Authors: Sabur Butt, Fazlourrahman Balouchzahi, Ahmad Imam Amjad, Maaz Amjad, Hector G. Ceballos, Salud Maria Jimenez-Zafra
Abstract: Hope is a complex and underexplored emotional state that plays a significant role in education, mental health, and social interaction. Unlike basic emotions, hope manifests in nuanced forms ranging from grounded optimism to exaggerated wishfulness or sarcasm, making it difficult for Natural Language Processing systems to detect accurately. This study introduces PolyHope V2, a multilingual, fine-grained hope speech dataset comprising over 30,000 annotated tweets in English and Spanish. This resource distinguishes between four hope subtypes Generalized, Realistic, Unrealistic, and Sarcastic and enhances existing datasets by explicitly labeling sarcastic instances. We benchmark multiple pretrained transformer models and compare them with large language models (LLMs) such as GPT 4 and Llama 3 under zero-shot and few-shot regimes. Our findings show that fine-tuned transformers outperform prompt-based LLMs, especially in distinguishing nuanced hope categories and sarcasm. Through qualitative analysis and confusion matrices, we highlight systematic challenges in separating closely related hope subtypes. The dataset and results provide a robust foundation for future emotion recognition tasks that demand greater semantic and contextual sensitivity across languages.
Authors: Brihi Joshi, Xiang Ren, Swabha Swayamdipta, Rik Koncel-Kedziorski, Tim Paek
Abstract: Language models prompted with a user description or persona can predict a user's preferences and opinions, but existing approaches to building personas -- based solely on a user's demographic attributes and/or prior judgments -- fail to capture the underlying reasoning behind said user judgments. We introduce PB&J (Psychology of Behavior and Judgments), a framework that improves LLM personas by incorporating rationales of why a user might make specific judgments. These rationales are LLM-generated, and aim to reason about a user's behavior on the basis of their experiences, personality traits or beliefs. This is done using psychological scaffolds -- structured frameworks grounded in theories such as the Big 5 Personality Traits and Primal World Beliefs -- that help provide structure to the generated rationales. Experiments on public opinion and movie preference prediction tasks demonstrate that LLM personas augmented with PB&J rationales consistently outperform methods using only a user's demographics and/or judgments. Additionally, LLM personas constructed using scaffolds describing user beliefs perform competitively with those using human-written rationales.
Authors: Zhuang Yu, Shiliang Sun, Jing Zhao, Tengfei Song, Hao Yang
Abstract: Multimodal Machine Translation (MMT) aims to improve translation quality by leveraging auxiliary modalities such as images alongside textual input. While recent advances in large-scale pre-trained language and vision models have significantly benefited unimodal natural language processing tasks, their effectiveness and role in MMT remain underexplored. In this work, we conduct a systematic study on the impact of pre-trained encoders and decoders in multimodal translation models. Specifically, we analyze how different training strategies, from training from scratch to using pre-trained and partially frozen components, affect translation performance under a unified MMT framework. Experiments are carried out on the Multi30K and CoMMuTE dataset across English-German and English-French translation tasks. Our results reveal that pre-training plays a crucial yet asymmetrical role in multimodal settings: pre-trained decoders consistently yield more fluent and accurate outputs, while pre-trained encoders show varied effects depending on the quality of visual-text alignment. Furthermore, we provide insights into the interplay between modality fusion and pre-trained components, offering guidance for future architecture design in multimodal translation systems.
Authors: Bang An, Shiyue Zhang, Mark Dredze
Abstract: Efforts to ensure the safety of large language models (LLMs) include safety fine-tuning, evaluation, and red teaming. However, despite the widespread use of the Retrieval-Augmented Generation (RAG) framework, AI safety work focuses on standard LLMs, which means we know little about how RAG use cases change a model's safety profile. We conduct a detailed comparative analysis of RAG and non-RAG frameworks with eleven LLMs. We find that RAG can make models less safe and change their safety profile. We explore the causes of this change and find that even combinations of safe models with safe documents can cause unsafe generations. In addition, we evaluate some existing red teaming methods for RAG settings and show that they are less effective than when used for non-RAG settings. Our work highlights the need for safety research and red-teaming methods specifically tailored for RAG LLMs.
Authors: Jianyu Liu, Hangyu Guo, Ranjie Duan, Xingyuan Bu, Yancheng He, Shilong Li, Hui Huang, Jiaheng Liu, Yucheng Wang, Chenchen Jing, Xingwei Qu, Xiao Zhang, Yingshui Tan, Yanan Wu, Jihao Gu, Yangguang Li, Jianke Zhu
Abstract: Multimodal Large Language Models (MLLMs) pose unique safety challenges due to their integration of visual and textual data, thereby introducing new dimensions of potential attacks and complex risk combinations. In this paper, we begin with a detailed analysis aimed at disentangling risks through step-by-step reasoning within multimodal inputs. We find that systematic multimodal risk disentanglement substantially enhances the risk awareness of MLLMs. Via leveraging the strong discriminative abilities of multimodal risk disentanglement, we further introduce \textbf{DREAM} (\textit{\textbf{D}isentangling \textbf{R}isks to \textbf{E}nhance Safety \textbf{A}lignment in \textbf{M}LLMs}), a novel approach that enhances safety alignment in MLLMs through supervised fine-tuning and iterative Reinforcement Learning from AI Feedback (RLAIF). Experimental results show that DREAM significantly boosts safety during both inference and training phases without compromising performance on normal tasks (namely oversafety), achieving a 16.17\% improvement in the SIUO safe\&effective score compared to GPT-4V. The data and code are available at https://github.com/Kizna1ver/DREAM.
Authors: Sijia Cheng, Wen-Yu Chang, Yun-Nung Chen
Abstract: The integration of dialogue agents into the sales domain requires a deep understanding of how these systems interact with users possessing diverse personas. This study explores the influence of user personas, defined using the Myers-Briggs Type Indicator (MBTI), on the interaction quality and performance of sales-oriented dialogue agents. Through large-scale testing and analysis, we assess the pre-trained agent's effectiveness, adaptability, and personalization capabilities across a wide range of MBTI-defined user types. Our findings reveal significant patterns in interaction dynamics, task completion rates, and dialogue naturalness, underscoring the future potential for dialogue agents to refine their strategies to better align with varying personality traits. This work not only provides actionable insights for building more adaptive and user-centric conversational systems in the sales domain but also contributes broadly to the field by releasing persona-defined user simulators. These simulators, unconstrained by domain, offer valuable tools for future research and demonstrate the potential for scaling personalized dialogue systems across diverse applications.
Authors: Jingjin Wang
Abstract: Retrieval Augmented Generation (RAG) has become the standard non-parametric approach for equipping Large Language Models (LLMs) with up-to-date knowledge and mitigating catastrophic forgetting common in continual learning. However, standard RAG, relying on independent passage retrieval, fails to capture the interconnected nature of human memory crucial for complex reasoning (associativity) and contextual understanding (sense-making). While structured RAG methods like HippoRAG utilize knowledge graphs (KGs) built from triples, the inherent context loss limits fidelity. We introduce PropRAG, a framework leveraging contextually rich propositions and a novel beam search algorithm over proposition paths to explicitly discover multi-step reasoning chains. Crucially, PropRAG's online retrieval process operates entirely without invoking generative LLMs, relying instead on efficient graph traversal and pre-computed embeddings. This avoids online LLM inference costs and potential inconsistencies during evidence gathering. LLMs are used effectively offline for high-quality proposition extraction and post-retrieval for answer generation. PropRAG achieves state-of-the-art zero-shot Recall@5 results on PopQA (55.3%), 2Wiki (93.7%), HotpotQA (97.0%), and MuSiQue (77.3%), alongside top F1 scores (e.g., 52.4% on MuSiQue). By improving evidence retrieval through richer representation and explicit, LLM-free online path finding, PropRAG advances non-parametric continual learning.
Authors: Wataru Kawakami, Keita Suzuki, Junichiro Iwasawa
Abstract: Large Language Models (LLMs) show potential in medicine, yet clinical adoption is hindered by concerns over factual accuracy, language-specific limitations (e.g., Japanese), and critically, their reliability when required to generate reasoning explanations -- a prerequisite for trust. This paper introduces Preferred-MedLLM-Qwen-72B, a 72B-parameter model optimized for the Japanese medical domain to achieve both high accuracy and stable reasoning. We employ a two-stage fine-tuning process on the Qwen2.5-72B base model: first, Continued Pretraining (CPT) on a comprehensive Japanese medical corpus instills deep domain knowledge. Second, Reasoning Preference Optimization (RPO), a preference-based method, enhances the generation of reliable reasoning pathways while preserving high answer accuracy. Evaluations on the Japanese Medical Licensing Exam benchmark (IgakuQA) show Preferred-MedLLM-Qwen-72B achieves state-of-the-art performance (0.868 accuracy), surpassing strong proprietary models like GPT-4o (0.866). Crucially, unlike baseline or CPT-only models which exhibit significant accuracy degradation (up to 11.5\% and 3.8\% respectively on IgakuQA) when prompted for explanations, our model maintains its high accuracy (0.868) under such conditions. This highlights RPO's effectiveness in stabilizing reasoning generation. This work underscores the importance of optimizing for reliable explanations alongside accuracy. We release the Preferred-MedLLM-Qwen-72B model weights to foster research into trustworthy LLMs for specialized, high-stakes applications.
Authors: Muhammad Mubashar, Shireen Kudukkil Manchingal, Fabio Cuzzolin
Abstract: Large Language Models (LLMs) are known to produce very high-quality tests and responses to our queries. But how much can we trust this generated text? In this paper, we study the problem of uncertainty quantification in LLMs. We propose a novel Random-Set Large Language Model (RSLLM) approach which predicts finite random sets (belief functions) over the token space, rather than probability vectors as in classical LLMs. In order to allow so efficiently, we also present a methodology based on hierarchical clustering to extract and use a budget of "focal" subsets of tokens upon which the belief prediction is defined, rather than using all possible collections of tokens, making the method scalable yet effective. RS-LLMs encode the epistemic uncertainty induced in their generation process by the size and diversity of its training set via the size of the credal sets associated with the predicted belief functions. The proposed approach is evaluated on CoQA and OBQA datasets using Llama2-7b, Mistral-7b and Phi-2 models and is shown to outperform the standard model in both datasets in terms of correctness of answer while also showing potential in estimating the second level uncertainty in its predictions and providing the capability to detect when its hallucinating.
Authors: Yinglong Yu, Hao Shen, Zhengyi Lyu, Qi He
Abstract: In response to the growing problem of misinformation in the context of globalization and informatization, this paper proposes a classification method for fact-check-worthiness estimation based on prompt tuning. We construct a model for fact-check-worthiness estimation at the methodological level using prompt tuning. By applying designed prompt templates to large language models, we establish in-context learning and leverage prompt tuning technology to improve the accuracy of determining whether claims have fact-check-worthiness, particularly when dealing with limited or unlabeled data. Through extensive experiments on public datasets, we demonstrate that the proposed method surpasses or matches multiple baseline methods in the classification task of fact-check-worthiness estimation assessment, including classical pre-trained models such as BERT, as well as recent popular large models like GPT-3.5 and GPT-4. Experiments show that the prompt tuning-based method proposed in this study exhibits certain advantages in evaluation metrics such as F1 score and accuracy, thereby effectively validating its effectiveness and advancement in the task of fact-check-worthiness estimation.
