Authors: Yongheng Sun, Yueh Z. Lee, Genevieve A. Woodard, Hongtu Zhu, Chunfeng Lian, Mingxia Liu
Abstract: Radiology report generation is crucial in medical imaging,but the manual annotation process by physicians is time-consuming and labor-intensive, necessitating the develop-ment of automatic report generation methods. Existingresearch predominantly utilizes Transformers to generateradiology reports, which can be computationally intensive,limiting their use in real applications. In this work, we presentR2Gen-Mamba, a novel automatic radiology report genera-tion method that leverages the efficient sequence processingof the Mamba with the contextual benefits of Transformerarchitectures. Due to lower computational complexity ofMamba, R2Gen-Mamba not only enhances training and in-ference efficiency but also produces high-quality reports.Experimental results on two benchmark datasets with morethan 210,000 X-ray image-report pairs demonstrate the ef-fectiveness of R2Gen-Mamba regarding report quality andcomputational efficiency compared with several state-of-the-art methods. The source code can be accessed online.
Authors: Yang Zhenyuan, Liu Zhengliang, Zhang Jing, Lu Cen, Tai Jiaxin, Zhong Tianyang, Li Yiwei, Zhao Siyan, Yao Teng, Liu Qing, Yang Jinlin, Liu Qixin, Li Zhaowei, Wang Kexin, Ma Longjun, Zhu Dajiang, Ren Yudan, Ge Bao, Zhang Wei, Qiang Ning, Zhang Tuo, Liu Tianming
Abstract: This study examines the capabilities of advanced Large Language Models (LLMs), particularly the o1 model, in the context of literary analysis. The outputs of these models are compared directly to those produced by graduate-level human participants. By focusing on two Nobel Prize-winning short stories, 'Nine Chapters' by Han Kang, the 2024 laureate, and 'Friendship' by Jon Fosse, the 2023 laureate, the research explores the extent to which AI can engage with complex literary elements such as thematic analysis, intertextuality, cultural and historical contexts, linguistic and structural innovations, and character development. Given the Nobel Prize's prestige and its emphasis on cultural, historical, and linguistic richness, applying LLMs to these works provides a deeper understanding of both human and AI approaches to interpretation. The study uses qualitative and quantitative evaluations of coherence, creativity, and fidelity to the text, revealing the strengths and limitations of AI in tasks typically reserved for human expertise. While LLMs demonstrate strong analytical capabilities, particularly in structured tasks, they often fall short in emotional nuance and coherence, areas where human interpretation excels. This research underscores the potential for human-AI collaboration in the humanities, opening new opportunities in literary studies and beyond.
Authors: Chandra Irugalbandara
Abstract: Extending Large Language Models (LLMs) to advanced applications requires reliable structured output generation. Existing methods which often rely on rigid JSON schemas, can lead to unreliable outputs, diminished reasoning capabilities, and increased computational overhead, limiting LLMs' adaptability for complex tasks. We introduce Meaning Typed Prompting (MTP), a technique for efficient structured output generation that integrates types, meanings, and abstractions, such as variables and classes, into the prompting process. By utilizing expressive type definitions, MTP enhances output clarity and reduces dependence on complex abstractions, simplifying development, and improving implementation efficiency. This enables LLMs to understand relationships and generate structured data more effectively. Empirical evaluations on multiple benchmarks demonstrate that MTP outperforms existing frameworks in accuracy, reliability, consistency, and token efficiency. We present Semantix, a framework that implements MTP, providing practical insights into its application.
Authors: Nicholas Walker
Abstract: Causal decoder-only transformer models used for generative language modelling, such as Generative Pre-trained Transformers (GPT), are trained to predict the next token in a sequence based only on its previous tokens. Despite this simple training objective, they have proved to be powerful AI tools. However, only predicting the next token results in top layer embedding vectors that are highly token-focused. There may be benefits in generating embedding vectors at each token position that better capture the overall meaning of longer sequences of future text. Recent studies matching brain scans with deep language models suggest that humans also predict upcoming words when listening or reading but consider multiple future tokens rather than just one. This research investigates a new pretraining method called Future Token Prediction (FTP). In FTP, a large transformer encoder generates top layer embedding vectors for each token position, which, instead of being passed to a language head, are linearly and expansively projected to a pseudo-sequence, which is cross attended to by a small transformer decoder to predict the next N tokens forward from that position in the sequence. The top layer embedding vectors from FTP models exhibit distinct properties compared to those from standard GPT models, varying smoothly along a text sequence as measured by cosine similarity between adjacent tokens. Text generated by FTP models show improved topic coherence compared to standard GPT-like models trained with the same prediction perplexity for the next single token. The vectors are shown to better represent the topic of text based on the results of text classification examples. On a toy, but complex, coding problem, FTP networks produce significantly better results than GPT networks.
Authors: Samar M. Magdy, Fakhraddin Alwajih, Sang Yun Kwon, Reem Abdel-Salam, Muhammad Abdul-Mageed
Abstract: Writing has long been considered a hallmark of human intelligence and remains a pinnacle task for artificial intelligence (AI) due to the intricate cognitive processes involved. Recently, rapid advancements in generative AI, particularly through the development of Large Language Models (LLMs), have significantly transformed the landscape of writing assistance. However, underrepresented languages like Arabic encounter significant challenges in the development of advanced AI writing tools, largely due to the limited availability of data. This scarcity constrains the training of effective models, impeding the creation of sophisticated writing assistance technologies. To address these issues, we present Gazelle, a comprehensive dataset for Arabic writing assistance. In addition, we offer an evaluation framework designed to enhance Arabic writing assistance tools. Our human evaluation of leading LLMs, including GPT-4, GPT-4o, Cohere Command R+, and Gemini 1.5 Pro, highlights their respective strengths and limitations in addressing the challenges of Arabic writing. Our findings underscore the need for continuous model training and dataset enrichment to manage the complexities of Arabic language processing, paving the way for more effective AI-powered Arabic writing tools.
Authors: Chia-Hsuan Lee, Hao Cheng, Mari Ostendorf
Abstract: Large language models (LLMs) have demonstrated self-improvement capabilities via feedback and refinement, but current small language models (SLMs) have had limited success in this area. Existing correction approaches often rely on distilling knowledge from LLMs, which imposes significant computation demands. In this work, we introduce CORRECTIONLM, a novel correction framework that enables SLMs to self-correct using in-context exemplars without LLM involvement. Applied to two dialogue state tracking (DST) tasks in low-resource settings, CORRECTIONLM achieves results similar to a state-of-the-art LLM at a small fraction of the computation costs.
Authors: Samuele Poppi, Zheng-Xin Yong, Yifei He, Bobbie Chern, Han Zhao, Aobo Yang, Jianfeng Chi
Abstract: Recent advancements in Large Language Models (LLMs) have sparked widespread concerns about their safety. Recent work demonstrates that safety alignment of LLMs can be easily removed by fine-tuning with a few adversarially chosen instruction-following examples, i.e., fine-tuning attacks. We take a further step to understand fine-tuning attacks in multilingual LLMs. We first discover cross-lingual generalization of fine-tuning attacks: using a few adversarially chosen instruction-following examples in one language, multilingual LLMs can also be easily compromised (e.g., multilingual LLMs fail to refuse harmful prompts in other languages). Motivated by this finding, we hypothesize that safety-related information is language-agnostic and propose a new method termed Safety Information Localization (SIL) to identify the safety-related information in the model parameter space. Through SIL, we validate this hypothesis and find that only changing 20% of weight parameters in fine-tuning attacks can break safety alignment across all languages. Furthermore, we provide evidence to the alternative pathways hypothesis for why freezing safety-related parameters does not prevent fine-tuning attacks, and we demonstrate that our attack vector can still jailbreak LLMs adapted to new languages.
Authors: Katherine Howitt, Sathvik Nair, Allison Dods, Robert Melvin Hopkins
Abstract: Humans develop their grammars by making structural generalizations from finite input. We ask how filler-gap dependencies, which share a structural generalization despite diverse surface forms, might arise from the input. We explicitly control the input to a neural language model (NLM) to uncover whether the model posits a shared representation for filler-gap dependencies. We show that while NLMs do have success differentiating grammatical from ungrammatical filler-gap dependencies, they rely on superficial properties of the input, rather than on a shared generalization. Our work highlights the need for specific linguistic inductive biases to model language acquisition.
Authors: Ashish Khisti, M. Reza Ebrahimi, Hassan Dbouk, Arash Behboodi, Roland Memisevic, Christos Louizos
Abstract: We consider multi-draft speculative sampling, where the proposal sequences are sampled independently from different draft models. At each step, a token-level draft selection scheme takes a list of valid tokens as input and produces an output token whose distribution matches that of the target model. Previous works have demonstrated that the optimal scheme (which maximizes the probability of accepting one of the input tokens) can be cast as a solution to a linear program. In this work we show that the optimal scheme can be decomposed into a two-step solution: in the first step an importance sampling (IS) type scheme is used to select one intermediate token; in the second step (single-draft) speculative sampling is applied to generate the output token. For the case of two identical draft models we further 1) establish a necessary and sufficient condition on the distributions of the target and draft models for the acceptance probability to equal one and 2) provide an explicit expression for the optimal acceptance probability. Our theoretical analysis also motives a new class of token-level selection scheme based on weighted importance sampling. Our experimental results demonstrate consistent improvements in the achievable block efficiency and token rates over baseline schemes in a number of scenarios.
Authors: Cl\'ea Chataigner, Afaf Ta\"ik, Golnoosh Farnadi
Abstract: Large language models (LLMs) are increasingly used as alternatives to traditional search engines given their capacity to generate text that resembles human language. However, this shift is concerning, as LLMs often generate hallucinations, misleading or false information that appears highly credible. In this study, we explore the phenomenon of hallucinations across multiple languages in freeform text generation, focusing on what we call multilingual hallucination gaps. These gaps reflect differences in the frequency of hallucinated answers depending on the prompt and language used. To quantify such hallucinations, we used the FactScore metric and extended its framework to a multilingual setting. We conducted experiments using LLMs from the LLaMA, Qwen, and Aya families, generating biographies in 19 languages and comparing the results to Wikipedia pages. Our results reveal variations in hallucination rates, especially between high and low resource languages, raising important questions about LLM multilingual performance and the challenges in evaluating hallucinations in multilingual freeform text generation.
Authors: Shrenik Bhansali, Alwin Jin, Tyler Lizzo, Larry Heck
Abstract: Large language models (LLMs) are essential in natural language processing (NLP) but are costly in data collection, pre-training, fine-tuning, and inference. Task-specific small language models (SLMs) offer a cheaper alternative but lack robustness and generalization. This paper proposes LEGO, a novel technique to extract SLMs from an LLM and recombine them. Using state-of-the-art LLM pruning strategies, we can create task- and user-specific SLM building blocks that are efficient for fine-tuning and inference while also preserving user data privacy. LEGO utilizes Federated Learning and a novel aggregation scheme for the LLM reconstruction, maintaining robustness without high costs and preserving user data privacy. We experimentally demonstrate the versatility of LEGO, showing its ability to enable model heterogeneity and mitigate the effects of data heterogeneity while maintaining LLM robustness.
Authors: Samuel Aeschbach, Rui Mata, Dirk U. Wulff
Abstract: Accurately capturing individual differences in semantic networks is fundamental to advancing our mechanistic understanding of semantic memory. Past empirical attempts to construct individual-level semantic networks from behavioral paradigms may be limited by data constraints. To assess these limitations and propose improved designs for the measurement of individual semantic networks, we conducted a recovery simulation investigating the psychometric properties underlying estimates of individual semantic networks obtained from two different behavioral paradigms: free associations and relatedness judgment tasks. Our results show that successful inference of semantic networks is achievable, but they also highlight critical challenges. Estimates of absolute network characteristics are severely biased, such that comparisons between behavioral paradigms and different design configurations are often not meaningful. However, comparisons within a given paradigm and design configuration can be accurate and generalizable when based on designs with moderate numbers of cues, moderate numbers of responses, and cue sets including diverse words. Ultimately, our results provide insights that help evaluate past findings on the structure of semantic networks and design new studies capable of more reliably revealing individual differences in semantic networks.
Authors: Junyi Ye, Jingyi Gu, Xinyun Zhao, Wenpeng Yin, Guiling Wang
Abstract: The mathematical capabilities of AI systems are complex and multifaceted. Most existing research has predominantly focused on the correctness of AI-generated solutions to mathematical problems. In this work, we argue that beyond producing correct answers, AI systems should also be capable of, or assist humans in, developing novel solutions to mathematical challenges. This study explores the creative potential of Large Language Models (LLMs) in mathematical reasoning, an aspect that has received limited attention in prior research. We introduce a novel framework and benchmark, CreativeMath, which encompasses problems ranging from middle school curricula to Olympic-level competitions, designed to assess LLMs' ability to propose innovative solutions after some known solutions have been provided. Our experiments demonstrate that, while LLMs perform well on standard mathematical tasks, their capacity for creative problem-solving varies considerably. Notably, the Gemini-1.5-Pro model outperformed other LLMs in generating novel solutions. This research opens a new frontier in evaluating AI creativity, shedding light on both the strengths and limitations of LLMs in fostering mathematical innovation, and setting the stage for future developments in AI-assisted mathematical discovery.
Authors: Fengchen Liu, Jordan Jung, Wei Feinstein, Jeff DAmbrogia, Gary Jung
Abstract: This paper introduces a novel approach to enhancing closed-domain Question Answering (QA) systems, focusing on the specific needs of the Lawrence Berkeley National Laboratory (LBL) Science Information Technology (ScienceIT) domain. Utilizing a rich dataset derived from the ScienceIT documentation, our study embarks on a detailed comparison of two fine-tuned large language models and five retrieval-augmented generation (RAG) models. Through data processing techniques, we transform the documentation into structured context-question-answer triples, leveraging the latest Large Language Models (AWS Bedrock, GCP PaLM2, Meta LLaMA2, OpenAI GPT-4, Google Gemini-Pro) for data-driven insights. Additionally, we introduce the Aggregated Knowledge Model (AKM), which synthesizes responses from the seven models mentioned above using K-means clustering to select the most representative answers. The evaluation of these models across multiple metrics offers a comprehensive look into their effectiveness and suitability for the LBL ScienceIT environment. The results demonstrate the potential benefits of integrating fine-tuning and retrieval-augmented strategies, highlighting significant performance improvements achieved with the AKM. The insights gained from this study can be applied to develop specialized QA systems tailored to specific domains.
Authors: Sudhanshu Agrawal, Wonseok Jeon, Mingu Lee
Abstract: Speculative decoding is a powerful technique that attempts to circumvent the autoregressive constraint of modern Large Language Models (LLMs). The aim of speculative decoding techniques is to improve the average inference time of a large, target model without sacrificing its accuracy, by using a more efficient draft model to propose draft tokens which are then verified in parallel. The number of draft tokens produced in each drafting round is referred to as the draft length and is often a static hyperparameter chosen based on the acceptance rate statistics of the draft tokens. However, setting a static draft length can negatively impact performance, especially in scenarios where drafting is expensive and there is a high variance in the number of tokens accepted. Adaptive Entropy-based Draft Length (AdaEDL) is a simple, training and parameter-free criteria which allows for early stopping of the token drafting process by approximating a lower bound on the expected acceptance probability of the drafted token based on the currently observed entropy of the drafted logits. We show that AdaEDL consistently outperforms static draft-length speculative decoding by 10%-57% as well as other training-free draft-stopping techniques by upto 10% in a variety of settings and datasets. At the same time, we show that AdaEDL is more robust than these techniques and preserves performance in high-sampling-temperature scenarios. Since it is training-free, in contrast to techniques that rely on the training of dataset-specific draft-stopping predictors, AdaEDL can seamlessly be integrated into a variety of pre-existing LLM systems.
Authors: Yiqing Xie, Wenxuan Zhou, Pradyot Prakash, Di Jin, Yuning Mao, Quintin Fettes, Arya Talebzadeh, Sinong Wang, Han Fang, Carolyn Rose, Daniel Fried, Hejia Zhang
Abstract: Factuality evaluation aims to detect factual errors produced by language models (LMs) and hence guide the development of more factual models. Towards this goal, we train a factuality evaluator, FenCE, that provides LM generators with claim-level factuality feedback. We conduct data augmentation on a combination of public judgment datasets to train FenCE to (1) generate textual critiques along with scores and (2) make claim-level judgment based on diverse source documents obtained by various tools. We then present a framework that leverages FenCE to improve the factuality of LM generators by constructing training data. Specifically, we generate a set of candidate responses, leverage FenCE to revise and score each response without introducing lesser-known facts, and train the generator by preferring highly scored revised responses. Experiments show that our data augmentation methods improve the evaluator's accuracy by 2.9% on LLM-AggreFact. With FenCE, we improve Llama3-8B-chat's factuality rate by 14.45% on FActScore, outperforming state-of-the-art factuality finetuning methods by 6.96%.
Authors: Xinyu Wang, Wenbo Zhang, Sarah Rajtmajer
Abstract: In today's global digital landscape, misinformation transcends linguistic boundaries, posing a significant challenge for moderation systems. While significant advances have been made in misinformation detection, the focus remains largely on monolingual high-resource contexts, with low-resource languages often overlooked. This survey aims to bridge that gap by providing a comprehensive overview of the current research on low-resource language misinformation detection in both monolingual and multilingual settings. We review the existing datasets, methodologies, and tools used in these domains, identifying key challenges related to: data resources, model development, cultural and linguistic context, real-world applications, and research efforts. We also examine emerging approaches, such as language-agnostic models and multi-modal techniques, while emphasizing the need for improved data collection practices, interdisciplinary collaboration, and stronger incentives for socially responsible AI research. Our findings underscore the need for robust, inclusive systems capable of addressing misinformation across diverse linguistic and cultural contexts.
Authors: Tanmay Parekh, Jeffrey Kwan, Jiarui Yu, Sparsh Johri, Hyosang Ahn, Sreya Muppalla, Kai-Wei Chang, Wei Wang, Nanyun Peng
Abstract: Social media is often the first place where communities discuss the latest societal trends. Prior works have utilized this platform to extract epidemic-related information (e.g. infections, preventive measures) to provide early warnings for epidemic prediction. However, these works only focused on English posts, while epidemics can occur anywhere in the world, and early discussions are often in the local, non-English languages. In this work, we introduce the first multilingual Event Extraction (EE) framework SPEED++ for extracting epidemic event information for a wide range of diseases and languages. To this end, we extend a previous epidemic ontology with 20 argument roles; and curate our multilingual EE dataset SPEED++ comprising 5.1K tweets in four languages for four diseases. Annotating data in every language is infeasible; thus we develop zero-shot cross-lingual cross-disease models (i.e., training only on English COVID data) utilizing multilingual pre-training and show their efficacy in extracting epidemic-related events for 65 diverse languages across different diseases. Experiments demonstrate that our framework can provide epidemic warnings for COVID-19 in its earliest stages in Dec 2019 (3 weeks before global discussions) from Chinese Weibo posts without any training in Chinese. Furthermore, we exploit our framework's argument extraction capabilities to aggregate community epidemic discussions like symptoms and cure measures, aiding misinformation detection and public attention monitoring. Overall, we lay a strong foundation for multilingual epidemic preparedness.
Authors: Zhisheng Lin, Yifu Liu, Zhiling Luo, Jinyang Gao, Yu Li
Abstract: The improvement in translating natural language to structured query language (SQL) can be attributed to the advancements in large language models (LLMs). Open-source LLMs, tailored for specific database dialects such as MySQL, have shown great performance. However, cloud service providers are looking for a unified database manager service (e.g., Cosmos DB from Azure, Amazon Aurora from AWS, Lindorm from AlibabaCloud) that can support multiple dialects. This requirement has led to the concept of multi-dialect query generation, which presents challenges to LLMs. These challenges include syntactic differences among dialects and imbalanced data distribution across multiple dialects. To tackle these challenges, we propose MoMQ, a novel Mixture-of-Experts-based multi-dialect query generation framework across both relational and non-relational databases. MoMQ employs a dialect expert group for each dialect and a multi-level routing strategy to handle dialect-specific knowledge, reducing interference during query generation. Additionally, a shared expert group is introduced to address data imbalance, facilitating the transfer of common knowledge from high-resource dialects to low-resource ones. Furthermore, we have developed a high-quality multi-dialect query generation benchmark that covers relational and non-relational databases such as MySQL, PostgreSQL, Cypher for Neo4j, and nGQL for NebulaGraph. Extensive experiments have shown that MoMQ performs effectively and robustly even in resource-imbalanced scenarios.
