Authors: Bryar A. Hassan, Shko M. Qader, Alla A. Hassan, Joan Lu, Aram M. Ahmed, Jafar Majidpour, Tarik A. Rashid
Abstract: Automating the extraction of concept hierarchies from free text is advantageous because manual generation is frequently labor- and resource-intensive. Free result, the whole procedure for concept hierarchy learning from free text entails several phases, including sentence-level text processing, sentence splitting, and tokenization. Lemmatization is after formal context analysis (FCA) to derive the pairings. Nevertheless, there could be a few uninteresting and incorrect pairings in the formal context. It may take a while to generate formal context; thus, size reduction formal context is necessary to weed out irrelevant and incorrect pairings to extract the concept lattice and hierarchies more quickly. This study aims to propose a framework for reducing formal context in extracting concept hierarchies from free text to reduce the ambiguity of the formal context. We achieve this by reducing the size of the formal context using a hybrid of a WordNet-based method and a frequency-based technique. Using 385 samples from the Wikipedia corpus and the suggested framework, tests are carried out to examine the reduced size of formal context, leading to concept lattice and concept hierarchy. With the help of concept lattice-invariants, the generated formal context lattice is compared to the normal one. In contrast to basic ones, the homomorphic between the resultant lattices retains up to 98% of the quality of the generating concept hierarchies, and the reduced concept lattice receives the structural connection of the standard one. Additionally, the new framework is compared to five baseline techniques to calculate the running time on random datasets with various densities. The findings demonstrate that, in various fill ratios, hybrid approaches of the proposed method outperform other indicated competing strategies in concept lattice performance.
Authors: Yifei Yuan, Zahra Abbasiantaeb, Yang Deng, Mohammad Aliannejadi
Abstract: Query understanding in Conversational Information Seeking (CIS) involves accurately interpreting user intent through context-aware interactions. This includes resolving ambiguities, refining queries, and adapting to evolving information needs. Large Language Models (LLMs) enhance this process by interpreting nuanced language and adapting dynamically, improving the relevance and precision of search results in real-time. In this tutorial, we explore advanced techniques to enhance query understanding in LLM-based CIS systems. We delve into LLM-driven methods for developing robust evaluation metrics to assess query understanding quality in multi-turn interactions, strategies for building more interactive systems, and applications like proactive query management and query reformulation. We also discuss key challenges in integrating LLMs for query understanding in conversational search systems and outline future research directions. Our goal is to deepen the audience's understanding of LLM-based conversational query understanding and inspire discussions to drive ongoing advancements in this field.
Authors: Rebecca M. M. Hicke, Sil Hamilton, David Mimno
Abstract: Sensory language expresses embodied experiences ranging from taste and sound to excitement and stomachache. This language is of interest to scholars from a wide range of domains including robotics, narratology, linguistics, and cognitive science. In this work, we explore whether language models, which are not embodied, can approximate human use of embodied language. We extend an existing corpus of parallel human and model responses to short story prompts with an additional 18,000 stories generated by 18 popular models. We find that all models generate stories that differ significantly from human usage of sensory language, but the direction of these differences varies considerably between model families. Namely, Gemini models use significantly more sensory language than humans along most axes whereas most models from the remaining five families use significantly less. Linear probes run on five models suggest that they are capable of identifying sensory language. However, we find preliminary evidence suggesting that instruction tuning may discourage usage of sensory language. Finally, to support further work, we release our expanded story dataset.
Authors: Hanqing Zeng, Yinglong Xia, Zhuokai Zhao, Gilbert Jiang, Qiang Zhang, Jiayi Liu, Lizhu Zhang, Xiangjun Fan, Benyu Zhang
Abstract: Fine-tuning pre-trained large language models (LLMs) presents a dual challenge of balancing parameter efficiency and model capacity. Existing methods like low-rank adaptations (LoRA) are efficient but lack flexibility, while Mixture-of-Experts (MoE) architectures enhance model capacity at the cost of more & under-utilized parameters. To address these limitations, we propose Structural Mixture of Residual Experts (S'MoRE), a novel framework that seamlessly integrates the efficiency of LoRA with the flexibility of MoE. Specifically, S'MoRE employs hierarchical low-rank decomposition of expert weights, yielding residuals of varying orders interconnected in a multi-layer structure. By routing input tokens through sub-trees of residuals, S'MoRE emulates the capacity of many experts by instantiating and assembling just a few low-rank matrices. We craft the inter-layer propagation of S'MoRE's residuals as a special type of Graph Neural Network (GNN), and prove that under similar parameter budget, S'MoRE improves "structural flexibility" of traditional MoE (or Mixture-of-LoRA) by exponential order. Comprehensive theoretical analysis and empirical results demonstrate that S'MoRE achieves superior fine-tuning performance, offering a transformative approach for efficient LLM adaptation.
Authors: Dogus Yuksel, Mehmet Cem Catalbas, Bora Oc
Abstract: As leading examples of large language models, ChatGPT and Gemini claim to provide accurate and unbiased information, emphasizing their commitment to political neutrality and avoidance of personal bias. This research investigates the political tendency of large language models and the existence of differentiation according to the query language. For this purpose, ChatGPT and Gemini were subjected to a political axis test using 14 different languages. The findings of the study suggest that these large language models do exhibit political tendencies, with both models demonstrating liberal and leftist biases. A comparative analysis revealed that Gemini exhibited a more pronounced liberal and left-wing tendency compared to ChatGPT. The study also found that these political biases varied depending on the language used for inquiry. The study delves into the factors that constitute political tendencies and linguistic differentiation, exploring differences in the sources and scope of educational data, structural and grammatical features of languages, cultural and political contexts, and the model's response to linguistic features. From this standpoint, and an ethical perspective, it is proposed that artificial intelligence tools should refrain from asserting a lack of political tendencies and neutrality, instead striving for political neutrality and executing user queries by incorporating these tendencies.
Authors: Yuehan Qin, Shawn Li, Yi Nian, Xinyan Velocity Yu, Yue Zhao, Xuezhe Ma
Abstract: Large language models (LLMs) have shown substantial capacity for generating fluent, contextually appropriate responses. However, they can produce hallucinated outputs, especially when a user query includes one or more false premises-claims that contradict established facts. Such premises can mislead LLMs into offering fabricated or misleading details. Existing approaches include pretraining, fine-tuning, and inference-time techniques that often rely on access to logits or address hallucinations after they occur. These methods tend to be computationally expensive, require extensive training data, or lack proactive mechanisms to prevent hallucination before generation, limiting their efficiency in real-time applications. We propose a retrieval-based framework that identifies and addresses false premises before generation. Our method first transforms a user's query into a logical representation, then applies retrieval-augmented generation (RAG) to assess the validity of each premise using factual sources. Finally, we incorporate the verification results into the LLM's prompt to maintain factual consistency in the final output. Experiments show that this approach effectively reduces hallucinations, improves factual accuracy, and does not require access to model logits or large-scale fine-tuning.
Authors: Sai Adith Senthil Kumar, Hao Yan, Saipavan Perepa, Murong Yue, Ziyu Yao
Abstract: Large Language Models (LLMs) are now increasingly widely used to simulate personas in virtual environments, leveraging their instruction-following capability. However, we discovered that even state-of-the-art LLMs cannot simulate personas with reversed performance (e.g., student personas with low proficiency in educational settings), which impairs the simulation diversity and limits the practical applications of the simulated environments. In this work, using mathematical reasoning as a representative scenario, we propose the first benchmark dataset for evaluating LLMs on simulating personas with reversed performance, a capability that we dub "counterfactual instruction following". We evaluate both open-weight and closed-source LLMs on this task and find that LLMs, including the OpenAI o1 reasoning model, all struggle to follow counterfactual instructions for simulating reversedly performing personas. Intersectionally simulating both the performance level and the race population of a persona worsens the effect even further. These results highlight the challenges of counterfactual instruction following and the need for further research.
Authors: Ye (Cheryl), Ma
Abstract: This study explores using Natural Language Processing (NLP) to analyze candidate comments for identifying problematic test items. We developed and validated machine learning models that automatically identify relevant negative feedback, evaluated approaches of incorporating psychometric features enhances model performance, and compared NLP-flagged items with traditionally flagged items. Results demonstrate that candidate feedback provides valuable complementary information to statistical methods, potentially improving test validity while reducing manual review burden. This research offers testing organizations an efficient mechanism to incorporate direct candidate experience into quality assurance processes.
Authors: Khai Phan Tran, Xue Li
Abstract: Document-level Relation Extraction (DocRE) involves identifying relations between entities across multiple sentences in a document. Evidence sentences, crucial for precise entity pair relationships identification, enhance focus on essential text segments, improving DocRE performance. However, existing evidence retrieval systems often overlook the collaborative nature among semantically similar entity pairs in the same document, hindering the effectiveness of the evidence retrieval task. To address this, we propose a novel evidence retrieval framework, namely CDER. CDER employs an attentional graph-based architecture to capture collaborative patterns and incorporates a dynamic sub-structure for additional robustness in evidence retrieval. Experimental results on the benchmark DocRE dataset show that CDER not only excels in the evidence retrieval task but also enhances overall performance of existing DocRE system.
Authors: Happy Buzaaba, Alexander Wettig, David Ifeoluwa Adelani, Christiane Fellbaum
Abstract: Large language models (LLMs) have achieved impressive results in a wide range of natural language applications. However, they often struggle to recognize low-resource languages, in particular African languages, which are not well represented in large training corpora. In this paper, we consider how to adapt LLMs to low-resource African languages. We find that combining curated data from African languages with high-quality English educational texts results in a training mix that substantially improves the model's performance on these languages. On the challenging IrokoBench dataset, our models consistently achieve the best performance amongst similarly sized baselines, particularly on knowledge-intensive multiple-choice questions (AfriMMLU). Additionally, on the cross-lingual question answering benchmark AfriQA, our models outperform the base model by over 10%. To better understand the role of English data during training, we translate a subset of 200M tokens into Swahili language and perform an analysis which reveals that the content of these data is primarily responsible for the strong performance. We release our models and data to encourage future research on African languages.
Authors: Lanrui Wang, Mingyu Zheng, Hongyin Tang, Zheng Lin, Yanan Cao, Jingang Wang, Xunliang Cai, Weiping Wang
Abstract: Processing structured tabular data, particularly lengthy tables, constitutes a fundamental yet challenging task for large language models (LLMs). However, existing long-context benchmarks primarily focus on unstructured text, neglecting the challenges of long and complex structured tables. To address this gap, we introduce NeedleInATable (NIAT), a novel task that treats each table cell as a "needle" and requires the model to extract the target cell under different queries. Evaluation results of mainstream LLMs on this benchmark show they lack robust long-table comprehension, often relying on superficial correlations or shortcuts for complex table understanding tasks, revealing significant limitations in processing intricate tabular data. To this end, we propose a data synthesis method to enhance models' long-table comprehension capabilities. Experimental results show that our synthesized training data significantly enhances LLMs' performance on the NIAT task, outperforming both long-context LLMs and long-table agent methods. This work advances the evaluation of LLMs' genuine long-structured table comprehension capabilities and paves the way for progress in long-context and table understanding applications.
Authors: Longguang Zhong, Fanqi Wan, Ziyi Yang, Guosheng Liang, Tianyuan Shi, Xiaojun Quan
Abstract: Heterogeneous model fusion enhances the performance of LLMs by integrating the knowledge and capabilities of multiple structurally diverse models. However, existing approaches often rely solely on selecting the best output for each prompt from source models, which underutilizes their full potential due to limited source knowledge and results in sparse optimization signals. To address this limitation, we propose FuseRL, a novel two-stage framework comprising FuseSFT and FusePO to maximize the utilization of source LLMs. FuseSFT establishes a robust initialization by integrating the strengths of heterogeneous source models through weighted supervised fine-tuning (SFT) on diverse outputs for each prompt. FusePO optimizes weighted preferences based on the outputs of multiple source models to enable superior alignment performance. Extensive experiments demonstrate the effectiveness of our framework across various preference alignment methods, including RLOO, DPO, and SimPO. Using Llama-3.1-8B-Instruct as the target model, our approach achieves state-of-the-art performance among 8B LLMs on the AlpacaEval-2 and Arena-Hard benchmarks. Further analysis suggests that FuseSFT regularizes the training process to reduce overfitting, while FusePO introduces dense and diverse signals for preference optimization.
Authors: Qingcheng Zeng, Weihao Xuan, Leyang Cui, Rob Voigt
Abstract: Large reasoning models (LRMs) have recently shown impressive capabilities in complex reasoning by leveraging increased test-time computation and exhibiting behaviors akin to human-like deliberation. Despite these advances, it remains an open question whether LRMs are better calibrated - particularly in their verbalized confidence - compared to instruction-tuned counterparts. In this paper, we investigate the calibration properties of LRMs trained via supervised fine-tuning distillation on long reasoning traces (henceforth SFT reasoning models) and outcome-based reinforcement learning for reasoning (henceforth RL reasoning models) across diverse domains. Our findings reveal that LRMs significantly outperform instruction-tuned models on complex reasoning tasks in both accuracy and confidence calibration. In contrast, we find surprising trends in the domain of factuality in particular. On factuality tasks, while Deepseek-R1 shows strong calibration behavior, smaller QwQ-32B shows no improvement over instruct models; moreover, SFT reasoning models display worse calibration (greater overconfidence) compared to instruct models. Our results provide evidence for a potentially critical role of reasoning-oriented RL training in improving LLMs' capacity for generating trustworthy, self-aware outputs.
Authors: Pedro Cisneros-Velarde
Abstract: In this paper, we show it is possible to bypass the safety guardrails of large language models (LLMs) through a humorous prompt including the unsafe request. In particular, our method does not edit the unsafe request and follows a fixed template -- it is simple to implement and does not need additional LLMs to craft prompts. Extensive experiments show the effectiveness of our method across different LLMs. We also show that both removing and adding more humor to our method can reduce its effectiveness -- excessive humor possibly distracts the LLM from fulfilling its unsafe request. Thus, we argue that LLM jailbreaking occurs when there is a proper balance between focus on the unsafe request and presence of humor.
