Digital Commons Network^™

On Autopilot? An Empirical Study Of Human-Ai Teaming And Review Practices In Open Source, Haoyu Gao, Peerachai Banyongrakkul, Hao Guan, Mansooreh Zahedi, Christoph Treude

Research Collection School Of Computing and Information Systems

Large Language Models (LLMs) increasingly automate software engineering tasks. While recent studies highlight the accelerated adoption of “AI as a teammate” in Open Source Software (OSS), developer interaction patterns remain under-explored. In this work, we investigated project-level guidelines and developers’ interactions with AI-assisted pull requests (PRs) by expanding the AIDev dataset to include finer-grained contributor code ownership and a comparative baseline of human-created PRs. We found that over 67.5% of AI-co-authored PRs originate from contributors without prior code ownership. Despite this, the majority of repositories lack guidelines for AI-coding agent usage. Notably, we observed a distinct interaction pattern: AI-co-authored PRs …

Distributional Vision-Language Alignment By Cauchy-Schwarz Divergence, Wenzhe Yin, Zehao Xiao, Pan Zhou, Shujian Yu, Jiayi Shen, Jan-Jakob Sonke, Stratis Gavves

Research Collection School Of Computing and Information Systems

Vision-language alignment is crucial for various downstream tasks such as cross-modal generation and retrieval. Previous multimodal approaches like CLIP utilize InfoNCE to maximize mutual information, primarily aligning pairwise samples across modalities while overlooking distributional differences. In addition, InfoNCE has inherent conflict in terms of alignment and uniformity in multimodality, leading to suboptimal alignment with modality gaps. To overcome the limitations, we propose CS-Aligner, a novel framework that performs distributional vision-language alignment by integrating Cauchy-Schwarz (CS) divergence with mutual information. CS-Aligner captures both the global distribution information of each modality and the pairwise semantic relationships. We find that the CS divergence …

Antecedents And Consequences Of Cognitive And Affective Trust In Wealth Management: Impacts On Wealth Manager Selection And Asset Under Management (Aum) Growth, Ting Hsi Chen

Dissertations and Theses Collection (Open Access)

Trust is key to wealth management decision-making; there is little empirical clarity on the relationships between different dimensions of trust and client behavior at each stage of the relationship. Previous works differentiated Cognitive trust, based on perceived competence, and Affective trust, based on relational and emotional connections. Nevertheless, the mechanisms under which these trust dimensions act during the pre- and post-account-opening phases have not yet been thoroughly studied. This research addresses this gap by exploring the stage-dependent influence of Cognitive and Affective trust on the central behavioral outcomes in wealth management.

The approach is a quantitative research based on a …

When Stress Matters Most: Developmental Timing And Socio-Ecological Stressors Among Mexican-Origin Adolescents From Low-Income Immigrant Families, Ka I Ip, Wen Wen, Sujin Lee, Wei Xiang Sim, Su Yeong Kim

Research Collection School of Social Sciences

This study investigates the dynamic, time-varying associations between multiple socio-ecological stressors and internalizing symptoms among Mexican-origin youth from low-income immigrant families. Grounded in a socioecological framework and employing time-varying effect modeling (TVEM), we examine how stressors at the interpersonal, family, and neighborhood levels differentially influence anxiety and depressive symptoms across early adolescence (ages 11–13), middle adolescence (ages 14–17), and late adolescence/emerging adulthood (ages 18–20). Participants included 604 Mexican-origin adolescents (54% female) from low-income immigrant families, assessed across three waves spanning nine years. Five distinct stressors were identified: discrimination, foreigner stress, economic stress, language brokering stress, and neighborhood violence/non-safety. Results from …

Derivation Of An Updated Brief Multivariable Prediction Model To Detect Panic-Related Anxiety In Emergency Department Patients With Cardiopulmonary Complaints, Sharon C. Sung, Felicia J. L. Ang, Arul Earnest, Leslie E. C. Lim, Shreshtha Jolly, Gilaine Rui Ng, A. John Rush, Marcus E. H. Ong

