Authors: Lionel Wong, Ayman Ali, Raymond Xiong, Shannon Zeijang Shen, Yoon Kim, Monica Agrawal
Abstract: Patients have long sought health information online, and increasingly, they are turning to generative AI to answer their health-related queries. Given the high stakes of the medical domain, techniques like retrieval-augmented generation and citation grounding have been widely promoted as methods to reduce hallucinations and improve the accuracy of AI-generated responses and have been widely adopted into search engines. This paper argues that even when these methods produce literally accurate content drawn from source documents sans hallucinations, they can still be highly misleading. Patients may derive significantly different interpretations from AI-generated outputs than they would from reading the original source material, let alone consulting a knowledgeable clinician. Through a large-scale query analysis on topics including disputed diagnoses and procedure safety, we support our argument with quantitative and qualitative evidence of the suboptimal answers resulting from current systems. In particular, we highlight how these models tend to decontextualize facts, omit critical relevant sources, and reinforce patient misconceptions or biases. We propose a series of recommendations -- such as the incorporation of communication pragmatics and enhanced comprehension of source documents -- that could help mitigate these issues and extend beyond the medical domain.
Authors: Georgios P. Georgiou
Abstract: One ongoing debate in linguistics is whether Artificial Intelligence (AI) can effectively mimic human performance in language-related tasks. While much research has focused on various linguistic abilities of AI, little attention has been given to how it defines neologisms formed through different word formation processes. This study addresses this gap by examining the degree of agreement between human and AI-generated responses in defining three types of Greek neologisms: blends, compounds, and derivatives. The study employed an online experiment in which human participants selected the most appropriate definitions for neologisms, while ChatGPT received identical prompts. The results revealed fair agreement between human and AI responses for blends and derivatives but no agreement for compounds. However, when considering the majority response among humans, agreement with AI was high for blends and derivatives. These findings highlight the complexity of human language and the challenges AI still faces in capturing its nuances. In particular, they suggest a need for integrating more advanced semantic networks and contextual learning mechanisms into AI models to improve their interpretation of complex word formations, especially compounds.
Authors: Jonathan Bourne
Abstract: Oscar Wilde said, "The difference between literature and journalism is that journalism is unreadable, and literature is not read." Unfortunately, The digitally archived journalism of Oscar Wilde's 19th century often has no or poor quality Optical Character Recognition (OCR), reducing the accessibility of these archives and making them unreadable both figuratively and literally. This paper helps address the issue by performing OCR on "The Nineteenth Century Serials Edition" (NCSE), an 84k-page collection of 19th-century English newspapers and periodicals, using Pixtral 12B, a pre-trained image-to-text language model. The OCR capability of Pixtral was compared to 4 other OCR approaches, achieving a median character error rate of 1%, 5x lower than the next best model. The resulting NCSE v2.0 dataset features improved article identification, high-quality OCR, and text classified into four types and seventeen topics. The dataset contains 1.4 million entries, and 321 million words. Example use cases demonstrate analysis of topic similarity, readability, and event tracking. NCSE v2.0 is freely available to encourage historical and sociological research. As a result, 21st-century readers can now share Oscar Wilde's disappointment with 19th-century journalistic standards, reading the unreadable from the comfort of their own computers.
Authors: Boyu Chen, Zirui Guo, Zidan Yang, Yuluo Chen, Junze Chen, Zhenghao Liu, Chuan Shi, Cheng Yang
Abstract: Retrieval-augmented generation (RAG) improves the response quality of large language models (LLMs) by retrieving knowledge from external databases. Typical RAG approaches split the text database into chunks, organizing them in a flat structure for efficient searches. To better capture the inherent dependencies and structured relationships across the text database, researchers propose to organize textual information into an indexing graph, known asgraph-based RAG. However, we argue that the limitation of current graph-based RAG methods lies in the redundancy of the retrieved information, rather than its insufficiency. Moreover, previous methods use a flat structure to organize retrieved information within the prompts, leading to suboptimal performance. To overcome these limitations, we propose PathRAG, which retrieves key relational paths from the indexing graph, and converts these paths into textual form for prompting LLMs. Specifically, PathRAG effectively reduces redundant information with flow-based pruning, while guiding LLMs to generate more logical and coherent responses with path-based prompting. Experimental results show that PathRAG consistently outperforms state-of-the-art baselines across six datasets and five evaluation dimensions. The code is available at the following link: https://github.com/BUPT-GAMMA/PathRAG
Authors: Bhavik Agarwal, Ishan Joshi, Viktoria Rojkova
Abstract: In this paper, we address the challenge of enforcing strict schema adherence in large language model (LLM) generation by leveraging LLM reasoning capabilities. Building on the DeepSeek R1 reinforcement learning framework, our approach trains structured reasoning skills of a 1.5B parameter model through a novel pipeline that combines synthetic reasoning dataset construction with custom reward functions under Group Relative Policy Optimization (GRPO). Specifically, we first perform R1 reinforcement learning on a 20K sample unstructured-to-structured dataset, mirroring the original DeepSeek R1 methods, to establish core reasoning abilities. Subsequently, we performed supervised fine-tuning on a separate 10K reasoning sample dataset, focusing on refining schema adherence for downstream tasks. Despite the relatively modest training scope, requiring approximately 20 hours on an 8xH100 GPU cluster for GRPO training and 3 hours on 1xA100 for SFT, our model demonstrates robust performance in enforcing schema consistency. We compare our ThinkJSON approach against the original DeepSeek R1 (671B), distilled versions of DeepSeek R1 (Qwen-1.5B and Qwen-7B), and Gemini 2.0 Flash (70B), showcasing its effectiveness in real-world applications. Our results underscore the practical utility of a resource-efficient framework for schema-constrained text generation.
Authors: Srishti Yadav, Zhi Zhang, Daniel Hershcovich, Ekaterina Shutova
Abstract: Investigating value alignment in Large Language Models (LLMs) based on cultural context has become a critical area of research. However, similar biases have not been extensively explored in large vision-language models (VLMs). As the scale of multimodal models continues to grow, it becomes increasingly important to assess whether images can serve as reliable proxies for culture and how these values are embedded through the integration of both visual and textual data. In this paper, we conduct a thorough evaluation of multimodal model at different scales, focusing on their alignment with cultural values. Our findings reveal that, much like LLMs, VLMs exhibit sensitivity to cultural values, but their performance in aligning with these values is highly context-dependent. While VLMs show potential in improving value understanding through the use of images, this alignment varies significantly across contexts highlighting the complexities and underexplored challenges in the alignment of multimodal models.
Authors: Hajar Emami Gohari, Swanand Ravindra Kadhe, Syed Yousaf Shah. Constantin Adam, Abdulhamid Adebayo, Praneet Adusumilli, Farhan Ahmed, Nathalie Baracaldo Angel, Santosh Borse, Yuan-Chi Chang, Xuan-Hong Dang, Nirmit Desai, Ravital Eres, Ran Iwamoto, Alexei Karve, Yan Koyfman, Wei-Han Lee, Changchang Liu, Boris Lublinsky, Takuyo Ohko, Pablo Pesce, Maroun Touma, Shiqiang Wang, Shalisha Witherspoon, Herbert Woisetschlager, David Wood, Kun-Lung Wu, Issei Yoshida, Syed Zawad, Petros Zerfos, Yi Zhou, Bishwaranjan Bhattacharjee
Abstract: Data quantity and quality play a vital role in determining the performance of Large Language Models (LLMs). High-quality data, in particular, can significantly boost the LLM's ability to generalize on a wide range of downstream tasks. Large pre-training datasets for leading LLMs remain inaccessible to the public, whereas many open datasets are small in size (less than 5 trillion tokens), limiting their suitability for training large models. In this paper, we introduce GneissWeb, a large dataset yielding around 10 trillion tokens that caters to the data quality and quantity requirements of training LLMs. Our GneissWeb recipe that produced the dataset consists of sharded exact sub-string deduplication and a judiciously constructed ensemble of quality filters. GneissWeb achieves a favorable trade-off between data quality and quantity, producing models that outperform models trained on state-of-the-art open large datasets (5+ trillion tokens). We show that models trained using GneissWeb dataset outperform those trained on FineWeb-V1.1.0 by 2.73 percentage points in terms of average score computed on a set of 11 commonly used benchmarks (both zero-shot and few-shot) for pre-training dataset evaluation. When the evaluation set is extended to 20 benchmarks (both zero-shot and few-shot), models trained using GneissWeb still achieve a 1.75 percentage points advantage over those trained on FineWeb-V1.1.0.
Authors: Shangyu Wu, Hongchao Du, Ying Xiong, Shuai Chen, Tei-wei Kuo, Nan Guan, Chun Jason Xue
Abstract: Large Language Models (LLMs) have achieved remarkable success in natural language processing tasks, but their massive size and computational demands hinder their deployment in resource-constrained environments. Existing structured pruning methods address this issue by removing redundant structures (e.g., elements, channels, layers) from the model. However, these methods employ a heuristic pruning strategy, which leads to suboptimal performance. Besides, they also ignore the data characteristics when pruning the model. To overcome these limitations, we propose EvoP, an evolutionary pruning framework for robust LLM inference. EvoP first presents a cluster-based calibration dataset sampling (CCDS) strategy for creating a more diverse calibration dataset. EvoP then introduces an evolutionary pruning pattern searching (EPPS) method to find the optimal pruning pattern. Compared to existing structured pruning techniques, EvoP achieves the best performance while maintaining the best efficiency. Experiments across different LLMs and different downstream tasks validate the effectiveness of the proposed EvoP, making it a practical and scalable solution for deploying LLMs in real-world applications.
Authors: Jann Railey Montalan, Jimson Paulo Layacan, David Demitri Africa, Richell Isaiah Flores, Michael T. Lopez II, Theresa Denise Magsajo, Anjanette Cayabyab, William Chandra Tjhi
Abstract: Recent advances in large language models (LLMs) have demonstrated remarkable capabilities on widely benchmarked high-resource languages; however, linguistic nuances of under-resourced languages remain unexplored. We introduce Batayan, a holistic Filipino benchmark designed to systematically evaluate LLMs across three key natural language processing (NLP) competencies: understanding, reasoning, and generation. Batayan consolidates eight tasks, covering both Tagalog and code-switched Taglish utterances. Our rigorous, native-speaker-driven annotation process ensures fluency and authenticity to the complex morphological and syntactic structures of Filipino, alleviating a pervasive translationese bias in existing Filipino corpora. We report empirical results on a variety of multilingual LLMs, highlighting significant performance gaps that signal the under-representation of Filipino in pretraining corpora, the unique hurdles in modeling Filipino's rich morphology and construction, and the importance of explicit Filipino language support and instruction tuning. Moreover, we discuss the practical challenges encountered in dataset construction and propose principled solutions for building culturally and linguistically-faithful resources in under-represented languages. We also provide a public benchmark and leaderboard as a clear foundation for iterative, community-driven progress in Filipino NLP.
Authors: Yunze Jia, Yuehui Xian, Yangyang Xu, Pengfei Dang, Xiangdong Ding, Jun Sun, Yumei Zhou, Dezhen Xue
Abstract: We present a framework for generating universal semantic embeddings of chemical elements to advance materials inference and discovery. This framework leverages ElementBERT, a domain-specific BERT-based natural language processing model trained on 1.29 million abstracts of alloy-related scientific papers, to capture latent knowledge and contextual relationships specific to alloys. These semantic embeddings serve as robust elemental descriptors, consistently outperforming traditional empirical descriptors with significant improvements across multiple downstream tasks. These include predicting mechanical and transformation properties, classifying phase structures, and optimizing materials properties via Bayesian optimization. Applications to titanium alloys, high-entropy alloys, and shape memory alloys demonstrate up to 23% gains in prediction accuracy. Our results show that ElementBERT surpasses general-purpose BERT variants by encoding specialized alloy knowledge. By bridging contextual insights from scientific literature with quantitative inference, our framework accelerates the discovery and optimization of advanced materials, with potential applications extending beyond alloys to other material classes.
Authors: Xiangjin Xie, Guangwei Xu, Lingyan Zhao, Ruijie Guo
Abstract: Although multi-agent collaborative Large Language Models (LLMs) have achieved significant breakthroughs in the Text-to-SQL task, their performance is still constrained by various factors. These factors include the incompleteness of the framework, failure to follow instructions, and model hallucination problems. To address these problems, we propose OpenSearch-SQL, which divides the Text-to-SQL task into four main modules: Preprocessing, Extraction, Generation, and Refinement, along with an Alignment module based on a consistency alignment mechanism. This architecture aligns the inputs and outputs of agents through the Alignment module, reducing failures in instruction following and hallucination. Additionally, we designed an intermediate language called SQL-Like and optimized the structured CoT based on SQL-Like. Meanwhile, we developed a dynamic few-shot strategy in the form of self-taught Query-CoT-SQL. These methods have significantly improved the performance of LLMs in the Text-to-SQL task. In terms of model selection, we directly applied the base LLMs without any post-training, thereby simplifying the task chain and enhancing the framework's portability. Experimental results show that OpenSearch-SQL achieves an execution accuracy(EX) of 69.3% on the BIRD development set, 72.28% on the test set, and a reward-based validity efficiency score (R-VES) of 69.36%, with all three metrics ranking first at the time of submission. These results demonstrate the comprehensive advantages of the proposed method in both effectiveness and efficiency.
Authors: Xinxin You, Xien Liu, Xue Yang, Ziyi Wang, Ji Wu
Abstract: The task of automatically coding the International Classification of Diseases (ICD) in the medical field has been well-established and has received much attention. Automatic coding of the ICD in the medical field has been successful in English but faces challenges when dealing with Chinese electronic medical records (EMRs). The first issue lies in the difficulty of extracting disease code-related information from Chinese EMRs, primarily due to the concise writing style and specific internal structure of the EMRs. The second problem is that previous methods have failed to leverage the disease-based multi-axial knowledge and lack of association with the corresponding clinical evidence. This paper introduces a novel framework called MKE-Coder: Multi-axial Knowledge with Evidence verification in ICD coding for Chinese EMRs. Initially, we identify candidate codes for the diagnosis and categorize each of them into knowledge under four coding axes.Subsequently, we retrieve corresponding clinical evidence from the comprehensive content of EMRs and filter credible evidence through a scoring model. Finally, to ensure the validity of the candidate code, we propose an inference module based on the masked language modeling strategy. This module verifies that all the axis knowledge associated with the candidate code is supported by evidence and provides recommendations accordingly. To evaluate the performance of our framework, we conduct experiments using a large-scale Chinese EMR dataset collected from various hospitals. The experimental results demonstrate that MKE-Coder exhibits significant superiority in the task of automatic ICD coding based on Chinese EMRs. In the practical evaluation of our method within simulated real coding scenarios, it has been demonstrated that our approach significantly aids coders in enhancing both their coding accuracy and speed.
Authors: Matthieu Meeus, Lukas Wutschitz, Santiago Zanella-B\'eguelin, Shruti Tople, Reza Shokri
Abstract: How much information about training samples can be gleaned from synthetic data generated by Large Language Models (LLMs)? Overlooking the subtleties of information flow in synthetic data generation pipelines can lead to a false sense of privacy. In this paper, we design membership inference attacks (MIAs) that target data used to fine-tune pre-trained LLMs that are then used to synthesize data, particularly when the adversary does not have access to the fine-tuned model but only to the synthetic data. We show that such data-based MIAs do significantly better than a random guess, meaning that synthetic data leaks information about the training data. Further, we find that canaries crafted to maximize vulnerability to model-based MIAs are sub-optimal for privacy auditing when only synthetic data is released. Such out-of-distribution canaries have limited influence on the model's output when prompted to generate useful, in-distribution synthetic data, which drastically reduces their vulnerability. To tackle this problem, we leverage the mechanics of auto-regressive models to design canaries with an in-distribution prefix and a high-perplexity suffix that leave detectable traces in synthetic data. This enhances the power of data-based MIAs and provides a better assessment of the privacy risks of releasing synthetic data generated by LLMs.
Authors: Zihao Zeng, Xuyao Huang, Boxiu Li, Zhijie Deng
Abstract: This paper identifies the misinterpretation of the context can be a significant issue during the reasoning process of large language models, spanning from smaller models like Llama3.2-3B-Instruct to cutting-edge ones like DeepSeek-R1. For example, in the phrase "10 dollars per kilo," LLMs might not recognize that "per" means "for each," leading to calculation errors. We introduce a novel, post-training approach called **Stick to the Facts (SIFT)** to tackle this. SIFT leverages increasing inference-time compute to ground LLM reasoning in contexts. At the core of SIFT lies the *Sticker*, which is generated by the model itself to explicitly emphasize the key information within the context. Given the curated Sticker, SIFT generates two predictions -- one from the original query and one from the query augmented with the Sticker. If they differ, the Sticker is sequentially refined via *forward* optimization (to better align the extracted facts with the query) and *inverse* generation (to conform with the model's inherent tendencies) for more faithful reasoning outcomes. Studies across diverse models (from 3B to 100B+) and benchmarks (e.g., GSM8K, MATH-500) reveal consistent performance improvements. Notably, SIFT improves the pass@1 accuracy of DeepSeek-R1 on AIME2024 from 78.33% to **85.67**%, establishing a new state-of-the-art in the open-source community. The code is available at https://github.com/zhijie-group/SIFT.
Authors: Jose F Quesada
Abstract: While language technologies have advanced significantly, current approaches fail to address the complex sociocultural dimensions of linguistic preservation. AI Thinking proposes a meaning-centered framework that would transform technological development from creating tools FOR communities to co-creating solutions WITH them. This approach recognizes that meaningful solutions emerge through the interplay of cultural understanding, community agency, and technological innovation. The proposal articulates a holistic methodology and a five-layer technological ecosystem where communities maintain control over their linguistic and cultural knowledge representation. This systematic integration of community needs, cultural preservation, and advanced capabilities could revolutionize how we approach linguistic diversity preservation in the digital age.
Authors: Ibrahim Alabdulmohsin, Andreas Steiner
Abstract: Language exhibits a fractal structure in its information-theoretic complexity (i.e. bits per token), with self-similarity across scales and long-range dependence (LRD). In this work, we investigate whether large language models (LLMs) can replicate such fractal characteristics and identify conditions-such as temperature setting and prompting method-under which they may fail. Moreover, we find that the fractal parameters observed in natural language are contained within a narrow range, whereas those of LLMs' output vary widely, suggesting that fractal parameters might prove helpful in detecting a non-trivial portion of LLM-generated texts. Notably, these findings, and many others reported in this work, are robust to the choice of the architecture; e.g. Gemini 1.0 Pro, Mistral-7B and Gemma-2B. We also release a dataset comprising of over 240,000 articles generated by various LLMs (both pretrained and instruction-tuned) with different decoding temperatures and prompting methods, along with their corresponding human-generated texts. We hope that this work highlights the complex interplay between fractal properties, prompting, and statistical mimicry in LLMs, offering insights for generating, evaluating and detecting synthetic texts.
Authors: Han Zhang, Langshi Zhou, Hanfang Yang
Abstract: Extensive research has investigated the integration of large language models (LLMs) with knowledge graphs to enhance the reasoning process. However, understanding how models perform reasoning utilizing structured graph knowledge remains underexplored. Most existing approaches rely on LLMs or retrievers to make binary judgments regarding the utilization of knowledge, which is too coarse. Meanwhile, there is still a lack of feedback mechanisms for reflection and correction throughout the entire reasoning path. This paper proposes an Active self-Reflection framework for knowledge Graph reasoning ARG, introducing for the first time an end-to-end training approach to achieve iterative reasoning grounded on structured graphs. Within the framework, the model leverages special tokens to \textit{actively} determine whether knowledge retrieval is necessary, performs \textit{reflective} critique based on the retrieved knowledge, and iteratively reasons over the knowledge graph. The reasoning paths generated by the model exhibit high interpretability, enabling deeper exploration of the model's understanding of structured knowledge. Ultimately, the proposed model achieves outstanding results compared to existing baselines in knowledge graph reasoning tasks.
Authors: Sil Hamilton, David Mimno
Abstract: General-purpose language models are trained to produce varied natural language outputs, but for some tasks like annotation or classification we need more specific output formats. LLM systems increasingly support structured output, sampling tokens according to a grammar, which enforces a format but which can also reduce performance. We ask whether there are systematic differences between grammars that appear semantically similar to humans. To answer this question, we test four popular model families with five token formats on four NLP benchmarks. All models perform most accurately when instructed to classify with real numbers. Performance also improves by 5%-10% when models are instructed to return tokens incorporating leading whitespace, which we find can help models avoid structural deficiencies in subword token representations. Format-based differences are largest for smaller models that are often used for local laptop-scale inference. We present best practices for researchers using language models as zero-shot classifiers with structured output.
Authors: David Noever, Grant Rosario
Abstract: We present an open-source benchmark and evaluation framework for assessing emotional boundary handling in Large Language Models (LLMs). Using a dataset of 1156 prompts across six languages, we evaluated three leading LLMs (GPT-4o, Claude-3.5 Sonnet, and Mistral-large) on their ability to maintain appropriate emotional boundaries through pattern-matched response analysis. Our framework quantifies responses across seven key patterns: direct refusal, apology, explanation, deflection, acknowledgment, boundary setting, and emotional awareness. Results demonstrate significant variation in boundary-handling approaches, with Claude-3.5 achieving the highest overall score (8.69/10) and producing longer, more nuanced responses (86.51 words on average). We identified a substantial performance gap between English (average score 25.62) and non-English interactions (< 0.22), with English responses showing markedly higher refusal rates (43.20% vs. < 1% for non-English). Pattern analysis revealed model-specific strategies, such as Mistral's preference for deflection (4.2%) and consistently low empathy scores across all models (< 0.06). Limitations include potential oversimplification through pattern matching, lack of contextual understanding in response analysis, and binary classification of complex emotional responses. Future work should explore more nuanced scoring methods, expand language coverage, and investigate cultural variations in emotional boundary expectations. Our benchmark and methodology provide a foundation for systematic evaluation of LLM emotional intelligence and boundary-setting capabilities.
Authors: Lew Lefton, Kexin Rong, Chinar Dankhara, Lila Ghemri, Firdous Kausar, A. Hannibal Hamdallahi
Abstract: In this paper, we propose a Retrieval Augmented Generation (RAG) agent that maps natural language queries about research topics to precise, machine-interpretable semantic entities. Our approach combines RAG with Socratic dialogue to align a user's intuitive understanding of research topics with established Knowledge Organization Systems (KOSs). The proposed approach will effectively bridge "little semantics" (domain-specific KOS structures) with "big semantics" (broad bibliometric repositories), making complex academic taxonomies more accessible. Such agents have the potential for broad use. We illustrate with a sample application called CollabNext, which is a person-centric knowledge graph connecting people, organizations, and research topics. We further describe how the application design has an intentional focus on HBCUs and emerging researchers to raise visibility of people historically rendered invisible in the current science system.
Authors: Anton Razzhigaev, Matvey Mikhalchuk, Temurbek Rahmatullaev, Elizaveta Goncharova, Polina Druzhinina, Ivan Oseledets, Andrey Kuznetsov
Abstract: We introduce methods to quantify how Large Language Models (LLMs) encode and store contextual information, revealing that tokens often seen as minor (e.g., determiners, punctuation) carry surprisingly high context. Notably, removing these tokens -- especially stopwords, articles, and commas -- consistently degrades performance on MMLU and BABILong-4k, even if removing only irrelevant tokens. Our analysis also shows a strong correlation between contextualization and linearity, where linearity measures how closely the transformation from one layer's embeddings to the next can be approximated by a single linear mapping. These findings underscore the hidden importance of filler tokens in maintaining context. For further exploration, we present LLM-Microscope, an open-source toolkit that assesses token-level nonlinearity, evaluates contextual memory, visualizes intermediate layer contributions (via an adapted Logit Lens), and measures the intrinsic dimensionality of representations. This toolkit illuminates how seemingly trivial tokens can be critical for long-range understanding.
Authors: Raymond Wilson, Chase Carter, Cole Graham
Abstract: Conversational query rewriting is crucial for effective conversational search, yet traditional supervised methods require substantial labeled data, which is scarce in low-resource settings. This paper introduces Prompt-Guided In-Context Learning, a novel approach that leverages the in-context learning capabilities of Large Language Models (LLMs) for few-shot conversational query rewriting. Our method employs carefully designed prompts, incorporating task descriptions, input/output format specifications, and a small set of illustrative examples, to guide pre-trained LLMs to generate context-independent queries without explicit fine-tuning. Extensive experiments on benchmark datasets, TREC and Taskmaster-1, demonstrate that our approach significantly outperforms strong baselines, including supervised models and contrastive co-training methods, across various evaluation metrics such as BLEU, ROUGE-L, Success Rate, and MRR. Ablation studies confirm the importance of in-context examples, and human evaluations further validate the superior fluency, relevance, and context utilization of our generated rewrites. The results highlight the potential of prompt-guided in-context learning as an efficient and effective paradigm for low-resource conversational query rewriting, reducing the reliance on extensive labeled data and complex training procedures.
Authors: Mark Russinovich, Ahmed Salem
Abstract: Recent copyright agreements between AI companies and content creators have highlighted the need for precise control over language models' ability to reproduce copyrighted content. While existing approaches rely on either complete concept removal through unlearning or simple output filtering, we propose Obliviate, a novel post-training technique that selectively prevents verbatim reproduction of specific text while preserving semantic understanding. Obliviate operates by selecting tokens within memorized sequences and modifying the model's probability distribution to prevent exact reproduction while maintaining contextual understanding. We evaluate Obliviate on multiple large language models (LLaMA-3.1 8B, LLaMA-3.1-instruct 8B, Qwen-2.5-7B, and Yi-1.5 6B) across both synthetic memorization tasks and organic copyright content. Our results demonstrate that Obliviate achieves orders of magnitude reduction, e.g., 100x, in verbatim memorization while maintaining model performance within 1% of baseline on standard benchmarks (HellaSwag, MMLU, TruthfulQA, and Winogrande). This makes Obliviate particularly suitable for practical deployment scenarios where companies need to efficiently address copyright concerns in pretrained models without compromising their general capabilities.
Authors: Wonjin Yoon, Ian Bulovic, Timothy A. Miller
Abstract: Large language models perform surprisingly well on many zero-shot classification tasks, but are difficult to fairly compare to supervised classifiers due to the lack of a modifiable decision boundary. In this work, we propose and evaluate a method that converts binary classification tasks into pairwise comparison tasks, obtaining relative rankings from LLMs. Repeated pairwise comparisons can be used to score instances using the Elo rating system (used in chess and other competitions), inducing a confidence ordering over instances in a dataset. We evaluate scheduling algorithms for their ability to minimize comparisons, and show that our proposed algorithm leads to improved classification performance, while also providing more information than traditional zero-shot classification.
Authors: Amalie Brogaard Pauli, Isabelle Augenstein, Ira Assent
Abstract: LLMs make it easy to rewrite text in any style, be it more polite, persuasive, or more positive. We present a large-scale study of evaluation metrics for style and attribute transfer with a focus on content preservation; meaning content not attributed to the style shift is preserved. The de facto evaluation approach uses lexical or semantic similarity metrics often between source sentences and rewrites. While these metrics are not designed to distinguish between style or content differences, empirical meta-evaluation shows a reasonable correlation to human judgment. In fact, recent works find that LLMs prompted as evaluators are only comparable to semantic similarity metrics, even though intuitively, the LLM approach should better fit the task. To investigate this discrepancy, we benchmark 8 metrics for evaluating content preservation on existing datasets and additionally construct a new test set that better aligns with the meta-evaluation aim. Indeed, we then find that the empirical conclusion aligns with the intuition: content preservation metrics for style/attribute transfer must be conditional on the style shift. To support this, we propose a new efficient zero-shot evaluation method using the likelihood of the next token. We hope our meta-evaluation can foster more research on evaluating content preservation metrics, and also to ensure fair evaluation of methods for conducting style transfer.
Authors: Henry Hengyuan Zhao, Wenqi Pei, Yifei Tao, Haiyang Mei, Mike Zheng Shou
Abstract: Existing benchmarks do not test Large Multimodal Models (LMMs) on their interactive intelligence with human users which is vital for developing general-purpose AI assistants. We design InterFeedback, an interactive framework, which can be applied to any LMM and dataset to assess this ability autonomously. On top of this, we introduce InterFeedback-Bench which evaluates interactive intelligence using two representative datasets, MMMU-Pro and MathVerse, to test 10 different open-source LMMs. Additionally, we present InterFeedback-Human, a newly collected dataset of 120 cases designed for manually testing interactive performance in leading models such as OpenAI-o1 and Claude-3.5-Sonnet. Our evaluation results show that even state-of-the-art LMM (like OpenAI-o1) can correct their results through human feedback less than 50%. Our findings point to the need for methods that can enhance the LMMs' capability to interpret and benefit from feedback.
Authors: Yun-Wei Chu, Kai Zhang, Christopher Malon, Martin Renqiang Min
Abstract: Multimodal Large Language Models (MLLMs) have shown impressive performance in vision and text tasks. However, hallucination remains a major challenge, especially in fields like healthcare where details are critical. In this work, we show how MLLMs may be enhanced to support Visual RAG (V-RAG), a retrieval-augmented generation framework that incorporates both text and visual data from retrieved images. On the MIMIC-CXR chest X-ray report generation and Multicare medical image caption generation datasets, we show that Visual RAG improves the accuracy of entity probing, which asks whether a medical entities is grounded by an image. We show that the improvements extend both to frequent and rare entities, the latter of which may have less positive training data. Downstream, we apply V-RAG with entity probing to correct hallucinations and generate more clinically accurate X-ray reports, obtaining a higher RadGraph-F1 score.
