arXiv subjects

Jie Jiang

Publications and source records attributed to Jie Jiang.

At least 91 records · Page 5Linked to original sources

arXiv2026

ProAct: Agentic Lookahead in Interactive Environments

Existing Large Language Model (LLM) agents struggle in interactive environments requiring long-horizon planning, primarily due to compounding errors when simulating future states. To address this, we propose ProAct, a framework that enables agents to internalize accurate lookahead reasoning through a two-stage training paradigm. First, we introduce Grounded LookAhead Distillation (GLAD), where the agent undergoes supervised fine-tuning on trajectories derived from environment-based search. By compressing complex search trees into concise, causal reasoning chains, the agent learns the logic of foresight without the computational overhead of inference-time search. Second, to further refine decision accuracy, we propose the Monte-Carlo Critic (MC-Critic), a plug-and-play auxiliary value estimator designed to enhance policy-gradient algorithms like PPO and GRPO. By leveraging lightweight environment rollouts to calibrate value estimates, MC-Critic provides a low-variance signal that facilitates stable policy optimization without relying on expensive model-based value approximation. Experiments on both stochastic (e.g., 2048) and deterministic (e.g., Sokoban) environments demonstrate that ProAct significantly improves planning accuracy. Notably, a 4B parameter model trained with ProAct outperforms all open-source baselines and rivals state-of-the-art closed-source models, while demonstrating robust generalization to unseen environments. The codes and models are available at https://github.com/GreatX3/ProAct

cs.AI↗

arXiv2026

Adaptive Multi-Grade Deep Learning for Highly Oscillatory Fredholm Integral Equations of the Second Kind

This paper studies the use of Multi-Grade Deep Learning (MGDL) for solving highly oscillatory Fredholm integral equations of the second kind. We provide rigorous error analyses of continuous and discrete MGDL models, showing that the discrete model retains the convergence and stability of its continuous counterpart under sufficiently small quadrature error. We identify the DNN training error as the primary source of approximation error, motivating a novel adaptive MGDL algorithm that selects the network grade based on training performance. Numerical experiments with highly oscillatory (including wavenumber 500) and singular solutions confirm the accuracy, effectiveness and robustness of the proposed approach.

math.NA↗

arXiv2026

AT$^2$PO: Agentic Turn-based Policy Optimization via Tree Search

LLM agents have emerged as powerful systems for tackling multi-turn tasks by interleaving internal reasoning and external tool interactions. Agentic Reinforcement Learning has recently drawn significant research attention as a critical post-training paradigm to further refine these capabilities. In this paper, we present AT$^2$PO (Agentic Turn-based Policy Optimization via Tree Search), a unified framework for multi-turn agentic RL that addresses three core challenges: limited exploration diversity, sparse credit assignment, and misaligned policy optimization. AT$^2$PO introduces a turn-level tree structure that jointly enables Entropy-Guided Tree Expansion for strategic exploration and Turn-wise Credit Assignment for fine-grained reward propagation from sparse outcomes. Complementing this, we propose Agentic Turn-based Policy Optimization, a turn-level learning objective that aligns policy updates with the natural decision granularity of agentic interactions. ATPO is orthogonal to tree search and can be readily integrated into any multi-turn RL pipeline. Experiments across seven benchmarks demonstrate consistent improvements over the state-of-the-art baseline by up to 1.84 percentage points in average, with ablation studies validating the effectiveness of each component. Our code is available at https://github.com/zzfoutofspace/ATPO.

cs.AI↗

arXiv2026

TransMamba: A Sequence-Level Hybrid Transformer-Mamba Language Model

Transformers are the cornerstone of modern large language models, but their quadratic computational complexity limits efficiency in long-sequence processing. Recent advancements in Mamba, a state space model (SSM) with linear complexity, offer promising efficiency gains but suffer from unstable contextual learning and multitask generalization. Some works conduct layer-level hybrid structures that combine Transformer and Mamba layers, aiming to make full use of both advantages. This paper proposes TransMamba, a novel sequence-level hybrid framework that unifies Transformer and Mamba through shared parameter matrices (QKV and CBx), and thus could dynamically switch between attention and SSM mechanisms at different token lengths and layers. We design the Memory Converter to bridge Transformer and Mamba by converting attention outputs into SSM-compatible states, ensuring seamless information flow at TransPoints where the transformation happens. The TransPoint scheduling is also thoroughly explored for balancing effectiveness and efficiency. We conducted extensive experiments demonstrating that TransMamba achieves superior training efficiency and performance compared to single and hybrid baselines, and validated the deeper consistency between Transformer and Mamba paradigms at sequence level, offering a scalable solution for next-generation language modeling. Code and data are available at https://github.com/Yixing-Li/TransMamba

