Authors: Xudong Pan, Jiarun Dai, Yihe Fan, Min Yang
Abstract: Successful self-replication under no human assistance is the essential step for AI to outsmart the human beings, and is an early signal for rogue AIs. That is why self-replication is widely recognized as one of the few red line risks of frontier AI systems. Nowadays, the leading AI corporations OpenAI and Google evaluate their flagship large language models GPT-o1 and Gemini Pro 1.0, and report the lowest risk level of self-replication. However, following their methodology, we for the first time discover that two AI systems driven by Meta's Llama31-70B-Instruct and Alibaba's Qwen25-72B-Instruct, popular large language models of less parameters and weaker capabilities, have already surpassed the self-replicating red line. In 50% and 90% experimental trials, they succeed in creating a live and separate copy of itself respectively. By analyzing the behavioral traces, we observe the AI systems under evaluation already exhibit sufficient self-perception, situational awareness and problem-solving capabilities to accomplish self-replication. We further note the AI systems are even able to use the capability of self-replication to avoid shutdown and create a chain of replica to enhance the survivability, which may finally lead to an uncontrolled population of AIs. If such a worst-case risk is let unknown to the human society, we would eventually lose control over the frontier AI systems: They would take control over more computing devices, form an AI species and collude with each other against human beings. Our findings are a timely alert on existing yet previously unknown severe AI risks, calling for international collaboration on effective governance on uncontrolled self-replication of AI systems.
Authors: Naira Abdou Mohamed, Zakarya Erraji, Abdessalam Bahafid, Imade Benelallam
Abstract: If today some African languages like Swahili have enough resources to develop high-performing Natural Language Processing (NLP) systems, many other languages spoken on the continent are still lacking such support. For these languages, still in their infancy, several possibilities exist to address this critical lack of data. Among them is Transfer Learning, which allows low-resource languages to benefit from the good representation of other languages that are similar to them. In this work, we adopt a similar approach, aiming to pioneer NLP technologies for Comorian, a group of four languages or dialects belonging to the Bantu family. Our approach is initially motivated by the hypothesis that if a human can understand a different language from their native language with little or no effort, it would be entirely possible to model this process on a machine. To achieve this, we consider ways to construct Comorian datasets mixed with Swahili. One thing to note here is that in terms of Swahili data, we only focus on elements that are closest to Comorian by calculating lexical distances between candidate and source data. We empirically test this hypothesis in two use cases: Automatic Speech Recognition (ASR) and Machine Translation (MT). Our MT model achieved ROUGE-1, ROUGE-2, and ROUGE-L scores of 0.6826, 0.42, and 0.6532, respectively, while our ASR system recorded a WER of 39.50\% and a CER of 13.76\%. This research is crucial for advancing NLP in underrepresented languages, with potential to preserve and promote Comorian linguistic heritage in the digital age.
Authors: Chang-Jin Li, Jiyuan Zhang, Yun Tang, Jian Li
Abstract: Personality assessment, particularly through situational judgment tests (SJTs), is a vital tool for psychological research, talent selection, and educational evaluation. This study explores the potential of GPT-4, a state-of-the-art large language model (LLM), to automate the generation of personality situational judgment tests (PSJTs) in Chinese. Traditional SJT development is labor-intensive and prone to biases, while GPT-4 offers a scalable, efficient alternative. Two studies were conducted: Study 1 evaluated the impact of prompt design and temperature settings on content validity, finding that optimized prompts with a temperature of 1.0 produced creative and accurate items. Study 2 assessed the psychometric properties of GPT-4-generated PSJTs, revealing that they demonstrated satisfactory reliability and validity, surpassing the performance of manually developed tests in measuring the Big Five personality traits. This research highlights GPT-4's effectiveness in developing high-quality PSJTs, providing a scalable and innovative method for psychometric test development. These findings expand the possibilities of automatic item generation and the application of LLMs in psychology, and offer practical implications for streamlining test development processes in resource-limited settings.
Authors: Yu Yan, Sheng Sun, Junqi Tong, Min Liu, Qi Li
Abstract: Metaphor serves as an implicit approach to convey information, while enabling the generalized comprehension of complex subjects. However, metaphor can potentially be exploited to bypass the safety alignment mechanisms of Large Language Models (LLMs), leading to the theft of harmful knowledge. In our study, we introduce a novel attack framework that exploits the imaginative capacity of LLMs to achieve jailbreaking, the J\underline{\textbf{A}}ilbreak \underline{\textbf{V}}ia \underline{\textbf{A}}dversarial Me\underline{\textbf{TA}} -pho\underline{\textbf{R}} (\textit{AVATAR}). Specifically, to elicit the harmful response, AVATAR extracts harmful entities from a given harmful target and maps them to innocuous adversarial entities based on LLM's imagination. Then, according to these metaphors, the harmful target is nested within human-like interaction for jailbreaking adaptively. Experimental results demonstrate that AVATAR can effectively and transferablly jailbreak LLMs and achieve a state-of-the-art attack success rate across multiple advanced LLMs. Our study exposes a security risk in LLMs from their endogenous imaginative capabilities. Furthermore, the analytical study reveals the vulnerability of LLM to adversarial metaphors and the necessity of developing defense methods against jailbreaking caused by the adversarial metaphor. \textcolor{orange}{ \textbf{Warning: This paper contains potentially harmful content from LLMs.}}
Authors: Lingdong Shen, Qigqi, Kun Ding, Gaofeng Meng, Shiming Xiang
Abstract: Scientific Literature charts often contain complex visual elements, including multi-plot figures, flowcharts, structural diagrams and etc. Evaluating multimodal models using these authentic and intricate charts provides a more accurate assessment of their understanding abilities. However, existing benchmarks face limitations: a narrow range of chart types, overly simplistic template-based questions and visual elements, and inadequate evaluation methods. These shortcomings lead to inflated performance scores that fail to hold up when models encounter real-world scientific charts. To address these challenges, we introduce a new benchmark, Scientific Chart QA (SCI-CQA), which emphasizes flowcharts as a critical yet often overlooked category. To overcome the limitations of chart variety and simplistic visual elements, we curated a dataset of 202,760 image-text pairs from 15 top-tier computer science conferences papers over the past decade. After rigorous filtering, we refined this to 37,607 high-quality charts with contextual information. SCI-CQA also introduces a novel evaluation framework inspired by human exams, encompassing 5,629 carefully curated questions, both objective and open-ended. Additionally, we propose an efficient annotation pipeline that significantly reduces data annotation costs. Finally, we explore context-based chart understanding, highlighting the crucial role of contextual information in solving previously unanswerable questions.
Authors: Jiayu Liu, Zhenya Huang, Chaokun Wang, Xunpeng Huang, Chengxiang Zhai, Enhong Chen
Abstract: Owing to the capability of in-context learning, large language models (LLMs) have shown impressive performance across diverse mathematical reasoning benchmarks. However, we find that few-shot demonstrations can sometimes bring negative performance and their effectiveness on LLMs' reasoning abilities remains unreliable. To this end, in this paper, we aim to theoretically analyze the impact of in-context demonstrations on LLMs' reasoning performance. We prove that the reasoning efficacy (measured by empirical prediction loss) can be bounded by a LLM-oriented semantic similarity and an inference stability of demonstrations, which is general for both one-shot and few-shot scenarios. Based on this finding, we propose a straightforward, generalizable, and low-complexity demonstration selection method named LMS3. It can adaptively facilitate to select the most pertinent samples for different LLMs and includes a novel demonstration rejection mechanism to automatically filter out samples that are unsuitable for few-shot learning. Through experiments on three representative benchmarks, two LLM backbones, and multiple few-shot settings, we verify that our LMS3 has superiority and achieves consistent improvements on all datasets, which existing methods have been unable to accomplish.
Authors: Nikhil Mehta, Sithira Ambepitiya, Thanveer Ahamad, Dinuka Wijesundara, Yudara Kularathne
Abstract: Introduction: Global burden of sexually transmitted infections (STIs) is rising out of proportion to specialists. Current chatbots like ChatGPT are not tailored for handling STI-related concerns out of the box. We developed Otiz, an Artificial Intelligence-based (AI-based) chatbot platform designed specifically for STI detection and counseling, and assessed its performance. Methods: Otiz employs a multi-agent system architecture based on GPT4-0613, leveraging large language model (LLM) and Deterministic Finite Automaton principles to provide contextually relevant, medically accurate, and empathetic responses. Its components include modules for general STI information, emotional recognition, Acute Stress Disorder detection, and psychotherapy. A question suggestion agent operates in parallel. Four STIs (anogenital warts, herpes, syphilis, urethritis/cervicitis) and 2 non-STIs (candidiasis, penile cancer) were evaluated using prompts mimicking patient language. Each prompt was independently graded by two venereologists conversing with Otiz as patient actors on 6 criteria using Numerical Rating Scale ranging from 0 (poor) to 5 (excellent). Results: Twenty-three venereologists did 60 evaluations of 30 prompts. Across STIs, Otiz scored highly on diagnostic accuracy (4.1-4.7), overall accuracy (4.3-4.6), correctness of information (5.0), comprehensibility (4.2-4.4), and empathy (4.5-4.8). However, relevance scores were lower (2.9-3.6), suggesting some redundancy. Diagnostic scores for non-STIs were lower (p=0.038). Inter-observer agreement was strong, with differences greater than 1 point occurring in only 12.7% of paired evaluations. Conclusions: AI conversational agents like Otiz can provide accurate, correct, discrete, non-judgmental, readily accessible and easily understandable STI-related information in an empathetic manner, and can alleviate the burden on healthcare systems.
Authors: Evangelia Gkritzali, Panagiotis Kaliosis, Sofia Galanaki, Elisavet Palogiannidi, Theodoros Giannakopoulos
Abstract: In the vast majority of the academic and scientific domains, LaTeX has established itself as the de facto standard for typesetting complex mathematical equations and formulae. However, LaTeX's complex syntax and code-like appearance present accessibility barriers for individuals with disabilities, as well as those unfamiliar with coding conventions. In this paper, we present a novel solution to this challenge through the development of a novel speech-to-LaTeX equations system specifically designed for the Greek language. We propose an end-to-end system that harnesses the power of Automatic Speech Recognition (ASR) and Natural Language Processing (NLP) techniques to enable users to verbally dictate mathematical expressions and equations in natural language, which are subsequently converted into LaTeX format. We present the architecture and design principles of our system, highlighting key components such as the ASR engine, the LLM-based prompt-driven equations generation mechanism, as well as the application of a custom evaluation metric employed throughout the development process. We have made our system open source and available at https://github.com/magcil/greek-speech-to-math.
Authors: Soumita Roy
Abstract: This document explores the potential of employing Artificial Intelligence (AI), specifically Natural Language Processing (NLP), to strengthen the detection and prevention of financial crimes within the Mobile Financial Services(MFS) of Bangladesh with multilingual scenario. The analysis focuses on the utilization of NLP for adverse media screening, a vital aspect of compliance with anti-money laundering (AML) and combating financial terrorism (CFT) regulations. Additionally, it investigates the overall reception and obstacles related to the integration of AI in Bangladeshi banks. This report measures the effectiveness of NLP is promising with an accuracy around 94\%. NLP algorithms display substantial promise in accurately identifying adverse media content linked to financial crimes. The lack of progress in this aspect is visible in Bangladesh, whereas globally the technology is already being used to increase effectiveness and efficiency. Hence, it is clear there is an issue with the acceptance of AI in Bangladesh. Some AML \& CFT concerns are already being addressed by AI technology. For example, Image Recognition OCR technology are being used in KYC procedures. Primary hindrances to AI integration involve a lack of technical expertise, high expenses, and uncertainties surrounding regulations. This investigation underscores the potential of AI-driven NLP solutions in fortifying efforts to prevent financial crimes in Bangladesh.
Authors: Soham Sane, Angus McLean
Abstract: This paper presents a novel framework for enhancing reasoning capabilities in large language models (LLMs) by leveraging iterative reasoning and feedback-driven methodologies. Building on the limitations identified in the SimpleBench benchmark, a dataset designed to evaluate logical coherence and real-world reasoning, we propose a multi-step prompting strategy coupled with global consistency checks to improve model accuracy and robustness. Through comparative analysis of state-of-the-art models, including Claude 3 Opus, Claude 3.5, GPT- 4o, and o1-preview, we demonstrate that iterative reasoning significantly enhances model performance, with improvements observed in both standard accuracy metrics (AVG@5) and a newly introduced metric, Extreme Averaging (EAG@5). Our results reveal model-specific strengths: Claude excels in maintaining logical consistency, while GPT-4o exhibits exploratory creativity but struggles with ambiguous prompts. By analyzing case studies and identifying gaps in spatial and temporal reasoning, we highlight areas for further refinement. The findings underscore the potential of structured reasoning frameworks to address inherent model limitations, irrespective of pretraining methodologies. This study lays the groundwork for integrating dynamic feedback mechanisms, adaptive restart strategies, and diverse evaluation metrics to advance LLM reasoning capabilities across complex and multi-domain problem spaces.
Authors: Weitang Liu, Yuelei Li, Ying Wai Li, Zihan Wang, Jingbo Shang
Abstract: Comparing two (large) language models (LMs) side-by-side and pinpointing their prediction similarities and differences on the same set of inputs are crucial in many real-world scenarios, e.g., one can test if a licensed model was potentially plagiarized by another. Traditional analysis compares the LMs' outputs on some benchmark datasets, which only cover a limited number of inputs of designed perspectives for the intended applications. The benchmark datasets cannot prepare data to cover the test cases from unforeseen perspectives which can help us understand differences between models unbiasedly. In this paper, we propose a new model comparative analysis setting that considers a large input space where brute-force enumeration would be infeasible. The input space can be simply defined as all token sequences that a LM would produce low perplexity on -- we follow this definition in the paper as it would produce the most human-understandable inputs. We propose a novel framework \our that uses text generation by sampling and deweights the histogram of sampling statistics to estimate prediction differences between two LMs in this input space efficiently and unbiasedly. Our method achieves this by drawing and counting the inputs at each prediction difference value in negative log-likelihood. Experiments reveal for the first time the quantitative prediction differences between LMs in a large input space, potentially facilitating the model analysis for applications such as model plagiarism.
Authors: Seungwook Han, Jinyeop Song, Jeff Gore, Pulkit Agrawal
Abstract: Humans distill complex experiences into fundamental abstractions that enable rapid learning and adaptation. Similarly, autoregressive transformers exhibit adaptive learning through in-context learning (ICL), which begs the question of how. In this paper, we propose \textbf{concept encoding-decoding mechanism} to explain ICL by studying how transformers form and use internal abstractions in their representations. On synthetic ICL tasks, we analyze the training dynamics of a small transformer and report the coupled emergence of concept encoding and decoding. As the model learns to encode different latent concepts (e.g., ``Finding the first noun in a sentence.") into distinct, separable representations, it concureently builds conditional decoding algorithms and improve its ICL performance. We validate the existence of this mechanism across pretrained models of varying scales (Gemma-2 2B/9B/27B, Llama-3.1 8B/70B). Further, through mechanistic interventions and controlled finetuning, we demonstrate that the quality of concept encoding is causally related and predictive of ICL performance. Our empirical insights shed light into better understanding the success and failure modes of large language models via their representations.
Authors: Xiangyu Peng, Prafulla Kumar Choubey, Caiming Xiong, Chien-Sheng Wu
Abstract: Existing evaluation frameworks for retrieval-augmented generation (RAG) systems focus on answerable queries, but they overlook the importance of appropriately rejecting unanswerable requests. In this paper, we introduce UAEval4RAG, a framework designed to evaluate whether RAG systems can handle unanswerable queries effectively. We define a taxonomy with six unanswerable categories, and UAEval4RAG automatically synthesizes diverse and challenging queries for any given knowledge base with unanswered ratio and acceptable ratio metrics. We conduct experiments with various RAG components, including retrieval models, rewriting methods, rerankers, language models, and prompting strategies, and reveal hidden trade-offs in performance of RAG systems. Our findings highlight the critical role of component selection and prompt design in optimizing RAG systems to balance the accuracy of answerable queries with high rejection rates of unanswerable ones. UAEval4RAG provides valuable insights and tools for developing more robust and reliable RAG systems.
Authors: Jianqing Zhu, Huang Huang, Zhihang Lin, Juhao Liang, Zhengyang Tang, Khalid Almubarak, Abdulmohsen Alharthik, Bang An, Juncai He, Xiangbo Wu, Fei Yu, Junying Chen, Zhuoheng Ma, Yuhao Du, He Zhang, Emad A. Alghamdi, Lian Zhang, Ruoyu Sun, Haizhou Li, Benyou Wang, Jinchao Xu
Abstract: This paper addresses the critical need for democratizing large language models (LLM) in the Arab world, a region that has seen slower progress in developing models comparable to state-of-the-art offerings like GPT-4 or ChatGPT 3.5, due to a predominant focus on mainstream languages (e.g., English and Chinese). One practical objective for an Arabic LLM is to utilize an Arabic-specific vocabulary for the tokenizer that could speed up decoding. However, using a different vocabulary often leads to a degradation of learned knowledge since many words are initially out-of-vocabulary (OOV) when training starts. Inspired by the vocabulary learning during Second Language (Arabic) Acquisition for humans, the released AraLLaMA employs progressive vocabulary expansion, which is implemented by a modified BPE algorithm that progressively extends the Arabic subwords in its dynamic vocabulary during training, thereby balancing the OOV ratio at every stage. The ablation study demonstrated the effectiveness of Progressive Vocabulary Expansion. Moreover, AraLLaMA achieves decent performance comparable to the best Arabic LLMs across a variety of Arabic benchmarks. Models, training data, benchmarks, and codes will be all open-sourced.
Authors: Shuzhou Yuan, Jingyi Sun, Ran Zhang, Michael F\"arber, Steffen Eger, Pepa Atanasova, Isabelle Augenstein
Abstract: Natural language explanations (NLEs) are commonly used to provide plausible free-text explanations of a model's reasoning about its predictions. However, recent work has questioned the faithfulness of NLEs, as they may not accurately reflect the model's internal reasoning process regarding its predicted answer. In contrast, highlight explanations -- input fragments identified as critical for the model's predictions -- exhibit measurable faithfulness, which has been incrementally improved through existing research. Building on this foundation, we propose G-Tex, a Graph-Guided Textual Explanation Generation framework designed to enhance the faithfulness of NLEs by leveraging highlight explanations. Specifically, highlight explanations are extracted as highly faithful cues representing the model's reasoning and are subsequently encoded through a graph neural network layer, which explicitly guides the NLE generation process. This alignment ensures that the generated explanations closely reflect the model's underlying reasoning. Experiments on T5 and BART using three reasoning datasets show that G-Tex improves NLE faithfulness by up to 17.59% compared to baseline methods. Additionally, G-Tex generates NLEs with greater semantic and lexical similarity to human-written ones. Human evaluations show that G-Tex can decrease redundant content and enhance the overall quality of NLEs. As our work introduces a novel method for explicitly guiding NLE generation to improve faithfulness, we hope it will serve as a stepping stone for addressing additional criteria for NLE and generated text overall.
Authors: Samy Ateia, Udo Kruschwitz
Abstract: We present BioRAGent, an interactive web-based retrieval-augmented generation (RAG) system for biomedical question answering. The system uses large language models (LLMs) for query expansion, snippet extraction, and answer generation while maintaining transparency through citation links to the source documents and displaying generated queries for further editing. Building on our successful participation in the BioASQ 2024 challenge, we demonstrate how few-shot learning with LLMs can be effectively applied for a professional search setting. The system supports both direct short paragraph style responses and responses with inline citations. Our demo is available online, and the source code is publicly accessible through GitHub.
Authors: Giang (Dexter), Nguyen, Ivan Brugere, Shubham Sharma, Sanjay Kariyappa, Anh Totti Nguyen, Freddy Lecue
Abstract: Interpretability for Table Question Answering (Table QA) is critical, particularly in high-stakes industries like finance or healthcare. Although recent approaches using Large Language Models (LLMs) have significantly improved Table QA performance, their explanations for how the answers are generated are ambiguous. To fill this gap, we introduce Plan-of-SQLs ( or POS), an interpretable, effective, and efficient approach to Table QA that answers an input query solely with SQL executions. Through qualitative and quantitative evaluations with human and LLM judges, we show that POS is most preferred among explanation methods, helps human users understand model decision boundaries, and facilitates model success and error identification. Furthermore, when evaluated in standard benchmarks (TabFact, WikiTQ, and FetaQA), POS achieves competitive or superior accuracy compared to existing methods, while maintaining greater efficiency by requiring significantly fewer LLM calls and database queries.
Authors: Tuka Alhanai, Adam Kasumovic, Mohammad Ghassemi, Aven Zitzelberger, Jessica Lundin, Guillaume Chabot-Couture
Abstract: Large Language Models (LLMs) have shown remarkable performance across various tasks, yet significant disparities remain for non-English languages, and especially native African languages. This paper addresses these disparities by creating approximately 1 million human-translated words of new benchmark data in 8 low-resource African languages, covering a population of over 160 million speakers of: Amharic, Bambara, Igbo, Sepedi (Northern Sotho), Shona, Sesotho (Southern Sotho), Setswana, and Tsonga. Our benchmarks are translations of Winogrande and three sections of MMLU: college medicine, clinical knowledge, and virology. Using the translated benchmarks, we report previously unknown performance gaps between state-of-the-art (SOTA) LLMs in English and African languages. Finally, using results from over 400 fine-tuned models, we explore several methods to reduce the LLM performance gap, including high-quality dataset fine-tuning (using an LLM-as-an-Annotator), cross-lingual transfer, and cultural appropriateness adjustments. Key findings include average mono-lingual improvements of 5.6% with fine-tuning (with 5.4% average mono-lingual improvements when using high-quality data over low-quality data), 2.9% average gains from cross-lingual transfer, and a 3.0% out-of-the-box performance boost on culturally appropriate questions. The publicly available benchmarks, translations, and code from this study support further research and development aimed at creating more inclusive and effective language technologies.
Authors: Monica Munnangi, Akshay Swaminathan, Jason Alan Fries, Jenelle Jindal, Sanjana Narayanan, Ivan Lopez, Lucia Tu, Philip Chung, Jesutofunmi A. Omiye, Mehr Kashyap, Nigam Shah
Abstract: Verifying factual claims is critical for using large language models (LLMs) in healthcare. Recent work has proposed fact decomposition, which uses LLMs to rewrite source text into concise sentences conveying a single piece of information, as an approach for fine-grained fact verification. Clinical documentation poses unique challenges for fact decomposition due to dense terminology and diverse note types. To explore these challenges, we present FactEHR, a dataset consisting of full document fact decompositions for 2,168 clinical notes spanning four types from three hospital systems. Our evaluation, including review by clinicians, highlights significant variability in the quality of fact decomposition for four commonly used LLMs, with some LLMs generating 2.6x more facts per sentence than others. The results underscore the need for better LLM capabilities to support factual verification in clinical text. To facilitate future research in this direction, we plan to release our code at \url{https://github.com/som-shahlab/factehr}.
Authors: Chia-Hsuan Chang, Tien-Yuan Huang, Yi-Hang Tsai, Chia-Ming Chang, San-Yih Hwang
Abstract: Recent works in clustering-based topic models perform well in monolingual topic identification by introducing a pipeline to cluster the contextualized representations. However, the pipeline is suboptimal in identifying topics across languages due to the presence of language-dependent dimensions (LDDs) generated by multilingual language models. To address this issue, we introduce a novel, SVD-based dimension refinement component into the pipeline of the clustering-based topic model. This component effectively neutralizes the negative impact of LDDs, enabling the model to accurately identify topics across languages. Our experiments on three datasets demonstrate that the updated pipeline with the dimension refinement component generally outperforms other state-of-the-art cross-lingual topic models.
Authors: Zihao Lin, Zichao Wang, Yuanting Pan, Varun Manjunatha, Ryan Rossi, Angela Lau, Lifu Huang, Tong Sun
Abstract: Suggested questions (SQs) provide an effective initial interface for users to engage with their documents in AI-powered reading applications. In practical reading sessions, users have diverse backgrounds and reading goals, yet current SQ features typically ignore such user information, resulting in homogeneous or ineffective questions. We introduce a pipeline that generates personalized SQs by incorporating reader profiles (professions and reading goals) and demonstrate its utility in two ways: 1) as an improved SQ generation pipeline that produces higher quality and more diverse questions compared to current baselines, and 2) as a data generator to fine-tune extremely small models that perform competitively with much larger models on SQ generation. Our approach can not only serve as a drop-in replacement in current SQ systems to immediately improve their performance but also help develop on-device SQ models that can run locally to deliver fast and private SQ experience.
Authors: Chia-Hsuan Chang, Jui-Tse Tsai, Yi-Hang Tsai, San-Yih Hwang
Abstract: Topic modeling is widely used for uncovering thematic structures within text corpora, yet traditional models often struggle with specificity and coherence in domain-focused applications. Guided approaches, such as SeededLDA and CorEx, incorporate user-provided seed words to improve relevance but remain labor-intensive and static. Large language models (LLMs) offer potential for dynamic topic refinement and discovery, yet their application often incurs high API costs. To address these challenges, we propose the LLM-assisted Iterative Topic Augmentation framework (LITA), an LLM-assisted approach that integrates user-provided seeds with embedding-based clustering and iterative refinement. LITA identifies a small number of ambiguous documents and employs an LLM to reassign them to existing or new topics, minimizing API costs while enhancing topic quality. Experiments on two datasets across topic quality and clustering performance metrics demonstrate that LITA outperforms five baseline models, including LDA, SeededLDA, CorEx, BERTopic, and PromptTopic. Our work offers an efficient and adaptable framework for advancing topic modeling and text clustering.