Research Collection School of Social Sciences

Background Patients with panic related-anxiety (i.e., panic attacks or panic disorder) frequently present to emergency departments (EDs) with cardiopulmonary complaints but are often undiagnosed, which can lead to recurrent visits and prolonged distress. This study aimed to derive a new symptom-based multivariable diagnostic prediction model to detect panic-related anxiety in ED patients with cardiopulmonary symptoms.Methods We conducted a single-blind prospective derivation study over 15 months in the ED of a major tertiary hospital in Singapore. Patients presenting with symptoms of palpitations, chest pain, dizziness, or difficulty breathing were assessed using the Structured Clinical Interview for DSM Disorders (SCID) to diagnose …

Are Less Hierarchical Firms Organized Around Stronger Cultures? Evidence From Big Data, Arianna Marchetti, Phanish Puranam

Research Collection Lee Kong Chian School Of Business

Research Summary: Are less hierarchical firms organized around stronger cultures instead? We analyze 1.5 million employee reviews on Glassdoor.com from 23,000 US-based firms, alongside data on managerial hierarchy estimated from 42 million professional social media profiles. Our findings confirm a negative association between managerial hierarchy and organizational culture strength. We explore two potential explanations for this association: Functional equivalence between the two, and culture fragmentation caused by managerial hierarchy. Multiple correlational tests show support for functional equivalence as a plausible explanation for the observed negative correlation. Our findings enhance our understanding of the complex relationships between organizational structure and culture …

The Effects Of Linguistic Ostracism On Job Performance. A Replication And An Extension, John Fiset, Devasheesh P. Bhave

Research Collection Lee Kong Chian School Of Business

We examined the phenomenon of linguistic ostracism—instances where a focal workgroup member perceives other members of their workgroup have rejected and/or excluded them by using a language they cannot comprehend. In a 2021 article, Fiset and Bhave observed that linguistic ostracism was related to two dimensions of job performance (interpersonal citizenship and deviance) and that disidentification served as an explanatory mechanism for the linguistic ostracism–job performance relationship. We constructively replicate and extend their work in several ways. First, we replicate prior effects on interpersonal citizenship and deviance and extend their work to focus on a third dimension of job performance: …

The Distinctiveness Effect: How Cross-Country Dissimilarities Influence Governance Decisions, Ilya Cuypers, Gokhan Ertug, Niels G. Noorderhaven, Korcan Kavusan

Research Collection Lee Kong Chian School Of Business

Transaction cost theory is one of the most commonly used theories to explain how firms govern their economic activities. In the context of cross-border collaborations, transaction cost theory proposes that dissimilarities between the partners’ home countries, which constitute a key source of behavioral uncertainty, affect how collaborations are governed (i.e. whether firms opt for equity joint ventures or non-equity alliances). Although many firms are likely to face more than one dimension of dissimilarity—for example, in terms language and religion, as well as culture—studies in transaction cost theory typically focus on how each source of dissimilarity impacts governance choices independently, in …

Grounding Is All You Need? Dual Temporal Grounding For Video Dialog, You Qin, Wei Ji, Xinze Lan, Hao Fei, Xun Yang, Dan Guo, Roger Zimmermann, Lizi Liao

Research Collection School Of Computing and Information Systems

In the realm of video dialog response generation, capturing both the essence of video content and the temporal nuances of conversation history is crucial. While some approaches rely on large-scale pretrained visual-language models, often neglecting temporal dynamics, others emphasize spatial-temporal relationships within videos but demand intricate object trajectory pre-extractions and overlook dialog temporal dynamics. This paper introduces the Dual Temporal Grounding-enhanced Video Dialog model (DTGVD), designed to bridge the gap between these two approaches. DTGVD uniquely integrates the strengths of both by emphasizing dual temporal relationships. It achieves this by predicting dialog turn-specific temporal regions, selectively filtering video content, and …

Prompt Tuning Without Labeled Samples For Zero-Shot Node Classification In Text-Attributed Graphs, Sethupathy Parameswaran, Suresh Sundaram, Yuan Fang

Research Collection School Of Computing and Information Systems

Node classification is a fundamental problem in information retrieval with many real-world applications, such as community detection in social networks, grouping articles published online and product categorization in e-commerce. Zero-shot node classification in text-attributed graphs (TAGs) presents a significant challenge, particularly due to the absence of labeled data. In this paper, we propose a novel Zero-shot Prompt Tuning (ZPT) framework to address this problem by leveraging a Universal Bimodal Conditional Generator (UBCG). Our approach begins with pre-training a graph-language model to capture both the graph structure and the associated textual descriptions of each node. Following this, a conditional generative model …