Authors: Yinglong Yu, Zhaopu Yao, Fang Yuan
Abstract: This study analyzes Chinese and English media reports on the Paris Olympics using topic modeling, Large Language Model (LLM) prompt engineering, and corpus phraseology methods to explore similarities and differences in discourse construction and attitudinal meanings. Common topics include the opening ceremony, athlete performance, and sponsorship brands. Chinese media focus on specific sports, sports spirit, doping controversies, and new technologies, while English media focus on female athletes, medal wins, and eligibility controversies. Chinese reports show more frequent prepositional co-occurrences and positive semantic prosody in describing the opening ceremony and sports spirit. English reports exhibit positive semantic prosody when covering female athletes but negative prosody in predicting opening ceremony reactions and discussing women's boxing controversies.
Authors: Atharva Kulkarni, Yuan Zhang, Joel Ruben Antony Moniz, Xiou Ge, Bo-Hsiang Tseng, Dhivya Piraviperumal, Swabha Swayamdipta, Hong Yu
Abstract: Hallucinations pose a significant obstacle to the reliability and widespread adoption of language models, yet their accurate measurement remains a persistent challenge. While many task- and domain-specific metrics have been proposed to assess faithfulness and factuality concerns, the robustness and generalization of these metrics are still untested. In this paper, we conduct a large-scale empirical evaluation of 6 diverse sets of hallucination detection metrics across 4 datasets, 37 language models from 5 families, and 5 decoding methods. Our extensive investigation reveals concerning gaps in current hallucination evaluation: metrics often fail to align with human judgments, take an overtly myopic view of the problem, and show inconsistent gains with parameter scaling. Encouragingly, LLM-based evaluation, particularly with GPT-4, yields the best overall results, and mode-seeking decoding methods seem to reduce hallucinations, especially in knowledge-grounded settings. These findings underscore the need for more robust metrics to understand and quantify hallucinations, and better strategies to mitigate them.
Authors: Tatsunori Tanaka, Fi Zheng, Kai Sato, Zhifeng Li, Yuanyun Zhang, Shi Li
Abstract: Clinical language models have achieved strong performance on downstream tasks by pretraining on domain specific corpora such as discharge summaries and medical notes. However, most approaches treat the electronic health record as a static document, neglecting the temporally-evolving and causally entwined nature of patient trajectories. In this paper, we introduce a novel temporal entailment pretraining objective for language models in the clinical domain. Our method formulates EHR segments as temporally ordered sentence pairs and trains the model to determine whether a later state is entailed by, contradictory to, or neutral with respect to an earlier state. Through this temporally structured pretraining task, models learn to perform latent clinical reasoning over time, improving their ability to generalize across forecasting and diagnosis tasks. We pretrain on a large corpus derived from MIMIC IV and demonstrate state of the art results on temporal clinical QA, early warning prediction, and disease progression modeling.
Authors: Fida Ullah, Muhammad Ahmad, Muhammad Tayyab Zamir, Muhammad Arif, Grigori sidorov, Edgardo Manuel Felipe River\'on, Alexander Gelbukh
Abstract: Named Entity Recognition (NER) plays a pivotal role in various Natural Language Processing (NLP) tasks by identifying and classifying named entities (NEs) from unstructured data into predefined categories such as person, organization, location, date, and time. While extensive research exists for high-resource languages and general domains, NER in Urdu particularly within domain-specific contexts like education remains significantly underexplored. This is Due to lack of annotated datasets for educational content which limits the ability of existing models to accurately identify entities such as academic roles, course names, and institutional terms, underscoring the urgent need for targeted resources in this domain. To the best of our knowledge, no dataset exists in the domain of the Urdu language for this purpose. To achieve this objective this study makes three key contributions. Firstly, we created a manually annotated dataset in the education domain, named EDU-NER-2025, which contains 13 unique most important entities related to education domain. Second, we describe our annotation process and guidelines in detail and discuss the challenges of labelling EDU-NER-2025 dataset. Third, we addressed and analyzed key linguistic challenges, such as morphological complexity and ambiguity, which are prevalent in formal Urdu texts.
Authors: {\TH}\'orir Hrafn Har{\dh}arson, Hrafn Loftsson, Stef\'an \'Olafsson
Abstract: The integration of language models in the legal domain holds considerable promise for streamlining processes and improving efficiency in managing extensive workloads. However, the specialized terminology, nuanced language, and formal style of legal texts can present substantial challenges. This study examines whether preference-based training techniques, specifically Reinforcement Learning from Human Feedback and Direct Preference Optimization, can enhance models' performance in generating Icelandic legal summaries that align with domain-specific language standards and user preferences. We compare models fine-tuned with preference training to those using conventional supervised learning. Results indicate that preference training improves the legal accuracy of generated summaries over standard fine-tuning but does not significantly enhance the overall quality of Icelandic language usage. Discrepancies between automated metrics and human evaluations further underscore the importance of qualitative assessment in developing language models for the legal domain.
Authors: Shuxiang Du, Ana Guerberof Arenas, Antonio Toral, Kyo Gerrits, Josep Marco Borillo
Abstract: This study examines the variability of Chat-GPT machine translation (MT) outputs across six different configurations in four languages,with a focus on creativity in a literary text. We evaluate GPT translations in different text granularity levels, temperature settings and prompting strategies with a Creativity Score formula. We found that prompting ChatGPT with a minimal instruction yields the best creative translations, with "Translate the following text into [TG] creatively" at the temperature of 1.0 outperforming other configurations and DeepL in Spanish, Dutch, and Chinese. Nonetheless, ChatGPT consistently underperforms compared to human translation (HT).
Authors: Pierre-Carl Langlais, Pavel Chizhov, Mattia Nee, Carlos Rosas Hinostroza, Matthieu Delsart, Ir\`ene Girard, Othman Hicheur, Anastasia Stasenko, Ivan P. Yamshchikov
Abstract: We introduce a new generation of small reasoning models for RAG, search, and source summarization. Pleias-RAG-350m and Pleias-RAG-1B are mid-trained on a large synthetic dataset emulating the retrieval of a wide variety of multilingual open sources from the Common Corpus. They provide native support for citation and grounding with literal quotes and reintegrate multiple features associated with RAG workflows, such as query routing, query reformulation, and source reranking. Pleias-RAG-350m and Pleias-RAG-1B outperform SLMs below 4 billion parameters on standardized RAG benchmarks (HotPotQA, 2wiki) and are competitive with popular larger models, including Qwen-2.5-7B, Llama-3.1-8B, and Gemma-3-4B. They are the only SLMs to date maintaining consistent RAG performance across leading European languages and ensuring systematic reference grounding for statements. Due to their size and ease of deployment on constrained infrastructure and higher factuality by design, the models unlock a range of new use cases for generative AI.
Authors: Ritesh Goru, Shanay Mehta, Prateek Jain
Abstract: Training Large Language Models ( LLMs) to generate explicit reasoning before they produce an answer has been shown to improve their performance across various tasks such as mathematics and coding. However, fine-tuning LLMs on multi-turn reasoning datasets presents a unique challenge: LLMs must generate reasoning tokens that are excluded from subsequent inputs to the LLM. This discrepancy prevents us from processing an entire conversation in a single forward pass-an optimization readily available when we fine-tune on a multi-turn non-reasoning dataset. This paper proposes a novel approach that overcomes this limitation through response token duplication and a custom attention mask that enforces appropriate visibility constraints. Our approach significantly reduces the training time and allows efficient fine-tuning on multi-turn reasoning datasets.
Authors: Guanqun Bi, Zhuang Chen, Zhoufu Liu, Hongkai Wang, Xiyao Xiao, Yuqiang Xie, Wen Zhang, Yongkang Huang, Yuxuan Chen, Libiao Peng, Yi Feng, Minlie Huang
Abstract: Automating structured clinical interviews could revolutionize mental healthcare accessibility, yet existing large language models (LLMs) approaches fail to align with psychiatric diagnostic protocols. We present MAGI, the first framework that transforms the gold-standard Mini International Neuropsychiatric Interview (MINI) into automatic computational workflows through coordinated multi-agent collaboration. MAGI dynamically navigates clinical logic via four specialized agents: 1) an interview tree guided navigation agent adhering to the MINI's branching structure, 2) an adaptive question agent blending diagnostic probing, explaining, and empathy, 3) a judgment agent validating whether the response from participants meet the node, and 4) a diagnosis Agent generating Psychometric Chain-of- Thought (PsyCoT) traces that explicitly map symptoms to clinical criteria. Experimental results on 1,002 real-world participants covering depression, generalized anxiety, social anxiety and suicide shows that MAGI advances LLM- assisted mental health assessment by combining clinical rigor, conversational adaptability, and explainable reasoning.
Authors: Shintaro Ozaki, Kazuki Hayashi, Yusuke Sakai, Jingun Kwon, Hidetaka Kamigaito, Katsuhiko Hayashi, Manabu Okumura, Taro Watanabe
Abstract: Generating images from prompts containing specific entities requires models to retain as much entity-specific knowledge as possible. However, fully memorizing such knowledge is impractical due to the vast number of entities and their continuous emergence. To address this, we propose Text-based Intelligent Generation with Entity prompt Refinement (TextTIGER), which augments knowledge on entities included in the prompts and then summarizes the augmented descriptions using Large Language Models (LLMs) to mitigate performance degradation from longer inputs. To evaluate our method, we introduce WiT-Cub (WiT with Captions and Uncomplicated Background-explanations), a dataset comprising captions, images, and an entity list. Experiments on four image generation models and five LLMs show that TextTIGER improves image generation performance in standard metrics (IS, FID, and CLIPScore) compared to caption-only prompts. Additionally, multiple annotators' evaluation confirms that the summarized descriptions are more informative, validating LLMs' ability to generate concise yet rich descriptions. These findings demonstrate that refining prompts with augmented and summarized entity-related descriptions enhances image generation capabilities. The code and dataset will be available upon acceptance.
Authors: Toghrul Abbasli, Kentaroh Toyoda, Yuan Wang, Leon Witt, Muhammad Asif Ali, Yukai Miao, Dan Li, Qingsong Wei
Abstract: Large Language Models (LLMs) have been transformative across many domains. However, hallucination -- confidently outputting incorrect information -- remains one of the leading challenges for LLMs. This raises the question of how to accurately assess and quantify the uncertainty of LLMs. Extensive literature on traditional models has explored Uncertainty Quantification (UQ) to measure uncertainty and employed calibration techniques to address the misalignment between uncertainty and accuracy. While some of these methods have been adapted for LLMs, the literature lacks an in-depth analysis of their effectiveness and does not offer a comprehensive benchmark to enable insightful comparison among existing solutions. In this work, we fill this gap via a systematic survey of representative prior works on UQ and calibration for LLMs and introduce a rigorous benchmark. Using two widely used reliability datasets, we empirically evaluate six related methods, which justify the significant findings of our review. Finally, we provide outlooks for key future directions and outline open challenges. To the best of our knowledge, this survey is the first dedicated study to review the calibration methods and relevant metrics for LLMs.
