arXiv subjects

Cheng Chen

Publications and source records attributed to Cheng Chen.

At least 127 records · Page 7Linked to original sources

arXiv2025

AgentEvolver: Towards Efficient Self-Evolving Agent System

Autonomous agents powered by large language models (LLMs) have the potential to significantly enhance human productivity by reasoning, using tools, and executing complex tasks in diverse environments. However, current approaches to developing such agents remain costly and inefficient, as they typically require manually constructed task datasets and reinforcement learning (RL) pipelines with extensive random exploration. These limitations lead to prohibitively high data-construction costs, low exploration efficiency, and poor sample utilization. To address these challenges, we present AgentEvolver, a self-evolving agent system that leverages the semantic understanding and reasoning capabilities of LLMs to drive autonomous agent learning. AgentEvolver introduces three synergistic mechanisms: (i) self-questioning, which enables curiosity-driven task generation in novel environments, reducing dependence on handcrafted datasets; (ii) self-navigating, which improves exploration efficiency through experience reuse and hybrid policy guidance; and (iii) self-attributing, which enhances sample efficiency by assigning differentiated rewards to trajectory states and actions based on their contribution. By integrating these mechanisms into a unified framework, AgentEvolver enables scalable, cost-effective, and continual improvement of agent capabilities. Preliminary experiments indicate that AgentEvolver achieves more efficient exploration, better sample utilization, and faster adaptation compared to traditional RL-based baselines.

cs.LG↗

arXiv2025

HyMiRec: A Hybrid Multi-interest Learning Framework for LLM-based Sequential Recommendation

Large language models (LLMs) have recently demonstrated strong potential for sequential recommendation. However, current LLM-based approaches face critical limitations in modeling users' long-term and diverse interests. First, due to inference latency and feature fetching bandwidth constraints, existing methods typically truncate user behavior sequences to include only the most recent interactions, resulting in the loss of valuable long-range preference signals. Second, most current methods rely on next-item prediction with a single predicted embedding, overlooking the multifaceted nature of user interests and limiting recommendation diversity. To address these challenges, we propose HyMiRec, a hybrid multi-interest sequential recommendation framework, which leverages a lightweight recommender to extracts coarse interest embeddings from long user sequences and an LLM-based recommender to captures refined interest embeddings. To alleviate the overhead of fetching features, we introduce a residual codebook based on cosine similarity, enabling efficient compression and reuse of user history embeddings. To model the diverse preferences of users, we design a disentangled multi-interest learning module, which leverages multiple interest queries to learn disentangles multiple interest signals adaptively, allowing the model to capture different facets of user intent. Extensive experiments are conducted on both benchmark datasets and a collected industrial dataset, demonstrating our effectiveness over existing state-of-the-art methods. Furthermore, online A/B testing shows that HyMiRec brings consistent improvements in real-world recommendation systems. Code is available at https://github.com/FireRedTeam/FireRedSeqRec.

cs.IR↗

arXiv2025

Cross-Scenario Unified Modeling of User Interests at Billion Scale

User interests on content platforms are inherently diverse, manifesting through complex behavioral patterns across heterogeneous scenarios such as search, feed browsing, and content discovery. Traditional recommendation systems typically prioritize business metric optimization within isolated specific scenarios, neglecting cross-scenario behavioral signals and struggling to integrate advanced techniques like LLMs at billion-scale deployments, which finally limits their ability to capture holistic user interests across platform touchpoints. We propose RED-Rec, an LLM-enhanced hierarchical Recommender Engine for Diversified scenarios, tailored for industry-level content recommendation systems. RED-Rec unifies user interest representations across multiple behavioral contexts by aggregating and synthesizing actions from varied scenarios, resulting in comprehensive item and user modeling. At its core, a two-tower LLM-powered framework enables nuanced, multifaceted representations with deployment efficiency, and a scenario-aware dense mixing and querying policy effectively fuses diverse behavioral signals to capture cross-scenario user intent patterns and express fine-grained, context-specific intents during serving. We validate RED-Rec through online A/B testing on hundreds of millions of users in RedNote through online A/B testing, showing substantial performance gains in both content recommendation and advertisement targeting tasks. We further introduce a million-scale sequential recommendation dataset, RED-MMU, for comprehensive offline training and evaluation. Our work advances unified user modeling, unlocking deeper personalization and fostering more meaningful user engagement in large-scale UGC platforms.

