Peiyu Liu

Ze-Feng Gao, Peiyu Liu, Wayne Xin Zhao et al.

Recently, Mixture-of-Experts (short as MoE) architecture has achieved remarkable success in increasing the model capacity of large-scale language models. However, MoE requires incorporating significantly more parameters than the base model being extended. In this paper, we propose building a parameter-efficient MoE architecture by sharing information among experts. We adopt the matrix product operator (MPO, a tensor decomposition from quantum many-body physics) to reconstruct the parameter matrix in the expert layer and increase model capacity for pre-trained language models by sharing parameters of the central tensor (containing the core information) among different experts while enabling the specificity through the auxiliary tensors (complementing the central tensor) of different experts. To address the unbalanced optimization issue, we further design the gradient mask strategy for the MPO-based MoE architecture. Extensive experiments based on T5 and GPT-2 show improved performance and efficiency of the pre-trained language model (27.2x reduction in total parameters for the superior model performance, compared with the Switch Transformers). Our code is publicly available at https://github.com/RUCAIBox/MPOE.

Wayne Xin Zhao, Kun Zhou, Junyi Li et al.

Language is essentially a complex, intricate system of human expressions governed by grammatical rules. It poses a significant challenge to develop capable AI algorithms for comprehending and grasping a language. As a major approach, language modeling has been widely studied for language understanding and generation in the past two decades, evolving from statistical language models to neural language models. Recently, pre-trained language models (PLMs) have been proposed by pre-training Transformer models over large-scale corpora, showing strong capabilities in solving various NLP tasks. Since researchers have found that model scaling can lead to performance improvement, they further study the scaling effect by increasing the model size to an even larger size. Interestingly, when the parameter scale exceeds a certain level, these enlarged language models not only achieve a significant performance improvement but also show some special abilities that are not present in small-scale language models. To discriminate the difference in parameter scale, the research community has coined the term large language models (LLM) for the PLMs of significant size. Recently, the research on LLMs has been largely advanced by both academia and industry, and a remarkable progress is the launch of ChatGPT, which has attracted widespread attention from society. The technical evolution of LLMs has been making an important impact on the entire AI community, which would revolutionize the way how we develop and use AI algorithms. In this survey, we review the recent advances of LLMs by introducing the background, key findings, and mainstream techniques. In particular, we focus on four major aspects of LLMs, namely pre-training, adaptation tuning, utilization, and capacity evaluation. Besides, we also summarize the available resources for developing LLMs and discuss the remaining issues for future directions.

Wayne Xin Zhao, Kun Zhou, Junyi Li et al.

Language is essentially a complex, intricate system of human expressions governed by grammatical rules. It poses a significant challenge to develop capable AI algorithms for comprehending and grasping a language. As a major approach, language modeling has been widely studied for language understanding and generation in the past two decades, evolving from statistical language models to neural language models. Recently, pre-trained language models (PLMs) have been proposed by pre-training Transformer models over large-scale corpora, showing strong capabilities in solving various NLP tasks. Since researchers have found that model scaling can lead to performance improvement, they further study the scaling effect by increasing the model size to an even larger size. Interestingly, when the parameter scale exceeds a certain level, these enlarged language models not only achieve a significant performance improvement but also show some special abilities that are not present in small-scale language models. To discriminate the difference in parameter scale, the research community has coined the term large language models (LLM) for the PLMs of significant size. Recently, the research on LLMs has been largely advanced by both academia and industry, and a remarkable progress is the launch of ChatGPT, which has attracted widespread attention from society. The technical evolution of LLMs has been making an important impact on the entire AI community, which would revolutionize the way how we develop and use AI algorithms. In this survey, we review the recent advances of LLMs by introducing the background, key findings, and mainstream techniques. In particular, we focus on four major aspects of LLMs, namely pre-training, adaptation tuning, utilization, and capacity evaluation. Besides, we also summarize the available resources for developing LLMs and discuss the remaining issues for future directions.

Peiyu Liu, Zikang Liu, Ze-Feng Gao et al.

