Authors: Martins Samuel Dogo
Abstract: Mis- and disinformation, commonly collectively called fake news, continue to menace society. Perhaps, the impact of this age-old problem is presently most plain in politics and healthcare. However, fake news is affecting an increasing number of domains. It takes many different forms and continues to shapeshift as technology advances. Though it arguably most widely spreads in textual form, e.g., through social media posts and blog articles. Thus, it is imperative to thwart the spread of textual misinformation, which necessitates its initial detection. This thesis contributes to the creation of representations that are useful for detecting misinformation. Firstly, it develops a novel method for extracting textual features from news articles for misinformation detection. These features harness the disparity between the thematic coherence of authentic and false news stories. In other words, the composition of themes discussed in both groups significantly differs as the story progresses. Secondly, it demonstrates the effectiveness of topic features for fake news detection, using classification and clustering. Clustering is particularly useful because it alleviates the need for a labelled dataset, which can be labour-intensive and time-consuming to amass. More generally, it contributes towards a better understanding of misinformation and ways of detecting it using Machine Learning and Natural Language Processing.
Authors: Liang Chen, Zekun Wang, Shuhuai Ren, Lei Li, Haozhe Zhao, Yunshui Li, Zefan Cai, Hongcheng Guo, Lei Zhang, Yizhe Xiong, Yichi Zhang, Ruoyu Wu, Qingxiu Dong, Ge Zhang, Jian Yang, Lingwei Meng, Shujie Hu, Yulong Chen, Junyang Lin, Shuai Bai, Andreas Vlachos, Xu Tan, Minjia Zhang, Wen Xiao, Aaron Yee, Tianyu Liu, Baobao Chang
Abstract: Building on the foundations of language modeling in natural language processing, Next Token Prediction (NTP) has evolved into a versatile training objective for machine learning tasks across various modalities, achieving considerable success. As Large Language Models (LLMs) have advanced to unify understanding and generation tasks within the textual modality, recent research has shown that tasks from different modalities can also be effectively encapsulated within the NTP framework, transforming the multimodal information into tokens and predict the next one given the context. This survey introduces a comprehensive taxonomy that unifies both understanding and generation within multimodal learning through the lens of NTP. The proposed taxonomy covers five key aspects: Multimodal tokenization, MMNTP model architectures, unified task representation, datasets \& evaluation, and open challenges. This new taxonomy aims to aid researchers in their exploration of multimodal intelligence. An associated GitHub repository collecting the latest papers and repos is available at https://github.com/LMM101/Awesome-Multimodal-Next-Token-Prediction
URLs: https://github.com/LMM101/Awesome-Multimodal-Next-Token-Prediction
Authors: Guangyao Dou
Abstract: Pre-trained Large Language Models (LLMs) have demonstrated remarkable capabilities but also pose risks by learning and generating copyrighted material, leading to significant legal and ethical concerns. In a potential real-world scenario, model owners may need to continuously address copyright infringement in order to address requests for content removal that emerge at different time points. One potential way of addressing this is via sequential unlearning, where copyrighted content is removed sequentially as new requests arise. Despite its practical relevance, sequential unlearning in the context of copyright infringement has not been rigorously explored in existing literature. To address this gap, we propose Stable Sequential Unlearning (SSU), a novel framework designed to unlearn copyrighted content from LLMs over multiple time steps. Our approach works by identifying and removing specific weight updates in the model's parameters that correspond to copyrighted content using task vectors. We improve unlearning efficacy by introducing random labeling loss and ensuring the model retains its general-purpose knowledge by adjusting targeted parameters with gradient-based weight saliency. Extensive experimental results show that SSU sometimes achieves an effective trade-off between unlearning efficacy and general-purpose language abilities, outperforming existing baselines, but it's not a cure-all for unlearning copyrighted material.
Authors: Tairan Fu, Raquel Ferrando, Javier Conde, Carlos Arriaga, Pedro Reviriego
Abstract: Large Language Models (LLMs) have achieved unprecedented performance on many complex tasks, being able, for example, to answer questions on almost any topic. However, they struggle with other simple tasks, such as counting the occurrences of letters in a word, as illustrated by the inability of many LLMs to count the number of "r" letters in "strawberry". Several works have studied this problem and linked it to the tokenization used by LLMs, to the intrinsic limitations of the attention mechanism, or to the lack of character-level training data. In this paper, we conduct an experimental study to evaluate the relations between the LLM errors when counting letters with 1) the frequency of the word and its components in the training dataset and 2) the complexity of the counting operation. We present a comprehensive analysis of the errors of LLMs when counting letter occurrences by evaluating a representative group of models over a large number of words. The results show a number of consistent trends in the models evaluated: 1) models are capable of recognizing the letters but not counting them; 2) the frequency of the word and tokens in the word does not have a significant impact on the LLM errors; 3) there is a positive correlation of letter frequency with errors, more frequent letters tend to have more counting errors, 4) the errors show a strong correlation with the number of letters or tokens in a word and 5) the strongest correlation occurs with the number of letters with counts larger than one, with most models being unable to correctly count words in which letters appear more than twice.
Authors: Xingyu Xiao, Peng Chen, Ben Qi, Hongru Zhao, Jingang Liang, Jiejuan Tong, Haitao Wang
Abstract: Human reliability analysis (HRA) is crucial for evaluating and improving the safety of complex systems. Recent efforts have focused on estimating human error probability (HEP), but existing methods often rely heavily on expert knowledge,which can be subjective and time-consuming. Inspired by the success of large language models (LLMs) in natural language processing, this paper introduces a novel two-stage framework for knowledge-driven reliability analysis, integrating IDHEAS and LLMs (KRAIL). This innovative framework enables the semi-automated computation of base HEP values. Additionally, knowledge graphs are utilized as a form of retrieval-augmented generation (RAG) for enhancing the framework' s capability to retrieve and process relevant data efficiently. Experiments are systematically conducted and evaluated on authoritative datasets of human reliability. The experimental results of the proposed methodology demonstrate its superior performance on base HEP estimation under partial information for reliability assessment.
Authors: Karishma Thakrar
Abstract: Graph Retrieval-Augmented Generation (GRAG or Graph RAG) architectures aim to enhance language understanding and generation by leveraging external knowledge. However, effectively capturing and integrating the rich semantic information present in textual and structured data remains a challenge. To address this, a novel GRAG framework is proposed to focus on enhancing subgraph representation and diversity within the knowledge graph. By improving graph density, capturing entity and relation information more effectively, and dynamically prioritizing relevant and diverse subgraphs, the proposed approach enables a more comprehensive understanding of the underlying semantic structure. This is achieved through a combination of de-duplication processes, two-step mean pooling of embeddings, query-aware retrieval considering unique nodes, and a Dynamic Similarity-Aware BFS (DSA-BFS) traversal algorithm. Integrating Graph Convolutional Networks (GCNs) and Large Language Models (LLMs) through hard prompting further enhances the learning of rich node and edge representations while preserving the hierarchical subgraph structure. Experimental results on multiple benchmark datasets demonstrate the effectiveness of the proposed GRAG framework, showcasing the significance of enhanced subgraph representation and diversity for improved language understanding and generation.
Authors: Laura V\'asquez-Rodr\'iguez, Nhung T. H. Nguyen, Piotr Przyby{\l}a, Matthew Shardlow, Sophia Ananiadou
Abstract: In this paper, we present the SimDoc system, a simplification model considering simplicity, readability, and discourse aspects, such as coherence. In the past decade, the progress of the Text Simplification (TS) field has been mostly shown at a sentence level, rather than considering paragraphs or documents, a setting from which most TS audiences would benefit. We propose a simplification system that is initially fine-tuned with professionally created corpora. Further, we include multiple objectives during training, considering simplicity, readability, and coherence altogether. Our contributions include the extension of professionally annotated simplification corpora by the association of existing annotations into (complex text, simple text, readability label) triples to benefit from readability during training. Also, we present a comparative analysis in which we evaluate our proposed models in a zero-shot, few-shot, and fine-tuning setting using document-level TS corpora, demonstrating novel methods for simplification. Finally, we show a detailed analysis of outputs, highlighting the difficulties of simplification at a document level.
Authors: Ratnesh Kumar Joshi, Sagnik Sengupta, Asif Ekbal
Abstract: Hallucination, a persistent challenge plaguing language models, undermines their efficacy and trustworthiness in various natural language processing endeavors by generating responses that deviate from factual accuracy or coherence. This paper addresses language model hallucination by integrating curated knowledge graph (KG) triples to anchor responses in empirical data. We meticulously select and integrate relevant KG triples tailored to specific contexts, enhancing factual grounding and alignment with input. Our contribution involves constructing a comprehensive KG repository from Wikipedia and refining data to spotlight essential information for model training. By imbuing language models with access to this curated knowledge, we aim to generate both linguistically fluent responses and deeply rooted in factual accuracy and context relevance. This integration mitigates hallucinations by providing a robust foundation of information, enabling models to draw upon a rich reservoir of factual data during response generation. Experimental evaluations demonstrate the effectiveness of multiple approaches in reducing hallucinatory responses, underscoring the role of curated knowledge graphs in improving the reliability and trustworthiness of language model outputs.
Authors: Ainesh Chatterjee, Samuel Miller, Nithin Parepally
Abstract: We introduce AgreeMate, a framework for training Large Language Models (LLMs) to perform strategic price negotiations through natural language. We apply recent advances to a negotiation setting where two agents (i.e. buyer or seller) use natural language to bargain on goods using coarse actions. Specifically, we present the performance of Large Language Models when used as agents within a decoupled (modular) bargaining architecture. We demonstrate that using prompt engineering, fine-tuning, and chain-of-thought prompting enhances model performance, as defined by novel metrics. We use attention probing to show model attention to semantic relationships between tokens during negotiations.
Authors: Yanlin Feng, Simone Papicchio, Sajjadur Rahman
Abstract: Retrieval from graph data is crucial for augmenting large language models (LLM) with both open-domain knowledge and private enterprise data, and it is also a key component in the recent GraphRAG system (edge et al., 2024). Despite decades of research on knowledge graphs and knowledge base question answering, leading LLM frameworks (e.g. Langchain and LlamaIndex) have only minimal support for retrieval from modern encyclopedic knowledge graphs like Wikidata. In this paper, we analyze the root cause and suggest that modern RDF knowledge graphs (e.g. Wikidata, Freebase) are less efficient for LLMs due to overly large schemas that far exceed the typical LLM context window, use of resource identifiers, overlapping relation types and lack of normalization. As a solution, we propose property graph views on top of the underlying RDF graph that can be efficiently queried by LLMs using Cypher. We instantiated this idea on Wikidata and introduced CypherBench, the first benchmark with 11 large-scale, multi-domain property graphs with 7.8 million entities and over 10,000 questions. To achieve this, we tackled several key challenges, including developing an RDF-to-property graph conversion engine, creating a systematic pipeline for text-to-Cypher task generation, and designing new evaluation metrics.
Authors: Si Wu, John Wieting, David A. Smith
Abstract: While a source sentence can be translated in many ways, most machine translation (MT) models are trained with only a single reference. Previous work has shown that using synthetic paraphrases can improve MT. This paper investigates best practices for employing multiple references by analyzing the semantic similarity among different English translations of world literature in the Par3 dataset. We classify the semantic similarity between paraphrases into three groups: low, medium, and high, and fine-tune two different LLMs (mT5-large and LLaMA-2-7B) for downstream MT tasks. Across different models, holding the total training instances constant, single-reference but more source texts only marginally outperforms multiple-reference with half of the source texts. Moreover, using paraphrases of medium and high semantic similarity outperforms an unfiltered dataset (+BLEU 0.3-0.5, +COMET 0.2-0.9, +chrF++ 0.25-0.32). Our code is publicly available on GitHub.
Authors: Chris Impey, Matthew Wenger, Nikhil Garuda, Shahriar Golchin, Sarah Stamer
Abstract: Assessing writing in large classes for formal or informal learners presents a significant challenge. Consequently, most large classes, particularly in science, rely on objective assessment tools such as multiple-choice quizzes, which have a single correct answer. The rapid development of AI has introduced the possibility of using large language models (LLMs) to evaluate student writing. An experiment was conducted using GPT-4 to determine if machine learning methods based on LLMs can match or exceed the reliability of instructor grading in evaluating short writing assignments on topics in astronomy. The audience consisted of adult learners in three massive open online courses (MOOCs) offered through Coursera. One course was on astronomy, the second was on astrobiology, and the third was on the history and philosophy of astronomy. The results should also be applicable to non-science majors in university settings, where the content and modes of evaluation are similar. The data comprised answers from 120 students to 12 questions across the three courses. GPT-4 was provided with total grades, model answers, and rubrics from an instructor for all three courses. In addition to evaluating how reliably the LLM reproduced instructor grades, the LLM was also tasked with generating its own rubrics. Overall, the LLM was more reliable than peer grading, both in aggregate and by individual student, and approximately matched instructor grades for all three online courses. The implication is that LLMs may soon be used for automated, reliable, and scalable grading of student science writing.
Authors: Jiacheng Hu, Xiaoxuan Liao, Jia Gao, Zhen Qi, Hongye Zheng, Chihang Wang
Abstract: This study proposes a large language model optimization method based on the improved LoRA fine-tuning algorithm, aiming to improve the accuracy and computational efficiency of the model in natural language processing tasks. We fine-tune the large language model through a low-rank adaptation strategy, which significantly reduces the consumption of computing resources while maintaining the powerful capabilities of the pre-trained model. The experiment uses the QQP task as the evaluation scenario. The results show that the improved LoRA algorithm shows significant improvements in accuracy, F1 score, and MCC compared with traditional models such as BERT, Roberta, T5, and GPT-4. In particular, in terms of F1 score and MCC, our model shows stronger robustness and discrimination ability, which proves the potential of the improved LoRA algorithm in fine-tuning large-scale pre-trained models. In addition, this paper also discusses the application prospects of the improved LoRA algorithm in other natural language processing tasks, emphasizing its advantages in multi-task learning and scenarios with limited computing resources. Future research can further optimize the LoRA fine-tuning strategy and expand its application in larger-scale pre-trained models to improve the generalization ability and task adaptability of the model.
Authors: Zhenqi Jia, Rui Liu
Abstract: Conversational Speech Synthesis (CSS) aims to effectively take the multimodal dialogue history (MDH) to generate speech with appropriate conversational prosody for target utterance. The key challenge of CSS is to model the interaction between the MDH and the target utterance. Note that text and speech modalities in MDH have their own unique influences, and they complement each other to produce a comprehensive impact on the target utterance. Previous works did not explicitly model such intra-modal and inter-modal interactions. To address this issue, we propose a new intra-modal and inter-modal context interaction scheme-based CSS system, termed III-CSS. Specifically, in the training phase, we combine the MDH with the text and speech modalities in the target utterance to obtain four modal combinations, including Historical Text-Next Text, Historical Speech-Next Speech, Historical Text-Next Speech, and Historical Speech-Next Text. Then, we design two contrastive learning-based intra-modal and two inter-modal interaction modules to deeply learn the intra-modal and inter-modal context interaction. In the inference phase, we take MDH and adopt trained interaction modules to fully infer the speech prosody of the target utterance's text content. Subjective and objective experiments on the DailyTalk dataset show that III-CSS outperforms the advanced baselines in terms of prosody expressiveness. Code and speech samples are available at https://github.com/AI-S2-Lab/I3CSS.
Authors: Xinkai Du, Quanjie Han, Chao Lv, Yan Liu, Yalin Sun, Hao Shu, Hongbo Shan, Maosong Sun
Abstract: Open-domain Question Answering (QA) has garnered substantial interest by combining the advantages of faithfully retrieved passages and relevant passages generated through Large Language Models (LLMs). However, there is a lack of definitive labels available to pair these sources of knowledge. In order to address this issue, we propose an unsupervised and simple framework called Bi-Reranking for Merging Generated and Retrieved Knowledge (BRMGR), which utilizes re-ranking methods for both retrieved passages and LLM-generated passages. We pair the two types of passages using two separate re-ranking methods and then combine them through greedy matching. We demonstrate that BRMGR is equivalent to employing a bipartite matching loss when assigning each retrieved passage with a corresponding LLM-generated passage. The application of our model yielded experimental results from three datasets, improving their performance by +1.7 and +1.6 on NQ and WebQ datasets, respectively, and obtaining comparable result on TriviaQA dataset when compared to competitive baselines.
Authors: Lei Yang, Shaoyang Xu, Deyi Xiong
Abstract: Large language models (LLMs) based on the Transformer architecture usually have their context length limited due to the high training cost. Recent advancements extend the context window by adjusting the scaling factors of RoPE and fine-tuning. However, suboptimal initialization of these factors results in increased fine-tuning costs and reduced performance at target length. To address these challenges, we propose an innovative RoPE-based fine-tuning framework that diverges from conventional scaling factors search. Specifically, we present a Divide-and-Conquer Incremental Search (DCIS) algorithm that strategically determines the better scaling factors. Further fine-tuning with the identified scaling factors effectively extends the context window of LLMs. Empirical results demonstrate that our methodology not only mitigates performance decay at extended target lengths but also allows the model to fine-tune on short contexts and generalize to long contexts, thereby reducing the cost of fine-tuning. The scaling factors obtained through DCIS can even perform effectively without fine-tuning. Further analysis of the search space reveals that DCIS achieves twice the search efficiency compared to other methods. We also examine the impact of the non-strictly increasing scaling factors utilized in DCIS and evaluate the general capabilities of LLMs across various context lengths.
Authors: Yilei Jiang, Yingshui Tan, Xiangyu Yue
Abstract: While Multimodal Large Language Models (MLLMs) have made remarkable progress in vision-language reasoning, they are also more susceptible to producing harmful content compared to models that focus solely on text. Existing defensive prompting techniques rely on a static, unified safety guideline that fails to account for the specific risks inherent in different multimodal contexts. To address these limitations, we propose RapGuard, a novel framework that uses multimodal chain-of-thought reasoning to dynamically generate scenario-specific safety prompts. RapGuard enhances safety by adapting its prompts to the unique risks of each input, effectively mitigating harmful outputs while maintaining high performance on benign tasks. Our experimental results across multiple MLLM benchmarks demonstrate that RapGuard achieves state-of-the-art safety performance, significantly reducing harmful content without degrading the quality of responses.
Authors: Liang Wang, Nan Yang, Xingxing Zhang, Xiaolong Huang, Furu Wei
Abstract: We introduce a bootstrapping approach to train long-context language models by exploiting their short-context capabilities only. Our method utilizes a simple agent workflow to synthesize diverse long-context instruction tuning data, thereby eliminating the necessity for manual data collection and annotation. The proposed data synthesis workflow requires only a short-context language model, a text retriever, and a document collection, all of which are readily accessible within the open-source ecosystem. Subsequently, language models are fine-tuned using the synthesized data to extend their context lengths. In this manner, we effectively transfer the short-context capabilities of language models to long-context scenarios through a bootstrapping process. We conduct experiments with the open-source Llama-3 family of models and demonstrate that our method can successfully extend the context length to up to 1M tokens, achieving superior performance across various benchmarks.