Thinkmatter: Panoramic-Aware Instructional Semantics For Monocular Vision-And-Language Navigation, Guangzhao Dai, Shuo Wang, Hao Zhao, Bin Zhu, Qianru Sun, Xiangbo Shu

Research Collection School Of Computing and Information Systems

Vision-and-Language Navigation in continuous environments (VLN-CE) requires an embodied robot to navigate the target destination following the natural language instruction. Most existing methods use panoramic RGB-D cameras for 360° observation of environments. However, these methods struggle in real-world applications because of the higher cost of panoramic RGB-D cameras. This paper studies a low-cost and practical VLN-CE setting, e.g., using monocular cameras of limited field of view, which means “Look Less” for visual observations and environment semantics. In this paper, we propose a ThinkMatter framework for monocular VLN-CE, where we motivate monocular robots to “Think More” by 1) generating novel views …

Memoryart: Enhancing Llms Via Multi-Memory Models With Adaptive Resonance Theory For Healthcare Agents, Renke Dai, Hebin Hu, Jiahui Zhang, Yilin Kang, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Though promising in healthcare consultation applications, large language models (LLMs) face critical limitations in retaining and utilizing long-term memory across multiturn interactions. In particular, existing memory enhancing paradigms are constrained by limited context windows and embedding-based retrieval, often failing to maintain task relevance and still suffering from memory prototype collapse in multi-turn healthcare consultation. To address these challenges, we propose a cognitively-inspired memory framework named MemoryART, which is grounded in Adaptive Resonance Theory (ART)—a cognitive and learning theory of how humans and animals adapt to dynamic environments. MemoryART employs three memory modules—working memory, episodic memory, and semantic memory to support …

Clique Annealing: Semi-Supervised Community Detection Under Crystallization Kinetics, Ling Cheng, Jiashu Pu, Ruicheng Liang, Qian Shao, Hezhe Qiao, Feida Zhu

Research Collection School Of Computing and Information Systems

Semi-supervised community detection seeks to find a specified community type when only few communities are labeled. Existing "select-then-refine" pipelines often start from mis-aligned cores and rely on Reinforcement-Learning or Generative Adversarial Network, increasing computational cost and limiting scalability. We address these issues with a unified energy framework under crystallization kinetics that jointly models energy, structure, and growth. Based on this perspective, we propose CLique ANNealing (CLANN), which first employs Nucleus Proposer to select candidate clique as community core under four physics-inspired criteria. A learning-free Transitive Annealer then iteratively merges neighboring cliques and repositions the nucleus, enabling spontaneous, scalable community growth. …

Artem: Enhancing Large Language Model Agents With Spatial-Temporal Episodic Memory, Cassandra Hui Ming Tan, Budhitama Subagdja, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Current large language models (LLMs) exhibit significant deficiencies in episodic memory tasks including encoding, storing, and retrieving specific information from temporally dependent events over a long period of time. Recent approaches to handle memory tasks in LLMs, such as in-context learning, retrieval-augmented generation (RAG), and fine-tuning, may resolve the long-term retention issues, but are still inadequate to handle tasks requiring chronological awareness of the stored information. We introduce Agentic Retrieval with Temporal-Episodic Memory (ARTEM), a hybrid LLM-based agent architecture integrating LLMs with a self-organizing neural network named Spatial-Temporal Episodic Memory (STEM), designed to handle episodic memory tasks. Our approach employs …

Llamoco: Instruction Tuning Of Large Language Models For Optimization Code Generation, Zeyuan Ma, Yue-Jiao Gong, Hongshu Guo, Jiacheng Chen, Yining Ma, Zhiguang Cao

Research Collection School Of Computing and Information Systems

Recently, combining the strength of large language models (LLMs) and Evolutionary Computation (EC) has shown promising results for addressing optimization problems. It typically involves either iterative next-step solution seeking or directly prompting LLMs to generate critical optimization codes. However, these methods often suffer from low computational efficiency, high sensitivity to prompt design, and a lack of domain-specific knowledge. We introduce LLaMoCo, the first instruction-tuning framework designed to adapt LLMs for solving optimization problems in a code-to-code manner. LLaMoCo features a comprehensive instruction set that includes code-style problem descriptions as input prompts and robust optimization codes from expert EC optimizers as …