Authors: Lei Shen, Xiaoyu Shen
Abstract: In recent years, multi-agent frameworks powered by large language models (LLMs) have advanced rapidly. Despite this progress, there is still a notable absence of benchmark datasets specifically tailored to evaluate their performance. To bridge this gap, we introduce Auto-SLURP, a benchmark dataset aimed at evaluating LLM-based multi-agent frameworks in the context of intelligent personal assistants. Auto-SLURP extends the original SLURP dataset -- initially developed for natural language understanding tasks -- by relabeling the data and integrating simulated servers and external services. This enhancement enables a comprehensive end-to-end evaluation pipeline, covering language understanding, task execution, and response generation. Our experiments demonstrate that Auto-SLURP presents a significant challenge for current state-of-the-art frameworks, highlighting that truly reliable and intelligent multi-agent personal assistants remain a work in progress. The dataset and related code are available at https://github.com/lorashen/Auto-SLURP/.
Authors: Pablo Miralles-Gonz\'alez, Javier Huertas-Tato, Alejandro Mart\'in, David Camacho
Abstract: Natural Language Inference (NLI) is a central task in natural language understanding with applications in fact-checking, question answering, and information retrieval. Despite its importance, current NLI systems heavily rely on supervised learning with datasets that often contain annotation artifacts and biases, limiting generalization and real-world applicability. In this work, we apply a reinforcement learning-based approach using Group Relative Policy Optimization (GRPO) for Chain-of-Thought (CoT) learning in NLI, eliminating the need for labeled rationales and enabling this type of training on more challenging datasets such as ANLI. We fine-tune 7B, 14B, and 32B language models using parameter-efficient techniques (LoRA and QLoRA), demonstrating strong performance across standard and adversarial NLI benchmarks. Our 32B AWQ-quantized model surpasses state-of-the-art results on 7 out of 11 adversarial sets$\unicode{x2013}$or on all of them considering our replication$\unicode{x2013}$within a 22GB memory footprint, showing that robust reasoning can be retained under aggressive quantization. This work provides a scalable and practical framework for building robust NLI systems without sacrificing inference quality.
Authors: Amir Zeldes, Nina Speransky, Nicholas Wagner, Caroline T. Schroeder
Abstract: Despite recent advances in digital resources for other Coptic dialects, especially Sahidic, Bohairic Coptic, the main Coptic dialect for pre-Mamluk, late Byzantine Egypt, and the contemporary language of the Coptic Church, remains critically under-resourced. This paper presents and evaluates the first syntactically annotated corpus of Bohairic Coptic, sampling data from a range of works, including Biblical text, saints' lives and Christian ascetic writing. We also explore some of the main differences we observe compared to the existing UD treebank of Sahidic Coptic, the classical dialect of the language, and conduct joint and cross-dialect parsing experiments, revealing the unique nature of Bohairic as a related, but distinct variety from the more often studied Sahidic.
Authors: Yusen Zhang, Wenliang Zheng, Aashrith Madasu, Peng Shi, Ryo Kamoi, Hao Zhou, Zhuoyang Zou, Shu Zhao, Sarkar Snigdha Sarathi Das, Vipul Gupta, Xiaoxin Lu, Nan Zhang, Ranran Haoran Zhang, Avitej Iyer, Renze Lou, Wenpeng Yin, Rui Zhang
Abstract: High-resolution image (HRI) understanding aims to process images with a large number of pixels, such as pathological images and agricultural aerial images, both of which can exceed 1 million pixels. Vision Large Language Models (VLMs) can allegedly handle HRIs, however, there is a lack of a comprehensive benchmark for VLMs to evaluate HRI understanding. To address this gap, we introduce HRScene, a novel unified benchmark for HRI understanding with rich scenes. HRScene incorporates 25 real-world datasets and 2 synthetic diagnostic datasets with resolutions ranging from 1,024 $\times$ 1,024 to 35,503 $\times$ 26,627. HRScene is collected and re-annotated by 10 graduate-level annotators, covering 25 scenarios, ranging from microscopic to radiology images, street views, long-range pictures, and telescope images. It includes HRIs of real-world objects, scanned documents, and composite multi-image. The two diagnostic evaluation datasets are synthesized by combining the target image with the gold answer and distracting images in different orders, assessing how well models utilize regions in HRI. We conduct extensive experiments involving 28 VLMs, including Gemini 2.0 Flash and GPT-4o. Experiments on HRScene show that current VLMs achieve an average accuracy of around 50% on real-world tasks, revealing significant gaps in HRI understanding. Results on synthetic datasets reveal that VLMs struggle to effectively utilize HRI regions, showing significant Regional Divergence and lost-in-middle, shedding light on future research.
Authors: Jared Moore, Declan Grabb, William Agnew, Kevin Klyman, Stevie Chancellor, Desmond C. Ong, Nick Haber
Abstract: Should a large language model (LLM) be used as a therapist? In this paper, we investigate the use of LLMs to *replace* mental health providers, a use case promoted in the tech startup and research space. We conduct a mapping review of therapy guides used by major medical institutions to identify crucial aspects of therapeutic relationships, such as the importance of a therapeutic alliance between therapist and client. We then assess the ability of LLMs to reproduce and adhere to these aspects of therapeutic relationships by conducting several experiments investigating the responses of current LLMs, such as `gpt-4o`. Contrary to best practices in the medical community, LLMs 1) express stigma toward those with mental health conditions and 2) respond inappropriately to certain common (and critical) conditions in naturalistic therapy settings -- e.g., LLMs encourage clients' delusional thinking, likely due to their sycophancy. This occurs even with larger and newer LLMs, indicating that current safety practices may not address these gaps. Furthermore, we note foundational and practical barriers to the adoption of LLMs as therapists, such as that a therapeutic alliance requires human characteristics (e.g., identity and stakes). For these reasons, we conclude that LLMs should not replace therapists, and we discuss alternative roles for LLMs in clinical therapy.
Authors: Hongyu Wang, Shuming Ma, Furu Wei
Abstract: Efficient deployment of 1-bit Large Language Models (LLMs) is hindered by activation outliers, which complicate quantization to low bit-widths. We introduce BitNet v2, a novel framework enabling native 4-bit activation quantization for 1-bit LLMs. To tackle outliers in attention and feed-forward network activations, we propose H-BitLinear, a module applying an online Hadamard transformation prior to activation quantization. This transformation smooths sharp activation distributions into more Gaussian-like forms, suitable for low-bit representation. Experiments show BitNet v2 trained from scratch with 8-bit activations matches BitNet b1.58 performance. Crucially, BitNet v2 achieves minimal performance degradation when trained with native 4-bit activations, significantly reducing memory footprint and computational cost for batched inference.
Authors: Yiming Wang, Pei Zhang, Jialong Tang, Haoran Wei, Baosong Yang, Rui Wang, Chenshu Sun, Feitong Sun, Jiran Zhang, Junxuan Wu, Qiqian Cang, Yichang Zhang, Fei Huang, Junyang Lin, Fei Huang, Jingren Zhou
Abstract: In this paper, we introduce PolyMath, a multilingual mathematical reasoning benchmark covering 18 languages and 4 easy-to-hard difficulty levels. Our benchmark ensures difficulty comprehensiveness, language diversity, and high-quality translation, making it a highly discriminative multilingual mathematical benchmark in the era of reasoning LLMs. We conduct a comprehensive evaluation for advanced LLMs and find that even Deepseek-R1-671B and Qwen-QwQ-32B, achieve only 43.4 and 41.8 benchmark scores, with less than 30% accuracy under the highest level. From a language perspective, our benchmark reveals several key challenges of LLMs in multilingual reasoning: (1) Reasoning performance varies widely across languages for current LLMs; (2) Input-output language consistency is low in reasoning LLMs and may be correlated with performance; (3) The thinking length differs significantly by language for current LLMs. Additionally, we demonstrate that controlling the output language in the instructions has the potential to affect reasoning performance, especially for some low-resource languages, suggesting a promising direction for improving multilingual capabilities in LLMs.
Authors: Wenyi Xiao, Leilei Gan, Weilong Dai, Wanggui He, Ziwei Huang, Haoyuan Li, Fangxun Shu, Zhelun Yu, Peng Zhang, Hao Jiang, Fei Wu
Abstract: Recent advances in large vision-language models (LVLMs) have revealed an \textit{overthinking} phenomenon, where models generate verbose reasoning across all tasks regardless of questions. To address this issue, we present \textbf{FAST}, a novel \textbf{Fa}st-\textbf{S}low \textbf{T}hinking framework that dynamically adapts reasoning depth based on question characteristics. Through empirical analysis, we establish the feasibility of fast-slow thinking in LVLMs by investigating how response length and data distribution affect performance. We develop FAST-GRPO with three components: model-based metrics for question characterization, an adaptive thinking reward mechanism, and difficulty-aware KL regularization. Experiments across seven reasoning benchmarks demonstrate that FAST achieves state-of-the-art accuracy with over 10\% relative improvement compared to the base model, while reducing token usage by 32.7-67.3\% compared to previous slow-thinking approaches, effectively balancing reasoning length and accuracy.
Authors: James D. Finch, Yasasvi Josyula, Jinho D. Choi
Abstract: In task-oriented dialogue (TOD) systems, Slot Schema Induction (SSI) is essential for automatically identifying key information slots from dialogue data without manual intervention. This paper presents a novel state-of-the-art (SoTA) approach that formulates SSI as a text generation task, where a language model incrementally constructs and refines a slot schema over a stream of dialogue data. To develop this approach, we present a fully automatic LLM-based TOD simulation method that creates data with high-quality state labels for novel task domains. Furthermore, we identify issues in SSI evaluation due to data leakage and poor metric alignment with human judgment. We resolve these by creating new evaluation data using our simulation method with human guidance and correction, as well as designing improved evaluation metrics. These contributions establish a foundation for future SSI research and advance the SoTA in dialogue understanding and system development.
Authors: Leandra Fichtel, Maximilian Splieth\"over, Eyke H\"ullermeier, Patricia Jimenez, Nils Klowait, Stefan Kopp, Axel-Cyrille Ngonga Ngomo, Amelie Robrecht, Ingrid Scharlau, Lutz Terfloth, Anna-Lisa Vollmer, Henning Wachsmuth
Abstract: The ability to generate explanations that are understood by explainees is the quintessence of explainable artificial intelligence. Since understanding depends on the explainee's background and needs, recent research has focused on co-constructive explanation dialogues, where the explainer continuously monitors the explainee's understanding and adapts explanations dynamically. We investigate the ability of large language models (LLMs) to engage as explainers in co-constructive explanation dialogues. In particular, we present a user study in which explainees interact with LLMs, of which some have been instructed to explain a predefined topic co-constructively. We evaluate the explainees' understanding before and after the dialogue, as well as their perception of the LLMs' co-constructive behavior. Our results indicate that current LLMs show some co-constructive behaviors, such as asking verification questions, that foster the explainees' engagement and can improve understanding of a topic. However, their ability to effectively monitor the current understanding and scaffold the explanations accordingly remains limited.