Authors: Kun Li, Tianhua Zhang, Xixin Wu, Hongyin Luo, James Glass, Helen Meng
Abstract: Knowledge Graphs (KGs) can serve as reliable knowledge sources for question answering (QA) due to their structured representation of knowledge. Existing research on the utilization of KG for large language models (LLMs) prevalently relies on subgraph retriever or iterative prompting, overlooking the potential synergy of LLMs' step-wise reasoning capabilities and KGs' structural nature. In this paper, we present DoG (Decoding on Graphs), a novel framework that facilitates a deep synergy between LLMs and KGs. We first define a concept, well-formed chain, which consists of a sequence of interrelated fact triplets on the KGs, starting from question entities and leading to answers. We argue that this concept can serve as a principle for making faithful and sound reasoning for KGQA. To enable LLMs to generate well-formed chains, we propose graph-aware constrained decoding, in which a constraint derived from the topology of the KG regulates the decoding process of the LLMs. This constrained decoding method ensures the generation of well-formed chains while making full use of the step-wise reasoning capabilities of LLMs. Based on the above, DoG, a training-free approach, is able to provide faithful and sound reasoning trajectories grounded on the KGs. Experiments across various KGQA tasks with different background KGs demonstrate that DoG achieves superior and robust performance. DoG also shows general applicability with various open-source LLMs.
Authors: Maarten Buyl, Alexander Rogiers, Sander Noels, Iris Dominguez-Catena, Edith Heiter, Raphael Romero, Iman Johary, Alexandru-Cristian Mara, Jefrey Lijffijt, Tijl De Bie
Abstract: Large language models (LLMs) are trained on vast amounts of data to generate natural language, enabling them to perform tasks like text summarization and question answering. These models have become popular in artificial intelligence (AI) assistants like ChatGPT and already play an influential role in how humans access information. However, the behavior of LLMs varies depending on their design, training, and use. In this paper, we uncover notable diversity in the ideological stance exhibited across different LLMs and languages in which they are accessed. We do this by prompting a diverse panel of popular LLMs to describe a large number of prominent and controversial personalities from recent world history, both in English and in Chinese. By identifying and analyzing moral assessments reflected in the generated descriptions, we find consistent normative differences between how the same LLM responds in Chinese compared to English. Similarly, we identify normative disagreements between Western and non-Western LLMs about prominent actors in geopolitical conflicts. Furthermore, popularly hypothesized disparities in political goals among Western models are reflected in significant normative differences related to inclusion, social inequality, and political scandals. Our results show that the ideological stance of an LLM often reflects the worldview of its creators. This raises important concerns around technological and regulatory efforts with the stated aim of making LLMs ideologically `unbiased', and it poses risks for political instrumentalization.
Authors: Donglin Di, Weinan Zhang, Yue Zhang, Fanglin Wang
Abstract: Making use of off-the-shelf resources of resource-rich languages to transfer knowledge for low-resource languages raises much attention recently. The requirements of enabling the model to reach the reliable performance lack well guided, such as the scale of required annotated data or the effective framework. To investigate the first question, we empirically investigate the cost-effectiveness of several methods to train the intent classification and slot-filling models for Indonesia (ID) from scratch by utilizing the English data. Confronting the second challenge, we propose a Bi-Confidence-Frequency Cross-Lingual transfer framework (BiCF), composed by ``BiCF Mixing'', ``Latent Space Refinement'' and ``Joint Decoder'', respectively, to tackle the obstacle of lacking low-resource language dialogue data. Extensive experiments demonstrate our framework performs reliably and cost-efficiently on different scales of manually annotated Indonesian data. We release a large-scale fine-labeled dialogue dataset (ID-WOZ) and ID-BERT of Indonesian for further research.
Authors: Seoyeon Kim, Huiseo Kim, Chanjun Park, Jinyoung Yeo, Dongha Lee
Abstract: Code-switching (CS), a phenomenon where multilingual speakers alternate between languages in a discourse, can convey subtle cultural and linguistic nuances that can be otherwise lost in translation. Recent state-of-the-art multilingual large language models (LLMs) demonstrate excellent multilingual abilities in various aspects including understanding CS, but the power of CS in eliciting language-specific knowledge is yet to be discovered. Therefore, we investigate the effectiveness of code-switching on a wide range of multilingual LLMs in terms of knowledge activation, or the act of identifying and leveraging knowledge for reasoning. To facilitate the research, we first present EnKoQA, a synthetic English-Korean CS question-answering dataset. We provide a comprehensive analysis on a variety of multilingual LLMs by subdividing activation process into knowledge identification and knowledge leveraging. Our experiments demonstrate that compared to English text, CS can faithfully activate knowledge inside LLMs, especially on language-specific domains. In addition, the performance gap between CS and English is larger in models that show excellent monolingual abilities, suggesting that there exists a correlation with CS and Korean proficiency.
Authors: ChaeHun Park, Hojun Cho, Jaegul Choo
Abstract: This paper explores integrating Automatic Speech Recognition (ASR) into natural language query systems to improve weather forecasting efficiency for Korean meteorologists. We address challenges in developing ASR systems for the Korean weather domain, specifically specialized vocabulary and Korean linguistic intricacies. To tackle these issues, we constructed an evaluation dataset of spoken queries recorded by native Korean speakers. Using this dataset, we assessed various configurations of a multilingual ASR model family, identifying performance limitations related to domain-specific terminology. We then implemented a simple text-to-speech-based data augmentation method, which improved the recognition of specialized terms while maintaining general-domain performance. Our contributions include creating a domain-specific dataset, comprehensive ASR model evaluations, and an effective augmentation technique. We believe our work provides a foundation for future advancements in ASR for the Korean weather forecasting domain.
Authors: Zezhong Wang, Xingshan Zeng, Weiwen Liu, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
Abstract: Supervised fine-tuning (SFT) is a common method to enhance the tool calling capabilities of Large Language Models (LLMs), with the training data often being synthesized. The current data synthesis process generally involves sampling a set of tools, formulating a requirement based on these tools, and generating the call statements. However, tools sampled randomly lack relevance, making them difficult to combine and thus reducing the diversity of the data. Additionally, current work overlooks the coherence between turns of dialogues, leading to a gap between the synthesized data and real-world scenarios. To address these issues, we propose a Graph-based Sampling strategy to sample more relevant tool combinations, and a Planned-generation strategy to create plans that guide the synthesis of coherent dialogues. We integrate these two strategies and enable multiple agents to synthesize the dialogue data interactively, resulting in our tool-calling data synthesis pipeline ToolFlow. Data quality assessments demonstrate improvements in the naturalness and coherence of our synthesized dialogues. Finally, we apply SFT on LLaMA-3.1-8B using 8,000 synthetic dialogues generated with ToolFlow. Results show that the model achieves tool-calling performance comparable to or even surpassing GPT-4, while maintaining strong general capabilities.
Authors: Chung-En Sun, Xiaodong Liu, Weiwei Yang, Tsui-Wei Weng, Hao Cheng, Aidan San, Michel Galley, Jianfeng Gao
Abstract: Recent research has shown that Large Language Models (LLMs) are vulnerable to automated jailbreak attacks, where adversarial suffixes crafted by algorithms appended to harmful queries bypass safety alignment and trigger unintended responses. Current methods for generating these suffixes are computationally expensive and have low Attack Success Rates (ASR), especially against well-aligned models like Llama2 and Llama3. To overcome these limitations, we introduce ADV-LLM, an iterative self-tuning process that crafts adversarial LLMs with enhanced jailbreak ability. Our framework significantly reduces the computational cost of generating adversarial suffixes while achieving nearly 100\% ASR on various open-source LLMs. Moreover, it exhibits strong attack transferability to closed-source models, achieving 99% ASR on GPT-3.5 and 49% ASR on GPT-4, despite being optimized solely on Llama3. Beyond improving jailbreak ability, ADV-LLM provides valuable insights for future safety alignment research through its ability to generate large datasets for studying LLM safety. Our code is available at: https://github.com/SunChungEn/ADV-LLM
Authors: Sergio Burdisso, Srikanth Madikeri, Petr Motlicek
Abstract: Efficiently deriving structured workflows from unannotated dialogs remains an underexplored and formidable challenge in computational linguistics. Automating this process could significantly accelerate the manual design of workflows in new domains and enable the grounding of large language models in domain-specific flowcharts, enhancing transparency and controllability. In this paper, we introduce Dialog2Flow (D2F) embeddings, which differ from conventional sentence embeddings by mapping utterances to a latent space where they are grouped according to their communicative and informative functions (i.e., the actions they represent). D2F allows for modeling dialogs as continuous trajectories in a latent space with distinct action-related regions. By clustering D2F embeddings, the latent space is quantized, and dialogs can be converted into sequences of region/action IDs, facilitating the extraction of the underlying workflow. To pre-train D2F, we build a comprehensive dataset by unifying twenty task-oriented dialog datasets with normalized per-turn action annotations. We also introduce a novel soft contrastive loss that leverages the semantic information of these actions to guide the representation learning process, showing superior performance compared to standard supervised contrastive loss. Evaluation against various sentence embeddings, including dialog-specific ones, demonstrates that D2F yields superior qualitative and quantitative results across diverse domains.
Authors: Hengxiang Zhang, Hongfu Gao, Qiang Hu, Guanhua Chen, Lili Yang, Bingyi Jing, Hongxin Wei, Bing Wang, Haifeng Bai, Lei Yang
Abstract: With the rapid development of Large language models (LLMs), understanding the capabilities of LLMs in identifying unsafe content has become increasingly important. While previous works have introduced several benchmarks to evaluate the safety risk of LLMs, the community still has a limited understanding of current LLMs' capability to recognize illegal and unsafe content in Chinese contexts. In this work, we present a Chinese safety benchmark (ChineseSafe) to facilitate research on the content safety of large language models. To align with the regulations for Chinese Internet content moderation, our ChineseSafe contains 205,034 examples across 4 classes and 10 sub-classes of safety issues. For Chinese contexts, we add several special types of illegal content: political sensitivity, pornography, and variant/homophonic words. Moreover, we employ two methods to evaluate the legal risks of popular LLMs, including open-sourced models and APIs. The results reveal that many LLMs exhibit vulnerability to certain types of safety issues, leading to legal risks in China. Our work provides a guideline for developers and researchers to facilitate the safety of LLMs. Our results are also available at https://huggingface.co/spaces/SUSTech/ChineseSafe-Benchmark.
URLs: https://huggingface.co/spaces/SUSTech/ChineseSafe-Benchmark.
Authors: Liangdong Wang, Bo-Wen Zhang, Chengwei Wu, Hanyu Zhao, Xiaofeng Shi, Shuhao Gu, Jijie Li, Quanyue Ma, TengFei Pan, Guang Liu
Abstract: We present CCI3.0-HQ (https://huggingface.co/datasets/BAAI/CCI3-HQ), a high-quality 500GB subset of the Chinese Corpora Internet 3.0 (CCI3.0)(https://huggingface.co/datasets/BAAI/CCI3-Data), developed using a novel two-stage hybrid filtering pipeline that significantly enhances data quality. To evaluate its effectiveness, we trained a 0.5B parameter model from scratch on 100B tokens across various datasets, achieving superior performance on 10 benchmarks in a zero-shot setting compared to CCI3.0, SkyPile, and WanjuanV1. The high-quality filtering process effectively distills the capabilities of the Qwen2-72B-instruct model into a compact 0.5B model, attaining optimal F1 scores for Chinese web data classification. We believe this open-access dataset will facilitate broader access to high-quality language models.
URLs: https://huggingface.co/datasets/BAAI/CCI3-HQ),, https://huggingface.co/datasets/BAAI/CCI3-Data),
Authors: Abdulfattah Safa, Tamta Kapanadze, Arda Uzuno\u{g}lu, G\"ozde G\"ul \c{S}ahin
Abstract: Recent advances in large language models have demonstrated promising capabilities in following simple instructions through instruction tuning. However, real-world tasks often involve complex, multi-step instructions that remain challenging for current NLP systems. Despite growing interest in this area, there lacks a comprehensive survey that systematically analyzes the landscape of complex instruction understanding and processing. Through a systematic review of the literature, we analyze available resources, representation schemes, and downstream tasks related to instructional text. Our study examines 177 papers, identifying trends, challenges, and opportunities in this emerging field. We provide AI/NLP researchers with essential background knowledge and a unified view of various approaches to complex instruction understanding, bridging gaps between different research directions and highlighting future research opportunities.
Authors: Zecheng Tang, Zechen Sun, Juntao Li, Qiaoming Zhu, Min Zhang
Abstract: Long-context models(LCMs) have shown great potential in processing long input sequences(even more than 100M tokens) conveniently and effectively. With significant progress, recent research has pointed out that LCMs can accurately locate token-level salient information within the context. Yet, the generation performance of these LCMs is far from satisfactory and might result in misaligned responses, such as hallucinations. To enhance the generation capability of LCMs, existing works have investigated the effects of data size and quality for both pre-training and instruction tuning. Though achieving meaningful improvement, previous methods fall short in either effectiveness or efficiency. In this paper, we introduce LOGO(Long cOntext aliGnment via efficient preference Optimization), a training strategy that first introduces preference optimization for long-context alignment. To overcome the GPU memory-bound issue caused by the long sequence, LOGO employs a reference-free preference optimization strategy and adopts a position synthesis method to construct the training data. By training with only 0.3B data on a single 8$\times$A800 GPU machine for 16 hours, LOGO allows the Llama-3-8B-Instruct-80K model to achieve comparable performance with GPT-4 in real-world long-context tasks while preserving the model's original capabilities on other tasks, e.g., language modeling and MMLU. Moreover, LOGO can extend the model's context window size while enhancing its generation performance.
Authors: Omar Naim, Nicholas Asher
Abstract: This paper explores the much discussed, possible explanatory link between attention weights (AW) in transformer models and predicted output. Contrary to intuition and early research on attention, more recent prior research has provided formal arguments and empirical evidence that AW are not explanatorily relevant. We show that the formal arguments are incorrect. We introduce and effectively compute efficient attention, which isolates the effective components of attention matrices in tasks and models in which AW play an explanatory role. We show that efficient attention has a causal role (provides minimally necessary and sufficient conditions) for predicting model output in NLP tasks requiring contextual information, and we show, contrary to [7], that efficient attention matrices are probability distributions and are effectively calculable. Thus, they should play an important part in the explanation of attention based model behavior. We offer empirical experiments in support of our method illustrating various properties of efficient attention with various metrics on four datasets.
Authors: Shuhao Gu, Jialing Zhang, Siyuan Zhou, Kevin Yu, Zhaohu Xing, Liangdong Wang, Zhou Cao, Jintao Jia, Zhuoyi Zhang, Yixuan Wang, Zhenchong Hu, Bo-Wen Zhang, Jijie Li, Dong Liang, Yingli Zhao, Yulong Ao, Yaoqi Liu, Fangxiang Feng, Guang Liu
Abstract: Vision-Language Models (VLMs) have recently made significant progress, but the limited scale and quality of open-source instruction data hinder their performance compared to closed-source models. In this work, we address this limitation by introducing Infinity-MM, a large-scale multimodal instruction dataset with 40 million samples, enhanced through rigorous quality filtering and deduplication. We also propose a synthetic instruction generation method based on open-source VLMs, using detailed image annotations and diverse question generation. Using this data, we trained a 2-billion-parameter VLM, Aquila-VL-2B, achieving state-of-the-art (SOTA) performance for models of similar scale. This demonstrates that expanding instruction data and generating synthetic data can significantly improve the performance of open-source models.
Authors: Krzysztof Ociepa, {\L}ukasz Flis, Krzysztof Wr\'obel, Adrian Gwo\'zdziej, Remigiusz Kinas
Abstract: We introduce Bielik 7B v0.1, a 7-billion-parameter generative text model for Polish language processing. Trained on curated Polish corpora, this model addresses key challenges in language model development through innovative techniques. These include Weighted Instruction Cross-Entropy Loss, which balances the learning of different instruction types, and Adaptive Learning Rate, which dynamically adjusts the learning rate based on training progress. To evaluate performance, we created the Open PL LLM Leaderboard and Polish MT-Bench, novel frameworks assessing various NLP tasks and conversational abilities. Bielik 7B v0.1 demonstrates significant improvements, achieving a 9 percentage point increase in average score compared to Mistral-7B-v0.1 on the RAG Reader task. It also excels in the Polish MT-Bench, particularly in Reasoning (6.15/10) and Role-playing (7.83/10) categories. This model represents a substantial advancement in Polish language AI, offering a powerful tool for diverse linguistic applications and setting new benchmarks in the field.
Authors: Adam Nohejl, Akio Hayakawa, Yusuke Ide, Taro Watanabe
Abstract: The tasks of lexical complexity prediction (LCP) and complex word identification (CWI) commonly presuppose that difficult to understand words are shared by the target population. Meanwhile, personalization methods have also been proposed to adapt models to individual needs. We verify that a recent Japanese LCP dataset is representative of its target population by partially replicating the annotation. By another reannotation we show that native Chinese speakers perceive the complexity differently due to Sino-Japanese vocabulary. To explore the possibilities of personalization, we compare competitive baselines trained on the group mean ratings and individual ratings in terms of performance for an individual. We show that the model trained on a group mean performs similarly to an individual model in the CWI task, while achieving good LCP performance for an individual is difficult. We also experiment with adapting a finetuned BERT model, which results only in marginal improvements across all settings.
Authors: Chien Van Nguyen, Huy Huu Nguyen, Thang M. Pham, Ruiyi Zhang, Hanieh Deilamsalehy, Puneet Mathur, Ryan A. Rossi, Trung Bui, Viet Dac Lai, Franck Dernoncourt, Thien Huu Nguyen
Abstract: Efficient long-context language modeling remains a significant challenge in Natural Language Processing (NLP). While Transformers dominate language tasks, they struggle with long sequences due to quadratic computational complexity in training and linearly scaling memory costs during inference. Recent State Space Models (SSMs) such as Mamba offer alternatives with constant memory usage, but they underperform in tasks requiring extensive in-context retrieval. We introduce Taipan, a novel hybrid architecture that combines Mamba-2 with Selective Attention Layers (SALs). These SALs identify tokens requiring long-range interactions, remove less important features, and then augment their representations using the attention module. This approach balances Mamba's efficiency with Transformer-like performance in memory-intensive tasks. By constraining the attention budget, Taipan extends accurate predictions to context lengths of up to 1 million tokens while preserving computational efficiency. Our experiments demonstrate Taipan's superior performance across various scales and tasks, offering a promising solution for efficient long-context language modeling.
Authors: Hawau Olamide Toyin, Hao Li, Hanan Aldarmaki
Abstract: Speech recognition and speech synthesis models are typically trained separately, each with its own set of learning objectives, training data, and model parameters, resulting in two distinct large networks. We propose a parameter-efficient approach to learning ASR and TTS jointly via a multi-task learning objective and shared parameters. Our evaluation demonstrates that the performance of our multi-task model is comparable to that of individually trained models while significantly saving computational and memory costs ($\sim$50\% reduction in the total number of parameters required for the two tasks combined). We experiment with English as a resource-rich language, and Arabic as a relatively low-resource language due to shortage of TTS data. Our models are trained with publicly available data, and both the training code and model checkpoints are openly available for further research.
Authors: David Thulke, Yingbo Gao, Rricha Jalota, Christian Dugast, Hermann Ney
Abstract: This paper explores the rapid development of a telephone call summarization system utilizing large language models (LLMs). Our approach involves initial experiments with prompting existing LLMs to generate summaries of telephone conversations, followed by the creation of a tailored synthetic training dataset utilizing stronger frontier models. We place special focus on the diversity of the generated data and on the ability to control the length of the generated summaries to meet various use-case specific requirements. The effectiveness of our method is evaluated using two state-of-the-art LLM-as-a-judge-based evaluation techniques to ensure the quality and relevance of the summaries. Our results show that fine-tuned Llama-2-7B-based summarization model performs on-par with GPT-4 in terms of factual accuracy, completeness and conciseness. Our findings demonstrate the potential for quickly bootstrapping a practical and efficient call summarization system.
Authors: Sanjay Singh, Amaresh Chakrabarti
Abstract: This paper proposes a framework for assessing the novelty of design problems using the SAPPhIRE model of causality. The novelty of a problem is measured as its minimum distance from the problems in a reference problem database. The distance is calculated by comparing the current problem and each reference past problem at the various levels of abstraction in the SAPPhIRE ontology. The basis for comparison is textual similarity. To demonstrate the applicability of the proposed framework, The current set of problems associated with an artifact, as collected from its stakeholders, were compared with the past set of problems, as collected from patents and other web sources, to assess the novelty of the current set. This approach is aimed at providing a better understanding of the degree of novelty of any given set of current problems by comparing them to similar problems available from historical records. Since manual assessment, the current mode of such assessments as reported in the literature, is a tedious process, to reduce time complexity and to afford better applicability for larger sets of problem statements, an automated assessment is proposed and used in this paper.
