Searcharxiv⌕ Search

arXiv subjects

Zhao Liu

Publications and source records attributed to Zhao Liu.

At least 19 recordsLinked to original sources

FLARE: A Full-Lifecycle Dense Supervision Paradigm for Long-Horizon Coding Agents via Generative Reward Model

While test-time scaling enhances Large Language Model (LLM) agents in long-horizon software engineering (SWE), sparse binary rewards (Pass/Fail) create a severe credit assignment crisis and waste failed exploratory trajectories. Current trajectory optimization and scaling methods are costly and structurally limited, relying on heuristic state reuse without causal diagnosis or delayed scalar scoring without actionable online guidance. We propose FLARE (Full-Lifecycle Alignment and Reward Engine), a novel dense supervision paradigm driven by a lightweight Generative Reward Model (GRM). First, RADAR, an offline causal-aware diagnostic framework, extracts high-fidelity, hindsight-free supervision through causal-chain backtracking to distill a GRM providing real-time, step-level risk feedback. Second, FLARE uses this GRM to continuously optimize the agent across its entire lifecycle. During inference, FLARE acts as an Active Scaffold, autonomously intercepting high-risk generation steps for localized breakpoint re-execution, drastically reducing compute overhead. During post-training, the GRM's structured signals serve as process-supervised reranking scores for Supervised Fine-Tuning (SFT) and step-level dense rewards for Reinforcement Learning (RL), mitigating policy collapse in sparse environments. Extensive evaluations show that FLARE establishes a new Pareto frontier across the agent lifecycle: FLARE (N=1) outperforms Global Rollout (N=5) with a 5x reduction in token consumption. Extending FLARE to training overcomes the sparse reward problem in long-horizon interactive tasks, delivering relative performance gains of 19.13% in SFT through process-aware data curation and a consistent 9.19% improvement in RL.

cs.CL↗

ANCHOR: Assessing Naturalistic Constraint Hierarchies and Ordering in Repository-Grounded Agentic Coding

For repository-grounded coding agents, a functionally correct patch is not necessarily an acceptable one: developers also constrain how the task must be carried out-modification scope, workflow order, validation, tool use, and reporting. This capability, which we call engineering constraint following, remains largely invisible to current evaluations. We introduce ANCHOR (Assessing Naturalistic Constraint Hierarchies and Ordering in Repository-Grounded Agentic Coding), a benchmark built on naturalistic, multilingual repository tasks spanning eight language groups. ANCHOR combines (i) a hierarchical taxonomy of 10 categories and 78 leaf types distilled from real developer interactions, (ii) an automated pipeline that instantiates task-grounded atomic constraints and composes them into typed AND/THEN/PRIORITY relations, and (iii) a trajectory-grounded evaluation framework that checks both the code diff and the execution trajectory with rule-based and LLM judges. Across 19 agent configurations spanning 2 scaffolds and 10 foundation models on 358 tasks, the strongest measured system, Claude Opus 4.8 under Cline, reaches 70.1% atomic compliance and 63.7% relation compliance, while 19.0% of tasks pass every scored checklist item under the strict all-item FCCR criterion. A five-engineer audit of constraints from 50 instances reports 94% feasibility, 94% repository consistency, and 96% realism. In the primary paired set (14 configurations), the mean signed change in task completion after adding constraints is +0.7 percentage points; individual changes vary in both directions. These results distinguish functional success from constraint compliance, with failures concentrated in workflow ordering, tool-use discipline, and reporting, and substantial interactions between models and scaffolds.

cs.CL↗

Hidden Coherence and Controlled Quasihole Motion in a Fermionic Laughlin State on a Superconducting Quantum Processor

Fractional quantum Hall (FQH) states support fractionalized excitations and geometric phases, yet experimentally resolving the many-body coherence encoded in their wave functions remains challenging. Here, we prepare a three-electron $ν=1/3$ Laughlin state on the Haldane sphere using seven superconducting qubits and reconstruct its many-body density matrix. Through Read's composite-boson construction, we reveal vortex-dressed coherence that remains appreciable over a larger separation range than electron coherence. The extended profile is suppressed in a Fock-diagonal reference constructed from the measured populations, demonstrating its dependence on off-diagonal many-body coherence. We further prepare a one-flux-added quasihole and characterize its fractional charge response against finite-size benchmarks. Controlled quasihole motion yields an effective geometric-phase signature from the reconstructed dominant-eigenstate trajectory. These complementary measurements provide microscopic access to the coherent structure and quasihole response of a finite fermionic Laughlin state, demonstrating an experimental route to probing strongly correlated FQH wave functions on programmable quantum processors.

