Searcharxiv⌕ Search

arXiv subjects

Xi Zhao

Publications and source records attributed to Xi Zhao.

At least 19 recordsLinked to original sources

Arbor: How Search Policy Shapes Agent Behavior in System Optimization

Achieving peak performance on modern accelerators requires coordinated optimization across kernels, operator fusion, KV-cache management, quantization, scheduling, and communication overlap, a process that demands weeks of expert effort per model and hardware platform. We present Arbor, a system that gives autonomous agents an explicit optimization tree to reason over, backtrack through, and deepen with experience, with pluggable search policies, scored bandit traversal, parallel agent swarms, and depth-first sweep, sharing a common harness of containerized isolation, automatic rollback, and end-to-end validation. Arbor generalizes across domains: +40% to +193% on six production inference models across two GPU generations, +42.8% on LLM training, and +41% on recommendation training, where agents autonomously wrote kernel-level and framework-level optimizations spanning thousands of lines of code. In controlled ablations where every search policy receives the same agent, tree, and tools, we find that search structure shapes both the magnitude and the nature of the optimizations. The bandit achieves +30.1% through targeted framework fixes, the sweep +26.4% through deep kernel rewrites, and the swarm +33.4% by compounding independent changes. An agent-driven loop that receives the same harness, tools, and information, but selects its own targets reaches +14%, writing 4x more code per percentage point of improvement than Arbor's structured search policies. We additionally introduce an evaluation framework for autonomous loop-based agents spanning efficiency, grounding, robustness, and depth, designed to characterize agent search behavior across long-running optimization campaigns.

cs.AI↗

CutAgent: Creative-Gene-Driven Planning and Experience-Guided Self-Refinement for High-CTR E-commerce Advertising Video Creation

E-commerce advertising video creation requires both high-level creative planning and reliable multi-modal production. However, existing methods often map product attributes directly to scripts or editing decisions without explicitly modeling advertising strategies, narrative structures, and production styles. They also have limited ability to reuse past production failures. To address these limitations, we propose $\textbf{CutAgent}$, a dual-agent framework that formulates advertising video creation as a Plan--Execute--Self-Evolve process. CutAgent comprises two collaborative agents, $\textbf{GeneAgent}$ and $\textbf{VideoAgent}$, connected by a structured creative gene that bridges product understanding and video production. GeneAgent maps multi-modal product inputs---textual attributes together with product images---to creative genes that specify narrative units and production rules. To train GeneAgent, we construct $\textbf{CutAgent-70K}$ with approximately 70K samples of product attributes/images, reverse-extracted creative genes, cleaned clip pools, and real-world CTR, and apply high-CTR supervised fine-tuning followed by CTR-guided GRPO. Guided by the generated gene, VideoAgent retrieves category-then-gene failure--correction experiences from a long-term memory as constraints for shot-script planning, then coordinates material retrieval, narration, subtitles, music, and timeline rendering. Independently, a reflection loop diagnoses quality defects, revises deficient production decisions, and writes verified corrections back to memory for later retrieval. Experiments on $\textbf{CutAgent-Eval}$ show that CutAgent outperforms representative baselines in e-commerce advertising video creation, achieving the highest online CTR of $\textbf{2.21\%}$.

cs.CV↗

UBASE: An AI Search Engine for Trillion-Scale Vector Data Management at ByteDance

Since 2016, UBASE has been the foundation of ByteDance's search infrastructure, scaling to more than 7,000 clusters and 300 PB of indexed data. Driven by the demands of AI workloads, UBASE has evolved from a text search engine into a unified AI search system supporting vector retrieval, lexical matching, and predicate filtering. Its largest deployment indexes nearly one trillion high-dimensional vectors. This scale exposes two central bottlenecks in AI-era retrieval: memory-intensive graph-index construction under sustained ingestion, and the prohibitive cost of keeping vector indexes entirely in memory. UBASE addresses these bottlenecks with two techniques. First, it introduces a quantization-aware vector kernel based on SymRaBitQ, a new symmetric quantization scheme with tight theoretical guarantees that allows index construction to run directly in the quantized space accurately and efficiently without retaining a copy of full-precision vectors. Second, it provides a hybrid storage engine that supports memory-resident, hybrid, and SSD-resident deployments, with fine-grained record-level caching to trade memory for latency under operational control. On large-scale benchmarks, UBASE improves throughput by up to 3x, reduces indexing memory by 80%, and lowers serving cost by up to 99.4% compared with prior systems, while supporting trillion-vector scale, write-heavy or latency-sensitive workloads in production.