Authors: William De Michele, Abel Armas Cervantes, Lea Frermann
Abstract: Business processes are fundamental to organizational operations, yet their optimization remains challenging due to the timeconsuming nature of manual process analysis. Our paper harnesses Large Language Models (LLMs) to automate value-added analysis, a qualitative process analysis technique that aims to identify steps in the process that do not deliver value. To date, this technique is predominantly manual, time-consuming, and subjective. Our method offers a more principled approach which operates in two phases: first, decomposing high-level activities into detailed steps to enable granular analysis, and second, performing a value-added analysis to classify each step according to Lean principles. This approach enables systematic identification of waste while maintaining the semantic understanding necessary for qualitative analysis. We develop our approach using 50 business process models, for which we collect and publish manual ground-truth labels. Our evaluation, comparing zero-shot baselines with more structured prompts reveals (a) a consistent benefit of structured prompting and (b) promising performance for both tasks. We discuss the potential for LLMs to augment human expertise in qualitative process analysis while reducing the time and subjectivity inherent in manual approaches.
Authors: Zijian Wang, Chang Xu
Abstract: Pre-trained large language models (LLMs) have been demonstrated to possess intrinsic reasoning capabilities that can emerge naturally when expanding the response space. However, the neural representation mechanisms underlying these intrinsic capabilities and approaches for their optimal utilization remain inadequately understood. In this work, we make the key discovery that a simple linear classifier can effectively detect intrinsic reasoning capabilities in LLMs' activation space, particularly within specific representation types and network layers. Based on this finding, we propose a classifier-guided search framework that strategically explore a tree-structured response space. In each node expansion, the classifier serves as a scoring and ranking mechanism that efficiently allocates computational resources by identifying and prioritizing more thoughtful reasoning directions for continuation. After completing the tree expansion, we collect answers from all branches to form a candidate answer pool. We propose a branch-aggregation selection method that marginalizes over all supporting branches by aggregating their thoughtfulness scores, thereby identifying the optimal answer from the pool. Experimental results show that our framework's comprehensive exploration not only covers valid reasoning chains but also effectively identifies them, achieving significant improvements across multiple arithmetic reasoning benchmarks.
Authors: Zhilin Wang, Yafu Li, Xiaoye Qu, Yu Cheng
Abstract: Continual fine-tuning of large language models (LLMs) suffers from catastrophic forgetting. Rehearsal-based methods mitigate this problem by retaining a small set of old data. Nevertheless, they still suffer inevitable performance loss. Although training separate experts for each task can help prevent forgetting, effectively assembling them remains a challenge. Some approaches use routers to assign tasks to experts, but in continual learning, they often require retraining for optimal performance. To address these challenges, we introduce the Sequential Ensemble of Experts (SEE) framework. SEE removes the need for an additional router, allowing each expert to independently decide whether a query should be handled. The framework employs distributed routing, and during continual fine-tuning, SEE only requires the training of new experts for incoming tasks rather than retraining the entire system. Experiments reveal that the SEE outperforms prior approaches, including multi-task learning, in continual fine-tuning. It also demonstrates remarkable generalization ability, as the expert can effectively identify out-of-distribution queries, which can then be directed to a more generalized model for resolution. This work highlights the promising potential of integrating routing and response mechanisms within each expert, paving the way for the future of distributed model ensembling.
Authors: Heather Lent, Erick Galinkin, Yiyi Chen, Jens Myrup Pedersen, Leon Derczynski, Johannes Bjerva
Abstract: As NLP models are used by a growing number of end-users, an area of increasing importance is NLP Security (NLPSec): assessing the vulnerability of models to malicious attacks and developing comprehensive countermeasures against them. While work at the intersection of NLP and cybersecurity has the potential to create safer NLP for all, accidental oversights can result in tangible harm (e.g., breaches of privacy or proliferation of malicious models). In this emerging field, however, the research ethics of NLP have not yet faced many of the long-standing conundrums pertinent to cybersecurity, until now. We thus examine contemporary works across NLPSec, and explore their engagement with cybersecurity's ethical norms. We identify trends across the literature, ultimately finding alarming gaps on topics like harm minimization and responsible disclosure. To alleviate these concerns, we provide concrete recommendations to help NLP researchers navigate this space more ethically, bridging the gap between traditional cybersecurity and NLP ethics, which we frame as ``white hat NLP''. The goal of this work is to help cultivate an intentional culture of ethical research for those working in NLP Security.
Authors: Zican Dong, Han Peng, Peiyu Liu, Wayne Xin Zhao, Dong Wu, Feng Xiao, Zhifeng Wang
Abstract: 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 demonstrations, the model consistently activates a sparse and stable subset of experts. 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 magnitudes of the outputs of activated experts, while the latter assesses the contribution of tokens based on representation similarities after and before routed experts. Experiments show that our method can achieve comparable performances and $2.99\times$ throughput under the same memory budget with full DeepSeek-R1 with only half the experts. Our code is available at https://github.com/RUCAIBox/EASYEP.
Authors: Karol Mikula, Mariana Sarkociov\'a Reme\v{s}\'ikov\'a
Abstract: In this paper, we present an algorithm for evaluating lexical similarity between a given language and several reference language clusters. As an input, we have a list of concepts and the corresponding translations in all considered languages. Moreover, each reference language is assigned to one of $c$ language clusters. For each of the concepts, the algorithm computes the distance between each pair of translations. Based on these distances, it constructs a weighted directed graph, where every vertex represents a language. After, it solves a graph diffusion equation with a Dirichlet boundary condition, where the unknown is a map from the vertex set to $\mathbb{R}^c$. The resulting coordinates are values from the interval $[0,1]$ and they can be interpreted as probabilities of belonging to each of the clusters or as a lexical similarity distribution with respect to the reference clusters. The distances between translations are calculated using phonetic transcriptions and a modification of the Damerau-Levenshtein distance. The algorithm can be useful in analyzing relationships between languages spoken in multilingual territories with a lot of mutual influences. We demonstrate this by presenting a case study regarding various European languages.
Authors: Zora Zhiruo Wang, Apurva Gandhi, Graham Neubig, Daniel Fried
Abstract: To succeed in common digital tasks such as web navigation, agents must carry out a variety of specialized tasks such as searching for products or planning a travel route. To tackle these tasks, agents can bootstrap themselves by learning task-specific skills online through interaction with the web environment. In this work, we demonstrate that programs are an effective representation for skills. We propose agent skill induction (ASI), which allows agents to adapt themselves by inducing, verifying, and utilizing program-based skills on the fly. We start with an evaluation on the WebArena agent benchmark and show that ASI outperforms the static baseline agent and its text-skill counterpart by 23.5% and 11.3% in success rate, mainly thanks to the programmatic verification guarantee during the induction phase. ASI also improves efficiency by reducing 10.7-15.3% of the steps over baselines, by composing primitive actions (e.g., click) into higher-level skills (e.g., search product). We then highlight the efficacy of ASI in remaining efficient and accurate under scaled-up web activities. Finally, we examine the generalizability of induced skills when transferring between websites, and find that ASI can effectively reuse common skills, while also updating incompatible skills to versatile website changes.
Authors: Xiaotian Ye, Mengqi Zhang, Shu Wu
Abstract: Knowledge is fundamental to the overall capabilities of Large Language Models (LLMs). The knowledge paradigm of a model, which dictates how it encodes and utilizes knowledge, significantly affects its performance. Despite the continuous development of LLMs under existing knowledge paradigms, issues within these frameworks continue to constrain model potential. This blog post highlight three critical open problems limiting model capabilities: (1) challenges in knowledge updating for LLMs, (2) the failure of reverse knowledge generalization (the reversal curse), and (3) conflicts in internal knowledge. We review recent progress made in addressing these issues and discuss potential general solutions. Based on observations in these areas, we propose a hypothetical paradigm based on Contextual Knowledge Scaling, and further outline implementation pathways that remain feasible within contemporary techniques. Evidence suggests this approach holds potential to address current shortcomings, serving as our vision for future model paradigms. This blog post aims to provide researchers with a brief overview of progress in LLM knowledge systems, while provide inspiration for the development of next-generation model architectures.
Authors: Angela Lopez-Cardona, Sebastian Idesis, Ioannis Arapakis
Abstract: Recently, the integration of cognitive neuroscience in Natural Language Processing (NLP) has gained significant attention. This article provides a critical and timely overview of recent advancements in leveraging cognitive signals, particularly Eye-tracking (ET) signals, to enhance Language Models (LMs) and Multimodal Large Language Models (MLLMs). By incorporating user-centric cognitive signals, these approaches address key challenges, including data scarcity and the environmental costs of training large-scale models. Cognitive signals enable efficient data augmentation, faster convergence, and improved human alignment. The review emphasises the potential of ET data in tasks like Visual Question Answering (VQA) and mitigating hallucinations in MLLMs, and concludes by discussing emerging challenges and research trends.
Authors: Seungwon Lim, Seungbeen Lee, Dongjun Min, Youngjae Yu
Abstract: Artificial agents are increasingly central to complex interactions and decision-making tasks, yet aligning their behaviors with desired human values remains an open challenge. In this work, we investigate how human-like personality traits influence agent behavior and performance within text-based interactive environments. We introduce PANDA: PersonalityAdapted Neural Decision Agents, a novel method for projecting human personality traits onto agents to guide their behavior. To induce personality in a text-based game agent, (i) we train a personality classifier to identify what personality type the agent's actions exhibit, and (ii) we integrate the personality profiles directly into the agent's policy-learning pipeline. By deploying agents embodying 16 distinct personality types across 25 text-based games and analyzing their trajectories, we demonstrate that an agent's action decisions can be guided toward specific personality profiles. Moreover, certain personality types, such as those characterized by higher levels of Openness, display marked advantages in performance. These findings underscore the promise of personality-adapted agents for fostering more aligned, effective, and human-centric decision-making in interactive environments.
Authors: Sheng Lu, Ilia Kuznetsov, Iryna Gurevych
Abstract: Peer review is central to academic publishing, but the growing volume of submissions is straining the process. This motivates the development of computational approaches to support peer review. While each review is tailored to a specific paper, reviewers often make assessments according to certain aspects such as Novelty, which reflect the values of the research community. This alignment creates opportunities for standardizing the reviewing process, improving quality control, and enabling computational support. While prior work has demonstrated the potential of aspect analysis for peer review assistance, the notion of aspect remains poorly formalized. Existing approaches often derive aspect sets from review forms and guidelines of major NLP venues, yet data-driven methods for aspect identification are largely underexplored. To address this gap, our work takes a bottom-up approach: we propose an operational definition of aspect and develop a data-driven schema for deriving fine-grained aspects from a corpus of peer reviews. We introduce a dataset of peer reviews augmented with aspects and show how it can be used for community-level review analysis. We further show how the choice of aspects can impact downstream applications, such as LLM-generated review detection. Our results lay a foundation for a principled and data-driven investigation of review aspects, and pave the path for new applications of NLP to support peer review.
Authors: Ming Liu, Massimo Poesio
Abstract: With the growth of the Internet, buying habits have changed, and customers have become more dependent on the online opinions of other customers to guide their purchases. Identifying fake reviews thus became an important area for Natural Language Processing (NLP) research. However, developing high-performance NLP models depends on the availability of large amounts of training data, which are often not available for low-resource languages or domains. In this research, we used large language models to generate datasets to train fake review detectors. Our approach was used to generate fake reviews in different domains (book reviews, restaurant reviews, and hotel reviews) and different languages (English and Chinese). Our results demonstrate that our data augmentation techniques result in improved performance at fake review detection for all domains and languages. The accuracy of our fake review detection model can be improved by 0.3 percentage points on DeRev TEST, 10.9 percentage points on Amazon TEST, 8.3 percentage points on Yelp TEST and 7.2 percentage points on DianPing TEST using the augmented datasets.
Authors: Natalia Loukachevitch, Natalia Tkachenko, Anna Lapanitsyna, Mikhail Tikhomirov, Nicolay Rusnachenko
Abstract: In this paper, we introduce the Dialogue Evaluation shared task on extraction of structured opinions from Russian news texts. The task of the contest is to extract opinion tuples for a given sentence; the tuples are composed of a sentiment holder, its target, an expression and sentiment from the holder to the target. In total, the task received more than 100 submissions. The participants experimented mainly with large language models in zero-shot, few-shot and fine-tuning formats. The best result on the test set was obtained with fine-tuning of a large language model. We also compared 30 prompts and 11 open source language models with 3-32 billion parameters in the 1-shot and 10-shot settings and found the best models and prompts.
Authors: Lilian Ngweta, Kiran Kate, Jason Tsay, Yara Rizk
Abstract: Large language models (LLMs) have gained popularity in recent years for their utility in various applications. However, they are sensitive to non-semantic changes in prompt formats, where small changes in the prompt format can lead to significant performance fluctuations. In the literature, this problem is commonly referred to as prompt brittleness. Previous research on prompt engineering has focused mainly on developing techniques for identifying the optimal prompt for specific tasks. Some studies have also explored the issue of prompt brittleness and proposed methods to quantify performance variations; however, no simple solution has been found to address this challenge. We propose Mixture of Formats (MOF), a simple and efficient technique for addressing prompt brittleness in LLMs by diversifying the styles used in the prompt few-shot examples. MOF was inspired by computer vision techniques that utilize diverse style datasets to prevent models from associating specific styles with the target variable. Empirical results show that our proposed technique reduces style-induced prompt brittleness in various LLMs while also enhancing overall performance across prompt variations and different datasets.
Authors: Chad Melton, Alex Sorokine, Steve Peterson
Abstract: Applications of generative Large Language Models LLMs are rapidly expanding across various domains, promising significant improvements in workflow efficiency and information retrieval. However, their implementation in specialized, high-stakes domains such as hazardous materials transportation is challenging due to accuracy and reliability concerns. This study evaluates the performance of three fine-tuned generative models, ChatGPT, Google's Vertex AI, and ORNL Retrieval Augmented Generation augmented LLaMA 2 and LLaMA in retrieving regulatory information essential for hazardous material transportation compliance in the United States. Utilizing approximately 40 publicly available federal and state regulatory documents, we developed 100 realistic queries relevant to route planning and permitting requirements. Responses were qualitatively rated based on accuracy, detail, and relevance, complemented by quantitative assessments of semantic similarity between model outputs. Results demonstrated that the RAG-augmented LLaMA models significantly outperformed Vertex AI and ChatGPT, providing more detailed and generally accurate information, despite occasional inconsistencies. This research introduces the first known application of RAG in transportation safety, emphasizing the need for domain-specific fine-tuning and rigorous evaluation methodologies to ensure reliability and minimize the risk of inaccuracies in high-stakes environments.