cs.IR↗

arXiv2025

HoloCine: Holistic Generation of Cinematic Multi-Shot Long Video Narratives

State-of-the-art text-to-video models excel at generating isolated clips but fall short of creating the coherent, multi-shot narratives, which are the essence of storytelling. We bridge this "narrative gap" with HoloCine, a model that generates entire scenes holistically to ensure global consistency from the first shot to the last. Our architecture achieves precise directorial control through a Window Cross-Attention mechanism that localizes text prompts to specific shots, while a Sparse Inter-Shot Self-Attention pattern (dense within shots but sparse between them) ensures the efficiency required for minute-scale generation. Beyond setting a new state-of-the-art in narrative coherence, HoloCine develops remarkable emergent abilities: a persistent memory for characters and scenes, and an intuitive grasp of cinematic techniques. Our work marks a pivotal shift from clip synthesis towards automated filmmaking, making end-to-end cinematic creation a tangible future. Our code is available at: https://holo-cine.github.io/.

cs.CV↗

arXiv2025

Enabling Self-Improving Agents to Learn at Test Time With Human-In-The-Loop Guidance

Large language model (LLM) agents often struggle in environments where rules and required domain knowledge frequently change, such as regulatory compliance and user risk screening. Current approaches, like offline fine-tuning and standard prompting, are insufficient because they cannot effectively adapt to new knowledge during actual operation. To address this limitation, we propose the Adaptive Reflective Interactive Agent (ARIA), an LLM agent framework designed specifically to continuously learn updated domain knowledge at test time. ARIA assesses its own uncertainty through structured self-dialogue, proactively identifying knowledge gaps and requesting targeted explanations or corrections from human experts. It then systematically updates an internal, timestamped knowledge repository with provided human guidance, detecting and resolving conflicting or outdated knowledge through comparisons and clarification queries. We evaluate ARIA on the realistic customer due diligence name screening task on TikTok Pay, alongside publicly available dynamic knowledge tasks. Results demonstrate significant improvements in adaptability and accuracy compared to baselines using standard offline fine-tuning and existing self-improving agents. ARIA is deployed within TikTok Pay serving over 150 million monthly active users, confirming its practicality and effectiveness for operational use in rapidly evolving environments.

cs.LG↗

arXiv2025

AI Knowledge Assist: An Automated Approach for the Creation of Knowledge Bases for Conversational AI Agents

The utilization of conversational AI systems by leveraging Retrieval Augmented Generation (RAG) techniques to solve customer problems has been on the rise with the rapid progress of Large Language Models (LLMs). However, the absence of a company-specific dedicated knowledge base is a major barrier to the integration of conversational AI systems in contact centers. To this end, we introduce AI Knowledge Assist, a system that extracts knowledge in the form of question-answer (QA) pairs from historical customer-agent conversations to automatically build a knowledge base. Fine-tuning a lightweight LLM on internal data demonstrates state-of-the-art performance, outperforming larger closed-source LLMs. More specifically, empirical evaluation on 20 companies demonstrates that the proposed AI Knowledge Assist system that leverages the LLaMA-3.1-8B model eliminates the cold-start gap in contact centers by achieving above 90% accuracy in answering information-seeking questions. This enables immediate deployment of RAG-powered chatbots.

cs.CL↗

arXiv2025

DACIP-RC: Domain Adaptive Continual Instruction Pre-Training via Reading Comprehension on Business Conversations

The rapid advancements in Large Language Models (LLMs) have enabled their adoption in real-world industrial scenarios for various natural language processing tasks. However, the high inference cost of large-scale LLMs makes their deployment impractical, necessitating the use of smaller models. Despite their efficiency, smaller LLMs lack robust zero-shot instruction-following capabilities across diverse domains, limiting their adaptability to dynamic user requirements. Traditional fine-tuning approaches exacerbate this issue by inducing catastrophic forgetting, reducing the model's generalization ability for unseen tasks. In this paper, we propose Domain Adaptive Continual Instruction Pre-Training via Reading Comprehension (DACIP-RC), a continual pre-training technique that enhances smaller LLMs' domain adaptability for business conversational tasks. Unlike conventional pre-training approaches that rely on next-token prediction, DACIP-RC generates diverse task instructions and responses via reading comprehension on conversation transcripts, enabling better instruction generalization. Our empirical evaluations demonstrate that DACIP-RC significantly improves zero-shot generalization across a wide range of business conversational tasks, including meeting summarization, action item generation, and call purpose identification. To the best of our knowledge, this is the first work to apply instruction pre-training on business conversational data, providing insights into how industries can leverage proprietary datasets for domain adaptation.