Authors: Meltem Aksoy
Abstract: Large language models (LLMs) have become integral tools in diverse domains, yet their moral reasoning capabilities across cultural and linguistic contexts remain underexplored. This study investigates whether multilingual LLMs, such as GPT-3.5-Turbo, GPT-4o-mini, Llama 3.1, and MistralNeMo, reflect culturally specific moral values or impose dominant moral norms, particularly those rooted in English. Using the updated Moral Foundations Questionnaire (MFQ-2) in eight languages, Arabic, Farsi, English, Spanish, Japanese, Chinese, French, and Russian, the study analyzes the models' adherence to six core moral foundations: care, equality, proportionality, loyalty, authority, and purity. The results reveal significant cultural and linguistic variability, challenging the assumption of universal moral consistency in LLMs. Although some models demonstrate adaptability to diverse contexts, others exhibit biases influenced by the composition of the training data. These findings underscore the need for culturally inclusive model development to improve fairness and trust in multilingual AI systems.
Authors: Lifeng Han, Kilian Evang, Archna Bhatia, Gosse Bouma, A. Seza Do\u{g}ru\"oz, Marcos Garcia, Voula Giouli, Joakim Nivre, Alexandre Rademacher
Abstract: Starting in 2003 when the first MWE workshop was held with ACL in Sapporo, Japan, this year, the joint workshop of MWE-UD co-located with the LREC-COLING 2024 conference marked the 20th anniversary of MWE workshop events over the past nearly two decades. Standing at this milestone, we look back to this workshop series and summarise the research topics and methodologies researchers have carried out over the years. We also discuss the current challenges that we are facing and the broader impacts/synergies of MWE research within the CL and NLP fields. Finally, we give future research perspectives. We hope this position paper can help researchers, students, and industrial practitioners interested in MWE get a brief but easy understanding of its history, current, and possible future.
Authors: JiaCheng Li
Abstract: With the explosive growth of Chinese text data and advancements in natural language processing technologies, Chinese text classification has become one of the key techniques in fields such as information retrieval and sentiment analysis, attracting increasing attention. This paper conducts a comparative study on three deep learning models:TextCNN, TextRNN, and FastText.specifically for Chinese text classification tasks. By conducting experiments on the THUCNews dataset, the performance of these models is evaluated, and their applicability in different scenarios is discussed.
Authors: Junying Chen, Zhenyang Cai, Ke Ji, Xidong Wang, Wanlong Liu, Rongsheng Wang, Jianye Hou, Benyou Wang
Abstract: The breakthrough of OpenAI o1 highlights the potential of enhancing reasoning to improve LLM. Yet, most research in reasoning has focused on mathematical tasks, leaving domains like medicine underexplored. The medical domain, though distinct from mathematics, also demands robust reasoning to provide reliable answers, given the high standards of healthcare. However, verifying medical reasoning is challenging, unlike those in mathematics. To address this, we propose verifiable medical problems with a medical verifier to check the correctness of model outputs. This verifiable nature enables advancements in medical reasoning through a two-stage approach: (1) using the verifier to guide the search for a complex reasoning trajectory for fine-tuning LLMs, (2) applying reinforcement learning (RL) with verifier-based rewards to enhance complex reasoning further. Finally, we introduce HuatuoGPT-o1, a medical LLM capable of complex reasoning, which outperforms general and medical-specific baselines using only 40K verifiable problems. Experiments show complex reasoning improves medical problem-solving and benefits more from RL. We hope our approach inspires advancements in reasoning across medical and other specialized domains.
Authors: Libo Zhang (National University of Defense Technology, Changsha, China), Zhaoning Zhang (National University of Defense Technology, Changsha, China), Baizhou Xu (National University of Defense Technology, Changsha, China), Songzhu Mei (National University of Defense Technology, Changsha, China), Dongsheng Li (National University of Defense Technology, Changsha, China)
Abstract: Due to the high resource demands of Large Language Models (LLMs), achieving widespread deployment on consumer-grade devices presents significant challenges. Typically, personal or consumer-grade devices, including servers configured prior to the era of large-scale models, generally have relatively weak GPUs and relatively strong CPUs. However, most current methods primarily depend on GPUs for computation. Therefore, we propose Dovetail, an approach that deploys the draft model on the GPU to generate draft tokens while allowing the target model to perform parallel verification on the CPU, thereby improving the utilization of all available hardware resources and occupying less inter-device communication bandwidth. Accordingly, we have redesigned the draft model to better align with heterogeneous hardware characteristics. To this end, we implemented several optimizations: reducing the number of draft tokens to mitigate latency in parallel verification, increasing the depth of the draft model to enhance its predictive capacity, and introducing DGF (Dynamic Gating Fusion) to improve the integration of features and token embeddings. In the HumanEval benchmark, Dovetail achieved an inference speed of 5.86 tokens per second for LLaMA2-Chat-7B using 3GB of VRAM, representing an approximately 2.77x improvement over CPU-only inference. Furthermore, the inference speed was increased to 8 tokens per second when utilizing 7GB of VRAM.
Authors: Kaiwen Zuo, Yirui Jiang
Abstract: Medical Large Language Models (MLLMs) have demonstrated potential in healthcare applications, yet their propensity for hallucinations -- generating medically implausible or inaccurate information -- presents substantial risks to patient care. This paper introduces MedHallBench, a comprehensive benchmark framework for evaluating and mitigating hallucinations in MLLMs. Our methodology integrates expert-validated medical case scenarios with established medical databases to create a robust evaluation dataset. The framework employs a sophisticated measurement system that combines automated ACHMI (Automatic Caption Hallucination Measurement in Medical Imaging) scoring with rigorous clinical expert evaluations and utilizes reinforcement learning methods to achieve automatic annotation. Through an optimized reinforcement learning from human feedback (RLHF) training pipeline specifically designed for medical applications, MedHallBench enables thorough evaluation of MLLMs across diverse clinical contexts while maintaining stringent accuracy standards. We conducted comparative experiments involving various models, utilizing the benchmark to establish a baseline for widely adopted large language models (LLMs). Our findings indicate that ACHMI provides a more nuanced understanding of the effects of hallucinations compared to traditional metrics, thereby highlighting its advantages in hallucination assessment. This research establishes a foundational framework for enhancing MLLMs' reliability in healthcare settings and presents actionable strategies for addressing the critical challenge of AI hallucinations in medical applications.
Authors: Ruixi Lin, Yang You
Abstract: In-context learning, which allows large language models to perform diverse tasks with a few demonstrations, is found to have imbalanced per-class prediction accuracy on multi-class text classification. Although notable output correction methods have been developed to tackle the issue and simultaneously improve downstream prediction accuracy, they may fail to answer the core interpretability challenges: why and which certain classes need corrections, and more importantly, a tailored correction for per-sample, per-class's probability. To address such interpretability gaps, we first find that the imbalance arises from certain classes consistently receiving high ICL output probabilities, whereas others receiving lower or mixed ranges, so the former is more frequently chosen, resulting in higher accuracy; more crucially, we find that these ranges have significantly varying degrees of influence on the accuracy bias, highlighting the need for precise, interpretable probability corrections by range. Motivated by this, we propose FuRud, a Fuzzy Rule Optimization based Debiasing method, that (1) detects which classes need corrections, and (2) for each correction-needed class, detects its probability ranges and applies asymmetric amplifications or reductions to correct them interpretably. Notably, across seven benchmark datasets, FuRud reduces the pairwise class accuracy bias (COBias) by more than half (56%), while achieving a relative increase of 21% in accuracy, outperforming state-of-the-art debiasing methods. Moreover, FuRud can optimize downstream tasks with as few as 10 optimization examples. Furthermore, FuRud can work for prompt formats that lead to highly skewed predictions. For example, FuRud greatly improves ICL outputs which use letter options, with 44% relative accuracy increase and 54% relative COBias reduction.
Authors: Ahmad Alfani Handoyo, Chung Tran, Dessi Puji Lestari, Sakriani Sakti
Abstract: Multilingual text-to-speech systems convert text into speech across multiple languages. In many cases, text sentences may contain segments in different languages, a phenomenon known as code-switching. This is particularly common in Indonesia, especially between Indonesian and English. Despite its significance, no research has yet developed a multilingual TTS system capable of handling code-switching between these two languages. This study addresses Indonesian-English code-switching in STEN-TTS. Key modifications include adding a language identification component to the text-to-phoneme conversion using finetuned BERT for per-word language identification, as well as removing language embedding from the base model. Experimental results demonstrate that the code-switching model achieves superior naturalness and improved speech intelligibility compared to the Indonesian and English baseline STEN-TTS models.
Authors: Tomek Rutowski, Elizabeth Shriberg, Amir Harati, Yang Lu, Ricardo Oliveira, Piotr Chlebek
Abstract: Deep learning models are rapidly gaining interest for real-world applications in behavioral health. An important gap in current literature is how well such models generalize over different populations. We study Natural Language Processing (NLP) based models to explore portability over two different corpora highly mismatched in age. The first and larger corpus contains younger speakers. It is used to train an NLP model to predict depression. When testing on unseen speakers from the same age distribution, this model performs at AUC=0.82. We then test this model on the second corpus, which comprises seniors from a retirement community. Despite the large demographic differences in the two corpora, we saw only modest degradation in performance for the senior-corpus data, achieving AUC=0.76. Interestingly, in the senior population, we find AUC=0.81 for the subset of patients whose health state is consistent over time. Implications for demographic portability of speech-based applications are discussed.
Authors: Dima Galat
Abstract: The recent proliferation of AI-generated content has prompted significant interest in developing reliable detection methods. This study explores techniques for identifying AI-generated text through sentence-level evaluation within hybrid articles. Our findings indicate that ChatGPT-3.5 Turbo exhibits distinct, repetitive probability patterns that enable consistent in-domain detection. Empirical tests show that minor textual modifications, such as rewording, have minimal impact on detection accuracy. These results provide valuable insights for advancing AI detection methodologies, offering a pathway toward robust solutions to address the complexities of synthetic text identification.
Authors: Jiawei Yu, Xiang Geng, Yuang Li, Mengxin Ren, Wei Tang, Jiahuan Li, Zhibin Lan, Min Zhang, Hao Yang, Shujian Huang, Jinsong Su
Abstract: Spoken named entity recognition (NER) aims to identify named entities from speech, playing an important role in speech processing. New named entities appear every day, however, annotating their Spoken NER data is costly. In this paper, we demonstrate that existing Spoken NER systems perform poorly when dealing with previously unseen named entities. To tackle this challenge, we propose a method for generating Spoken NER data based on a named entity dictionary (NED) to reduce costs. Specifically, we first use a large language model (LLM) to generate sentences from the sampled named entities and then use a text-to-speech (TTS) system to generate the speech. Furthermore, we introduce a noise metric to filter out noisy data. To evaluate our approach, we release a novel Spoken NER benchmark along with a corresponding NED containing 8,853 entities. Experiment results show that our method achieves state-of-the-art (SOTA) performance in the in-domain, zero-shot domain adaptation, and fully zero-shot settings. Our data will be available at https://github.com/DeepLearnXMU/HeardU.
Authors: Yunfan Zhang, Changlun Li, Yuyu Luo, Nan Tang
Abstract: Missing value is a critical issue in data science, significantly impacting the reliability of analyses and predictions. Missing value imputation (MVI) is a longstanding problem because it highly relies on domain knowledge. Large language models (LLMs) have emerged as a promising tool for data cleaning, including MVI for tabular data, offering advanced capabilities for understanding and generating content. However, despite their promise, existing LLM techniques such as in-context learning and Chain-of-Thought (CoT) often fall short in guiding LLMs to perform complex reasoning for MVI, particularly when imputing derived missing values, which require mathematical formulas and data relationships across rows and columns. This gap underscores the need for further advancements in LLM methodologies to enhance their reasoning capabilities for more reliable imputation outcomes. To fill this gap, we propose SketchFill, a novel sketch-based method to guide LLMs in generating accurate formulas to impute missing numerical values. Our experimental results demonstrate that SketchFill significantly outperforms state-of-the-art methods, achieving 56.2% higher accuracy than CoT-based methods and 78.8% higher accuracy than MetaGPT. This sets a new standard for automated data cleaning and advances the field of MVI for numerical values.
Authors: Senbin Zhu, Chenyuan He, Hongde Liu, Pengcheng Dong, Hanjie Zhao, Yuchen Yan, Yuxiang Jia, Hongying Zan, Min Peng
Abstract: In recent years, fine-grained sentiment analysis in finance has gained significant attention, but the scarcity of entity-level datasets remains a key challenge. To address this, we have constructed the largest English and Chinese financial entity-level sentiment analysis datasets to date. Building on this foundation, we propose a novel two-stage sentiment analysis approach called Self-aware In-context Learning Correction (SILC). The first stage involves fine-tuning a base large language model to generate pseudo-labeled data specific to our task. In the second stage, we train a correction model using a GNN-based example retriever, which is informed by the pseudo-labeled data. This two-stage strategy has allowed us to achieve state-of-the-art performance on the newly constructed datasets, advancing the field of financial sentiment analysis. In a case study, we demonstrate the enhanced practical utility of our data and methods in monitoring the cryptocurrency market. Our datasets and code are available at https://github.com/NLP-Bin/SILC-EFSA.
Authors: Simona Frenda, Andrea Piergentili, Beatrice Savoldi, Marco Madeddu, Martina Rosola, Silvia Casola, Chiara Ferrando, Viviana Patti, Matteo Negri, Luisa Bentivogli
Abstract: Gender-fair language aims at promoting gender equality by using terms and expressions that include all identities and avoid reinforcing gender stereotypes. Implementing gender-fair strategies is particularly challenging in heavily gender-marked languages, such as Italian. To address this, the Gender-Fair Generation challenge intends to help shift toward gender-fair language in written communication. The challenge, designed to assess and monitor the recognition and generation of gender-fair language in both mono- and cross-lingual scenarios, includes three tasks: (1) the detection of gendered expressions in Italian sentences, (2) the reformulation of gendered expressions into gender-fair alternatives, and (3) the generation of gender-fair language in automatic translation from English to Italian. The challenge relies on three different annotated datasets: the GFL-it corpus, which contains Italian texts extracted from administrative documents provided by the University of Brescia; GeNTE, a bilingual test set for gender-neutral rewriting and translation built upon a subset of the Europarl dataset; and Neo-GATE, a bilingual test set designed to assess the use of non-binary neomorphemes in Italian for both fair formulation and translation tasks. Finally, each task is evaluated with specific metrics: average of F1-score obtained by means of BERTScore computed on each entry of the datasets for task 1, an accuracy measured with a gender-neutral classifier, and a coverage-weighted accuracy for tasks 2 and 3.
Authors: Asma Ben Abacha, Wen-wai Yim, Yujuan Fu, Zhaoyi Sun, Meliha Yetisgen, Fei Xia, Thomas Lin
Abstract: Several studies showed that Large Language Models (LLMs) can answer medical questions correctly, even outperforming the average human score in some medical exams. However, to our knowledge, no study has been conducted to assess the ability of language models to validate existing or generated medical text for correctness and consistency. In this paper, we introduce MEDEC (https://github.com/abachaa/MEDEC), the first publicly available benchmark for medical error detection and correction in clinical notes, covering five types of errors (Diagnosis, Management, Treatment, Pharmacotherapy, and Causal Organism). MEDEC consists of 3,848 clinical texts, including 488 clinical notes from three US hospital systems that were not previously seen by any LLM. The dataset has been used for the MEDIQA-CORR shared task to evaluate seventeen participating systems [Ben Abacha et al., 2024]. In this paper, we describe the data creation methods and we evaluate recent LLMs (e.g., o1-preview, GPT-4, Claude 3.5 Sonnet, and Gemini 2.0 Flash) for the tasks of detecting and correcting medical errors requiring both medical knowledge and reasoning capabilities. We also conducted a comparative study where two medical doctors performed the same task on the MEDEC test set. The results showed that MEDEC is a sufficiently challenging benchmark to assess the ability of models to validate existing or generated notes and to correct medical errors. We also found that although recent LLMs have a good performance in error detection and correction, they are still outperformed by medical doctors in these tasks. We discuss the potential factors behind this gap, the insights from our experiments, the limitations of current evaluation metrics, and share potential pointers for future research.
Authors: Fangyi Chen, Gongbo Zhang, Yilu Fang, Yifan Peng, Chunhua Weng
Abstract: Objective: Extracting PICO elements -- Participants, Intervention, Comparison, and Outcomes -- from clinical trial literature is essential for clinical evidence retrieval, appraisal, and synthesis. Existing approaches do not distinguish the attributes of PICO entities. This study aims to develop a named entity recognition (NER) model to extract PICO entities with fine granularities. Materials and Methods: Using a corpus of 2,511 abstracts with PICO mentions from 4 public datasets, we developed a semi-supervised method to facilitate the training of a NER model, FinePICO, by combining limited annotated data of PICO entities and abundant unlabeled data. For evaluation, we divided the entire dataset into two subsets: a smaller group with annotations and a larger group without annotations. We then established the theoretical lower and upper performance bounds based on the performance of supervised learning models trained solely on the small, annotated subset and on the entire set with complete annotations, respectively. Finally, we evaluated FinePICO on both the smaller annotated subset and the larger, initially unannotated subset. We measured the performance of FinePICO using precision, recall, and F1. Results: Our method achieved precision/recall/F1 of 0.567/0.636/0.60, respectively, using a small set of annotated samples, outperforming the baseline model (F1: 0.437) by more than 16\%. The model demonstrates generalizability to a different PICO framework and to another corpus, which consistently outperforms the benchmark in diverse experimental settings (p-value \textless0.001). Conclusion: This study contributes a generalizable and effective semi-supervised approach to named entity recognition leveraging large unlabeled data together with small, annotated data. It also initially supports fine-grained PICO extraction.
Authors: Jiaao Chen, Diyi Yang
Abstract: We present Dynamic Skill Adaptation (DSA), an adaptive and dynamic framework to adapt novel and complex skills to Large Language Models (LLMs). Compared with previous work which learns from human-curated and static data in random orders, we propose to first automatically generate and organize the training data by mimicking the learning pathways of human and then dynamically tailor the training data based on the training dynamics. Specifically, inspired by the learning structures and teaching strategies in the human education system, we first construct a skill graph by decomposing complex skills into sub-skills and arranging them based on their dependencies in human syllables. For every skill, we utilize LLMs to generate both textbook-like data which contains detailed descriptions of skills for pre-training and exercise-like data which targets at explicitly utilizing the skills to solve problems for instruction-tuning. Furthermore, during the instruction-tuning, we dynamically update the training data which down-weight easy-to-learn examples, generate more complex examples, and filter out data with errors. Experiments on large language models such as LLAMA and Mistral demonstrate the effectiveness of our proposed methods in adapting math reasoning skills and social study skills.