Authors: Alessio Tosolini, Claire Bowern
Abstract: A continued issue for those working with computational tools and endangered and under-resourced languages is the lower accuracy of results for languages with smaller amounts of data. We attempt to ameliorate this issue by using data augmentation methods to increase corpus size, comparing augmentation to hyperparameter tuning for multilingual forced alignment. Unlike text augmentation methods, audio augmentation does not lead to substantially increased performance. Hyperparameter tuning, on the other hand, results in substantial improvement without (for this amount of data) infeasible additional training time. For languages with small to medium amounts of training data, this is a workable alternative to adapting models from high-resource languages.
Authors: Liang-Hsuan Tseng, Yi-Chang Chen, Kuan-Yi Lee, Da-Shan Shiu, Hung-yi Lee
Abstract: Large Language Models (LLMs) excel in text-based natural language processing tasks but remain constrained by their reliance on textual inputs and outputs. To enable more natural human-LLM interaction, recent progress have focused on deriving a spoken language model (SLM) that can not only listen but also generate speech. To achieve this, a promising direction is to conduct speech-text joint modeling. However, recent SLM still lag behind text LLM due to the modality mismatch. One significant mismatch can be the sequence lengths between speech and text tokens. To address this, we introduce Text-Aligned Speech Tokenization and Embedding (TASTE), a method that directly addresses the modality gap by aligning speech token with the corresponding text transcription during the tokenization stage. We propose a method that can achieve this through the special aggregation mechanism and with speech reconstruction as the training objective. We conduct extensive experiments and show that TASTE can preserve essential paralinguistic information while dramatically reducing the token sequence length. Furthermore, by leveraging TASTE, we can adapt text-based LLMs into effective SLMs with parameter-efficient fine-tuning techniques such as Low-Rank Adaptation (LoRA). Experimental results on benchmark tasks, including SALMON and StoryCloze, demonstrate that TASTE-based SLMs perform similarly to previous full-finetuning methods. To our knowledge, TASTE is the first end-to-end approach that utilizes a reconstruction objective to automatically learn a text-aligned speech tokenization and embedding suitable for spoken language modeling. Our demo, code, and models are publicly available at https://github.com/mtkresearch/TASTE-SpokenLM.
Authors: Bibek Paudel, Alexander Lyzhov, Preetam Joshi, Puneet Anand
Abstract: This paper introduces a comprehensive system for detecting hallucinations in large language model (LLM) outputs in enterprise settings. We present a novel taxonomy of LLM responses specific to hallucination in enterprise applications, categorizing them into context-based, common knowledge, enterprise-specific, and innocuous statements. Our hallucination detection model HDM-2 validates LLM responses with respect to both context and generally known facts (common knowledge). It provides both hallucination scores and word-level annotations, enabling precise identification of problematic content. To evaluate it on context-based and common-knowledge hallucinations, we introduce a new dataset HDMBench. Experimental results demonstrate that HDM-2 out-performs existing approaches across RagTruth, TruthfulQA, and HDMBench datasets. This work addresses the specific challenges of enterprise deployment, including computational efficiency, domain specialization, and fine-grained error identification. Our evaluation dataset, model weights, and inference code are publicly available.
Authors: Zhouhang Xie, Junda Wu, Yiran Shen, Yu Xia, Xintong Li, Aaron Chang, Ryan Rossi, Sachin Kumar, Bodhisattwa Prasad Majumder, Jingbo Shang, Prithviraj Ammanabrolu, Julian McAuley
Abstract: Personalized preference alignment for large language models (LLMs), the process of tailoring LLMs to individual users' preferences, is an emerging research direction spanning the area of NLP and personalization. In this survey, we present an analysis of works on personalized alignment and modeling for LLMs. We introduce a taxonomy of preference alignment techniques, including training time, inference time, and additionally, user-modeling based methods. We provide analysis and discussion on the strengths and limitations of each group of techniques and then cover evaluation, benchmarks, as well as open problems in the field.
Authors: Israfel Salazar, Manuel Fern\'andez Burda, Shayekh Bin Islam, Arshia Soltani Moakhar, Shivalika Singh, Fabian Farestam, Angelika Romanou, Danylo Boiko, Dipika Khullar, Mike Zhang, Dominik Krzemi\'nski, Jekaterina Novikova, Lu\'isa Shimabucoro, Joseph Marvin Imperial, Rishabh Maheshwary, Sharad Duwal, Alfonso Amayuelas, Swati Rajwal, Jebish Purbey, Ahmed Ruby, Nicholas Popovi\v{c}, Marek Suppa, Azmine Toushik Wasi, Ram Mohan Rao Kadiyala, Olga Tsymboi, Maksim Kostritsya, Bardia Soltani Moakhar, Gabriel da Costa Merlin, Ot\'avio Ferracioli Coletti, Maral Jabbari Shiviari, MohammadAmin farahani fard, Silvia Fernandez, Mar\'ia Grandury, Dmitry Abulkhanov, Drishti Sharma, Andre Guarnier De Mitri, Leticia Bossatto Marchezi, Johan Obando-Ceron, Nazar Kohut, Beyza Ermis, Desmond Elliott, Enzo Ferrante, Sara Hooker, Marzieh Fadaee
Abstract: The evaluation of vision-language models (VLMs) has mainly relied on English-language benchmarks, leaving significant gaps in both multilingual and multicultural coverage. While multilingual benchmarks have expanded, both in size and languages, many rely on translations of English datasets, failing to capture cultural nuances. In this work, we propose Kaleidoscope, as the most comprehensive exam benchmark to date for the multilingual evaluation of vision-language models. Kaleidoscope is a large-scale, in-language multimodal benchmark designed to evaluate VLMs across diverse languages and visual inputs. Kaleidoscope covers 18 languages and 14 different subjects, amounting to a total of 20,911 multiple-choice questions. Built through an open science collaboration with a diverse group of researchers worldwide, Kaleidoscope ensures linguistic and cultural authenticity. We evaluate top-performing multilingual vision-language models and find that they perform poorly on low-resource languages and in complex multimodal scenarios. Our results highlight the need for progress on culturally inclusive multimodal evaluation frameworks.
Authors: Atharva Pandey, Kshitij Dubey, Rahul Sharma, Amit Sharma
Abstract: Despite great performance on Olympiad-level reasoning problems, frontier large language models can still struggle on high school math when presented with novel problems outside standard benchmarks. Going beyond final accuracy, we propose a deductive consistency metric to analyze chain-of-thought output from language models (LMs).Formally, deductive reasoning involves two subtasks: understanding a set of input premises and inferring the conclusions that follow from them. The proposed metric studies LMs' performance on these subtasks, with the goal of explaining LMs' reasoning errors on novel problems: how well do LMs understand input premises with increasing context lengths, and how well can they infer conclusions over multiple reasoning hops? Since existing benchmarks may be memorized, we develop a pipeline to evaluate LMs' deductive consistency on novel, perturbed versions of benchmark problems. On novel grade school math problems (GSM-8k), we find that LMs are fairly robust to increasing number of input premises, but suffer significant accuracy decay as the number of reasoning hops is increased. Interestingly, these errors are masked in the original benchmark as all models achieve near 100% accuracy. As we increase the number of solution steps using a synthetic dataset, prediction over multiple hops still remains the major source of error compared to understanding input premises. Other factors, such as shifts in language style or natural propagation of early errors do not explain the trends. Our analysis provides a new view to characterize LM reasoning -- as computations over a window of input premises and reasoning hops -- that can provide unified evaluation across problem domains.
Authors: Gabriel Grand, Joshua B. Tenenbaum, Vikash K. Mansinghka, Alexander K. Lew, Jacob Andreas
Abstract: While test-time reasoning enables language models to tackle complex tasks, searching or planning in natural language can be slow, costly, and error-prone. But even when LMs struggle to emulate the precise reasoning steps needed to solve a problem, they often excel at describing its abstract structure--both how to verify solutions and how to search for them. This paper introduces DisCIPL, a method for "self-steering" LMs where a Planner model generates a task-specific inference program that is executed by a population of Follower models. Our approach equips LMs with the ability to write recursive search procedures that guide LM inference, enabling new forms of verifiable and efficient reasoning. When instantiated with a small Follower (e.g., Llama-3.2-1B), DisCIPL matches (and sometimes outperforms) much larger models, including GPT-4o and o1, on challenging constrained generation tasks. In decoupling planning from execution, our work opens up a design space of highly-parallelized Monte Carlo inference strategies that outperform standard best-of-N sampling, require no finetuning, and can be implemented automatically by existing LMs.
Authors: Elan Markowitz, Krupa Galiya, Greg Ver Steeg, Aram Galstyan
Abstract: Knowledge graphs have emerged as a popular method for injecting up-to-date, factual knowledge into large language models (LLMs). This is typically achieved by converting the knowledge graph into text that the LLM can process in context. While multiple methods of encoding knowledge graphs have been proposed, the impact of this textualization process on LLM performance remains under-explored. We introduce KG-LLM-Bench, a comprehensive and extensible benchmark spanning five knowledge graph understanding tasks, and evaluate how different encoding strategies affect performance across various base models. Our extensive experiments with seven language models and five textualization strategies provide insights for optimizing LLM performance on KG reasoning tasks.
Authors: Jiacheng Liu, Taylor Blanton, Yanai Elazar, Sewon Min, YenSung Chen, Arnavi Chheda-Kothary, Huy Tran, Byron Bischoff, Eric Marsh, Michael Schmitz, Cassidy Trier, Aaron Sarnat, Jenna James, Jon Borchardt, Bailey Kuehl, Evie Cheng, Karen Farley, Sruthi Sreeram, Taira Anderson, David Albright, Carissa Schoenick, Luca Soldaini, Dirk Groeneveld, Rock Yuren Pang, Pang Wei Koh, Noah A. Smith, Sophie Lebrecht, Yejin Choi, Hannaneh Hajishirzi, Ali Farhadi, Jesse Dodge
Abstract: We present OLMoTrace, the first system that traces the outputs of language models back to their full, multi-trillion-token training data in real time. OLMoTrace finds and shows verbatim matches between segments of language model output and documents in the training text corpora. Powered by an extended version of infini-gram (Liu et al., 2024), our system returns tracing results within a few seconds. OLMoTrace can help users understand the behavior of language models through the lens of their training data. We showcase how it can be used to explore fact checking, hallucination, and the creativity of language models. OLMoTrace is publicly available and fully open-source.
Authors: Jiali Cheng, Hadi Amiri
Abstract: Machine Unlearning removes specific knowledge about training data samples from an already trained model. It has significant practical benefits, such as purging private, inaccurate, or outdated information from trained models without the need for complete re-training. Unlearning within a multimodal setting presents unique challenges due to the complex dependencies between different data modalities and the expensive cost of training on large multimodal datasets and architectures. This paper presents the first machine unlearning approach for multimodal data and models, titled MultiDelete, which is designed to decouple associations between unimodal data points during unlearning without losing the overall representation strength of the trained model. MultiDelete advocates for three key properties for effective multimodal unlearning: (a): modality decoupling, which effectively decouples the association between individual unimodal data points marked for deletion, rendering them as unrelated data points, (b): multimodal knowledge retention, which retains the multimodal representation post-unlearning, and (c): unimodal knowledge retention, which retains the unimodal representation postunlearning. MultiDelete is efficient to train and is not constrained by using a strongly convex loss -- a common restriction among existing baselines. Experiments on two architectures and four datasets, including image-text and graph-text datasets, show that MultiDelete gains an average improvement of 17.6 points over best performing baseline in unlearning multimodal samples, can maintain the multimodal and unimodal knowledge of the original model post unlearning, and can provide better protection to unlearned data against adversarial attacks.
Authors: Yiming Tang, Yi Fan, Chenxiao Yu, Tiankai Yang, Yue Zhao, Xiyang Hu
Abstract: The integration of large language models (LLMs) into information retrieval systems introduces new attack surfaces, particularly for adversarial ranking manipulations. We present StealthRank, a novel adversarial ranking attack that manipulates LLM-driven product recommendation systems while maintaining textual fluency and stealth. Unlike existing methods that often introduce detectable anomalies, StealthRank employs an energy-based optimization framework combined with Langevin dynamics to generate StealthRank Prompts (SRPs)-adversarial text sequences embedded within product descriptions that subtly yet effectively influence LLM ranking mechanisms. We evaluate StealthRank across multiple LLMs, demonstrating its ability to covertly boost the ranking of target products while avoiding explicit manipulation traces that can be easily detected. Our results show that StealthRank consistently outperforms state-of-the-art adversarial ranking baselines in both effectiveness and stealth, highlighting critical vulnerabilities in LLM-driven recommendation systems.
Authors: Liyuan Mao, Haoran Xu, Amy Zhang, Weinan Zhang, Chenjia Bai
Abstract: A generalizable reward model is crucial in Reinforcement Learning from Human Feedback (RLHF) as it enables correctly evaluating unseen prompt-response pairs. However, existing reward models lack this ability, as they are typically trained by increasing the reward gap between chosen and rejected responses, while overlooking the prompts that the responses are conditioned on. Consequently, when the trained reward model is evaluated on prompt-response pairs that lie outside the data distribution, neglecting the effect of prompts may result in poor generalization of the reward model. To address this issue, we decompose the reward value into two independent components: prompt-free reward and prompt-related reward. Prompt-free reward represents the evaluation that is determined only by responses, while the prompt-related reward reflects the reward that derives from both the prompt and the response. We extract these two components from an information-theoretic perspective, which requires no extra models. Subsequently, we propose a new reward learning algorithm by prioritizing data samples based on their prompt-free reward values. Through toy examples, we demonstrate that the extracted prompt-free and prompt-related rewards effectively characterize two parts of the reward model. Further, standard evaluations show that our method improves both the alignment performance and the generalization capability of the reward model.