Authors: Gwen Yidou Weng, Benjie Wang, Guy Van den Broeck
Abstract: As large language models (LMs) advance, there is an increasing need to control their outputs to align with human values (e.g., detoxification) or desired attributes (e.g., personalization, topic). However, autoregressive models focus on next-token predictions and struggle with global properties that require looking ahead. Existing solutions either tune or post-train LMs for each new attribute - expensive and inflexible - or approximate the Expected Attribute Probability (EAP) of future sequences by sampling or training, which is slow and unreliable for rare attributes. We introduce TRACE (Tractable Probabilistic Reasoning for Adaptable Controllable gEneration), a novel framework that efficiently computes EAP and adapts to new attributes through tractable probabilistic reasoning and lightweight control. TRACE distills a Hidden Markov Model (HMM) from an LM and pairs it with a small classifier to estimate attribute probabilities, enabling exact EAP computation over the HMM's predicted futures. This EAP is then used to reweigh the LM's next-token probabilities for globally compliant continuations. Empirically, TRACE achieves state-of-the-art results in detoxification with only 10% decoding overhead, adapts to 76 low-resource personalized LLMs within seconds, and seamlessly extends to composite attributes.
Authors: Xinyu Chen, Yunxin Li, Haoyuan Shi, Baotian Hu, Wenhan Luo, Yaowei Wang, Min Zhang
Abstract: Assessing the video comprehension capabilities of multimodal AI systems can effectively measure their understanding and reasoning abilities. Most video evaluation benchmarks are limited to a single language, typically English, and predominantly feature videos rooted in Western cultural contexts. In this paper, we present VideoVista-CulturalLingo, the first video evaluation benchmark designed to bridge cultural, linguistic, and domain divide in video comprehension. Our work differs from existing benchmarks in the following ways: 1) Cultural diversity, incorporating cultures from China, North America, and Europe; 2) Multi-linguistics, with questions presented in Chinese and English-two of the most widely spoken languages; and 3) Broad domain, featuring videos sourced from hundreds of human-created domains. VideoVista-CulturalLingo contains 1,389 videos and 3,134 QA pairs, and we have evaluated 24 recent open-source or proprietary video large models. From the experiment results, we observe that: 1) Existing models perform worse on Chinese-centric questions than Western-centric ones, particularly those related to Chinese history; 2) Current open-source models still exhibit limitations in temporal understanding, especially in the Event Localization task, achieving a maximum score of only 45.2%; 3) Mainstream models demonstrate strong performance in general scientific questions, while open-source models demonstrate weak performance in mathematics.
Authors: Yongkang Li, Panagiotis Eustratiadis, Simon Lupart, Evangelos Kanoulas
Abstract: This paper concerns corpus poisoning attacks in dense information retrieval, where an adversary attempts to compromise the ranking performance of a search algorithm by injecting a small number of maliciously generated documents into the corpus. Our work addresses two limitations in the current literature. First, attacks that perform adversarial gradient-based word substitution search do so in the discrete lexical space, while retrieval itself happens in the continuous embedding space. We thus propose an optimization method that operates in the embedding space directly. Specifically, we train a perturbation model with the objective of maintaining the geometric distance between the original and adversarial document embeddings, while also maximizing the token-level dissimilarity between the original and adversarial documents. Second, it is common for related work to have a strong assumption that the adversary has prior knowledge about the queries. In this paper, we focus on a more challenging variant of the problem where the adversary assumes no prior knowledge about the query distribution (hence, unsupervised). Our core contribution is an adversarial corpus attack that is fast and effective. We present comprehensive experimental results on both in- and out-of-domain datasets, focusing on two related tasks: a top-1 attack and a corpus poisoning attack. We consider attacks under both a white-box and a black-box setting. Notably, our method can generate successful adversarial examples in under two minutes per target document; four times faster compared to the fastest gradient-based word substitution methods in the literature with the same hardware. Furthermore, our adversarial generation method generates text that is more likely to occur under the distribution of natural text (low perplexity), and is therefore more difficult to detect.
Authors: Yasmine Omri, Parth Shroff, Thierry Tambe
Abstract: The exponential growth of Large Multimodal Models (LMMs) has driven advancements in cross-modal reasoning but at significant computational costs. In this work, we focus on visual language models. We highlight the redundancy and inefficiency in current vision encoders, and seek to construct an adaptive compression method for multimodal data. In this work, we characterize a panoply of visual token selection and merging approaches through both benchmarking and qualitative analysis. In particular, we demonstrate that simple cluster-level token aggregation outperforms prior state-of-the-art works in token selection and merging, including merging at the vision encoder level and attention-based approaches. We underline the redundancy in current vision encoders, and shed light on several puzzling trends regarding principles of visual token selection through cross-modal attention visualizations. This work is a first effort towards more effective encoding and processing of high-dimensional data, and paves the way for more scalable and sustainable multimodal systems.
Authors: Girish A. Koushik, Diptesh Kanojia, Helen Treharne, Aditya Joshi
Abstract: Social media memes are a challenging domain for hate detection because they intertwine visual and textual cues into culturally nuanced messages. We introduce a novel framework, CAMU, which leverages large vision-language models to generate more descriptive captions, a caption-scoring neural network to emphasise hate-relevant content, and parameter-efficient fine-tuning of CLIP's text encoder for an improved multimodal understanding of memes. Experiments on publicly available hateful meme datasets show that simple projection layer fine-tuning yields modest gains, whereas selectively tuning deeper text encoder layers significantly boosts performance on all evaluation metrics. Moreover, our approach attains high accuracy (0.807) and F1-score (0.806) on the Hateful Memes dataset, at par with the existing SoTA framework while being much more efficient, offering practical advantages in real-world scenarios that rely on fixed decision thresholds. CAMU also achieves the best F1-score of 0.673 on the MultiOFF dataset for offensive meme identification, demonstrating its generalisability. Additional analyses on benign confounders reveal that robust visual grounding and nuanced text representations are crucial for reliable hate and offence detection. We will publicly release CAMU along with the resultant models for further research. Disclaimer: This paper includes references to potentially disturbing, hateful, or offensive content due to the nature of the task.
Authors: Chaoran Chen, Zhiping Zhang, Ibrahim Khalilov, Bingcan Guo, Simret A Gebreegziabher, Yanfang Ye, Ziang Xiao, Yaxing Yao, Tianshi Li, Toby Jia-Jun Li
Abstract: The rise of Large Language Models (LLMs) has revolutionized Graphical User Interface (GUI) automation through LLM-powered GUI agents, yet their ability to process sensitive data with limited human oversight raises significant privacy and security risks. This position paper identifies three key risks of GUI agents and examines how they differ from traditional GUI automation and general autonomous agents. Despite these risks, existing evaluations focus primarily on performance, leaving privacy and security assessments largely unexplored. We review current evaluation metrics for both GUI and general LLM agents and outline five key challenges in integrating human evaluators for GUI agent assessments. To address these gaps, we advocate for a human-centered evaluation framework that incorporates risk assessments, enhances user awareness through in-context consent, and embeds privacy and security considerations into GUI agent design and evaluation.
Authors: Isadora White, Kolby Nottingham, Ayush Maniar, Max Robinson, Hansen Lillemark, Mehul Maheshwari, Lianhui Qin, Prithviraj Ammanabrolu
Abstract: Collaboration is ubiquitous and essential in day-to-day life -- from exchanging ideas, to delegating tasks, to generating plans together. This work studies how LLMs can adaptively collaborate to perform complex embodied reasoning tasks. To this end we introduce MINDcraft, an easily extensible platform built to enable LLM agents to control characters in the open-world game of Minecraft; and MineCollab, a benchmark to test the different dimensions of embodied and collaborative reasoning. An experimental study finds that the primary bottleneck in collaborating effectively for current state-of-the-art agents is efficient natural language communication, with agent performance dropping as much as 15% when they are required to communicate detailed task completion plans. We conclude that existing LLM agents are ill-optimized for multi-agent collaboration, especially in embodied scenarios, and highlight the need to employ methods beyond in-context and imitation learning. Our website can be found here: https://mindcraft-minecollab.github.io/
Authors: Yiwei Zha
Abstract: Financial sectors are rapidly adopting language model technologies, yet evaluating specialized RAG systems in this domain remains challenging. This paper introduces SMARTFinRAG, addressing three critical gaps in financial RAG assessment: (1) a fully modular architecture where components can be dynamically interchanged during runtime; (2) a document-centric evaluation paradigm generating domain-specific QA pairs from newly ingested financial documents; and (3) an intuitive interface bridging research-implementation divides. Our evaluation quantifies both retrieval efficacy and response quality, revealing significant performance variations across configurations. The platform's open-source architecture supports transparent, reproducible research while addressing practical deployment challenges faced by financial institutions implementing RAG systems.
Authors: Leena G Pillai, D. Muhammad Noorul Mubarak, Elizabeth Sherly
Abstract: Speech production is a complex sequential process which involve the coordination of various articulatory features. Among them tongue being a highly versatile active articulator responsible for shaping airflow to produce targeted speech sounds that are intellectual, clear, and distinct. This paper presents a novel approach for predicting tongue and lip articulatory features involved in a given speech acoustics using a stacked Bidirectional Long Short-Term Memory (BiLSTM) architecture, combined with a one-dimensional Convolutional Neural Network (CNN) for post-processing with fixed weights initialization. The proposed network is trained with two datasets consisting of simultaneously recorded speech and Electromagnetic Articulography (EMA) datasets, each introducing variations in terms of geographical origin, linguistic characteristics, phonetic diversity, and recording equipment. The performance of the model is assessed in Speaker Dependent (SD), Speaker Independent (SI), corpus dependent (CD) and cross corpus (CC) modes. Experimental results indicate that the proposed model with fixed weights approach outperformed the adaptive weights initialization with in relatively minimal number of training epochs. These findings contribute to the development of robust and efficient models for articulatory feature prediction, paving the way for advancements in speech production research and applications.
