arXiv subjects

Ziqi Yin

Publications and source records attributed to Ziqi Yin.

18 recordsLinked to original sources

arXiv2026

Jagle: Building a Large-Scale Japanese Multimodal Post-Training Dataset for Vision-Language Models

Developing vision-language models (VLMs) that generalize across diverse tasks requires large-scale training datasets with diverse content. In English, such datasets are typically constructed by aggregating and curating numerous existing visual question answering (VQA) resources. However, this strategy does not readily extend to other languages, where VQA datasets remain limited in both scale and domain coverage. In this work, we introduce Jagle, the largest Japanese multimodal post-training dataset to date, comprising approximately 9.2 million instances across diverse tasks. Rather than relying on existing VQA datasets, we collect heterogeneous source data, including images, image-text pairs, and PDF documents, and generate VQA pairs through multiple strategies such as VLM-based QA generation, translation, and text rendering. Experiments demonstrate that a 2.2B model trained with Jagle surpasses InternVL3.5-2B in average score across ten Japanese evaluation tasks and comes within five points of Qwen3-VL-2B-Instruct. Moreover, combining Jagle with FineVision improves English performance over training with FineVision alone. These benefits generalize across model backbones and scales: LLM-jp-4-VL 9B beta, a 9B-scale model built on the Japanese-centric LLM-jp-4-8B-instruct, achieves Japanese performance comparable to Qwen3-VL-8B-Instruct. To facilitate reproducibility and future research, we release the dataset, models, and code.

cs.CV↗

arXiv2026

LiteTopK: Exploiting the Curse of Dimensionality for a Fused Indexer-TopK Kernel in Long-Context Sparse Attention

Indexer-TopK, the operation to compute the scores and select the top-k candidates, is widely used by sparse attention algorithms in large language models and vector retrieval in recommendation systems and vector databases. However, existing GPU-based Indexer-TopK kernels like DeepSeek Sparse Attention (DSA) remain inefficient due to excessive global memory traffic, costly synchronization, and prohibitive memory overhead. In this study, inspired by the curse of dimensionality phenomenon, we first observe that sparse attention scores exhibit a score concentration phenomenon, where scores tend to fall within a narrow range. Based on this observation, we propose LITETOPK, an efficient fused Indexer-TopK kernel. LITETOPK first samples a small subset of data to estimate query-data score ranges, then partitions candidates into bins accordingly. This organization allows the LITETOPK kernel to maintain a tight approximate threshold online, write back only promising candidates, reduce unnecessary I/O and memory overhead while preserving exact Top-k correctness. Building on LITETOPK, we further propose LITEDSA, which exploits the similarity of top-k candidate sets among neighboring tokens. LITEDSA packs neighboring tokens' candidates for joint computation and masks out extra scores for each query, thereby reducing memory traffic while preserving correctness. Experimental results in a real-world deployment environ ment with eight B200 GPUs show that LITETOPK+LITEDSA accelerates the prefill stage of GLM 5.2 by 1.35x, with no performance loss and lower memory overhead.

cs.LG↗

arXiv2026

Efficient Graph Indexing for Interval-Aware Vector Search

Interval-aware Approximate Nearest Neighbor (ANN) search arises in applications where each object is associated with a numeric value or interval, and queries must satisfy both vector-similarity and interval constraints. Existing methods are typically tailored to a single query semantics, such as interval-filtered ANN search, and therefore require multiple specialized indexes to support diverse workloads, leading to substantial indexing and memory overhead. To address this limitation, we propose the Unified Interval-aware Relative Neighborhood Graph (URNG), a unified graph framework for interval-aware ANN search. URNG preserves the monotonic searchability of relative-neighborhood-graph based ANN indexes while additionally ensuring structural heredity over query-induced subgraphs, enabling a single index to support multiple interval-aware query semantics. Building on this framework, we develop UG, a practical graph index that efficiently approximates URNG through unified interval-aware pruning and iterative repair, together with a query algorithm for interval-aware ANN search. Extensive experiments on 5 datasets show that UG consistently achieves a strong accuracy-efficiency trade-off across diverse interval-aware workloads while maintaining competitive index construction cost and memory usage.

cs.DB↗

arXiv2026

BBC: Improving Large-k Approximate Nearest Neighbor Search with a Bucket-based Result Collector

