arXiv subjects

Jiawei Chen

Publications and source records attributed to Jiawei Chen.

At least 127 records · Page 7Linked to original sources

arXiv2025

DiSTAR: Diffusion over a Scalable Token Autoregressive Representation for Speech Generation

Recent attempts to interleave autoregressive (AR) sketchers with diffusion-based refiners over continuous speech representations have shown promise, but they remain brittle under distribution shift and offer limited levers for controllability. We introduce DISTAR, a zero-shot text-to-speech framework that operates entirely in a discrete residual vector quantization (RVQ) code space and tightly couples an AR language model with a masked diffusion model, without forced alignment or a duration predictor. Concretely, DISTAR drafts block-level RVQ tokens with an AR language model and then performs parallel masked-diffusion infilling conditioned on the draft to complete the next block, yielding long-form synthesis with blockwise parallelism while mitigating classic AR exposure bias. The discrete code space affords explicit control at inference: DISTAR produces high-quality audio under both greedy and sample-based decoding using classifier-free guidance, supports trade-offs between robustness and diversity, and enables variable bit-rate and controllable computation via RVQ layer pruning at test time. Extensive experiments and ablations demonstrate that DISTAR surpasses state-of-the-art zero-shot TTS systems in robustness, naturalness, and speaker/style consistency, while maintaining rich output diversity. Audio samples are provided on https://anonymous.4open.science/w/DiSTAR_demo.

eess.AS↗

arXiv2025

HatLLM: Hierarchical Attention Masking for Enhanced Collaborative Modeling in LLM-based Recommendation

Recent years have witnessed a surge of research on leveraging large language models (LLMs) for sequential recommendation. LLMs have demonstrated remarkable potential in inferring users' nuanced preferences through fine-grained semantic reasoning. However, they also exhibit a notable limitation in effectively modeling collaborative signals, i.e., behavioral correlations inherent in users' historical interactions. Our empirical analysis further reveals that the attention mechanisms in LLMs tend to disproportionately focus on tokens within the same item, thereby impeding the capture of cross-item correlations. To address this limitation, we propose a novel hierarchical attention masking strategy for LLM-based recommendation, termed HatLLM. Specifically, in shallow layers, HatLLM masks attention between tokens from different items, facilitating intra-item semantic understanding; in contrast, in deep layers, HatLLM masks attention within items, thereby compelling the model to capture cross-item correlations. This progressive, layer-wise approach enables LLMs to jointly model both token-level and item-level dependencies. Extensive experiments on three real-world datasets demonstrate that HatLLM achieves significant performance gains (9.13% on average) over existing LLM-based methods.

cs.IR↗

arXiv2025

Heptapod: Language Modeling on Visual Signals

We introduce Heptapod, an image autoregressive model that adheres to the foundational principles of language modeling. Heptapod employs \textbf{causal attention}, \textbf{eliminates reliance on CFG}, and \textbf{eschews the trend of semantic tokenizers}. Our key innovation is \textit{next 2D distribution prediction}: a causal Transformer with reconstruction-focused visual tokenizer, learns to predict the distribution over the entire 2D spatial grid of images at each timestep. This learning objective unifies the sequential modeling of autoregressive framework with the holistic self-supervised learning of masked autoencoding, enabling the model to capture comprehensive image semantics via generative training. On the ImageNet generation benchmark, Heptapod achieves an FID of $2.70$, significantly outperforming previous causal autoregressive approaches. We hope our work inspires a principled rethinking of language modeling on visual signals and beyond.

cs.CV↗

arXiv2025

Learning Counterfactual Outcomes Under Rank Preservation

Counterfactual inference aims to estimate the counterfactual outcome at the individual level given knowledge of an observed treatment and the factual outcome, with broad applications in fields such as epidemiology, econometrics, and management science. Previous methods rely on a known structural causal model (SCM) or assume the homogeneity of the exogenous variable and strict monotonicity between the outcome and exogenous variable. In this paper, we propose a principled approach for identifying and estimating the counterfactual outcome. We first introduce a simple and intuitive rank preservation assumption to identify the counterfactual outcome without relying on a known structural causal model. Building on this, we propose a novel ideal loss for theoretically unbiased learning of the counterfactual outcome and further develop a kernel-based estimator for its empirical estimation. Our theoretical analysis shows that the rank preservation assumption is not stronger than the homogeneity and strict monotonicity assumptions, and shows that the proposed ideal loss is convex, and the proposed estimator is unbiased. Extensive semi-synthetic and real-world experiments are conducted to demonstrate the effectiveness of the proposed method.