Authors: Haonan Chen, Liang Wang, Nan Yang, Yutao Zhu, Ziliang Zhao, Furu Wei, Zhicheng Dou
Abstract: Synthetic data generation has become an increasingly popular way of training models without the need for large, manually labeled datasets. For tasks like text embedding, synthetic data offers diverse and scalable training examples, significantly reducing the cost of human annotation. However, most current approaches rely heavily on proprietary models like GPT-4, which are expensive and inefficient for generating large-scale embedding data. In this paper, we introduce SPEED, a framework that aligns open-source small models (8B) to efficiently generate large-scale synthetic embedding data. Through supervised fine-tuning, preference optimization, and self-improvement, SPEED enables small open-source models to produce high-quality data. Remarkably, SPEED uses only less than 1/10 of the GPT API calls, outperforming the state-of-the-art embedding model E5_mistral when both are trained solely on their synthetic data. Using this efficient generator, we conduct a comprehensive study on how various factors within the alignment pipeline impact data quality and reveal the scaling law for synthetic embedding data.
Authors: Wenhong Zhu, Zhiwei He, Xiaofeng Wang, Pengfei Liu, Rui Wang
Abstract: Aligning language models (LMs) with human preferences has become a key area of research, enabling these models to meet diverse user needs better. Inspired by weak-to-strong generalization, where a strong LM fine-tuned on labels generated by a weaker model can consistently outperform its weak supervisor, we extend this idea to model alignment. In this work, we observe that the alignment behavior in weaker models can be effectively transferred to stronger models and even exhibit an amplification effect. Based on this insight, we propose a method called Weak-to-Strong Preference Optimization (WSPO), which achieves strong model alignment by learning the distribution differences before and after the alignment of the weak model. Experiments demonstrate that WSPO delivers outstanding performance, improving the win rate of Qwen2-7B-Instruct on Arena-Hard from 39.70 to 49.60, achieving a remarkable 47.04 length-controlled win rate on AlpacaEval 2, and scoring 7.33 on MT-bench. Our results suggest that using the weak model to elicit a strong model with a high alignment ability is feasible.
Authors: Esteban Garces Arias, Hannah Blocher, Julian Rodemann, Meimingwei Li, Christian Heumann, Matthias A{\ss}enmacher
Abstract: Open-ended text generation has become a prominent task in natural language processing due to the rise of powerful (large) language models. However, evaluating the quality of these models and the employed decoding strategies remains challenging because of trade-offs among widely used metrics such as coherence, diversity, and perplexity. Decoding methods often excel in some metrics while underperforming in others, complicating the establishment of a clear ranking. In this paper, we present novel ranking strategies within this multicriteria framework. Specifically, we employ benchmarking approaches based on partial orderings and present a new summary metric designed to balance existing automatic indicators, providing a more holistic evaluation of text generation quality. Furthermore, we discuss the alignment of these approaches with human judgments. Our experiments demonstrate that the proposed methods offer a robust way to compare decoding strategies, exhibit similarities with human preferences, and serve as valuable tools in guiding model selection for open-ended text generation tasks. Finally, we suggest future directions for improving evaluation methodologies in text generation. Our codebase, datasets, and models are publicly available.
Authors: Yuyang Ding, Xinyu Shi, Xiaobo Liang, Juntao Li, Qiaoming Zhu, Min Zhang
Abstract: The availability of high-quality data is one of the most important factors in improving the reasoning capability of LLMs. Existing works have demonstrated the effectiveness of creating more instruction data from seed questions or knowledge bases. Recent research indicates that continually scaling up data synthesis from strong models (e.g., GPT-4) can further elicit reasoning performance. Though promising, the open-sourced community still lacks high-quality data at scale and scalable data synthesis methods with affordable costs. To address this, we introduce ScaleQuest, a scalable and novel data synthesis method that utilizes "small-size" (e.g., 7B) open-source models to generate questions from scratch without the need for seed data with complex augmentation constraints. With the efficient ScaleQuest, we automatically constructed a mathematical reasoning dataset consisting of 1 million problem-solution pairs, which are more effective than existing open-sourced datasets. It can universally increase the performance of mainstream open-source models (i.e., Mistral, Llama3, DeepSeekMath, and Qwen2-Math) by achieving 29.2% to 46.4% gains on MATH. Notably, simply fine-tuning the Qwen2-Math-7B-Base model with our dataset can even surpass Qwen2-Math-7B-Instruct, a strong and well-aligned model on closed-source data, and proprietary models such as GPT-4-Turbo and Claude-3.5 Sonnet.
Authors: Ran Zhang, Wei Zhao, Steffen Eger
Abstract: Recent research has focused on literary machine translation (MT) as a new challenge in MT. However, the evaluation of literary MT remains an open problem. We contribute to this ongoing discussion by introducing LITEVAL-CORPUS, a paragraph-level parallel corpus comprising multiple verified human translations and outputs from 9 MT systems, which totals over 2k paragraphs and includes 13k annotated sentences across four language pairs, costing 4.5k Euro. This corpus enables us to (i) examine the consistency and adequacy of multiple annotation schemes, (ii) compare evaluations by students and professionals, and (iii) assess the effectiveness of LLM-based metrics. We find that Multidimensional Quality Metrics (MQM), as the de facto standard in non-literary human MT evaluation, is inadequate for literary translation: While Best-Worst Scaling (BWS) with students and Scalar Quality Metric (SQM) with professional translators prefer human translations at rates of ~82% and ~94%, respectively, MQM with student annotators prefers human professional translations over the translations of the best-performing LLMs in only ~42% of cases. While automatic metrics generally show a moderate correlation with human MQM and SQM, they struggle to accurately identify human translations, with rates of at most ~20%. Our overall evaluation indicates that human professional translations consistently outperform LLM translations, where even the most recent LLMs tend to produce more literal and less diverse translations compared to human translations. However, newer LLMs such as GPT-4o perform substantially better than older ones.
Authors: Rita Ramos, Everlyn Asiko Chimoto, Maartje ter Hoeve, Natalie Schluter
Abstract: We introduce GrammaMT, a grammatically-aware prompting approach for machine translation that uses Interlinear Glossed Text (IGT), a common form of linguistic description providing morphological and lexical annotations for source sentences. GrammaMT proposes three prompting strategies: gloss-shot, chain-gloss and model-gloss. All are training-free, requiring only a few examples that involve minimal effort to collect, and making them well-suited for low-resource setups. Experiments show that GrammaMT enhances translation performance on open-source instruction-tuned LLMs for various low- to high-resource languages across three benchmarks: (1) the largest IGT corpus, (2) the challenging 2023 SIGMORPHON Shared Task data over endangered languages, and (3) even in an out-of-domain setting with FLORES. Moreover, ablation studies reveal that leveraging gloss resources could substantially boost MT performance (by over 17 BLEU points) if LLMs accurately generate or access input sentence glosses.
Authors: Chenxin An, Jun Zhang, Ming Zhong, Lei Li, Shansan Gong, Yao Luo, Jingjing Xu, Lingpeng Kong
Abstract: Advancements in distributed training and efficient attention mechanisms have significantly expanded the context window sizes of large language models (LLMs). However, recent work reveals that the effective context lengths of open-source LLMs often fall short, typically not exceeding half of their training lengths. In this work, we attribute this limitation to the left-skewed frequency distribution of relative positions formed in LLMs pretraining and post-training stages, which impedes their ability to effectively gather distant information. To address this challenge, we introduce ShifTed Rotray position embeddING (STRING). STRING shifts well-trained positions to overwrite the original ineffective positions during inference, enhancing performance within their existing training lengths. Experimental results show that without additional training, STRING dramatically improves the performance of the latest large-scale models, such as Llama3.1 70B and Qwen2 72B, by over 10 points on popular long-context benchmarks RULER and InfiniteBench, establishing new state-of-the-art results for open-source LLMs. Compared to commercial models, Llama 3.1 70B with \method even achieves better performance than GPT-4-128K and clearly surpasses Claude 2 and Kimi-chat.
Authors: Md. Khairul Islam, Andrew Wang, Tianhao Wang, Yangfeng Ji, Judy Fox, Jieyu Zhao
Abstract: Differential privacy (DP) is applied when fine-tuning pre-trained large language models (LLMs) to limit leakage of training examples. While most DP research has focused on improving a model's privacy-utility tradeoff, some find that DP can be unfair to or biased against underrepresented groups. In this work, we show the impact of DP on bias in LLMs through empirical analysis. Differentially private training can increase the model bias against protected groups w.r.t AUC-based bias metrics. DP makes it more difficult for the model to differentiate between the positive and negative examples from the protected groups and other groups in the rest of the population. Our results also show that the impact of DP on bias is not only affected by the privacy protection level but also the underlying distribution of the dataset.
Authors: Yingjie Li, Yun Luo, Xiaotian Xie, Yue Zhang
Abstract: Large language models (LLMs) have exhibited impressive zero-shot performance on inference tasks. However, LLMs may suffer from spurious correlations between input texts and output labels, which limits LLMs' ability to reason based purely on general language understanding. In other words, LLMs may make predictions primarily based on premise or hypothesis, rather than both components. To address this problem that may lead to unexpected performance degradation, we propose task calibration (TC), a zero-shot and inference-only calibration method inspired by mutual information which recovers LLM performance through task reformulation. TC encourages LLMs to reason based on both premise and hypothesis, while mitigating the models' over-reliance on individual premise or hypothesis for inference. Experimental results show that TC achieves a substantial improvement on 13 inference tasks in the zero-shot setup. We further validate the effectiveness of TC in few-shot setups and various natural language understanding tasks. Further analysis indicates that TC is also robust to prompt templates and has the potential to be integrated with other calibration methods.
Authors: Wei He, Zhiheng Xi, Wanxu Zhao, Xiaoran Fan, Yiwen Ding, Zifei Shan, Tao Gui, Qi Zhang, Xuanjing Huang
Abstract: Solving complex chart Q&A tasks requires advanced visual reasoning abilities in multimodal large language models (MLLMs). Recent studies highlight that these abilities consist of two main parts: recognizing key information from visual inputs and conducting reasoning over it. Thus, a promising approach to enhance MLLMs is to construct relevant training data focusing on the two aspects. However, collecting and annotating complex charts and questions is costly and time-consuming, and ensuring the quality of annotated answers remains a challenge. In this paper, we propose Code-as-Intermediary Translation (CIT), a cost-effective, efficient and easily scalable data synthesis method for distilling visual reasoning abilities from LLMs to MLLMs. The code serves as an intermediary that translates visual chart representations into textual representations, enabling LLMs to understand cross-modal information. Specifically, we employ text-based synthesizing techniques to construct chart-plotting code and produce ReachQA, a dataset containing 3k reasoning-intensive charts and 20k Q&A pairs to enhance both recognition and reasoning abilities. Experiments show that when fine-tuned with our data, models not only perform well on chart-related benchmarks, but also demonstrate improved multimodal reasoning abilities on general mathematical benchmarks like MathVista. The code and dataset are publicly available at https://github.com/hewei2001/ReachQA.
Authors: Zhengkai Lin, Zhihang Fu, Kai Liu, Liang Xie, Binbin Lin, Wenxiao Wang, Deng Cai, Yue Wu, Jieping Ye
Abstract: While large language models (LLMs) showcase unprecedented capabilities, they also exhibit certain inherent limitations when facing seemingly trivial tasks. A prime example is the recently debated "reversal curse", which surfaces when models, having been trained on the fact "A is B", struggle to generalize this knowledge to infer that "B is A". In this paper, we examine the manifestation of the reversal curse across various tasks and delve into both the generalization abilities and the problem-solving mechanisms of LLMs. This investigation leads to a series of significant insights: (1) LLMs are able to generalize to "B is A" when both A and B are presented in the context as in the case of a multiple-choice question. (2) This generalization ability is highly correlated to the structure of the fact "A is B" in the training documents. For example, this generalization only applies to biographies structured in "[Name] is [Description]" but not to "[Description] is [Name]". (3) We propose and verify the hypothesis that LLMs possess an inherent bias in fact recalling during knowledge application, which explains and underscores the importance of the document structure to successful learning. (4) The negative impact of this bias on the downstream performance of LLMs can hardly be mitigated through training alone. Based on these intriguing findings, our work not only presents a novel perspective for interpreting LLMs' generalization abilities from their intrinsic working mechanism but also provides new insights for the development of more effective learning methods for LLMs.
Authors: Sirui Chen, Shu Yu, Shengjie Zhao, Chaochao Lu
Abstract: Self-consciousness, the introspection of one's existence and thoughts, represents a high-level cognitive process. As language models advance at an unprecedented pace, a critical question arises: Are these models becoming self-conscious? Drawing upon insights from psychological and neural science, this work presents a practical definition of self-consciousness for language models and refines ten core concepts. Our work pioneers an investigation into self-consciousness in language models by, for the first time, leveraging causal structural games to establish the functional definitions of the ten core concepts. Based on our definitions, we conduct a comprehensive four-stage experiment: quantification (evaluation of ten leading models), representation (visualization of self-consciousness within the models), manipulation (modification of the models' representation), and acquisition (fine-tuning the models on core concepts). Our findings indicate that although models are in the early stages of developing self-consciousness, there is a discernible representation of certain concepts within their internal mechanisms. However, these representations of self-consciousness are hard to manipulate positively at the current stage, yet they can be acquired through targeted fine-tuning. Our datasets and code are at https://github.com/OpenCausaLab/SelfConsciousness.
Authors: Artur Kiulian, Anton Polishko, Mykola Khandoga, Yevhen Kostiuk, Guillermo Gabrielli, {\L}ukasz Gaga{\l}a, Fadi Zaraket, Qusai Abu Obaida, Hrishikesh Garud, Wendy Wing Yee Mak, Dmytro Chaplynskyi, Selma Belhadj Amor, Grigol Peradze
Abstract: In this paper, we propose a model-agnostic cost-effective approach to developing bilingual base large language models (LLMs) to support English and any target language. The method includes vocabulary expansion, initialization of new embeddings, model training and evaluation. We performed our experiments with three languages, each using a non-Latin script - Ukrainian, Arabic, and Georgian. Our approach demonstrates improved language performance while reducing computational costs. It mitigates the disproportionate penalization of underrepresented languages, promoting fairness and minimizing adverse phenomena such as code-switching and broken grammar. Additionally, we introduce new metrics to evaluate language quality, revealing that vocabulary size significantly impacts the quality of generated text.
Authors: Maya K. Nachesa, Vlad Niculae
Abstract: Speech recognition performance varies by language, domain, and speaker characteristics such as accent, and fine-tuning a model on any of these categories may lead to catastrophic forgetting. $k$ nearest neighbor search ($k$NN), first proposed for neural sequence decoders for natural language generation (NLG) and machine translation (MT), is a non-parametric method that can instead adapt by building an external datastore that can then be searched during inference time, without training the underlying model. We show that Whisper, a transformer end-to-end speech model, benefits from $k$NN. We investigate the differences between the speech and text setups. We discuss implications for speaker adaptation, and analyze improvements by gender, accent, and age.
Authors: Aryo Pradipta Gema, Chen Jin, Ahmed Abdulaal, Tom Diethe, Philip Teare, Beatrice Alex, Pasquale Minervini, Amrutha Saseendran
Abstract: Large Language Models (LLMs) often hallucinate, producing unfaithful or factually incorrect outputs by misrepresenting the provided context or incorrectly recalling internal knowledge. Recent studies have identified specific attention heads within the Transformer architecture, known as retrieval heads, responsible for extracting relevant contextual information. We hypothesise that masking these retrieval heads can induce hallucinations and that contrasting the outputs of the base LLM and the masked LLM can reduce hallucinations. To this end, we propose Decoding by Contrasting Retrieval Heads (DeCoRe), a novel training-free decoding strategy that amplifies information found in the context and model parameters. DeCoRe mitigates potentially hallucinated responses by dynamically contrasting the outputs of the base LLM and the masked LLM, using conditional entropy as a guide. Our extensive experiments confirm that DeCoRe significantly improves performance on tasks requiring high contextual faithfulness, such as summarisation (XSum by 18.6%), instruction following (MemoTrap by 10.9%), and open-book question answering (NQ-Open by 2.4% and NQ-Swap by 5.5%).
Authors: Shafkat Farabi, Tharindu Ranasinghe, Diptesh Kanojia, Yu Kong, Marcos Zampieri
Abstract: Sarcasm is a rhetorical device that is used to convey the opposite of the literal meaning of an utterance. Sarcasm is widely used on social media and other forms of computer-mediated communication motivating the use of computational models to identify it automatically. While the clear majority of approaches to sarcasm detection have been carried out on text only, sarcasm detection often requires additional information present in tonality, facial expression, and contextual images. This has led to the introduction of multimodal models, opening the possibility to detect sarcasm in multiple modalities such as audio, images, text, and video. In this paper, we present the first comprehensive survey on multimodal sarcasm detection - henceforth MSD - to date. We survey papers published between 2018 and 2023 on the topic, and discuss the models and datasets used for this task. We also present future research directions in MSD.
Authors: Omer Nahum, Nitay Calderon, Orgad Keller, Idan Szpektor, Roi Reichart
Abstract: NLP benchmarks rely on standardized datasets for training and evaluating models and are crucial for advancing the field. Traditionally, expert annotations ensure high-quality labels; however, the cost of expert annotation does not scale well with the growing demand for larger datasets required by modern models. While crowd-sourcing provides a more scalable solution, it often comes at the expense of annotation precision and consistency. Recent advancements in large language models (LLMs) offer new opportunities to enhance the annotation process, particularly for detecting label errors in existing datasets. In this work, we consider the recent approach of LLM-as-a-judge, leveraging an ensemble of LLMs to flag potentially mislabeled examples. Through a case study of four datasets from the TRUE benchmark, covering different tasks and domains, we empirically analyze the labeling quality of existing datasets, and compare expert, crowd-sourced, and our LLM-based annotations in terms of agreement, label quality, and efficiency, demonstrating the strengths and limitations of each annotation method. Our findings reveal a substantial number of label errors, which, when corrected, induce a significant upward shift in reported model performance. This suggests that many of the LLMs so-called mistakes are due to label errors rather than genuine model failures. Additionally, we discuss the implications of mislabeled data and propose methods to mitigate them in training to improve model performance.
Authors: Taido Purason, Hele-Andra Kuulmets, Mark Fishel
Abstract: The advancement of large language models (LLMs) has predominantly focused on high-resource languages, leaving low-resource languages, such as those in the Finno-Ugric family, significantly underrepresented. This paper addresses this gap by focusing on V\~oro, Livonian, and Komi. We cover almost the entire cycle of LLM creation, from data collection to instruction tuning and evaluation. Our contributions include developing multilingual base and instruction-tuned models; creating evaluation benchmarks, including the smugri-MT-bench multi-turn conversational benchmark; and conducting human evaluation. We intend for this work to promote linguistic diversity, ensuring that lesser-resourced languages can benefit from advancements in NLP.
Authors: Leif Azzopardi, Yashar Moshfeghi
Abstract: Auditing Large Language Models (LLMs) to discover their biases and preferences is an emerging challenge in creating Responsible Artificial Intelligence (AI). While various methods have been proposed to elicit the preferences of such models, countermeasures have been taken by LLM trainers, such that LLMs hide, obfuscate or point blank refuse to disclosure their positions on certain subjects. This paper presents PRISM, a flexible, inquiry-based methodology for auditing LLMs - that seeks to illicit such positions indirectly through task-based inquiry prompting rather than direct inquiry of said preferences. To demonstrate the utility of the methodology, we applied PRISM on the Political Compass Test, where we assessed the political leanings of twenty-one LLMs from seven providers. We show LLMs, by default, espouse positions that are economically left and socially liberal (consistent with prior work). We also show the space of positions that these models are willing to espouse - where some models are more constrained and less compliant than others - while others are more neutral and objective. In sum, PRISM can more reliably probe and audit LLMs to understand their preferences, biases and constraints.
