arXiv subjects

Yuyang Dong

Publications and source records attributed to Yuyang Dong.

15 recordsLinked to original sources

arXiv2026

Evolution of the intertwining correlated topological phases in iron-based superconductor Fe(Te,Se)

Multiple topological electronic phases can coexist within a single quantum material and induce different topological superconducting states, offering deeper insights into interplay of topological superconducting states and Majorana modes, which may also be influenced and modified by correlation effect. Iron-based superconductors, with both topological states and correlation effect, is an ideal platform to study these phenomena. Here, with high resolution angle resolved photoelectron spectroscopy, we directly resolve two distinct intertwining topological states in iron-based superconductor Co-doped Fe(Te,Se), and study their evolution with electron doping. We identify a region where both topological insulator surface states and topological Dirac semimetal states intersect the Fermi level. The topological states are affected by the strong correlation effect and are isolated from trivial bulk states. The evolution between distinct topological phases offers a good opportunity to study various Majorana modes from different superconducting phases according to theoretical analysis. Our findings establish an ideal platform for exploring the interaction between multiple topological superconducting states and the related Majorana modes.

cond-mat.supr-con↗

arXiv2026

Learning Deliberately, Acting Intuitively: Unlocking Test-Time Reasoning in Multimodal LLMs

Reasoning is essential for large language models (LLMs), especially in complex tasks such as mathematical problem solving. However, multimodal reasoning still faces challenges in modality alignment and training scalability, as many existing methods rely on additional annotations or complex rule-based rewards. To address these issues, we propose the Deliberate-to-Intuitive reasoning framework (D2I), which improves the understanding and reasoning abilities of multimodal LLMs (MLLMs) without extra annotations or complex rewards. During training, D2I uses deliberate reasoning strategies supervised only by rule-based format rewards to enhance modality alignment. During inference, it shifts to intuitive reasoning by removing these explicit strategies, allowing the model to implicitly apply the acquired abilities in its responses. D2I outperforms baselines on both in-domain and out-of-domain benchmarks, highlighting the effectiveness of format rewards in fostering transferable multimodal reasoning skills and suggesting the benefit of decoupling training-time reasoning depth from test-time response flexibility.

cs.CV↗

arXiv2026

Adaptive Layer Selection for Layer-Wise Token Pruning in LLM Inference

Due to the prevalence of large language models (LLMs), key-value (KV) cache reduction for LLM inference has received remarkable attention. Among numerous works that have been proposed in recent years, layer-wise token pruning approaches, which select a subset of tokens at particular layers to retain in KV cache and prune others, are one of the most popular schemes. They primarily adopt a set of pre-defined layers, at which tokens are selected. Such design is inflexible in the sense that the accuracy significantly varies across tasks and deteriorates in harder tasks such as KV retrieval. In this paper, we propose ASL, a training-free method that adaptively chooses the selection layer for KV cache reduction, exploiting the variance of token ranks ordered by attention score. The proposed method balances the performance across different tasks while meeting the user-specified KV budget requirement. ASL operates during the prefilling stage and can be jointly used with existing KV cache reduction methods such as SnapKV to optimize the decoding stage. By evaluations on the InfiniteBench, RULER, and NIAH benchmarks, we show that ASL, equipped with one-shot token selection, adaptively trades inference speed for accuracy, outperforming state-of-the-art layer-wise token pruning methods in difficult tasks.

cs.CL↗

arXiv2026

SCAN: Semantic Document Layout Analysis for Textual and Visual Retrieval-Augmented Generation

With the increasing adoption of Large Language Models (LLMs) and Vision-Language Models (VLMs), rich document analysis technologies for applications like Retrieval-Augmented Generation (RAG) and visual RAG are gaining significant attention. Recent research indicates that using VLMs yields better RAG performance, but processing rich documents remains a challenge since a single page contains large amounts of information. In this paper, we present SCAN (SemantiC Document Layout ANalysis), a novel approach that enhances both textual and visual Retrieval-Augmented Generation (RAG) systems that work with visually rich documents. It is a VLM-friendly approach that identifies document components with appropriate semantic granularity, balancing context preservation with processing efficiency. SCAN uses a coarse-grained semantic approach that divides documents into coherent regions covering contiguous components. We trained the SCAN model by fine-tuning object detection models on an annotated dataset. Our experimental results across English and Japanese datasets demonstrate that applying SCAN improves end-to-end textual RAG performance by up to 9.4 points and visual RAG performance by up to 10.4 points, outperforming conventional approaches and even commercial document processing solutions.