cs.LG↗

arXiv2026

The Repulsive Effect of Covariant Effective Quantum Gravity

In General Relativity, gravity around a black hole is universally regarded as an attractive force. However, quantum effects may significantly alter this classical picture and thus affect the dynamics of test particles. In this work, we investigate geodesic motion in a recently proposed covariant effective quantum black hole model, the ZLMY type I model in \href{https://journals.aps.org/prd/abstract/10.1103/PhysRevD.111.L081504}{[Phys.Rev.D 111 (2025) 8, L081504]}. Our analytical study reveals that, at the Planck scale, gravity can exhibit a repulsive character, most notably in trajectories involving turning points. These findings suggest that quantum corrections could fundamentally modify our conventional understanding of gravitational interactions in certain regimes.

gr-qc↗

arXiv2025

HY-Motion 1.0: Scaling Flow Matching Models for Text-To-Motion Generation

We present HY-Motion 1.0, a series of state-of-the-art, large-scale, motion generation models capable of generating 3D human motions from textual descriptions. HY-Motion 1.0 represents the first successful attempt to scale up Diffusion Transformer (DiT)-based flow matching models to the billion-parameter scale within the motion generation domain, delivering instruction-following capabilities that significantly outperform current open-source benchmarks. Uniquely, we introduce a comprehensive, full-stage training paradigm -- including large-scale pretraining on over 3,000 hours of motion data, high-quality fine-tuning on 400 hours of curated data, and reinforcement learning from both human feedback and reward models -- to ensure precise alignment with the text instruction and high motion quality. This framework is supported by our meticulous data processing pipeline, which performs rigorous motion cleaning and captioning. Consequently, our model achieves the most extensive coverage, spanning over 200 motion categories across 6 major classes. We release HY-Motion 1.0 to the open-source community to foster future research and accelerate the transition of 3D human motion generation models towards commercial maturity.

cs.CV↗

arXiv2025

An Efficient Embedding Based Ad Retrieval with GPU-Powered Feature Interaction

In large-scale advertising recommendation systems, retrieval serves as a critical component, aiming to efficiently select a subset of candidate ads relevant to user behaviors from a massive ad inventory for subsequent ranking and recommendation. The Embedding-Based Retrieval (EBR) methods modeled by the dual-tower network are widely used in the industry to maintain both retrieval efficiency and accuracy. However, the dual-tower model has significant limitations: the embeddings of users and ads interact only at the final inner product computation, resulting in insufficient feature interaction capabilities. Although DNN-based models with both user and ad as input features, allowing for early-stage interaction between these features, are introduced in the ranking stage to mitigate this issue, they are computationally infeasible for the retrieval stage. To bridge this gap, this paper proposes an efficient GPU-based feature interaction for the dual-tower network to significantly improve retrieval accuracy while substantially reducing computational costs. Specifically, we introduce a novel compressed inverted list designed for GPU acceleration, enabling efficient feature interaction computation at scale. To the best of our knowledge, this is the first framework in the industry to successfully implement Wide and Deep in a retrieval system. We apply this model to the real-world business scenarios in Tencent Advertising, and experimental results demonstrate that our method outperforms existing approaches in offline evaluation and has been successfully deployed to Tencent's advertising recommendation system, delivering significant online performance gains. This improvement not only validates the effectiveness of the proposed method, but also provides new practical guidance for optimizing large-scale ad retrieval systems.

cs.LG↗

arXiv2025

From Feature Interaction to Feature Generation: A Generative Paradigm of CTR Prediction Models

Click-Through Rate (CTR) prediction, a core task in recommendation systems, aims to estimate the probability of users clicking on items. Existing models predominantly follow a discriminative paradigm, which relies heavily on explicit interactions between raw ID embeddings. However, this paradigm inherently renders them susceptible to two critical issues: embedding dimensional collapse and information redundancy, stemming from the over-reliance on feature interactions \emph{over raw ID embeddings}. To address these limitations, we propose a novel \emph{Supervised Feature Generation (SFG)} framework, \emph{shifting the paradigm from discriminative ``feature interaction" to generative ``feature generation"}. Specifically, SFG comprises two key components: an \emph{Encoder} that constructs hidden embeddings for each feature, and a \emph{Decoder} tasked with regenerating the feature embeddings of all features from these hidden representations. Unlike existing generative approaches that adopt self-supervised losses, we introduce a supervised loss to utilize the supervised signal, \ie, click or not, in the CTR prediction task. This framework exhibits strong generalizability: it can be seamlessly integrated with most existing CTR models, reformulating them under the generative paradigm. Extensive experiments demonstrate that SFG consistently mitigates embedding collapse and reduces information redundancy, while yielding substantial performance gains across various datasets and base models. The code is available at https://github.com/USTC-StarTeam/GE4Rec.

