Authors: Pepijn de Reus, Ana Oprescu, Jelle Zuidema
Abstract: This study examines quantisation and pruning strategies to reduce energy consumption in code Large Language Models (LLMs) inference. Using StarCoder2, we observe increased energy demands with quantization due to lower throughput and some accuracy losses. Conversely, pruning reduces energy usage but impairs performance. The results highlight challenges and trade-offs in LLM model compression. We suggest future work on hardware-optimized quantization to enhance efficiency with minimal loss in accuracy.
Authors: Grace Sng, Yanming Zhang, Klaus Mueller
Abstract: The increasing use of large language models (LLMs) in causal discovery as a substitute for human domain experts highlights the need for optimal model selection. This paper presents the first hallucination survey of popular LLMs for causal discovery. We show that hallucinations exist when using LLMs in causal discovery so the choice of LLM is important. We propose using Retrieval Augmented Generation (RAG) to reduce hallucinations when quality data is available. Additionally, we introduce a novel method employing multiple LLMs with an arbiter in a debate to audit edges in causal graphs, achieving a comparable reduction in hallucinations to RAG.
Authors: Yu Peng, Zewen Long, Fangming Dong, Congyi Li, Shu Wu, Kai Chen
Abstract: The advent of large language models (LLMs) has spurred the development of numerous jailbreak techniques aimed at circumventing their security defenses against malicious attacks. An effective jailbreak approach is to identify a domain where safety generalization fails, a phenomenon known as mismatched generalization. In this paper, we introduce two novel jailbreak methods based on mismatched generalization: natural language games and custom language games, both of which effectively bypass the safety mechanisms of LLMs, with various kinds and different variants, making them hard to defend and leading to high attack rates. Natural language games involve the use of synthetic linguistic constructs and the actions intertwined with these constructs, such as the Ubbi Dubbi language. Building on this phenomenon, we propose the custom language games method: by engaging with LLMs using a variety of custom rules, we successfully execute jailbreak attacks across multiple LLM platforms. Extensive experiments demonstrate the effectiveness of our methods, achieving success rates of 93% on GPT-4o, 89% on GPT-4o-mini and 83% on Claude-3.5-Sonnet. Furthermore, to investigate the generalizability of safety alignments, we fine-tuned Llama-3.1-70B with the custom language games to achieve safety alignment within our datasets and found that when interacting through other language games, the fine-tuned models still failed to identify harmful content. This finding indicates that the safety alignment knowledge embedded in LLMs fails to generalize across different linguistic formats, thus opening new avenues for future research in this area.
Authors: Weiqing He, Bojian Hou, Tianqi Shang, Davoud Ataee Tarzanagh, Qi Long, Li Shen
Abstract: The widespread adoption of large language models (LLMs) has created an urgent need for robust tools to detect LLM-generated text, especially in light of \textit{paraphrasing} techniques that often evade existing detection methods. To address this challenge, we present a novel semantic-enhanced framework for detecting LLM-generated text (SEFD) that leverages a retrieval-based mechanism to fully utilize text semantics. Our framework improves upon existing detection methods by systematically integrating retrieval-based techniques with traditional detectors, employing a carefully curated retrieval mechanism that strikes a balance between comprehensive coverage and computational efficiency. We showcase the effectiveness of our approach in sequential text scenarios common in real-world applications, such as online forums and Q\&A platforms. Through comprehensive experiments across various LLM-generated texts and detection methods, we demonstrate that our framework substantially enhances detection accuracy in paraphrasing scenarios while maintaining robustness for standard LLM-generated content.
Authors: Max Lovitt, Haotian Ma, Song Wang, Yifan Peng
Abstract: With social media communities increasingly becoming places where suicidal individuals post and congregate, natural language processing presents an exciting avenue for the development of automated suicide risk assessment systems. However, past efforts suffer from a lack of labeled data and class imbalances within the available labeled data. To accommodate this task's imperfect data landscape, we propose a semi-supervised framework that leverages labeled (n=500) and unlabeled (n=1,500) data and expands upon the self-training algorithm with a novel pseudo-label acquisition process designed to handle imbalanced datasets. To further ensure pseudo-label quality, we manually verify a subset of the pseudo-labeled data that was not predicted unanimously across multiple trials of pseudo-label generation. We test various models to serve as the backbone for this framework, ultimately deciding that RoBERTa performs the best. Ultimately, by leveraging partially validated pseudo-labeled data in addition to ground-truth labeled data, we substantially improve our model's ability to assess suicide risk from social media posts.
Authors: Nay Myat Min, Long H. Pham, Yige Li, Jun Sun
Abstract: Recent studies reveal that Large Language Models (LLMs) are susceptible to backdoor attacks, where adversaries embed hidden triggers that manipulate model responses. Existing backdoor defense methods are primarily designed for vision or classification tasks, and are thus ineffective for text generation tasks, leaving LLMs vulnerable. We introduce Internal Consistency Regularization (CROW), a novel defense using consistency regularization finetuning to address layer-wise inconsistencies caused by backdoor triggers. CROW leverages the intuition that clean models exhibit smooth, consistent transitions in hidden representations across layers, whereas backdoored models show noticeable fluctuation when triggered. By enforcing internal consistency through adversarial perturbations and regularization, CROW neutralizes backdoor effects without requiring clean reference models or prior trigger knowledge, relying only on a small set of clean data. This makes it practical for deployment across various LLM architectures. Experimental results demonstrate that CROW consistently achieves a significant reductions in attack success rates across diverse backdoor strategies and tasks, including negative sentiment, targeted refusal, and code injection, on models such as Llama-2 (7B, 13B), CodeLlama (7B, 13B) and Mistral-7B, while preserving the model's generative capabilities.
Authors: Sonny George, Chris Sypherd, Dylan Cashman
Abstract: Large language model (LLM) agents show promise in an increasing number of domains. In many proposed applications, it is expected that the agent reasons over accumulated experience presented in an input prompt. We propose the OEDD (Operationalize Experience Despite Distraction) corpus, a human-annotator-validated body of scenarios with pre-scripted agent histories where the agent must make a decision based on disparate experiential information in the presence of a distractor. We evaluate three state-of-the-art LLMs (GPT-3.5 Turbo, GPT-4o, and Gemini 1.5 Pro) using a minimal chain-of-thought prompting strategy and observe that when (1) the input context contains over 1,615 tokens of historical interactions, (2) a crucially decision-informing premise is the rightful conclusion over two disparate environment premises, and (3) a trivial, but distracting red herring fact follows, all LLMs perform worse than random choice at selecting the better of two actions. Our code and test corpus are publicly available at: https://github.com/sonnygeorge/OEDD .
Authors: Nigar Alishzade, Jamaladdin Hasanov
Abstract: Sign language processing technology development relies on extensive and reliable datasets, instructions, and ethical guidelines. We present a comprehensive Azerbaijani Sign Language Dataset (AzSLD) collected from diverse sign language users and linguistic parameters to facilitate advancements in sign recognition and translation systems and support the local sign language community. The dataset was created within the framework of a vision-based AzSL translation project. This study introduces the dataset as a summary of the fingerspelling alphabet and sentence- and word-level sign language datasets. The dataset was collected from signers of different ages, genders, and signing styles, with videos recorded from two camera angles to capture each sign in full detail. This approach ensures robust training and evaluation of gesture recognition models. AzSLD contains 30,000 videos, each carefully annotated with accurate sign labels and corresponding linguistic translations. The dataset is accompanied by technical documentation and source code to facilitate its use in training and testing. This dataset offers a valuable resource of labeled data for researchers and developers working on sign language recognition, translation, or synthesis. Ethical guidelines were strictly followed throughout the project, with all participants providing informed consent for collecting, publishing, and using the data.
Authors: Eta Yang
Abstract: Sign language translation, especially in gloss-free paradigm, is confronting a dilemma of impracticality and unsustainability due to growing resource-intensive methodologies. Contemporary state-of-the-arts (SOTAs) have significantly hinged on pretrained sophiscated backbones such as Large Language Models (LLMs), embedding sources, or extensive datasets, inducing considerable parametric and computational inefficiency for sustainable use in real-world scenario. Despite their success, following this research direction undermines the overarching mission of this domain to create substantial value to bridge hard-hearing and common populations. Committing to the prevailing trend of LLM and Natural Language Processing (NLP) studies, we pursue a profound essential change in architecture to achieve ground-up improvements without external aid from pretrained models, prior knowledge transfer, or any NLP strategies considered not-from-scratch. Introducing Signformer, a from-scratch Feather-Giant transforming the area towards Edge AI that redefines extremities of performance and efficiency with LLM-competence and edgy-deployable compactness. In this paper, we present nature analysis of sign languages to inform our algorithmic design and deliver a scalable transformer pipeline with convolution and attention novelty. We achieve new 2nd place on leaderboard with a parametric reduction of 467-1807x against the finests as of 2024 and outcompete almost every other methods in a lighter configuration of 0.57 million parameters.
Authors: Gabriel Chua, Shing Yee Chan, Shaun Khoo
Abstract: Large Language Models are prone to off-topic misuse, where users may prompt these models to perform tasks beyond their intended scope. Current guardrails, which often rely on curated examples or custom classifiers, suffer from high false-positive rates, limited adaptability, and the impracticality of requiring real-world data that is not available in pre-production. In this paper, we introduce a flexible, data-free guardrail development methodology that addresses these challenges. By thoroughly defining the problem space qualitatively and passing this to an LLM to generate diverse prompts, we construct a synthetic dataset to benchmark and train off-topic guardrails that outperform heuristic approaches. Additionally, by framing the task as classifying whether the user prompt is relevant with respect to the system prompt, our guardrails effectively generalize to other misuse categories, including jailbreak and harmful prompts. Lastly, we further contribute to the field by open-sourcing both the synthetic dataset and the off-topic guardrail models, providing valuable resources for developing guardrails in pre-production environments and supporting future research and development in LLM safety.