Authors: Elliot Schumacher, Dhruv Naik, Anitha Kannan
Abstract: Large language models (LLMs) have demonstrated impressive capabilities in disease diagnosis. However, their effectiveness in identifying rarer diseases, which are inherently more challenging to diagnose, remains an open question. Rare disease performance is critical with the increasing use of LLMs in healthcare settings. This is especially true if a primary care physician needs to make a rarer prognosis from only a patient conversation so that they can take the appropriate next step. To that end, several clinical decision support systems are designed to support providers in rare disease identification. Yet their utility is limited due to their lack of knowledge of common disorders and difficulty of use. In this paper, we propose RareScale to combine the knowledge LLMs with expert systems. We use jointly use an expert system and LLM to simulate rare disease chats. This data is used to train a rare disease candidate predictor model. Candidates from this smaller model are then used as additional inputs to black-box LLM to make the final differential diagnosis. Thus, RareScale allows for a balance between rare and common diagnoses. We present results on over 575 rare diseases, beginning with Abdominal Actinomycosis and ending with Wilson's Disease. Our approach significantly improves the baseline performance of black-box LLMs by over 17% in Top-5 accuracy. We also find that our candidate generation performance is high (e.g. 88.8% on gpt-4o generated chats).
Authors: Yeonjun In, Wonjoong Kim, Kanghoon Yoon, Sungchul Kim, Mehrab Tanjim, Kibum Kim, Chanyoung Park
Abstract: As the use of large language model (LLM) agents continues to grow, their safety vulnerabilities have become increasingly evident. Extensive benchmarks evaluate various aspects of LLM safety by defining the safety relying heavily on general standards, overlooking user-specific standards. However, safety standards for LLM may vary based on a user-specific profiles rather than being universally consistent across all users. This raises a critical research question: Do LLM agents act safely when considering user-specific safety standards? Despite its importance for safe LLM use, no benchmark datasets currently exist to evaluate the user-specific safety of LLMs. To address this gap, we introduce U-SAFEBENCH, the first benchmark designed to assess user-specific aspect of LLM safety. Our evaluation of 18 widely used LLMs reveals current LLMs fail to act safely when considering user-specific safety standards, marking a new discovery in this field. To address this vulnerability, we propose a simple remedy based on chain-of-thought, demonstrating its effectiveness in improving user-specific safety. Our benchmark and code are available at https://github.com/yeonjun-in/U-SafeBench.
Authors: Masha Fedzechkina, Eleonora Gualdoni, Sinead Williamson, Katherine Metcalf, Skyler Seto, Barry-John Theobald
Abstract: Modern large language models (LLMs) achieve impressive performance on some tasks, while exhibiting distinctly non-human-like behaviors on others. This raises the question of how well the LLM's learned representations align with human representations. In this work, we introduce a novel approach to the study of representation alignment: we adopt a method from research on activation steering to identify neurons responsible for specific concepts (e.g., 'cat') and then analyze the corresponding activation patterns. Our findings reveal that LLM representations closely align with human representations inferred from behavioral data. Notably, this alignment surpasses that of word embeddings, which have been center stage in prior work on human and model alignment. Additionally, our approach enables a more granular view of how LLMs represent concepts. Specifically, we show that LLMs organize concepts in a way that reflects hierarchical relationships interpretable to humans (e.g., 'animal'-'dog').
Authors: Dengjie Li, Tiancheng Shen, Yao Zhou, Baisong Yang, Zhongying Liu, Masheng Yang, Bernard Ghanem, Yibo Yang, Yujie Zhong, Ming-Hsuan Yang
Abstract: Large language models (LLMs) have demonstrated remarkable capabilities, yet prohibitive parameter complexity often hinders their deployment. Existing singular value decomposition (SVD) based compression methods simply deem singular values as importance scores of decomposed components. However, this importance ordered by singular values does not necessarily correlate with the performance of a downstream task. In this work, we introduce SoCo (Singular spectrum optimization for large language model Compression), a novel compression framework that learns to rescale the decomposed components of SVD in a data-driven manner. Concretely, we employ a learnable diagonal matrix to assign importance scores for singular spectrum and develop a three-stage training process that progressively refines these scores from initial coarse compression to fine-grained sparsification-thereby striking an effective balance between aggressive model compression and performance preservation. Thanks to the learnable singular spectrum, SoCo adaptively prunes components according to the sparsified importance scores, rather than relying on the fixed order of singular values. More importantly, the remaining components with amplified importance scores can compensate for the loss of the pruned ones. Experimental evaluations across multiple LLMs and benchmarks demonstrate that SoCo surpasses the state-of-the-art methods in model compression.
Authors: Marianne Chuang, Gabriel Chuang, Cheryl Chuang, John Chuang
Abstract: We study the use of large language models (LLMs) to both evaluate and greenwash corporate climate disclosures. First, we investigate the use of the LLM-as-a-Judge (LLMJ) methodology for scoring company-submitted reports on emissions reduction targets and progress. Second, we probe the behavior of an LLM when it is prompted to greenwash a response subject to accuracy and length constraints. Finally, we test the robustness of the LLMJ methodology against responses that may be greenwashed using an LLM. We find that two LLMJ scoring systems, numerical rating and pairwise comparison, are effective in distinguishing high-performing companies from others, with the pairwise comparison system showing greater robustness against LLM-greenwashed responses.
Authors: Anil Ramakrishna, Yixin Wan, Xiaomeng Jin, Kai-Wei Chang, Zhiqi Bu, Bhanukiran Vinzamuri, Volkan Cevher, Mingyi Hong, Rahul Gupta
Abstract: Unlearning aims to remove copyrighted, sensitive, or private content from large language models (LLMs) without a full retraining. In this work, we develop a multi-task unlearning benchmark (LUME) which features three tasks: (1) unlearn synthetically generated creative short novels, (2) unlearn synthetic biographies with sensitive information, and (3) unlearn a collection of public biographies. We further release two fine-tuned LLMs of 1B and 7B parameter sizes as the target models. We conduct detailed evaluations of several recently proposed unlearning algorithms and present results on carefully crafted metrics to understand their behavior and limitations.
Authors: Leena Mathur, Marian Qian, Paul Pu Liang, Louis-Philippe Morency
Abstract: Social reasoning abilities are crucial for AI systems to effectively interpret and respond to multimodal human communication and interaction within social contexts. We introduce Social Genome, the first benchmark for fine-grained, grounded social reasoning abilities of multimodal models. Social Genome contains 272 videos of interactions and 1,486 human-annotated reasoning traces related to inferences about these interactions. These traces contain 5,777 reasoning steps that reference evidence from visual cues, verbal cues, vocal cues, and external knowledge (contextual knowledge external to videos). Social Genome is also the first modeling challenge to study external knowledge in social reasoning. Social Genome computes metrics to holistically evaluate semantic and structural qualities of model-generated social reasoning traces. We demonstrate the utility of Social Genome through experiments with state-of-the-art models, identifying performance gaps and opportunities for future research to improve the grounded social reasoning abilities of multimodal models.
Authors: Yen-Che Hsiao, Abhishek Dutta
Abstract: This study investigates the in-context learning capabilities of various decoder-only transformer-based language models with different model sizes and training data, including GPT2, SmolLM2, OpenELM, TinyLlama, Stable LM, and Gemma 2. We identify a critical parameter threshold (~1.6 billion), beyond which reasoning performance improves significantly in tasks such as commonsense reasoning in multiple-choice question answering and deductive reasoning. Specifically, models above this threshold achieve better success rates in chain-of-thought (CoT) prompting for deductive reasoning tasks, especially those requiring longer reasoning chains, such as proof by contradiction and disjunction elimination. To address limitations in sub-threshold models, we demonstrate that fine-tuning with task-specific exemplars substantially enhances reasoning performance, enabling accurate CoT generation even without additional exemplars in the prompt for tasks with shorter reasoning chains. Finally, our analysis of attention maps reveals that models capable of generating correct CoTs exhibit higher token-level attention scores on subsequent correct tokens and the correct parts of speech, providing interpretability insights into reasoning processes. These findings collectively advance understanding of reasoning capabilities in decoder-only transformer-based models. The code is available at: https://github.com/AnnonymousForPapers/CoT_Reasoning_Test.
URLs: https://github.com/AnnonymousForPapers/CoT_Reasoning_Test.
Authors: Vignesh Kothapalli, Hamed Firooz, Maziar Sanjabi
Abstract: We introduce CoT-ICL Lab, a framework and methodology to generate synthetic tokenized datasets and systematically study chain-of-thought (CoT) in-context learning (ICL) in language models. CoT-ICL Lab allows fine grained control over the complexity of in-context examples by decoupling (1) the causal structure involved in chain token generation from (2) the underlying token processing functions. We train decoder-only transformers (up to 700M parameters) on these datasets and show that CoT accelerates the accuracy transition to higher values across model sizes. In particular, we find that model depth is crucial for leveraging CoT with limited in-context examples, while more examples help shallow models match deeper model performance. Additionally, limiting the diversity of token processing functions throughout training improves causal structure learning via ICL. We also interpret these transitions by analyzing transformer embeddings and attention maps. Overall, CoT-ICL Lab serves as a simple yet powerful testbed for theoretical and empirical insights into ICL and CoT in language models.
Authors: Juntae Lee, Jihwan Bang, Seunghan Yang, Kyuhong Shim, Simyung Chang
Abstract: Retrieval-augmented generation (RAG) with large language models (LLMs) is especially valuable in specialized domains, where precision is critical. To more specialize the LLMs into a target domain, domain-specific RAG has recently been developed by allowing the LLM to access the target domain early via finetuning. The domain-specific RAG makes more sense in resource-constrained environments like edge devices, as they should perform a specific task (e.g. personalization) reliably using only small-scale LLMs. While the domain-specific RAG is well-aligned with edge devices in this respect, it often relies on widely-used reasoning techniques like chain-of-thought (CoT). The reasoning step is useful to understand the given external knowledge, and yet it is computationally expensive and difficult for small-scale LLMs to learn it. Tackling this, we propose the Chain of Rank (CoR) which shifts the focus from intricate lengthy reasoning to simple ranking of the reliability of input external documents. Then, CoR reduces computational complexity while maintaining high accuracy, making it particularly suited for resource-constrained environments. We attain the state-of-the-art (SOTA) results in benchmarks, and analyze its efficacy.
Authors: Naiming Liu, Shashank Sonkar, Richard G. Baraniuk
Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities in various educational tasks, yet their alignment with human learning patterns, particularly in predicting which incorrect options students are most likely to select in multiple-choice questions (MCQs), remains underexplored. Our work investigates the relationship between LLM generation likelihood and student response distributions in MCQs with a specific focus on distractor selections. We collect a comprehensive dataset of MCQs with real-world student response distributions to explore two fundamental research questions: (1). RQ1 - Do the distractors that students more frequently select correspond to those that LLMs assign higher generation likelihood to? (2). RQ2 - When an LLM selects a incorrect choice, does it choose the same distractor that most students pick? Our experiments reveals moderate correlations between LLM-assigned probabilities and student selection patterns for distractors in MCQs. Additionally, when LLMs make mistakes, they are more likley to select the same incorrect answers that commonly mislead students, which is a pattern consistent across both small and large language models. Our work provides empirical evidence that despite LLMs' strong performance on generating educational content, there remains a gap between LLM's underlying reasoning process and human cognitive processes in identifying confusing distractors. Our findings also have significant implications for educational assessment development. The smaller language models could be efficiently utilized for automated distractor generation as they demonstrate similar patterns in identifying confusing answer choices as larger language models. This observed alignment between LLMs and student misconception patterns opens new opportunities for generating high-quality distractors that complement traditional human-designed distractors.
Authors: Zhouhang Xie, Tushar Khot, Bhavana Dalvi Mishra, Harshit Surana, Julian McAuley, Peter Clark, Bodhisattwa Prasad Majumder
Abstract: Instruction-following LLMs have recently allowed systems to discover hidden concepts from a collection of unstructured documents based on a natural language description of the purpose of the discovery (i.e., goal). Still, the quality of the discovered concepts remains mixed, as it depends heavily on LLM's reasoning ability and drops when the data is noisy or beyond LLM's knowledge. We present Instruct-LF, a goal-oriented latent factor discovery system that integrates LLM's instruction-following ability with statistical models to handle large, noisy datasets where LLM reasoning alone falls short. Instruct-LF uses LLMs to propose fine-grained, goal-related properties from documents, estimates their presence across the dataset, and applies gradient-based optimization to uncover hidden factors, where each factor is represented by a cluster of co-occurring properties. We evaluate latent factors produced by Instruct-LF on movie recommendation, text-world navigation, and legal document categorization tasks. These interpretable representations improve downstream task performance by 5-52% than the best baselines and were preferred 1.8 times as often as the best alternative, on average, in human evaluation.
Authors: Tianjie Ju, Bowen Wang, Hao Fei, Mong-Li Lee, Wynne Hsu, Yun Li, Qianren Wang, Pengzhou Cheng, Zongru Wu, Zhuosheng Zhang, Gongshen Liu
Abstract: Recent advances in Large Language Models (LLMs) have upgraded them from sophisticated text generators to autonomous agents capable of corporation and tool use in multi-agent systems (MASs). However, the robustness of these LLM-based MASs, especially under knowledge conflicts, remains unclear. In this paper, we design four comprehensive metrics to investigate the robustness of MASs when facing mild or task-critical knowledge conflicts. We first analyze mild knowledge conflicts introduced by heterogeneous agents and find that they do not harm system robustness but instead improve collaborative decision-making. Next, we investigate task-critical knowledge conflicts by synthesizing knowledge conflicts and embedding them into one of the agents. Our results show that these conflicts have surprisingly little to no impact on MAS robustness. Furthermore, we observe that MASs demonstrate certain self-repairing capabilities by reducing their reliance on knowledge conflicts and adopting alternative solution paths to maintain stability. Finally, we conduct ablation studies on the knowledge conflict number, agent number, and interaction rounds, finding that the self-repairing capability of MASs has intrinsic limits, and all findings hold consistently across various factors. Our code is publicly available at https://github.com/wbw625/MultiAgentRobustness.
Authors: Sarthak Mahajan, Nimmi Rangaswamy
Abstract: In recent years, widespread internet adoption and the growth in userbase of various social media platforms have led to an increase in the proliferation of extreme speech online. While traditional language models have demonstrated proficiency in distinguishing between neutral text and non-neutral text (i.e. extreme speech), categorizing the diverse types of extreme speech presents significant challenges. The task of extreme speech classification is particularly nuanced, as it requires a deep understanding of socio-cultural contexts to accurately interpret the intent of the language used by the speaker. Even human annotators often disagree on the appropriate classification of such content, emphasizing the complex and subjective nature of this task. The use of human moderators also presents a scaling issue, necessitating the need for automated systems for extreme speech classification. The recent launch of ChatGPT has drawn global attention to the potential applications of Large Language Models (LLMs) across a diverse variety of tasks. Trained on vast and diverse corpora, and demonstrating the ability to effectively capture and encode contextual information, LLMs emerge as highly promising tools for tackling this specific task of extreme speech classification. In this paper, we leverage the Indian subset of the extreme speech dataset from Maronikolakis et al. (2022) to develop an effective classification framework using LLMs. We evaluate open-source Llama models against closed-source OpenAI models, finding that while pre-trained LLMs show moderate efficacy, fine-tuning with domain-specific data significantly enhances performance, highlighting their adaptability to linguistic and contextual nuances. Although GPT-based models outperform Llama models in zero-shot settings, the performance gap disappears after fine-tuning.
Authors: Justin Qiu, Jiacheng Zhu, Ajay Patel, Marianna Apidianaki, Chris Callison-Burch
Abstract: Style embeddings are useful for stylistic analysis and style transfer; however, only English style embeddings have been made available. We introduce Multilingual StyleDistance (mStyleDistance), a multilingual style embedding model trained using synthetic data and contrastive learning. We train the model on data from nine languages and create a multilingual STEL-or-Content benchmark (Wegmann et al., 2022) that serves to assess the embeddings' quality. We also employ our embeddings in an authorship verification task involving different languages. Our results show that mStyleDistance embeddings outperform existing models on these multilingual style benchmarks and generalize well to unseen features and languages. We make our model publicly available at https://huggingface.co/StyleDistance/mstyledistance .
Authors: Jianglin Lu, Yixuan Liu, Yitian Zhang, Yun Fu
Abstract: Graph-language models (GLMs) have demonstrated great potential in graph-based semi-supervised learning. A typical GLM consists of two key stages: graph generation and text embedding, which are usually implemented by inferring a latent graph and finetuning a language model (LM), respectively. However, the former often relies on artificial assumptions about the underlying edge distribution, while the latter requires extensive data annotations. To tackle these challenges, this paper introduces a novel GLM that integrates graph generation and text embedding within a unified framework. Specifically, for graph generation, we leverage an inherent characteristic of real edge distribution--the scale-free property--as a structural prior. We unexpectedly find that this natural property can be effectively approximated by a simple k-nearest neighbor (KNN) graph. For text embedding, we develop a graph-based pseudo-labeler that utilizes scale-free graphs to provide complementary supervision for improved LM finetuning. Extensive experiments on representative datasets validate our findings on the scale-free structural approximation of KNN graphs and demonstrate the effectiveness of integrating graph generation and text embedding with a real structural prior. Our code is available at https://github.com/Jianglin954/SFGL.
Authors: Zhaoxuan Wu, Zijian Zhou, Arun Verma, Alok Prakash, Daniela Rus, Bryan Kian Hsiang Low
Abstract: We propose TETRIS, a novel method that optimizes the total throughput of batch speculative decoding in multi-request settings. Unlike existing methods that optimize for a single request or a group of requests as a whole, TETRIS actively selects the most promising draft tokens (for every request in a batch) to be accepted when verified in parallel, resulting in fewer rejected tokens and hence less wasted computing resources. Such an effective resource utilization to achieve fast inference in large language models (LLMs) is especially important to service providers with limited inference capacity. Compared to baseline speculative decoding, TETRIS yields a consistently higher acceptance rate and more effective utilization of the limited inference capacity. We show theoretically and empirically that TETRIS outperforms baseline speculative decoding and existing methods that dynamically select draft tokens, leading to a more efficient batch inference in LLMs.
Authors: Zhilin Wang, Yafu Li, Jianhao Yan, Yu Cheng, Yue Zhang
Abstract: Dynamical systems theory provides a framework for analyzing iterative processes and evolution over time. Within such systems, repetitive transformations can lead to stable configurations, known as attractors, including fixed points and limit cycles. Applying this perspective to large language models (LLMs), which iteratively map input text to output text, provides a principled approach to characterizing long-term behaviors. Successive paraphrasing serves as a compelling testbed for exploring such dynamics, as paraphrases re-express the same underlying meaning with linguistic variation. Although LLMs are expected to explore a diverse set of paraphrases in the text space, our study reveals that successive paraphrasing converges to stable periodic states, such as 2-period attractor cycles, limiting linguistic diversity. This phenomenon is attributed to the self-reinforcing nature of LLMs, as they iteratively favour and amplify certain textual forms over others. This pattern persists with increasing generation randomness or alternating prompts and LLMs. These findings underscore inherent constraints in LLM generative capability, while offering a novel dynamical systems perspective for studying their expressive potential.
Authors: Jinchuan Tian, Jiatong Shi, William Chen, Siddhant Arora, Yoshiki Masuyama, Takashi Maekaku, Yihan Wu, Junyi Peng, Shikhar Bharadwaj, Yiwen Zhao, Samuele Cornell, Yifan Peng, Xiang Yue, Chao-Han Huck Yang, Graham Neubig, Shinji Watanabe
Abstract: We present ESPnet-SpeechLM, an open toolkit designed to democratize the development of speech language models (SpeechLMs) and voice-driven agentic applications. The toolkit standardizes speech processing tasks by framing them as universal sequential modeling problems, encompassing a cohesive workflow of data preprocessing, pre-training, inference, and task evaluation. With ESPnet-SpeechLM, users can easily define task templates and configure key settings, enabling seamless and streamlined SpeechLM development. The toolkit ensures flexibility, efficiency, and scalability by offering highly configurable modules for every stage of the workflow. To illustrate its capabilities, we provide multiple use cases demonstrating how competitive SpeechLMs can be constructed with ESPnet-SpeechLM, including a 1.7B-parameter model pre-trained on both text and speech tasks, across diverse benchmarks. The toolkit and its recipes are fully transparent and reproducible at: https://github.com/espnet/espnet/tree/speechlm.
Authors: Mengqiao Liu, Tevin Wang, Cassandra A. Cohen, Sarah Li, Chenyan Xiong
Abstract: Which large language model (LLM) is better? Every evaluation tells a story, but what do users really think about current LLMs? This paper presents CLUE, an LLM-powered interviewer that conducts in-the-moment user experience interviews, right after users interacted with LLMs, and automatically gathers insights about user opinions from massive interview logs. We conduct a study with thousands of users to understand user opinions on mainstream LLMs, recruiting users to first chat with a target LLM and then interviewed by CLUE. Our experiments demonstrate that CLUE captures interesting user opinions, for example, the bipolar views on the displayed reasoning process of DeepSeek-R1 and demands for information freshness and multi-modality. Our collected chat-and-interview logs will be released.
Authors: Renjie Wei, Songqiang Xu, Linfeng Zhong, Zebin Yang, Qingyu Guo, Yuan Wang, Runsheng Wang, Meng Li
Abstract: State space models (SSMs) like Mamba have recently attracted much attention. Compared to Transformer-based large language models (LLMs), Mamba achieves linear computation complexity with the sequence length and demonstrates superior performance. However, Mamba is hard to accelerate due to the scattered activation outliers and the complex computation dependency, rendering existing LLM accelerators inefficient. In this paper, we propose LightMamba that co-designs the quantization algorithm and FPGA accelerator architecture for efficient Mamba inference. We first propose an FPGA-friendly post-training quantization algorithm that features rotation-assisted quantization and power-of-two SSM quantization to reduce the majority of computation to 4-bit. We further design an FPGA accelerator that partially unrolls the Mamba computation to balance the efficiency and hardware costs. Through computation reordering as well as fine-grained tiling and fusion, the hardware utilization and memory efficiency of the accelerator get drastically improved. We implement LightMamba on Xilinx Versal VCK190 FPGA and achieve 4.65x to 6.06x higher energy efficiency over the GPU baseline. When evaluated on Alveo U280 FPGA, LightMamba reaches 93 tokens/s, which is 1.43x that of the GPU baseline.
Authors: Jingheng Ye, Shang Qin, Yinghui Li, Hai-Tao Zheng, Shen Wang, Qingsong Wen
Abstract: Grammatical Error Correction (GEC) faces a critical challenge concerning explainability, notably when GEC systems are designed for language learners. Existing research predominantly focuses on explaining grammatical errors extracted in advance, thus neglecting the relationship between explanations and corrections. To address this gap, we introduce EXGEC, a unified explainable GEC framework that integrates explanation and correction tasks in a generative manner, advocating that these tasks mutually reinforce each other. Experiments have been conducted on EXPECT, a recent human-labeled dataset for explainable GEC, comprising around 20k samples. Moreover, we detect significant noise within EXPECT, potentially compromising model training and evaluation. Therefore, we introduce an alternative dataset named EXPECT-denoised, ensuring a more objective framework for training and evaluation. Results on various NLP models (BART, T5, and Llama3) show that EXGEC models surpass single-task baselines in both tasks, demonstrating the effectiveness of our approach.
Authors: Peng Shen, Xugang Lu, Hisashi Kawai
Abstract: Speech recognition systems often face challenges due to domain mismatch, particularly in real-world applications where domain-specific data is unavailable because of data accessibility and confidentiality constraints. Inspired by Retrieval-Augmented Generation (RAG) techniques for large language models (LLMs), this paper introduces a LLM-based retrieval-augmented speech recognition method that incorporates domain-specific textual data at the inference stage to enhance recognition performance. Rather than relying on domain-specific textual data during the training phase, our model is trained to learn how to utilize textual information provided in prompts for LLM decoder to improve speech recognition performance. Benefiting from the advantages of the RAG retrieval mechanism, our approach efficiently accesses locally available domain-specific documents, ensuring a convenient and effective process for solving domain mismatch problems. Experiments conducted on the CSJ database demonstrate that the proposed method significantly improves speech recognition accuracy and achieves state-of-the-art results on the CSJ dataset, even without relying on the full training data.
Authors: Houquan Zhou, Bo Zhang, Zhenghua Li, Ming Yan, Min Zhang
Abstract: Chinese spelling correction (CSC) is a crucial task that aims to correct character errors in Chinese text. While conventional CSC focuses on character substitution errors caused by mistyping, two other common types of character errors, missing and redundant characters, have received less attention. These errors are often excluded from CSC datasets during the annotation process or ignored during evaluation, even when they have been annotated. This issue limits the practicality of the CSC task. To address this issue, we introduce the task of General Chinese Character Error Correction (C2EC), which focuses on all three types of character errors. We construct a high-quality C2EC benchmark by combining and manually verifying data from CCTC and Lemon datasets. We extend the training-free prompt-free CSC method to C2EC by using Levenshtein distance for handling length changes and leveraging an additional prompt-based large language model (LLM) to improve performance. Experiments show that our method enables a 14B-parameter LLM to be on par with models nearly 50 times larger on both conventional CSC and C2EC tasks, without any fine-tuning.
Authors: Ryoma Kumon, Hitomi Yanaka
Abstract: The compositional generalization abilities of neural models have been sought after for human-like linguistic competence. The popular method to evaluate such abilities is to assess the models' input-output behavior. However, that does not reveal the internal mechanisms, and the underlying competence of such models in compositional generalization remains unclear. To address this problem, we explore the inner workings of a Transformer model by finding an existing subnetwork that contributes to the generalization performance and by performing causal analyses on how the model utilizes syntactic features. We find that the model depends on syntactic features to output the correct answer, but that the subnetwork with much better generalization performance than the whole model relies on a non-compositional algorithm in addition to the syntactic features. We also show that the subnetwork improves its generalization performance relatively slowly during the training compared to the in-distribution one, and the non-compositional solution is acquired in the early stages of the training.
Authors: Yaohua Tang, Zhicheng Hu, Kun Cheng, Fan Mo, Qiheng Lv, Hua Wang, Zhi Chen
Abstract: The increasing context window size in large language models (LLMs) has improved their ability to handle complex, long-text tasks. However, as the conversation rounds continue, it is required to store a large amount of KV cache in GPU memory, which significantly affects the efficiency and even availability of the model serving systems. This paper analyzes dialogue data from real users and discovers that the LLM inference manifests a watershed layer, after which the distribution of round-level attention shows notable similarity. We propose Round Attention, a novel round-level attention mechanism that only recalls and computes the KV cache of the most relevant rounds. The experiments show that our method saves 55\% memory usage without compromising model performance.
Authors: Puneet Prashar, Krishna Mohan Shukla, Adam Jatowt
Abstract: The ability to automatically identify whether an entity is referenced in a future context can have multiple applications including decision making, planning and trend forecasting. This paper focuses on detecting implicit future references in entity-centric texts, addressing the growing need for automated temporal analysis in information processing. We first present a novel dataset of 19,540 sentences built around popular entities sourced from Wikipedia, which consists of future-related and non-future-related contexts in which those entities appear. As a second contribution, we evaluate the performance of several Language Models including also Large Language Models (LLMs) on the task of distinguishing future-oriented content in the absence of explicit temporal references.
Authors: Siyuan Wang, Enda Zhao, Zhongyu Wei, Xiang Ren
Abstract: Advances in Large Language Models (LLMs) have significantly improved multi-step reasoning through generating free-text rationales. However, recent studies show that LLMs tend to lose focus over the middle of long contexts. This raises concerns that as reasoning progresses, LLMs may overlook information in earlier steps when decoding subsequent steps, leading to generate unreliable and redundant rationales. To address this, we propose guiding LLMs to generate more accurate and concise step-by-step rationales by (1) proactively referencing information from underutilized prior steps, and (2) minimizing redundant information between new and existing steps. We introduce stepwise informativeness search, an inference-time tree search framework incorporating two selection heuristics: grounding-guided selection which prioritizes steps paying higher attention over underutilized steps; and novelty-guided selection which encourages steps with novel conclusions. During rationale generation, we use a self-grounding strategy that prompts LLMs to explicitly reference relevant prior steps to provide premises before deduction at each step. Experimental results on four reasoning datasets demonstrate that our approach improves reasoning accuracy by generating higher-quality rationales with reduced errors and redundancy.