cs.LG↗

arXiv2025

LayerCake: Token-Aware Contrastive Decoding within Large Language Model Layers

Large language models (LLMs) excel at natural language understanding and generation but remain vulnerable to factual errors, limiting their reliability in knowledge-intensive tasks. While decoding-time strategies provide a promising efficient solution without training, existing methods typically treat token-level and layer-level signals in isolation, overlooking the joint dynamics between them. In this work, we introduce a token-aware, layer-localized contrastive decoding method that aligns specific token types with their most influential transformer layers to improve factual generation. Through empirical attention analysis, we identify two key patterns: punctuation tokens receive dominant attention in early layers, while conceptual tokens govern semantic reasoning in intermediate layers. By selectively suppressing attention to these token types at their respective depths, we achieve the induction of controlled factual degradation and derive contrastive signals to guide the final factual decoding. Our method requires no additional training or model modification, and experiments demonstrate that our method consistently improves factuality across multiple LLMs and various benchmarks.

cs.AI↗

arXiv2025

MIRG-RL: Multi-Image Reasoning and Grounding with Reinforcement Learning

Multi-image reasoning and grounding require understanding complex cross-image relationships at both object levels and image levels. Current Large Visual Language Models (LVLMs) face two critical challenges: the lack of cross-image reasoning capabilities and insufficient cross-image reference reward modeling. To address these issues, we propose a unified framework - Multi-Image Reasoning and Grounding with Reinforcement Learning (MIRG-RL). Specifically, our two-stage training paradigm combines supervised fine-tuning with annotated trajectories and image-aware reinforcement learning optimization, progressively developing multi-image reasoning capabilities. Furthermore, we innovatively propose a method for constructing the trajectory data, which integrates object-level and image-level annotation information, and use this method to generate a lightweight reasoning-enhanced dataset. To effectively resolve cross-image ambiguities, we design an image-aware RL policy with dual reward functions for objects and images. Experiments demonstrate that MIRG-RL achieves state-of-the-art (SOTA) performance in multi-image grounding benchmarks, attaining 64.82% on cross-image reasoning tasks - exceeding the previous best method by 1%. The code and dataset have been released at https://github.com/ZEUS2035/MIRG-RL.

cs.CV↗

arXiv2025

ConsistentChat: Building Skeleton-Guided Consistent Multi-Turn Dialogues for Large Language Models from Scratch

Current instruction data synthesis methods primarily focus on single-turn instructions and often neglect cross-turn coherence, resulting in context drift and reduced task completion rates in extended conversations. To address this limitation, we propose Skeleton-Guided Multi-Turn Dialogue Generation, a framework that constrains multi-turn instruction synthesis by explicitly modeling human conversational intent. It operates in two stages: (1) Intent Modeling, which captures the global structure of human dialogues by assigning each conversation to one of nine well-defined intent trajectories, ensuring a coherent and goal-oriented information flow; and (2) Skeleton Generation, which constructs a structurally grounded sequence of user queries aligned with the modeled intent, thereby serving as a scaffold that constrains and guides the downstream instruction synthesis process. Based on this process, we construct ConsistentChat, a multi-turn instruction dataset with approximately 15,000 multi-turn conversations and 224,392 utterances. Experiments on the Light, Topdial, and MT-Eval benchmarks show that models fine-tuned on ConsistentChat achieve a 20-30% improvement in chat consistency and up to a 15% increase in task success rate, significantly outperforming models trained on existing single-turn and multi-turn instruction datasets.

cs.CL↗

arXiv2025

Motion of spinning particles around a quantum-corrected black hole without Cauchy horizons

In this paper, we investigate the motion of spinning particles around a covariant quantum-corrected black hole without a Cauchy horizon within the framework of effective quantum gravity, and examine the influence of quantum gravitational effects on the motion of these spinning particles. First, we employ the Mathisson-Papapetrou-Dixon equations to derive the 4-momentum and 4-velocity of spinning particles, and introduce the effective potential for radial motion using the components of the 4-momentum. We find that an increase in the quantum parameter $ζ$ leads to a decrease in the effective potential, while the spin $S$ significantly affects the magnitude of the effective potential. Then, through the effective potential, we investigate the properties of circular orbits and the innermost stable circular orbit, and discuss the timelike condition that spinning particles must satisfy when moving around the black hole. Finally, we study the trajectories of spinning particles on bound orbits around the quantum-corrected black hole and compare them with those around other covariant quantum-corrected black holes. The results show that the trajectories of spinning particles in this quantum-corrected black hole model are weakly influenced by $ζ$, making them almost indistinguishable from those in the Schwarzschild black hole, but they can be distinguished from other covariant quantum-corrected models under certain initial conditions. These results contribute to our understanding of black hole properties under quantum corrections.

