Author name cluster

Zhefeng Wang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

10 papers

2 author rows

AAAI Conference 2026 Conference Paper

Discovering Decoupled Functional Modules in Large Language Models

Yanke Yu
Jin Li
Ying Sun
Ping Li
Zhefeng Wang
Yi Zheng

Understanding the internal functional organization of Large Language Models (LLMs) is crucial for improving their trustworthiness and performance. However, how LLMs organize different functions into modules remains highly unexplored. To bridge this gap, we formulate a function module discovery problem and propose an Unsupervised LLM Cross-layer MOdule Discovery (ULCMOD) framework that simultaneously disentangles the large set of neurons in the entire LLM into modules while discovering the topics of input samples related to these modules. Our framework introduces a novel objective function and an efficient Iterative Decoupling (IterD) algorithm. Extensive experiments show that our method discovers high-quality, disentangled modules that capture more meaningful semantic information and achieve superior performance in various downstream tasks. Moreover, our qualitative analysis reveals that the discovered modules show function comprehensiveness, function hierarchy, and clear function spatial arrangement within LLMs. Our work provides a novel tool for interpreting LLMs' function modules, filling a critical gap in LLMs' interpretability research.

PDF Details DOI

ICML Conference 2025 Conference Paper

Efficiently Serving Large Multimodal Models Using EPD Disaggregation

Gursimran Singh
Xinglu Wang
Yifan Hu
Timothy Tin Long Yu
Linzi Xing
Wei Jiang
Zhefeng Wang
Xiaolong Bai

Large Multimodal Models (LMMs) extend Large Language Models (LLMs) by handling diverse inputs such as images, audio, and video, but at the cost of adding a multimodal encoding stage that increases both computational and memory overhead. This step negatively affects key Service Level Objectives (SLOs), such as time to first token (TTFT) and time per output token (TPOT). We introduce Encode-Prefill-Decode (EPD) Disaggregation, a novel framework that separates the encoding, prefill, and decode stages onto dedicated resources. Unlike current systems, which bundle encoding and prefill together, our approach decouples these steps, unlocking new opportunities and optimizations. These include a mechanism to cache multimedia tokens for efficient transfer, a novel way to parallelize the encoding load within a request, a module for optimal resource allocation for disaggregated serving, and a novel role-switching method to handle changing workload characteristics. Experimental evaluations with popular LMMs show substantial gains in memory efficiency (up to 15$\times$ lower peak memory utilization), batch sizes (up to 22$\times$ larger), 10$\times$ more images per request, and 2. 2$\times$ larger KV caches. Furthermore, it leads to significant improvements in SLO attainment (up to 90–100% improvement) and TTFT (up to 71% reduction), compared to systems that do not disaggregate. The code is available at https: //github. com/vbdi/epdserve.

Details

AAAI Conference 2025 Conference Paper

Multi-Branch Self-Drafting for LLM Inference Acceleration

Zipeng Gao
Qingrong Xia
Tong Xu
Xinyu Duan
Zhi Zheng
Zhefeng Wang
Enhong Chen

The autoregressive decoding paradigm endows large language models (LLMs) with superior language generation capabilities; however, its step-by-step decoding process inherently limits decoding speed. To mitigate these constraints, the prevalent “draft and validation” strategy enables parallel validation of candidate drafts, allowing LLMs to decode multiple tokens simultaneously during one model forward propagation. However, existing methodologies for obtaining drafts often incur additional overhead in communication or training process, or statistical biases from the corpus. To this end, we propose an innovative draft generation and maintenance approach that leverages the capabilities of LLM itself. Specifically, we extend the autoregressive decoding paradigm to a multi-branch drafting procedure, which can efficiently generate draft sequences without any additional models or training process, while preserving the quality of the generated content by maintaining LLM parameters. Experiments across various open-source benchmarks show that our method generates 2.0 to 3.2 tokens per forward step and achieves around 2 times improvement of end-to-end throughput compared to the autoregressive decoding strategy.