cs.DB↗

Beyond Score Accuracy: Examining the Diagnostic Quality of LLM-Generated Structured Assessment in Higher Education

As Large Language Models (LLMs) are increasingly adopted for automated grading and feedback in higher education, their structured outputs, including multi-dimensional rubric scores, detailed feedback comments, and improvement suggestions, create an appearance of thorough analytic evaluation. This study examines whether these outputs deliver what they appear to offer. Using the JorGPT dataset of 3,041 student responses to 50 open-ended computer science questions, scored by both human instructors and three commercial LLMs, we identify three systematic discrepancies between the apparent and actual quality of LLM-generated grading and feedback. The sub-dimension scores are highly correlated (r = 0.82-0.99, VIF up to 45), providing redundant rather than independent diagnostic information. The textual feedback rarely detects student misconceptions (5-7% vs. 15-31% for teachers), functioning as a coverage checklist rather than a diagnostic instrument. The feedback tone remains uniformly positive regardless of response quality, lacking the severity modulation observed in human feedback. Additionally, grading accuracy varies significantly by knowledge domain, with procedural topics most reliable. These findings provide empirically grounded guidance on which aspects of LLM-generated grading and feedback can be relied upon and which require continued human oversight.

cs.CY↗

Sparse and weak-measurement certification of graph-edge entanglement in PXP scar wavepackets

An imperfect many-body revival does not by itself certify the entanglement of the returning state. We give a finite-record protocol for graph-edge localizable entanglement along scar wavepackets of a graph-dressed PXP chain. The target cluster state has nonzero energy variance, so neither an exact target eigenstate nor a dark-state embedding is assumed. Fresh binary probes of the undeformed Hamiltonians Pauli terms have a fully separable explanation, even within the dressed blockade sector. Adding local graph-stabilizer probes makes established entanglement witnesses accessible with simultaneous confidence bounds. For a specified square-root instrument, an imposed worst-case disturbance budget fixes the strength that minimizes the equal-allocation Hoeffding sufficient sampling cost. A local commutator bound accounts for finite-duration ancilla pulses while the Hamiltonian remains active. We test the protocol on chains through twenty spins, under perturbed dynamics, and with independent implementations. At the first twenty-spin return, synthetic weak records certify all nineteen graph edges. Generator, two-color, bounded-weight, and full-group benchmarks separate this localizable resource from genuine multipartite certification. The protocol measures recoverable entanglement in a known scar wavepacket, with explicit calibration and limits on its physical and statistical interpretation.

quant-ph↗

Nondemolition filtering of an embedded cluster-state scar under continuous local monitoring

Identifying a low-entanglement eigenstate inside a many-body spectrum and preserving it during measurement are distinct tasks. We construct an explicit local ring Hamiltonian with an exact cluster-state eigenvector and study continuous monitoring of its stabilizer defects. For arbitrary mixed inputs, the conditional cluster fidelity is the initial target weight divided by the no-observed-click probability. A positive defect-operator gap gives finite-time bounds that hold for noncommuting Hamiltonian dynamics, nonnormal effective generators and imperfect detection. At fixed total monitoring rate, the guaranteed exponent falls inversely with system size; high conditional fidelity does not remove the preparation cost set by the initial overlap. Exact diagonalization up to eleven qubits gives finite-size evidence for a cluster-state outlier in a chaotic spectral background. Independent matrix and trajectory calculations verify the dynamics and a conservative coherent-error bound. This construction specializes established scar embedding and nondemolition verification frameworks, with explicit measurement assumptions, finite-time guarantees and resource limitations.