gr-qc↗

arXiv2025

MobileRAG: Enhancing Mobile Agent with Retrieval-Augmented Generation

Smartphones have become indispensable in people's daily lives, permeating nearly every aspect of modern society. With the continuous advancement of large language models (LLMs), numerous LLM-based mobile agents have emerged. These agents are capable of accurately parsing diverse user queries and automatically assisting users in completing complex or repetitive operations. However, current agents 1) heavily rely on the comprehension ability of LLMs, which can lead to errors caused by misoperations or omitted steps during tasks, 2) lack interaction with the external environment, often terminating tasks when an app cannot fulfill user queries, and 3) lack memory capabilities, requiring each instruction to reconstruct the interface and being unable to learn from and correct previous mistakes. To alleviate the above issues, we propose MobileRAG, a mobile agents framework enhanced by Retrieval-Augmented Generation (RAG), which includes InterRAG, LocalRAG, and MemRAG. It leverages RAG to more quickly and accurately identify user queries and accomplish complex and long-sequence mobile tasks. Additionally, to more comprehensively assess the performance of MobileRAG, we introduce MobileRAG-Eval, a more challenging benchmark characterized by numerous complex, real-world mobile tasks that require external knowledge assistance. Extensive experimental results on MobileRAG-Eval demonstrate that MobileRAG can easily handle real-world mobile tasks, achieving 10.3\% improvement over state-of-the-art methods with fewer operational steps. Our code is publicly available at: https://github.com/liuxiaojieOutOfWorld/MobileRAG_arxiv

cs.CL↗

arXiv2025

BloomScene: Lightweight Structured 3D Gaussian Splatting for Crossmodal Scene Generation

With the widespread use of virtual reality applications, 3D scene generation has become a new challenging research frontier. 3D scenes have highly complex structures and need to ensure that the output is dense, coherent, and contains all necessary structures. Many current 3D scene generation methods rely on pre-trained text-to-image diffusion models and monocular depth estimators. However, the generated scenes occupy large amounts of storage space and often lack effective regularisation methods, leading to geometric distortions. To this end, we propose BloomScene, a lightweight structured 3D Gaussian splatting for crossmodal scene generation, which creates diverse and high-quality 3D scenes from text or image inputs. Specifically, a crossmodal progressive scene generation framework is proposed to generate coherent scenes utilizing incremental point cloud reconstruction and 3D Gaussian splatting. Additionally, we propose a hierarchical depth prior-based regularization mechanism that utilizes multi-level constraints on depth accuracy and smoothness to enhance the realism and continuity of the generated scenes. Ultimately, we propose a structured context-guided compression mechanism that exploits structured hash grids to model the context of unorganized anchor attributes, which significantly eliminates structural redundancy and reduces storage overhead. Comprehensive experiments across multiple scenes demonstrate the significant potential and advantages of our framework compared with several baselines.

cs.CV↗

arXiv2025

CoCoNUTS: Concentrating on Content while Neglecting Uninformative Textual Styles for AI-Generated Peer Review Detection

The growing integration of large language models (LLMs) into the peer review process presents potential risks to the fairness and reliability of scholarly evaluation. While LLMs offer valuable assistance for reviewers with language refinement, there is growing concern over their use to generate substantive review content. Existing general AI-generated text detectors are vulnerable to paraphrasing attacks and struggle to distinguish between surface language refinement and substantial content generation, suggesting that they primarily rely on stylistic cues. When applied to peer review, this limitation can result in unfairly suspecting reviews with permissible AI-assisted language enhancement, while failing to catch deceptively humanized AI-generated reviews. To address this, we propose a paradigm shift from style-based to content-based detection. Specifically, we introduce CoCoNUTS, a content-oriented benchmark built upon a fine-grained dataset of AI-generated peer reviews, covering six distinct modes of human-AI collaboration. Furthermore, we develop CoCoDet, an AI review detector via a multi-task learning framework, designed to achieve more accurate and robust detection of AI involvement in review content. Our work offers a practical foundation for evaluating the use of LLMs in peer review, and contributes to the development of more precise, equitable, and reliable detection methods for real-world scholarly applications. Our code and data will be publicly available at https://github.com/Y1hanChen/COCONUTS.

cs.CL↗

arXiv2025

From Pixels to Tokens: Revisiting Object Hallucinations in Large Vision-Language Models