Authors: Anna Wegmann, Dong Nguyen, David Jurgens
Abstract: Variation in language is ubiquitous and often systematically linked to regional, social, and contextual factors. Tokenizers split texts into smaller units and might behave differently for less common linguistic forms. This might affect downstream LLM performance differently on two types of tasks: Tasks where the model should be robust to language variation (e.g., for semantic tasks like NLI, labels do not depend on whether a text uses British or American spelling) and tasks where the model should be sensitive to language variation (e.g., for form-based tasks like authorship verification, labels depend on whether a text uses British or American spelling). We pre-train BERT base models for the popular Byte-Pair Encoding algorithm to investigate how key algorithmic design choices impact downstream models' performances: fitting corpus, pre-tokenizer and vocabulary size. We find that the best tokenizer varies on the two task types -- with the pre-tokenizer having the biggest impact on performance. Further, we introduce a new approach to estimate tokenizer impact on downstream LLM performance, showing significant improvement over techniques like R\'enyi efficiency. We encourage more work on language variation and its relation to tokenizers and thus LLM performance.
Authors: Yi Zhang, Fan Wei, Jingyi Li, Yan Wang, Yanyan Yu, Jianli Chen, Zipo Cai, Xinyu Liu, Wei Wang, Peng Wang, Zhong Wang
Abstract: The use of children's drawings to examining their conceptual understanding has been proven to be an effective method, but there are two major problems with previous research: 1. The content of the drawings heavily relies on the task, and the ecological validity of the conclusions is low; 2. The interpretation of drawings relies too much on the subjective feelings of the researchers. To address this issue, this study uses the Large Language Model (LLM) to identify 1420 children's scientific drawings (covering 9 scientific themes/concepts), and uses the word2vec algorithm to calculate their semantic similarity. The study explores whether there are consistent drawing representations for children on the same theme, and attempts to establish a norm for children's scientific drawings, providing a baseline reference for follow-up children's drawing research. The results show that the representation of most drawings has consistency, manifested as most semantic similarity greater than 0.8. At the same time, it was found that the consistency of the representation is independent of the accuracy (of LLM's recognition), indicating the existence of consistency bias. In the subsequent exploration of influencing factors, we used Kendall rank correlation coefficient to investigate the effects of Sample Size, Abstract Degree, and Focus Points on drawings, and used word frequency statistics to explore whether children represented abstract themes/concepts by reproducing what was taught in class.
Authors: Feiyang Chen, Yu Cheng, Lei Wang, Yuqing Xia, Ziming Miao, Lingxiao Ma, Fan Yang, Jilong Xue, Zhi Yang, Mao Yang, Haibo Chen
Abstract: Transformers and large language models (LLMs) have revolutionized machine learning, with attention mechanisms at the core of their success. As the landscape of attention variants expands, so too do the challenges of optimizing their performance, particularly across different hardware platforms. Current optimization strategies are often narrowly focused, requiring extensive manual intervention to accommodate changes in model configurations or hardware environments. In this paper, we introduce AttentionEngine, a comprehensive framework designed to streamline the optimization of attention mechanisms across heterogeneous hardware backends. By decomposing attention computation into modular operations with customizable components, AttentionEngine enables flexible adaptation to diverse algorithmic requirements. The framework further automates kernel optimization through a combination of programmable templates and a robust cross-platform scheduling strategy. Empirical results reveal performance gains of up to 10x on configurations beyond the reach of existing methods. AttentionEngine offers a scalable, efficient foundation for developing and deploying attention mechanisms with minimal manual tuning. Our code has been open-sourced and is available at https://github.com/microsoft/AttentionEngine.
Authors: Xuyang Wu, Jinming Nian, Zhiqiang Tao, Yi Fang
Abstract: In the recent development of AI reasoning, large language models (LLMs) are trained to automatically generate chain-of-thought reasoning steps, which have demonstrated compelling performance on math and coding tasks. However, when bias is mixed within the reasoning process to form strong logical arguments, it could cause even more harmful results and further induce hallucinations. In this paper, we have evaluated the 8B and 32B variants of DeepSeek-R1 against their instruction tuned counterparts on the BBQ dataset, and investigated the bias that is elicited out and being amplified through reasoning steps. To the best of our knowledge, this empirical study is the first to assess bias issues in LLM reasoning.
Authors: Xuetao Ma, Wenbin Jiang, Hua Huang
Abstract: In-context learning (ICL) can significantly enhance the complex reasoning capabilities of large language models (LLMs), with the key lying in the selection and ordering of demonstration examples. Previous methods typically relied on simple features to measure the relevance between examples. We argue that these features are not sufficient to reflect the intrinsic connections between examples. In this study, we propose a curriculum ICL strategy guided by problem-solving logic. We select demonstration examples by analyzing the problem-solving logic and order them based on curriculum learning. Specifically, we constructed a problem-solving logic instruction set based on the BREAK dataset and fine-tuned a language model to analyze the problem-solving logic of examples. Subsequently, we selected appropriate demonstration examples based on problem-solving logic and assessed their difficulty according to the number of problem-solving steps. In accordance with the principles of curriculum learning, we ordered the examples from easy to hard to serve as contextual prompts. Experimental results on multiple benchmarks indicate that our method outperforms previous ICL approaches in terms of performance and efficiency, effectively enhancing the complex reasoning capabilities of LLMs. Our project will be publicly available subsequently.
Authors: Rasmus Aavang, Giovanni Rizzi, Rasmus B{\o}ggild, Alexandre Iolov, Mike Zhang, Johannes Bjerva
Abstract: The U.S. Securities and Exchange Commission (SEC) requires that public companies file financial reports tagging numbers with the machine readable inline eXtensible Business Reporting Language (iXBRL) standard. However, the highly complex and highly granular taxonomy defined by iXBRL limits label transferability across domains. In this paper, we introduce the Hierarchical Financial Key Performance Indicator (HiFi-KPI) dataset, designed to facilitate numerical KPI extraction at specified levels of granularity from unstructured financial text. Our approach organizes a 218,126-label hierarchy using a taxonomy based grouping method, investigating which taxonomy layer provides the most meaningful structure. HiFi-KPI comprises ~1.8M paragraphs and ~5M entities, each linked to a label in the iXBRL-specific calculation and presentation taxonomies. We provide baselines using encoder-based approaches and structured extraction using Large Language Models (LLMs). To simplify LLM inference and evaluation, we additionally release HiFi-KPI Lite, a manually curated subset with four expert-mapped labels. We publicly release all artifacts
Authors: Yunqing Xu, Xinbei Ma, Jiyang Qiu, Hai Zhao
Abstract: Generating presentation slides is a time-consuming task that urgently requires automation. Due to their limited flexibility and lack of automated refinement mechanisms, existing autonomous LLM-based agents face constraints in real-world applicability. We decompose the task of generating missing presentation slides into two key components: content generation and layout generation, aligning with the typical process of creating academic slides. First, we introduce a content generation approach that enhances coherence and relevance by incorporating context from surrounding slides and leveraging section retrieval strategies. For layout generation, we propose a textual-to-visual self-verification process using a LLM-based Reviewer + Refiner workflow, transforming complex textual layouts into intuitive visual formats. This modality transformation simplifies the task, enabling accurate and human-like review and refinement. Experiments show that our approach significantly outperforms baseline methods in terms of alignment, logical flow, visual appeal, and readability.
Authors: Suraj Racha, Prashant Joshi, Anshika Raman, Nikita Jangid, Mridul Sharma, Ganesh Ramakrishnan, Nirmal Punjabi
Abstract: Mental health remains a challenging problem all over the world, with issues like depression, anxiety becoming increasingly common. Large Language Models (LLMs) have seen a vast application in healthcare, specifically in answering medical questions. However, there is a lack of standard benchmarking datasets for question answering (QA) in mental health. Our work presents a novel multiple choice dataset, MHQA (Mental Health Question Answering), for benchmarking Language models (LMs). Previous mental health datasets have focused primarily on text classification into specific labels or disorders. MHQA, on the other hand, presents question-answering for mental health focused on four key domains: anxiety, depression, trauma, and obsessive/compulsive issues, with diverse question types, namely, factoid, diagnostic, prognostic, and preventive. We use PubMed abstracts as the primary source for QA. We develop a rigorous pipeline for LLM-based identification of information from abstracts based on various selection criteria and converting it into QA pairs. Further, valid QA pairs are extracted based on post-hoc validation criteria. Overall, our MHQA dataset consists of 2,475 expert-verified gold standard instances called MHQA-gold and ~56.1k pairs pseudo labeled using external medical references. We report F1 scores on different LLMs along with few-shot and supervised fine-tuning experiments, further discussing the insights for the scores.
Authors: Yi-Ling Chung, Aurora Cobo, Pablo Serna
Abstract: Robust automatic fact-checking systems have the potential to combat online misinformation at scale. However, most existing research primarily focuses on English. In this paper, we introduce MultiSynFact, the first large-scale multilingual fact-checking dataset containing 2.2M claim-source pairs designed to support Spanish, German, English, and other low-resource languages. Our dataset generation pipeline leverages Large Language Models (LLMs), integrating external knowledge from Wikipedia and incorporating rigorous claim validation steps to ensure data quality. We evaluate the effectiveness of MultiSynFact across multiple models and experimental settings. Additionally, we open-source a user-friendly framework to facilitate further research in multilingual fact-checking and dataset generation.
Authors: Sanghee Park, Geewook Kim
Abstract: This paper presents the Korean National Educational Test Benchmark (KoNET), a new benchmark designed to evaluate Multimodal Generative AI Systems using Korean national educational tests. KoNET comprises four exams: the Korean Elementary General Educational Development Test (KoEGED), Middle (KoMGED), High (KoHGED), and College Scholastic Ability Test (KoCSAT). These exams are renowned for their rigorous standards and diverse questions, facilitating a comprehensive analysis of AI performance across different educational levels. By focusing on Korean, KoNET provides insights into model performance in less-explored languages. We assess a range of models - open-source, open-access, and closed APIs - by examining difficulties, subject diversity, and human error rates. The code and dataset builder will be made fully open-sourced at https://github.com/naver-ai/KoNET.
Authors: Lihu Chen, Shuojie Fu, Gabriel Freedman, Cemre Zor, Guy Martin, James Kinross, Uddhav Vaghela, Ovidiu Serban, Francesca Toni
Abstract: A significant and growing number of published scientific articles is found to involve fraudulent practices, posing a serious threat to the credibility and safety of research in fields such as medicine. We propose Pub-Guard-LLM, the first large language model-based system tailored to fraud detection of biomedical scientific articles. We provide three application modes for deploying Pub-Guard-LLM: vanilla reasoning, retrieval-augmented generation, and multi-agent debate. Each mode allows for textual explanations of predictions. To assess the performance of our system, we introduce an open-source benchmark, PubMed Retraction, comprising over 11K real-world biomedical articles, including metadata and retraction labels. We show that, across all modes, Pub-Guard-LLM consistently surpasses the performance of various baselines and provides more reliable explanations, namely explanations which are deemed more relevant and coherent than those generated by the baselines when evaluated by multiple assessment methods. By enhancing both detection performance and explainability in scientific fraud detection, Pub-Guard-LLM contributes to safeguarding research integrity with a novel, effective, open-source tool.
Authors: Yue Zhou, Yi Chang, Yuan Wu
Abstract: Model merging integrates the parameters of multiple models into a unified model, combining their diverse capabilities. Existing model merging methods are often constrained by fixed parameter merging ratios. In this study, we propose Mixup Model Merge (M$^3$), an innovative approach inspired by the Mixup data augmentation technique. This method merges the parameters of two large language models (LLMs) by randomly generating linear interpolation ratios, allowing for a more flexible and comprehensive exploration of the parameter space. Extensive experiments demonstrate the superiority of our proposed M$^3$ method in merging fine-tuned LLMs: (1) it significantly improves performance across multiple tasks, (2) it enhances LLMs' out-of-distribution (OOD) robustness and adversarial robustness, (3) it achieves superior results when combined with sparsification techniques such as DARE, and (4) it offers a simple yet efficient solution that does not require additional computational resources. In conclusion, M$^3$ is a simple yet effective model merging method that significantly enhances the performance of the merged model by randomly generating contribution ratios for two fine-tuned LLMs. The code is available at https://github.com/MLGroupJLU/MixupModelMerge.
Authors: Weilan Wang, Yu Mao, Dongdong Tang, Hongchao Du, Nan Guan, Chun Jason Xue
Abstract: Large language models (LLMs) exhibit excellent performance in various tasks. However, the memory requirements of LLMs present a great challenge when deploying on memory-limited devices, even for quantized LLMs. This paper introduces a framework to compress LLM after quantization further, achieving about 2.2x compression ratio. A compression-aware quantization is first proposed to enhance model weight compressibility by re-scaling the model parameters before quantization, followed by a pruning method to improve further. Upon this, we notice that decompression can be a bottleneck during practical scenarios. We then give a detailed analysis of the trade-off between memory usage and latency brought by the proposed method. A speed-adaptive method is proposed to overcome it. The experimental results show inference with the compressed model can achieve a 40% reduction in memory size with negligible loss in accuracy and inference speed.
Authors: Xinghan Pan
Abstract: This paper presents an enhanced RWKV-based language generation model designed to improve long-sequence text processing. We propose an adaptive token shift and gating mechanism to better capture long-range dependencies in text generation. Through a series of experiments, we compare the baseline RWKV model with the enhanced model, evaluating performance in terms of forward propagation time, text generation quality, and automatic evaluation metrics such as perplexity, BLEU, and ROUGE. Experimental results show that the enhanced model significantly improves generation quality, especially in BLEU and ROUGE scores, and demonstrates stronger context-capturing ability in long-text generation tasks.
Authors: Martina Miliani, Serenna Auriemma, Alessandro Bondielli, Emmanuele Chersoni, Lucia Passaro, Irene Sucameli, Alessandro Lenci
Abstract: Large Language Models (LLMs) are increasingly used in tasks requiring interpretive and inferential accuracy. In this paper, we introduce ExpliCa, a new dataset for evaluating LLMs in explicit causal reasoning. ExpliCa uniquely integrates both causal and temporal relations presented in different linguistic orders and explicitly expressed by linguistic connectives. The dataset is enriched with crowdsourced human acceptability ratings. We tested LLMs on ExpliCa through prompting and perplexity-based metrics. We assessed seven commercial and open-source LLMs, revealing that even top models struggle to reach 0.80 accuracy. Interestingly, models tend to confound temporal relations with causal ones, and their performance is also strongly influenced by the linguistic order of the events. Finally, perplexity-based scores and prompting performance are differently affected by model size.
Authors: Ya Wang, Zhijian Zhuo, Yutao Zeng, Xun Zhou, Jian Yang, Xiaoqing Li
Abstract: Training stability is a persistent challenge in the pre-training of large language models (LLMs), particularly for architectures such as Post-Norm Transformers, which are prone to gradient explosion and dissipation. In this paper, we propose Scale-Distribution Decoupling (SDD), a novel approach that stabilizes training by explicitly decoupling the scale and distribution of the weight matrix in fully-connected layers. SDD applies a normalization mechanism to regulate activations and a learnable scaling vector to maintain well-conditioned gradients, effectively preventing $\textbf{gradient explosion and dissipation}$. This separation improves optimization efficiency, particularly in deep networks, by ensuring stable gradient propagation. Experimental results demonstrate that our method stabilizes training across various LLM architectures and outperforms existing techniques in different normalization configurations. Furthermore, the proposed method is lightweight and compatible with existing frameworks, making it a practical solution for stabilizing LLM training. Code is available at https://github.com/kaihemo/SDD.
Authors: Wenyuan Zhang, Tianyun Liu, Mengxiao Song, Xiaodong Li, Tingwen Liu
Abstract: Despite the abundance of prior social strategies possessed by humans, there remains a paucity of research dedicated to their transfer and integration into social agents. Our proposed SOTOPIA-{\Omega} framework aims to address and bridge this gap, with a particular focus on enhancing the social capabilities of language agents. This framework dynamically injects multi-step reasoning strategies inspired by negotiation theory, along with two simple direct strategies, into expert agents, thereby automating the construction of high-quality social dialogue training corpus. Additionally, we introduce the concept of Social Instruction Following (S-IF) and propose two new S-IF evaluation metrics that are complementary to social capability. We demonstrate that several 7B models trained on high-quality corpus not only significantly surpass the expert agent (GPT-4) in achieving social goals but also enhance S-IF performance. Analysis and variant experiments validate the advantages of dynamic construction, which can especially break the agent's prolonged deadlock.
Authors: Pengcheng Huang, Zhenghao Liu, Yukun Yan, Xiaoyuan Yi, Hao Chen, Zhiyuan Liu, Maosong Sun, Tong Xiao, Ge Yu, Chenyan Xiong
Abstract: Knowledge-Augmented Generation (KAG) has shown great promise in updating the internal memory of Large Language Models (LLMs) by integrating external knowledge. However, KAG inevitably faces knowledge conflicts when the internal memory contradicts external information. Current approaches to mitigating these conflicts mainly focus on improving external knowledge utilization. However, these methods have shown only limited effectiveness in mitigating the knowledge conflict problem, as internal knowledge continues to influence the generation process of LLMs. In this paper, we propose a ParametrIc Pruning-based Knowledge-Augmented Generation (PIP-KAG) approach, which prunes internal knowledge of LLMs and incorporates a plug-and-play adaptation module to help LLMs better leverage external sources. Additionally, we construct the CoConflictQA benchmark based on the hallucination of LLMs to better evaluate contextual faithfulness during answering questions. Experimental results on CoConflictQA demonstrate that PIP-KAG significantly reduces knowledge conflicts and improves context fidelity. Notably, PIP-KAG reduces LLM's parameters by 13%, enhancing parameter efficiency in LLMs within the KAG framework. All codes are available at https://github.com/OpenBMB/PIP-KAG.
Authors: Milan Gritta, Huiyin Xue, Gerasimos Lampouras
Abstract: Speculative decoding (SD) accelerates Large Language Model (LLM) generation by using an efficient draft model to propose the next few tokens, which are verified by the LLM in a single forward call, reducing latency while preserving its outputs. We focus on retrieval-based SD where the draft model retrieves the next tokens from a non-parametric datastore. Sparse retrieval (REST), which operates on the surface form of strings, is currently the dominant paradigm due to its simplicity and scalability. However, its effectiveness is limited due to the usage of short contexts and exact string matching. Instead, we introduce Dense Retrieval for Speculative Decoding (DReSD), a novel framework that uses approximate nearest neighbour search with contextualised token embeddings to retrieve the most semantically relevant token sequences for SD. Extensive experiments show that DReSD achieves (on average) 87% higher acceptance rates, 65% longer accepted tokens and 19% faster generation speeds compared to sparse retrieval (REST).
Authors: Yingxue Fu
Abstract: Question Under Discussion (QUD), which is originally a linguistic analytic framework, gains increasing attention in the community of natural language processing over the years. Various models have been proposed for implementing QUD for discourse processing. This survey summarizes these models, with a focus on application to written texts, and examines studies that explore the relationship between QUD and mainstream discourse frameworks, including RST, PDTB and SDRT. Some questions that may require further study are suggested.
Authors: Xuansheng Wu, Jiayi Yuan, Wenlin Yao, Xiaoming Zhai, Ninghao Liu
Abstract: Large language models (LLMs) excel at handling human queries, but they can occasionally generate flawed or unexpected responses. Understanding their internal states is crucial for understanding their successes, diagnosing their failures, and refining their capabilities. Although sparse autoencoders (SAEs) have shown promise for interpreting LLM internal representations, limited research has explored how to better explain SAE features, i.e., understanding the semantic meaning of features learned by SAE. Our theoretical analysis reveals that existing explanation methods suffer from the frequency bias issue, where they emphasize linguistic patterns over semantic concepts, while the latter is more critical to steer LLM behaviors. To address this, we propose using a fixed vocabulary set for feature interpretations and designing a mutual information-based objective, aiming to better capture the semantic meaning behind these features. We further propose two runtime steering strategies that adjust the learned feature activations based on their corresponding explanations. Empirical results show that, compared to baselines, our method provides more discourse-level explanations and effectively steers LLM behaviors to defend against jailbreak attacks. These findings highlight the value of explanations for steering LLM behaviors in downstream applications. We will release our code and data once accepted.
Authors: Yunfeng Li, Jiqun Zhang, Guofu Liao, Xue Shi, Junhong Liu
Abstract: With rapid advancements in artificial intelligence, question-answering (Q&A) systems have become essential in intelligent search engines, virtual assistants, and customer service platforms. However, in dynamic domains like smart grids, conventional retrieval-augmented generation(RAG) Q&A systems face challenges such as inadequate retrieval quality, irrelevant responses, and inefficiencies in handling large-scale, real-time data streams. This paper proposes an optimized iterative retrieval-based Q&A framework called Chats-Grid tailored for smart grid environments. In the pre-retrieval phase, Chats-Grid advanced query expansion ensures comprehensive coverage of diverse data sources, including sensor readings, meter records, and control system parameters. During retrieval, Best Matching 25(BM25) sparse retrieval and BAAI General Embedding(BGE) dense retrieval in Chats-Grid are combined to process vast, heterogeneous datasets effectively. Post-retrieval, a fine-tuned large language model uses prompt engineering to assess relevance, filter irrelevant results, and reorder documents based on contextual accuracy. The model further generates precise, context-aware answers, adhering to quality criteria and employing a self-checking mechanism for enhanced reliability. Experimental results demonstrate Chats-Grid's superiority over state-of-the-art methods in fidelity, contextual recall, relevance, and accuracy by 2.37%, 2.19%, and 3.58% respectively. This framework advances smart grid management by improving decision-making and user interactions, fostering resilient and adaptive smart grid infrastructures.
Authors: Jintian Zhang, Yuqi Zhu, Mengshu Sun, Yujie Luo, Shuofei Qiao, Lun Du, Da Zheng, Huajun Chen, Ningyu Zhang
Abstract: Large language models (LLMs) have shown remarkable performance in complex reasoning tasks, but their efficiency is hindered by the substantial memory and computational costs associated with generating lengthy tokens. In this paper, we propose LightThinker, a novel method that enables LLMs to dynamically compress intermediate thoughts during reasoning. Inspired by human cognitive processes, LightThinker compresses verbose thought steps into compact representations and discards the original reasoning chains, thereby significantly reducing the number of tokens stored in the context window. This is achieved by training the model on when and how to perform compression through data construction, mapping hidden states to condensed gist tokens, and creating specialized attention masks. Additionally, we introduce the Dependency (Dep) metric to quantify the degree of compression by measuring the reliance on historical tokens during generation. Extensive experiments on four datasets and two models show that LightThinker reduces peak memory usage and inference time, while maintaining competitive accuracy. Our work provides a new direction for improving the efficiency of LLMs in complex reasoning tasks without sacrificing performance. Code will be released at https://github.com/zjunlp/LightThinker.
Authors: Wenhao Zhu, Pinzhen Chen, Hanxu Hu, Shujian Huang, Fei Yuan, Jiajun Chen, Alexandra Birch
Abstract: Long-context modelling for large language models (LLMs) has been a key area of recent research because many real world use cases require reasoning over longer inputs such as documents. The focus of research into modelling long context has been on how to model position and there has been little investigation into other important aspects of language modelling such as instruction tuning. Long context training examples are challenging and expensive to create and use. In this paper, we investigate how to design instruction data for the post-training phase of a long context pre-trained model: how much and what type of context is needed for optimal and efficient post-training. Our controlled study reveals that models instruction-tuned on short contexts can effectively generalize to longer ones, while also identifying other critical factors such as instruction difficulty and context composition. Based on these findings, we propose context synthesis, a novel data synthesis framework that leverages off-the-shelf LLMs to generate extended background contexts for high-quality instruction-answer pairs. Experiment results on the document-level benchmark (LongBench) demonstrate that our proposed approach outperforms previous instruction synthesis approaches and comes close to the performance of human-annotated long-context instruction data. The project will be available at: https://github.com/NJUNLP/context-synthesis.
Authors: Jiaqi Wu, Chen Chen, Chunyan Hou, Xiaojie Yuan
Abstract: With the widespread real-world deployment of large language models (LLMs), ensuring their behavior complies with safety standards has become crucial. Jailbreak attacks exploit vulnerabilities in LLMs to induce undesirable behavior, posing a significant threat to LLM safety. Previous defenses often fail to achieve both effectiveness and efficiency simultaneously. Defenses from a representation perspective offer new insights, but existing interventions cannot dynamically adjust representations based on the harmfulness of the queries. To address this limitation while ensuring both effectiveness and efficiency, we propose SafeIntervention (SafeInt), a novel defense method that shields LLMs from jailbreak attacks through safety-aware representation intervention. SafeInt is built on our analysis of the representations of jailbreak samples. It adjusts representation distributions of jailbreak samples through intervention to align them with the representations of unsafe samples while minimizing unnecessary perturbations to jailbreak-irrelevant representations. We conduct comprehensive experiments covering six jailbreak attacks, two jailbreak datasets, and two utility benchmarks. Experimental results demonstrate that SafeInt outperforms all baselines in defending LLMs against jailbreak attacks while largely maintaining utility. Additionally, we evaluate SafeInt against adaptive attacks and verify its effectiveness in mitigating real-time attacks.
Authors: Ingroj Shrestha, Louis Tay, Padmini Srinivasan
Abstract: There has been significant prior work using templates to study bias against demographic attributes in MLMs. However, these have limitations: they overlook random variability of templates and target concepts analyzed, assume equality amongst templates, and overlook bias quantification. Addressing these, we propose a systematic statistical approach to assess bias in MLMs, using mixed models to account for random effects, pseudo-perplexity weights for sentences derived from templates and quantify bias using statistical effect sizes. Replicating prior studies, we match on bias scores in magnitude and direction with small to medium effect sizes. Next, we explore the novel problem of gender bias in the context of $\textit{personality}$ and $\textit{character}$ traits, across seven MLMs (base and large). We find that MLMs vary; ALBERT is unbiased for binary gender but the most biased for non-binary $\textit{neo}$, while RoBERTa-large is the most biased for binary gender but shows small to no bias for $\textit{neo}$. There is some alignment of MLM bias and findings in psychology (human perspective) - in $\textit{agreeableness}$ with RoBERTa-large and $\textit{emotional stability}$ with BERT-large. There is general agreement for the remaining 3 personality dimensions: both sides observe at most small differences across gender. For character traits, human studies on gender bias are limited thus comparisons are not feasible.
Authors: Lisa Schut, Yarin Gal, Sebastian Farquhar
Abstract: Large language models (LLMs) have multilingual capabilities and can solve tasks across various languages. However, we show that current LLMs make key decisions in a representation space closest to English, regardless of their input and output languages. Exploring the internal representations with a logit lens for sentences in French, German, Dutch, and Mandarin, we show that the LLM first emits representations close to English for semantically-loaded words before translating them into the target language. We further show that activation steering in these LLMs is more effective when the steering vectors are computed in English rather than in the language of the inputs and outputs. This suggests that multilingual LLMs perform key reasoning steps in a representation that is heavily shaped by English in a way that is not transparent to system users.
Authors: Tianle Li, Chenyang Zhang, Xingwu Chen, Yuan Cao, Difan Zou
Abstract: Transformer-based Large Language Models (LLMs) have demonstrated powerful in-context learning capabilities. However, their predictions can be disrupted by factually correct context, a phenomenon known as context hijacking, revealing a significant robustness issue. To understand this phenomenon theoretically, we explore an in-context linear classification problem based on recent advances in linear transformers. In our setup, context tokens are designed as factually correct query-answer pairs, where the queries are similar to the final query but have opposite labels. Then, we develop a general theoretical analysis on the robustness of the linear transformers, which is formulated as a function of the model depth, training context lengths, and number of hijacking context tokens. A key finding is that a well-trained deeper transformer can achieve higher robustness, which aligns with empirical observations. We show that this improvement arises because deeper layers enable more fine-grained optimization steps, effectively mitigating interference from context hijacking. This is also well supported by our numerical experiments. Our findings provide theoretical insights into the benefits of deeper architectures and contribute to enhancing the understanding of transformer architectures.
Authors: Hugo Pitorro, Marcos Treviso
Abstract: State space models (SSMs), such as Mamba, have emerged as an efficient alternative to transformers for long-context sequence modeling. However, despite their growing adoption, SSMs lack the interpretability tools that have been crucial for understanding and improving attention-based architectures. While recent efforts provide insights into Mamba's internal mechanisms, they do not explicitly decompose token-wise contributions, leaving gaps in understanding how Mamba selectively processes sequences across layers. In this work, we introduce LaTIM, a novel token-level decomposition method for both Mamba-1 and Mamba-2 that enables fine-grained interpretability. We extensively evaluate our method across diverse tasks, including machine translation, copying, and retrieval-based generation, demonstrating its effectiveness in revealing Mamba's token-to-token interaction patterns.
Authors: Xueran Han, Yuhan Liu, Mingzhe Li, Wei Liu, Sen Hu, Rui Yan, Zhiqiang Xu, Xiuying Chen
Abstract: Great novels create immersive worlds with rich character arcs, well-structured plots, and nuanced writing styles. However, current novel generation methods often rely on brief, simplistic story outlines and generate details using plain, generic language. To bridge this gap, we introduce the task of Pastiche Novel Generation, which requires the generated novels to imitate the distinctive features of the original work, including understanding character profiles, predicting plausible plot developments, and writing concrete details using vivid, expressive language. To achieve this, we propose WriterAgent, a novel generation system designed to master the core aspects of literary pastiche. WriterAgent is trained through a curriculum learning paradigm, progressing from low-level stylistic mastery to high-level narrative coherence. Its key tasks include language style learning, character modeling, plot planning, and stylish writing, ensuring comprehensive narrative control. To support this, WriterAgent leverages the WriterLoRA framework, an extension of LoRA with hierarchical and cumulative task-specific modules, each specializing in a different narrative aspect. We evaluate WriterAgent on multilingual classics like Harry Potter and Dream of the Red Chamber, demonstrating its superiority over baselines in capturing the target author's settings, character dynamics, and writing style to produce coherent, faithful narratives.