Authors: Yin Wu, Zhengxuan Zhang, Fuling Wang, Yuyu Luo, Hui Xiong, Nan Tang
Abstract: Misinformation continues to pose a significant challenge in today's information ecosystem, profoundly shaping public perception and behavior. Among its various manifestations, Out-of-Context (OOC) misinformation is particularly obscure, as it distorts meaning by pairing authentic images with misleading textual narratives. Existing methods for detecting OOC misinformation predominantly rely on coarse-grained similarity metrics between image-text pairs, which often fail to capture subtle inconsistencies or provide meaningful explainability. While multi-modal large language models (MLLMs) demonstrate remarkable capabilities in visual reasoning and explanation generation, they have not yet demonstrated the capacity to address complex, fine-grained, and cross-modal distinctions necessary for robust OOC detection. To overcome these limitations, we introduce EXCLAIM, a retrieval-based framework designed to leverage external knowledge through multi-granularity index of multi-modal events and entities. Our approach integrates multi-granularity contextual analysis with a multi-agent reasoning architecture to systematically evaluate the consistency and integrity of multi-modal news content. Comprehensive experiments validate the effectiveness and resilience of EXCLAIM, demonstrating its ability to detect OOC misinformation with 4.3% higher accuracy compared to state-of-the-art approaches, while offering explainable and actionable insights.
Authors: Yikuan Xia, Jiazun Chen, Yirui Zhan, Suifeng Zhao, Weipeng Jiang, Chaorui Zhang, Wei Han, Bo Bai, Jun Gao
Abstract: Large language models (LLMs) excel in question-answering (QA) tasks, and retrieval-augmented generation (RAG) enhances their precision by incorporating external evidence from diverse sources like web pages, databases, and knowledge graphs. However, current RAG methods rely on agent-specific strategies for individual data sources, posing challenges low-resource or black-box environments and complicates operations when evidence is fragmented across sources. To address these limitations, we propose ER-RAG, a framework that unifies evidence integration across heterogeneous data sources using the Entity-Relationship (ER) model. ER-RAG standardizes entity retrieval and relationship querying through ER-based APIs with GET and JOIN operations. It employs a two-stage generation process: first, a preference optimization module selects optimal sources; second, another module constructs API chains based on source schemas. This unified approach allows efficient fine-tuning and seamless integration across diverse data sources. ER-RAG demonstrated its effectiveness by winning all three tracks of the 2024 KDDCup CRAG Challenge, achieving performance on par with commercial RAG pipelines using an 8B LLM backbone. It outperformed hybrid competitors by 3.1% in LLM score and accelerated retrieval by 5.5X.
Authors: Jiaming Luo, Weiyi Luo, Guoqing Sun, Mengchen Zhu, Haifeng Tang, Kunyao Lan, Mengyue Wu, Kenny Q. Zhu
Abstract: Designing effective debt collection systems is crucial for improving operational efficiency and reducing costs in the financial industry. However, the challenges of maintaining script diversity, contextual relevance, and coherence make this task particularly difficult. This paper presents a debt collection system based on real debtor-collector data from a major commercial bank. We construct a script library from real-world debt collection conversations, and propose a two-stage retrieval based response system for contextual relevance. Experimental results show that our system improves script diversity, enhances response relevance, and achieves practical deployment efficiency through knowledge distillation. This work offers a scalable and automated solution, providing valuable insights for advancing debt collection practices in real-world applications.
Authors: Dang Nguyen, Chenhao Tan
Abstract: Understanding and mitigating biases is critical for the adoption of large language models (LLMs) in high-stakes decision-making. We introduce Admissions and Hiring, decision tasks with hypothetical applicant profiles where a person's race can be inferred from their name, as simplified test beds for racial bias. We show that Gemma 2B Instruct and LLaMA 3.2 3B Instruct exhibit strong biases. Gemma grants admission to 26% more White than Black applicants, and LLaMA hires 60% more Asian than White applicants. We demonstrate that these biases are resistant to prompt engineering: multiple prompting strategies all fail to promote fairness. In contrast, using distributed alignment search, we can identify "race subspaces" within model activations and intervene on them to debias model decisions. Averaging the representation across all races within the subspaces reduces Gemma's bias by 37-57%. Finally, we examine the generalizability of Gemma's race subspaces, and find limited evidence for generalization, where changing the prompt format can affect the race representation. Our work suggests mechanistic approaches may provide a promising venue for improving the fairness of LLMs, but a universal race representation remains elusive.
Authors: Jiali Cheng, Hadi Amiri
Abstract: Machine Unlearning aims to remove undesired information from trained models without requiring full retraining from scratch. Despite recent advancements, their underlying loss landscapes and optimization dynamics received less attention. In this paper, we investigate and analyze machine unlearning through the lens of mode connectivity - the phenomenon where independently trained models can be connected by smooth low-loss paths in the parameter space. We define and study mode connectivity in unlearning across a range of overlooked conditions, including connections between different unlearning methods, models trained with and without curriculum learning, and models optimized with first-order and secondorder techniques. Our findings show distinct patterns of fluctuation of different evaluation metrics along the curve, as well as the mechanistic (dis)similarity between unlearning methods. To the best of our knowledge, this is the first study on mode connectivity in the context of machine unlearning.
Authors: Chenrui Fan, Ming Li, Lichao Sun, Tianyi Zhou
Abstract: We find that the response length of reasoning LLMs, whether trained by reinforcement learning or supervised learning, drastically increases for ill-posed questions with missing premises (MiP), ending up with redundant and ineffective thinking. This newly introduced scenario exacerbates the general overthinking issue to a large extent, which we name as the MiP-Overthinking. Such failures are against the ``test-time scaling law'' but have been widely observed on multiple datasets we curated with MiP, indicating the harm of cheap overthinking and a lack of critical thinking. Surprisingly, LLMs not specifically trained for reasoning exhibit much better performance on the MiP scenario, producing much shorter responses that quickly identify ill-posed queries. This implies a critical flaw of the current training recipe for reasoning LLMs, which does not encourage efficient thinking adequately, leading to the abuse of thinking patterns. To further investigate the reasons behind such failures, we conduct fine-grained analyses of the reasoning length, overthinking patterns, and location of critical thinking on different types of LLMs. Moreover, our extended ablation study reveals that the overthinking is contagious through the distillation of reasoning models' responses. These results improve the understanding of overthinking and shed novel insights into mitigating the problem.
Authors: Li An, Yujian Liu, Yepeng Liu, Yang Zhang, Yuheng Bu, Shiyu Chang
Abstract: Watermarking has emerged as a promising technique for detecting texts generated by LLMs. Current research has primarily focused on three design criteria: high quality of the watermarked text, high detectability, and robustness against removal attack. However, the security against spoofing attacks remains relatively understudied. For example, a piggyback attack can maliciously alter the meaning of watermarked text-transforming it into hate speech-while preserving the original watermark, thereby damaging the reputation of the LLM provider. We identify two core challenges that make defending against spoofing difficult: (1) the need for watermarks to be both sensitive to semantic-distorting changes and insensitive to semantic-preserving edits, and (2) the contradiction between the need to detect global semantic shifts and the local, auto-regressive nature of most watermarking schemes. To address these challenges, we propose a semantic-aware watermarking algorithm that post-hoc embeds watermarks into a given target text while preserving its original meaning. Our method introduces a semantic mapping model, which guides the generation of a green-red token list, contrastively trained to be sensitive to semantic-distorting changes and insensitive to semantic-preserving changes. Experiments on two standard benchmarks demonstrate strong robustness against removal attacks and security against spoofing attacks, including sentiment reversal and toxic content insertion, while maintaining high watermark detectability. Our approach offers a significant step toward more secure and semantically aware watermarking for LLMs. Our code is available at https://github.com/UCSB-NLP-Chang/contrastive-watermark.
URLs: https://github.com/UCSB-NLP-Chang/contrastive-watermark.
Authors: Chrisantha Fernando, Dylan Banarse, Simon Osindero
Abstract: This paper explores an intrinsic motivation for mutual awareness, hypothesizing that humans possess a fundamental drive to understand \textit{and to be understood} even in the absence of extrinsic rewards. Through simulations of the perceptual crossing paradigm, we explore the effect of various internal reward functions in reinforcement learning agents. The drive to understand is implemented as an active inference type artificial curiosity reward, whereas the drive to be understood is implemented through intrinsic rewards for imitation, influence/impressionability, and sub-reaction time anticipation of the other. Results indicate that while artificial curiosity alone does not lead to a preference for social interaction, rewards emphasizing reciprocal understanding successfully drive agents to prioritize interaction. We demonstrate that this intrinsic motivation can facilitate cooperation in tasks where only one agent receives extrinsic reward for the behaviour of the other.
Authors: Xiaohua Feng, Yuyuan Li, Chengye Wang, Junlin Liu, Li Zhang, Chaochao Chen
Abstract: Driven by privacy protection laws and regulations, unlearning in Large Language Models (LLMs) is gaining increasing attention. However, current research often neglects the interpretability of the unlearning process, particularly concerning sample-level unlearning difficulty. Existing studies typically assume a uniform unlearning difficulty across samples. This simplification risks attributing the performance of unlearning algorithms to sample selection rather than the algorithm's design, potentially steering the development of LLM unlearning in the wrong direction. Thus, we investigate the relationship between LLM unlearning and sample characteristics, with a focus on unlearning difficulty. Drawing inspiration from neuroscience, we propose a Memory Removal Difficulty ($\mathrm{MRD}$) metric to quantify sample-level unlearning difficulty. Using $\mathrm{MRD}$, we analyze the characteristics of hard-to-unlearn versus easy-to-unlearn samples. Furthermore, we propose an $\mathrm{MRD}$-based weighted sampling method to optimize existing unlearning algorithms, which prioritizes easily forgettable samples, thereby improving unlearning efficiency and effectiveness. We validate the proposed metric and method using public benchmarks and datasets, with results confirming its effectiveness.
Authors: Xiaohua Feng, Yuyuan Li, Huwei Ji, Jiaming Zhang, Li Zhang, Tianyu Du, Chaochao Chen
Abstract: Despite advances in Preference Alignment (PA) for Large Language Models (LLMs), mainstream methods like Reinforcement Learning with Human Feedback (RLHF) face notable challenges. These approaches require high-quality datasets of positive preference examples, which are costly to obtain and computationally intensive due to training instability, limiting their use in low-resource scenarios. LLM unlearning technique presents a promising alternative, by directly removing the influence of negative examples. However, current research has primarily focused on empirical validation, lacking systematic quantitative analysis. To bridge this gap, we propose a framework to explore the relationship between PA and LLM unlearning. Specifically, we introduce a bi-level optimization-based method to quantify the impact of unlearning specific negative examples on PA performance. Our analysis reveals that not all negative examples contribute equally to alignment improvement when unlearned, and the effect varies significantly across examples. Building on this insight, we pose a crucial question: how can we optimally select and weight negative examples for unlearning to maximize PA performance? To answer this, we propose a framework called Unlearning to Align (U2A), which leverages bi-level optimization to efficiently select and unlearn examples for optimal PA performance. We validate the proposed method through extensive experiments, with results confirming its effectiveness.
Authors: Yoshihiro Yamada
Abstract: Transformers have driven remarkable breakthroughs in natural language processing and computer vision, yet their standard attention mechanism still imposes O(N^2) complexity, hindering scalability to longer sequences. We introduce Circular-convolutional ATtention (CAT), a Fourier-based approach that efficiently applies circular convolutions to reduce complexity without sacrificing representational power. CAT achieves O(NlogN) computations, requires fewer learnable parameters by streamlining fully-connected layers, and introduces no heavier operations, resulting in consistent accuracy improvements and about a 10% speedup in naive PyTorch implementations on large-scale benchmarks such as ImageNet-1k and WikiText-103. Grounded in an engineering-isomorphism framework, CAT's design not only offers practical efficiency and ease of implementation but also provides insights to guide the development of next-generation, high-performance Transformer architectures. Finally, our ablation studies highlight the key conditions underlying CAT's success, shedding light on broader principles for scalable attention mechanisms.
Authors: Yuxin Wang, Yiran Guo, Yining Zheng, Zhangyue Yin, Shuo Chen, Jie Yang, Jiajun Chen, Xuanjing Huang, Xipeng Qiu
Abstract: The integration of tool learning with Large Language Models (LLMs) has expanded their capabilities in handling complex tasks by leveraging external tools. However, existing benchmarks for tool learning inadequately address critical real-world personalized scenarios, particularly those requiring multi-hop reasoning and inductive knowledge adaptation in dynamic environments. To bridge this gap, we introduce FamilyTool, a novel benchmark grounded in a family-based knowledge graph (KG) that simulates personalized, multi-hop tool use scenarios. FamilyTool challenges LLMs with queries spanning 1 to 3 relational hops (e.g., inferring familial connections and preferences) and incorporates an inductive KG setting where models must adapt to unseen user preferences and relationships without re-training, a common limitation in prior approaches that compromises generalization. We further propose KGETool: a simple KG-augmented evaluation pipeline to systematically assess LLMs' tool use ability in these settings. Experiments reveal significant performance gaps in state-of-the-art LLMs, with accuracy dropping sharply as hop complexity increases and inductive scenarios exposing severe generalization deficits. These findings underscore the limitations of current LLMs in handling personalized, evolving real-world contexts and highlight the urgent need for advancements in tool-learning frameworks. FamilyTool serves as a critical resource for evaluating and advancing LLM agents' reasoning, adaptability, and scalability in complex, dynamic environments. Code and dataset are available at Github.
Authors: Zhixuan Lin, Johan Obando-Ceron, Xu Owen He, Aaron Courville
Abstract: The recently proposed Forgetting Transformer (FoX) incorporates a forget gate into softmax attention and has shown consistently better or on-par performance compared to the standard RoPE-based Transformer. Notably, many attention heads in FoX tend to forget quickly, causing their output at each timestep to rely primarily on the local context. Based on this observation, we propose Adaptive Computation Pruning (ACP) for FoX, a method that dynamically prunes computations involving input-output dependencies that are strongly decayed by the forget gate. This is achieved using a dynamically set pruning threshold that ensures that the pruned attention weights remain negligible. We apply ACP to language model pretraining with FoX and show it consistently reduces the number of FLOPs in softmax attention by around 70% across different model sizes and context lengths, resulting in a roughly 10% to 35% improvement in training throughput. Furthermore, longer context lengths yield greater computational savings. All these speed improvements are achieved without any performance degradation. We also perform several analyses to provide deeper insights into our method, such as examining the pruning patterns and analyzing the distribution of FLOP savings across different attention heads. Our code is available at https://github.com/zhixuan-lin/arctic-fox.