Although Approximate Nearest Neighbor (ANN) search has been extensively studied, large-k ANN queries that aim to retrieve a large number of nearest neighbors remain underexplored, despite their numerous real-world applications. Existing ANN methods face significant performance degradation for such queries. In this work, we first investigate the reasons for the performance degradation of quantization-based ANN indexes: (1) the inefficiency of existing top-k collectors, which incurs significant overhead in candidate maintenance, and (2) the reduced pruning effectiveness of quantization methods, which leads to a costly re-ranking process. To address this, we propose a novel bucket-based result collector (BBC) to enhance the efficiency of existing quantization-based ANN indexes for large-k ANN queries. BBC introduces two key components: (1) a bucket-based result buffer that organizes candidates into buckets by their distances to the query. This design reduces ranking costs and improves cache efficiency, enabling high performance maintenance of a candidate superset and a lightweight final selection of top-k results. (2) two re-ranking algorithms tailored for different types of quantization methods, which accelerate their re-ranking process by reducing either the number of candidate objects to be re-ranked or cache misses. Extensive experiments on real-world datasets demonstrate that BBC accelerates existing quantization-based ANN methods by up to 3.8x at recall@k = 0.95 for large-k ANN queries.

cs.DB↗

arXiv2026

RNSG: A Range-Aware Graph Index for Efficient Range-Filtered Approximate Nearest Neighbor Search

Range-filtered approximate nearest neighbor (RFANN) search is a fundamental operation in modern data systems. Given a set of objects, each with a vector and a numerical attribute, an RFANN query retrieves the nearest neighbors to a query vector among those objects whose numerical attributes fall within the range specified by the query. Existing state-of-the-art methods for RFANN search often require constructing multiple range-specific graph indexes to achieve high query performance, which incurs significant indexing overhead. To address this, we first establish a novel graph indexing theory, the range-aware relative neighborhood graph (RRNG), which jointly considers spatial and attribute proximity. We prove that the RRNG satisfies two crucial properties: (1) monotonic search-ability, which ensures correct nearest neighbor retrieval via beam search; and (2) structural heredity, which guarantees that any range-induced subgraph remains a valid RRNG, thus enabling efficient search with a single graph index. Based on this theoretical foundation, we propose a new graph index called RNSG as a practical solution that efficiently approximates RRNG. We develop fast algorithms for both constructing the RNSG index and processing RFANN queries with it. Extensive experiments on five real-world datasets show that RNSG achieves significantly higher query performance with a more compact index and lower construction cost than existing state-of-the-art methods.

cs.DB↗

arXiv2026

Contact-Guided 3D Genome Structure Generation of E. coli via Diffusion Transformers

In this study, we present a conditional diffusion-transformer framework for generating ensembles of three-dimensional Escherichia coli genome conformations guided by Hi-C contact maps. Instead of producing a single deterministic structure, we formulate genome reconstruction as a conditional generative modeling problem that samples heterogeneous conformations whose ensemble-averaged contacts are consistent with the input Hi-C data. A synthetic dataset is constructed using coarse-grained molecular dynamics simulations to generate chromatin ensembles and corresponding Hi-C maps under circular topology. Our models operate in a latent diffusion setting with a variational autoencoder that preserves per-bin alignment and supports replication-aware representations. Hi-C information is injected through a transformer-based encoder and cross-attention, enforcing a physically interpretable one-way constraint from Hi-C to structure. The model is trained using a flow-matching objective for stable optimization. On held-out ensembles, generated structures reproduce the input Hi-C distance-decay and structural correlation metrics while maintaining substantial conformational diversity, demonstrating the effectiveness of diffusion-based generative modeling for ensemble-level 3D genome reconstruction.

cs.LG↗

arXiv2025

δ-EMG: A Monotonic Graph Index for Approximate Nearest Neighbor Search

Approximate nearest neighbor (ANN) search in high-dimensional spaces is a foundational component of many modern retrieval and recommendation systems. Currently, almost all algorithms follow an $ε$-Recall-Bounded principle when comparing performance: they require the ANN search results to achieve a recall of more than $1-ε$ and then compare query-per-second (QPS) performance. However, this approach only accounts for the recall of true positive results and does not provide guarantees on the deviation of incorrect results. To address this limitation, we focus on an Error-Bounded ANN method, which ensures that the returned results are a $(1/δ)$-approximation of the true values. Our approach adopts a graph-based framework. To enable Error-Bounded ANN search, we propose a $δ$-EMG (Error-bounded Monotonic Graph), which, for the first time, provides a provable approximation for arbitrary queries. By enforcing a $δ$-monotonic geometric constraint during graph construction, $δ$-EMG ensures that any greedy search converges to a $(1/δ)$-approximate neighbor without backtracking. Building on this foundation, we design an error-bounded top-$k$ ANN search algorithm that adaptively controls approximation accuracy during query time. To make the framework practical at scale, we introduce $δ$-EMQG (Error-bounded Monotonic Quantized Graph), a localized and degree-balanced variant with near-linear construction complexity. We further integrate vector quantization to accelerate distance computation while preserving theoretical guarantees. Extensive experiments on the ANN-Benchmarks dataset demonstrate the effectiveness of our approach. Under a recall requirement of 0.99, our algorithm achieves 19,000 QPS on the SIFT1M dataset, outperforming other methods by more than 40\%.