Authors: Narek Maloyan, Dmitry Namiot
Abstract: LLM as judge systems used to assess text quality code correctness and argument strength are vulnerable to prompt injection attacks. We introduce a framework that separates content author attacks from system prompt attacks and evaluate five models Gemma 3.27B Gemma 3.4B Llama 3.2 3B GPT 4 and Claude 3 Opus on four tasks with various defenses using fifty prompts per condition. Attacks achieved up to seventy three point eight percent success smaller models proved more vulnerable and transferability ranged from fifty point five to sixty two point six percent. Our results contrast with Universal Prompt Injection and AdvPrompter We recommend multi model committees and comparative scoring and release all code and datasets
Authors: KimiTeam, Ding Ding, Zeqian Ju, Yichong Leng, Songxiang Liu, Tong Liu, Zeyu Shang, Kai Shen, Wei Song, Xu Tan, Heyi Tang, Zhengtao Wang, Chu Wei, Yifei Xin, Xinran Xu, Jianwei Yu, Yutao Zhang, Xinyu Zhou, Y. Charles, Jun Chen, Yanru Chen, Yulun Du, Weiran He, Zhenxing Hu, Guokun Lai, Qingcheng Li, Yangyang Liu, Weidong Sun, Jianzhou Wang, Yuzhi Wang, Yuefeng Wu, Yuxin Wu, Dongchao Yang, Hao Yang, Ying Yang, Zhilin Yang, Aoxiong Yin, Ruibin Yuan, Yutong Zhang, Zaida Zhou
Abstract: We present Kimi-Audio, an open-source audio foundation model that excels in audio understanding, generation, and conversation. We detail the practices in building Kimi-Audio, including model architecture, data curation, training recipe, inference deployment, and evaluation. Specifically, we leverage a 12.5Hz audio tokenizer, design a novel LLM-based architecture with continuous features as input and discrete tokens as output, and develop a chunk-wise streaming detokenizer based on flow matching. We curate a pre-training dataset that consists of more than 13 million hours of audio data covering a wide range of modalities including speech, sound, and music, and build a pipeline to construct high-quality and diverse post-training data. Initialized from a pre-trained LLM, Kimi-Audio is continual pre-trained on both audio and text data with several carefully designed tasks, and then fine-tuned to support a diverse of audio-related tasks. Extensive evaluation shows that Kimi-Audio achieves state-of-the-art performance on a range of audio benchmarks including speech recognition, audio understanding, audio question answering, and speech conversation. We release the codes, model checkpoints, as well as the evaluation toolkits in https://github.com/MoonshotAI/Kimi-Audio.
Authors: Zhangdie Yuan, Eric Chamoun, Rami Aly, Chenxi Whitehouse, Andreas Vlachos
Abstract: This paper introduces PRobELM (Plausibility Ranking Evaluation for Language Models), a benchmark designed to assess language models' ability to discern more plausible from less plausible scenarios through their parametric knowledge. While benchmarks such as TruthfulQA emphasise factual accuracy or truthfulness, and others such as COPA explore plausible scenarios without explicitly incorporating world knowledge, PRobELM seeks to bridge this gap by evaluating models' capabilities to prioritise plausible scenarios that leverage world knowledge over less plausible alternatives. This design allows us to assess the potential of language models for downstream use cases such as literature-based discovery where the focus is on identifying information that is likely but not yet known. Our benchmark is constructed from a dataset curated from Wikidata edit histories, tailored to align the temporal bounds of the training data for the evaluated models. PRobELM facilitates the evaluation of language models across multiple prompting types, including statement, text completion, and question-answering. Experiments with 10 models of various sizes and architectures on the relationship between model scales, training recency, and plausibility performance, reveal that factual accuracy does not directly correlate with plausibility performance and that up-to-date training data enhances plausibility assessment across different model architectures.
Authors: Minghan Li, Xilun Chen, Ari Holtzman, Beidi Chen, Jimmy Lin, Wen-tau Yih, Xi Victoria Lin
Abstract: Large language models (LLMs) often hallucinate and lack the ability to provide attribution for their generations. Semi-parametric LMs, such as kNN-LM, approach these limitations by refining the output of an LM for a given prompt using its nearest neighbor matches in a non-parametric data store. However, these models often exhibit slow inference speeds and produce non-fluent texts. In this paper, we introduce Nearest Neighbor Speculative Decoding (NEST), a novel semi-parametric language modeling approach that is capable of incorporating real-world text spans of arbitrary length into the LM generations and providing attribution to their sources. NEST performs token-level retrieval at each inference step to compute a semi-parametric mixture distribution and identify promising span continuations in a corpus. It then uses an approximate speculative decoding procedure that accepts a prefix of the retrieved span or generates a new token. NEST significantly enhances the generation quality and attribution rate of the base LM across a variety of knowledge-intensive tasks, surpassing the conventional kNN-LM method and performing competitively with in-context retrieval augmentation. In addition, NEST substantially improves the generation speed, achieving a 1.8x speedup in inference time when applied to Llama-2-Chat 70B. Code will be released at https://github.com/facebookresearch/NEST/tree/main.
Authors: Peitao Han, Lis Kanashiro Pereira, Fei Cheng, Wan Jou She, Eiji Aramaki
Abstract: Existing in-context learning (ICL) methods for relation extraction (RE) often prioritize language similarity over structural similarity, which can lead to overlooking entity relationships. To address this, we propose an AMR-enhanced retrieval-based ICL method for RE. Our model retrieves in-context examples based on semantic structure similarity between task inputs and training samples. Evaluations on four standard English RE datasets show that our model outperforms baselines in the unsupervised setting across all datasets. In the supervised setting, it achieves state-of-the-art results on three datasets and competitive results on the fourth.
Authors: Daniil Gurgurov, Tanja B\"aumel, Tatiana Anikina
Abstract: Multilingual Large Language Models (LLMs) have gained large popularity among Natural Language Processing (NLP) researchers and practitioners. These models, trained on huge datasets, show proficiency across various languages and demonstrate effectiveness in numerous downstream tasks. This paper navigates the landscape of multilingual LLMs, providing an introductory overview of their technical aspects. It explains underlying architectures, objective functions, pre-training data sources, and tokenization methods. This work explores the unique features of different model types: encoder-only (mBERT, XLM-R), decoder-only (XGLM, PALM, BLOOM, GPT-3), and encoder-decoder models (mT5, mBART). Additionally, it addresses one of the significant limitations of multilingual LLMs - the curse of multilinguality - and discusses current attempts to overcome it.
Authors: Jieyong Kim, Hyunseo Kim, Hyunjin Cho, SeongKu Kang, Buru Chang, Jinyoung Yeo, Dongha Lee
Abstract: Recent advancements in Large Language Models (LLMs) have demonstrated exceptional performance across a wide range of tasks, generating significant interest in their application to recommendation systems. However, existing methods have not fully capitalized on the potential of LLMs, often constrained by limited input information or failing to fully utilize their advanced reasoning capabilities. To address these limitations, we introduce EXP3RT, a novel LLM-based recommender designed to leverage rich preference information contained in user and item reviews. EXP3RT is basically fine-tuned through distillation from a teacher LLM to perform three key tasks in order: EXP3RT first extracts and encapsulates essential subjective preferences from raw reviews, aggregates and summarizes them according to specific criteria to create user and item profiles. It then generates detailed step-by-step reasoning followed by predicted rating, i.e., reasoning-enhanced rating prediction, by considering both subjective and objective information from user/item profiles and item descriptions. This personalized preference reasoning from EXP3RT enhances rating prediction accuracy and also provides faithful and reasonable explanations for recommendation. Extensive experiments show that EXP3RT outperforms existing methods on both rating prediction and candidate item reranking for top-k recommendation, while significantly enhancing the explainability of recommendation systems.
Authors: Leo Yeykelis, Kaavya Pichai, James J. Cummings, Byron Reeves
Abstract: This report analyzes the potential for large language models (LLMs) to expedite accurate replication and generalization of published research about message effects in marketing. LLM-powered participants (personas) were tested by replicating 133 experimental findings from 14 papers containing 45 recent studies published in the Journal of Marketing. For each study, the measures, stimuli, and sampling specifications were used to generate prompts for LLMs to act as unique personas. The AI personas, 19,447 in total across all of the studies, generated complete datasets and statistical analyses were then compared with the original human study results. The LLM replications successfully reproduced 76% of the original main effects (84 out of 111), demonstrating strong potential for AI-assisted replication. The overall replication rate including interaction effects was 68% (90 out of 133). Furthermore, a test of how human results generalized to different participant samples, media stimuli, and measures showed that replication results can change when tests go beyond the parameters of the original human studies. Implications are discussed for the replication and generalizability crises in social science, the acceleration of theory building in media and marketing psychology, and the practical advantages of rapid message testing for consumer products. Limitations of AI replications are addressed with respect to complex interaction effects, biases in AI models, and establishing benchmarks for AI metrics in marketing research.
Authors: Leitian Tao, Yixuan Li
Abstract: The burgeoning capabilities of large language models (LLMs) have underscored the need for alignment to ensure these models act in accordance with human values and intentions. Existing alignment frameworks present constraints either in the form of expensive human effort or high computational costs. This paper explores a promising middle ground, where we employ a weak LLM that is significantly less resource-intensive than top-tier models, yet offers more automation than purely human feedback. We present a systematic study to evaluate and understand weak LLM's ability to generate feedback for alignment. Our empirical findings demonstrate that weak LLMs can provide feedback that rivals or even exceeds that of fully human-annotated data. Our study indicates a minimized impact of model size on feedback efficacy, shedding light on a scalable and sustainable alignment strategy. To deepen our understanding of alignment under weak LLM feedback, we conduct a series of qualitative and quantitative analyses, offering novel insights into the quality discrepancies between human feedback vs. weak LLM feedback.
Authors: Ningyuan Xi, Xiaoyu Wang, Yetao Wu, Teng Chen, Qingqing Gu, Yue Zhao, Jinxian Qu, Zhonglin Jiang, Yong Chen, Luo Ji
Abstract: Large Language Model can reasonably understand and generate human expressions but may lack of thorough thinking and reasoning mechanisms. Recently there have been several studies which enhance the thinking ability of language models but most of them are not data-driven or training-based. In this paper, we are motivated by the cognitive mechanism in the natural world, and design a novel model architecture called TaS which allows it to first consider the thoughts and then express the response based upon the query. We design several pipelines to annotate or generate the thought contents from prompt-response samples, then add language heads in a middle layer which behaves as the thinking layer. We train the language model by the thoughts-augmented data and successfully let the thinking layer automatically generate reasonable thoughts and finally output more reasonable responses. Both qualitative examples and quantitative results validate the effectiveness and performance of TaS. Our code is available at https://anonymous.4open.science/r/TadE.
Authors: Yifei Ming, Senthil Purushwalkam, Shrey Pandit, Zixuan Ke, Xuan-Phi Nguyen, Caiming Xiong, Shafiq Joty
Abstract: Ensuring faithfulness to context in large language models (LLMs) and retrieval-augmented generation (RAG) systems is crucial for reliable deployment in real-world applications, as incorrect or unsupported information can erode user trust. Despite advancements on standard benchmarks, faithfulness hallucination-where models generate responses misaligned with the provided context-remains a significant challenge. In this work, we introduce FaithEval, a novel and comprehensive benchmark tailored to evaluate the faithfulness of LLMs in contextual scenarios across three diverse tasks: unanswerable, inconsistent, and counterfactual contexts. These tasks simulate real-world challenges where retrieval mechanisms may surface incomplete, contradictory, or fabricated information. FaithEval comprises 4.9K high-quality problems in total, validated through a rigorous four-stage context construction and validation framework, employing both LLM-based auto-evaluation and human validation. Our extensive study across a wide range of open-source and proprietary models reveals that even state-of-the-art models often struggle to remain faithful to the given context, and that larger models do not necessarily exhibit improved faithfulness.Project is available at: https://github.com/SalesforceAIResearch/FaithEval.