quant-ph↗

GRACE: Grounded Reasoning via Adapter Composition and Evidence-Aware Calibration for Educational Visual Question Answering

Educational visual question answering, or VQA, requires models to solve curriculum-oriented multiple-choice questions using both language and visual evidence. Compared with conventional open-ended VQA, educational examples often include structured assessment metadata, diagrams or image contexts, and semantically close answer options, creating strong opportunities for question-option shortcuts. We develop and evaluate a parameter-efficient adaptation framework for a frozen multimodal large language model in this setting. We introduce GRACE, Grounded Reasoning via Adapter Composition and Evidence-Aware Calibration, a framework that uses the pedagogical state of each question to specialize lightweight language and vision adaptation. The state combines inference-visible subject, grouped skill, grade, visual-context, question-intent, and option-structure cues. GRACE uses factor-specific prompts and lightweight visual adapters, then applies evidence-aware option calibration to score all candidates under a shared multimodal context. On ScienceQA, GRACE improves a shared-adapter baseline from 90.5 percent to 93.1 percent overall accuracy and from 88.7 percent to 91.2 percent on image-context questions. Removing pedagogical composition, option calibration, or the visual adapter reduces overall accuracy by 1.4, 1.0, and 1.5 points, respectively. These controlled results show that structured educational state is an effective routing signal for parameter-efficient multimodal adaptation.

cs.MM↗

GuideGround: VLM-guided Semantic Understanding and Viewpoint-aware Reasoning for 3D Visual Grounding

3D visual grounding aims to localize the target object in a 3D scene from a natural language query, requiring both fine-grained semantic understanding and viewpoint-dependent spatial reasoning. Existing methods typically formulate semantic understanding as an auxiliary closed-set object classification task and rely on multi-view feature aggregation for viewpoint reasoning, limiting semantic generalization and weakening viewpoint-specific evidence. We observe that vision-language models naturally provide complementary capabilities through open-vocabulary semantic understanding and global scene perception. Based on this insight, we propose GuideGround, a VLM-guided framework that complements rather than replaces task-specific grounding models by leveraging VLMs for semantic enhancement and viewpoint-specific hypothesis verification. Specifically, we replace auxiliary closed-set object classification with VLM-generated object semantic descriptions to enhance semantic understanding. Meanwhile, instead of directly aggregating multi-view representations, we preserve viewpoint-specific grounding hypotheses through per-view grounding and explicitly verify them using VLMs across candidate viewpoints. Extensive experiments on the ReferIt3D benchmark demonstrate that GuideGround consistently outperforms previous state-of-the-art methods. Comprehensive ablation studies further confirm the effectiveness of both the proposed semantic understanding and viewpoint reasoning strategies.

cs.CV↗

Möbius-Guided Diagonal-Gate Compilation with Native Multiqubit Controlled-Phase Gates on Neutral-Atom Processors

Diagonal gates are ubiquitous primitives in quantum algorithms, from phase oracles, hypergraph-state preparation, and multi-control logic to Hamiltonian simulation of spin models and digitized lattice field theories, where Ising interactions and local potential terms are diagonal in the encoded basis. Standard compilers, however, often lower diagonal structure into one- and two-qubit gates before neutral-atom hardware can exploit native Rydberg-mediated multiqubit controlled-phase operations. We propose a Möbius-guided compiler that maps a diagonal phase function to a phase hypergraph via subset-lattice Möbius inversion. The hypergraph retains the support and angle of each many-body phase term, allowing sparse or local high-order structure to be routed as native multiqubit controlled-phase candidates when feasible and decomposed otherwise. The neutral-atom scheduler accounts for atom motion, interaction-zone constraints, blockade feasibility, and error costs, enabling a direct comparison between native high-order execution and decomposed alternatives. Benchmarks against routed ZAP and ZX-calculus baselines show improved estimated success for algorithmic instances with exploitable three- and four-body phase terms, and comparable performance on predominantly two-body instances. These results provide a feasible compilation strategy for more fully exploiting the native capabilities of neutral-atom hardware, using atom reconfigurability and Rydberg-mediated multiqubit phase operations as practical resources for more efficient quantum computation.

