation during training to provide continuous learning signals. We derive three evolution directions that influence environment difficulty from the multi-turn learning objective and then implement evolution along these directions through a loop-engineered multi-agent harness. Quantitative rollout experiments with Hy4 preview, Claude Opus 5, and GPT-5.6 Sol show that environment evolution consistently produces more difficult environments. We validate its effectiveness on Qwen3.6-27B and Qwen3.6-35B-A3B through simple long-horizon RL training, improving their performance by 14.4 and 18.0 percentage points on Terminal-Bench 2.1, respectively.

cs.AI↗

arXiv2026

Commutators of a class of $\Psi$-bounded oscillation operators on abstract weighted Orlicz-Morrey spaces endowed with ball-basis and applications

In our previous work \cite{SZ2026}, we established a class of abstract Orlicz-Morrey spaces endowed with a ball-basis and introduced the notion of $\Psi$-bounded oscillation operators. The present paper further develops this theory by investigating sparse domination and weighted estimates for such operators and their commutators within the framework of abstract Orlicz-Morrey spaces endowed with a ball-basis. We obtain pointwise sparse domination results for the operators and establish their boundedness on weighted Orlicz-Morrey spaces. The hypotheses of these conclusions differ from those in the classical Orlicz-Morrey setting and do not rely on dyadic decompositions of Euclidean structure. Moreover, we study the duality of these spaces and prove that the dual of the associated block space is exactly the Orlicz-Morrey space under consideration.

math.FA↗

arXiv2026

Deep Research Pretraining via Predictive Navigation

Deep research agents are often trained on expensive, environment-grounded tool-use trajectories that require repeated retrieval, document inspection, and report evaluation. We introduce Deep Research Pretraining (DRP), an offline framework that derives predictive navigation supervision from naturally occurring evidence structures. Given a citation-bearing or hyperlinked passage, DRP constructs a proxy research objective, recovers linked evidence and graph-related alternatives, and converts them into search-open-write trajectories. This teaches models what to search for, which documents to inspect, and how to synthesize evidence, without a live retrieval environment or executed policy rollout. We instantiate DRP on scholarly citation graphs (DRP-Paper) and Wikipedia hyperlinks (DRP-Web), continually pretrain separate Qwen3-14B-Base models on 1B tokens, and fine-tune them on controlled fractions of 13K agent trajectories. Across five independently sampled subsets at each low-data budget, both variants consistently outperform matched no-DRP models on DeepResearch Bench. With one quarter of the SFT data, DRP-Web even surpasses a fixed no-DRP full-data checkpoint, with gains transferring to ResearchQA, WebWalkerQA, and SimpleQA. Starting from matched low-data SFT checkpoints, the DRP-Web advantage also persists through subsequent agentic RL. Source-matched and evidence-mismatch controls indicate that these improvements arise from evidence-conditioned navigation rather than domain exposure or agent-format imitation. DRP thus provides a promising complementary approach to trajectory-based agent training.

cs.CL↗

arXiv2026

Abstract Measure Grand Lebesgue Spaces and Applications

We introduce abstract measure grand Lebesgue spaces endowed with ball basis structures and investigate their fundamental properties, together with the behavior of $\mathfrak{BO}$ operators. By exploiting sparse domination for $\mathfrak{BO}$ operators and norm inequalities for sparse operators, we derive norm estimates for this class of operators. As applications, we establish that $\mathfrak{BO}$ operators encompass the maximal operators, Calder\'on-Zygmund operators on homogeneous spaces and Carleson operators.

math.FA↗

arXiv2026

SARA: Unlocking Multilingual Knowledge in Mixture-of-Experts via Semantically Anchored Routing Alignment

Sparse Mixture-of-Experts (MoE) architectures have emerged as an increasingly influential paradigm as they offer a strategic balance between parameter scalability and computational efficiency. However, low-resource languages, which suffer from a scarcity of high-quality training data, often have their tokens routed to different experts than those predominantly activated by high-resource inputs, which limits cross-lingual expert sharing. This cross-lingual routing divergence consequently hinders their efficacy in multilingual contexts. To address this issue, we propose SARA (Semantically Anchored Routing Alignment), a framework designed to transfer specialized capabilities from high-resource languages as anchors to low-resource languages. SARA explicitly aligns the routing distribution of multilingual inputs with high-resource semantic anchors using a symmetric Jensen-Shannon (JS) divergence constraint. Unlike traditional distillation methods that operate on output logits, SARA directly aligns the internal routing distributions of MoE layers, encouraging mechanistic consistency in expert selection across languages. We conduct experiments on 2 LLMs across 5 low-resource languages and 3 benchmarks. Experiment results demonstrate that SARA outperforms standard instruction tuning, e.g., +0.8% on Qwen3-30B-A3B and +1.2% on Phi-3.5-MoE-instruct on Global-MMLU. Further analyses show that SARA effectively addresses performance bottlenecks in low-resource languages, providing a scalable pathway to enhance multilingual capabilities in sparse architectures.