Despite the superior performance, Large Language Models~(LLMs) require significant computational resources for deployment and use. To overcome this issue, quantization methods have been widely applied to reduce the memory footprint of LLMs as well as increasing the inference rate. However, a major challenge is that low-bit quantization methods often lead to performance degradation. It is important to understand how quantization impacts the capacity of LLMs. Different from previous studies focused on overall performance, this work aims to investigate the impact of quantization on \emph{emergent abilities}, which are important characteristics that distinguish LLMs from small language models. Specially, we examine the abilities of in-context learning, chain-of-thought reasoning, and instruction-following in quantized LLMs. Our empirical experiments show that these emergent abilities still exist in 4-bit quantization models, while 2-bit models encounter severe performance degradation on the test of these abilities. To improve the performance of low-bit models, we conduct two special experiments: (1) fine-gained impact analysis that studies which components (or substructures) are more sensitive to quantization, and (2) performance compensation through model fine-tuning. Our work derives a series of important findings to understand the impact of quantization on emergent abilities, and sheds lights on the possibilities of extremely low-bit quantization for LLMs.

Hongpeng Lin, Ludan Ruan, Wenke Xia et al.

To facilitate the research on intelligent and human-like chatbots with multi-modal context, we introduce a new video-based multi-modal dialogue dataset, called TikTalk. We collect 38K videos from a popular video-sharing platform, along with 367K conversations posted by users beneath them. Users engage in spontaneous conversations based on their multi-modal experiences from watching videos, which helps recreate real-world chitchat context. Compared to previous multi-modal dialogue datasets, the richer context types in TikTalk lead to more diverse conversations, but also increase the difficulty in capturing human interests from intricate multi-modal information to generate personalized responses. Moreover, external knowledge is more frequently evoked in our dataset. These facts reveal new challenges for multi-modal dialogue models. We quantitatively demonstrate the characteristics of TikTalk, propose a video-based multi-modal chitchat task, and evaluate several dialogue baselines. Experimental results indicate that the models incorporating large language models (LLM) can generate more diverse responses, while the model utilizing knowledge graphs to introduce external knowledge performs the best overall. Furthermore, no existing model can solve all the above challenges well. There is still a large room for future improvements, even for LLM with visual extensions. Our dataset is available at \url{https://ruc-aimind.github.io/projects/TikTalk/}.

Yifan Xia, Zichen Xie, Peiyu Liu et al.

While the automated detection of cryptographic API misuses has progressed significantly, its precision diminishes for intricate targets due to the reliance on manually defined patterns. Large Language Models (LLMs) offer a promising context-aware understanding to address this shortcoming, yet the stochastic nature and the hallucination issue pose challenges to their applications in precise security analysis. This paper presents the first systematic study to explore LLMs' application in cryptographic API misuse detection. Our findings are noteworthy: The instability of directly applying LLMs results in over half of the initial reports being false positives. Despite this, the reliability of LLM-based detection could be significantly enhanced by aligning detection scopes with realistic scenarios and employing a novel code and analysis validation technique, achieving a nearly 90% detection recall. This improvement substantially surpasses traditional methods and leads to the discovery of previously unknown vulnerabilities in established benchmarks. Nevertheless, we identify recurring failure patterns that illustrate current LLMs' blind spots. Leveraging these findings, we deploy an LLM-based detection system and uncover 63 new vulnerabilities (47 confirmed, 7 already fixed) in open-source Java and Python repositories, including prominent projects like Apache.

Tong Ye, Hang Yu, Tengfei Ma et al.

Large Language Models have demonstrated remarkable progress in general-purpose capabilities and can achieve strong performance in specific domains through fine-tuning on domain-specific data. However, acquiring high-quality data for target domains remains a significant challenge. Existing data synthesis approaches follow a deductive paradigm, heavily relying on explicit domain descriptions expressed in natural language and careful prompt engineering, limiting their applicability in real-world scenarios where domains are difficult to describe or formally articulate. In this work, we tackle the underexplored problem of domain-specific data synthesis through an inductive paradigm, where the target domain is defined only through a set of reference examples, particularly when domain characteristics are difficult to articulate in natural language. We propose a novel framework, DOMINO, that learns a minimal sufficient domain representation from reference samples and leverages it to guide the generation of domain-aligned synthetic data. DOMINO integrates prompt tuning with a contrastive disentanglement objective to separate domain-level patterns from sample-specific noise, mitigating overfitting while preserving core domain characteristics. Theoretically, we prove that DOMINO expands the support of the synthetic data distribution, ensuring greater diversity. Empirically, on challenging coding benchmarks where domain definitions are implicit, fine-tuning on data synthesized by DOMINO improves Pass@1 accuracy by up to 4.63\% over strong, instruction-tuned backbones, demonstrating its effectiveness and robustness. This work establishes a new paradigm for domain-specific data synthesis, enabling practical and scalable domain adaptation without manual prompt design or natural language domain specifications.

