论文检索

Zezhen Ding, Zhen Tan, Jiheng Zhang, Tianlong Chen

Optimization modeling and solving are fundamental to the application of Operations Research (OR) in real-world decision making, yet the process of translating natural language problem descriptions into formal models and solver code remains highly expertise intensive. While recent advances in large language models (LLMs) have opened new opportunities for automation, the generalization ability and data efficiency of existing LLM-based methods are still limited, asmost require vast amounts of annotated or synthetic data, resulting in high costs and scalability barriers. In this work, we present OR-R1, a data-efficient training framework for automated optimization modeling and solving. OR-R1 first employs supervised fine-tuning (SFT) to help the model acquire the essential reasoning patterns for problem formulation and code generation from limited labeled data. In addition, it improves the capability and consistency through Test-Time Group Relative Policy Optimization (TGRPO). This two-stage design enables OR-R1 to leverage both scarce labeled and abundant unlabeled data for effective learning. Experiments show that OR-R1 achieves state-of-the-art performance with an average solving accuracy of 67.7%, using only 1/10 the synthetic data required by prior methods such as ORLM, exceeding ORLM’s solving accuracy by up to 4.2%. Remarkably, OR-R1 outperforms ORLM by over 2.4% with just 100 synthetic samples. Furthermore, TGRPO contributes an additional 3.1%–6.4% improvement in accuracy, significantly narrowing the gap between single-attempt (Pass@1) and multi-attempt (Pass@8) performance from 13% to 7%. Extensive evaluations across diverse real-world benchmarks demonstrate that OR-R1 provides a robust, scalable, and cost-effective solution for automated OR optimization problem modeling and solving, lowering the expertise and data barriers for industrial OR applications.

Ke Ding, Brian Parker, Jiayu Wen

Pre-training large language models on genomic sequences has become a powerful approach for learning biologically meaningful representations. While masked language modeling (MLM)-based approaches, such as DNABERT and Nucleotide Transformer (NT), achieve strong performance, they are hindered by inefficiencies due to partial token supervision, pre-training/fine-tuning mismatches, and high computational costs. We introduce NucEL, the first ELECTRA-style pre-training framework for genomic foundation models, which overcomes these challenges. Through a discriminator network identifying tokens modified by a generator, NucEL achieves comprehensive token-level supervision across all sequence positions, thereby markedly improving training efficiency relative to the partial supervision of masked positions inherent in MLM frameworks. By integrating ModernBERT’s architectural advancements, including hybrid local-global attention and flash attention mechanisms, NucEL establishes an optimized BERT architecture for genomic sequence modeling. Unlike traditional methods that tokenize genomic sequences into 6-mers, NucEL implements single-nucleotide tokenization, enabling fine-grained resolution and improving both efficiency and interpretability. Pre-trained on the human genome only, NucEL achieves state-of-the-art performance on benchmark datasets across diverse downstream tasks in both human and non-human species, including regulatory element identification (e.g., promoters, enhancers), transcription factor binding prediction in human and mouse, open chromatin region classification, and histone modification profiles, surpassing MLM-based models of similar size and rivaling models 25 times larger, such as NT. Ablation studies provide critical insights into tokenization and masking strategies, optimizing ELECTRA-style pretraining for DNA sequences. Attention analyses reveal NucEL’s superior ability to capture biologically relevant sequence motifs compared to NT, offering valuable insights into its hierarchical learning process and regulatory element modeling capabilities. This work highlights the potential of ELECTRA-style pretraining as an efficient and effective strategy for advancing genomic representation learning with broad implications for future genomic research.

Xiaolei Diao, Zhihan Zhou, Lida Shi, Ting Wang, Ruihua Qi, Daqian Shi, Hao Xu

Constructing historical language models (LMs) plays a crucial role in aiding archaeological provenance studies and understanding ancient cultures. However, existing resources present major challenges for training effective LMs on historical texts. First, the scarcity of historical language samples renders unsupervised learning approaches based on large text corpora highly inefficient, hindering effective pre-training. Moreover, due to the considerable temporal gap and complex evolution of ancient scripts, the absence of comprehensive character encoding schemes limits the digitization and computational processing of ancient texts, particularly in early Chinese writing. To address these challenges, we introduce InteChar, a unified and extensible character list that integrates unencoded oracle bone characters with traditional and modern Chinese. InteChar enables consistent digitization and representation of historical texts, providing a foundation for robust modeling of ancient scripts. To evaluate the effectiveness of InteChar, we construct the Oracle Corpus Set (OracleCS), an ancient Chinese corpus that combines expert-annotated samples with LLM-assisted data augmentation, centered on Chinese oracle bone inscriptions. Extensive experiments show that models trained with InteChar on OracleCS achieve substantial improvements across various historical language understanding tasks, confirming the effectiveness of our approach and establishing a solid foundation for future research in ancient Chinese NLP.