Authors: A M Muntasir Rahman, Junyi Ye, Wei Yao, Wenpeng Yin, Guiling Wang
Abstract: Consider the math problem: "Lily received 3 cookies from her best friend yesterday and ate 5 for breakfast. Today, her friend gave her 3 more cookies. How many cookies does Lily have now?" Many large language models (LLMs) in previous research approach this problem by calculating the answer "1" using the equation "3 - 5 + 3." However, from a human perspective, we recognize the inherent flaw in this problem: Lily cannot eat 5 cookies if she initially only had 3. This discrepancy prompts a key question: Are current LLMs merely Blind Solver that apply mathematical operations without deeper reasoning, or can they function as Logical Thinker capable of identifying logical inconsistencies? To explore this question, we propose a benchmark dataset, FaultyMath, which includes faulty math problems of rich diversity: i) multiple mathematical categories, e.g., algebra, geometry, number theory, etc., ii) varying levels of difficulty, and iii) different origins of faultiness -- ranging from violations of common sense and ambiguous statements to mathematical contradictions and more. We evaluate a broad spectrum of LLMs, including open-source, closed-source, and math-specialized models, using FaultyMath across three dimensions: (i) How accurately can the models detect faulty math problems without being explicitly prompted to do so? (ii) When provided with hints -- either correct or misleading -- about the validity of the problems, to what extent do LLMs adapt to become reliable Logical Thinker? (iii) How trustworthy are the explanations generated by LLMs when they recognize a math problem as flawed? Through extensive experimentation and detailed analysis, our results demonstrate that existing LLMs largely function as Blind Solver and fall short of the reasoning capabilities required to perform as Logical Thinker.
Authors: Jort Vincenti, Karim Abdel Sadek, Joan Velja, Matteo Nulli, Metod Jazbec
Abstract: Increasing the size of large language models (LLMs) has been shown to lead to better performance. However, this comes at the cost of slower and more expensive inference. Early-exiting is a promising approach for improving the efficiency of LLM inference by enabling next token prediction at intermediate layers. Yet, the large vocabulary size in modern LLMs makes the confidence estimation required for exit decisions computationally expensive, diminishing the efficiency gains. To address this, we propose dynamically pruning the vocabulary at test time for each token. Specifically, the vocabulary is pruned at one of the initial layers, and the smaller vocabulary is then used throughout the rest of the forward pass. Our experiments demonstrate that such post-hoc dynamic vocabulary pruning improves the efficiency of confidence estimation in early-exit LLMs while maintaining competitive performance.
Authors: Yujuan Velvin Fu, Giridhar Kaushik Ramachandran, Namu Park, Kevin Lybarger, Fei Xia, Ozlem Uzuner, Meliha Yetisgen
Abstract: Large language models (LLMs) such as ChatGPT are fine-tuned on large and diverse instruction-following corpora, and can generalize to new tasks. However, those instruction-tuned LLMs often perform poorly in specialized medical natural language understanding (NLU) tasks that require domain knowledge, granular text comprehension, and structured data extraction. To bridge the gap, we: (1) propose a unified prompting format for 7 important NLU tasks, % through span extraction and multi-choice question-answering (QA), (2) curate an instruction-tuning dataset, MNLU-Instruct, utilizing diverse existing open-source medical NLU corpora, and (3) develop BioMistral-NLU, a generalizable medical NLU model, through fine-tuning BioMistral on MNLU-Instruct. We evaluate BioMistral-NLU in a zero-shot setting, across 6 important NLU tasks, from two widely adopted medical NLU benchmarks: Biomedical Language Understanding Evaluation (BLUE) and Biomedical Language Understanding and Reasoning Benchmark (BLURB). Our experiments show that our BioMistral-NLU outperforms the original BioMistral, as well as the proprietary LLMs - ChatGPT and GPT-4. Our dataset-agnostic prompting strategy and instruction tuning step over diverse NLU tasks enhance LLMs' generalizability across diverse medical NLU tasks. Our ablation experiments show that instruction-tuning on a wider variety of tasks, even when the total number of training instances remains constant, enhances downstream zero-shot generalization.
Authors: Jipeng Zhang, Jianshu Zhang, Yuanzhe Li, Renjie Pi, Rui Pan, Runtao Liu, Ziqiang Zheng, Tong Zhang
Abstract: Large Language Models (LLMs) demonstrate strong proficiency in generating code for high-resource programming languages (HRPLs) like Python but struggle significantly with low-resource programming languages (LRPLs) such as Racket or D. This performance gap deepens the digital divide, preventing developers using LRPLs from benefiting equally from LLM advancements and reinforcing disparities in innovation within underrepresented programming communities. While generating additional training data for LRPLs is promising, it faces two key challenges: manual annotation is labor-intensive and costly, and LLM-generated LRPL code is often of subpar quality. The underlying cause of this issue is the gap between natural language to programming language gap (NL-PL Gap), which is especially pronounced in LRPLs due to limited aligned data. In this work, we introduce a novel approach called Bridge-Coder, which leverages LLMs' intrinsic capabilities to enhance the performance on LRPLs. Our method consists of two key stages. Bridge Generation, where we create high-quality dataset by utilizing LLMs' general knowledge understanding, proficiency in HRPLs, and in-context learning abilities. Then, we apply the Bridged Alignment, which progressively improves the alignment between NL instructions and LRPLs. Experimental results across multiple LRPLs show that Bridge-Coder significantly enhances model performance, demonstrating the effectiveness and generalization of our approach. Furthermore, we offer a detailed analysis of the key components of our method, providing valuable insights for future work aimed at addressing the challenges associated with LRPLs.
Authors: Yujuan Fu, Ozlem Uzuner, Meliha Yetisgen, Fei Xia
Abstract: Large language models (LLMs) have demonstrated great performance across various benchmarks, showing potential as general-purpose task solvers. However, as LLMs are typically trained on vast amounts of data, a significant concern in their evaluation is data contamination, where overlap between training data and evaluation datasets inflates performance assessments. While multiple approaches have been developed to identify data contamination, these approaches rely on specific assumptions that may not hold universally across different settings. To bridge this gap, we systematically review 47 papers on data contamination detection, categorize the underlying assumptions, and assess whether they have been rigorously validated. We identify and analyze eight categories of assumptions and test three of them as case studies. Our analysis reveals that when classifying instances used for pretraining LLMs, detection approaches based on these three assumptions perform close to random guessing, suggesting that current LLMs learn data distributions rather than memorizing individual instances. Overall, this work underscores the importance of approaches clearly stating their underlying assumptions and testing their validity across various scenarios.
Authors: Fang Wang, Shenglin Yin, Xiaoying Bai, Minghao Hu, Tianwei Yan, Yi Liang
Abstract: Multi-modal Entity Linking (MEL) is a fundamental component for various downstream tasks. However, existing MEL datasets suffer from small scale, scarcity of topic types and limited coverage of tasks, making them incapable of effectively enhancing the entity linking capabilities of multi-modal models. To address these obstacles, we propose a dataset construction pipeline and publish $M^3EL$, a large-scale dataset for MEL. $M^3EL$ includes 79,625 instances, covering 9 diverse multi-modal tasks, and 5 different topics. In addition, to further improve the model's adaptability to multi-modal tasks, We propose a modality-augmented training strategy. Utilizing $M^3EL$ as a corpus, train the $\textit{CLIP}_{\textit{ND}}$ model based on $\textit{CLIP} (\textit{ViT}-\textit{B}-\textit{32})$, and conduct a comparative analysis with an existing multi-modal baselines. Experimental results show that the existing models perform far below expectations (ACC of 49.4%-75.8%), After analysis, it was obtained that small dataset sizes, insufficient modality task coverage, and limited topic diversity resulted in poor generalisation of multi-modal models. Our dataset effectively addresses these issues, and the $\textit{CLIP}_{\textit{ND}}$ model fine-tuned with $M^3EL$ shows a significant improvement in accuracy, with an average improvement of 9.3% to 25% across various tasks. Our dataset is available at https://anonymous.4open.science/r/M3EL.
Authors: Junxiao Shen, Khadija Khaldi, Enmin Zhou, Hemant Bhaskar Surale, Amy Karlson
Abstract: Text entry with word-gesture keyboards (WGK) is emerging as a popular method and becoming a key interaction for Extended Reality (XR). However, the diversity of interaction modes, keyboard sizes, and visual feedback in these environments introduces divergent word-gesture trajectory data patterns, thus leading to complexity in decoding trajectories into text. Template-matching decoding methods, such as SHARK^2, are commonly used for these WGK systems because they are easy to implement and configure. However, these methods are susceptible to decoding inaccuracies for noisy trajectories. While conventional neural-network-based decoders (neural decoders) trained on word-gesture trajectory data have been proposed to improve accuracy, they have their own limitations: they require extensive data for training and deep-learning expertise for implementation. To address these challenges, we propose a novel solution that combines ease of implementation with high decoding accuracy: a generalizable neural decoder enabled by pre-training on large-scale coarsely discretized word-gesture trajectories. This approach produces a ready-to-use WGK decoder that is generalizable across mid-air and on-surface WGK systems in augmented reality (AR) and virtual reality (VR), which is evident by a robust average Top-4 accuracy of 90.4% on four diverse datasets. It significantly outperforms SHARK^2 with a 37.2% enhancement and surpasses the conventional neural decoder by 7.4%. Moreover, the Pre-trained Neural Decoder's size is only 4 MB after quantization, without sacrificing accuracy, and it can operate in real-time, executing in just 97 milliseconds on Quest 3.
Authors: Junxiao Shen, Roger Boldu, Arpit Kalla, Michael Glueck, Hemant Bhaskar Surale Amy Karlson
Abstract: Text entry is a critical capability for any modern computing experience, with lightweight augmented reality (AR) glasses being no exception. Designed for all-day wearability, a limitation of lightweight AR glass is the restriction to the inclusion of multiple cameras for extensive field of view in hand tracking. This constraint underscores the need for an additional input device. We propose a system to address this gap: a ring-based mid-air gesture typing technique, RingGesture, utilizing electrodes to mark the start and end of gesture trajectories and inertial measurement units (IMU) sensors for hand tracking. This method offers an intuitive experience similar to raycast-based mid-air gesture typing found in VR headsets, allowing for a seamless translation of hand movements into cursor navigation. To enhance both accuracy and input speed, we propose a novel deep-learning word prediction framework, Score Fusion, comprised of three key components: a) a word-gesture decoding model, b) a spatial spelling correction model, and c) a lightweight contextual language model. In contrast, this framework fuses the scores from the three models to predict the most likely words with higher precision. We conduct comparative and longitudinal studies to demonstrate two key findings: firstly, the overall effectiveness of RingGesture, which achieves an average text entry speed of 27.3 words per minute (WPM) and a peak performance of 47.9 WPM. Secondly, we highlight the superior performance of the Score Fusion framework, which offers a 28.2% improvement in uncorrected Character Error Rate over a conventional word prediction framework, Naive Correction, leading to a 55.2% improvement in text entry speed for RingGesture. Additionally, RingGesture received a System Usability Score of 83 signifying its excellent usability.
Authors: Thea Aviss
Abstract: In this paper we present APEX-Embedding-7B (Advanced Processing for Epistemic eXtraction), a 7-billion parameter decoder-only text Feature Extraction Model, specifically designed for Document Retrieval-Augmented Generation (RAG) tasks. Our approach employs two training techniques that yield an emergent improvement in factual focus: (1) Pre-convergence interrupted fine-tuning using Structured Entity Relationship Maps as training data input: designed to shift the model's attention and create a bias towards factual content rather than semantic style - this enhances plain text performance despite not being directly trained for it; and (2) Model-Aware Contrastive Sampling, creating a balanced and evenly distributed collation map of hard and soft negatives directly informed by the base model's competency. This combined methodology yields significant improvements, enhancing plain text query/document pair retrieval to achieve an absolute rank@1 accuracy of 90.86% (an increase of 6.26% compared to the next leading model) in our evaluation, and reducing training data input context size by an average of 37.71% compared to plain text for both queries and document texts. Based on our evaluations, our model establishes a new state-of-the-art standard in text feature extraction for longer context document retrieval tasks.
Authors: Ivo Verhoeven, Pushkar Mishra, Ekaterina Shutova
Abstract: This paper introduces misinfo-general, a benchmark dataset for evaluating misinformation models' ability to perform out-of-distribution generalisation. Misinformation changes rapidly, much quicker than moderators can annotate at scale, resulting in a shift between the training and inference data distributions. As a result, misinformation models need to be able to perform out-of-distribution generalisation, an understudied problem in existing datasets. We identify 6 axes of generalisation-time, event, topic, publisher, political bias, misinformation type-and design evaluation procedures for each. We also analyse some baseline models, highlighting how these fail important desiderata.
Authors: Jiacong Zhou, Xianyun Wang, Jun Yu
Abstract: Aligning large language models with human preferences improves interaction quality and safety by ensuring outputs better reflect human values. A promising strategy involves Reinforcement Learning from Human Feedback (RLHF), starting with collecting and ranking responses generated by a supervised fine-tuning model to refine alignment. Current methods (DPO) focus on learning from pairwise preference data, categorizing responses into preferred and less preferred pairs, and optimizing by maximizing pairwise margins. Recent studies have uncovered a substantial discrepancy between the theoretical aspirations of preference learning and its real-world results. Current preference alignment techniques underperform expectations, with ranking accuracies below $60\%$ on standard datasets. This suggests existing methods inadequately capture ideal preference relationships within sequences. To address this challenge, this paper introduces \underline{D}irect \underline{R}anking \underline{P}reference \underline{O}ptimization (DRPO), a novel method that views human preference alignment as a Learning-to-Rank (LTR) task. DRPO leverages NDCG, a widely used LTR metric, to optimize the ranking of responses within lists based on preference data, thereby enhancing ranking accuracies. Due to the nondifferentiability of NDCG, we propose diffNDCG loss, a differentiable approximation facilitated by a sorting network to simulate NDCG. Furthermore, to improve the quality of generated response, we propose a novel margin-based Adaptive Rank Policy Score. Extensive experiments have shown that DRPO outperforms existing baseline methods, enhancing the quality of the generated responses.
Authors: Wei Ai, Jianbin Li, Ze Wang, Jiayi Du, Tao Meng, Yuntao Shou, Keqin Li
Abstract: Graph contrastive learning (GCL) has been widely applied to text classification tasks due to its ability to generate self-supervised signals from unlabeled data, thus facilitating model training. However, existing GCL-based text classification methods often suffer from negative sampling bias, where similar nodes are incorrectly paired as negative pairs. This can lead to over-clustering, where instances of the same class are divided into different clusters. To address the over-clustering issue, we propose an innovative GCL-based method of graph contrastive learning via cluster-refined negative sampling for semi-supervised text classification, namely ClusterText. Firstly, we combine the pre-trained model Bert with graph neural networks to learn text representations. Secondly, we introduce a clustering refinement strategy, which clusters the learned text representations to obtain pseudo labels. For each text node, its negative sample set is drawn from different clusters. Additionally, we propose a self-correction mechanism to mitigate the loss of true negative samples caused by clustering inconsistency. By calculating the Euclidean distance between each text node and other nodes within the same cluster, distant nodes are still selected as negative samples. Our proposed ClusterText demonstrates good scalable computing, as it can effectively extract important information from from a large amount of data. Experimental results demonstrate the superiority of ClusterText in text classification tasks.
Authors: Mayank Nagda, Phil Ostheimer, Sophie Fellenz
Abstract: Topic models are a popular approach for extracting semantic information from large document collections. However, recent studies suggest that the topics generated by these models often do not align well with human intentions. While metadata such as labels and authorship information is available, it has not yet been effectively incorporated into neural topic models. To address this gap, we introduce FANToM, a novel method for aligning neural topic models with both labels and authorship information. FANToM allows for the inclusion of this metadata when available, producing interpretable topics and author distributions for each topic. Our approach demonstrates greater expressiveness than conventional topic models by learning the alignment between labels, topics, and authors. Experimental results show that FANToM improves upon existing models in terms of both topic quality and alignment. Additionally, it identifies author interests and similarities.
Authors: Jingsheng Gao, Linxu Li, Weiyuan Li, Yuzhuo Fu, Bin Dai
Abstract: RAG systems consist of multiple modules to work together. However, these modules are usually separately trained. We argue that a system like RAG that incorporates multiple modules should be jointly optimized to achieve optimal performance. To demonstrate this, we design a specific pipeline called \textbf{SmartRAG} that includes a policy network and a retriever. The policy network can serve as 1) a decision maker that decides when to retrieve, 2) a query rewriter to generate a query most suited to the retriever, and 3) an answer generator that produces the final response with/without the observations. We then propose to jointly optimize the whole system using a reinforcement learning algorithm, with the reward designed to encourage the system to achieve the best performance with minimal retrieval cost. When jointly optimized, all the modules can be aware of how other modules are working and thus find the best way to work together as a complete system. Empirical results demonstrate that the jointly optimized SmartRAG can achieve better performance than separately optimized counterparts.
Authors: Elyas Obbad, Iddah Mlauzi, Brando Miranda, Rylan Schaeffer, Kamal Obbad, Suhana Bedi, Sanmi Koyejo
Abstract: Data selection is crucial for optimizing language model (LM) performance on specific tasks, yet most existing methods fail to effectively consider the target task distribution. Current approaches either ignore task-specific requirements entirely or rely on approximations that fail to capture the nuanced patterns needed for tasks like Autoformalization or code generation. Methods that do consider the target distribution often rely on simplistic, sometimes noisy, representations, like hashed n-gram features, which can lead to collisions and introduce noise. We introduce ZIP-FIT, a data selection framework that uses gzip compression to directly measure alignment between potential training data and the target task distribution. In extensive evaluations on Autoformalization and Python code generation, ZIP-FIT significantly outperforms leading baselines like DSIR and D4. Models trained on ZIP-FIT-selected data achieve their lowest cross-entropy loss up to 85.1\% faster than baselines, demonstrating that better task alignment leads to more efficient learning. In addition, ZIP-FIT performs selection up to 65.8\% faster than DSIR and two orders of magnitude faster than D4. Notably, ZIP-FIT shows that smaller, well-aligned datasets often outperform larger but less targeted ones, demonstrating that a small amount of higher quality data is superior to a large amount of lower quality data. Our results imply that task-aware data selection is crucial for efficient domain adaptation, and that compression offers a principled way to measure task alignment. By showing that targeted data selection can dramatically improve task-specific performance, our work provides new insights into the relationship between data quality, task alignment, and model learning efficiency.
Authors: Sudarshan Srinivasan, Maria Mahbub, Amir Sadovnik
Abstract: This position paper proposes a novel approach to advancing NLP security by leveraging Large Language Models (LLMs) as engines for generating diverse adversarial attacks. Building upon recent work demonstrating LLMs' effectiveness in creating word-level adversarial examples, we argue for expanding this concept to encompass a broader range of attack types, including adversarial patches, universal perturbations, and targeted attacks. We posit that LLMs' sophisticated language understanding and generation capabilities can produce more effective, semantically coherent, and human-like adversarial examples across various domains and classifier architectures. This paradigm shift in adversarial NLP has far-reaching implications, potentially enhancing model robustness, uncovering new vulnerabilities, and driving innovation in defense mechanisms. By exploring this new frontier, we aim to contribute to the development of more secure, reliable, and trustworthy NLP systems for critical applications.
Authors: Margaret Kroll, Kelsey Kraus
Abstract: The emergence of powerful LLMs has led to a paradigm shift in abstractive summarization of spoken documents. The properties that make LLMs so valuable for this task -- creativity, ability to produce fluent speech, and ability to abstract information from large corpora -- also present new challenges to evaluating their content. Quick, cost-effective automatic evaluations such as ROUGE and BERTScore offer promise, but do not yet show competitive performance when compared to human evaluations. We draw on methodologies from the social sciences to propose an evaluation paradigm for spoken document summarization explicitly tailored for generative AI content. We provide detailed evaluation criteria and best practices guidelines to ensure robustness in the experimental design, replicability, and trustworthiness of human evaluation studies. We additionally include two case studies that show how these human-in-the-loop evaluation methods have been implemented at a major U.S. technology company.
Authors: Michael Noukhovitch, Shengyi Huang, Sophie Xhonneux, Arian Hosseini, Rishabh Agarwal, Aaron Courville
Abstract: The dominant paradigm for RLHF is online and on-policy RL: synchronously generating from the large language model (LLM) policy, labelling with a reward model, and learning using feedback on the LLM's own outputs. While performant, this paradigm is computationally inefficient. Inspired by classical deep RL literature, we propose separating generation and learning in RLHF. This enables asynchronous generation of new samples while simultaneously training on old samples, leading to faster training and more compute-optimal scaling. However, asynchronous training relies on an underexplored regime, online but off-policy RLHF: learning on samples from previous iterations of our model. To understand the challenges in this regime, we investigate a fundamental question: how much off-policyness can we tolerate for asynchronous training to speed up learning but maintain performance? Among several RLHF algorithms we tested, we find that online DPO is most robust to off-policy data, and robustness increases with the scale of the policy model. We study further compute optimizations for asynchronous RLHF but find that they come at a performance cost, giving rise to a trade-off. Finally, we verify the scalability of asynchronous RLHF by training LLaMA 3.1 8B on an instruction-following task 40% faster than a synchronous run while matching final performance.