Peiyu Liu, Ze-Feng Gao, Yushuo Chen et al.

In this paper, we propose a highly parameter-efficient approach to scaling pre-trained language models (PLMs) to a deeper model depth. Unlike prior work that shares all parameters or uses extra blocks, we design a more capable parameter-sharing architecture based on matrix product operator (MPO). MPO decomposition can reorganize and factorize the information of a parameter matrix into two parts: the major part that contains the major information (central tensor) and the supplementary part that only has a small proportion of parameters (auxiliary tensors). Based on such a decomposition, our architecture shares the central tensor across all layers for reducing the model size and meanwhile keeps layer-specific auxiliary tensors (also using adapters) for enhancing the adaptation flexibility. To improve the model training, we further propose a stable initialization algorithm tailored for the MPO-based architecture. Extensive experiments have demonstrated the effectiveness of our proposed model in reducing the model size and achieving highly competitive performance.

Peiyu Liu, Junming Liu, Lirong Fu et al.

Recently, ChatGPT has attracted great attention from the code analysis domain. Prior works show that ChatGPT has the capabilities of processing foundational code analysis tasks, such as abstract syntax tree generation, which indicates the potential of using ChatGPT to comprehend code syntax and static behaviors. However, it is unclear whether ChatGPT can complete more complicated real-world vulnerability management tasks, such as the prediction of security relevance and patch correctness, which require an all-encompassing understanding of various aspects, including code syntax, program semantics, and related manual comments. In this paper, we explore ChatGPT's capabilities on 6 tasks involving the complete vulnerability management process with a large-scale dataset containing 70,346 samples. For each task, we compare ChatGPT against SOTA approaches, investigate the impact of different prompts, and explore the difficulties. The results suggest promising potential in leveraging ChatGPT to assist vulnerability management. One notable example is ChatGPT's proficiency in tasks like generating titles for software bug reports. Furthermore, our findings reveal the difficulties encountered by ChatGPT and shed light on promising future directions. For instance, directly providing random demonstration examples in the prompt cannot consistently guarantee good performance in vulnerability management. By contrast, leveraging ChatGPT in a self-heuristic way -- extracting expertise from demonstration examples itself and integrating the extracted expertise in the prompt is a promising research direction. Besides, ChatGPT may misunderstand and misuse the information in the prompt. Consequently, effectively guiding ChatGPT to focus on helpful information rather than the irrelevant content is still an open problem.

Tao Yang, Xuefeng Jiang, Wei Li et al.

Existing research on sensor data anomaly detection for industrial sensor networks still has several inherent limitations. First, most detection models usually consider centralized detection. Thus, all sensor data have to be uploaded to the control center for analysis, leading to a heavy traffic load. However, industrial sensor networks have high requirements for reliable and real-time communication. The heavy traffic load may cause communication delays or packets lost by corruption. Second, there are complex spatial and temporal features in industrial sensor data. The full extraction of such features plays a key role in improving detection performance.To solve the limitations above, this paper develops a cloud-edge collaborative data anomaly detection approach for industrial sensor networks that mainly consists of a sensor data detection model deployed at individual edges and a sensor data analysis model deployed in the cloud. The former is implemented using Gaussian and Bayesian algorithms, which effectively filter the substantial volume of sensor data generated during the normal operation of the industrial sensor network, thereby reducing traffic load. It only uploads all the sensor data to the sensor data analysis model for further analysis when the network is in an anomalous state. The latter based on GCRL is developed by inserting Long Short-Term Memory network (LSTM) into Graph Convolutional Network (GCN), which can effectively extract the spatial and temporal features of the sensor data for anomaly detection.

Tong Ye, Lingfei Wu, Tengfei Ma et al.

Automatically generating function summaries for binaries is an extremely valuable but challenging task, since it involves translating the execution behavior and semantics of the low-level language (assembly code) into human-readable natural language. However, most current works on understanding assembly code are oriented towards generating function names, which involve numerous abbreviations that make them still confusing. To bridge this gap, we focus on generating complete summaries for binary functions, especially for stripped binary (no symbol table and debug information in reality). To fully exploit the semantics of assembly code, we present a control flow graph and pseudo code guided binary code summarization framework called CP-BCS. CP-BCS utilizes a bidirectional instruction-level control flow graph and pseudo code that incorporates expert knowledge to learn the comprehensive binary function execution behavior and logic semantics. We evaluate CP-BCS on 3 different binary optimization levels (O1, O2, and O3) for 3 different computer architectures (X86, X64, and ARM). The evaluation results demonstrate CP-BCS is superior and significantly improves the efficiency of reverse engineering.