Authors: Shuai Niu, Jing Ma, Hongzhan Lin, Liang Bai, Zhihua Wang, Yida Xu, Yunya Song, Xian Yang
Abstract: Interpretation is critical for disease diagnosis, but existing models struggle to balance predictive accuracy with human-understandable rationales. While large language models (LLMs) offer strong reasoning abilities, their clinical use is limited by high computational costs and restricted multimodal reasoning ability. Small language models (SLMs) are efficient but lack advanced reasoning for integrating multimodal medical data. In addition, both LLMs and SLMs lack of domain knowledge for trustworthy reasoning. Therefore, we propose ClinRaGen, enhancing SLMs by leveraging LLM-derived reasoning ability via rationale distillation and domain knowledge injection for trustworthy multimodal rationale generation. Key innovations include a sequential rationale distillation framework that equips SLMs with LLM-comparable mutlimodal reasoning abilities, and a knowledge-augmented attention mechanism that jointly unifies multimodal representation from time series and textual data in a same encoding space, enabling it naturally interpreted by SLMs while incorporating domain knowledge for reliable rationale generation. Experiments on real-world medical datasets show that ClinRaGen achieves state-of-the-art performance in disease diagnosis and rationale generation, demonstrating the effectiveness of combining LLM-driven reasoning with knowledge augmentation for improved interpretability.
Authors: Atsuki Yamaguchi, Terufumi Morishita, Aline Villavicencio, Nikolaos Aletras
Abstract: Vocabulary expansion (VE) is the de-facto approach to language adaptation of large language models (LLMs) by adding new tokens and continuing pre-training on target data. While this is effective for base models trained on unlabeled data, it poses challenges for chat models trained to follow instructions through labeled conversation data. Directly adapting the latter with VE on target unlabeled data may result in forgetting chat abilities. While ideal, target chat data is often unavailable or costly to create for low-resource languages, and machine-translated alternatives are not always effective. To address this issue, previous work proposed using a base and chat model from the same family. This method first adapts the base LLM with VE on target unlabeled data and then converts it to a chat model by adding a chat vector (CV) derived from the weight difference between the source base and chat models. We propose ElChat, a new language adaptation method for chat LLMs that adapts a chat model directly on target unlabeled data, without a base model. It elicits chat abilities by injecting information from the source chat model. ElChat offers more robust and competitive target language and safety performance while achieving superior English, chat, and instruction-following abilities compared to CV.
Authors: Hichem Ammar Khodja, Fr\'ed\'eric B\'echet, Quentin Brabant, Alexis Nasr, Gw\'enol\'e Lecorv\'e
Abstract: This paper explores the robustness of language models (LMs) to variations in the temporal context within factual knowledge. It examines whether LMs can correctly associate a temporal context with a past fact valid over a defined period, by asking them to differentiate correct from incorrect contexts. The accuracy of LMs is analyzed along two dimensions: the distance of the incorrect context from the validity period and the granularity of the context. To this end, a dataset called TimeStress is introduced, enabling the evaluation of 18 diverse LMs. Results reveal that the best LM achieves perfect accuracy for only 6% of the studied facts, with critical errors that humans would not make. This work highlights the limitations of current LMs in temporal representation. We provide all data and code for further research.
Authors: Xiaoqian Liu, Ke Wang, Yongbin Li, Yuchuan Wu, Wentao Ma, Aobo Kong, Fei Huang, Jianbin Jiao, Junge Zhang
Abstract: Large Language Models (LLMs) have shown impressive reasoning capabilities in well-defined problems with clear solutions, such as mathematics and coding. However, they still struggle with complex real-world scenarios like business negotiations, which require strategic reasoning-an ability to navigate dynamic environments and align long-term goals amidst uncertainty. Existing methods for strategic reasoning face challenges in adaptability, scalability, and transferring strategies to new contexts. To address these issues, we propose explicit policy optimization (EPO) for strategic reasoning, featuring an LLM that provides strategies in open-ended action space and can be plugged into arbitrary LLM agents to motivate goal-directed behavior. To improve adaptability and policy transferability, we train the strategic reasoning model via multi-turn reinforcement learning (RL) using process rewards and iterative self-play, without supervised fine-tuning (SFT) as a preliminary step. Experiments across social and physical domains demonstrate EPO's ability of long-term goal alignment through enhanced strategic reasoning, achieving state-of-the-art performance on social dialogue and web navigation tasks. Our findings reveal various collaborative reasoning mechanisms emergent in EPO and its effectiveness in generating novel strategies, underscoring its potential for strategic reasoning in real-world applications. Code and data are available at https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/EPO.
URLs: https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/EPO.
Authors: George Drayson, Emine Yilmaz, Vasileios Lampos
Abstract: As Large Language Models (LLMs) become increasingly prevalent, their generated outputs are proliferating across the web, risking a future where machine-generated content dilutes human-authored text. Since online data is the primary resource for LLM pre-training, subsequent models could be trained on an unknown portion of synthetic samples. This will lead to model collapse, a degenerative process whereby LLMs reinforce their own errors, and ultimately yield a declining performance. In this study, we investigate the impact of decoding strategy on model collapse, analysing the characteristics of text at each model generation, the similarity to human references, and the resulting model performance. Using the decoding strategies that lead to the most significant degradation, we evaluate model collapse in more realistic scenarios where the origin of the data (human or synthetic) is unknown. We train a machine-generated text detector and propose an importance sampling approach to alleviate model collapse. Our method is validated on two LLM variants (GPT-2 and SmolLM2) on the open-ended text generation task. We demonstrate that it can not only prevent model collapse but also improve performance when sufficient human-authored samples are present. We release our code at https://github.com/GeorgeDrayson/model_collapse.
Authors: Jiuding Sun, Jing Huang, Sidharth Baskaran, Karel D'Oosterlinck, Christopher Potts, Michael Sklar, Atticus Geiger
Abstract: Mechanistic interpretability has made great strides in identifying neural network features (e.g., directions in hidden activation space) that mediate concepts(e.g., the birth year of a person) and enable predictable manipulation. Distributed alignment search (DAS) leverages supervision from counterfactual data to learn concept features within hidden states, but DAS assumes we can afford to conduct a brute force search over potential feature locations. To address this, we present HyperDAS, a transformer-based hypernetwork architecture that (1) automatically locates the token-positions of the residual stream that a concept is realized in and (2) constructs features of those residual stream vectors for the concept. In experiments with Llama3-8B, HyperDAS achieves state-of-the-art performance on the RAVEL benchmark for disentangling concepts in hidden states. In addition, we review the design decisions we made to mitigate the concern that HyperDAS (like all powerful interpretabilty methods) might inject new information into the target model rather than faithfully interpreting it.
Authors: Minhu Park, Hongseok Oh, Eunkyung Choi, Wonseok Hwang
Abstract: Recently, building retrieval-augmented generation (RAG) systems to enhance the capability of large language models (LLMs) has become a common practice. Especially in the legal domain, previous judicial decisions play a significant role under the doctrine of stare decisis which emphasizes the importance of making decisions based on (retrieved) prior documents. However, the overall performance of RAG system depends on many components: (1) retrieval corpora, (2) retrieval algorithms, (3) rerankers, (4) LLM backbones, and (5) evaluation metrics. Here we propose LRAGE, an open-source tool for holistic evaluation of RAG systems focusing on the legal domain. LRAGE provides GUI and CLI interfaces to facilitate seamless experiments and investigate how changes in the aforementioned five components affect the overall accuracy. We validated LRAGE using multilingual legal benches including Korean (KBL), English (LegalBench), and Chinese (LawBench) by demonstrating how the overall accuracy changes when varying the five components mentioned above. The source code is available at https://github.com/hoorangyee/LRAGE.
Authors: Haowei Lin, Xiangyu Wang, Ruilin Yan, Baizhou Huang, Haotian Ye, Jianhua Zhu, Zihao Wang, James Zou, Jianzhu Ma, Yitao Liang
Abstract: With powerful large language models (LLMs) demonstrating superhuman reasoning capabilities, a critical question arises: Do LLMs genuinely reason, or do they merely recall answers from their extensive, web-scraped training datasets? Publicly released benchmarks inevitably become contaminated once incorporated into subsequent LLM training sets, undermining their reliability as faithful assessments. To address this, we introduce KUMO, a generative evaluation framework designed specifically for assessing reasoning in LLMs. KUMO synergistically combines LLMs with symbolic engines to dynamically produce diverse, multi-turn reasoning tasks that are partially observable and adjustable in difficulty. Through an automated pipeline, KUMO continuously generates novel tasks across open-ended domains, compelling models to demonstrate genuine generalization rather than memorization. We evaluated 23 state-of-the-art LLMs on 5,000 tasks across 100 domains created by KUMO, benchmarking their reasoning abilities against university students. Our findings reveal that many LLMs have outperformed university-level performance on easy reasoning tasks, and reasoning-scaled LLMs reach university-level performance on complex reasoning challenges. Moreover, LLM performance on KUMO tasks correlates strongly with results on newly released real-world reasoning benchmarks, underscoring KUMO's value as a robust, enduring assessment tool for genuine LLM reasoning capabilities.
Authors: Akram Mustafa, Usman Naseem, Mostafa Rahimi Azghadi
Abstract: Clinical document classification is essential for converting unstructured medical texts into standardised ICD-10 diagnoses, yet it faces challenges due to complex medical language, privacy constraints, and limited annotated datasets. Large Language Models (LLMs) offer promising improvements in accuracy and efficiency for this task. This study evaluates the performance and consistency of eight LLMs; four reasoning (Qwen QWQ, Deepseek Reasoner, GPT o3 Mini, Gemini 2.0 Flash Thinking) and four non-reasoning (Llama 3.3, GPT 4o Mini, Gemini 2.0 Flash, Deepseek Chat); in classifying clinical discharge summaries using the MIMIC-IV dataset. Using cTAKES to structure clinical narratives, models were assessed across three experimental runs, with majority voting determining final predictions. Results showed that reasoning models outperformed non-reasoning models in accuracy (71% vs 68%) and F1 score (67% vs 60%), with Gemini 2.0 Flash Thinking achieving the highest accuracy (75%) and F1 score (76%). However, non-reasoning models demonstrated greater stability (91% vs 84% consistency). Performance varied across ICD-10 codes, with reasoning models excelling in complex cases but struggling with abstract categories. Findings indicate a trade-off between accuracy and consistency, suggesting that a hybrid approach could optimise clinical coding. Future research should explore multi-label classification, domain-specific fine-tuning, and ensemble methods to enhance model reliability in real-world applications.
Authors: Shuming Ma, Hongyu Wang, Shaohan Huang, Xingxing Zhang, Ying Hu, Ting Song, Yan Xia, Furu Wei
Abstract: We introduce BitNet b1.58 2B4T, the first open-source, native 1-bit Large Language Model (LLM) at the 2-billion parameter scale. Trained on a corpus of 4 trillion tokens, the model has been rigorously evaluated across benchmarks covering language understanding, mathematical reasoning, coding proficiency, and conversational ability. Our results demonstrate that BitNet b1.58 2B4T achieves performance on par with leading open-weight, full-precision LLMs of similar size, while offering significant advantages in computational efficiency, including substantially reduced memory footprint, energy consumption, and decoding latency. To facilitate further research and adoption, the model weights are released via Hugging Face along with open-source inference implementations for both GPU and CPU architectures.