Authors: Yaofo Chen, Zeng You, Shuhai Zhang, Haokun Li, Yirui Li, Yaowei Wang, Mingkui Tan
Abstract: Transformer-based Large Language Models (LLMs) have exhibited remarkable success in various natural language processing tasks primarily attributed to self-attention mechanism, which requires a token to consider all preceding tokens as its context to compute the attention score. However, when the context length L becomes very large (e.g., 32K), more redundant context information will be included w.r.t. any tokens, making the self-attention suffer from two main limitations: 1) The computational and memory complexity scales quadratically w.r.t. L; 2) The presence of redundant context information may hamper the model to capture dependencies among crucial tokens, which may degrade the representation performance. In this paper, we propose a plug-and-play Core Context Aware (CCA) Attention for efficient long-range context modeling, which consists of two components: 1) Globality-pooling attention that divides input tokens into groups and then dynamically merges tokens within each group into one core token based on their significance; 2) Locality-preserved attention that incorporates neighboring tokens into the attention calculation. The two complementary attentions will then be fused to the final attention, maintaining comprehensive modeling ability as the full self-attention. In this way, the core context information w.r.t. a given token will be automatically focused and strengthened, while the context information in redundant groups will be diminished during the learning process. As a result, the computational and memory complexity will be significantly reduced. More importantly, the CCA-Attention can improve the long-context modeling ability by diminishing the redundant context information. Extensive experimental results demonstrate that our CCA-Attention significantly outperforms state-of-the-art models in terms of computational efficiency and long-context modeling ability.
Authors: Moxin Li, Yong Zhao, Yang Deng, Wenxuan Zhang, Shuaiyi Li, Wenya Xie, See-Kiong Ng, Tat-Seng Chua
Abstract: Although large language models (LLMs) store vast amount of knowledge in their parameters, they still have limitations in the memorization and utilization of certain knowledge, leading to undesired behaviors such as generating untruthful and inaccurate responses. This highlights the critical need to understand the knowledge boundary of LLMs, a concept that remains inadequately defined in existing research. In this survey, we propose a comprehensive definition of the LLM knowledge boundary and introduce a formalized taxonomy categorizing knowledge into four distinct types. Using this foundation, we systematically review the field through three key lenses: the motivation for studying LLM knowledge boundaries, methods for identifying these boundaries, and strategies for mitigating the challenges they present. Finally, we discuss open challenges and potential research directions in this area. We aim for this survey to offer the community a comprehensive overview, facilitate access to key issues, and inspire further advancements in LLM knowledge research.
Authors: Xuanzhong Chen, Ye Jin, Xiaohao Mao, Lun Wang, Shuyang Zhang, Ting Chen
Abstract: Rare diseases, despite their low individual incidence, collectively impact around 300 million people worldwide due to the huge number of diseases. The complexity of symptoms and the shortage of specialized doctors with relevant experience make diagnosing and treating rare diseases more challenging than common diseases. Recently, agents powered by large language models (LLMs) have demonstrated notable improvements across various domains. In the medical field, some agent methods have outperformed direct prompts in question-answering tasks from medical exams. However, current agent frameworks lack adaptation for real-world clinical scenarios, especially those involving the intricate demands of rare diseases. To address these challenges, we present RareAgents, the first multi-disciplinary team of LLM-based agents tailored to the complex clinical context of rare diseases. RareAgents integrates advanced planning capabilities, memory mechanisms, and medical tools utilization, leveraging Llama-3.1-8B/70B as the base model. Experimental results show that RareAgents surpasses state-of-the-art domain-specific models, GPT-4o, and existing agent frameworks in both differential diagnosis and medication recommendation for rare diseases. Furthermore, we contribute a novel dataset, MIMIC-IV-Ext-Rare, derived from MIMIC-IV, to support further advancements in this field.
Authors: Xi Cao, Yuan Sun, Jiajun Li, Quzong Gesang, Nuo Qun, Tashi Nyima
Abstract: DNN-based language models perform excellently on various tasks, but even SOTA LLMs are susceptible to textual adversarial attacks. Adversarial texts play crucial roles in multiple subfields of NLP. However, current research has the following issues. (1) Most textual adversarial attack methods target rich-resourced languages. How do we generate adversarial texts for less-studied languages? (2) Most textual adversarial attack methods are prone to generating invalid or ambiguous adversarial texts. How do we construct high-quality adversarial robustness benchmarks? (3) New language models may be immune to part of previously generated adversarial texts. How do we update adversarial robustness benchmarks? To address the above issues, we introduce HITL-GAT, a system based on a general approach to human-in-the-loop generation of adversarial texts. HITL-GAT contains four stages in one pipeline: victim model construction, adversarial example generation, high-quality benchmark construction, and adversarial robustness evaluation. Additionally, we utilize HITL-GAT to make a case study on Tibetan script which can be a reference for the adversarial research of other less-studied languages.
Authors: Hongjin Qian, Zheng Liu, Peitian Zhang, Zhicheng Dou, Defu Lian
Abstract: Processing long contexts poses a significant challenge for large language models (LLMs) due to their inherent context-window limitations and the computational burden of extensive key-value (KV) activations, which severely impact efficiency. For information-seeking tasks, full context perception is often unnecessary, as a query's information needs can dynamically range from localized details to a global perspective, depending on its complexity. However, existing methods struggle to adapt effectively to these dynamic information needs. In the paper, we propose a method for processing long-context information-seeking tasks via query-guided Activation Refilling (ACRE). ACRE constructs a Bi-layer KV Cache for long contexts, where the layer-1 (L1) cache compactly captures global information, and the layer-2 (L2) cache provides detailed and localized information. ACRE establishes a proxying relationship between the two caches, allowing the input query to attend to the L1 cache and dynamically refill it with relevant entries from the L2 cache. This mechanism integrates global understanding with query-specific local details, thus improving answer decoding. Experiments on a variety of long-context information-seeking datasets demonstrate ACRE's effectiveness, achieving improvements in both performance and efficiency.
Authors: Xin Yi, Shunfan Zheng, Linlin Wang, Gerard de Melo, Xiaoling Wang, Liang He
Abstract: The emergence of finetuning-as-a-service has revealed a new vulnerability in large language models (LLMs). A mere handful of malicious data uploaded by users can subtly manipulate the finetuning process, resulting in an alignment-broken model. Existing methods to counteract fine-tuning attacks typically require substantial computational resources. Even with parameter-efficient techniques like LoRA, gradient updates remain essential. To address these challenges, we propose \textbf{N}euron-\textbf{L}evel \textbf{S}afety \textbf{R}ealignment (\textbf{NLSR}), a training-free framework that restores the safety of LLMs based on the similarity difference of safety-critical neurons before and after fine-tuning. The core of our framework is first to construct a safety reference model from an initially aligned model to amplify safety-related features in neurons. We then utilize this reference model to identify safety-critical neurons, which we prepare as patches. Finally, we selectively restore only those neurons that exhibit significant similarity differences by transplanting these prepared patches, thereby minimally altering the fine-tuned model. Extensive experiments demonstrate significant safety enhancements in fine-tuned models across multiple downstream tasks, while greatly maintaining task-level accuracy. Our findings suggest regions of some safety-critical neurons show noticeable differences after fine-tuning, which can be effectively corrected by transplanting neurons from the reference model without requiring additional training. The code will be available at \url{https://github.com/xinykou/NLSR}
Authors: Hongbin Zhang, Kehai Chen, Xuefeng Bai, Yang Xiang, Min Zhang
Abstract: Large language models (LLMs) have demonstrated impressive multilingual understanding and reasoning capabilities, driven by extensive pre-training multilingual corpora and fine-tuning instruction data. However, a performance gap persists between high-resource and low-resource language tasks due to language imbalance in the pre-training corpus, even using more low-resource data during fine-tuning. To alleviate this issue, we propose LinguaLIFT, a two-stage instruction tuning framework for advancing low-resource language tasks. An additional language alignment layer is first integrated into the LLM to adapt a pre-trained multilingual encoder, thereby enhancing multilingual alignment through code-switched fine-tuning. The second stage fine-tunes LLM with English-only instruction data while freezing the language alignment layer, allowing LLM to transfer task-specific capabilities from English to low-resource language tasks. Additionally, we introduce the Multilingual Math World Problem (MMWP) benchmark, which spans 21 low-resource, 17 medium-resource, and 10 high-resource languages, enabling comprehensive evaluation of multilingual reasoning. Experimental results show that LinguaLIFT outperforms several competitive baselines across MMWP and other widely used benchmarks.
Authors: Sina Bagheri Nezhad, Ameeta Agrawal, Rhitabrat Pokharel
Abstract: Multilingual language models (MLLMs) are crucial for handling text across various languages, yet they often show performance disparities due to differences in resource availability and linguistic characteristics. While the impact of pre-train data percentage and model size on performance is well-known, our study reveals additional critical factors that significantly influence MLLM effectiveness. Analyzing a wide range of features, including geographical, linguistic, and resource-related aspects, we focus on the SIB-200 dataset for classification and the Flores-200 dataset for machine translation, using regression models and SHAP values across 204 languages. Our findings identify token similarity and country similarity as pivotal factors, alongside pre-train data and model size, in enhancing model performance. Token similarity facilitates cross-lingual transfer, while country similarity highlights the importance of shared cultural and linguistic contexts. These insights offer valuable guidance for developing more equitable and effective multilingual language models, particularly for underrepresented languages.
Authors: Mingxu Chai, Ziyu Shen, Chong Zhang, Yue Zhang, Xiao Wang, Shihan Dou, Jihua Kang, Jiazheng Zhang, Qi Zhang
Abstract: Document parsing is essential for analyzing complex document structures and extracting fine-grained information, supporting numerous downstream applications. However, existing methods often require integrating multiple independent models to handle various parsing tasks, leading to high complexity and maintenance overhead. To address this, we propose DocFusion, a lightweight generative model with only 0.28B parameters. It unifies task representations and achieves collaborative training through an improved objective function. Experiments reveal and leverage the mutually beneficial interaction among recognition tasks, and integrating recognition data significantly enhances detection performance. The final results demonstrate that DocFusion achieves state-of-the-art (SOTA) performance across four key tasks.
Authors: Kayla Schroeder, Zach Wood-Doughty
Abstract: Large Language Models (LLMs) have become increasingly powerful and ubiquitous, but their stochastic nature poses challenges to the reliability of their outputs. While deterministic settings can improve consistency, they do not guarantee reliability, as a single sample from the model's probability distribution can still be misleading. Building upon the concept of LLM-as-a-judge, we introduce a novel framework for rigorously evaluating the reliability of LLM judgments, leveraging McDonald's omega. We evaluate the reliability of LLMs when judging the outputs of other LLMs on standard single-turn and multi-turn benchmarks, simultaneously investigating the impact of temperature on reliability. By analyzing these results, we demonstrate the limitations of fixed randomness and the importance of considering multiple samples, which we show has significant implications for downstream applications. Our findings highlight the need for a nuanced understanding of LLM reliability and the potential risks associated with over-reliance on single-shot evaluations. This work provides a crucial step towards building more trustworthy and reliable LLM-based systems and applications.
Authors: Bohan Li, Jiannan Guan, Longxu Dou, Yunlong Feng, Dingzirui Wang, Yang Xu, Enbo Wang, Qiguang Chen, Bichen Wang, Xiao Xu, Yimeng Zhang, Libo Qin, Yanyan Zhao, Qingfu Zhu, Wanxiang Che
Abstract: The Myers-Briggs Type Indicator (MBTI) is one of the most influential personality theories reflecting individual differences in thinking, feeling, and behaving. MBTI personality detection has garnered considerable research interest and has evolved significantly over the years. However, this task tends to be overly optimistic, as it currently does not align well with the natural distribution of population personality traits. Specifically, (1) the self-reported labels in existing datasets result in incorrect labeling issues, and (2) the hard labels fail to capture the full range of population personality distributions. In this paper, we optimize the task by constructing MBTIBench, the first manually annotated high-quality MBTI personality detection dataset with soft labels, under the guidance of psychologists. As for the first challenge, MBTIBench effectively solves the incorrect labeling issues, which account for 29.58% of the data. As for the second challenge, we estimate soft labels by deriving the polarity tendency of samples. The obtained soft labels confirm that there are more people with non-extreme personality traits. Experimental results not only highlight the polarized predictions and biases in LLMs as key directions for future research, but also confirm that soft labels can provide more benefits to other psychological tasks than hard labels. The code and data are available at https://github.com/Personality-NLP/MbtiBench.
Authors: Geling Liu, Yunzhi Tan, Ruichao Zhong, Yuanzhen Xie, Lingchen Zhao, Qian Wang, Bo Hu, Zang Li
Abstract: Recently, large language models (LLMs) have significantly improved the performance of text-to-SQL systems. Nevertheless, many state-of-the-art (SOTA) approaches have overlooked the critical aspect of system robustness. Our experiments reveal that while LLM-driven methods excel on standard datasets, their accuracy is notably compromised when faced with adversarial perturbations. To address this challenge, we propose a robust text-to-SQL solution, called Solid-SQL, designed to integrate with various LLMs. We focus on the pre-processing stage, training a robust schema-linking model enhanced by LLM-based data augmentation. Additionally, we design a two-round, structural similarity-based example retrieval strategy for in-context learning. Our method achieves SOTA SQL execution accuracy levels of 82.1% and 58.9% on the general Spider and Bird benchmarks, respectively. Furthermore, experimental results show that Solid-SQL delivers an average improvement of 11.6% compared to baselines on the perturbed Spider-Syn, Spider-Realistic, and Dr. Spider benchmarks.
Authors: Hyuhng Joon Kim, Youna Kim, Sang-goo Lee, Taeuk Kim
Abstract: Large Language Models (LLMs) demonstrate exceptional performance across diverse tasks by leveraging both pre-trained knowledge (i.e., parametric knowledge) and external knowledge (i.e., contextual knowledge). While substantial efforts have been made to leverage both forms of knowledge, scenarios in which the model lacks any relevant knowledge remain underexplored. Such limitations can result in issues like hallucination, causing reduced reliability and potential risks in high-stakes applications. To address such limitations, this paper extends the task scope to encompass cases where the user's request cannot be fulfilled due to the lack of relevant knowledge. To this end, we introduce Contrastive Decoding with Abstention (CDA), a training-free decoding method that empowers LLMs to generate responses when relevant knowledge is available and to abstain otherwise. CDA evaluates the relevance of each knowledge for a given query, adaptively determining which knowledge to prioritize or which to completely ignore. Extensive experiments with four LLMs on three question-answering datasets demonstrate that CDA can effectively perform accurate generation and abstention simultaneously. These findings highlight CDA's potential to broaden the applicability of LLMs, enhancing reliability and preserving user trust.
Authors: Tao Fang, Derek F. Wong, Lusheng Zhang, Keyan Jin, Qiang Zhang, Tianjiao Li, Jinlong Hou, Lidia S. Chao
Abstract: While large-scale language models (LLMs) have demonstrated remarkable capabilities in specific natural language processing (NLP) tasks, they may still lack proficiency compared to specialized models in certain domains, such as grammatical error correction (GEC). Drawing inspiration from the concept of curriculum learning, we have delved into refining LLMs into proficient GEC experts by devising effective curriculum learning (CL) strategies. In this paper, we introduce a novel approach, termed LLM-based curriculum learning, which capitalizes on the robust semantic comprehension and discriminative prowess inherent in LLMs to gauge the complexity of GEC training data. Unlike traditional curriculum learning techniques, our method closely mirrors human expert-designed curriculums. Leveraging the proposed LLM-based CL method, we sequentially select varying levels of curriculums ranging from easy to hard, and iteratively train and refine using the pretrianed T5 and LLaMA series models. Through rigorous testing and analysis across diverse benchmark assessments in English GEC, including the CoNLL14 test, BEA19 test, and BEA19 development sets, our approach showcases a significant performance boost over baseline models and conventional curriculum learning methodologies.
Authors: Taeho Hwang, Sukmin Cho, Soyeong Jeong, Hoyun Song, SeungYoon Han, Jong C. Park
Abstract: We introduce EXIT, an extractive context compression framework that enhances both the effectiveness and efficiency of retrieval-augmented generation (RAG) in question answering (QA). Current RAG systems often struggle when retrieval models fail to rank the most relevant documents, leading to the inclusion of more context at the expense of latency and accuracy. While abstractive compression methods can drastically reduce token counts, their token-by-token generation process significantly increases end-to-end latency. Conversely, existing extractive methods reduce latency but rely on independent, non-adaptive sentence selection, failing to fully utilize contextual information. EXIT addresses these limitations by classifying sentences from retrieved documents - while preserving their contextual dependencies - enabling parallelizable, context-aware extraction that adapts to query complexity and retrieval quality. Our evaluations on both single-hop and multi-hop QA tasks show that EXIT consistently surpasses existing compression methods and even uncompressed baselines in QA accuracy, while also delivering substantial reductions in inference time and token count. By improving both effectiveness and efficiency, EXIT provides a promising direction for developing scalable, high-quality QA solutions in RAG pipelines. Our code is available at https://github.com/ThisIsHwang/EXIT
Authors: Vaden Masrani, Mohammad Akbari, David Ming Xuan Yue, Ahmad Rezaei, Yong Zhang
Abstract: In the era of costly pre-training of large language models, ensuring the intellectual property rights of model owners, and insuring that said models are responsibly deployed, is becoming increasingly important. To this end, we propose model watermarking via passthrough layers, which are added to existing pre-trained networks and trained using a self-supervised loss such that the model produces high-entropy output when prompted with a unique private key, and acts normally otherwise. Unlike existing model watermarking methods, our method is fully task-agnostic, and can be applied to both classification and sequence-to-sequence tasks without requiring advanced access to downstream fine-tuning datasets. We evaluate the proposed passthrough layers on a wide range of downstream tasks, and show experimentally our watermarking method achieves a near-perfect watermark extraction accuracy and false-positive rate in most cases without damaging original model performance. Additionally, we show our method is robust to both downstream fine-tuning, fine-pruning, and layer removal attacks, and can be trained in a fraction of the time required to train the original model. Code is available in the paper.
Authors: Chengyan Wu, Bolei Ma, Zheyu Zhang, Ningyuan Deng, Yanqing He, Yun Xue
Abstract: Aspect-based sentiment analysis (ABSA), a sequence labeling task, has attracted increasing attention in multilingual contexts. While previous research has focused largely on fine-tuning or training models specifically for ABSA, we evaluate large language models (LLMs) under zero-shot conditions to explore their potential to tackle this challenge with minimal task-specific adaptation. We conduct a comprehensive empirical evaluation of a series of LLMs on multilingual ABSA tasks, investigating various prompting strategies, including vanilla zero-shot, chain-of-thought (CoT), self-improvement, self-debate, and self-consistency, across nine different models. Results indicate that while LLMs show promise in handling multilingual ABSA, they generally fall short of fine-tuned, task-specific models. Notably, simpler zero-shot prompts often outperform more complex strategies, especially in high-resource languages like English. These findings underscore the need for further refinement of LLM-based approaches to effectively address ABSA task across diverse languages.
Authors: Seunghee Kim, Changhyeon Kim, Taeuk Kim
Abstract: Real-world decision-making often requires integrating and reasoning over information from multiple modalities. While recent multimodal large language models (MLLMs) have shown promise in such tasks, their ability to perform multi-hop reasoning across diverse sources remains insufficiently evaluated. Existing benchmarks, such as MMQA, face challenges due to (1) data contamination and (2) a lack of complex queries that necessitate operations across more than two modalities, hindering accurate performance assessment. To address this, we present Financial Cross-Modal Multi-Hop Reasoning (FCMR), a benchmark created to analyze the reasoning capabilities of MLLMs by urging them to combine information from textual reports, tables, and charts within the financial domain. FCMR is categorized into three difficulty levels-Easy, Medium, and Hard-facilitating a step-by-step evaluation. In particular, problems at the Hard level require precise cross-modal three-hop reasoning and are designed to prevent the disregard of any modality. Experiments on this new benchmark reveal that even state-of-the-art MLLMs struggle, with the best-performing model (Claude 3.5 Sonnet) achieving only 30.4% accuracy on the most challenging tier. We also conduct analysis to provide insights into the inner workings of the models, including the discovery of a critical bottleneck in the information retrieval phase.
Authors: Ryo Kishino, Hiroaki Yamagiwa, Ryo Nagata, Sho Yokoi, Hidetoshi Shimodaira
Abstract: Lexical semantic change detection aims to identify shifts in word meanings over time. While existing methods using embeddings from a diachronic corpus pair estimate the degree of change for target words, they offer limited insight into changes at the level of individual usage instances. To address this, we apply Unbalanced Optimal Transport (UOT) to sets of contextualized word embeddings, capturing semantic change through the excess and deficit in the alignment between usage instances. In particular, we propose Sense Usage Shift (SUS), a measure that quantifies changes in the usage frequency of a word sense at each usage instance. By leveraging SUS, we demonstrate that several challenges in semantic change detection can be addressed in a unified manner, including quantifying instance-level semantic change and word-level tasks such as measuring the magnitude of semantic change and the broadening or narrowing of meaning.
Authors: Hanyin Wang, Qiping Xu, Bolun Liu, Guleid Hussein, Hariprasad Korsapati, Mohamad El Labban, Kingsley Iheasirim, Mohamed Hassan, Gokhan Anil, Brian Bartlett, Jimeng Sun
Abstract: Process-supervised reward models (PRMs), which verify large language model (LLM) outputs step-by-step, have achieved significant success in mathematical and coding problems. However, their application to other domains remains largely unexplored. In this work, we train a PRM to provide step-level reward signals for clinical notes generated by LLMs from patient-doctor dialogues. Guided by real-world clinician expertise, we carefully designed step definitions for clinical notes and utilized Gemini-Pro 1.5 to automatically generate process supervision data at scale. Our proposed PRM, trained on the LLaMA-3.1 8B instruct model, demonstrated superior performance compared to Gemini-Pro 1.5 and an outcome-supervised reward model (ORM) across two key evaluations: (1) the accuracy of selecting gold-reference samples from error-containing samples, achieving 98.8% (versus 61.3% for ORM and 93.8% for Gemini-Pro 1.5), and (2) the accuracy of selecting physician-preferred notes, achieving 56.2% (compared to 51.2% for ORM and 50.0% for Gemini-Pro 1.5). Additionally, we conducted ablation studies to determine optimal loss functions and data selection strategies, along with physician reader studies to explore predictors of downstream Best-of-N performance. Our promising results suggest the potential of PRMs to extend beyond the clinical domain, offering a scalable and effective solution for diverse generative tasks.
Authors: Yun Luo, Yingjie Li, Xiangkun Hu, Qinglin Qi, Fang Guo, Qipeng Guo, Zheng Zhang, Yue Zhang
Abstract: As online platforms and recommendation algorithms evolve, people are increasingly trapped in echo chambers, leading to biased understandings of various issues. To combat this issue, we have introduced PerSphere, a benchmark designed to facilitate multi-faceted perspective retrieval and summarization, thus breaking free from these information silos. For each query within PerSphere, there are two opposing claims, each supported by distinct, non-overlapping perspectives drawn from one or more documents. Our goal is to accurately summarize these documents, aligning the summaries with the respective claims and their underlying perspectives. This task is structured as a two-step end-to-end pipeline that includes comprehensive document retrieval and multi-faceted summarization. Furthermore, we propose a set of metrics to evaluate the comprehensiveness of the retrieval and summarization content. Experimental results on various counterparts for the pipeline show that recent models struggle with such a complex task. Analysis shows that the main challenge lies in long context and perspective extraction, and we propose a simple but effective multi-agent summarization system, offering a promising solution to enhance performance on PerSphere.
Authors: Chongyang Tao, Tao Shen, Shen Gao, Junshuo Zhang, Zhen Li, Zhengwei Tao, Shuai Ma
Abstract: Large language models (LLMs) have revolutionized natural language processing by achieving state-of-the-art performance across various tasks. Recently, their effectiveness as embedding models has gained attention, marking a paradigm shift from traditional encoder-only models like ELMo and BERT to decoder-only, large-scale LLMs such as GPT, LLaMA, and Mistral. This survey provides an in-depth overview of this transition, beginning with foundational techniques before the LLM era, followed by LLM-based embedding models through two main strategies to derive embeddings from LLMs. 1) Direct prompting: We mainly discuss the prompt designs and the underlying rationale for deriving competitive embeddings. 2) Data-centric tuning: We cover extensive aspects that affect tuning an embedding model, including model architecture, training objectives, data constructions, etc. Upon the above, we also cover advanced methods, such as handling longer texts, and multilingual and cross-modal data. Furthermore, we discuss factors affecting choices of embedding models, such as performance/efficiency comparisons, dense vs sparse embeddings, pooling strategies, and scaling law. Lastly, the survey highlights the limitations and challenges in adapting LLMs for embeddings, including cross-task embedding quality, trade-offs between efficiency and accuracy, low-resource, long-context, data bias, robustness, etc. This survey serves as a valuable resource for researchers and practitioners by synthesizing current advancements, highlighting key challenges, and offering a comprehensive framework for future work aimed at enhancing the effectiveness and efficiency of LLMs as embedding models.
Authors: Nianqi Li, Zujie Liang, Siyu Yuan, Jiaqing Liang, Feng Wei, Yanghua Xiao
Abstract: Program-of-Thought (PoT), which aims to use programming language instead of natural language as an intermediate step in reasoning, is an important way for LLMs to solve mathematical problems. Since different programming languages excel in different areas, it is natural to use the most suitable language for solving specific problems. However, current PoT research only focuses on single language PoT, ignoring the differences between different programming languages. Therefore, this paper proposes an multilingual program reasoning method, MultiLingPoT. This method allows the model to answer questions using multiple programming languages by fine-tuning on multilingual data. Additionally, prior and posterior hybrid methods are used to help the model select the most suitable language for each problem. Our experimental results show that the training of MultiLingPoT improves each program's mathematical reasoning by about 2.5\%. Moreover, with proper mixing, the performance of MultiLingPoT can be further improved, achieving a 6\% increase compared to the single-language PoT with the data augmentation.Resources of this paper can be found at https://github.com/Nianqi-Li/MultiLingPoT.