quant-ph↗

ZAP: Zoned Architecture and Performant Compiler for Field Programmable Atom Array

The scalability of neutral-atom quantum computing is increasingly limited by a compiler--architecture challenge: logical circuits must be mapped onto dynamically reconfigurable atom arrays while controlling crosstalk, transport overhead, and hardware constraints. To address this problem, we present ZAP, a co-designed zoned architecture and deterministic compiler for field-programmable atom arrays. ZAP partitions the array into storage and entanglement zones and combines hardware-aware ASAP-separate scheduling, look-ahead placement, and conflict-aware routing in a single-pass compilation flow, thereby avoiding the repeated global search used in prior approaches. Evaluated on structured quantum benchmarks and random 3-regular circuits, ZAP consistently delivers multi-order-of-magnitude compilation speedups while maintaining competitive or superior execution quality. Relative to ZAC and PowerMove, ZAP typically reduces compilation time from tens of seconds to below 0.1~s and achieves speedups exceeding 1,000$\times$; relative to Enola, the speedup exceeds 10,000$\times$ on the evaluated suite. ZAP's fidelity gains are most pronounced on structured workloads with irregular connectivity and nonuniform qubit reuse, where its scheduling and placement decisions more effectively suppress crosstalk and limit transport-related loss, while on random circuits it remains competitive and preserves the same scalability advantage. These results show that hardware-structured, non-iterative compilation provides a practical path toward fast, scalable, and noise-aware neutral-atom quantum computing.

quant-ph↗

VisualNeo: Bridging the Gap between Visual Query Interfaces and Graph Query Engines

Visual Graph Query Interfaces (VQIs) empower non-programmers to query graph data by constructing visual queries intuitively. Devising efficient technologies in Graph Query Engines (GQEs) for interactive search and exploration has also been studied for years. However, these two vibrant scientific fields are traditionally independent of each other, causing a vast barrier for users who wish to explore the full-stack operations of graph querying. In this demonstration, we propose a novel VQI system built upon Neo4j called VisualNeo that facilities an efficient subgraph query in large graph databases. VisualNeo inherits several advanced features from recent advanced VQIs, which include the data-driven gui design and canned pattern generation. Additionally, it embodies a database manager module in order that users can connect to generic Neo4j databases. It performs query processing through the Neo4j driver and provides an aesthetic query result exploration.

cs.DB↗

KD-CVG: A Knowledge-Driven Approach for Creative Video Generation

Creative Generation (CG) leverages generative models to automatically produce advertising content that highlights product features, and it has been a significant focus of recent research. However, while CG has advanced considerably, most efforts have concentrated on generating advertising text and images, leaving Creative Video Generation (CVG) relatively underexplored. This gap is largely due to two major challenges faced by Text-to-Video (T2V) models: (a) \textbf{ambiguous semantic alignment}, where models struggle to accurately correlate product selling points with creative video content, and (b) \textbf{inadequate motion adaptability}, resulting in unrealistic movements and distortions. To address these challenges, we develop a comprehensive Advertising Creative Knowledge Base (ACKB) as a foundational resource and propose a knowledge-driven approach (KD-CVG) to overcome the knowledge limitations of existing models. KD-CVG consists of two primary modules: Semantic-Aware Retrieval (SAR) and Multimodal Knowledge Reference (MKR). SAR utilizes the semantic awareness of graph attention networks and reinforcement learning feedback to enhance the model's comprehension of the connections between selling points and creative videos. Building on this, MKR incorporates semantic and motion priors into the T2V model to address existing knowledge gaps. Extensive experiments have demonstrated KD-CVG's superior performance in achieving semantic alignment and motion adaptability, validating its effectiveness over other state-of-the-art methods. The code and dataset will be open source at https://kdcvg.github.io/KDCVG/.