Authors: Vladimir Bataev
Abstract: Training speech recognition systems on noisy transcripts is a significant challenge in industrial pipelines, where datasets are enormous and ensuring accurate transcription for every instance is difficult. In this work, we introduce novel loss functions to mitigate the impact of transcription errors in RNN-Transducer models. Our Star-Transducer loss addresses deletion errors by incorporating "skip frame" transitions in the loss lattice, restoring over 90% of the system's performance compared to models trained with accurate transcripts. The Bypass-Transducer loss uses "skip token" transitions to tackle insertion errors, recovering more than 60% of the quality. Finally, the Target-Robust Transducer loss merges these approaches, offering robust performance against arbitrary errors. Experimental results demonstrate that the Target-Robust Transducer loss significantly improves RNN-T performance on noisy data by restoring over 70% of the quality compared to well-transcribed data.
Authors: Jifang Wang, Xue Yang, Longyue Wang, Zhenran Xu, Yiyu Wang, Yaowei Wang, Weihua Luo, Kaifu Zhang, Baotian Hu, Min Zhang
Abstract: Conditional image generation has gained significant attention for its ability to personalize content. However, the field faces challenges in developing task-agnostic, reliable, and explainable evaluation metrics. This paper introduces CIGEval, a unified agentic framework for comprehensive evaluation of conditional image generation tasks. CIGEval utilizes large multimodal models (LMMs) as its core, integrating a multi-functional toolbox and establishing a fine-grained evaluation framework. Additionally, we synthesize evaluation trajectories for fine-tuning, empowering smaller LMMs to autonomously select appropriate tools and conduct nuanced analyses based on tool outputs. Experiments across seven prominent conditional image generation tasks demonstrate that CIGEval (GPT-4o version) achieves a high correlation of 0.4625 with human assessments, closely matching the inter-annotator correlation of 0.47. Moreover, when implemented with 7B open-source LMMs using only 2.3K training trajectories, CIGEval surpasses the previous GPT-4o-based state-of-the-art method. Case studies on GPT-4o image generation highlight CIGEval's capability in identifying subtle issues related to subject consistency and adherence to control guidance, indicating its great potential for automating evaluation of image generation tasks with human-level reliability.
Authors: Boyuan Zheng, Michael Y. Fatemi, Xiaolong Jin, Zora Zhiruo Wang, Apurva Gandhi, Yueqi Song, Yu Gu, Jayanth Srinivasa, Gaowen Liu, Graham Neubig, Yu Su
Abstract: To survive and thrive in complex environments, humans have evolved sophisticated self-improvement mechanisms through environment exploration, hierarchical abstraction of experiences into reuseable skills, and collaborative construction of an ever-growing skill repertoire. Despite recent advancements, autonomous web agents still lack crucial self-improvement capabilities, struggling with procedural knowledge abstraction, refining skills, and skill composition. In this work, we introduce SkillWeaver, a skill-centric framework enabling agents to self-improve by autonomously synthesizing reusable skills as APIs. Given a new website, the agent autonomously discovers skills, executes them for practice, and distills practice experiences into robust APIs. Iterative exploration continually expands a library of lightweight, plug-and-play APIs, significantly enhancing the agent's capabilities. Experiments on WebArena and real-world websites demonstrate the efficacy of SkillWeaver, achieving relative success rate improvements of 31.8% and 39.8%, respectively. Additionally, APIs synthesized by strong agents substantially enhance weaker agents through transferable skills, yielding improvements of up to 54.3% on WebArena. These results demonstrate the effectiveness of honing diverse website interactions into APIs, which can be seamlessly shared among various web agents.
Authors: Andreas Hochlehnert, Hardik Bhatnagar, Vishaal Udandarao, Samuel Albanie, Ameya Prabhu, Matthias Bethge
Abstract: Reasoning has emerged as the next major frontier for language models (LMs), with rapid advances from both academic and industrial labs. However, this progress often outpaces methodological rigor, with many evaluations relying on benchmarking practices that lack transparency, robustness, or statistical grounding. In this work, we conduct a comprehensive empirical study and find that current mathematical reasoning benchmarks are highly sensitive to subtle implementation choices - including decoding parameters, random seeds, prompt formatting, and even hardware and software-framework configurations. Performance gains reported in recent studies frequently hinge on unclear comparisons or unreported sources of variance. To address these issues, we propose a standardized evaluation framework with clearly defined best practices and reporting standards. Using this framework, we reassess recent methods and find that reinforcement learning (RL) approaches yield only modest improvements - far below prior claims - and are prone to overfitting, especially on small-scale benchmarks like AIME24. In contrast, supervised finetuning (SFT) methods show consistently stronger generalization. To foster reproducibility, we release all code, prompts, and model outputs, for reasoning benchmarks, establishing more rigorous foundations for future work.
Authors: Yiting Lu, Jiakang Yuan, Zhen Li, Shitian Zhao, Qi Qin, Xinyue Li, Le Zhuo, Licheng Wen, Dongyang Liu, Yuewen Cao, Xiangchao Yan, Xin Li, Botian Shi, Tao Chen, Zhibo Chen, Lei Bai, Bo Zhang, Peng Gao
Abstract: We propose OmniCaptioner, a versatile visual captioning framework for generating fine-grained textual descriptions across a wide variety of visual domains. Unlike prior methods limited to specific image types (e.g., natural images or geometric visuals), our framework provides a unified solution for captioning natural images, visual text (e.g., posters, UIs, textbooks), and structured visuals (e.g., documents, tables, charts). By converting low-level pixel information into semantically rich textual representations, our framework bridges the gap between visual and textual modalities. Our results highlight three key advantages: (i) Enhanced Visual Reasoning with LLMs, where long-context captions of visual modalities empower LLMs, particularly the DeepSeek-R1 series, to reason effectively in multimodal scenarios; (ii) Improved Image Generation, where detailed captions improve tasks like text-to-image generation and image transformation; and (iii) Efficient Supervised Fine-Tuning (SFT), which enables faster convergence with less data. We believe the versatility and adaptability of OmniCaptioner can offer a new perspective for bridging the gap between language and visual modalities.
Authors: Nikhil Shivakumar Nayak, Krishnateja Killamsetty, Ligong Han, Abhishek Bhandwaldar, Prateek Chanda, Kai Xu, Hao Wang, Aldo Pareja, Oleg Silkin, Mustafa Eyceoz, Akash Srivastava
Abstract: Continual learning in large language models (LLMs) is prone to catastrophic forgetting, where adapting to new tasks significantly degrades performance on previously learned ones. Existing methods typically rely on low-rank, parameter-efficient updates that limit the model's expressivity and introduce additional parameters per task, leading to scalability issues. To address these limitations, we propose a novel continual full fine-tuning approach leveraging adaptive singular value decomposition (SVD). Our method dynamically identifies task-specific low-rank parameter subspaces and constrains updates to be orthogonal to critical directions associated with prior tasks, thus effectively minimizing interference without additional parameter overhead or storing previous task gradients. We evaluate our approach extensively on standard continual learning benchmarks using both encoder-decoder (T5-Large) and decoder-only (LLaMA-2 7B) models, spanning diverse tasks including classification, generation, and reasoning. Empirically, our method achieves state-of-the-art results, up to 7% higher average accuracy than recent baselines like O-LoRA, and notably maintains the model's general linguistic capabilities, instruction-following accuracy, and safety throughout the continual learning process by reducing forgetting to near-negligible levels. Our adaptive SVD framework effectively balances model plasticity and knowledge retention, providing a practical, theoretically grounded, and computationally scalable solution for continual learning scenarios in large language models.
Authors: Manuel Faysse, Patrick Fernandes, Nuno M. Guerreiro, Ant\'onio Loison, Duarte M. Alves, Caio Corro, Nicolas Boizard, Jo\~ao Alves, Ricardo Rei, Pedro H. Martins, Antoni Bigata Casademunt, Fran\c{c}ois Yvon, Andr\'e F. T. Martins, Gautier Viaud, C\'eline Hudelot, Pierre Colombo
Abstract: We introduce CroissantLLM, a 1.3B language model pretrained on a set of 3T English and French tokens, to bring to the research and industrial community a high-performance, fully open-sourced bilingual model that runs swiftly on consumer-grade local hardware. To that end, we pioneer the approach of training an intrinsically bilingual model with a 1:1 English-to-French pretraining data ratio, a custom tokenizer, and bilingual finetuning datasets. We release the training dataset, notably containing a French split with manually curated, high-quality, and varied data sources. To assess performance outside of English, we craft a novel benchmark, FrenchBench, consisting of an array of classification and generation tasks, covering various orthogonal aspects of model performance in the French Language. Additionally, rooted in transparency and to foster further Large Language Model research, we release codebases, and dozens of checkpoints across various model sizes, training data distributions, and training steps, as well as fine-tuned Chat models, and strong translation models. We evaluate our model through the FMTI framework, and validate 81 % of the transparency criteria, far beyond the scores of even most open initiatives. This work enriches the NLP landscape, breaking away from previous English-centric work in order to strengthen our understanding of multilinguality in language models.
Authors: Nicholas Meade, Arkil Patel, Siva Reddy
Abstract: Recent work has developed optimization procedures to find token sequences, called adversarial triggers, which can elicit unsafe responses from aligned language models. These triggers are believed to be highly transferable, i.e., a trigger optimized on one model can jailbreak other models. In this paper, we concretely show that such adversarial triggers are not consistently transferable. We extensively investigate trigger transfer amongst 13 open models and observe poor and inconsistent transfer. Our experiments further reveal a significant difference in robustness to adversarial triggers between models Aligned by Preference Optimization (APO) and models Aligned by Fine-Tuning (AFT). We find that APO models are extremely hard to jailbreak even when the trigger is optimized directly on the model. On the other hand, while AFT models may appear safe on the surface, exhibiting refusals to a range of unsafe instructions, we show that they are highly susceptible to adversarial triggers. Lastly, we observe that most triggers optimized on AFT models also generalize to new unsafe instructions from five diverse domains, further emphasizing their vulnerability. Overall, our work highlights the need for more comprehensive safety evaluations for aligned language models.
Authors: Simone Barandoni, Filippo Chiarello, Lorenzo Cascone, Emiliano Marrale, Salvatore Puccio
Abstract: In the rapidly evolving landscape of Natural Language Processing (NLP), Large Language Models (LLMs) have emerged as powerful tools for many tasks, such as extracting valuable insights from vast amounts of textual data. In this study, we conduct a comparative analysis of LLMs for the extraction of travel customer needs from TripAdvisor and Reddit posts. Leveraging a diverse range of models, including both open-source and proprietary ones such as GPT-4 and Gemini, we aim to elucidate their strengths and weaknesses in this specialized domain. Through an evaluation process involving metrics such as BERTScore, ROUGE, and BLEU, we assess the performance of each model in accurately identifying and summarizing customer needs. Our findings highlight the efficacy of opensource LLMs, particularly Mistral 7B, in achieving comparable performance to larger closed models while offering affordability and customization benefits. Additionally, we underscore the importance of considering factors such as model size, resource requirements, and performance metrics when selecting the most suitable LLM for customer needs analysis tasks. Overall, this study contributes valuable insights for businesses seeking to leverage advanced NLP techniques to enhance customer experience and drive operational efficiency in the travel industry.
Authors: Hanyang Zhong, Liman Wang, Wenting Cao, Zeyuan Sun
Abstract: This paper examines the role of cognitive biases in the decision-making processes of large language models (LLMs), challenging the conventional goal of eliminating all biases. When properly balanced, we show that certain cognitive biases can enhance decision-making efficiency through rational deviations and heuristic shortcuts. By introducing heuristic moderation and an abstention option, which allows LLMs to withhold responses when uncertain, we reduce error rates, improve decision accuracy, and optimize decision rates. Using the Balance Rigor and Utility (BRU) dataset, developed through expert collaboration, our findings demonstrate that targeted inspection of cognitive biases aligns LLM decisions more closely with human reasoning, enhancing reliability and suggesting strategies for future improvements. This approach offers a novel way to leverage cognitive biases to improve the practical utility of LLMs across various applications.
Authors: Yuni Susanti, Nina Holsmoelle
Abstract: This study explores the capability of Large Language Models (LLMs) to evaluate causality in causal graphs generated by conventional statistical causal discovery methods-a task traditionally reliant on manual assessment by human subject matter experts. To bridge this gap in causality assessment, LLMs are employed to evaluate the causal relationships by determining whether a causal connection between variable pairs can be inferred from textual context. Our study compares two approaches: (1) prompting-based method for zero-shot and few-shot causal inference and, (2) fine-tuning language models for the causal relation prediction task. While prompt-based LLMs have demonstrated versatility across various NLP tasks, our experiments on biomedical and general-domain datasets show that fine-tuned models consistently outperform them, achieving up to a 20.5-point improvement in F1 score-even when using smaller-parameter language models. These findings provide valuable insights into the strengths and limitations of both approaches for causal graph evaluation.
Authors: Feijie Wu, Xiaoze Liu, Haoyu Wang, Xingchen Wang, Lu Su, Jing Gao
Abstract: Reinforcement learning with human feedback (RLHF) fine-tunes a pretrained large language model (LLM) using user preference data, enabling it to generate content aligned with human preferences. However, due to privacy concerns, users may be reluctant to share sensitive preference data. To address this, we propose utilizing Federated Learning (FL) techniques, allowing large-scale preference collection from diverse real-world users without requiring them to transmit data to a central server. Our federated RLHF methods (i.e., FedBis and FedBiscuit) encode each client's preferences into binary selectors and aggregate them to capture common preferences. In particular, FedBiscuit overcomes key challenges, such as preference heterogeneity and reward hacking, through innovative solutions like grouping clients with similar preferences to reduce heterogeneity and using multiple binary selectors to enhance LLM output quality. To evaluate the performance of the proposed methods, we establish the first federated RLHF benchmark with a heterogeneous human preference dataset. Experimental results show that by integrating the LLM with aggregated client preferences, FedBis and FedBiscuit significantly enhance the professionalism and readability of the generated content.