Hallucinations in large vision-language models (LVLMs) are a significant challenge, i.e., generating objects that are not presented in the visual input, which impairs their reliability. Recent studies often attribute hallucinations to a lack of understanding of visual input, yet ignore a more fundamental issue: the model's inability to effectively extract or decouple visual features. In this paper, we revisit the hallucinations in LVLMs from an architectural perspective, investigating whether the primary cause lies in the visual encoder (feature extraction) or the modal alignment module (feature decoupling). Motivated by our findings on the preliminary investigation, we propose a novel tuning strategy, PATCH, to mitigate hallucinations in LVLMs. This plug-and-play method can be integrated into various LVLMs, utilizing adaptive virtual tokens to extract object features from bounding boxes, thereby addressing hallucinations caused by insufficient decoupling of visual features. PATCH achieves state-of-the-art performance on multiple multi-modal hallucination datasets. We hope this approach provides researchers with deeper insights into the underlying causes of hallucinations in LVLMs, fostering further advancements and innovation in this field.

cs.CL↗

arXiv2025

COPO: Consistency-Aware Policy Optimization

Reinforcement learning has significantly enhanced the reasoning capabilities of Large Language Models (LLMs) in complex problem-solving tasks. Recently, the introduction of DeepSeek R1 has inspired a surge of interest in leveraging rule-based rewards as a low-cost alternative for computing advantage functions and guiding policy optimization. However, a common challenge observed across many replication and extension efforts is that when multiple sampled responses under a single prompt converge to identical outcomes, whether correct or incorrect, the group-based advantage degenerates to zero. This leads to vanishing gradients and renders the corresponding samples ineffective for learning, ultimately limiting training efficiency and downstream performance. To address this issue, we propose a consistency-aware policy optimization framework that introduces a structured global reward based on outcome consistency, the global loss based on it ensures that, even when model outputs show high intra-group consistency, the training process still receives meaningful learning signals, which encourages the generation of correct and self-consistent reasoning paths from a global perspective. Furthermore, we incorporate an entropy-based soft blending mechanism that adaptively balances local advantage estimation with global optimization, enabling dynamic transitions between exploration and convergence throughout training. Our method introduces several key innovations in both reward design and optimization strategy. We validate its effectiveness through substantial performance gains on multiple mathematical reasoning benchmarks, highlighting the proposed framework's robustness and general applicability. Code of this work has been released at https://github.com/hijih/copo-code.git.

cs.LG↗

arXiv2025

Breaking the Top-$K$ Barrier: Advancing Top-$K$ Ranking Metrics Optimization in Recommender Systems

In the realm of recommender systems (RS), Top-$K$ ranking metrics such as NDCG@$K$ are the gold standard for evaluating recommendation performance. However, during the training of recommendation models, optimizing NDCG@$K$ poses significant challenges due to its inherent discontinuous nature and the intricate Top-$K$ truncation. Recent efforts to optimize NDCG@$K$ have either overlooked the Top-$K$ truncation or suffered from high computational costs and training instability. To overcome these limitations, we propose SoftmaxLoss@$K$ (SL@$K$), a novel recommendation loss tailored for NDCG@$K$ optimization. Specifically, we integrate the quantile technique to handle Top-$K$ truncation and derive a smooth upper bound for optimizing NDCG@$K$ to address discontinuity. The resulting SL@$K$ loss has several desirable properties, including theoretical guarantees, ease of implementation, computational efficiency, gradient stability, and noise robustness. Extensive experiments on four real-world datasets and three recommendation backbones demonstrate that SL@$K$ outperforms existing losses with a notable average improvement of 6.03%. The code is available at https://github.com/Tiny-Snow/IR-Benchmark.

cs.IR↗

arXiv2025

SmartFlow: A CFD-solver-agnostic deep reinforcement learning framework for computational fluid dynamics on HPC platforms

Deep reinforcement learning (DRL) is emerging as a powerful tool for fluid-dynamics research, encompassing active flow control, autonomous navigation, turbulence modeling and discovery of novel numerical schemes. We introduce SmartFlow, a CFD-solver-agnostic framework for both single- and multi-agent DRL algorithms that can easily integrate with MPI-parallel CPU and GPU-accelerated solvers. Built on Relexi and SmartSOD2D, SmartFlow uses the SmartSim infrastructure library and our newly developed SmartRedis-MPI library to enable asynchronous, low-latency, in-memory communication between CFD solvers and Python-based DRL algorithms. SmartFlow leverages PyTorch's Stable-Baselines3 for training, which provides a modular, Gym-like environment API. We demonstrate its versatility via three case studies: single-agent synthetic-jet control for drag reduction in a cylinder flow simulated by the high-order FLEXI solver, multi-agent cylinder wake control using the GPU-accelerated spectral-element code SOD2D, and multi-agent wall-model learning for large-eddy simulation with the finite-difference solver CaLES. SmartFlow's CFD-solver-agnostic design and seamless HPC integration is promising to accelerate RL-driven fluid-mechanics studies.

