arXiv subjects

Yue Dai

Publications and source records attributed to Yue Dai.

At least 19 recordsLinked to original sources

arXiv2026

StateFlow: Dual-State Recurrent Modeling for Long-Horizon Time Series Forecasting

Long-horizon multivariate time series forecasting (LTSF) remains challenging due to non-stationarity, regime shifts, and error accumulation. The Variability-Aware Recursive Neural Network (VARNN) is designed to track such variability by maintaining a residual-memory state driven by one-step prediction errors. However, its original formulation is limited to one-step sequence regression and does not directly support multi-step forecasting. In this work, we extend VARNN to long-horizon forecasting and introduce StateFlow, a recurrent forecasting framework that uses VARNN as a dual-state recurrent backbone to capture two complementary signals from the lookback sequence: a hidden-state trajectory representing primary temporal dynamics, including trend, seasonality, level changes, and recurring patterns, and a residual-memory trajectory representing structured local prediction deviations, driven from a nonlinear recurrent transformation of errors between one-step base predictions and observed values. A chunk-based decoder separately summarizes these trajectories and maps them to the future horizon for direct multi-step forecasting. We further employ a two-stage optimization strategy that first trains the VARNN encoder through a one-step base prediction objective to optimize the internal representations over the lookback sequence, and then trains a horizon-specific decoder for direct multi-step forecasting. Experiments on standard LTSF benchmarks show that StateFlow achieves competitive performance against strong linear, recurrent, convolutional, and Transformer-based baselines while preserving linear recurrent encoding and a compact model design.

cs.LG↗

arXiv2026

Temporal Aware Pruning for Efficient Diffusion-based Video Generation

Video diffusion models have recently enabled high-quality video generation with ViT-based architectures, but remain computationally intensive because generation requires attention computation over long spatiotemporal sequences. Token pruning has proven effective for ViTs and VLMs. However, most prior pruning methods are attention-based and operate per frame, failing to ensure the vital temporal coherence across frames in video generation tasks. In practice, naively adopting attention-only pruning causes noticeable degradation due to worsened background consistency, flickering, and reduced image quality. To address this, we propose TAPE, a training-free Temporal Aware Pruning for Efficient diffusion-based video generation. TAPE (i) applies temporal smoothing to align token-importance across adjacent frames and suppress selection jitter; and (ii) performs token reselection in selected layers to align token pruning with layers' diverse semantic focus and avoid error accumulation in specific areas; it also (iii) adopt a timestep-level budget scheduling that prunes aggressively at early noisy steps and relaxes pruning during fidelity-critical refinement. The experimental results show that TAPE delivers significant speedups while preserving high visual fidelity, outperforming prior token reduction approaches.

cs.CV↗

arXiv2026

Accelerating 3D Gaussian Splatting using Tensor Cores

3D Gaussian Splatting (3DGS) has become a leading technique for real-time neural rendering and 3D scene reconstruction, but its rendering cost remains too high for many latency-sensitive scenarios. In particular, the rasterization stage in 3DGS dominates end-to-end rendering time, during which the renderer repeatedly evaluates each Gaussian's contribution to each covered pixel, making this stage compute-bound. At the same time, modern GPUs provide high-throughput Tensor Cores for low-precision matrix operations, yet existing 3DGS systems execute rasterization entirely on CUDA cores and leave Tensor Cores idle. We find that 3DGS rendering can be executed in FP16 with negligible quality degradation, suggesting a promising opportunity for Tensor Core acceleration. However, exploiting Tensor Cores for 3DGS is non-trivial because rasterization does not naturally match their execution model. Existing 3DGS rasterization is expressed as irregular per-pixel scalar operations, whereas Tensor Cores require dense, regular, and reuse-rich matrix workloads. Moreover, conventional tile-by-tile execution fails to exploit Gaussian reuse across neighboring tiles, resulting in repeated data loading and thus high data movement overhead. To this end, we present TensorGS, a 3DGS acceleration framework using Tensor Cores. TensorGS tensorizes the dominant rasterization computation into Tensor-Core-compatible matrix operations and introduces cross-tile grouping to improve Gaussian reuse, amortize overhead, and increase Tensor Core utilization. Experimental results show that TensorGS improves end-to-end rendering performance by 1.65$\times$ while preserving image quality.

