SLShenghua Liu

Shenghua Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Shenghua Liu has a total of 110 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    From Trajectories to Experience: A Survey of Experience-Driven LLM Agents

    by Xiaohui Yan, Yingchen Zhang, Shiyao Liu, Xiaofei Huang, Yuxin Yue, Zaiyu Xia, Jiayu Yao, Xinjie Chen, Yun Hu, Furong Li, et al.

    Large language model (LLM) agents can plan, use tools, and recover from errors within a task, yet many carry little from one task to the next and repeat mistakes they have already resolved. The root cause is a lack of experience accumulation: trajectories record what failed and how the agent recovered, but this information is seldom turned into guidance for later tasks. Research on this capability spans reflection, agent memory, workflow learning, and skill acquisition, but these lines describe similar content in different terms and lack a common account of experience itself, leaving basic questions open: what counts as experience, how it should be accumulated, used, and revised, and how to tell whether it improves later behavior. This survey addresses these questions with a unified framework. We define agent experience as action-guiding information grounded in task trajectories and retained to guide later episodes where it applies, whether as a trajectory precedent, a semantic lesson, or a reusable procedure. We focus on explicitly accessible experience, including its use as a training signal for internalization into model parameters. We describe each unit of experience by its guidance, applicability conditions, execution evidence, and reliability evaluation, so that memory entries, workflows, and skills become directly comparable, and follow experience through four lifecycle functions---formation, organization, utilization, and maintenance---that determine how it is accumulated, used, and revised. To judge reported gains, we distinguish three improvement claims---fixed experience reuse, feedback-driven sustained improvement, and improvement of the learning mechanism---by the evidence each requires. Current studies most clearly support fixed experience reuse, chiefly on tasks with tests or reference answers; evidence for the other two claims remains partial. Our analysis yields recommendations for seven application settings and a central design principle: keep guidance connected to its applicability conditions and to the evidence that supports or challenges it. Finally, we outline eight open problems, from learning with human expert records to testing whether gains accumulate, toward agents whose improvement is cumulative, transferable, and correctable.

  • HighlightBench: Benchmarking and Diagnosing Markup-Driven Table Reasoning in Scientific Documents

    by Lexin Wang, Shenghua Liu, Yiwei Wang, Yujun Cai, Yuyao Ge, Jiayu Yao, Jiafeng Guo, Xueqi Cheng

  • Focusing by Contrastive Attention: Enhancing VLMs’ Visual Reasoning

    by Yuyao Ge, Shenghua Liu, Yiwei Wang, Lingrui Mei, Baolong Bi, Xuanshan Zhou, Jiayu Yao, Jiafeng Guo, Xueqi Cheng

  • Open Access

    PromptCD: Test-Time Behavior Enhancement via Polarity-Prompt Contrastive Decoding

    by Baolong Bi, Yuyao Ge, Shenghua Liu, Yuchen He, Siqian Tong, Lizhe Chen, Lingrui Mei, Zehao Li, Yiwei Wang, Yujun Cai, et al.

    Reliable AI systems require large language models (LLMs) to exhibit behaviors aligned with human preferences and values. However, most existing alignment approaches operate at training time and rely on additional high-quality data, incurring significant computational and annotation costs. While recent work has shown that contrastive decoding can leverage a model's internal distributions to improve specific capabilities, its applicability remains limited to narrow behavioral scopes and scenarios. In this work, we introduce Polarity-Prompt Contrastive Decoding (PromptCD), a test-time behavior control method that generalizes contrastive decoding to broader enhancement settings. PromptCD constructs paired positive and negative guiding prompts for a target behavior and contrasts model responses-specifically token-level probability distributions in LLMs and visual attention patterns in VLMs-to reinforce desirable outcomes. This formulation extends contrastive decoding to a wide range of enhancement objectives and is applicable to both LLMs and Vision-Language Models (VLMs) without additional training. For LLMs, experiments on the "3H" alignment objectives (helpfulness, honesty, and harmlessness) demonstrate consistent and substantial improvements, indicating that post-trained models can achieve meaningful self-enhancement purely at test time. For VLMs, we further analyze contrastive effects on visual attention, showing that PromptCD significantly improves VQA performance by reinforcing behavior-consistent visual grounding. Collectively, these results highlight PromptCD as a simple, general, and cost-efficient strategy for reliable behavior control across modalities.