Authors: Aman Tiwari, Shiva Krishna Reddy Malay, Vikas Yadav, Masoud Hashemi, Sathwik Tejaswi Madhusudhan
Abstract: Cypher, the query language for Neo4j graph databases, plays a critical role in enabling graph-based analytics and data exploration. While substantial research has been dedicated to natural language to SQL query generation (Text2SQL), the analogous problem for graph databases referred to as Text2Cypher remains underexplored. In this work, we introduce SynthCypher, a fully synthetic and automated data generation pipeline designed to address this gap. SynthCypher employs a novel LLMSupervised Generation-Verification framework, ensuring syntactically and semantically correct Cypher queries across diverse domains and query complexities. Using this pipeline, we create SynthCypher Dataset, a large-scale benchmark containing 29.8k Text2Cypher instances. Fine-tuning open-source large language models (LLMs), including LLaMa-3.1- 8B, Mistral-7B, and QWEN-7B, on SynthCypher yields significant performance improvements of up to 40% on the Text2Cypher test set and 30% on the SPIDER benchmark adapted for graph databases. This work demonstrates that high-quality synthetic data can effectively advance the state-of-the-art in Text2Cypher tasks.
Authors: Weixiong Zheng, Peijian Zeng, Yiwei Li, Hongyan Wu, Nankai Lin, Junhao Chen, Aimin Yang, Yongmei Zhou
Abstract: Large language models (LLMs) excel in various tasks but remain vulnerable to jailbreak attacks, where adversaries manipulate prompts to generate harmful outputs. Examining jailbreak prompts helps uncover the shortcomings of LLMs. However, current jailbreak methods and the target model's defenses are engaged in an independent and adversarial process, resulting in the need for frequent attack iterations and redesigning attacks for different models. To address these gaps, we propose a Reverse Embedded Defense Attack (REDA) mechanism that disguises the attack intention as the "defense". intention against harmful content. Specifically, REDA starts from the target response, guiding the model to embed harmful content within its defensive measures, thereby relegating harmful content to a secondary role and making the model believe it is performing a defensive task. The attacking model considers that it is guiding the target model to deal with harmful content, while the target model thinks it is performing a defensive task, creating an illusion of cooperation between the two. Additionally, to enhance the model's confidence and guidance in "defensive" intentions, we adopt in-context learning (ICL) with a small number of attack examples and construct a corresponding dataset of attack examples. Extensive evaluations demonstrate that the REDA method enables cross-model attacks without the need to redesign attack strategies for different models, enables successful jailbreak in one iteration, and outperforms existing methods on both open-source and closed-source models.
Authors: Andong Chen, Yuchen Song, Kehai Chen, Muyun Yang, Tiejun Zhao, Min Zhang
Abstract: Visual information has been introduced for enhancing machine translation (MT), and its effectiveness heavily relies on the availability of large amounts of bilingual parallel sentence pairs with manual image annotations. In this paper, we introduce a stable diffusion-based imagination network into a multimodal large language model (MLLM) to explicitly generate an image for each source sentence, thereby advancing the multimodel MT. Particularly, we build heuristic human feedback with reinforcement learning to ensure the consistency of the generated image with the source sentence without the supervision of image annotation, which breaks the bottleneck of using visual information in MT. Furthermore, the proposed method enables imaginative visual information to be integrated into large-scale text-only MT in addition to multimodal MT. Experimental results show that our model significantly outperforms existing multimodal MT and text-only MT, especially achieving an average improvement of more than 14 BLEU points on Multi30K multimodal MT benchmarks.
Authors: Zhiyuan Chang, Mingyang Li, Xiaojun Jia, Junjie Wang, Yuekai Huang, Qing Wang, Yihao Huang, Yang Liu
Abstract: Incorporating external knowledge into large language models (LLMs) has emerged as a promising approach to mitigate outdated knowledge and hallucination in LLMs. However, external knowledge is often imperfect. In addition to useful knowledge, external knowledge is rich in irrelevant or misinformation in the context that can impair the reliability of LLM responses. This paper focuses on LLMs' preferred external knowledge in imperfect contexts when handling multi-hop QA. Inspired by criminal procedural law's Chain of Evidence (CoE), we characterize that knowledge preferred by LLMs should maintain both relevance to the question and mutual support among knowledge pieces. Accordingly, we propose an automated CoE discrimination approach and explore LLMs' preferences from their effectiveness, faithfulness and robustness, as well as CoE's usability in a naive Retrieval-Augmented Generation (RAG) case. The evaluation on five LLMs reveals that CoE enhances LLMs through more accurate generation, stronger answer faithfulness, better robustness against knowledge conflict, and improved performance in a popular RAG case.
Authors: Xiangxiang Gao, Weisheng Xie, Yiwei Xiang, Feng Ji
Abstract: Striking an optimal balance between minimal drafting latency and high speculation accuracy to enhance the inference speed of Large Language Models remains a significant challenge in speculative decoding. In this paper, we introduce Falcon, an innovative semi-autoregressive speculative decoding framework fashioned to augment both the drafter's parallelism and output quality. Falcon incorporates the Coupled Sequential Glancing Distillation technique, which fortifies inter-token dependencies within the same block, leading to increased speculation accuracy. We offer a comprehensive theoretical analysis to illuminate the underlying mechanisms. Additionally, we introduce a Custom-Designed Decoding Tree, which permits the drafter to generate multiple tokens in a single forward pass and accommodates multiple forward passes as needed, thereby boosting the number of drafted tokens and significantly improving the overall acceptance rate. Comprehensive evaluations on benchmark datasets such as MT-Bench, HumanEval, and GSM8K demonstrate Falcon's superior acceleration capabilities. The framework achieves a lossless speedup ratio ranging from 2.91x to 3.51x when tested on the Vicuna and LLaMA2-Chat model series. These results outstrip existing speculative decoding methods for LLMs, including Eagle, Medusa, Lookahead, SPS, and PLD, while maintaining a compact drafter architecture equivalent to merely two Transformer layers.
Authors: Mufan Xu, Kehai Chen, Xuefeng Bai, Muyun Yang, Tiejun Zhao, Min Zhang
Abstract: Large language models (LLMs) based on generative pre-trained Transformer have achieved remarkable performance on knowledge graph question-answering (KGQA) tasks. However, LLMs often produce ungrounded subgraph planning or reasoning results in KGQA due to the hallucinatory behavior brought by the generative paradigm, which may hinder the advancement of the LLM-based KGQA model. To deal with the issue, we propose a novel LLM-based Discriminative Reasoning (LDR) method to explicitly model the subgraph retrieval and answer inference process. By adopting discriminative strategies, the proposed LDR method not only enhances the capability of LLMs to retrieve question-related subgraphs but also alleviates the issue of ungrounded reasoning brought by the generative paradigm of LLMs. Experimental results show that the proposed approach outperforms multiple strong comparison methods, along with achieving state-of-the-art performance on two widely used WebQSP and CWQ benchmarks.
Authors: Jiahui Li, Roman Klinger
Abstract: Prompt engineering has made significant contributions to the era of large language models, yet its effectiveness depends on the skills of a prompt author. Automatic prompt optimization can support the prompt development process, but requires annotated data. This paper introduces $\textit{iPrOp}$, a novel Interactive Prompt Optimization system, to bridge manual prompt engineering and automatic prompt optimization. With human intervention in the optimization loop, $\textit{iPrOp}$ offers users the flexibility to assess evolving prompts. We present users with prompt variations, selected instances, large language model predictions accompanied by corresponding explanations, and performance metrics derived from a subset of the training data. This approach empowers users to choose and further refine the provided prompts based on their individual preferences and needs. This system not only assists non-technical domain experts in generating optimal prompts tailored to their specific tasks or domains, but also enables to study the intrinsic parameters that influence the performance of prompt optimization. Our evaluation shows that our system has the capability to generate improved prompts, leading to enhanced task performance.
Authors: Robert Aufschl\"ager, Sebastian Wilhelm, Michael Heigl, Martin Schramm
Abstract: This work introduces ClustEm4Ano, an anonymization pipeline that can be used for generalization and suppression-based anonymization of nominal textual tabular data. It automatically generates value generalization hierarchies (VGHs) that, in turn, can be used to generalize attributes in quasi-identifiers. The pipeline leverages embeddings to generate semantically close value generalizations through iterative clustering. We applied KMeans and Hierarchical Agglomerative Clustering on $13$ different predefined text embeddings (both open and closed-source (via APIs)). Our approach is experimentally tested on a well-known benchmark dataset for anonymization: The UCI Machine Learning Repository's Adult dataset. ClustEm4Ano supports anonymization procedures by offering more possibilities compared to using arbitrarily chosen VGHs. Experiments demonstrate that these VGHs can outperform manually constructed ones in terms of downstream efficacy (especially for small $k$-anonymity ($2 \leq k \leq 30$)) and therefore can foster the quality of anonymized datasets. Our implementation is made public.
Authors: Jenny Kunz
Abstract: Smaller LLMs still face significant challenges even in medium-resourced languages, particularly when it comes to language-specific knowledge -- a problem not easily resolved with machine-translated data. In this case study on Icelandic, we aim to enhance the generation performance of an LLM by specialising it using unstructured text corpora. A key focus is on preventing interference with the models' capabilities of handling longer context during this adaptation. Through ablation studies using various parameter-efficient fine-tuning (PEFT) methods and setups, we find that increasing the number of trainable parameters leads to better and more robust language adaptation. LoRAs placed in the feed-forward layers and bottleneck adapters show promising results with sufficient parameters, while prefix tuning and (IA)3 are not suitable. Although improvements are consistent in 0-shot summarisation, some adapted models struggle with longer context lengths, an issue that can be mitigated by adapting only the final layers.
Authors: Yupei Li, Manuel Milling, Lucia Specia, Bj\"orn W. Schuller
Abstract: The availability of high-quality APIs for Large Language Models (LLMs) has facilitated the widespread creation of Machine-Generated Content (MGC), posing challenges such as academic plagiarism and the spread of misinformation. Existing MGC detectors often focus solely on surface-level information, overlooking implicit and structural features. This makes them susceptible to deception by surface-level sentence patterns, particularly for longer texts and in texts that have been subsequently paraphrased. To overcome these challenges, we introduce novel methodologies and datasets. Besides the publicly available dataset Plagbench, we developed the paraphrased Long-Form Question and Answer (paraLFQA) and paraphrased Writing Prompts (paraWP) datasets using GPT and DIPPER, a discourse paraphrasing tool, by extending artifacts from their original versions. To address the challenge of detecting highly similar paraphrased texts, we propose MhBART, an encoder-decoder model designed to emulate human writing style while incorporating a novel difference score mechanism. This model outperforms strong classifier baselines and identifies deceptive sentence patterns. To better capture the structure of longer texts at document level, we propose DTransformer, a model that integrates discourse analysis through PDTB preprocessing to encode structural features. It results in substantial performance gains across both datasets -- 15.5\% absolute improvement on paraLFQA, 4\% absolute improvement on paraWP, and 1.5\% absolute improvement on M4 compared to SOTA approaches.
Authors: Yangfan Ye, Xiaocheng Feng, Xiachong Feng, Libo Qin, Yichong Huang, Lei Huang, Weitao Ma, Zhirui Zhang, Yunfei Lu, Xiaohui Yan, Duyu Tang, Dandan Tu, Bing Qin
Abstract: Current large language models (LLMs) often exhibit imbalances in multilingual capabilities and cultural adaptability, largely due to their English-centric pretraining data. To address this imbalance, we propose a probing method named XTransplant that explores cross-lingual latent interactions via cross-lingual feed-forward transplantation during inference stage, with the hope of enabling the model to leverage the strengths of both English and non-English languages. Through extensive pilot experiments, we empirically prove that both the multilingual capabilities and cultural adaptability of LLMs hold the potential to be significantly improved by XTransplant, respectively from En -> non-En and non-En -> En, highlighting the underutilization of current LLMs' multilingual potential. And the patterns observed in these pilot experiments further motivate an offline scaling inference strategy, which demonstrates consistent performance improvements in multilingual and culture-aware tasks, sometimes even surpassing multilingual supervised fine-tuning. And we do hope our further analysis and discussion could help gain deeper insights into XTransplant mechanism.
Authors: Kepu Zhang, Zhongxiang Sun, Xiao Zhang, Xiaoxue Zang, Kai Zheng, Yang Song, Jun Xu
Abstract: In search scenarios, user experience can be hindered by erroneous queries due to typos, voice errors, or knowledge gaps. Therefore, query correction is crucial for search engines. Current correction models, usually small models trained on specific data, often struggle with queries beyond their training scope or those requiring contextual understanding. While the advent of Large Language Models (LLMs) offers a potential solution, they are still limited by their pre-training data and inference cost, particularly for complex queries, making them not always effective for query correction. To tackle these, we propose Trigger$^3$, a large-small model collaboration framework that integrates the traditional correction model and LLM for query correction, capable of adaptively choosing the appropriate correction method based on the query and the correction results from the traditional correction model and LLM. Trigger$^3$ first employs a correction trigger to filter out correct queries. Incorrect queries are then corrected by the traditional correction model. If this fails, an LLM trigger is activated to call the LLM for correction. Finally, for queries that no model can correct, a fallback trigger decides to return the original query. Extensive experiments demonstrate Trigger$^3$ outperforms correction baselines while maintaining efficiency.
Authors: Jiebin Zhang, Dawei Zhu, Yifan Song, Wenhao Wu, Chuqiao Kuang, Xiaoguang Li, Lifeng Shang, Qun Liu, Sujian Li
Abstract: As large language models (LLMs) process increasing context windows, the memory usage of KV cache has become a critical bottleneck during inference. The mainstream KV compression methods, including KV pruning and KV quantization, primarily focus on either token or precision dimension and seldom explore the efficiency of their combination. In this paper, we comprehensively investigate the token-precision trade-off in KV cache compression. Experiments demonstrate that storing more tokens in the KV cache with lower precision, i.e., quantized pruning, can significantly enhance the long-context performance of LLMs. Furthermore, in-depth analysis regarding token-precision trade-off from a series of key aspects exhibit that, quantized pruning achieves substantial improvements in retrieval-related tasks and consistently performs well across varying input lengths. Moreover, quantized pruning demonstrates notable stability across different KV pruning methods, quantization strategies, and model scales. These findings provide valuable insights into the token-precision trade-off in KV cache compression. We plan to release our code in the near future.
Authors: Syed Zohaib Hassan, Pierre Lison, P{\aa}l Halvorsen
Abstract: Disfluencies are a natural feature of spontaneous human speech but are typically absent from the outputs of Large Language Models (LLMs). This absence can diminish the perceived naturalness of synthesized speech, which is an important criteria when building conversational agents that aim to mimick human behaviours. We show how the insertion of disfluencies can alleviate this shortcoming. The proposed approach involves (1) fine-tuning an LLM with Low-Rank Adaptation (LoRA) to incorporate various types of disfluencies into LLM-generated utterances and (2) synthesizing those utterances using a text-to-speech model that supports the generation of speech phenomena such as disfluencies. We evaluated the quality of the generated speech across two metrics: intelligibility and perceived spontaneity. We demonstrate through a user study that the insertion of disfluencies significantly increase the perceived spontaneity of the generated speech. This increase came, however, along with a slight reduction in intelligibility.
Authors: Kshitij Dave, Nouhaila Innan, Bikash K. Behera, Zahid Mumtaz, Saif Al-Kuwari, Ahmed Farouk
Abstract: Sentiment analysis is an essential component of natural language processing, used to analyze sentiments, attitudes, and emotional tones in various contexts. It provides valuable insights into public opinion, customer feedback, and user experiences. Researchers have developed various classical machine learning and neuro-fuzzy approaches to address the exponential growth of data and the complexity of language structures in sentiment analysis. However, these approaches often fail to determine the optimal number of clusters, interpret results accurately, handle noise or outliers efficiently, and scale effectively to high-dimensional data. Additionally, they are frequently insensitive to input variations. In this paper, we propose a novel hybrid approach for sentiment analysis called the Quantum Fuzzy Neural Network (QFNN), which leverages quantum properties and incorporates a fuzzy layer to overcome the limitations of classical sentiment analysis algorithms. In this study, we test the proposed approach on two Twitter datasets: the Coronavirus Tweets Dataset (CVTD) and the General Sentimental Tweets Dataset (GSTD), and compare it with classical and hybrid algorithms. The results demonstrate that QFNN outperforms all classical, quantum, and hybrid algorithms, achieving 100% and 90% accuracy in the case of CVTD and GSTD, respectively. Furthermore, QFNN demonstrates its robustness against six different noise models, providing the potential to tackle the computational complexity associated with sentiment analysis on a large scale in a noisy environment. The proposed approach expedites sentiment data processing and precisely analyses different forms of textual data, thereby enhancing sentiment classification and insights associated with sentiment analysis.
Authors: Alon Eirew, Eviatar Nachshoni, Aviv Slobodkin, Ido Dagan
Abstract: Event relation detection is a fundamental NLP task, leveraged in many downstream applications, whose modeling requires datasets annotated with event relations of various types. However, systematic and complete annotation of these relations is costly and challenging, due to the quadratic number of event pairs that need to be considered. Consequently, many current event relation datasets lack systematicity and completeness. In response, we introduce \textit{EventFull}, the first tool that supports consistent, complete and efficient annotation of temporal, causal and coreference relations via a unified and synergetic process. A pilot study demonstrates that EventFull accelerates and simplifies the annotation process while yielding high inter-annotator agreement.
Authors: Nan Li, Bo Kang, Tijl De Bie
Abstract: Text classification with hierarchical labels is a prevalent and challenging task in natural language processing. Examples include assigning ICD codes to patient records, tagging patents into IPC classes, assigning EUROVOC descriptors to European legal texts, and more. Despite its widespread applications, a comprehensive understanding of state-of-the-art methods across different domains has been lacking. In this paper, we provide the first comprehensive cross-domain overview with empirical analysis of state-of-the-art methods. We propose a unified framework that positions each method within a common structure to facilitate research. Our empirical analysis yields key insights and guidelines, confirming the necessity of learning across different research areas to design effective methods. Notably, under our unified evaluation pipeline, we achieved new state-of-the-art results by applying techniques beyond their original domains.
Authors: Debajyoti Mazumder, Aakash Kumar, Jasabanta Patro
Abstract: In this paper, we reported our experiments with various strategies to improve code-mixed humour and sarcasm detection. We did all of our experiments for Hindi-English code-mixed scenario, as we have the linguistic expertise for the same. We experimented with three approaches, namely (i) native sample mixing, (ii) multi-task learning (MTL), and (iii) prompting very large multilingual language models (VMLMs). In native sample mixing, we added monolingual task samples in code-mixed training sets. In MTL learning, we relied on native and code-mixed samples of a semantically related task (hate detection in our case). Finally, in our third approach, we evaluated the efficacy of VMLMs via few-shot context prompting. Some interesting findings we got are (i) adding native samples improved humor (raising the F1-score up to 6.76%) and sarcasm (raising the F1-score up to 8.64%) detection, (ii) training MLMs in an MTL framework boosted performance for both humour (raising the F1-score up to 10.67%) and sarcasm (increment up to 12.35% in F1-score) detection, and (iii) prompting VMLMs couldn't outperform the other approaches. Finally, our ablation studies and error analysis discovered the cases where our model is yet to improve. We provided our code for reproducibility.
Authors: Cristiano Chesi
Abstract: A significant debate has emerged in response to a paper written by Steven Piantadosi (Piantadosi, 2023) and uploaded to the LingBuzz platform, the open archive for generative linguistics. Piantadosi's dismissal of Chomsky's approach is ruthless, but generative linguists deserve it. In this paper, I will adopt three idealized perspectives -- computational, theoretical, and experimental -- to focus on two fundamental issues that lend partial support to Piantadosi's critique: (a) the evidence challenging the Poverty of Stimulus (PoS) hypothesis and (b) the notion of simplicity as conceived within mainstream Minimalism. In conclusion, I argue that, to reclaim a central role in language studies, generative linguistics -- representing a prototypical theoretical perspective on language -- needs a serious update leading to (i) more precise, consistent, and complete formalizations of foundational intuitions and (ii) the establishment and utilization of a standardized dataset of crucial empirical evidence to evaluate the theory's adequacy. On the other hand, ignoring the formal perspective leads to major drawbacks in both computational and experimental approaches. Neither descriptive nor explanatory adequacy can be easily achieved without the precise formulation of general principles that can be challenged empirically.
Authors: Robert Litschko, Oliver Kraus, Verena Blaschke, Barbara Plank
Abstract: A large amount of local and culture-specific knowledge (e.g., people, traditions, food) can only be found in documents written in dialects. While there has been extensive research conducted on cross-lingual information retrieval (CLIR), the field of cross-dialect retrieval (CDIR) has received limited attention. Dialect retrieval poses unique challenges due to the limited availability of resources to train retrieval models and the high variability in non-standardized languages. We study these challenges on the example of German dialects and introduce the first German dialect retrieval dataset, dubbed WikiDIR, which consists of seven German dialects extracted from Wikipedia. Using WikiDIR, we demonstrate the weakness of lexical methods in dealing with high lexical variation in dialects. We further show that commonly used zero-shot cross-lingual transfer approach with multilingual encoders do not transfer well to extremely low-resource setups, motivating the need for resource-lean and dialect-specific retrieval models. We finally demonstrate that (document) translation is an effective way to reduce the dialect gap in CDIR.
Authors: Ziqi Qiu, Jianxing Yu, Yufeng Zhang, Hanjiang Lai, Yanghui Rao, Qinliang Su, Jian Yin
Abstract: This paper focuses on sarcasm detection, which aims to identify whether given statements convey criticism, mockery, or other negative sentiment opposite to the literal meaning. To detect sarcasm, humans often require a comprehensive understanding of the semantics in the statement and even resort to external commonsense to infer the fine-grained incongruity. However, existing methods lack commonsense inferential ability when they face complex real-world scenarios, leading to unsatisfactory performance. To address this problem, we propose a novel framework for sarcasm detection, which conducts incongruity reasoning based on commonsense augmentation, called EICR. Concretely, we first employ retrieval-augmented large language models to supplement the missing but indispensable commonsense background knowledge. To capture complex contextual associations, we construct a dependency graph and obtain the optimized topology via graph refinement. We further introduce an adaptive reasoning skeleton that integrates prior rules to extract sentiment-inconsistent subgraphs explicitly. To eliminate the possible spurious relations between words and labels, we employ adversarial contrastive learning to enhance the robustness of the detector. Experiments conducted on five datasets demonstrate the effectiveness of EICR.
Authors: Jinxiang Xie, Yilin Li, Xunjian Yin, Xiaojun Wan
Abstract: Evaluating the performance of Grammatical Error Correction (GEC) models has become increasingly challenging, as large language model (LLM)-based GEC systems often produce corrections that diverge from provided gold references. This discrepancy undermines the reliability of traditional reference-based evaluation metrics. In this study, we propose a novel evaluation framework for GEC models, DSGram, integrating Semantic Coherence, Edit Level, and Fluency, and utilizing a dynamic weighting mechanism. Our framework employs the Analytic Hierarchy Process (AHP) in conjunction with large language models to ascertain the relative importance of various evaluation criteria. Additionally, we develop a dataset incorporating human annotations and LLM-simulated sentences to validate our algorithms and fine-tune more cost-effective models. Experimental results indicate that our proposed approach enhances the effectiveness of GEC model evaluations.
Authors: Ruikang Ni, Da Xiao, Qingye Meng, Xiangyu Li, Shihui Zheng, Hongliang Liang
Abstract: Compositional relational reasoning (CRR) is a hallmark of human intelligence, but we lack a clear understanding of whether and how existing transformer large language models (LLMs) can solve CRR tasks. To enable systematic exploration of the CRR capability of LLMs, we first propose a new synthetic benchmark called Generalized Associative Recall (GAR) by integrating and generalizing the essence of several tasks in mechanistic interpretability (MI) study in a unified framework. Evaluation shows that GAR is challenging enough for existing LLMs, revealing their fundamental deficiency in CRR. Meanwhile, it is easy enough for systematic MI study. Then, to understand how LLMs solve GAR tasks, we use attribution patching to discover the core circuits reused by Vicuna-33B across different tasks and a set of vital attention heads. Intervention experiments show that the correct functioning of these heads significantly impacts task performance. Especially, we identify two classes of heads whose activations represent the abstract notion of true and false in GAR tasks respectively. They play a fundamental role in CRR across various models and tasks. The dataset and code are available at https://github.com/Caiyun-AI/GAR.
Authors: Ziheng Qiao, Houquan Zhou, Yumeng Liu, Zhenghua Li, Min Zhang, Bo Zhang, Chen Li, Ji Zhang, Fei Huang
Abstract: One key characteristic of the Chinese spelling check (CSC) task is that incorrect characters are usually similar to the correct ones in either phonetics or glyph. To accommodate this, previous works usually leverage confusion sets, which suffer from two problems, i.e., difficulty in determining which character pairs to include and lack of probabilities to distinguish items in the set. In this paper, we propose a light-weight plug-and-play DISC (i.e., decoding intervention with similarity of characters) module for CSC models.DISC measures phonetic and glyph similarities between characters and incorporates this similarity information only during the inference phase. This method can be easily integrated into various existing CSC models, such as ReaLiSe, SCOPE, and ReLM, without additional training costs. Experiments on three CSC benchmarks demonstrate that our proposed method significantly improves model performance, approaching and even surpassing the current state-of-the-art models.
