arXiv subjects

Huan Yu

Publications and source records attributed to Huan Yu.

At least 37 records · Page 2Linked to original sources

arXiv2026

Search-P1: Path-Centric Reward Shaping for Stable and Efficient Agentic RAG Training

Retrieval-Augmented Generation (RAG) enhances large language models (LLMs) by incorporating external knowledge, yet traditional single-round retrieval struggles with complex multi-step reasoning. Agentic RAG addresses this by enabling LLMs to dynamically decide when and what to retrieve, but current RL-based training methods suffer from sparse outcome rewards that discard intermediate signals and low sample efficiency where failed samples contribute nothing. We propose Search-P1, a framework that introduces path-centric reward shaping for agentic RAG training, comprising two key components: (1) Path-Centric Reward, which evaluates the structural quality of reasoning trajectories through order-agnostic step coverage and soft scoring that extracts learning signals even from failed samples, and (2) Dual-Track Path Scoring with offline-generated reference planners that assesses paths from both self-consistency and reference-alignment perspectives. Experiments on multiple QA benchmarks demonstrate that Search-P1 achieves significant improvements over Search-R1 and other strong baselines, with an average accuracy gain of 7.7 points.

cs.CL↗

arXiv2026

Towards Faithful Industrial RAG: A Reinforced Co-adaptation Framework for Advertising QA

Industrial advertising question answering (QA) is a high-stakes task in which hallucinated content, particularly fabricated URLs, can lead to financial loss, compliance violations, and legal risk. Although Retrieval-Augmented Generation (RAG) is widely adopted, deploying it in production remains challenging because industrial knowledge is inherently relational, frequently updated, and insufficiently aligned with generation objectives. We propose a reinforced co-adaptation framework that jointly optimizes retrieval and generation through two components: (1) Graph-aware Retrieval (GraphRAG), which models entity-relation structure over a high-citation knowledge subgraph for multi-hop, domain-specific evidence selection; and (2) evidence-constrained reinforcement learning via Group Relative Policy Optimization (GRPO) with multi-dimensional rewards covering faithfulness, style compliance, safety, and URL validity. Experiments on an internal advertising QA dataset show consistent gains across expert-judged dimensions including accuracy, completeness, and safety, while reducing the hallucination rate by 72\%. A two-week online A/B test demonstrates a 28.6\% increase in like rate, a 46.2\% decrease in dislike rate, and a 92.7\% reduction in URL hallucination. The system has been running in production for over half a year and has served millions of QA interactions.

cs.CL↗

arXiv2026

SCoTER: Structured Chain-of-Thought Transfer for Enhanced Recommendation

Harnessing the reasoning power of Large Language Models (LLMs) for recommender systems is hindered by two fundamental challenges. First, current approaches lack a mechanism for automated, data-driven discovery of effective reasoning patterns, relying instead on brittle manual templates or unstable zero-shot prompting. Second, they employ structure-collapsing integration: direct prompting incurs prohibitive online inference costs, while feature extraction collapses reasoning chains into single vectors, discarding stepwise logic. To address these challenges, we propose SCoTER (Structured Chain-of-Thought Transfer for Enhanced Recommendation), a unified framework that treats pattern discovery and structure-aware transfer as a jointly optimized problem. Specifically, SCoTER operationalizes this through two synergistic components: a Generate-Validate-Mine (GVM) pipeline for automated pattern discovery and a structure-preserving integration architecture that transfers stepwise logic to efficient models. Empirically, experiments on four benchmarks demonstrate consistent improvements across diverse backbones. Moreover, in production deployment on the Tencent Advertising Platform, SCoTER achieved a 2.14\% lift in Gross Merchandise Value (GMV) while eliminating online LLM inference costs. Overall, SCoTER presents a practical and unified framework for integrating structured LLM reasoning into recommender systems, validated by consistent improvements in both offline benchmarks and online production environments.

cs.IR↗

arXiv2026

S-GRec: Personalized Semantic-Aware Generative Recommendation with Asymmetric Advantage