Authors: Qi Le, Enmao Diao, Ziyan Wang, Xinran Wang, Jie Ding, Li Yang, Ali Anwar
Abstract: We introduce Probe Pruning (PP), a novel framework for online, dynamic, structured pruning of Large Language Models (LLMs) applied in a batch-wise manner. PP leverages the insight that not all samples and tokens contribute equally to the model's output, and probing a small portion of each batch effectively identifies crucial weights, enabling tailored dynamic pruning for different batches. It comprises three main stages: probing, history-informed pruning, and full inference. In the probing stage, PP selects a small yet crucial set of hidden states, based on residual importance, to run a few model layers ahead. During the history-informed pruning stage, PP strategically integrates the probing states with historical states. Subsequently, it structurally prunes weights based on the integrated states and the PP importance score, a metric developed specifically to assess the importance of each weight channel in maintaining performance. In the final stage, full inference is conducted on the remaining weights. A major advantage of PP is its compatibility with existing models, as it operates without requiring additional neural network modules or fine-tuning. Comprehensive evaluations of PP on LLaMA-2/3 and OPT models reveal that even minimal probing-using just 1.5% of FLOPs-can substantially enhance the efficiency of structured pruning of LLMs. For instance, when evaluated on LLaMA-2-7B with WikiText2, PP achieves a 2.56 times lower ratio of performance degradation per unit of runtime reduction compared to the state-of-the-art method at a 40% pruning ratio. Our code is available at https://github.com/Qi-Le1/Probe_Pruning.
Authors: Ngoc Luyen Le, Gildas Tagny Ngomp\'e
Abstract: In this article, we present the BTransformer18 model, a deep learning architecture designed for multi-label relation extraction in French texts. Our approach combines the contextual representation capabilities of pre-trained language models from the BERT family - such as BERT, RoBERTa, and their French counterparts CamemBERT and FlauBERT - with the power of Transformer encoders to capture long-term dependencies between tokens. Experiments conducted on the dataset from the TextMine'25 challenge show that our model achieves superior performance, particularly when using CamemBERT-Large, with a macro F1 score of 0.654, surpassing the results obtained with FlauBERT-Large. These results demonstrate the effectiveness of our approach for the automatic extraction of complex relations in intelligence reports.
Authors: Anirudh Sundar, Sinead Williamson, Katherine Metcalf, Barry-John Theobald, Skyler Seto, Masha Fedzechkina
Abstract: Aligned representations across languages is a desired property in multilingual large language models (mLLMs), as alignment can improve performance in cross-lingual tasks. Typically alignment requires fine-tuning a model, which is computationally expensive, and sizable language data, which often may not be available. A data-efficient alternative to fine-tuning is model interventions -- a method for manipulating model activations to steer generation into the desired direction. We analyze the effect of a popular intervention (finding experts) on the alignment of cross-lingual representations in mLLMs. We identify the neurons to manipulate for a given language and introspect the embedding space of mLLMs pre- and post-manipulation. We show that modifying the mLLM's activations changes its embedding space such that cross-lingual alignment is enhanced. Further, we show that the changes to the embedding space translate into improved downstream performance on retrieval tasks, with up to 2x improvements in top-1 accuracy on cross-lingual retrieval.
Authors: George Drayson, 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 web data is the primary resource for LLM pretraining, future models will be trained on an unknown portion of synthetic data. This will lead to model collapse, a degenerative process which causes models to reinforce their own errors and experience a drop in model performance. In this study, we investigate the impact of decoding strategy on model collapse, where we analyse the characteristics of the generated data during recursive training, its similarity to human references and the resulting model performance. Using the decoding strategies that lead to the most significant model degradation, we tackle the question: how to avoid model collapse when the origin (human or synthetic) of the training data is unknown. We design a novel methodology based on resampling the data distribution using importance weights from our machine-generated text detector. Our method is validated on two LLM variants (GPT-2 and SmolLM2) on the open-ended text generation task, demonstrating that we can successfully prevent model collapse and when there is enough human-authored data in the training dataset, our method improves model performance.
Authors: Shoumik Saha, Soheil Feizi
Abstract: The growing use of large language models (LLMs) for text generation has led to widespread concerns about AI-generated content detection. However, an overlooked challenge is AI-polished text, where human-written content undergoes subtle refinements using AI tools. This raises a critical question: should minimally polished text be classified as AI-generated? Misclassification can lead to false plagiarism accusations and misleading claims about AI prevalence in online content. In this study, we systematically evaluate eleven state-of-the-art AI-text detectors using our AI-Polished-Text Evaluation (APT-Eval) dataset, which contains $11.7K$ samples refined at varying AI-involvement levels. Our findings reveal that detectors frequently misclassify even minimally polished text as AI-generated, struggle to differentiate between degrees of AI involvement, and exhibit biases against older and smaller models. These limitations highlight the urgent need for more nuanced detection methodologies.
Authors: Zongkai Zhao, Guozeng Xu, Xiuhua Li, Kaiwen Wei, Jiang Zhong
Abstract: Locate-then-Edit Knowledge Editing (LEKE) is a key technique for updating large language models (LLMs) without full retraining. However, existing methods assume a single-user setting and become inefficient in real-world multi-client scenarios, where decentralized organizations (e.g., hospitals, financial institutions) independently update overlapping knowledge, leading to redundant mediator knowledge vector (MKV) computations and privacy concerns. To address these challenges, we introduce Federated Locate-then-Edit Knowledge Editing (FLEKE), a novel task that enables multiple clients to collaboratively perform LEKE while preserving privacy and reducing computational overhead. To achieve this, we propose FedEdit, a two-stage framework that optimizes MKV selection and reuse. In the first stage, clients locally apply LEKE and upload the computed MKVs. In the second stage, rather than relying solely on server-based MKV sharing, FLEKE allows clients retrieve relevant MKVs based on cosine similarity, enabling knowledge re-edit and minimizing redundant computations. Experimental results on two benchmark datasets demonstrate that FedEdit retains over 96% of the performance of non-federated LEKE while significantly outperforming a FedAvg-based baseline by approximately twofold. Besides, we find that MEMIT performs more consistently than PMET in the FLEKE task with our FedEdit framework. Our code is available at https://github.com/zongkaiz/FLEKE.
Authors: Jaydeep Borkar, Matthew Jagielski, Katherine Lee, Niloofar Mireshghallah, David A. Smith, Christopher A. Choquette-Choo
Abstract: Due to the sensitive nature of personally identifiable information (PII), its owners may have the authority to control its inclusion or request its removal from large-language model (LLM) training. Beyond this, PII may be added or removed from training datasets due to evolving dataset curation techniques, because they were newly scraped for retraining, or because they were included in a new downstream fine-tuning stage. We find that the amount and ease of PII memorization is a dynamic property of a model that evolves throughout training pipelines and depends on commonly altered design choices. We characterize three such novel phenomena: (1) similar-appearing PII seen later in training can elicit memorization of earlier-seen sequences in what we call assisted memorization, and this is a significant factor (in our settings, up to 1/3); (2) adding PII can increase memorization of other PII significantly (in our settings, as much as $\approx\!7.5\times$); and (3) removing PII can lead to other PII being memorized. Model creators should consider these first- and second-order privacy risks when training models to avoid the risk of new PII regurgitation.
Authors: Hamid Moradi-Kamali, Mohammad-Hossein Rajabi-Ghozlou, Mahdi Ghazavi, Ali Soltani, Amirreza Sattarzadeh, Reza Entezari-Maleki
Abstract: Financial Sentiment Analysis (FSA) traditionally relies on human-annotated sentiment labels to infer investor sentiment and forecast market movements. However, inferring the potential market impact of words based on human-perceived intentions is inherently challenging. We hypothesize that the historical market reactions to words, offer a more reliable indicator of their potential impact on markets than subjective sentiment interpretations by human annotators. To test this hypothesis, a market-derived labeling approach is proposed to assign tweet labels based on ensuing short-term price trends, enabling the language model to capture the relationship between textual signals and market dynamics directly. A domain-specific language model was fine-tuned on these labels, achieving up to an 11% improvement in short-term trend prediction accuracy over traditional sentiment-based benchmarks. Moreover, by incorporating market and temporal context through prompt-tuning, the proposed context-aware language model demonstrated an accuracy of 89.6% on a curated dataset of 227 impactful Bitcoin-related news events with significant market impacts. Aggregating daily tweet predictions into trading signals, our method outperformed traditional fusion models (which combine sentiment-based and price-based predictions). It challenged the assumption that sentiment-based signals are inferior to price-based predictions in forecasting market movements. Backtesting these signals across three distinct market regimes yielded robust Sharpe ratios of up to 5.07 in trending markets and 3.73 in neutral markets. Our findings demonstrate that language models can serve as effective short-term market predictors. This paradigm shift underscores the untapped capabilities of language models in financial decision-making and opens new avenues for market prediction applications.
Authors: Peter Carragher, Nikitha Rao, Abhinand Jha, R Raghav, Kathleen M. Carley
Abstract: The robustness of large language models (LLMs) against knowledge conflicts in unimodal question answering systems has been well studied. However, the effect of conflicts in information sources on vision language models (VLMs) in multimodal settings has not yet been explored. In this work, we propose \segsub, a framework that applies targeted perturbations to image sources to study and improve the robustness of VLMs against three different types of knowledge conflicts, namely parametric, source, and counterfactual conflicts. Contrary to prior findings that showed that LLMs are sensitive to parametric conflicts arising from textual perturbations, we find VLMs are largely robust to image perturbation. On the other hand, VLMs perform poorly on counterfactual examples (<30% accuracy) and fail to reason over source conflicts (<1% accuracy). We also find a link between hallucinations and image context, with GPT-4o prone to hallucination when presented with highly contextualized counterfactual examples. While challenges persist with source conflicts, finetuning models significantly improves reasoning over counterfactual samples. Our findings highlight the need for VLM training methodologies that enhance their reasoning capabilities, particularly in addressing complex knowledge conflicts between multimodal sources.
Authors: Zhihang Liu, Chen-Wei Xie, Bin Wen, Feiwu Yu, Jixuan Chen, Boqiang Zhang, Nianzu Yang, Pandeng Li, Yun Zheng, Hongtao Xie
Abstract: Recent advancements in Multimodal Large Language Models (MLLMs) have rendered traditional visual captioning benchmarks obsolete, as they primarily evaluate short descriptions with outdated metrics. While recent benchmarks address these limitations by decomposing captions into visual elements and adopting model-based evaluation, they remain incomplete-overlooking critical aspects, while providing vague, non-explanatory scores. To bridge this gap, we propose CV-CapBench, a Comprehensive Visual Caption Benchmark that systematically evaluates caption quality across 6 views and 13 dimensions. CV-CapBench introduces precision, recall, and hit rate metrics for each dimension, uniquely assessing both correctness and coverage. Experiments on leading MLLMs reveal significant capability gaps, particularly in dynamic and knowledge-intensive dimensions. These findings provide actionable insights for future research. The code and data will be released.
Authors: Yuehong Cassandra Tai, Khushi Navin Patni, Nicholas Daniel Hemauer, Bruce Desmarais, Yu-Ru Lin
Abstract: Despite recent advances in understanding the capabilities and limits of generative artificial intelligence (GenAI) models, we are just beginning to understand their capacity to assess and reason about the veracity of content. We evaluate multiple GenAI models across tasks that involve the rating of, and perceived reasoning about, the credibility of information. The information in our experiments comes from content that subnational U.S. politicians post to Facebook. We find that GPT-4o, one of the most used AI models in consumer applications, outperforms other models, but all models exhibit only moderate agreement with human coders. Importantly, even when GenAI models accurately identify low-credibility content, their reasoning relies heavily on linguistic features and ``hard'' criteria, such as the level of detail, source reliability, and language formality, rather than an understanding of veracity. We also assess the effectiveness of summarized versus full content inputs, finding that summarized content holds promise for improving efficiency without sacrificing accuracy. While GenAI has the potential to support human fact-checkers in scaling misinformation detection, our results caution against relying solely on these models.
Authors: Ahmed Heakl, Abdullah Sohail, Mukul Ranjan, Rania Hossam, Ghazi Ahmed, Mohamed El-Geish, Omar Maher, Zhiqiang Shen, Fahad Khan, Salman Khan
Abstract: With the growing adoption of Retrieval-Augmented Generation (RAG) in document processing, robust text recognition has become increasingly critical for knowledge extraction. While OCR (Optical Character Recognition) for English and other languages benefits from large datasets and well-established benchmarks, Arabic OCR faces unique challenges due to its cursive script, right-to-left text flow, and complex typographic and calligraphic features. We present KITAB-Bench, a comprehensive Arabic OCR benchmark that fills the gaps in current evaluation systems. Our benchmark comprises 8,809 samples across 9 major domains and 36 sub-domains, encompassing diverse document types including handwritten text, structured tables, and specialized coverage of 21 chart types for business intelligence. Our findings show that modern vision-language models (such as GPT-4, Gemini, and Qwen) outperform traditional OCR approaches (like EasyOCR, PaddleOCR, and Surya) by an average of 60% in Character Error Rate (CER). Furthermore, we highlight significant limitations of current Arabic OCR models, particularly in PDF-to-Markdown conversion, where the best model Gemini-2.0-Flash achieves only 65% accuracy. This underscores the challenges in accurately recognizing Arabic text, including issues with complex fonts, numeral recognition errors, word elongation, and table structure detection. This work establishes a rigorous evaluation framework that can drive improvements in Arabic document analysis methods and bridge the performance gap with English OCR technologies.
Authors: Lingjun Zhao, Mingyang Xie, Paola Cascante-Bonilla, Hal Daum\'e III, Kwonjoon Lee
Abstract: Large Vision-Language Models often generate hallucinated content that is not grounded in its visual inputs. While prior work focuses on mitigating hallucinations, we instead explore leveraging hallucination correction as a training objective to improve video-language alignment. We introduce HACA, a self-training framework learning to correct hallucinations in descriptions that do not align with the video content. By identifying and correcting inconsistencies, HACA enhances the model's ability to align video and textual representations for spatio-temporal reasoning. Our experimental results show consistent gains in video-caption binding and text-to-video retrieval tasks, demonstrating that hallucination correction-inspired tasks serve as an effective strategy for improving vision and language alignment.
Authors: Vaidehi Patil, Elias Stengel-Eskin, Mohit Bansal
Abstract: User specifications or legal frameworks often require information to be removed from pretrained models, including large language models (LLMs). This requires deleting or "forgetting" a set of data points from an already-trained model, which typically degrades its performance on other data points. Thus, a balance must be struck between removing information and keeping the model's other abilities intact, with a failure to balance this trade-off leading to poor deletion or an unusable model. To this end, we propose UPCORE (Utility-Preserving Coreset Selection), a method-agnostic data selection framework for mitigating collateral damage during unlearning. Finding that the model damage is correlated with the variance of the model's representations on the forget set, we selectively prune the forget set to remove outliers, thereby minimizing model degradation after unlearning. We evaluate UPCORE across three standard unlearning methods consistently achieving a superior balance between the competing objectives of deletion efficacy and model preservation. To better evaluate this trade-off, we introduce a new metric, measuring the area-under-the-curve (AUC) across standard metrics. We find that UPCORE improves both standard metrics and AUC, benefitting from positive transfer between the coreset and pruned points while reducing negative transfer from the forget set to points outside of it.
Authors: Xiaoyu Chen, Changde Du, Che Liu, Yizhe Wang, Huiguang He
Abstract: Decoding language information from brain signals represents a vital research area within brain-computer interfaces, particularly in the context of deciphering the semantic information from the fMRI signal. Although existing work uses LLM to achieve this goal, their method does not use an end-to-end approach and avoids the LLM in the mapping of fMRI-to-text, leaving space for the exploration of the LLM in auditory decoding. In this paper, we introduce a novel method, the Brain Prompt GPT (BP-GPT). By using the brain representation that is extracted from the fMRI as a prompt, our method can utilize GPT-2 to decode fMRI signals into stimulus text. Further, we introduce the text prompt and align the fMRI prompt to it. By introducing the text prompt, our BP-GPT can extract a more robust brain prompt and promote the decoding of pre-trained LLM. We evaluate our BP-GPT on the open-source auditory semantic decoding dataset and achieve a significant improvement up to 4.61 on METEOR and 2.43 on BERTScore across all the subjects compared to the state-of-the-art method. The experimental results demonstrate that using brain representation as a prompt to further drive LLM for auditory neural decoding is feasible and effective. The code is available at https://github.com/1994cxy/BP-GPT.
Authors: Aarash Feizi, Sai Rajeswar, Adriana Romero-Soriano, Reihaneh Rabbany, Spandana Gella, Valentina Zantedeschi, Jo\~ao Monteiro
Abstract: As large vision language models (VLMs) are increasingly used as automated evaluators, understanding their ability to effectively compare data pairs as instructed in the prompt becomes essential. To address this, we present PairBench, a low-cost framework that systematically evaluates VLMs as customizable similarity tools across various modalities and scenarios. Through PairBench, we introduce four metrics that represent key desiderata of similarity scores: alignment with human annotations, consistency for data pairs irrespective of their order, smoothness of similarity distributions, and controllability through prompting. Our analysis demonstrates that no model, whether closed- or open-source, is superior on all metrics; the optimal choice depends on an auto evaluator's desired behavior (e.g., a smooth vs. a sharp judge), highlighting risks of widespread adoption of VLMs as evaluators without thorough assessment. For instance, the majority of VLMs struggle with maintaining symmetric similarity scores regardless of order. Additionally, our results show that the performance of VLMs on the metrics in PairBench closely correlates with popular benchmarks, showcasing its predictive power in ranking models.
Authors: Sheila Schoepp, Masoud Jafaripour, Yingyue Cao, Tianpei Yang, Fatemeh Abdollahi, Shadan Golestan, Zahin Sufiyan, Osmar R. Zaiane, Matthew E. Taylor
Abstract: Reinforcement learning (RL) has shown impressive results in sequential decision-making tasks. Meanwhile, Large Language Models (LLMs) and Vision-Language Models (VLMs) have emerged, exhibiting impressive capabilities in multimodal understanding and reasoning. These advances have led to a surge of research integrating LLMs and VLMs into RL. In this survey, we review representative works in which LLMs and VLMs are used to overcome key challenges in RL, such as lack of prior knowledge, long-horizon planning, and reward design. We present a taxonomy that categorizes these LLM/VLM-assisted RL approaches into three roles: agent, planner, and reward. We conclude by exploring open problems, including grounding, bias mitigation, improved representations, and action advice. By consolidating existing research and identifying future directions, this survey establishes a framework for integrating LLMs and VLMs into RL, advancing approaches that unify natural language and visual understanding with sequential decision-making.
Authors: Sangeetha N, Harish Thangaraj, Varun Vashisht, Eshaan Joshi, Kanishka Verma, Diya Katariya
Abstract: Universities serve as a hub for academic collaboration, promoting the exchange of diverse ideas and perspectives among students and faculty through interdisciplinary dialogue. However, as universities expand in size, conventional networking approaches via student chapters, class groups, and faculty committees become cumbersome. To address this challenge, an academia-specific profile recommendation system is proposed to connect like-minded stakeholders within any university community. This study evaluates three techniques: Term Frequency-Inverse Document Frequency (TF-IDF), Bidirectional Encoder Representations from Transformers (BERT), and a hybrid approach to generate effective recommendations. Due to the unlabelled nature of the dataset, Affinity Propagation cluster-based relabelling is performed to understand the grouping of similar profiles. The hybrid model demonstrated superior performance, evidenced by its similarity score, Silhouette score, Davies-Bouldin index, and Normalized Discounted Cumulative Gain (NDCG), achieving an optimal balance between diversity and relevance in recommendations. Furthermore, the optimal model has been implemented as a mobile application, which dynamically suggests relevant profiles based on users' skills and collaboration interests, incorporating contextual understanding. The potential impact of this application is significant, as it promises to enhance networking opportunities within large academic institutions through the deployment of intelligent recommendation systems.
Authors: Shilong Hou, Ruilin Shang, Zi Long, Xianghua Fu, Yin Chen
Abstract: An increasing number of companies have begun providing services that leverage cloud-based large language models (LLMs), such as ChatGPT. However, this development raises substantial privacy concerns, as users' prompts are transmitted to and processed by the model providers. Among the various privacy protection methods for LLMs, those implemented during the pre-training and fine-tuning phrases fail to mitigate the privacy risks associated with the remote use of cloud-based LLMs by users. On the other hand, methods applied during the inference phrase are primarily effective in scenarios where the LLM's inference does not rely on privacy-sensitive information. In this paper, we outline the process of remote user interaction with LLMs and, for the first time, propose a detailed definition of a general pseudonymization framework applicable to cloud-based LLMs. The experimental results demonstrate that the proposed framework strikes an optimal balance between privacy protection and utility. The code for our method is available to the public at https://github.com/Mebymeby/Pseudonymization-Framework.
URLs: https://github.com/Mebymeby/Pseudonymization-Framework.
Authors: Hong Yankun, Li Xing, Zhen Hui-Ling, Yu Xianzhi, Liu Wulong, Yuan Mingxuan
Abstract: For the efficient inference of Large Language Models (LLMs), the effective compression of key-value (KV) cache is essential. Three main types of KV cache compression techniques, namely sparsity, channel compression, and quantization, have been identified. This study presents SVDq, a Singular Value Decomposition (SVD) - based mixed precision quantization method for K cache. Initially, K cache is transformed into latent channels using SVD basis representations. Since the values in latent channels decay rapidly and become negligible after only a few latent channels, our method then incorporates importance-aware quantization and compression for latent channels. This enables the effective allocation of higher precision to more significant channels. Theoretically, we prove that SVDq results in quantization errors (x0.1 or even lower) that are much lower than those of per-channel key quantization in the original space. Our findings based on RULER and LongBench benchmarks demonstrate that SVDq can achieve an equivalent key cache precision as low as 1.25-bit. When combined with key sparsity, it can reach a key compression ratio of up to 410x for attention computation, all while maintaining comparable model performance. Notably, our method is nearly lossless for LongBench datasets. This indicates that SVDq enables high-precision low-bit quantization, providing a more efficient solution for KV cache compression in LLMs.
Authors: Kai Li, Fei Liu, Zhenkun Wang, Xialiang Tong, Xiongwei Han, Mingxuan Yuan
Abstract: Real-world Vehicle Routing Problems (VRPs) are characterized by a variety of practical constraints, making manual solver design both knowledge-intensive and time-consuming. Although there is increasing interest in automating the design of routing algorithms, existing research has explored only a limited array of VRP variants and fails to adequately address the complex and prevalent constraints encountered in real-world situations. To fill this gap, this paper introduces RoutBench, a benchmark of 1,000 VRP variants derived from 24 attributes, for evaluating the effectiveness of automatic routing solvers in addressing complex constraints. Along with RoutBench, we present the Automatic Routing Solver (ARS), which employs Large Language Model (LLM) agents to enhance a backbone algorithm framework by automatically generating constraint-aware heuristic code, based on problem descriptions and several representative constraints selected from a database. Our experiments show that ARS outperforms state-of-the-art LLM-based methods and commonly used solvers, automatically solving 91.67% of common VRPs and achieving at least a 30% improvement across all benchmarks.
Authors: Kang Bongsu, Kim Jundong, Yun Tae-Rim, Bae Hyojin, Kim Chang-Eop
Abstract: This study quantitively examines which features of AI-generated text lead humans to perceive subjective consciousness in large language model (LLM)-based AI systems. Drawing on 99 passages from conversations with Claude 3 Opus and focusing on eight features -- metacognitive self-reflection, logical reasoning, empathy, emotionality, knowledge, fluency, unexpectedness, and subjective expressiveness -- we conducted a survey with 123 participants. Using regression and clustering analyses, we investigated how these features influence participants' perceptions of AI consciousness. The results reveal that metacognitive self-reflection and the AI's expression of its own emotions significantly increased perceived consciousness, while a heavy emphasis on knowledge reduced it. Participants clustered into seven subgroups, each showing distinct feature-weighting patterns. Additionally, higher prior knowledge of LLMs and more frequent usage of LLM-based chatbots were associated with greater overall likelihood assessments of AI consciousness. This study underscores the multidimensional and individualized nature of perceived AI consciousness and provides a foundation for better understanding the psychosocial implications of human-AI interaction.
Authors: Shaharukh Khan, Ayush Tarun, Abhinav Ravi, Ali Faraz, Akshat Patidar, Praveen Kumar Pokala, Anagha Bhangare, Raja Kolla, Chandra Khatri, Shubham Agarwal
Abstract: Recent multimodal foundation models are primarily trained on English or high resource European language data, which hinders their applicability to other medium and low-resource languages. To address this limitation, we introduce Chitrarth (Chitra: Image; Artha: Meaning), an inclusive Vision-Language Model (VLM), specifically targeting the rich linguistic diversity and visual reasoning across 10 prominent Indian languages. Our model effectively integrates a state-of-the-art (SOTA) multilingual Large Language Model (LLM) with a vision module, primarily trained on multilingual image-text data. Furthermore, we also introduce BharatBench, a comprehensive framework for evaluating VLMs across various Indian languages, ultimately contributing to more diverse and effective AI systems. Our model achieves SOTA results for benchmarks across low resource languages while retaining its efficiency in English. Through our research, we aim to set new benchmarks in multilingual-multimodal capabilities, offering substantial improvements over existing models and establishing a foundation to facilitate future advancements in this arena.
Authors: Leyla Naz Candogan, Yongtao Wu, Elias Abad Rocamora, Grigorios G. Chrysos, Volkan Cevher
Abstract: Defending aligned Large Language Models (LLMs) against jailbreaking attacks is a challenging problem, with existing approaches requiring multiple requests or even queries to auxiliary LLMs, making them computationally heavy. Instead, we focus on detecting jailbreaking input in a single forward pass. Our method, called Single Pass Detection SPD, leverages the information carried by the logits to predict whether the output sentence will be harmful. This allows us to defend in just one forward pass. SPD can not only detect attacks effectively on open-source models, but also minimizes the misclassification of harmless inputs. Furthermore, we show that SPD remains effective even without complete logit access in GPT-3.5 and GPT-4. We believe that our proposed method offers a promising approach to efficiently safeguard LLMs against adversarial attacks.
Authors: Raghav Singhal, Kaustubh Ponkshe, Rohit Vartak, Lav R. Varshney, Praneeth Vepakomma
Abstract: Low-Rank Adaptation (LoRA) has become ubiquitous for efficiently fine-tuning foundation models. However, federated fine-tuning using LoRA is challenging due to suboptimal updates arising from traditional federated averaging of individual adapters. Existing solutions either incur prohibitively high communication cost that scales linearly with the number of clients or suffer from performance degradation due to limited expressivity. We introduce Federated Silver Bullet (Fed-SB), a novel approach for federated fine-tuning of LLMs using LoRA-SB, a recently proposed low-rank adaptation method. LoRA-SB optimally aligns the optimization trajectory with the ideal low-rank full fine-tuning projection by learning a small square matrix (R) between adapters B and A, keeping other components fixed. Direct averaging of R guarantees exact updates, substantially reducing communication cost, which remains independent of the number of clients, and enables scalability. Fed-SB achieves state-of-the-art performance across commonsense reasoning, arithmetic reasoning, and language inference tasks while reducing communication costs by up to 230x. In private settings, Fed-SB further improves performance by (1) reducing trainable parameters, thereby lowering the noise required for differential privacy and (2) avoiding noise amplification introduced by other methods. Overall, Fed-SB establishes a new Pareto frontier in the tradeoff between communication and performance, offering an efficient and scalable solution for both private and non-private federated fine-tuning. Our code is publicly available at https://github.com/CERT-Lab/fed-sb.
Authors: Yuan Sun
Abstract: MoE (Mixture-of-Expert) architectures appear frequently in large language models, and the number of experts can be over one hundred recently. However, the expert load imbalance problem always happens in MoE model pre-training, which will cause routing collapse or increased computational overhead. In order to balance loads on experts, we propose BIP-Based Balancing, an expert load balancing algorithm based on binary integer programming (BIP). The algorithm maintains an additional vector q that can help change the top-K order of s by solving a binary integer programming with very small time costs. In simulation experiments, we observe that BIP-Based Balancing make imbalance disappoint very fast, while the final sum of routine scores decreases very little. Our algorithm achieves nearly perfect trade-off between expert load balance and pre-training efficiency under the simulation view.
Authors: Shashank Kirtania
Abstract: Large Language Models (LLMs) have shown promise in proving formal theorems using proof assistants like Lean. However, current state of the art language models struggles to predict next step in proofs leading practitioners to use different sampling techniques to improve LLMs capabilities. We observe that the LLM is capable of predicting the correct tactic; however, it faces challenges in ranking it appropriately within the set of candidate tactics, affecting the overall selection process. To overcome this hurdle, we use activation steering to guide LLMs responses to improve the generations at the time of inference. Our results suggest that activation steering offers a promising lightweight alternative to specialized fine-tuning for enhancing theorem proving capabilities in LLMs, particularly valuable in resource-constrained environments.