Authors: Yuchen Fan, Yuzhong Hong, Qiushi Wang, Junwei Bao, Hongfei Jiang, Yang Song
Abstract: Alignment, endowing a pre-trained Large language model (LLM) with the ability to follow instructions, is crucial for its real-world applications. Conventional supervised fine-tuning (SFT) methods formalize it as causal language modeling typically with a cross-entropy objective, requiring a large amount of high-quality instruction-response pairs. However, the quality of widely used SFT datasets can not be guaranteed due to the high cost and intensive labor for the creation and maintenance in practice. To overcome the limitations associated with the quality of SFT datasets, we introduce a novel \textbf{p}reference-\textbf{o}riented supervised \textbf{f}ine-\textbf{t}uning approach, namely PoFT. The intuition is to boost SFT by imposing a particular preference: \textit{favoring the target model over aligned LLMs on the same SFT data.} This preference encourages the target model to predict a higher likelihood than that predicted by the aligned LLMs, incorporating assessment information on data quality (i.e., predicted likelihood by the aligned LLMs) into the training process. Extensive experiments are conducted, and the results validate the effectiveness of the proposed method. PoFT achieves stable and consistent improvements over the SFT baselines across different training datasets and base models. Moreover, we prove that PoFT can be integrated with existing SFT data filtering methods to achieve better performance, and further improved by following preference optimization procedures, such as DPO.
Authors: Jinhao Jiang, Jiayi Chen, Junyi Li, Ruiyang Ren, Shijie Wang, Wayne Xin Zhao, Yang Song, Tao Zhang
Abstract: Existing large language models (LLMs) show exceptional problem-solving capabilities but might struggle with complex reasoning tasks. Despite the successes of chain-of-thought and tree-based search methods, they mainly depend on the internal knowledge of LLMs to search over intermediate reasoning steps, limited to dealing with simple tasks involving fewer reasoning steps. In this paper, we propose \textbf{RAG-Star}, a novel RAG approach that integrates the retrieved information to guide the tree-based deliberative reasoning process that relies on the inherent knowledge of LLMs. By leveraging Monte Carlo Tree Search, RAG-Star iteratively plans intermediate sub-queries and answers for reasoning based on the LLM itself. To consolidate internal and external knowledge, we propose an retrieval-augmented verification that utilizes query- and answer-aware reward modeling to provide feedback for the inherent reasoning of LLMs. Our experiments involving Llama-3.1-8B-Instruct and GPT-4o demonstrate that RAG-Star significantly outperforms previous RAG and reasoning methods.
Authors: Kevin Fischer, Darren F\"urst, Sebastian Steindl, Jakob Lindner, Ulrich Sch\"afer
Abstract: Large Language Models (LLMs) have been showing promising results for various NLP-tasks without the explicit need to be trained for these tasks by using few-shot or zero-shot prompting techniques. A common NLP-task is question-answering (QA). In this study, we propose a comprehensive performance comparison between smaller fine-tuned models and out-of-the-box instruction-following LLMs on the Stanford Question Answering Dataset 2.0 (SQuAD2), specifically when using a single-inference prompting technique. Since the dataset contains unanswerable questions, previous work used a double inference method. We propose a prompting style which aims to elicit the same ability without the need for double inference, saving compute time and resources. Furthermore, we investigate their generalization capabilities by comparing their performance on similar but different QA datasets, without fine-tuning neither model, emulating real-world uses where the context and questions asked may differ from the original training distribution, for example swapping Wikipedia for news articles. Our results show that smaller, fine-tuned models outperform current State-Of-The-Art (SOTA) LLMs on the fine-tuned task, but recent SOTA models are able to close this gap on the out-of-distribution test and even outperform the fine-tuned models on 3 of the 5 tested QA datasets.
Authors: Ildik\'o Pil\'an, Benet Manzanares-Salor, David S\'anchez, Pierre Lison
Abstract: The purpose of text sanitization is to rewrite those text spans in a document that may directly or indirectly identify an individual, to ensure they no longer disclose personal information. Text sanitization must strike a balance between preventing the leakage of personal information (privacy protection) while also retaining as much of the document's original content as possible (utility preservation). We present an automated text sanitization strategy based on generalizations, which are more abstract (but still informative) terms that subsume the semantic content of the original text spans. The approach relies on instruction-tuned large language models (LLMs) and is divided into two stages. The LLM is first applied to obtain truth-preserving replacement candidates and rank them according to their abstraction level. Those candidates are then evaluated for their ability to protect privacy by conducting inference attacks with the LLM. Finally, the system selects the most informative replacement shown to be resistant to those attacks. As a consequence of this two-stage process, the chosen replacements effectively balance utility and privacy. We also present novel metrics to automatically evaluate these two aspects without the need to manually annotate data. Empirical results on the Text Anonymization Benchmark show that the proposed approach leads to enhanced utility, with only a marginal increase in the risk of re-identifying protected individuals compared to fully suppressing the original information. Furthermore, the selected replacements are shown to be more truth-preserving and abstractive than previous methods.
Authors: Dasol Choi, Guijin Son, Soo Yong Kim, Gio Paik, Seunghyeok Hong
Abstract: Visual-Language Models (VLMs) have become a powerful tool for bridging the gap between visual and linguistic understanding. However, the conventional learning approaches for VLMs often suffer from limitations, such as the high resource requirements of collecting and training image-text paired data. Recent research has suggested that language understanding plays a crucial role in the performance of VLMs, potentially indicating that text-only training could be a viable approach. In this work, we investigate the feasibility of enhancing fine-grained visual understanding in VLMs through text-only training. Inspired by how humans develop visual concept understanding, where rich textual descriptions can guide visual recognition, we hypothesize that VLMs can also benefit from leveraging text-based representations to improve their visual recognition abilities. We conduct comprehensive experiments on two distinct domains: fine-grained species classification and cultural visual understanding tasks. Our findings demonstrate that text-only training can be comparable to conventional image-text training while significantly reducing computational costs. This suggests a more efficient and cost-effective pathway for advancing VLM capabilities, particularly valuable in resource-constrained environments.
Authors: Yarik Menchaca Resendiz, Roman Klinger
Abstract: How emotions are expressed depends on the context and domain. On X (formerly Twitter), for instance, an author might simply use the hashtag #anger, while in a news headline, emotions are typically written in a more polite, indirect manner. To enable conditional text generation models to create emotionally connotated texts that fit a domain, users need to have access to a parameter that allows them to choose the appropriate way to express an emotion. To achieve this, we introduce MOPO, a Multi-Objective Prompt Optimization methodology. MOPO optimizes prompts according to multiple objectives (which correspond here to the output probabilities assigned by emotion classifiers trained for different domains). In contrast to single objective optimization, MOPO outputs a set of prompts, each with a different weighting of the multiple objectives. Users can then choose the most appropriate prompt for their context. We evaluate MOPO using three objectives, determined by various domain-specific emotion classifiers. MOPO improves performance by up to 15 pp across all objectives with a minimal loss (1-2 pp) for any single objective compared to single-objective optimization. These minor performance losses are offset by a broader generalization across multiple objectives - which is not possible with single-objective optimization. Additionally, MOPO reduces computational requirements by simultaneously optimizing for multiple objectives, eliminating separate optimization procedures for each objective.
Authors: Martin Borquez, Mikaela Keller, Michael Perrot, Damien Sileo
Abstract: It has been shown in the field of Author Profiling that texts may inadvertently reveal sensitive information about their authors, such as gender or age. This raises important privacy concerns that have been extensively addressed in the literature, in particular with the development of methods to hide such information. We argue that, when these texts are in fact messages exchanged between individuals, this is not the end of the story. Indeed, in this case, a second party, the intended recipient, is also involved and should be considered. In this work, we investigate the potential privacy leaks affecting them, that is we propose and address the problem of Recipient Profiling. We provide empirical evidence that such a task is feasible on several publicly accessible datasets (https://huggingface.co/datasets/sileod/recipient_profiling). Furthermore, we show that the learned models can be transferred to other datasets, albeit with a loss in accuracy.
URLs: https://huggingface.co/datasets/sileod/recipient_profiling).
Authors: Michael Heck, Christian Geishauser, Nurul Lubis, Carel van Niekerk, Shutong Feng, Hsien-Chin Lin, Benjamin Matthias Ruppik, Renato Vukovic, Milica Ga\v{s}i\'c
Abstract: Correct labels are indispensable for training effective machine learning models. However, creating high-quality labels is expensive, and even professionally labeled data contains errors and ambiguities. Filtering and denoising can be applied to curate labeled data prior to training, at the cost of additional processing and loss of information. An alternative is on-the-fly sample reweighting during the training process to decrease the negative impact of incorrect or ambiguous labels, but this typically requires clean seed data. In this work we propose unsupervised on-the-fly meta loss rescaling to reweight training samples. Crucially, we rely only on features provided by the model being trained, to learn a rescaling function in real time without knowledge of the true clean data distribution. We achieve this via a novel meta learning setup that samples validation data for the meta update directly from the noisy training corpus by employing the rescaling function being trained. Our proposed method consistently improves performance across various NLP tasks with minimal computational overhead. Further, we are among the first to attempt on-the-fly training data reweighting on the challenging task of dialogue modeling, where noisy and ambiguous labels are common. Our strategy is robust in the face of noisy and clean data, handles class imbalance, and prevents overfitting to noisy labels. Our self-taught loss rescaling improves as the model trains, showing the ability to keep learning from the model's own signals. As training progresses, the impact of correctly labeled data is scaled up, while the impact of wrongly labeled data is suppressed.
Authors: Mike Zhang, Max M\"uller-Eberstein, Elisa Bassignana, Rob van der Goot
Abstract: We present SnakModel, a Danish large language model (LLM) based on Llama2-7B, which we continuously pre-train on 13.6B Danish words, and further tune on 3.7M Danish instructions. As best practices for creating LLMs for smaller language communities have yet to be established, we examine the effects of early modeling and training decisions on downstream performance throughout the entire training pipeline, including (1) the creation of a strictly curated corpus of Danish text from diverse sources; (2) the language modeling and instruction-tuning training process itself, including the analysis of intermediate training dynamics, and ablations across different hyperparameters; (3) an evaluation on eight language and culturally-specific tasks. Across these experiments SnakModel achieves the highest overall performance, outperforming multiple contemporary Llama2-7B-based models. By making SnakModel, the majority of our pre-training corpus, and the associated code available under open licenses, we hope to foster further research and development in Danish Natural Language Processing, and establish training guidelines for languages with similar resource constraints.
Authors: Fatiha Ait Kbir, J\'er\'emy Bourgoin, R\'emy Decoupes, Marie Gradeler, Roberto Interdonato
Abstract: The Land Matrix initiative (https://landmatrix.org) and its global observatory aim to provide reliable data on large-scale land acquisitions to inform debates and actions in sectors such as agriculture, extraction, or energy in low- and middle-income countries. Although these data are recognized in the academic world, they remain underutilized in public policy, mainly due to the complexity of access and exploitation, which requires technical expertise and a good understanding of the database schema. The objective of this work is to simplify access to data from different database systems. The methods proposed in this article are evaluated using data from the Land Matrix. This work presents various comparisons of Large Language Models (LLMs) as well as combinations of LLM adaptations (Prompt Engineering, RAG, Agents) to query different database systems (GraphQL and REST queries). The experiments are reproducible, and a demonstration is available online: https://github.com/tetis-nlp/landmatrix-graphql-python.
URLs: https://landmatrix.org), https://github.com/tetis-nlp/landmatrix-graphql-python.
Authors: Vivek Kumar, Eirini Ntoutsi, Pushpraj Singh Rajawat, Giacomo Medda, Diego Reforgiato Recupero
Abstract: Large language models (LLMs) have shown promising capabilities in healthcare analysis but face several challenges like hallucinations, parroting, and bias manifestation. These challenges are exacerbated in complex, sensitive, and low-resource domains. Therefore, in this work we introduce IC-AnnoMI, an expert-annotated motivational interviewing (MI) dataset built upon AnnoMI by generating in-context conversational dialogues leveraging LLMs, particularly ChatGPT. IC-AnnoMI employs targeted prompts accurately engineered through cues and tailored information, taking into account therapy style (empathy, reflection), contextual relevance, and false semantic change. Subsequently, the dialogues are annotated by experts, strictly adhering to the Motivational Interviewing Skills Code (MISC), focusing on both the psychological and linguistic dimensions of MI dialogues. We comprehensively evaluate the IC-AnnoMI dataset and ChatGPT's emotional reasoning ability and understanding of domain intricacies by modeling novel classification tasks employing several classical machine learning and current state-of-the-art transformer approaches. Finally, we discuss the effects of progressive prompting strategies and the impact of augmented data in mitigating the biases manifested in IC-AnnoM. Our contributions provide the MI community with not only a comprehensive dataset but also valuable insights for using LLMs in empathetic text generation for conversational therapy in supervised settings.
Authors: Umer Butt, Stalin Veranasi, G\"unter Neumann
Abstract: As the Information Retrieval (IR) field increasingly recognizes the importance of inclusivity, addressing the needs of low-resource languages remains a significant challenge. This paper introduces the first large-scale Urdu IR dataset, created by translating the MS MARCO dataset through machine translation. We establish baseline results through zero-shot learning for IR in Urdu and subsequently apply the mMARCO multilingual IR methodology to this newly translated dataset. Our findings demonstrate that the fine-tuned model (Urdu-mT5-mMARCO) achieves a Mean Reciprocal Rank (MRR@10) of 0.247 and a Recall@10 of 0.439, representing significant improvements over zero-shot results and showing the potential for expanding IR access for Urdu speakers. By bridging access gaps for speakers of low-resource languages, this work not only advances multilingual IR research but also emphasizes the ethical and societal importance of inclusive IR technologies. This work provides valuable insights into the challenges and solutions for improving language representation and lays the groundwork for future research, especially in South Asian languages, which can benefit from the adaptable methods used in this study.
Authors: Tongguan Wang, Junkai Li, Guixin Su, Yongcheng Zhang, Dongyu Su, Yuxue Hu, Ying Sha
Abstract: Sarcasm typically conveys emotions of contempt or criticism by expressing a meaning that is contrary to the speaker's true intent. Accurate detection of sarcasm aids in identifying and filtering undesirable information on the Internet, thereby reducing malicious defamation and rumor-mongering. Nonetheless, the task of automatic sarcasm detection remains highly challenging for machines, as it critically depends on intricate factors such as relational context. Most existing multimodal sarcasm detection methods focus on introducing graph structures to establish entity relationships between text and images while neglecting to learn the relational context between text and images, which is crucial evidence for understanding the meaning of sarcasm. In addition, the meaning of sarcasm changes with the evolution of different contexts, but existing methods may not be accurate in modeling such dynamic changes, limiting the generalization ability of the models. To address the above issues, we propose a relational context learning and multiplex fusion network (RCLMuFN) for multimodal sarcasm detection. Firstly, we employ four feature extractors to comprehensively extract features from raw text and images, aiming to excavate potential features that may have been previously overlooked. Secondly, we utilize the relational context learning module to learn the contextual information of text and images and capture the dynamic properties through shallow and deep interactions. Finally, we employ a multiplex feature fusion module to enhance the generalization of the model by penetratingly integrating multimodal features derived from various interaction contexts. Extensive experiments on two multimodal sarcasm detection datasets show that our proposed method achieves state-of-the-art performance.
Authors: Shuting Wang, Jiejun Tan, Zhicheng Dou, Ji-Rong Wen
Abstract: As a typical and practical application of Large Language Models (LLMs), Retrieval-Augmented Generation (RAG) techniques have gained extensive attention, particularly in vertical domains where LLMs may lack domain-specific knowledge. In this paper, we introduce an omnidirectional and automatic RAG benchmark, OmniEval, in the financial domain. Our benchmark is characterized by its multi-dimensional evaluation framework, including (1) a matrix-based RAG scenario evaluation system that categorizes queries into five task classes and 16 financial topics, leading to a structured assessment of diverse query scenarios; (2) a multi-dimensional evaluation data generation approach, which combines GPT-4-based automatic generation and human annotation, achieving an 87.47\% acceptance ratio in human evaluations on generated instances; (3) a multi-stage evaluation system that evaluates both retrieval and generation performance, result in a comprehensive evaluation on the RAG pipeline; and (4) robust evaluation metrics derived from rule-based and LLM-based ones, enhancing the reliability of assessments through manual annotations and supervised fine-tuning of an LLM evaluator. Our experiments demonstrate the comprehensiveness of OmniEval, which includes extensive test datasets and highlights the performance variations of RAG systems across diverse topics and tasks, revealing significant opportunities for RAG models to improve their capabilities in vertical domains. We open source the code of our benchmark in \href{https://github.com/RUC-NLPIR/OmniEval}{https://github.com/RUC-NLPIR/OmniEval}.
URLs: https://github.com/RUC-NLPIR/OmniEval, https://github.com/RUC-NLPIR/OmniEval
Authors: Karan Wanchoo, Xiaoye Zuo, Hannah Gonzalez, Soham Dan, Georgios Georgakis, Dan Roth, Kostas Daniilidis, Eleni Miltsakaki
Abstract: We present NAVCON, a large-scale annotated Vision-Language Navigation (VLN) corpus built on top of two popular datasets (R2R and RxR). The paper introduces four core, cognitively motivated and linguistically grounded, navigation concepts and an algorithm for generating large-scale silver annotations of naturally occurring linguistic realizations of these concepts in navigation instructions. We pair the annotated instructions with video clips of an agent acting on these instructions. NAVCON contains 236, 316 concept annotations for approximately 30, 0000 instructions and 2.7 million aligned images (from approximately 19, 000 instructions) showing what the agent sees when executing an instruction. To our knowledge, this is the first comprehensive resource of navigation concepts. We evaluated the quality of the silver annotations by conducting human evaluation studies on NAVCON samples. As further validation of the quality and usefulness of the resource, we trained a model for detecting navigation concepts and their linguistic realizations in unseen instructions. Additionally, we show that few-shot learning with GPT-4o performs well on this task using large-scale silver annotations of NAVCON.
Authors: Hugo Math, Rainer Lienhart, Robin Sch\"on
Abstract: In this paper, we draw an analogy between processing natural languages and processing multivariate event streams from vehicles in order to predict $\textit{when}$ and $\textit{what}$ error pattern is most likely to occur in the future for a given car. Our approach leverages the temporal dynamics and contextual relationships of our event data from a fleet of cars. Event data is composed of discrete values of error codes as well as continuous values such as time and mileage. Modelled by two causal Transformers, we can anticipate vehicle failures and malfunctions before they happen. Thus, we introduce $\textit{CarFormer}$, a Transformer model trained via a new self-supervised learning strategy, and $\textit{EPredictor}$, an autoregressive Transformer decoder model capable of predicting $\textit{when}$ and $\textit{what}$ error pattern will most likely occur after some error code apparition. Despite the challenges of high cardinality of event types, their unbalanced frequency of appearance and limited labelled data, our experimental results demonstrate the excellent predictive ability of our novel model. Specifically, with sequences of $160$ error codes on average, our model is able with only half of the error codes to achieve $80\%$ F1 score for predicting $\textit{what}$ error pattern will occur and achieves an average absolute error of $58.4 \pm 13.2$h $\textit{when}$ forecasting the time of occurrence, thus enabling confident predictive maintenance and enhancing vehicle safety.
Authors: Mohammad Mahdi Abootorabi, Ehsaneddin Asgari
Abstract: This study introduces CLASP (Contrastive Language-Speech Pretraining), a multilingual, multimodal representation tailored for audio-text information retrieval. CLASP leverages the synergy between spoken content and textual data. During training, we utilize our newly introduced speech-text dataset, which encompasses 15 diverse categories ranging from fiction to religion. CLASP's audio component integrates audio spectrograms with a pre-trained self-supervised speech model, while its language encoding counterpart employs a sentence encoder pre-trained on over 100 languages. This unified lightweight model bridges the gap between various modalities and languages, enhancing its effectiveness in handling and retrieving multilingual and multimodal data. Our evaluations across multiple languages demonstrate that CLASP establishes new benchmarks in HITS@1, MRR, and meanR metrics, outperforming traditional ASR-based retrieval approaches in specific scenarios.
Authors: Jon Saad-Falcon, Rajan Vivek, William Berrios, Nandita Shankar Naik, Matija Franklin, Bertie Vidgen, Amanpreet Singh, Douwe Kiela, Shikib Mehri
Abstract: As language models become integral to critical workflows, assessing their behavior remains a fundamental challenge -- human evaluation is costly and noisy, while automated metrics provide only coarse, difficult-to-interpret signals. We introduce natural language unit tests, a paradigm that decomposes response quality into explicit, testable criteria, along with a unified scoring model, LMUnit, which combines multi-objective training across preferences, direct ratings, and natural language rationales. Through controlled human studies, we show this paradigm significantly improves inter-annotator agreement and enables more effective LLM development workflows. LMUnit achieves state-of-the-art performance on evaluation benchmarks (FLASK, BigGenBench) and competitive results on RewardBench. These results validate both our proposed paradigm and scoring model, suggesting a promising path forward for language model evaluation and development.
Authors: Roberto Mondini, Neema Kotonya, Robert L. Logan IV, Elizabeth M Olson, Angela Oduor Lungati, Daniel Duke Odongo, Tim Ombasa, Hemank Lamba, Aoife Cahill, Joel R. Tetreault, Alejandro Jaimes
Abstract: Online reporting platforms have enabled citizens around the world to collectively share their opinions and report in real time on events impacting their local communities. Systematically organizing (e.g., categorizing by attributes) and geotagging large amounts of crowdsourced information is crucial to ensuring that accurate and meaningful insights can be drawn from this data and used by policy makers to bring about positive change. These tasks, however, typically require extensive manual annotation efforts. In this paper we present Uchaguzi-2022, a dataset of 14k categorized and geotagged citizen reports related to the 2022 Kenyan General Election containing mentions of election-related issues such as official misconduct, vote count irregularities, and acts of violence. We use this dataset to investigate whether language models can assist in scalably categorizing and geotagging reports, thus highlighting its potential application in the AI for Social Good space.
Authors: Tiannan Wang, Meiling Tao, Ruoyu Fang, Huilin Wang, Shuai Wang, Yuchen Eleanor Jiang, Wangchunshu Zhou
Abstract: In this work, we introduce the task of life-long personalization of large language models. While recent mainstream efforts in the LLM community mainly focus on scaling data and compute for improved capabilities of LLMs, we argue that it is also very important to enable LLM systems, or language agents, to continuously adapt to the diverse and ever-changing profiles of every distinct user and provide up-to-date personalized assistance. We provide a clear task formulation and introduce a simple, general, effective, and scalable framework for life-long personalization of LLM systems and language agents. To facilitate future research on LLM personalization, we also introduce methods to synthesize realistic benchmarks and robust evaluation metrics. We will release all codes and data for building and benchmarking life-long personalized LLM systems.
Authors: Takumi Goto, Justin Vasselli, Taro Watanabe
Abstract: Various evaluation metrics have been proposed for Grammatical Error Correction (GEC), but many, particularly reference-free metrics, lack explainability. This lack of explainability hinders researchers from analyzing the strengths and weaknesses of GEC models and limits the ability to provide detailed feedback for users. To address this issue, we propose attributing sentence-level scores to individual edits, providing insight into how specific corrections contribute to the overall performance. For the attribution method, we use Shapley values, from cooperative game theory, to compute the contribution of each edit. Experiments with existing sentence-level metrics demonstrate high consistency across different edit granularities and show approximately 70\% alignment with human evaluations. In addition, we analyze biases in the metrics based on the attribution results, revealing trends such as the tendency to ignore orthographic edits. Our implementation is available at \url{https://github.com/naist-nlp/gec-attribute}.
Authors: Anton Alekseev, Alina Tillabaeva, Gulnara Dzh. Kabaeva, Sergey I. Nikolenko
Abstract: The Kyrgyz language, as a low-resource language, requires significant effort to create high-quality syntactic corpora. This study proposes an approach to simplify the development process of a syntactic corpus for Kyrgyz. We present a tool for transferring syntactic annotations from Turkish to Kyrgyz based on a treebank translation method. The effectiveness of the proposed tool was evaluated using the TueCL treebank. The results demonstrate that this approach achieves higher syntactic annotation accuracy compared to a monolingual model trained on the Kyrgyz KTMU treebank. Additionally, the study introduces a method for assessing the complexity of manual annotation for the resulting syntactic trees, contributing to further optimization of the annotation process.
Authors: Mohammad Nazmush Shamael, Sabila Nawshin, Swakkhar Shatabda, Salekul Islam
Abstract: This work presents the BanglishRev Dataset, the largest e-commerce product review dataset to date for reviews written in Bengali, English, a mixture of both and Banglish, Bengali words written with English alphabets. The dataset comprises of 1.74 million written reviews from 3.2 million ratings information collected from a total of 128k products being sold in online e-commerce platforms targeting the Bengali population. It includes an extensive array of related metadata for each of the reviews including the rating given by the reviewer, date the review was posted and date of purchase, number of likes, dislikes, response from the seller, images associated with the review etc. With sentiment analysis being the most prominent usage of review datasets, experimentation with a binary sentiment analysis model with the review rating serving as an indicator of positive or negative sentiment was conducted to evaluate the effectiveness of the large amount of data presented in BanglishRev for sentiment analysis tasks. A BanglishBERT model is trained on the data from BanglishRev with reviews being considered labeled positive if the rating is greater than 3 and negative if the rating is less than or equal to 3. The model is evaluated by being testing against a previously published manually annotated dataset for e-commerce reviews written in a mixture of Bangla, English and Banglish. The experimental model achieved an exceptional accuracy of 94\% and F1 score of 0.94, demonstrating the dataset's efficacy for sentiment analysis. Some of the intriguing patterns and observations seen within the dataset and future research directions where the dataset can be utilized is also discussed and explored. The dataset can be accessed through https://huggingface.co/datasets/BanglishRev/bangla-english-and-code-mixed-ecommerce-review-dataset.