Baoxing Jiang, Shehui Liang, Peiyu Liu et al.

Aspect sentiment triplet extraction (ASTE) is a crucial subtask of aspect-based sentiment analysis (ABSA) that aims to comprehensively identify sentiment triplets. Previous research has focused on enhancing ASTE through innovative table-filling strategies. However, these approaches often overlook the multi-perspective nature of language expressions, resulting in a loss of valuable interaction information between aspects and opinions. To address this limitation, we propose a framework that leverages both a basic encoder, primarily based on BERT, and a particular encoder comprising a Bi-LSTM network and graph convolutional network (GCN ). The basic encoder captures the surface-level semantics of linguistic expressions, while the particular encoder extracts deeper semantics, including syntactic and lexical information. By modeling the dependency tree of comments and considering the part-of-speech and positional information of words, we aim to capture semantics that are more relevant to the underlying intentions of the sentences. An interaction strategy combines the semantics learned by the two encoders, enabling the fusion of multiple perspectives and facilitating a more comprehensive understanding of aspect--opinion relationships. Experiments conducted on benchmark datasets demonstrate the state-of-the-art performance of our proposed framework.

Sha Yuan, Shuai Zhao, Jiahong Leng et al.

Compared with the domain-specific model, the vision-language pre-training models (VLPMs) have shown superior performance on downstream tasks with fast fine-tuning process. For example, ERNIE-ViL, Oscar and UNIMO trained VLPMs with a uniform transformers stack architecture and large amounts of image-text paired data, achieving remarkable results on downstream tasks such as image-text reference(IR and TR), vision question answering (VQA) and image captioning (IC) etc. During the training phase, VLPMs are always fed with a combination of multiple public datasets to meet the demand of large-scare training data. However, due to the unevenness of data distribution including size, task type and quality, using the mixture of multiple datasets for model training can be problematic. In this work, we introduce a large-scale multi-modal corpora named WuDaoMM, totally containing more than 650M image-text pairs. Specifically, about 600 million pairs of data are collected from multiple webpages in which image and caption present weak correlation, and the other 50 million strong-related image-text pairs are collected from some high-quality graphic websites. We also release a base version of WuDaoMM with 5 million strong-correlated image-text pairs, which is sufficient to support the common cross-modal model pre-training. Besides, we trained both an understanding and a generation vision-language (VL) model to test the dataset effectiveness. The results show that WuDaoMM can be applied as an efficient dataset for VLPMs, especially for the model in text-to-image generation task. The data is released at https://data.wudaoai.cn

Zican Dong, Peiyu Liu, Junyi Li et al.

Recently, large language models (LLMs) have shown remarkable reasoning abilities by producing long reasoning traces. However, as the sequence length grows, the key-value (KV) cache expands linearly, incurring significant memory and computation costs. Existing KV cache eviction methods mitigate this issue by discarding less important KV pairs, but often fail to capture complex KV dependencies, resulting in performance degradation. To better balance efficiency and performance, we introduce ForesightKV, a training-based KV cache eviction framework that learns to predict which KV pairs to evict during long-text generations. We first design the Golden Eviction algorithm, which identifies the optimal eviction KV pairs at each step using future attention scores. These traces and the scores at each step are then distilled via supervised training with a Pairwise Ranking Loss. Furthermore, we formulate cache eviction as a Markov Decision Process and apply the GRPO algorithm to mitigate the significant language modeling loss increase on low-entropy tokens. Experiments on AIME2024 and AIME2025 benchmarks of three reasoning models demonstrate that ForesightKV consistently outperforms prior methods under only half the cache budget, while benefiting synergistically from both supervised and reinforcement learning approaches.

Yihan Wang, Peiyu Liu, Runyu Chen et al.