Authors: DeepSeek-AI, Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dongjie Ji, Erhang Li, Fangyun Lin, Fucong Dai, Fuli Luo, Guangbo Hao, Guanting Chen, Guowei Li, H. Zhang, Han Bao, Hanwei Xu, Haocheng Wang, Haowei Zhang, Honghui Ding, Huajian Xin, Huazuo Gao, Hui Li, Hui Qu, J. L. Cai, Jian Liang, Jianzhong Guo, Jiaqi Ni, Jiashi Li, Jiawei Wang, Jin Chen, Jingchang Chen, Jingyang Yuan, Junjie Qiu, Junlong Li, Junxiao Song, Kai Dong, Kai Hu, Kaige Gao, Kang Guan, Kexin Huang, Kuai Yu, Lean Wang, Lecong Zhang, Lei Xu, Leyi Xia, Liang Zhao, Litong Wang, Liyue Zhang, Meng Li, Miaojun Wang, Mingchuan Zhang, Minghua Zhang, Minghui Tang, Mingming Li, Ning Tian, Panpan Huang, Peiyi Wang, Peng Zhang, Qiancheng Wang, Qihao Zhu, Qinyu Chen, Qiushi Du, R. J. Chen, R. L. Jin, Ruiqi Ge, Ruisong Zhang, Ruizhe Pan, Runji Wang, Runxin Xu, Ruoyu Zhang, Ruyi Chen, S. S. Li, Shanghao Lu, Shangyan Zhou, Shanhuang Chen, Shaoqing Wu, Shengfeng Ye, Shengfeng Ye, Shirong Ma, Shiyu Wang, Shuang Zhou, Shuiping Yu, Shunfeng Zhou, Shuting Pan, T. Wang, Tao Yun, Tian Pei, Tianyu Sun, W. L. Xiao, Wangding Zeng, Wanjia Zhao, Wei An, Wen Liu, Wenfeng Liang, Wenjun Gao, Wenqin Yu, Wentao Zhang, X. Q. Li, Xiangyue Jin, Xianzu Wang, Xiao Bi, Xiaodong Liu, Xiaohan Wang, Xiaojin Shen, Xiaokang Chen, Xiaokang Zhang, Xiaosha Chen, Xiaotao Nie, Xiaowen Sun, Xiaoxiang Wang, Xin Cheng, Xin Liu, Xin Xie, Xingchao Liu, Xingkai Yu, Xinnan Song, Xinxia Shan, Xinyi Zhou, Xinyu Yang, Xinyuan Li, Xuecheng Su, Xuheng Lin, Y. K. Li, Y. Q. Wang, Y. X. Wei, Y. X. Zhu, Yang Zhang, Yanhong Xu, Yanhong Xu, Yanping Huang, Yao Li, Yao Zhao, Yaofeng Sun, Yaohui Li, Yaohui Wang, Yi Yu, Yi Zheng, Yichao Zhang, Yifan Shi, Yiliang Xiong, Ying He, Ying Tang, Yishi Piao, Yisong Wang, Yixuan Tan, Yiyang Ma, Yiyuan Liu, Yongqiang Guo, Yu Wu, Yuan Ou, Yuchen Zhu, Yuduan Wang, Yue Gong, Yuheng Zou, Yujia He, Yukun Zha, Yunfan Xiong, Yunxian Ma, Yuting Yan, Yuxiang Luo, Yuxiang You, Yuxuan Liu, Yuyang Zhou, Z. F. Wu, Z. Z. Ren, Zehui Ren, Zhangli Sha, Zhe Fu, Zhean Xu, Zhen Huang, Zhen Zhang, Zhenda Xie, Zhengyan Zhang, Zhewen Hao, Zhibin Gou, Zhicheng Ma, Zhigang Yan, Zhihong Shao, Zhipeng Xu, Zhiyu Wu, Zhongyu Zhang, Zhuoshu Li, Zihui Gu, Zijia Zhu, Zijun Liu, Zilin Li, Ziwei Xie, Ziyang Song, Ziyi Gao, Zizheng Pan
Abstract: We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks. The model checkpoints are available at https://github.com/deepseek-ai/DeepSeek-V3.
Authors: Shuo Wang, Chihang Wang, Jia Gao, Zhen Qi, Hongye Zheng, Xiaoxuan Liao
Abstract: This study proposes a knowledge distillation algorithm based on large language models and feature alignment, aiming to effectively transfer the knowledge of large pre-trained models into lightweight student models, thereby reducing computational costs while maintaining high model performance. Different from the traditional soft label distillation method, this method introduces a multi-layer feature alignment strategy to deeply align the intermediate features and attention mechanisms of the teacher model and the student model, maximally retaining the semantic expression ability and context modeling ability of the teacher model. In terms of method design, a multi-task loss function is constructed, including feature matching loss, attention alignment loss, and output distribution matching loss, to ensure multi-level information transfer through joint optimization. The experiments were comprehensively evaluated on the GLUE data set and various natural language processing tasks. The results show that the proposed model performs very close to the state-of-the-art GPT-4 model in terms of evaluation indicators such as perplexity, BLEU, ROUGE, and CER. At the same time, it far exceeds baseline models such as DeBERTa, XLNet, and GPT-3, showing significant performance improvements and computing efficiency advantages. Research results show that the feature alignment distillation strategy is an effective model compression method that can significantly reduce computational overhead and storage requirements while maintaining model capabilities. Future research can be further expanded in the directions of self-supervised learning, cross-modal feature alignment, and multi-task transfer learning to provide more flexible and efficient solutions for the deployment and optimization of deep learning models.
Authors: Shiwen Ni, Hao Cheng, Min Yang
Abstract: Legal question answering (QA) has attracted increasing attention from people seeking legal advice, which aims to retrieve the most applicable answers from a large-scale database of question-answer pairs. Previous methods mainly use a dual-encoder architecture to learn dense representations of both questions and answers. However, these methods could suffer from lacking domain knowledge and sufficient labeled training data. In this paper, we propose a three-stage (\underline{p}re-training, \underline{f}ine-tuning and \underline{r}e-ranking) framework for \underline{l}egal \underline{QA} (called PFR-LQA), which promotes the fine-grained text representation learning and boosts the performance of dense retrieval with the dual-encoder architecture. Concretely, we first conduct domain-specific pre-training on legal questions and answers through a self-supervised training objective, allowing the pre-trained model to be adapted to the legal domain. Then, we perform task-specific fine-tuning of the dual-encoder on legal question-answer pairs by using the supervised learning objective, leading to a high-quality dual-encoder for the specific downstream QA task. Finally, we employ a contextual re-ranking objective to further refine the output representations of questions produced by the document encoder, which uses contextual similarity to increase the discrepancy between the anchor and hard negative samples for better question re-ranking. We conduct extensive experiments on a manually annotated legal QA dataset. Experimental results show that our PFR-LQA method achieves better performance than the strong competitors for legal question answering.
Authors: Asahi Hentona, Jun Baba, Shiki Sato, Reina Akama
Abstract: User willingness is a crucial element in the sales talk process that affects the achievement of the salesperson's or sales system's objectives. Despite the importance of user willingness, to the best of our knowledge, no previous study has addressed the development of automated sales talk dialogue systems that explicitly consider user willingness. A major barrier is the lack of sales talk datasets with reliable user willingness data. Thus, in this study, we developed a user willingness-aware sales talk collection by leveraging the ecological validity concept, which is discussed in the field of human-computer interaction. Our approach focused on three types of user willingness essential in real sales interactions. We created a dialogue environment that closely resembles real-world scenarios to elicit natural user willingness, with participants evaluating their willingness at the utterance level from multiple perspectives. We analyzed the collected data to gain insights into practical user willingness-aware sales talk strategies. In addition, as a practical application of the constructed dataset, we developed and evaluated a sales dialogue system aimed at enhancing the user's intent to purchase.
Authors: Hua Farn, Hsuan Su, Shachi H Kumar, Saurav Sahay, Shang-Tse Chen, Hung-yi Lee
Abstract: Fine-tuning large language models (LLMs) for downstream tasks is a widely adopted approach, but it often leads to safety degradation in safety-aligned LLMs. Currently, many solutions address this issue by incorporating additional safety data, which can be impractical in many cases. In this paper, we address the question: How can we improve downstream task performance while preserving safety in LLMs without relying on additional safety data? We propose a simple and effective method that maintains the inherent safety of LLMs while enhancing their downstream task performance: merging the weights of pre- and post-fine-tuned safety-aligned models. Experimental results across various downstream tasks, models, and merging methods demonstrate that this approach effectively mitigates safety degradation while improving downstream task performance, offering a practical solution for adapting safety-aligned LLMs.
Authors: Zhe Yang, Yichang Zhang, Yudong Wang, Ziyao Xu, Junyang Lin, Zhifang Sui
Abstract: Large Language Models (LLMs) can correct their self-generated responses, but a decline in accuracy after self-correction is also witnessed. To have a deeper understanding of self-correction, we endeavor to decompose, evaluate, and analyze the self-correction behaviors of LLMs. By enumerating and analyzing answer correctness before and after self-correction, we decompose the self-correction capability into confidence (being confident to correct answers) and critique (turning wrong answers to correct) capabilities, and propose two metrics from a probabilistic perspective to measure these 2 capabilities, along with another metric for overall self-correction capability evaluation. Based on our decomposition and evaluation metrics, we conduct extensive experiments and draw some empirical conclusions. For example, we find different models can exhibit distinct behaviors: some models are confident while others are more critical. We also find the trade-off between the two capabilities (i.e. improving one can lead to a decline in the other) when manipulating model self-correction behavior by prompts or in-context learning. Further, we find a simple yet efficient strategy to improve self-correction capability by transforming Supervision Fine-Tuning (SFT) data format, and our strategy outperforms vanilla SFT in both capabilities and achieves much higher accuracy after self-correction. Our code will be publicly available on GitHub.
Authors: Surangika Ranathungaa, Shravan Nayak, Shih-Ting Cindy Huang, Yanke Mao, Tong Su, Yun-Hsiang Ray Chan, Songchen Yuan, Anthony Rinaldi, Annie En-Shiun Lee
Abstract: Neural Machine Translation (NMT) systems built on multilingual sequence-to-sequence Language Models (msLMs) fail to deliver expected results when the amount of parallel data for a language, as well as the language's representation in the model are limited. This restricts the capabilities of domain-specific NMT systems for low-resource languages (LRLs). As a solution, parallel data from auxiliary domains can be used either to fine-tune or to further pre-train the msLM. We present an evaluation of the effectiveness of these two techniques in the context of domain-specific LRL-NMT. We also explore the impact of domain divergence on NMT model performance. We recommend several strategies for utilizing auxiliary parallel data in building domain-specific NMT models for LRLs.
Authors: Xiang Huang, Jiayu Shen, Shanshan Huang, Sitao Cheng, Xiaxia Wang, Yuzhong Qu
Abstract: Semantic parsing, which converts natural language questions into logic forms, plays a crucial role in reasoning within structured environments. However, existing methods encounter two significant challenges: reliance on extensive manually annotated datasets and limited generalization capability to unseen examples. To tackle these issues, we propose Targeted Synthetic Data Generation (TARGA), a practical framework that dynamically generates high-relevance synthetic data without manual annotation. Starting from the pertinent entities and relations of a given question, we probe for the potential relevant queries through layer-wise expansion and cross-layer combination. Then we generate corresponding natural language questions for these constructed queries to jointly serve as the synthetic demonstrations for in-context learning. Experiments on multiple knowledge base question answering (KBQA) datasets demonstrate that TARGA, using only a 7B-parameter model, substantially outperforms existing non-fine-tuned methods that utilize close-sourced model, achieving notable improvements in F1 scores on GrailQA(+7.7) and KBQA-Agent(+12.2). Furthermore, TARGA also exhibits superior sample efficiency, robustness, and generalization capabilities under non-I.I.D. settings.
Authors: Sanjukta Ghosh
Abstract: This study compares the performance of AI-generated and human-written product descriptions using a multifaceted evaluation model. We analyze descriptions for 100 products generated by four AI models (Gemma 2B, LLAMA, GPT2, and ChatGPT 4) with and without sample descriptions, against human-written descriptions. Our evaluation metrics include sentiment, readability, persuasiveness, Search Engine Optimization(SEO), clarity, emotional appeal, and call-to-action effectiveness. The results indicate that ChatGPT 4 performs the best. In contrast, other models demonstrate significant shortcomings, producing incoherent and illogical output that lacks logical structure and contextual relevance. These models struggle to maintain focus on the product being described, resulting in disjointed sentences that do not convey meaningful information. This research provides insights into the current capabilities and limitations of AI in the creation of content for e-Commerce.
Authors: Cassandra Daniels, Koffka Khan
Abstract: This research investigates the performance of various machine learning algorithms (CNN, LSTM, VADER, and RoBERTa) for sentiment analysis of Twitter data related to imported food items in Trinidad and Tobago. The study addresses three primary research questions: the comparative accuracy and efficiency of the algorithms, the optimal configurations for each model, and the potential applications of the optimized models in a live system for monitoring public sentiment and its impact on the import bill. The dataset comprises tweets from 2018 to 2024, divided into imbalanced, balanced, and temporal subsets to assess the impact of data balancing and the COVID-19 pandemic on sentiment trends. Ten experiments were conducted to evaluate the models under various configurations. Results indicated that VADER outperformed the other models in both multi-class and binary sentiment classifications. The study highlights significant changes in sentiment trends pre- and post-COVID-19, with implications for import policies.
Authors: Kumud Tripathi, Raj Gothi, Pankaj Wasnik
Abstract: Automatic speech recognition has recently seen a significant advancement with large foundational models such as Whisper. However, these models often struggle to perform well in low-resource languages, such as Indian languages. This paper explores two novel approaches to enhance Whisper's multilingual speech recognition performance in Indian languages. First, we propose prompt-tuning with language family information, which enhances Whisper's accuracy in linguistically similar languages. Second, we introduce a novel tokenizer that reduces the number of generated tokens, thereby accelerating Whisper's inference speed. Our extensive experiments demonstrate that the tokenizer significantly reduces inference time, while prompt-tuning enhances accuracy across various Whisper model sizes, including Small, Medium, and Large. Together, these techniques achieve a balance between optimal WER and inference speed.
Authors: Tomer Ullman
Abstract: Illusions are entertaining, but they are also a useful diagnostic tool in cognitive science, philosophy, and neuroscience. A typical illusion shows a gap between how something "really is" and how something "appears to be", and this gap helps us understand the mental processing that lead to how something appears to be. Illusions are also useful for investigating artificial systems, and much research has examined whether computational models of perceptions fall prey to the same illusions as people. Here, I invert the standard use of perceptual illusions to examine basic processing errors in current vision language models. I present these models with illusory-illusions, neighbors of common illusions that should not elicit processing errors. These include such things as perfectly reasonable ducks, crooked lines that truly are crooked, circles that seem to have different sizes because they are, in fact, of different sizes, and so on. I show that many current vision language systems mistakenly see these illusion-illusions as illusions. I suggest that such failures are part of broader failures already discussed in the literature.
Authors: Rongfeng Su, Changqing Xu, Xinyi Wu, Feng Xu, Xie Chen, Lan Wangt, Nan Yan
Abstract: Previous studies have demonstrated that emotional features from a single acoustic sentiment label can enhance depression diagnosis accuracy. Additionally, according to the Emotion Context-Insensitivity theory and our pilot study, individuals with depression might convey negative emotional content in an unexpectedly calm manner, showing a high degree of inconsistency in emotional expressions during natural conversations. So far, few studies have recognized and leveraged the emotional expression inconsistency for depression detection. In this paper, a multimodal cross-attention method is presented to capture the Acoustic-Textual Emotional Inconsistency (ATEI) information. This is achieved by analyzing the intricate local and long-term dependencies of emotional expressions across acoustic and textual domains, as well as the mismatch between the emotional content within both domains. A Transformer-based model is then proposed to integrate this ATEI information with various fusion strategies for detecting depression. Furthermore, a scaling technique is employed to adjust the ATEI feature degree during the fusion process, thereby enhancing the model's ability to discern patients with depression across varying levels of severity. To best of our knowledge, this work is the first to incorporate emotional expression inconsistency information into depression detection. Experimental results on a counseling conversational dataset illustrate the effectiveness of our method.
Authors: Anurag Mishra
Abstract: Neural Machine Translation (NMT) models have shown remarkable performance but remain largely opaque in their decision making processes. The interpretability of these models, especially their internal attention mechanisms, is critical for building trust and verifying that these systems behave as intended. In this work, we introduce a systematic framework to quantitatively evaluate the explainability of an NMT model attention patterns by comparing them against statistical alignments and correlating them with standard machine translation quality metrics. We present a set of metrics attention entropy and alignment agreement and validate them on an English-German test subset from WMT14 using a pre trained mT5 model. Our results indicate that sharper attention distributions correlate with improved interpretability but do not always guarantee better translation quality. These findings advance our understanding of NMT explainability and guide future efforts toward building more transparent and reliable machine translation systems.
Authors: Alex Beutel, Kai Xiao, Johannes Heidecke, Lilian Weng
Abstract: Automated red teaming can discover rare model failures and generate challenging examples that can be used for training or evaluation. However, a core challenge in automated red teaming is ensuring that the attacks are both diverse and effective. Prior methods typically succeed in optimizing either for diversity or for effectiveness, but rarely both. In this paper, we provide methods that enable automated red teaming to generate a large number of diverse and successful attacks. Our approach decomposes the task into two steps: (1) automated methods for generating diverse attack goals and (2) generating effective attacks for those goals. While we provide multiple straightforward methods for generating diverse goals, our key contributions are to train an RL attacker that both follows those goals and generates diverse attacks for those goals. First, we demonstrate that it is easy to use a large language model (LLM) to generate diverse attacker goals with per-goal prompts and rewards, including rule-based rewards (RBRs) to grade whether the attacks are successful for the particular goal. Second, we demonstrate how training the attacker model with multi-step RL, where the model is rewarded for generating attacks that are different from past attempts further increases diversity while remaining effective. We use our approach to generate both prompt injection attacks and prompts that elicit unsafe responses. In both cases, we find that our approach is able to generate highly-effective and considerably more diverse attacks than past general red-teaming approaches.
Authors: Yuan Zhao, Rui Liu, Gaoxiang Cong
Abstract: Automatic Video Dubbing (AVD) generates speech aligned with lip motion and facial emotion from scripts. Recent research focuses on modeling multimodal context to enhance prosody expressiveness but overlooks two key issues: 1) Multiscale prosody expression attributes in the context influence the current sentence's prosody. 2) Prosody cues in context interact with the current sentence, impacting the final prosody expressiveness. To tackle these challenges, we propose M2CI-Dubber, a Multiscale Multimodal Context Interaction scheme for AVD. This scheme includes two shared M2CI encoders to model the multiscale multimodal context and facilitate its deep interaction with the current sentence. By extracting global and local features for each modality in the context, utilizing attention-based mechanisms for aggregation and interaction, and employing an interaction-based graph attention network for fusion, the proposed approach enhances the prosody expressiveness of synthesized speech for the current sentence. Experiments on the Chem dataset show our model outperforms baselines in dubbing expressiveness. The code and demos are available at \textcolor[rgb]{0.93,0.0,0.47}{https://github.com/AI-S2-Lab/M2CI-Dubber}.
