arXiv subjects

Kai Liu

Publications and source records attributed to Kai Liu.

At least 19 recordsLinked to original sources

arXiv2026

Hunyuan-A13B Technical Report

We present Hunyuan-A13B, an open-source large language model based on a Mixture-of-Experts architecture. It contains 80 billion total parameters but activates only 13 billion during inference, balancing model capability, computational efficiency, and deployment cost. The model is pretrained on a rigorously filtered 20T-token corpus with enhanced STEM data curation, improving factual reliability and reasoning ability. High-quality supervised fine-tuning and large-scale reinforcement learning further enhance its overall performance. Hunyuan-A13B also introduces a dual-mode Chain-of-Thought framework that adapts reasoning depth to task complexity: fast thinking for routine queries and slow thinking for complex, multi-step problems. Evaluations show competitive performance across mathematics, science, programming, general language understanding, and agent tasks, often approaching that of much larger models. Its high inference throughput makes it suitable for latency-sensitive applications. We release Hunyuan-A13B to support open research and practical LLM deployment.

cs.AI↗

arXiv2026

DualWAM: Dual-System World Action Models for Asynchronous Global Planning and Local Refinement

World Action Models (WAMs) jointly generate robot actions and predict future world states, transferring priors from video pretraining to robot control. However, future visual prediction is computationally expensive, so existing WAMs often rely on long action chunks to amortize inference cost across control steps, at the cost of closed-loop responsiveness. We present \method, a dual-system WAM that preserves broader-horizon world-action generation while enabling high-frequency closed-loop action updates by decoupling global planning and local refinement. \systwo periodically performs high-noise bidirectional denoising over a broader world-action chunk to establish a global plan, while wrist-only \sysone extracts a temporally aligned short window from the intermediate denoising state and completes low-noise refinement using the latest wrist observations, which provide action-aligned cues about local geometry, motion, and contact during interaction. The two systems operate asynchronously along a shared denoising trajectory: each global plan is reused across multiple local updates, while \sysone repeatedly incorporates fresh interaction feedback. Across zero-shot manipulation tasks on Franka and Galbot, \method improves success over the strongest evaluated baseline by 4.5 percentage points on average, while achieving a 16.6$\times$ critical-path speedup. Further studies show that role-matched egocentric and UMI data improve success by 14 percentage points, and that the decoupled design naturally supports edge--cloud deployment with substantially lower communication overhead than the baseline.

cs.RO↗

arXiv2026

Sample-Guided Exact Top-K Selection for Long-Context Sparse Attention

Sparse attention bounds downstream attention work by retaining a fixed-size subset of indexed tokens, but its standalone exact Top-$K$ stage must still process materialized score rows whose length grows with context. Production radix selectors discover their first actionable boundary only after a complete-row pass, forcing another row-scale traversal before exact refinement. We observe that locating a compact upper tail requires substantially less resolution than identifying the exact rank boundary, and that fixed-stride partial views of the current row remain calibrated to the corresponding complete-row rank across ragged lengths. We present HPC-Ops Top-K, a sample-guided exact selector for ragged sparse-attention score rows. A fixed-stride view proposes a row-local coarse boundary; the mandatory complete-row pass certifies its sufficiency, forms the admitted candidate set, and initializes exact FP32 refinement over the unresolved frontier. A nested secondary boundary and exact recovery handle underfilled proposals before any output is committed, so sampling controls common-path work but never correctness. The GPU implementation fuses complete-row certification and candidate formation, and combines persistent, KV-split, and direct-exact execution behind graph-capturable ragged-row dispatch. We evaluate HPC-Ops Top-K on indexer scores from Hy4-Preview. It outperforms the fastest verified external exact baseline by $1.29$--$1.75\times$ across 20 operator configurations, with a $1.55\times$ geometric-mean speedup. It further achieves $1.36\times$ and $1.48\times$ speedups on two framework-derived sparse-attention traces. The implementation is available in HPC-Ops, Tencent's open-source high-performance operator library for LLM inference, at https://github.com/Tencent/hpc-ops.

cs.DC↗

arXiv2026

Reproducible capillary fluctuation analysis of solid-liquid interfaces for stiffness and anisotropy calculations