cs.GR↗

arXiv2026

BEAM: Binary Expert Activation Masking for Dynamic Routing in MoE

Mixture-of-Experts (MoE) architectures enhance the efficiency of large language models by activating only a subset of experts per token. However, standard MoE employs a fixed Top-K routing strategy, leading to redundant computation and suboptimal inference latency. Existing acceleration methods either require costly retraining with architectural changes or suffer from severe performance drop at high sparsity due to train-inference mismatch. To address these limitations, we propose BEAM (Binary Expert Activation Masking), a novel method that learns token-adaptive expert selection via trainable binary masks. With a straight-through estimator and an auxiliary regularization loss, BEAM induces dynamic expert sparsity through end-to-end training while maintaining model capability. We further implement an efficient custom CUDA kernel for BEAM, ensuring seamless integration with the vLLM inference framework. Experiments show that BEAM retains over 98\% of the original model's performance while reducing MoE layer FLOPs by up to 85\%, achieving up to 2.5$\times$ faster decoding and 1.4$\times$ higher throughput, demonstrating its effectiveness as a practical, plug-and-play solution for efficient MoE inference.

cs.AI↗

arXiv2026

Clore: Interactive Pathology Image Segmentation with Click-based Local Refinement

Recent advancements in deep learning-based interactive segmentation methods have significantly improved pathology image segmentation. Most existing approaches utilize user-provided positive and negative clicks to guide the segmentation process. However, these methods primarily rely on iterative global updates for refinement, which lead to redundant re-prediction and often fail to capture fine-grained structures or correct subtle errors during localized adjustments. To address this limitation, we propose the Click-based Local Refinement (Clore) pipeline, a simple yet efficient method designed to enhance interactive segmentation. The key innovation of Clore lies in its hierarchical interaction paradigm: the initial clicks drive global segmentation to rapidly outline large target regions, while subsequent clicks progressively refine local details to achieve precise boundaries. This approach not only improves the ability to handle fine-grained segmentation tasks but also achieves high-quality results with fewer interactions. Experimental results on four datasets demonstrate that Clore achieves the best balance between segmentation accuracy and interaction cost, making it an effective solution for efficient and accurate interactive pathology image segmentation.

cs.CV↗

arXiv2026

Epitaxial Growth and Anomalous Hall Effect in High-Quality Altermagnetic $\alpha$-MnTe Thin Films

The recent identification of $\alpha$-MnTe as a candidate altermagnet has attracted considerable interest, particularly for its potential application in magnetic random-access memory. However, the development of high-quality thin films - essential for practical implementation - has remained limited. Here, we report the epitaxial growth of centimeter-scale $\alpha$-MnTe thin films on InP(111) substrates via molecular beam epitaxy (MBE). Through X-ray diffraction (XRD) analysis, we construct a MnTe phase diagram that provides clear guidance for stabilizing the pure $\alpha$-MnTe phase, revealing that it is favored under high Te/Mn flux ratios and elevated growth temperatures. Cross-sectional electron microscopy confirms an atomically sharp film-substrate interface, consistent with a layer-by-layer epitaxial growth mode. Remarkably, these high-quality $\alpha$-MnTe films exhibit a pronounced anomalous Hall effect (AHE) originating from Berry curvature, despite a net magnetic moment approaching zero - a signature of robust altermagnetic character. Our work establishes a viable route for synthesizing wafer-scale $\alpha$-MnTe thin films and highlights their promise for altermagnet-based spintronics and magnetic sensing.

cond-mat.mtrl-sci↗

arXiv2026

Magnetic field-induced non-trivial Lifshitz transition in TaCo2Te2

Magnetic-field-driven Lifshitz transitions are typically considered zero-temperature phenomena involving Fermi-surface reconstruction without symmetry breaking. Here, we report an unconventional Lifshitz transition in TaCo2Te2 that emerges exclusively within a narrow finite-temperature window under cooperative tuning by both temperature and magnetic field. Bulk-sensitive transport and thermoelectric measurements demonstrate continuous Fermi-surface renormalization at low temperatures, where the transition is sharply triggered by a critical magnetic field. Crucially, neutron diffraction reveals the absence of structural or magnetic phase transitions, while angle-resolved photoemission spectroscopy shows no spectral anomalies in electronic structure without magnetic field. These observations constrain the mechanism to a Zeeman-driven process invisible to equilibrium probes, establishing a paradigm where Fermi-surface topology is jointly controlled by temperature and magnetic field.