Authors: Situo Zhang, Hankun Wang, Da Ma, Zichen Zhu, Lu Chen, Kunyao Lan, Kai Yu
Abstract: Speculative Decoding (SD) is a popular lossless technique for accelerating the inference of Large Language Models (LLMs). We show that the decoding speed of SD frameworks with static draft structures can be significantly improved by incorporating context-aware adaptive draft structures. However, current studies on adaptive draft structures are limited by their performance, modeling approaches, and applicability. In this paper, we introduce AdaEAGLE, the first SD framework that explicitly models adaptive draft structures. AdaEAGLE leverages the Lightweight Draft Length Predictor (LDLP) module to explicitly predict the optimal number of draft tokens during inference to guide the draft model. It achieves comparable speedup results without manual thresholds and allows for deeper, more specialized optimizations. Moreover, together with threshold-based strategies, AdaEAGLE achieves a $1.62\times$ speedup over the vanilla AR decoding and outperforms fixed-length SotA baseline while maintaining output quality.
Authors: Y. Lu, A. Harati, T. Rutowski, R. Oliveira, P. Chlebek, E. Shriberg
Abstract: Depression is a global health concern with a critical need for increased patient screening. Speech technology offers advantages for remote screening but must perform robustly across patients. We have described two deep learning models developed for this purpose. One model is based on acoustics; the other is based on natural language processing. Both models employ transfer learning. Data from a depression-labeled corpus in which 11,000 unique users interacted with a human-machine application using conversational speech is used. Results on binary depression classification have shown that both models perform at or above AUC=0.80 on unseen data with no speaker overlap. Performance is further analyzed as a function of test subset characteristics, finding that the models are generally robust over speaker and session variables. We conclude that models based on these approaches offer promise for generalized automated depression screening.
Authors: Yang Chen, Shuai Fu, Yu Zhang
Abstract: Soft prompt learning methods are effective for adapting vision-language models (VLMs) to downstream tasks. Nevertheless, empirical evidence reveals a tendency of existing methods that they overfit seen classes and exhibit degraded performance on unseen classes. This limitation is due to the inherent bias in the training data towards the seen classes. To address this issue, we propose a novel soft prompt learning method, named Mixture-of-Prompts Distillation (MoPD), which can effectively transfer useful knowledge from hard prompts manually hand-crafted (a.k.a. teacher prompts) to the learnable soft prompt (a.k.a. student prompt), thereby enhancing the generalization ability of soft prompts on unseen classes. Moreover, the proposed MoPD method utilizes a gating network that learns to select hard prompts used for prompt distillation. Extensive experiments demonstrate that the proposed MoPD method outperforms state-of-the-art baselines especially on on unseen classes.
Authors: Yabing Wang, Zhuotao Tian, Qingpei Guo, Zheng Qin, Sanping Zhou, Ming Yang, Le Wang
Abstract: Visual Grounding aims to localize the referring object in an image given a natural language expression. Recent advancements in DETR-based visual grounding methods have attracted considerable attention, as they directly predict the coordinates of the target object without relying on additional efforts, such as pre-generated proposal candidates or pre-defined anchor boxes. However, existing research primarily focuses on designing stronger multi-modal decoder, which typically generates learnable queries by random initialization or by using linguistic embeddings. This vanilla query generation approach inevitably increases the learning difficulty for the model, as it does not involve any target-related information at the beginning of decoding. Furthermore, they only use the deepest image feature during the query learning process, overlooking the importance of features from other levels. To address these issues, we propose a novel approach, called RefFormer. It consists of the query adaption module that can be seamlessly integrated into CLIP and generate the referential query to provide the prior context for decoder, along with a task-specific decoder. By incorporating the referential query into the decoder, we can effectively mitigate the learning difficulty of the decoder, and accurately concentrate on the target object. Additionally, our proposed query adaption module can also act as an adapter, preserving the rich knowledge within CLIP without the need to tune the parameters of the backbone network. Extensive experiments demonstrate the effectiveness and efficiency of our proposed method, outperforming state-of-the-art approaches on five visual grounding benchmarks.
Authors: Yang Du, Yuqi Liu, Qin Jin
Abstract: Cross-modal (e.g. image-text, video-text) retrieval is an important task in information retrieval and multimodal vision-language understanding field. Temporal understanding makes video-text retrieval more challenging than image-text retrieval. However, we find that the widely used video-text benchmarks have shortcomings in comprehensively assessing abilities of models, especially in temporal understanding, causing large-scale image-text pre-trained models can already achieve comparable zero-shot performance with video-text pre-trained models. In this paper, we introduce RTime, a novel temporal-emphasized video-text retrieval dataset. We first obtain videos of actions or events with significant temporality, and then reverse these videos to create harder negative samples. We then recruit annotators to judge the significance and reversibility of candidate videos, and write captions for qualified videos. We further adopt GPT-4 to extend more captions based on human-written captions. Our RTime dataset currently consists of 21k videos with 10 captions per video, totalling about 122 hours. Based on RTime, we propose three retrieval benchmark tasks: RTime-Origin, RTime-Hard, and RTime-Binary. We further enhance the use of harder-negatives in model training, and benchmark a variety of video-text models on RTime. Extensive experiment analysis proves that RTime indeed poses new and higher challenges to video-text retrieval. We release our RTime dataset\footnote{\url{https://github.com/qyr0403/Reversed-in-Time}} to further advance video-text retrieval and multimodal understanding research.
Authors: Wenjing Chen
Abstract: With the rapid development of multimodal learning, the image-text matching task, as a bridge connecting vision and language, has become increasingly important. Based on existing research, this study proposes an innovative visual semantic embedding model, Multi-Headed Consensus-Aware Visual-Semantic Embedding (MH-CVSE). This model introduces a multi-head self-attention mechanism based on the consensus-aware visual semantic embedding model (CVSE) to capture information in multiple subspaces in parallel, significantly enhancing the model's ability to understand and represent the complex relationship between images and texts. In addition, we adopt a parameterized feature fusion strategy to flexibly integrate feature information at different levels, further improving the model's expressive power. In terms of loss function design, the MH-CVSE model adopts a dynamic weight adjustment strategy to dynamically adjust the weight according to the loss value itself, so that the model can better balance the contribution of different loss terms during training. At the same time, we introduce a cosine annealing learning rate strategy to help the model converge more stably in the later stages of training. Extensive experimental verification on the Flickr30k dataset shows that the MH-CVSE model achieves better performance than previous methods in both bidirectional image and text retrieval tasks, fully demonstrating its effectiveness and superiority.
Authors: Jingcheng Hu, Houyi Li, Yinmin Zhang, Zili Wang, Shuigeng Zhou, Xiangyu Zhang, Heung-Yeung Shum
Abstract: We propose novel attention architectures, Multi-matrix Factorization Attention (MFA) and MFA-Key-Reuse (MFA-KR). Existing variants for standard Multi-Head Attention (MHA), including SOTA methods like MLA, fail to maintain as strong performance under stringent Key-Value cache (KV cache) constraints. MFA enhances model capacity by efficiently scaling up both the number and dimension of attention heads through low-rank matrix factorization in the Query-Key (QK) circuit. Extending MFA, MFA-KR further reduces memory requirements by repurposing the key cache as value through value projection re-parameterization. MFA's design enables strong model capacity when working under tight KV cache budget, while MFA-KR is suitable for even harsher KV cache limits with minor performance trade-off. Notably, in our extensive and large-scale experiments, the proposed architecture outperforms MLA and performs comparably to MHA, while reducing KV cache usage by up to 56% and 93.7%, respectively.
Authors: Quang Hoang Trung, Le Trung Hoang, Nguyen Van Hoang Phuc
Abstract: Efficient text retrieval is critical for applications such as legal document analysis, particularly in specialized contexts like Japanese legal systems. Existing retrieval methods often underperform in such domain-specific scenarios, necessitating tailored approaches. In this paper, we introduce a novel two-phase text retrieval pipeline optimized for Japanese legal datasets. Our method leverages advanced language models to achieve state-of-the-art performance, significantly improving retrieval efficiency and accuracy. To further enhance robustness and adaptability, we incorporate an ensemble model that integrates multiple retrieval strategies, resulting in superior outcomes across diverse tasks. Extensive experiments validate the effectiveness of our approach, demonstrating strong performance on both Japanese legal datasets and widely recognized benchmarks like MS-MARCO. Our work establishes new standards for text retrieval in domain-specific and general contexts, providing a comprehensive solution for addressing complex queries in legal and multilingual environments.
Authors: Taewhan Kim, Soeun Lee, Si-Woo Kim, Dong-Jin Kim
Abstract: Recent lightweight image captioning models using retrieved data mainly focus on text prompts. However, previous works only utilize the retrieved text as text prompts, and the visual information relies only on the CLIP visual embedding. Because of this issue, there is a limitation that the image descriptions inherent in the prompt are not sufficiently reflected in the visual embedding space. To tackle this issue, we propose ViPCap, a novel retrieval text-based visual prompt for lightweight image captioning. ViPCap leverages the retrieved text with image information as visual prompts to enhance the ability of the model to capture relevant visual information. By mapping text prompts into the CLIP space and generating multiple randomized Gaussian distributions, our method leverages sampling to explore randomly augmented distributions and effectively retrieves the semantic features that contain image information. These retrieved features are integrated into the image and designated as the visual prompt, leading to performance improvements on the datasets such as COCO, Flickr30k, and NoCaps. Experimental results demonstrate that ViPCap significantly outperforms prior lightweight captioning models in efficiency and effectiveness, demonstrating the potential for a plug-and-play solution.
Authors: Aleksandar Terzi\'c, Michael Hersche, Giacomo Camposampiero, Thomas Hofmann, Abu Sebastian, Abbas Rahimi
Abstract: Selective state-space models (SSMs) are an emerging alternative to the Transformer, offering the unique advantage of parallel training and sequential inference. Although these models have shown promising performance on a variety of tasks, their formal expressiveness and length generalization properties remain underexplored. In this work, we provide insight into the workings of selective SSMs by analyzing their expressiveness and length generalization performance on regular language tasks, i.e., finite-state automaton (FSA) emulation. We address certain limitations of modern SSM-based architectures by introducing the Selective Dense State-Space Model (SD-SSM), the first selective SSM that exhibits perfect length generalization on a set of various regular language tasks using a single layer. It utilizes a dictionary of dense transition matrices, a softmax selection mechanism that creates a convex combination of dictionary matrices at each time step, and a readout consisting of layer normalization followed by a linear map. We then proceed to evaluate variants of diagonal selective SSMs by considering their empirical performance on commutative and non-commutative automata. We explain the experimental results with theoretical considerations. Our code is available at https://github.com/IBM/selective-dense-state-space-model.
URLs: https://github.com/IBM/selective-dense-state-space-model.
Authors: Sang-gil Lee, Zhifeng Kong, Arushi Goel, Sungwon Kim, Rafael Valle, Bryan Catanzaro
Abstract: Recent years have seen significant progress in Text-To-Audio (TTA) synthesis, enabling users to enrich their creative workflows with synthetic audio generated from natural language prompts. Despite this progress, the effects of data, model architecture, training objective functions, and sampling strategies on target benchmarks are not well understood. With the purpose of providing a holistic understanding of the design space of TTA models, we set up a large-scale empirical experiment focused on diffusion and flow matching models. Our contributions include: 1) AF-Synthetic, a large dataset of high quality synthetic captions obtained from an audio understanding model; 2) a systematic comparison of different architectural, training, and inference design choices for TTA models; 3) an analysis of sampling methods and their Pareto curves with respect to generation quality and inference speed. We leverage the knowledge obtained from this extensive analysis to propose our best model dubbed Elucidated Text-To-Audio (ETTA). When evaluated on AudioCaps and MusicCaps, ETTA provides improvements over the baselines trained on publicly available data, while being competitive with models trained on proprietary data. Finally, we show ETTA's improved ability to generate creative audio following complex and imaginative captions -- a task that is more challenging than current benchmarks.
Authors: Omar M. Safa, Mahmoud M. Abdelaziz, Mustafa Eltawy, Mohamed Mamdouh, Moamen Gharib, Salaheldin Eltenihy, Nagia M. Ghanem, Mohamed M. Ismail
Abstract: Machine Unlearning has emerged as a critical area in artificial intelligence, addressing the need to selectively remove learned data from machine learning models in response to data privacy regulations. This paper provides a comprehensive comparative analysis of six state-of-theart unlearning techniques applied to image and text classification tasks. We evaluate their performance, efficiency, and compliance with regulatory requirements, highlighting their strengths and limitations in practical scenarios. By systematically analyzing these methods, we aim to provide insights into their applicability, challenges,and tradeoffs, fostering advancements in the field of ethical and adaptable machine learning.
Authors: Sijia Chen, Baochun Li
Abstract: Large language models (LLMs) have been routinely used to solve various tasks using step-by-step reasoning. However, the structure of intermediate reasoning steps, or thoughts, is rigid and unidirectional, such as chains, trees, or acyclic-directed graphs. Consequently, the resulting inflexible and forward-only reasoning may not address challenging tasks and fail when the LLM frequently gives false responses, i.e., ``hallucinations''. This paper proposes a new reasoning framework, called Thought Rollback (TR), allowing LLMs to adaptively build thought structure while maintaining effective reasoning toward problem-solving under ``hallucinations''. The core mechanism of TR is rolling back thoughts, which allows LLMs to perform error analysis on thoughts, and thus roll back to any previously mistaken thought for revision. Subsequently, by including such trial-and-error in the prompt to guide the LLM, each rollback leads to one more reliable reasoning path. Therefore, starting with a simple prompt without human annotations, LLM with TR adaptively and gradually explores thoughts for a correct solution. Comprehensive experiments on mathematical problems and multi-task reasoning demonstrate the state-of-the-art performance of TR in terms of problem-solving rate and interaction cost. For instance, the solving rate of GPT-4 with TR outperforms the current best by $9\%$ on the MATH dataset.
Authors: Qiushi Sun, Kanzhi Cheng, Zichen Ding, Chuanyang Jin, Yian Wang, Fangzhi Xu, Zhenyu Wu, Chengyou Jia, Liheng Chen, Zhoumianze Liu, Ben Kao, Guohao Li, Junxian He, Yu Qiao, Zhiyong Wu
Abstract: Graphical User Interface (GUI) agents powered by Vision-Language Models (VLMs) have demonstrated human-like computer control capability. Despite their utility in advancing digital automation, a critical bottleneck persists: collecting high-quality trajectory data for training. Common practices for collecting such data rely on human supervision or synthetic data generation through executing pre-defined tasks, which are either resource-intensive or unable to guarantee data quality. Moreover, these methods suffer from limited data diversity and significant gaps between synthetic data and real-world environments. To address these challenges, we propose OS-Genesis, a novel GUI data synthesis pipeline that reverses the conventional trajectory collection process. Instead of relying on pre-defined tasks, OS-Genesis enables agents first to perceive environments and perform step-wise interactions, then retrospectively derive high-quality tasks to enable trajectory-level exploration. A trajectory reward model is then employed to ensure the quality of the generated trajectories. We demonstrate that training GUI agents with OS-Genesis significantly improves their performance on highly challenging online benchmarks. In-depth analysis further validates OS-Genesis's efficiency and its superior data quality and diversity compared to existing synthesis methods. Our codes, data, and checkpoints are available at \href{https://qiushisun.github.io/OS-Genesis-Home/}{OS-Genesis Homepage}.
Authors: Ananth Balashankar, Ziteng Sun, Jonathan Berant, Jacob Eisenstein, Michael Collins, Adrian Hutter, Jong Lee, Chirag Nagpal, Flavien Prost, Aradhana Sinha, and Ananda Theertha Suresh, Ahmad Beirami
Abstract: Language model alignment has become a critical step in training modern generative language models. The goal of alignment is to finetune a reference model such that the win rate of a sample from the aligned model over a sample from the reference model is high, subject to a KL divergence constraint. Today, we are increasingly using inference-time algorithms (e.g., Best-of-N, controlled decoding, tree search) to decode from language models rather than standard sampling. However, the alignment objective does not capture such inference-time decoding procedures. We show that the existing alignment framework is sub-optimal in view of such inference-time methods. We then modify the alignment objective and propose a framework for inference-aware alignment (IAPO). We prove that for any inference-time decoding algorithm, the optimal solution that optimizes the inference-time win rate of the aligned policy against the reference policy is the solution to the typical RLHF problem with a transformation of the reward. This motivates us to provide the KL-regularized calibrate-and-transform RL (CTRL) algorithm to solve this problem, which involves a reward calibration step and a KL-regularized reward maximization step with a transformation of the calibrated reward. We particularize our study to two important inference-time strategies: best-of-N sampling and best-of-N jailbreaking, where N responses are sampled from the model and the one with the highest or lowest reward is selected. We propose specific transformations for these strategies and demonstrate that our framework offers significant improvements over existing state-of-the-art methods for language model alignment. Empirically, we outperform baselines that are designed without taking inference-time decoding into consideration by 8-12% and 4-9% on inference-time win rates over the Anthropic helpfulness and harmlessness dialog benchmark datasets.
Authors: Yuqi Zhu, Xiaohan Wang, Jing Chen, Shuofei Qiao, Yixin Ou, Yunzhi Yao, Shumin Deng, Huajun Chen, Ningyu Zhang
Abstract: This paper presents an exhaustive quantitative and qualitative evaluation of Large Language Models (LLMs) for Knowledge Graph (KG) construction and reasoning. We engage in experiments across eight diverse datasets, focusing on four representative tasks encompassing entity and relation extraction, event extraction, link prediction, and question-answering, thereby thoroughly exploring LLMs' performance in the domain of construction and inference. Empirically, our findings suggest that LLMs, represented by GPT-4, are more suited as inference assistants rather than few-shot information extractors. Specifically, while GPT-4 exhibits good performance in tasks related to KG construction, it excels further in reasoning tasks, surpassing fine-tuned models in certain cases. Moreover, our investigation extends to the potential generalization ability of LLMs for information extraction, leading to the proposition of a Virtual Knowledge Extraction task and the development of the corresponding VINE dataset. Based on these empirical findings, we further propose AutoKG, a multi-agent-based approach employing LLMs and external sources for KG construction and reasoning. We anticipate that this research can provide invaluable insights for future undertakings in the field of knowledge graphs. The code and datasets are in https://github.com/zjunlp/AutoKG.