Bin Deng, Huifang Ma, Ruijia Zhang, Meihuizi Jia, Rui Bing

Protein-Protein Interactions (PPIs) prediction is crucial for understanding cellular functions and disease mechanisms. Existing deep learning–based methods primarily rely on direct interaction within the PPI network to update protein representations. However, (1) such networks overlook the potential associations between functionally similar proteins, limiting the smoothing capability of Graph Neural Networks (GNNs) in learning representations for similar nodes. (2) Additionally, most approaches fail to adequately model the latent dependencies among interaction types (edge labels), which hinders their performance in PPI prediction tasks. To address these limitations, we propose TELC-PPI, a topology-enhanced and label correlation-aware model for protein-protein interactions prediction. Specifically, TELC-PPI first identifies similar proteins by leveraging both the topological information of the PPI network and the label distributions of nodes, constructing similarity edges. Then, it incorporates label co-occurrence statistics into the learning of label embeddings. Experimental results on multiple datasets and under various data split settings demonstrate that TELC-PPI significantly outperforms existing methods, validating the effectiveness of our model design.

Shengqi Dang, Fu Chai, Jiaxin Li, Chao Yuan, Wei Ye, Nan Cao

The rise of 3D generative models has enabled automatic 3D geometry and texture synthesis from multimodal inputs (e.g., text or images). However, these methods often ignore physical constraints and manufacturability considerations. In this work, we address the challenge of producing 3D designs that are both lightweight and self-supporting. We present DensiCrafter, a framework for generating lightweight, self-supporting 3D hollow structures by optimizing the density field. Starting from coarse voxel grids produced by Trellis, we interpret these as continuous density fields to optimize and introduce three differentiable, physically constrained, and simulation-free loss terms. Additionally, a mass regularization penalizes unnecessary material, while a restricted optimization domain preserves the outer surface. Our method seamlessly integrates with pretrained Trellis-based models (e.g., Trellis, DSO) without any architectural changes. In extensive evaluations, we achieve up to 43% reduction in material mass on the text-to-3D task. Compared to state-of-the-art baselines, our method could improve the stability and maintain high geometric fidelity. Real-world 3D-printing experiments confirm that our hollow designs can be reliably fabricated and could be self-supporting.

Longchao Da, David Isele, Hua Wei, Manish Saroya

Being able to anticipate the motion of surrounding agents is essential for the safe operation of autonomous driving systems in dynamic situations. While various methods have been proposed for trajectory prediction, the current evaluation practices still rely on error-based metrics (e.g., ADE, FDE), which reveal the accuracy from a post-hoc view but ignore the actual effect the predictor brings to the self-driving vehicles (SDVs), especially in complex interactive scenarios: a high-quality predictor not only chases accuracy, but should also captures all possible directions a neighbor agent might move, to support the SDVs' cautious decision-making. Given that the existing metrics hardly account for this standard, in our work, we propose a comprehensive pipeline that adaptively evaluates the predictor's performance by two dimensions: accuracy and diversity. Based on the criticality of the driving scenario, these two dimensions are dynamically combined and result in a final score for the predictor's performance. Extensive experiments on a closed-loop benchmark using a real-world dataset show that our pipeline yields a more reasonable evaluation than traditional metrics by better reflecting the correlation of the predictors' evaluation with the autonomous vehicles' driving performance. This evaluation pipeline shows a robust way to select a predictor that potentially contributes most to the SDV's driving performance.