cond-mat.mtrl-sci↗

arXiv2025

Let It Flow: Agentic Crafting on Rock and Roll, Building the ROME Model within an Open Agentic Learning Ecosystem

Agentic crafting requires LLMs to operate in real-world environments over multiple turns by taking actions, observing outcomes, and iteratively refining artifacts. Despite its importance, the open-source community lacks a principled, end-to-end ecosystem to streamline agent development. We introduce the Agentic Learning Ecosystem (ALE), a foundational infrastructure that optimizes the production pipeline for agentic model. ALE consists of three components: ROLL, a post-training framework for weight optimization; ROCK, a sandbox environment manager for trajectory generation; and iFlow CLI, an agent framework for efficient context engineering. We release ROME, an open-source agent grounded by ALE and trained on over one million trajectories. Our approach includes data composition protocols for synthesizing complex behaviors and a novel policy optimization algorithm, Interaction-Perceptive Agentic Policy Optimization (IPA), which assigns credit over semantic interaction chunks rather than individual tokens to improve long-horizon training stability. Empirically, we evaluate ROME within a structured setting and introduce Terminal Bench Pro, a benchmark with improved scale and contamination control. ROME demonstrates strong performance across benchmarks like SWE-bench Verified and Terminal Bench, proving the effectiveness of ALE.

cs.AI↗

arXiv2025

Non-altermagnetic spin texture in MnTe

Recently, altermagnets have emerged as promising candidates in spintronics, uniquely combining large spin-polarized electronic states with zero net magnetization. A prominent example is $\alpha$-MnTe, whose altermagnetic spin splitting, i.e., the degeneracy lift in momentum space induced by collinear magnetic order, has been experimentally observed. However, the direct evidence of its $g$-wave spin polarization, the key property for altermagnetic spintronics, is thus far lacking. By combining high-resolution spin- and angle-resolved photoemission spectroscopy (SARPES) with first-principles calculations, we reveal a $k_z$-independent, Rashba-like spin texture in $\alpha$-MnTe. Our results indicate that the observed spin polarization is primarily governed by spin-orbit coupling, whereas the magnetic order contributes to the splitting of energy bands but plays a much less dominant role in spin polarization due to the multi-domain nature. From this result, we further establish a way to prescreen altermagnet candidates that favor the formation of large antiferromagnetic domains based on symmetry analysis. Our work elucidates the interplay between magnetic order and spin-orbit coupling in governing spin polarization in altermagnet candidates, and thereby advances the materials design paradigm for spin-functional devices.

cond-mat.mtrl-sci↗

arXiv2025

FlashOmni: A Unified Sparse Attention Engine for Diffusion Transformers

Multi-Modal Diffusion Transformers (DiTs) demonstrate exceptional capabilities in visual synthesis, yet their deployment remains constrained by substantial computational demands. To alleviate this bottleneck, many sparsity-based acceleration methods have been proposed. However, their diverse sparsity patterns often require customized kernels for high-performance inference, limiting universality. We propose FlashOmni, a unified sparse attention engine compatible with arbitrary DiT architectures. FlashOmni introduces flexible sparse symbols to standardize the representation of a wide range of sparsity strategies, such as feature caching and block-sparse skipping. This unified abstraction enables the execution of diverse sparse computations within a single attention kernel. In addition, FlashOmni designs optimized sparse GEMMs for attention blocks, leveraging sparse symbols to eliminate redundant computations and further improve efficiency. Experiments demonstrate that FlashOmni delivers near-linear, closely matching the sparsity ratio speedup (1:1) in attention and GEMM-$Q$, and achieves 2.5$\times$-3.8$\times$ acceleration in GEMM-$O$ (max peaking at about 87.5% of the theoretical limit). Applied with a multi-granularity sparsity strategy, it enables the Hunyuan model (33K) to achieve about 1.5$\times$ end-to-end acceleration without degrading visual quality.