physics.flu-dyn↗

arXiv2025

Near-Wall Scaling and Separation Prediction of a Rotation-Based Subgrid-Scale Stress Model

This paper presents an in-depth analysis of a novel subgrid-scale stress model proposed in 2022, which utilizes the rotational part of the velocity gradient as the velocity scale for computing eddy viscosity. This study investigates the near-wall asymptotic behavior and separation prediction capability of this model for the first time. Two canonical flows--fully-developed turbulent channel flow and periodic hill flow--are selected for analysis. The eddy viscosity predicted by this model correlates well with the visualized vortices and exhibits an asymptotic behavior of O(y) near the walls. The dimensionless eddy viscosity, like that of the Wall-Adapting Local Eddy Viscosity (WALE) subgrid model, remains within a small numerical range of 10^-2 to 10^-4. The power spectral density results reveal the asymptotic behavior of the velocity scale in the dissipation range, following a -10/3 scaling law. Additionally, this model predicts velocity profiles more accurately than the Smagorinsky model, even when using Van Driest damping. For the periodic hill case, this model predicts the reattachment point with only a 6.9% error, compared to 14.0% for the Smagorinsky model and 16.4% for the Smagorinsky model with Van Driest damping. In near-wall regions with separation, this model achieves even greater accuracy in Reynolds stress prediction than the WALE model, demonstrating its superior potential for separated flow simulations.

physics.flu-dyn↗

arXiv2025

A numerical investigation of sweep effects on turbulent flow over iced wings

This study employs an improved delayed detached eddy simulation (AMD-IDDES) method to investigate the flow characteristics of an iced swept wing. The AMD-IDDES approach incorporates an anisotropic minimum-dissipation (AMD) model into the traditional IDDES framework to better account for local flow anisotropy and energy dissipation characteristics. The method is validated against the traditional IDDES approach, and the results show that AMD-IDDES more accurately captures the evolution of vortices and separated shear layers. Compared with an iced straight wing, a swept wing exhibits distinct aerodynamic behavior driven mainly by the sweep angle. The sweep angle induces strong spanwise flow, which reshapes the separation region and transforms the flow from two-dimensional to three-dimensional. This spanwise motion significantly alters vortex development and enhances the complexity of the unsteady flow. The shear layer separation, initiated by a Kelvin-Helmholtz instability, dominates the unsteady aerodynamic response. Although wingtip effects remain secondary, their interaction with the leading-edge vortex still contributes to lowering the dominant frequency. Overall, this study highlights that the aerodynamic forces of an iced swept wing are primarily governed by sweep-induced spanwise flow and associated shear layer dynamics, with minimal influence from the root and tip regions.

physics.flu-dyn↗

arXiv2025

Periodic orbits and gravitational waveforms in quantum-corrected black hole spacetimes

In this paper, we study the periodic orbits of massive particles around two quantum-corrected black holes proposed in effective quantum gravity, and explore the quantum gravity effect on both the particle orbits and the associated gravitational wave signals. First, we analyze the geodesic motion of the massive particle around the black holes. We then study two important types of bound orbits of the massive particles, the marginally bound orbit and the innermost stable circular orbit. We find that, for the first black hole, increasing the quantum parameter $ζ$ leads to larger orbital radii and reduced angular momenta for both orbits. In contrast, the second black hole shows $ζ$-independent orbital radii and angular momenta. By analyzing the effective potential, we determine the allowed range of the energy and the angular momentum for bound orbits, with $ζ$-dependence only for the first black hole. We further investigate periodic orbits with a fixed energy for both black holes, revealing that the parameter $ζ$ similarly affects the orbits, although its effect is negligible in the second black hole. Finally, we calculate the gravitational waves emitted by the periodic orbits. The results demonstrate that increasing $ζ$ leads to a significant phase delay for the first black hole, while only inducing a subtle phase advance for the second one. Therefore, we conclude that the first black hole can be distinguished from the Schwarzschild one through gravitational wave observations, whereas the second one cannot be effectively distinguished when the quantum correction is weak.

gr-qc↗