Authors: Ziqiao Ma, Jiayi Pan, Joyce Chai
Abstract: The ability to connect language units to their referents in the physical world, referred to as grounding, is crucial to learning and understanding grounded meanings of words. While humans demonstrate fast mapping in new word learning, it remains unclear whether modern vision-language models can truly represent language with their grounded meanings and how grounding may further bootstrap new word learning. To this end, we introduce Grounded Open Vocabulary Acquisition (GOVA) to examine grounding and bootstrapping in open-world language learning. As an initial attempt, we propose object-oriented BERT (OctoBERT), a novel visually-grounded language model by pre-training on image-text pairs highlighting grounding as an objective. Through extensive experiments and analysis, we demonstrate that OctoBERT is a more coherent and fast grounded word learner, and that the grounding ability acquired during pre-training helps the model to learn unseen words more rapidly and robustly. Our code is available at https://github.com/sled-group/world-to-words
Authors: Ziqiao Ma, Jacob Sansom, Run Peng, Joyce Chai
Abstract: Large Language Models (LLMs) have generated considerable interest and debate regarding their potential emergence of Theory of Mind (ToM). Several recent inquiries reveal a lack of robust ToM in these models and pose a pressing demand to develop new benchmarks, as current ones primarily focus on different aspects of ToM and are prone to shortcuts and data leakage. In this position paper, we seek to answer two road-blocking questions: (1) How can we taxonomize a holistic landscape of machine ToM? (2) What is a more effective evaluation protocol for machine ToM? Following psychological studies, we taxonomize machine ToM into 7 mental state categories and delineate existing benchmarks to identify under-explored aspects of ToM. We argue for a holistic and situated evaluation of ToM to break ToM into individual components and treat LLMs as an agent who is physically situated in environments and socially situated in interactions with humans. Such situated evaluation provides a more comprehensive assessment of mental states and potentially mitigates the risk of shortcuts and data leakage. We further present a pilot study in a grid world setup as a proof of concept. We hope this position paper can facilitate future research to integrate ToM with LLMs and offer an intuitive means for researchers to better position their work in the landscape of ToM. Project page: https://github.com/Mars-tin/awesome-theory-of-mind
Authors: Mengyang Chen, Lingwei Wei, Han Cao, Wei Zhou, Songlin Hu
Abstract: Large Language Models (LLMs) have garnered significant attention for their powerful ability in natural language understanding and reasoning. In this paper, we present a comprehensive empirical study to explore the performance of LLMs on misinformation detection tasks. This study stands as the pioneering investigation into the understanding capabilities of multiple LLMs regarding both content and propagation across social media platforms. Our empirical studies on eight misinformation detection datasets show that LLM-based detectors can achieve comparable performance in text-based misinformation detection but exhibit notably constrained capabilities in comprehending propagation structure compared to existing models in propagation-based misinformation detection. Our experiments further demonstrate that LLMs exhibit great potential to enhance existing misinformation detection models. These findings highlight the potential ability of LLMs to detect misinformation.
Authors: Souradip Chakraborty, Jiahao Qiu, Hui Yuan, Alec Koppel, Furong Huang, Dinesh Manocha, Amrit Singh Bedi, Mengdi Wang
Abstract: Reinforcement Learning from Human Feedback (RLHF) aligns language models to human preferences by employing a singular reward model derived from preference data. However, such an approach overlooks the rich diversity of human preferences inherent in data collected from multiple users. In this work, we first derive an impossibility result of alignment with single reward RLHF, thereby highlighting its insufficiency in representing diverse human preferences. To provide an equitable solution to the problem, we learn a mixture of preference distributions via an expectation-maximization algorithm and propose a MaxMin alignment objective for policy learning inspired by the Egalitarian principle in social choice theory to better represent diverse human preferences. We elucidate the connection of our proposed approach to distributionally robust optimization and general utility RL, thereby highlighting the generality and robustness of our proposed solution. We present comprehensive experimental results on small-scale (GPT-2) and large-scale language models (with Tulu2-7B) and show the efficacy of the proposed approach in the presence of diversity among human preferences. Our algorithm achieves an average improvement of more than 16% in win-rates over conventional RLHF algorithms and improves the win-rate (accuracy) for minority groups by over 33% without compromising the performance of majority groups, showcasing the robustness and fairness of our approach. We remark that our findings in this work are not only limited to language models but also extend to reinforcement learning in general.
Authors: Aliakbar Nafar, Kristen Brent Venable, Parisa Kordjamshidi
Abstract: This paper considers the challenges Large Language Models (LLMs) face when reasoning over text that includes information involving uncertainty explicitly quantified via probability values. This type of reasoning is relevant to a variety of contexts ranging from everyday conversations to medical decision-making. Despite improvements in the mathematical reasoning capabilities of LLMs, they still exhibit significant difficulties when it comes to probabilistic reasoning. To deal with this problem, we introduce the Bayesian Linguistic Inference Dataset (BLInD), a new dataset specifically designed to test the probabilistic reasoning capabilities of LLMs. We use BLInD to find out the limitations of LLMs for tasks involving probabilistic reasoning. In addition, we present several prompting strategies that map the problem to different formal representations, including Python code, probabilistic algorithms, and probabilistic logical programming. We conclude by providing an evaluation of our methods on BLInD and an adaptation of a causal reasoning question-answering dataset. Our empirical results highlight the effectiveness of our proposed strategies for multiple LLMs.
Authors: Jialin Hua, Liangqing Luo, Weiying Ping, Yan Liao, Chunhai Tao, Xuewen Lub
Abstract: Open information extraction (OIE) aims to extract surface relations and their corresponding arguments from natural language text, irrespective of domain. This paper presents an innovative OIE model, APRCOIE, tailored for Chinese text. Diverging from previous models, our model generates extraction patterns autonomously. The model defines a new pattern form for Chinese OIE and proposes an automated pattern generation methodology. In that way, the model can handle a wide array of complex and diverse Chinese grammatical phenomena. We design a preliminary filter based on tensor computing to conduct the extraction procedure efficiently. To train the model, we manually annotated a large-scale Chinese OIE dataset. In the comparative evaluation, we demonstrate that APRCOIE outperforms state-of-the-art Chinese OIE models and significantly expands the boundaries of achievable OIE performance. The code of APRCOIE and the annotated dataset are released on GitHub (https://github.com/jialin666/APRCOIE_v1)
Authors: Taishi Nakamura, Mayank Mishra, Simone Tedeschi, Yekun Chai, Jason T Stillerman, Felix Friedrich, Prateek Yadav, Tanmay Laud, Vu Minh Chien, Terry Yue Zhuo, Diganta Misra, Ben Bogin, Xuan-Son Vu, Marzena Karpinska, Arnav Varma Dantuluri, Wojciech Kusa, Tommaso Furlanello, Rio Yokota, Niklas Muennighoff, Suhas Pai, Tosin Adewumi, Veronika Laippala, Xiaozhe Yao, Adalberto Junior, Alpay Ariyak, Aleksandr Drozd, Jordan Clive, Kshitij Gupta, Liangyu Chen, Qi Sun, Ken Tsui, Noah Persaud, Nour Fahmy, Tianlong Chen, Mohit Bansal, Nicolo Monti, Tai Dang, Ziyang Luo, Tien-Tung Bui, Roberto Navigli, Virendra Mehta, Matthew Blumberg, Victor May, Huu Nguyen, Sampo Pyysalo
Abstract: Pretrained language models are an integral part of AI applications, but their high computational cost for training limits accessibility. Initiatives such as Bloom and StarCoder aim to democratize access to pretrained models for collaborative community development. Despite these efforts, such models encounter challenges such as limited multilingual capabilities, risks of catastrophic forgetting during continual pretraining, and the high costs of training models from scratch, alongside the need to align with AI safety standards and regulatory frameworks. This paper presents Aurora-M, a 15B parameter multilingual open-source model trained on English, Finnish, Hindi, Japanese, Vietnamese, and code. Continually pretrained from StarCoderPlus on 435B additional tokens, Aurora-M surpasses 2T tokens in total training token count. It is the first open-source multilingual model fine-tuned on human-reviewed safety instructions, thus aligning its development not only with conventional red-teaming considerations, but also with the specific concerns articulated in the Biden-Harris Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. We evaluate Aurora-M across a wide range of tasks and languages, showcasing its robustness against catastrophic forgetting and its superior performance in multilingual settings, particularly in safety evaluations. We open-source Aurora-M and its variants to encourage responsible open-source development of large language models at https://huggingface.co/aurora-m.
Authors: Alexander Loth, Martin Kappes, Marc-Oliver Pahl
Abstract: Fake news significantly influence our society. They impact consumers, voters, and many other societal groups. While Fake News exist for a centuries, Generative AI brings fake news on a new level. It is now possible to automate the creation of masses of high-quality individually targeted Fake News. On the other end, Generative AI can also help detecting Fake News. Both fields are young but developing fast. This survey provides a comprehensive examination of the research and practical use of Generative AI for Fake News detection and creation in 2024. Following the Structured Literature Survey approach, the paper synthesizes current results in the following topic clusters 1) enabling technologies, 2) creation of Fake News, 3) case study social media as most relevant distribution channel, 4) detection of Fake News, and 5) deepfakes as upcoming technology. The article also identifies current challenges and open issues.
Authors: Alexandre Bonlarron, Jean-Charles R\'egin
Abstract: Constrained text generation remains a challenging task, particularly when dealing with hard constraints. Traditional NLP approaches prioritize generating meaningful and coherent output. Also, the current state-of-the-art methods often lack the expressiveness and constraint satisfaction capabilities to handle such tasks effectively. Recently, an approach for generating constrained sentences in CP has been proposed in (Bonlarron et al, 2023). This ad-hoc model to solve the sentences generation problem under MNREAD rules proved neithertheless to be computationaly and structuraly unsuitable to deal with other more constrained problems. In this paper, a novel more generic approach is introduced to tackle many of these previously untractable problems, and illustrated here with the quite untractable sentences generation problem following RADNER rules. More precisely, this paper presents the CPTextGen Framework. This framework considers a constrained text generation problem as a discrete combinatorial optimization problem. It is solved by a constraint programming method that combines linguistic properties (e.g., n-grams or language level) with other more classical constraints (e.g., the number of characters, syllables). Eventually, a curation phase allows for selecting the best-generated sentences according to perplexity using an LLM. The effectiveness of this approach is demonstrated by tackling a new, more tediously constrained text generation problem: the iconic RADNER sentences problem. This problem aims to generate sentences respecting a set of quite strict rules defined by their use in vision and clinical research. Thanks to our CP-based approach, many new strongly constrained sentences have been successfully generated. This highlights our approach's potential to handle unreasonably constrained text generation scenarios.
Authors: Zhijie Nie, Richong Zhang, Zhanyu Wu
Abstract: Text embeddings from large language models (LLMs) have achieved excellent results in tasks such as information retrieval, semantic textual similarity, etc. In this work, we show an interesting finding: when feeding a text into the LLM-based embedder, the obtained text embedding will be able to be aligned with the key tokens in the input text. We first fully analyze this phenomenon on eight LLM-based embedders and show that this phenomenon is universal and is not affected by model architecture, training strategy, and embedding method. With a deeper analysis, we find that the main change in embedding space between these embedders and their LLM backbones is in the first principal component. By adjusting the first principal component, we can align text embedding with the key tokens. Finally, we give several examples to demonstrate the vast application potential of this finding: (1) we propose a simple and practical sparse retrieval method based on the aligned tokens, which can achieve 80% of the dense retrieval effect of the same model while reducing the computation significantly; (2) we show that our findings provide a novel perspective to help understand novel technologies (e.g., instruction-following embedding) and fuzzy concepts (e.g., semantic relatedness vs. similarity) in this field.
Authors: Sundesh Donthi, Maximilian Spencer, Om Patel, Joon Doh, Eid Rodan
Abstract: For large language models (LLMs) like NLLB and GPT, translating idioms remains a challenge. Our goal is to enhance translation fidelity by improving LLM processing of idiomatic language while preserving the original linguistic style. This has a significant social impact, as it preserves cultural nuances and ensures translated texts retain their intent and emotional resonance, fostering better cross-cultural communication. Previous work has utilized knowledge bases like IdiomKB by providing the LLM with the meaning of an idiom to use in translation. Although this method yielded better results than a direct translation, it is still limited in its ability to preserve idiomatic writing style across languages. In this research, we expand upon the knowledge base to find corresponding idioms in the target language. Our research performs translations using two methods: The first method employs the SentenceTransformers model to semantically generate cosine similarity scores between the meanings of the original and target language idioms, selecting the best idiom (Cosine Similarity method). The second method uses an LLM to find a corresponding idiom in the target language for use in the translation (LLM-generated idiom method). As a baseline, we performed a direct translation without providing additional information. Human evaluations on the English -> Chinese, and Chinese -> English show the Cosine Similarity Lookup method out-performed others in all GPT4o translations. To further build upon IdiomKB, we developed a low-resource Urdu dataset containing Urdu idioms and their translations. Despite dataset limitations, the Cosine Similarity Lookup method shows promise, potentially overcoming language barriers and enabling the exploration of diverse literary works in Chinese and Urdu.(LoResLM @ COLING Preprint)
Authors: Junteng Liu, Shiqi Chen, Yu Cheng, Junxian He
Abstract: While large language models (LLMs) have demonstrated remarkable abilities across various fields, hallucination remains a significant challenge. Recent studies have explored hallucinations through the lens of internal representations, proposing mechanisms to decipher LLMs' adherence to facts. However, these approaches often fail to generalize to out-of-distribution data, leading to concerns about whether internal representation patterns reflect fundamental factual awareness, or only overfit spurious correlations on the specific datasets. In this work, we investigate whether a universal truthfulness hyperplane that distinguishes the model's factually correct and incorrect outputs exists within the model. To this end, we scale up the number of training datasets and conduct an extensive evaluation -- we train the truthfulness hyperplane on a diverse collection of over 40 datasets and examine its cross-task, cross-domain, and in-domain generalization. Our results indicate that increasing the diversity of the training datasets significantly enhances the performance in all scenarios, while the volume of data samples plays a less critical role. This finding supports the optimistic hypothesis that a universal truthfulness hyperplane may indeed exist within the model, offering promising directions for future research.
Authors: Adamu Lawan, Juhua Pu, Haruna Yunusa, Aliyu Umar, Muhammad Lawan
Abstract: Aspect-based Sentiment Analysis (ABSA) evaluates sentiments toward specific aspects of entities within the text. However, attention mechanisms and neural network models struggle with syntactic constraints. The quadratic complexity of attention mechanisms also limits their adoption for capturing long-range dependencies between aspect and opinion words in ABSA. This complexity can lead to the misinterpretation of irrelevant contextual words, restricting their effectiveness to short-range dependencies. To address the above problem, we present a novel approach to enhance long-range dependencies between aspect and opinion words in ABSA (MambaForGCN). This approach incorporates syntax-based Graph Convolutional Network (SynGCN) and MambaFormer (Mamba-Transformer) modules to encode input with dependency relations and semantic information. The Multihead Attention (MHA) and Selective State Space model (Mamba) blocks in the MambaFormer module serve as channels to enhance the model with short and long-range dependencies between aspect and opinion words. We also introduce the Kolmogorov-Arnold Networks (KANs) gated fusion, an adaptive feature representation system that integrates SynGCN and MambaFormer and captures non-linear, complex dependencies. Experimental results on three benchmark datasets demonstrate MambaForGCN's effectiveness, outperforming state-of-the-art (SOTA) baseline models.
Authors: Haishuo Fang, Xiaodan Zhu, Iryna Gurevych
Abstract: Deploying LLM-based agents in real-life applications often faces a critical challenge: the misalignment between agents' behavior and user intent. Such misalignment may lead agents to unintentionally execute critical actions that carry negative outcomes (e.g., accidentally triggering a "buy-now" in web shopping), resulting in undesirable or even irreversible consequences. Although addressing these issues is crucial, the preemptive detection and correction of misaligned actions remains relatively underexplored. To fill this gap, we introduce InferAct, a novel approach that leverages the belief reasoning ability of LLMs, grounded in Theory-of-Mind, to detect misaligned actions before execution. Once the misalignment is detected, InferAct alerts users for timely correction, preventing adverse outcomes and enhancing the reliability of LLM agents' decision-making processes. Experiments on three widely used tasks demonstrate that InferAct achieves up to 20% improvements on Marco-F1 against baselines in misaligned action detection. An in-depth evaluation of misalignment correction further highlights InferAct's effectiveness in improving agent alignment.
Authors: Soroosh Tayebi Arasteh, Mahshad Lotfinia, Keno Bressem, Robert Siepmann, Lisa Adams, Dyke Ferber, Christiane Kuhl, Jakob Nikolas Kather, Sven Nebelung, Daniel Truhn
Abstract: Large language models (LLMs) often generate outdated or inaccurate information based on static training datasets. Retrieval augmented generation (RAG) mitigates this by integrating outside data sources. While previous RAG systems used pre-assembled, fixed databases with limited flexibility, we have developed Radiology RAG (RadioRAG), an end-to-end framework that retrieves data from authoritative radiologic online sources in real-time. We evaluate the diagnostic accuracy of various LLMs when answering radiology-specific questions with and without access to additional online information via RAG. Using 80 questions from the RSNA Case Collection across radiologic subspecialties and 24 additional expert-curated questions with reference standard answers, LLMs (GPT-3.5-turbo, GPT-4, Mistral-7B, Mixtral-8x7B, and Llama3 [8B and 70B]) were prompted with and without RadioRAG in a zero-shot inference scenario RadioRAG retrieved context-specific information from www.radiopaedia.org in real-time. Accuracy was investigated. Statistical analyses were performed using bootstrapping. The results were further compared with human performance. RadioRAG improved diagnostic accuracy across most LLMs, with relative accuracy increases ranging up to 54% for different LLMs. It matched or exceeded non-RAG models and the human radiologist in question answering across radiologic subspecialties, particularly in breast imaging and emergency radiology. However, the degree of improvement varied among models; GPT-3.5-turbo and Mixtral-8x7B-instruct-v0.1 saw notable gains, while Mistral-7B-instruct-v0.2 showed no improvement, highlighting variability in RadioRAG's effectiveness. LLMs benefit when provided access to domain-specific data beyond their training data. For radiology, RadioRAG establishes a robust framework that substantially improves diagnostic accuracy and factuality in radiological question answering.
Authors: Shican Wu, Xiao Ma, Dehui Luo, Lulu Li, Xiangcheng Shi, Xin Chang, Xiaoyun Lin, Ran Luo, Chunlei Pei, Zhi-Jian Zhao, Jinlong Gong
Abstract: Literature research, vital for scientific work, faces the challenge of the surging torrent of information in the vast ocean of literature exceeding researchers' processing capabilities. To address this issue, we present an automated review generation method based on Large Language Models (LLMs), aimed at overcoming efficiency bottlenecks in literature processing and reducing cognitive load. Our statistically validated evaluation framework demonstrates that the generated reviews match or exceed manual quality, offering broad applicability across research fields due to minimal domain knowledge requirements. In a case study on propane dehydrogenation (PDH) catalysts, our method swiftly analyzed 343 articles, averaging seconds per article per LLM account, producing comprehensive reviews spanning 35 topics. Extended analysis of 1041 articles provided deep insights into catalysts' composition, structure, and performance. Recognizing LLMs' hallucinations, we implemented a multi-layered quality control strategy, effectively mitigating risks and ensuring reliability, as quantitatively demonstrated through manual verification. Expert verification confirms the accuracy and citation integrity of generated reviews, demonstrating LLM hallucination risks reduced to below 0.5\% with over 95\% confidence. Released Windows application enables one-click review generation, aiding researchers in tracking advancements and recommending literature. This approach showcases LLMs' role in enhancing scientific research productivity and sets the stage for further exploration.