quant-ph↗

Finite-momentum superconductivity with singlet-triplet mixing in an altermagnetic metal: A pairing instability analysis

We analyze the pairing instability of an altermagnetic metal on a square lattice driven by an attractive nearest-neighbor interaction. This interaction enables multiple pairing channels, including even-parity extended $s$-wave and $d$-wave states, as well as two odd-parity $p$-wave channels. We verify that altermagnetic spin-splitting in the single-particle dispersion gives rise to finite-momentum pairing between electrons with unlike spins, in agreement with earlier predictions. Quite unexpectedly, this pairing typically emerges across multiple channels with mixed parity. Consequently, the resulting finite-momentum Fulde-Ferrell-Larkin-Ovchinnikov (FFLO) superconducting phase is expected to exhibit a multi-component order parameter featuring singlet-triplet mixing. We examine several forms of altermagnetism, specifically $d_{xy}$-wave and $d_{x^{2}-y^{2}}$-wave altermagnetic couplings, and present the corresponding phase diagrams. Additionally, we study triplet pairing between electrons with identical spins and find that it always occurs at zero center-of-mass momentum. Although it is unfavorable in the regime of weak altermagnetic coupling and low electron filling, it can dominate the phase diagram when the altermagnetic coupling is sufficiently strong. The influence of on-site attractive interactions on mixed-parity pairing is also explored.

cond-mat.supr-con↗

From Latents to Wires: Surgical Post-Editing on Large Language Models

Given a large language model (LLM), can whoever holds the weights name a semantic target (e.g., the model's identity), locate the model components that produce it, and edit them so that the target no longer appears while other capability is preserved? We call such an edit on a trained model a post-edit. We present L2W (latents to wires), a framework that performs surgical post-edits for named semantic targets. For localization, L2W uses Jacobian lens (J-lens) attribution to score components against the semantic target. For surgical removal, because LLM mechanisms are redundant (i.e., a semantic target may have multiple components producing it), L2W runs Counterexample-Guided Causal Cut (CGCC) until the target no longer appears. CGCC first cumulatively closes model components, treating each surviving expression of the target as a counterexample that exposes the next components to close, and then reopens some of them to preserve capability. In a controlled experiment with an implanted behavioural watermark, L2W removes the watermark, and its localization lands on the model region the implant changed. Across three model configurations, L2W removes model-metadata (e.g., identity) self-claims in all nine runs, and adult-content refusal in all three, with no held-out target residual. L2W further composes two post-edits on a text-to-image model: one removes the refusal of requested nudity, and a second removes the nude rendering the first exposes. The results support post-editing as a complement to post-training: post-training installs preferred behaviours, and post-editing removes named unwanted ones.

cs.AI↗

Higher order anharmonicity and polymorphism in the lattice thermal conductivity of transition metal dichalcogenides

We present a comprehensive {\it ab initio} study of the effect of higher-order anharmonicity and polymorphism on the lattice thermal conductivity, ${\bm κ}$, of single-layer transition metal dichalcogenides. We show that four-phonon processes are responsible for an additional suppression of ${\bm κ}$ ranging from 15\% to 35\% in the ground-state 2H phase. Conversely, their effect is almost negligible on the thermodynamically competitive \1Tp\ phase. The ${\bm κ}$ difference between the 2H higher-conductivity phase and the \1Tp\ lower-conductivity phase can be exploited for the design of thermal switches that may find applications in thermal management and in the development of phonon-based logic elements, particularly in the case of tellurides, where field-controlled ultrafast and reversible phase transitions have already been demonstrated experimentally.