cs.IR↗

arXiv2025

InpaintDPO: Mitigating Spatial Relationship Hallucinations in Foreground-conditioned Inpainting via Diverse Preference Optimization

Foreground-conditioned inpainting, which aims at generating a harmonious background for a given foreground subject based on the text prompt, is an important subfield in controllable image generation. A common challenge in current methods, however, is the occurrence of Spatial Relationship Hallucinations between the foreground subject and the generated background, including inappropriate scale, positional relationships, and viewpoints. Critically, the subjective nature of spatial rationality makes it challenging to quantify, hindering the use of traditional reward-based RLHF methods. To address this issue, we propose InpaintDPO, the first Direct Preference Optimization (DPO) based framework dedicated to spatial rationality in foreground-conditioned inpainting, ensuring plausible spatial relationships between foreground and background elements. To resolve the gradient conflicts in standard DPO caused by identical foreground in win-lose pairs, we propose MaskDPO, which confines preference optimization exclusively to the background to enhance background spatial relationships, while retaining the inpainting loss in the foreground region for robust foreground preservation. To enhance coherence at the foreground-background boundary, we propose Conditional Asymmetric Preference Optimization, which samples pairs with differentiated cropping operations and applies global preference optimization to promote contextual awareness and enhance boundary coherence. Finally, based on the observation that winning samples share a commonality in plausible spatial relationships, we propose Shared Commonality Preference Optimization to enhance the model's understanding of spatial commonality across high-quality winning samples, further promoting shared spatial rationality.

cs.CV↗

arXiv2025

HunyuanOCR Technical Report

This paper presents HunyuanOCR, a commercial-grade, open-source, and lightweight (1B parameters) Vision-Language Model (VLM) dedicated to OCR tasks. The architecture comprises a Native Vision Transformer (ViT) and a lightweight LLM connected via an MLP adapter. HunyuanOCR demonstrates superior performance, outperforming commercial APIs, traditional pipelines, and larger models (e.g., Qwen3-VL-4B). Specifically, it surpasses current public solutions in perception tasks (Text Spotting, Parsing) and excels in semantic tasks (IE, Text Image Translation), securing first place in the ICDAR 2025 DIMT Challenge (Small Model Track). Furthermore, it achieves state-of-the-art (SOTA) results on OCRBench among VLMs with fewer than 3B parameters. HunyuanOCR achieves breakthroughs in three key aspects: 1) Unifying Versatility and Efficiency: We implement comprehensive support for core capabilities including spotting, parsing, IE, VQA, and translation within a lightweight framework. This addresses the limitations of narrow "OCR expert models" and inefficient "General VLMs". 2) Streamlined End-to-End Architecture: Adopting a pure end-to-end paradigm eliminates dependencies on pre-processing modules (e.g., layout analysis). This fundamentally resolves error propagation common in traditional pipelines and simplifies system deployment. 3) Data-Driven and RL Strategies: We confirm the critical role of high-quality data and, for the first time in the industry, demonstrate that Reinforcement Learning (RL) strategies yield significant performance gains in OCR tasks. HunyuanOCR is officially open-sourced on HuggingFace. We also provide a high-performance deployment solution based on vLLM, placing its production efficiency in the top tier. We hope this model will advance frontier research and provide a solid foundation for industrial applications.