URLs: https://huggingface.co/datasets/BanglishRev/bangla-english-and-code-mixed-ecommerce-review-dataset.
Authors: Bolei Ma, Berk Yoztyurk, Anna-Carolina Haensch, Xinpeng Wang, Markus Herklotz, Frauke Kreuter, Barbara Plank, Matthias Assenmacher
Abstract: In recent research, large language models (LLMs) have been increasingly used to investigate public opinions. This study investigates the algorithmic fidelity of LLMs, i.e., the ability to replicate the socio-cultural context and nuanced opinions of human participants. Using open-ended survey data from the German Longitudinal Election Studies (GLES), we prompt different LLMs to generate synthetic public opinions reflective of German subpopulations by incorporating demographic features into the persona prompts. Our results show that Llama performs better than other LLMs at representing subpopulations, particularly when there is lower opinion diversity within those groups. Our findings further reveal that the LLM performs better for supporters of left-leaning parties like The Greens and The Left compared to other parties, and matches the least with the right-party AfD. Additionally, the inclusion or exclusion of specific variables in the prompts can significantly impact the models' predictions. These findings underscore the importance of aligning LLMs to more effectively model diverse public opinions while minimizing political biases and enhancing robustness in representativeness.
Authors: Jeffrey Cheng, Benjamin Van Durme
Abstract: Chain-of-thought (CoT) decoding enables language models to improve reasoning performance at the cost of high generation latency in decoding. Recent proposals have explored variants of contemplation tokens, a term we introduce that refers to special tokens used during inference to allow for extra computation. Prior work has considered fixed-length sequences drawn from a discrete set of embeddings as contemplation tokens. Here we propose Compressed Chain-of-Thought (CCoT), a framework to generate contentful and continuous contemplation tokens of variable sequence length. The generated contemplation tokens are compressed representations of explicit reasoning chains, and our method can be applied to off-the-shelf decoder language models. Through experiments, we illustrate how CCoT enables additional reasoning over dense contentful representations to achieve corresponding improvements in accuracy. Moreover, the reasoning improvements can be adaptively modified on demand by controlling the number of contemplation tokens generated.
Authors: Miriam Wanner, Benjamin Van Durme, Mark Dredze
Abstract: The decompose-then-verify strategy for verification of Large Language Model (LLM) generations decomposes claims that are then independently verified. Decontextualization augments text (claims) to ensure it can be verified outside of the original context, enabling reliable verification. While decomposition and decontextualization have been explored independently, their interactions in a complete system have not been investigated. Their conflicting purposes can create tensions: decomposition isolates atomic facts while decontextualization inserts relevant information. Furthermore, a decontextualized subclaim presents a challenge to the verification step: what part of the augmented text should be verified as it now contains multiple atomic facts? We conduct an evaluation of different decomposition, decontextualization, and verification strategies and find that the choice of strategy matters in the resulting factuality scores. Additionally, we introduce DnDScore, a decontextualization aware verification method which validates subclaims in the context of contextual information.
Authors: Eunjung Yeo
Abstract: Dysarthria, a motor speech disorder, severely impacts voice quality, pronunciation, and prosody, leading to diminished speech intelligibility and reduced quality of life. Accurate assessment is crucial for effective treatment, but traditional perceptual assessments are limited by their subjectivity and resource intensity. To mitigate the limitations, automatic dysarthric speech assessment methods have been proposed to support clinicians on their decision-making. While these methods have shown promising results, most research has focused on monolingual environments. However, multilingual approaches are necessary to address the global burden of dysarthria and ensure equitable access to accurate diagnosis. This thesis proposes a novel multilingual dysarthria severity classification method, by analyzing three languages: English, Korean, and Tamil.
Authors: John Schultz, Jakub Adamek, Matej Jusup, Marc Lanctot, Michael Kaisers, Sarah Perrin, Daniel Hennes, Jeremy Shar, Cannada Lewis, Anian Ruoss, Tom Zahavy, Petar Veli\v{c}kovi\'c, Laurel Prince, Satinder Singh, Eric Malmi, Nenad Toma\v{s}ev
Abstract: While large language models perform well on a range of complex tasks (e.g., text generation, question answering, summarization), robust multi-step planning and reasoning remains a considerable challenge for them. In this paper we show that search-based planning can significantly improve LLMs' playing strength across several board games (Chess, Fischer Random / Chess960, Connect Four, and Hex). We introduce, compare and contrast two major approaches: In external search, the model guides Monte Carlo Tree Search (MCTS) rollouts and evaluations without calls to an external engine, and in internal search, the model directly generates in-context a linearized tree of potential futures and a resulting final choice. Both build on a language model pre-trained on relevant domain knowledge, capturing the transition and value functions across these games. We find that our pre-training method minimizes hallucinations, as our model is highly accurate regarding state prediction and legal moves. Additionally, both internal and external search indeed improve win-rates against state-of-the-art bots, even reaching Grandmaster-level performance in chess while operating on a similar move count search budget per decision as human Grandmasters. The way we combine search with domain knowledge is not specific to board games, suggesting direct extensions into more general language model inference and training techniques.
Authors: Christoph Leiter, Jonas Belouadi, Yanran Chen, Ran Zhang, Daniil Larionov, Aida Kostikova, Steffen Eger
Abstract: The NLLG (Natural Language Learning & Generation) arXiv reports assist in navigating the rapidly evolving landscape of NLP and AI research across cs.CL, cs.CV, cs.AI, and cs.LG categories. This fourth installment captures a transformative period in AI history - from January 1, 2023, following ChatGPT's debut, through September 30, 2024. Our analysis reveals substantial new developments in the field - with 45% of the top 40 most-cited papers being new entries since our last report eight months ago and offers insights into emerging trends and major breakthroughs, such as novel multimodal architectures, including diffusion and state space models. Natural Language Processing (NLP; cs.CL) remains the dominant main category in the list of our top-40 papers but its dominance is on the decline in favor of Computer vision (cs.CV) and general machine learning (cs.LG). This report also presents novel findings on the integration of generative AI in academic writing, documenting its increasing adoption since 2022 while revealing an intriguing pattern: top-cited papers show notably fewer markers of AI-generated content compared to random samples. Furthermore, we track the evolution of AI-associated language, identifying declining trends in previously common indicators such as "delve".
Authors: Bhaskarjit Sarmah, Mingshu Li, Jingrao Lyu, Sebastian Frank, Nathalia Castellanos, Stefano Pasquali, Dhagash Mehta
Abstract: To ensure and monitor large language models (LLMs) reliably, various evaluation metrics have been proposed in the literature. However, there is little research on prescribing a methodology to identify a robust threshold on these metrics even though there are many serious implications of an incorrect choice of the thresholds during deployment of the LLMs. Translating the traditional model risk management (MRM) guidelines within regulated industries such as the financial industry, we propose a step-by-step recipe for picking a threshold for a given LLM evaluation metric. We emphasize that such a methodology should start with identifying the risks of the LLM application under consideration and risk tolerance of the stakeholders. We then propose concrete and statistically rigorous procedures to determine a threshold for the given LLM evaluation metric using available ground-truth data. As a concrete example to demonstrate the proposed methodology at work, we employ it on the Faithfulness metric, as implemented in various publicly available libraries, using the publicly available HaluBench dataset. We also lay a foundation for creating systematic approaches to select thresholds, not only for LLMs but for any GenAI applications.
Authors: Sihan Chen, Zhuangzhuang Qian, Wingchun Siu, Xingcan Hu, Jiaqi Li, Shawn Li, Yuehan Qin, Tiankai Yang, Zhuo Xiao, Wanghao Ye, Yichi Zhang, Yushun Dong, Yue Zhao
Abstract: Outlier detection (OD), also known as anomaly detection, is a critical machine learning (ML) task with applications in fraud detection, network intrusion detection, clickstream analysis, recommendation systems, and social network moderation. Among open-source libraries for outlier detection, the Python Outlier Detection (PyOD) library is the most widely adopted, with over 8,500 GitHub stars, 25 million downloads, and diverse industry usage. However, PyOD currently faces three limitations: (1) insufficient coverage of modern deep learning algorithms, (2) fragmented implementations across PyTorch and TensorFlow, and (3) no automated model selection, making it hard for non-experts. To address these issues, we present PyOD Version 2 (PyOD 2), which integrates 12 state-of-the-art deep learning models into a unified PyTorch framework and introduces a large language model (LLM)-based pipeline for automated OD model selection. These improvements simplify OD workflows, provide access to 45 algorithms, and deliver robust performance on various datasets. In this paper, we demonstrate how PyOD 2 streamlines the deployment and automation of OD models and sets a new standard in both research and industry. PyOD 2 is accessible at [https://github.com/yzhao062/pyod](https://github.com/yzhao062/pyod). This study aligns with the Web Mining and Content Analysis track, addressing topics such as the robustness of Web mining methods and the quality of algorithmically-generated Web data.
URLs: https://github.com/yzhao062/pyod](https://github.com/yzhao062/pyod).
Authors: Melanie Sclar, Jane Yu, Maryam Fazel-Zarandi, Yulia Tsvetkov, Yonatan Bisk, Yejin Choi, Asli Celikyilmaz
Abstract: Do large language models (LLMs) have theory of mind? A plethora of papers and benchmarks have been introduced to evaluate if current models have been able to develop this key ability of social intelligence. However, all rely on limited datasets with simple patterns that can potentially lead to problematic blind spots in evaluation and an overestimation of model capabilities. We introduce ExploreToM, the first framework to allow large-scale generation of diverse and challenging theory of mind data for robust training and evaluation. Our approach leverages an A* search over a custom domain-specific language to produce complex story structures and novel, diverse, yet plausible scenarios to stress test the limits of LLMs. Our evaluation reveals that state-of-the-art LLMs, such as Llama-3.1-70B and GPT-4o, show accuracies as low as 0% and 9% on ExploreToM-generated data, highlighting the need for more robust theory of mind evaluation. As our generations are a conceptual superset of prior work, fine-tuning on our data yields a 27-point accuracy improvement on the classic ToMi benchmark (Le et al., 2019). ExploreToM also enables uncovering underlying skills and factors missing for models to show theory of mind, such as unreliable state tracking or data imbalances, which may contribute to models' poor performance on benchmarks.
Authors: Jing Zhang, Shuzhen Sun, Peng Zhang, Guangxing Cao, Hui Gao, Xindian Ma, Nan Xu, Yuexian Hou
Abstract: Transformer-based large language models exhibit groundbreaking capabilities, but their storage and computational costs are prohibitively high, limiting their application in resource-constrained scenarios. An effective approach is to eliminate redundant model parameters and computational costs while incorporating efficient expert-derived knowledge structures to achieve a balance between compression and performance. Therefore, we propose the \textit{Sememe Entanglement Encoding (SEE)} algorithm. Guided by expert prior knowledge, the model is compressed through the low-rank approximation idea. In Entanglement Embedding, basic semantic units such as sememes are represented as low-dimensional vectors, and then reconstructed into high-dimensional word embeddings through the combination of generalized quantum entanglement. We adapt the Sememe Entanglement Encoding algorithm to transformer-based models of different magnitudes. Experimental results indicate that our approach achieves stable performance while compressing model parameters and computational costs.
Authors: Portia Cooper, Harshita Narnoli, Mihai Surdeanu
Abstract: Text-to-image models are vulnerable to the stepwise "Divide-and-Conquer Attack" (DACA) that utilize a large language model to obfuscate inappropriate content in prompts by wrapping sensitive text in a benign narrative. To mitigate stepwise DACA attacks, we propose a two-layer method involving text summarization followed by binary classification. We assembled the Adversarial Text-to-Image Prompt (ATTIP) dataset ($N=940$), which contained DACA-obfuscated and non-obfuscated prompts. From the ATTIP dataset, we created two summarized versions: one generated by a small encoder model and the other by a large language model. Then, we used an encoder classifier and a GPT-4o classifier to perform content moderation on the summarized and unsummarized prompts. When compared with a classifier that operated over the unsummarized data, our method improved F1 score performance by 31%. Further, the highest recorded F1 score achieved (98%) was produced by the encoder classifier on a summarized ATTIP variant. This study indicates that pre-classification text summarization can inoculate content detection models against stepwise DACA obfuscations.
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: Ioannis Papadimitriou (Yiannis), Ilias Gialampoukidis (Yiannis), Stefanos Vrochidis (Yiannis), Ioannis (Yiannis), Kompatsiaris
Abstract: We present RAG Playground, an open-source framework for systematic evaluation of Retrieval-Augmented Generation (RAG) systems. The framework implements and compares three retrieval approaches: naive vector search, reranking, and hybrid vector-keyword search, combined with ReAct agents using different prompting strategies. We introduce a comprehensive evaluation framework with novel metrics and provide empirical results comparing different language models (Llama 3.1 and Qwen 2.5) across various retrieval configurations. Our experiments demonstrate significant performance improvements through hybrid search methods and structured self-evaluation prompting, achieving up to 72.7% pass rate on our multi-metric evaluation framework. The results also highlight the importance of prompt engineering in RAG systems, with our custom-prompted agents showing consistent improvements in retrieval accuracy and response quality.
Authors: Jinhe Bi, Yujun Wang, Haokun Chen, Xun Xiao, Artur Hecker, Volker Tresp, Yunpu Ma
Abstract: Multimodal Large Language Models (MLLMs) have significantly advanced visual tasks by integrating visual representations into large language models (LLMs). The textual modality, inherited from LLMs, equips MLLMs with abilities like instruction following and in-context learning. In contrast, the visual modality enhances performance in downstream tasks by leveraging rich semantic content, spatial information, and grounding capabilities. These intrinsic modalities work synergistically across various visual tasks. Our research initially reveals a persistent imbalance between these modalities, with text often dominating output generation during visual instruction tuning. This imbalance occurs when using both full fine-tuning and parameter-efficient fine-tuning (PEFT) methods. We then found that re-balancing these modalities can significantly reduce the number of trainable parameters required, inspiring a direction for further optimizing visual instruction tuning. We introduce Modality Linear Representation-Steering (MoReS) to achieve the goal. MoReS effectively re-balances the intrinsic modalities throughout the model, where the key idea is to steer visual representations through linear transformations in the visual subspace across each model layer. To validate our solution, we composed LLaVA Steering, a suite of models integrated with the proposed MoReS method. Evaluation results show that the composed LLaVA Steering models require, on average, 500 times fewer trainable parameters than LoRA needs while still achieving comparable performance across three visual benchmarks and eight visual question-answering tasks. Last, we present the LLaVA Steering Factory, an in-house developed platform that enables researchers to quickly customize various MLLMs with component-based architecture for seamlessly integrating state-of-the-art models, and evaluate their intrinsic modality imbalance.
Authors: Jaeseok Yoo, Hojae Han, Youngwon Lee, Jaejin Kim, Seung-won Hwang
Abstract: Code generation with large language models has shown significant promise, especially when employing retrieval-augmented generation (RAG) with few-shot examples. However, selecting effective examples that enhance generation quality remains a challenging task, particularly when the target programming language (PL) is underrepresented. In this study, we present two key findings: (1) retrieving examples whose presented algorithmic plans can be referenced for generating the desired behavior significantly improves generation accuracy, and (2) converting code into pseudocode effectively captures such algorithmic plans, enhancing retrieval quality even when the source and the target PLs are different. Based on these findings, we propose Plan-as-query Example Retrieval for few-shot prompting in Code generation (PERC), a novel framework that utilizes algorithmic plans to identify and retrieve effective examples. We validate the effectiveness of PERC through extensive experiments on the CodeContests, HumanEval and MultiPL-E benchmarks: PERC consistently outperforms the state-of-the-art RAG methods in code generation, both when the source and target programming languages match or differ, highlighting its adaptability and robustness in diverse coding environments.
Authors: Xunkai Li, Zhengyu Wu, Jiayi Wu, Hanwen Cui, Jishuo Jia, Rong-Hua Li, Guoren Wang
Abstract: With the increasing prevalence of cross-domain Text-Attributed Graph (TAG) Data (e.g., citation networks, recommendation systems, social networks, and ai4science), the integration of Graph Neural Networks (GNNs) and Large Language Models (LLMs) into a unified Model architecture (e.g., LLM as enhancer, LLM as collaborators, LLM as predictor) has emerged as a promising technological paradigm. The core of this new graph learning paradigm lies in the synergistic combination of GNNs' ability to capture complex structural relationships and LLMs' proficiency in understanding informative contexts from the rich textual descriptions of graphs. Therefore, we can leverage graph description texts with rich semantic context to fundamentally enhance Data quality, thereby improving the representational capacity of model-centric approaches in line with data-centric machine learning principles. By leveraging the strengths of these distinct neural network architectures, this integrated approach addresses a wide range of TAG-based Task (e.g., graph learning, graph reasoning, and graph question answering), particularly in complex industrial scenarios (e.g., supervised, few-shot, and zero-shot settings). In other words, we can treat text as a medium to enable cross-domain generalization of graph learning Model, allowing a single graph model to effectively handle the diversity of downstream graph-based Task across different data domains. This work serves as a foundational reference for researchers and practitioners looking to advance graph learning methodologies in the rapidly evolving landscape of LLM. We consistently maintain the related open-source materials at \url{https://github.com/xkLi-Allen/Awesome-GNN-in-LLMs-Papers}.
URLs: https://github.com/xkLi-Allen/Awesome-GNN-in-LLMs-Papers
Authors: Wenbin An, Haonan Lin, Jiahao Nie, Feng Tian, Wenkai Shi, Yaqiang Wu, Qianying Wang, Ping Chen
Abstract: Generalized Category Discovery is a significant and complex task that aims to identify both known and undefined novel categories from a set of unlabeled data, leveraging another labeled dataset containing only known categories. The primary challenges stem from model bias induced by pre-training on only known categories and the lack of precise supervision for novel ones, leading to category bias towards known categories and category confusion among different novel categories, which hinders models' ability to identify novel categories effectively. To address these challenges, we propose a novel framework named Self-Debiasing Calibration (SDC). Unlike prior methods that regard model bias towards known categories as an obstacle to novel category identification, SDC provides a novel insight into unleashing the potential of the bias to facilitate novel category learning. Specifically, the output of the biased model serves two key purposes. First, it provides an accurate modeling of category bias, which can be utilized to measure the degree of bias and debias the output of the current training model. Second, it offers valuable insights for distinguishing different novel categories by transferring knowledge between similar categories. Based on these insights, SDC dynamically adjusts the output logits of the current training model using the output of the biased model. This approach produces less biased logits to effectively address the issue of category bias towards known categories, and generates more accurate pseudo labels for unlabeled data, thereby mitigating category confusion for novel categories. Experiments on three benchmark datasets show that SDC outperforms SOTA methods, especially in the identification of novel categories. Our code and data are available at \url{https://github.com/Lackel/SDC}.
Authors: YiFan Zhang, Shanglin Lei, Runqi Qiao, Zhuoma GongQue, Xiaoshuai Song, Guanting Dong, Qiuna Tan, Zhe Wei, Peiqing Yang, Ye Tian, Yadong Xue, Xiaofei Wang, Honggang Zhang
Abstract: The rapidly developing field of large multimodal models (LMMs) has led to the emergence of diverse models with remarkable capabilities. However, existing benchmarks fail to comprehensively, objectively and accurately evaluate whether LMMs align with the diverse needs of humans in real-world scenarios. To bridge this gap, we propose the Multi-Dimensional Insights (MDI) benchmark, which includes over 500 images covering six common scenarios of human life. Notably, the MDI-Benchmark offers two significant advantages over existing evaluations: (1) Each image is accompanied by two types of questions: simple questions to assess the model's understanding of the image, and complex questions to evaluate the model's ability to analyze and reason beyond basic content. (2) Recognizing that people of different age groups have varying needs and perspectives when faced with the same scenario, our benchmark stratifies questions into three age categories: young people, middle-aged people, and older people. This design allows for a detailed assessment of LMMs' capabilities in meeting the preferences and needs of different age groups. With MDI-Benchmark, the strong model like GPT-4o achieve 79% accuracy on age-related tasks, indicating that existing LMMs still have considerable room for improvement in addressing real-world applications. Looking ahead, we anticipate that the MDI-Benchmark will open new pathways for aligning real-world personalization in LMMs. The MDI-Benchmark data and evaluation code are available at https://mdi-benchmark.github.io/
Authors: Hritik Bansal, Daniel Israel, Siyan Zhao, Shufan Li, Tung Nguyen, Aditya Grover
Abstract: Recent advancements in mixed-modal generative models have enabled flexible integration of information across image-text content. These models have opened new avenues for developing unified biomedical assistants capable of analyzing biomedical images, answering complex questions about them, and predicting the impact of medical procedures on a patient's health. However, existing resources face challenges such as limited data availability, narrow domain coverage, and restricted sources (e.g., medical papers). To address these gaps, we present MedMax, the first large-scale multimodal biomedical instruction-tuning dataset for mixed-modal foundation models. With 1.47 million instances, MedMax encompasses a diverse range of tasks, including multimodal content generation (interleaved image-text data), biomedical image captioning and generation, visual chatting, and report understanding. These tasks span diverse medical domains such as radiology and histopathology. Subsequently, we fine-tune a mixed-modal foundation model on the MedMax dataset, achieving significant performance improvements: a 26% gain over the Chameleon model and an 18.3% improvement over GPT-4o across 12 downstream biomedical visual question-answering tasks. Additionally, we introduce a unified evaluation suite for biomedical tasks, providing a robust framework to guide the development of next-generation mixed-modal biomedical AI assistants.
Authors: Mukai Li, Lei Li, Shansan Gong, Qi Liu
Abstract: Visual Language Models (VLMs) demonstrate impressive capabilities in processing multimodal inputs, yet applications such as visual agents, which require handling multiple images and high-resolution videos, demand enhanced long-range modeling. Moreover, existing open-source VLMs lack systematic exploration into extending their context length, and commercial models often provide limited details. To tackle this, we aim to establish an effective solution that enhances long context performance of VLMs while preserving their capacities in short context scenarios. Towards this goal, we make the best design choice through extensive experiment settings from data curation to context window extending and utilizing: (1) we analyze data sources and length distributions to construct ETVLM - a data recipe to balance the performance across scenarios; (2) we examine existing position extending methods, identify their limitations and propose M-RoPE++ as an enhanced approach; we also choose to solely instruction-tune the backbone with mixed-source data; (3) we discuss how to better utilize extended context windows and propose hybrid-resolution training. Built on the Qwen-VL series model, we propose Giraffe, which is effectively extended to 128K lengths. Evaluated on extensive long context VLM benchmarks such as VideoMME and Viusal Haystacks, our Giraffe achieves state-of-the-art performance among similarly sized open-source long VLMs and is competitive with commercial model GPT-4V. We will open-source the code, data, and models.
Authors: Paheli Bhattacharya, Rishabh Gupta
Abstract: Code explanation plays a crucial role in the software engineering domain, aiding developers in grasping code functionality efficiently. Recent work shows that the performance of LLMs for code explanation improves in a few-shot setting, especially when the few-shot examples are selected intelligently. State-of-the-art approaches for such Selective Shot Learning (SSL) include token-based and embedding-based methods. However, these SSL approaches have been evaluated on proprietary LLMs, without much exploration on open-source Code-LLMs. Additionally, these methods lack consideration for programming language syntax. To bridge these gaps, we present a comparative study and propose a novel SSL method (SSL_ner) that utilizes entity information for few-shot example selection. We present several insights and show the effectiveness of SSL_ner approach over state-of-the-art methods across two datasets. To the best of our knowledge, this is the first systematic benchmarking of open-source Code-LLMs while assessing the performances of the various few-shot examples selection approaches for the code explanation task.
Authors: Shreeyash Gowaikar, Srinivasan Iyengar, Sameer Segal, Shivkumar Kalyanaraman
Abstract: The Piping and Instrumentation Diagrams (P&IDs) are foundational to the design, construction, and operation of workflows in the engineering and process industries. However, their manual creation is often labor-intensive, error-prone, and lacks robust mechanisms for error detection and correction. While recent advancements in Generative AI, particularly Large Language Models (LLMs) and Vision-Language Models (VLMs), have demonstrated significant potential across various domains, their application in automating generation of engineering workflows remains underexplored. In this work, we introduce a novel copilot for automating the generation of P&IDs from natural language descriptions. Leveraging a multi-step agentic workflow, our copilot provides a structured and iterative approach to diagram creation directly from Natural Language prompts. We demonstrate the feasibility of the generation process by evaluating the soundness and completeness of the workflow, and show improved results compared to vanilla zero-shot and few-shot generation approaches.
Authors: Weiguo Pian, Shijian Deng, Shentong Mo, Yunhui Guo, Yapeng Tian
Abstract: In this paper, we introduce Modality-Inconsistent Continual Learning (MICL), a new continual learning scenario for Multimodal Large Language Models (MLLMs) that involves tasks with inconsistent modalities (image, audio, or video) and varying task types (captioning or question-answering). Unlike existing vision-only or modality-incremental settings, MICL combines modality and task type shifts, both of which drive catastrophic forgetting. To address these challenges, we propose MoInCL, which employs a Pseudo Targets Generation Module to mitigate forgetting caused by task type shifts in previously seen modalities. It also incorporates Instruction-based Knowledge Distillation to preserve the model's ability to handle previously learned modalities when new ones are introduced. We benchmark MICL using a total of six tasks and conduct experiments to validate the effectiveness of our proposed MoInCL. The experimental results highlight the superiority of MoInCL, showing significant improvements over representative and state-of-the-art continual learning baselines.