Authors: Lilian Wanzare, Joel Okutoyi, Maurine Kang'ahi, Mildred Ayere
Abstract: Kenyan Sign Language (KSL) is the primary language used by the deaf community in Kenya. It is the medium of instruction from Pre-primary 1 to university among deaf learners, facilitating their education and academic achievement. Kenyan Sign Language is used for social interaction, expression of needs, making requests and general communication among persons who are deaf in Kenya. However, there exists a language barrier between the deaf and the hearing people in Kenya. Thus, the innovation on AI4KSL is key in eliminating the communication barrier. Artificial intelligence for KSL is a two-year research project (2023-2024) that aims to create a digital open-access AI of spontaneous and elicited data from a representative sample of the Kenyan deaf community. The purpose of this study is to develop AI assistive technology dataset that translates English to KSL as a way of fostering inclusion and bridging language barriers among deaf learners in Kenya. Specific objectives are: Build KSL dataset for spoken English and video recorded Kenyan Sign Language and to build transcriptions of the KSL signs to a phonetic-level interface of the sign language. In this paper, the methodology for building the dataset is described. Data was collected from 48 teachers and tutors of the deaf learners and 400 learners who are Deaf. Participants engaged mainly in sign language elicitation tasks through reading and singing. Findings of the dataset consisted of about 14,000 English sentences with corresponding KSL Gloss derived from a pool of about 4000 words and about 20,000 signed KSL videos that are either signed words or sentences. The second level of data outcomes consisted of 10,000 split and segmented KSL videos. The third outcome of the dataset consists of 4,000 transcribed words into five articulatory parameters according to HamNoSys system.
Authors: Kexin Feng, Theodora Chaspari
Abstract: This study investigates explainable machine learning algorithms for identifying depression from speech. Grounded in evidence from speech production that depression affects motor control and vowel generation, pre-trained vowel-based embeddings, that integrate semantically meaningful linguistic units, are used. Following that, an ensemble learning approach decomposes the problem into constituent parts characterized by specific depression symptoms and severity levels. Two methods are explored: a "bottom-up" approach with 8 models predicting individual Patient Health Questionnaire-8 (PHQ-8) item scores, and a "top-down" approach using a Mixture of Experts (MoE) with a router module for assessing depression severity. Both methods depict performance comparable to state-of-the-art baselines, demonstrating robustness and reduced susceptibility to dataset mean/median values. System explainability benefits are discussed highlighting their potential to assist clinicians in depression diagnosis and screening.
Authors: Qinsi Wang, Saeed Vahidian, Hancheng Ye, Jianyang Gu, Jianyi Zhang, Yiran Chen
Abstract: Large language models (LLMs) with billions of parameters have sparked a new wave of exciting AI applications. However, their high computational costs and memory demands during inference pose significant challenges. Adaptive sparse activation inference, which activates only a small number of neurons for each token, offers a novel way to accelerate model inference without degrading performance, showing great potential for resource-constrained hardware devices. Nevertheless, existing methods predict activated neurons based on individual tokens with additional MLP, which involve frequent changes in activation maps and resource calls, limiting the acceleration benefits of sparse activation. In this paper, we introduce CoreInfer, an MLP-free adaptive sparse activation inference method based on sentence-level prediction. Specifically, we propose the concept of sentence-wise core neurons, which refers to the subset of neurons most critical for a given sentence, and empirically demonstrate its effectiveness. To determine the core neurons, we explore the correlation between core neurons and the sentence's semantics. Remarkably, we discovered that core neurons exhibit both stability and similarity in relation to the sentence's semantics -- an insight overlooked by previous studies. Building on this finding, we further design two semantic-based methods for predicting core neurons to fit different input scenarios. In CoreInfer, the core neurons are determined during the pre-filling stage and fixed during the encoding stage, enabling zero-cost sparse inference. We evaluated the model generalization and task generalization of CoreInfer across various models and tasks. Notably, on an NVIDIA TITAN XP GPU, CoreInfer achieved a 10.33 times and 2.72 times speedup compared to the Huggingface implementation and PowerInfer, respectively.
Authors: Shanchao Liang, Nan Jiang, Shangshu Qian, Lin Tan
Abstract: Web development involves turning UI designs into functional webpages, which can be difficult for both beginners and experienced developers due to the complexity of HTML's hierarchical structures and styles. While Large Language Models (LLMs) have shown promise in generating source code, two major challenges persist in UI-to-HTML code generation: (1) effectively representing HTML's hierarchical structure for LLMs, and (2) bridging the gap between the visual nature of UI designs and the text-based format of HTML code. To tackle these challenges, we introduce Waffle, a new fine-tuning strategy that uses a structure-aware attention mechanism to improve LLMs' understanding of HTML's structure and a contrastive fine-tuning approach to align LLMs' understanding of UI images and HTML code. Models fine-tuned with Waffle show up to 9.00 pp (percentage point) higher HTML match, 0.0982 higher CW-SSIM, 32.99 higher CLIP, and 27.12 pp higher LLEM on our new benchmark WebSight-Test and an existing benchmark Design2Code, outperforming current fine-tuning methods.
Authors: Chris Yuhao Liu, Liang Zeng, Jiacai Liu, Rui Yan, Jujie He, Chaojie Wang, Shuicheng Yan, Yang Liu, Yahui Zhou
Abstract: In this report, we introduce a collection of methods to enhance reward modeling for LLMs, focusing specifically on data-centric techniques. We propose effective data selection and filtering strategies for curating high-quality open-source preference datasets, culminating in the Skywork-Reward data collection, which contains only 80K preference pairs -- significantly smaller than existing datasets. Using this curated dataset, we developed the Skywork-Reward model series -- Skywork-Reward-Gemma-27B and Skywork-Reward-Llama-3.1-8B -- with the former currently holding the top position on the RewardBench leaderboard. Notably, our techniques and datasets have directly enhanced the performance of many top-ranked models on RewardBench, highlighting the practical impact of our contributions in real-world preference learning applications.
Authors: Yifei Yang, Zouying Cao, Qiguang Chen, Libo Qin, Dongjie Yang, Hai Zhao, Zhi Chen
Abstract: The development of large language models (LLMs) has significantly expanded model sizes, resulting in substantial GPU memory requirements during inference. The key and value storage of the attention map in the KV (key-value) cache accounts for more than 80\% of this memory consumption. Nowadays, most existing KV cache compression methods focus on intra-layer compression within a single Transformer layer but few works consider layer-wise compression. In this paper, we propose a plug-and-play method called \textit{KVSharer}, which shares the KV cache between layers to achieve layer-wise compression. Rather than intuitively sharing based on higher similarity, we discover a counterintuitive phenomenon: sharing dissimilar KV caches better preserves the model performance. Experiments show that \textit{KVSharer} can reduce KV cache computation by 30\%, thereby lowering memory consumption without significantly impacting model performance and it can also achieve at least 1.3 times generation acceleration. Additionally, we verify that \textit{KVSharer} is compatible with existing intra-layer KV cache compression methods, and combining both can further save memory.
Authors: Jean-Marc Luck, Anita Mehta
Abstract: We present a model of speech perception which takes into account effects of correlations between sounds. Words in this model correspond to the attractors of a suitably chosen descent dynamics. The resulting lexicon is rich in short words, and much less so in longer ones, as befits a reasonable word length distribution. We separately examine the decryption of short and long words in the presence of mishearings. In the regime of short words, the algorithm either quickly retrieves a word, or proposes another valid word. In the regime of longer words, the behaviour is markedly different. While the successful decryption of words continues to be relatively fast, there is a finite probability of getting lost permanently, as the algorithm wanders round the landscape of suitable words without ever settling on one.
Authors: Woosung Koh, Jang Han Yoon, MinHyung Lee, Youngjin Song, Jaegwan Cho, Jaehyun Kang, Taehyeon Kim, Se-young Yun, Youngjae Yu, Bongshin Lee
Abstract: Generating high-quality charts with Large Language Models presents significant challenges due to limited data and the high cost of scaling through human curation. Instruction, data, and code triplets are scarce and expensive to manually curate as their creation demands technical expertise. To address this scalability issue, we introduce a reference-free automatic feedback generator, which eliminates the need for costly human intervention. Our novel framework, $C^2$, consists of (1) an automatic feedback provider (ChartAF) and (2) a diverse, reference-free dataset (ChartUIE-8K). Quantitative results are compelling: in our first experiment, 74% of respondents strongly preferred, and 10% preferred, the results after feedback. The second post-feedback experiment demonstrates that ChartAF outperforms nine baselines. Moreover, ChartUIE-8K significantly improves data diversity by increasing queries, datasets, and chart types by 5982%, 1936%, and 91%, respectively, over benchmarks. Finally, an LLM user study revealed that 94% of participants preferred ChartUIE-8K's queries, with 93% deeming them aligned with real-world use cases. Core contributions are available as open-source at an anonymized project site, with ample qualitative examples.
Authors: Vasiliki Papanikou, Panagiotis Papadakos, Theodora Karamanidou, Thanos G. Stavropoulos, Evaggelia Pitoura, Panayiotis Tsaparas
Abstract: In this paper, we present a comprehensive survey on the pervasive issue of medical misinformation in social networks from the perspective of information technology. The survey aims at providing a systematic review of related research and helping researchers and practitioners navigate through this fast-changing field. Specifically, we first present manual and automatic approaches for fact-checking. We then explore fake news detection methods, using content, propagation features, or source features, as well as mitigation approaches for countering the spread of misinformation. We also provide a detailed list of several datasets on health misinformation and of publicly available tools. We conclude the survey with a discussion on the open challenges and future research directions in the battle against health misinformation.
Authors: Ankit Singh Rawat, Veeranjaneyulu Sadhanala, Afshin Rostamizadeh, Ayan Chakrabarti, Wittawat Jitkrittum, Vladimir Feinberg, Seungyeon Kim, Hrayr Harutyunyan, Nikunj Saunshi, Zachary Nado, Rakesh Shivanna, Sashank J. Reddi, Aditya Krishna Menon, Rohan Anil, Sanjiv Kumar
Abstract: A primary challenge in large language model (LLM) development is their onerous pre-training cost. Typically, such pre-training involves optimizing a self-supervised objective (such as next-token prediction) over a large corpus. This paper explores a promising paradigm to improve LLM pre-training efficiency and quality by suitably leveraging a small language model (SLM). In particular, this paradigm relies on an SLM to both (1) provide soft labels as additional training supervision, and (2) select a small subset of valuable ("informative" and "hard") training examples. Put together, this enables an effective transfer of the SLM's predictive distribution to the LLM, while prioritizing specific regions of the training data distribution. Empirically, this leads to reduced LLM training time compared to standard training, while improving the overall quality. Theoretically, we develop a statistical framework to systematically study the utility of SLMs in enabling efficient training of high-quality LLMs. In particular, our framework characterizes how the SLM's seemingly low-quality supervision can enhance the training of a much more capable LLM. Furthermore, it also highlights the need for an adaptive utilization of such supervision, by striking a balance between the bias and variance introduced by the SLM-provided soft labels. We corroborate our theoretical framework by improving the pre-training of an LLM with 2.8B parameters by utilizing a smaller LM with 1.5B parameters on the Pile dataset.
Authors: Yuxing Chen, Weijie Wang, Sylvain Lobry, Camille Kurtz
Abstract: Large language models (LLMs) are being used in data science code generation tasks, but they often struggle with complex sequential tasks, leading to logical errors. Their application to geospatial data processing is particularly challenging due to difficulties in incorporating complex data structures and spatial constraints, effectively utilizing diverse function calls, and the tendency to hallucinate less-used geospatial libraries. To tackle these problems, we introduce GeoAgent, a new interactive framework designed to help LLMs handle geospatial data processing more effectively. GeoAgent pioneers the integration of a code interpreter, static analysis, and Retrieval-Augmented Generation (RAG) techniques within a Monte Carlo Tree Search (MCTS) algorithm, offering a novel approach to geospatial data processing. In addition, we contribute a new benchmark specifically designed to evaluate the LLM-based approach in geospatial tasks. This benchmark leverages a variety of Python libraries and includes both single-turn and multi-turn tasks such as data acquisition, data analysis, and visualization. By offering a comprehensive evaluation among diverse geospatial contexts, this benchmark sets a new standard for developing LLM-based approaches in geospatial data analysis tasks. Our findings suggest that relying solely on knowledge of LLM is insufficient for accurate geospatial task programming, which requires coherent multi-step processes and multiple function calls. Compared to the baseline LLMs, the proposed GeoAgent has demonstrated superior performance, yielding notable improvements in function calls and task completion. In addition, these results offer valuable insights for the future development of LLM agents in automatic geospatial data analysis task programming.
Authors: Zheyuan Zhang
Abstract: Substantial research on deep learning-based emergent communication uses the referential game framework, specifically the Lewis signaling game, however we argue that successful communication in this game typically only need one or two effective symbols (i.e. message length) because of a sampling pitfall in the training data. To address this issue, we provide a theoretical analysis and introduce a combinatorial algorithm SolveMinSym (SMS) to determine the minimum number of symbols for successful communication min(|M|) in the Lewis signaling game. We use SMS algorithm to create datasets with different min(|M|) to empirically show that higher min(|M|) for the training data increases the number of effective symbols in the emergent language.
Authors: Qiao Jin, Nicholas Wan, Robert Leaman, Shubo Tian, Zhizheng Wang, Yifan Yang, Zifeng Wang, Guangzhi Xiong, Po-Ting Lai, Qingqing Zhu, Benjamin Hou, Maame Sarfo-Gyamfi, Gongbo Zhang, Aidan Gilson, Balu Bhasuran, Zhe He, Aidong Zhang, Jimeng Sun, Chunhua Weng, Ronald M. Summers, Qingyu Chen, Yifan Peng, Zhiyong Lu
Abstract: Large language models (LLMs) represent a transformative class of AI tools capable of revolutionizing various aspects of healthcare by generating human-like responses across diverse contexts and adapting to novel tasks following human instructions. Their potential application spans a broad range of medical tasks, such as clinical documentation, matching patients to clinical trials, and answering medical questions. In this primer paper, we propose an actionable guideline to help healthcare professionals more efficiently utilize LLMs in their work, along with a set of best practices. This approach consists of several main phases, including formulating the task, choosing LLMs, prompt engineering, fine-tuning, and deployment. We start with the discussion of critical considerations in identifying healthcare tasks that align with the core capabilities of LLMs and selecting models based on the selected task and data, performance requirements, and model interface. We then review the strategies, such as prompt engineering and fine-tuning, to adapt standard LLMs to specialized medical tasks. Deployment considerations, including regulatory compliance, ethical guidelines, and continuous monitoring for fairness and bias, are also discussed. By providing a structured step-by-step methodology, this tutorial aims to equip healthcare professionals with the tools necessary to effectively integrate LLMs into clinical practice, ensuring that these powerful technologies are applied in a safe, reliable, and impactful manner.
Authors: Miranda Christ, Sam Gunn, Tal Malkin, Mariana Raykova
Abstract: The recent explosion of high-quality language models has necessitated new methods for identifying AI-generated text. Watermarking is a leading solution and could prove to be an essential tool in the age of generative AI. Existing approaches embed watermarks at inference and crucially rely on the large language model (LLM) specification and parameters being secret, which makes them inapplicable to the open-source setting. In this work, we introduce the first watermarking scheme for open-source LLMs. Our scheme works by modifying the parameters of the model, but the watermark can be detected from just the outputs of the model. Perhaps surprisingly, we prove that our watermarks are unremovable under certain assumptions about the adversary's knowledge. To demonstrate the behavior of our construction under concrete parameter instantiations, we present experimental results with OPT-6.7B and OPT-1.3B. We demonstrate robustness to both token substitution and perturbation of the model parameters. We find that the stronger of these attacks, the model-perturbation attack, requires deteriorating the quality score to 0 out of 100 in order to bring the detection rate down to 50%.
Authors: Sha Li, Revanth Gangi Reddy, Khanh Duy Nguyen, Qingyun Wang, May Fung, Chi Han, Jiawei Han, Kartik Natarajan, Clare R. Voss, Heng Ji
Abstract: Complex news events, such as natural disasters and socio-political conflicts, require swift responses from the government and society. Relying on historical events to project the future is insufficient as such events are sparse and do not cover all possible conditions and nuanced situations. Simulation of these complex events can help better prepare and reduce the negative impact. We develop a controllable complex news event simulator guided by both the event schema representing domain knowledge about the scenario and user-provided assumptions representing case-specific conditions. As event dynamics depend on the fine-grained social and cultural context, we further introduce a geo-diverse commonsense and cultural norm-aware knowledge enhancement component. To enhance the coherence of the simulation, apart from the global timeline of events, we take an agent-based approach to simulate the individual character states, plans, and actions. By incorporating the schema and cultural norms, our generated simulations achieve much higher coherence and appropriateness and are received favorably by participants from a humanitarian assistance organization.
Authors: Xiaoqiang Wang, Bang Liu
Abstract: Large language models (LLMs) and large multimodal models (LMMs) have shown great potential in automating complex tasks like web browsing and gaming. However, their ability to generalize across diverse applications remains limited, hindering broader utility. To address this challenge, we present OSCAR: Operating System Control via state-Aware reasoning and Re-planning. OSCAR is a generalist agent designed to autonomously navigate and interact with various desktop and mobile applications through standardized controls, such as mouse and keyboard inputs, while processing screen images to fulfill user commands. OSCAR translates human instructions into executable Python code, enabling precise control over graphical user interfaces (GUIs). To enhance stability and adaptability, OSCAR operates as a state machine, equipped with error-handling mechanisms and dynamic task re-planning, allowing it to efficiently adjust to real-time feedback and exceptions. We demonstrate OSCAR's effectiveness through extensive experiments on diverse benchmarks across desktop and mobile platforms, where it transforms complex workflows into simple natural language commands, significantly boosting user productivity. Our code will be open-source upon publication.
Authors: Zhangheng Li, Keen You, Haotian Zhang, Di Feng, Harsh Agrawal, Xiujun Li, Mohana Prasad Sathya Moorthy, Jeff Nichols, Yinfei Yang, Zhe Gan
Abstract: Building a generalist model for user interface (UI) understanding is challenging due to various foundational issues, such as platform diversity, resolution variation, and data limitation. In this paper, we introduce Ferret-UI 2, a multimodal large language model (MLLM) designed for universal UI understanding across a wide range of platforms, including iPhone, Android, iPad, Webpage, and AppleTV. Building on the foundation of Ferret-UI, Ferret-UI 2 introduces three key innovations: support for multiple platform types, high-resolution perception through adaptive scaling, and advanced task training data generation powered by GPT-4o with set-of-mark visual prompting. These advancements enable Ferret-UI 2 to perform complex, user-centered interactions, making it highly versatile and adaptable for the expanding diversity of platform ecosystems. Extensive empirical experiments on referring, grounding, user-centric advanced tasks (comprising 9 subtasks $\times$ 5 platforms), GUIDE next-action prediction dataset, and GUI-World multi-platform benchmark demonstrate that Ferret-UI 2 significantly outperforms Ferret-UI, and also shows strong cross-platform transfer capabilities.
Authors: Jialu Li, Yuanzhen Li, Neal Wadhwa, Yael Pritch, David E. Jacobs, Michael Rubinstein, Mohit Bansal, Nataniel Ruiz
Abstract: We introduce the concept of a generative infinite game, a video game that transcends the traditional boundaries of finite, hard-coded systems by using generative models. Inspired by James P. Carse's distinction between finite and infinite games, we leverage recent advances in generative AI to create Unbounded: a game of character life simulation that is fully encapsulated in generative models. Specifically, Unbounded draws inspiration from sandbox life simulations and allows you to interact with your autonomous virtual character in a virtual world by feeding, playing with and guiding it - with open-ended mechanics generated by an LLM, some of which can be emergent. In order to develop Unbounded, we propose technical innovations in both the LLM and visual generation domains. Specifically, we present: (1) a specialized, distilled large language model (LLM) that dynamically generates game mechanics, narratives, and character interactions in real-time, and (2) a new dynamic regional image prompt Adapter (IP-Adapter) for vision models that ensures consistent yet flexible visual generation of a character across multiple environments. We evaluate our system through both qualitative and quantitative analysis, showing significant improvements in character life simulation, user instruction following, narrative coherence, and visual consistency for both characters and the environments compared to traditional related approaches.