Text-to-SQL has recently achieved impressive progress, yet remains difficult to apply effectively in real-world scenarios. This gap stems from the reliance on single static workflows, fundamentally limiting scalability to out-of-distribution and long-tail scenarios. Instead of requiring users to select suitable methods through extensive experimentation, we attempt to enable systems to adaptively construct workflows at inference time. Through theoretical and empirical analysis, we demonstrate that optimal dynamic policies consistently outperform the best static workflow, with performance gains fundamentally driven by heterogeneity across candidate workflows. Motivated by this, we propose SquRL, a reinforcement learning framework that enhances LLMs' reasoning capability in adaptive workflow construction. We design a rule-based reward function and introduce two effective training mechanisms: dynamic actor masking to encourage broader exploration, and pseudo rewards to improve training efficiency. Experiments on widely-used Text-to-SQL benchmarks demonstrate that dynamic workflow construction consistently outperforms the best static workflow methods, with especially pronounced gains on complex and out-of-distribution queries. The codes are available at https://github.com/Satissss/SquRL

Yihan Wang, Peiyu Liu, Xin Yang

Schema linking is a critical bottleneck in applying existing Text-to-SQL models to real-world, large-scale, multi-database environments. Through error analysis, we identify two major challenges in schema linking: (1) Database Retrieval: accurately selecting the target database from a large schema pool, while effectively filtering out irrelevant ones; and (2) Schema Item Grounding: precisely identifying the relevant tables and columns within complex and often redundant schemas for SQL generation. Based on these, we introduce LinkAlign, a novel framework tailored for large-scale databases with thousands of fields. LinkAlign comprises three key steps: multi-round semantic enhanced retrieval and irrelevant information isolation for Challenge 1, and schema extraction enhancement for Challenge 2. Each stage supports both Agent and Pipeline execution modes, enabling balancing efficiency and performance via modular design. To enable more realistic evaluation, we construct AmbiDB, a synthetic dataset designed to reflect the ambiguity of real-world schema linking. Experiments on widely-used Text-to-SQL benchmarks demonstrate that LinkAlign consistently outperforms existing baselines on all schema linking metrics. Notably, it improves the overall Text-to-SQL pipeline and achieves a new state-of-the-art score of 33.09% on the Spider 2.0-Lite benchmark using only open-source LLMs, ranking first on the leaderboard at the time of submission. The codes are available at https://github.com/Satissss/LinkAlign

Xiaolin Hu, Xiang Cheng, Peiyu Liu et al.

Low-rank adaptation (LoRA) reduces the computational and memory demands of fine-tuning large language models (LLMs) by approximating updates with low-rank matrices. However, low-rank approximation in two-dimensional space fails to capture high-dimensional structures within the target matrix. Recently, tensor decomposition methods have been explored for fine-tuning LLMs, leveraging their ability to extract structured information. Yet, these approaches primarily rely on random initialization, and the impact of initialization on tensor adaptation remains underexplored. In this paper, we reveal that random initialization significantly diverges from the validation loss achieved by full fine-tuning. To address this, we propose Weight-Decomposed Tensor Adaptation (DoTA), which leverages the Matrix Product Operator (MPO) decomposition of pre-trained weights for effective initialization in fine-tuning LLMs. Additionally, we introduce QDoTA, a quantized version of DoTA designed for 4-bit quantization. Experiments on commonsense and arithmetic reasoning tasks show that DoTA outperforms random initialization methods with fewer parameters. QDoTA further reduces memory consumption and achieves comparable performance to DoTA on commonsense reasoning tasks. We will release our code to support future research.

Peiyu Liu, Junping Du, Yingxia Shao et al.

Relational extraction is one of the basic tasks related to information extraction in the field of natural language processing, and is an important link and core task in the fields of information extraction, natural language understanding, and information retrieval. None of the existing relation extraction methods can effectively solve the problem of triple overlap. The CasAug model proposed in this paper based on the CasRel framework combined with the semantic enhancement mechanism can solve this problem to a certain extent. The CasAug model enhances the semantics of the identified possible subjects by adding a semantic enhancement mechanism, First, based on the semantic coding of possible subjects, pre-classify the possible subjects, and then combine the subject lexicon to calculate the semantic similarity to obtain the similar vocabulary of possible subjects. According to the similar vocabulary obtained, each word in different relations is calculated through the attention mechanism. For the contribution of the possible subject, finally combine the relationship pre-classification results to weight the enhanced semantics of each relationship to find the enhanced semantics of the possible subject, and send the enhanced semantics combined with the possible subject to the object and relationship extraction module. Complete the final relation triplet extraction. The experimental results show that, compared with the baseline model, the CasAug model proposed in this paper has improved the effect of relation extraction, and CasAug's ability to deal with overlapping problems and extract multiple relations is also better than the baseline model, indicating that the semantic enhancement mechanism proposed in this paper It can further reduce the judgment of redundant relations and alleviate the problem of triple overlap.