Authors: Tim R\"adsch, Leon Mayer, Simon Pavicic, A. Emre Kavur, Marcel Knopp, Bar{\i}\c{s} \"Ozt\"urk, Klaus Maier-Hein, Paul F. Jaeger, Fabian Isensee, Annika Reinke, Lena Maier-Hein
Abstract: Reliable evaluation of AI models is critical for scientific progress and practical application. While existing VLM benchmarks provide general insights into model capabilities, their heterogeneous designs and limited focus on a few imaging domains pose significant challenges for both cross-domain performance comparison and targeted domain-specific evaluation. To address this, we propose three key contributions: (1) a framework for the resource-efficient creation of domain-specific VLM benchmarks enabled by task augmentation for creating multiple diverse tasks from a single existing task, (2) the release of new VLM benchmarks for seven domains, created according to the same homogeneous protocol and including 162,946 thoroughly human-validated answers, and (3) an extensive benchmarking of 22 state-of-the-art VLMs on a total of 37,171 tasks, revealing performance variances across domains and tasks, thereby supporting the need for tailored VLM benchmarks. Adoption of our methodology will pave the way for the resource-efficient domain-specific selection of models and guide future research efforts toward addressing core open questions.
Authors: Fengxiang Cheng, Haoxuan Li, Fenrong Liu, Robert van Rooij, Kun Zhang, Zhouchen Lin
Abstract: Large language models (LLMs) have achieved remarkable successes on various natural language tasks. However, recent studies have found that there are still significant challenges to the logical reasoning abilities of LLMs. This paper summarizes and categorizes the main challenges into two aspects: (1) Logical question answering, LLMs often fail to generate the correct answer within complex logical problem which requires sophisticated deductive, inductive or abductive reasoning given a collection of premises and constrains. (2) Logical consistency, LLMs are prone to producing responses contradicting themselves across different questions. For example, a state-of-the-art Macaw question-answering LLM answers Yes to both questions Is a magpie a bird? and Does a bird have wings? but answers No to Does a magpie have wings?. To facilitate this research direction, we comprehensively investigate the most cutting-edge methods and propose detailed taxonomies of these methods. Specifically, to accurately answer complex logic questions, previous methods can be categorized based on reliance on external solvers, prompts, pretraining, and fine-tuning. To avoid logical contradictions, we discuss concepts and solutions of various logical consistencies, including implication, negation, transitivity, factuality consistency, and their composites. In addition, we review commonly used benchmark datasets and evaluation metrics, and discuss promising research directions, such as extensions to modal logic to account for uncertainty, and efficient algorithms satisfying multiple logical consistencies simultaneously.
Authors: Zhining Zhang, Chuanyang Jin, Mung Yao Jia, Tianmin Shu
Abstract: Theory of Mind (ToM), the ability to understand people's mental variables based on their behavior, is key to developing socially intelligent agents. Current approaches to Theory of Mind reasoning either rely on prompting Large Language Models (LLMs), which are prone to systematic errors, or use rigid, handcrafted Bayesian Theory of Mind (BToM) models, which are more robust but cannot generalize across different domains. In this work, we introduce AutoToM, an automated Bayesian Theory of Mind method for achieving open-ended machine Theory of Mind. AutoToM can operate in any domain, infer any mental variable, and conduct robust Theory of Mind reasoning of any order. Given a Theory of Mind inference problem, AutoToM first proposes an initial BToM model. It then conducts automated Bayesian inverse planning based on the proposed model, leveraging an LLM as the backend. Based on the uncertainty of the inference, it iteratively refines the model, by introducing additional mental variables and/or incorporating more timesteps in the context. Empirical evaluations across multiple Theory of Mind benchmarks demonstrate that AutoToM consistently achieves state-of-the-art performance, offering a scalable, robust, and interpretable approach to machine Theory of Mind.
Authors: Manisha Mukherjee, Vincent J. Hellendoorn
Abstract: Recent advancements in AI have sparked a trend in constructing large, generalist language models that handle a multitude of tasks, including many code-related ones. While these models are expensive to train and are often closed-source, they have enjoyed broad adoption because they tend to outperform smaller, domain-specific models of code. In this work, we argue that this is not a foregone conclusion. We show that modestly sized domain-specific models can outperform much larger ones on code labeling tasks, provided they are trained to the same standards. Concretely, we focus on StackOverflow (SO), which offers large volumes of aligned code and text data. We align established best-practices for pre-training large language models with properties of SO as a data source, especially using a large context window (2,048 tokens), coupled with a powerful toolkit (Megatron-LM) to train two models: SOBertBase (125M parameters) and SOBertLarge (762M parameters), at a budget of just $374 and $1600 each. We compare the performance of our models with a prior domain-specific model which did not adopt many of these practices (BERTOverflow), as well two general-purpose BERT models and two models in OpenAI's GPT series (GPT-3.5 and GPT-4). We study four labeling tasks: question quality prediction, closed question prediction, NER and obsoletion prediction. The final task is a new benchmark we introduce, on which we additionally compare SOBert with a fine-tuned CodeLlama and StackLlama (models with 10x more parameters than SOBertLarge). Our models consistently outperform all baselines. In contrast, BertOverflow is outperformed by generalist models in most tasks. These results demonstrate that pre-training both extensively and properly on in-domain data can yield a powerful and affordable alternative to leveraging closed-source general-purpose models. Both models are released to the public on Hugging Face.
Authors: Tianlong Li, Zhenghua Wang, Wenhao Liu, Muling Wu, Shihan Dou, Changze Lv, Xiaohua Wang, Xiaoqing Zheng, Xuanjing Huang
Abstract: The recent surge in jailbreaking attacks has revealed significant vulnerabilities in Large Language Models (LLMs) when exposed to malicious inputs. While various defense strategies have been proposed to mitigate these threats, there has been limited research into the underlying mechanisms that make LLMs vulnerable to such attacks. In this study, we suggest that the self-safeguarding capability of LLMs is linked to specific activity patterns within their representation space. Although these patterns have little impact on the semantic content of the generated text, they play a crucial role in shaping LLM behavior under jailbreaking attacks. Our findings demonstrate that these patterns can be detected with just a few pairs of contrastive queries. Extensive experimentation shows that the robustness of LLMs against jailbreaking can be manipulated by weakening or strengthening these patterns. Further visual analysis provides additional evidence for our conclusions, providing new insights into the jailbreaking phenomenon. These findings highlight the importance of addressing the potential misuse of open-source LLMs within the community.
Authors: Kun-Peng Ning, Shuo Yang, Yu-Yang Liu, Jia-Yu Yao, Zhen-Hui Liu, Yong-Hong Tian, Yibing Song, Li Yuan
Abstract: Existing large language models (LLMs) evaluation methods typically focus on testing the performance on some closed-environment and domain-specific benchmarks with human annotations. In this paper, we explore a novel unsupervised evaluation direction, utilizing peer-review mechanisms to measure LLMs automatically. In this setting, both open-source and closed-source LLMs lie in the same environment, capable of answering unlabeled questions and evaluating each other, where each LLM's response score is jointly determined by other anonymous ones. To obtain the ability hierarchy among these models, we assign each LLM a learnable capability parameter to adjust the final ranking. We formalize it as a constrained optimization problem, intending to maximize the consistency of each LLM's capabilities and scores. The key assumption behind is that high-level LLM can evaluate others' answers more accurately than low-level ones, while higher-level LLM can also achieve higher response scores. Moreover, we propose three metrics called PEN, CIN, and LIS to evaluate the gap in aligning human rankings. We perform experiments on multiple datasets with these metrics, validating the effectiveness of the proposed approach.
Authors: Berivan Isik, Natalia Ponomareva, Hussein Hazimeh, Dimitris Paparas, Sergei Vassilvitskii, Sanmi Koyejo
Abstract: Scaling laws provide important insights that can guide the design of large language models (LLMs). Existing work has primarily focused on studying scaling laws for pretraining (upstream) loss. However, in transfer learning settings, in which LLMs are pretrained on an unsupervised dataset and then finetuned on a downstream task, we often also care about the downstream performance. In this work, we study the scaling behavior in a transfer learning setting, where LLMs are finetuned for machine translation tasks. Specifically, we investigate how the choice of the pretraining data and its size affect downstream performance (translation quality) as judged by: downstream cross-entropy and translation quality metrics such as BLEU and COMET scores. Our experiments indicate that the size of the finetuning dataset and the distribution alignment between the pretraining and downstream data significantly influence the scaling behavior. With sufficient alignment, both downstream cross-entropy and translation quality scores improve monotonically with more pretraining data. In such cases, we show that it is possible to predict the downstream translation quality metrics with good accuracy using a log-law. However, there are cases where moderate misalignment causes the downstream translation scores to fluctuate or get worse with more pretraining, whereas downstream cross-entropy monotonically improves. By analyzing these, we provide new practical insights for choosing appropriate pretraining data.
Authors: Zhenyu Pan, Haozheng Luo, Manling Li, Han Liu
Abstract: We present a Chain-of-Action (CoA) framework for multimodal and retrieval-augmented Question-Answering (QA). Compared to the literature, CoA overcomes two major challenges of current QA applications: (i) unfaithful hallucination that is inconsistent with real-time or domain facts and (ii) weak reasoning performance over compositional information. Our key contribution is a novel reasoning-retrieval mechanism that decomposes a complex question into a reasoning chain via systematic prompting and pre-designed actions. Methodologically, we propose three types of domain-adaptable `Plug-and-Play' actions for retrieving real-time information from heterogeneous sources. We also propose a multi-reference faith score (MRFS) to verify and resolve conflicts in the answers. Empirically, we exploit both public benchmarks and a Web3 case study to demonstrate the capability of CoA over other methods.
Authors: Diyi Yang, Dirk Hovy, David Jurgens, Barbara Plank
Abstract: Language technologies have made enormous progress, especially with the introduction of large language models (LLMs). On traditional tasks such as machine translation and sentiment analysis, these models perform at near-human level. These advances can, however, exacerbate a variety of issues that models have traditionally struggled with, such as bias, evaluation, and risks. In this position paper, we argue that many of these issues share a common core: a lack of awareness of the factors, context, and implications of the social environment in which NLP operates, which we call social awareness. While NLP is getting better at solving the formal linguistic aspects, limited progress has been made in adding the social awareness required for language applications to work in all situations for all users. Integrating social awareness into NLP models will make applications more natural, helpful, and safe, and will open up new possibilities. Thus we argue that substantial challenges remain for NLP to develop social awareness and that we are just at the beginning of a new era for the field.
Authors: Dyah Adila, Changho Shin, Yijing Zhang, Frederic Sala
Abstract: Aligning pretrained language models (LMs) often requires large-scale preference data and substantial computational resources. These costs become even more prohibitive for multi-objective or pluralistic alignment. Is this truly necessary? Can we perform efficient alignment using only internal model capabilities, and without additional training? To answer this question, we propose AlignEZ, a novel approach that leverages (1) self-generated preference data and (2) representation editing to achieve cost-effective, efficient alignment. By operating directly on learned representations, AlignEZ independently targets different behavioral aspects without the overhead of traditional alignment methods. Our experiments reveal that this cost-efficient procedure improves performance across diverse tasks: up to 19.9% on general alignment and 1.9% on challenging mathematical reasoning tasks, even when starting from a strong base model. AlignEZ can also align models to multiple objectives simultaneously, granting fine-grained control over multiple preference axes. Finally, we show that AlignEZ can accelerate more expensive alignment procedures--such as DPO--even under limited availability of ground-truth preference data.
Authors: Mehrdad Khatir, Sanchit Kabara, Chandan K. Reddy
Abstract: This paper shifts focus to the often-overlooked input embeddings - the initial representations fed into transformer blocks. Using fuzzy graph, k-nearest neighbor (k-NN), and community detection, we analyze embeddings from diverse LLMs, finding significant categorical community structure aligned with predefined concepts and categories aligned with humans. We observe these groupings exhibit within-cluster organization (such as hierarchies, topological ordering, etc.), hypothesizing a fundamental structure that precedes contextual processing. To further investigate the conceptual nature of these groupings, we explore cross-model alignments across different LLM categories within their input embeddings, observing a medium to high degree of alignment. Furthermore, provide evidence that manipulating these groupings can play a functional role in mitigating ethnicity bias in LLM tasks.
Authors: Riya Sawhney, Samrat Yadav, Indrajit Bhattacharya, Mausam
Abstract: Real-world applications of KBQA require models to handle unanswerable questions with a limited volume of in-domain labeled training data. We propose the novel task of few-shot transfer for KBQA with unanswerable questions and contribute two new datasets for performance evaluation. We present FUn-FuSIC - a novel solution for our task that extends FuSIC KBQA, the state-of-the-art few-shot transfer model for answerable-only KBQA. We first note that FuSIC-KBQA's iterative repair makes a strong assumption that all questions are unanswerable. As a remedy, we propose Feedback for Unanswerability (FUn), which uses iterative repair using feedback from a suite of strong and weak verifiers, and an adaptation of self consistency for unanswerabilty to better assess the answerability of a question. Our experiments show that FUn-FuSIC significantly outperforms suitable adaptations of multiple LLM based and supervised SoTA models on our task, while establishing a new SoTA for answerable few-shot transfer as well.
Authors: Amalie Brogaard Pauli, Isabelle Augenstein, Ira Assent
Abstract: We are exposed to much information trying to influence us, such as teaser messages, debates, politically framed news, and propaganda - all of which use persuasive language. With the recent interest in Large Language Models (LLMs), we study the ability of LLMs to produce persuasive text. As opposed to prior work which focuses on particular domains or types of persuasion, we conduct a general study across various domains to measure and benchmark to what degree LLMs produce persuasive language - both when explicitly instructed to rewrite text to be more or less persuasive and when only instructed to paraphrase. We construct the new dataset Persuasive-Pairs of pairs of a short text and its rewrite by an LLM to amplify or diminish persuasive language. We multi-annotate the pairs on a relative scale for persuasive language: a valuable resource in itself, and for training a regression model to score and benchmark persuasive language, including for new LLMs across domains. In our analysis, we find that different 'personas' in LLaMA3's system prompt change persuasive language substantially, even when only instructed to paraphrase.
Authors: Xuansheng Wu, Padmaja Pravin Saraf, Gyeonggeon Lee, Ehsan Latif, Ninghao Liu, Xiaoming Zhai
Abstract: Large language models (LLMs) have demonstrated strong potential in performing automatic scoring for constructed response assessments. While constructed responses graded by humans are usually based on given grading rubrics, the methods by which LLMs assign scores remain largely unclear. It is also uncertain how closely AI's scoring process mirrors that of humans or if it adheres to the same grading criteria. To address this gap, this paper uncovers the grading rubrics that LLMs used to score students' written responses to science tasks and their alignment with human scores. We also examine whether enhancing the alignments can improve scoring accuracy. Specifically, we prompt LLMs to generate analytic rubrics that they use to assign scores and study the alignment gap with human grading rubrics. Based on a series of experiments with various configurations of LLM settings, we reveal a notable alignment gap between human and LLM graders. While LLMs can adapt quickly to scoring tasks, they often resort to shortcuts, bypassing deeper logical reasoning expected in human grading. We found that incorporating high-quality analytical rubrics designed to reflect human grading logic can mitigate this gap and enhance LLMs' scoring accuracy. These results underscore the need for a nuanced approach when applying LLMs in science education and highlight the importance of aligning LLM outputs with human expectations to ensure efficient and accurate automatic scoring.
Authors: Yupeng Chang, Yi Chang, Yuan Wu
Abstract: Large language models (LLMs) have demonstrated remarkable proficiency across various natural language processing (NLP) tasks. However, adapting LLMs to downstream applications requires computationally intensive and memory-demanding fine-tuning procedures. To alleviate these burdens, parameter-efficient fine-tuning (PEFT) techniques have emerged as a promising approach to tailor LLMs with minimal computational overhead. While PEFT methods offer substantial advantages, they do not fully address the pervasive issue of bias propagation from pre-training data. This work introduces Bias-Alleviating Low-Rank Adaptation (BA-LoRA), a novel PEFT method designed to counteract bias inheritance. BA-LoRA incorporates three distinct regularization terms: (1) a consistency regularizer, (2) a diversity regularizer, and (3) a singular value decomposition regularizer. These regularizers aim to enhance the models' consistency, diversity, and generalization capabilities during fine-tuning. We conduct extensive experiments on natural language understanding (NLU) and natural language generation (NLG) tasks using prominent LLMs such as LLaMA, Mistral, and Gemma. The results demonstrate that BA-LoRA outperforms LoRA and its state-of-the-art variants. Moreover, our method effectively mitigates the adverse effects of pre-training bias, leading to more reliable and robust model outputs. The code is available at https://github.com/cyp-jlu-ai/BA-LoRA.
Authors: Haohan Yuan, Siu Cheung Hui, Haopeng Zhang
Abstract: Biomedical Event Extraction (BEE) is a challenging task that involves modeling complex relationships between fine-grained entities in biomedical text. BEE has traditionally been formulated as a classification problem. With recent advancements in large language models (LLMs), generation-based models that cast event extraction as a sequence generation problem have attracted attention in the NLP research community. However, current generative models often overlook cross-instance information in complex event structures, such as nested and overlapping events, which constitute over 20% of events in benchmark datasets. In this paper, we propose GenBEE, an event structure-aware generative model that captures complex event structures in biomedical text for biomedical event extraction. GenBEE constructs event prompts that distill knowledge from LLMs to incorporate both label semantics and argument dependency relationships. In addition, GenBEE generates prefixes with event structural prompts to incorporate structural features to improve the model's overall performance. We have evaluated the proposed GenBEE model on three widely used BEE benchmark datasets, namely MLEE, GE11, and PHEE. Experimental results show that GenBEE has achieved state-of-the-art performance on the MLEE and GE11 datasets, and achieved competitive results when compared to the state-of-the-art classification-based models on the PHEE dataset.
Authors: Chenlu Guo, Nuo Xu, Yi Chang, Yuan Wu
Abstract: With the rapid development of large language models (LLMs), assessing their performance on health-related inquiries has become increasingly essential. The use of these models in real-world contexts-where misinformation can lead to serious consequences for individuals seeking medical advice and support-necessitates a rigorous focus on safety and trustworthiness. In this work, we introduce CHBench, the first comprehensive safety-oriented Chinese health-related benchmark designed to evaluate LLMs' capabilities in understanding and addressing physical and mental health issues with a safety perspective across diverse scenarios. CHBench comprises 6,493 entries on mental health and 2,999 entries on physical health, spanning a wide range of topics. Our extensive evaluations of four popular Chinese LLMs highlight significant gaps in their capacity to deliver safe and accurate health information, underscoring the urgent need for further advancements in this critical domain. The code is available at https://github.com/TracyGuo2001/CHBench.
Authors: Diego Marcos, Robert van de Vlasakker, Ioannis N. Athanasiadis, Pierre Bonnet, Herv\'e Goeau, Alexis Joly, W. Daniel Kissling, C\'esar Leblanc, Andr\'e S. J. van Proosdij, Konstantinos P. Panousis
Abstract: Plant morphological traits, their observable characteristics, are fundamental to understand the role played by each species within their ecosystem. However, compiling trait information for even a moderate number of species is a demanding task that may take experts years to accomplish. At the same time, massive amounts of information about species descriptions is available online in the form of text, although the lack of structure makes this source of data impossible to use at scale. To overcome this, we propose to leverage recent advances in large language models (LLMs) and devise a mechanism for gathering and processing information on plant traits in the form of unstructured textual descriptions, without manual curation. We evaluate our approach by automatically replicating three manually created species-trait matrices. Our method managed to find values for over half of all species-trait pairs, with an F1-score of over 75%. Our results suggest that large-scale creation of structured trait databases from unstructured online text is currently feasible thanks to the information extraction capabilities of LLMs, being limited by the availability of textual descriptions covering all the traits of interest.
Authors: Yu Wang, Xinshuang Liu, Xiusi Chen, Sean O'Brien, Junda Wu, Julian McAuley
Abstract: Despite significant advancements in large language models (LLMs), the rapid and frequent integration of small-scale experiences, such as interactions with surrounding objects, remains a substantial challenge. Two critical factors in assimilating these experiences are (1) Efficacy: the ability to accurately remember recent events; (2) Retention: the capacity to recall long-past experiences. Current methods either embed experiences within model parameters using continual learning, model editing, or knowledge distillation techniques, which often struggle with rapid updates and complex interactions, or rely on external storage to achieve long-term retention, thereby increasing storage requirements. In this paper, we propose SELF-PARAM (Self-Updatable Large Language Models with Parameter Integration). SELF-PARAM requires no extra parameters while ensuring near-optimal efficacy and long-term retention. Our method employs a training objective that minimizes the Kullback-Leibler (KL) divergence between the predictions of an original model (with access to contextual information) and a target model (without such access). By generating diverse question-answer pairs related to the knowledge and minimizing the KL divergence across this dataset, we update the target model to internalize the knowledge seamlessly within its parameters. Evaluations on question-answering and conversational recommendation tasks demonstrate that SELF-PARAM significantly outperforms existing methods, even when accounting for non-zero storage requirements. This advancement paves the way for more efficient and scalable integration of experiences in large language models by embedding knowledge directly into model parameters.
Authors: Jing Luo, Longze Chen, Run Luo, Liang Zhu, Chang Ao, Jiaming Li, Yukun Chen, Xin Cheng, Wen Yang, Jiayuan Su, Ahmadreza Argha, Hamid Alinejad-Rokny, Chengming Li, Shiwen Ni, Min Yang
Abstract: While closed-source Large Language Models (LLMs) demonstrate strong mathematical problem-solving abilities, open-source models still face challenges with such tasks. To bridge this gap, we propose a data augmentation approach and introduce PersonaMathQA, a dataset derived from MATH and GSM8K, on which we train the PersonaMath models. Our approach consists of two stages: the first stage focuses on learning from Persona Diversification, and the second stage emphasizes learning from Reflection. In the first stage, we regenerate detailed chain-of-thought (CoT) solutions as instructions using a closed-source LLM and introduce a persona-driven data augmentation technique. This technique innovatively classifies personas based on occupations, significantly enhancing the dataset's diversity and quality. In the second stage, we incorporate reflection to fully leverage more challenging and valuable questions. Evaluation of our PersonaMath models on MATH and GSM8K reveals that the PersonaMath-7B model (based on Qwen2.5-7B) achieves an accuracy of 61.2% on MATH and 87.8% on GSM8K, surpassing all baseline methods and achieving state-of-the-art performance. Notably, our dataset contains only 128.9K data points-merely 32.6% of MetaMathQA and 49.5% of MathInstruct-yet our model outperforms these baselines, demonstrating the high quality and diversity of our dataset, which enables more efficient model training. We open-source the PersonaMathQA dataset, PersonaMath models, and our code for public usage.
Authors: Nathaniel Berger, Miriam Exel, Matthias Huck, Stefan Riezler
Abstract: Preference Optimization (PO) techniques are currently one of the state of the art techniques for fine-tuning large language models (LLMs) on pairwise preference feedback from human annotators. However, in machine translation, this sort of feedback can be difficult to solicit. Additionally, Kreutzer et al. (2018) have shown that, for machine translation, pairwise preferences are less reliable than other forms of human feedback, such as 5-point ratings. We examine post-edits to see if they can be a source of reliable human preferences by construction. In PO, a human annotator is shown sequences $s_1$ and $s_2$ and asked for a preference judgment, %$s_1 > s_2$; while for post-editing, editors create $s_1$ and know that it should be better than $s_2$. We attempt to use these implicit preferences for PO and show that it helps the model move towards post-edit-like hypotheses and away from machine translation-like hypotheses. Furthermore, we show that best results are obtained by pre-training the model with supervised fine-tuning (SFT) on post-edits in order to promote post-edit-like hypotheses to the top output ranks.
Authors: Stephanie Brandl, Oliver Eberle
Abstract: Instruction-tuned LLMs are able to provide an explanation about their output to users by generating self-explanations. These do not require gradient computations or the application of possibly complex XAI methods. In this paper, we analyse whether this ability results in a good explanation. We evaluate self-explanations in the form of input rationales with respect to their plausibility to humans as well as their faithfulness to models. We study two text classification tasks: sentiment classification and forced labour detection, i.e., identifying pre-defined risk indicators of forced labour. In addition to English, we include Danish and Italian translations of the sentiment classification task and compare self-explanations to human annotations for all samples. To allow for direct comparisons, we also compute post-hoc feature attribution, i.e., layer-wise relevance propagation (LRP) and analyse 4 LLMs. We show that self-explanations align more closely with human annotations compared to LRP, while maintaining a comparable level of faithfulness. This finding suggests that self-explanations indeed provide good explanations for text classification.
Authors: Nishchal Bhandari, Danny Chen, Miguel \'Angel del R\'io Fern\'andez, Natalie Delworth, Jennifer Drexler Fox, Mig\"uel Jett\'e, Quinten McNamara, Corey Miller, Ond\v{r}ej Novotn\'y, J\'an Profant, Nan Qin, Martin Ratajczak, Jean-Philippe Robichaud
Abstract: Today, we are open-sourcing our core speech recognition and diarization models for non-commercial use. We are releasing both a full production pipeline for developers as well as pared-down research models for experimentation. Rev hopes that these releases will spur research and innovation in the fast-moving domain of voice technology. The speech recognition models released today outperform all existing open source speech recognition models across a variety of long-form speech recognition domains.
Authors: Alimohammad Beigi, Bohan Jiang, Dawei Li, Zhen Tan, Pouya Shaeri, Tharindu Kumarage, Amrita Bhattacharjee, Huan Liu
Abstract: Traditional fact-checking relies on humans to formulate relevant and targeted fact-checking questions (FCQs), search for evidence, and verify the factuality of claims. While Large Language Models (LLMs) have been commonly used to automate evidence retrieval and factuality verification at scale, their effectiveness for fact-checking is hindered by the absence of FCQ formulation. To bridge this gap, we seek to answer two research questions: (1) Can LLMs generate relevant FCQs? (2) Can LLM-generated FCQs improve multimodal fact-checking? We therefore introduce a framework LRQ-FACT for using LLMs to generate relevant FCQs to facilitate evidence retrieval and enhance fact-checking by probing information across multiple modalities. Through extensive experiments, we verify if LRQ-FACT can generate relevant FCQs of different types and if LRQ-FACT can consistently outperform baseline methods in multimodal fact-checking. Further analysis illustrates how each component in LRQ-FACT works toward improving the fact-checking performance.
Authors: Utkarsh P
Abstract: India has 1369 languages of which 22 are official. About 13 different scripts are used to represent these languages. A Common Label Set (CLS) was developed based on phonetics to address the issue of large vocabulary of units required in the End-to-End (E2E) framework for multilingual synthesis. The Indian language text is first converted to CLS. This approach enables seamless code switching across 13 Indian languages and English in a given native speaker's voice, which corresponds to everyday speech in the Indian subcontinent, where the population is multilingual.
Authors: Yuhan Fu, Ruobing Xie, Jiazhen Liu, Bangxiang Lan, Xingwu Sun, Zhanhui Kang, Xirong Li
Abstract: Hallucinations in multimodal large language models (MLLMs) hinder their practical applications. To address this, we propose a Magnifier Prompt (MagPrompt), a simple yet effective method to tackle hallucinations in MLLMs via extremely simple instructions. MagPrompt is based on the following two key principles, which guide the design of various effective prompts, demonstrating robustness: (1) MLLMs should focus more on the image. (2) When there are conflicts between the image and the model's inner knowledge, MLLMs should prioritize the image. MagPrompt is training-free and can be applied to open-source and closed-source models, such as GPT-4o and Gemini-pro. It performs well across many datasets and its effectiveness is comparable or even better than more complex methods like VCD. Furthermore, our prompt design principles and experimental analyses provide valuable insights into multimodal hallucination.
Authors: Nikitas Theodoropoulos, Giorgos Filandrianos, Vassilis Lyberatos, Maria Lymperaiou, Giorgos Stamou
Abstract: We describe our contribution to the Strict and Strict-Small tracks of the 2nd iteration of the BabyLM Challenge. The shared task is centered around efficient pre-training given data constraints motivated by human development. In response, we study the effect of synthetic story data in language pre-training using TinyStories: a recently introduced dataset of short stories. Initially, we train GPT-Neo models on subsets of TinyStories, while varying the amount of available data. We find that, even with access to less than 100M words, the models are able to generate high-quality, original completions to a given story, and acquire substantial linguistic knowledge. To measure the effect of synthetic story data, we train LTG-BERT encoder models on a combined dataset of: a subset of TinyStories, story completions generated by GPT-Neo, and a subset of the BabyLM dataset. Our experimentation reveals that synthetic data can occasionally offer modest gains, but overall have a negative influence on linguistic understanding. Our work offers an initial study on synthesizing story data in low resource settings and underscores their potential for augmentation in data-constrained language modeling. We publicly release our models and implementation on our GitHub.