cs.CL↗

arXiv2026

Spectral bounds for distance coloring and packing parameters of graphs via semidefinite programming

Using methods from spectral graph theory and semidefinite programming, we obtain sharp spectral bounds for several graph parameters related to distance colorings and packing, including the injective chromatic number, the open packing number, the injective chromatic index, and the strong chromatic index. The new spectral bounds improve several existing combinatorial bounds. Furthermore, we apply the obtained eigenvalue bounds on the first two parameters to estimate the code covering number and the open packing number of hypercubes, obtaining new exact values and strengthened bounds regarding the existing literature. The obtained results illustrate the power of combining spectral and semidefinite programming tools for tackling coloring and packing problems in graph theory and coding theory.

math.CO↗

arXiv2026

Toward Scalable Terminal Task Synthesis via Skill Graphs

Terminal agents have demonstrated strong potential for autonomous command-line execution, yet their training remains constrained by the scarcity of high-quality and diverse execution trajectories. Existing approaches mitigate this bottleneck by synthesizing large-scale terminal task instances for trajectory sampling. However, they primarily focus on scaling the number of tasks while providing limited control over the diversity of execution trajectories that agents actually experience during training. In this paper, we present SkillSynth, an automated framework for terminal task synthesis built on a scenario-mediated skill graph. SkillSynth first constructs a large-scale skill graph, where scenarios serve as intermediate transition nodes that connect diverse command-line skills. It then samples paths from this graph as abstractions of real-world workflows, and uses a multi-agent harness to instantiate them into executable task instances. By grounding task synthesis in graph-sampled workflow paths, SkillSynth explicitly controls the diversity of minimal execution trajectories required to solve the synthesized tasks. Experiments on Terminal-Bench demonstrate the effectiveness of SkillSynth. Moreover, task instances synthesized by SkillSynth have been adopted to train Hy3 Preview, contributing to its enhanced agentic capabilities in terminal-based settings.

cs.AI↗

arXiv2026

Fractals of Simple Random Walks in Two Dimensions: A Monte Carlo Study

We present a Monte Carlo study of the fractal geometry of clusters formed by discrete-time simple random walks (sRW) of $L^2$ steps on a periodic square $L\times L$ lattice. We verify with high precision that the asymptotic behavior of the cluster mass follows $M/L^2 \simeq (\ln L)^{-1} [\fracπ{2}+b (\ln L)^{-2}]$, with $b\approx -(π/2)^{-2}$, demonstrating marginal ``logarithmic fractals". We further determine the fractal dimension of the hull to be $d_{\rm hull}=1.333\,29(14)=4/3$, in excellent agreement with the prediction of Schramm-Loewner evolution ($\rm SLE_{8/3}$) for the Brownian frontier universality class. More importantly, we analyze the chemical distance $S$ spanning the cluster and obtain strong evidence that it asymptotically scales as $S\sim L(\ln L)^{1/4}$, lying exactly on the theoretical upper bound for the chemical distance for level-set percolation clusters on the two-dimensional Gaussian free field. Our numerical results show that the sRW cluster exhibits a conformally invariant external frontier and contains highly efficient asymptotically linear connective paths.

cond-mat.stat-mech↗

arXiv2026

Incentivizing Parametric Knowledge via Reinforcement Learning with Verifiable Rewards for Cross-Cultural Entity Translation

Cross-cultural entity translation remains challenging for large language models (LLMs) as literal or phonetic renderings are usually yielded instead of culturally appropriate translations in context. However, relevant knowledge may already be encoded in model parameters during large-scale pre-training. To incentivize the effective use of parametric knowledge, we propose EA-RLVR (Entity-Anchored Reinforcement Learning with Verifiable Rewards), a training framework that optimizes cross-cultural entity translation without relying on external knowledge bases. EA-RLVR anchors supervision on a verifiable, entity-level reward signal and incorporates lightweight structural gates to stabilize optimization. This design steers the model toward learning a robust reasoning process rather than merely imitating reference translations. We evaluate EA-RLVR on XC-Translate and observe consistent improvements in both entity translation accuracy and out-of-domain generalization. Specifically, training on merely 7k samples boosts Qwen3-14B's entity translation accuracy from 23.66\% to 31.87\% on a 50k test set comprising entirely unseen entities. The learned entity translation ability also transfers to general translation, yielding +1.35 XCOMET on WMT24++, which scales to +1.59 with extended optimization. Extensive analyses of $pass@k$ dynamics and reward formulations attribute these gains to superior sampling efficiency and a stable optimization landscape.