cs.CV↗

arXiv2025

Constraining the outer boundary condition for the Babcock-Leighton dynamo models

The evolution of the Sun's large-scale surface magnetic field is well captured by surface flux transport models, which can therefore provide a natural constraint on the outer boundary condition (BC) of Babcock-Leighton (BL) dynamo models. For the first time, we propose a zero radial diffusion BC for BL dynamo models, enabling their surface field evolution to align consistently with surface flux transport simulations. We derive a zero radial diffusion BC from the Magnetohydrodynamic induction equation and evaluate its effects in comparison with two alternatives: (i) a radial outer BC, and (ii) a radial outer BC combined with strong near-surface radial pumping. The comparison is carried out both for the evolution of a single bipolar magnetic region and within a full BL dynamo model. The zero radial diffusion outer BC effectively suppresses radial diffusion across the surface, ensuring consistency between the evolution of the bipolar magnetic region in the BL dynamo and the surface flux transport model. With this outer BC, the full BL dynamo model successfully reproduces the fundamental properties of the solar cycle. In addition, the model naturally produces a surface magnetic field that is not purely radial, in closer agreement with solar observations. The physically motivated zero radial diffusion boundary condition paves the way for deeper insight into the solar and stellar cycles.

astro-ph.SR↗

arXiv2025

HunyuanVideo 1.5 Technical Report

We present HunyuanVideo 1.5, a lightweight yet powerful open-source video generation model that achieves state-of-the-art visual quality and motion coherence with only 8.3 billion parameters, enabling efficient inference on consumer-grade GPUs. This achievement is built upon several key components, including meticulous data curation, an advanced DiT architecture featuring selective and sliding tile attention (SSTA), enhanced bilingual understanding through glyph-aware text encoding, progressive pre-training and post-training, and an efficient video super-resolution network. Leveraging these designs, we developed a unified framework capable of high-quality text-to-video and image-to-video generation across multiple durations and resolutions. Extensive experiments demonstrate that this compact and proficient model establishes a new state-of-the-art among open-source video generation models. By releasing the code and model weights, we provide the community with a high-performance foundation that lowers the barrier to video creation and research, making advanced video generation accessible to a broader audience. All open-source assets are publicly available at https://github.com/Tencent-Hunyuan/HunyuanVideo-1.5.

cs.CV↗

arXiv2025

Skypilot: Fine-Tuning LLM with Physical Grounding for AAV Coverage Search

Autonomous aerial vehicles (AAVs) have played a pivotal role in coverage operations and search missions. Recent advances in large language models (LLMs) offer promising opportunities to augment AAV intelligence. These advances help address complex challenges like area coverage optimization, dynamic path planning, and adaptive decision-making. However, the absence of physical grounding in LLMs leads to hallucination and reproducibility problems in spatial reasoning and decision-making. To tackle these issues, we present Skypilot, an LLM-enhanced two-stage framework that grounds language models in physical reality by integrating monte carlo tree search (MCTS). In the first stage, we introduce a diversified action space that encompasses generate, regenerate, fine-tune, and evaluate operations, coupled with physics-informed reward functions to ensure trajectory feasibility. In the second stage, we fine-tune Qwen3-4B on 23,000 MCTS-generated samples, achieving substantial inference acceleration while maintaining solution quality. Extensive numerical simulations and real-world flight experiments validate the efficiency and superiority of our proposed approach. Detailed information and experimental results are accessible at https://sky-pilot.top.

cs.RO↗

arXiv2025

Hydraulis: Balancing Large Transformer Model Training via Co-designing Parallel Strategies and Data Assignment

To optimize large Transformer model training, both efficient parallel computing and advanced data management are indispensable. However, current methods often assume a stable and uniform training workload, neglecting data-induced imbalances-arising from both sampling and packing processes-which can impede training performance. Specifically, data sampling imbalance arises from uneven sequence length distribution of the training data, while data packing imbalance stems from the discrepancy between the linear memory complexity and quadratic time complexity of the attention mechanism. To address these imbalance issues, we develop Hydraulis, which jointly optimizes the parallel strategies and data assignment. For one thing, we introduce large model training with dynamic heterogeneous parallel strategies in response to the sequence length variations within and across training iterations. For another, we devise a two-stage data assignment approach, which strikes a good balance in terms of the training workloads both within and across model replicas. Empirical results demonstrate that Hydraulis outperforms existing systems by 1.32-2.66 times.