cs.CV↗

GTSR: Subsurface Scattering Awared 3D Gaussians for Translucent Surface Reconstruction

Reconstructing translucent objects from multi-view images is a difficult problem. Previously, researchers have used differentiable path tracing and the neural implicit field, which require relatively large computational costs. Recently, many works have achieved good reconstruction results for opaque objects based on a 3DGS pipeline with much higher efficiency. However, such methods have difficulty dealing with translucent objects, because they do not consider the optical properties of translucent objects. In this paper, we propose a novel 3DGS-based pipeline (GTSR) to reconstruct the surface geometry of translucent objects. GTSR combines two sets of Gaussians, surface and interior Gaussians, which are used to model the surface and scattering color when lights pass translucent objects. To render the appearance of translucent objects, we introduce a method that uses the Fresnel term to blend two sets of Gaussians. Furthermore, to improve the reconstructed details of non-contour areas, we introduce the Disney BSDF model with deferred rendering to enhance constraints of the normal and depth. Experimental results demonstrate that our method outperforms baseline reconstruction methods on the NeuralTO Syn dataset while showing great real-time rendering performance. We also extend the dataset with new translucent objects of varying material properties and demonstrate our method can adapt to different translucent materials.

cs.CV↗

GEM: A Native Graph-based Index for Multi-Vector Retrieval

In multi-vector retrieval, both queries and data are represented as sets of high-dimensional vectors, enabling finer-grained semantic matching and improving retrieval quality over single-vector approaches. However, its practical adoption is held back by the lack of effective indexing algorithms. Existing work, attempting to reuse standard single-vector indexes, often fails to preserve multi-vector semantics or remains slow. In this work, we present GEM, a native indexing framework for multi-vector representations. The core idea is to construct a proximity graph directly over vector sets, preserving their fine-grained semantics while enabling efficient navigation. First, GEM designs a set-level clustering scheme. It associates each vector set with only its most informative clusters, effectively reducing redundancy without hurting semantic coverage. Then, it builds local proximity graphs within clusters and bridges them into a globally navigable structure. To handle the non-metric nature of multi-vector similarity, GEM decouples the graph construction metric from the final relevance score and injects semantic shortcuts to guide efficient navigation toward relevant regions. At query time, GEM launches beam search from multiple entry points and prunes paths early using cluster cues. To further enhance efficiency, a quantized distance estimation technique is used for both indexing and search. Across in-domain, out-of-domain, and multi-modal benchmarks, GEM achieves up to 16x speedup over state-of-the-art methods while matching or improving accuracy.

cs.IR↗

Fueling Dynamics towards Tunable Liquid Metal Machine

Self-propelled liquid metal-aluminum hybrid machines represent a promising class of autonomous motion systems capable of sustained movement without external power sources. While interactions between machines and their environment inevitably occur, the fundamental question of how spatial confinement affects the motion dynamics and the controllability of speed, direction, and lifetime of such liquid metal machines (LMMs) remains underexplored. Understanding these confined dynamics is essential for practical applications. Here, we present a comprehensive investigation of the non-symmetrical fueling principle governing the direction-tuning effect in LMMs. By confining LMMs within one-dimensional semi-open channels, we thoroughly disclose their impact and turning dynamics with different end obstacles throughout their lifecycle, with particular focus on fuel region morphological evolution, overall motion, and local flow characteristics after reaction times exceeding one hour. Utilizing ultra-high-speed imaging techniques, we systematically clarify how fuel region evolution and end-obstacle interactions influence symmetry-breaking mechanisms and reciprocating dynamics. Our findings reveal complex interactions between material properties, charge transfer processes, and fluid dynamics during end-turning processes, establishing a theoretical foundation for LMM driving dynamics. Beyond the theoretical mechanisms, we further demonstrate that LMM exhibits efficient heat and mass transfer capabilities, paving the way for applications in controlled transport systems and autonomous robotics.