cs.CL↗

arXiv2026

WRAP++: Web discoveRy Amplified Pretraining

Synthetic data rephrasing has emerged as a powerful technique for enhancing knowledge acquisition during large language model (LLM) pretraining. However, existing approaches operate at the single-document level, rewriting individual web pages in isolation. This confines synthesized examples to intra-document knowledge, missing cross-document relationships and leaving facts with limited associative context. We propose WRAP++ (Web discoveRy Amplified Pretraining), which amplifies the associative context of factual knowledge by discovering cross-document relationships from web hyperlinks and synthesizing joint QA over each discovered document pair. Concretely, WRAP++ discovers high-confidence relational motifs including dual-links and co-mentions, and synthesizes QA that requires reasoning across both documents. This produces relational knowledge absent from either source document alone, creating diverse entry points to the same facts. Because the number of valid entity pairs grows combinatorially, this discovery-driven synthesis also amplifies data scale far beyond single-document rewriting. Instantiating WRAP++ on Wikipedia, we amplify ~8.4B tokens of raw text into 80B tokens of cross-document QA data. On SimpleQA, OLMo-based models at both 7B and 32B scales trained with WRAP++ substantially outperform single-document approaches and exhibit sustained scaling gains, underscoring the advantage of cross-document knowledge discovery and amplification.

cs.CL↗

arXiv2026

PolicyLong: Towards On-Policy Context Extension

Extending LLM context windows is hindered by scarce high-quality long-context data. Recent methods synthesize data with genuine long-range dependencies via information-theoretic verification, selecting contexts that reduce a base model's predictive entropy. However, their single-pass offline construction with a fixed model creates a fundamental off-policy gap: the static screening landscape misaligns with the model's evolving capabilities, causing the training distribution to drift. We propose PolicyLong, shifting data construction towards a dynamic on-policy paradigm. By iteratively re-executing data screening (entropy computation, retrieval, and verification) using the current model, PolicyLong ensures the training distribution tracks evolving capabilities, yielding an emergent self-curriculum. Crucially, both positive and hard negative contexts derive from the current model's entropy landscape, co-evolving what the model learns to exploit and resist. Experiments on RULER, HELMET, and LongBench-v2 (Qwen2.5-3B) show PolicyLong consistently outperforms EntropyLong and NExtLong, with gains growing at longer contexts (e.g., +2.54 at 128K on RULER), confirming the value of on-policy data evolution.

cs.LG↗

arXiv2026

Spectral bounds for the independence number of graphs and even uniform hypergraphs

In this paper, we give spectral upper bounds for the independence number of even uniform hypergraphs and graphs, extend the Hoffman bound to even uniform hypergraphs, and give a simple spectral condition for determining the independence number, the Shannon capacity and the Lovász number of a graph. The Hoffman bound on the Lovász number is also extended from regular graphs to general graphs.

math.CO↗

arXiv2026

Near-Light Color Photometric Stereo for Mono-Chromatic Non-Lambertian Surfaces

Color photometric stereo enables single-shot surface reconstruction, extending conventional photometric stereo that requires multiple images of a static scene under varying illumination to dynamic scenarios. However, most existing approaches assume ideal distant lighting and Lambertian reflectance, leaving more practical near-light conditions and non-Lambertian surfaces underexplored. To overcome this limitation, we propose a framework that leverages neural implicit representations for depth and BRDF modeling under the assumption of mono-chromaticity (uniform chromaticity and homogeneous material), which alleviates the inherent ill-posedness of color photometric stereo and allows for detailed surface recovery from just one image. Furthermore, we design a compact optical tactile sensor to validate our approach. Experiments on both synthetic and real-world datasets demonstrate that our method achieves accurate and robust surface reconstruction.

cs.CV↗

arXiv2026

Backdoor Sentinel: Detecting and Detoxifying Backdoors in Diffusion Models via Temporal Noise Consistency