Authors: Hongjin Qian, Zheng Liu, Peitian Zhang, Kelong Mao, Defu Lian, Zhicheng Dou, Tiejun Huang
Abstract: Processing long contexts presents a significant challenge for large language models (LLMs). While recent advancements allow LLMs to handle much longer contexts than before (e.g., 32K or 128K tokens), it is computationally expensive and can still be insufficient for many applications. Retrieval-Augmented Generation (RAG) is considered a promising strategy to address this problem. However, conventional RAG methods face inherent limitations because of two underlying requirements: 1) explicitly stated queries, and 2) well-structured knowledge. These conditions, however, do not hold in general long-context processing tasks. In this work, we propose MemoRAG, a novel RAG framework empowered by global memory-augmented retrieval. MemoRAG features a dual-system architecture. First, it employs a light but long-range system to create a global memory of the long context. Once a task is presented, it generates draft answers, providing useful clues for the retrieval tools to locate relevant information within the long context. Second, it leverages an expensive but expressive system, which generates the final answer based on the retrieved information. Building upon this fundamental framework, we realize the memory module in the form of KV compression, and reinforce its memorization and cluing capacity from the Generation quality's Feedback (a.k.a. RLGF). In our experiments, MemoRAG achieves superior performances across a variety of long-context evaluation tasks, not only complex scenarios where traditional RAG methods struggle, but also simpler ones where RAG is typically applied.
Authors: Ilya Gusev
Abstract: We introduce a benchmark for evaluating the role-playing capabilities of language models. Our approach leverages different language models to simulate users in dynamic, multi-turn conversations and assess the resulting dialogues. Our methodology involves three main components: a player model that adopts a specific character role, an interrogator model that simulates user behavior in a specific situation, and a judge model ensemble that evaluates conversation quality with 3 metrics: character consistency, entertainment value, and language fluency. We evaluated more than 40 models in both English and Russian, with each model participating in 64 conversations with 8 characters and 8 situations. We conducted experiments comparing automated evaluations with human annotations to validate our approach, demonstrating strong correlations across multiple criteria. This work provides a foundation for a robust and dynamic evaluation of different model capabilities in interactive scenarios.
Authors: Tommaso Giorgi, Lorenzo Cima, Tiziano Fagni, Marco Avvenuti, Stefano Cresci
Abstract: The rise of online platforms exacerbated the spread of hate speech, demanding scalable and effective detection. However, the accuracy of hate speech detection systems heavily relies on human-labeled data, which is inherently susceptible to biases. While previous work has examined the issue, the interplay between the characteristics of the annotator and those of the target of the hate are still unexplored. We fill this gap by leveraging an extensive dataset with rich socio-demographic information of both annotators and targets, uncovering how human biases manifest in relation to the target's attributes. Our analysis surfaces the presence of widespread biases, which we quantitatively describe and characterize based on their intensity and prevalence, revealing marked differences. Furthermore, we compare human biases with those exhibited by persona-based LLMs. Our findings indicate that while persona-based LLMs do exhibit biases, these differ significantly from those of human annotators. Overall, our work offers new and nuanced results on human biases in hate speech annotations, as well as fresh insights into the design of AI-driven hate speech detection systems.
Authors: Zhiyuan Peng, Jinming Nian, Alexandre Evfimievski, Yi Fang
Abstract: Large Language Models (LLMs) are widely used in Conversational AI systems to generate responses to user inquiries. However, many natural questions lack well-defined answers. While existing studies primarily focus on question types such as false premises, they often overlook out-of-scope questions, where the provided document is semantically highly similar to the query but does not contain the required answer. In this paper, we propose a guided hallucination-based method to efficiently generate a diverse set of out-of-scope questions from a given document corpus. We then evaluate multiple LLMs based on their effectiveness in confusion detection and appropriate response generation. Furthermore, we introduce an improved method for detecting such out-of-scope questions, enhancing the reliability of LLM-based question-answering systems.
Authors: Aoi Ito, Kota Dohi, Yohei Kawaguchi
Abstract: This paper presents CLaSP, a novel model for retrieving time-series signals using natural language queries that describe signal characteristics. The ability to search time-series signals based on descriptive queries is essential in domains such as industrial diagnostics, where data scientists often need to find signals with specific characteristics. However, existing methods rely on sketch-based inputs, predefined synonym dictionaries, or domain-specific manual designs, limiting their scalability and adaptability. CLaSP addresses these challenges by employing contrastive learning to map time-series signals to natural language descriptions. Unlike prior approaches, it eliminates the need for predefined synonym dictionaries and leverages the rich contextual knowledge of large language models (LLMs). Using the TRUCE and SUSHI datasets, which pair time-series signals with natural language descriptions, we demonstrate that CLaSP achieves high accuracy in retrieving a variety of time series patterns based on natural language queries.
Authors: Gabriel Chua, Shing Yee Chan, Shaun Khoo
Abstract: Large Language Models (LLMs) are prone to off-topic misuse, where users may prompt these models to perform tasks beyond their intended scope. Current guardrails, which often rely on curated examples or custom classifiers, suffer from high false-positive rates, limited adaptability, and the impracticality of requiring real-world data that is not available in pre-production. In this paper, we introduce a flexible, data-free guardrail development methodology that addresses these challenges. By thoroughly defining the problem space qualitatively and passing this to an LLM to generate diverse prompts, we construct a synthetic dataset to benchmark and train off-topic guardrails that outperform heuristic approaches. Additionally, by framing the task as classifying whether the user prompt is relevant with respect to the system prompt, our guardrails effectively generalize to other misuse categories, including jailbreak and harmful prompts. Lastly, we further contribute to the field by open-sourcing both the synthetic dataset and the off-topic guardrail models, providing valuable resources for developing guardrails in pre-production environments and supporting future research and development in LLM safety.
Authors: Meher Bhardwaj, Hrishikesh Ethari, Dennis Singh Moirangthem
Abstract: The SQL-to-text generation task traditionally uses template base, Seq2Seq, tree-to-sequence, and graph-to-sequence models. Recent models take advantage of pre-trained generative language models for this task in the Seq2Seq framework. However, treating SQL as a sequence of inputs to the pre-trained models is not optimal. In this work, we put forward a new SQL intermediate representation called EzSQL to align SQL with the natural language text sequence. EzSQL simplifies the SQL queries and brings them closer to natural language text by modifying operators and keywords, which can usually be described in natural language. EzSQL also removes the need for set operators. Our proposed SQL-to-text generation model uses EzSQL as the input to a pre-trained generative language model for generating the text descriptions. We demonstrate that our model is an effective state-of-the-art method to generate text narrations from SQL queries on the WikiSQL and Spider datasets. We also show that by generating pretraining data using our SQL-to-text generation model, we can enhance the performance of Text-to-SQL parsers.
Authors: Chulun Zhou, Qiujing Wang, Mo Yu, Xiaoqian Yue, Rui Lu, Jiangnan Li, Yifan Zhou, Shunchi Zhang, Jie Zhou, Wai Lam
Abstract: Theory-of-Mind (ToM) is a fundamental psychological capability that allows humans to understand and interpret the mental states of others. Humans infer others' thoughts by integrating causal cues and indirect clues from broad contextual information, often derived from past interactions. In other words, human ToM heavily relies on the understanding about the backgrounds and life stories of others. Unfortunately, this aspect is largely overlooked in existing benchmarks for evaluating machines' ToM capabilities, due to their usage of short narratives without global context, especially personal background of characters. In this paper, we verify the importance of comprehensive contextual understanding about personal backgrounds in ToM and assess the performance of LLMs in such complex scenarios. To achieve this, we introduce CharToM benchmark, comprising 1,035 ToM questions based on characters from classic novels. Our human study reveals a significant disparity in performance: the same group of educated participants performs dramatically better when they have read the novels compared to when they have not. In parallel, our experiments on state-of-the-art LLMs, including the very recent o1 and DeepSeek-R1 models, show that LLMs still perform notably worse than humans, despite that they have seen these stories during pre-training. This highlights the limitations of current LLMs in capturing the nuanced contextual information required for ToM reasoning.
Authors: Mingyang Song, Zhaochen Su, Xiaoye Qu, Jiawei Zhou, Yu Cheng
Abstract: Process-level Reward Models (PRMs) are crucial for complex reasoning and decision-making tasks, where each intermediate step plays an important role in the reasoning process. Since language models are prone to various types of errors during the reasoning process, PRMs are required to possess nuanced capabilities for detecting various implicit error types in real-world scenarios. However, current benchmarks primarily focus on step correctness, failing to evaluate PRMs' performance systematically. To address this gap, we introduce PRMBench, a process-level benchmark specifically designed to assess the fine-grained error detection capabilities of PRMs. PRMBench comprises 6,216 carefully designed problems and 83,456 step-level labels, evaluating models across multiple dimensions, including simplicity, soundness, and sensitivity. In our experiments on 15 models, spanning both open-source PRMs and closed-source large language models prompted as critic models, we uncover significant weaknesses in current PRMs. These findings underscore the challenges inherent in process-level evaluation and highlight key directions for future research. We hope PRMBench can be a robust bench for advancing research on PRM evaluation and development.
Authors: Tianshi Zheng, Jiayang Cheng, Chunyang Li, Haochen Shi, Zihao Wang, Jiaxin Bai, Yangqiu Song, Ginny Y. Wong, Simon See
Abstract: Modern large language models (LLMs) employ various forms of logical inference, both implicitly and explicitly, when addressing reasoning tasks. Understanding how to optimally leverage these inference paradigms is critical for advancing LLMs' reasoning capabilities. This paper adopts an exploratory approach by introducing a controlled evaluation environment for analogical reasoning -- a fundamental cognitive task -- that is systematically parameterized across three dimensions: modality (textual, visual, symbolic), difficulty (easy, medium, hard), and task format (multiple-choice or free-text generation). We analyze the comparative dynamics of inductive, abductive, and deductive inference pipelines across these dimensions, and demonstrate that our findings generalize to broader in-context learning tasks. Additionally, we investigate advanced paradigms such as hypothesis selection, verification, and refinement, revealing their potential to scale up logical inference in LLM reasoning. This exploratory study provides a foundation for future research in enhancing LLM reasoning through systematic logical inference strategies. Resources are available at https://github.com/HKUST-KnowComp/LogiDynamics.
Authors: Nicola Cecere, Andrea Bacciu, Ignacio Fern\'andez Tob\'ias, Amin Mantrach
Abstract: Uncertainty quantification (UQ) in Large Language Models (LLMs) is essential for their safe and reliable deployment, particularly in critical applications where incorrect outputs can have serious consequences. Current UQ methods typically rely on querying the model multiple times using non-zero temperature sampling to generate diverse outputs for uncertainty estimation. However, the impact of selecting a given temperature parameter is understudied, and our analysis reveals that temperature plays a fundamental role in the quality of uncertainty estimates. The conventional approach of identifying optimal temperature values requires expensive hyperparameter optimization (HPO) that must be repeated for each new model-dataset combination. We propose Monte Carlo Temperature (MCT), a robust sampling strategy that eliminates the need for temperature calibration. Our analysis reveals that: 1) MCT provides more robust uncertainty estimates across a wide range of temperatures, 2) MCT improves the performance of UQ methods by replacing fixed-temperature strategies that do not rely on HPO, and 3) MCT achieves statistical parity with oracle temperatures, which represent the ideal outcome of a well-tuned but computationally expensive HPO process. These findings demonstrate that effective UQ can be achieved without the computational burden of temperature parameter calibration.
Authors: Andrei-Marius Avram, Marian Lupa\c{s}cu, Dumitru-Clementin Cercel, Ionu\c{t} Mironic\u{a}, \c{S}tefan Tr\u{a}u\c{s}an-Matu
Abstract: This paper presents UniBERT, a compact multilingual language model that leverages an innovative training framework integrating three components: masked language modeling, adversarial training, and knowledge distillation. Pre-trained on a meticulously curated Wikipedia corpus spanning 107 languages, UniBERT is designed to reduce the computational demands of large-scale models while maintaining competitive performance across various natural language processing tasks. Comprehensive evaluations on four tasks -- named entity recognition, natural language inference, question answering, and semantic textual similarity -- demonstrate that our multilingual training strategy enhanced by an adversarial objective significantly improves cross-lingual generalization. Specifically, UniBERT models show an average relative improvement of 7.72% over traditional baselines, which achieved an average relative improvement of only 1.17%, with statistical analysis confirming the significance of these gains (p-value = 0.0181). This work highlights the benefits of combining adversarial training and knowledge distillation to build scalable and robust language models, thereby advancing the field of multilingual and cross-lingual natural language processing.
Authors: Xuechen Liang, Meiling Tao, Yinghui Xia, Jianhui Wang, Kun Li, Yijin Wang, Jingsong Yang, Tianyu Shi, Yuantao Wang, Miao Zhang, Xueqian Wang
Abstract: Large language models (LLMs) have made significant advances in the field of natural language processing, but they still face challenges such as continuous decision-making, lack of long-term memory, and limited context windows in dynamic environments. To address these issues, this paper proposes an innovative framework Memory-Enhanced Agents with Reflective Self-improvement. The MARS framework comprises three agents: the User, the Assistant, and the Checker. By integrating iterative feedback, reflective mechanisms, and a memory optimization mechanism based on the Ebbinghaus forgetting curve, it significantly enhances the agents capabilities in handling multi-tasking and long-span information.
Authors: Iuri Macocco, Nora Graichen, Gemma Boleda, Marco Baroni
Abstract: We study last-layer outlier dimensions, i.e. dimensions that display extreme activations for the majority of inputs. We show that outlier dimensions arise in many different modern language models, and trace their function back to the heuristic of constantly predicting frequent words. We further show how a model can block this heuristic when it is not contextually appropriate, by assigning a counterbalancing weight mass to the remaining dimensions, and we investigate which model parameters boost outlier dimensions and when they arise during training. We conclude that outlier dimensions are a specialized mechanism discovered by many distinct models to implement a useful token prediction heuristic.
Authors: Abdul Sittar, Luka Golob, Mateja Smiljanic
Abstract: This study explores the generation and evaluation of synthetic fake news through fact based manipulations using large language models (LLMs). We introduce a novel methodology that extracts key facts from real articles, modifies them, and regenerates content to simulate fake news while maintaining coherence. To assess the quality of the generated content, we propose a set of evaluation metrics coherence, dissimilarity, and correctness. The research also investigates the application of synthetic data in fake news classification, comparing traditional machine learning models with transformer based models such as BERT. Our experiments demonstrate that transformer models, especially BERT, effectively leverage synthetic data for fake news detection, showing improvements with smaller proportions of synthetic data. Additionally, we find that fact verification features, which focus on identifying factual inconsistencies, provide the most promising results in distinguishing synthetic fake news. The study highlights the potential of synthetic data to enhance fake news detection systems, offering valuable insights for future research and suggesting that targeted improvements in synthetic data generation can further strengthen detection models.