cs.AI↗

arXiv2026

Towards Principled Design of Mixture-of-Experts Language Models under Memory and Inference Constraints

Modern Mixture-of-Experts (MoE) language models are designed based on total parameters (memory footprint) and active parameters (inference cost). However, we find these two factors alone are insufficient to describe an optimal architecture. Through a systematic study, we demonstrate that MoE performance is primarily determined by total parameters ($N_{total}$) and expert sparsity ($s:=n_{exp}/n_{topk}$). Moreover, $n_{exp}$ and $n_{topk}$ do not "cancel out" within the sparsity ratio; instead, a larger total number of experts slightly penalizes performance by forcing a reduction in core model dimensions (depth and width) to meet memory constraints. This motivates a simple principle for MoE design which maximizes $N_{total}$ while minimizing $s$ (maximizing $n_{topk}$) and $n_{exp}$ under the given constraints. Our findings provide a robust framework for resolving architectural ambiguity and guiding MoE design.

cs.CL↗

arXiv2025

Pseudogap and Fermi arc induced by Fermi surface nesting in a centrosymmetric skyrmion magnet

Skyrmions in noncentrosymmetric materials are believed to occur due to the Dzyaloshinskii-Moriya interaction. By contrast, the skyrmion formation mechanism in centrosymmetric materials remains elusive. Here, we reveal the intrinsic electronic structure of the centrosymmetric GdRu2Si2 by selectively measuring magnetic domains using angle-resolved photoemission spectroscopy (ARPES). We found robust Fermi surface (FS) nesting, consistent with the magnetic modulation q-vector detected by the previous resonant x-ray scattering measurements. The pseudogap opens at the nested FS portions, which vary for different magnetic domains. The anomalous pseudogap disconnects the FS to generate Fermi arcs with twofold symmetry. These results indicate that the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction plays a decisive role in generating the screw spin modulation responsible for the skyrmion formation in GdRu2Si2. Furthermore, we demonstrate the flexible nature of magnetism in GdRu2Si2 by manipulating magnetic domains with magnetic field and temperature cyclings, providing potential future applications for data storage and processing devices.

cond-mat.mtrl-sci↗

arXiv2024

Jellyfish: A Large Language Model for Data Preprocessing

This paper explores the utilization of LLMs for data preprocessing (DP), a crucial step in the data mining pipeline that transforms raw data into a clean format conducive to easy processing. Whereas the use of LLMs has sparked interest in devising universal solutions to DP, recent initiatives in this domain typically rely on GPT APIs, raising inevitable data breach concerns. Unlike these approaches, we consider instruction-tuning local LLMs (7 -- 13B models) as universal DP task solvers that operate on a local, single, and low-priced GPU, ensuring data security and enabling further customization. We select a collection of datasets across four representative DP tasks and construct instruction tuning data using data configuration, knowledge injection, and reasoning data distillation techniques tailored to DP. By tuning Mistral-7B, Llama 3-8B, and OpenOrca-Platypus2-13B, our models, namely, Jellyfish-7B/8B/13B, deliver competitiveness compared to GPT-3.5/4 models and strong generalizability to unseen tasks while barely compromising the base models' abilities in NLP tasks. Meanwhile, Jellyfish offers enhanced reasoning capabilities compared to GPT-3.5. Our models are available at: https://huggingface.co/NECOUDBFM/Jellyfish . Our instruction dataset is available at: https://huggingface.co/datasets/NECOUDBFM/Jellyfish-Instruct .

cs.AI↗

arXiv2024

Large Language Models as Data Preprocessors

Large Language Models (LLMs), typified by OpenAI's GPT, have marked a significant advancement in artificial intelligence. Trained on vast amounts of text data, LLMs are capable of understanding and generating human-like text across a diverse range of topics. This study expands on the applications of LLMs, exploring their potential in data preprocessing, a critical stage in data mining and analytics applications. Aiming at tabular data, we delve into the applicability of state-of-the-art LLMs such as GPT-4 and GPT-4o for a series of preprocessing tasks, including error detection, data imputation, schema matching, and entity matching. Alongside showcasing the inherent capabilities of LLMs, we highlight their limitations, particularly in terms of computational expense and inefficiency. We propose an LLM-based framework for data preprocessing, which integrates cutting-edge prompt engineering techniques, coupled with traditional methods like contextualization and feature selection, to improve the performance and efficiency of these models. The effectiveness of LLMs in data preprocessing is evaluated through an experimental study spanning a variety of public datasets. GPT-4 emerged as a standout, achieving 100\% accuracy or F1 score on 4 of these datasets, suggesting LLMs' immense potential in these tasks. Despite certain limitations, our study underscores the promise of LLMs in this domain and anticipates future developments to overcome current hurdles.