Authors: Yihan Lin, Zhirong Bella Yu, Simon Lee
Abstract: Synthetic Electronic Health Records (EHRs) offer a valuable opportunity to create privacy preserving and harmonized structured data, supporting numerous applications in healthcare. Key benefits of synthetic data include precise control over the data schema, improved fairness and representation of patient populations, and the ability to share datasets without concerns about compromising real individuals privacy. Consequently, the AI community has increasingly turned to Large Language Models (LLMs) to generate synthetic data across various domains. However, a significant challenge in healthcare is ensuring that synthetic health records reliably generalize across different hospitals, a long standing issue in the field. In this work, we evaluate the current state of commercial LLMs for generating synthetic data and investigate multiple aspects of the generation process to identify areas where these models excel and where they fall short. Our main finding from this work is that while LLMs can reliably generate synthetic health records for smaller subsets of features, they struggle to preserve realistic distributions and correlations as the dimensionality of the data increases, ultimately limiting their ability to generalize across diverse hospital settings.
Authors: Junshu Pan, Wei Shen, Shulin Huang, Qiji Zhou, Yue Zhang
Abstract: Direct Preference Optimization (DPO) simplifies reinforcement learning from human feedback (RLHF) for large language models (LLMs) by directly optimizing human preferences without an explicit reward model. We find that during DPO training, the reference model plays the role of a data weight adjuster. However, the common practice of initializing the policy and reference models identically in DPO can lead to inefficient data utilization and impose a performance ceiling. Meanwhile, the lack of a reference model in Simple Preference Optimization (SimPO) reduces training robustness and necessitates stricter conditions to prevent catastrophic forgetting. In this work, we propose Pre-DPO, a simple yet effective DPO-based training paradigm that enhances preference optimization performance by leveraging a guiding reference model. This reference model provides foresight into the optimal policy state achievable through the training preference data, serving as a guiding mechanism that adaptively assigns higher weights to samples more suitable for the model and lower weights to those less suitable. Extensive experiments on AlpacaEval 2.0 and Arena-Hard v0.1 benchmarks demonstrate that Pre-DPO consistently improves the performance of both DPO and SimPO, without relying on external models or additional data.
Authors: Tom Zehle, Moritz Schlager, Timo Hei{\ss}, Matthias Feurer
Abstract: Large language models (LLMs) have revolutionized natural language processing by solving a wide range of tasks simply guided by a prompt. Yet their performance is highly sensitive to prompt formulation. While automated prompt optimization addresses this challenge by finding optimal prompts, current methods require a substantial number of LLM calls and input tokens, making prompt optimization expensive. We introduce CAPO (Cost-Aware Prompt Optimization), an algorithm that enhances prompt optimization efficiency by integrating AutoML techniques. CAPO is an evolutionary approach with LLMs as operators, incorporating racing to save evaluations and multi-objective optimization to balance performance with prompt length. It jointly optimizes instructions and few-shot examples while leveraging task descriptions for improved robustness. Our extensive experiments across diverse datasets and LLMs demonstrate that CAPO outperforms state-of-the-art discrete prompt optimization methods in 11/15 cases with improvements up to 21%p. Our algorithm achieves better performances already with smaller budgets, saves evaluations through racing, and decreases average prompt length via a length penalty, making it both cost-efficient and cost-aware. Even without few-shot examples, CAPO outperforms its competitors and generally remains robust to initial prompts. CAPO represents an important step toward making prompt optimization more powerful and accessible by improving cost-efficiency.
Authors: Muskan Garg, Shaina Raza, Shebuti Rayana, Xingyi Liu, Sunghwan Sohn
Abstract: Despite substantial progress in healthcare applications driven by large language models (LLMs), growing concerns around data privacy, and limited resources; the small language models (SLMs) offer a scalable and clinically viable solution for efficient performance in resource-constrained environments for next-generation healthcare informatics. Our comprehensive survey presents a taxonomic framework to identify and categorize them for healthcare professionals and informaticians. The timeline of healthcare SLM contributions establishes a foundational framework for analyzing models across three dimensions: NLP tasks, stakeholder roles, and the continuum of care. We present a taxonomic framework to identify the architectural foundations for building models from scratch; adapting SLMs to clinical precision through prompting, instruction fine-tuning, and reasoning; and accessibility and sustainability through compression techniques. Our primary objective is to offer a comprehensive survey for healthcare professionals, introducing recent innovations in model optimization and equipping them with curated resources to support future research and development in the field. Aiming to showcase the groundbreaking advancements in SLMs for healthcare, we present a comprehensive compilation of experimental results across widely studied NLP tasks in healthcare to highlight the transformative potential of SLMs in healthcare. The updated repository is available at Github
Authors: Xiaoyu Tian, Sitong Zhao, Haotian Wang, Shuaiting Chen, Yiping Peng, Yunjie Ji, Han Zhao, Xiangang Li
Abstract: Although large language models (LLMs) have recently achieved remarkable performance on various complex reasoning benchmarks, the academic community still lacks an in-depth understanding of base model training processes and data quality. To address this, we construct a large-scale, difficulty-graded reasoning dataset containing approximately 3.34 million unique queries of varying difficulty levels and about 40 million distilled responses generated by multiple models over several passes. Leveraging pass rate and Coefficient of Variation (CV), we precisely select the most valuable training data to enhance reasoning capability. Notably, we observe a training pattern shift, indicating that reasoning-focused training based on base models requires higher learning rates for effective training. Using this carefully selected data, we significantly improve the reasoning capabilities of the base model, achieving a pass rate of 79.2\% on the AIME2024 mathematical reasoning benchmark. This result surpasses most current distilled models and closely approaches state-of-the-art performance. We provide detailed descriptions of our data processing, difficulty assessment, and training methodology, and have publicly released all datasets and methods to promote rapid progress in open-source long-reasoning LLMs. The dataset is available at: https://huggingface.co/datasets/a-m-team/AM-DeepSeek-Distilled-40M
URLs: https://huggingface.co/datasets/a-m-team/AM-DeepSeek-Distilled-40M
Authors: Yuanchang Ye, Weiyan Wen
Abstract: This study addresses the critical challenge of hallucination mitigation in Large Vision-Language Models (LVLMs) for Visual Question Answering (VQA) tasks through a Split Conformal Prediction (SCP) framework. While LVLMs excel in multi-modal reasoning, their outputs often exhibit hallucinated content with high confidence, posing risks in safety-critical applications. We propose a model-agnostic uncertainty quantification method that integrates dynamic threshold calibration and cross-modal consistency verification. By partitioning data into calibration and test sets, the framework computes nonconformity scores to construct prediction sets with statistical guarantees under user-defined risk levels ($\alpha$). Key innovations include: (1) rigorous control of \textbf{marginal coverage} to ensure empirical error rates remain strictly below $\alpha$; (2) dynamic adjustment of prediction set sizes inversely with $\alpha$, filtering low-confidence outputs; (3) elimination of prior distribution assumptions and retraining requirements. Evaluations on benchmarks (ScienceQA, MMMU) with eight LVLMs demonstrate that SCP enforces theoretical guarantees across all $\alpha$ values. The framework achieves stable performance across varying calibration-to-test split ratios, underscoring its robustness for real-world deployment in healthcare, autonomous systems, and other safety-sensitive domains. This work bridges the gap between theoretical reliability and practical applicability in multi-modal AI systems, offering a scalable solution for hallucination detection and uncertainty-aware decision-making.
Authors: Ahmad AbdulKader, Kareem Nassar, Mohamed El-Geish, Daniel Galvez, Chetan Patil
Abstract: We propose using cascaded classifiers for a keyword spotting (KWS) task on narrow-band (NB), 8kHz audio acquired in non-IID environments -- a more challenging task than most state-of-the-art KWS systems face. We present a model that incorporates Deep Neural Networks (DNNs), cascading, multiple-feature representations, and multiple-instance learning. The cascaded classifiers handle the task's class imbalance and reduce power consumption on computationally-constrained devices via early termination. The KWS system achieves a false negative rate of 6% at an hourly false positive rate of 0.75
Authors: Hongyu Wang, Jiayu Xu, Senwei Xie, Ruiping Wang, Jialin Li, Zhaojie Xie, Bin Zhang, Chuyan Xiong, Xilin Chen
Abstract: Multilingual capability is an essential aspect for large multimodal models, since they are usually deployed across various countries and languages. However, most existing benchmarks for multilingual multimodal reasoning struggle to differentiate between models of varying performance; even language models without visual capabilities can easily achieve high scores. This leaves a comprehensive evaluation of leading multilingual multimodal models largely unexplored. In this work, we introduce M4U, a novel and challenging benchmark for assessing the capability of multi-discipline multilingual multimodal understanding and reasoning. M4U contains 10k samples covering 64 disciplines across 16 subfields in Science, Engineering, and Healthcare in six languages. Using M4U, we conduct extensive evaluations of leading Large Multimodal Models (LMMs) and Large Language Models (LLMs) with external tools. The evaluation results demonstrate that the state-of-the-art model, GPT-4o, achieves only 47.6% average accuracy on M4U. Additionally, we observe that the leading LMMs exhibit significant language preferences. Our in-depth analysis indicates that leading LMMs, including GPT-4o, struggle to perform reasoning using multilingual information present in both visual and textual context. Specifically, they suffer performance degradation when prompted with cross-lingual multimodal questions. Our code and dataset is public available.
Authors: Ognjen Kundacina, Vladimir Vincan, Dragisa Miskovic
Abstract: This paper introduces a novel two-stage active learning (AL) pipeline for automatic speech recognition (ASR), combining unsupervised and supervised AL methods. The first stage utilizes unsupervised AL by using x-vectors clustering for diverse sample selection from unlabeled speech data, thus establishing a robust initial dataset for the subsequent supervised AL. The second stage incorporates a supervised AL strategy, with a batch AL method specifically developed for ASR, aimed at selecting diverse and informative batches of samples. Here, sample diversity is also achieved using x-vectors clustering, while the most informative samples are identified using a Bayesian AL method tailored for ASR with an adaptation of Monte Carlo dropout to approximate Bayesian inference. This approach enables precise uncertainty estimation, thereby enhancing ASR model training with significantly reduced data requirements. Our method has shown superior performance compared to competing methods on homogeneous, heterogeneous, and OOD test sets, demonstrating that strategic sample selection and innovative Bayesian modeling can substantially optimize both labeling effort and data utilization in deep learning-based ASR applications.
Authors: Lecheng Kong, Jiarui Feng, Hao Liu, Chengsong Huang, Jiaxin Huang, Yixin Chen, Muhan Zhang
Abstract: Foundation models, such as Large Language Models (LLMs) or Large Vision Models (LVMs), have emerged as one of the most powerful tools in the respective fields. However, unlike text and image data, graph data do not have a definitive structure, posing great challenges to developing a Graph Foundation Model (GFM). For example, current attempts at designing general graph models either transform graph data into a language format for LLM-based prediction or still train a GNN model with LLM as an assistant. The former can handle unlimited tasks, while the latter captures graph structure much better -- yet, no existing work can achieve both simultaneously. In this paper, we identify three key desirable properties of a GFM: self-supervised pretraining, fluidity in tasks, and graph awareness. To account for these properties, we extend the conventional language modeling to the graph domain and propose a novel generative graph language model GOFA to solve the problem. The model interleaves randomly initialized GNN layers into a frozen pre-trained LLM so that the semantic and structural modeling abilities are organically combined. GOFA is pre-trained on newly proposed graph-level next-word prediction, question-answering, and structural tasks to obtain the above GFM properties. The pre-trained model is further fine-tuned on downstream tasks to obtain task-solving ability. The fine-tuned model is evaluated on various downstream tasks, demonstrating a strong ability to solve structural and contextual problems in zero-shot scenarios. The code is available at https://github.com/JiaruiFeng/GOFA.