cs.LG↗

arXiv2025

A Survey on MLLM-based Visually Rich Document Understanding: Methods, Challenges, and Emerging Trends

Visually Rich Document Understanding (VRDU) has become a pivotal area of research, driven by the need to automatically interpret documents that contain intricate visual, textual, and structural elements. Recently, Multimodal Large Language Models (MLLMs) have demonstrated significant promise in this domain, including both OCR-based and OCR-free approaches for information extraction from document images. This survey reviews recent advances in MLLM-based VRDU, highlighting emerging trends and promising research directions with a focus on two key aspects: (1) techniques for representing and integrating textual, visual, and layout features; (2) training paradigms, including pretraining, instruction tuning, and training strategies. Moreover, we address challenges such as data scarcity, handling multi-page and multilingual documents, and integrating emerging trends such as Retrieval-Augmented Generation and agentic frameworks. Our analysis offers a roadmap for advancing MLLM-based VRDU toward more scalable, reliable, and adaptable systems.

cs.CV↗

arXiv2025

Enhancing Document Key Information Localization Through Data Augmentation

The Visually Rich Form Document Intelligence and Understanding (VRDIU) Track B focuses on the localization of key information in document images. The goal is to develop a method capable of localizing objects in both digital and handwritten documents, using only digital documents for training. This paper presents a simple yet effective approach that includes a document augmentation phase and an object detection phase. Specifically, we augment the training set of digital documents by mimicking the appearance of handwritten documents. Our experiments demonstrate that this pipeline enhances the models' generalization ability and achieves high performance in the competition.

cs.CV↗

arXiv2025

Graph-Based Multimodal Contrastive Learning for Chart Question Answering

Chart question answering (ChartQA) is challenged by the heterogeneous composition of chart elements and the subtle data patterns they encode. This work introduces a novel joint multimodal scene graph framework that explicitly models the relationships among chart components and their underlying structures. The framework integrates both visual and textual graphs to capture structural and semantic characteristics, while a graph contrastive learning strategy aligns node representations across modalities enabling their seamless incorporation into a transformer decoder as soft prompts. Moreover, a set of tailored Chain of Thought (CoT) prompts is proposed to enhance multimodal large language models (MLLMs) in zero-s ot scenarios by mitigating hallucinations. Extensive evaluations on benchmarks including ChartQA, OpenCQA, and ChartX demonstrate significant performance improvements and validate the efficacy of the proposed approach.

cs.CL↗

arXiv2024

Using Reinforcement Learning to Guide Graph State Generation for Photonic Quantum Computers

Photonic quantum computer (PQC) is an emerging and promising quantum computing paradigm that has gained momentum in recent years. In PQC, which leverages the measurement-based quantum computing (MBQC) model, computations are executed by performing measurements on photons in graph states (i.e., sets of entangled photons) that are generated before measurements. The graph state in PQC is generated deterministically by quantum emitters. The generation process is achieved by applying a sequence of quantum gates to quantum emitters. In this process, i) the time required to complete the process, ii) the number of quantum emitters used, and iii) the number of CZ gates performed between emitters greatly affect the fidelity of the generated graph state. However, prior work for determining the generation sequence only focuses on optimizing the number of quantum emitters. Moreover, identifying the optimal generation sequence has vast search space. To this end, we propose RLGS, a novel compilation framework to identify optimal generation sequences that optimize the three metrics. Experimental results show that RLGS achieves an average reduction in generation time of 31.1%, 49.6%, and 57.5% for small, medium, and large graph states compared to the baseline.

quant-ph↗

arXiv2024

ChuLo: Chunk-Level Key Information Representation for Long Document Understanding