The capillary fluctuation method (CFM) is widely used to compute solid--liquid interfacial properties from atomistic simulations, but its accuracy depends on choices in interface construction, wave-vector selection, sampling, and simulation geometry. Here, we develop a diagnostics-driven workflow for reproducible CFM calculations using pure Al as a representative system. Employing both ribbon models and thick two-dimensional references, we show that apparent linearity of the fluctuation spectrum alone does not ensure reliable stiffness or anisotropy estimates. Instead, a reliable CFM analysis requires a fitting window consistent with both temporal sampling and continuum capillary-wave assumptions, systematic sensitivity tests of the interface identification procedure, explicit propagation of replica variability, and independent verification of model-thickness convergence. We further propose a practical thickness-selection rule based on coexistence-temperature consistency, which enables finite-size effects to be controlled while retaining the substantial computational efficiency of ribbon geometries. By making the main sources of uncertainty explicit and diagnosable, the proposed workflow improves the reliability of the CFM as a quantitative tool and provides a foundation for its broader application to complex solid--liquid interfaces.

cond-mat.mtrl-sci↗

arXiv2026

Frame phase retrievability and state distinguishability of quantum channels

This survey introduces the role of frame phase retrieval in pure-state identification and information preservation by quantum channels. For unit vectors, the lift $x\mapsto xx^*$ identifies vectors that differ only by a global phase with the same rank-one quantum state, converts frame intensities into linear functionals of the lifted state, and makes the adjoint pullback of output observables an operator-valued measurement on the input system. From this viewpoint, a channel is phase retrievable exactly when it is injective on pure states. We develop this correspondence through Choi-rank-two linear combinations of Kraus operators, higher-rank relative spectra, structural obstructions, and constructions with prescribed Choi rank. We distinguish injective identification from full tomography, perfect one-shot discrimination, zero-error classical communication, and exact quantum correction. Twirling channels then provide a structured setting in which commutants, irreducible dimensions, multiplicities, orbit-frame orthogonality, coding indices, and phase-retrievable subspaces can be read from group representations.

quant-ph↗

arXiv2026

ET-Prune: Evidence-Aware Dynamic Budgeting for Visual Token Pruning in Text-Rich MLLMs

Visual token pruning reduces the inference cost of multimodal large language models, but a fixed token ratio is poorly matched to text-rich inputs. In OCR-centric tasks, decisive evidence can be a small number, label, or field whose relevance is specified by the question; indiscriminate pruning can erase that evidence while retaining visually salient but irrelevant regions. We present ET-Prune, a training-free framework that casts pruning as evidence allocation. It derives question-conditioned evidence from a decoder-side partial query-key block, safeguards text-like spatial regions, and converts evidence uncertainty and density into a sample-specific token floor. Three progressive middle-layer events then move the sequence toward this budget, retaining more tokens for diffuse or text-dense evidence and pruning concentrated evidence more aggressively. At the observed point estimates from one deterministic pass per configuration, ET-Prune leads or ties among pruned methods in all six backbone-benchmark comparisons at roughly half tokens. On OCRBench-v2, it leads the strongest pruned baselines by 1.80 and 0.68 percentage points on Qwen3-VL-8B and InternVL3.5-8B, respectively, while retaining about half of the visual tokens; on MMBench v1.1, it reaches 0.8467 circular exact-matching accuracy versus 0.8437 for Vanilla at 54.45% average visual-token retention. These results show a favorable observed quality-cost trade-off for evidence-aware dynamic budgeting in text-rich multimodal inference.

cs.CV↗

arXiv2026

Strain-driven orbital-selective reconstruction and bicollinear-to-stripe evolution in FeTe

FeTe, as a representative parent material among iron-based superconductors, provides an ideal platform for exploring the interplay among orbital-selective correlations, magnetism, and unconventional superconductivity. However, a unified picture of the correlated electronic structure and magnetism of FeTe under strain remains to be fully clarified. Here, combining density functional theory plus dynamical mean-field theory and Heisenberg model analysis, we uncover an orbital-selective reconstruction of the correlated electronic structure and reveal a strain-driven trajectory from bicollinear to stripe antiferromagnetism (AFM) via an intermediate competing staggered $n$-mer AFM regime in FeTe. Moderate strain gives rise to a regime where more coherent quasiparticles coexist with suppressed local moments. Further strain drives FeTe into an incoherent correlated regime with robust local moments and Fe-$3d_{z^2}$-dominated low-energy states. These results establish a strain-driven trajectory across distinct magnetic and correlated electronic states in FeTe.