Authors: Skyler Seto, Maartje ter Hoeve, He Bai, Natalie Schluter, David Grangier
Abstract: Large language models are trained on massive scrapes of the web, as required by current scaling laws. Most progress is made for English, given its abundance of high-quality pretraining data. For most other languages, however, such high quality pretraining data is unavailable. In this work, we study how to boost pretrained model performance in a data constrained target language by enlisting data from an auxiliary language for which high quality data is available. We study this by quantifying the performance gap between training with data in a data-rich auxiliary language compared with training in the target language, exploring the benefits of translation systems, studying the limitations of model scaling for data constrained languages, and proposing new methods for upsampling data from the auxiliary language. Our results show that stronger auxiliary datasets result in performance gains without modification to the model or training objective for close languages, and, in particular, that performance gains due to the development of more information-rich English pretraining datasets can extend to targeted language settings with limited data.
Authors: Ning Ding, Yehui Tang, Haochen Qin, Zhenli Zhou, Chao Xu, Lin Li, Kai Han, Heng Liao, Yunhe Wang
Abstract: In order to reduce the computational complexity of large language models, great efforts have been made to to improve the efficiency of transformer models such as linear attention and flash-attention. However, the model size and corresponding computational complexity are constantly scaled up in pursuit of higher performance. In this work, we present MemoryFormer, a novel transformer architecture which significantly reduces the computational complexity (FLOPs) from a new perspective. We eliminate nearly all the computations of the transformer model except for the necessary computation required by the multi-head attention operation. This is made possible by utilizing an alternative method for feature transformation to replace the linear projection of fully-connected layers. Specifically, we first construct a group of in-memory lookup tables that store a large amount of discrete vectors to replace the weight matrix used in linear projection. We then use a hash algorithm to retrieve a correlated subset of vectors dynamically based on the input embedding. The retrieved vectors combined together will form the output embedding, which provides an estimation of the result of matrix multiplication operation in a fully-connected layer. Compared to conducting matrix multiplication, retrieving data blocks from memory is a much cheaper operation which requires little computations. We train MemoryFormer from scratch and conduct extensive experiments on various benchmarks to demonstrate the effectiveness of the proposed model.
Authors: Yijiong Yu
Abstract: Recent advancements in the field of large language models, particularly through the Chain of Thought (CoT) approach, have demonstrated significant improvements in solving complex problems. However, existing models either tend to sacrifice detailed reasoning for brevity due to user preferences, or require extensive and expensive training data to learn complicated reasoning ability, limiting their potential in solving complex tasks. To bridge this gap, following the concept of scaling test-time, we propose a simple method by encouraging models to adopt a more patient reasoning style without the need of introducing new knowledge or skills. To employ a preference optimization approach, we generate detailed reasoning processes as positive examples and simple answers as negative examples, thereby training the model to favor thoroughness in its responses. Our results demonstrate a performance increase of up to 6.7% on GSM8k with training just on a lightweight dataset.
Authors: Yunkee Chae, Eunsik Shin, Hwang Suntae, Seungryeol Paik, Kyogu Lee
Abstract: Lyrics generation presents unique challenges, particularly in achieving precise syllable control while adhering to song form structures such as verses and choruses. Conventional line-by-line approaches often lead to unnatural phrasing, underscoring the need for more granular syllable management. We propose a framework for lyrics generation that enables multi-level syllable control at the word, phrase, line, and paragraph levels, aware of song form. Our approach generates complete lyrics conditioned on input text and song form, ensuring alignment with specified syllable constraints. Generated lyrics samples are available at: https://tinyurl.com/lyrics9999
Authors: Hyun Ryu, Eric Kim
Abstract: Efficient inference in large language models (LLMs) has become a critical focus as their scale and complexity grow. Traditional autoregressive decoding, while effective, suffers from computational inefficiencies due to its sequential token generation process. Speculative decoding addresses this bottleneck by introducing a two-stage framework: drafting and verification. A smaller, efficient model generates a preliminary draft, which is then refined by a larger, more sophisticated model. This paper provides a comprehensive survey of speculative decoding methods, categorizing them into draft-centric and model-centric approaches. We discuss key ideas associated with each method, highlighting their potential for scaling LLM inference. This survey aims to guide future research in optimizing speculative decoding and its integration into real-world LLM applications.
Authors: Jiawei Yu, Yuang Li, Xiaosong Qiao, Huan Zhao, Xiaofeng Zhao, Wei Tang, Min Zhang, Hao Yang, Jinsong Su
Abstract: Text-to-speech (TTS) models have been widely adopted to enhance automatic speech recognition (ASR) systems using text-only corpora, thereby reducing the cost of labeling real speech data. Existing research primarily utilizes additional text data and predefined speech styles supported by TTS models. In this paper, we propose Hard-Synth, a novel ASR data augmentation method that leverages large language models (LLMs) and advanced zero-shot TTS. Our approach employs LLMs to generate diverse in-domain text through rewriting, without relying on additional text data. Rather than using predefined speech styles, we introduce a hard prompt selection method with zero-shot TTS to clone speech styles that the ASR model finds challenging to recognize. Experiments demonstrate that Hard-Synth significantly enhances the Conformer model, achieving relative word error rate (WER) reductions of 6.5\%/4.4\% on LibriSpeech dev/test-other subsets. Additionally, we show that Hard-Synth is data-efficient and capable of reducing bias in ASR.
Authors: Zichen Wen, Dadi Guo, Huishuai Zhang
Abstract: As large language models (LLMs) rapidly advance and integrate into daily life, the privacy risks they pose are attracting increasing attention. We focus on a specific privacy risk where LLMs may help identify the authorship of anonymous texts, which challenges the effectiveness of anonymity in real-world systems such as anonymous peer review systems. To investigate these risks, we present AIDBench, a new benchmark that incorporates several author identification datasets, including emails, blogs, reviews, articles, and research papers. AIDBench utilizes two evaluation methods: one-to-one authorship identification, which determines whether two texts are from the same author; and one-to-many authorship identification, which, given a query text and a list of candidate texts, identifies the candidate most likely written by the same author as the query text. We also introduce a Retrieval-Augmented Generation (RAG)-based method to enhance the large-scale authorship identification capabilities of LLMs, particularly when input lengths exceed the models' context windows, thereby establishing a new baseline for authorship identification using LLMs. Our experiments with AIDBench demonstrate that LLMs can correctly guess authorship at rates well above random chance, revealing new privacy risks posed by these powerful models. The source code and data will be made publicly available after acceptance.
Authors: Xu Zou
Abstract: Recently, generative pre-trained models have made significant strides, particularly highlighted by the release of ChatGPT and GPT-4, which exhibit superior cross-domain capabilities. However, these models still face challenges on constrained writing tasks like poem generation under open-domain titles. In response to this challenge, we introduce Block Inverse Prompting (BIPro) constrained generation framework. BIPro leverages two block inverse prompting methods, revise and rewrite, that mimic the process of human text writing using block generative models. It significantly improves the zero-shot generation quality on the formidable constrained generation task of open-domain traditional-form Chinese poem generation. Based on a less powerful block generative model GLM-10B-Chinese, poems composed via BIPro without priming or additional training outperform both most advanced direct generative systems like GPT-4 or GLM-4 and best domain-specific systems such as Yusheng, Shisanbai, or Baidu Poetry Helper in human evaluation by proficient poets. Finally, BIPro considerably narrows the gap between AI-generated works and short-listed human literary arts in another human evaluation, unveiling the promising potential of block generative models in improving the quality of constrained generation.
Authors: Zhibo Chu, Zichong Wang, Qitao Qin
Abstract: Large Language Models (LLMs) exhibit impressive problem-solving skills across many tasks, but they still underperform compared to humans in various downstream applications, such as text-to-SQL. On the BIRD benchmark leaderboard, human performance achieves an accuracy of 92.96\%, whereas the top-performing method reaches only 72.39\%. Notably, these state-of-the-art (SoTA) methods predominantly rely on in-context learning to simulate human-like reasoning. However, they overlook a critical human skill: continual learning. Inspired by the educational practice of maintaining mistake notebooks during our formative years, we propose LPE-SQL (Leveraging Prior Experience: An Expandable Auxiliary Knowledge Base for Text-to-SQL), a novel framework designed to augment LLMs by enabling continual learning without requiring parameter fine-tuning. LPE-SQL consists of four modules that \textbf{i)} retrieve relevant entries, \textbf{ii)} efficient sql generation, \textbf{iii)} generate the final result through a cross-consistency mechanism and \textbf{iv)} log successful and failed tasks along with their reasoning processes or reflection-generated tips. Importantly, the core module of LPE-SQL is the fourth one, while the other modules employ foundational methods, allowing LPE-SQL to be easily integrated with SoTA technologies to further enhance performance. Our experimental results demonstrate that this continual learning approach yields substantial performance gains, with the smaller Llama-3.1-70B model with surpassing the performance of the larger Llama-3.1-405B model using SoTA methods.
Authors: Jo\~ao Gon\c{c}alves
Abstract: This paper presents CAALM-TC (Combining Autoregressive and Autoencoder Language Models for Text Classification), a novel method that enhances text classification by integrating autoregressive and autoencoder language models. Autoregressive large language models such as Open AI's GPT, Meta's Llama or Microsoft's Phi offer promising prospects for content analysis practitioners, but they generally underperform supervised BERT based models for text classification. CAALM leverages autoregressive models to generate contextual information based on input texts, which is then combined with the original text and fed into an autoencoder model for classification. This hybrid approach capitalizes on the extensive contextual knowledge of autoregressive models and the efficient classification capabilities of autoencoders. Experimental results on four benchmark datasets demonstrate that CAALM consistently outperforms existing methods, particularly in tasks with smaller datasets and more abstract classification objectives. The findings indicate that CAALM offers a scalable and effective solution for automated content analysis in social science research that minimizes sample size requirements.