Yongxuan Wu, Runyu Chen, Peiyu Liu et al.

Long-context understanding poses significant challenges in natural language processing, particularly for real-world dialogues characterized by speech-based elements, high redundancy, and uneven information density. Although large language models (LLMs) achieve impressive results on existing benchmarks, these datasets fail to reflect the complexities of such texts, limiting their applicability to practical scenarios. To bridge this gap, we construct the first spoken long-text dataset, derived from live streams, designed to reflect the redundancy-rich and conversational nature of real-world scenarios. We construct tasks in three categories: retrieval-dependent, reasoning-dependent, and hybrid. We then evaluate both popular LLMs and specialized methods to assess their ability to understand long-contexts in these tasks. Our results show that current methods exhibit strong task-specific preferences and perform poorly on highly redundant inputs, with no single method consistently outperforming others. We propose a new baseline that better handles redundancy in spoken text and achieves strong performance across tasks. Our findings highlight key limitations of current methods and suggest future directions for improving long-context understanding. Finally, our benchmark fills a gap in evaluating long-context spoken language understanding and provides a practical foundation for developing real-world e-commerce systems. The code and benchmark are available at https://github.com/Yarayx/livelongbench.

Haoran Yang, Yinan Zhang, Wenjie Zhang et al.

Nowadays, industrial hybrid modeling which integrates both mechanistic modeling and machine learning-based modeling techniques has attracted increasing interest from scholars due to its high accuracy, low computational cost, and satisfactory interpretability. Nevertheless, the existing industrial hybrid modeling methods still face two main limitations. First, current research has mainly focused on applying a single machine learning method to one specific task, failing to develop a comprehensive machine learning architecture suitable for modeling tasks, which limits their ability to effectively represent complex industrial scenarios. Second, industrial datasets often contain underlying associations (e.g., monotonicity or periodicity) that are not adequately exploited by current research, which can degrade model's predictive performance. To address these limitations, this paper proposes the Recurrent Perceptron-based Channel Attention Transformer Encoder (RP-CATE), with three distinctive characteristics: 1: We developed a novel architecture by replacing the self-attention mechanism with channel attention and incorporating our proposed Recurrent Perceptron (RP) Module into Transformer, achieving enhanced effectiveness for industrial modeling tasks compared to the original Transformer. 2: We proposed a new data type called Pseudo-Image Data (PID) tailored for channel attention requirements and developed a cyclic sliding window method for generating PID. 3: We introduced the concept of Pseudo-Sequential Data (PSD) and a method for converting industrial datasets into PSD, which enables the RP Module to capture the underlying associations within industrial dataset more effectively. An experiment aimed at hybrid modeling in chemical engineering was conducted by using RP-CATE and the experimental results demonstrate that RP-CATE achieves the best performance compared to other baseline models.

Yihan Wang, Peiyu Liu, Runyu Chen et al.

Text-to-SQL technology has evolved rapidly, with diverse academic methods achieving impressive results. However, deploying these techniques in real-world systems remains challenging due to limited integration tools. Despite these advances, we introduce Squrve, a unified, modular, and extensive Text-to-SQL framework designed to bring together research advances and real-world applications. Squrve first establishes a universal execution paradigm that standardizes invocation interfaces, then proposes a multi-actor collaboration mechanism based on seven abstracted effective atomic actor components. Experiments on widely adopted benchmarks demonstrate that the collaborative workflows consistently outperform the original individual methods, thereby opening up a new effective avenue for tackling complex real-world queries. The codes are available at https://github.com/Satissss/Squrve.

Han Peng, Peiyu Liu, Zican Dong et al.

Diffusion language models (DLMs) have emerged as a promising alternative to the long-dominant autoregressive (AR) paradigm, offering a parallelable decoding process that could yield greater efficiency. Yet, in practice, current open-source DLMs often underperform their AR counterparts in speed, limiting their real-world utility. This work presents a systematic study of DLM efficiency, identifying key issues in prior evaluation methods. Through empirical benchmarking and a theoretical analysis, we demonstrate that AR models generally achieve higher throughput, while DLMs consistently lag. We also investigate acceleration strategies, finding that techniques like dual cache and parallel decoding mainly offer gains at small batch sizes, with their benefits diminishing upon scaling. Our findings underscore the necessity of robust evaluation methods and improved acceleration strategies to advance research on DLMs.