cond-mat.supr-con↗

arXiv2026

CrossMambaTuning: Synergistic Spatial and Cross-Layer Adaptation for Machine Vision Compression

To reduce deployment cost and retraining overhead, adapting pretrained learned image compression (LIC) models to downstream machine vision tasks has attracted growing attention. However, existing methods typically insert fine-tuning modules independently into frozen backbones, lacking explicit mechanisms for cross-layer coordination. To address this limitation, we propose a novel framework named CrossMambaTuning, which integrates State Space Models with cross-layer interaction mechanisms for parameter-efficient fine-tuning. Specifically, we design an efficient Mamba adapter equipped with task-specific prompts and multi-scale branching to precisely capture both local features and global dependencies. Furthermore, we introduce a Scale-Invariant Cross-Layer Adapter (SICA) utilizing a parameter-sharing strategy to fuse task information across different scales and reduce redundancy. Extensive experiments demonstrate that CrossMambaTuning achieves state-of-the-art (SOTA) performance on multiple machine vision tasks, reducing parameter overhead by 72\% compared to SOTA methods. Code is available at https://github.com/rsr1123/CrossMambaTuning.

cs.CV↗

arXiv2026

Bipolarized Weyl semimetals and quantum crystal valley Hall effect in two-dimensional altermagnetic materials

Magnetism and topology are two major areas of condensed matter physics. The combination of magnetism and topology gives rise to more novel physical effects, which have attracted strongly theoretical and experimental attention. Recently, the concept of altermagnetism has been introduced, characterized by a dual nature: real-space antiparallel spins with zero total magnetic moment and reciprocal-space anisotropic spin polarization. The amalgamation of altermagnetism with topology may lead to the emergence of previously unobserved topological phases and the associated physical effects. In this study, utilizing a four-band lattice model that incorporates altermagnetism and spin group symmetry, we demonstrate that type-I, type-II, and type-III bipolarized Weyl semimetals can exist in altermagnetic systems. Through the first-principles electronic structure calculations, we predict four ideal two-dimensional type-I altermagnetic bipolarized Weyl semimetals Fe$_2$WTe$_4$ and Fe$_2$MoZ$_4$ (Z=S, Se, Te). More significantly, we introduce the quantum crystal valley Hall effect, a phenomenon achievable in three of these materials namely Fe$_2$WTe$_4$, Fe$_2$MoS$_4$, and Fe$_2$MoSe$_4$, when spin-orbit coupling is considered. Therefore, our work not only enriches the topological phases but also provides a material platform for studying novel topological phases in altermagnetism.

cond-mat.mtrl-sci↗

arXiv2026

ACQ: A Deployed Two-Stage Framework for Automated Creative Quota Allocation in Large-Scale Online Advertising

In digital advertising, demand-side platforms (DSPs) allow advertisers to create multiple ad creatives from a single photo for real-time bidding. While increasing the number of creatives can improve bidding opportunities, it cannot scale indefinitely, and the incremental advertising revenue typically exhibits diminishing returns as more creatives are generated. This raises a practical problem for DSPs: how to automatically determine an appropriate creative quota for each photo at scale. To address this problem, we propose Automated Creatives Quota (ACQ), a deployed two-stage framework for creative quota allocation in large-scale online advertising. In the first stage, ACQ predicts quota-conditioned expected revenue using a multi-task model built on an unbalanced binary tree, which is designed to handle the highly skewed revenue distribution across quota levels. In the second stage, ACQ formulates quota allocation under global capacity constraints as a multiple-choice knapsack problem (MCKP) and solves it with an efficient dual-based algorithm. Extensive offline experiments and online experiments on Kuaishou's advertising delivery platform demonstrate the effectiveness of ACQ, achieving a 6.20% increase in platform advertising revenue.

cs.AI↗

arXiv2026

GigaBrain-0.7: Scaling Embodied Foundation Models to Emergent Capabilities with a Three-System Architecture