Authors: Yige Yuan, Bingbing Xu, Hexiang Tan, Fei Sun, Teng Xiao, Wei Li, Huawei Shen, Xueqi Cheng
Abstract: Confidence calibration in LLMs, i.e., aligning their self-assessed confidence with the actual accuracy of their responses, enabling them to self-evaluate the correctness of their outputs. However, current calibration methods for LLMs typically estimate two scalars to represent overall response confidence and correctness, which is inadequate for long-form generation where the response includes multiple atomic facts and may be partially confident and correct. These methods also overlook the relevance of each fact to the query. To address these challenges, we propose a Fact-Level Calibration framework that operates at a finer granularity, calibrating confidence to relevance-weighted correctness at the fact level. Furthermore, comprehensive analysis under the framework inspired the development of Confidence-Guided Fact-level Self-Correction ($\textbf{ConFix}$), which uses high-confidence facts within a response as additional knowledge to improve low-confidence ones. Extensive experiments across four datasets and six models demonstrate that ConFix effectively mitigates hallucinations without requiring external knowledge sources such as retrieval systems.
Authors: Lucie Charlotte Magister, Katherine Metcalf, Yizhe Zhang, Maartje ter Hoeve
Abstract: Large Language Models (LLMs) have quickly become an invaluable assistant for a variety of tasks. However, their effectiveness is constrained by their ability to tailor responses to human preferences and behaviors via personalization. Prior work in LLM personalization has largely focused on style transfer or incorporating small factoids about the user, as knowledge injection remains an open challenge. In this paper, we explore injecting knowledge of prior conversations into LLMs to enable future work on less redundant, personalized conversations. We identify two real-world constraints: (1) conversations are sequential in time and must be treated as such during training, and (2) per-user personalization is only viable in parameter-efficient settings. To this aim, we propose PLUM, a pipeline performing data augmentation for up-sampling conversations as question-answer pairs, that are then used to finetune a low-rank adaptation adapter with a weighted cross entropy loss. Even in this first exploration of the problem, we perform competitively with baselines such as RAG, attaining an accuracy of 81.5% across 100 conversations.
Authors: Dat Van-Thanh Nguyen, Tin Van Huynh, Kiet Van Nguyen, Ngan Luu-Thuy Nguyen
Abstract: Natural Language Inference (NLI) is a task within Natural Language Processing (NLP) that holds value for various AI applications. However, there have been limited studies on Natural Language Inference in Vietnamese that explore the concept of joint models. Therefore, we conducted experiments using various combinations of contextualized language models (CLM) and neural networks. We use CLM to create contextualized work presentations and use Neural Networks for classification. Furthermore, we have evaluated the strengths and weaknesses of each joint model and identified the model failure points in the Vietnamese context. The highest F1 score in this experiment, up to 82.78\% in the benchmark dataset (ViNLI). By conducting experiments with various models, the most considerable size of the CLM is XLM-R (355M). That combination has consistently demonstrated superior performance compared to fine-tuning strong pre-trained language models like PhoBERT (+6.58\%), mBERT (+19.08\%), and XLM-R (+0.94\%) in terms of F1-score. This article aims to introduce a novel approach or model that attains improved performance for Vietnamese NLI. Overall, we find that the joint approach of CLM and neural networks is simple yet capable of achieving high-quality performance, which makes it suitable for applications that require efficient resource utilization.
Authors: Muhammad Sharif, Jiangyan Yi, Muhammad Shoaib
Abstract: The language called Balti belongs to the Sino-Tibetan, specifically the Tibeto-Burman language family. It is understood with variations, across populations in India, China, Pakistan, Nepal, Tibet, Burma, and Bhutan, influenced by local cultures and producing various dialects. Considering the diverse cultural, socio-political, religious, and geographical impacts, it is important to step forward unifying the dialects, the basis of common root, lexica, and phonological perspectives, is vital. In the era of globalization and the increasingly frequent developments in AI technology, understanding the diversity and the efforts of dialect unification is important to understanding commonalities and shortening the gaps impacted by unavoidable circumstances. This article analyzes and examines how artificial intelligence AI in the essence of Large Language Models LLMs, can assist in analyzing, documenting, and standardizing the endangered Balti Language, based on the efforts made in different dialects so far.
Authors: Salvatore Mario Carta, Stefano Chessa, Giulia Contu, Andrea Corriga, Andrea Deidda, Gianni Fenu, Luca Frigau, Alessandro Giuliani, Luca Grassi, Marco Manolo Manca, Mirko Marras, Francesco Mola, Bastianino Mossa, Piergiorgio Mura, Marco Ortu, Leonardo Piano, Simone Pisano, Alessia Pisu, Alessandro Sebastian Podda, Livio Pompianu, Simone Seu, Sandro Gabriele Tiddia
Abstract: Minority languages are vital to preserving cultural heritage, yet they face growing risks of extinction due to limited digital resources and the dominance of artificial intelligence models trained on high-resource languages. This white paper proposes a framework to generate linguistic tools for low-resource languages, focusing on data creation to support the development of language models that can aid in preservation efforts. Sardinian, an endangered language, serves as the case study to demonstrate the framework's effectiveness. By addressing the data scarcity that hinders intelligent applications for such languages, we contribute to promoting linguistic diversity and support ongoing efforts in language standardization and revitalization through modern technologies.
Authors: Haonan Wang, Qian Liu, Chao Du, Tongyao Zhu, Cunxiao Du, Kenji Kawaguchi, Tianyu Pang
Abstract: Extending context window sizes allows large language models (LLMs) to process longer sequences and handle more complex tasks. Rotary Positional Embedding (RoPE) has become the de facto standard due to its relative positional encoding properties that benefit long-context training. However, we observe that using RoPE with BFloat16 format results in numerical issues, causing it to deviate from its intended relative positional encoding, especially in long-context scenarios. This issue arises from BFloat16's limited precision and accumulates as context length increases, with the first token contributing significantly to this problem. To address this, we develop AnchorAttention, a plug-and-play attention method that alleviates numerical issues caused by BFloat16, improves long-context capabilities, and speeds up training. AnchorAttention reduces unnecessary attention computations, maintains semantic coherence, and boosts computational efficiency by treating the first token as a shared anchor with a consistent position ID, making it visible to all documents within the training context. Experiments on three types of LLMs demonstrate that AnchorAttention significantly improves long-context performance and reduces training time by over 50\% compared to standard full attention mechanisms, while preserving the original LLM's capabilities on general tasks. Our code is available at https://github.com/haonan3/AnchorContext.
Authors: Ryan Lee, Alexander Spangher, Xuezhe Ma
Abstract: A patent must be deemed novel and non-obvious in order to be granted by the US Patent Office (USPTO). If it is not, a US patent examiner will cite the prior work, or prior art, that invalidates the novelty and issue a non-final rejection. Predicting what claims of the invention should change given the prior art is an essential and crucial step in securing invention rights, yet has not been studied before as a learnable task. In this work we introduce the PatentEdits dataset, which contains 105K examples of successful revisions that overcome objections to novelty. We design algorithms to label edits sentence by sentence, then establish how well these edits can be predicted with large language models (LLMs). We demonstrate that evaluating textual entailment between cited references and draft sentences is especially effective in predicting which inventive claims remained unchanged or are novel in relation to prior art.
Authors: John D. Hastings, Sherri Weitl-Harms, Joseph Doty, Zachary L. Myers, Warren Thompson
Abstract: This research explores the application of large language models (LLMs) to generate synthetic datasets for Product Desirability Toolkit (PDT) testing, a key component in evaluating user sentiment and product experience. Utilizing gpt-4o-mini, a cost-effective alternative to larger commercial LLMs, three methods, Word+Review, Review+Word, and Supply-Word, were each used to synthesize 1000 product reviews. The generated datasets were assessed for sentiment alignment, textual diversity, and data generation cost. Results demonstrated high sentiment alignment across all methods, with Pearson correlations ranging from 0.93 to 0.97. Supply-Word exhibited the highest diversity and coverage of PDT terms, although with increased generation costs. Despite minor biases toward positive sentiments, in situations with limited test data, LLM-generated synthetic data offers significant advantages, including scalability, cost savings, and flexibility in dataset production.
Authors: Mingyu Jin, Weidi Luo, Sitao Cheng, Xinyi Wang, Wenyue Hua, Ruixiang Tang, William Yang Wang, Yongfeng Zhang
Abstract: Large Language Models (LLMs) have demonstrated strong performance in handling complex tasks requiring both extensive knowledge and reasoning abilities. However, the existing LLM inference pipeline operates as an opaque process without explicit separation between knowledge retrieval and reasoning steps, making the model's decision-making process unclear and disorganized. This ambiguity can lead to issues such as hallucinations and knowledge forgetting, which significantly impact the reliability of LLMs in high-stakes domains. In this paper, we propose a new inference paradigm that decomposes the complex inference process into two distinct and clear actions: (1) memory recall: which retrieves relevant knowledge, and (2) reasoning: which performs logical steps based on the recalled knowledge. To facilitate this decomposition, we introduce two special tokens memory and reason, guiding the model to distinguish between steps that require knowledge retrieval and those that involve reasoning. Our experiment results show that this decomposition not only improves model performance but also enhances the interpretability of the inference process, enabling users to identify sources of error and refine model responses effectively. The code is available at https://github.com/MingyuJ666/Disentangling-Memory-and-Reasoning.
URLs: https://github.com/MingyuJ666/Disentangling-Memory-and-Reasoning.