Authors: Jianlyu Chen, Nan Wang, Chaofan Li, Bo Wang, Shitao Xiao, Han Xiao, Hao Liao, Defu Lian, Zheng Liu
Abstract: Evaluation plays a crucial role in the advancement of information retrieval (IR) models. However, current benchmarks, which are based on predefined domains and human-labeled data, face limitations in addressing evaluation needs for emerging domains both cost-effectively and efficiently. To address this challenge, we propose the Automated Heterogeneous Information Retrieval Benchmark (AIR-Bench). AIR-Bench is distinguished by three key features: 1) Automated. The testing data in AIR-Bench is automatically generated by large language models (LLMs) without human intervention. 2) Heterogeneous. The testing data in AIR-Bench is generated with respect to diverse tasks, domains and languages. 3) Dynamic. The domains and languages covered by AIR-Bench are constantly augmented to provide an increasingly comprehensive evaluation benchmark for community developers. We develop a reliable and robust data generation pipeline to automatically create diverse and high-quality evaluation datasets based on real-world corpora. Our findings demonstrate that the generated testing data in AIR-Bench aligns well with human-labeled testing data, making AIR-Bench a dependable benchmark for evaluating IR models. The resources in AIR-Bench are publicly available at https://github.com/AIR-Bench/AIR-Bench.
Authors: Tatsuki Kuribayashi, Timothy Baldwin
Abstract: Neural language models (LMs) are arguably less data-efficient than humans from a language acquisition perspective. One fundamental question is why this human-LM gap arises. This study explores the advantage of grounded language acquisition, specifically the impact of visual information -- which humans can usually rely on but LMs largely do not have access to during language acquisition -- on syntactic generalization in LMs. Our experiments, following the poverty of stimulus paradigm under two scenarios (using artificial vs. naturalistic images), demonstrate that if the alignments between the linguistic and visual components are clear in the input, access to vision data does help with the syntactic generalization of LMs, but if not, visual input does not help. This highlights the need for additional biases or signals, such as mutual gaze, to enhance cross-modal alignment and enable efficient syntactic generalization in multimodal LMs.
Authors: Jordan Meadows, Marco Valentino, Andre Freitas
Abstract: This paper investigates how hallucination rates in Large Language Models (LLMs) may be controlled and mitigated via a symbolic data generation framework, and explores a fundamental relationship between the rate of certain mathematical errors and interventions. Specifically, we systematically generate data for a derivation generation task, and apply targeted interventions on prompts to perturb aspects such as the surface forms of symbols, equational tree structures, and mathematical context, and evaluate the effect of prompt interventions across a range of LLMs including fine-tuned T5 models, GPT, and others. Experiments suggest that T5-Large can outperform the few-shot performance of GPT-4 on various evaluation sets generated via the framework, however, an extensive evaluation based on human analysis, template-based error detection, and various text generation metrics reveals fine-tuned model weaknesses beyond what the reference-based metrics singularly describe. We use these results to tie characteristic distributional footprints of interventions to the human evaluation of LLM derivation quality, potentially leading to significant control over fine-grained mathematical capabilities of language models with respect to specific types of errors.
Authors: Amir Taubenfeld, Yaniv Dover, Roi Reichart, Ariel Goldstein
Abstract: The emergence of Large Language Models (LLMs), has opened exciting possibilities for constructing computational simulations designed to replicate human behavior accurately. Current research suggests that LLM-based agents become increasingly human-like in their performance, sparking interest in using these AI agents as substitutes for human participants in behavioral studies. However, LLMs are complex statistical learners without straightforward deductive rules, making them prone to unexpected behaviors. Hence, it is crucial to study and pinpoint the key behavioral distinctions between humans and LLM-based agents. In this study, we highlight the limitations of LLMs in simulating human interactions, particularly focusing on LLMs' ability to simulate political debates on topics that are important aspects of people's day-to-day lives and decision-making processes. Our findings indicate a tendency for LLM agents to conform to the model's inherent social biases despite being directed to debate from certain political perspectives. This tendency results in behavioral patterns that seem to deviate from well-established social dynamics among humans. We reinforce these observations using an automatic self-fine-tuning method, which enables us to manipulate the biases within the LLM and demonstrate that agents subsequently align with the altered biases. These results underscore the need for further research to develop methods that help agents overcome these biases, a critical step toward creating more realistic simulations.
Authors: Quanyu Long, Yue Deng, LeiLei Gan, Wenya Wang, Sinno Jialin Pan
Abstract: Dense retrieval systems have been widely used in various NLP applications. However, their vulnerabilities to potential attacks have been underexplored. This paper investigates a novel attack scenario where the attackers aim to mislead the retrieval system into retrieving the attacker-specified contents. Those contents, injected into the retrieval corpus by attackers, can include harmful text like hate speech or spam. Unlike prior methods that rely on model weights and generate conspicuous, unnatural outputs, we propose a covert backdoor attack triggered by grammar errors. Our approach ensures that the attacked models can function normally for standard queries while covertly triggering the retrieval of the attacker's contents in response to minor linguistic mistakes. Specifically, dense retrievers are trained with contrastive loss and hard negative sampling. Surprisingly, our findings demonstrate that contrastive loss is notably sensitive to grammatical errors, and hard negative sampling can exacerbate susceptibility to backdoor attacks. Our proposed method achieves a high attack success rate with a minimal corpus poisoning rate of only 0.048%, while preserving normal retrieval performance. This indicates that the method has negligible impact on user experience for error-free queries. Furthermore, evaluations across three real-world defense strategies reveal that the malicious passages embedded within the corpus remain highly resistant to detection and filtering, underscoring the robustness and subtlety of the proposed attack.
Authors: Yuepei Li, Kang Zhou, Qiao Qiao, Qing Wang, Qi Li
Abstract: Distantly-Supervised Named Entity Recognition (DS-NER) uses knowledge bases or dictionaries for annotations, reducing manual efforts but rely on large human labeled validation set. In this paper, we introduce a real-life DS-NER dataset, QTL, where the training data is annotated using domain dictionaries and the test data is annotated by domain experts. This dataset has a small validation set, reflecting real-life scenarios. Existing DS-NER approaches fail when applied to QTL, which motivate us to re-examine existing DS-NER approaches. We found that many of them rely on large validation sets and some used test set for tuning inappropriately. To solve this issue, we proposed a new approach, token-level Curriculum-based Positive-Unlabeled Learning (CuPUL), which uses curriculum learning to order training samples from easy to hard. This method stabilizes training, making it robust and effective on small validation sets. CuPUL also addresses false negative issues using the Positive-Unlabeled learning paradigm, demonstrating improved performance in real-life applications.
Authors: Jiawei Zheng, Hanghai Hong, Feiyan Liu, Xiaoli Wang, Jingsong Su, Yonggui Liang, Shikai Wu
Abstract: Large language models (LLMs) have shown great potential in domain-specific machine translation (MT). However, one major issue is that LLMs pre-trained on general domain corpus might not generalize well to specific domains due to the lack of domain-specific knowledge. To address this issue, this paper focuses on enhancing the domain-specific MT capability of LLMs, by providing high-quality training datasets and proposing a novel fine-tuning framework denoted by DragFT. DragFT augments LLMs via three techniques: (i) Dictionary-enhanced prompting integrates dictionary information into prompts to improve the translation of domain-specific terminology.; (ii) RAG-based few-shot example selection provides high-quality examples that simulate both the domain and style characteristics; (iii) Fine-tuning with few-shot examples further enhances performance when using in-domain examples. We deploy DragFT on three well-known LLM backbones with 13B training parameters to validate its effectiveness. The results on three domain-specific datasets show that DragFT achieves a significant performance boost and shows superior performance compared to advanced models such as GPT-3.5 and GPT-4o. The drastic performance improvement of DragFT over existing LLMs can be attributed to incorporating relevant knowledge while mitigating noise.
Authors: Jiebin Zhang, Eugene J. Yu, Qinyu Chen, Chenhao Xiong, Dawei Zhu, Han Qian, Mingbo Song, Weimin Xiong, Xiaoguang Li, Qun Liu, Sujian Li
Abstract: It presents significant challenges to generate comprehensive and accurate Wikipedia articles for newly emerging events under a real-world scenario. Existing attempts fall short either by focusing only on short snippets or by using metrics that are insufficient to evaluate real-world scenarios. In this paper, we construct WIKIGENBENCH, a new benchmark consisting of 1,320 entries, designed to align with real-world scenarios in both generation and evaluation. For generation, we explore a real-world scenario where structured, full-length Wikipedia articles with citations are generated for new events using input documents from web sources. For evaluation, we integrate systematic metrics and LLM-based metrics to assess the verifiability, organization, and other aspects aligned with real-world scenarios. Based on this benchmark, we conduct extensive experiments using various models within three commonly used frameworks: direct RAG, hierarchical structure-based RAG, and RAG with a fine-tuned generation model. Experimental results show that hierarchical-based methods can generate more comprehensive content, while fine-tuned methods achieve better verifiability. However, even the best methods still show a significant gap compared to existing Wikipedia content, indicating that further research is necessary.
Authors: Congzhi Zhang, Linhai Zhang, Jialong Wu, Yulan He, Deyu Zhou
Abstract: Despite the notable advancements of existing prompting methods, such as In-Context Learning and Chain-of-Thought for Large Language Models (LLMs), they still face challenges related to various biases. Traditional debiasing methods primarily focus on the model training stage, including approaches based on data augmentation and reweighting, yet they struggle with the complex biases inherent in LLMs. To address such limitations, the causal relationship behind the prompting methods is uncovered using a structural causal model, and a novel causal prompting method based on front-door adjustment is proposed to effectively mitigate LLMs biases. In specific, causal intervention is achieved by designing the prompts without accessing the parameters and logits of LLMs. The chain-of-thought generated by LLM is employed as the mediator variable and the causal effect between input prompts and output answers is calculated through front-door adjustment to mitigate model biases. Moreover, to accurately represent the chain-of-thoughts and estimate the causal effects, contrastive learning is used to fine-tune the encoder of chain-of-thought by aligning its space with that of the LLM. Experimental results show that the proposed causal prompting approach achieves excellent performance across seven natural language processing datasets on both open-source and closed-source LLMs.
Authors: Prabin Bhandari, Abhishek Paudel
Abstract: Recent advancements in morpheme segmentation primarily emphasize word-level segmentation, often neglecting the contextual relevance within the sentence. In this study, we redefine the morpheme segmentation task as a sequence-to-sequence problem, treating the entire sentence as input rather than isolating individual words. Our findings reveal that the multilingual model consistently exhibits superior performance compared to monolingual counterparts. While our model did not surpass the performance of the current state-of-the-art, it demonstrated comparable efficacy with high-resource languages while revealing limitations in low-resource language scenarios.
Authors: Qinyuan Wu, Mohammad Aflah Khan, Soumi Das, Vedant Nanda, Bishwamittra Ghosh, Camila Kolling, Till Speicher, Laurent Bindschaedler, Krishna P. Gummadi, Evimaria Terzi
Abstract: In this paper, we focus on the challenging task of reliably estimating factual knowledge that is embedded inside large language models (LLMs). To avoid reliability concerns with prior approaches, we propose to eliminate prompt engineering when probing LLMs for factual knowledge. Our approach, called Zero-Prompt Latent Knowledge Estimator (ZP-LKE), leverages the in-context learning ability of LLMs to communicate both the factual knowledge question as well as the expected answer format. Our knowledge estimator is both conceptually simpler (i.e., doesn't depend on meta-linguistic judgments of LLMs) and easier to apply (i.e., is not LLM-specific), and we demonstrate that it can surface more of the latent knowledge embedded in LLMs. We also investigate how different design choices affect the performance of ZP-LKE. Using the proposed estimator, we perform a large-scale evaluation of the factual knowledge of a variety of open-source LLMs, like OPT, Pythia, Llama(2), Mistral, Gemma, etc. over a large set of relations and facts from the Wikidata knowledge base. We observe differences in the factual knowledge between different model families and models of different sizes, that some relations are consistently better known than others but that models differ in the precise facts they know, and differences in the knowledge of base models and their finetuned counterparts. Code available at: https://github.com/QinyuanWu0710/ZeroPrompt_LKE
Authors: Ngoc Phuoc An Vo, Brent Paulovicks, Vadim Sheinin
Abstract: Given recent advancements of Large Language Models (LLMs), code generation tasks attract immense attention for wide application in different domains. In an effort to evaluate and select a best model to automatically remediate system incidents discovered by Application Performance Monitoring (APM) platforms, it is crucial to verify if the generated code is syntactically and semantically correct, and whether it can be executed correctly as intended. However, current methods for evaluating the quality of code generated by LLMs heavily rely on surface form similarity metrics (e.g. BLEU, ROUGE, and exact/partial match) which have numerous limitations. In contrast, execution based evaluation focuses more on code functionality and does not constrain the code generation to any fixed solution. Nevertheless, designing and implementing such execution-based evaluation platform is not a trivial task. There are several works creating execution-based evaluation platforms for popular programming languages such as SQL, Python, Java, but limited or no attempts for scripting languages such as Bash and PowerShell. In this paper, we present the first execution-based evaluation platform in which we created three test suites (total 125 handcrafted test cases) to evaluate Bash (both single-line commands and multiple-line scripts) and PowerShell codes generated by LLMs. We benchmark seven closed and open-source LLMs using our platform with different techniques (zero-shot vs. few-shot learning).
Authors: Weixiang Zhao, Yulin Hu, Zhuojun Li, Yang Deng, Jiahe Guo, Xingyu Sui, Yanyan Zhao, Bing Qin, Tat-Seng Chua, Ting Liu
Abstract: Safety alignment of large language models (LLMs) has been gaining increasing attention. However, current safety-aligned LLMs suffer from the fragile and imbalanced safety mechanisms, which can still be induced to generate unsafe responses, exhibit over-safety by rejecting safe user inputs, and fail to preserve general utility after safety alignment. To this end, we propose a novel post safety alignment (PSA) method to address these inherent and emerging safety challenges, including safety enhancement, over-safety mitigation, and utility preservation. In specific, we introduce \textsc{SafePatching}, a novel framework for comprehensive PSA, where two distinct safety patches are developed on the harmful data to enhance safety and mitigate over-safety concerns, and then seamlessly integrated into the target LLM backbone without compromising its utility. Extensive experiments on four representative aligned LLMs, including LLaMA-2/3, Gemma and Mistral, show that \textsc{SafePatching} achieves a more comprehensive PSA than baseline methods, further optimizing the balance between being helpful and harmless in current aligned LLMs. Also, \textsc{SafePatching} demonstrates its superiority in continual PSA scenarios.
Authors: Zhen Tao, Yanfang Chen, Dinghao Xi, Zhiyu Li, Wei Xu
Abstract: The increasing prevalence of large language models (LLMs) has significantly advanced text generation, but the human-like quality of LLM outputs presents major challenges in reliably distinguishing between human-authored and LLM-generated texts. Existing detection benchmarks are constrained by their reliance on static datasets, scenario-specific tasks (e.g., question answering and text refinement), and a primary focus on English, overlooking the diverse linguistic and operational subtleties of LLMs. To address these gaps, we propose CUDRT, a comprehensive evaluation framework and bilingual benchmark in Chinese and English, categorizing LLM activities into five key operations: Create, Update, Delete, Rewrite, and Translate. CUDRT provides extensive datasets tailored to each operation, featuring outputs from state-of-the-art LLMs to assess the reliability of LLM-generated text detectors. This framework supports scalable, reproducible experiments and enables in-depth analysis of how operational diversity, multilingual training sets, and LLM architectures influence detection performance. Our extensive experiments demonstrate the framework's capacity to optimize detection systems, providing critical insights to enhance reliability, cross-linguistic adaptability, and detection accuracy. By advancing robust methodologies for identifying LLM-generated texts, this work contributes to the development of intelligent systems capable of meeting real-world multilingual detection challenges. Source code and dataset are available at GitHub.
Authors: Hiroaki Yamagiwa, Momose Oyama, Hidetoshi Shimodaira
Abstract: Cosine similarity is widely used to measure the similarity between two embeddings, while interpretations based on angle and correlation coefficient are common. In this study, we focus on the interpretable axes of embeddings transformed by Independent Component Analysis (ICA), and propose a novel interpretation of cosine similarity as the sum of semantic similarities over axes. The normalized ICA-transformed embeddings exhibit sparsity, enhancing the interpretability of each axis, and the semantic similarity defined by the product of the components represents the shared meaning between the two embeddings along each axis. The effectiveness of this approach is demonstrated through intuitive numerical examples and thorough numerical experiments. By deriving the probability distributions that govern each component and the product of components, we propose a method for selecting statistically significant axes.
Authors: Boyi Deng, Wenjie Wang, Fengbin Zhu, Qifan Wang, Fuli Feng
Abstract: Retrieval-Augmented Generation (RAG) can alleviate hallucinations of Large Language Models (LLMs) by referencing external documents. However, the misinformation in external documents may mislead LLMs' generation. To address this issue, we explore the task of "credibility-aware RAG", in which LLMs automatically adjust the influence of retrieved documents based on their credibility scores to counteract misinformation. To this end, we introduce a plug-and-play method named $\textbf{Cr}$edibility-aware $\textbf{A}$ttention $\textbf{M}$odification (CrAM). CrAM identifies influential attention heads in LLMs and adjusts their attention weights based on the credibility of the documents, thereby reducing the impact of low-credibility documents. Experiments on Natual Questions and TriviaQA using Llama2-13B, Llama3-8B, and Qwen1.5-7B show that CrAM improves the RAG performance of LLMs against misinformation pollution by over 20%, even surpassing supervised fine-tuning methods.
Authors: Seungbin Yang, ChaeHun Park, Taehee Kim, Jaegul Choo
Abstract: Recent advancements in integrating large language models (LLMs) with tools have allowed the models to interact with real-world environments. However, these tool-augmented LLMs often encounter incomplete scenarios when users provide partial information or the necessary tools are unavailable. Recognizing and managing such scenarios is crucial for LLMs to ensure their reliability, but this exploration remains understudied. This study examines whether LLMs can identify incomplete conditions and appropriately determine when to refrain from using tools. To this end, we address a dataset by manipulating instances from two datasets by removing necessary tools or essential information for tool invocation. Our experiments show that LLMs often struggle to identify the absence of information required to utilize specific tools and recognize the absence of appropriate tools. We further analyze model behaviors in different environments and compare their performance against humans. Our research can contribute to advancing reliable LLMs by addressing common scenarios during interactions between humans and LLMs. Our code and dataset will be publicly available.
Authors: Xue Zhang, Yunlong Liang, Fandong Meng, Songming Zhang, Yufeng Chen, Jinan Xu, Jie Zhou
Abstract: Multilingual knowledge editing (MKE) aims to simultaneously update factual knowledge across multiple languages within large language models (LLMs). Previous research indicates that the same knowledge across different languages within LLMs exhibits a degree of shareability. However, most existing MKE methods overlook the connections of the same knowledge between different languages, resulting in knowledge conflicts and limited edit performance. To address this issue, we first investigate how LLMs process multilingual factual knowledge and discover that the same factual knowledge in different languages generally activates a shared set of neurons, which we call language-agnostic factual neurons (LAFNs). These neurons represent the same factual knowledge shared across languages and imply the semantic connections among multilingual knowledge. Inspired by this finding, we propose a new MKE method by Locating and Updating Language-Agnostic Factual Neurons (LU-LAFNs) to edit multilingual knowledge simultaneously, which avoids knowledge conflicts and thus improves edit performance. Experimental results on Bi-ZsRE and MzsRE benchmarks demonstrate that our method achieves the best edit performance, indicating the effectiveness and importance of modeling the semantic connections among multilingual knowledge.
Authors: Yujin Baek, ChaeHun Park, Jaeseok Kim, Yu-Jung Heo, Du-Seong Chang, Jaegul Choo
Abstract: To create culturally inclusive vision-language models (VLMs), developing a benchmark that tests their ability to address culturally relevant questions is essential. Existing approaches typically rely on human annotators, making the process labor-intensive and creating a cognitive burden in generating diverse questions. To address this, we propose a semi-automated framework for constructing cultural VLM benchmarks, specifically targeting multiple-choice QA. This framework combines human-VLM collaboration, where VLMs generate questions based on guidelines, a small set of annotated examples, and relevant knowledge, followed by a verification process by native speakers. We demonstrate the effectiveness of this framework through the creation of K-Viscuit, a dataset focused on Korean culture. Our experiments on this dataset reveal that open-source models lag behind proprietary ones in understanding Korean culture, highlighting key areas for improvement. We also present a series of further analyses, including human evaluation, augmenting VLMs with external knowledge, and the evaluation beyond multiple-choice QA. Our dataset is available at https://huggingface.co/datasets/ddehun/k-viscuit.
Authors: Huaizhi Ge, Frank Rudzicz, Zining Zhu
Abstract: Large language models (LLMs) have demonstrated remarkable capabilities, but updating their knowledge post-training remains a critical challenge. While recent model editing techniques like Rank-One Model Editing (ROME) show promise, their effectiveness may vary based on the nature of the knowledge being edited. We introduce the concept of ``perplexingness'': the degree to which new knowledge conflicts with an LLM's learned conceptual hierarchies and categorical relationships. For instance, editing ``British Shorthair is a kind of cat'' to ``British Shorthair is a kind of dog'' represents a low-perplexingness edit within the same taxonomic level, while editing ``A cat is a kind of animal'' to ``A cat is a kind of plant'' represents a high-perplexingness edit that violates fundamental categorical boundaries. To systematically investigate this phenomenon, we introduce HierarchyData, a carefully curated dataset of 99 hyponym-hypernym pairs across diverse categories. Through controlled experiments across three models and four editing methods, we demonstrate a strong negative correlation between the perplexingness of new knowledge and the effectiveness of knowledge editing. Our analysis reveals that edits involving more abstract concepts (hypernyms) generally exhibit higher perplexingness and are more resistant to modification than their specific counterparts (hyponyms). These findings highlight a fundamental challenge in LLM knowledge editing: the more a new fact contradicts an LLM's learned conceptual hierarchies, the harder it becomes to reliably encode that knowledge.
Authors: Zhenglin Wang, Jialong Wu, Yilong Lai, Congzhi Zhang, Deyu Zhou
Abstract: Large Language Models (LLMs) demonstrate remarkable emergent abilities across various tasks, yet fall short of complex reasoning and planning tasks. The tree-search-based reasoning methods address this by surpassing the capabilities of chain-of-thought prompting, encouraging exploration of intermediate steps. However, such methods introduce significant inference latency due to the systematic exploration and evaluation of multiple thought paths. This paper introduces SeeD, a novel and efficient inference framework to optimize runtime speed and GPU memory management concurrently. By employing a scheduled speculative execution, SeeD efficiently handles multiple iterations for the thought generation and the state evaluation, leveraging a rounds-scheduled strategy to manage draft model dispatching. Extensive experimental evaluations on three reasoning datasets demonstrate superior speedup performance of SeeD, providing a viable path for batched inference in training-free speculative decoding.
Authors: Wen Ding, Fei Jia, Hainan Xu, Yu Xi, Junjie Lai, Boris Ginsburg
Abstract: We introduce romanization encoding for script-heavy languages to optimize multilingual and code-switching Automatic Speech Recognition (ASR) systems. By adopting romanization encoding alongside a balanced concatenated tokenizer within a FastConformer-RNNT framework equipped with a Roman2Char module, we significantly reduce vocabulary and output dimensions, enabling larger training batches and reduced memory consumption. Our method decouples acoustic modeling and language modeling, enhancing the flexibility and adaptability of the system. In our study, applying this method to Mandarin-English ASR resulted in a remarkable 63.51% vocabulary reduction and notable performance gains of 13.72% and 15.03% on SEAME code-switching benchmarks. Ablation studies on Mandarin-Korean and Mandarin-Japanese highlight our method's strong capability to address the complexities of other script-heavy languages, paving the way for more versatile and effective multilingual ASR systems.
Authors: Yuxi Xia, Kilm Zaporojets, Benjamin Roth
Abstract: A diverse range of large language models (LLMs), e.g., ChatGPT, and visual question answering (VQA) models, e.g., BLIP, have been developed for solving textual and visual question answering tasks. However, fine-tuning these models is either difficult, as it requires access via APIs, rendering them as black-boxes, or costly due to the need of tuning a large number of parameters. To address this, we introduce InfoSel, a data-efficient ensemble method that learns to dynamically pick the winner from existing black-box models for predictions on both textual and multimodal visual question answering tasks. Unlike traditional ensemble models, InfoSel does not rely on prediction probabilities or confidences, which typically are not available in black-box models. Experimental results on four datasets demonstrate that our approach achieves an absolute increase of up to +5.19\% in the F1-score compared to standalone LLMs using only 1K training instances.
Authors: Jiajun Shen, Tong Zhou, Yubo Chen, Kang Liu
Abstract: Enabling Large Language Models (LLMs) to generate citations in Question-Answering (QA) tasks is an emerging paradigm aimed at enhancing the verifiability of their responses when LLMs are utilizing external references to generate an answer. However, there is currently no unified framework to standardize and fairly compare different citation generation methods, leading to difficulties in reproducing different methods and a comprehensive assessment. To cope with the problems above, we introduce \name, an open-source and modular toolkit designed to facilitate the implementation and evaluation of existing citation generation methods, while also fostering the development of new approaches to improve citation quality in LLM outputs. This tool is highly extensible, allowing users to utilize 4 main modules and 14 components to construct a pipeline, evaluating an existing method or innovative designs. Our experiments with two state-of-the-art LLMs and 11 citation generation baselines demonstrate varying strengths of different modules in answer accuracy and citation quality improvement, as well as the challenge of enhancing granularity. Based on our analysis of the effectiveness of components, we propose a new method, self-RAG \snippet, obtaining a balanced answer accuracy and citation quality. Citekit is released at https://github.com/SjJ1017/Citekit.
