Source author record

Wenxuan Huang

Wenxuan Huang appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Computer Vision Artificial Intelligence cond-mat.dis-nn cond-mat.stat-mech physics.comp-ph eess.IV Human-Computer Interaction math-ph math.MP

Catalog footprint

What is connected

8works

9topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

Flow-OPD: On-Policy Distillation for Flow Matching Models

Existing Flow Matching (FM) text-to-image models suffer from two critical bottlenecks under multi-task alignment: the reward sparsity induced by scalar-valued rewards, and the gradient interference arising from jointly optimizing heterogeneous objectives, which together give rise to a 'seesaw effect' of competing metrics and pervasive reward hacking. Inspired by the success of On-Policy Distillation (OPD) in the large language model community, we propose Flow-OPD, the first unified post-training framework that integrates on-policy distillation into Flow Matching models. Flow-OPD adopts a two-stage alignment strategy: it first cultivates domain-specialized teacher models via single-reward GRPO fine-tuning, allowing each expert to reach its performance ceiling in isolation; it then establishes a robust initial policy through a Flow-based Cold-Start scheme and seamlessly consolidates heterogeneous expertise into a single student via a three-step orchestration of on-policy sampling, task-routing labeling, and dense trajectory-level supervision. We further introduce Manifold Anchor Regularization (MAR), which leverages a task-agnostic teacher to provide full-data supervision that anchors generation to a high-quality manifold, effectively mitigating the aesthetic degradation commonly observed in purely RL-driven alignment. Built upon Stable Diffusion 3.5 Medium, Flow-OPD raises the GenEval score from 63 to 92 and the OCR accuracy from 59 to 94, yielding an overall improvement of roughly 10 points over vanilla GRPO, while preserving image fidelity and human-preference alignment and exhibiting an emergent 'teacher-surpassing' effect. These results establish Flow-OPD as a scalable alignment paradigm for building generalist text-to-image models. The codes and weights will be released in: https://github.com/CostaliyA/Flow-OPD .

preprint2026arXiv

OpenSearch-VL: An Open Recipe for Frontier Multimodal Search Agents

Deep search has become a crucial capability for frontier multimodal agents, enabling models to solve complex questions through active search, evidence verification, and multi-step reasoning. Despite rapid progress, top-tier multimodal search agents remain difficult to reproduce, largely due to the absence of open high-quality training data, transparent trajectory synthesis pipelines, or detailed training recipes. To this end, we introduce OpenSearch-VL, a fully open-source recipe for training frontier multimodal deep search agents with agentic reinforcement learning. First, we curated a dedicated pipeline to construct high-quality training data through Wikipedia path sampling, fuzzy entity rewriting, and source-anchor visual grounding, which jointly reduce shortcuts and one-step retrieval collapse. Based on this pipeline, we curate two training datasets, SearchVL-SFT-36k for SFT and SearchVL-RL-8k for RL. Besides, we design a diverse tool environment that unifies text search, image search, OCR, cropping, sharpening, super-resolution, and perspective correction, enabling agents to combine active perception with external knowledge acquisition. Finally, we propose a multi-turn fatal-aware GRPO training algorithm that handles cascading tool failures by masking post-failure tokens while preserving useful pre-failure reasoning through one-sided advantage clamping. Built on this recipe, OpenSearch-VL delivers substantial performance gains, with over 10-point average improvements across seven benchmarks, and achieves results comparable to proprietary commercial models on several tasks. We will release all data, code, and models to support open research on multimodal deep search agents.

preprint2026arXiv

SCOPE: Structured Decomposition and Conditional Skill Orchestration for Complex Image Generation

While text-to-image models have made strong progress in visual fidelity, faithfully realizing complex visual intents remains challenging because many requirements must be tracked across grounding, generation, and verification. We refer to these requirements as semantic commitments and formalize their lifecycle discontinuity as the Conceptual Rift, where commitments may be locally resolved or checked but fail to remain identifiable as the same operational units throughout the generation lifecycle. To address this, we propose SCOPE, a specification-guided skill orchestration framework that maintains semantic commitments in an evolving structured specification and conditionally invokes retrieval, reasoning, and repair skills around unresolved or violated commitments. To evaluate commitment-level intent realization, we introduce Gen-Arena, a human-annotated benchmark with entity- and constraint-level specifications, together with Entity-Gated Intent Pass Rate (EGIP), a strict entity-first pass criterion. SCOPE substantially outperforms all evaluated baselines on Gen-Arena, achieving 0.60 EGIP, and further achieves strong results on WISE-V (0.907) and MindBench (0.61), demonstrating the effectiveness of persistent commitment tracking for complex image generation.

preprint2026arXiv

UniCorn: Towards Self-Improving Unified Multimodal Models through Self-Generated Supervision