Generative recommendation models sequence generation to produce items end-to-end, but training from behavioral logs often provides weak supervision on underlying user intent. Although Large Language Models (LLMs) offer rich semantic priors that could supply such supervision, direct adoption in industrial recommendation is hindered by two obstacles: semantic signals can conflict with platform business objectives, and LLM inference is prohibitively expensive at scale. This paper presents S-GRec, a semantic-aware framework that decouples an online lightweight generator from an offline LLM-based semantic judge for train-time supervision. S-GRec introduces a two-stage Personalized Semantic Judge (PSJ) that produces interpretable aspect evidence and learns user-conditional aggregation from pairwise feedback, yielding stable semantic rewards. To prevent semantic supervision from deviating from business goals, Asymmetric Advantage Policy Optimization (A2PO) anchors optimization on business rewards (e.g., eCPM) and injects semantic advantages only when they are consistent. Extensive experiments on public benchmarks and a large-scale production system validate both effectiveness and scalability, including statistically significant gains in CTR and a 1.19\% lift in GMV in online A/B tests, without requiring real-time LLM inference.

cs.IR↗

arXiv2026

Recurrent Preference Memory for Efficient Long-Sequence Generative Recommendation

Generative recommendation (GenRec) models typically model user behavior via full attention, but scaling to lifelong sequences is hindered by prohibitive computational costs and noise accumulation from stochastic interactions. To address these challenges, we introduce Rec2PM, a framework that compresses long user interaction histories into compact Preference Memory tokens. Unlike traditional recurrent methods that suffer from serial training, Rec2PM employs a novel self-referential teacher-forcing strategy: it leverages a global view of the history to generate reference memories, which serve as supervision targets for parallelized recurrent updates. This allows for fully parallel training while maintaining the capability for iterative updates during inference. Additionally, by representing memory as token embeddings rather than extensive KV caches, Rec2PM achieves extreme storage efficiency. Experiments on large-scale benchmarks show that Rec2PM significantly reduces inference latency and memory footprint while achieving superior accuracy compared to full-sequence models. Analysis reveals that the Preference Memory functions as a denoising Information Bottleneck, effectively filtering interaction noise to capture robust long-term interests.

cs.IR↗

arXiv2026

Reasoning to Rank: An End-to-End Solution for Exploiting Large Language Models for Recommendation

Recommender systems are tasked to infer users' evolving preferences and rank items aligned with their intents, which calls for in-depth reasoning beyond pattern-based scoring. Recent efforts start to leverage large language models (LLMs) for recommendation, but how to effectively optimize the model for improved recommendation utility is still under explored. In this work, we propose Reasoning to Rank, an end-to-end training framework that internalizes recommendation utility optimization into the learning of step-by-step reasoning in LLMs. To avoid position bias in LLM reasoning and enable direct optimization of the reasoning process, our framework performs reasoning at the user-item level and employs reinforcement learning for end-to-end training of the LLM. Experiments on three Amazon datasets and a large-scale industrial dataset showed consistent gains over strong conventional and LLM-based solutions. Extensive in-depth analyses validate the necessity of the key components in the proposed framework and shed lights on the future developments of this line of work.

cs.IR↗

arXiv2026

GPR: Towards a Generative Pre-trained One-Model Paradigm for Large-Scale Advertising Recommendation