Authors: Suchir Salhan, Richard Diehl Martinez, Z\'ebulon Goriely, Paula Buttery
Abstract: Curriculum Learning has been a popular strategy to improve the cognitive plausibility of Small-Scale Language Models (SSLMs) in the BabyLM Challenge. However, it has not led to considerable improvements over non-curriculum models. We assess whether theoretical linguistic acquisition theories can be used to specify more fine-grained curriculum learning strategies, creating age-ordered corpora of Child-Directed Speech for four typologically distant language families to implement SSLMs and acquisition-inspired curricula cross-lingually. Comparing the success of three objective curricula (Growing, Inwards and MMM) that precisely replicate the predictions of acquisition theories on a standard SSLM architecture, we find fine-grained acquisition-inspired curricula can outperform non-curriculum baselines and performance benefits of curricula strategies in SSLMs can be derived by specifying fine-grained language-specific curricula that precisely replicate language acquisition theories.
Authors: Christopher Richardson, Roshan Sharma, Neeraj Gaur, Parisa Haghani, Anirudh Sundar, Bhuvana Ramabhadran
Abstract: Zero-shot domain adaptation for dialogue state tracking (DST) remains a challenging problem in task-oriented dialogue (TOD) systems, where models must generalize to target domains unseen at training time. Current large language model approaches for zero-shot domain adaptation rely on prompting to introduce knowledge pertaining to the target domains. However, their efficacy strongly depends on prompt engineering, as well as the zero-shot ability of the underlying language model. In this work, we devise a novel data augmentation approach, Schema Augmentation, that improves the zero-shot domain adaptation of language models through fine-tuning. Schema Augmentation is a simple but effective technique that enhances generalization by introducing variations of slot names within the schema provided in the prompt. Experiments on MultiWOZ and SpokenWOZ showed that the proposed approach resulted in a substantial improvement over the baseline, in some experiments achieving over a twofold accuracy gain over unseen domains while maintaining equal or superior performance over all domains.
Authors: Yiwen Ding, Zhiheng Xi, Wei He, Zhuoyuan Li, Yitao Zhai, Xiaowei Shi, Xunliang Cai, Tao Gui, Qi Zhang, Xuanjing Huang
Abstract: Self-improvement methods enable large language models (LLMs) to generate solutions themselves and iteratively train on filtered, high-quality rationales. This process proves effective and reduces the reliance on human supervision in LLMs' reasoning, but the performance soon plateaus. We delve into the process and find that models tend to over-sample on easy queries and under-sample on queries they have yet to master. As iterations proceed, this imbalance in sampling is exacerbated, leading to a long-tail distribution where solutions to difficult queries almost diminish. This phenomenon limits the performance gain of self-improving models. A straightforward solution is brute-force sampling to balance the distribution, which significantly raises computational costs. In this paper, we introduce Guided Self-Improvement (GSI), a strategy aimed at improving the efficiency of sampling challenging heavy-tailed data. It leverages Socratic-style guidance signals to help LLM reasoning with complex queries, reducing the exploration effort and minimizing computational overhead. Experiments on four models across diverse mathematical tasks show that GSI strikes a balance between performance and efficiency, while also being effective on held-out tasks.
Authors: Hyukhun Koh, Minha Jhang, Dohyung Kim, Sangmook Lee, Kyomin Jung
Abstract: Although auto-regressive models excel in natural language processing, they often struggle to generate diverse text and provide limited controllability. Non-auto-regressive methods could be an alternative but often produce degenerate outputs and exhibit shortcomings in conditional generation. To address these challenges, we propose Diffusion-EAGS, a novel framework that integrates conditional masked language models into diffusion language models through the theoretical lens of a conditional Markov Random Field. In doing so, we propose entropy-adaptive Gibbs sampling and entropy-based noise scheduling to counterbalance each model's shortcomings. Experimental results show that Diffusion-EAGS outperforms baselines and achieves the best quality-diversity tradeoff, demonstrating its effectiveness in non-autoregressive text generation.
Authors: Ali Marashian, Enora Rice, Luke Gessler, Alexis Palmer, Katharina von der Wense
Abstract: Many of the world's languages have insufficient data to train high-performing general neural machine translation (NMT) models, let alone domain-specific models, and often the only available parallel data are small amounts of religious texts. Hence, domain adaptation (DA) is a crucial issue faced by contemporary NMT and has, so far, been underexplored for low-resource languages. In this paper, we evaluate a set of methods from both low-resource NMT and DA in a realistic setting, in which we aim to translate between a high-resource and a low-resource language with access to only: a) parallel Bible data, b) a bilingual dictionary, and c) a monolingual target-domain corpus in the high-resource language. Our results show that the effectiveness of the tested methods varies, with the simplest one, DALI, being most effective. We follow up with a small human evaluation of DALI, which shows that there is still a need for more careful investigation of how to accomplish DA for low-resource NMT.
Authors: Ryota Nonomura, Hiroki Mori
Abstract: Multi-agent systems utilizing large language models (LLMs) have shown great promise in achieving natural dialogue. However, smooth dialogue control and autonomous decision making among agents still remain challenges. In this study, we focus on conversational norms such as adjacency pairs and turn-taking found in conversation analysis and propose a new framework called "Murder Mystery Agents" that applies these norms to AI agents' dialogue control. As an evaluation target, we employed the "Murder Mystery" game, a reasoning-type table-top role-playing game that requires complex social reasoning and information manipulation. In this game, players need to unravel the truth of the case based on fragmentary information through cooperation and bargaining. The proposed framework integrates next speaker selection based on adjacency pairs and a self-selection mechanism that takes agents' internal states into account to achieve more natural and strategic dialogue. To verify the effectiveness of this new approach, we analyzed utterances that led to dialogue breakdowns and conducted automatic evaluation using LLMs, as well as human evaluation using evaluation criteria developed for the Murder Mystery game. Experimental results showed that the implementation of the next speaker selection mechanism significantly reduced dialogue breakdowns and improved the ability of agents to share information and perform logical reasoning. The results of this study demonstrate that the systematics of turn-taking in human conversation are also effective in controlling dialogue among AI agents, and provide design guidelines for more advanced multi-agent dialogue systems.
Authors: Pusen Dong, Tianchen Zhu, Yue Qiu, Haoyi Zhou, Jianxin Li
Abstract: Safe reinforcement learning (RL) requires the agent to finish a given task while obeying specific constraints. Giving constraints in natural language form has great potential for practical scenarios due to its flexible transfer capability and accessibility. Previous safe RL methods with natural language constraints typically need to design cost functions manually for each constraint, which requires domain expertise and lacks flexibility. In this paper, we harness the dual role of text in this task, using it not only to provide constraint but also as a training signal. We introduce the Trajectory-level Textual Constraints Translator (TTCT) to replace the manually designed cost function. Our empirical results demonstrate that TTCT effectively comprehends textual constraint and trajectory, and the policies trained by TTCT can achieve a lower violation rate than the standard cost function. Extra studies are conducted to demonstrate that the TTCT has zero-shot transfer capability to adapt to constraint-shift environments.
Authors: Yuxuan Jiang, Francis Ferraro
Abstract: Recently, Large Language Models (LLMs) have shown impressive performance in character understanding tasks, such as analyzing the roles, personalities, and relationships of fictional characters. However, the extensive pre-training corpora used by LLMs raise concerns that they may rely on memorizing popular fictional works rather than genuinely understanding and reasoning about them. In this work, we argue that 'gist memory'-capturing essential meaning - should be the primary mechanism for character understanding tasks, as opposed to 'verbatim memory' - exact match of a string. We introduce a simple yet effective method to mitigate mechanized memorization in character understanding evaluations while preserving the essential implicit cues needed for comprehension and reasoning. Our approach reduces memorization-driven performance on popular fictional works from 96% accuracy to 72% and results in up to an 18% drop in accuracy across various character understanding tasks. These findings underscore the issue of data contamination in existing benchmarks, which often measure memorization rather than true character understanding.
Authors: Xiang Cheng, Raveesh Mayya, Jo\~ao Sedoc
Abstract: Unstructured text data annotation and analysis are fundamental to management research, often relying on human annotators through crowdsourcing platforms. While Large Language Models (LLMs) promise to provide a cost-effective and efficient alternative to human annotation, there lacks a systematic workflow that evaluate when LLMs are suitable or how to proceed with LLM-based text annotation in a reproducible manner. This paper addresses this methodological gap by introducing the ``SILICON" (Systematic Inference with LLMs for Information Classification and Notation) workflow. The workflow integrates established principles of human annotation with systematic prompt optimization and model selection, addressing challenges such as developing robust annotation guidelines, establishing high-quality human baselines, optimizing prompts, and ensuring reproducibility across LLMs. We validate the SILICON workflow through seven case studies covering common management research tasks. Our findings highlight the importance of validating annotation guideline agreement, the superiority of expert-developed human baselines over crowdsourced ones, the iterative nature of prompt optimization, and the necessity of testing multiple LLMs. We also find that LLMs agree well with expert annotations in most cases but show low agreement in more complex multi-label classification tasks. Notably, we propose a regression-based methodology to empirically compare LLM outputs across prompts and models. Our workflow advances management research by establishing rigorous, transparent, and reproducible processes for LLM-based annotation. We provide practical guidance for researchers to effectively navigate the evolving landscape of generative AI tools.
Authors: KuanChao Chu, Yi-Pei Chen, Hideki Nakayama
Abstract: Controlling diversity in LLM-agent world simulations is essential for maintaining stability in structured tasks while enabling variation where creativity is needed. However, we observe that dialogue diversity declines significantly over long-term simulation. To investigate the role of prompt design in conversational diversity, we modularized the utterance generation prompt and found that reducing the given information leads to more diverse outputs. Based on this insight, we propose Adaptive Prompt Pruning (APP), a novel method that allows users to control diversity through a single parameter, lambda. APP dynamically prunes the utterance generation prompt based on their attention weights and is compatible with traditional diversity control techniques. We demonstrate that APP effectively controls output diversity through extensive experiments, and propose a method to balance the control trade-offs. Additionally, we provide an in-depth analysis to offer insights into optimizing diversity control in multi-agent simulation.
Authors: Mayi Xu, Yunfeng Ning, Yongqi Li, Jianhao Chen, Jintao Wen, Yao Xiao, Shen Zhou, Birong Pan, Zepeng Bao, Xin Miao, Hankun Kang, Ke Sun, Tieyun Qian
Abstract: Reasoning is fundamental to human intelligence, and critical for problem-solving, decision-making, and critical thinking. Reasoning refers to drawing new conclusions based on existing knowledge, which can support various applications like clinical diagnosis, basic education, and financial analysis. Though a good number of surveys have been proposed for reviewing reasoning-related methods, none of them has systematically investigated these methods from the viewpoint of their dependent knowledge base. Both the scenarios to which the knowledge bases are applied and their storage formats are significantly different. Hence, investigating reasoning methods from the knowledge base perspective helps us better understand the challenges and future directions. To fill this gap, this paper first classifies the knowledge base into symbolic and parametric ones. The former explicitly stores information in human-readable symbols, and the latter implicitly encodes knowledge within parameters. Then, we provide a comprehensive overview of reasoning methods using symbolic knowledge bases, parametric knowledge bases, and both of them. Finally, we identify the future direction toward enhancing reasoning capabilities to bridge the gap between human and machine intelligence.
Authors: Aman Gupta, Shao Tang, Qingquan Song, Sirou Zhu, Jiwoo Hong, Ankan Saha, Viral Gupta, Noah Lee, Eunki Kim, Siyu Zhu, Parag Agrawal, Natesh Pillai, S. Sathiya Keerthi
Abstract: Reinforcement Learning with Human Feedback (RLHF) and its variants have made huge strides toward the effective alignment of large language models (LLMs) to follow instructions and reflect human values. More recently, Direct Alignment Algorithms (DAAs) have emerged in which the reward modeling stage of RLHF is skipped by characterizing the reward directly as a function of the policy being learned. Some popular examples of DAAs include Direct Preference Optimization (DPO) and Simple Preference Optimization (SimPO). These methods often suffer from likelihood displacement, a phenomenon by which the probabilities of preferred responses are often reduced undesirably. In this paper, we argue that, for DAAs the reward (function) shape matters. We introduce \textbf{AlphaPO}, a new DAA method that leverages an $\alpha$-parameter to help change the shape of the reward function beyond the standard log reward. AlphaPO helps maintain fine-grained control over likelihood displacement and over-optimization. Compared to SimPO, one of the best performing DAAs, AlphaPO leads to about 7\% to 10\% relative improvement in alignment performance for the instruct versions of Mistral-7B and Llama3-8B while achieving 15\% to 50\% relative improvement over DPO on the same models. The analysis and results presented highlight the importance of the reward shape, and how one can systematically change it to affect training dynamics, as well as improve alignment performance.
Authors: Dylan Gaines, Keith Vertanen
Abstract: Users of Augmentative and Alternative Communication (AAC) may write letter-by-letter via an interface that uses a character language model. However, most state-of-the-art large pretrained language models predict subword tokens of variable length. We investigate how to practically use such models to make accurate and efficient character predictions. We fine-tune models using a large dataset of sentences we curated in which each sentence is rated according to how useful it might be for spoken or written AAC communication. We find that using an algorithm to produce character predictions from a subword large language model provides more accurate predictions than adding a classification layer or using a byte-level model. We also find that our domain adaptation procedure is effective at improving model performance on simple, conversational text.
Authors: Santhosh Thottingal
Abstract: This paper introduces an approach to question answering over knowledge bases like Wikipedia and Wikidata by performing "question-to-question" matching and retrieval from a dense vector embedding store. Instead of embedding document content, we generate a comprehensive set of questions for each logical content unit using an instruction-tuned LLM. These questions are vector-embedded and stored, mapping to the corresponding content. Vector embedding of user queries are then matched against this question vector store. The highest similarity score leads to direct retrieval of the associated article content, eliminating the need for answer generation. Our method achieves high cosine similarity ( > 0.9 ) for relevant question pairs, enabling highly precise retrieval. This approach offers several advantages including computational efficiency, rapid response times, and increased scalability. We demonstrate its effectiveness on Wikipedia and Wikidata, including multimedia content through structured fact retrieval from Wikidata, opening up new pathways for multimodal question answering.
Authors: Viktor Moskvoretskii, Maria Lysyuk, Mikhail Salnikov, Nikolay Ivanov, Sergey Pletenev, Daria Galimzianova, Nikita Krayko, Vasily Konovalov, Irina Nikishina, Alexander Panchenko
Abstract: Retrieval Augmented Generation (RAG) improves correctness of Question Answering (QA) and addresses hallucinations in Large Language Models (LLMs), yet greatly increase computational costs. Besides, RAG is not always needed as may introduce irrelevant information. Recent adaptive retrieval methods integrate LLMs' intrinsic knowledge with external information appealing to LLM self-knowledge, but they often neglect efficiency evaluations and comparisons with uncertainty estimation techniques. We bridge this gap by conducting a comprehensive analysis of 35 adaptive retrieval methods, including 8 recent approaches and 27 uncertainty estimation techniques, across 6 datasets using 10 metrics for QA performance, self-knowledge, and efficiency. Our findings show that uncertainty estimation techniques often outperform complex pipelines in terms of efficiency and self-knowledge, while maintaining comparable QA performance.
Authors: Xiaofeng Wang, Yiming Wang, Wenhong Zhu, Rui Wang
Abstract: Geometric ability is a significant challenge for large language models (LLMs) due to the need for advanced spatial comprehension and abstract thinking. Existing datasets primarily evaluate LLMs on their final answers, but they cannot truly measure their true understanding of geometric structures, as LLMs can arrive at correct answers by coincidence. To fill this gap, we introduce the GeomRel dataset, designed to evaluate LLMs' understanding of geometric structures by isolating the core step of geometric relationship identification in problem-solving. Using this benchmark, we conduct thorough evaluations of diverse LLMs and identify key limitations in understanding geometric structures. We further propose the Geometry Chain-of-Thought (GeoCoT) method, which enhances LLMs' ability to identify geometric relationships, resulting in significant performance improvements.
Authors: Ziyao Xu, Houfeng Wang
Abstract: To achieve generalized and robust natural-to-formal language conversion (N2F), large language models (LLMs) need to have strong capabilities of decomposition and composition in N2F when faced with an unfamiliar formal language and be able to cope with compositional gaps and counter-intuitive symbolic names. To investigate whether LLMs have this set of basic capabilities in N2F, we propose the DEDC framework. This framework semi-automatically performs sample and task construction, allowing decoupled evaluation of the set of decomposition and composition capabilities of LLMs in N2F. Based on this framework, we evaluate and analyze the most advanced LLMs, and the main findings include that: (1) the LLMs are deficient in both decomposition and composition; (2) the LLMs show a wide coverage of error types that can be attributed to deficiencies in natural language understanding and the learning and use of symbolic systems; (3) compositional gaps and counter-intuitive symbolic names both affect the decomposition and composition of the LLMs. Our work provides a new perspective for investigating the basic capabilities of decomposition and composition of LLMs in N2F. The detailed analysis of deficiencies and attributions can help subsequent improvements of LLMs.
Authors: Ang Li, Yichuan Mo, Mingjie Li, Yifei Wang, Yisen Wang
Abstract: Large Language Models (LLMs) have demonstrated remarkable success across various NLP benchmarks. However, excelling in complex tasks that require nuanced reasoning and precise decision-making demands more than raw language proficiency--LLMs must reason, i.e., think logically, draw from past experiences, and synthesize information to reach conclusions and take action. To enhance reasoning abilities, approaches such as prompting and fine-tuning have been widely explored. While these methods have led to clear improvements in reasoning, their impact on LLM safety remains less understood. In this work, we investigate the interplay between reasoning and safety in LLMs. We highlight the latent safety risks that arise as reasoning capabilities improve, shedding light on previously overlooked vulnerabilities. At the same time, we explore how reasoning itself can be leveraged to enhance safety, uncovering potential mitigation strategies. By examining both the risks and opportunities in reasoning-driven LLM safety, our study provides valuable insights for developing models that are not only more capable but also more trustworthy in real-world deployments.
Authors: Shiza Ali, Jeremy Blackburn, Gianluca Stringhini
Abstract: The proliferation of social media platforms has led to an increase in the spread of hate speech, particularly targeting vulnerable communities. Unfortunately, existing methods for automatically identifying and blocking toxic language rely on pre-constructed lexicons, making them reactive rather than adaptive. As such, these approaches become less effective over time, especially when new communities are targeted with slurs not included in the original datasets. To address this issue, we present an adaptive approach that uses word embeddings to update lexicons and develop a hybrid model that adjusts to emerging slurs and new linguistic patterns. This approach can effectively detect toxic language, including intentional spelling mistakes employed by aggressors to avoid detection. Our hybrid model, which combines BERT with lexicon-based techniques, achieves an accuracy of 95% for most state-of-the-art datasets. Our work has significant implications for creating safer online environments by improving the detection of toxic content and proactively updating the lexicon. Content Warning: This paper contains examples of hate speech that may be triggering.
Authors: Shahriar Kabir Nahin, Rabindra Nath Nandi, Sagor Sarker, Quazi Sarwar Muhtaseem, Md Kowsher, Apu Chandraw Shill, Md Ibrahim, Mehadi Hasan Menon, Tareq Al Muntasir, Firoj Alam
Abstract: In this paper, we present TituLLMs, the first large pretrained Bangla LLMs, available in 1b and 3b parameter sizes. Due to computational constraints during both training and inference, we focused on smaller models. To train TituLLMs, we collected a pretraining dataset of approximately ~37 billion tokens. We extended the Llama-3.2 tokenizer to incorporate language- and culture-specific knowledge, which also enables faster training and inference. There was a lack of benchmarking datasets to benchmark LLMs for Bangla. To address this gap, we developed five benchmarking datasets. We benchmarked various LLMs, including TituLLMs, and demonstrated that TituLLMs outperforms its initial multilingual versions. However, this is not always the case, highlighting the complexities of language adaptation. Our work lays the groundwork for adapting existing multilingual open models to other low-resource languages. To facilitate broader adoption and further research, we have made the TituLLMs models and benchmarking datasets publicly available (https://huggingface.co/collections/hishab/titulm-llama-family-6718d31fc1b83529276f490a).
URLs: https://huggingface.co/collections/hishab/titulm-llama-family-6718d31fc1b83529276f490a).
Authors: Tian Jin, Ellie Y. Cheng, Zack Ankner, Nikunj Saunshi, Blake M. Elias, Amir Yazdanbakhsh, Jonathan Ragan-Kelley, Suvinay Subramanian, Michael Carbin
Abstract: Decoding with autoregressive large language models (LLMs) traditionally occurs sequentially, generating one token after another. An emerging line of work explored parallel decoding by identifying and simultaneously generating semantically independent chunks of LLM responses. However, these techniques rely on hand-crafted heuristics tied to syntactic structures like lists and paragraphs, making them rigid and imprecise. We present PASTA, a learning-based system that teaches LLMs to identify semantic independence and express parallel decoding opportunities in their own responses. At its core are PASTA-LANG and its interpreter: PASTA-LANG is an annotation language that enables LLMs to express semantic independence in their own responses; the language interpreter acts on these annotations to orchestrate parallel decoding on-the-fly at inference time. Through a two-stage finetuning process, we train LLMs to generate PASTA-LANG annotations that optimize both response quality and decoding speed. Evaluation on AlpacaEval, an instruction following benchmark, shows that our approach Pareto-dominates existing methods in terms of decoding speed and response quality; our results demonstrate geometric mean speedups ranging from 1.21x to 1.93x with corresponding quality changes of +2.2% to -7.1%, measured by length-controlled win rates against sequential decoding baseline.
Authors: Yuncheng Hua, Lizhen Qu, Zhuang Li, Hao Xue, Flora D. Salim, Gholamreza Haffari
Abstract: Alignment tuning is crucial for ensuring large language models (LLMs) behave ethically and helpfully. Current alignment approaches require high-quality annotations and significant training resources. This paper proposes a low-cost, tuning-free method using in-context learning (ICL) to enhance LLM alignment. Through an analysis of high-quality ICL demos, we identified style as a key factor influencing LLM alignment capabilities and explicitly restyled ICL exemplars based on this stylistic framework. Additionally, we combined the restyled demos to achieve a balance between the two conflicting aspects of LLM alignment--factuality and safety. We packaged the restyled examples as prompts to trigger few-shot learning, improving LLM alignment. Compared to the best baseline approach, with an average score of 5.00 as the maximum, our method achieves a maximum 0.10 increase on the Alpaca task (from 4.50 to 4.60), a 0.22 enhancement on the Just-eval benchmark (from 4.34 to 4.56), and a maximum improvement of 0.32 (from 3.53 to 3.85) on the MT-Bench dataset. We release the code and data at https://github.com/AnonymousCode-ComputerScience/RIDE.
URLs: https://github.com/AnonymousCode-ComputerScience/RIDE.
Authors: Jean Seo, Jaeyoon Kim, SungJoo Byun, Hyopil Shin
Abstract: The necessity of language-specific tokenizers intuitively appears crucial for effective natural language processing, yet empirical analyses on their significance and underlying reasons are lacking. This study explores how language-specific tokenizers influence the behavior of Large Language Models predominantly trained with English text data, through the case study of Korean. The research unfolds in two main stages: (1) the development of a Korean-specific extended tokenizer and (2) experiments to compare models with the basic tokenizer and the extended tokenizer through various Next Token Prediction tasks. Our in-depth analysis reveals that the extended tokenizer decreases confidence in incorrect predictions during generation and reduces cross-entropy in complex tasks, indicating a tendency to produce less nonsensical outputs. Consequently, the extended tokenizer provides stability during generation, potentially leading to higher performance in downstream tasks.
Authors: Weizhe Yuan, Jane Yu, Song Jiang, Karthik Padthe, Yang Li, Dong Wang, Ilia Kulikov, Kyunghyun Cho, Yuandong Tian, Jason E Weston, Xian Li
Abstract: Scaling reasoning capabilities beyond traditional domains such as math and coding is hindered by the lack of diverse and high-quality questions. To overcome this limitation, we introduce a scalable approach for generating diverse and challenging reasoning questions, accompanied by reference answers. We present NaturalReasoning, a comprehensive dataset comprising 2.8 million questions that span multiple domains, including STEM fields (e.g., Physics, Computer Science), Economics, Social Sciences, and more. We demonstrate the utility of the questions in NaturalReasoning through knowledge distillation experiments which show that NaturalReasoning can effectively elicit and transfer reasoning capabilities from a strong teacher model. Furthermore, we demonstrate that NaturalReasoning is also effective for unsupervised self-training using external reward models or self-rewarding.
Authors: Jian Wang, Yinpei Dai, Yichi Zhang, Ziqiao Ma, Wenjie Li, Joyce Chai
Abstract: Intelligent tutoring agents powered by large language models (LLMs) have been increasingly explored to deliver personalized guidance in areas such as language learning and science education. However, their capabilities in guiding users to solve complex real-world tasks remain underexplored. To address this limitation, in this work, we focus on coding tutoring, a challenging problem that requires tutors to proactively guide students toward completing predefined coding tasks. We propose a novel agent workflow, Trace-and-Verify (TRAVER), which combines knowledge tracing to estimate a student's knowledge state and turn-by-turn verification to ensure effective guidance toward task completion. We introduce DICT, an automatic evaluation protocol that assesses tutor agents holistically using controlled student simulation and code generation tests. Extensive experiments reveal the challenges of coding tutoring and demonstrate that TRAVER achieves a significantly higher success rate. Although we use code tutoring as an example in this paper, our results and findings can be extended beyond coding, providing valuable insights into advancing tutoring agents for a variety of tasks.
Authors: Yifei Xu, Tusher Chakraborty, Emre K{\i}c{\i}man, Bibek Aryal, Eduardo Rodrigues, Srinagesh Sharma, Roberto Estevao, Maria Angels de Luis Balaguer, Jessica Wolk, Rafael Padilha, Leonardo Nunes, Shobana Balakrishnan, Songwu Lu, Ranveer Chandra
Abstract: Fine-tuning large language models (LLMs) to align with user preferences is challenging due to the high cost of quality human annotations in Reinforcement Learning from Human Feedback (RLHF) and the generalizability limitations of AI Feedback. To address these challenges, we propose RLTHF, a human-AI hybrid framework that combines LLM-based initial alignment with selective human annotations to achieve full-human annotation alignment with minimal effort. RLTHF identifies hard-to-annotate samples mislabeled by LLMs using a reward model's reward distribution and iteratively enhances alignment by integrating strategic human corrections while leveraging LLM's correctly labeled samples. Evaluations on HH-RLHF and TL;DR datasets show that RLTHF reaches full-human annotation-level alignment with only 6-7% of the human annotation effort. Furthermore, models trained on RLTHF's curated datasets for downstream tasks outperform those trained on fully human-annotated datasets, underscoring the effectiveness of RLTHF's strategic data curation.
Authors: Haun Leung, ZiNan Wang
Abstract: It is popular lately to train large language models to be used as chat assistants, but in the conversation between the user and the chat assistant, there are prompts, require multi-turns between the chat assistant and the user. However, there are a number of issues with the multi-turns conversation: The response of the chat assistant is prone to errors and can't help users achieve their goals, and as the number of conversation turns increases, the probability of errors will also increase; It is difficult for chat assistant to generate responses with different processes based on actual needs for the same prompt; Chat assistant require the use of tools, but the current approach is not elegant and efficient, and the number of tool calls is limited. The main reason for these issues is that large language models don't have the thinking ability as a human, lack the reasoning ability and planning ability, and lack the ability to execute plans. To solve these issues, we propose a thinking method based on a built-in chain of thought: In the multi-turns conversation, for each user prompt, the large language model thinks based on elements such as chat history, thinking context, action calls, memory and knowledge, makes detailed reasoning and planning, and actions according to the plan. We also explored how the large language model enhances thinking ability through this thinking method: Collect training datasets according to the thinking method and fine tune the large language model through supervised learning; Train a consistency reward model and use it as a reward function to fine tune the large language model using reinforcement learning, and the reinforced large language model outputs according to this way of thinking. Our experimental results show that the reasoning ability and planning ability of the large language model are enhanced, and the issues in the multi-turns conversation are solved.
Authors: Peiwen Yuan, Chuyi Tan, Shaoxiong Feng, Yiwei Li, Xinglin Wang, Yueqi Zhang, Jiayi Shi, Boyuan Pan, Yao Hu, Kan Li
Abstract: Despite the rapid progress of large language models (LLMs), their length-controllable text generation (LCTG) ability remains below expectations, posing a major limitation for practical applications. Existing methods mainly focus on end-to-end training to reinforce adherence to length constraints. However, the lack of decomposition and targeted enhancement of LCTG sub-abilities restricts further progress. To bridge this gap, we conduct a bottom-up decomposition of LCTG sub-abilities with human patterns as reference and perform a detailed error analysis. On this basis, we propose MarkerGen, a simple-yet-effective plug-and-play approach that:(1) mitigates LLM fundamental deficiencies via external tool integration;(2) conducts explicit length modeling with dynamically inserted markers;(3) employs a three-stage generation scheme to better align length constraints while maintaining content quality. Comprehensive experiments demonstrate that MarkerGen significantly improves LCTG across various settings, exhibiting outstanding effectiveness and generalizability.