Diffusion models have been widely deployed in AIGC services, but their reliance on opaque training data exposes them to backdoor attacks. In practical auditing scenarios, auditors are typically unable to access model parameters due to intellectual property protection, making white-box or query-intensive detection impractical. After detection, existing detoxification approaches are trapped in a dilemma between detoxification effectiveness and generation quality for service providers. We reveal Temporal Noise Consistency (TNC), a previously unreported phenomenon in which backdoor activation disrupts the consistency of noise predictions between adjacent diffusion timesteps within specific temporal segments, while clean inputs remain stable. Based on this finding, we propose TNC-Defense, a closed-loop framework for gray-box backdoor detection and model repair. Specifically, TNC-Detect (for auditors) uses inference-stage adjacent-noise statistics to detect backdoors and precisely localize anomalous timesteps without model-weight access. TNC-Detox (for service providers) utilizes these locations to perform trigger-agnostic, timestep-aware correction of the generation path, suppressing backdoor behavior while reducing detoxification cost. Across five representative backdoor attacks and state-of-the-art defenses, TNC-Defense improves the average detection accuracy by $11\%$ with negligible additional overhead, and invalidates an average of $98.5\%$ of triggered samples with only a mild degradation in generation quality. Our code is publicly available at: https://github.com/binzhwang/TNC-Defense.

cs.CR↗

arXiv2025

Spectral characterizations of local structures of graphs and hypergraphs

In this paper, we give the relationship between spectral radius and local structures of graphs and hypergraphs. Our work shows that certain local subgraphs (subhypergraphs) must occur when the spectral radius ratio is large. We also give spectral bounds on the local vector chromatic number in terms of tensor eigenvalues of graphs.

math.CO↗

arXiv2025

Challenging Multilingual LLMs: A New Taxonomy and Benchmark for Unraveling Hallucination in Translation

Large Language Models (LLMs) have advanced machine translation but remain vulnerable to hallucinations. Unfortunately, existing MT benchmarks are not capable of exposing failures in multilingual LLMs. To disclose hallucination in multilingual LLMs, we introduce a diagnostic framework with a taxonomy that separates Instruction Detachment from Source Detachment. Guided by this taxonomy, we create HalloMTBench, a multilingual, human-verified benchmark across 11 English-to-X directions. We employed 4 frontier LLMs to generate candidates and scrutinize these candidates with an ensemble of LLM judges, and expert validation. In this way, we curate 5,435 high-quality instances. We have evaluated 17 LLMs on HalloMTBench. Results reveal distinct ``hallucination triggers'' -- unique failure patterns reflecting model scale, source length sensitivity, linguistic biases, and Reinforcement-Learning (RL) amplified language mixing. HalloMTBench offers a forward-looking testbed for diagnosing LLM translation failures. HalloMTBench is available in https://huggingface.co/collections/AIDC-AI/marco-mt.

cs.CL↗

arXiv2025

$M^2PO$: Multi-Perspective Multi-Pair Preference Optimization for Machine Translation

Aligning Large Language Models (LLMs) with human preferences is pivotal for Machine Translation (MT), yet current approaches are often hindered by misleading reward signals. Our analysis reveals that prevailing Quality Estimation (QE) models exhibit a systematic blind spot toward partial errors, specifically partial hallucinations and omissions, often favoring superficially fluent but unfaithful translations. To address this issue, we propose $M^2PO$ (Multi-Perspective Multi-Pair Preference Optimization), a data-centric framework for preference optimization in machine translation. First, to correct the bias toward fluency, $M^2PO$ uses a dual-perspective mechanism that decouples semantic fidelity from fluency and prioritizes faithfulness through a curriculum strategy. Second, after correcting this bias, partial errors fall between perfect and severely incorrect translations, making them difficult to learn through standard best-versus-worst comparisons. We therefore introduce a multi-pair objective that leverages the full candidate list to capture these fine-grained error signals. Experiments on WMT23, WMT24, and FLORES-200 show that $M^2PO$ enables a 9B model to outperform leading open-source baselines and achieve parity with proprietary models such as GPT-4o and Gemini-2.0-Flash, demonstrating strong potential for efficient and high-fidelity LLM-based translation. Our code and dataset will be released.

cs.CL↗

arXiv2025

Spherical two-distance sets and graph eigenvalues

A set of unit vectors in $\mathbb{R}^d$ is a called a spherical two-distance set if the inner products of distinct vectors only take two values. In this paper, we give explicit correspondence between spherical two-distance sets and graphs with specific spectral properties, and derive some bounds on the maximum size of spherical two-distance sets.

math.CO↗