Authors: Yao Tao, Yehui Tang, Yun Wang, Mingjian Zhu, Hailin Hu, Yunhe Wang
Abstract: Despite the recent success of large language models (LLMs), LLMs are particularly challenging in long-sequence inference scenarios due to the quadratic computational complexity of the attention mechanism. Inspired by the interpretability theory of feature attribution in neural network models, we observe that not all tokens have the same contribution. Based on this observation, we propose a novel token pruning framework, namely Saliency-driven Dynamic Token Pruning (SDTP), to gradually and dynamically prune redundant tokens based on the input context. Specifically, a lightweight saliency-driven prediction module is designed to estimate the importance score of each token with its hidden state, which is added to different layers of the LLM to hierarchically prune redundant tokens. Furthermore, a ranking-based optimization strategy is proposed to minimize the ranking divergence of the saliency score and the predicted importance score. Extensive experiments have shown that our framework is generalizable to various models and datasets. By hierarchically pruning 65\% of the input tokens, our method greatly reduces 33\% $\sim$ 47\% FLOPs and achieves speedup up to 1.75$\times$ during inference, while maintaining comparable performance. We further demonstrate that SDTP can be combined with KV cache compression method for further compression.
Authors: Yifei Yu, Qian-Wen Zhang, Lingfeng Qiao, Di Yin, Fang Li, Jie Wang, Zengxi Chen, Suncong Zheng, Xiaolong Liang, Xing Sun
Abstract: Evaluating the ability of large language models (LLMs) to handle extended contexts is critical, particularly for retrieving information relevant to specific queries embedded within lengthy inputs. We introduce Sequential-NIAH, a benchmark specifically designed to evaluate the capability of LLMs to extract sequential information items (known as needles) from long contexts. The benchmark comprises three types of needle generation pipelines: synthetic, real, and open-domain QA. It includes contexts ranging from 8K to 128K tokens in length, with a dataset of 14,000 samples (2,000 reserved for testing). To facilitate evaluation on this benchmark, we trained a synthetic data-driven evaluation model capable of evaluating answer correctness based on chronological or logical order, achieving an accuracy of 99.49% on synthetic test data. We conducted experiments on six well-known LLMs, revealing that even the best-performing model achieved a maximum accuracy of only 63.15%. Further analysis highlights the growing challenges posed by increasing context lengths and the number of needles, underscoring substantial room for improvement. Additionally, noise robustness experiments validate the reliability of the benchmark, making Sequential-NIAH an important reference for advancing research on long text extraction capabilities of LLMs.
Authors: Elisabeth Fittschen, Sabrina Li, Tom Lippincott, Leshem Choshen, Craig Messner
Abstract: Large language models (LLMs) have shown potential as tools for scientific discovery. This has engendered growing interest in their use in humanistic disciplines, such as historical linguistics and literary studies. These fields often construct arguments on the basis of delineations like genre, or more inflexibly, time period. Although efforts have been made to restrict inference to specific domains via fine-tuning or model editing, we posit that the only true guarantee is domain-restricted pretraining -- typically, a data- and compute-expensive proposition. We show that efficient pretraining techniques can produce useful models over corpora too large for easy manual inspection but too small for "typical" LLM approaches. We employ a novel date-attribution pipeline in order to obtain a temporally-segmented dataset of five 10-million-word slices. We train two corresponding five-model batteries over these corpus segments, efficient pretraining and Llama3-8B parameter efficiently finetuned. We find that the pretrained models are faster to train than the finetuned baselines and that they better respect the historical divisions of our corpus. Emphasizing speed and precision over a-historical comprehensiveness enables a number of novel approaches to hypothesis discovery and testing in our target fields. Taking up diachronic linguistics as a testbed, we show that our method enables the detection of a diverse set of phenomena, including en masse lexical change, non-lexical (grammatical and morphological) change, and word sense introduction/obsolescence. We provide a ready-to-use pipeline that allows extension of our approach to other target fields with only minimal adaptation.
Authors: Sanchit Kabra, Akshita Jha, Chandan K. Reddy
Abstract: Recent advances in large-scale generative language models have shown that reasoning capabilities can significantly improve model performance across a variety of tasks. However, the impact of reasoning on a model's ability to mitigate stereotypical responses remains largely underexplored. In this work, we investigate the crucial relationship between a model's reasoning ability and fairness, and ask whether improved reasoning capabilities can mitigate harmful stereotypical responses, especially those arising due to shallow or flawed reasoning. We conduct a comprehensive evaluation of multiple open-source LLMs, and find that larger models with stronger reasoning abilities exhibit substantially lower stereotypical bias on existing fairness benchmarks. Building on this insight, we introduce ReGiFT -- Reasoning Guided Fine-Tuning, a novel approach that extracts structured reasoning traces from advanced reasoning models and infuses them into models that lack such capabilities. We use only general-purpose reasoning and do not require any fairness-specific supervision for bias mitigation. Notably, we see that models fine-tuned using ReGiFT not only improve fairness relative to their non-reasoning counterparts but also outperform advanced reasoning models on fairness benchmarks. We also analyze how variations in the correctness of the reasoning traces and their length influence model fairness and their overall performance. Our findings highlight that enhancing reasoning capabilities is an effective, fairness-agnostic strategy for mitigating stereotypical bias caused by reasoning flaws.
Authors: Nathana\"el Beau, Beno\^it Crabb\'e
Abstract: As text and code resources have expanded, large-scale pre-trained models have shown promising capabilities in code generation tasks, typically employing supervised fine-tuning with problem statement-program pairs. However, increasing model size and data volume for performance gains also raises computational demands and risks of overfitting. Addressing these challenges, we present RETROcode, a novel adaptation of the RETRO architecture \cite{RETRO} for sequence-to-sequence models, utilizing a large code database as an auxiliary scaling method. This approach, diverging from simply enlarging model and dataset sizes, allows RETROcode to leverage a vast code database for prediction, enhancing the model's efficiency by integrating extensive memory. Our findings indicate that RETROcode not only outperforms similar-sized traditional architectures on test sets but also approaches the effectiveness of the much larger Codex model, despite being trained from scratch on a substantially smaller dataset.
Authors: Seunghyun Ji, Soowon Lee
Abstract: Masked language modeling is a widely used method for learning language representations, where the model predicts a randomly masked word in each input. However, this approach typically considers only a single correct answer during training, ignoring the variety of plausible alternatives that humans might choose. This issue becomes more pronounced when the input text is short, as the possible word distribution tends to have higher entropy, potentially causing the model to become overconfident in its predictions. To mitigate this, we propose a novel confidence regularizer that adaptively adjusts the regularization strength based on the input length. Experiments on the GLUE and SQuAD benchmarks show that our method improves both accuracy and expected calibration error
Authors: Rijul Magu, Arka Dutta, Sean Kim, Ashiqur R. KhudaBukhsh, Munmun De Choudhury
Abstract: Large Language Models (LLMs) have been shown to demonstrate imbalanced biases against certain groups. However, the study of unprovoked targeted attacks by LLMs towards at-risk populations remains underexplored. Our paper presents three novel contributions: (1) the explicit evaluation of LLM-generated attacks on highly vulnerable mental health groups; (2) a network-based framework to study the propagation of relative biases; and (3) an assessment of the relative degree of stigmatization that emerges from these attacks. Our analysis of a recently released large-scale bias audit dataset reveals that mental health entities occupy central positions within attack narrative networks, as revealed by a significantly higher mean centrality of closeness (p-value = 4.06e-10) and dense clustering (Gini coefficient = 0.7). Drawing from sociological foundations of stigmatization theory, our stigmatization analysis indicates increased labeling components for mental health disorder-related targets relative to initial targets in generation chains. Taken together, these insights shed light on the structural predilections of large language models to heighten harmful discourse and highlight the need for suitable approaches for mitigation.
Authors: Usama Muneeb, Mesrob I. Ohannessian
Abstract: We consider scenarios where a very accurate (often small) predictive model using restricted features is available when training a full-featured (often larger) model. This restricted model may be thought of as side-information'', and can come either from an auxiliary dataset or from the same dataset by forcing the restriction. How can the restricted model be useful to the full model? To answer this, we introduce a methodology called Induced Model Matching (IMM). IMM aligns the context-restricted, or induced, version of the large model with the restricted model. We relate IMM to approaches such as noising, which is implicit in addressing the problem, and reverse knowledge distillation from weak teachers, which is explicit but does not exploit restriction being the nature of the weakness. We show that these prior methods can be thought of as approximations to IMM and can be problematic in terms of consistency. Experimentally, we first motivate IMM using logistic regression as a toy example. We then explore it in language modeling, the application that initially inspired it, and demonstrate it on both LSTM and transformer full models, using bigrams as restricted models. We lastly give a simple RL example, which shows that POMDP policies can help learn better MDP policies. The IMM principle is thus generally applicable in common scenarios where restricted data is cheaper to collect or restricted models are easier to learn.
Authors: Atsuyuki Miyai, Jingkang Yang, Jingyang Zhang, Yifei Ming, Qing Yu, Go Irie, Yixuan Li, Hai Li, Ziwei Liu, Kiyoharu Aizawa
Abstract: This paper introduces a novel task to evaluate the robust understanding capability of Large Multimodal Models (LMMs), termed $\textbf{Unsolvable Problem Detection (UPD)}$. Multiple-choice question answering (MCQA) is widely used to assess the understanding capability of LMMs, but it does not guarantee that LMMs truly comprehend the answer. UPD assesses the LMM's ability to withhold answers when encountering unsolvable problems of MCQA, verifying whether the model truly understands the answer. UPD encompasses three problems: Absent Answer Detection (AAD), Incompatible Answer Set Detection (IASD), and Incompatible Visual Question Detection (IVQD), covering unsolvable cases like answer-lacking or incompatible choices and image-question mismatches. For the evaluation, we introduce the MM-UPD Bench, a benchmark for assessing performance across various ability dimensions. Our experiments reveal that even most LMMs, which demonstrate adequate performance on existing benchmarks, struggle significantly with MM-UPD, underscoring a novel aspect of trustworthiness that current benchmarks have overlooked. A detailed analysis shows that LMMs have different bottlenecks and chain-of-thought and self-reflection improved performance for LMMs with the bottleneck in their LLM capability. We hope our insights will enhance the broader understanding and development of more reliable LMMs.
Authors: Xisen Jin, Xiang Ren
Abstract: Large Language models (LLMs) suffer from forgetting of upstream data when fine-tuned. Despite efforts on mitigating forgetting, few have investigated whether, and how forgotten upstream examples are dependent on newly learned tasks. Insights on such dependencies enable efficient and targeted mitigation of forgetting. In this paper, we empirically analyze forgetting that occurs in $N$ upstream examples of language modeling or instruction-tuning after fine-tuning LLMs on one of $M$ new tasks, visualized in $M\times N$ matrices. We show that the matrices are often well-approximated with low-rank matrices, indicating the dominance of simple associations between the learned tasks and forgotten upstream examples. Leveraging the analysis, we predict forgetting of upstream examples when fine-tuning on unseen tasks with matrix completion over the empirical associations. This enables fast identification of most forgotten examples without expensive inference on the entire upstream data. The approach, despite simplicity, outperforms prior approaches that learn semantic relationships of learned tasks and upstream examples with LMs for predicting forgetting. We demonstrate the practical utility of our analysis by showing statistically significantly reduced forgetting as we upweight predicted examples for replay at fine-tuning. Project page: https://inklab.usc.edu/lm-forgetting-prediction/
Authors: Silin Meng, Yiwei Wang, Cheng-Fu Yang, Nanyun Peng, Kai-Wei Chang
Abstract: Path planning is a fundamental scientific problem in robotics and autonomous navigation, requiring the derivation of efficient routes from starting to destination points while avoiding obstacles. Traditional algorithms like A* and its variants are capable of ensuring path validity but suffer from significant computational and memory inefficiencies as the state space grows. Conversely, large language models (LLMs) excel in broader environmental analysis through contextual understanding, providing global insights into environments. However, they fall short in detailed spatial and temporal reasoning, often leading to invalid or inefficient routes. In this work, we propose LLM-A*, an new LLM based route planning method that synergistically combines the precise pathfinding capabilities of A* with the global reasoning capability of LLMs. This hybrid approach aims to enhance pathfinding efficiency in terms of time and space complexity while maintaining the integrity of path validity, especially in large-scale scenarios. By integrating the strengths of both methodologies, LLM-A* addresses the computational and memory limitations of conventional algorithms without compromising on the validity required for effective pathfinding.
Authors: Weilin Cai, Juyong Jiang, Fan Wang, Jing Tang, Sunghun Kim, Jiayi Huang
Abstract: Large language models (LLMs) have garnered unprecedented advancements across diverse fields, ranging from natural language processing to computer vision and beyond. The prowess of LLMs is underpinned by their substantial model size, extensive and diverse datasets, and the vast computational power harnessed during training, all of which contribute to the emergent abilities of LLMs (e.g., in-context learning) that are not present in small models. Within this context, the mixture of experts (MoE) has emerged as an effective method for substantially scaling up model capacity with minimal computation overhead, gaining significant attention from academia and industry. Despite its growing prevalence, there lacks a systematic and comprehensive review of the literature on MoE. This survey seeks to bridge that gap, serving as an essential resource for researchers delving into the intricacies of MoE. We first briefly introduce the structure of the MoE layer, followed by proposing a new taxonomy of MoE. Next, we overview the core designs for various MoE models including both algorithmic and systemic aspects, alongside collections of available open-source implementations, hyperparameter configurations and empirical evaluations. Furthermore, we delineate the multifaceted applications of MoE in practice, and outline some potential directions for future research. To facilitate ongoing updates and the sharing of cutting-edge advances in MoE research, we have established a resource repository at https://github.com/withinmiaov/A-Survey-on-Mixture-of-Experts-in-LLMs.
URLs: https://github.com/withinmiaov/A-Survey-on-Mixture-of-Experts-in-LLMs.