As an intelligent infrastructure connecting users with commercial content, advertising recommendation systems play a central role in information flow and value creation within the digital economy. However, existing multi-stage advertising recommendation systems suffer from objective misalignment and error propagation, making it difficult to achieve global optimality, while unified generative recommendation models still struggle to meet the demands of practical industrial applications. To address these issues, we propose GPR (Generative Pre-trained Recommender), the first one-model framework that redefines advertising recommendation as an end-to-end generative task, replacing the traditional cascading paradigm with a unified generative approach. To realize GPR, we introduce three key innovations spanning unified representation, network architecture, and training strategy. First, we design a unified input schema and tokenization method tailored to advertising scenarios, mapping both ads and organic content into a shared multi-level semantic ID space, thereby enhancing semantic alignment and modeling consistency across heterogeneous data. Second, we develop the Heterogeneous Hierarchical Decoder (HHD), a dual-decoder architecture that decouples user intent modeling from ad generation, achieving a balance between training efficiency and inference flexibility while maintaining strong modeling capacity. Finally, we propose a multi-stage joint training strategy that integrates Multi-Token Prediction (MTP), Value-Aware Fine-Tuning and the Hierarchy Enhanced Policy Optimization (HEPO) algorithm, forming a complete generative recommendation pipeline that unifies interest modeling, value alignment, and policy optimization. GPR has been fully deployed in the Tencent Weixin Channels advertising system, delivering significant improvements in key business metrics including GMV and CTCVR.

cs.IR↗

arXiv2026

DiffuReason: Bridging Latent Reasoning and Generative Refinement for Sequential Recommendation

Latent reasoning has emerged as a promising paradigm for sequential recommendation, enabling models to capture complex user intent through multi-step deliberation. Yet existing approaches often rely on deterministic latent chains that accumulate noise and overlook the uncertainty inherent in user intent, and they are typically trained in staged pipelines that hinder joint optimization and exploration. To address these challenges, we propose DiffuReason, a unified "Think-then-Diffuse" framework for sequential recommendation. It integrates multi-step Thinking Tokens for latent reasoning, diffusion-based refinement for denoising intermediate representations, and end-to-end Group Relative Policy Optimization (GRPO) alignment to optimize for ranking performance. In the Think stage, the model generates Thinking Tokens that reason over user history to form an initial intent hypothesis. In the Diffuse stage, rather than treating this hypothesis as the final output, we refine it through a diffusion process that models user intent as a probabilistic distribution, providing iterative denoising against reasoning noise. Finally, GRPO-based reinforcement learning enables the reasoning and refinement modules to co-evolve throughout training, without the constraints of staged optimization. Extensive experiments on four benchmarks demonstrate that DiffuReason consistently improves diverse backbone architectures. Online A/B tests on a large-scale industrial platform further validate its practical effectiveness.

cs.IR↗

arXiv2026

Internalizing Multi-Agent Reasoning for Accurate and Efficient LLM-based Recommendation

Large Language Models (LLMs) are reshaping recommender systems by leveraging extensive world knowledge and semantic reasoning to interpret user intent. However, effectively integrating these capabilities with collaborative signals while avoiding prohibitive inference latency remains a critical bottleneck. To address this, we propose a trajectory-driven internalization framework to develop a Single-agent Trajectory-Aligned Recommender (STAR). Specifically, to internalize complex reasoning capabilities into a single efficient model, we first design a multi-agent teacher system capable of multi-turn tool usage and reflection. This teacher utilizes a Collaborative Signal Translation mechanism to explicitly convert latent behavioral patterns into descriptive natural language evidence to enhance reasoning accuracy. Subsequently, a trajectory-driven distillation pipeline transfers this agentic logic, including planning, tool usage, and self-reflection, into the compact STAR model. Extensive experiments demonstrate that STAR surpasses its teacher by 8.7% to 39.5% while eliminating iterative latency, paving the way for real-time, reasoning-enhanced recommendation.

cs.IR↗

arXiv2026

Spend Search Where It Pays: Value-Guided Structured Sampling and Optimization for Generative Recommendation

Generative recommendation via autoregressive models has unified retrieval and ranking into a single conditional generation framework. However, fine-tuning these models with Reinforcement Learning (RL) often suffers from a fundamental probability-reward mismatch. Conventional likelihood-dominated decoding (e.g., beam search) exhibits a myopic bias toward locally probable prefixes, which causes two critical failures: (1) insufficient exploration, where high-reward items in low-probability branches are prematurely pruned and rarely sampled, and (2) advantage compression, where trajectories sharing high-probability prefixes receive highly correlated rewards with low within-group variance, yielding a weak comparative signal for RL. To address these challenges, we propose V-STAR, a Value-guided Sampling and Tree-structured Advantage Reinforcement framework. V-STAR forms a self-evolving loop via two synergistic components. First, a Value-Guided Efficient Decoding (VED) is developed to identify decisive nodes and selectively deepen high-potential prefixes. This improves exploration efficiency without exhaustive tree search. Second, we propose Sibling-GRPO, which exploits the induced tree topology to compute sibling-relative advantages and concentrates learning signals on decisive branching decisions. Extensive experiments on both offline and online datasets demonstrate that V-STAR outperforms state-of-the-art baselines, delivering superior accuracy and candidate-set diversity under strict latency constraints.