Transformer-based models have achieved remarkable success in various Natural Language Processing (NLP) tasks, yet their ability to handle long documents is constrained by computational limitations. Traditional approaches, such as truncating inputs, sparse self-attention, and chunking, attempt to mitigate these issues, but they often lead to information loss and hinder the model's ability to capture long-range dependencies. In this paper, we introduce ChuLo, a novel chunk representation method for long document understanding that addresses these limitations. Our ChuLo groups input tokens using unsupervised keyphrase extraction, emphasizing semantically important keyphrase based chunks to retain core document content while reducing input length. This approach minimizes information loss and improves the efficiency of Transformer-based models. Preserving all tokens in long document understanding, especially token classification tasks, is important to ensure that fine-grained annotations, which depend on the entire sequence context, are not lost. We evaluate our method on multiple long document classification tasks and long document token classification tasks, demonstrating its effectiveness through comprehensive qualitative and quantitative analysis. Our implementation is open-sourced on https://github.com/adlnlp/Chulo.

cs.CL↗

arXiv2024

MSG-Chart: Multimodal Scene Graph for ChartQA

Automatic Chart Question Answering (ChartQA) is challenging due to the complex distribution of chart elements with patterns of the underlying data not explicitly displayed in charts. To address this challenge, we design a joint multimodal scene graph for charts to explicitly represent the relationships between chart elements and their patterns. Our proposed multimodal scene graph includes a visual graph and a textual graph to jointly capture the structural and semantical knowledge from the chart. This graph module can be easily integrated with different vision transformers as inductive bias. Our experiments demonstrate that incorporating the proposed graph module enhances the understanding of charts' elements' structure and semantics, thereby improving performance on publicly available benchmarks, ChartQA and OpenCQA.

cs.CL↗

arXiv2024

Improving GPU Multi-Tenancy Through Dynamic Multi-Instance GPU Reconfiguration

Continuous learning (CL) has emerged as one of the most popular deep learning paradigms deployed in modern cloud GPUs. Specifically, CL has the capability to continuously update the model parameters (through model retraining) and use the updated model (if available) to serve overtime arriving inference requests. It is generally beneficial to co-locate the retraining and inference together to enable timely model updates and avoid model transfer overheads. This brings the need for GPU sharing among retraining and inferences. Meanwhile, multiple CL workloads can share the modern GPUs in the cloud, leading to multi-tenancy execution. In this paper, we observe that prior GPU-sharing techniques are not optimized for multi-tenancy CL workloads. Specifically, they do not coherently consider the accuracy of the retraining model and the inference service level objective (SLO) attainment. Moreover, they cannot accommodate the overtime dynamics (e.g., inference arrival intensity) in CL execution. In this paper, we propose MIGRator, a novel GPU reconfiguration runtime that dynamically performs GPU reconfiguration for multi-tenancy CL workloads. MIGRator is based on the recent NVIDIA multi-instance GPU (MIG) to mitigate resource contention and formulates the reconfiguration optimization into Integer Linear Programming (ILP) to dynamically identify, reconfigure, and allocate the GPU instances. MIGRator leverages the "Goodput" metric in the ILP objective function to consider both inference SLO attainment and model accuracy in the reconfiguration exploration. We evaluate MIGRator using representative multi-tenancy CL workloads. The results show our approach outperforms the state-of-the-art GPU sharing techniques (i.e., Ekya, Astraea, and PARIS) by 17\%, 21\%, and 20\%, respectively.

cs.DC↗

arXiv2024

Observation of floating surface state in obstructed atomic insulator candidate NiP$_2$

Obstructed atomic insulator is recently proposed as an unconventional material, in which electric charge centers localized at sites away from the atoms. A half-filling surface state would emerge at specific interfaces cutting through these charge centers and avoid intersecting any atoms. In this article, we utilized angle-resolved photoemission spectroscopy and density functional theory calculations to study one of the obstructed atomic insulator candidates, NiP$_2$. A floating surface state with large effective mass that is isolated from all bulk states is resolved on the (100) cleavage plane, distinct from previously reported surface states in obstructed atomic insulators that are merged into bulk bands. Density functional theory calculation results elucidate that this floating surface state is originated from the obstructed Wannier charge centers, albeit underwent surface reconstruction that splits the half-filled obstructed surface state. Our findings not only shed lights on the spectroscopy study of obstructed atomic insulators and obstructed surface states, but also provide possible route for development of new catalysts.

cond-mat.mtrl-sci↗