Tong Ye, Weigang Huang, Xuhong Zhang et al.

Large Language Models (LLMs), particularly Code LLMs, have demonstrated impressive performance in code generation. Current research primarily focuses on the correctness of generated code, while efficiency remains less explored. Recent works have focused on modifying the initial version of the code to improve its efficiency. However, such refinements are limited by the algorithmic design and overall logic of the initial code, resulting in only incremental improvements. In contrast, when human developers write high-quality code, they typically begin by designing several potential solutions at the logical level, evaluating various algorithms and their complexities, and then proceeding to implement and optimize the solution. In this study, we introduce \tool: \uline{L}arge \uline{L}anguage \uline{M}odel for Code \uline{Effi}ciency, a novel framework that enables LLMs to generate code that balances both efficiency and correctness. Specifically, \tool divides the efficiency optimization process into two domains: algorithmic exploration in the logic domain and implementation optimization in the code domain. The correctness of the code is then guaranteed through a synthetic test case refinement process. This approach, which prioritizes efficiency before ensuring correctness, offers a new paradigm for efficient code generation. Experiments demonstrate that \tool consistently improves both efficiency and correctness, achieving new state-of-the-art performance in code efficiency benchmarks across various LLM backbones.

Peiyu Liu, Ze-Feng Gao, Wayne Xin Zhao et al.

Key-value~(KV) caching is an important technique to accelerate the inference of large language models~(LLMs), but incurs significant memory overhead. To compress the size of KV cache, existing methods often compromise precision or require extra data for calibration, limiting their practicality in LLM deployment. In this paper, we introduce \textbf{DecoQuant}, a novel data-free low-bit quantization technique based on tensor decomposition methods, to effectively compress KV cache. Our core idea is to adjust the outlier distribution of the original matrix by performing tensor decomposition, so that the quantization difficulties are migrated from the matrix to decomposed local tensors. Specially, we find that outliers mainly concentrate on small local tensors, while large tensors tend to have a narrower value range. Based on this finding, we propose to apply low-bit quantization to the large tensor, while maintaining high-precision representation for the small tensor. Furthermore, we utilize the proposed quantization method to compress the KV cache of LLMs to accelerate the inference and develop an efficient dequantization kernel tailored specifically for DecoQuant. Through extensive experiments, DecoQuant demonstrates remarkable efficiency gains, showcasing up to a $\sim$75\% reduction in memory footprint while maintaining comparable generation quality.

Zican Dong, Han Peng, Peiyu Liu et al.

Mixture-of-Experts (MoE) models achieve a favorable trade-off between performance and inference efficiency by activating only a subset of experts. However, the memory overhead of storing all experts remains a major limitation, especially in large-scale MoE models such as DeepSeek-R1(671B). In this study, we investigate domain specialization and expert redundancy in large-scale MoE models and uncover a consistent behavior we term few-shot expert localization, with only a few in-domain demonstrations, the model consistently activates a sparse and stable subset of experts on tasks within the same domain. Building on this observation, we propose a simple yet effective pruning framework, EASY-EP, that leverages a few domain-specific demonstrations to identify and retain only the most relevant experts. EASY-EP comprises two key components: output-aware expert importance assessment and expert-level token contribution estimation. The former evaluates the importance of each expert for the current token by considering the gating scores and L2 norm of the outputs of activated experts, while the latter assesses the contribution of tokens based on representation similarities before and after routed experts. Experiments on DeepSeek-R1 and DeepSeek-V3-0324 show that our method can achieve comparable performances and $2.99\times$ throughput under the same memory budget with full model with only half the experts.

Jiahui Xiang, Tong Ye, Peiyu Liu et al.

Modelica is a widely adopted language for simulating complex physical systems, yet effective model creation and optimization require substantial domain expertise. Although large language models (LLMs) have demonstrated promising capabilities in code generation, their application to modeling remains largely unexplored. To address this gap, we have developed benchmark datasets specifically designed to evaluate the performance of LLMs in generating Modelica component models and test cases. Our evaluation reveals substantial limitations in current LLMs, as the generated code often fails to simulate successfully. To overcome these challenges, we propose a specialized workflow that integrates supervised fine-tuning, graph retrieval-augmented generation, and feedback optimization to improve the accuracy and reliability of Modelica code generation. The evaluation results demonstrate significant performance gains: the maximum improvement in pass@1 reached 0.3349 for the component generation task and 0.2457 for the test case generation task. This research underscores the potential of LLMs to advance intelligent modeling tools and offers valuable insights for future developments in system modeling and engineering applications.