Authors: Chanseo Lee, Sonu Kumar, Kimon A. Vogt, Sam Meraj, Antonia Vogt
Abstract: The increasing demand for multilingual capabilities in healthcare underscores the need for AI models adept at processing diverse languages, particularly in clinical documentation and decision-making. Arabic, with its complex morphology, syntax, and diglossia, poses unique challenges for natural language processing (NLP) in medical contexts. This case study evaluates Sporo AraSum, a language model tailored for Arabic clinical documentation, against JAIS, the leading Arabic NLP model. Using synthetic datasets and modified PDQI-9 metrics modified ourselves for the purposes of assessing model performances in a different language. The study assessed the models' performance in summarizing patient-physician interactions, focusing on accuracy, comprehensiveness, clinical utility, and linguistic-cultural competence. Results indicate that Sporo AraSum significantly outperforms JAIS in AI-centric quantitative metrics and all qualitative attributes measured in our modified version of the PDQI-9. AraSum's architecture enables precise and culturally sensitive documentation, addressing the linguistic nuances of Arabic while mitigating risks of AI hallucinations. These findings suggest that Sporo AraSum is better suited to meet the demands of Arabic-speaking healthcare environments, offering a transformative solution for multilingual clinical workflows. Future research should incorporate real-world data to further validate these findings and explore broader integration into healthcare systems.
Authors: S. Chapagain, Y. Zhao, T. K. Rohleen, S. M. Hamdi, S. F. Boubrahimi, R. E. Flinn, E. M. Lund, D. Klooster, J. R. Scheer, C. J. Cascalheira
Abstract: Individuals who identify as sexual and gender minorities, including lesbian, gay, bisexual, transgender, queer, and others (LGBTQ+) are more likely to experience poorer health than their heterosexual and cisgender counterparts. One primary source that drives these health disparities is minority stress (i.e., chronic and social stressors unique to LGBTQ+ communities' experiences adapting to the dominant culture). This stress is frequently expressed in LGBTQ+ users' posts on social media platforms. However, these expressions are not just straightforward manifestations of minority stress. They involve linguistic complexity (e.g., idiom or lexical diversity), rendering them challenging for many traditional natural language processing methods to detect. In this work, we designed a hybrid model using Graph Neural Networks (GNN) and Bidirectional Encoder Representations from Transformers (BERT), a pre-trained deep language model to improve the classification performance of minority stress detection. We experimented with our model on a benchmark social media dataset for minority stress detection (LGBTQ+ MiSSoM+). The dataset is comprised of 5,789 human-annotated Reddit posts from LGBTQ+ subreddits. Our approach enables the extraction of hidden linguistic nuances through pretraining on a vast amount of raw data, while also engaging in transductive learning to jointly develop representations for both labeled training data and unlabeled test data. The RoBERTa-GCN model achieved an accuracy of 0.86 and an F1 score of 0.86, surpassing the performance of other baseline models in predicting LGBTQ+ minority stress. Improved prediction of minority stress expressions on social media could lead to digital health interventions to improve the wellbeing of LGBTQ+ people-a community with high rates of stress-sensitive health problems.
Authors: Hermann Kroll, Pascal Sackhoff, Bill Matthias Thang, Maha Ksouri, Wolf-Tilo Balke
Abstract: Digital libraries that maintain extensive textual collections may want to further enrich their content for certain downstream applications, e.g., building knowledge graphs, semantic enrichment of documents, or implementing novel access paths. All of these applications require some text processing, either to identify relevant entities, extract semantic relationships between them, or to classify documents into some categories. However, implementing reliable, supervised workflows can become quite challenging for a digital library because suitable training data must be crafted, and reliable models must be trained. While many works focus on achieving the highest accuracy on some benchmarks, we tackle the problem from a digital library practitioner. In other words, we also consider trade-offs between accuracy and application costs, dive into training data generation through distant supervision and large language models such as ChatGPT, LLama, and Olmo, and discuss how to design final pipelines. Therefore, we focus on relation extraction and text classification, using the showcase of eight biomedical benchmarks.
Authors: Junghoon Kim, Junmo Lee, Yeonjun In, Kanghoon Yoon, Chanyoung Park
Abstract: Social graph-based fake news detection aims to identify news articles containing false information by utilizing social contexts, e.g., user information, tweets and comments. However, conventional methods are evaluated under less realistic scenarios, where the model has access to future knowledge on article-related and context-related data during training. In this work, we newly formalize a more realistic evaluation scheme that mimics real-world scenarios, where the data is temporality-aware and the detection model can only be trained on data collected up to a certain point in time. We show that the discriminative capabilities of conventional methods decrease sharply under this new setting, and further propose DAWN, a method more applicable to such scenarios. Our empirical findings indicate that later engagements (e.g., consuming or reposting news) contribute more to noisy edges that link real news-fake news pairs in the social graph. Motivated by this, we utilize feature representations of engagement earliness to guide an edge weight estimator to suppress the weights of such noisy edges, thereby enhancing the detection performance of DAWN. Through extensive experiments, we demonstrate that DAWN outperforms existing fake news detection methods under real-world environments. The source code is available at https://github.com/LeeJunmo/DAWN.
Authors: Claire Bonial, Stephanie M. Lukin, Mitchell Abrams, Anthony Baker, Lucia Donatelli, Ashley Foots, Cory J. Hayes, Cassidy Henry, Taylor Hudson, Matthew Marge, Kimberly A. Pollard, Ron Artstein, David Traum, Clare R. Voss
Abstract: In this paper, we describe the development of symbolic representations annotated on human-robot dialogue data to make dimensions of meaning accessible to autonomous systems participating in collaborative, natural language dialogue, and to enable common ground with human partners. A particular challenge for establishing common ground arises in remote dialogue (occurring in disaster relief or search-and-rescue tasks), where a human and robot are engaged in a joint navigation and exploration task of an unfamiliar environment, but where the robot cannot immediately share high quality visual information due to limited communication constraints. Engaging in a dialogue provides an effective way to communicate, while on-demand or lower-quality visual information can be supplemented for establishing common ground. Within this paradigm, we capture propositional semantics and the illocutionary force of a single utterance within the dialogue through our Dialogue-AMR annotation, an augmentation of Abstract Meaning Representation. We then capture patterns in how different utterances within and across speaker floors relate to one another in our development of a multi-floor Dialogue Structure annotation schema. Finally, we begin to annotate and analyze the ways in which the visual modalities provide contextual information to the dialogue for overcoming disparities in the collaborators' understanding of the environment. We conclude by discussing the use-cases, architectures, and systems we have implemented from our annotations that enable physical robots to autonomously engage with humans in bi-directional dialogue and navigation.
Authors: Shang Liu, Yu Pan, Guanting Chen, Xiaocheng Li
Abstract: Learning a reward model (RM) from human preferences has been an important component in aligning large language models (LLMs). The canonical setup of learning RMs from pairwise preference data is rooted in the classic Bradley-Terry (BT) model that accepts binary feedback, i.e., the label being either Response 1 is better than Response 2, or the opposite. Such a setup inevitably discards potentially useful samples (such as "tied" between the two responses) and loses more fine-grained information (such as "slightly better"). In this paper, we propose a framework for learning RMs under ordinal feedback which generalizes the case of binary preference feedback to any arbitrary granularity. Specifically, we first identify a marginal unbiasedness condition, which generalizes the assumption of the BT model in the existing binary feedback setting. The condition validates itself via the sociological concept of the wisdom of the crowd. Under the condition, we develop a natural probability model for pairwise preference data under ordinal feedback and analyze its properties. We prove the statistical benefits of ordinal feedback in terms of reducing the Rademacher complexity compared to the case of binary feedback. The proposed learning objective and the theory also extend to hinge loss and direct policy optimization (DPO). In particular, the theoretical analysis may be of independent interest when applying to a seemingly unrelated problem of knowledge distillation to interpret the bias-variance trade-off therein. The framework also sheds light on writing guidance for human annotators. Our numerical experiments validate that fine-grained feedback leads to better reward learning for both in-distribution and out-of-distribution settings. Further experiments show that incorporating a certain proportion of samples with tied preference boosts RM learning.
Authors: Stephanie M. Lukin, Claire Bonial, Matthew Marge, Taylor Hudson, Cory J. Hayes, Kimberly A. Pollard, Anthony Baker, Ashley N. Foots, Ron Artstein, Felix Gervits, Mitchell Abrams, Cassidy Henry, Lucia Donatelli, Anton Leuski, Susan G. Hill, David Traum, Clare R. Voss
Abstract: We introduce the Situated Corpus Of Understanding Transactions (SCOUT), a multi-modal collection of human-robot dialogue in the task domain of collaborative exploration. The corpus was constructed from multiple Wizard-of-Oz experiments where human participants gave verbal instructions to a remotely-located robot to move and gather information about its surroundings. SCOUT contains 89,056 utterances and 310,095 words from 278 dialogues averaging 320 utterances per dialogue. The dialogues are aligned with the multi-modal data streams available during the experiments: 5,785 images and 30 maps. The corpus has been annotated with Abstract Meaning Representation and Dialogue-AMR to identify the speaker's intent and meaning within an utterance, and with Transactional Units and Relations to track relationships between utterances to reveal patterns of the Dialogue Structure. We describe how the corpus and its annotations have been used to develop autonomous human-robot systems and enable research in open questions of how humans speak to robots. We release this corpus to accelerate progress in autonomous, situated, human-robot dialogue, especially in the context of navigation tasks where details about the environment need to be discovered.