  • Open Access

    Gated Differentiable Working Memory for Long-Context Language Modeling

    by Lingrui Mei, Shenghua Liu, Yiwei Wang, Yuyao Ge, Baolong Bi, Jiayu Yao, Jun Wan, Ziling Yin, Jiafeng Guo, Xueqi Cheng

    Long contexts challenge transformers: attention scores dilute across thousands of tokens, critical information is often lost in the middle, and models struggle to adapt to novel patterns at inference time. Recent work on test-time adaptation addresses this by maintaining a form of working memory -- transient parameters updated on the current context -- but existing approaches rely on uniform write policies that waste computation on low-utility regions and suffer from high gradient variance across semantically heterogeneous contexts. In this work, we reframe test-time adaptation as a budget-constrained memory consolidation problem, focusing on which parts of the context should be consolidated into working memory under limited computation. We propose Gdwm (Gated Differentiable Working Memory), a framework that introduces a write controller to gate the consolidation process. The controller estimates Contextual Utility, an information-theoretic measure of long-range contextual dependence, and allocates gradient steps accordingly while maintaining global coverage. Experiments on ZeroSCROLLS and LongBench v2 demonstrate that Gdwm achieves comparable or superior performance with 4$\times$ fewer gradient steps than uniform baselines, establishing a new efficiency-performance Pareto frontier for test-time adaptation.

  • Open Access

    Reward and Guidance through Rubrics: Promoting Exploration to Improve Multi-Domain Reasoning

    by Baolong Bi, Shenghua Liu, Yiwei Wang, Siqian Tong, Lingrui Mei, Yuyao Ge, Yilong Xu, Jiafeng Guo, Xueqi Cheng

    Recent advances in reinforcement learning (RL) have significantly improved the complex reasoning capabilities of large language models (LLMs). Despite these successes, existing methods mainly focus on single-domain RL (e.g., mathematics) with verifiable rewards (RLVR), and their reliance on purely online RL frameworks restricts the exploration space, thereby limiting reasoning performance. In this paper, we address these limitations by leveraging rubrics to provide both fine-grained reward signals and offline guidance. We propose $\textbf{RGR-GRPO}$ (Reward and Guidance through Rubrics), a rubric-driven RL framework for multi-domain reasoning. RGR-GRPO enables LLMs to receive dense and informative rewards while exploring a larger solution space during GRPO training. Extensive experiments across 14 benchmarks spanning multiple domains demonstrate that RGR-GRPO consistently outperforms RL methods that rely solely on alternative reward schemes or offline guidance. Compared with verifiable online RL baseline, RGR-GRPO achieves average improvements of +7.0%, +5.4%, +8.4%, and +6.6% on mathematics, physics, chemistry, and general reasoning tasks, respectively. Notably, RGR-GRPO maintains stable entropy fluctuations during off-policy training and achieves superior pass@k performance, reflecting sustained exploration and effective breakthrough beyond existing performance bottlenecks.

  • Open Access

    Not in Sync: Unveiling Temporal Bias in Audio Chat Models

    by Jiayu Yao, Shenghua Liu, Yiwei Wang, Rundong Cheng, Lingrui Mei, Baolong Bi, Zhen Xiong, Xueqi Cheng

    Large Audio Language Models (LALMs) are increasingly applied to audio understanding and multimodal reasoning, yet their ability to locate when events occur remains underexplored. We present the first systematic study of temporal bias in LALMs, revealing a key limitation in their timestamp prediction. For example, when asked "At which second does the lecturer introduce the key formula?", models often predict timestamps that are consistently earlier or later than the ground truth. Through controlled experiments on timestamped datasets, we find that temporal bias (i) is prevalent across datasets and models, (ii) increases with audio length - even accumulating to tens of seconds in extended recordings, and (iii) varies across event types and positions. We quantify this effect with the Temporal Bias Index (TBI), measuring systematic misalignment in predicted event timings, and complement it with a visualization framework. Our findings highlight a fundamental limitation in current LALMs and call for the development of temporally robust architectures.