Authors: Congchi Yin, Feng Li, Shu Zhang, Zike Wang, Jun Shao, Piji Li, Jianhua Chen, Xun Jiang
Abstract: The clinical diagnosis of most mental disorders primarily relies on the conversations between psychiatrist and patient. The creation of such diagnostic conversation datasets is promising to boost the AI mental healthcare community. However, directly collecting the conversations in real diagnosis scenarios is near impossible due to stringent privacy and ethical considerations. To address this issue, we seek to synthesize diagnostic conversation by exploiting anonymized patient cases that are easier to access. Specifically, we design a neuro-symbolic multi-agent framework for synthesizing the diagnostic conversation of mental disorders with large language models. It takes patient case as input and is capable of generating multiple diverse conversations with one single patient case. The framework basically involves the interaction between a doctor agent and a patient agent, and generates conversations under symbolic control via a dynamic diagnosis tree. By applying the proposed framework, we develop the largest Chinese mental disorders diagnosis dataset MDD-5k. This dataset is built upon 1000 real, anonymized patient cases by cooperating with Shanghai Mental Health Center and comprises 5000 high-quality long conversations with diagnosis results and treatment opinions as labels. To the best of our knowledge, it's also the first labeled dataset for Chinese mental disorders diagnosis. Human evaluation demonstrates the proposed MDD-5k dataset successfully simulates human-like diagnostic process of mental disorders.
Authors: Ond\v{r}ej Pra\v{z}\'ak, Miloslav Konop\'ik, Pavel Kr\'al
Abstract: Coreference resolution, the task of identifying expressions in text that refer to the same entity, is a critical component in various natural language processing applications. This paper presents a novel end-to-end neural coreference resolution system utilizing the CorefUD 1.1 dataset, which spans 17 datasets across 12 languages. The proposed model is based on the standard end-to-end neural coreference resolution system. We first establish baseline models, including monolingual and cross-lingual variations, and then propose several extensions to enhance performance across diverse linguistic contexts. These extensions include cross-lingual training, incorporation of syntactic information, a Span2Head model for optimized headword prediction, and advanced singleton modeling. We also experiment with headword span representation and long-documents modeling through overlapping segments. The proposed extensions, particularly the heads-only approach, singleton modeling, and long document prediction, significantly improve performance across most datasets. We also perform zero-shot cross-lingual experiments, highlighting the potential and limitations of cross-lingual transfer in coreference resolution. Our findings contribute to the development of robust and scalable coreference systems for multilingual coreference resolution. Finally, we evaluate our model on the CorefUD 1.1 test set and surpass the best model from the CRAC 2023 shared task of comparable size by a large margin.
Authors: Junhui He, Shangyu Wu, Weidong Wen, Chun Jason Xue, Qingan Li
Abstract: Deploying large language models (LLMs) on edge devices presents significant challenges due to the substantial computational overhead and memory requirements. Activation sparsification can mitigate these resource challenges by reducing the number of activated neurons during inference. Existing methods typically employ thresholding-based sparsification based on the statistics of activation tensors. However, they do not model the impact of activation sparsification on performance, resulting in suboptimal performance degradation. To address the limitations, this paper reformulates the activation sparsification problem to explicitly capture the relationship between activation sparsity and model performance. Then, this paper proposes CHESS, a general activation sparsification approach via CHannel-wise thrEsholding and Selective Sparsification. First, channel-wise thresholding assigns a unique threshold to each activation channel in the feed-forward network (FFN) layers. Then, selective sparsification involves applying thresholding-based activation sparsification to specific layers within the attention modules. Finally, we detail the implementation of sparse kernels to accelerate LLM inference. Experimental results demonstrate that the proposed CHESS achieves lower performance degradation over eight downstream tasks while activating fewer parameters than existing methods, thus speeding up the LLM inference by up to 1.27x.
Authors: Kuiyun Chen, Yanbin Wei
Abstract: Narrative systems, such as dialogue and storytelling systems, often utilize persona profiles to enhance personalized interactions. Existing persona profiles frequently exhibit biases, posing risks to system integrity and fairness. To address this, we introduce the UPCS framework, which categorizes character descriptions into eight dimensions, including bias mitigation strategies. Experimental results demonstrate UPCS's superiority in accuracy, diversity, bias elimination, and user satisfaction, marking a significant advancement in persona construction for reliable narrative systems.
Authors: Ruya Jiang, Chun Wang, Weihong Deng
Abstract: The complexities of table structures and question logic make table-based question answering (TQA) tasks challenging for Large Language Models (LLMs), often requiring task simplification before solving. This paper reveals that the reasoning process during task simplification may be more valuable than the simplified tasks themselves and aims to improve TQA performance by leveraging LLMs' reasoning capabilities. We propose a Seek-and-Solve pipeline that instructs the LLM to first seek relevant information and then answer questions, integrating these two stages at the reasoning level into a coherent Seek-and-Solve Chain of Thought (SS-CoT). Additionally, we distill a single-step TQA-solving prompt from this pipeline, using demonstrations with SS-CoT paths to guide the LLM in solving complex TQA tasks under In-Context Learning settings. Our experiments show that our approaches result in improved performance and reliability while being efficient. Our findings emphasize the importance of eliciting LLMs' reasoning capabilities to handle complex TQA tasks effectively.
Authors: HuangChao Xu, Baohua Zhang, Zhong Jin, Tiannian Zhu, Quansheng Wu, Hongming Weng
Abstract: Large language models (LLMs), such as ChatGPT, have demonstrated impressive performance in the text generation task, showing the ability to understand and respond to complex instructions. However, the performance of naive LLMs in speciffc domains is limited due to the scarcity of domain-speciffc corpora and specialized training. Moreover, training a specialized large-scale model necessitates signiffcant hardware resources, which restricts researchers from leveraging such models to drive advances. Hence, it is crucial to further improve and optimize LLMs to meet speciffc domain demands and enhance their scalability. Based on the condensed matter data center, we establish a material knowledge graph (MaterialsKG) and integrate it with literature. Using large language models and prompt learning, we develop a specialized dialogue system for topological materials called TopoChat. Compared to naive LLMs, TopoChat exhibits superior performance in structural and property querying, material recommendation, and complex relational reasoning. This system enables efffcient and precise retrieval of information and facilitates knowledge interaction, thereby encouraging the advancement on the ffeld of condensed matter materials.
Authors: Zitian Gao, Boye Niu, Xuzheng He, Haotian Xu, Hongzhang Liu, Aiwei Liu, Xuming Hu, Lijie Wen
Abstract: We propose SC-MCTS*: a novel Monte Carlo Tree Search (MCTS) reasoning algorithm for Large Language Models (LLMs), significantly improves both reasoning accuracy and speed. Our motivation comes from: 1. Previous MCTS LLM reasoning works often overlooked its biggest drawback--slower speed compared to CoT; 2. Previous research mainly used MCTS as a tool for LLM reasoning on various tasks with limited quantitative analysis or ablation studies of its components from reasoning interpretability perspective. 3. The reward model is the most crucial component in MCTS, however previous work has rarely conducted in-depth study or improvement of MCTS's reward models. Thus, we conducted extensive ablation studies and quantitative analysis on components of MCTS, revealing the impact of each component on the MCTS reasoning performance of LLMs. Building on this, (i) we designed a highly interpretable reward model based on the principle of contrastive decoding and (ii) achieved an average speed improvement of 51.9% per node using speculative decoding. Additionally, (iii) we improved UCT node selection strategy and backpropagation used in previous works, resulting in significant performance improvement. We outperformed o1-mini by an average of 17.4% on the Blocksworld multi-step reasoning dataset using Llama-3.1-70B with SC-MCTS*. Our code is available at https://github.com/zitian-gao/SC-MCTS.
Authors: Si-An Chen, Lesly Miculicich, Julian Martin Eisenschlos, Zifeng Wang, Zilong Wang, Yanfei Chen, Yasuhisa Fujii, Hsuan-Tien Lin, Chen-Yu Lee, Tomas Pfister
Abstract: Recent advancements in language models (LMs) have notably enhanced their ability to reason with tabular data, primarily through program-aided mechanisms that manipulate and analyze tables. However, these methods often require the entire table as input, leading to scalability challenges due to the positional bias or context length constraints. In response to these challenges, we introduce TableRAG, a Retrieval-Augmented Generation (RAG) framework specifically designed for LM-based table understanding. TableRAG leverages query expansion combined with schema and cell retrieval to pinpoint crucial information before providing it to the LMs. This enables more efficient data encoding and precise retrieval, significantly reducing prompt lengths and mitigating information loss. We have developed two new million-token benchmarks from the Arcade and BIRD-SQL datasets to thoroughly evaluate TableRAG's effectiveness at scale. Our results demonstrate that TableRAG's retrieval design achieves the highest retrieval quality, leading to the new state-of-the-art performance on large-scale table understanding.
Authors: Anbang Wang, Difei Mei, Zhichao Zhang, Xiuxiu Bai, Ran Yao, Zewen Fang, Min Hu, Zhirui Cao, Haitao Sun, Yifeng Guo, Hongyao Zhou, Yu Guo
Abstract: This paper presents ReverseNER, a method aimed at overcoming the limitation of large language models (LLMs) in zero-shot named entity recognition (NER) tasks, arising from their reliance on pre-provided demonstrations. ReverseNER tackles this challenge by constructing a reliable example library composed of dozens of entity-labeled sentences, generated through the reverse process of NER. Specifically, while conventional NER methods label entities in a sentence, ReverseNER features reversing the process by using an LLM to generate entities from their definitions and subsequently expand them into full sentences. During the entity expansion process, the LLM is guided to generate sentences by replicating the structures of a set of specific \textsl{feature sentences}, extracted from the task sentences by clustering. This expansion process produces dozens of entity-labeled task-relevant sentences. After constructing the example library, the method selects several semantically similar entity-labeled examples for each task sentence as references to facilitate the LLM's entity recognition. We also propose an entity-level self-consistency scoring mechanism to improve NER performance with LLMs. Experiments show that ReverseNER significantly outperforms other zero-shot NER methods with LLMs, marking a notable improvement in NER for domains without labeled data, while declining computational resource consumption.
Authors: Shan Zhong, Jiahao Zeng, Yongxin Yu, Bohong Lin
Abstract: This paper proposes a Clustering, Labeling, then Augmenting framework that significantly enhances performance in Semi-Supervised Text Classification (SSTC) tasks, effectively addressing the challenge of vast datasets with limited labeled examples. Unlike traditional SSTC approaches that rely on a predefined small set of labeled data to generate pseudo-labels for the unlabeled data, this framework innovatively employs clustering to select representative "landmarks" for labeling. These landmarks subsequently act as intermediaries in an ensemble of augmentation techniques, including Retrieval-Augmented Generation (RAG), Large Language Model (LLMs)-based rewriting, and synonym substitution, to generate synthetic labeled data without making pseudo-labels for the unlabeled data. Empirical results show that even in complex text document classification scenarios involving over 100 categories, our method achieves state-of-the-art accuracies of 95.41% on the Reuters dataset and 82.43% on the Web of Science dataset. Our approach significantly reduces the reliance on human labeling efforts and the associated expenses, while simultaneously ensuring high data quality and minimizing privacy risks. The finetuning results further show the efficiency of fine-tuning LLMs for text classification tasks, highlighting a robust solution for leveraging limited labeled data.
Authors: Chih-Kai Yang, Yu-Kuan Fu, Chen-An Li, Yi-Cheng Lin, Yu-Xiang Lin, Wei-Chih Chen, Ho Lam Chung, Chun-Yi Kuan, Wei-Ping Huang, Ke-Han Lu, Tzu-Quan Lin, Hsiu-Hsuan Wang, En-Pei Hu, Chan-Jan Hsu, Liang-Hsuan Tseng, I-Hsiang Chiu, Ulin Sanga, Xuanjun Chen, Po-chun Hsu, Shu-wen Yang, Hung-yi Lee
Abstract: This technical report presents our initial attempt to build a spoken large language model (LLM) for Taiwanese Mandarin, specifically tailored to enable real-time, speech-to-speech interaction in multi-turn conversations. Our end-to-end model incorporates a decoder-only transformer architecture and aims to achieve seamless interaction while preserving the conversational flow, including full-duplex capabilities allowing simultaneous speaking and listening. The paper also details the training process, including data preparation with synthesized dialogues and adjustments for real-time interaction. We also developed a platform to evaluate conversational fluency and response coherence in multi-turn dialogues. We hope the release of the report can contribute to the future development of spoken LLMs in Taiwanese Mandarin.
Authors: Albert Kornilov, Tatiana Shavrina
Abstract: Recent advances in language modeling have demonstrated significant improvements in zero-shot capabilities, including in-context learning, instruction following, and machine translation for extremely under-resourced languages (Tanzer et al., 2024). However, many languages with limited written resources rely primarily on formal descriptions of grammar and vocabulary. In this paper, we introduce a set of benchmarks to evaluate how well models can extract and classify information from the complex descriptions found in linguistic grammars. We present a Retrieval-Augmented Generation (RAG)-based approach that leverages these descriptions for downstream tasks such as machine translation. Our benchmarks encompass linguistic descriptions for 248 languages across 142 language families, focusing on typological features from WALS and Grambank. This set of benchmarks offers the first comprehensive evaluation of language models' in-context ability to accurately interpret and extract linguistic features, providing a critical resource for scaling NLP to low-resource languages. The code and data are publicly available at \url{https://github.com/al-the-eigenvalue/RAG-on-grammars}.
Authors: Yijiong Yu
Abstract: It has been well-known that Chain-of-Thought can remarkably enhance LLMs' performance on complex tasks. However, because it also introduces slower inference speeds and higher computational costs, many researches have attempted to use implicit CoT, which does not need LLMs to explicitly generate the intermediate steps. However, the invisible reasoning process leaves us a doubt that, can implicit CoT really be equal to explicit CoT? Therefore, in this study, we address this question through experiments. We probe the information of intermediate steps from the model's hidden states when it is either trained or prompted to perform implicit CoT. The results surprisingly indicate that when prompted, LLMs hardly think about intermediate steps, suggesting they may just rely on experience rather than strict step-by-step reasoning. But when trained, they indeed calculate intermediate steps. Moreover, in both situations, we find the effect of using implicit CoT is susceptible to the format of the problem, reaffirming the current deficiency of implicit CoT.
Authors: Zefang Liu
Abstract: Incident response (IR) is a critical aspect of cybersecurity, requiring rapid decision-making and coordinated efforts to address cyberattacks effectively. Leveraging large language models (LLMs) as intelligent agents offers a novel approach to enhancing collaboration and efficiency in IR scenarios. This paper explores the application of LLM-based multi-agent collaboration using the Backdoors & Breaches framework, a tabletop game designed for cybersecurity training. We simulate real-world IR dynamics through various team structures, including centralized, decentralized, and hybrid configurations. By analyzing agent interactions and performance across these setups, we provide insights into optimizing multi-agent collaboration for incident response. Our findings highlight the potential of LLMs to enhance decision-making, improve adaptability, and streamline IR processes, paving the way for more effective and coordinated responses to cyber threats.
Authors: Xi Cao, Quzong Gesang, Yuan Sun, Nuo Qun, Tashi Nyima
Abstract: Language models based on deep neural networks are vulnerable to textual adversarial attacks. While rich-resource languages like English are receiving focused attention, Tibetan, a cross-border language, is gradually being studied due to its abundant ancient literature and critical language strategy. Currently, there are several Tibetan adversarial text generation methods, but they do not fully consider the textual features of Tibetan script and overestimate the quality of generated adversarial texts. To address this issue, we propose a novel Tibetan adversarial text generation method called TSCheater, which considers the characteristic of Tibetan encoding and the feature that visually similar syllables have similar semantics. This method can also be transferred to other abugidas, such as Devanagari script. We utilize a self-constructed Tibetan syllable visual similarity database called TSVSDB to generate substitution candidates and adopt a greedy algorithm-based scoring mechanism to determine substitution order. After that, we conduct the method on eight victim language models. Experimentally, TSCheater outperforms existing methods in attack effectiveness, perturbation magnitude, semantic similarity, visual similarity, and human acceptance. Finally, we construct the first Tibetan adversarial robustness evaluation benchmark called AdvTS, which is generated by existing methods and proofread by humans.
Authors: Mohammed Q. Shormani
Abstract: This study sets out to answer one major question: Can ChatGPT capture swearing nuances? It presents an empirical study on the ability of ChatGPT to translate Arabic oath expressions into English. 30 Arabic oath expressions were collected from the literature. These 30 oaths were first translated via ChatGPT and then analyzed and compared to the human translation in terms of types of gaps left unfulfilled by ChatGPT. Specifically, the gaps involved are: religious gap, cultural gap, both religious and cultural gaps, no gap, using non-oath particles, redundancy and noncapturing of Arabic script diacritics. It concludes that ChatGPT translation of oaths is still much unsatisfactory, unveiling the need of further developments of ChatGPT, and the inclusion of Arabic data on which ChatGPT should be trained including oath expressions, oath nuances, rituals, and practices.
Authors: Md. Tariquzzaman, Audwit Nafi Anam, Naimul Haque, Mohsinul Kabir, Hasan Mahmud, Md Kamrul Hasan
Abstract: Data augmentation involves generating synthetic samples that resemble those in a given dataset. In resource-limited fields where high-quality data is scarce, augmentation plays a crucial role in increasing the volume of training data. This paper introduces a Bangla Text Data Augmentation (BDA) Framework that uses both pre-trained models and rule-based methods to create new variants of the text. A filtering process is included to ensure that the new text keeps the same meaning as the original while also adding variety in the words used. We conduct a comprehensive evaluation of the framework's effectiveness in Bangla text classification tasks. Our framework achieved significant improvement in F1 scores across five distinct datasets, delivering performance equivalent to models trained on 100% of the data while utilizing only 50% of the training dataset. Additionally, we explore the impact of data scarcity by progressively reducing the training data and augmenting it through BDA, resulting in notable F1 score enhancements. The study offers a thorough examination of BDA's performance, identifying key factors for optimal results and addressing its limitations through detailed analysis.
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: Yifan Lu, Yigeng Zhou, Jing Li, Yequan Wang, Xuebo Liu, Daojing He, Fangming Liu, Min Zhang
Abstract: Multi-hop question answering (MHQA) poses a significant challenge for large language models (LLMs) due to the extensive knowledge demands involved. Knowledge editing, which aims to precisely modify the LLMs to incorporate specific knowledge without negatively impacting other unrelated knowledge, offers a potential solution for addressing MHQA challenges with LLMs. However, current solutions struggle to effectively resolve issues of knowledge conflicts. Most parameter-preserving editing methods are hindered by inaccurate retrieval and overlook secondary editing issues, which can introduce noise into the reasoning process of LLMs. In this paper, we introduce KEDKG, a novel knowledge editing method that leverages a dynamic knowledge graph for MHQA, designed to ensure the reliability of answers. KEDKG involves two primary steps: dynamic knowledge graph construction and knowledge graph augmented generation. Initially, KEDKG autonomously constructs a dynamic knowledge graph to store revised information while resolving potential knowledge conflicts. Subsequently, it employs a fine-grained retrieval strategy coupled with an entity and relation detector to enhance the accuracy of graph retrieval for LLM generation. Experimental results on benchmarks show that KEDKG surpasses previous state-of-the-art models, delivering more accurate and reliable answers in environments with dynamic information.