cs.AI↗

arXiv2024

Fermi Surface Nesting Driving the RKKY Interaction in the Centrosymmetric Skyrmion Magnet Gd2PdSi3

The magnetic skyrmions generated in a centrosymmetric crystal were recently first discovered in Gd2PdSi3. In light of this, we observe the electronic structure by angle-resolved photoemission spectroscopy (ARPES) and unveil its direct relationship with the magnetism in this compound. The Fermi surface and band dispersions are demonstrated to have a good agreement with the density functional theory (DFT) calculations carried out with careful consideration of the crystal superstructure. Most importantly, we find that the three-dimensional Fermi surface has extended nesting which matches well the q-vector of the magnetic order detected by recent scattering measurements. The consistency we find among ARPES, DFT, and the scattering measurements suggests the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction involving itinerant electrons to be the formation mechanism of skyrmions in Gd2PdSi3.

cond-mat.mtrl-sci↗

arXiv2024

Anomalous Fermi pockets on Hund's metal surface of Sr2RuO4 induced by the correlation-enhanced spin-orbit coupling

The electronic structure of the topmost layer in Sr2RuO4 in the close vicinity of the Fermi level is investigated by angle-resolved photoemission spectroscopy (ARPES) with a 7-eV laser. We find that the spin-orbit coupling (SOC) predicted as 100 meV by the density functional theory (DFT) calculations is enormously enhanced in a real material up to 250 meV, even more than that of bulk state (200 meV), by the electron-correlation effect increased by the octahedral rotation in the crystal structure. This causes the formation of highly orbital-mixing small Fermi pockets and reasonably explains why the orbital-selective Mott transition (OSMT) is not realized in perovskite oxides with crystal distortion. Interestingly, Hund's metal feature allows the quasiparticle generation only near EF, restricting the spectral gap opening derived by band hybridization within an extremely small binding energy (< 10 meV). Furthermore, it causes coherent-incoherent crossover, making the Fermi pockets disappear at elevated temperatures. The anomalous Fermi pockets are characterized by the dichotomy of the orbital-isolating Hund's coupling and the orbital-mixing SOC, which is key to understanding the nature of Sr2RuO4.

cond-mat.str-el↗

arXiv2024

Spontaneous gap opening and potential excitonic states in an ideal Dirac semimetal Ta$_2$Pd$_3$Te$_5$

The opening of an energy gap in the electronic structure generally indicates the presence of interactions. In materials with low carrier density and short screening length, long-range Coulomb interaction favors the spontaneous formation of electron-hole pairs, so-called excitons, opening an excitonic gap at the Fermi level. Excitonic materials host unique phenomenons associated with pair excitations. However, there is still no generally recognized single-crystal material with excitonic order, which is, therefore, awaited in condensed matter physics. Here, we show that excitonic states may exist in the quasi-one-dimensional material Ta$_2$Pd$_3$Te$_5$, which has an almost ideal Dirac-like band structure, with Dirac point located exactly at Fermi level. We find that an energy gap appears at 350 K, and it grows with decreasing temperature. The spontaneous gap opening is absent in a similar material Ta$_2$Ni$_3$Te$_5$. Intriguingly, the gap is destroyed by the potassium deposition on the crystal, likely due to extra-doped carriers. Furthermore, we observe a pair of in-gap flat bands, which is an analog of the impurity states in a superconducting gap. All these observations can be properly explained by an excitonic order, providing Ta$_2$Pd$_3$Te$_5$ as a new and promising candidate realizing excitonic states.