While Unified Multimodal Models (UMMs) have achieved remarkable success in cross-modal comprehension, a significant gap persists in their ability to leverage such internal knowledge for high-quality generation. We formalize this discrepancy as Conduction Aphasia, a phenomenon where models accurately interpret multimodal inputs but struggle to translate that understanding into faithful and controllable synthesis. To address this, we propose UniCorn, a simple yet elegant self-improvement framework that eliminates the need for external data or teacher supervision. By partitioning a single UMM into three collaborative roles: Proposer, Solver, and Judge, UniCorn generates high-quality interactions via self-play and employs cognitive pattern reconstruction to distill latent understanding into explicit generative signals. To validate the restoration of multimodal coherence, we introduce UniCycle, a cycle-consistency benchmark based on a Text to Image to Text reconstruction loop. Extensive experiments demonstrate that UniCorn achieves comprehensive and substantial improvements over the base model across six general image generation benchmarks. Notably, it achieves SOTA performance on TIIF(73.8), DPG(86.8), CompBench(88.5), and UniCycle while further delivering substantial gains of +5.0 on WISE and +6.5 on OneIG. These results highlight that our method significantly enhances T2I generation while maintaining robust comprehension, demonstrating the scalability of fully self-supervised refinement for unified multimodal intelligence.

preprint2026arXiv

VideoSeeker: Incentivizing Instance-level Video Understanding via Native Agentic Tool Invocation

Large Vision-Language Models (LVLMs) have shown significant progress in video understanding, yet they face substantial challenges in tasks requiring precise spatiotemporal localization at the instance level. Existing methods primarily rely on text prompts for human-model interaction, but these prompts struggle to provide precise spatial and temporal references, resulting in poor user experience. Furthermore, current approaches typically decouple visual perception from language reasoning, centering reasoning around language rather than visual content, which limits the model's ability to proactively perceive fine-grained visual evidence. To address these challenges, we propose VideoSeeker, a novel paradigm for instance-level video understanding through visual prompts. VideoSeeker seamlessly integrates agentic reasoning with instance-level video understanding tasks, enabling the model to proactively perceive and retrieve relevant video segments on demand. We construct a four-stage fully automated data synthesis pipeline to efficiently generate large-scale, high-quality instance-level video data. We internalize tool-calling and proactive perception capabilities into the model via cold-start supervision and RL training, building a powerful video understanding model. Experiments demonstrate that our model achieves an average improvement of +13.7% over baselines on instance-level video understanding tasks, surpassing powerful closed-source models such as GPT-4o and Gemini-2.5-Pro, while also showing effective transferability on general video understanding benchmarks. The relevant datasets and code will be released publicly.

preprint2021arXiv

cGANs for Cartoon to Real-life Images

The image-to-image translation is a learning task to establish a visual mapping between an input and output image. The task has several variations differentiated based on the purpose of the translation, such as synthetic to real translation, photo to caricature translation, and many others. The problem has been tackled using different approaches, either through traditional computer vision methods, as well as deep learning approaches in recent trends. One approach currently deemed popular and effective is using the conditional generative adversarial network, also known shortly as cGAN. It is adapted to perform image-to-image translation tasks with typically two networks: a generator and a discriminator. This project aims to evaluate the robustness of the Pix2Pix model by applying the Pix2Pix model to datasets consisting of cartoonized images. Using the Pix2Pix model, it should be possible to train the network to generate real-life images from the cartoonized images.

preprint2016arXiv

Constructing and proving the ground state of a generalized Ising model by the cluster tree optimization algorithm

Generalized Ising models, also known as cluster expansions, are an important tool in many areas of condensed-matter physics and materials science, as they are often used in the study of lattice thermodynamics, solid-solid phase transitions, magnetic and thermal properties of solids, and fluid mechanics. However, the problem of finding the global ground state of generalized Ising model has remained unresolved, with only a limited number of results for simple systems known. We propose a method to efficiently find the periodic ground state of a generalized Ising model of arbitrary complexity by a new algorithm which we term cluster tree optimization. Importantly, we are able to show that even in the case of an aperiodic ground state, our algorithm produces a sequence of states with energy converging to the true ground state energy, with a provable bound on error. Compared to the current state-of-the-art polytope method, this algorithm eliminates the necessity of introducing an exponential number of variables to counter frustration, and thus significantly improves tractability. We believe that the cluster tree algorithm offers an intuitive and efficient approach to finding and proving ground states of generalized Ising Hamiltonians of arbitrary complexity, which will help validate assumptions regarding local vs. global optimality in lattice models, as well as offer insights into the low-energy behavior of highly frustrated systems.

preprint2016arXiv

Finding and proving the exact ground state of a generalized Ising model by convex optimization and MAX-SAT

Lattice models, also known as generalized Ising models or cluster expansions, are widely used in many areas of science and are routinely applied to alloy thermodynamics, solid-solid phase transitions, magnetic and thermal properties of solids, and fluid mechanics, among others. However, the problem of finding the true global ground state of a lattice model, which is essential for all of the aforementioned applications, has remained unresolved, with only a limited number of results for highly simplified systems known. In this article, we present the first general algorithm to find the exact ground states of complex lattice models and to prove their global optimality, resolving this fundamental problem in condensed matter and materials theory. We transform the infinite-discrete-optimization problem into a pair of combinatorial optimization (MAX-SAT) and non-smooth convex optimization (MAX-MIN) problems, which provide upper and lower bounds on the ground state energy respectively. By systematically converging these bounds to each other, we find and prove the exact ground state of realistic Hamiltonians whose solutions are completely intractable via traditional methods. Considering that currently such Hamiltonians are solved using simulated annealing and genetic algorithms that are often unable to find the true global energy minimum, and never able to prove the optimality of their result, our work opens the door to resolving long-standing uncertainties in lattice models of physical phenomena.

Institution

Affiliation not imported yet

This author record came from a source that does not expose affiliation metadata. Once the author claims the profile or we enrich the record from another provider, this section will link to the concrete institution.

Topic footprint