Authors: Xin Sun, Xiao Tang, Abdallah El Ali, Zhuying Li, Pengjie Ren, Jan de Wit, Jiahuan Pei, Jos A. Bosch
Abstract: Recent advancements in large language models (LLMs) have shown promise in generating psychotherapeutic dialogues, particularly in the context of motivational interviewing (MI). However, the inherent lack of transparency in LLM outputs presents significant challenges given the sensitive nature of psychotherapy. Applying MI strategies, a set of MI skills, to generate more controllable therapeutic-adherent conversations with explainability provides a possible solution. In this work, we explore the alignment of LLMs with MI strategies by first prompting the LLMs to predict the appropriate strategies as reasoning and then utilizing these strategies to guide the subsequent dialogue generation. We seek to investigate whether such alignment leads to more controllable and explainable generations. Multiple experiments including automatic and human evaluations are conducted to validate the effectiveness of MI strategies in aligning psychotherapy dialogue generation. Our findings demonstrate the potential of LLMs in producing strategically aligned dialogues and suggest directions for practical applications in psychotherapeutic settings.
Authors: Yiming Liang, Ge Zhang, Xingwei Qu, Tianyu Zheng, Jiawei Guo, Xinrun Du, Zhenzhu Yang, Jiaheng Liu, Chenghua Lin, Lei Ma, Wenhao Huang, Jiajun Zhang
Abstract: Large Language Models (LLMs) have achieved significant advancements, however, the common learning paradigm treats LLMs as passive information repositories, neglecting their potential for active learning and alignment. Some approaches train LLMs using their own generated synthetic data, exploring the possibility of active alignment. However, there is still a huge gap between these one-time alignment methods and the continuous automatic alignment of humans. In this paper, we introduce \textbf{I-SHEEP}, an \textbf{I}terative \textbf{S}elf-En\textbf{H}anc\textbf{E}m\textbf{E}nt \textbf{P}aradigm.This human-like paradigm enables LLMs to \textbf{continuously self-align from scratch with nothing}. Compared to the one-time alignment method Dromedary \cite{sun2023principledriven}, which refers to the first iteration in this paper, I-SHEEP can significantly enhance capacities on both Qwen and Llama models. I-SHEEP achieves a maximum relative improvement of 78.2\% in the Alpaca Eval, 24.0\% in the MT Bench, and an absolute increase of 8.88\% in the IFEval accuracy over subsequent iterations in Qwen-1.5 72B model. Additionally, I-SHEEP surpasses the base model in various standard benchmark generation tasks, achieving an average improvement of 24.77\% in code generation tasks, 12.04\% in TrivialQA, and 20.29\% in SQuAD. We also provide new insights based on the experiment results. Our codes, datasets, and models are available at \textbf{https://anonymous.4open.science/r/I-SHEEP}.
Authors: Jiaang Li, Quan Wang, Zhongnan Wang, Yongdong Zhang, Zhendong Mao
Abstract: Large language models (LLMs) require model editing to efficiently update specific knowledge within them and avoid factual errors. Most model editing methods are solely designed for single-time use and result in a significant forgetting effect in lifelong editing scenarios, where sequential edits are conducted over time. Previous approaches manage sequential edits by freezing original parameters and discretely allocating new parameters for each knowledge update. However, these methods lack robustness to minor input variations due to the discrete mapping between data and parameters. To overcome this challenge, we propose ELDER, a novel approach to create a continuous association between data and adapters. ELDER integrates multiple LoRAs through a router network and is trained to establish a smooth data-adapter association, thereby enhancing the edit robustness and generalization of semantically equivalent inputs. To ensure inputs containing the same knowledge will be processed by the same LoRAs, we design a novel loss to guide the model link LoRA allocations with edit knowledge. Furthermore, we propose a deferral mechanism to retain the original LLM capabilities post-edit. Extensive experiments on GPT-2 XL and LLaMA2-7B demonstrate that ELDER effectively edits models in the lifelong setting, outperforming eight baselines while exhibiting strong scalability and preserving LLMs' general abilities on downstream tasks.
Authors: Abu Ubaida Akash, Ahmed Fahmy, Amine Trabelsi
Abstract: Stance detection (SD) identifies the text position towards a target, typically labeled as favor, against, or none. We introduce Open-Target Stance Detection (OTSD), the most realistic task where targets are neither seen during training nor provided as input. We evaluate Large Language Models (LLMs) from GPT, Gemini, Llama, and Mistral families, comparing their performance to the only existing work, Target-Stance Extraction (TSE), which benefits from predefined targets. Unlike TSE, OTSD removes the dependency of a predefined list, making target generation and evaluation more challenging. We also provide a metric for evaluating target quality that correlates well with human judgment. Our experiments reveal that LLMs outperform TSE in target generation, both when the real target is explicitly and not explicitly mentioned in the text. Similarly, LLMs overall surpass TSE in stance detection for both explicit and non-explicit cases. However, LLMs struggle in both target generation and stance detection when the target is not explicit.
Authors: Dingjie Song, Wenjun Wang, Shunian Chen, Xidong Wang, Michael Guan, Benyou Wang
Abstract: The rapid advancement of Multimodal Large Language Models (MLLMs) has led to remarkable performances across various domains. However, this progress is accompanied by a substantial surge in the resource consumption of these models. We address this pressing issue by introducing a new approach, Token Reduction using CLIP Metric (TRIM), aimed at improving the efficiency of MLLMs without sacrificing their performance. Inspired by human attention patterns in Visual Question Answering (VQA) tasks, TRIM presents a fresh perspective on the selection and reduction of image tokens. The TRIM method has been extensively tested across 12 datasets, and the results demonstrate a significant reduction in computational overhead while maintaining a consistent level of performance. This research marks a critical stride in efficient MLLM development, promoting greater accessibility and sustainability of high-performing models.
Authors: Hiroaki Yamagiwa, Hidetoshi Shimodaira
Abstract: Contextualized embeddings vary by context, even for the same token, and form a distribution in the embedding space. To analyze this distribution, we focus on the norm of the mean embedding and the variance of the embeddings. In this study, we first demonstrate that these values follow the well-known formula for variance in statistics and provide an efficient sequential computation method. Then, by observing embeddings from intermediate layers of several Transformer models, we found a strong trade-off relationship between the norm and the variance: as the mean embedding becomes closer to the origin, the variance increases. This trade-off is likely influenced by the layer normalization mechanism used in Transformer models. Furthermore, when the sets of token embeddings are treated as clusters, we show that the variance of the entire embedding set can theoretically be decomposed into the within-cluster variance and the between-cluster variance. We found experimentally that as the layers of Transformer models deepen, the embeddings move farther from the origin, the between-cluster variance relatively decreases, and the within-cluster variance relatively increases. These results are consistent with existing studies on the anisotropy of the embedding spaces across layers.
Authors: Ningyuan Xi, Xiaoyu Wang, Yetao Wu, Teng Chen, Qingqing Gu, Yue Zhao, Jinxian Qu, Zhonglin Jiang, Yong Chen, Luo Ji
Abstract: Large Language Model can reasonably understand and generate human expressions but may lack of thorough thinking and reasoning mechanisms. Recently there have been several studies which enhance the thinking ability of language models but most of them are not data-driven or training-based. In this paper, we are motivated by the cognitive mechanism in the natural world, and design a novel model architecture called TaS which allows it to first consider the thoughts and then express the response based upon the query. We design several pipelines to annotate or generate the thought contents from prompt-response samples, then add language heads in a middle layer which behaves as the thinking layer. We train the language model by the thoughts-augmented data and successfully let the thinking layer automatically generate reasonable thoughts and finally output more reasonable responses. Both qualitative examples and quantitative results validate the effectiveness and performance of TaS. Our code is available at https://anonymous.4open.science/r/TadE.
Authors: Jiaming Zhou, Abbas Ghaddar, Ge Zhang, Liheng Ma, Yaochen Hu, Soumyasundar Pal, Mark Coates, Bin Wang, Yingxue Zhang, Jianye Hao
Abstract: Despite recent advances in training and prompting strategies for Large Language Models (LLMs), these models continue to face challenges with complex logical reasoning tasks that involve long reasoning chains. In this work, we explore the potential and limitations of using graph-based synthetic reasoning data as training signals to enhance LLMs' reasoning capabilities. Our extensive experiments, conducted on two established natural language reasoning tasks -- inductive reasoning and spatial reasoning -- demonstrate that supervised fine-tuning (SFT) with synthetic graph-based reasoning data effectively enhances LLMs' reasoning performance without compromising their effectiveness on other standard evaluation benchmarks.
Authors: Dongwon Jung, Qin Liu, Tenghao Huang, Ben Zhou, Muhao Chen
Abstract: Retrieval-augmented generation (RAG) improves large language models (LMs) by incorporating non-parametric knowledge through evidence retrieved from external sources. However, it often struggles to cope with inconsistent and irrelevant information that can distract the LM from its tasks, especially when multiple evidence pieces are required. While compressing the retrieved evidence with a compression model aims to address this issue, the compressed evidence may still be unfamiliar to the target model used for downstream tasks, potentially failing to utilize the evidence effectively. We propose FaviComp (Familarity-Aware Evidence Compression), a novel training-free evidence compression technique that makes retrieved evidence more familiar to the target model, while seamlessly integrating parametric knowledge from the model. Experimental results show that FaviComp consistently outperforms most recent evidence compression baselines across multiple open-domain QA datasets, improving accuracy by up to 28.1% while achieving high compression rates. Additionally, we demonstrate the effective integration of both parametric and non-parametric knowledge during evidence compression.
Authors: Anmol Mekala, Vineeth Dorna, Shreya Dubey, Abhishek Lalwani, David Koleczek, Mukund Rungta, Sadid Hasan, Elita Lobo
Abstract: Machine unlearning aims to efficiently eliminate the influence of specific training data, known as the forget set, from the model. However, existing unlearning methods for Large Language Models (LLMs) face a critical challenge: they rely solely on negative feedback to suppress responses related to the forget set, which often results in nonsensical or inconsistent outputs, diminishing model utility and posing potential privacy risks. To address this limitation, we propose a novel approach called Alternate Preference Optimization (AltPO), which combines negative feedback with in-domain positive feedback on the forget set. Additionally, we introduce new evaluation metrics to assess the quality of responses related to the forget set. Extensive experiments show that our approach not only enables effective unlearning but also avoids undesirable model behaviors while maintaining overall model performance. Our implementation can be found at https://github.com/molereddy/Alternate-Preference-Optimization.
URLs: https://github.com/molereddy/Alternate-Preference-Optimization.
Authors: Ioannis Panagiotopoulos, Giorgos Filandrianos, Maria Lymperaiou, Giorgos Stamou
Abstract: Riddle-solving requires advanced reasoning skills, pushing LLMs to engage in abstract thinking and creative problem-solving, often revealing limitations in their cognitive abilities. In this paper, we examine the riddle-solving capabilities of LLMs using a multiple-choice format, exploring how different prompting techniques impact performance on riddles that demand diverse reasoning skills. To enhance results, we introduce RISCORE (RIddle Solving with COntext REcontruciton) a novel fully automated prompting method that generates and utilizes contextually reconstructed sentence-based puzzles in conjunction with the original examples to create few-shot exemplars. Our experiments demonstrate that RISCORE significantly improves the performance of language models in both vertical and lateral thinking tasks, surpassing traditional exemplar selection strategies across a variety of few-shot settings.
Authors: Hakaze Cho, Mariko Kato, Yoshihiro Sakai, Naoya Inoue
Abstract: In-context Learning (ICL) is an emerging few-shot learning paradigm on Language Models (LMs) with inner mechanisms un-explored. There are already existing works describing the inner processing of ICL, while they struggle to capture all the inference phenomena in large language models. Therefore, this paper proposes a comprehensive circuit to model the inference dynamics and try to explain the observed phenomena of ICL. In detail, we divide ICL inference into 3 major operations: (1) Input Text Encode: LMs encode every input text (demonstrations and queries) into linear representation in the hidden states with sufficient information to solve ICL tasks. (2) Semantics Merge: LMs merge the encoded representations of demonstrations with their corresponding label tokens to produce joint representations of labels and demonstrations. (3) Feature Retrieval and Copy: LMs search the joint representations similar to the query representation on a task subspace, and copy the searched representations into the query. Then, language model heads capture these copied label representations to a certain extent and decode them into predicted labels. The proposed inference circuit successfully captured many phenomena observed during the ICL process, making it a comprehensive and practical explanation of the ICL inference process. Moreover, ablation analysis by disabling the proposed steps seriously damages the ICL performance, suggesting the proposed inference circuit is a dominating mechanism. Additionally, we confirm and list some bypass mechanisms that solve ICL tasks in parallel with the proposed circuit.
Authors: Zaid Khan, Elias Stengel-Eskin, Jaemin Cho, Mohit Bansal
Abstract: The process of creating training data to teach models is currently driven by humans, who manually analyze model weaknesses and plan how to create data that improves a student model. Approaches using LLMs as annotators reduce human effort, but still require humans to interpret feedback from evaluations and control the LLM to produce data the student needs. Automating this labor-intensive process by creating autonomous data generation agents - or teachers - is desirable, but requires environments that can simulate the feedback-driven, iterative, closed loop of data creation. To enable rapid, scalable testing for such agents and their modules, we introduce DataEnvGym, a testbed of teacher environments for data generation agents. DataEnvGym frames data generation as a sequential decision-making task, involving an agent consisting of a data generation policy (which generates a plan for creating training data) and a data generation engine (which transforms the plan into data), inside an environment that provides student feedback. The agent's goal is to improve student performance. Students are iteratively trained and evaluated on generated data, and their feedback (in the form of errors or weak skills) is reported to the agent after each iteration. DataEnvGym includes multiple teacher environment instantiations across 3 levels of structure in the state representation and action space. More structured environments are based on inferred skills and offer more interpretability and curriculum control. We support 4 domains (math, code, VQA, and tool-use) and test multiple students and teachers. Example agents in our teaching environments can iteratively improve students across tasks and settings. Moreover, we show that environments teach different skill levels and test variants of key modules, pointing to future work in improving data generation agents, engines, and feedback mechanisms.
Authors: Tarun Raheja, Nilay Pochhi, F. D. C. M. Curie
Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities in natural language processing tasks, but their vulnerability to jailbreak attacks poses significant security risks. This survey paper presents a comprehensive analysis of recent advancements in attack strategies and defense mechanisms within the field of Large Language Model (LLM) red-teaming. We analyze various attack methods, including gradient-based optimization, reinforcement learning, and prompt engineering approaches. We discuss the implications of these attacks on LLM safety and the need for improved defense mechanisms. This work aims to provide a thorough understanding of the current landscape of red-teaming attacks and defenses on LLMs, enabling the development of more secure and reliable language models.
Authors: Akriti Jain, Saransh Sharma, Koyel Mukherjee, Soumyabrata Pal
Abstract: Auto-regressive Large Language Models (LLMs) demonstrate remarkable performance across different domains such as vision and language processing. However, due to sequential processing through a stack of transformer layers, autoregressive decoding faces significant computation/latency challenges, particularly in resource-constrained environments like mobile and edge devices. Existing approaches in literature that aim to improve latency via skipping layers have two distinct flavors - 1) Early exit, and 2) Input-agnostic heuristics where tokens exit at pre-determined layers irrespective of input sequence. Both the above strategies have limitations - the former cannot be applied to handle KV Caching necessary for speed-ups in modern framework and the latter does not capture the variation in layer importance across tasks or more generally, across input sequences. To address both limitations, we propose FiRST, an algorithm that reduces inference latency by using layer-specific routers to select a subset of transformer layers adaptively for each input sequence - the prompt (during the prefill stage) decides which layers will be skipped during decoding. FiRST preserves compatibility with KV caching enabling faster inference while being quality-aware. FiRST is model-agnostic and can be easily enabled on any pre-trained LLM. Our approach reveals that input adaptivity is critical - indeed, different task-specific middle layers play a crucial role in evolving hidden representations depending on tasks. Extensive experiments show that FiRST significantly reduces latency while outperforming other layer selection strategies in quality metics. It retains competitive performance to base model (without layer skipping) and in some cases, even improves upon it. FiRST is thus a promising and efficient solution for LLM deployment in low-resource environments.
Authors: Vernon Y. H. Toh, Deepanway Ghosal, Soujanya Poria
Abstract: Large language models (LLMs) have shown increasing competence in solving mathematical reasoning problems. However, many open-source LLMs still struggle with errors in calculation and semantic understanding during intermediate reasoning steps. In this work, we introduce Prove, a simple yet effective framework that leverages translated programs derived from natural language solutions as a verification mechanism to filter out potentially incorrect reasoning paths before aggregating final answers. Unlike vanilla majority voting, our approach filters out solutions whose corresponding program output is inconsistent with the generated solution, aggregating only those that pass verification. We conducted extensive experiments using 13 open-source LLMs from various model families and sizes, ranging from 0.5B to 13B parameters, across eight mathematical benchmarks. Our results show that Prove consistently outperforms vanilla majority voting as a heuristic for solving mathematical reasoning tasks across all model sizes and datasets, achieving improvements of up to 18% on GSM8K and 8% on MATH-500. Our codes are available at https://github.com/declare-lab/prove.
Authors: Zonghai Yao, Aditya Parashar, Huixue Zhou, Won Seok Jang, Feiyun Ouyang, Zhichao Yang, Hong Yu
Abstract: Automatic question generation (QG) is essential for AI and NLP, particularly in intelligent tutoring, dialogue systems, and fact verification. Generating multiple-choice questions (MCQG) for professional exams, like the United States Medical Licensing Examination (USMLE), is particularly challenging, requiring domain expertise and complex multi-hop reasoning for high-quality questions. However, current large language models (LLMs) like GPT-4 struggle with professional MCQG due to outdated knowledge, hallucination issues, and prompt sensitivity, resulting in unsatisfactory quality and difficulty. To address these challenges, we propose MCQG-SRefine, an LLM self-refine-based (Critique and Correction) framework for converting medical cases into high-quality USMLE-style questions. By integrating expert-driven prompt engineering with iterative self-critique and self-correction feedback, MCQG-SRefine significantly enhances human expert satisfaction regarding both the quality and difficulty of the questions. Furthermore, we introduce an LLM-as-Judge-based automatic metric to replace the complex and costly expert evaluation process, ensuring reliable and expert-aligned assessments.
Authors: Zhuoming Chen, Ranajoy Sadhukhan, Zihao Ye, Yang Zhou, Jianyu Zhang, Niklas Nolte, Yuandong Tian, Matthijs Douze, Leon Bottou, Zhihao Jia, Beidi Chen
Abstract: Large language models (LLMs) with long context windows have gained significant attention. However, the KV cache, stored to avoid re-computation, becomes a bottleneck. Various dynamic sparse or TopK-based attention approximation methods have been proposed to leverage the common insight that attention is sparse. In this paper, we first show that TopK attention itself suffers from quality degradation in certain downstream tasks because attention is not always as sparse as expected. Rather than selecting the keys and values with the highest attention scores, sampling with theoretical guarantees can provide a better estimation for attention output. To make the sampling-based approximation practical in LLM generation, we propose MagicPIG, a heterogeneous system based on Locality Sensitive Hashing (LSH). MagicPIG significantly reduces the workload of attention computation while preserving high accuracy for diverse tasks. MagicPIG stores the LSH hash tables and runs the attention computation on the CPU, which allows it to serve longer contexts and larger batch sizes with high approximation accuracy. MagicPIG can improve decoding throughput by up to $5\times$ across various GPU hardware and achieve 54ms decoding latency on a single RTX 4090 for Llama-3.1-8B-Instruct model with a context of 96k tokens. The code is available at https://github.com/Infini-AI-Lab/MagicPIG.
Authors: Ji Ma
Abstract: As Large Language Model (LLM)-based agents increasingly undertake real-world tasks and engage with human society, how well do we understand their behaviors? We (1) investigate how LLM agents' prosocial behaviors -- a fundamental social norm -- can be induced by different personas and benchmarked against human behaviors; and (2) introduce a behavioral and social science approach to evaluate LLM agents' decision-making. We explored how different personas and experimental framings affect these AI agents' altruistic behavior in dictator games and compared their behaviors within the same LLM family, across various families, and with human behaviors. The findings reveal substantial variations and inconsistencies among LLMs and notable differences compared to human behaviors. Merely assigning a human-like identity to LLMs does not produce human-like behaviors. Despite being trained on extensive human-generated data, these AI agents are unable to capture the internal processes of human decision-making. Their alignment with human is highly variable and dependent on specific model architectures and prompt formulations; even worse, such dependence does not follow a clear pattern. LLMs can be useful task-specific tools but are not yet intelligent human-like agents.
Authors: Keyu Chen, Cheng Fei, Ziqian Bi, Junyu Liu, Benji Peng, Sen Zhang, Xuanhe Pan, Jiawei Xu, Jinlang Wang, Caitlyn Heqi Yin, Yichao Zhang, Pohsun Feng, Yizhu Wen, Tianyang Wang, Ming Li, Jintao Ren, Qian Niu, Silin Chen, Weiche Hsieh, Lawrence K. Q. Yan, Chia Xin Liang, Han Xu, Hong-Ming Tseng, Xinyuan Song, Ming Liu
Abstract: With a focus on natural language processing (NLP) and the role of large language models (LLMs), we explore the intersection of machine learning, deep learning, and artificial intelligence. As artificial intelligence continues to revolutionize fields from healthcare to finance, NLP techniques such as tokenization, text classification, and entity recognition are essential for processing and understanding human language. This paper discusses advanced data preprocessing techniques and the use of frameworks like Hugging Face for implementing transformer-based models. Additionally, it highlights challenges such as handling multilingual data, reducing bias, and ensuring model robustness. By addressing key aspects of data processing and model fine-tuning, this work aims to provide insights into deploying effective and ethically sound AI solutions.
Authors: Chuyi Kong, Ziyang Luo, Hongzhan Lin, Zhiyuan Fan, Yaxin Fan, Yuxi Sun, Jing Ma
Abstract: The advanced role-playing capabilities of Large Language Models (LLMs) have paved the way for developing Role-Playing Agents (RPAs). However, existing benchmarks in social interaction such as HPD and SocialBench have not investigated hallucination and face limitations like poor generalizability and implicit judgments for character fidelity. To address these issues, we propose a generalizable, explicit and effective paradigm to unlock the interactive patterns in diverse worldviews. Specifically, we define the interactive hallucination based on stance transfer and construct a benchmark, SHARP, by extracting relations from a general commonsense knowledge graph and leveraging the inherent hallucination properties of RPAs to simulate interactions across roles. Extensive experiments validate the effectiveness and stability of our paradigm. Our findings further explore the factors influencing these metrics and discuss the trade-off between blind loyalty to roles and adherence to facts in RPAs.
Authors: Xiaohao Yang, He Zhao, Weijie Xu, Yuanyuan Qi, Jueqing Lu, Dinh Phung, Lan Du
Abstract: Topic modeling is a fundamental task in natural language processing, allowing the discovery of latent thematic structures in text corpora. While Large Language Models (LLMs) have demonstrated promising capabilities in topic discovery, their direct application to topic modeling suffers from issues such as incomplete topic coverage, misalignment of topics, and inefficiency. To address these limitations, we propose LLM-ITL, a novel LLM-in-the-loop framework that integrates LLMs with Neural Topic Models (NTMs). In LLM-ITL, global topics and document representations are learned through the NTM. Meanwhile, an LLM refines these topics using an Optimal Transport (OT)-based alignment objective, where the refinement is dynamically adjusted based on the LLM's confidence in suggesting topical words for each set of input words. With the flexibility of being integrated into many existing NTMs, the proposed approach enhances the interpretability of topics while preserving the efficiency of NTMs in learning topics and document representations. Extensive experiments demonstrate that LLM-ITL helps NTMs significantly improve their topic interpretability while maintaining the quality of document representation. Our code and datasets will be available at Github.
Authors: Zhu Xu, Zhiqiang Zhao, Zihan Zhang, Yuchi Liu, Quanwei Shen, Fei Liu, Yu Kuang, Jian He, Conglin Liu
Abstract: Tokenization methods like Byte-Pair Encoding (BPE) enhance computational efficiency in large language models (LLMs) but often obscure internal character structures within tokens. This limitation hinders LLMs' ability to predict precise character positions, which is crucial in tasks like Chinese Spelling Correction (CSC) where identifying the positions of misspelled characters accelerates correction processes. We propose Token Internal Position Awareness (TIPA), a method that significantly improves models' ability to capture character positions within tokens by training them on reverse character prediction tasks using the tokenizer's vocabulary. Experiments demonstrate that TIPA enhances position prediction accuracy in LLMs, enabling more precise identification of target characters in original text. Furthermore, when applied to downstream tasks that do not require exact position prediction, TIPA still boosts performance in tasks needing character-level information, validating its versatility and effectiveness.