Authors: Xiangzhe Xu, Zian Su, Jinyao Guo, Kaiyuan Zhang, Zhenting Wang, Xiangyu Zhang
Abstract: Recent advances in code-specific large language models (LLMs) have greatly enhanced code generation and refinement capabilities. However, the safety of code LLMs remains under-explored, posing potential risks as insecure code generated by these models may introduce vulnerabilities into real-world systems. Previous work proposes to collect security-focused instruction-tuning dataset from real-world vulnerabilities. It is constrained by the data sparsity of vulnerable code, and has limited applicability in the iterative post-training workflows of modern LLMs. In this paper, we propose ProSec, a novel proactive security alignment approach designed to align code LLMs with secure coding practices. ProSec systematically exposes the vulnerabilities in a code LLM by synthesizing error-inducing coding scenarios from Common Weakness Enumerations (CWEs), and generates fixes to vulnerable code snippets, allowing the model to learn secure practices through advanced preference learning objectives. The scenarios synthesized by ProSec triggers 25 times more vulnerable code than a normal instruction-tuning dataset, resulting in a security-focused alignment dataset 7 times larger than the previous work. Experiments show that models trained with ProSec is 29.2% to 35.5% more secure compared to previous work, with a marginal negative effect of less than 2 percentage points on model's utility.
Authors: Xuechen Zhang, Xiangyu Chang, Mingchen Li, Amit Roy-Chowdhury, Jiasi Chen, Samet Oymak
Abstract: The attention mechanism within the transformer architecture enables the model to weigh and combine tokens based on their relevance to the query. While self-attention has enjoyed major success, it notably treats all queries $q$ in the same way by applying the mapping $V^\top\text{softmax}(Kq)$, where $V,K$ are the value and key embeddings respectively. In this work, we argue that this uniform treatment hinders the ability to control contextual sparsity and relevance. As a solution, we introduce the $\textit{Selective Self-Attention}$ (SSA) layer that augments the softmax nonlinearity with a principled temperature scaling strategy. By controlling temperature, SSA adapts the contextual sparsity of the attention map to the query embedding and its position in the context window. Through theory and experiments, we demonstrate that this alleviates attention dilution, aids the optimization process, and enhances the model's ability to control softmax spikiness of individual queries. We also incorporate temperature scaling for value embeddings and show that it boosts the model's ability to suppress irrelevant/noisy tokens. Notably, SSA is a lightweight method which introduces less than 0.5% new parameters through a weight-sharing strategy and can be fine-tuned on existing LLMs. Extensive empirical evaluations demonstrate that SSA-equipped models achieve a noticeable and consistent accuracy improvement on language modeling benchmarks.
Authors: David Brandfonbrener, Nikhil Anand, Nikhil Vyas, Eran Malach, Sham Kakade
Abstract: While scaling laws provide a reliable methodology for predicting train loss across compute scales for a single data distribution, less is known about how these predictions should change as we change the distribution. In this paper, we derive a strategy for predicting one loss from another and apply it to predict across different pre-training datasets and from pre-training data to downstream task data. Our predictions extrapolate well even at 20x the largest FLOP budget used to fit the curves. More precisely, we find that there are simple shifted power law relationships between (1) the train losses of two models trained on two separate datasets when the models are paired by training compute (train-to-train), (2) the train loss and the test loss on any downstream distribution for a single model (train-to-test), and (3) the test losses of two models trained on two separate train datasets (test-to-test). The results hold up for pre-training datasets that differ substantially (some are entirely code and others have no code at all) and across a variety of downstream tasks. Finally, we find that in some settings these shifted power law relationships can yield more accurate predictions than extrapolating single-dataset scaling laws.
Authors: Mircea Lic\u{a}, Ojas Shirekar, Baptiste Colle, Chirag Raman
Abstract: Contemporary embodied agents, such as Voyager in Minecraft, have demonstrated promising capabilities in open-ended individual learning. However, when powered with open large language models (LLMs), these agents often struggle with rudimentary tasks, even when fine-tuned on domain-specific knowledge. Inspired by human cultural learning, we present \collabvoyager, a novel framework that enhances Voyager with lifelong collaborative learning through explicit perspective-taking. \collabvoyager introduces three key innovations: (1) theory of mind representations linking percepts, beliefs, desires, and actions; (2) natural language communication between agents; and (3) semantic memory of task and environment knowledge and episodic memory of collaboration episodes. These advancements enable agents to reason about their and others' mental states, empirically addressing two prevalent failure modes: false beliefs and faulty task executions. In mixed-expertise Minecraft experiments, \collabvoyager agents outperform Voyager counterparts, significantly improving task completion rate by $66.6\% (+39.4\%)$ for collecting one block of dirt and $70.8\% (+20.8\%)$ for collecting one wood block. They exhibit emergent behaviors like knowledge transfer from expert to novice agents and collaborative code correction. \collabvoyager agents also demonstrate the ability to adapt to out-of-distribution tasks by using their previous experiences and beliefs obtained through collaboration. In this open-ended social learning paradigm, \collabvoyager paves the way for the democratic development of embodied AI, where agents learn in deployment from both peer and environmental feedback.
Authors: Zhuohan Gu, Jiayi Yao, Kuntai Du, Junchen Jiang
Abstract: As large language models (LLMs) show impressive performance on complex tasks, they still struggle with longer contextual understanding and high computational costs. To balance efficiency and quality, we introduce LLMSteer, a fine-tuning-free framework that enhances LLMs through query-independent attention steering. Tested on popular LLMs and datasets, LLMSteer narrows the performance gap with baselines by 65.9% and reduces the runtime delay by up to 4.8x compared to recent attention steering methods.
Authors: Siyuan Jin, Zhendong Bei, Bichao Chen, Yong Xia
Abstract: Traditional banks face significant challenges in digital transformation, primarily due to legacy system constraints and fragmented ownership. Recent incidents show that such fragmentation often results in superficial incident resolutions, leaving root causes unaddressed and causing recurring failures. We introduce a novel approach to post-incident analysis, integrating knowledge-based GenAI agents with the "Five Whys" technique to examine problem descriptions and change request data. This method uncovered that approximately 70% of the incidents previously attributed to management or vendor failures were due to underlying internal code issues. We present a case study to show the impact of our method. By scanning over 5,000 projects, we identified over 400 files with a similar root cause. Overall, we leverage the knowledge-based agents to automate and elevate root cause analysis, transforming it into a more proactive process. These agents can be applied across other phases of the software development lifecycle, further improving development processes.
Authors: Gang Zhao, Ximing Zhang, Chenji Lu, Hui Zhao, Tianshu Wu, Pengjie Wang, Jian Xu, Bo Zheng
Abstract: Effective query-item relevance modeling is pivotal for enhancing user experience and safeguarding user satisfaction in e-commerce search systems. Recently, benefiting from the vast inherent knowledge, Large Language Model (LLM) approach demonstrates strong performance and long-tail generalization ability compared with previous neural-based specialized relevance learning methods. Though promising, current LLM-based methods encounter the following inadequacies in practice: First, the massive parameters and computational demands make it difficult to be deployed online. Second, distilling LLM models to online models is a feasible direction, but the LLM relevance modeling is a black box, and its rich intrinsic knowledge is difficult to extract and apply online. To improve the interpretability of LLM and boost the performance of online relevance models via LLM, we propose an Explainable LLM-driven Multi-dimensional Distillation framework for e-commerce relevance learning, which comprises two core components: (1) An Explainable LLM for relevance modeling (ELLM-rele), which decomposes the relevance learning into intermediate steps and models relevance learning as a Chain-of-Thought (CoT) reasoning, thereby enhancing both interpretability and performance of LLM. (2) A Multi-dimensional Knowledge Distillation (MKD) architecture that transfers the knowledge of ELLM-rele to current deployable interaction-based and representation-based student models from both the relevance score distribution and CoT reasoning aspects. Through distilling the probabilistic and CoT reasoning knowledge, MKD improves both the semantic interaction and long-tail generalization abilities of student models. Extensive offline evaluations and online experiments on Taobao search ad scene demonstrate that our proposed framework significantly enhances e-commerce relevance learning performance and user experience.
Authors: Ziyang Luo, Haoning Wu, Dongxu Li, Jing Ma, Mohan Kankanhalli, Junnan Li
Abstract: Large multimodal models (LMMs) with advanced video analysis capabilities have recently garnered significant attention. However, most evaluations rely on traditional methods like multiple-choice questions in benchmarks such as VideoMME and LongVideoBench, which are prone to lack the depth needed to capture the complex demands of real-world users. To address this limitation-and due to the prohibitive cost and slow pace of human annotation for video tasks-we introduce VideoAutoArena, an arena-style benchmark inspired by LMSYS Chatbot Arena's framework, designed to automatically assess LMMs' video analysis abilities. VideoAutoArena utilizes user simulation to generate open-ended, adaptive questions that rigorously assess model performance in video understanding. The benchmark features an automated, scalable evaluation framework, incorporating a modified ELO Rating System for fair and continuous comparisons across multiple LMMs. To validate our automated judging system, we construct a 'gold standard' using a carefully curated subset of human annotations, demonstrating that our arena strongly aligns with human judgment while maintaining scalability. Additionally, we introduce a fault-driven evolution strategy, progressively increasing question complexity to push models toward handling more challenging video analysis scenarios. Experimental results demonstrate that VideoAutoArena effectively differentiates among state-of-the-art LMMs, providing insights into model strengths and areas for improvement. To further streamline our evaluation, we introduce VideoAutoBench as an auxiliary benchmark, where human annotators label winners in a subset of VideoAutoArena battles. We use GPT-4o as a judge to compare responses against these human-validated answers. Together, VideoAutoArena and VideoAutoBench offer a cost-effective, and scalable framework for evaluating LMMs in user-centric video analysis.