cond-mat.mtrl-sci↗

Altermagnetism and Superconductivity: A Short Historical Review

This review is organized into three parts. In the first part, we explore the deep interconnections among three seemingly unrelated concepts in condensed matter physics: electronic liquid crystal phases, multipole expansions, and altermagnetism. At the heart of these phenomena lies a shared foundation: spin-momentum locking in the nonrelativistic regime. Originally proposed in the context of electronic liquid crystal phases, nonrelativistic spin-momentum locking was later elegantly incorporated into the formalism of multipole expansions. This framework can be further extended across multiple atomic sites, making it particularly effective for describing altermagnets, which host localized magnetic moments with anisotropic magnetization densities distributed over sublattices. In the second part, we examine superconducting phenomena associated with altermagnetism from three complementary perspectives. First, we investigate superconductivity associated with nonrelativistic spin-momentum locked Fermi surfaces, the unifying theme of the first part, highlighting a rich variety of unconventional superconducting states. These include finite-momentum pairing, $d$-wave and spin-triplet superconductivity, and topological Bogoliubov Fermi surfaces, among others. We then review superconductivity emerging from either static altermagnetic order or altermagnetic fluctuations. Finally, we discuss the possible competition and intertwining between altermagnetic order and superconductivity, illustrated using the repulsive Hubbard model. Additional related topics are addressed in the concluding part. Overall, this work offers both an accessible introduction to the newly identified magnetic order known as altermagnetism and a conceptual guide for researchers aiming to harness the ensuing unconventional superconductivity in the development of future quantum technologies.

cond-mat.supr-con↗

Topological Order Without Band Topology in Moiré Graphene

The discovery of zero-field fractional Chern insulators (FCIs) in moiré materials has attracted intense interest in the interplay between topology and correlations. Here, we demonstrate that fractionalized topological order can emerge under realistic conditions even within a topologically trivial moiré band. By projecting long-range Coulomb interactions into a trivial band of twisted multilayer graphene, we identify a set of incompressible FCI ground states exhibiting fractional quantized Hall conductance. Their Laughlin-like behavior is further confirmed through the particle-cut entanglement spectrum. We trace the origin of this phase to the strongly inhomogeneous distribution of quantum geometry within the moiré Brillouin zone, which reshapes interaction effects independently of the band topology. Extending this heuristic quantum geometric mechanism, we demonstrate that similarly unexpected Laughlin-like FCIs can also be stabilized in higher-Chern-number moiré bands under experimentally accessible conditions. Our results establish realistic scenarios under which many-body topological order can emerge independently of single-particle band topology.

cond-mat.mes-hall↗

StreamSoccer: Event-Driven Memory for Streaming Soccer Commentary

Streaming video understanding requires models to causally update state as video arrives and organize growing history into semantic units that can evolve, persist, and be recalled under bounded computation and memory. This challenge is pronounced in live soccer commentary, where a system must describe completed events, summarize recent play, recall earlier events, or remain silent using only information available before each utterance. We present StreamSoccer, an event-driven system that uses event memory as its intermediate representation. A fixed-budget active memory integrates the stream; completed event states are retained locally and consolidated into retrievable historical records. A unified generator uses current, recent, and historical context to produce three commentary modes, while a rule-assisted scheduler selects a mode or silence. Unlike streaming video-language models organized around frames, visual tokens, or caches, and soccer-commentary methods based on predefined clips or output timestamps, StreamSoccer explicitly models event lifecycles. We construct a three-track streaming soccer commentary dataset and a layered evaluation protocol. At common reference anchors, StreamSoccer obtains CIDEr scores of 38.62, 23.96, and 17.39 for current-event, recent-window, and historical-memory commentary, ranking first on the current-event and historical-memory tracks and second on recent-window. Controlled ablations show that local completed events improve all tracks and that the full system performs best on all three. Across 174 raw-video runs on 58 matches, per-minute RTF p95 ranges from 0.10 to 0.22 without sustained growth with match history. These results indicate that event memory supports streaming soccer commentary across temporal scopes while controlling long-history computation.