Fluid Agency In Ai Systems: A Case For Functional Equivalence In Copyright, Patent, And Tort, Anirban Mukherjee, Hannah H. Chang

Research Collection Lee Kong Chian School Of Business

Modern Artificial Intelligence (AI) systems exhibit fluid agency in multi-step workflows: lacking human-like consciousness or culpability, yet they display behavior that is (i) stochastic (probabilistic and path‑dependent), (ii) dynamic (co‑evolving with user interaction), and (iii) adaptive (able to reorient across contexts). These properties generate valuable outputs but collapse attribution, irreducibly entangling human and machine inputs. Doctrines that assume traceable provenance—authorship, inventorship, and liability—fracture under this unmappability, yielding ownership gaps and moral “crumple zones.”This Article argues that only functional equivalence stabilizes doctrine under unmappability: Where provenance is indeterminate, legal frameworks should treat human and AI contributions as equivalent for allocating rights …

Purified Zero-Shot Sketch-Based Image Retrieval, Yang Zhou, Jingru Yang, Jin Wang, Kaixiang Huang, Guodong Lu, Shengfeng He

Research Collection School Of Computing and Information Systems

Sketches, as a new solution in multimedia systems that can replace natural language, are characterized by sparse visual cues such as simple strokes that differ significantly from natural images containing complex elements such as background, foreground, and texture. This misalignment poses substantial challenges for zero-shot sketch-based image retrieval (ZS-SBIR). Prior approaches match sketches to full images and tend to overlook redundant elements in natural images, leading to model distraction and semantic ambiguity. To address this issue, we introduce a distraction-agnostic framework, purified cross-domain matching (PuXIM), which operates on a straightforward principle: masking and matching. We devise a visual-cross-linguistic (VxL) sampler …

International Law, The Courts, And The Political Branches Of Singapore: Painting A Complete Picture, Benjamin Joshua Ong

Research Collection Yong Pung How School Of Law

In line with Singapore's vision of the separation of powers, the courts' duty is primarily to give effect to domestic law; the political branches take the lead in engaging with international law. A study of Singapore's interface with international law would therefore be incomplete were it to consider only the courts' role and not the political branches' model of international law as primarily a guarantor of Singapore's sovereignty and standing as a participant on the international stage. The political branches have been circumspect in engaging with international law in other areas, such as human rights, preferring a specifically Singaporean vision …

Digital Communications Between Firms And Investors: Impact Of Explanatory Responses On Investor Engagement In Online Financial Q&A, Runyu Wang, Zili Zhang, Keng Siau, Ziqiong Zhang

Research Collection School Of Computing and Information Systems

The emerging trend of digital communications between firms and investors through online question-and-answer (Q&A) platforms is recognized as a vital strategy for managing investor relations, contributing to enhanced market efficiency and information transparency through increased information exchange. Potential investors can seek responses from firm managers to address their information needs, thereby mitigating market uncertainties. To provide foundational insights, we conduct a survey of investors to assess their awareness, usage, and perceptions of firm-investor Q&A platforms. In the subsequent empirical study, we specifically focus on the substance of managers’ responses, which are primarily aimed at clarifying firm events or information. In …

When Less Language Is More: Language-Reasoning Disentanglement Makes Llms Better Multilingual Reasoners, Weixiang Zhao, Jiahe Guo, Yang Deng, Tongtong Wu, Wenxuan Zhang, Yulin Hu, Xingyu Sui, Yanyan Zhao, Wanxiang Che, Bing Qin, Tat-Seng Chua, Ting Liu

Research Collection School Of Computing and Information Systems

Multilingual reasoning remains a significant challenge for large language models (LLMs), with performance disproportionately favoring high-resource languages. Drawing inspiration from cognitive neuroscience, which suggests that human reasoning functions largely independently of language processing, we hypothesize that LLMs similarly encode reasoning and language as separable components that can be disentangled to enhance multilingual reasoning. To evaluate this, we perform a causal intervention by ablating language-specific representations at inference time. Experiments on 10 open-weight LLMs spanning 11 typologically diverse languages show that this language-specific ablation consistently boosts multilingual reasoning performance. Layer-wise analyses further confirm that language and reasoning representations can be effectively …