Authors: Stefano Scanzio, Francesco Velluto, Matteo Rosani, Lukasz Wisniewski, Gianluca Cena
Abstract: Executable QR codes, also known as eQR codes or just sQRy, are a special kind of QR codes that embed programs conceived to run on mobile devices like smartphones. Since the program is directly encoded in binary form within the QR code, it can be executed even when the reading device is not provided with Internet access. The applications of this technology are manifold, and range from smart user guides to advisory systems. The first programming language made available for eQR is QRtree, which enables the implementation of decision trees aimed, for example, at guiding the user in operating/maintaining a complex machinery or for reaching a specific location. In this work, an additional language is proposed, we term QRind, which was specifically devised for Industry. It permits to integrate distinct computational blocks into the QR code, e.g., machine learning models to enable predictive maintenance and algorithms to ease machinery usage. QRind permits the Industry 4.0/5.0 paradigms to be implemented, in part, also in those cases where Internet is unavailable.
Authors: Houssam Eddine-Othman Lachemat, Akli Abbas, Nourredine Oukas, Yassine El Kheir, Samia Haboussi, Absar Showdhury Shammur
Abstract: The paper introduces and publicly releases (Data download link available after acceptance) CAFE -- the first Code-switching dataset between Algerian dialect, French, and english languages. The CAFE speech data is unique for (a) its spontaneous speaking style in vivo human-human conversation capturing phenomena like code-switching and overlapping speech, (b) addresses distinct linguistic challenges in North African Arabic dialect; (c) the CAFE captures dialectal variations from various parts of Algeria within different sociolinguistic contexts. CAFE data contains approximately 37 hours of speech, with a subset, CAFE-small, of 2 hours and 36 minutes released with manual human annotation including speech segmentation, transcription, explicit annotation of code-switching points, overlapping speech, and other events such as noises, and laughter among others. The rest approximately 34.58 hours contain pseudo label transcriptions. In addition to the data release, the paper also highlighted the challenges of using state-of-the-art Automatic Speech Recognition (ASR) models such as Whisper large-v2,3 and PromptingWhisper to handle such content. Following, we benchmark CAFE data with the aforementioned Whisper models and show how well-designed data processing pipelines and advanced decoding techniques can improve the ASR performance in terms of Mixed Error Rate (MER) of 0.310, Character Error Rate (CER) of 0.329 and Word Error Rate (WER) of 0.538.
Authors: Jiacheng Liang, Zian Wang, Lauren Hong, Shouling Ji, Ting Wang
Abstract: To mitigate the misuse of large language models (LLMs), such as disinformation, automated phishing, and academic cheating, there is a pressing need for the capability of identifying LLM-generated texts. Watermarking emerges as one promising solution: it plants statistical signals into LLMs' generative processes and subsequently verifies whether LLMs produce given texts. Various watermarking methods (``watermarkers'') have been proposed; yet, due to the lack of unified evaluation platforms, many critical questions remain under-explored: i) What are the strengths/limitations of various watermarkers, especially their attack robustness? ii) How do various design choices impact their robustness? iii) How to optimally operate watermarkers in adversarial environments? To fill this gap, we systematize existing LLM watermarkers and watermark removal attacks, mapping out their design spaces. We then develop WaterPark, a unified platform that integrates 10 state-of-the-art watermarkers and 12 representative attacks. More importantly, leveraging WaterPark, we conduct a comprehensive assessment of existing watermarkers, unveiling the impact of various design choices on their attack robustness. For instance, a watermarker's resilience to increasingly intensive attacks hinges on its context dependency. We further explore the best practices to operate watermarkers in adversarial environments. For instance, using a generic detector alongside a watermark-specific detector improves the security of vulnerable watermarkers. We believe our study sheds light on current LLM watermarking techniques while WaterPark serves as a valuable testbed to facilitate future research.
Authors: Gaurav Verma, Rachneet Kaur, Nishan Srishankar, Zhen Zeng, Tucker Balch, Manuela Veloso
Abstract: State-of-the-art multimodal web agents, powered by Multimodal Large Language Models (MLLMs), can autonomously execute many web tasks by processing user instructions and interacting with graphical user interfaces (GUIs). Current strategies for building web agents rely on (i) the generalizability of underlying MLLMs and their steerability via prompting, and (ii) large-scale fine-tuning of MLLMs on web-related tasks. However, web agents still struggle to automate tasks on unseen websites and domains, limiting their applicability to enterprise-specific and proprietary platforms. Beyond generalization from large-scale pre-training and fine-tuning, we propose building agents for few-shot adaptability using human demonstrations. We introduce the AdaptAgent framework that enables both proprietary and open-weights multimodal web agents to adapt to new websites and domains using few human demonstrations (up to 2). Our experiments on two popular benchmarks -- Mind2Web & VisualWebArena -- show that using in-context demonstrations (for proprietary models) or meta-adaptation demonstrations (for meta-learned open-weights models) boosts task success rate by 3.36% to 7.21% over non-adapted state-of-the-art models, corresponding to a relative increase of 21.03% to 65.75%. Furthermore, our additional analyses (a) show the effectiveness of multimodal demonstrations over text-only ones, (b) shed light on the influence of different data selection strategies during meta-learning on the generalization of the agent, and (c) demonstrate the effect of number of few-shot examples on the web agent's success rate. Overall, our results unlock a complementary axis for developing widely applicable multimodal web agents beyond large-scale pre-training and fine-tuning, emphasizing few-shot adaptability.
Authors: Ga\v{s}per Begu\v{s}, Thomas Lu, Zili Wang
Abstract: Computational models of syntax are predominantly text-based. Here we propose that the most basic first step in the evolution of syntax can be modeled directly from raw speech in a fully unsupervised way. We focus on one of the most ubiquitous and elementary suboperation of syntax -- concatenation. We introduce spontaneous concatenation: a phenomenon where convolutional neural networks (CNNs) trained on acoustic recordings of individual words start generating outputs with two or even three words concatenated without ever accessing data with multiple words in the input. We replicate this finding in several independently trained models with different hyperparameters and training data. Additionally, networks trained on two words learn to embed words into novel unobserved word combinations. We also show that the concatenated outputs contain precursors to compositionality. To our knowledge, this is a previously unreported property of CNNs trained in the ciwGAN/fiwGAN setting on raw speech and has implications both for our understanding of how these architectures learn as well as for modeling syntax and its evolution in the brain from raw acoustic inputs. We also propose a potential neural mechanism called disinhibition that outlines a possible neural pathway towards concatenation and compositionality and suggests our modeling is useful for generating testable prediction for biological and artificial neural processing of speech.
Authors: Ming Dong, Yujing Chen, Miao Zhang, Hao Sun, Tingting He
Abstract: Chinese Spell Checking (CSC) is a widely used technology, which plays a vital role in speech to text (STT) and optical character recognition (OCR). Most of the existing CSC approaches relying on BERT architecture achieve excellent performance. However, limited by the scale of the foundation model, BERT-based method does not work well in few-shot scenarios, showing certain limitations in practical applications. In this paper, we explore using an in-context learning method named RS-LLM (Rich Semantic based LLMs) to introduce large language models (LLMs) as the foundation model. Besides, we study the impact of introducing various Chinese rich semantic information in our framework. We found that by introducing a small number of specific Chinese rich semantic structures, LLMs achieve better performance than the BERT-based model on few-shot CSC task. Furthermore, we conduct experiments on multiple datasets, and the experimental results verified the superiority of our proposed framework.
Authors: Xuechen Zhang, Zijian Huang, Ege Onur Taga, Carlee Joe-Wong, Samet Oymak, Jiasi Chen
Abstract: Recent successes in natural language processing have led to the proliferation of large language models (LLMs) by multiple providers. Each LLM offering has different inference accuracy, monetary cost, and latency, and their accuracy further depends on the exact wording of the question (i.e., the specific prompt). At the same time, users often have a limit on monetary budget and latency to answer all their questions, and they do not know which LLMs to choose for each question to meet their accuracy and long term budget requirements. To navigate this rich design space, we propose TREACLE ($\underline{T}$hrifty $\underline{Rea}$soning via $\underline{C}$ontext-Aware $\underline{L}$LM and Prompt S$\underline{e}$lection), a reinforcement learning policy that jointly selects the model and prompting scheme while respecting the user's monetary cost and latency constraints. TREACLE uses the problem context, including question text embeddings (reflecting the type or difficulty of a query) and the response history (reflecting the consistency of previous responses) to make smart decisions. Our evaluations on standard reasoning datasets (GSM8K, CSQA, and LLC) with various LLMs and prompts show that TREACLE enables cost savings of up to 85% compared to baselines, while maintaining high accuracy. Importantly, it provides the user with the ability to gracefully trade off accuracy for cost.
Authors: Bowen Ping, Shuo Wang, Hanqing Wang, Xu Han, Yuzhuang Xu, Yukun Yan, Yun Chen, Baobao Chang, Zhiyuan Liu, Maosong Sun
Abstract: Fine-tuning is a crucial process for adapting large language models (LLMs) to diverse applications. In certain scenarios, such as multi-tenant serving, deploying multiple LLMs becomes necessary to meet complex demands. Recent studies suggest decomposing a fine-tuned LLM into a base model and corresponding delta weights, which are then compressed using low-rank or low-bit approaches to reduce costs. In this work, we observe that existing low-rank and low-bit compression methods can significantly harm the model performance for task-specific fine-tuned LLMs (e.g., WizardMath for math problems). Motivated by the long-tail distribution of singular values in the delta weights, we propose a delta quantization approach using mixed-precision. This method employs higher-bit representation for singular vectors corresponding to larger singular values. We evaluate our approach on various fine-tuned LLMs, including math LLMs, code LLMs, chat LLMs, and even VLMs. Experimental results demonstrate that our approach performs comparably to full fine-tuned LLMs, surpassing both low-rank and low-bit baselines by a considerable margin. Additionally, we show that our method is compatible with various backbone LLMs, such as Llama-2, Llama-3, and Mistral, highlighting its generalizability.