cs.IR↗

arXiv2025

VIR-Bench: Evaluating Geospatial and Temporal Understanding of MLLMs via Travel Video Itinerary Reconstruction

Recent advances in multimodal large language models (MLLMs) have significantly enhanced video understanding capabilities, opening new possibilities for practical applications. Yet current video benchmarks focus largely on indoor scenes or short-range outdoor activities, leaving the challenges associated with long-distance travel largely unexplored. Mastering extended geospatial-temporal trajectories is critical for next-generation MLLMs, underpinning real-world tasks such as embodied-AI planning and navigation. To bridge this gap, we present VIR-Bench, a novel benchmark consisting of 200 travel videos that frames itinerary reconstruction as a challenging task designed to evaluate and push forward MLLMs' geospatial-temporal intelligence. Experimental results reveal that state-of-the-art MLLMs, including proprietary ones, struggle to achieve high scores, underscoring the difficulty of handling videos that span extended spatial and temporal scales. Moreover, we conduct an in-depth case study in which we develop a prototype travel-planning agent that leverages the insights gained from VIR-Bench. The agent's markedly improved itinerary recommendations verify that our evaluation protocol not only benchmarks models effectively but also translates into concrete performance gains in user-facing applications.

cs.CV↗

arXiv2025

MACE Foundation Models for Lattice Dynamics: A Benchmark Study on Double Halide Perovskites

Recent developments in materials informatics and artificial intelligence has led to the emergence of foundational energy models for material chemistry, as represented by the suite of MACE-based foundation models, bringing a significant breakthrough in universal potentials for inorganic solids. As to all method developments in computational materials science, performance benchmarking against existing high-level data with focusing on specific applications, is critically needed to understand the limitations in the models, thus facilitating the ongoing improvements in the model development process, and occasionally, leading to significant conceptual leaps in materials theory. Here, using our own published DFT (Density Functional Theory) database of room-temperature dynamic stability and vibrational anharmonicity for $\sim2000$ cubic halide double perovskites, we benchmarked the performances of four different variants of the MACE foundation models for screening the dynamic stabilities of inorganic solids. Our analysis shows that, as anticipated, the model accuracy improves with more training data. The dynamic stabilities of weakly anharmonic materials (as predicted by DFT) are more accurately reproduced by the foundation model, than those highly anharmonic and dynamically unstable ones. The predominant source of error in predicting the dynamic stability arises predominantly from the amplification of errors in atomic forces when predicting the harmonic phonon properties through the computation of the Hessian matrix, less so is the contribution from possible differences in the range of the configurational spaces that are sampled by DFT and the foundation model in molecular dynamics. We hope that our present findings will stimulate future works towards more physics-inspired approaches in assessing the accuracy of foundation models for atomistic modelling.

cond-mat.mtrl-sci↗

arXiv2025

Harmony: A Human-Aware, Responsive, Modular Assistant with a Locally Deployed Large Language Model

Large Language Models (LLMs) offer powerful capabilities for natural language understanding, enabling more intelligent smart home assistants. However, existing systems often rely on cloud-based LLMs, raising concerns around user privacy and system dependency on external connectivity. In this work, we present Harmony, a privacy-preserving and robust smart home assistant powered by the locally deployable Llama3-8B model. Beyond protecting user data, Harmony also addresses reliability challenges of smaller models, such as hallucination and instruction misinterpretation, through structured prompting and modular agent design. Experimental results in both virtual environments and user studies show that Harmony achieves performance comparable to GPT-4-based systems, while enabling offline, proactive, and personalized smart home interaction.