cs.CV↗

Robust Privacy: Inference-Stage Privacy through Certified Robustness

An adversary observing a model's released prediction can infer sensitive attributes of the queried input, or even reconstruct representatives of the model's training data. The inference interface thus acts as a side channel for privacy leakage. We introduce Robust Privacy (RP), an inference-stage privacy notion inspired by certified robustness: if a model's prediction is provably invariant within a radius-$R$ neighborhood around an input $x$ with confidence at least $1-α$, then $x$ enjoys $(R,α)$-Robust Privacy, under which we prove that any adversary observing the released prediction has at most $α/2$ advantage in distinguishing $x$ from any input within distance $R$ of $x$. Building on RP, we formalize Robust Attribute Privacy (RAP), an attribute-level privacy notion that characterizes the set of sensitive-attribute values that remain compatible with a released prediction. On a classification task, RP increases the median length of the RAP-compatible inference interval from $23.50$ to $29.96$, reducing attribute-inference precision. Model inversion attacks, often treated as a training-stage threat, in fact rely on fine-grained input-output dependence signals leaked through the inference interface; RP masks these signals at the inference stage, reducing attack success rate (ASR) from $73\%$ to $4\%$ on a black-box inversion attack. This direct targeting of the leakage channel enables RP to dominate DP-SGD and randomized response in the privacy-utility tradeoff space: RP retains $98.4\%$ accuracy at $21\%$ ASR, whereas DP-SGD must drop accuracy to $61.7\%$ to reach a comparable ASR. Across both experiments, increasing the smoothing sample size $N$ at fixed noise scale strengthens privacy and improves utility together. Finally, we examine model distillation as a scope boundary and show that RP mitigates attribute-level and instance-level inference-stage privacy leakage.

cs.LG↗

DiffGRM: Diffusion-based Generative Recommendation Model

Generative recommendation (GR) is an emerging paradigm that represents each item via a tokenizer as an n-digit semantic ID (SID) and predicts the next item by autoregressively generating its SID conditioned on the user's history. However, two structural properties of SIDs make ARMs ill-suited. First, intra-item consistency: the n digits jointly specify one item, yet the left-to-right causality trains each digit only under its prefix and blocks bidirectional cross-digit evidence, collapsing supervision to a single causal path. Second, inter-digit heterogeneity: digits differ in semantic granularity and predictability, while the uniform next-token objective assigns equal weight to all digits, overtraining easy digits and undertraining hard digits. To address these two issues, we propose DiffGRM, a diffusion-based GR model that replaces the autoregressive decoder with a masked discrete diffusion model (MDM), thereby enabling bidirectional context and any-order parallel generation of SID digits for recommendation. Specifically, we tailor DiffGRM in three aspects: (1) tokenization with Parallel Semantic Encoding (PSE) to decouple digits and balance per-digit information; (2) training with On-policy Coherent Noising (OCN) that prioritizes uncertain digits via coherent masking to concentrate supervision on high-value signals; and (3) inference with Confidence-guided Parallel Denoising (CPD) that fills higher-confidence digits first and generates diverse Top-K candidates. Experiments show consistent gains over strong generative and discriminative recommendation baselines on multiple datasets, improving NDCG@10 by 6.9%-15.5%. Code is available at https://github.com/liuzhao09/DiffGRM.

cs.IR↗

LIBAD: A Multimodal Anomaly Detection Benchmark for Li-Ion Battery Electrode Manufacturing

Multimodal industrial anomaly detection has largely focused on discrete products using strongly correlated RGB and 3D observations, leaving continuous process manufacturing and weakly correlated sensing modalities underexplored. We introduce LIBAD, the first multimodal anomaly detection benchmark for Li-ion battery electrode manufacturing. Collected from real roll-to-roll production lines, LIBAD provides aligned double-sided visible-light imaging, high-resolution X-ray radiography, and inline-compatible low-resolution X-ray radiography. Electrode patches in LIBAD exhibit highly homogeneous material appearance, while defect evidence can be strong in one modality but weak or absent in another, resulting in pronounced cross-modal anomaly inconsistency. Benchmarks of representative methods under the inline-compatible visible-light and low-resolution X-ray setting exhibit limited transferability and consistently high false-positive rates. We therefore propose DA-Core, a memory-based method that jointly considers feature-space coverage and local density of normal features during coreset selection, allowing compact memory banks to better preserve fine-grained normal variations. With a coreset ratio of 0.05, DA-Core reduces FPR95 from 60.4% to 54.3% compared with standard farthest point sampling. At this ratio, DA-Core also outperforms the best standard coreset result (obtained at 0.20) while reducing inference time by 43.9%. These results suggest that both the data distribution of normal features and the modality relationship itself require explicit consideration when designing anomaly detection methods for process manufacturing.