Vision-language-action (VLA) models have become a dominant paradigm for generalist embodied agents, demonstrating strong complex and long-horizon task completion in structured settings. Yet it remains an open question whether current VLA systems can benefit from more effective architectural design, scale to substantially larger and more heterogeneous data regimes, and achieve broader generalization across tasks and embodiments. To this end, we present GigaBrain-0.7, an embodied foundation model with substantially improved generalization across diverse robot embodiments. Specifically, GigaBrain-0.7 unifies understanding, prediction, and action through a three-system architecture, scales pretraining to over 37,000 hours of heterogeneous embodied data, and introduces one-stage alignment training that jointly optimizes vision-language understanding and multi-embodiment action generation. Compared with the preceding GigaBrain-0 series and prior state-of-the-art models including $π_{0.5}$, GigaBrain-0.7 achieves substantial improvements in foundation zero-shot capabilities, language-conditioned instruction following, and post-training task success rates. In particular, on our in-house Maker H01 platform and mainstream robot embodiments, GigaBrain-0.7 demonstrates strong task adaptability and completion ability across both home and industrial scenarios. All training code and pretrained model weights will be released.

cs.RO↗

arXiv2026

Layer-Number-Controlled Symmetry Breaking and Surface-State Transport in Rhombohedral Graphene Multilayers

Rhombohedral multilayer graphene hosts layer-polarized flat bands, providing an intriguing platform for correlated and topological electronic states; however, the role of layer number in governing symmetry breaking and surface screening remains elusive. Here we prepare rhombohedral graphene multilayers and systematically conduct electrical transport measurements. We uncover an unconventional layer dependence of phase transitions: the critical displacement field (D$_{c}$) for the layer-antiferromagnetic (LAF)-to-semimetal transitions remains constant across tetralayer to hexalayer graphene, whereas the D$_{c}$ for semimetal-to-layer-polarized-insulator (LPI) transition increases with layer number, defying unscreened Coulomb interaction models. In hexalayer graphene, surface-state-dominated transport emerges, with Landau levels (LLs) and resistive peaks selectively controlled by adjacent gates, a signature of strong interlayer screening absent in thinner stacks. High magnetic fields reveal valley-layer-locked LLs and dissipative states possibly from interlayer backscattering, highlighting the presence of decoupled surface states. Our findings establish layer number as a key tuning knob for engineering correlated and topological phases in rhombohedral graphene multilayers.

cond-mat.mes-hall↗

arXiv2026

VOICE: A Vision-Omics Foundation Model Integrating Direct and Retrieval-Based Prediction of In-situ Single-Cell Gene Expression

Spatial transcriptomics can resolve gene expression at single-cell resolution, but it is costly, limited to targeted panels of a few hundred to a few thousand genes, and applicable to only a small number of samples. H&E imaging, by contrast, is cheap and collected routinely at scale. This makes predicting single-cell expression directly from morphology a practical way to bring molecular analysis to large tissue archives. We therefore present VOICE, a multimodal foundation model that predicts single-cell gene expression from H&E images using paired Xenium data. VOICE first aligns cell centered H&E morphology from a pathology foundation model with single-cell expression embeddings from a transcriptome foundation model, trained using contrastive learning over 23 million cells. Next it predicts expression through two branches. One branch directly regresses expression from morphology. The other branch retrieves measured expression from similar reference cells, recovering genes that do not have morphological signal. Because genes vary in morphological predictability, VOICE fuses the two branches with a per-gene weight. After training, VOICE generalizes to heldout patients, slides, and partially overlapping gene panels from Xenium, and it consistently outperforms prior single-cell expression prediction methods on seven metrics.

cs.CV↗

arXiv2026

KV-Skill: Forging Expertise in the Model's Native Language

Task knowledge is commonly stored either as text in the prompt or as an update to model weights. Text is modular but must be interpreted on every use, while weight adaptation makes the resulting capability difficult to load, remove, or share independently. We introduce KV-Skill, a design space of external factorized operators that a frozen language model reads through a lightweight interface. KV-Skill supports two complementary paths. Registration converts an authored text skill into a text-derived operator and trains a shared per-backbone interface. Reward learning develops a compact latent operator directly from task outcomes, with or without an authored skill. Neither path adds positions to the prompt. Across ten benchmarks and four backbones from three model families, converting text to a KV-Skill consistently makes the same procedural knowledge more effective. On Qwen3.5-4B LiveMath, registration reaches 77.2 accuracy, compared with 23.4 for the source text skill, 52.0 for SkillOpt, and 64.5 for SoftSkill. Under matched reward training and parameter budgets, KV-Skill gives the best result in seven of eight matched settings against soft prefixes, prefix tuning, and LoRA. A post-hoc rank analysis further shows that text-derived operators retain nearly all of their benefit with one task-aligned direction per injection layer, while matched random directions fail. Finally, one shared interface retains three independently loadable KV-Skills without measurable forgetting. These results show that task knowledge can be acquired from text or experience, compressed into an external operator, and deployed separately from the backbone. Code is available at: https://github.com/shawnzhg/KV-Skill

cs.LG↗

arXiv2026

EvoHIL: Self-Evolving Reward and Flow-Matched Policy Optimization for Robust Human-in-the-Loop Reinforcement Learning