Authors: Sara Ghaboura, Ahmed Heakl, Omkar Thawakar, Ali Alharthi, Ines Riahi, Abduljalil Saif, Jorma Laaksonen, Fahad S. Khan, Salman Khan, Rao M. Anwer
Abstract: Recent years have witnessed a significant interest in developing large multimodal models (LMMs) capable of performing various visual reasoning and understanding tasks. This has led to the introduction of multiple LMM benchmarks to evaluate LMMs on different tasks. However, most existing LMM evaluation benchmarks are predominantly English-centric. In this work, we develop a comprehensive LMM evaluation benchmark for the Arabic language to represent a large population of over 400 million speakers. The proposed benchmark, named CAMEL-Bench, comprises eight diverse domains and 38 sub-domains including, multi-image understanding, complex visual perception, handwritten document understanding, video understanding, medical imaging, plant diseases, and remote sensing-based land use understanding to evaluate broad scenario generalizability. Our CAMEL-Bench comprises around 29,036 questions that are filtered from a larger pool of samples, where the quality is manually verified by native speakers to ensure reliable model assessment. We conduct evaluations of both closed-source, including GPT-4 series, and open-source LMMs. Our analysis reveals the need for substantial improvement, especially among the best open-source models, with even the closed-source GPT-4o achieving an overall score of 62%. Our benchmark and evaluation scripts are open-sourced.
Authors: Ali Modarressi, Ayyoob Imani, Mohsen Fayyaz, Hinrich Sch\"utze
Abstract: Large language models (LLMs) have significantly advanced the field of natural language processing (NLP) through their extensive parameters and comprehensive data utilization. However, existing LLMs lack a dedicated memory unit, limiting their ability to explicitly store and retrieve knowledge for various tasks. In this paper, we propose RET-LLM a novel framework that equips LLMs with a general write-read memory unit, allowing them to extract, store, and recall knowledge from the text as needed for task performance. Inspired by Davidsonian semantics theory, we extract and save knowledge in the form of triplets. The memory unit is designed to be scalable, aggregatable, updatable, and interpretable. Through qualitative evaluations, we demonstrate the superiority of our proposed framework over baseline approaches in question answering tasks. Moreover, our framework exhibits robust performance in handling temporal-based question answering tasks, showcasing its ability to effectively manage time-dependent information.
Authors: Zouying Cao, Yifei Yang, Hai Zhao
Abstract: Large Language Models (LLMs) suffer from huge number of parameters, which restricts their deployment on edge devices. Weight sharing is one promising solution that encourages weight reuse, effectively reducing memory usage with less performance drop. However, current weight sharing techniques primarily focus on small-scale models like BERT and employ coarse-grained sharing rules, e.g., layer-wise. This becomes limiting given the prevalence of LLMs and sharing an entire layer or block obviously diminishes the flexibility of weight sharing. In this paper, we present a perspective on head-wise shareable attention for large language models. We further propose two memory-efficient methods that share parameters across attention heads, with a specific focus on LLMs. Both of them use the same dynamic strategy to select the shared weight matrices. The first method directly reuses the pre-trained weights without retraining, denoted as $\textbf{DirectShare}$. The second method first post-trains with constraint on weight matrix similarity and then shares, denoted as $\textbf{PostShare}$. Experimental results reveal our head-wise shared models still maintain satisfactory capabilities, demonstrating the feasibility of fine-grained weight sharing applied to LLMs.
Authors: Lang Cao, Jimeng Sun, Adam Cross
Abstract: Rare diseases affect millions worldwide but often face limited research focus due to their low prevalence. This results in prolonged diagnoses and a lack of approved therapies. Recent advancements in Large Language Models (LLMs) have shown promise in automating the extraction of medical information, offering potential to improve medical diagnosis and management. However, most LLMs lack professional medical knowledge, especially concerning rare diseases, and struggle to handle the latest rare disease information. They also cannot effectively manage rare disease data and are not directly suitable for diagnosis and management tasks. Our objective is to create an end-to-end system called AutoRD, which automates the extraction of information from medical texts about rare diseases, focusing on entities and their relations. AutoRD integrates up-to-date structured knowledge and demonstrates superior performance in rare disease extraction tasks. We conduct various experiments to evaluate AutoRD's performance, aiming to surpass common LLMs and traditional methods.
Authors: Yixin Wan, Kai-Wei Chang
Abstract: Social biases can manifest in language agency. While several studies approached agency-related bias in human-written language, very limited research has investigated such biases in Large Language Model (LLM)-generated content. In addition, previous works often rely on string-matching techniques to identify agentic and communal words within texts, which fall short of accurately classifying language agency. We introduce the novel Language Agency Bias Evaluation (LABE) benchmark, which comprehensively evaluates biases in LLMs by analyzing agency levels attributed to different demographic groups in model generations. LABE leverages 5,400 template-based prompts, an accurate agency classifier, and corresponding bias metrics to test for gender, racial, and intersectional language agency biases in LLMs on 3 text generation tasks: biographies, professor reviews, and reference letters. We also contribute the Language Agency Classification (LAC) dataset, consisting of 3,724 agentic and communal sentences. Using LABE, we unveil language agency social biases in 3 recent LLMs: ChatGPT, Llama3, and Mistral. We observe that: (1) LLM generations tend to demonstrate greater gender bias than human-written texts; (2) Models demonstrate remarkably higher levels of intersectional bias than the other bias aspects. Those who are at the intersection of gender and racial minority groups--such as Black females--are consistently described by texts with lower levels of agency, aligning with real-world social inequalities; (3) Among the 3 LLMs investigated, Llama3 demonstrates the greatest overall bias; (4) Not only does prompt-based mitigation fail to resolve language agency bias in LLMs, but it frequently leads to the exacerbation of biases in generated texts.
Authors: Jo\~ao Monteiro, \'Etienne Marcotte, Pierre-Andr\'e No\"el, Valentina Zantedeschi, David V\'azquez, Nicolas Chapados, Christopher Pal, Perouz Taslakian
Abstract: In-context learning (ICL) approaches typically leverage prompting to condition decoder-only language model generation on reference information. Just-in-time processing of a context is inefficient due to the quadratic cost of self-attention operations, and caching is desirable. However, caching transformer states can easily require almost as much space as the model parameters. When the right context isn't known in advance, caching ICL can be challenging. This work addresses these limitations by introducing models that, inspired by the encoder-decoder architecture, use cross-attention to condition generation on reference text without the prompt. More precisely, we leverage pre-trained decoder-only models and only train a small number of added layers. We use Question-Answering (QA) as a testbed to evaluate the ability of our models to perform conditional generation and observe that they outperform ICL, are comparable to fine-tuned prompted LLMs, and drastically reduce the space footprint relative to standard KV caching by two orders of magnitude.
Authors: Yunjia Qi, Hao Peng, Xiaozhi Wang, Bin Xu, Lei Hou, Juanzi Li
Abstract: Large language models (LLMs) usually fall short on information extraction (IE) tasks and struggle to follow the complex instructions of IE tasks. This primarily arises from LLMs not being aligned with humans, as mainstream alignment datasets typically do not include IE data. In this paper, we introduce ADELIE (Aligning large language moDELs on Information Extraction), an aligned LLM that effectively solves various IE tasks, including closed IE, open IE, and on-demand IE. We first collect and construct a high-quality alignment corpus IEInstruct for IE. Then we train ADELIE_SFT using instruction tuning on IEInstruct. We further train ADELIE_SFT with direct preference optimization (DPO) objective, resulting in ADELIE_DPO. Extensive experiments on various held-out IE datasets demonstrate that our models (ADELIE_SFT and ADELIE_DPO) achieve state-of-the-art (SoTA) performance among open-source models. We further explore the general capabilities of ADELIE, and experimental results reveal that their general capabilities do not exhibit a noticeable decline. We will release the code, data, and models to facilitate further research.
Authors: Jianqiao Lu, Zhiyang Dou, Hongru Wang, Zeyu Cao, Jianbo Dai, Yingjia Wan, Zhijiang Guo
Abstract: In this work, we propose a novel method named \textbf{Auto}mated \textbf{P}rocess-\textbf{S}upervised \textbf{V}erifier (\textbf{\textsc{AutoPSV}}) to enhance the reasoning capabilities of large language models (LLMs) by automatically annotating the reasoning steps. \textsc{AutoPSV} begins by training a verification model on the correctness of final answers, enabling it to generate automatic process annotations. This verification model assigns a confidence score to each reasoning step, indicating the probability of arriving at the correct final answer from that point onward. We detect relative changes in the verification's confidence scores across reasoning steps to automatically annotate the reasoning process, enabling error detection even in scenarios where ground truth answers are unavailable. This alleviates the need for numerous manual annotations or the high computational costs associated with model-induced annotation approaches. We experimentally validate that the step-level confidence changes learned by the verification model trained on the final answer correctness can effectively identify errors in the reasoning steps. We demonstrate that the verification model, when trained on process annotations generated by \textsc{AutoPSV}, exhibits improved performance in selecting correct answers from multiple LLM-generated outputs. Notably, we achieve substantial improvements across five datasets in mathematics and commonsense reasoning. The source code of \textsc{AutoPSV} is available at \url{https://github.com/rookie-joe/AutoPSV}.
Authors: Hongfu Gao, Feipeng Zhang, Wenyu Jiang, Jun Shu, Feng Zheng, Hongxin Wei
Abstract: Large language models (LLMs) have shown impressive performance on downstream tasks by in-context learning (ICL), which heavily relies on the quality of demonstrations selected from a large set of annotated examples. Recent works claim that in-context learning is robust to noisy demonstrations in text classification. In this work, we show that, on text generation tasks, noisy annotations significantly hurt the performance of in-context learning. To circumvent the issue, we propose a simple and effective approach called Local Perplexity Ranking (LPR), which replaces the "noisy" candidates with their nearest neighbors that are more likely to be clean. Our method is motivated by analyzing the perplexity deviation caused by noisy labels and decomposing perplexity into inherent perplexity and matching perplexity. Our key idea behind LPR is thus to decouple the matching perplexity by performing the ranking among the neighbors in semantic space. Our approach can prevent the selected demonstrations from including mismatched input-label pairs while preserving the effectiveness of the original selection methods. Extensive experiments demonstrate the effectiveness of LPR, improving the EM score by up to 18.75 on common benchmarks with noisy annotations. Our code is available at https://github.com/ml-stat-Sustech/Local-Perplexity-Ranking.
URLs: https://github.com/ml-stat-Sustech/Local-Perplexity-Ranking.
Authors: Roberto Ceraolo, Dmitrii Kharlapenko, Ahmad Khan, Am\'elie Reymond, Rada Mihalcea, Bernhard Sch\"olkopf, Mrinmaya Sachan, Zhijing Jin
Abstract: The recent development of Large Language Models (LLMs) has changed our role in interacting with them. Instead of primarily testing these models with questions we already know the answers to, we now use them to explore questions where the answers are unknown to us. This shift, which hasn't been fully addressed in existing datasets, highlights the growing need to understand naturally occurring human questions - that are more complex, open-ended, and reflective of real-world needs. To this end, we present NatQuest, a collection of 13,500 naturally occurring questions from three diverse sources: human-to-search-engine queries, human-to-human interactions, and human-to-LLM conversations. Our comprehensive collection enables a rich understanding of human curiosity across various domains and contexts. Our analysis reveals a significant presence of causal questions (up to 42%) within the dataset, for which we develop an iterative prompt improvement framework to identify all causal queries, and examine their unique linguistic properties, cognitive complexity, and source distribution. We also lay the groundwork to explore LLM performance on these questions and provide six efficient classification models to identify causal questions at scale for future work.
Authors: Kiseung Kim, Jay-Yoon Lee
Abstract: The Retrieval Augmented Generation (RAG) framework utilizes a combination of parametric knowledge and external knowledge to demonstrate state-of-the-art performance on open-domain question answering tasks. However, the RAG framework suffers from performance degradation when the query is accompanied by irrelevant contexts. In this work, we propose the RE-RAG framework, which introduces a relevance estimator (RE) that not only provides relative relevance between contexts as previous rerankers did, but also provides confidence, which can be used to classify whether given context is useful for answering the given question. We propose a weakly supervised method for training the RE simply utilizing question-answer data without any labels for correct contexts. We show that RE trained with a small generator (sLM) can not only improve the sLM fine-tuned together with RE but also improve previously unreferenced large language models (LLMs). Furthermore, we investigate new decoding strategies that utilize the proposed confidence measured by RE such as choosing to let the user know that it is "unanswerable" to answer the question given the retrieved contexts or choosing to rely on LLM's parametric knowledge rather than unrelated contexts.
Authors: Renhao Li, Minghuan Tan, Derek F. Wong, Min Yang
Abstract: In recent years, instruction fine-tuning (IFT) on large language models (LLMs) has garnered considerable attention to enhance model performance on unseen tasks. Attempts have been made on automatic construction and effective selection for IFT data. However, we posit that previous methods have not fully harnessed the potential of LLMs for enhancing data quality. The responses within IFT data could be further enhanced by leveraging the capabilities of LLMs themselves. In this paper, we propose CoEvol, an LLM-based multi-agent cooperation framework for the improvement of responses to instructions. To effectively refine the responses, we develop an iterative framework following a debate-advise-edit-judge paradigm. A two-stage multi-agent debate strategy is further devised to ensure the diversity and reliability of editing suggestions within the framework. Empirically, models equipped with CoEvol outperform competitive baselines evaluated by MT-Bench and AlpacaEval, demonstrating its effectiveness in enhancing instruction-following capabilities for LLMs.
Authors: Saranya Venkatraman, Nafis Irtiza Tripto, Dongwon Lee
Abstract: The rise of unifying frameworks that enable seamless interoperability of Large Language Models (LLMs) has made LLM-LLM collaboration for open-ended tasks a possibility. Despite this, there have not been efforts to explore such collaborative writing. We take the next step beyond human-LLM collaboration to explore this multi-LLM scenario by generating the first exclusively LLM-generated collaborative stories dataset called CollabStory. We focus on single-author ($N=1$) to multi-author (up to $N=5$) scenarios, where multiple LLMs co-author stories. We generate over 32k stories using open-source instruction-tuned LLMs. Further, we take inspiration from the PAN tasks that have set the standard for human-human multi-author writing tasks and analysis. We extend their authorship-related tasks for multi-LLM settings and present baselines for LLM-LLM collaboration. We find that current baselines are not able to handle this emerging scenario. Thus, CollabStory is a resource that could help propel an understanding as well as the development of techniques to discern the use of multiple LLMs. This is crucial to study in the context of writing tasks since LLM-LLM collaboration could potentially overwhelm ongoing challenges related to plagiarism detection, credit assignment, maintaining academic integrity in educational settings, and addressing copyright infringement concerns. We make our dataset and code available at \texttt{\url{https://github.com/saranya-venkatraman/multi_llm_story_writing}}.
URLs: https://github.com/saranya-venkatraman/multi_llm_story_writing
Authors: Ujjwala Anantheswaran, Himanshu Gupta, Kevin Scaria, Shreyas Verma, Chitta Baral, Swaroop Mishra
Abstract: Large Language Models (LLMs) excel at various tasks, including solving math word problems (MWPs), but struggle with real-world problems containing irrelevant information. To address this, we propose a prompting framework that generates adversarial variants of MWPs by adding irrelevant variables. We introduce a dataset, PROBLEMATHIC, containing both adversarial and non-adversarial MWPs. Our experiments reveal that LLMs are susceptible to distraction by numerical noise, resulting in an average relative performance drop of ~26% on adversarial MWPs. To mitigate this, we fine-tune LLMs (Llama-2, Mistral) on the adversarial samples from our dataset. Fine-tuning on adversarial training instances improves performance on adversarial MWPs by ~8%, indicating increased robustness to noise and improved ability to identify relevant data for reasoning. Finally, to assess the generalizability of our prompting framework, we introduce GSM-8K-Adv, an adversarial variant of the GSM-8K benchmark. LLMs continue to struggle when faced with adversarial information, reducing performance by up to 6%.
Authors: Andrea Posada, Daniel Rueckert, Felix Meissen, Philip M\"uller
Abstract: Since the Transformer architecture emerged, language model development has grown, driven by their promising potential. Releasing these models into production requires properly understanding their behavior, particularly in sensitive domains like medicine. Despite this need, the medical literature still lacks practical assessment of pre-trained language models, which are especially valuable in settings where only consumer-grade computational resources are available. To address this gap, we have conducted a comprehensive survey of language models in the medical field and evaluated a subset of these for medical text classification and conditional text generation. The subset includes 53 models with 110 million to 13 billion parameters, spanning the Transformer-based model families and knowledge domains. Different approaches are employed for text classification, including zero-shot learning, enabling tuning without the need to train the model. These approaches are helpful in our target settings, where many users of language models find themselves. The results reveal remarkable performance across the tasks and datasets evaluated, underscoring the potential of certain models to contain medical knowledge, even without domain specialization. This study thus advocates for further exploration of model applications in medical contexts, particularly in computational resource-constrained settings, to benefit a wide range of users. The code is available on https://github.com/anpoc/Language-models-in-medicine.
Authors: Mihai Masala, Denis C. Ilie-Ablachim, Alexandru Dima, Dragos Corlatescu, Miruna Zavelca, Ovio Olaru, Simina Terian, Andrei Terian, Marius Leordeanu, Horia Velicu, Marius Popescu, Mihai Dascalu, Traian Rebedea
Abstract: In recent years, Large Language Models (LLMs) have achieved almost human-like performance on various tasks. While some LLMs have been trained on multilingual data, most of the training data is in English; hence, their performance in English greatly exceeds other languages. To our knowledge, we are the first to collect and translate a large collection of texts, instructions, and benchmarks and train, evaluate, and release open-source LLMs tailored for Romanian. We evaluate our methods on four different categories, including academic benchmarks, MT-Bench (manually translated), and a professionally built historical, cultural, and social benchmark adapted to Romanian. We argue for the usefulness and high performance of RoLLMs by obtaining state-of-the-art results across the board. We publicly release all resources (i.e., data, training and evaluation code, models) to support and encourage research on Romanian LLMs while concurrently creating a generalizable recipe, adequate for other low or less-resourced languages.
Authors: Yixin Wan, Di Wu, Haoran Wang, Kai-Wei Chang
Abstract: Prompt-based "diversity interventions" are commonly adopted to improve the diversity of Text-to-Image (T2I) models depicting individuals with various racial or gender traits. However, will this strategy result in nonfactual demographic distribution, especially when generating real historical figures. In this work, we propose DemOgraphic FActualIty Representation (DoFaiR), a benchmark to systematically quantify the trade-off between using diversity interventions and preserving demographic factuality in T2I models. DoFaiR consists of 756 meticulously fact-checked test instances to reveal the factuality tax of various diversity prompts through an automated evidence-supported evaluation pipeline. Experiments on DoFaiR unveil that diversity-oriented instructions increase the number of different gender and racial groups in DALLE-3's generations at the cost of historically inaccurate demographic distributions. To resolve this issue, we propose Fact-Augmented Intervention (FAI), which instructs a Large Language Model (LLM) to reflect on verbalized or retrieved factual information about gender and racial compositions of generation subjects in history, and incorporate it into the generation context of T2I models. By orienting model generations using the reflected historical truths, FAI significantly improves the demographic factuality under diversity interventions while preserving diversity.
Authors: Kazuaki Furumai, Roberto Legaspi, Julio Vizcarra, Yudai Yamazaki, Yasutaka Nishimura, Sina J. Semnani, Kazushi Ikeda, Weiyan Shi, Monica S. Lam
Abstract: Persuasion plays a pivotal role in a wide range of applications from health intervention to the promotion of social good. Persuasive chatbots employed responsibly for social good can be an enabler of positive individual and social change. Existing methods rely on fine-tuning persuasive chatbots with task-specific training data which is costly, if not infeasible, to collect. Furthermore, they employ only a handful of pre-defined persuasion strategies. We propose PersuaBot, a zero-shot chatbot based on Large Language Models (LLMs) that is factual and more persuasive by leveraging many more nuanced strategies. PersuaBot uses an LLM to first generate natural responses, from which the strategies used are extracted. To combat hallucination of LLMs, Persuabot replace any unsubstantiated claims in the response with retrieved facts supporting the extracted strategies. We applied our chatbot, PersuaBot, to three significantly different domains needing persuasion skills: donation solicitation, recommendations, and health intervention. Our experiments on simulated and human conversations show that our zero-shot approach is more persuasive than prior work, while achieving factual accuracy surpassing state-of-the-art knowledge-oriented chatbots.
Authors: Aleksander Ficek, Jiaqi Zeng, Oleksii Kuchaiev
Abstract: Parameter-Efficient Fine-Tuning (PEFT) and Retrieval-Augmented Generation (RAG) have become popular methods for adapting large language models while minimizing compute requirements. In this paper, we apply PEFT methods (P-tuning, Adapters, and LoRA) to a modified Retrieval-Enhanced Transformer (RETRO) and a baseline GPT model across several sizes, ranging from 823 million to 48 billion parameters. We show that RETRO models outperform GPT models in zero-shot settings due to their unique pre-training process but GPT models have higher performance potential with PEFT. Additionally, our study indicates that 8B parameter models strike an optimal balance between cost and performance and P-tuning lags behind other PEFT techniques. We further provide a comparative analysis between applying PEFT to an Instruction-tuned RETRO model and base RETRO model. This work presents the first comprehensive comparison of various PEFT methods integrated with RAG, applied to both GPT and RETRO models, highlighting their relative performance.