physics.app-ph↗

MobiFlow: Real-World Mobile Agent Benchmarking through Trajectory Fusion

Mobile agents can autonomously complete user-assigned tasks through GUI interactions. However, existing mainstream evaluation benchmarks, such as AndroidWorld, operate by connecting to a system-level Android emulator and provide evaluation signals based on the state of system resources. In real-world mobile-agent scenarios, however, many third-party applications do not expose system-level APIs to determine whether a task has succeeded, leading to a mismatch between benchmarks and real-world usage and making it difficult to evaluate model performance accurately. To address these issues, we propose MobiFlow, an evaluation framework built on tasks drawn from arbitrary third-party applications. Using an efficient graph-construction algorithm based on multi-trajectory fusion, MobiFlow can effectively compress the state space, support dynamic interaction, and better align with real-world third-party application scenarios. MobiFlow covers 20 widely used third-party applications and comprises 240 diverse real-world tasks, with enriched evaluation metrics. Compared with AndroidWorld, MobiFlow's evaluation results show higher alignment with human assessments and can guide the training of future GUI-based models under real workloads.

cs.AI↗

Reasoning Palette: Modulating Reasoning via Latent Contextualization for Controllable Exploration for (V)LMs

Exploration capacity shapes both inference-time performance and reinforcement learning (RL) training for large (vision-) language models, as stochastic sampling often yields redundant reasoning paths with little high-level diversity. This paper proposes Reasoning Palette, a novel latent-modulation framework that endows the model with a stochastic latent variable for strategic contextualization, guiding its internal planning prior to token generation. This latent context is inferred from the mean-pooled embedding of a question-answer pair via a variational autoencoder (VAE), where each sampled latent potentially encodes a distinct reasoning context. During inference, a sampled latent is decoded into learnable token prefixes and prepended to the input prompt, modulating the model's internal reasoning trajectory. In this way, the model performs internal sampling over reasoning strategies prior to output generation, which shapes the style and structure of the entire response sequence. A brief supervised fine-tuning (SFT) warm-up phase allows the model to adapt to this latent conditioning. Within RL optimization, Reasoning Palette facilitates structured exploration by enabling on-demand injection for diverse reasoning modes, significantly enhancing exploration efficiency and sustained learning capability. Experiments across multiple reasoning benchmarks demonstrate that our method enables interpretable and controllable control over the (vision-) language model's strategic behavior, thereby achieving consistent performance gains over standard RL methods.

cs.CV↗

Beyond Training: Enabling Self-Evolution of Agents with MOBIMEM

Large Language Model (LLM) agents are increasingly deployed to automate complex workflows in mobile and desktop environments. However, current model-centric agent architectures struggle to self-evolve post-deployment: improving personalization, capability, and efficiency typically requires continuous model retraining/fine-tuning, which incurs prohibitive computational overheads and suffers from an inherent trade-off between model accuracy and inference efficiency. To enable iterative self-evolution without model retraining, we propose MOBIMEM, a memory-centric agent system. MOBIMEM first introduces three specialized memory primitives to decouple agent evolution from model weights: (1) Profile Memory uses a lightweight distance-graph (DisGraph) structure to align with user preferences, resolving the accuracy-latency trade-off in user profile retrieval; (2) Experience Memory employs multi-level templates to instantiate execution logic for new tasks, ensuring capability generalization; and (3) Action Memory records fine-grained interaction sequences, reducing the reliance on expensive model inference. Building upon this memory architecture, MOBIMEM further integrates a suite of OS-inspired services to orchestrate execution: a scheduler that coordinates parallel sub-task execution and memory operations; an agent record-and-replay (AgentRR) mechanism that enables safe and efficient action reuse; and a context-aware exception handling that ensures graceful recovery from user interruptions and runtime errors. Evaluation on AndroidWorld and top-50 apps shows that MOBIMEM achieves 83.1% profile alignment with 23.83 ms retrieval time (280x faster than GraphRAG baselines), improves task success rates by up to 50.3%, and reduces end-to-end latency by up to 9x on mobile devices.

cs.AI↗