cs.CL↗

arXiv2025

ElasticMoE: An Efficient Auto Scaling Method for Mixture-of-Experts Models

Mixture-of-Experts (MoE) models promise efficient scaling of large language models (LLMs) by activating only a small subset of experts per token, but their parallelized inference pipelines make elastic serving challenging. Existing strategies fall short: horizontal scaling provisions entire replicas of the current configuration, often tens to hundreds of accelerators, leading to coarse granularity, long provisioning delays, and costly overprovisioning. Vertical scaling offers finer adjustments but typically requires instance restarts, incurring downtime. These limitations make current approaches ill-suited for the bursty, short-lived traffic patterns common in cloud deployments. We present ElasticMoE, an elastic scaling framework for MoE LLMs that achieves fine-grained, low-latency, and zero-downtime scaling. ElasticMoE decouples inference execution from memory operations, enabling scaling steps to proceed concurrently with serving. An HBM Management Module (HMM) reuses weights and KV caches via zero-copy remapping, while high-bandwidth peer-to-peer transfers bring newly added accelerators online without interrupting service. A virtual memory based expert redistribution mechanism migrates MoE experts without costly buffer reallocations, reducing peak memory usage during expert parallelism reconfiguration. Our evaluation on Ascend NPUs with three popular MoE LLMs shows that ElasticMoE achieves up to 9x lower scale-up latency, up to 2x better throughput during scaling, and significantly improves SLO attainment compared to baselines. By enabling fine-grained, concurrent scaling with minimal disruption, ElasticMoE advances the practicality of deploying massive MoE LLMs in dynamic cloud environments.

cs.DC↗

arXiv2025

Early Prediction of In-Hospital ICU Mortality Using Innovative First-Day Data: A Review

The intensive care unit (ICU) manages critically ill patients, many of whom face a high risk of mortality. Early and accurate prediction of in-hospital mortality within the first 24 hours of ICU admission is crucial for timely clinical interventions, resource optimization, and improved patient outcomes. Traditional scoring systems, while useful, often have limitations in predictive accuracy and adaptability. Objective: This review aims to systematically evaluate and benchmark innovative methodologies that leverage data available within the first day of ICU admission for predicting in-hospital mortality. We focus on advancements in machine learning, novel biomarker applications, and the integration of diverse data types.

cs.LG↗

arXiv2025

Non-excitonic mechanism for electronic and structural phase transitions in Ta2Ni(Se,S)5

We present a first-principles study based on density functional theory (DFT) on the electronic and structural properties of Ta2NiSe5, a layered transition metal chalcogenide that has been considered as a possible candidate for an excitonic insulator. Our systematic DFT results however provide a non-excitonic mechanism for the experimentally observed electronic and structural phase transitions in Ta2NiSe5, in particular explaining why sulfur substitution of selenium reduces the distortion angle in the low-temperature phase and potassium dosing closes the gap in the electronic structure. Moreover, the calculations show that these two effects couple to each other. Further, our first-principles calculations predict several changes in both the crystal structure and electronic structure under the effects of uniform charge dosing and uniaxial strain, which could be tested experimentally.

cond-mat.mtrl-sci↗

arXiv2025

MARS2 2025 Challenge on Multimodal Reasoning: Datasets, Methods, Results, Discussion, and Outlook

This paper reviews the MARS2 2025 Challenge on Multimodal Reasoning. We aim to bring together different approaches in multimodal machine learning and LLMs via a large benchmark. We hope it better allows researchers to follow the state-of-the-art in this very dynamic area. Meanwhile, a growing number of testbeds have boosted the evolution of general-purpose large language models. Thus, this year's MARS2 focuses on real-world and specialized scenarios to broaden the multimodal reasoning applications of MLLMs. Our organizing team released two tailored datasets Lens and AdsQA as test sets, which support general reasoning in 12 daily scenarios and domain-specific reasoning in advertisement videos, respectively. We evaluated 40+ baselines that include both generalist MLLMs and task-specific models, and opened up three competition tracks, i.e., Visual Grounding in Real-world Scenarios (VG-RS), Visual Question Answering with Spatial Awareness (VQA-SA), and Visual Reasoning in Creative Advertisement Videos (VR-Ads). Finally, 76 teams from the renowned academic and industrial institutions have registered and 40+ valid submissions (out of 1200+) have been included in our ranking lists. Our datasets, code sets (40+ baselines and 15+ participants' methods), and rankings are publicly available on the MARS2 workshop website and our GitHub organization page https://github.com/mars2workshop/, where our updates and announcements of upcoming events will be continuously provided.