cond-mat.mtrl-sci↗

arXiv2023

DeepJoin: Joinable Table Discovery with Pre-trained Language Models

Due to the usefulness in data enrichment for data analysis tasks, joinable table discovery has become an important operation in data lake management. Existing approaches target equi-joins, the most common way of combining tables for creating a unified view, or semantic joins, which tolerate misspellings and different formats to deliver more join results. They are either exact solutions whose running time is linear in the sizes of query column and target table repository or approximate solutions lacking precision. In this paper, we propose Deepjoin, a deep learning model for accurate and efficient joinable table discovery. Our solution is an embedding-based retrieval, which employs a pre-trained language model (PLM) and is designed as one framework serving both equi- and semantic joins. We propose a set of contextualization options to transform column contents to a text sequence. The PLM reads the sequence and is fine-tuned to embed columns to vectors such that columns are expected to be joinable if they are close to each other in the vector space. Since the output of the PLM is fixed in length, the subsequent search procedure becomes independent of the column size. With a state-of-the-art approximate nearest neighbor search algorithm, the search time is logarithmic in the repository size. To train the model, we devise the techniques for preparing training data as well as data augmentation. The experiments on real datasets demonstrate that by training on a small subset of a corpus, Deepjoin generalizes to large datasets and its precision consistently outperforms other approximate solutions'. Deepjoin is even more accurate than an exact solution to semantic joins when evaluated with labels from experts. Moreover, when equipped with a GPU, Deepjoin is up to two orders of magnitude faster than existing solutions.

cs.DB↗

arXiv2023

Testing Electron-phonon Coupling for the Superconductivity in Kagome Metal $\rm{CsV_3Sb_5}$

In crystalline materials, electron-phonon coupling (EPC) is a ubiquitous many-body interaction that drives conventional Bardeen-Cooper-Schrieffer superconductivity. Recently, in a new kagome metal $\rm{CsV_3Sb_5}$, superconductivity that possibly intertwines with time-reversal and spatial symmetry-breaking orders is observed. Density functional theory calculations predicted weak EPC strength,$λ$, supporting an unconventional pairing mechanism in $\rm{CsV_3Sb_5}$. However, experimental determination of $λ$ is still missing, hindering a microscopic understanding of the intertwined ground state of $\rm{CsV_3Sb_5}$. Here, using 7-eV laser-based angle-resolved photoemission spectroscopy and Eliashberg function analysis, we determine an intermediate $λ$=0.45~0.6 at T=6 K for both Sb 5p and V 3d electronic bands, which can support a conventional superconducting transition temperature on the same magnitude of experimental value in $\rm{CsV_3Sb_5}$. Remarkably, the EPC on the V 3d-band enhances to $λ$~0.75 as the superconducting transition temperature elevated to 4.4 K in $\rm{Cs(V_{0.93}Nb_{0.07})_3Sb_5}$. Our results provide an important clue to understand the pairing mechanism in the Kagome superconductor $\rm{CsV_3Sb_5}$.

cond-mat.supr-con↗

arXiv2022

Table Enrichment System for Machine Learning

Data scientists are constantly facing the problem of how to improve prediction accuracy with insufficient tabular data. We propose a table enrichment system that enriches a query table by adding external attributes (columns) from data lakes and improves the accuracy of machine learning predictive models. Our system has four stages, join row search, task-related table selection, row and column alignment, and feature selection and evaluation, to efficiently create an enriched table for a given query table and a specified machine learning task. We demonstrate our system with a web UI to show the use cases of table enrichment.

cs.IR↗

arXiv2021

Efficient Joinable Table Discovery in Data Lakes: A High-Dimensional Similarity-Based Approach

Finding joinable tables in data lakes is key procedure in many applications such as data integration, data augmentation, data analysis, and data market. Traditional approaches that find equi-joinable tables are unable to deal with misspellings and different formats, nor do they capture any semantic joins. In this paper, we propose PEXESO, a framework for joinable table discovery in data lakes. We embed textual values as high-dimensional vectors and join columns under similarity predicates on high-dimensional vectors, hence to address the limitations of equi-join approaches and identify more meaningful results. To efficiently find joinable tables with similarity, we propose a block-and-verify method that utilizes pivot-based filtering. A partitioning technique is developed to cope with the case when the data lake is large and the index cannot fit in main memory. An experimental evaluation on real datasets shows that our solution identifies substantially more tables than equi-joins and outperforms other similarity-based options, and the join results are useful in data enrichment for machine learning tasks. The experiments also demonstrate the efficiency of the proposed method.

cs.IR↗