cs.AI↗

arXiv2026

ChainRec: An Agentic Recommender Learning to Route Tool Chains for Diverse and Evolving Interests

Large language models (LLMs) are increasingly integrated into recommender systems, motivating recent interest in agentic and reasoning-based recommendation. However, most existing approaches still rely on fixed workflows, applying the same reasoning procedure across diverse recommendation scenarios. In practice, user contexts vary substantially-for example, in cold-start settings or during interest shifts, so an agent should adaptively decide what evidence to gather next rather than following a scripted process. To address this, we propose ChainRec, an agentic recommender that uses a planner to dynamically select reasoning tools. ChainRec builds a standardized Tool Agent Library from expert trajectories. It then trains a planner using supervised fine-tuning and preference optimization to dynamically select tools, decide their order, and determine when to stop. Experiments on AgentRecBench across Amazon, Yelp, and Goodreads show that ChainRec consistently improves Avg HR@{1,3,5} over strong baselines, with especially notable gains in cold-start and evolving-interest scenarios. Ablation studies further validate the importance of tool standardization and preference-optimized planning.

cs.IR↗

arXiv2026

Efficient Cross-Architecture Knowledge Transfer for Large-Scale Online User Response Prediction

Deploying new architectures in large-scale user response prediction systems incurs high model switching costs due to expensive retraining on massive historical data and performance degradation under data retention constraints. Existing knowledge distillation methods struggle with architectural heterogeneity and the prohibitive cost of transferring large embedding tables. We propose CrossAdapt, a two-stage framework for efficient cross-architecture knowledge transfer. The offline stage enables rapid embedding transfer via dimension-adaptive projections without iterative training, combined with progressive network distillation and strategic sampling to reduce computational cost. The online stage introduces asymmetric co-distillation, where students update frequently while teachers update infrequently, together with a distribution-aware adaptation mechanism that dynamically balances historical knowledge preservation and fast adaptation to evolving data. Experiments on three public datasets show that CrossAdapt achieves 0.27-0.43% AUC improvements while reducing training time by 43-71%. Large-scale deployment on Tencent WeChat Channels (~10M daily samples) further demonstrates its effectiveness, significantly mitigating AUC degradation, LogLoss increase, and prediction bias compared to standard distillation baselines.

cs.AI↗

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

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

Event-Triggered Regulation of Mixed-Autonomy Traffic Under Varying Traffic Conditions

Modeling and congestion mitigation of mixed-autonomy traffic systems consisting of human-driven vehicles (HVs) and autonomous vehicles (AVs) have become increasingly critical with the rapid development of autonomous driving technology. This paper develops an event-triggered control (ETC) framework for mitigating congestion in such systems, which are modeled using an extended Aw-Rascle-Zhang (ARZ) formulation consisting of coupled 4 x 4 hyperbolic partial differential equations (PDEs). Ramp metering is employed as the boundary actuation mechanism. To reduce computational and communication burdens while avoiding excessive ramp signal changes, we design the ETC strategy based on the backstepping method, together with an observer-based ETC formulation for practical implementation under limited sensing. Rigorous Lyapunov analysis ensures exponential convergence and avoidance of Zeno behavior. Extensive simulations validate the proposed approach under diverse traffic scenarios, including varying AV penetration rates, different spacing policies, multiple demand levels, and non-recurrent congestion patterns. Results show that ETC not only stabilizes mixed traffic flows but also significantly reduces control updates, improving driver comfort, and roadway safety. Higher AV penetration rates lead to longer release time and fewer triggering events, indicating the positive impact of AVs in mitigating traffic congestion while reducing computational resource usage. Compared to continuous backstepping controllers, the proposed ETC achieves near-equivalent stabilization performance with far fewer controller updates, resulting in longer signal release time that reduces driver distraction, which demonstrates great potential for ETC applications in traffic management.