Authors: Marek Kadl\v{c}\'ik, Michal \v{S}tef\'anik
Abstract: Recent language models achieve impressive results in tasks involving complex multistep reasoning, but scaling these capabilities further traditionally requires expensive collection of more annotated data. In this work, we explore the potential of improving models' reasoning capabilities without new data, merely using automated feedback to the validity of their predictions in arithmetic reasoning (self-training). In systematic experimentation across six different arithmetic reasoning datasets, we find that models can substantially improve in both single-round (offline) and online self-training, reaching a correct result in +13.9% and +25.9% more cases, respectively, underlining the importance of actuality of self-training feedback. We further find that in the single-round, offline self-training, traditional supervised training can deliver gains comparable to preference optimization, but in online self-training, preference optimization methods largely outperform supervised training thanks to their superior stability and robustness on unseen types of problems.
Authors: Guorun Wang, Lucia Specia
Abstract: Text-to-image models are known to propagate social biases. For example, when prompted to generate images of people in certain professions, these models tend to systematically generate specific genders or ethnicities. In this paper, we show that this bias is already present in the text encoder of the model and introduce a Mixture-of-Experts approach by identifying text-encoded bias in the latent space and then creating a Bias-Identification Gate mechanism. More specifically, we propose MoESD (Mixture of Experts Stable Diffusion) with BiAs (Bias Adapters) to mitigate gender bias in text-to-image models. We also demonstrate that introducing an arbitrary special token to the prompt is essential during the mitigation process. With experiments focusing on gender bias, we show that our approach successfully mitigates gender bias while maintaining image quality.
Authors: Rachneet Sachdeva, Yixiao Song, Mohit Iyyer, Iryna Gurevych
Abstract: Long-form question answering (LFQA) aims to provide thorough and in-depth answers to complex questions, enhancing comprehension. However, such detailed responses are prone to hallucinations and factual inconsistencies, challenging their faithful evaluation. This work introduces HaluQuestQA, the first hallucination dataset with localized error annotations for human-written and model-generated LFQA answers. HaluQuestQA comprises 698 QA pairs with 1.8k span-level error annotations for five different error types by expert annotators, along with preference judgments. Using our collected data, we thoroughly analyze the shortcomings of long-form answers and find that they lack comprehensiveness and provide unhelpful references. We train an automatic feedback model on this dataset that predicts error spans with incomplete information and provides associated explanations. Finally, we propose a prompt-based approach, Error-informed refinement, that uses signals from the learned feedback model to refine generated answers, which we show reduces errors and improves answer quality across multiple models. Furthermore, humans find answers generated by our approach comprehensive and highly prefer them (84%) over the baseline answers.
Authors: Hongjin Su, Howard Yen, Mengzhou Xia, Weijia Shi, Niklas Muennighoff, Han-yu Wang, Haisu Liu, Quan Shi, Zachary S. Siegel, Michael Tang, Ruoxi Sun, Jinsung Yoon, Sercan O. Arik, Danqi Chen, Tao Yu
Abstract: Existing retrieval benchmarks primarily consist of information-seeking queries (e.g., aggregated questions from search engines) where keyword or semantic-based retrieval is usually sufficient. However, many complex real-world queries require in-depth reasoning to identify relevant documents that go beyond surface form matching. For example, finding documentation for a coding question requires understanding the logic and syntax of the functions involved. To better benchmark retrieval on such challenging queries, we introduce BRIGHT, the first text retrieval benchmark that requires intensive reasoning to retrieve relevant documents. Our dataset consists of 1,384 real-world queries spanning diverse domains, such as economics, psychology, mathematics, and coding. These queries are drawn from naturally occurring and carefully curated human data. Extensive evaluation reveals that even state-of-the-art retrieval models perform poorly on BRIGHT. The leading model on the MTEB leaderboard (Muennighoff et al., 2023), which achieves a score of 59.0 nDCG@10, produces a score of nDCG@10 of 18.3 on BRIGHT. We show that incorporating explicit reasoning about the query improves retrieval performance by up to 12.2 points. Moreover, incorporating retrieved documents from the top-performing retriever boosts question-answering performance by over 6.6 points. We believe that BRIGHT paves the way for future research on retrieval systems in more realistic and challenging settings.
Authors: Vaskar Nath, Dylan Slack, Jeff Da, Yuntao Ma, Hugh Zhang, Spencer Whitehead, Sean Hendryx
Abstract: Techniques that learn improved representations via offline data or self-supervised objectives have shown impressive results in traditional reinforcement learning (RL). Nevertheless, it is unclear how improved representation learning can benefit reinforcement learning from human feedback (RLHF) on language models (LMs). In this work, we propose training reward models (RMs) in a contrastive, $\textit{goal-conditioned}$ fashion by increasing the representation similarity of future states along sampled preferred trajectories and decreasing the similarity along randomly sampled dispreferred trajectories. This objective significantly improves RM performance by up to 0.09 AUROC across challenging benchmarks, such as MATH and GSM8k. These findings extend to general alignment as well -- on the Helpful-Harmless dataset, we observe $2.3\%$ increase in accuracy. Beyond improving reward model performance, we show this way of training RM representations enables improved $\textit{steerability}$ because it allows us to evaluate the likelihood of an action achieving a particular goal-state (e.g., whether a solution is correct or helpful). Leveraging this insight, we find that we can filter up to $55\%$ of generated tokens during majority voting by discarding trajectories likely to end up in an "incorrect" state, which leads to significant cost savings. We additionally find that these representations can perform fine-grained control by conditioning on desired future goal-states. For example, we show that steering a Llama 3 model towards helpful generations with our approach improves helpfulness by $9.6\%$ over a supervised-fine-tuning trained baseline. Similarly, steering the model towards complex generations improves complexity by $21.6\%$ over the baseline. Overall, we find that training RMs in this contrastive, goal-conditioned fashion significantly improves performance and enables model steerability.
Authors: Sa\"uc Abadal Lloret, Shehzaad Dhuliawala, Keerthiram Murugesan, Mrinmaya Sachan
Abstract: We present ALT (ALignment with Textual feedback), an approach that aligns language models with user preferences expressed in text. We argue that text offers greater expressiveness, enabling users to provide richer feedback than simple comparative preferences and this richer feedback can lead to more efficient and effective alignment. ALT aligns the model by conditioning its generation on the textual feedback. Our method relies solely on language modeling techniques and requires minimal hyper-parameter tuning, though it still presents the main benefits of RL-based alignment algorithms and can effectively learn from textual feedback. We explore the efficacy and efficiency of textual feedback across different tasks such as toxicity reduction, summarization, and dialog response generation. We find that ALT outperforms PPO for the task of toxicity reduction while being able to match its performance on summarization with only 20% of the samples. We also explore how ALT can be used with feedback provided by an existing LLM where we explore an LLM providing constrained and unconstrained textual feedback. We also outline future directions to align models with natural language feedback.
Authors: Howe Tissue, Venus Wang, Lu Wang
Abstract: We find that the cross-entropy loss curves of neural language models empirically adhere to a scaling law with learning rate (LR) annealing over training steps: $$L(s) = L_0 + A\cdot S_1^{-\alpha} - C\cdot S_2,$$ where $L(s)$ is the validation loss at step $s$, $S_1$ is the area under the LR curve, $S_2$ is the LR annealing area, and $L_0$, $A$, $C$, $\alpha$ are constant parameters. This formulation takes into account two factors: (1) power-law scaling over data size, and (2) the additional loss reduction during LR annealing. Therefore, this formulation can describe the full loss curve at each step, rather than the single loss point at the end of training. Applying the scaling law with LR annealing and fitting only one or two training curves, we can accurately predict the loss at any given step across any learning rate scheduler (LRS). This approach significantly reduces computational cost in formulating scaling laws while providing more accuracy and expressiveness for training dynamics. Extensive experiments demonstrate that our findings hold across a range of hyper-parameters and model architectures, and our equation can extend to scaling effect of model sizes. Moreover, our formulation provides accurate theoretical verification and explanation for empirical results observed in numerous previous studies, particularly those focusing on LR schedule and annealing. We believe that this work is promising to enhance the understanding of LLM training dynamics while greatly democratizing scaling laws, and it can guide researchers in refining training strategies (e.g. critical LRS) for further LLMs.
Authors: Anna M\'esz\'aros, Szilvia Ujv\'ary, Wieland Brendel, Patrik Reizinger, Ferenc Husz\'ar
Abstract: LLMs show remarkable emergent abilities, such as inferring concepts from presumably out-of-distribution prompts, known as in-context learning. Though this success is often attributed to the Transformer architecture, our systematic understanding is limited. In complex real-world data sets, even defining what is out-of-distribution is not obvious. To better understand the OOD behaviour of autoregressive LLMs, we focus on formal languages, which are defined by the intersection of rules. We define a new scenario of OOD compositional generalization, termed rule extrapolation. Rule extrapolation describes OOD scenarios, where the prompt violates at least one rule. We evaluate rule extrapolation in formal languages with varying complexity in linear and recurrent architectures, the Transformer, and state space models to understand the architectures' influence on rule extrapolation. We also lay the first stones of a normative theory of rule extrapolation, inspired by the Solomonoff prior in algorithmic information theory.
Authors: Xiang Liu, Peijie Dong, Xuming Hu, Xiaowen Chu
Abstract: Current long-context benchmarks primarily focus on retrieval-based tests, requiring Large Language Models (LLMs) to locate specific information within extensive input contexts, such as the needle-in-a-haystack (NIAH) benchmark. Long-context generation refers to the ability of a language model to generate coherent and contextually accurate text that spans across lengthy passages or documents. While recent studies show strong performance on NIAH and other retrieval-based long-context benchmarks, there is a significant lack of benchmarks for evaluating long-context generation capabilities. To bridge this gap and offer a comprehensive assessment, we introduce a synthetic benchmark, LongGenBench, which allows for flexible configurations of customized generation context lengths. LongGenBench advances beyond traditional benchmarks by redesigning the format of questions and necessitating that LLMs respond with a single, cohesive long-context answer. Upon extensive evaluation using LongGenBench, we observe that: (1) both API accessed and open source models exhibit performance degradation in long-context generation scenarios, ranging from 1.2% to 47.1%; (2) different series of LLMs exhibit varying trends of performance degradation, with the Gemini-1.5-Flash model showing the least degradation among API accessed models, and the Qwen2 series exhibiting the least degradation in LongGenBench among open source models.
Authors: Ziru Chen, Shijie Chen, Yuting Ning, Qianheng Zhang, Boshi Wang, Botao Yu, Yifei Li, Zeyi Liao, Chen Wei, Zitong Lu, Vishal Dey, Mingyi Xue, Frazier N. Baker, Benjamin Burns, Daniel Adu-Ampratwum, Xuhui Huang, Xia Ning, Song Gao, Yu Su, Huan Sun
Abstract: The advancements of language language models (LLMs) have piqued growing interest in developing LLM-based language agents to automate scientific discovery end-to-end, which has sparked both excitement and skepticism about their true capabilities. In this work, we call for rigorous assessment of agents on individual tasks in a scientific workflow before making bold claims on end-to-end automation. To ensure the scientific authenticity and real-world relevance of our benchmark, we extract 102 tasks from 44 peer-reviewed publications in four disciplines and engage nine subject matter experts to validate them. We unify the target output for every task to a self-contained Python program file and employ an array of evaluation metrics to examine the generated programs, execution results, and costs. Each task goes through multiple rounds of manual validation by annotators and subject matter experts to ensure its annotation quality and scientific plausibility. We also propose two effective strategies to mitigate data contamination concerns. Using our benchmark, we evaluate five open-weight and proprietary LLMs, each with three frameworks: direct prompting, OpenHands CodeAct, and self-debug. Given three attempts for each task, the best-performing agent can only solve 32.4% of the tasks independently and 34.3% with expert-provided knowledge. In addition, we evaluate OpenAI o1 with direct prompting and self-debug, which demonstrates the effectiveness of increasing inference-time compute. Still, our results underscore the limitations of current language agents in generating code for data-driven discovery, let alone end-to-end automation for scientific research.
Authors: Jun Shern Chan, Neil Chowdhury, Oliver Jaffe, James Aung, Dane Sherburn, Evan Mays, Giulio Starace, Kevin Liu, Leon Maksin, Tejal Patwardhan, Lilian Weng, Aleksander M\k{a}dry
Abstract: We introduce MLE-bench, a benchmark for measuring how well AI agents perform at machine learning engineering. To this end, we curate 75 ML engineering-related competitions from Kaggle, creating a diverse set of challenging tasks that test real-world ML engineering skills such as training models, preparing datasets, and running experiments. We establish human baselines for each competition using Kaggle's publicly available leaderboards. We use open-source agent scaffolds to evaluate several frontier language models on our benchmark, finding that the best-performing setup--OpenAI's o1-preview with AIDE scaffolding--achieves at least the level of a Kaggle bronze medal in 16.9% of competitions. In addition to our main results, we investigate various forms of resource scaling for AI agents and the impact of contamination from pre-training. We open-source our benchmark code (github.com/openai/mle-bench/) to facilitate future research in understanding the ML engineering capabilities of AI agents.
Authors: Peter Schafhalter, Shun Liao, Yanqi Zhou, Chih-Kuan Yeh, Arun Kandoor, James Laudon
Abstract: Domain-specific adaptation is critical to maximizing the performance of pre-trained language models (PLMs) on one or multiple targeted tasks, especially under resource-constrained use cases, such as edge devices. However, existing methods often struggle to balance domain-specific performance, retention of general knowledge, and efficiency for training and inference. To address these challenges, we propose Modular Domain Experts (MoDE). MoDE is a mixture-of-experts architecture that augments a general PLMs with modular, domain-specialized experts. These experts are trained independently and composed together via a lightweight training process. In contrast to standard low-rank adaptation methods, each MoDE expert consists of several transformer layers which scale better with more training examples and larger parameter counts. Our evaluation demonstrates that MoDE achieves comparable target performances to full parameter fine-tuning while achieving 1.65% better retention performance. Moreover, MoDE's architecture enables flexible sharding configurations and improves training speeds by up to 38% over state-of-the-art distributed training configurations.
Authors: Rana Muhammad Shahroz Khan, Pingzhi Li, Sukwon Yun, Zhenyu Wang, Shahriar Nirjon, Chau-Wai Wong, Tianlong Chen
Abstract: As large language models (LLMs) increasingly shape the AI landscape, fine-tuning pretrained models has become more popular than in the pre-LLM era for achieving optimal performance in domain-specific tasks. However, pretrained LLMs such as ChatGPT are periodically evolved, i.e., model parameters are frequently updated), making it challenging for downstream users with limited resources to keep up with fine-tuning the newest LLMs for their domain application. Even though fine-tuning costs have nowadays been reduced thanks to the innovations of parameter-efficient fine-tuning such as LoRA, not all downstream users have adequate computing for frequent personalization. Moreover, access to fine-tuning datasets, particularly in sensitive domains such as healthcare, could be time-restrictive, making it crucial to retain the knowledge encoded in earlier fine-tuned rounds for future adaptation. In this paper, we present PortLLM, a training-free framework that (i) creates an initial lightweight model update patch to capture domain-specific knowledge, and (ii) allows a subsequent seamless plugging for the continual personalization of evolved LLM at minimal cost. Our extensive experiments cover seven representative datasets, from easier question-answering tasks {BoolQ, SST2} to harder reasoning tasks {WinoGrande, GSM8K}, and models including {Mistral-7B, Llama2, Llama3.1, and Gemma2}, validating the portability of our designed model patches and showcasing the effectiveness of our proposed framework. For instance, PortLLM achieves comparable performance to LoRA fine-tuning with reductions of up to 12.2x in GPU memory usage. Finally, we provide theoretical justifications to understand the portability of our model update patches, which offers new insights into the theoretical dimension of LLMs' personalization.
Authors: Yan Li, Soyeon Caren Han, Yue Dai, Feiqi Cao
Abstract: Transformer-based models have achieved remarkable success in various Natural Language Processing (NLP) tasks, yet their ability to handle long documents is constrained by computational limitations. Traditional approaches, such as truncating inputs, sparse self-attention, and chunking, attempt to mitigate these issues, but they often lead to information loss and hinder the model's ability to capture long-range dependencies. In this paper, we introduce ChuLo, a novel chunk representation method for long document classification that addresses these limitations. Our ChuLo groups input tokens using unsupervised keyphrase extraction, emphasizing semantically important keyphrase based chunk to retain core document content while reducing input length. This approach minimizes information loss and improves the efficiency of Transformer-based models. Preserving all tokens in long document understanding, especially token classification tasks, is especially important to ensure that fine-grained annotations, which depend on the entire sequence context, are not lost. We evaluate our method on multiple long document classification tasks and long document token classification tasks, demonstrating its effectiveness through comprehensive qualitative and quantitative analyses.
Authors: Fai Leui Chan, Duke Nguyen, Aditya Joshi
Abstract: This paper explores the challenges of detecting LGBTQIA+ hate speech of large language models across multiple languages, including English, Italian, Chinese and (code-switched) English-Tamil, examining the impact of machine translation and whether the nuances of hate speech are preserved across translation. We examine the hate speech detection ability of zero-shot and fine-tuned GPT. Our findings indicate that: (1) English has the highest performance and the code-switching scenario of English-Tamil being the lowest, (2) fine-tuning improves performance consistently across languages whilst translation yields mixed results. Through simple experimentation with original text and machine-translated text for hate speech detection along with a qualitative error analysis, this paper sheds light on the socio-cultural nuances and complexities of languages that may not be captured by automatic translation.
Authors: Lance Calvin Lim Gamboa, Mark Lee
Abstract: Work on bias in pretrained language models (PLMs) focuses on bias evaluation and mitigation and fails to tackle the question of bias attribution and explainability. We propose a novel metric, the $\textit{bias attribution score}$, which draws from information theory to measure token-level contributions to biased behavior in PLMs. We then demonstrate the utility of this metric by applying it on multilingual PLMs, including models from Southeast Asia which have not yet been thoroughly examined in bias evaluation literature. Our results confirm the presence of sexist and homophobic bias in Southeast Asian PLMs. Interpretability and semantic analyses also reveal that PLM bias is strongly induced by words relating to crime, intimate relationships, and helping among other discursive categories, suggesting that these are topics where PLMs strongly reproduce bias from pretraining data and where PLMs should be used with more caution.
Authors: Nelson Vadori, Antoine Gorceix, Bastien Le Chenadec, Ahmad Rammal, Manuela Veloso
Abstract: In this paper, we study the ability of large language models to learn specific mathematical rules such as distributivity or simplifying equations. We present an empirical analysis of their ability to generalize these rules, as well as to reuse them in the context of word problems. For this purpose, we provide a rigorous methodology to build synthetic data incorporating such rules, and perform fine-tuning of large language models on such data. Our experiments show that our model can learn and generalize these rules to some extent, as well as suitably reuse them in the context of word problems.
Authors: Yang Zhong, Jiangang Hao, Michael Fauss, Chen Li, Yuan Wang
Abstract: The recent revolutionary advance in generative AI enables the generation of realistic and coherent texts by large language models (LLMs). Despite many existing evaluation metrics on the quality of the generated texts, there is still a lack of rigorous assessment of how well LLMs perform in complex and demanding writing assessments. This study examines essays generated by ten leading LLMs for the analytical writing assessment of the Graduate Record Exam (GRE). We assessed these essays using both human raters and the e-rater automated scoring engine as used in the GRE scoring pipeline. Notably, the top-performing Gemini and GPT-4o received an average score of 4.78 and 4.67, respectively, falling between "generally thoughtful, well-developed analysis of the issue and conveys meaning clearly" and "presents a competent analysis of the issue and conveys meaning with acceptable clarity" according to the GRE scoring guideline. We also evaluated the detection accuracy of these essays, with detectors trained on essays generated by the same and different LLMs.
Authors: Bowen Wei, Ziwei Zhu
Abstract: Deep neural networks have achieved remarkable performance in various text-based tasks but often lack interpretability, making them less suitable for applications where transparency is critical. To address this, we propose ProtoLens, a novel prototype-based model that provides fine-grained, sub-sentence level interpretability for text classification. ProtoLens uses a Prototype-aware Span Extraction module to identify relevant text spans associated with learned prototypes and a Prototype Alignment mechanism to ensure prototypes are semantically meaningful throughout training. By aligning the prototype embeddings with human-understandable examples, ProtoLens provides interpretable predictions while maintaining competitive accuracy. Extensive experiments demonstrate that ProtoLens outperforms both prototype-based and non-interpretable baselines on multiple text classification benchmarks. Code and data are available at \url{https://anonymous.4open.science/r/ProtoLens-CE0B/}.