Xing Cui, Yueying Zou, Zekun Li, Peipei Li, Xinyuan Xu, Xuannan Liu, Huaibo Huang

Real-world multimodal misinformation often arises from mixed forgery sources, requiring dynamic reasoning and adaptive verification. However, existing methods mainly rely on static pipelines and limited tool usage, limiting their ability to handle such complexity and diversity. To address this challenge, we propose T2Agent, a novel misinformation detection agent that incorporates an extensible toolkit with Monte Carlo Tree Search (MCTS). The toolkit consists of modular tools such as web search, forgery detection, and consistency analysis. Each tool is described using standardized templates, enabling seamless integration and future expansion. To avoid inefficiency from using all tools simultaneously, a greedy search-based selector is proposed to identify a task-relevant subset. This subset then serves as the action space for MCTS to dynamically collect evidence and perform multi-source verification. To better align MCTS with the multi-source nature of misinformation detection, T2Agent extends traditional MCTS with multi-source verification, which decomposes the task into coordinated subtasks targeting different forgery sources. A dual reward mechanism containing a reasoning trajectory score and a confidence score is further proposed to encourage a balance between exploration across mixed forgery sources and exploitation for more reliable evidence. We conduct ablation studies to confirm the effectiveness of the tree search mechanism and tool usage. Extensive experiments further show that T2Agent consistently outperforms existing baselines on challenging mixed-source multimodal misinformation benchmarks, demonstrating its strong potential as a training-free detector.

Xing Cui, Jingzheng Wu, Wenxiang Ou, Tianyue Luo, Zhiyuan Li, Xiang Ling

Vulnerability-Fixing Commit Identification(VFCI) is a critical task in software security maintenance that aims to automatically identify code commits that patch security vulnerabilities. However, existing approaches face challenges in handling low-quality commit messages and entangled commits, which limit their identification performance. To address these issues, we propose VFCionX, a novel VFCI framework that integrates large and small language models in a collaborative architecture. VFCionX consists of three core modules: Message Classifier, Patch Classifier, and Ensemble Classifier. The Message Classifier employs a multi-source contextual augmentation strategy to enhance the quality of commit messages and fine-tunes the Qwen2.5-1.5B model, significantly improving classification performance in the textual modality. The Patch Classifier combines heuristic rules with a Qwen2.5-Coder-7B-driven file selector to filter noise from entangled commits, and incorporates a line-level feature extractor based on CodeBERT and CNN to capture local pattern differences between added and deleted code lines. The Ensemble Classifier integrates predictions from both channels using the AdaBoost algorithm, enhancing model robustness and generalization. Experimental results on five popular C/C++ repositories comprising 24,630 commits show that VFCionX achieves an F1-score of 81.47%, outperforming the best baseline by 9.42%. Ablation studies validate the effectiveness of each component, while sensitivity analysis reveals optimal parameter settings for balancing performance and noise resilience. This work provides a new and effective solution for robust vulnerability patch identification.

Wei Cui, Lukai Fan, Zhenghua Chen, Min Wu, Shili Xiang, Haixia Wang, Bing Li

Sensory Temporal Action Detection (STAD) aims to localize and classify human actions within long, untrimmed sequences captured by non-visual sensors such as WiFi or inertial measurement units (IMUs). Unlike video-based TAD, STAD poses unique challenges due to the low-dimensional, noisy, and heterogeneous nature of sensory data, as well as the real-time and resource constraints on edge devices. While recent STAD models have improved detection performance, their high computational cost hampers practical deployment. In this paper, we propose SlimSTAD, a simple yet effective framework that achieves both high accuracy and low latency for STAD. SlimSTAD features a novel Decoupled Channel Modeling (DCM) encoder, which preserves modality-specific temporal features and enables efficient inter-channel aggregation via lightweight graph attention. An anchor-free cascade predictor then refines action boundaries and class predictions in a two-stage design without dense proposals. Experiments on two real-world datasets demonstrate that SlimSTAD outperforms strong video-derived and sensory baselines by an average of 2.1 mAP, while significantly reducing GFLOPs, parameters, and latency, validating its effectiveness for real-world, edge-aware STAD deployment.

Kaiwen Cheng, Yutian Liu, Zhiwei Nie, Mujie Lin, Yanzhen Hou, Yiheng Tao, Chang Liu, Jie Chen, Youdong Mao, Yonghong Tian

Understanding the structural dynamics of biomolecules is crucial for uncovering biological functions. As molecular dynamics (MD) simulation data becomes more available, deep generative models have been developed to synthesize realistic MD trajectories. However, existing methods produce fixed-length trajectories by jointly denoising high-dimensional spatiotemporal representations, which conflicts with MD’s frame-by-frame integration process and fails to capture time-dependent conformational diversity. Inspired by MD's sequential nature, we introduce a new probabilistic autoregressive (ProAR) framework for trajectory generation. ProAR uses a dual-network system that models each frame as a multivariate Gaussian distribution and employs an anti-drifting sampling strategy to reduce cumulative errors. This approach captures conformational uncertainty and time-coupled structural changes while allowing flexible generation of trajectories of arbitrary length. Experiments on ATLAS, a large-scale protein MD dataset, demonstrate that for long trajectory generation, our model achieves a 7.5% reduction in reconstruction RMSE and an average 25.8% improvement in conformation change accuracy compared to previous state-of-the-art methods. For conformation sampling task, it performs comparably to specialized time-independent models, providing a flexible and dependable alternative to standard MD simulations.