PDF Details DOI

TIST Journal 2024 Journal Article

Multimodal Dialogue Systems via Capturing Context-aware Dependencies and Ordinal Information of Semantic Elements

Weidong He
Zhi Li
Hao Wang
Tong Xu
Zhefeng Wang
Baoxing Huai
Nicholas Jing Yuan
Enhong Chen

The topic of multimodal conversation systems has recently garnered significant attention across various industries, including travel and retail, among others. While pioneering works in this field have shown promising performance, they often focus solely on context information at the utterance level, overlooking the context-aware dependencies of multimodal semantic elements like words and images. Furthermore, the ordinal information of images, which indicates the relevance between visual context and users’ demands, remains underutilized during the integration of visual content. Additionally, the exploration of how to effectively utilize corresponding attributes provided by users when searching for desired products is still largely unexplored. To address these challenges, we propose PMATE, a P osition-aware M ultimodal di A logue system with seman T ic E lements. Specifically, to obtain semantic representations at the element level, we first unfold the multimodal historical utterances and devise a position-aware multimodal element-level encoder. This component considers all images that may be relevant to the current turn and introduces a novel position-aware image selector to choose related images before fusing the information from the two modalities. Finally, we present a knowledge-aware two-stage decoder and an attribute-enhanced image searcher for the tasks of generating textual responses and selecting image responses, respectively. We extensively evaluate our model on two large-scale multimodal dialogue datasets, and the results of our experiments demonstrate that our approach outperforms several baseline methods.

Details DOI

AAAI Conference 2023 Conference Paper

Distantly-Supervised Named Entity Recognition with Adaptive Teacher Learning and Fine-Grained Student Ensemble

Xiaoye Qu
Jun Zeng
Daizong Liu
Zhefeng Wang
Baoxing Huai
Pan Zhou

Distantly-Supervised Named Entity Recognition (DS-NER) effectively alleviates the data scarcity problem in NER by automatically generating training samples. Unfortunately, the distant supervision may induce noisy labels, thus undermining the robustness of the learned models and restricting the practical application. To relieve this problem, recent works adopt self-training teacher-student frameworks to gradually refine the training labels and improve the generalization ability of NER models. However, we argue that the performance of the current self-training frameworks for DS-NER is severely underestimated by their plain designs, including both inadequate student learning and coarse-grained teacher updating. Therefore, in this paper, we make the first attempt to alleviate these issues by proposing: (1) adaptive teacher learning comprised of joint training of two teacher-student networks and considering both consistent and inconsistent predictions between two teachers, thus promoting comprehensive student learning. (2) fine-grained student ensemble that updates each fragment of the teacher model with a temporal moving average of the corresponding fragment of the student, which enhances consistent predictions on each model fragment against noise. To verify the effectiveness of our proposed method, we conduct experiments on four DS-NER datasets. The experimental results demonstrate that our method significantly surpasses previous SOTA methods. The code is available at https://github.com/zenhjunpro/ATSEN.

PDF Details DOI

IJCAI Conference 2022 Conference Paper

Efficient Document-level Event Extraction via Pseudo-Trigger-aware Pruned Complete Graph

Tong Zhu
Xiaoye Qu
Wenliang Chen
Zhefeng Wang
Baoxing Huai
Nicholas Yuan
Min Zhang

Most previous studies of document-level event extraction mainly focus on building argument chains in an autoregressive way, which achieves a certain success but is inefficient in both training and inference. In contrast to the previous studies, we propose a fast and lightweight model named as PTPCG. In our model, we design a novel strategy for event argument combination together with a non-autoregressive decoding algorithm via pruned complete graphs, which are constructed under the guidance of the automatically selected pseudo triggers. Compared to the previous systems, our system achieves competitive results with 19. 8% of parameters and much lower resource consumption, taking only 3. 8% GPU hours for training and up to 8. 5 times faster for inference. Besides, our model shows superior compatibility for the datasets with (or without) triggers and the pseudo triggers can be the supplements for annotated triggers to make further improvements. Codes are available at https: //github. com/Spico197/DocEE.

PDF Details DOI

AAAI Conference 2021 Conference Paper

Cross-Oilfield Reservoir Classification via Multi-Scale Sensor Knowledge Transfer

Zhi Li
Zhefeng Wang
Zhicheng Wei
Xiangguang Zhou
Yijun Wang
Baoxing Huai
Qi Liu
Nicholas Jing Yuan