Authors: Jinghan He, Kuan Zhu, Haiyun Guo, Junfeng Fang, Zhenglin Hua, Yuheng Jia, Ming Tang, Tat-Seng Chua, Jinqiao Wang
Abstract: Large vision-language models (LVLMs) have made substantial progress in integrating large language models (LLMs) with visual inputs, enabling advanced multimodal reasoning. Despite their success, a persistent challenge is hallucination-where generated text fails to accurately reflect visual content-undermining both accuracy and reliability. Existing methods focus on alignment training or decoding refinements but primarily address symptoms at the generation stage without probing the underlying causes. In this work, we investigate the internal mechanisms driving hallucination in LVLMs, with an emphasis on the multi-head attention module. Specifically, we introduce Vision-aware Head Divergence (VHD), a metric that quantifies the sensitivity of attention head outputs to visual context. Based on this, our findings reveal the presence of vision-aware attention heads that are more attuned to visual information; however, the model's overreliance on its prior language patterns is closely related to hallucinations. Building on these insights, we propose Vision-aware Head Reinforcement (VHR), a training-free approach to mitigate hallucination by enhancing the role of vision-aware attention heads. Extensive experiments demonstrate that our method achieves superior performance compared to state-of-the-art approaches in mitigating hallucinations, while maintaining high efficiency with negligible additional time overhead.
Authors: Weijia Shi, Xiaochuang Han, Chunting Zhou, Weixin Liang, Xi Victoria Lin, Luke Zettlemoyer, Lili Yu
Abstract: We present LMFusion, a framework for empowering pretrained text-only large language models (LLMs) with multimodal generative capabilities, enabling them to understand and generate both text and images in arbitrary sequences. LMFusion leverages existing Llama-3's weights for processing texts autoregressively while introducing additional and parallel transformer modules for processing images with diffusion. During training, the data from each modality is routed to its dedicated modules: modality-specific feedforward layers, query-key-value projections, and normalization layers process each modality independently, while the shared self-attention layers allow interactions across text and image features. By freezing the text-specific modules and only training the image-specific modules, LMFusion preserves the language capabilities of text-only LLMs while developing strong visual understanding and generation abilities. Compared to methods that pretrain multimodal generative models from scratch, our experiments demonstrate that, LMFusion improves image understanding by 20% and image generation by 3.6% using only 50% of the FLOPs while maintaining Llama-3's language capabilities. We also demonstrate that this framework can adapt existing vision-language models with multimodal generation ability. Overall, this framework not only leverages existing computational investments in text-only LLMs but also enables the parallel development of language and vision capabilities, presenting a promising direction for efficient multimodal model development.
Authors: Megan Ung, Alicia Sun, Samuel J. Bell, Bhaktipriya Radharapu, Levent Sagun, Adina Williams
Abstract: Large language models (LLMs) are often fine-tuned for use on downstream tasks, though this can degrade capabilities learned during previous training. This phenomenon, often referred to as catastrophic forgetting, has important potential implications for the safety of deployed models. In this work, we first show that models trained on downstream tasks forget their safety tuning to a greater extent than models trained in the opposite order. Second, we show that forgetting disproportionately impacts safety information about certain groups. To quantify this phenomenon, we define a new metric we term biased forgetting. We conduct a systematic evaluation of the effects of task ordering on forgetting and apply mitigations that can help the model recover from the forgetting observed. We hope our findings can better inform methods for chaining the finetuning of LLMs in continual learning settings to enable training of safer and less toxic models.
Authors: Zirong Chen, Elizabeth Chason, Noah Mladenovski, Erin Wilson, Kristin Mullen, Stephen Martini, Meiyi Ma
Abstract: Emergency response services are vital for enhancing public safety by safeguarding the environment, property, and human lives. As frontline members of these services, 9-1-1 dispatchers have a direct impact on response times and the overall effectiveness of emergency operations. However, traditional dispatcher training methods, which rely on role-playing by experienced personnel, are labor-intensive, time-consuming, and often neglect the specific needs of underserved communities. To address these challenges, we introduce Sim911, the first training simulation for 9-1-1 dispatchers powered by Large Language Models (LLMs). Sim911 enhances training through three key technical innovations: (1) knowledge construction, which utilizes archived 9-1-1 call data to generate simulations that closely mirror real-world scenarios; (2) context-aware controlled generation, which employs dynamic prompts and vector bases to ensure that LLM behavior aligns with training objectives; and (3) validation with looped correction, which filters out low-quality responses and refines the system performance.
Authors: Rushendra Sidibomma, Pransh Patwa, Parth Patwa, Aman Chadha, Vinija Jain, Amitava Das
Abstract: The detection of hate speech has become increasingly important in combating online hostility and its real-world consequences. Despite recent advancements, there is limited research addressing hate speech detection in Devanagari-scripted languages, where resources and tools are scarce. While large language models (LLMs) have shown promise in language-related tasks, traditional fine-tuning approaches are often infeasible given the size of the models. In this paper, we propose a Parameter Efficient Fine tuning (PEFT) based solution for hate speech detection and target identification. We evaluate multiple LLMs on the Devanagari dataset provided by (Thapa et al., 2025), which contains annotated instances in 2 languages - Hindi and Nepali. The results demonstrate the efficacy of our approach in handling Devanagari-scripted content.
Authors: Yujie Lin, Jingyao Liu, Yan Gao, Ante Wang, Jinsong Su
Abstract: Metaphor detection, a critical task in natural language processing, involves identifying whether a particular word in a sentence is used metaphorically. Traditional approaches often rely on supervised learning models that implicitly encode semantic relationships based on metaphor theories. However, these methods often suffer from a lack of transparency in their decision-making processes, which undermines the reliability of their predictions. Recent research indicates that LLMs (large language models) exhibit significant potential in metaphor detection. Nevertheless, their reasoning capabilities are constrained by predefined knowledge graphs. To overcome these limitations, we propose DMD, a novel dual-perspective framework that harnesses both implicit and explicit applications of metaphor theories to guide LLMs in metaphor detection and adopts a self-judgment mechanism to validate the responses from the aforementioned forms of guidance. In comparison to previous methods, our framework offers more transparent reasoning processes and delivers more reliable predictions. Experimental results prove the effectiveness of DMD, demonstrating state-of-the-art performance across widely-used datasets.
Authors: Rishiraj Saha Roy, Chris Hinze, Joel Schlotthauer, Farzad Naderi, Viktor Hangya, Andreas Foltyn, Luzian Hahn, Fabian Kuech
Abstract: Conversational question answering (ConvQA) is a convenient means of searching over RDF knowledge graphs (KGs), where a prevalent approach is to translate natural language questions to SPARQL queries. However, SPARQL has certain shortcomings: (i) it is brittle for complex intents and conversational questions, and (ii) it is not suitable for more abstract needs. Instead, we propose a novel two-pronged system where we fuse: (i) SQL-query results over a database automatically derived from the KG, and (ii) text-search results over verbalizations of KG facts. Our pipeline supports iterative retrieval: when the results of any branch are found to be unsatisfactory, the system can automatically opt for further rounds. We put everything together in a retrieval augmented generation (RAG) setup, where an LLM generates a coherent response from accumulated search results. We demonstrate the superiority of our proposed system over several baselines on a knowledge graph of BMW automobiles.
Authors: Xiaoyang Hu, Richard L. Lewis
Abstract: Cognitive tasks originally developed for humans are now increasingly used to study language models. While applying these tasks is often straightforward, interpreting their results can be challenging. In particular, when a model underperforms, it is often unclear whether this results from a limitation in the cognitive ability being tested or a failure to understand the task itself. A recent study argues that GPT 3.5's declining performance on 2-back and 3-back tasks reflects a working memory capacity limit similar to humans (Gong et al., 2024). By analyzing a range of open-source language models of varying performance levels on these tasks, we show that the poor performance instead reflects a limitation in task comprehension and task set maintenance. In addition, we challenge the best-performing model with progressively harder versions of the task (up to 10-back) and experiment with alternative prompting strategies, before analyzing model attentions. Our larger aim is to contribute to the ongoing conversation around refining methodologies for the cognitive evaluation of language models.
Authors: Jiarui Liu, Iman Ouzzani, Wenkai Li, Lechen Zhang, Tianyue Ou, Houda Bouamor, Zhijing Jin, Mona Diab
Abstract: The field of machine translation has achieved significant advancements, yet domain-specific terminology translation, particularly in AI, remains challenging. We introduced GIST, a large-scale multilingual AI terminology dataset containing 5K terms extracted from top AI conference papers spanning 2000 to 2023. The terms were translated into Arabic, Chinese, French, Japanese, and Russian using a hybrid framework that combines LLMs for extraction with human expertise for translation. The dataset's quality was benchmarked against existing resources, demonstrating superior translation accuracy through crowdsourced evaluation. GIST was integrated into translation workflows using post-translation refinement methods that required no retraining, where LLM prompting consistently improved BLEU and COMET scores. A web demonstration on the ACL Anthology platform highlights its practical application, showcasing improved accessibility for non-English speakers. This work aims to address critical gaps in AI terminology resources and fosters global inclusivity and collaboration in AI research.
Authors: Shani Goren, Oren Kalinsky, Tomer Stav, Yuri Rapoport, Yaron Fairstein, Ram Yazdi, Nachshon Cohen, Alexander Libov, Guy Kushilevitz
Abstract: The rise of LLMs has deflected a growing portion of human-computer interactions towards LLM-based chatbots. The remarkable abilities of these models allow users to interact using long, diverse natural language text covering a wide range of topics and styles. Phrasing these messages is a time and effort consuming task, calling for an autocomplete solution to assist users. We introduce the task of chatbot interaction autocomplete. We present ChaI-TeA: CHat InTEraction Autocomplete; An autcomplete evaluation framework for LLM-based chatbot interactions. The framework includes a formal definition of the task, coupled with suitable datasets and metrics. We use the framework to evaluate After formally defining the task along with suitable datasets and metrics, we test 9 models on the defined auto completion task, finding that while current off-the-shelf models perform fairly, there is still much room for improvement, mainly in ranking of the generated suggestions. We provide insights for practitioners working on this task and open new research directions for researchers in the field. We release our framework to serve as a foundation for future research.
Authors: Yizhi Li, Ruibin Yuan, Ge Zhang, Yinghao Ma, Xingran Chen, Hanzhi Yin, Chenghao Xiao, Chenghua Lin, Anton Ragni, Emmanouil Benetos, Norbert Gyenge, Roger Dannenberg, Ruibo Liu, Wenhu Chen, Gus Xia, Yemin Shi, Wenhao Huang, Zili Wang, Yike Guo, Jie Fu
Abstract: Self-supervised learning (SSL) has recently emerged as a promising paradigm for training generalisable models on large-scale data in the fields of vision, text, and speech. Although SSL has been proven effective in speech and audio, its application to music audio has yet to be thoroughly explored. This is partially due to the distinctive challenges associated with modelling musical knowledge, particularly tonal and pitched characteristics of music. To address this research gap, we propose an acoustic Music undERstanding model with large-scale self-supervised Training (MERT), which incorporates teacher models to provide pseudo labels in the masked language modelling (MLM) style acoustic pre-training. In our exploration, we identified an effective combination of teacher models, which outperforms conventional speech and audio approaches in terms of performance. This combination includes an acoustic teacher based on Residual Vector Quantisation - Variational AutoEncoder (RVQ-VAE) and a musical teacher based on the Constant-Q Transform (CQT). Furthermore, we explore a wide range of settings to overcome the instability in acoustic language model pre-training, which allows our designed paradigm to scale from 95M to 330M parameters. Experimental results indicate that our model can generalise and perform well on 14 music understanding tasks and attain state-of-the-art (SOTA) overall scores.
Authors: Martin Weyssow, Xin Zhou, Kisub Kim, David Lo, Houari Sahraoui
Abstract: Large language models (LLMs) demonstrate impressive capabilities to generate accurate code snippets given natural language intents in a zero-shot manner, i.e., without the need for specific fine-tuning. While prior studies have highlighted the advantages of fine-tuning LLMs, this process incurs high computational costs, making it impractical in resource-scarce environments, particularly for models with billions of parameters. To address these challenges, previous research explored in-context learning (ICL) and retrieval-augmented generation (RAG) as strategies to guide the LLM generative process with task-specific prompt examples. However, ICL and RAG introduce inconveniences, such as the need for designing contextually relevant prompts and the absence of learning task-specific parameters, thereby limiting downstream task performance. In this context, we foresee parameter-efficient fine-tuning (PEFT) as a promising approach to efficiently specialize LLMs to task-specific data while maintaining reasonable resource consumption. In this paper, we deliver a comprehensive study of PEFT techniques for LLMs in the context of automated code generation. Our comprehensive investigation of PEFT techniques for LLMs reveals their superiority and potential over ICL and RAG across a diverse set of LLMs and three representative Python code generation datasets: Conala, CodeAlpacaPy, and APPS. Furthermore, our study highlights the potential for tuning larger LLMs and significant reductions in memory usage by combining PEFT with quantization. Therefore, this study opens opportunities for broader applications of PEFT in software engineering scenarios. Our code is available at https://github.com/martin-wey/peft-llm-code/.
Authors: Xingyou Song, Oscar Li, Chansoo Lee, Bangding Yang, Daiyi Peng, Sagi Perel, Yutian Chen
Abstract: Regression is a powerful tool to accurately predict the outcome metric of a system given a set of parameters, but has traditionally been restricted to methods which are only applicable to a specific task. In this paper, we propose OmniPred, a framework for training language models as universal end-to-end regressors over $(x,y)$ data from arbitrary formats. Using data sourced from Google Vizier, one of the largest proprietary blackbox optimization databases in the world, our extensive experiments demonstrate that language models are capable of very precise numerical regression using only textual representations of mathematical parameters and values, and if given the opportunity to train at scale over multiple tasks, can significantly outperform traditional regression models.
Authors: Martin Weyssow, Aton Kamanda, Xin Zhou, Houari Sahraoui
Abstract: Evaluating the alignment of large language models (LLMs) with user-defined coding preferences is a challenging endeavour that requires a deep assessment of LLMs' outputs. Existing methods and benchmarks rely primarily on automated metrics and static analysis tools, which often fail to capture the nuances of user instructions and LLM outputs. To address this gap, we propose using the LLM-as-a-Judge methodology to evaluate the alignment of LLMs with coding preferences. Based on this approach, we present CodeUltraFeedback, a comprehensive dataset designed to facilitate the evaluation and improvement of LLM alignment. CodeUltraFeedback consists of 10,000 coding instructions, each annotated with four responses generated from a diverse pool of 14 LLMs. These responses are ranked based on five distinct coding preferences using GPT-3.5 as a judge, providing both numerical scores and detailed textual feedback. Our analysis of CodeUltraFeedback reveals that responses from GPT-3.5 and GPT-4 are generally preferred over those from open-weight LLMs, highlighting significant differences in alignment between closed and open-weight models. In turn, we explore the usage of CodeUltraFeedback as feedback data to fine-tune and align CodeLlama-7B-Instruct using supervised fine-tuning (SFT) and reinforcement learning from AI feedback (RLAIF) with direct preference optimization (DPO). The resulting aligned CodeLlama-7B-Instruct model outperforms larger LLMs in terms of alignment with coding preferences and shows improved functional correctness on the HumanEval+ benchmark compared to the original instruct model. Therefore, our contributions bridge the gap in preference tuning of LLMs for code and set the stage for further advancements in model alignment and RLAIF in automated software engineering.
Authors: Rui Pan, Xiang Liu, Shizhe Diao, Renjie Pi, Jipeng Zhang, Chi Han, Tong Zhang
Abstract: The machine learning community has witnessed impressive advancements since large language models (LLMs) first appeared. Yet, their massive memory consumption has become a significant roadblock to large-scale training. For instance, a 7B model typically requires at least 60 GB of GPU memory with full parameter training, which presents challenges for researchers without access to high-resource environments. Parameter Efficient Fine-Tuning techniques such as Low-Rank Adaptation (LoRA) have been proposed to alleviate this problem. However, in most large-scale fine-tuning settings, their performance does not reach the level of full parameter training because they confine the parameter search to a low-rank subspace. Attempting to complement this deficiency, we investigate the layerwise properties of LoRA on fine-tuning tasks and observe an unexpected but consistent skewness of weight norms across different layers. Utilizing this key observation, a surprisingly simple training strategy is discovered, which outperforms both LoRA and full parameter training in a wide range of settings with memory costs as low as LoRA. We name it Layerwise Importance Sampled AdamW (LISA), a promising alternative for LoRA, which applies the idea of importance sampling to different layers in LLMs and randomly freezes most middle layers during optimization. Experimental results show that with similar or less GPU memory consumption, LISA surpasses LoRA or even full parameter tuning in downstream fine-tuning tasks, where LISA consistently outperforms LoRA by over 10%-35% in terms of MT-Bench score while achieving on-par or better performance in MMLU, AGIEval and WinoGrande. On large models, specifically LLaMA-2-70B, LISA surpasses LoRA on MT-Bench, GSM8K, and PubMedQA, demonstrating its effectiveness across different domains.
Authors: Richard Ren, Steven Basart, Adam Khoja, Alice Gatti, Long Phan, Xuwang Yin, Mantas Mazeika, Alexander Pan, Gabriel Mukobi, Ryan H. Kim, Stephen Fitz, Dan Hendrycks
Abstract: As artificial intelligence systems grow more powerful, there has been increasing interest in "AI safety" research to address emerging and future risks. However, the field of AI safety remains poorly defined and inconsistently measured, leading to confusion about how researchers can contribute. This lack of clarity is compounded by the unclear relationship between AI safety benchmarks and upstream general capabilities (e.g., general knowledge and reasoning). To address these issues, we conduct a comprehensive meta-analysis of AI safety benchmarks, empirically analyzing their correlation with general capabilities across dozens of models and providing a survey of existing directions in AI safety. Our findings reveal that many safety benchmarks highly correlate with both upstream model capabilities and training compute, potentially enabling "safetywashing"--where capability improvements are misrepresented as safety advancements. Based on these findings, we propose an empirical foundation for developing more meaningful safety metrics and define AI safety in a machine learning research context as a set of clearly delineated research goals that are empirically separable from generic capabilities advancements. In doing so, we aim to provide a more rigorous framework for AI safety research, advancing the science of safety evaluations and clarifying the path towards measurable progress.