Guanjie Cheng, Boyi Li, Peihan Wu, Feiyi Chen, Xinkui Zhao, Mengying Zhu, Shuiguang Deng

The wide spreading of Internet of Things (IoT) sensors generates vast spatio-temporal data streams, but ensuring data credibility is a critical yet unsolved challenge for applications like smart homes. While spatio-temporal graph (STG) models are a leading paradigm for such data, they often fall short in dynamic, human-centric environments due to two fundamental limitations: (1) their reliance on static graph topologies, which fail to capture physical, event-driven dynamics, and (2) their tendency to confuse spurious correlations with true causality, undermining robustness in human-centric environments. To address these gaps, we propose the Dynamic Causal Spatio-Temporal Graph Network (DyC-STG), a novel framework designed for real-time data credibility analysis in IoT. Our framework features two synergistic contributions: an event-driven dynamic graph module that adapts the graph topology in real-time to reflect physical state changes, and a causal reasoning module to distill causally-aware representations by strictly enforcing temporal precedence. To facilitate the research in this domain we release two new real-world datasets. Comprehensive experiments show that DyC-STG establishes a new state-of-the-art, outperforming the strongest baselines by 1.4 percentage points and achieving an F1-Score of up to 0.930.

Yunhao Chen, Shujie Wang, Difan Zou, Xingjun Ma

As diffusion probabilistic models (DPMs) become central to Generative AI (GenAI), understanding their memorization behavior is essential for evaluating risks such as data leakage, copyright infringement, and trustworthiness. While prior research finds conditional DPMs highly susceptible to data extraction attacks using explicit prompts, unconditional models are often assumed to be safe. We challenge this view by introducing Surrogate condItional Data Extraction (SIDE), a general framework that constructs data-driven surrogate conditions to enable targeted extraction from any DPM. Through extensive experiments on CIFAR-10, CelebA, ImageNet, and LAION-5B, we show that SIDE can successfully extract training data from so-called safe unconditional models, outperforming baseline attacks even on conditional models. Complementing these findings, we present a unified theoretical framework based on informative labels, demonstrating that all forms of conditioning, explicit or surrogate, amplify memorization. Our work redefines the threat landscape for DPMs, establishing precise conditioning as a fundamental vulnerability and setting a new, stronger benchmark for model privacy evaluation.

Yujie Chen, Tengfei Ma, Yuansheng Liu, Leyi Wei, Shu Wu, Dongsheng Cao, Yiping Liu, Xiangxiang Zeng

Identifying suitable reaction conditions is critical for chemical synthesis, as they directly affect yield, selectivity, and transformation feasibility. While recent methods have shown promising results, most approaches either encode reactants and products independently or rely on rule-based reaction graphs, both of which constrain the ability of the model to capture condition-relevant structural transformations. In this work, we propose TRACE, a transformation-aware graph refinement framework for reaction condition prediction. TRACE constructs atom-level joint graphs that integrate both reactant and product structures to represent condition-relevant transformations. A structure-aware encoder enriches atom features with local chemical context, followed by a dynamic interaction refinement module that adaptively infers task-specific edges. To further guide the model toward condition-relevant patterns, a mechanism regularized graph encoder incorporates reaction center information, enabling more accurate modeling of transformation mechanisms. Experiments on benchmark datasets show that TRACE achieves state-of-the-art performance across multiple condition types. The integration of transformation-aware refinement leads to improvements in prediction accuracy and generalization, while maintaining robust performance in challenging and realistic synthesis planning scenarios.