cs.CV↗

Zero-Clustering Geometry in Realistic Fractional Quantum Hall Wave Functions

The clustering pattern of zeros in the ground state of a fractional quantum Hall system is a defining feature of its topological properties. We analyze the geometrical fluctuations of the zeros around individual electrons and propose to use the displacement ratio of the zeros to visualize and measure the distance of a realistic state to a model wave function. The distribution of the zero displacement ratio behaves like an order parameter in the transition from a Laughlin phase to a topologically trivial one. The statistical comparison between quantum Hall states belonging to different Jain sequences leads to a composite fermion fluid description of the $ν= 1/5$ ground state with long-range Coulomb interaction that agrees almost perfectly for as few as $3$-$5$ electrons, overcoming the long-standing difficulties of accommodating the competing liquid and crystal orders at short distances.

cond-mat.str-el↗

DADF: A Distribution-Aware Debiasing Framework for Watch-Time Regression in Recommender Systems

Watch-time predictors in short-video recommender systems can be approximately calibrated by their own scores while still overestimating short observations and underestimating long ones. We study whether this label-space mean shrinkage contains inference-time-predictable residual structure that can be corrected without replacing a mature first-stage model. We propose DADF, a distribution-aware second-stage framework that applies multiplicative correction to a frozen watch-time predictor. DADF stabilizes long-tailed correction targets with group-specific transformations, uses video duration to route specialized correction experts, and incorporates auxiliary engagement representations. Duration is used only to index heterogeneous residual distributions, not treated as the cause of the observed pattern. Experiments on KuaiRec and WeChat21 with seven first-stage backbones, together with a large-scale industrial ranking system, show that DADF reduces offline MAE by 4.33% and improves XAUC by 4.01% on average. In production, it reduces MAE by 12.57%. Three online A/B tests across full ranking, rough ranking, and degraded serving improve average time spent per device by 0.649%, 0.235%, and 0.199%, respectively, and all three integrations were subsequently deployed to 100% of traffic. These results show that DADF is a practical, model-agnostic plug-in for correcting predictable conditional residuals while preserving the serving interface of mature first-stage models. Code is available at https://github.com/liuzhao09/DADF.

cs.IR↗

Multi-Decoder OneRec: Controllable Generative Retrieval for Multi-Objective Industrial Recommendation

Industrial recommender systems build candidate pools by assigning explicit quotas to objective-specific retrieval routes. This design offers quota control but increasingly fragments modeling, training, and serving as the route set grows. Semantic-ID-based generative retrieval provides a unified alternative, yet a single decoder entangles objective policies and limits candidate complementarity. We propose Multi-Decoder OneRec, a controllable framework that combines shared representations, isolated objective adaptation, and coordinated decoding. All objectives share a user-context module and the General Decoder, while each objective adds an isolated, parameter-efficient LoRA expert. During training, exposure-sample next-token prediction (NTP) updates the shared base, target-filtered NTP updates the event-based experts, and Kullback-Leibler (KL)-regularized policy optimization updates the Watch-time expert; gradient routing isolates these updates, and the General Decoder supplies a stop-gradient reference. At inference, explicit route quotas allocate the fixed budget and Multi-Decoder Constrained Beam Search reduces cross-route overlap. We publicly release Kwai26, a large-scale multi-objective benchmark with 1.31 billion raw item-level records, 31.85 million Item-ID entries, and 25.03 million items with valid Semantic IDs, together with predefined splits and an evaluation protocol. Under the same 512-item retrieval budget, Multi-Decoder OneRec improves over the single-decoder OneRec baseline by 1.69%-5.62% across four Recall@512 metrics. In a production A/B test, it yields relative gains of 0.37% in app usage time per device, 0.19% in Day-7 retained users, 0.19% in devices with at least one share, and 2.09% in new-content Cold-Start. These results show that generative retrieval can combine shared modeling with objective-specific control and complementary candidate generation.

cs.IR↗

CAPTS: Channel-Aware, Preference-Aligned Trigger Selection for Multi-Channel Item-to-Item Retrieval