Reservoir classification is an essential step for the exploration and production process in the oil and gas industry. An appropriate automatic reservoir classification will not only reduce the manual workloads of experts, but also help petroleum companies to make optimal decisions efficiently, which in turn will dramatically reduce the costs. Existing methods mainly focused on generating reservoir classification in a single geological block but failed to work well on a new oilfield block. Indeed, how to transfer the subsurface characteristics and make accurate reservoir classification across the geological oilfields is a very important but challenging problem. To that end, in this paper, we present a focused study on the cross-oilfield reservoir classification task. Specifically, we first propose a Multi-scale Sensor Extraction (MSE) module to extract the multi-scale feature representations of geological characteristics from multivariate well logs. Furthermore, we design an encoder-decoder module, i. e. , Specific Feature Learning (SFL), to take advantage of specific information of both oilfields. Then, we develop a Knowledge- Attentive Transfer (KAT) module to learn the feature-invariant representation and transfer the geological knowledge from a source oilfield to a target oilfield. Finally, we evaluate our approaches by conducting extensive experiments with realworld industrial datasets. The experimental results clearly demonstrate the effectiveness of our proposed approaches to transfer the geological knowledge and generate the crossoilfield reservoir classifications.

PDF Details

AAAI Conference 2021 Conference Paper

Read, Retrospect, Select: An MRC Framework to Short Text Entity Linking

Yingjie Gu
Xiaoye Qu
Zhefeng Wang
Baoxing Huai
Nicholas Jing Yuan
Xiaolin Gui

Entity linking (EL) for the rapidly growing short text (e. g. search queries and news titles) is critical to industrial applications. Most existing approaches relying on adequate context for long text EL are not effective for the concise and sparse short text. In this paper, we propose a novel framework called Multi-turn Multiple-choice Machine reading comprehension (M3) to solve the short text EL from a new perspective: a query is generated for each ambiguous mention exploiting its surrounding context, and an option selection module is employed to identify the golden entity from candidates using the query. In this way, M3 framework sufficiently interacts limited context with candidate entities during the encoding process, as well as implicitly considers the dissimilarities inside the candidate bunch in the selection stage. In addition, we design a two-stage verifier incorporated into M3 to address the commonly existed unlinkable problem in short text. To further consider the topical coherence and interdependence among referred entities, M3 leverages a multi-turn fashion to deal with mentions in a sequence manner by retrospecting historical cues. Evaluation shows that our M3 framework achieves the state-of-the-art performance on five Chinese and English datasets for the real-world short text EL.

PDF Details

AAAI Conference 2019 Conference Paper

Robust Estimation of Similarity Transformation for Visual Object Tracking

Yang Li
Jianke Zhu
Steven C.H. Hoi
Wenjie Song
Zhefeng Wang
Hantang Liu

Most of existing correlation filter-based tracking approaches only estimate simple axis-aligned bounding boxes, and very few of them is capable of recovering the underlying similarity transformation. To tackle this challenging problem, in this paper, we propose a new correlation filter-based tracker with a novel robust estimation of similarity transformation on the large displacements. In order to efficiently search in such a large 4-DoF space in real-time, we formulate the problem into two 2-DoF sub-problems and apply an efficient Block Coordinates Descent solver to optimize the estimation result. Specifically, we employ an efficient phase correlation scheme to deal with both scale and rotation changes simultaneously in log-polar coordinates. Moreover, a variant of correlation filter is used to predict the translational motion individually. Our experimental results demonstrate that the proposed tracker achieves very promising prediction performance compared with the state-of-the-art visual object tracking methods while still retaining the advantages of high efficiency and simplicity in conventional correlation filter-based tracking methods.

PDF Details

IJCAI Conference 2015 Conference Paper

Maximizing the Coverage of Information Propagation in Social Networks

Zhefeng Wang
Enhong Chen
Qi Liu
Yu Yang
Yong Ge
Biao Chang

Social networks, due to their popularity, have been studied extensively these years. A rich body of these studies is related to influence maximization, which aims to select a set of seed nodes for maximizing the expected number of active nodes at the end of the process. However, the set of active nodes can not fully represent the true coverage of information propagation. A node may be informed of the information when any of its neighbours become active and try to activate it, though this node (namely informed node) is still inactive. Therefore, we need to consider both active nodes and informed nodes that are aware of the information when we study the coverage of information propagation in a network. Along this line, in this paper we propose a new problem called Information Coverage Maximization that aims to maximize the expected number of both active nodes and informed ones. After we prove that this problem is NP-hard and submodular in the independent cascade model, we design two algorithms to solve it. Extensive experiments on three real-world data sets demonstrate the performance of the proposed algorithms.

PDF Details