Yuhang Chen, Jie Sun, Jialin Fan, Jian Sun

Traffic simulation is essential for validating the safety and reliability of autonomous driving systems, yet data-driven simulation methods often struggle with distribution shifts, limiting their generalizability across diverse datasets (domains). To address this, we present Causal Driving Pattern Transfer (CDPT), a novel two-stage knowledge distillation framework built upon diffusion model to enhance cross-domain generalizability. In Phase I, we implement hybrid self-distillation within the source domain by integrating feature-, response-, and contrastive-level distillation, which enables the model to decompose complex driving behaviors into their core causal components, including scene-conditioned driven patterns, multi-agent interaction dynamics and casual saliency. In Phase II, we introduce a continual distillation strategy: few-shot samples from the target domain are used to initiate generation of diverse synthetic scenarios, allowing the student model to continually adapt to novel environments without retraining on large-scale data. Extensive experiments demonstrate that CDPT achieves strong generalization in both open-loop and closed-loop simulations, effectively generating realistic, interaction-aware behaviors that are critical for scalable and reliable autonomous driving testing.

Xuanzhong Chen, Ye Jin, Xiaohao Mao, Lun Wang, Shuyang Zhang, Ting Chen

Rare diseases, despite their low individual incidence, collectively impact around 300 million people worldwide due to the vast number of diseases. The involvement of multiple organs and systems, and the shortage of specialized doctors with relevant experience, make diagnosing and treating rare diseases more challenging than common diseases. Recently, agents powered by large language models (LLMs) have demonstrated notable applications across various domains. In the medical field, some agent methods have outperformed direct prompts in question-answering tasks from medical examinations. However, current agent frameworks are not well-adapted to real-world clinical scenarios, especially those involving the complex demands of rare diseases. To bridge this gap, we introduce RareAgents, the first LLM-driven multi-disciplinary team decision-support tool designed specifically for the complex clinical context of rare diseases. RareAgents integrates advanced Multidisciplinary Team (MDT) coordination, memory mechanisms, and medical tools utilization, leveraging Llama-3.1-8B/70B as the base model. Experimental results show that RareAgents outperforms state-of-the-art domain-specific models, GPT-4o, and current agent frameworks in diagnosis and treatment for rare diseases. Furthermore, we contribute a novel rare disease dataset, MIMIC-IV-Ext-Rare, to facilitate further research in this field.

Shaorong Chen, Jingbo Zhou, Jun Xia

The discovery of novel proteins relies on sensitive protein identification, for which de novo peptide sequencing (DNPS) from mass spectra is a crucial approach. While deep learning has advanced DNPS, existing models inadequately enforce the fundamental mass consistency constraint—that a predicted peptide's mass must match the experimental measured precursor mass. Previous DNPS methods often treat this critical information as a simple input feature or use it in post-processing, leading to numerous implausible predictions that do not adhere to this fundamental physical property. To address this limitation, we introduce DiffuNovo, a novel regressor-guided diffusion model for de novo peptide sequencing that provides explicit peptide-level mass control. Our approach integrates the mass constraint at two critical stages: during training, a novel peptide-level mass loss guides model optimization, while at inference, regressor-based guidance from gradient-based updates in the latent space steers the generation to compel the predicted peptide adheres to the mass constraint. Comprehensive evaluations on established benchmarks demonstrate that DiffuNovo surpasses state-of-the-art methods in DNPS accuracy. Additionally, as the first DNPS model to employ a diffusion model as its core backbone, DiffuNovo leverages the powerful controllability of diffusion architecture and achieves a significant reduction in mass error, thereby producing much more physically plausible peptides. These innovations represent a substantial advancement toward robust and broadly applicable DNPS. The source code is available in the supplementary material.

Qiyuan Chen, Ajay Annamareddy, Ying-Fei Li, Dane Morgan, Bu Wang

Disordered materials such as glasses, unlike crystals, lack long‑range atomic order and have no periodic unit cells, yielding a high‑dimensional configuration space with widely varying properties. The complexity not only increases computational costs for atomistic simulations but also makes it difficult for generative AI models to deliver accurate property predictions and realistic structure generation. In this work, we introduce GlassVAE, a hierarchical graph variational autoencoder that uses graph representations to learn compact, translation‑, and permutation‑invariant embeddings of atomic configurations. The resulting structured latent space not only enables efficient generation of novel, physically plausible structures but also supports exploration of the glass energy landscape. To enforce structural realism and physical fidelity, we augment GlassVAE with two physics‑informed regularizers: a radial distribution function (RDF) loss that captures characteristic short‑ and medium‑range ordering and an energy regression loss that reflects the broad configurational energetics. Both theoretical analysis and experimental results highlight the critical impact of these regularizers. By encoding high‑dimensional atomistic data into a compact latent vector and decoding it into structures with accurate energy predictions, GlassVAE provides a fast, physics‑aware path for modeling and designing disordered materials.