cs.CV↗

arXiv2025

Visualizing Electronic Structure of Twisted Bilayer MoTe2 in Devices

The pursuit of emergent quantum phenomena lies at the forefront of modern condensed matter physics. A recent breakthrough in this arena is the discovery of the fractional quantum anomalous Hall effect (FQAHE) in twisted bilayer MoTe2 (tbMoTe2), marking a paradigm shift and establishing a versatile platform for exploring the intricate interplay among topology, magnetism, and electron correlations. While significant progress has been made through both optical and electrical transport measurements, direct experimental insights into the electronic structure - crucial for understanding and modeling this system - have remained elusive. Here, using spatially and angle-resolved photoemission spectroscopy (μ-ARPES), we directly map the electronic band structure of tbMoTe2. We identify the valence band maximum, whose partial filling underlies the FQAHE, at the K points, situated approximately 150 meV above the Γ valley. By fine-tuning the doping level via in-situ alkali metal deposition, we also resolve the conduction band minimum at the K point, providing direct evidence that tbMoTe2 exhibits a direct band gap - distinct from all previously known moire bilayer transition metal dichalcogenide systems. These results offer critical insights for theoretical modeling and advance our understanding of fractionalized excitations and correlated topological phases in this emergent quantum material.

cond-mat.mes-hall↗

arXiv2025

UniSearch: Rethinking Search System with a Unified Generative Architecture

Modern search systems play a crucial role in facilitating information acquisition. Traditional search engines typically rely on a cascaded architecture, where results are retrieved through recall, pre-ranking, and ranking stages. The complexity of designing and maintaining multiple modules makes it difficult to achieve holistic performance gains. Recent advances in generative recommendation have motivated the exploration of unified generative search as an alternative. However, existing approaches are not genuinely end-to-end: they typically train an item encoder to tokenize candidates first and then optimize a generator separately, leading to objective inconsistency and limited generalization. To address these limitations, we propose UniSearch, a unified generative search framework for Kuaishou Search. UniSearch replaces the cascaded pipeline with an end-to-end architecture that integrates a Search Generator and a Video Encoder. The Generator produces semantic identifiers of relevant items given a user query, while the Video Encoder learns latent item embeddings and provides their tokenized representations. A unified training framework jointly optimizes both components, enabling mutual enhancement and improving representation quality and generation accuracy. Furthermore, we introduce Search Preference Optimization (SPO), which leverages a reward model and real user feedback to better align generation with user preferences. Extensive experiments on industrial-scale datasets, together with online A/B testing in both short-video and live search scenarios, demonstrate the strong effectiveness and deployment potential of UniSearch. Notably, its deployment in live search yields the largest single-experiment improvement in recent years of our product's history, highlighting its practical value for real-world applications.

cs.IR↗

arXiv2025

Evaluating LLMs Without Oracle Feedback: Agentic Annotation Evaluation Through Unsupervised Consistency Signals

Large Language Models (LLMs), when paired with prompt-based tasks, have significantly reduced data annotation costs and reliance on human annotators. However, evaluating the quality of their annotations remains challenging in dynamic, unsupervised environments where oracle feedback is scarce and conventional methods fail. To address this challenge, we propose a novel agentic annotation paradigm, where a student model collaborates with a noisy teacher (the LLM) to assess and refine annotation quality without relying on oracle feedback. The student model, acting as an unsupervised feedback mechanism, employs a user preference-based majority voting strategy to evaluate the consistency of the LLM outputs. To systematically measure the reliability of LLM-generated annotations, we introduce the Consistent and Inconsistent (CAI) Ratio, a novel unsupervised evaluation metric. The CAI Ratio not only quantifies the annotation quality of the noisy teacher under limited user preferences but also plays a critical role in model selection, enabling the identification of robust LLMs in dynamic, unsupervised environments. Applied to ten open-domain NLP datasets across four LLMs, the CAI Ratio demonstrates a strong positive correlation with LLM accuracy, establishing it as an essential tool for unsupervised evaluation and model selection in real-world settings.

cs.CL↗

arXiv2025

Low-frequency interlayer phonon dynamics and photoinduced terahertz absorption in black phosphorus