  • Open Access

    Focusing by Contrastive Attention: Enhancing VLMs' Visual Reasoning

    by Yuyao Ge, Shenghua Liu, Yiwei Wang, Lingrui Mei, Baolong Bi, Xuanshan Zhou, Jiayu Yao, Jiafeng Guo, Xueqi Cheng

    Vision-Language Models (VLMs) have demonstrated remarkable success across diverse visual tasks, yet their performance degrades in complex visual environments. While existing enhancement approaches require additional training, rely on external segmentation tools, or operate at coarse-grained levels, they overlook the innate ability within VLMs. To bridge this gap, we investigate VLMs' attention patterns and discover that: (1) visual complexity strongly correlates with attention entropy, negatively impacting reasoning performance; (2) attention progressively refines from global scanning in shallow layers to focused convergence in deeper layers, with convergence degree determined by visual complexity. (3) Theoretically, we prove that the contrast of attention maps between general queries and task-specific queries enables the decomposition of visual signal into semantic signals and visual noise components. Building on these insights, we propose Contrastive Attention Refinement for Visual Enhancement (CARVE), a training-free method that extracts task-relevant visual signals through attention contrasting at the pixel level. Extensive experiments demonstrate that CARVE consistently enhances performance, achieving up to 75% improvement on open-source models. Our work provides critical insights into the interplay between visual complexity and attention mechanisms, offering an efficient pathway for improving visual reasoning with contrasting attention.

  • Open Access

    A Survey of Context Engineering for Large Language Models

    by Lingrui Mei, Jiayu Yao, Yuyao Ge, Yiwei Wang, Baolong Bi, Yujun Cai, Jiazhi Liu, Mingyu Li, Zhong-Zhi Li, Duzhen Zhang, et al.

    The performance of Large Language Models (LLMs) is fundamentally determined by the contextual information provided during inference. This survey introduces Context Engineering, a formal discipline that transcends simple prompt design to encompass the systematic optimization of information payloads for LLMs. We present a comprehensive taxonomy decomposing Context Engineering into its foundational components and the sophisticated implementations that integrate them into intelligent systems. We first examine the foundational components: context retrieval and generation, context processing and context management. We then explore how these components are architecturally integrated to create sophisticated system implementations: retrieval-augmented generation (RAG), memory systems and tool-integrated reasoning, and multi-agent systems. Through this systematic analysis of over 1400 research papers, our survey not only establishes a technical roadmap for the field but also reveals a critical research gap: a fundamental asymmetry exists between model capabilities. While current models, augmented by advanced context engineering, demonstrate remarkable proficiency in understanding complex contexts, they exhibit pronounced limitations in generating equally sophisticated, long-form outputs. Addressing this gap is a defining priority for future research. Ultimately, this survey provides a unified framework for both researchers and engineers advancing context-aware AI.


  • Open Access

    a1: Steep Test-time Scaling Law via Environment Augmented Generation

    by Lingrui Mei, Shenghua Liu, Yiwei Wang, Baolong Bi, Yuyao Ge, Jun Wan, Yurong Wu, Xueqi Cheng

    Large Language Models (LLMs) have made remarkable breakthroughs in reasoning, yet continue to struggle with hallucinations, logical errors, and inability to self-correct during complex multi-step tasks. Current approaches like chain-of-thought prompting offer limited reasoning capabilities that fail when precise step validation is required. We propose Environment Augmented Generation (EAG), a framework that enhances LLM reasoning through: (1) real-time environmental feedback validating each reasoning step, (2) dynamic branch exploration for investigating alternative solution paths when faced with errors, and (3) experience-based learning from successful reasoning trajectories. Unlike existing methods, EAG enables deliberate backtracking and strategic replanning through tight integration of execution feedback with branching exploration. Our a1-32B model achieves state-of-the-art performance among similar-sized models across all benchmarks, matching larger models like o1 on competition mathematics while outperforming comparable models by up to 24.4 percentage points. Analysis reveals EAG's distinctive scaling pattern: initial token investment in environment interaction yields substantial long-term performance dividends, with advantages amplifying proportionally to task complexity. EAG's theoretical framework demonstrates how environment interactivity and systematic branch exploration together establish a new paradigm for reliable machine reasoning, particularly for problems requiring precise multi-step calculation and logical verification.