Authors: Zongqi Wang, Baoyuan Wu, Jingyuan Deng, Yujiu Yang
Abstract: Embeddings as a Service (EaaS) is emerging as a crucial role in AI applications. Unfortunately, EaaS is vulnerable to model extraction attacks, highlighting the urgent need for copyright protection. Although some preliminary works propose applying embedding watermarks to protect EaaS, recent research reveals that these watermarks can be easily removed. Hence, it is crucial to inject robust watermarks resistant to watermark removal attacks. Existing watermarking methods typically inject a target embedding into embeddings through linear interpolation when the text contains triggers. However, this mechanism results in each watermarked embedding having the same component, which makes the watermark easy to identify and eliminate. Motivated by this, in this paper, we propose a novel embedding-specific watermarking (ESpeW) mechanism to offer robust copyright protection for EaaS. Our approach involves injecting unique, yet readily identifiable watermarks into each embedding. Watermarks inserted by ESpeW are designed to maintain a significant distance from one another and to avoid sharing common components, thus making it significantly more challenging to remove the watermarks. Extensive experiments on four popular datasets demonstrate that ESpeW can even watermark successfully against a highly aggressive removal strategy without sacrificing the quality of embeddings. Code is available at https://github.com/liudan193/ESpeW.
Authors: Qiqi Chen, Xinpeng Wang, Philipp Mondorf, Michael A. Hedderich, Barbara Plank
Abstract: Tree of Thoughts (ToT) is a reasoning strategy for Large Language Models (LLMs) that employs a generator to suggest reasoning steps and a discriminator to decide which steps to implement. ToT demonstrates strong performance on reasoning tasks, often surpassing simple methods such as Input-Output (IO) prompting and Chain-of-Thought (CoT) reasoning. However, ToT does not consistently outperform such simpler methods across all models, leaving large knowledge gaps on the conditions under which ToT is most beneficial. In this paper, we analyze the roles of the generator and discriminator separately to better understand the conditions when ToT is beneficial. We find that the generator plays a more critical role than the discriminator in driving the success of ToT. Scaling the generator leads to notable improvements in ToT performance, even when using a smaller model as the discriminator, whereas scaling the discriminator with a fixed generator yields only marginal gains. Our results show that models across different scales exhibit comparable discrimination capabilities, yet differ significantly in their generative performance for ToT.
Authors: Rui Xue, Xipeng Shen, Ruozhou Yu, Xiaorui Liu
Abstract: Learning from Text-Attributed Graphs (TAGs) has attracted significant attention due to its wide range of real-world applications. The rapid evolution of language models (LMs) has revolutionized the way we process textual data, which indicates a strong potential to replace shallow text embedding generally used in Graph Neural Networks (GNNs). However, we find that existing LM approaches that exploit text information in graphs suffer from inferior computation and data efficiency. In this study, we introduce LEADING, a novel and efficient approach for end-to-end fine-tuning of language models on TAGs. To enhance data efficiency, LEADING efficiently transfers rich knowledge from LMs to downstream graph learning tasks with limited labeled data by employing end-to-end training of LMs and GNNs in a semi-supervised learning setting. To address associated computation efficiency issues, it introduces two techniques: neighbor decoupling targeting LMs and implicit graph modeling targeting GNNs, respectively. Our proposed approach demonstrates superior performance, achieving state-of-the-art (SOTA) results on the ogbn-arxiv leaderboard, while maintaining computation cost and memory overhead comparable to graph-less fine-tuning of LMs. Through comprehensive experiments, we showcase its superior computation and data efficiency, presenting a promising solution for various LMs and graph learning tasks on TAGs.
Authors: Weizheng Lu, Jing Zhang, Ju Fan, Zihao Fu, Yueguo Chen, Xiaoyong Du
Abstract: Tables, typically two-dimensional and structured to store large amounts of data, are essential in daily activities like database queries, spreadsheet manipulations, web table question answering, and image table information extraction. Automating these table-centric tasks with Large Language Models (LLMs) or Visual Language Models (VLMs) offers significant public benefits, garnering interest from academia and industry. This survey provides a comprehensive overview of table-related tasks, examining both user scenarios and technical aspects. It covers traditional tasks like table question answering as well as emerging fields such as spreadsheet manipulation and table data analysis. We summarize the training techniques for LLMs and VLMs tailored for table processing. Additionally, we discuss prompt engineering, particularly the use of LLM-powered agents, for various table-related tasks. Finally, we highlight several challenges, including diverse user input when serving and slow thinking using chain-of-thought.
Authors: Zhen Ye, Zeqian Ju, Haohe Liu, Xu Tan, Jianyi Chen, Yiwen Lu, Peiwen Sun, Jiahao Pan, Weizhen Bian, Shulin He, Wei Xue, Qifeng Liu, Yike Guo
Abstract: Recent progress in large-scale zero-shot speech synthesis has been significantly advanced by language models and diffusion models. However, the generation process of both methods is slow and computationally intensive. Efficient speech synthesis using a lower computing budget to achieve quality on par with previous work remains a significant challenge. In this paper, we present FlashSpeech, a large-scale zero-shot speech synthesis system with approximately 5\% of the inference time compared with previous work. FlashSpeech is built on the latent consistency model and applies a novel adversarial consistency training approach that can train from scratch without the need for a pre-trained diffusion model as the teacher. Furthermore, a new prosody generator module enhances the diversity of prosody, making the rhythm of the speech sound more natural. The generation processes of FlashSpeech can be achieved efficiently with one or two sampling steps while maintaining high audio quality and high similarity to the audio prompt for zero-shot speech generation. Our experimental results demonstrate the superior performance of FlashSpeech. Notably, FlashSpeech can be about 20 times faster than other zero-shot speech synthesis systems while maintaining comparable performance in terms of voice quality and similarity. Furthermore, FlashSpeech demonstrates its versatility by efficiently performing tasks like voice conversion, speech editing, and diverse speech sampling. Audio samples can be found in https://flashspeech.github.io/.
Authors: Xinzhe Li
Abstract: Tool use, planning, and feedback learning are currently three prominent paradigms for developing Large Language Model (LLM)-based agents across various tasks. Although numerous frameworks have been devised for each paradigm, their intricate workflows and inconsistent taxonomy create challenges in understanding and reviewing the frameworks across different paradigms. This survey introduces a unified taxonomy to systematically review and discuss these frameworks. Specifically, 1) the taxonomy defines environments/tasks, common LLM-profiled roles or LMPRs (policy models, evaluators, and dynamic models), and universally applicable workflows found in prior work, and 2) it enables a comparison of key perspectives on the implementations of LMPRs and workflow designs across different agent paradigms and frameworks. 3) Finally, we identify three limitations in existing workflow designs and systematically discuss the future work. Resources have been made publicly available at in our GitHub repository https://github.com/xinzhel/LLM-Agent-Survey.
Authors: Moayed Haji-Ali, Willi Menapace, Aliaksandr Siarohin, Guha Balakrishnan, Sergey Tulyakov, Vicente Ordonez
Abstract: Generating ambient sounds is a challenging task due to data scarcity and often insufficient caption quality, making it difficult to employ large-scale generative models for the task. In this work, we tackle this problem by introducing two new models. First, we propose AutoCap, a high-quality and efficient automatic audio captioning model. By using a compact audio representation and leveraging audio metadata, AutoCap substantially enhances caption quality, reaching a CIDEr score of 83.2, marking a 3.2% improvement from the best available captioning model at four times faster inference speed. Second, we propose GenAu, a scalable transformer-based audio generation architecture that we scale up to 1.25B parameters. Using AutoCap to generate caption clips from existing audio datasets, we demonstrate the benefits of data scaling with synthetic captions as well as model size scaling. When compared to state-of-the-art audio generators trained at similar size and data scale, GenAu obtains significant improvements of 4.7% in FAD score, 22.7% in IS, and 13.5% in CLAP score, indicating significantly improved quality of generated audio compared to previous works. Moreover, we propose an efficient and scalable pipeline for collecting audio datasets, enabling us to compile 57M ambient audio clips, forming AutoReCap-XL, the largest available audio-text dataset, at 90 times the scale of existing ones. Our code, model checkpoints, and dataset are publicly available.
Authors: Yuren Mao, Yuhang Ge, Yijiang Fan, Wenyi Xu, Yu Mi, Zhonghao Hu, Yunjun Gao
Abstract: Low-Rank Adaptation~(LoRA), which updates the dense neural network layers with pluggable low-rank matrices, is one of the best performed parameter efficient fine-tuning paradigms. Furthermore, it has significant advantages in cross-task generalization and privacy-preserving. Hence, LoRA has gained much attention recently, and the number of related literature demonstrates exponential growth. It is necessary to conduct a comprehensive overview of the current progress on LoRA. This survey categorizes and reviews the progress from the perspectives of (1) downstream adaptation improving variants that improve LoRA's performance on downstream tasks; (2) cross-task generalization methods that mix multiple LoRA plugins to achieve cross-task generalization; (3) efficiency-improving methods that boost the computation-efficiency of LoRA; (4) data privacy-preserving methods that use LoRA in federated learning; (5) application. Besides, this survey also discusses the future directions in this field. At last, we provide a Github page~\footnote{\href{https://github.com/ZJU-LLMs/Awesome-LoRAs.git}{https://github.com/ZJU-LLMs/Awesome-LoRAs.git}} for readers to check the updates and initiate discussions on this survey paper.
URLs: https://github.com/ZJU-LLMs/Awesome-LoRAs.git, https://github.com/ZJU-LLMs/Awesome-LoRAs.git
Authors: Nikolaus Howe, Ian McKenzie, Oskar Hollinsworth, Micha{\l} Zajac, Tom Tseng, Aaron Tucker, Pierre-Luc Bacon, Adam Gleave
Abstract: Language models exhibit scaling laws, whereby increasing model and dataset size yields predictable decreases in negative log likelihood, unlocking a dazzling array of capabilities. This phenomenon spurs many companies to train ever larger models in pursuit of ever improved performance. Yet, these models are vulnerable to adversarial inputs such as ``jailbreaks'' and prompt injections that induce models to perform undesired behaviors, posing a growing risk as models become more capable. Prior work indicates that computer vision models become more robust with model and data scaling, raising the question: does language model robustness also improve with scale? We study this question empirically in the classification setting, finding that without explicit defense training, larger models tend to be modestly more robust on most tasks, though the effect is not reliable. Even with the advantage conferred by scale, undefended models remain easy to attack in absolute terms, and we thus turn our attention to explicitly training models for adversarial robustness, which we show to be a much more compute-efficient defense than scaling model size alone. In this setting, we also observe that adversarially trained larger models generalize faster and better to modified attacks not seen during training when compared with smaller models. Finally, we analyze the offense/defense balance of increasing compute, finding parity in some settings and an advantage for offense in others, suggesting that adversarial training alone is not sufficient to solve robustness, even at greater model scales.
Authors: Edward Y. Chang
Abstract: Biases and errors in human-labeled data present significant challenges for machine learning, especially in supervised learning reliant on potentially flawed ground truth data. These flaws, including diagnostic errors and societal biases, risk being propagated and amplified through models trained using maximum likelihood estimation. We present the Reflective LLM Dialogue Framework RLDF, which leverages structured adversarial dialogues between multiple instances of a single LLM or different LLMs to uncover diverse perspectives and correct inconsistencies. By conditioning LLMs to adopt opposing stances, RLDF enables systematic bias detection through conditional statistics, information theory, and divergence metrics. Experiments show RLDF successfully identifies potential biases in public content while exposing limitations in human-labeled data. Our framework supports measurable progress tracking and explainable remediation actions, offering a scalable approach for improving content neutrality through transparent, multi-perspective analysis.
Authors: Desiree Heim, Christian Jilek, Adrian Ulges, Andreas Dengel
Abstract: Current publicly available knowledge work data collections lack diversity, extensive annotations, and contextual information about the users and their documents. These issues hinder objective and comparable data-driven evaluations and optimizations of knowledge work assistance systems. Due to the considerable resources needed to collect such data in real-life settings and the necessity of data censorship, collecting such a dataset appears nearly impossible. For this reason, we propose a configurable, multi-agent knowledge work dataset generator. This system simulates collaborative knowledge work among agents producing Large Language Model-generated documents and accompanying data traces. Additionally, the generator captures all background information, given in its configuration or created during the simulation process, in a knowledge graph. Finally, the resulting dataset can be utilized and shared without privacy or confidentiality concerns. This paper introduces our approach's design and vision and focuses on generating authentic knowledge work documents using Large Language Models. Our study involving human raters who assessed 53% of the generated and 74% of the real documents as realistic demonstrates the potential of our approach. Furthermore, we analyze the authenticity criteria mentioned in the participants' comments and elaborate on potential improvements for identified common issues.
Authors: Kazuki Matsuda, Yuiga Wada, Komei Sugiura
Abstract: In this work, we address the challenge of developing automatic evaluation metrics for image captioning, with a particular focus on robustness against hallucinations. Existing metrics are often inadequate for handling hallucinations, primarily due to their limited ability to compare candidate captions with multifaceted reference captions. To address this shortcoming, we propose DENEB, a novel supervised automatic evaluation metric specifically robust against hallucinations. DENEB incorporates the Sim-Vec Transformer, a mechanism that processes multiple references simultaneously, thereby efficiently capturing the similarity between an image, a candidate caption, and reference captions. To train DENEB, we construct the diverse and balanced Nebula dataset comprising 32,978 images, paired with human judgments provided by 805 annotators. We demonstrated that DENEB achieves state-of-the-art performance among existing LLM-free metrics on the FOIL, Composite, Flickr8K-Expert, Flickr8K-CF, Nebula, and PASCAL-50S datasets, validating its effectiveness and robustness against hallucinations.
Authors: Yuandong Tian
Abstract: We prove rich algebraic structures of the solution space for 2-layer neural networks with quadratic activation and $L_2$ loss, trained on reasoning tasks in Abelian group (e.g., modular addition). Such a rich structure enables analytical construction of global optimal solutions from partial solutions that only satisfy part of the loss, despite its high nonlinearity. We coin the framework as CoGO (Composing Global Optimizers). Specifically, we show that the weight space over different numbers of hidden nodes of the 2-layer network is equipped with a semi-ring algebraic structure, and the loss function to be optimized consists of monomial potentials, which are ring homomorphism, allowing partial solutions to be composed into global ones by ring addition and multiplication. Our experiments show that around $95\%$ of the solutions obtained by gradient descent match exactly our theoretical constructions. Although the global optimizers constructed only required a small number of hidden nodes, our analysis on gradient dynamics shows that over-parameterization asymptotically decouples training dynamics and is beneficial. We further show that training dynamics favors simpler solutions under weight decay, and thus high-order global optimizers such as perfect memorization are unfavorable.
Authors: Jiawei Liu, Thanh Nguyen, Mingyue Shang, Hantian Ding, Xiaopeng Li, Yu Yu, Varun Kumar, Zijian Wang
Abstract: Large Language Models (LLMs) have recently demonstrated remarkable coding capabilities. However, assessing code generation based on well-formed properties and aligning it with developer preferences remains challenging. In this paper, we explore two key questions under the new challenge of code preference learning: (i) How do we train models to predict meaningful preferences for code? and (ii) How do human and LLM preferences align with verifiable code properties and developer code tastes? To this end, we propose CodeFavor, a framework for training pairwise code preference models from synthetic evolution data, including code commits and code critiques. To evaluate code preferences, we introduce CodePrefBench, a benchmark comprising 1364 rigorously curated code preference tasks to cover three verifiable properties-correctness, efficiency, and security-along with human preference. Our evaluation shows that CodeFavor holistically improves the accuracy of model-based code preferences by up to 28.8%. Meanwhile, CodeFavor models can match the performance of models with 6-9x more parameters while being 34x more cost-effective. We also rigorously validate the design choices in CodeFavor via a comprehensive set of controlled experiments. Furthermore, we discover the prohibitive costs and limitations of human-based code preference: despite spending 23.4 person-minutes on each task, 15.1-40.3% of tasks remain unsolved. Compared to model-based preference, human preference tends to be more accurate under the objective of code correctness, while being sub-optimal for non-functional objectives.
Authors: Joost de Winter, Dimitra Dodou, Yke Bauke Eisma
Abstract: The processes underlying human cognition are often divided into System 1, which involves fast, intuitive thinking, and System 2, which involves slow, deliberate reasoning. Previously, large language models were criticized for lacking the deeper, more analytical capabilities of System 2. In September 2024, OpenAI introduced the o1 model series, designed to handle System 2-like reasoning. While OpenAI's benchmarks are promising, independent validation is still needed. In this study, we tested the o1-preview model twice on the Dutch 'Mathematics B' final exam. It scored a near-perfect 76 and 74 out of 76 points. For context, only 24 out of 16,414 students in the Netherlands achieved a perfect score. By comparison, the GPT-4o model scored 66 and 62 out of 76, well above the Dutch students' average of 40.63 points. Neither model had access to the exam figures. Since there was a risk of model contami-nation (i.e., the knowledge cutoff for o1-preview and GPT-4o was after the exam was published online), we repeated the procedure with a new Mathematics B exam that was published after the cutoff date. The results again indicated that o1-preview performed strongly (97.8th percentile), which suggests that contamination was not a factor. We also show that there is some variability in the output of o1-preview, which means that sometimes there is 'luck' (the answer is correct) or 'bad luck' (the output has diverged into something that is incorrect). We demonstrate that the self-consistency approach, where repeated prompts are given and the most common answer is selected, is a useful strategy for identifying the correct answer. It is concluded that while OpenAI's new model series holds great potential, certain risks must be considered.
Authors: Long Li, Weiwen Xu, Jiayan Guo, Ruochen Zhao, Xinxuan Li, Yuqian Yuan, Boqiang Zhang, Yuming Jiang, Yifei Xin, Ronghao Dang, Deli Zhao, Yu Rong, Tian Feng, Lidong Bing
Abstract: Effective research ideation is a critical step for scientific research. However, the exponential increase in scientific literature makes it challenging for researchers to stay current with recent advances and identify meaningful research directions. Recent developments in large language models~(LLMs) suggest a promising avenue for automating the generation of novel research ideas. However, existing methods for idea generation either trivially prompt LLMs or directly expose LLMs to extensive literature without indicating useful information. Inspired by the research process of human researchers, we propose a Chain-of-Ideas~(CoI) agent, an LLM-based agent that organizes relevant literature in a chain structure to effectively mirror the progressive development in a research domain. This organization facilitates LLMs to capture the current advancements in research, thereby enhancing their ideation capabilities. Furthermore, we propose Idea Arena, an evaluation protocol that can comprehensively evaluate idea generation methods from different perspectives, aligning closely with the preferences of human researchers. Experimental results indicate that the CoI agent consistently outperforms other methods and shows comparable quality as humans in research idea generation. Moreover, our CoI agent is budget-friendly, with a minimum cost of \$0.50 to generate a candidate idea and its corresponding experimental design.
Authors: Nghia Hieu Nguyen, Tho Thanh Quan, Ngan Luu-Thuy Nguyen
Abstract: Text-based VQA is a challenging task that requires machines to use scene texts in given images to yield the most appropriate answer for the given question. The main challenge of text-based VQA is exploiting the meaning and information from scene texts. Recent studies tackled this challenge by considering the spatial information of scene texts in images via embedding 2D coordinates of their bounding boxes. In this study, we follow the definition of meaning from linguistics to introduce a novel method that effectively exploits the information from scene texts written in Vietnamese. Experimental results show that our proposed method obtains state-of-the-art results on two large-scale Vietnamese Text-based VQA datasets. The implementation can be found at this link.
Authors: Moraes Pablo, Rodr\'iguez M\'onica, Peters Christopher, Sodre Hiago, Mazondo Ahilen, Sandin Vincent, Barcelona Sebastian, Moraes William, Fern\'andez Santiago, Assun\c{c}\~ao Nathalie, de Vargas Bruna, D\"ornbach Tobias, Kelbouscas Andr\'e, Grando Ricardo
Abstract: This study investigates the interaction between humanoid robots NAO and Pepper, emphasizing their potential applications in educational settings. NAO, widely used in education, and Pepper, designed for social interactions, of er new opportunities for autonomous communication and collaboration. Through a series of programmed interactions, the robots demonstrated their ability to communicate and coordinate actions autonomously, highlighting their potential as tools for enhancing learning environments. The research also explores the integration of emerging technologies, such as artificial intelligence, into these systems, allowing robots to learn from each other and adapt their behavior. The findings suggest that NAO and Pepper can significantly contribute to both technical learning and the development of social and emotional skills in students, of ering innovative pedagogical approaches through the use of humanoid robotics.