Authors: Zifeng Wang, Benjamin Danek, Ziwei Yang, Zheng Chen, Jimeng Sun
Abstract: Data science plays a critical role in biomedical research, but it requires professionals with expertise in coding and medical data analysis. Large language models (LLMs) have shown great potential in supporting medical tasks and performing well in general coding tests. However, existing evaluations fail to assess their capability in biomedical data science, particularly in handling diverse data types such as genomics and clinical datasets. To address this gap, we developed a benchmark of data science coding tasks derived from the analyses of 39 published studies. This benchmark comprises 293 coding tasks (128 in Python and 165 in R) performed on real-world TCGA-type genomics and clinical data. Our findings reveal that the vanilla prompting of LLMs yields suboptimal performances due to drawbacks in following input instructions, understanding target data, and adhering to standard analysis practices. Next, we benchmarked six cutting-edge LLMs and advanced adaptation methods, finding two methods to be particularly effective: chain-of-thought prompting, which provides a step-by-step plan for data analysis, which led to a 21% code accuracy improvement (56.6% versus 35.3%); and self-reflection, enabling LLMs to refine the buggy code iteratively, yielding an 11% code accuracy improvement (45.5% versus 34.3%). Building on these insights, we developed a platform that integrates LLMs into the data science workflow for medical professionals. In a user study with five medical professionals, we found that while LLMs cannot fully automate programming tasks, they significantly streamline the programming process. We found that 80% of their submitted code solutions were incorporated from LLM-generated code, with up to 96% reuse in some cases. Our analysis highlights the potential of LLMs to enhance data science efficiency in biomedical research when integrated into expert workflows.
Authors: Yihua Shao, Yan Gu, Siyu Chen, Haiyang Liu, Zijian Ling, Minxi Yan, Ziyang Yan, Chenyu Zhang, Michele Magno, Haotong Qin, Yan Wang, Jingcai Guo, Ling Shao, Hao Tang
Abstract: Large language models (LLMs) show impressive performance in solving complex language tasks. However, its large number of parameters presents significant challenges for the deployment. So, compressing LLMs to low bits can enable to deploy on resource-constrained devices. To address this problem, we propose gradient-aware weight quantization (GWQ), the first quantization approach for low-bit weight quantization that leverages gradients to localize outliers, requiring only a minimal amount of calibration data for outlier detection. GWQ retains the top 1\% outliers preferentially at FP16 precision, while the remaining non-outlier weights are stored in a low-bit. We widely evaluate GWQ on different task include language modeling, grounding detection, massive multitask language understanding and vision-language question and answering. Results show that models quantified by GWQ performs better than other quantization method. During quantization process, GWQ only need one calibration set to realize effective quant. Also, GWQ achieves 1.2x inference speedup in comparison to the original model and effectively reduces the inference memory.
Authors: Pan Wang, Qiang Zhou, Yawen Wu, Tianlong Chen, Jingtong Hu
Abstract: Multimodal Sentiment Analysis (MSA) leverages heterogeneous modalities, such as language, vision, and audio, to enhance the understanding of human sentiment. While existing models often focus on extracting shared information across modalities or directly fusing heterogeneous modalities, such approaches can introduce redundancy and conflicts due to equal treatment of all modalities and the mutual transfer of information between modality pairs. To address these issues, we propose a Disentangled-Language-Focused (DLF) multimodal representation learning framework, which incorporates a feature disentanglement module to separate modality-shared and modality-specific information. To further reduce redundancy and enhance language-targeted features, four geometric measures are introduced to refine the disentanglement process. A Language-Focused Attractor (LFA) is further developed to strengthen language representation by leveraging complementary modality-specific information through a language-guided cross-attention mechanism. The framework also employs hierarchical predictions to improve overall accuracy. Extensive experiments on two popular MSA datasets, CMU-MOSI and CMU-MOSEI, demonstrate the significant performance gains achieved by the proposed DLF framework. Comprehensive ablation studies further validate the effectiveness of the feature disentanglement module, language-focused attractor, and hierarchical predictions. Our code is available at https://github.com/pwang322/DLF.
Authors: Luo Ji, Feixiang Guo, Teng Chen, Qingqing Gu, Xiaoyu Wang, Ningyuan Xi, Yihong Wang, Peng Yu, Yue Zhao, Hongyang Lei, Zhonglin Jiang, Yong Chen
Abstract: Despite the recent advancement in Retrieval-Augmented Generation (RAG) systems, most retrieval methodologies are often developed for factual retrieval, which assumes query and positive documents are semantically similar. In this paper, we instead propose and study a more challenging type of retrieval task, called hidden rationale retrieval, in which query and document are not similar but can be inferred by reasoning chains, logic relationships, or empirical experiences. To address such problems, an instruction-tuned Large language model (LLM) with a cross-encoder architecture could be a reasonable choice. To further strengthen pioneering LLM-based retrievers, we design a special instruction that transforms the retrieval task into a generative task by prompting LLM to answer a binary-choice question. The model can be fine-tuned with direct preference optimization (DPO). The framework is also optimized for computational efficiency with no performance degradation. We name this retrieval framework by RaHoRe and verify its zero-shot and fine-tuned performance superiority on Emotional Support Conversation (ESC), compared with previous retrieval works. Our study suggests the potential to employ LLM as a foundation for a wider scope of retrieval tasks. Our codes, models, and datasets are available on https://github.com/flyfree5/LaHoRe.
Authors: Yuhui Zhang, Yuchang Su, Yiming Liu, Xiaohan Wang, James Burgess, Elaine Sui, Chenyu Wang, Josiah Aklilu, Alejandro Lozano, Anjiang Wei, Ludwig Schmidt, Serena Yeung-Levy
Abstract: The rapid development of vision language models (VLMs) demands rigorous and reliable evaluation. However, current visual question answering (VQA) benchmarks often depend on open-ended questions, making accurate evaluation difficult due to the variability in natural language responses. To address this, we introduce AutoConverter, an agentic framework that automatically converts these open-ended questions into multiple-choice format, enabling objective evaluation while reducing the costly multiple-choice question creation process. Our experiments demonstrate that AutoConverter can generate correct and challenging multiple-choice questions, with VLMs demonstrating consistently similar or lower accuracy on these questions compared to human-created ones. Using AutoConverter, we construct VMCBench, a benchmark created by transforming 20 existing VQA datasets into a unified multiple-choice format, totaling 9,018 questions. We comprehensively evaluate 33 state-of-the-art VLMs on VMCBench, setting a new standard for scalable, consistent, and reproducible VLM evaluation.
Authors: Jiakang Yuan, Xiangchao Yan, Shiyang Feng, Bo Zhang, Tao Chen, Botian Shi, Wanli Ouyang, Yu Qiao, Lei Bai, Bowen Zhou
Abstract: The scientific research paradigm is undergoing a profound transformation owing to the development of Artificial Intelligence (AI). Recent works demonstrate that various AI-assisted research methods can largely improve research efficiency by improving data analysis, accelerating computation, and fostering novel idea generation. To further move towards the ultimate goal (i.e., automatic scientific research), in this paper, we introduce Dolphin, a closed-loop LLM-driven framework to enhance the automation level of scientific research. Dolphin first generates novel ideas based on feedback from previous experiments and relevant papers ranked by the topic and task attributes. Then, the generated ideas can be implemented using a code template refined and debugged with the designed exception-traceback-guided local code structure. Finally, Dolphin automatically analyzes the results of each idea and feeds the results back to the next round of idea generation. Experiments are conducted on the benchmark datasets of different topics and a subset of MLE-bench. Results show that Dolphin can continuously improve the performance of the input topic in a loop. We highlight that Dolphin can automatically propose methods that are comparable to the state-of-the-art in some tasks such as 3D point classification.
Authors: Qiguang Chen, Libo Qin, Jinhao Liu, Dengyun Peng, Jiannan Guan, Peng Wang, Mengkang Hu, Yuhang Zhou, Te Gao, Wanxiang Che
Abstract: Recent advancements in reasoning with large language models (RLLMs), such as OpenAI-O1 and DeepSeek-R1, have demonstrated their impressive capabilities in complex domains like mathematics and coding. A central factor in their success lies in the application of long chain-of-thought (Long CoT) characteristics, which enhance reasoning abilities and enable the solution of intricate problems. However, despite these developments, a comprehensive survey on Long CoT is still lacking, limiting our understanding of its distinctions from traditional short chain-of-thought (Short CoT) and complicating ongoing debates on issues like "overthinking" and "test-time scaling." This survey seeks to fill this gap by offering a unified perspective on Long CoT. (1) We first distinguish Long CoT from Short CoT and introduce a novel taxonomy to categorize current reasoning paradigms. (2) Next, we explore the key characteristics of Long CoT: deep reasoning, extensive exploration, and feasible reflection, which enable models to handle more complex tasks and produce more efficient, coherent outcomes compared to the shallower Short CoT. (3) We then investigate key phenomena such as the emergence of Long CoT with these characteristics, including overthinking, and test-time scaling, offering insights into how these processes manifest in practice. (4) Finally, we identify significant research gaps and highlight promising future directions, including the integration of multi-modal reasoning, efficiency improvements, and enhanced knowledge frameworks. By providing a structured overview, this survey aims to inspire future research and further the development of logical reasoning in artificial intelligence.
Authors: Rana Muhammad Shahroz Khan, Dongwen Tang, Pingzhi Li, Kai Wang, Tianlong Chen
Abstract: Parameter generation has emerged as a novel paradigm for neural network development, offering an alternative to traditional neural network training by synthesizing high-quality model weights directly. In the context of Low-Rank Adaptation (LoRA) for evolving ($\textit{i.e.}$, constantly updated) large language models (LLMs), this approach promises efficient adaptation without costly retraining. However, existing methods face critical limitations in simultaneously achieving scalability and controllability. In this paper, we introduce $\texttt{ORAL}$, a novel $\textbf{conditional recurrent diffusion}$ framework that addresses these challenges. $\texttt{ORAL}$ incorporates a novel conditioning mechanism that integrates model architecture and textual task specifications, enabling the generation of task-specific LoRA parameters that can seamlessly transfer across evolving foundation models. Our approach successfully scales to billions-of-parameter LLMs and maintains controllability. Through extensive experiments across seven language tasks, four vision tasks, and three multimodal tasks using five pre-trained LLMs, we demonstrate that $\texttt{ORAL}$ generates high-quality LoRA parameters that achieve comparable or superior performance to vanilla trained counterparts.
Authors: Wenjun Zeng, Dana Kurniawan, Ryan Mullins, Yuchi Liu, Tamoghna Saha, Dirichi Ike-Njoku, Jindong Gu, Yiwen Song, Cai Xu, Jingjing Zhou, Aparna Joshi, Shravan Dheep, Mani Malek, Hamid Palangi, Joon Baek, Rick Pereira, Karthik Narasimhan
Abstract: We introduce ShieldGemma 2, a 4B parameter image content moderation model built on Gemma 3. This model provides robust safety risk predictions across the following key harm categories: Sexually Explicit, Violence \& Gore, and Dangerous Content for synthetic images (e.g. output of any image generation model) and natural images (e.g. any image input to a Vision-Language Model). We evaluated on both internal and external benchmarks to demonstrate state-of-the-art performance compared to LlavaGuard \citep{helff2024llavaguard}, GPT-4o mini \citep{hurst2024gpt}, and the base Gemma 3 model \citep{gemma_2025} based on our policies. Additionally, we present a novel adversarial data generation pipeline which enables a controlled, diverse, and robust image generation. ShieldGemma 2 provides an open image moderation tool to advance multimodal safety and responsible AI development.
Authors: Marios Kokkodis, Richard Demsyn-Jones, Vijay Raghavan
Abstract: Are traditional classification approaches irrelevant in this era of AI hype? We show that there are multiclass classification problems where predictive models holistically outperform LLM prompt-based frameworks. Given text and images from home-service project descriptions provided by Thumbtack customers, we build embeddings-based softmax models that predict the professional category (e.g., handyman, bathroom remodeling) associated with each problem description. We then compare against prompts that ask state-of-the-art LLM models to solve the same problem. We find that the embeddings approach outperforms the best LLM prompts in terms of accuracy, calibration, latency, and financial cost. In particular, the embeddings approach has 49.5% higher accuracy than the prompting approach, and its superiority is consistent across text-only, image-only, and text-image problem descriptions. Furthermore, it yields well-calibrated probabilities, which we later use as confidence signals to provide contextualized user experience during deployment. On the contrary, prompting scores are overly uninformative. Finally, the embeddings approach is 14 and 81 times faster than prompting in processing images and text respectively, while under realistic deployment assumptions, it can be up to 10 times cheaper. Based on these results, we deployed a variation of the embeddings approach, and through A/B testing we observed performance consistent with our offline analysis. Our study shows that for multiclass classification problems that can leverage proprietary datasets, an embeddings-based approach may yield unequivocally better results. Hence, scientists, practitioners, engineers, and business leaders can use our study to go beyond the hype and consider appropriate predictive models for their classification use cases.
Authors: Gleb Rodionov, Roman Garipov, Alina Shutova, George Yakushev, Vage Egiazarian, Anton Sinitsin, Denis Kuznedelev, Dan Alistarh
Abstract: Large Language Models (LLMs) have demonstrated the ability to tackle increasingly complex tasks through advanced reasoning, long-form content generation, and tool use. Solving these tasks often involves long inference-time computations. In human problem solving, a common strategy to expedite work is collaboration: by dividing the problem into sub-tasks, exploring different strategies concurrently, etc. Recent research has shown that LLMs can also operate in parallel by implementing explicit cooperation frameworks, such as voting mechanisms or the explicit creation of independent sub-tasks that can be executed in parallel. However, each of these frameworks may not be suitable for all types of tasks, which can hinder their applicability. In this work, we propose a different design approach: we run LLM "workers" in parallel , allowing them to synchronize via a concurrently-updated attention cache and prompt these workers to decide how best to collaborate. Our approach allows the instances to come up with their own collaboration strategy for the problem at hand, all the while "seeing" each other's partial progress in the concurrent cache. We implement this approach via Hogwild! Inference: a parallel LLM inference engine where multiple instances of the same LLM run in parallel with the same attention cache, with "instant" access to each other's generated tokens. Hogwild! inference takes advantage of Rotary Position Embeddings (RoPE) to avoid recomputation while improving parallel hardware utilization. We find that modern reasoning-capable LLMs can perform inference with shared Key-Value cache out of the box, without additional fine-tuning.