Gen Li, Peiyu Liu

Vision-Language Models (VLMs) excel at visual reasoning but still struggle with integrating external knowledge. Retrieval-Augmented Generation (RAG) is a promising solution, but current methods remain inefficient and often fail to maintain high answer quality. To address these challenges, we propose VideoSpeculateRAG, an efficient VLM-based RAG framework built on two key ideas. First, we introduce a speculative decoding pipeline: a lightweight draft model quickly generates multiple answer candidates, which are then verified and refined by a more accurate heavyweight model, substantially reducing inference latency without sacrificing correctness. Second, we identify a major source of error - incorrect entity recognition in retrieved knowledge - and mitigate it with a simple yet effective similarity-based filtering strategy that improves entity alignment and boosts overall answer accuracy. Experiments demonstrate that VideoSpeculateRAG achieves comparable or higher accuracy than standard RAG approaches while accelerating inference by approximately 2x. Our framework highlights the potential of combining speculative decoding with retrieval-augmented reasoning to enhance efficiency and reliability in complex, knowledge-intensive multimodal tasks.

Tong Ye, Lingfei Wu, Tengfei Ma et al.

Automatically generating human-readable text describing the functionality of a program is the intent of source code summarization. Although neural language models achieve significant performance in this field, they are limited by their inability to access external knowledge. To address this limitation, an emerging trend is combining neural models with external knowledge through retrieval methods. Previous methods have relied on the sentence-level retrieval paradigm on the encoder side. However, this paradigm is coarse-grained, noise-filled and cannot directly take advantage of the high-quality retrieved summary tokens on the decoder side. In this paper, we propose a fine-grained Token-level retrieval-augmented mechanism (Tram) on the decoder side rather than the encoder side to enhance the performance of neural models and produce more low-frequency tokens in generating summaries. Furthermore, to overcome the challenge of token-level retrieval in capturing contextual code semantics, we also propose integrating code semantics into individual summary tokens. The results of extensive experiments and human evaluation show that our token-level retrieval-augmented approach significantly improves performance and is more interpretable.

Peiyu Liu, Ze-Feng Gao, Wayne Xin Zhao et al.

This paper presents a novel pre-trained language models (PLM) compression approach based on the matrix product operator (short as MPO) from quantum many-body physics. It can decompose an original matrix into central tensors (containing the core information) and auxiliary tensors (with only a small proportion of parameters). With the decomposed MPO structure, we propose a novel fine-tuning strategy by only updating the parameters from the auxiliary tensors, and design an optimization algorithm for MPO-based approximation over stacked network architectures. Our approach can be applied to the original or the compressed PLMs in a general way, which derives a lighter network and significantly reduces the parameters to be fine-tuned. Extensive experiments have demonstrated the effectiveness of the proposed approach in model compression, especially the reduction in finetuning parameters (91% reduction on average).

Yuqi Huo, Manli Zhang, Guangzhen Liu et al.

Multi-modal pre-training models have been intensively explored to bridge vision and language in recent years. However, most of them explicitly model the cross-modal interaction between image-text pairs, by assuming that there exists strong semantic correlation between the text and image modalities. Since this strong assumption is often invalid in real-world scenarios, we choose to implicitly model the cross-modal correlation for large-scale multi-modal pre-training, which is the focus of the Chinese project `WenLan' led by our team. Specifically, with the weak correlation assumption over image-text pairs, we propose a two-tower pre-training model called BriVL within the cross-modal contrastive learning framework. Unlike OpenAI CLIP that adopts a simple contrastive learning method, we devise a more advanced algorithm by adapting the latest method MoCo into the cross-modal scenario. By building a large queue-based dictionary, our BriVL can incorporate more negative samples in limited GPU resources. We further construct a large Chinese multi-source image-text dataset called RUC-CAS-WenLan for pre-training our BriVL model. Extensive experiments demonstrate that the pre-trained BriVL model outperforms both UNITER and OpenAI CLIP on various downstream tasks.