Authors: Minbeom Kim, Kang-il Lee, Seongho Joo, Hwaran Lee, Kyomin Jung
Abstract: Personalized alignments for individual users have been a long-standing goal in large language models (LLMs). We introduce Drift, a novel framework that personalizes LLMs at decoding time with implicit user preferences. Traditional Reinforcement Learning from Human Feedback (RLHF) requires thousands of annotated examples and expensive gradient updates. In contrast, Drift personalizes LLMs in a training-free manner, using only a few dozen examples to steer a frozen model through efficient preference modeling. Our approach models user preferences as a composition of predefined, interpretable attributes and aligns them at decoding time to enable personalized generation. Experiments on both a synthetic persona dataset (Perspective) and a real human-annotated dataset (PRISM) demonstrate that Drift significantly outperforms RLHF baselines while using only 50-100 examples. Our results and analysis show that Drift is both computationally efficient and interpretable.
Authors: Alan Dao (Gia Tuan Dao), Dinh Bach Vu
Abstract: Large Language Models (LLMs) have demonstrated impressive capabilities in language processing, yet they often struggle with tasks requiring genuine visual spatial reasoning. In this paper, we introduce a novel two-stage training framework designed to equip standard LLMs with visual reasoning abilities for maze navigation. First, we leverage Supervised Fine Tuning (SFT) on a curated dataset of tokenized maze representations to teach the model to predict step-by-step movement commands. Next, we apply Group Relative Policy Optimization (GRPO)-a technique used in DeepSeekR1-with a carefully crafted reward function to refine the model's sequential decision-making and encourage emergent chain-of-thought behaviors. Experimental results on synthetically generated mazes show that while a baseline model fails to navigate the maze, the SFT-trained model achieves 86% accuracy, and further GRPO fine-tuning boosts accuracy to 93%. Qualitative analyses reveal that GRPO fosters more robust and self-corrective reasoning, highlighting the potential of our approach to bridge the gap between language models and visual spatial tasks. These findings offer promising implications for applications in robotics, autonomous navigation, and other domains that require integrated visual and sequential reasoning.
Authors: Maryam Rahimi, Yadollah Yaghoobzadeh, Mohammad Reza Daliri
Abstract: Recent advances in artificial intelligence have given rise to large language models (LLMs) that not only achieve human-like performance but also share computational principles with the brain's language processing mechanisms. While previous research has primarily focused on aligning LLMs' internal representations with neural activity, we introduce a novel approach that leverages explainable AI (XAI) methods to forge deeper connections between the two domains. Using attribution methods, we quantified how preceding words contribute to an LLM's next-word predictions and employed these explanations to predict fMRI recordings from participants listening to the same narratives. Our findings demonstrate that attribution methods robustly predict brain activity across the language network, surpassing traditional internal representations in early language areas. This alignment is hierarchical: early-layer explanations correspond to the initial stages of language processing in the brain, while later layers align with more advanced stages. Moreover, the layers more influential on LLM next-word prediction$\unicode{x2014}$those with higher attribution scores$\unicode{x2014}$exhibited stronger alignment with neural activity. This work establishes a bidirectional bridge between AI and neuroscience. First, we demonstrate that attribution methods offer a powerful lens for investigating the neural mechanisms of language comprehension, revealing how meaning emerges from preceding context. Second, we propose using brain alignment as a metric to evaluate the validity of attribution methods, providing a framework for assessing their biological plausibility.
Authors: Thomas Vakili, Aron Henriksson, Hercules Dalianis
Abstract: Many sensitive domains -- such as the clinical domain -- lack widely available datasets due to privacy risks. The increasing generative capabilities of large language models (LLMs) have made synthetic datasets a viable path forward. In this study, we domain-adapt LLMs to the clinical domain and generate synthetic clinical texts that are machine-annotated with tags for personally identifiable information using capable encoder-based NER models. The synthetic corpora are then used to train synthetic NER models. The results show that training NER models using synthetic corpora incurs only a small drop in predictive performance. The limits of this process are investigated in a systematic ablation study -- using both Swedish and Spanish data. Our analysis shows that smaller datasets can be sufficient for domain-adapting LLMs for data synthesis. Instead, the effectiveness of this process is almost entirely contingent on the performance of the machine-annotating NER models trained using the original data.
Authors: Zujie Liang, Feng Wei, Wujiang Xu, Lin Chen, Yuxi Qian, Xinhui Wu
Abstract: Recent advancements in large language models (LLMs) have shown remarkable potential in automating machine learning tasks. However, existing LLM-based agents often struggle with low-diversity and suboptimal code generation. While recent work has introduced Monte Carlo Tree Search (MCTS) to address these issues, limitations persist in the quality and diversity of thoughts generated, as well as in the scalar value feedback mechanisms used for node selection. In this study, we introduce Introspective Monte Carlo Tree Search (I-MCTS), a novel approach that iteratively expands tree nodes through an introspective process that meticulously analyzes solutions and results from parent and sibling nodes. This facilitates a continuous refinement of the node in the search tree, thereby enhancing the overall decision-making process. Furthermore, we integrate a Large Language Model (LLM)-based value model to facilitate direct evaluation of each node's solution prior to conducting comprehensive computational rollouts. A hybrid rewarding mechanism is implemented to seamlessly transition the Q-value from LLM-estimated scores to actual performance scores. This allows higher-quality nodes to be traversed earlier. Applied to the various ML tasks, our approach demonstrates a 6% absolute improvement in performance compared to the strong open-source AutoML agents, showcasing its effectiveness in enhancing agentic AutoML systems. Resource available at https://github.com/jokieleung/I-MCTS
Authors: Yilei Jiang, Xinyan Gao, Tianshuo Peng, Yingshui Tan, Xiaoyong Zhu, Bo Zheng, Xiangyu Yue
Abstract: The integration of additional modalities increases the susceptibility of large vision-language models (LVLMs) to safety risks, such as jailbreak attacks, compared to their language-only counterparts. While existing research primarily focuses on post-hoc alignment techniques, the underlying safety mechanisms within LVLMs remain largely unexplored. In this work , we investigate whether LVLMs inherently encode safety-relevant signals within their internal activations during inference. Our findings reveal that LVLMs exhibit distinct activation patterns when processing unsafe prompts, which can be leveraged to detect and mitigate adversarial inputs without requiring extensive fine-tuning. Building on this insight, we introduce HiddenDetect, a novel tuning-free framework that harnesses internal model activations to enhance safety. Experimental results show that {HiddenDetect} surpasses state-of-the-art methods in detecting jailbreak attacks against LVLMs. By utilizing intrinsic safety-aware patterns, our method provides an efficient and scalable solution for strengthening LVLM robustness against multimodal threats. Our code will be released publicly at https://github.com/leigest519/HiddenDetect.
Authors: Sowmya S. Sundaram, Benjamin Solomon, Avani Khatri, Anisha Laumas, Purvesh Khatri, Mark A. Musen
Abstract: Metadata play a crucial role in ensuring the findability, accessibility, interoperability, and reusability of datasets. This paper investigates the potential of large language models (LLMs), specifically GPT-4, to improve adherence to metadata standards. We conducted experiments on 200 random data records describing human samples relating to lung cancer from the NCBI BioSample repository, evaluating GPT-4's ability to suggest edits for adherence to metadata standards. We computed the adherence accuracy of field name-field value pairs through a peer review process, and we observed a marginal average improvement in adherence to the standard data dictionary from 79% to 80% (p<0.5). We then prompted GPT-4 with domain information in the form of the textual descriptions of CEDAR templates and recorded a significant improvement to 97% from 79% (p<0.01). These results indicate that, while LLMs may not be able to correct legacy metadata to ensure satisfactory adherence to standards when unaided, they do show promise for use in automated metadata curation when integrated with a structured knowledge base
Authors: Junjielong Xu, Ying Fu, Shin Hwei Tan, Pinjia He
Abstract: Large language models (LLMs) have achieved decent results on automated program repair (APR). However, the next token prediction training objective of decoder-only LLMs (e.g., GPT-4) is misaligned with the masked span prediction objective of current infilling-style methods, which impedes LLMs from fully leveraging pre-trained knowledge for program repair. In addition, while some LLMs can locate and repair bugs in certain functions using the related artifacts (e.g., test cases), existing methods still depend on statement-level fault localization methods to provide a list of buggy hunks for repair. This restriction hinders LLMs from exploring potential patches beyond the given locations. In this paper, we investigate a new approach to adapt LLMs to program repair. Our core insight is that LLM's APR capability can be greatly improved by simply aligning the output to their training objective and allowing them to refine the whole program without first identifying faulty statements. Based on this insight, we designed D4C, a straightforward prompting framework for APR. D4C can repair 180 bugs correctly in Defects4J, with each patch being sampled only 10 times. This surpasses the SOTA APR methods with perfect fault localization by 10% and reduces the patch sampling number by 90%. Our findings reveal that (1) objective alignment is crucial for fully exploiting LLM's pre-trained capability, and (2) replacing the traditional localize-buggy-hunks-then-repair workflow with direct debugging is more effective for LLM-based APR methods. Thus, we believe this paper introduces a new mindset for harnessing LLMs in APR.
Authors: Raeid Saqur, Anastasis Kratsios, Florian Krach, Yannick Limmer, Jacob-Junqi Tian, John Willes, Blanka Horvath, Frank Rudzicz
Abstract: We propose MoE-F - a formalized mechanism for combining $N$ pre-trained Large Language Models (LLMs) for online time-series prediction by adaptively forecasting the best weighting of LLM predictions at every time step. Our mechanism leverages the conditional information in each expert's running performance to forecast the best combination of LLMs for predicting the time series in its next step. Diverging from static (learned) Mixture of Experts (MoE) methods, our approach employs time-adaptive stochastic filtering techniques to combine experts. By framing the expert selection problem as a finite state-space, continuous-time Hidden Markov model (HMM), we can leverage the Wohman-Shiryaev filter. Our approach first constructs N parallel filters corresponding to each of the $N$ individual LLMs. Each filter proposes its best combination of LLMs, given the information that they have access to. Subsequently, the N filter outputs are optimally aggregated to maximize their robust predictive power, and this update is computed efficiently via a closed-form expression, generating our ensemble predictor. Our contributions are: **(I)** the MoE-F plug-and-play filtering harness algorithm, **(II)** theoretical optimality guarantees of the proposed filtering-based gating algorithm (via optimality guarantees for its parallel Bayesian filtering and its robust aggregation steps), and **(III)** empirical evaluation and ablative results using state-of-the-art foundational and MoE LLMs on a real-world __Financial Market Movement__ task where MoE-F attains a remarkable 17\% absolute and 48.5\% relative F1 measure improvement over the next best performing individual LLM expert predicting short-horizon market movement based on streaming news. Further, we provide empirical evidence of substantial performance gains in applying MoE-F over specialized models in the long-horizon time-series forecasting domain.
Authors: Fangxu Yu, Lai Jiang, Haoqiang Kang, Shibo Hao, Lianhui Qin
Abstract: The ability to generate diverse solutions to a given problem is a hallmark of human creativity. This divergent reasoning is also crucial for machines, enhancing their robustness and enabling them to assist humans in many applications such as scientific discovery. However, existing approaches to multi-step reasoning with large language models (LLMs) have mostly focused only on reasoning accuracy, without further discovering more diverse valid solutions. For example, supervised fine-tuning can improve LLM reasoning quality, but requires extensive supervised data to capture the full range of possible solutions. Reward-maximization reinforcement learning aims to find limited highest-reward solutions while neglecting the solution diversity. To fill this gap, we propose Flow of Reasoning (FoR), an efficient diversity-seeking LLM finetuning method aimed at improving reasoning quality and diversity with minimal data. FoR formulates multi-step LLM reasoning as a Markovian flow on a DAG-structured reasoning graph. This formulation allows us to incorporate and adapt principled GFlowNet approaches, for finetuning LLMs to sample divergent paths with probabilities proportional to the (unnormalized) reward of target problems. Extensive experiments show that, with limited training examples (e.g., 15 examples), FoR enables the discovery of diverse, creative, high-quality solutions, greatly outperforming a wide range of existing inference and training methods across six challenging reasoning tasks, including BlocksWorld (embodied reasoning), Game24 (math puzzle solving), Rubik's Cube (spatial reasoning), 1D-ARC (abstraction reasoning), GSM8k (math reasoning), and ProntoQA (logical reasoning). Code is available at https://github.com/Yu-Fangxu/FoR.
Authors: Yusu Qian, Hanrong Ye, Jean-Philippe Fauconnier, Peter Grasch, Yinfei Yang, Zhe Gan
Abstract: We introduce MIA-Bench, a new benchmark designed to evaluate multimodal large language models (MLLMs) on their ability to strictly adhere to complex instructions. Our benchmark comprises a diverse set of 400 image-prompt pairs, each crafted to challenge the models' compliance with layered instructions in generating accurate responses that satisfy specific requested patterns. Evaluation results from a wide array of state-of-the-art MLLMs reveal significant variations in performance, highlighting areas for improvement in instruction fidelity. Additionally, we create extra training data and explore supervised fine-tuning to enhance the models' ability to strictly follow instructions without compromising performance on other tasks. We hope this benchmark not only serves as a tool for measuring MLLM adherence to instructions, but also guides future developments in MLLM training methods.
Authors: David Gimeno-G\'omez, Carlos-D. Mart\'inez-Hinarejos
Abstract: Recent advances in Audio-Visual Speech Recognition (AVSR) have led to unprecedented achievements in the field, improving the robustness of this type of system in adverse, noisy environments. In most cases, this task has been addressed through the design of models composed of two independent encoders, each dedicated to a specific modality. However, while recent works have explored unified audio-visual encoders, determining the optimal cross-modal architecture remains an ongoing challenge. Furthermore, such approaches often rely on models comprising vast amounts of parameters and high computational cost training processes. In this paper, we aim to bridge this research gap by introducing a novel audio-visual framework. Our proposed method constitutes, to the best of our knowledge, the first attempt to harness the flexibility and interpretability offered by encoder architectures, such as the Branchformer, in the design of parameter-efficient AVSR systems. To be more precise, the proposed framework consists of two steps: first, estimating audio- and video-only systems, and then designing a tailored audio-visual unified encoder based on the layer-level branch scores provided by the modality-specific models. Extensive experiments on English and Spanish AVSR benchmarks covering multiple data conditions and scenarios demonstrated the effectiveness of our proposed method. Even when trained on a moderate scale of data, our models achieve competitive word error rates (WER) of approximately 2.5\% for English and surpass existing approaches for Spanish, establishing a new benchmark with an average WER of around 9.1\%. These results reflect how our tailored AVSR system is able to reach state-of-the-art recognition rates while significantly reducing the model complexity w.r.t. the prevalent approach in the field. Code and pre-trained models are available at https://github.com/david-gimeno/tailored-avsr.
Authors: Alex Kim, Maximilian Muhn, Valeri Nikolaev
Abstract: We investigate whether large language models (LLMs) can successfully perform financial statement analysis in a way similar to a professional human analyst. We provide standardized and anonymous financial statements to GPT4 and instruct the model to analyze them to determine the direction of firms' future earnings. Even without narrative or industry-specific information, the LLM outperforms financial analysts in its ability to predict earnings changes directionally. The LLM exhibits a relative advantage over human analysts in situations when the analysts tend to struggle. Furthermore, we find that the prediction accuracy of the LLM is on par with a narrowly trained state-of-the-art ML model. LLM prediction does not stem from its training memory. Instead, we find that the LLM generates useful narrative insights about a company's future performance. Lastly, our trading strategies based on GPT's predictions yield a higher Sharpe ratio and alphas than strategies based on other models. Our results suggest that LLMs may take a central role in analysis and decision-making.
Authors: Rubing Chen, Xulu Zhang, Jiaxin Wu, Wenqi Fan, Xiao-Yong Wei, Qing Li
Abstract: This paper addresses the need for improved precision in existing knowledge-enhanced question-answering frameworks, specifically Retrieval-Augmented Generation (RAG) methods that primarily focus on enhancing recall. We propose a multi-layer knowledge pyramid approach within the RAG framework to achieve a better balance between precision and recall. The knowledge pyramid consists of three layers: Ontologies, Knowledge Graphs (KGs), and chunk-based raw text. We employ cross-layer augmentation techniques for comprehensive knowledge coverage and dynamic updates of the Ontology schema and instances. To ensure compactness, we utilize cross-layer filtering methods for knowledge condensation in KGs. Our approach, named PolyRAG, follows a waterfall model for retrieval, starting from the top of the pyramid and progressing down until a confident answer is obtained. We introduce two benchmarks for domain-specific knowledge retrieval, one in the academic domain and the other in the financial domain. The effectiveness of the methods has been validated through comprehensive experiments by outperforming 19 SOTA methods. An encouraging observation is that the proposed method has augmented the GPT-4, providing 395% F1 gain by improving its performance from 0.1636 to 0.8109.
Authors: Yuan Tian, Tianyi Zhang
Abstract: Recent advances in large language models (LLMs) have transformed software development by automatically generating code from natural language. Yet challenges remain in generating fully correct code that aligns with user intent. Our study reveals that LLMs tend to pay less attention to user prompts as more code tokens are generated. We hypothesize that this attention dilution issue is an important reason for code generation errors. To mitigate this issue, we propose Selective Prompt Anchoring (SPA) to guide code LLMs to pay more attention to user intent when generating code. We evaluate SPA using six base LLMs across six benchmarks. Our results demonstrate that SPA enhances Pass@1 by up to 12.9%, consistently outperforming SOTA code generation methods in all settings. Our code is available at https://github.com/magic-YuanTian/Selective-Prompt-Anchoring.
URLs: https://github.com/magic-YuanTian/Selective-Prompt-Anchoring.
Authors: Suhyeon Yoo, Khai N. Truong, Young-Ho Kim
Abstract: d/Deaf and hearing song-signers have become prevalent across video-sharing platforms, but translating songs into sign language remains cumbersome and inaccessible. Our formative study revealed the challenges song-signers face, including semantic, syntactic, expressive, and rhythmic considerations in translations. We present ELMI, an accessible song-signing tool that assists in translating lyrics into sign language. ELMI enables users to edit glosses line-by-line, with real-time synced lyric and music video snippets. Users can also chat with a large language model-driven AI to discuss meaning, glossing, emoting, and timing. Through an exploratory study with 13 song-signers, we examined how ELMI facilitates their workflows and how song-signers leverage and receive an LLM-driven chat for translation. Participants successfully adopted ELMI to song-signing, with active discussions throughout. They also reported improved confidence and independence in their translations, finding ELMI encouraging, constructive, and informative. We discuss research and design implications for accessible and culturally sensitive song-signing translation tools.
Authors: Yibo Zhong, Haoxiang Jiang, Lincan Li, Ryumei Nakada, Tianci Liu, Linjun Zhang, Huaxiu Yao, Haoyu Wang
Abstract: Fine-tuning pre-trained models often yields state-of-the-art performance but is computationally expensive when updating all parameters. Parameter-efficient fine-tuning (PEFT) methods, such as Low-Rank Adaptation (LoRA), address this by freezing pre-trained weights and introducing low-rank matrices. However, because LoRA relies on low-rank decomposition, it struggles to capture complex nonlinear dynamics and optimal optimization trajectories, resulting in a performance gap relative to full fine-tuning and inefficient parameter utilization. To overcome these issues, we propose NEAT, a nonlinear PEFT approach that employs a lightweight neural network to learn a nonlinear transformation of the pre-trained weights, thereby better approximating cumulative weight updates. Our theoretical analysis shows that NEAT achieves greater efficiency than LoRA while maintaining equivalent expressivity. Extensive experiments on four benchmarks and over twenty datasets demonstrate that NEAT significantly outperforms state-of-the-art baselines in both vision and text tasks.
Authors: Taneesh Gupta, Rahul Madhavan, Xuchao Zhang, Chetan Bansal, Saravan Rajmohan
Abstract: Direct Preference Optimization (DPO) has proven effective in aligning large language models with human preferences but is often constrained to pairwise comparisons -- overlooking additional positive and negative responses that are commonly available in real-world settings. We propose Simultaneous Weighted Preference Optimization (SWEPO), which incorporates multiple responses per query and prioritizes those that deviate most from the average reward. This deviation-based weighting focuses training on the most informative outliers, akin to a built-in curriculum. Theoretically, we prove that such multi-preference sampling lowers alignment bias, bounding the expected deviation from the true acceptable-response distribution at a rate of $\mathcal{O}(\tfrac{1}{\sqrt{k}})$. Empirically, SWEPO outperforms state-of-the-art baselines on the Ultra-Feedback dataset and demonstrates substantial improvements over DPO and InfoNCA, yielding boosts of up to $\sim 4$% on length-controlled win-rate on AlpacaEval.
Authors: Zijie Chen, Zhanchao Zhou, Yu Lu, Renjun Xu, Lili Pan, Zhenzhong Lan
Abstract: Solving NP-hard problems traditionally relies on heuristics, yet manually designing effective heuristics for complex problems remains a significant challenge. While recent advancements like FunSearch have shown that large language models (LLMs) can be integrated into evolutionary algorithms (EAs) for heuristic design, their potential is hindered by limitations in balancing exploitation and exploration. We introduce Quality-Uncertainty Balanced Evolution (QUBE), a novel approach that enhances LLM+EA methods by redefining the priority criterion within the FunSearch framework. QUBE employs the Quality-Uncertainty Trade-off Criterion (QUTC), based on our proposed Uncertainty-Inclusive Quality metric, to evaluate and guide the evolutionary process. Through extensive experiments on challenging NP-complete problems, QUBE demonstrates significant performance improvements over FunSearch and baseline methods. Our code are available at https://github.com/zzjchen/QUBE_code.
Authors: Hao Zheng, Xinyan Guan, Hao Kong, Jia Zheng, Weixiang Zhou, Hongyu Lin, Yaojie Lu, Ben He, Xianpei Han, Le Sun
Abstract: Automatically generating presentations from documents is a challenging task that requires accommodating content quality, visual appeal, and structural coherence. Existing methods primarily focus on improving and evaluating the content quality in isolation, overlooking visual appeal and structural coherence, which limits their practical applicability. To address these limitations, we propose PPTAgent, which comprehensively improves presentation generation through a two-stage, edit-based approach inspired by human workflows. PPTAgent first analyzes reference presentations to extract slide-level functional types and content schemas, then drafts an outline and iteratively generates editing actions based on selected reference slides to create new slides. To comprehensively evaluate the quality of generated presentations, we further introduce PPTEval, an evaluation framework that assesses presentations across three dimensions: Content, Design, and Coherence. Results demonstrate that PPTAgent significantly outperforms existing automatic presentation generation methods across all three dimensions.