Authors: Zhuofeng Li, Zixing Gou, Xiangnan Zhang, Zhongyuan Liu, Sirui Li, Yuntong Hu, Chen Ling, Zheng Zhang, Liang Zhao
Abstract: Text-Attributed Graphs (TAGs) augment graph structures with natural language descriptions, facilitating detailed depictions of data and their interconnections across various real-world settings. However, existing TAG datasets predominantly feature textual information only at the nodes, with edges typically represented by mere binary or categorical attributes. This lack of rich textual edge annotations significantly limits the exploration of contextual relationships between entities, hindering deeper insights into graph-structured data. To address this gap, we introduce Textual-Edge Graphs Datasets and Benchmark (TEG-DB), a comprehensive and diverse collection of benchmark textual-edge datasets featuring rich textual descriptions on nodes and edges. The TEG-DB datasets are large-scale and encompass a wide range of domains, from citation networks to social networks. In addition, we conduct extensive benchmark experiments on TEG-DB to assess the extent to which current techniques, including pre-trained language models, graph neural networks, and their combinations, can utilize textual node and edge information. Our goal is to elicit advancements in textual-edge graph research, specifically in developing methodologies that exploit rich textual node and edge descriptions to enhance graph analysis and provide deeper insights into complex real-world networks. The entire TEG-DB project is publicly accessible as an open-source repository on Github, accessible at https://github.com/Zhuofeng-Li/TEG-Benchmark.
Authors: Sean Welleck, Amanda Bertsch, Matthew Finlayson, Hailey Schoelkopf, Alex Xie, Graham Neubig, Ilia Kulikov, Zaid Harchaoui
Abstract: One of the most striking findings in modern research on large language models (LLMs) is that scaling up compute during training leads to better results. However, less attention has been given to the benefits of scaling compute during inference. This survey focuses on these inference-time approaches. We explore three areas under a unified mathematical formalism: token-level generation algorithms, meta-generation algorithms, and efficient generation. Token-level generation algorithms, often called decoding algorithms, operate by sampling a single token at a time or constructing a token-level search space and then selecting an output. These methods typically assume access to a language model's logits, next-token distributions, or probability scores. Meta-generation algorithms work on partial or full sequences, incorporating domain knowledge, enabling backtracking, and integrating external information. Efficient generation methods aim to reduce token costs and improve the speed of generation. Our survey unifies perspectives from three research communities: traditional natural language processing, modern LLMs, and machine learning systems.
Authors: Luohe Shi, Hongyi Zhang, Yao Yao, Zuchao Li, Hai Zhao
Abstract: Large Language Models (LLMs), epitomized by ChatGPT's release in late 2022, have revolutionized various industries with their advanced language comprehension. However, their efficiency is challenged by the Transformer architecture's struggle with handling long texts. KV Cache has emerged as a pivotal solution to this issue, converting the time complexity of token generation from quadratic to linear, albeit with increased GPU memory overhead proportional to conversation length. With the development of the LLM community and academia, various KV Cache compression methods have been proposed. In this review, we dissect the various properties of KV Cache and elaborate on various methods currently used to optimize the KV Cache space usage of LLMs. These methods span the pre-training phase, deployment phase, and inference phase, and we summarize the commonalities and differences among these methods. Additionally, we list some metrics for evaluating the long-text capabilities of large language models, from both efficiency and capability perspectives. Our review thus sheds light on the evolving landscape of LLM optimization, offering insights into future advancements in this dynamic field. Links to the papers mentioned in this review can be found in our Github Repo https://github.com/zcli-charlie/Awesome-KV-Cache.
Authors: Yufei Tao, Adam Hiatt, Erik Haake, Antonie J. Jetter, Ameeta Agrawal
Abstract: Large language models (LLMs) have demonstrated remarkable progress in leveraging diverse knowledge sources. This study investigates how nine widely used LLMs allocate knowledge between local context and global parameters when answering open-ended questions in knowledge-consistent scenarios. We introduce a novel dataset, WikiAtomic, and systematically vary context sizes to analyze how LLMs prioritize and utilize the provided information and their parametric knowledge in knowledge-consistent scenarios. Additionally, we also study their tendency to hallucinate under varying context sizes. Our findings reveal consistent patterns across models, including a consistent reliance on both contextual (around 70%) and parametric (around 30%) knowledge, and a decrease in hallucinations with increasing context. These insights highlight the importance of more effective context organization and developing models that use input more deterministically for robust performance.
Authors: Luohe Shi, Yao Yao, Zuchao Li, Lefei Zhang, Hai Zhao
Abstract: Large language models (LLMs) have rapidly advanced and demonstrated impressive capabilities. In-Context Learning (ICL) and Parameter-Efficient Fine-Tuning (PEFT) are currently two mainstream methods for augmenting LLMs to downstream tasks. ICL typically constructs a few-shot learning scenario, either manually or by setting up a Retrieval-Augmented Generation (RAG) system, helping models quickly grasp domain knowledge or question-answering patterns without changing model parameters. However, this approach involves trade-offs, such as slower inference speed and increased space occupancy. PEFT assists the model in adapting to tasks through minimal parameter modifications, but the training process still demands high hardware requirements, even with a small number of parameters involved. To address these challenges, we propose Reference Trustable Decoding (RTD), a paradigm that allows models to quickly adapt to new tasks without fine-tuning, maintaining low inference costs. RTD constructs a reference datastore from the provided training examples and optimizes the LLM's final vocabulary distribution by flexibly selecting suitable references based on the input, resulting in more trustable responses and enabling the model to adapt to downstream tasks at a low cost. Experimental evaluations on various LLMs using different benchmarks demonstrate that RTD establishes a new paradigm for augmenting models to downstream tasks. Furthermore, our method exhibits strong orthogonality with traditional methods, allowing for concurrent usage. Our code can be found at https://github.com/ShiLuohe/ReferenceTrustableDecoding
URLs: https://github.com/ShiLuohe/ReferenceTrustableDecoding
Authors: Daniel P. Jeong, Saurabh Garg, Zachary C. Lipton, Michael Oberst
Abstract: Several recent works seek to develop foundation models specifically for medical applications, adapting general-purpose large language models (LLMs) and vision-language models (VLMs) via continued pretraining on publicly available biomedical corpora. These works typically claim that such domain-adaptive pretraining (DAPT) improves performance on downstream medical tasks, such as answering medical licensing exam questions. In this paper, we compare seven public "medical" LLMs and two VLMs against their corresponding base models, arriving at a different conclusion: all medical VLMs and nearly all medical LLMs fail to consistently improve over their base models in the zero-/few-shot prompting regime for medical question-answering (QA) tasks. For instance, across the tasks and model pairs we consider in the 3-shot setting, medical LLMs only outperform their base models in 12.1% of cases, reach a (statistical) tie in 49.8% of cases, and are significantly worse than their base models in the remaining 38.2% of cases. Our conclusions are based on (i) comparing each medical model head-to-head, directly against the corresponding base model; (ii) optimizing the prompts for each model separately; and (iii) accounting for statistical uncertainty in comparisons. While these basic practices are not consistently adopted in the literature, our ablations show that they substantially impact conclusions. Our findings suggest that state-of-the-art general-domain models may already exhibit strong medical knowledge and reasoning capabilities, and offer recommendations to strengthen the conclusions of future studies.
Authors: Pedram Hosseini, Jessica M. Sin, Bing Ren, Bryceton G. Thomas, Elnaz Nouri, Ali Farahanchi, Saeed Hassanpour
Abstract: There is a lack of benchmarks for evaluating large language models (LLMs) in long-form medical question answering (QA). Most existing medical QA evaluation benchmarks focus on automatic metrics and multiple-choice questions. While valuable, these benchmarks fail to fully capture or assess the complexities of real-world clinical applications where LLMs are being deployed. Furthermore, existing studies on evaluating long-form answer generation in medical QA are primarily closed-source, lacking access to human medical expert annotations, which makes it difficult to reproduce results and enhance existing baselines. In this work, we introduce a new publicly available benchmark featuring real-world consumer medical questions with long-form answer evaluations annotated by medical doctors. We performed pairwise comparisons of responses from various open and closed-source medical and general-purpose LLMs based on criteria such as correctness, helpfulness, harmfulness, and bias. Additionally, we performed a comprehensive LLM-as-a-judge analysis to study the alignment between human judgments and LLMs. Our preliminary results highlight the strong potential of open LLMs in medical QA compared to leading closed models. Code & Data: https://github.com/lavita-ai/medical-eval-sphere
Authors: Bin Xu, Yiguan Lin, Yinghao Li, Yang Gao
Abstract: Large language models demonstrate exceptional performance in simple code generation tasks but still face challenges in tackling complex problems. These challenges may stem from insufficient reasoning and problem decomposition capabilities. To address this issue, we propose a reasoning-augmented data generation process, SRA-MCTS, which guides the model to autonomously generate high-quality intermediate reasoning paths. This creates a positive feedback loop, enabling continuous improvement. Our method operates entirely through the model itself without requiring additional supervision. By synthesizing natural language reasoning paths and translating them into executable code, the approach ensures analytical accuracy and enhances the success rate in solving complex tasks. Experimental results show that, even without additional supervisory signals, our method achieves performance improvements across different model scales, demonstrating the significant potential of self-improvement in small models. Furthermore, the method remains robust when traditional Chain-of-Thought (CoT) approaches exhibit performance degradation, with notable improvements observed in diversity metrics such as pass@10. We encourage further exploration of reasoning processes within training data to enhance the ability of language models to address complex problems.
Authors: Jai Doshi, Asa Cooper Stickland
Abstract: Large language model unlearning aims to remove harmful information that LLMs have learnt to prevent their use for malicious purposes. LLMU and RMU have been proposed as two methods for LLM unlearning, achieving impressive results on unlearning benchmarks. We study in detail the efficacy of these methods by evaluating their impact on general model capabilities on the WMDP benchmark as well as a biology benchmark we create. Our experiments show that RMU generally leads to better preservation of model capabilities, for similar or better unlearning. We further test the robustness of these methods and find that doing 5-shot prompting or rephrasing the question in simple ways can lead to an over ten-fold increase in accuracy on unlearning benchmarks. Finally, we show that training on unrelated data can almost completely recover pre-unlearning performance, demonstrating that these methods fail at truly unlearning. The code is available at: https://github.com/JaiDoshi/Knowledge-Erasure.