Multi-Task Vehicle Routing Solver Via Mixture Of Specialized Experts Under State-Decomposable Mdp, Yuxin Pan, Zhiguang Cao, Chengyang Gu, Liu Liu, Peilin Zhao, Yize Chen, Fangzhen Lin

Research Collection School Of Computing and Information Systems

Existing neural methods for multi-task vehicle routing problems (VRPs) typically learn unified solvers to handle multiple constraints simultaneously. However, they often underutilize the compositional structure of VRP variants, each derivable from a common set of basis VRP variants. This critical oversight causes unified solvers to miss out the potential benefits of basis solvers, each specialized for a basis VRP variant. To overcome this limitation, we propose a framework that enables unified solvers to perceive the shared-component nature across VRP variants by proactively reusing basis solvers, while mitigating the exponential growth of trained neural solvers. Specifically, we introduce a State-Decomposable MDP …

Robust Hallucination Detection In Llms Via Adaptive Token Selection, Mengjia Niu, Hamed Haddadi, Guansong Pang

Research Collection School Of Computing and Information Systems

Hallucinations in large language models (LLMs) pose significant safety concerns that impede their broader deployment. Recent research in hallucination detection has demonstrated that LLMs’ internal representations contain truthfulness hints, which can be harnessed for detector training. However, the performance of these detectors is heavily dependent on the internal representations of predetermined tokens, fluctuating considerably when working on free-form generations with varying lengths and sparse distributions of hallucinated entities. To address this, we propose HaMI, a novel approach that enables robust detection of hallucinations through adaptive selection and learning of critical tokens that are most indicative of hallucinations. We achieve this …

Sempo: Lightweight Foundation Models For Time Series Forecasting, Hui He, Kun Yi, Yuanchi Ma, Qi Zhang, Zhengdong Niu, Guansong Pang

Research Collection School Of Computing and Information Systems

The recent boom of large pre-trained models witnesses remarkable success in developing foundation models (FMs) for time series forecasting. Despite impressive performance across diverse downstream forecasting tasks, existing time series FMs possess massive network architectures and require substantial pre-training on large-scale datasets, which significantly hinders their deployment in resource-constrained environments. In response to this growing tension between versatility and affordability, we propose SEMPO, a novel lightweight foundation model that requires pretraining on relatively small-scale data, yet exhibits strong general time series forecasting. Concretely, SEMPO comprises two key modules: 1) energy-aware SpEctral decomposition module, that substantially improves the utilization of pre-training …

Towards Inclusive Digital Futures Of Cultural Heritage: Insights From A Critical Discourse Analysis Of Unesco Dialogues, Shiqing Huang, Keng Siau, Xiaoting Chen

Research Collection School Of Computing and Information Systems

Digital technologies are shaping many aspects of cultural heritage, but very little research has examined the implications of digital transformation. Drawing on concepts from Fairclough’s three-dimensional critical discourse analysis, this research examines the discourse using seven online dialogues (available on the UNESCO website) between 18 professionals who have different backgrounds and cultures to identify social practices related to the digital transformation of cultural heritage. We identify four digital transformation discourse types in professional dialogues: documentation, management, interpretation, and interaction. We also identify seven main groups: memory institutions including libraries, archives, and museums (LAMs), governments, international organizations, art and creative supporters, …

Jury-And-Judge Chain-Of-Thought For Uncovering Toxic Data In 3d Visual Grounding, Kaixiang Huang, Qifeng Zhang, Jin Wang, Jingru Yang, Yang Zhou, Huan Yu, Guodong Lu, Shengfeng He

Research Collection School Of Computing and Information Systems

3D Visual Grounding (3DVG) faces persistent challenges due to coarse scene-level observations and logically inconsistent annotations, which introduce ambiguities that compromise data quality and hinder effective model supervision. To address these challenges, we introduce Refer-Judge, a novel framework that harnesses the reasoning capabilities of Multimodal Large Language Models (MLLMs) to identify and mitigate toxic data. At the core of Refer-Judge is a Jury-and-Judge Chain-of-Thought paradigm, inspired by the deliberative process of the judicial system. This framework targets the root causes of annotation noise: jurors collaboratively assess 3DVG samples from diverse perspectives, providing structured, multi-faceted evaluations. Judges then consolidate these insights …