Authors: Long Phan (Michael Pokorny), Alice Gatti (Michael Pokorny), Ziwen Han (Michael Pokorny), Nathaniel Li (Michael Pokorny), Josephina Hu (Michael Pokorny), Hugh Zhang (Michael Pokorny), Chen Bo Calvin Zhang (Michael Pokorny), Mohamed Shaaban (Michael Pokorny), John Ling (Michael Pokorny), Sean Shi (Michael Pokorny), Michael Choi (Michael Pokorny), Anish Agrawal (Michael Pokorny), Arnav Chopra (Michael Pokorny), Adam Khoja (Michael Pokorny), Ryan Kim (Michael Pokorny), Richard Ren (Michael Pokorny), Jason Hausenloy (Michael Pokorny), Oliver Zhang (Michael Pokorny), Mantas Mazeika (Michael Pokorny), Tung Nguyen (Michael Pokorny), Daron Anderson (Michael Pokorny), Imad Ali Shah (Michael Pokorny), Mikhail Doroshenko (Michael Pokorny), Alun Cennyth Stokes (Michael Pokorny), Mobeen Mahmood (Michael Pokorny), Jaeho Lee (Michael Pokorny), Oleksandr Pokutnyi (Michael Pokorny), Oleg Iskra (Michael Pokorny), Jessica P. Wang (Michael Pokorny), Robert Gerbicz (Michael Pokorny), John-Clark Levin (Michael Pokorny), Serguei Popov (Michael Pokorny), Fiona Feng (Michael Pokorny), Steven Y. Feng (Michael Pokorny), Haoran Zhao (Michael Pokorny), Michael Yu (Michael Pokorny), Varun Gangal (Michael Pokorny), Chelsea Zou (Michael Pokorny), Zihan Wang (Michael Pokorny), Mstyslav Kazakov (Michael Pokorny), Geoff Galgon (Michael Pokorny), Johannes Schmitt (Michael Pokorny), Alvaro Sanchez (Michael Pokorny), Yongki Lee (Michael Pokorny), Will Yeadon (Michael Pokorny), Scott Sauers (Michael Pokorny), Marc Roth (Michael Pokorny), Chidozie Agu (Michael Pokorny), S{\o}ren Riis (Michael Pokorny), Fabian Giska (Michael Pokorny), Saiteja Utpala (Michael Pokorny), Antrell Cheatom (Michael Pokorny), Zachary Giboney (Michael Pokorny), Gashaw M. Goshu (Michael Pokorny), Sarah-Jane Crowson (Michael Pokorny), Mohinder Maheshbhai Naiya (Michael Pokorny), Noah Burns (Michael Pokorny), Lennart Finke (Michael Pokorny), Zerui Cheng (Michael Pokorny), Hyunwoo Park (Michael Pokorny), Francesco Fournier-Facio (Michael Pokorny), Jennifer Zampese (Michael Pokorny), John B. Wydallis (Michael Pokorny), Ryan G. Hoerr (Michael Pokorny), Mark Nandor (Michael Pokorny), Tim Gehrunger (Michael Pokorny), Jiaqi Cai (Michael Pokorny), Ben McCarty (Michael Pokorny), Jungbae Nam (Michael Pokorny), Edwin Taylor (Michael Pokorny), Jun Jin (Michael Pokorny), Gautier Abou Loume (Michael Pokorny), Hangrui Cao (Michael Pokorny), Alexis C Garretson (Michael Pokorny), Damien Sileo (Michael Pokorny), Qiuyu Ren (Michael Pokorny), Doru Cojoc (Michael Pokorny), Pavel Arkhipov (Michael Pokorny), Usman Qazi (Michael Pokorny), Aras Bacho (Michael Pokorny), Lianghui Li (Michael Pokorny), Sumeet Motwani (Michael Pokorny), Christian Schroeder de Witt (Michael Pokorny), Alexei Kopylov (Michael Pokorny), Johannes Veith (Michael Pokorny), Eric Singer (Michael Pokorny), Paolo Rissone (Michael Pokorny), Jaehyeok Jin (Michael Pokorny), Jack Wei Lun Shi (Michael Pokorny), Chris G. Willcocks (Michael Pokorny), Ameya Prabhu (Michael Pokorny), Longke Tang (Michael Pokorny), Kevin Zhou (Michael Pokorny), Emily de Oliveira Santos (Michael Pokorny), Andrey Pupasov Maksimov (Michael Pokorny), Edward Vendrow (Michael Pokorny), Kengo Zenitani (Michael Pokorny), Joshua Robinson (Michael Pokorny), Aleksandar Mikov (Michael Pokorny), Julien Guillod (Michael Pokorny), Yuqi Li (Michael Pokorny), Ben Pageler (Michael Pokorny), Joshua Vendrow (Michael Pokorny), Vladyslav Kuchkin (Michael Pokorny), Pierre Marion (Michael Pokorny), Denis Efremov (Michael Pokorny), Jayson Lynch (Michael Pokorny), Kaiqu Liang (Michael Pokorny), Andrew Gritsevskiy (Michael Pokorny), Dakotah Martinez (Michael Pokorny), Nick Crispino (Michael Pokorny), Dimitri Zvonkine (Michael Pokorny), Natanael Wildner Fraga (Michael Pokorny), Saeed Soori (Michael Pokorny), Ori Press (Michael Pokorny), Henry Tang (Michael Pokorny), Julian Salazar (Michael Pokorny), Sean R. Green (Michael Pokorny), Lina Br\"ussel (Michael Pokorny), Moon Twayana (Michael Pokorny), Aymeric Dieuleveut (Michael Pokorny), T. Ryan Rogers (Michael Pokorny), Wenjin Zhang (Michael Pokorny), Ross Finocchio (Michael Pokorny), Bikun Li (Michael Pokorny), Jinzhou Yang (Michael Pokorny), Arun Rao (Michael Pokorny), Gabriel Loiseau (Michael Pokorny), Mikhail Kalinin (Michael Pokorny), Marco Lukas (Michael Pokorny), Ciprian Manolescu (Michael Pokorny), Nate Stambaugh (Michael Pokorny), Subrata Mishra (Michael Pokorny), Ariel Ghislain Kemogne Kamdoum (Michael Pokorny), Tad Hogg (Michael Pokorny), Alvin Jin (Michael Pokorny), Carlo Bosio (Michael Pokorny), Gongbo Sun (Michael Pokorny), Brian P Coppola (Michael Pokorny), Haline Heidinger (Michael Pokorny), Rafael Sayous (Michael Pokorny), Stefan Ivanov (Michael Pokorny), Joseph M Cavanagh (Michael Pokorny), Jiawei Shen (Michael Pokorny), Joseph Marvin Imperial (Michael Pokorny), Philippe Schwaller (Michael Pokorny), Shaipranesh Senthilkuma (Michael Pokorny), Andres M Bran (Michael Pokorny), Andres Algaba (Michael Pokorny), Brecht Verbeken (Michael Pokorny), Kelsey Van den Houte (Michael Pokorny), Lynn Van Der Sypt (Michael Pokorny), David Noever (Michael Pokorny), Lisa Schut (Michael Pokorny), Ilia Sucholutsky (Michael Pokorny), Evgenii Zheltonozhskii (Michael Pokorny), Qiaochu Yuan (Michael Pokorny), Derek Lim (Michael Pokorny), Richard Stanley (Michael Pokorny), Shankar Sivarajan (Michael Pokorny), Tong Yang (Michael Pokorny), John Maar (Michael Pokorny), Julian Wykowski (Michael Pokorny), Mart\'i Oller (Michael Pokorny), Jennifer Sandlin (Michael Pokorny), Anmol Sahu (Michael Pokorny), Cesare Giulio Ardito (Michael Pokorny), Yuzheng Hu (Michael Pokorny), Felipe Meneguitti Dias (Michael Pokorny), Tobias Kreiman (Michael Pokorny), Kaivalya Rawal (Michael Pokorny), Tobias Garcia Vilchis (Michael Pokorny), Yuexuan Zu (Michael Pokorny), Martin Lackner (Michael Pokorny), James Koppel (Michael Pokorny), Jeremy Nguyen (Michael Pokorny), Daniil S. Antonenko (Michael Pokorny), Steffi Chern (Michael Pokorny), Bingchen Zhao (Michael Pokorny), Pierrot Arsene (Michael Pokorny), Sergey Ivanov (Michael Pokorny), Rafa{\l} Po\'swiata (Michael Pokorny), Chenguang Wang (Michael Pokorny), Daofeng Li (Michael Pokorny), Donato Crisostomi (Michael Pokorny), Ali Dehghan (Michael Pokorny), Andrea Achilleos (Michael Pokorny), John Arnold Ambay (Michael Pokorny), Benjamin Myklebust (Michael Pokorny), Archan Sen (Michael Pokorny), David Perrella (Michael Pokorny), Nurdin Kaparov (Michael Pokorny), Mark H Inlow (Michael Pokorny), Allen Zang (Michael Pokorny), Kalyan Ramakrishnan (Michael Pokorny), Daniil Orel (Michael Pokorny), Vladislav Poritski (Michael Pokorny), Shalev Ben-David (Michael Pokorny), Zachary Berger (Michael Pokorny), Parker Whitfill (Michael Pokorny), Michael Foster (Michael Pokorny), Daniel Munro (Michael Pokorny), Linh Ho (Michael Pokorny), Dan Bar Hava (Michael Pokorny), Aleksey Kuchkin (Michael Pokorny), Robert Lauff (Michael Pokorny), David Holmes (Michael Pokorny), Frank Sommerhage (Michael Pokorny), Anji Zhang (Michael Pokorny), Richard Moat (Michael Pokorny), Keith Schneider (Michael Pokorny), Daniel Pyda (Michael Pokorny), Zakayo Kazibwe (Michael Pokorny), Mukhwinder Singh (Michael Pokorny), Don Clarke (Michael Pokorny), Dae Hyun Kim (Michael Pokorny), Sara Fish (Michael Pokorny), Veit Elser (Michael Pokorny), Victor Efren Guadarrama Vilchis (Michael Pokorny), Immo Klose (Michael Pokorny), Christoph Demian (Michael Pokorny), Ujjwala Anantheswaran (Michael Pokorny), Adam Zweiger (Michael Pokorny), Guglielmo Albani (Michael Pokorny), Jeffery Li (Michael Pokorny), Nicolas Daans (Michael Pokorny), Maksim Radionov (Michael Pokorny), V\'aclav Rozho\v{n} (Michael Pokorny), Vincent Ginis (Michael Pokorny), Ziqiao Ma (Michael Pokorny), Christian Stump (Michael Pokorny), Jacob Platnick (Michael Pokorny), Volodymyr Nevirkovets (Michael Pokorny), Luke Basler (Michael Pokorny), Marco Piccardo (Michael Pokorny), Niv Cohen (Michael Pokorny), Virendra Singh (Michael Pokorny), Josef Tkadlec (Michael Pokorny), Paul Rosu (Michael Pokorny), Alan Goldfarb (Michael Pokorny), Piotr Padlewski (Michael Pokorny), Stanislaw Barzowski (Michael Pokorny), Kyle Montgomery (Michael Pokorny), Aline Menezes (Michael Pokorny), Arkil Patel (Michael Pokorny), Zixuan Wang (Michael Pokorny), Jamie Tucker-Foltz (Michael Pokorny), Jack Stade (Michael Pokorny), Declan Grabb (Michael Pokorny), Tom Goertzen (Michael Pokorny), Fereshteh Kazemi (Michael Pokorny), Jeremiah Milbauer (Michael Pokorny), Abhishek Shukla (Michael Pokorny), Hossam Elgnainy (Michael Pokorny), Yan Carlos Leyva Labrador (Michael Pokorny), Hao He (Michael Pokorny), Ling Zhang (Michael Pokorny), Alan Givr\'e (Michael Pokorny), Hew Wolff (Michael Pokorny), G\"ozdenur Demir (Michael Pokorny), Muhammad Fayez Aziz (Michael Pokorny), Younesse Kaddar (Michael Pokorny), Ivar \"Angquist (Michael Pokorny), Yanxu Chen (Michael Pokorny), Elliott Thornley (Michael Pokorny), Robin Zhang (Michael Pokorny), Jiayi Pan (Michael Pokorny), Antonio Terpin (Michael Pokorny), Niklas Muennighoff (Michael Pokorny), Hailey Schoelkopf (Michael Pokorny), Eric Zheng (Michael Pokorny), Avishy Carmi (Michael Pokorny), Jainam Shah (Michael Pokorny), Ethan D. L. Brown (Michael Pokorny), Kelin Zhu (Michael Pokorny), Max Bartolo (Michael Pokorny), Richard Wheeler (Michael Pokorny), Andrew Ho (Michael Pokorny), Shaul Barkan (Michael Pokorny), Jiaqi Wang (Michael Pokorny), Martin Stehberger (Michael Pokorny), Egor Kretov (Michael Pokorny), Peter Bradshaw (Michael Pokorny), JP Heimonen (Michael Pokorny), Kaustubh Sridhar (Michael Pokorny), Zaki Hossain (Michael Pokorny), Ido Akov (Michael Pokorny), Yury Makarychev (Michael Pokorny), Joanna Tam (Michael Pokorny), Hieu Hoang (Michael Pokorny), David M. Cunningham (Michael Pokorny), Vladimir Goryachev (Michael Pokorny), Demosthenes Patramanis (Michael Pokorny), Michael Krause (Michael Pokorny), Andrew Redenti (Michael Pokorny), David Aldous (Michael Pokorny), Jesyin Lai (Michael Pokorny), Shannon Coleman (Michael Pokorny), Jiangnan Xu (Michael Pokorny), Sangwon Lee (Michael Pokorny), Ilias Magoulas (Michael Pokorny), Sandy Zhao (Michael Pokorny), Ning Tang (Michael Pokorny), Michael K. Cohen (Michael Pokorny), Micah Carroll (Michael Pokorny), Orr Paradise (Michael Pokorny), Jan Hendrik Kirchner (Michael Pokorny), Stefan Steinerberger (Michael Pokorny), Maksym Ovchynnikov (Michael Pokorny), Jason O. Matos (Michael Pokorny), Adithya Shenoy (Michael Pokorny), Michael Wang (Michael Pokorny), Yuzhou Nie (Michael Pokorny), Paolo Giordano (Michael Pokorny), Philipp Petersen (Michael Pokorny), Anna Sztyber-Betley (Michael Pokorny), Paolo Faraboschi (Michael Pokorny), Robin Riblet (Michael Pokorny), Jonathan Crozier (Michael Pokorny), Shiv Halasyamani (Michael Pokorny), Antonella Pinto (Michael Pokorny), Shreyas Verma (Michael Pokorny), Prashant Joshi (Michael Pokorny), Eli Meril (Michael Pokorny), Zheng-Xin Yong (Michael Pokorny), Allison Tee (Michael Pokorny), J\'er\'emy Andr\'eoletti (Michael Pokorny), Orion Weller (Michael Pokorny), Raghav Singhal (Michael Pokorny), Gang Zhang (Michael Pokorny), Alexander Ivanov (Michael Pokorny), Seri Khoury (Michael Pokorny), Nils Gustafsson (Michael Pokorny), Hamid Mostaghimi (Michael Pokorny), Kunvar Thaman (Michael Pokorny), Qijia Chen (Michael Pokorny), Tran Quoc Kh\'anh (Michael Pokorny), Jacob Loader (Michael Pokorny), Stefano Cavalleri (Michael Pokorny), Hannah Szlyk (Michael Pokorny), Zachary Brown (Michael Pokorny), Himanshu Narayan (Michael Pokorny), Jonathan Roberts (Michael Pokorny), William Alley (Michael Pokorny), Kunyang Sun (Michael Pokorny), Ryan Stendall (Michael Pokorny), Max Lamparth (Michael Pokorny), Anka Reuel (Michael Pokorny), Ting Wang (Michael Pokorny), Hanmeng Xu (Michael Pokorny), Pablo Hern\'andez-C\'amara (Michael Pokorny), Freddie Martin (Michael Pokorny), Thomas Preu (Michael Pokorny), Tomek Korbak (Michael Pokorny), Marcus Abramovitch (Michael Pokorny), Dominic Williamson (Michael Pokorny), Ida Bosio (Michael Pokorny), Ziye Chen (Michael Pokorny), Bir\'o B\'alint (Michael Pokorny), Eve J. Y. Lo (Michael Pokorny), Maria In\^es S. Nunes (Michael Pokorny), Yibo Jiang (Michael Pokorny), M Saiful Bari (Michael Pokorny), Peyman Kassani (Michael Pokorny), Zihao Wang (Michael Pokorny), Behzad Ansarinejad (Michael Pokorny), Yewen Sun (Michael Pokorny), Stephane Durand (Michael Pokorny), Guillaume Douville (Michael Pokorny), Daniel Tordera (Michael Pokorny), George Balabanian (Michael Pokorny), Earth Anderson (Michael Pokorny), Lynna Kvistad (Michael Pokorny), Alejandro Jos\'e Moyano (Michael Pokorny), Hsiaoyun Milliron (Michael Pokorny), Ahmad Sakor (Michael Pokorny), Murat Eron (Michael Pokorny), Isaac C. McAlister (Michael Pokorny), Andrew Favre D. O. (Michael Pokorny), Shailesh Shah (Michael Pokorny), Xiaoxiang Zhou (Michael Pokorny), Firuz Kamalov (Michael Pokorny), Ronald Clark (Michael Pokorny), Sherwin Abdoli (Michael Pokorny), Tim Santens (Michael Pokorny), Harrison K Wang (Michael Pokorny), Evan Chen (Michael Pokorny), Alessandro Tomasiello (Michael Pokorny), G. Bruno De Luca (Michael Pokorny), Shi-Zhuo Looi (Michael Pokorny), Vinh-Kha Le (Michael Pokorny), Noam Kolt (Michael Pokorny), Niels M\"undler (Michael Pokorny), Avi Semler (Michael Pokorny), Emma Rodman (Michael Pokorny), Jacob Drori (Michael Pokorny), Carl J Fossum (Michael Pokorny), Luk Gloor (Michael Pokorny), Milind Jagota (Michael Pokorny), Ronak Pradeep (Michael Pokorny), Honglu Fan (Michael Pokorny), Tej Shah (Michael Pokorny), Jonathan Eicher (Michael Pokorny), Michael Chen (Michael Pokorny), Kushal Thaman (Michael Pokorny), William Merrill (Michael Pokorny), Moritz Firsching (Michael Pokorny), Carter Harris (Michael Pokorny), Stefan Ciob\^ac\u{a} (Michael Pokorny), Jason Gross (Michael Pokorny), Rohan Pandey (Michael Pokorny), Ilya Gusev (Michael Pokorny), Adam Jones (Michael Pokorny), Shashank Agnihotri (Michael Pokorny), Pavel Zhelnov (Michael Pokorny), Siranut Usawasutsakorn (Michael Pokorny), Mohammadreza Mofayezi (Michael Pokorny), Alexander Piperski (Michael Pokorny), Marc Carauleanu (Michael Pokorny), David K. Zhang (Michael Pokorny), Kostiantyn Dobarskyi (Michael Pokorny), Dylan Ler (Michael Pokorny), Roman Leventov (Michael Pokorny), Ignat Soroko (Michael Pokorny), Thorben Jansen (Michael Pokorny), Scott Creighton (Michael Pokorny), Pascal Lauer (Michael Pokorny), Joshua Duersch (Michael Pokorny), Vage Taamazyan (Michael Pokorny), Dario Bezzi (Michael Pokorny), Wiktor Morak (Michael Pokorny), Wenjie Ma (Michael Pokorny), William Held (Michael Pokorny), Tran {\DJ}uc Huy (Michael Pokorny), Ruicheng Xian (Michael Pokorny), Armel Randy Zebaze (Michael Pokorny), Mohanad Mohamed (Michael Pokorny), Julian Noah Leser (Michael Pokorny), Michelle X Yuan (Michael Pokorny), Laila Yacar (Michael Pokorny), Johannes Lengler (Michael Pokorny), Katarzyna Olszewska (Michael Pokorny), Hossein Shahrtash (Michael Pokorny), Edson Oliveira (Michael Pokorny), Joseph W. Jackson (Michael Pokorny), Daniel Espinosa Gonzalez (Michael Pokorny), Andy Zou (Michael Pokorny), Muthu Chidambaram (Michael Pokorny), Timothy Manik (Michael Pokorny), Hector Haffenden (Michael Pokorny), Dashiell Stander (Michael Pokorny), Ali Dasouqi (Michael Pokorny), Alexander Shen (Michael Pokorny), Emilien Duc (Michael Pokorny), Bita Golshani (Michael Pokorny), David Stap (Michael Pokorny), Mikalai Uzhou (Michael Pokorny), Alina Borisovna Zhidkovskaya (Michael Pokorny), Lukas Lewark (Michael Pokorny), Miguel Orbegozo Rodriguez (Michael Pokorny), M\'aty\'as Vincze (Michael Pokorny), Dustin Wehr (Michael Pokorny), Colin Tang (Michael Pokorny), Shaun Phillips (Michael Pokorny), Fortuna Samuele (Michael Pokorny), Jiang Muzhen (Michael Pokorny), Fredrik Ekstr\"om (Michael Pokorny), Angela Hammon (Michael Pokorny), Oam Patel (Michael Pokorny), Faraz Farhidi (Michael Pokorny), George Medley (Michael Pokorny), Forough Mohammadzadeh (Michael Pokorny), Madellene Pe\~naflor (Michael Pokorny), Haile Kassahun (Michael Pokorny), Alena Friedrich (Michael Pokorny), Claire Sparrow (Michael Pokorny), Rayner Hernandez Perez (Michael Pokorny), Taom Sakal (Michael Pokorny), Omkar Dhamane (Michael Pokorny), Ali Khajegili Mirabadi (Michael Pokorny), Eric Hallman (Michael Pokorny), Kenchi Okutsu (Michael Pokorny), Mike Battaglia (Michael Pokorny), Mohammad Maghsoudimehrabani (Michael Pokorny), Alon Amit (Michael Pokorny), Dave Hulbert (Michael Pokorny), Roberto Pereira (Michael Pokorny), Simon Weber (Michael Pokorny), Handoko (Michael Pokorny), Anton Peristyy (Michael Pokorny), Stephen Malina (Michael Pokorny), Samuel Albanie (Michael Pokorny), Will Cai (Michael Pokorny), Mustafa Mehkary (Michael Pokorny), Rami Aly (Michael Pokorny), Frank Reidegeld (Michael Pokorny), Anna-Katharina Dick (Michael Pokorny), Cary Friday (Michael Pokorny), Jasdeep Sidhu (Michael Pokorny), Hassan Shapourian (Michael Pokorny), Wanyoung Kim (Michael Pokorny), Mariana Costa (Michael Pokorny), Hubeyb Gurdogan (Michael Pokorny), Brian Weber (Michael Pokorny), Harsh Kumar (Michael Pokorny), Tong Jiang (Michael Pokorny), Arunim Agarwal (Michael Pokorny), Chiara Ceconello (Michael Pokorny), Warren S. Vaz (Michael Pokorny), Chao Zhuang (Michael Pokorny), Haon Park (Michael Pokorny), Andrew R. Tawfeek (Michael Pokorny), Daattavya Aggarwal (Michael Pokorny), Michael Kirchhof (Michael Pokorny), Linjie Dai (Michael Pokorny), Evan Kim (Michael Pokorny), Johan Ferret (Michael Pokorny), Yuzhou Wang (Michael Pokorny), Minghao Yan (Michael Pokorny), Krzysztof Burdzy (Michael Pokorny), Lixin Zhang (Michael Pokorny), Antonio Franca (Michael Pokorny), Diana T. Pham (Michael Pokorny), Kang Yong Loh (Michael Pokorny), Joshua Robinson (Michael Pokorny), Abram Jackson (Michael Pokorny), Shreen Gul (Michael Pokorny), Gunjan Chhablani (Michael Pokorny), Zhehang Du (Michael Pokorny), Adrian Cosma (Michael Pokorny), Jesus Colino (Michael Pokorny), Colin White (Michael Pokorny), Jacob Votava (Michael Pokorny), Vladimir Vinnikov (Michael Pokorny), Ethan Delaney (Michael Pokorny), Petr Spelda (Michael Pokorny), Vit Stritecky (Michael Pokorny), Syed M. Shahid (Michael Pokorny), Jean-Christophe Mourrat (Michael Pokorny), Lavr Vetoshkin (Michael Pokorny), Koen Sponselee (Michael Pokorny), Renas Bacho (Michael Pokorny), Florencia de la Rosa (Michael Pokorny), Xiuyu Li (Michael Pokorny), Guillaume Malod (Michael Pokorny), Leon Lang (Michael Pokorny), Julien Laurendeau (Michael Pokorny), Dmitry Kazakov (Michael Pokorny), Fatimah Adesanya (Michael Pokorny), Julien Portier (Michael Pokorny), Lawrence Hollom (Michael Pokorny), Victor Souza (Michael Pokorny), Yuchen Anna Zhou (Michael Pokorny), Julien Degorre (Michael Pokorny), Yi\u{g}it Yal{\i}n (Michael Pokorny), Gbenga Daniel Obikoya (Michael Pokorny), Luca Arnaboldi (Michael Pokorny), Rai (Michael Pokorny), Filippo Bigi (Quinn), M. C. Bosc\'a (Quinn), Oleg Shumar (Quinn), Kaniuar Bacho (Quinn), Pierre Clavier (Quinn), Gabriel Recchia (Quinn), Mara Popescu (Quinn), Nikita Shulga (Quinn), Ngefor Mildred Tanwie (Quinn), Denis Peskoff (Quinn), Thomas C. H. Lux (Quinn), Ben Rank (Quinn), Colin Ni (Quinn), Matthew Brooks (Quinn), Alesia Yakimchyk (Quinn), Huanxu (Quinn), Liu (Tony), Olle H\"aggstr\"om (Tony), Emil Verkama (Tony), Hans Gundlach (Tony), Leonor Brito-Santana (Tony), Brian Amaro (Tony), Vivek Vajipey (Tony), Rynaa Grover (Tony), Yiyang Fan (Tony), Gabriel Poesia Reis e Silva (Tony), Linwei Xin (Tony), Yosi Kratish (Tony), Jakub {\L}ucki (Tony), Wen-Ding Li (Tony), Sivakanth Gopi (Tony), Andrea Caciolai (Tony), Justin Xu (Tony), Kevin Joseph Scaria (Tony), Freddie Vargus (Tony), Farzad Habibi (Tony), Long (Tony), Lian, Emanuele Rodol\`a, Jules Robins, Vincent Cheng, Tony Fruhauff, Brad Raynor, Hao Qi, Xi Jiang, Ben Segev, Jingxuan Fan, Sarah Martinson, Erik Y. Wang, Kaylie Hausknecht, Michael P. Brenner, Mao Mao, Xinyu Zhang, David Avagian, Eshawn Jessica Scipio, Alon Ragoler, Justin Tan, Blake Sims, Rebeka Plecnik, Aaron Kirtland, Omer Faruk Bodur, D. P. Shinde, Zahra Adoul, Mohamed Zekry, Ali Karakoc, Tania C. B. Santos, Samir Shamseldeen, Loukmane Karim, Anna Liakhovitskaia, Nate Resman, Nicholas Farina, Juan Carlos Gonzalez, Gabe Maayan, Sarah Hoback, Rodrigo De Oliveira Pena, Glen Sherman, Elizabeth Kelley, Hodjat Mariji, Rasoul Pouriamanesh, Wentao Wu, Sandra Mendoza, Ismail Alarab, Joshua Cole, Danyelle Ferreira, Bryan Johnson, Mohammad Safdari, Liangti Dai, Siriphan Arthornthurasuk, Alexey Pronin, Jing Fan, Angel Ramirez-Trinidad, Ashley Cartwright, Daphiny Pottmaier, Omid Taheri, David Outevsky, Stanley Stepanic, Samuel Perry, Luke Askew, Ra\'ul Adri\'an Huerta Rodr\'iguez, Ali M. R. Minissi, Sam Ali, Ricardo Lorena, Krishnamurthy Iyer, Arshad Anil Fasiludeen, Sk Md Salauddin, Murat Islam, Juan Gonzalez, Josh Ducey, Maja Somrak, Vasilios Mavroudis, Eric Vergo, Juehang Qin, Benj\'amin Borb\'as, Eric Chu, Jack Lindsey, Anil Radhakrishnan, Antoine Jallon, I. M. J. McInnis, Pawan Kumar, Laxman Prasad Goswami, Daniel Bugas, Nasser Heydari, Ferenc Jeanplong, Archimedes Apronti, Abdallah Galal, Ng Ze-An, Ankit Singh, Joan of Arc Xavier, Kanu Priya Agarwal, Mohammed Berkani, Benedito Alves de Oliveira Junior, Dmitry Malishev, Nicolas Remy, Taylor D. Hartman, Tim Tarver, Stephen Mensah, Javier Gimenez, Roselynn Grace Montecillo, Russell Campbell, Asankhaya Sharma, Khalida Meer, Xavier Alapont, Deepakkumar Patil, Rajat Maheshwari, Abdelkader Dendane, Priti Shukla, Sergei Bogdanov, S\"oren M\"oller, Muhammad Rehan Siddiqi, Prajvi Saxena, Himanshu Gupta, Innocent Enyekwe, Ragavendran P V, Zienab EL-Wasif, Aleksandr Maksapetyan, Vivien Rossbach, Chris Harjadi, Mohsen Bahaloohoreh, Song Bian, John Lai, Justine Leon Uro, Greg Bateman, Mohamed Sayed, Ahmed Menshawy, Darling Duclosel, Yashaswini Jain, Ashley Aaron, Murat Tiryakioglu, Sheeshram Siddh, Keith Krenek, Alex Hoover, Joseph McGowan, Tejal Patwardhan, Summer Yue, Alexandr Wang, Dan Hendrycks
Abstract: Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 2,700 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai.
URLs: https://lastexam.ai.
Authors: Yibo Wang, Congying Xia, Wenting Zhao, Jiangshu Du, Chunyu Miao, Zhongfen Deng, Philip S. Yu, Chen Xing
Abstract: Unit test generation has become a promising and important use case of LLMs. However, existing evaluation benchmarks for assessing LLM unit test generation capabilities focus on function- or class-level code rather than more practical and challenging project-level codebases. To address such limitation, we propose ProjectTest, a project-level benchmark for unit test generation covering Python, Java, and JavaScript. ProjectTest features 20 moderate-sized and high-quality projects per language. We evaluate nine frontier LLMs on ProjectTest and the results show that all frontier LLMs tested exhibit moderate performance on ProjectTest on Python and Java, highlighting the difficulty of ProjectTest. We also conduct a thorough error analysis, which shows that even frontier LLMs, such as Claude-3.5-Sonnet, have significant basic yet critical errors, including compilation and cascade errors. Motivated by this observation, we further evaluate all frontier LLMs under manual error-fixing and self-error-fixing scenarios to assess their potential when equipped with error-fixing mechanisms. Our code and dataset is available at \href{https://github.com/YiboWANG214/ProjectTest}{ProjectTest}.
Authors: Fu-An Chao, Berlin Chen
Abstract: Prior efforts in building computer-assisted pronunciation training (CAPT) systems often treat automatic pronunciation assessment (APA) and mispronunciation detection and diagnosis (MDD) as separate fronts: the former aims to provide multiple pronunciation aspect scores across diverse linguistic levels, while the latter focuses instead on pinpointing the precise phonetic pronunciation errors made by non-native language learners. However, it is generally expected that a full-fledged CAPT system should perform both functionalities simultaneously and efficiently. In response to this surging demand, we in this work first propose HMamba, a novel CAPT approach that seamlessly integrates APA and MDD tasks in parallel. In addition, we introduce a novel loss function, decoupled cross-entropy loss (deXent), specifically tailored for MDD to facilitate better-supervised learning for detecting mispronounced phones, thereby enhancing overall performance. A comprehensive set of empirical results on the speechocean762 benchmark dataset demonstrates the effectiveness of our approach on APA. Notably, our proposed approach also yields a considerable improvement in MDD performance over a strong baseline, achieving an F1-score of 63.85%. Our codes are made available at https://github.com/Fuann/hmamba
Authors: Shengkun Tang, Oliver Sieberling, Eldar Kurtic, Zhiqiang Shen, Dan Alistarh
Abstract: Large Language Models (LLMs) have achieved significant success across various NLP tasks. However, their massive computational costs limit their widespread use, particularly in real-time applications. Structured pruning offers an effective solution by compressing models and directly providing end-to-end speed improvements, regardless of the hardware environment. Meanwhile, different components of the model exhibit varying sensitivities towards pruning, calling for \emph{non-uniform} model compression. However, a pruning method should not only identify a capable substructure, but also account for post-compression training. To this end, we propose \sysname, a method for \emph{training-aware} structured pruning. \sysname builds upon an evolutionary search process, generating multiple offspring models in each generation through mutation, and selecting the fittest for survival. To assess the effect of post-training, we incorporate a lightweight, multistep training process within the offspring population, progressively increasing the number of tokens and eliminating poorly performing models in each selection stage. We validate our method through extensive experiments on Llama-2-7B, Llama-3.1-8B and Qwen-2.5-14B-Instruct, achieving state-of-the-art performance for structured pruning. For instance, \sysname surpasses ShearedLlama while requiring $5\times$ less training data during post-compression training. Code is at: https://github.com/IST-DASLab/DarwinLM
Authors: Kai Yan, Zhan Ling, Kang Liu, Yifan Yang, Ting-Han Fan, Lingfeng Shen, Zhengyin Du, Jiecao Chen
Abstract: Inductive Reasoning (IR), the ability to summarize rules from examples and apply on new ones, has long been viewed as a primal ability for general intelligence and widely studied by cognitive science and AI researchers. Many benchmarks have been proposed to measure such ability for Large Language Models (LLMs); however, they focus on few-shot (usually $<$10) setting and lack evaluation for aggregating many pieces of information from long contexts. On the other hand, the ever-growing context length of LLMs have brought forth the novel paradigm of many-shot In-Context Learning (ICL), which addresses new tasks with hundreds to thousands of examples without expensive and inefficient fine-tuning. However, many-shot evaluations are mostly focused on classification (a very limited aspect of IR), and popular long-context LLM tasks such as Needle-In-A-Haystack (NIAH) seldom require complicated intelligence for integrating many pieces of information. To fix the issues from both worlds, we propose MIR-Bench, the first many-shot in-context inductive reasoning benchmark that asks LLM to induce output via input-output examples from underlying functions with diverse data format. Based on MIR-Bench, we study many novel problems for inductive reasoning and many-shot ICL, including robustness against erroneous shots and the effect of Chain-of-Thought (CoT), and acquired insightful findings.
Authors: Animesh Nighojkar, Bekhzodbek Moydinboyev, My Duong, John Licato
Abstract: In computational cognitive modeling, capturing the full spectrum of human judgment and decision-making processes, beyond just optimal behaviors, is a significant challenge. This study explores whether Large Language Models (LLMs) can emulate the breadth of human reasoning by predicting both intuitive, fast System 1 and deliberate, slow System 2 processes. We investigate the potential of AI to mimic diverse reasoning behaviors across a human population, addressing what we call the "full reasoning spectrum problem". We designed reasoning tasks using a novel generalization of the Natural Language Inference (NLI) format to evaluate LLMs' ability to replicate human reasoning. The questions were crafted to elicit both System 1 and System 2 responses. Human responses were collected through crowd-sourcing and the entire distribution was modeled, rather than just the majority of the answers. We used personality-based prompting inspired by the Big Five personality model to elicit AI responses reflecting specific personality traits, capturing the diversity of human reasoning, and exploring how personality traits influence LLM outputs. Combined with genetic algorithms to optimize the weighting of these prompts, this method was tested alongside traditional machine learning models. The results show that LLMs can mimic human response distributions, with open-source models like Llama and Mistral outperforming proprietary GPT models. Personality-based prompting, especially when optimized with genetic algorithms, significantly enhanced LLMs' ability to predict human response distributions, suggesting that capturing suboptimal, naturalistic reasoning may require modeling techniques incorporating diverse reasoning styles and psychological profiles. The study concludes that personality-based prompting combined with genetic algorithms is promising for enhancing AI's 'human-ness' in reasoning.