eess.SY↗

arXiv2025

Shared Control for Vehicle Lane-Changing with Uncertain Driver Behaviors

Lane changes are common yet challenging driving maneuvers that require continuous decision-making and dynamic interaction with surrounding vehicles. Relying solely on human drivers for lane-changing can lead to traffic disturbances due to the stochastic nature of human behavior and its variability under different task demands. Such uncertainties may significantly degrade traffic string stability, which is critical for suppressing disturbance propagation and ensuring smooth merging of the lane-changing vehicles. This paper presents a human-automation shared lane-changing control framework that preserves driver authority while allowing automated assistance to achieve stable maneuvers in the presence of driver's behavioral uncertainty. Human driving behavior is modeled as a Markov jump process with transitions driven by task difficulty, providing a tractable representation of stochastic state switching. Based on this model, we first design a nominal stabilizing controller that guarantees stochastic ${L}_2$ string stability under imperfect mode estimation. To further balance performance and automated effort, we then develop a Minimal Intervention Controller (MIC) that retains acceptable stability while limiting automation. Simulations using lane-changing data from the NGSIM dataset verify that the nominal controller reduces speed perturbations and shorten lane-changing time, while the MIC further reduces automated effort and enhances comfort but with moderate stability and efficiency loss. Validations on the TGSIM dataset with SAE Level 2 vehicles show that the MIC enables earlier lane changes than Level 2 control while preserving driver authority with a slight stability compromise. These findings highlight the potential of shared control strategies to balance stability, efficiency, and driver acceptance.

eess.SY↗

arXiv2025

Robust LLM Training Infrastructure at ByteDance

The training scale of large language models (LLMs) has reached tens of thousands of GPUs and is still continuously expanding, enabling faster learning of larger models. Accompanying the expansion of the resource scale is the prevalence of failures (CUDA error, NaN values, job hang, etc.), which poses significant challenges to training stability. Any large-scale LLM training infrastructure should strive for minimal training interruption, efficient fault diagnosis, and effective failure tolerance to enable highly efficient continuous training. This paper presents ByteRobust, a large-scale GPU infrastructure management system tailored for robust and stable training of LLMs. It exploits the uniqueness of LLM training process and gives top priorities to detecting and recovering failures in a routine manner. Leveraging parallelisms and characteristics of LLM training, ByteRobust enables high-capacity fault tolerance, prompt fault demarcation, and localization with an effective data-driven approach, comprehensively ensuring continuous and efficient training of LLM tasks. ByteRobust is deployed on a production GPU platform and achieves 97% ETTR for a three-month training job on 9,600 GPUs.

cs.LG↗

arXiv2025

A 3D Generation Framework from Cross Modality to Parameterized Primitive

Recent advancements in AI-driven 3D model generation have leveraged cross modality, yet generating models with smooth surfaces and minimizing storage overhead remain challenges. This paper introduces a novel multi-stage framework for generating 3D models composed of parameterized primitives, guided by textual and image inputs. In the framework, A model generation algorithm based on parameterized primitives, is proposed, which can identifies the shape features of the model constituent elements, and replace the elements with parameterized primitives with high quality surface. In addition, a corresponding model storage method is proposed, it can ensure the original surface quality of the model, while retaining only the parameters of parameterized primitives. Experiments on virtual scene dataset and real scene dataset demonstrate the effectiveness of our method, achieving a Chamfer Distance of 0.003092, a VIoU of 0.545, a F1-Score of 0.9139 and a NC of 0.8369, with primitive parameter files approximately 6KB in size. Our approach is particularly suitable for rapid prototyping of simple models.

cs.GR↗