Cssa-Fusion: Channel Selective And Spatial Alignment Infrared-Visible Image Fusion, Zhen Li, Zhi Zeng, Zhongrui Xiao, Ming Wen, Zhiyuan Zhang, Yibin Tian

Research Collection School Of Computing and Information Systems

Infrared-visible image fusion aims to integrate complementary information from two modalities to generate images with enriched semantic content. However, existing methods often neglect two critical aspects: the design of a local–global feature enhancement architecture and spatial alignment. To address these challenges, we propose Channel Selective and Spatial Alignment Fusion (CSSA-Fusion), a novel framework composed of two synergistic modules. The first is a selective channel and redundancy suppression module, which introduces a dual-branch selective channel attention mechanism to jointly capture local saliency and global channel importance for enhanced feature representation, and an informativeness–redundancy separation strategy to suppress redundant information while preserving …

A Partition Cover Approach To Tokenization, Jia Peng Lim, Shawn Tan, Davin Choo, Hady Wirawan Lauw

Research Collection School Of Computing and Information Systems

Tokenization is the process of encoding strings into tokens of a fixed vocabulary size, and is widely utilized in Natural Language Processing applications. The leading tokenization algorithm today is Byte Pair Encoding (BPE), which formulates the tokenization problem as a compression problem and tackles it by performing sequences of merges. In this work, we formulate tokenization as an optimization objective, show that it is NP-hard via a simple reduction from vertex cover, and propose a polynomial-time greedy algorithm GreedTok. Our formulation naturally relaxes to the well-studied weighted maximum coverage problem which has a simple -approximation algorithm GreedWMC. Through empirical evaluations …

Sheetpedia: A 300k-Spreadsheet Corpus For Spreadsheet Intelligence And Llm Fine-Tuning, Zailong Tian, Zhuoheng Han, Houfeng Wang, Lizi Liao

Research Collection School Of Computing and Information Systems

Spreadsheets are widely used for data analysis and reporting, yet their complex structure and formula logic pose significant challenges for AI systems. We introduce Sheetpedia, a large-scale corpus of over 290,000 diverse spreadsheets (from 324,000+ workbooks) compiled from enterprise email archives and online forums. We detail a rigorous collection and preprocessing pipeline (integrating the Enron email spreadsheet archive and the Fuse web corpus, plus a new crawl of Excel forums) to standardize formats, filter languages, and remove duplicates. Sheetpedia provides extensive coverage of real formulas and annotations – addressing a gap left by prior table datasets (e.g. web tables used …

Contx: Scene Context Prediction Via Context Bank And Layout Perception, Jingxin Liang, Yangyang Xu, Haorui Song, Yuan Lu, Yuhui Deng, Yiyi Long, Yan Huang, Shengxin Liu, Jianbo Jiao, Shengfeng He

Research Collection School Of Computing and Information Systems

Scene context prediction, which seeks to infer unknown contextual information from isolated object properties, currently faces limitations due to predominant reliance on pixel-wise supervision that overlooks real-world context priors. To address this, we present ContX, a context-prior-driven, coarse-to-fine model. ContX distinctively integrates explicit linguistic-contextual knowledge in two key ways. First, it proposes a linguistic guided context bank, leveraging linguistic-statistical contextual data to guide the rationality of segmentation shapes and foster meaningful inter-class contextual interactions. Second, ContX augments contextual comprehension by correlating layouts with linguistic descriptions, enhancing layout perception through a multi-modal strategy. Comprehensive experiments demonstrate ContX's superiority and versatility, outperforming …

Backdoorllm: A Comprehensive Benchmark For Backdoor Attacks And Defenses On Large Language Models, Yige Li, Hanxun Huang, Yunhan Zhao, Xingjun Ma, Jun Sun

Research Collection School Of Computing and Information Systems

Generative large language models (LLMs) have achieved state-of-the-art results on a wide range of tasks, yet they remain susceptible to backdoor attacks: carefully crafted triggers in the input can manipulate the model to produce adversaryspecified outputs. While prior research has predominantly focused on backdoor risks in vision and classification settings, the vulnerability of LLMs in open-ended text generation remains underexplored. To fill this gap, we introduce BackdoorLLM1 , the first comprehensive benchmark for systematically evaluating backdoor threats in text-generation LLMs. BackdoorLLM provides: (i) a unified repository of benchmarks with a standardized training and evaluation pipeline; (ii) a diverse suite of …