cs.HC↗

arXiv2025

DEG: Efficient Hybrid Vector Search Using the Dynamic Edge Navigation Graph

Bimodal data, such as image-text pairs, has become increasingly prevalent in the digital era. The Hybrid Vector Query (HVQ) is an effective approach for querying such data and has recently garnered considerable attention from researchers. It calculates similarity scores for objects represented by two vectors using a weighted sum of each individual vector's similarity, with a query-specific parameter $α$ to determine the weight. Existing methods for HVQ typically construct Approximate Nearest Neighbors Search (ANNS) indexes with a fixed $α$ value. This leads to significant performance degradation when the query's $α$ dynamically changes based on the different scenarios and needs. In this study, we introduce the Dynamic Edge Navigation Graph (DEG), a graph-based ANNS index that maintains efficiency and accuracy with changing $α$ values. It includes three novel components: (1) a greedy Pareto frontier search algorithm to compute a candidate neighbor set for each node, which comprises the node's approximate nearest neighbors for all possible $α$ values; (2) a dynamic edge pruning strategy to determine the final edges from the candidate set and assign each edge an active range. This active range enables the dynamic use of the Relative Neighborhood Graph's pruning strategy based on the query's $α$ values, skipping redundant edges at query time and achieving a better accuracy-efficiency trade-off; and (3) an edge seed method that accelerates the querying process. Extensive experiments on real-world datasets show that DEG demonstrates superior performance compared to existing methods under varying $α$ values.

cs.DB↗

arXiv2024

LIST: Learning to Index Spatio-Textual Data for Embedding based Spatial Keyword Queries

With the proliferation of spatio-textual data, Top-k KNN spatial keyword queries (TkQs), which return a list of objects based on a ranking function that considers both spatial and textual relevance, have found many real-life applications. To efficiently handle TkQs, many indexes have been developed, but the effectiveness of TkQ is limited. To improve effectiveness, several deep learning models have recently been proposed, but they suffer severe efficiency issues and there are no efficient indexes specifically designed to accelerate the top-k search process for these deep learning models. To tackle these issues, we consider embedding based spatial keyword queries, which capture the semantic meaning of query keywords and object descriptions in two separate embeddings to evaluate textual relevance. Although various models can be used to generate these embeddings, no indexes have been specifically designed for such queries. To fill this gap, we propose LIST, a novel machine learning based Approximate Nearest Neighbor Search index that Learns to Index the Spatio-Textual data. LIST utilizes a new learning-to-cluster technique to group relevant queries and objects together while separating irrelevant queries and objects. There are two key challenges in building an effective and efficient index, i.e., the absence of high-quality labels and the unbalanced clustering results. We develop a novel pseudo-label generation technique to address the two challenges. Additionally, we introduce a learning based spatial relevance model that can integrates with various text relevance models to form a lightweight yet effective relevance for reranking objects retrieved by LIST.

cs.IR↗

arXiv2024

Should We Respect LLMs? A Cross-Lingual Study on the Influence of Prompt Politeness on LLM Performance

We investigate the impact of politeness levels in prompts on the performance of large language models (LLMs). Polite language in human communications often garners more compliance and effectiveness, while rudeness can cause aversion, impacting response quality. We consider that LLMs mirror human communication traits, suggesting they align with human cultural norms. We assess the impact of politeness in prompts on LLMs across English, Chinese, and Japanese tasks. We observed that impolite prompts often result in poor performance, but overly polite language does not guarantee better outcomes. The best politeness level is different according to the language. This phenomenon suggests that LLMs not only reflect human behavior but are also influenced by language, particularly in different cultural contexts. Our findings highlight the need to factor in politeness for cross-cultural natural language processing and LLM usage.

cs.CL↗

arXiv2023

Fairness-aware Maximal Biclique Enumeration on Bipartite Graphs

Maximal biclique enumeration is a fundamental problem in bipartite graph data analysis. Existing biclique enumeration methods mainly focus on non-attributed bipartite graphs and also ignore the \emph{fairness} of graph attributes. In this paper, we introduce the concept of fairness into the biclique model for the first time and study the problem of fairness-aware biclique enumeration. Specifically, we propose two fairness-aware biclique models, called \nonesidebc~and \ntwosidebc~respectively. To efficiently enumerate all {\nonesidebc}s, we first present two non-trivial pruning techniques, called fair $α$-$β$ core pruning and colorful fair $α$-$β$ core pruning, to reduce the graph size without losing accuracy. Then, we develop a branch and bound algorithm, called \onesideFBCEM, to enumerate all single-side fair bicliques on the reduced bipartite graph. To further improve the efficiency, we propose an efficient branch and bound algorithm with a carefully-designed combinatorial enumeration technique. Note that all of our techniques can also be extended to enumerate all bi-side fair bicliques. We also extend the two fairness-aware biclique models by constraining the ratio of the number of vertices of each attribute to the total number of vertices and present corresponding enumeration algorithms. Extensive experimental results on five large real-world datasets demonstrate our methods' efficiency, effectiveness, and scalability.