Authors: Daeyong Kwon, SeungHeon Doh, Juhan Nam
Abstract: Intent classification is a text understanding task that identifies user needs from input text queries. While intent classification has been extensively studied in various domains, it has not received much attention in the music domain. In this paper, we investigate intent classification models for music discovery conversation, focusing on pre-trained language models. Rather than only predicting functional needs: intent classification, we also include a task for classifying musical needs: musical attribute classification. Additionally, we propose a method of concatenating previous chat history with just single-turn user queries in the input text, allowing the model to understand the overall conversation context better. Our proposed model significantly improves the F1 score for both user intent and musical attribute classification, and surpasses the zero-shot and few-shot performance of the pretrained Llama 3 model.
Authors: Sneha Singhania, Silviu Cucerzan, Allen Herring, Sujay Kumar Jauhar
Abstract: Capturing fresh information in near real-time and using it to augment existing large language models (LLMs) is essential to generate up-to-date, grounded, and reliable output. This problem becomes particularly challenging when LLMs are used for informational tasks in rapidly evolving fields, such as Web search related to recent or unfolding events involving entities, where generating temporally relevant responses requires access to up-to-the-hour news sources. However, the information modeled by the parametric memory of LLMs is often outdated, and Web results from prototypical retrieval systems may fail to capture the latest relevant information and struggle to handle conflicting reports in evolving news. To address this challenge, we present the NEON framework, designed to extract emerging entity interactions -- such as events or activities -- as described in news articles. NEON constructs an entity-centric timestamped knowledge graph that captures such interactions, thereby facilitating enhanced QA capabilities related to news events. Our framework innovates by integrating open Information Extraction (openIE) style tuples into LLMs to enable in-context retrieval-augmented generation. This integration demonstrates substantial improvements in QA performance when tackling temporal, entity-centric search queries. Through NEON, LLMs can deliver more accurate, reliable, and up-to-date responses.
Authors: Hongbin Ye, Honghao Gui, Aijia Zhang, Tong Liu, Weiqiang Jia
Abstract: This paper introduces CooperKGC, a novel framework challenging the conventional solitary approach of large language models (LLMs) in knowledge graph construction (KGC). CooperKGC establishes a collaborative processing network, assembling a team capable of concurrently addressing entity, relation, and event extraction tasks. Experimentation demonstrates that fostering collaboration within CooperKGC enhances knowledge selection, correction, and aggregation capabilities across multiple rounds of interactions.
Authors: Jian Gu, Aldeida Aleti, Chunyang Chen, Hongyu Zhang
Abstract: Language Models (LMs) have become widely used in software engineering, especially for tasks such as code generation, where they are referred to as code LMs. These models have proven effective in generating code, making it easier for developers to automate coding activities. However, research has highlighted a significant limitation: despite their effectiveness, LMs often produce code that is incorrect, buggy, or not fully functional. Updating these models with limited data can be prohibitively challenging, yet it is essential to maximize their utility. This may require hot-fix techniques (updating models with limited data) to resolve. In this paper, we propose \ul{M}odel \ul{I}mprovement via \ul{N}euron \ul{T}argeting (\textsc{MINT}), a novel approach for repairing code LMs. MINT leverages the semantic property of language models to perform neuron-level repairs in a novel way. Further, by analyzing the relationships between the model's latent representations, the incorrect outputs, and the desired outputs, \textsc{MINT} determines which neurons are worth updating. This approach ensures that only the neurons crucial to the model's failure are targeted, avoiding unnecessary changes and allowing for a more efficient and precise repair process. \textsc{MINT} is effective, efficient, and reliable, capable of correcting a neural model by patching a minimum number of neurons (usually one or two neurons). Our approach is evaluated on three coding tasks: line-level code generation, shellcode generation, and intent-to-bash translation. The experimental results demonstrate that the proposed approach significantly outperforms the state-of-the-art in both effectiveness and efficiency measures. In addition, we analyze and discuss the side effects of model repair techniques, including the balance between generalization and specificity, and the performance after multiple repairs in succession.
Authors: Yang Cao
Abstract: In this paper, we propose Singular Values and Orthonormal Regularized Singular Vectors Adaptation, or SORSA, a novel PEFT method. Each SORSA adapter consists of two main parts: trainable principal singular weights $W_p = U_p \text{diag}(S_p) V^\top_p$, and frozen residual weights $W_r = U_r \text{diag}(S_r) V^\top_r$. These parts are initialized by performing singular value decomposition (SVD) on pre-trained weights. Moreover, we implement and analyze an orthonormal regularizer, which we prove could decrease the condition number of $W_p$ and make the optimization more efficient. SORSA adapters could be merged during inference, thus eliminating any inference latency. We also introduce a method to analyze the variation of the parameters by performing SVD and discuss and analyze SORSA's superiority in minimizing the alteration in the SVD aspect. After all, SORSA shows a faster convergence than LoRA and PiSSA in our experiments. On the GSM-8K benchmark, Llama 2 7B adapted using SORSA achieved 56.03% accuracy, surpassing LoRA (42.30%), AdaLoRA (47.30%), Full FT (49.05%), and PiSSA (53.07%). On the MATH benchmark, SORSA achieved 10.36% accuracy, outperforming LoRA (5.50%), AdaLoRA (6.48%), Full FT (7.22%), and PiSSA (7.44%). We conclude that SORSA offers a new perspective on parameter-efficient fine-tuning, demonstrating remarkable performance.
Authors: Saber Malekmohammadi, Golnoosh Farnadi
Abstract: A significant approach in natural language processing involves large-scale pre-training models on general domain data followed by their adaptation to specific tasks or domains. As models grow in size, full fine-tuning all of their parameters becomes increasingly impractical. To address this, some methods for low-rank task adaptation of language models have been proposed, e.g., LoRA and FLoRA. These methods keep the pre-trained model weights fixed and incorporate trainable low-rank decomposition matrices into some layers of the transformer architecture, called adapters. This approach significantly reduces the number of trainable parameters required for downstream tasks compared to full fine-tuning all parameters. In this work, we look at low-rank adaptation from the lens of data privacy. We show theoretically that the low-rank adaptation used in LoRA and FLoRA is equivalent to injecting some random noise into the batch gradients w.r.t the adapter parameters, and we quantify the variance of the injected noise. By establishing a Berry-Esseen type bound on the total variation distance between distribution of the injected noise and a Gaussian distribution with the same variance, we show that the dynamics of low-rank adaptation is close to that of differentially private fine-tuning of the adapters. Finally, using Johnson-Lindenstrauss lemma, we show that when augmented with gradient scaling, low-rank adaptation is very close to performing DPSGD algorithm with a fixed noise scale to fine-tune the adapters. These theoretical findings suggest that unlike other existing fine-tuning algorithms, low-rank adaptation provides privacy w.r.t the fine-tuning data implicitly.
Authors: Gen Luo, Xue Yang, Wenhan Dou, Zhaokai Wang, Jiawen Liu, Jifeng Dai, Yu Qiao, Xizhou Zhu
Abstract: In this paper, we focus on monolithic Multimodal Large Language Models (MLLMs) that integrate visual encoding and language decoding into a single LLM. In particular, we identify that existing pre-training strategies for monolithic MLLMs often suffer from unstable optimization or catastrophic forgetting. To address this issue, our core idea is to embed a new visual parameter space into a pre-trained LLM, thereby stably learning visual knowledge from noisy data while freezing the LLM. Based on this principle, we present Mono-InternVL, a novel monolithic MLLM that seamlessly integrates a set of visual experts via a multimodal mixture-of-experts structure. Moreover, we propose an innovative pre-training strategy to maximize the visual capability of Mono-InternVL, namely Endogenous Visual Pre-training (EViP). In particular, EViP is designed as a progressive learning process for visual experts, which aims to fully exploit the visual knowledge from noisy data to high-quality data. To validate our approach, we conduct extensive experiments on 16 benchmarks. Experimental results confirm the superior performance of Mono-InternVL than existing monolithic MLLMs on 13 of 16 multimodal benchmarks, e.g., +80 points over Emu3 on OCRBench. Compared to the modular baseline, i.e., InternVL-1.5, Mono-InternVL still retains comparable multimodal performance while reducing up to 67% first token latency. Code and model are released at https://huggingface.co/OpenGVLab/Mono-InternVL-2B.
Authors: Haokun Liu, Yangqiaoyu Zhou, Mingxuan Li, Chenfei Yuan, Chenhao Tan
Abstract: AI holds promise for transforming scientific processes, including hypothesis generation. Prior work on hypothesis generation can be broadly categorized into theory-driven and data-driven approaches. While both have proven effective in generating novel and plausible hypotheses, it remains an open question whether they can complement each other. To address this, we develop the first method that combines literature-based insights with data to perform LLM-powered hypothesis generation. We apply our method on five different datasets and demonstrate that integrating literature and data outperforms other baselines (8.97\% over few-shot, 15.75\% over literature-based alone, and 3.37\% over data-driven alone). Additionally, we conduct the first human evaluation to assess the utility of LLM-generated hypotheses in assisting human decision-making on two challenging tasks: deception detection and AI generated content detection. Our results show that human accuracy improves significantly by 7.44\% and 14.19\% on these tasks, respectively. These findings suggest that integrating literature-based and data-driven approaches provides a comprehensive and nuanced framework for hypothesis generation and could open new avenues for scientific inquiry.
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.