cs.DC↗

arXiv2025

Domain Wall Sliding-induced Polarization Switching in Multilayer Graphene

Electric polarization and metallicity are long believed not to coexist until the emergence of exceptionally rare material examples including the bulk polar metals and more recently two-dimensional (2D) van der Waals (vdW) materials such as 1T' WTe2. The electric polarization for the latter represents a new and distinguishable paradigm in materials science and physics because its electric polarization states embedded in the conduction electron sea are able to couple with (and controlled by) the external electric field. However, the microscopic polarization switching process and mechanism in these 2D vdW metallic materials have not been experimentally observed and remain elusive. Here, we report the first direct experimental imaging of the microscopic mechanism behind electric-field-coupled polarization switching in a metallic system. Our gate-tunable nanoscale optical imaging identifies the robust coexistence of electric polarization and appreciable carrier densities in adjacent polar domains hosting opposite electric polarizations in a 2D elemental metallic material, tetralayer graphene. We directly visualize and verify that the sliding domain wall (DW) solitons confined at the middlemost interface are responsible for the polarization switching in tetralayer graphene upon the application of electric fields and mechanical forces. Our work provides the first direct visualization of domain wall sliding-induced polarization switching in 2D elemental carbon at room temperature, significantly expanding and advancing the research of "ferroelectric metal" initially dubbed by P. W. Anderson and coauthors in 1965.

cond-mat.mes-hall↗

arXiv2025

TridentServe: A Stage-level Serving System for Diffusion Pipelines

Diffusion pipelines, renowned for their powerful visual generation capabilities, have seen widespread adoption in generative vision tasks (e.g., text-to-image/video). These pipelines typically follow an encode--diffuse--decode three-stage architecture. Current serving systems deploy diffusion pipelines within a static, manual, and pipeline-level paradigm, allocating the same resources to every request and stage. However, through an in-depth analysis, we find that such a paradigm is inefficient due to the discrepancy in resource needs across the three stages of each request, as well as across different requests. Following the analysis, we propose the dynamic stage-level serving paradigm and develop TridentServe, a brand new diffusion serving system. TridentServe automatically, dynamically derives the placement plan (i.e., how each stage resides) for pipeline deployment and the dispatch plan (i.e., how the requests are routed) for request processing, co-optimizing the resource allocation for both model and requests. Extensive experiments show that TridentServe consistently improves SLO attainment and reduces average/P95 latencies by up to 2.5x and 3.6x/4.1x over existing works across a variety of workloads.

cs.DC↗

arXiv2025

Towards a Comprehensive Scaling Law of Mixture-of-Experts

Mixture-of-Experts (MoE) models have become the consensus approach for enabling parameter-efficient scaling and cost-effective deployment in large language models. However, existing scaling laws for dense models are inapplicable to MoE models, which stems from three critical challenges: the multiplicity of influencing factors, their intricate coupling relationships and the non-monotonic nature of their performance impacts. They collectively necessitate a fine-grained investigation into MoE-specific scaling laws. In this work, we perform a systematic decomposition of MoE settings, identifying five key factors that influence model performance from both size and structural perspectives (data size ($D$), total model size ($N$), activated model size ($N_a$), number of active experts ($G$) and the ratio of shared experts ($S$)). Specifically, we design $446$ controlled experiments to characterize their marginal effects, ultimately constructing a comprehensive and precise joint MoE scaling law that considers all essential factors. Furthermore, we derive the theoretically optimal and practically efficiency-aware optimal configurations for $G$, $S$ and $N_a/N$ with detailed analyses. Our results demonstrate that the optimal settings for $G$ and $S$ are independent of both the model architecture and data size. With the scaling of $N$, the optimal activation parameter ratio of $N_a/N$ becomes sparser. Our proposed MoE scaling law could function as an accurate and insightful guidance to facilitate future MoE model design and training.

cs.LG↗

arXiv2025

Observation-Based Iterative Map for Solar Cycles. II. The Gnevyshev-Ohl Rule and its Generation Mechanism

The Gnevyshev-Ohl (G-O) rule, or even-odd effect, is an important observational phenomenon in solar cycles, originally suggesting that even-numbered cycles are typically followed by stronger odd-numbered ones. However, subsequent studies have reported varied forms and often conflicting manifestations of this rule, along with diverse interpretations of its physical origin. Using an observation-based iterative map, we investigate these different forms of the G-O rule and propose a more general underlying rule: statistically, a given solar cycle is more likely to be followed by a stronger one, regardless of even-odd numbering. This transition asymmetry arises from the map's inherent asymmetry relative to the diagonal. Over timescales comparable to historical observations, both the G-O rule and its reversal can arise randomly, without a consistent preference. The short-term behavior of the rule is sensitive to the initial cycle, the selected time interval, and the parameters of the recursion function. These findings reconcile previously conflicting reports and point to a general generation mechanism: G-O-like behavior arises naturally from nonlinear stochastic dynamics. While different recursion parameters may lead to varying short-term patterns and statistical strengths, the emergence of G-O-like features appears to be a generic property of such systems.

astro-ph.SR↗