Authors: Yanhu Wang, Muhammad Muzammil Afzal, Zhengyang Li, Jie Zhou, Chenyuan Feng, Shuaishuai Guo, Tony Q. S. Quek
Abstract: Traditional base station siting (BSS) methods rely heavily on drive testing and user feedback, which are laborious and require extensive expertise in communication, networking, and optimization. As large language models (LLMs) and their associated technologies advance, particularly in the realms of prompt engineering and agent engineering, network optimization will witness a revolutionary approach. This approach entails the strategic use of well-crafted prompts to infuse human experience and knowledge into these sophisticated LLMs, and the deployment of autonomous agents as a communication bridge to seamlessly connect the machine language based LLMs with human users using natural language. Furthermore, our proposed framework incorporates retrieval-augmented generation (RAG) to enhance the system's ability to acquire domain-specific knowledge and generate solutions, thereby enabling the customization and optimization of the BSS process. This integration represents the future paradigm of artificial intelligence (AI) as a service and AI for more ease. This research first develops a novel LLM-empowered BSS optimization framework, and heuristically proposes three different potential implementations: the strategies based on Prompt-optimized LLM (PoL), LLM-empowered autonomous BSS agent (LaBa), and Cooperative multiple LLM-based autonomous BSS agents (CLaBa). Through evaluation on real-world data, the experiments demonstrate that prompt-assisted LLMs and LLM-based agents can generate more efficient and reliable network deployments, noticeably enhancing the efficiency of BSS optimization and reducing trivial manual participation.
Authors: Pei-Fu Guo, Yun-Da Tsai, Shou-De Lin
Abstract: Prompt optimization algorithms for Large Language Models (LLMs) excel in multi-step reasoning but still lack effective uncertainty estimation. This paper introduces a benchmark dataset to evaluate uncertainty metrics, focusing on Answer, Correctness, Aleatoric, and Epistemic Uncertainty. Through analysis of models like GPT-3.5-Turbo and Meta-Llama-3.1-8B-Instruct, we show that current metrics align more with Answer Uncertainty, which reflects output confidence and diversity, rather than Correctness Uncertainty, highlighting the need for improved metrics that are optimization-objective-aware to better guide prompt optimization. Our code and dataset are available at https://github.com/0Frett/PO-Uncertainty-Benchmarking.
URLs: https://github.com/0Frett/PO-Uncertainty-Benchmarking.
Authors: Weihao Ye, Qiong Wu, Wenhao Lin, Yiyi Zhou
Abstract: Recent progress in Multimodal Large Language Models(MLLMs) often use large image tokens to compensate the visual shortcoming of MLLMs, which not only exhibits obvious redundancy but also greatly exacerbates the already high computation. Token pruning is an effective solution for speeding up MLLMs, but when and how to drop tokens still remains a challenge. In this paper, we propose a novel and training-free approach for the effective visual token pruning of MLLMs, termed FitPrune, which can quickly produce a complete pruning recipe for MLLMs according to a pre-defined budget. Specifically, FitPrune considers token pruning as a statistical problem of MLLM and its objective is to find out an optimal pruning scheme that can minimize the divergence of the attention distributions before and after pruning. In practice, FitPrune can be quickly accomplished based on the attention statistics from a small batch of inference data, avoiding the expensive trials of MLLMs. According to the pruning recipe, an MLLM can directly remove the redundant visual tokens of different examples during inference. To validate FitPrune, we apply it to a set of recent MLLMs, including LLaVA-1.5, LLaVA-HR and LLaVA-NEXT, and conduct extensive experiments on a set of benchmarks. The experimental results show that our FitPrune can not only reduce the computational complexity to a large extent, while retaining high performance, e.g., -54.9% FLOPs for LLaVA-NEXT with only 0.5% accuracy drop. Notably, the pruning recipe can be obtained in about 5 minutes. Our code is available at https://github.com/ywh187/FitPrune.
Authors: Ying Fang, Xiaofei Li
Abstract: This paper works on streaming automatic speech recognition (ASR). Mamba, a recently proposed state space model, has demonstrated the ability to match or surpass Transformers in various tasks while benefiting from a linear complexity advantage. We explore the efficiency of Mamba encoder for streaming ASR and propose an associated lookahead mechanism for leveraging controllable future information. Additionally, a streaming-style unimodal aggregation (UMA) method is implemented, which automatically detects token activity and streamingly triggers token output, and meanwhile aggregates feature frames for better learning token representation. Based on UMA, an early termination (ET) method is proposed to further reduce recognition latency. Experiments conducted on two Mandarin Chinese datasets demonstrate that the proposed model achieves competitive ASR performance in terms of both recognition accuracy and latency.
Authors: Yadong Li, Haoze Sun, Mingan Lin, Tianpeng Li, Guosheng Dong, Tao Zhang, Bowen Ding, Wei Song, Zhenglin Cheng, Yuqi Huo, Song Chen, Xu Li, Da Pan, Shusen Zhang, Xin Wu, Zheng Liang, Jun Liu, Tao Zhang, Keer Lu, Yaqi Zhao, Yanjun Shen, Fan Yang, Kaicheng Yu, Tao Lin, Jianhua Xu, Zenan Zhou, Weipeng Chen
Abstract: The salient multimodal capabilities and interactive experience of GPT-4o highlight its critical role in practical applications, yet it lacks a high-performing open-source counterpart. In this paper, we introduce Baichuan-omni, the first open-source 7B Multimodal Large Language Model (MLLM) adept at concurrently processing and analyzing modalities of image, video, audio, and text, while delivering an advanced multimodal interactive experience and strong performance. We propose an effective multimodal training schema starting with 7B model and proceeding through two stages of multimodal alignment and multitask fine-tuning across audio, image, video, and text modal. This approach equips the language model with the ability to handle visual and audio data effectively. Demonstrating strong performance across various omni-modal and multimodal benchmarks, we aim for this contribution to serve as a competitive baseline for the open-source community in advancing multimodal understanding and real-time interaction.
Authors: Zifan Liu, Amin Karbasi, Theodoros Rekatsinas
Abstract: Finetuning foundation models for specific tasks is an emerging paradigm in modern machine learning. The efficacy of task-specific finetuning largely depends on the selection of appropriate training data. We present TSDS (Task-Specific Data Selection), a framework to select data for task-specific model finetuning, guided by a small but representative set of examples from the target task. To do so, we formulate data selection for task-specific finetuning as an optimization problem with a distribution alignment loss based on optimal transport to capture the discrepancy between the selected data and the target distribution. In addition, we add a regularizer to encourage the diversity of the selected data and incorporate kernel density estimation into the regularizer to reduce the negative effects of near-duplicates among the candidate data. We connect our optimization problem to nearest neighbor search and design efficient algorithms to compute the optimal solution based on approximate nearest neighbor search techniques. We evaluate our method on data selection for both continued pretraining and instruction tuning of language models. We show that instruction tuning using data selected by our method with a 1% selection ratio often outperforms using the full dataset and beats the baseline selection methods by 1.5 points in F1 score on average.
Authors: Mingan Lin, Fan Yang, Yanjun Shen, Haoze Sun, Tianpeng Li, Tao Zhang, Chenzheng Zhu, Tao Zhang, Miao Zheng, Xu Li, Yijie Zhou, Mingyang Chen, Yanzhao Qin, Youquan Li, Hao Liang, Fei Li, Yadong Li, Mang Wang, Guosheng Dong, Kun Fang, Jianhua Xu, Bin Cui, Wentao Zhang, Zenan Zhou, Weipeng Chen
Abstract: We introduce Baichuan Alignment, a detailed analysis of the alignment techniques employed in the Baichuan series of models. This represents the industry's first comprehensive account of alignment methodologies, offering valuable insights for advancing AI research. We investigate the critical components that enhance model performance during the alignment process, including optimization methods, data strategies, capability enhancements, and evaluation processes. The process spans three key stages: Prompt Augmentation System(PAS), Supervised Fine-Tuning(SFT), and Preference Alignment. The problems encountered, the solutions applied, and the improvements made are thoroughly recorded. Through comparisons across well-established benchmarks, we highlight the technological advancements enabled by Baichuan Alignment. Baichuan-Instruct is an internal model, while Qwen2-Nova-72B and Llama3-PBM-Nova-70B are instruct versions of the Qwen2-72B and Llama-3-70B base models, optimized through Baichuan Alignment. Baichuan-Instruct demonstrates significant improvements in core capabilities, with user experience gains ranging from 17% to 28%, and performs exceptionally well on specialized benchmarks. In open-source benchmark evaluations, both Qwen2-Nova-72B and Llama3-PBM-Nova-70B consistently outperform their respective official instruct versions across nearly all datasets. This report aims to clarify the key technologies behind the alignment process, fostering a deeper understanding within the community. Llama3-PBM-Nova-70B model is available at https://huggingface.co/PKU-Baichuan-MLSystemLab/Llama3-PBM-Nova-70B.
URLs: https://huggingface.co/PKU-Baichuan-MLSystemLab/Llama3-PBM-Nova-70B.
Authors: Muzhi Li, Cehao Yang, Chengjin Xu, Zixing Song, Xuhui Jiang, Jian Guo, Ho-fung Leung, Irwin King
Abstract: Inductive knowledge graph completion (KGC) aims to predict missing triples with unseen entities. Recent works focus on modeling reasoning paths between the head and tail entity as direct supporting evidence. However, these methods depend heavily on the existence and quality of reasoning paths, which limits their general applicability in different scenarios. In addition, we observe that latent type constraints and neighboring facts inherent in KGs are also vital in inferring missing triples. To effectively utilize all useful information in KGs, we introduce CATS, a novel context-aware inductive KGC solution. With sufficient guidance from proper prompts and supervised fine-tuning, CATS activates the strong semantic understanding and reasoning capabilities of large language models to assess the existence of query triples, which consist of two modules. First, the type-aware reasoning module evaluates whether the candidate entity matches the latent entity type as required by the query relation. Then, the subgraph reasoning module selects relevant reasoning paths and neighboring facts, and evaluates their correlation to the query triple. Experiment results on three widely used datasets demonstrate that CATS significantly outperforms state-of-the-art methods in 16 out of 18 transductive, inductive, and few-shot settings with an average absolute MRR improvement of 7.2%.
Authors: Khurram Yamin, Shantanu Gupta, Gaurav R. Ghosal, Zachary C. Lipton, Bryan Wilder
Abstract: In this work, we investigate the causal reasoning abilities of large language models (LLMs) through the representative problem of inferring causal relationships from narratives. We find that even state-of-the-art language models rely on unreliable shortcuts, both in terms of the narrative presentation and their parametric knowledge. For example, LLMs tend to determine causal relationships based on the topological ordering of events (i.e., earlier events cause later ones), resulting in lower performance whenever events are not narrated in their exact causal order. Similarly, we demonstrate that LLMs struggle with long-term causal reasoning and often fail when the narratives are long and contain many events. Additionally, we show LLMs appear to rely heavily on their parametric knowledge at the expense of reasoning over the provided narrative. This degrades their abilities whenever the narrative opposes parametric knowledge. We extensively validate these failure modes through carefully controlled synthetic experiments, as well as evaluations on real-world narratives. Finally, we observe that explicitly generating a causal graph generally improves performance while naive chain-of-thought is ineffective. Collectively, our results distill precise failure modes of current state-of-the-art models and can pave the way for future techniques to enhance causal reasoning in LLMs.
Authors: Chaeyun Jang, Hyungi Lee, Jungtaek Kim, Juho Lee
Abstract: Fine-tuning pre-trained models for downstream tasks is a widely adopted technique known for its adaptability and reliability across various domains. Despite its conceptual simplicity, fine-tuning entails several troublesome engineering choices, such as selecting hyperparameters and determining checkpoints from an optimization trajectory. To tackle the difficulty of choosing the best model, one effective solution is model fusion, which combines multiple models in a parameter space. However, we observe a large discrepancy between loss and metric landscapes during the fine-tuning of pre-trained language models. Building on this observation, we introduce a novel model fusion technique that optimizes both the desired metric and loss through multi-objective Bayesian optimization. In addition, to effectively select hyperparameters, we establish a two-stage procedure by integrating Bayesian optimization processes into our framework. Experiments across various downstream tasks show considerable performance improvements using our Bayesian optimization-guided method.
Authors: Dingyuan Shi, Yong Wang, Hangyu Li, Xiangxiang Chu
Abstract: Diffusion models have shown remarkable success in text-to-image generation, making alignment methods for these models increasingly important. A key challenge is the sparsity of preference labels, which are typically available only at the terminal of denoising trajectories. This raises the issue of how to assign credit across denoising steps based on these sparse labels. In this paper, we propose Denoised Distribution Estimation (DDE), a novel method for credit assignment. Unlike previous approaches that rely on auxiliary models or hand-crafted schemes, DDE derives its strategy more explicitly. The proposed DDE directly estimates the terminal denoised distribution from the perspective of each step. It is equipped with two estimation strategies and capable of representing the entire denoising trajectory with a single model inference. Theoretically and empirically, we show that DDE prioritizes optimizing the middle part of the denoising trajectory, resulting in a novel and effective credit assignment scheme. Extensive experiments demonstrate that our approach achieves superior performance, both quantitatively and qualitatively.
Authors: Dawei Li, Bohan Jiang, Liangjie Huang, Alimohammad Beigi, Chengshuai Zhao, Zhen Tan, Amrita Bhattacharjee, Yuxuan Jiang, Canyu Chen, Tianhao Wu, Kai Shu, Lu Cheng, Huan Liu
Abstract: Assessment and evaluation have long been critical challenges in artificial intelligence (AI) and natural language processing (NLP). However, traditional methods, whether matching-based or embedding-based, often fall short of judging subtle attributes and delivering satisfactory results. Recent advancements in Large Language Models (LLMs) inspire the "LLM-as-a-judge" paradigm, where LLMs are leveraged to perform scoring, ranking, or selection across various tasks and applications. This paper provides a comprehensive survey of LLM-based judgment and assessment, offering an in-depth overview to advance this emerging field. We begin by giving detailed definitions from both input and output perspectives. Then we introduce a comprehensive taxonomy to explore LLM-as-a-judge from three dimensions: what to judge, how to judge and where to judge. Finally, we compile benchmarks for evaluating LLM-as-a-judge and highlight key challenges and promising directions, aiming to provide valuable insights and inspire future research in this promising research area. Paper list and more resources about LLM-as-a-judge can be found at \url{https://github.com/llm-as-a-judge/Awesome-LLM-as-a-judge} and \url{https://llm-as-a-judge.github.io}.
URLs: https://github.com/llm-as-a-judge/Awesome-LLM-as-a-judge, https://llm-as-a-judge.github.io
Authors: Lanxiang Hu, Qiyu Li, Anze Xie, Nan Jiang, Ion Stoica, Haojian Jin, Hao Zhang
Abstract: Evaluating the reasoning abilities of large language models (LLMs) is challenging. Existing benchmarks often depend on static datasets, which are vulnerable to data contamination and may get saturated over time, or on binary live human feedback that conflates reasoning with other abilities. As the most prominent dynamic benchmark, Chatbot Arena evaluates open-ended questions in real-world settings, but lacks the granularity in assessing specific reasoning capabilities. We introduce GameArena, a dynamic benchmark designed to evaluate LLM reasoning capabilities through interactive gameplay with humans. GameArena consists of three games designed to test specific reasoning capabilities (e.g., deductive and inductive reasoning), while keeping participants entertained and engaged. We analyze the gaming data retrospectively to uncover the underlying reasoning processes of LLMs and measure their fine-grained reasoning capabilities. We collect over 2000 game sessions and provide detailed assessments of various reasoning capabilities for five state-of-the-art LLMs. Our user study with 100 participants suggests that GameArena improves user engagement compared to Chatbot Arena. For the first time, GameArena enables the collection of step-by-step LLM reasoning data in the wild.
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: Zijun Chen, Wenbo Hu, Guande He, Zhijie Deng, Zheng Zhang, Richang Hong
Abstract: Multimodal large language models (MLLMs) combine visual and textual data for tasks such as image captioning and visual question answering. Proper uncertainty calibration is crucial, yet challenging, for reliable use in areas like healthcare and autonomous driving. This paper investigates representative MLLMs, focusing on their calibration across various scenarios, including before and after visual fine-tuning, as well as before and after multimodal training of the base LLMs. We observed miscalibration in their performance, and at the same time, no significant differences in calibration across these scenarios. We also highlight how uncertainty differs between text and images and how their integration affects overall uncertainty. To better understand MLLMs' miscalibration and their ability to self-assess uncertainty, we construct the IDK (I don't know) dataset, which is key to evaluating how they handle unknowns. Our findings reveal that MLLMs tend to give answers rather than admit uncertainty, but this self-assessment improves with proper prompt adjustments. Finally, to calibrate MLLMs and enhance model reliability, we propose techniques such as temperature scaling and iterative prompt optimization. Our results provide insights into improving MLLMs for effective and responsible deployment in multimodal applications. Code and IDK dataset: https://github.com/hfutml/Calibration-MLLM.
Authors: Huaijie Wang, Shibo Hao, Hanze Dong, Shenao Zhang, Yilin Bao, Ziran Yang, Yi Wu
Abstract: Improving the multi-step reasoning ability of large language models (LLMs) with offline reinforcement learning (RL) is essential for quickly adapting them to complex tasks. While Direct Preference Optimization (DPO) has shown promise in aligning LLMs with human preferences, it is less suitable for multi-step reasoning tasks because (1) DPO relies on paired preference data, which is not readily available for multi-step reasoning tasks, and (2) it treats all tokens uniformly, making it ineffective for credit assignment in multi-step reasoning tasks, which often come with sparse reward. In this work, we propose OREO (Offline Reasoning Optimization), an offline RL method for enhancing LLM multi-step reasoning. Building on insights from previous works of maximum entropy reinforcement learning, it jointly learns a policy model and value function by optimizing the soft Bellman Equation. We show in principle that it reduces the need to collect pairwise data and enables better credit assignment. Empirically, OREO surpasses existing offline learning methods on multi-step reasoning benchmarks, including mathematical reasoning tasks (GSM8K, MATH) and embodied agent control (ALFWorld). The approach can be extended to a multi-iteration framework when additional resources are available. Furthermore, the learned value function can be leveraged to guide the tree search for free, which can further boost performance during test time.
Authors: Chao Deng, Jiale Yuan, Pi Bu, Peijie Wang, Zhong-Zhi Li, Jian Xu, Xiao-Hui Li, Yuan Gao, Jun Song, Bo Zheng, Cheng-Lin Liu
Abstract: Large vision language models (LVLMs) have improved the document understanding capabilities remarkably, enabling the handling of complex document elements, longer contexts, and a wider range of tasks. However, existing document understanding benchmarks have been limited to handling only a small number of pages and fail to provide a comprehensive analysis of layout elements locating. In this paper, we first define three primary task categories: Long Document Understanding, numerical Reasoning, and cross-element Locating, and then propose a comprehensive benchmark, LongDocURL, integrating above three primary tasks and comprising 20 sub-tasks categorized based on different primary tasks and answer evidences. Furthermore, we develop a semi-automated construction pipeline and collect 2,325 high-quality question-answering pairs, covering more than 33,000 pages of documents, significantly outperforming existing benchmarks. Subsequently, we conduct comprehensive evaluation experiments on both open-source and closed-source models across 26 different configurations, revealing critical performance gaps in this field.