The strong interlayer coupling in black phosphorus (BP), arising from wavefunction overlap between layers, is critical for understanding its electronic and optical properties. Here, we utilize terahertz (THz) spectroscopy to study phonon dynamics in BP. We identify two peaks at 6 and 8.5 meV in steady-state THz spectra, which are attributed to low-frequency interlayer phonon modes. Both modes exhibit anharmonic phonon coupling behavior below 150 K, manifesting as temperature-dependent red-shifts. Using time-resolved THz spectroscopy, we observe significantly increased THz absorption under photoexcitation, arising from the transient enhancements of the extended Drude component and the oscillator strengths of interlayer phonons in non-equilibrium. These findings highlight the critical role of interlayer phonon dynamics in understanding many-body physics in BP.

cond-mat.str-el↗

arXiv2025

Realization of a doped quantum antiferromagnet with dipolar tunnelings in a Rydberg tweezer array

Doping an antiferromagnetic Mott insulator is central to our understanding of a variety of phenomena in strongly-correlated electrons, including high-temperature superconductors. To describe the competition between tunneling $t$ of hole dopants and antiferromagnetic (AFM) spin interactions $J$, theoretical and numerical studies often focus on the paradigmatic $t$-$J$ model, and the direct analog quantum simulation of this model in the relevant regime of high-particle density has long been sought. Here, we realize a doped quantum antiferromagnet with next-nearest neighbour (NNN) tunnelings $t'$ and hard-core bosonic holes using a Rydberg tweezer platform. We utilize coherent dynamics between three Rydberg levels, encoding spins and holes, to implement a tunable bosonic $t$-$J$-$V$ model allowing us to study previously inaccessible parameter regimes. We observe dynamical phase separation between hole and spin domains for $|t/J|\ll 1$, and demonstrate the formation of repulsively bound hole pairs in a variety of spin backgrounds. The interference between NNN tunnelings $t'$ and perturbative pair tunneling gives rise to light and heavy pairs depending on the sign of $t$. Using the single-site control allows us to study the dynamics of a single hole in 2D square lattice (anti)ferromagnets. The model we implement extends the toolbox of Rydberg tweezer experiments beyond spin-1/2 models to a larger class of $t$-$J$ and spin-$1$ models.

quant-ph↗

arXiv2025

Elucidating Rectified Flow with Deterministic Sampler: Polynomial Discretization Complexity for Multi and One-step Models

Recently, rectified flow (RF)-based models have achieved state-of-the-art performance in many areas for both the multi-step and one-step generation. However, only a few theoretical works analyze the discretization complexity of RF-based models. Existing works either focus on flow-based models with stochastic samplers or establish complexity results that exhibit exponential dependence on problem parameters. In this work, under the realistic bounded support assumption, we prove the first polynomial discretization complexity for multi-step and one-step RF-based models with a deterministic sampler simultaneously. For the multi-step setting, inspired by the predictor-corrector framework of diffusion models, we introduce a Langevin process as a corrector and show that RF-based models can achieve better polynomial discretization complexity than diffusion models. To achieve this result, we conduct a detailed analysis of the RF-based model and explain why it is better than previous popular models, such as variance preserving (VP) and variance exploding (VE)-based models. Based on the observation of multi-step RF-based models, we further provide the first polynomial discretization complexity result for one-step RF-based models, improving upon prior results for one-step diffusion-based models. These findings mark the first step toward theoretically understanding the impressive empirical performance of RF-based models in both multi-step and one-step generation.

cs.LG↗

arXiv2025

Beyond Adapter Retrieval: Latent Geometry-Preserving Composition via Sparse Task Projection

Recent advances in parameter-efficient transfer learning have demonstrated the utility of composing LoRA adapters from libraries of pretrained modules. However, most existing approaches rely on simple retrieval heuristics or uniform averaging, which overlook the latent structure of task relationships in representation space. We propose a new framework for adapter reuse that moves beyond retrieval, formulating adapter composition as a geometry-aware sparse reconstruction problem. Specifically, we represent each task by a latent prototype vector derived from the base model's encoder and aim to approximate the target task prototype as a sparse linear combination of retrieved reference prototypes, under an $\ell_1$-regularized optimization objective. The resulting combination weights are then used to blend the corresponding LoRA adapters, yielding a composite adapter tailored to the target task. This formulation not only preserves the local geometric structure of the task representation manifold, but also promotes interpretability and efficient reuse by selecting a minimal set of relevant adapters. We demonstrate the effectiveness of our approach across multiple domains-including medical image segmentation, medical report generation and image synthesis. Our results highlight the benefit of coupling retrieval with latent geometry-aware optimization for improved zero-shot generalization.

cs.LG↗