Letian Chen, Runhan Shi, Gufeng Yu, Yang Yang

Aligning molecular sequence representations (e.g., SMILES notations) with textual descriptions is critical for applications spanning drug discovery, materials design, and automated chemical literature analysis. Existing methodologies typically treat molecular captioning (molecule-to-text) and text-based molecular design (text-to-molecule) as separate tasks, relying on supervised fine-tuning or contrastive learning pipelines. These approaches face three key limitations: (i) conventional metrics like BLEU prioritize linguistic fluency over chemical accuracy, (ii) training datasets frequently contain chemically ambiguous narratives with incomplete specifications, and (iii) independent optimization of generation directions leads to bidirectional inconsistency. To address these issues, we propose RTMol, a bidirectional alignment framework that unifies molecular captioning and text-to-SMILES generation through self-supervised round-trip learning. The framework introduces novel round-trip evaluation metrics and enables unsupervised training for molecular captioning without requiring paired molecule-text corpora. Experiments demonstrate that RTMol enhances bidirectional alignment performance by up to 47% across various LLMs, establishing an effective paradigm for joint molecule-text understanding and generation.

Junyang Chen, Yueqian Li, Ka Chung Ng, Huan Wang, Liang-Jie Zhang

The rapid proliferation of social media platforms has led to a surge in multimodal fake news, where deceptive content often combines text and images to mislead audiences. Traditional unimodal detection methods struggle to address the complexity of such content, necessitating holistic multimodal approaches. While the latest advancements in Multimodal Large Language Models (MLLMs) offer new opportunities for enhancing detection performance by analyzing multi-dimensional features, including source credibility, cross-modal contradictions, emotional bias, and manipulative writing patterns, these methods suffer from a key flaw: a susceptibility to hallucinations or erroneous reasoning, which can lead to flawed conclusions and ultimately biased detection results. We propose the Multimodal Fake News Detection via Multi-perspective Rationale Generation and Verification (MMRGV) model to mitigate this challenge. Our method employs a cross-verification mechanism to screen and reconcile contradictions among different rationales, thereby preserving the LLM's analytical advantages while mitigating the impact of erroneous reasoning or hallucinations on the final detection. Subsequently, these optimized rationales are fused via an adaptive weighting strategy to output a robust final prediction. Extensive experiments on three benchmark datasets (Twitter, Weibo, and GossipCop) demonstrate the superiority of our method, achieving state-of-the-art accuracy of 0.9972, 0.9663, and 0.8772, respectively, and significantly outperforming existing baselines. These results validate the effectiveness of multi-perspective rationale generation and cross-verification in enhancing multimodal fake news detection, offering a resilient solution to combat misinformation in the era of generative AI.

Shuyu Chang, Haiping Huang, Yanjun Zhang, Yujin Huang, Fu Xiao, Leo Yu Zhang

Code models are increasingly adopted in software development but remain vulnerable to backdoor attacks via poisoned training data. Existing backdoor attacks on code models face a fundamental trade-off between transferability and stealthiness. Static trigger-based attacks insert fixed dead code patterns that transfer well across models and datasets but are easily detected by code-specific defenses. In contrast, dynamic trigger-based attacks adaptively generate context-aware triggers to evade detection but suffer from poor cross-dataset transferability. Moreover, they rely on unrealistic assumptions of identical data distributions between poisoned and victim training data, limiting their practicality. To overcome these limitations, we propose Sharpness-aware Transferable Adversarial Backdoor (STAB), a novel attack that achieves both transferability and stealthiness without requiring complete victim data. STAB is motivated by the observation that adversarial perturbations in flat regions of the loss landscape transfer more effectively across datasets than those in sharp minima. To this end, we train a surrogate model using Sharpness-Aware Minimization to guide model parameters toward flat loss regions, and employ Gumbel-Softmax optimization to enable differentiable search over discrete trigger tokens for generating context-aware adversarial triggers. Experiments across three datasets and two code models show that STAB outperforms prior attacks in terms of transferability and stealthiness. It achieves a 73.2% average attack success rate after defense, outperforming static trigger–based attacks that fail under defense. STAB also surpasses the best dynamic trigger–based attack by 12.4% in cross-dataset attack success rate and maintains performance on clean inputs.