cs.DS↗

arXiv2022

Graph Attention MLP with Reliable Label Utilization

Graph neural networks (GNNs) have recently achieved state-of-the-art performance in many graph-based applications. Despite the high expressive power, they typically need to perform an expensive recursive neighborhood expansion in multiple training epochs and face a scalability issue. Moreover, most of them are inflexible since they are restricted to fixed-hop neighborhoods and insensitive to actual receptive field demands for different nodes. We circumvent these limitations by introducing a scalable and flexible Graph Attention Multilayer Perceptron (GAMLP). With the separation of the non-linear transformation and feature propagation, GAMLP significantly improves the scalability and efficiency by performing the propagation procedure in a pre-compute manner. With three principled receptive field attention, each node in GAMLP is flexible and adaptive in leveraging the propagated features over the different sizes of reception field. We conduct extensive evaluations on the three large open graph benchmarks (e.g., ogbn-papers100M, ogbn-products and ogbn-mag), demonstrating that GAMLP not only achieves the state-of-art performance, but also additionally provide high scalability and efficiency.

cs.LG↗

arXiv2022

Graph Attention Multi-Layer Perceptron

Graph neural networks (GNNs) have achieved great success in many graph-based applications. However, the enormous size and high sparsity level of graphs hinder their applications under industrial scenarios. Although some scalable GNNs are proposed for large-scale graphs, they adopt a fixed $K$-hop neighborhood for each node, thus facing the over-smoothing issue when adopting large propagation depths for nodes within sparse regions. To tackle the above issue, we propose a new GNN architecture -- Graph Attention Multi-Layer Perceptron (GAMLP), which can capture the underlying correlations between different scales of graph knowledge. We have deployed GAMLP in Tencent with the Angel platform, and we further evaluate GAMLP on both real-world datasets and large-scale industrial datasets. Extensive experiments on these 14 graph datasets demonstrate that GAMLP achieves state-of-the-art performance while enjoying high scalability and efficiency. Specifically, it outperforms GAT by 1.3\% regarding predictive accuracy on our large-scale Tencent Video dataset while achieving up to $50\times$ training speedup. Besides, it ranks top-1 on both the leaderboards of the largest homogeneous and heterogeneous graph (i.e., ogbn-papers100M and ogbn-mag) of Open Graph Benchmark.

cs.LG↗

arXiv2022

Model Degradation Hinders Deep Graph Neural Networks

Graph Neural Networks (GNNs) have achieved great success in various graph mining tasks.However, drastic performance degradation is always observed when a GNN is stacked with many layers. As a result, most GNNs only have shallow architectures, which limits their expressive power and exploitation of deep neighborhoods.Most recent studies attribute the performance degradation of deep GNNs to the \textit{over-smoothing} issue. In this paper, we disentangle the conventional graph convolution operation into two independent operations: \textit{Propagation} (\textbf{P}) and \textit{Transformation} (\textbf{T}).Following this, the depth of a GNN can be split into the propagation depth ($D_p$) and the transformation depth ($D_t$). Through extensive experiments, we find that the major cause for the performance degradation of deep GNNs is the \textit{model degradation} issue caused by large $D_t$ rather than the \textit{over-smoothing} issue mainly caused by large $D_p$. Further, we present \textit{Adaptive Initial Residual} (AIR), a plug-and-play module compatible with all kinds of GNN architectures, to alleviate the \textit{model degradation} issue and the \textit{over-smoothing} issue simultaneously. Experimental results on six real-world datasets demonstrate that GNNs equipped with AIR outperform most GNNs with shallow architectures owing to the benefits of both large $D_p$ and $D_t$, while the time costs associated with AIR can be ignored.

cs.LG↗

arXiv2022

GRAND+: Scalable Graph Random Neural Networks

Graph neural networks (GNNs) have been widely adopted for semi-supervised learning on graphs. A recent study shows that the graph random neural network (GRAND) model can generate state-of-the-art performance for this problem. However, it is difficult for GRAND to handle large-scale graphs since its effectiveness relies on computationally expensive data augmentation procedures. In this work, we present a scalable and high-performance GNN framework GRAND+ for semi-supervised graph learning. To address the above issue, we develop a generalized forward push (GFPush) algorithm in GRAND+ to pre-compute a general propagation matrix and employ it to perform graph data augmentation in a mini-batch manner. We show that both the low time and space complexities of GFPush enable GRAND+ to efficiently scale to large graphs. Furthermore, we introduce a confidence-aware consistency loss into the model optimization of GRAND+, facilitating GRAND+'s generalization superiority. We conduct extensive experiments on seven public datasets of different sizes. The results demonstrate that GRAND+ 1) is able to scale to large graphs and costs less running time than existing scalable GNNs, and 2) can offer consistent accuracy improvements over both full-batch and scalable GNNs across all datasets.

cs.LG↗