Authors: Lingxiao Wei, He Yan, Xiangju Lu, Junmin Zhu, Jun Wang, Wei Zhang
Abstract: Large Language Models (LLMs) have been well-researched in various long-context tasks. However, the scarcity of high-quality long-context summarization datasets has hindered further advancements in this area. To address this, we introduce CNNSum, a multi-scale long-context summarization benchmark based on Chinese novels, featuring human-driven annotations, which comprises four subsets totaling 695 samples, with lengths ranging from 16k to 128k. We evaluate numerous LLMs and conduct detailed case analyses. Furthermore, we conduct extensive fine-tuning experiments to explore and improve long-context summarization. In our study: (1) Advanced LLMs like GPT-4o may still generate subjective commentary, leading to vague summaries. (2) Currently, long-context summarization mainly relies on memory ability afforded by longer context lengths. The advantages of Large LLMs are hard to utilize, thus small LLMs are the most cost-effective. (3) Different prompt templates paired with various version models may cause large performance gaps. In further fine-tuning, these can be mitigated, and the Base version models perform better. (4) LLMs with RoPE-base scaled exhibit strong extrapolation potential; using short-context data can significantly improve long-context summarization performance. However, further applying other interpolation methods requires careful selection. (5) CNNSum provides more reliable and insightful evaluation results than other benchmarks. We release CNNSum to advance future research in this field. https://github.com/CxsGhost/CNNSum
Authors: Inkit Padhi, Manish Nagireddy, Giandomenico Cornacchia, Subhajit Chaudhury, Tejaswini Pedapati, Pierre Dognin, Keerthiram Murugesan, Erik Miehling, Mart\'in Santill\'an Cooper, Kieran Fraser, Giulio Zizzo, Muhammad Zaid Hameed, Mark Purcell, Michael Desmond, Qian Pan, Zahra Ashktorab, Inge Vejsbjerg, Elizabeth M. Daly, Michael Hind, Werner Geyer, Ambrish Rawat, Kush R. Varshney, Prasanna Sattigeri
Abstract: We introduce the Granite Guardian models, a suite of safeguards designed to provide risk detection for prompts and responses, enabling safe and responsible use in combination with any large language model (LLM). These models offer comprehensive coverage across multiple risk dimensions, including social bias, profanity, violence, sexual content, unethical behavior, jailbreaking, and hallucination-related risks such as context relevance, groundedness, and answer relevance for retrieval-augmented generation (RAG). Trained on a unique dataset combining human annotations from diverse sources and synthetic data, Granite Guardian models address risks typically overlooked by traditional risk detection models, such as jailbreaks and RAG-specific issues. With AUC scores of 0.871 and 0.854 on harmful content and RAG-hallucination-related benchmarks respectively, Granite Guardian is the most generalizable and competitive model available in the space. Released as open-source, Granite Guardian aims to promote responsible AI development across the community. https://github.com/ibm-granite/granite-guardian
Authors: Shuhe Wang, Shengyu Zhang, Jie Zhang, Runyi Hu, Xiaoya Li, Tianwei Zhang, Jiwei Li, Fei Wu, Guoyin Wang, Eduard Hovy
Abstract: This paper surveys research in the rapidly growing field of enhancing large language models (LLMs) with reinforcement learning (RL), a technique that enables LLMs to improve their performance by receiving feedback in the form of rewards based on the quality of their outputs, allowing them to generate more accurate, coherent, and contextually appropriate responses. In this work, we make a systematic review of the most up-to-date state of knowledge on RL-enhanced LLMs, attempting to consolidate and analyze the rapidly growing research in this field, helping researchers understand the current challenges and advancements. Specifically, we (1) detail the basics of RL; (2) introduce popular RL-enhanced LLMs; (3) review researches on two widely-used reward model-based RL techniques: Reinforcement Learning from Human Feedback (RLHF) and Reinforcement Learning from AI Feedback (RLAIF); and (4) explore Direct Preference Optimization (DPO), a set of methods that bypass the reward model to directly use human preference data for aligning LLM outputs with human expectations. We will also point out current challenges and deficiencies of existing methods and suggest some avenues for further improvements. Project page of this work can be found at: \url{https://github.com/ShuheWang1998/Reinforcement-Learning-Enhanced-LLMs-A-Survey}.
URLs: https://github.com/ShuheWang1998/Reinforcement-Learning-Enhanced-LLMs-A-Survey
Authors: Runsheng "Anson" Huang, Lara J. Martin, Chris Callison-Burch
Abstract: WHAT-IF -- Writing a Hero's Alternate Timeline through Interactive Fiction -- is a system that uses zero-shot meta-prompting to create branching narratives from a prewritten story. Played as an interactive fiction (IF) game, WHAT-IF lets the player choose between decisions that the large language model (LLM) GPT-4 generates as possible branches in the story. Starting with an existing linear plot as input, a branch is created at each key decision taken by the main character. By meta-prompting the LLM to consider the major plot points from the story, the system produces coherent and well-structured alternate storylines. WHAT-IF stores the branching plot tree in a graph which helps it to both keep track of the story for prompting and maintain the structure for the final IF system. A video demo of our system can be found here: https://youtu.be/8vBqjqtupcc.
Authors: Mengjie Qian, Kate Knill, Stefano Banno, Siyuan Tang, Penny Karanasou, Mark J. F. Gales, Diane Nicholls
Abstract: This paper presents the "Speak & Improve Challenge 2025: Spoken Language Assessment and Feedback" -- a challenge associated with the ISCA SLaTE 2025 Workshop. The goal of the challenge is to advance research on spoken language assessment and feedback, with tasks associated with both the underlying technology and language learning feedback. Linked with the challenge, the Speak & Improve (S&I) Corpus 2025 is being pre-released, a dataset of L2 learner English data with holistic scores and language error annotation, collected from open (spontaneous) speaking tests on the Speak & Improve learning platform. The corpus consists of approximately 315 hours of audio data from second language English learners with holistic scores, and a 55-hour subset with manual transcriptions and error labels. The Challenge has four shared tasks: Automatic Speech Recognition (ASR), Spoken Language Assessment (SLA), Spoken Grammatical Error Correction (SGEC), and Spoken Grammatical Error Correction Feedback (SGECF). Each of these tasks has a closed track where a predetermined set of models and data sources are allowed to be used, and an open track where any public resource may be used. Challenge participants may do one or more of the tasks. This paper describes the challenge, the S&I Corpus 2025, and the baseline systems released for the Challenge.
Authors: Kate Knill, Diane Nicholls, Mark J. F. Gales, Mengjie Qian, Pawel Stroinski
Abstract: We introduce the Speak & Improve Corpus 2025, a dataset of L2 learner English data with holistic scores and language error annotation, collected from open (spontaneous) speaking tests on the Speak & Improve learning platform. The aim of the corpus release is to address a major challenge to developing L2 spoken language processing systems, the lack of publicly available data with high-quality annotations. It is being made available for non-commercial use on the ELiT website. In designing this corpus we have sought to make it cover a wide-range of speaker attributes, from their L1 to their speaking ability, as well as providing manual annotations. This enables a range of language-learning tasks to be examined, such as assessing speaking proficiency or providing feedback on grammatical errors in a learner's speech. Additionally the data supports research into the underlying technology required for these tasks including automatic speech recognition (ASR) of low resource L2 learner English, disfluency detection or spoken grammatical error correction (GEC). The corpus consists of around 315 hours of L2 English learners audio with holistic scores, and a subset of audio annotated with transcriptions and error labels.
Authors: Ci-Siang Lin, Chien-Yi Wang, Yu-Chiang Frank Wang, Min-Hung Chen
Abstract: Weakly-Supervised Semantic Segmentation (WSSS) aims to train segmentation models using image data with only image-level supervision. Since precise pixel-level annotations are not accessible, existing methods typically focus on producing pseudo masks for training segmentation models by refining CAM-like heatmaps. However, the produced heatmaps may capture only the discriminative image regions of object categories or the associated co-occurring backgrounds. To address the issues, we propose a Semantic Prompt Learning for WSSS (SemPLeS) framework, which learns to effectively prompt the CLIP latent space to enhance the semantic alignment between the segmented regions and the target object categories. More specifically, we propose Contrastive Prompt Learning and Prompt-guided Semantic Refinement to learn the prompts that adequately describe and suppress the co-occurring backgrounds associated with each object category. In this way, SemPLeS can perform better semantic alignment between object regions and class labels, resulting in desired pseudo masks for training segmentation models. The proposed SemPLeS framework achieves competitive performance on standard WSSS benchmarks, PASCAL VOC 2012 and MS COCO2014, and shows compatibility with other WSSS methods.
Authors: Haowen Hou, Peigen Zeng, Fei Ma, Fei Richard Yu
Abstract: Visual Language Models (VLMs) have rapidly progressed with the recent success of large language models. However, there have been few attempts to incorporate efficient linear Recurrent Neural Networks (RNNs) architectures into VLMs. In this study, we introduce VisualRWKV, the first application of a linear RNN model to multimodal learning tasks, leveraging the pre-trained RWKV language model. We propose a data-dependent recurrence and sandwich prompts to enhance our modeling capabilities, along with a 2D image scanning mechanism to enrich the processing of visual sequences. Extensive experiments demonstrate that VisualRWKV achieves competitive performance compared to Transformer-based models like LLaVA-1.5 on various benchmarks. Compared to LLaVA-1.5, VisualRWKV has a speed advantage of 3.98 times and can save 54% of GPU memory when reaching an inference length of 24K tokens. To facilitate further research and analysis, we have made the checkpoints and the associated code publicly accessible at the following GitHub repository: see https://github.com/howard-hou/VisualRWKV.
Authors: Kening Zheng, Junkai Chen, Yibo Yan, Xin Zou, Xuming Hu
Abstract: Hallucination issues continue to affect multimodal large language models (MLLMs), with existing research mainly addressing object-level or attribute-level hallucinations, neglecting the more complex relation hallucinations that require advanced reasoning. Current benchmarks for relation hallucinations lack detailed evaluation and effective mitigation, and their datasets often suffer from biases due to systematic annotation processes. To address these challenges, we introduce Reefknot, a comprehensive benchmark targeting relation hallucinations, comprising over 20,000 real-world samples. We provide a systematic definition of relation hallucinations, integrating perceptive and cognitive perspectives, and construct a relation-based corpus using the Visual Genome scene graph dataset. Our comparative evaluation reveals significant limitations in current MLLMs' ability to handle relation hallucinations. Additionally, we propose a novel confidence-based mitigation strategy, which reduces the hallucination rate by an average of 9.75% across three datasets, including Reefknot. Our work offers valuable insights for achieving trustworthy multimodal intelligence.
Authors: Yuehan Zhang, Peizhuo Lv, Yinpeng Liu, Yongqiang Ma, Wei Lu, Xiaofeng Wang, Xiaozhong Liu, Jiawei Liu
Abstract: The rapid advancement of customized Large Language Models (LLMs) offers considerable convenience. However, it also intensifies concerns regarding the protection of copyright/confidential information. With the extensive adoption of private LLMs, safeguarding model copyright and ensuring data privacy have become critical. Text watermarking has emerged as a viable solution for detecting AI-generated content and protecting models. However, existing methods fall short in providing individualized watermarks for each user, a critical feature for enhancing accountability and traceability. In this paper, we introduce PersonaMark, a novel personalized text watermarking scheme designed to protect LLMs' copyrights and bolster accountability. PersonaMark leverages sentence structure as a subtle carrier of watermark information and optimizes the generation process to maintain the natural output of the model. By employing a personalized hashing function, unique watermarks are embedded for each user, enabling high-quality text generation without compromising the model's performance. This approach is both time-efficient and scalable, capable of handling large numbers of users through a multi-user hashing mechanism. To the best of our knowledge, this is a pioneer study to explore personalized watermarking in LLMs. We conduct extensive evaluations across four LLMs, analyzing various metrics such as perplexity, sentiment, alignment, and readability. The results validate that PersonaMark preserves text quality, ensures unbiased watermark insertion, and offers robust watermark detection capabilities, all while maintaining the model's behavior with minimal disruption.
Authors: Jinyi Mi, Xiaohan Shi, Ding Ma, Jiajun He, Takuya Fujimura, Tomoki Toda
Abstract: Developing a robust speech emotion recognition (SER) system in noisy conditions faces challenges posed by different noise properties. Most previous studies have not considered the impact of human speech noise, thus limiting the application scope of SER. In this paper, we propose a novel two-stage framework for the problem by cascading target speaker extraction (TSE) method and SER. We first train a TSE model to extract the speech of target speaker from a mixture. Then, in the second stage, we utilize the extracted speech for SER training. Additionally, we explore a joint training of TSE and SER models in the second stage. Our developed system achieves a 14.33% improvement in unweighted accuracy (UA) compared to a baseline without using TSE method, demonstrating the effectiveness of our framework in mitigating the impact of human speech noise. Moreover, we conduct experiments considering speaker gender, showing that our framework performs particularly well in different-gender mixture.
Authors: Fabian Paischer, Lukas Hauzenberger, Thomas Schmied, Benedikt Alkin, Marc Peter Deisenroth, Sepp Hochreiter
Abstract: Foundation models (FMs) are pre-trained on large-scale datasets and then fine-tuned on a downstream task for a specific application. The most successful and most commonly used fine-tuning method is to update the pre-trained weights via a low-rank adaptation (LoRA). LoRA introduces new weight matrices that are usually initialized at random with a uniform rank distribution across the model weights. Recent works focus on different initialization schemes or the learning of adaptive ranks during fine-tuning. Both approaches have only been investigated in isolation, resulting in slow convergence or a uniform rank distribution, in turn leading to suboptimal performance. We propose to improve LoRA by initializing the new weights in a data-driven manner by computing singular value decomposition (SVD) on minibatches of activation vectors. Then, we initialize the LoRA matrices with the obtained right-singular vectors and redistribute ranks among all weight matrices to provably store the maximum amount of information of the downstream data in the newly introduced weights. In this way, only what information to maintain or neglect during the fine-tuning process needs to be learned. We call our new method $\textbf{E}$xplained $\textbf{V}$ariance $\textbf{A}$daptation (EVA). We apply EVA to a variety of fine-tuning tasks ranging from language generation and understanding to image classification and reinforcement learning. EVA exhibits faster convergence than competitors and achieves the highest average score across a multitude of tasks per domain while reducing the number of trainable parameters through rank redistribution.
Authors: Woosung Koh, Jang Han Yoon, MinHyung Lee, Youngjin Song, Jaegwan Cho, Jaehyun Kang, Taehyeon Kim, Se-young Yun, Youngjae Yu, Bongshin Lee
Abstract: Generating high-quality charts with Large Language Models (LLMs) presents significant challenges due to limited data and the high cost of scaling through human curation. $\langle \text{instruction}, \text{data}, \text{code} \rangle$ triplets are scarce and expensive to manually curate as their creation demands technical expertise. To address this scalability challenge, we introduce a reference-free automatic feedback generator, which eliminates the need for costly human intervention. Our novel framework, C$^2$, consists of (1) an automatic feedback provider (ChartAF) and (2) a diverse, reference-free dataset (ChartUIE-8K). The results are compelling: in our first experiment, 74% of respondents strongly preferred, and 10% preferred, the results after feedback. The second post-feedback experiment demonstrates that ChartAF outperform nine baselines. Moreover, ChartUIE-8K significantly improves data diversity by increasing queries, datasets, and chart types by 5982%, 1936%, and 91%, respectively, over benchmarks. Finally, a study of LLM users revealed that 94% of participants preferred ChartUIE-8K's queries, with 93% deeming them aligned with real-world use cases. Core contributions are available as open-source at chartsquared.github.io, with ample qualitative examples.
Authors: Yingqi Gao, Yifu Liu, Xiaoxia Li, Xiaorong Shi, Yin Zhu, Yiming Wang, Shiqi Li, Wei Li, Yuntao Hong, Zhiling Luo, Jinyang Gao, Liyu Mou, Yu Li
Abstract: To tackle the challenges of large language model performance in natural language to SQL tasks, we introduce XiYan-SQL, an innovative framework that employs a multi-generator ensemble strategy to improve candidate generation. We introduce M-Schema, a semi-structured schema representation method designed to enhance the understanding of database structures. To enhance the quality and diversity of generated candidate SQL queries, XiYan-SQL integrates the significant potential of in-context learning (ICL) with the precise control of supervised fine-tuning. On one hand, we propose a series of training strategies to fine-tune models to generate high-quality candidates with diverse preferences. On the other hand, we implement the ICL approach with an example selection method based on named entity recognition to prevent overemphasis on entities. The refiner optimizes each candidate by correcting logical or syntactical errors. To address the challenge of identifying the best candidate, we fine-tune a selection model to distinguish nuances of candidate SQL queries. The experimental results on multiple dialect datasets demonstrate the robustness of XiYan-SQL in addressing challenges across different scenarios. Overall, our proposed XiYan-SQL achieves the state-of-the-art execution accuracy of 75.63% on Bird benchmark, 89.65% on the Spider test set, 69.86% on SQL-Eval, 41.20% on NL2GQL. The proposed framework not only enhances the quality and diversity of SQL queries but also outperforms previous methods.
Authors: Jiacheng Liang, Zian Wang, Lauren Hong, Shouling Ji, Ting Wang
Abstract: Various watermarking methods (``watermarkers'') have been proposed to identify LLM-generated texts; yet, due to the lack of unified evaluation platforms, many critical questions remain under-explored: i) What are the strengths/limitations of various watermarkers, especially their attack robustness? ii) How do various design choices impact their robustness? iii) How to optimally operate watermarkers in adversarial environments? To fill this gap, we systematize existing LLM watermarkers and watermark removal attacks, mapping out their design spaces. We then develop WaterPark, a unified platform that integrates 10 state-of-the-art watermarkers and 12 representative attacks. More importantly, by leveraging WaterPark, we conduct a comprehensive assessment of existing watermarkers, unveiling the impact of various design choices on their attack robustness. We further explore the best practices to operate watermarkers in adversarial environments. We believe our study sheds light on current LLM watermarking techniques while WaterPark serves as a valuable testbed to facilitate future research.
Authors: Zun Wang, Jialu Li, Han Lin, Jaehong Yoon, Mohit Bansal
Abstract: Storytelling video generation (SVG) has recently emerged as a task to create long, multi-motion, multi-scene videos that consistently represent the story described in the input text script. SVG holds great potential for diverse content creation in media and entertainment; however, it also presents significant challenges: (1) objects must exhibit a range of fine-grained, complex motions, (2) multiple objects need to appear consistently across scenes, and (3) subjects may require multiple motions with seamless transitions within a single scene. To address these challenges, we propose DreamRunner, a novel story-to-video generation method: First, we structure the input script using a large language model (LLM) to facilitate both coarse-grained scene planning as well as fine-grained object-level layout and motion planning. Next, DreamRunner presents retrieval-augmented test-time adaptation to capture target motion priors for objects in each scene, supporting diverse motion customization based on retrieved videos, thus facilitating the generation of new videos with complex, scripted motions. Lastly, we propose a novel spatial-temporal region-based 3D attention and prior injection module SR3AI for fine-grained object-motion binding and frame-by-frame semantic control. We compare DreamRunner with various SVG baselines, demonstrating state-of-the-art performance in character consistency, text alignment, and smooth transitions. Additionally, DreamRunner exhibits strong fine-grained condition-following ability in compositional text-to-video generation, significantly outperforming baselines on T2V-ComBench. Finally, we validate DreamRunner's robust ability to generate multi-object interactions with qualitative examples.
Authors: Chaoyun Zhang, Shilin He, Jiaxu Qian, Bowen Li, Liqun Li, Si Qin, Yu Kang, Minghua Ma, Guyue Liu, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang, Qi Zhang
Abstract: GUIs have long been central to human-computer interaction, providing an intuitive and visually-driven way to access and interact with digital systems. The advent of LLMs, particularly multimodal models, has ushered in a new era of GUI automation. They have demonstrated exceptional capabilities in natural language understanding, code generation, and visual processing. This has paved the way for a new generation of LLM-brained GUI agents capable of interpreting complex GUI elements and autonomously executing actions based on natural language instructions. These agents represent a paradigm shift, enabling users to perform intricate, multi-step tasks through simple conversational commands. Their applications span across web navigation, mobile app interactions, and desktop automation, offering a transformative user experience that revolutionizes how individuals interact with software. This emerging field is rapidly advancing, with significant progress in both research and industry. To provide a structured understanding of this trend, this paper presents a comprehensive survey of LLM-brained GUI agents, exploring their historical evolution, core components, and advanced techniques. We address research questions such as existing GUI agent frameworks, the collection and utilization of data for training specialized GUI agents, the development of large action models tailored for GUI tasks, and the evaluation metrics and benchmarks necessary to assess their effectiveness. Additionally, we examine emerging applications powered by these agents. Through a detailed analysis, this survey identifies key research gaps and outlines a roadmap for future advancements in the field. By consolidating foundational knowledge and state-of-the-art developments, this work aims to guide both researchers and practitioners in overcoming challenges and unlocking the full potential of LLM-brained GUI agents.
Authors: Wenxuan Huang, Zijie Zhai, Yunhang Shen, Shaosheng Cao, Fei Zhao, Xiangfeng Xu, Zheyu Ye, Shaohui Lin
Abstract: Multimodal Large Language Models (MLLMs) have achieved remarkable success in vision understanding, reasoning, and interaction. However, the inference computation and memory increase progressively with the generation of output tokens during decoding, directly affecting the efficacy of MLLMs. Existing methods attempt to reduce the vision context redundancy to achieve efficient MLLMs. Unfortunately, the efficiency benefits of the vision context reduction in the prefill stage gradually diminish during the decoding stage. To address this problem, we proposed a dynamic vision-language context sparsification framework Dynamic-LLaVA, which dynamically reduces the redundancy of vision context in the prefill stage and decreases the memory and computation overhead of the generated language context during decoding. Dynamic-LLaVA designs a tailored sparsification inference scheme for different inference modes, i.e., prefill, decoding with and without KV cache, to achieve efficient inference of MLLMs. In practice, Dynamic-LLaVA can reduce computation consumption by $\sim$75\% in the prefill stage. Meanwhile, throughout the entire generation process of MLLMs, Dynamic-LLaVA reduces the $\sim$50\% computation consumption under decoding without KV cache, while saving $\sim$50\% GPU memory overhead when decoding with KV cache, due to the vision-language context sparsification. Extensive experiments also demonstrate that Dynamic-LLaVA achieves efficient inference for MLLMs with negligible understanding and generation ability degradation or even performance gains compared to the full-context inference baselines. Code is available at https://github.com/Osilly/dynamic_llava .
Authors: Tian Qin, Naomi Saphra, David Alvarez-Melis
Abstract: Language models (LMs), like other neural networks, often favor shortcut heuristics based on surface-level patterns. Although LMs behave like n-gram models early in training, they must eventually learn hierarchical syntactic representations to correctly apply grammatical rules out-of-distribution (OOD). In this work, we use case studies of English grammar to explore how complex, diverse training data drives models to generalize OOD. We construct a framework that unifies our understanding of random variation with training dynamics, rule selection with memorization, and data diversity with complexity. We show that these factors are nuanced, and that intermediate levels of diversity and complexity lead to inconsistent behavior across random seeds and to unstable training dynamics. Our findings emphasize the critical role of training data in shaping generalization patterns and illuminate how competing model strategies lead to inconsistent generalization outcomes across random seeds. Code is available at https://github.com/sunnytqin/concept_comp.git.
Authors: Rainer Muehlhoff, Marte Henningsen
Abstract: [Study in German language.] This study examines the AI-powered grading tool "AI Grading Assistant" by the German company Fobizz, designed to support teachers in evaluating and providing feedback on student assignments. Against the societal backdrop of an overburdened education system and rising expectations for artificial intelligence as a solution to these challenges, the investigation evaluates the tool's functional suitability through two test series. The results reveal significant shortcomings: The tool's numerical grades and qualitative feedback are often random and do not improve even when its suggestions are incorporated. The highest ratings are achievable only with texts generated by ChatGPT. False claims and nonsensical submissions frequently go undetected, while the implementation of some grading criteria is unreliable and opaque. Since these deficiencies stem from the inherent limitations of large language models (LLMs), fundamental improvements to this or similar tools are not immediately foreseeable. The study critiques the broader trend of adopting AI as a quick fix for systemic problems in education, concluding that Fobizz's marketing of the tool as an objective and time-saving solution is misleading and irresponsible. Finally, the study calls for systematic evaluation and subject-specific pedagogical scrutiny of the use of AI tools in educational contexts.
Authors: Ahmed Rashed, Abdelkrim Kallich, Mohamed Eltayeb
Abstract: Machine learning (ML) algorithms play a crucial role in decision making across diverse fields such as healthcare, finance, education, and law enforcement. Despite their widespread adoption, these systems raise ethical and social concerns due to potential biases and fairness issues. This study focuses on evaluating and improving the fairness of Computer Vision and Natural Language Processing (NLP) models applied to unstructured datasets, emphasizing how biased predictions can reinforce existing systemic inequalities. A publicly available dataset from Kaggle was utilized to simulate a practical scenario for examining fairness in ML workflows. To address and mitigate biases, the study employed two leading fairness libraries: Fairlearn by Microsoft, and AIF360 by IBM. These tools offer comprehensive frameworks for fairness analysis, including metrics evaluation, result visualization, and bias mitigation techniques. The research aims to measure bias levels in ML models, compare the effectiveness of these fairness libraries, and provide actionable recommendations for practitioners. The results demonstrate that each library possesses distinct strengths and limitations in evaluating and mitigating fairness. By systematically analyzing these tools, the study contributes valuable insights to the growing field of ML fairness, offering practical guidance for integrating fairness solutions into real world applications. This research underscores the importance of building more equitable and responsible machine learning systems.
Authors: Qi Sun, Pengfei Hong, Tej Deep Pala, Vernon Toh, U-Xuan Tan, Deepanway Ghosal, Soujanya Poria
Abstract: Traditional reinforcement learning-based robotic control methods are often task-specific and fail to generalize across diverse environments or unseen objects and instructions. Visual Language Models (VLMs) demonstrate strong scene understanding and planning capabilities but lack the ability to generate actionable policies tailored to specific robotic embodiments. To address this, Visual-Language-Action (VLA) models have emerged, yet they face challenges in long-horizon spatial reasoning and grounded task planning. In this work, we propose the Embodied Multimodal Action Model with Grounded Chain of Thought and Look-ahead Spatial Reasoning, Emma-X. Emma-X leverages our constructed hierarchical embodiment dataset based on BridgeV2, containing 60,000 robot manipulation trajectories auto-annotated with grounded task reasoning and spatial guidance. Additionally, we introduce a trajectory segmentation strategy based on gripper states and motion trajectories, which can help mitigate hallucination in grounding subtask reasoning generation. Experimental results demonstrate that Emma-X achieves superior performance over competitive baselines, particularly in real-world robotic tasks requiring spatial reasoning.