Large-scale industrial recommender systems commonly adopt multi-channel retrieval for candidate generation, combining direct user-to-item (U2I) retrieval with two-hop user-to-item-to-item (U2I2I) pipelines. In U2I2I, the system selects a small set of historical interactions as triggers to seed downstream item-to-item (I2I) retrieval across multiple channels. In production, triggers are often selected using rule-based policies or learned scorers and tuned in a channel-by-channel manner. However, these practices face two persistent challenges: biased value attribution that values triggers by on-trigger feedback rather than their downstream utility as retrieval seeds, and uncoordinated multi-channel routing where channels select triggers independently under a shared quota, increasing cross-channel overlap. To address these challenges, we propose Channel-Aware, Preference-Aligned Trigger Selection (CAPTS), a unified and flexible framework that treats multi-channel trigger selection as a learnable routing problem. CAPTS introduces a Value Attribution Module (VAM) that provides look-ahead supervision by crediting each trigger with the subsequent engagement generated by items retrieved from it on each I2I channel, and a Channel-Adaptive Trigger Routing (CATR) module that coordinates trigger-to-channel assignment to maximize the overall value of multi-channel retrieval. Extensive offline experiments and large-scale online A/B tests on Kwai, Kuaishou's international short-video platform, show that CAPTS consistently improves multi-channel recall offline and delivers a +0.351% lift in average time spent per device online.

cs.IR↗

Dynamic Defense Profiling Enables Cognitive Jailbreak of Text-to-Image Models

Text-to-Image (T2I) generative models have achieved remarkable progress in synthesizing high-quality visual content, yet they remain vulnerable to adversarial misuse, particularly in generating Not-Safe-For-Work (NSFW) images. Most existing jailbreak attacks primarily rely on heuristic prompt engineering or black-box optimization, treating model feedback as a binary signal (success or failure). This coarse-grained paradigm overlooks the rich information embedded in diverse failure modes, such as textual refusal, visual blocking, and semantic sanitization, resulting in inefficient exploration and severe semantic collapse. In this paper, we propose MIND, a cognitive jailbreak framework that reframes adversarial prompt generation as a belief-state inference problem over latent defense mechanisms. Instead of blindly searching for bypass prompts, MIND actively models the target system's latent defense mechanisms by interpreting multi-modal feedback as high-density signals. Specifically, the framework integrates three core components: (1) a Multi-modal Judge for fine-grained feedback decomposition, (2) a Defense Profiler for iterative belief updating, and (3) a Meta-Memory module for retrieving historically effective attack strategies. These components are unified within a reasoning-driven evolutionary optimization process, enabling adaptive and semantically consistent jailbreak generation. Extensive experiments on the I2P benchmark demonstrate the effectiveness of MIND. Under six representative pre-processing and post-processing defense settings applied to the Stable Diffusion v1.5 model, MIND achieves an Attack Success Rate (ASR) of 95.62%, significantly outperforming existing methods. Additionally, the effectiveness of the proposed framework is validated across four widely used commercial T2I systems, achieving the highest ASR of 91.58% on Wan-2.5.

cs.AI↗

RECAP: Feedback-Driven Streaming Semantic User Profiles for Short-Video Recommendation

Language-based user profiles convert long behavioral histories into explicit semantic representations for recommendation. However, most profile generators are optimized in an open loop: they may summarize past behavior fluently, but are not directly trained to improve future recommendation. We study this problem in real-world short-video recommendation, where user behaviors continuously arrive as streams and profiles must be incrementally updated under limited capacity. This requires maintaining a consistent bounded profile state and constructing profile-targeted semantic feedback from industrial implicit behavior logs. We propose RECAP, an offline closed-loop framework for optimizing streaming structured semantic profiles with historical recommendation feedback. RECAP maintains each profile as a bounded structured memory by combining LLM-based semantic updates with deterministic lifecycle and capacity control. RECAP constructs profile-targeted semantic feedback by filtering label-consistent behavior pairs with an LLM judge and training a dual-tower evaluator whose matching score serves as a GRPO reward. Experiments on Kuaishou short-video data show that RECAP improves uAUC by 0.0084 and Recall@2000 by about 4.9% over the base generator. Further analyses confirm the benefits of feedback construction and policy optimization, and show more grounded refinement and user-level abstraction in profile updates. A seven-day online A/B test further shows a statistically significant 0.139% improvement in average application usage time per user.

cs.IR↗