Authors: Sungmin Cha, Sungjun Cho, Dasol Hwang, Moontae Lee
Abstract: Large Language Models (LLMs) have demonstrated strong reasoning and memorization capabilities via pretraining on massive textual corpora. However, this poses risk of privacy and copyright violations, highlighting the need for efficient machine unlearning methods that remove sensitive data without retraining from scratch. While Gradient Ascent (GA) is commonly used to unlearn by reducing the likelihood of generating unwanted content, it leads to unstable optimization and catastrophic forgetting of retrained knowledge. We find that combining GA with low-rank adaptation results in poor trade-offs between computational cost and generative performance. To address these challenges, we propose Low-rank Knowledge Unlearning (LoKU), a novel framework that enables robust and efficient unlearning for LLMs. First, we introduce Inverted Hinge Loss, which suppresses unwanted tokens while maintaining fluency by boosting the probability of the next most likely token. Second, we develop a data-adaptive initialization for LoRA adapters via low-rank approximation weighted with relative Fisher information, thereby focusing updates on parameters critical for removing targeted knowledge. Experiments on the Training Data Extraction Challenge dataset using GPT-Neo models as well as on the TOFU benchmark with Phi-1.5B and Llama2-7B models demonstrate that our approach effectively removes sensitive information while maintaining reasoning and generative capabilities with minimal impact. Our implementation can be found in https://github.com/csm9493/efficient-llm-unlearning.
Authors: Jungeum Kim, Sean O'Hagan, Veronika Rockova
Abstract: This work is concerned with conformal prediction in contemporary applications (including generative AI) where a black-box model has been trained on data that are not accessible to the user. Mirroring split-conformal inference, we design a wrapper around a black-box algorithm which calibrates conformity scores. This calibration is local and proceeds in two stages by first adaptively partitioning the predictor space into groups and then calibrating sectionally group by group. Adaptive partitioning (self-grouping) is achieved by fitting a robust regression tree to the conformity scores on the calibration set. This new tree variant is designed in such a way that adding a single new observation does not change the tree fit with overwhelmingly large probability. This add-one-in robustness property allows us to conclude a finite sample group-conditional coverage guarantee, a refinement of the marginal guarantee. In addition, unlike traditional split-conformal inference, adaptive splitting and within-group calibration yields adaptive bands which can stretch and shrink locally. We demonstrate benefits of local tightening on several simulated as well as real examples using non-parametric regression. Finally, we consider two contemporary classification applications for obtaining uncertainty quantification around GPT-4o predictions. We conformalize skin disease diagnoses based on self-reported symptoms as well as predicted states of U.S. legislators based on summaries of their ideology. We demonstrate substantial local tightening of the uncertainty sets while attaining similar marginal coverage.
Authors: Syeda Nahida Akter, Shrimai Prabhumoye, John Kamalu, Sanjeev Satheesh, Eric Nyberg, Mostofa Patwary, Mohammad Shoeybi, Bryan Catanzaro
Abstract: The utility of synthetic data to enhance pretraining data quality and hence to improve downstream task accuracy has been widely explored in recent large language models (LLMs). Yet, these approaches fall inadequate in complex, multi-hop and mathematical reasoning tasks as the synthetic data typically fails to add complementary knowledge to the existing raw corpus. In this work, we propose a novel large-scale and diverse Math Informed syNthetic Dialogue (MIND) generation method that improves the mathematical reasoning ability of LLMs. Specifically, using MIND, we generate synthetic conversations based on OpenWebMath (OWM), resulting in a new math corpus, MIND-OWM. Our experiments with different conversational settings reveal that incorporating knowledge gaps between dialog participants is essential for generating high-quality math data. We further identify an effective way to format and integrate synthetic and raw data during pretraining to maximize the gain in mathematical reasoning, emphasizing the need to restructure raw data rather than use it as-is. Compared to pretraining just on raw data, a model pretrained on MIND-OWM shows significant boost in mathematical reasoning (GSM8K: +13.42%, MATH: +2.30%), including superior performance in specialized knowledge (MMLU: +4.55%, MMLU-STEM: +4.28%) and general purpose reasoning tasks (GENERAL REASONING: +2.51%).
Authors: Shi Dong, Xiaobei Niu, Rui Zhong, Zhifeng Wang, Mingzhang Zuo
Abstract: Accurate annotation of educational resources is crucial for effective personalized learning and resource recommendation in online education. However, fine-grained knowledge labels often overlap or share similarities, making it difficult for existing multi-label classification methods to differentiate them. The label distribution imbalance due to sparsity of human annotations further intensifies these challenges. To address these issues, this paper introduces RR2QC, a novel Retrieval Reranking method to multi-label Question Classification by leveraging label semantics and meta-label refinement. First, RR2QC improves the pre-training strategy by utilizing semantic relationships within and across label groups. Second, it introduces a class center learning task to align questions with label semantics during downstream training. Finally, this method decomposes labels into meta-labels and uses a meta-label classifier to rerank the retrieved label sequences. In doing so, RR2QC enhances the understanding and prediction capability of long-tail labels by learning from meta-labels that frequently appear in other labels. Additionally, a mathematical LLM is used to generate solutions for questions, extracting latent information to further refine the model's insights. Experimental results show that RR2QC outperforms existing methods in Precision@K and F1 scores across multiple educational datasets, demonstrating its effectiveness for online education applications. The code and datasets are available at https://github.com/78Erii/RR2QC.
Authors: Anton Alyakin, Jaden Stryker, Daniel Alexander Alber, Karl L. Sangwon, Jin Vivian Lee, Brandon Duderstadt, Akshay Save, David Kurland, Spencer Frome, Shrutika Singh, Jeff Zhang, Eunice Yang, Ki Yun Park, Cordelia Orillac, Aly A. Valliani, Sean Neifert, Albert Liu, Aneek Patel, Christopher Livia, Darryl Lau, Ilya Laufer, Peter A. Rozman, Eveline Teresa Hidalgo, Howard Riina, Rui Feng, Todd Hollon, Yindalon Aphinyanaphongs, John G. Golfinos, Laura Snyder, Eric Leuthardt, Douglas Kondziolka, Eric Karl Oermann
Abstract: Leading vision-language models (VLMs) are trained on general Internet content, overlooking scientific journals' rich, domain-specific knowledge. Training on specialty-specific literature could yield high-performance, task-specific tools, enabling generative AI to match generalist models in specialty publishing, educational, and clinical tasks. We created NeuroPubs, a multimodal dataset of 23,000 Neurosurgery Publications articles (134M words, 78K image-caption pairs). Using NeuroPubs, VLMs generated publication-ready graphical abstracts (70% of 100 abstracts) and board-style questions indistinguishable from human-written ones (54% of 89,587 questions). We used these questions to train CNS-Obsidian, a 34B-parameter VLM. In a blinded, randomized controlled trial, our model demonstrated non-inferiority to then state-of-the-art GPT-4o in neurosurgical differential diagnosis (clinical utility, 40.62% upvotes vs. 57.89%, p=0.1150; accuracy, 59.38% vs. 65.79%, p=0.3797). Our pilot study demonstrates how training generative AI models on specialty-specific journal content - without large-scale internet data - results in high-performance academic and clinical tools, enabling domain-tailored AI across diverse fields.
Authors: Ayushi Mishra, Yang Bai, Priyadarshan Narayanasamy, Nakul Garg, Nirupam Roy
Abstract: Integrating spatial context into large language models (LLMs) has the potential to revolutionize human-computer interaction, particularly in wearable devices. In this work, we present a novel system architecture that incorporates spatial speech understanding into LLMs, enabling contextually aware and adaptive applications for wearable technologies. Our approach leverages microstructure-based spatial sensing to extract precise Direction of Arrival (DoA) information using a monaural microphone. To address the lack of existing dataset for microstructure-assisted speech recordings, we synthetically create a dataset called OmniTalk by using the LibriSpeech dataset. This spatial information is fused with linguistic embeddings from OpenAI's Whisper model, allowing each modality to learn complementary contextual representations. The fused embeddings are aligned with the input space of LLaMA-3.2 3B model and fine-tuned with lightweight adaptation technique LoRA to optimize for on-device processing. SING supports spatially-aware automatic speech recognition (ASR), achieving a mean error of $25.72^\circ$-a substantial improvement compared to the 88.52$^\circ$ median error in existing work-with a word error rate (WER) of 5.3. SING also supports soundscaping, for example, inference how many people were talking and their directions, with up to 5 people and a median DoA error of 16$^\circ$. Our system demonstrates superior performance in spatial speech understanding while addressing the challenges of power efficiency, privacy, and hardware constraints, paving the way for advanced applications in augmented reality, accessibility, and immersive experiences.
Authors: Yash Savani, Asher Trockman, Zhili Feng, Avi Schwarzschild, Alexander Robey, Marc Finzi, J. Zico Kolter
Abstract: Frontier models that generate extended reasoning traces inadvertently produce rich token sequences that can facilitate model distillation. Recognizing this vulnerability, model owners may seek sampling strategies that limit the effectiveness of distillation without compromising model performance. Antidistillation sampling provides exactly this capability. By strategically modifying a model's next-token probability distribution, antidistillation sampling poisons reasoning traces, rendering them significantly less effective for distillation while preserving the model's practical utility. For further details, see https://antidistillation.com.
Authors: Ling You, Wenxuan Huang, Xinni Xie, Xiangyi Wei, Bangyan Li, Shaohui Lin, Yang Li, Changbo Wang
Abstract: Soccer is a globally popular sporting event, typically characterized by long matches and distinctive highlight moments. Recent advances in Multimodal Large Language Models (MLLMs) offer promising capabilities in temporal grounding and video understanding, soccer commentary generation often requires precise temporal localization and semantically rich descriptions over long-form video. However, existing soccer MLLMs often rely on the temporal a priori for caption generation, so they cannot process the soccer video end-to-end. While some traditional approaches follow a two-step paradigm that is complex and fails to capture the global context to achieve suboptimal performance. To solve the above issues, we present TimeSoccer, the first end-to-end soccer MLLM for Single-anchor Dense Video Captioning (SDVC) in full-match soccer videos. TimeSoccer jointly predicts timestamps and generates captions in a single pass, enabling global context modeling across 45-minute matches. To support long video understanding of soccer matches, we introduce MoFA-Select, a training-free, motion-aware frame compression module that adaptively selects representative frames via a coarse-to-fine strategy, and incorporates complementary training paradigms to strengthen the model's ability to handle long temporal sequences. Extensive experiments demonstrate that our TimeSoccer achieves State-of-The-Art (SoTA) performance on the SDVC task in an end-to-end form, generating high-quality commentary with accurate temporal alignment and strong semantic relevance.