Human-in-the-loop reinforcement learning (HIL-RL) enables robots to learn contact-rich manipulation from limited real-world interaction, but deployment exposes three coupled limitations: static visual reward models fail under scene changes; independently sampled actions cause temporally inconsistent motion; and vision-based policies remain sensitive to appearance shifts. We present EvoHIL, a unified framework that adapts the reward model, action generator, and visual do main within a staged human-in-the-loop learning process. First, self-evolving reward (SER) adapts the success classifier from human-confirmed positives and provisional weak negatives. Second, Action Flow Stabilization (AFS) generates temporally coherent action chunks through flow matching, grounding policy updates in executed action prefixes and demonstrated behavior. Third, retention-aware offline fine-tuning replays relit interaction data while anchoring the AFS actor-critic to prior behavior, adapting the visual domain without additional robot interaction. Across six manipulation tasks on Franka FR3 and SO-101 arms under a controlled lighting shift, EvoHIL improves task success, agreement with human-confirmation labels, motion smoothness, and completion time relative to human-in-the-loop and imitation baselines.Project page: https://anonymous4366.github.io/EvoHIL/

cs.RO↗

arXiv2026

Screening phonon-mediated superconductors from static orbital Hamiltonians

The first-principles search for superconductors is severely limited by the high cost of electron-phonon coupling (EPC) calculations. Here we develop a low-cost, physically transparent framework that identifies strong-EPC materials directly from static orbital-based Hamiltonians without explicit phonon perturbation calculations. Verification using density functional perturbation theory (DFPT) for representative superconductors shows that the framework captures semi-quantitatively the EPC scale at substantially lower computational cost. Applied to more than 36,000 compounds in the MattKeyBond database, it identifies 34 dynamically stable superconducting candidates with calculated $T_c > 10$ K after DFPT verification. These candidates reveal two distinct routes to relatively high-$T_c$ superconductivity: a metallized covalent $σ$-bond route that is more favorable for achieving high-$T_c$ superconductors, and a Fermi-level density-of-states accumulation route that can enhance $T_c$ but usually to a more limited extent.

cond-mat.supr-con↗

arXiv2026

Decoding the Micromagnetic Hamiltonian from Magnetic Fingerprints

Extracting intrinsic magnetic Hamiltonians directly from magnetometry is challenging due to the high dimensionality of the parameter space and the degeneracy induced by ensemble averaging. Here, we introduce a collection of deep convolutional neural networks (CNNs) to extract the full phenomenological micromagnetic Hamiltonian directly from the magnetic fingerprints encoded within First-Order Reversal Curves (FORCs). We validate this approach via closed-loop verification, re-creating the input magnetometry for both simulated and experimental FORCs. To mitigate false positives, we deploy an `Alice--Bob' parallel network that quantifies prediction uncertainty based on solely the information in FORCs without any additional ground-truth knowledge. This framework provides a robust, machine-learning-assisted approach to unravel the underlying spin behaviors in complex magnetic systems

cond-mat.mtrl-sci↗

arXiv2026

Bridging Crystal Structure and Material Properties via Bond-Centric Descriptors

Although chemical bonding is the fundamental mechanistic bridge connecting atomic structure to macroscopic material properties, current data-driven materials science largely treats it as an implicit "black box". Existing machine learning (ML) models rely predominantly on geometric coordinates, forcing them to implicitly relearn complex quantum mechanics from scratch. This lack of intermediate physical features limits model interpretability and generalizability, particularly when training data is scarce. To solve this problem, we introduce MattKeyBond, a bond-centric materials database that explicitly maps the local electronic landscape and bonding interactions of materials. Building on this, we propose Bonding Attractivity (BA), a novel element-specific descriptor that quantifies the intrinsic capability of atoms to form covalent networks. By providing pre-calculated, energy-dimensional bonding descriptors, MattKeyBond transforms the implicit "black box" into physically interpretable features. This strategy relieves ML models from the burden of deducing physical laws from pure geometry, enabling accurate predictions even with limited data and seamlessly integrating electronic structure theory into modern AI workflows.

cond-mat.mtrl-sci↗