Authors: Zifeng Wang, Hanyin Wang, Benjamin Danek, Ying Li, Christina Mack, Hoifung Poon, Yajun Wang, Pranav Rajpurkar, Jimeng Sun
Abstract: The integration of Large Language Models (LLMs) into medical applications has sparked widespread interest across the healthcare industry, from drug discovery and development to clinical decision support, assisting telemedicine, medical devices, and healthcare insurance applications. This perspective paper aims to discuss the inner workings of building LLM-powered medical AI applications and introduces a comprehensive framework for their development. We review existing literature and outline the unique challenges of applying LLMs in specialized medical contexts. Additionally, we introduce a three-step framework to organize medical LLM research activities: 1) Modeling: breaking down complex medical workflows into manageable steps for developing medical-specific models; 2) Optimization: optimizing the model performance with crafted prompts and integrating external knowledge and tools, and 3) System engineering: decomposing complex tasks into subtasks and leveraging human expertise for building medical AI applications. Furthermore, we offer a detailed use case playbook that describes various LLM-powered medical AI applications, such as optimizing clinical trial design, enhancing clinical decision support, and advancing medical imaging analysis. Finally, we discuss various challenges and considerations for building medical AI applications with LLMs, such as handling hallucination issues, data ownership and compliance, privacy, intellectual property considerations, compute cost, sustainability issues, and responsible AI requirements.
Authors: Zhehao Zhang, Ryan A. Rossi, Branislav Kveton, Yijia Shao, Diyi Yang, Hamed Zamani, Franck Dernoncourt, Joe Barrow, Tong Yu, Sungchul Kim, Ruiyi Zhang, Jiuxiang Gu, Tyler Derr, Hongjie Chen, Junda Wu, Xiang Chen, Zichao Wang, Subrata Mitra, Nedim Lipka, Nesreen Ahmed, Yu Wang
Abstract: Personalization of Large Language Models (LLMs) has recently become increasingly important with a wide range of applications. Despite the importance and recent progress, most existing works on personalized LLMs have focused either entirely on (a) personalized text generation or (b) leveraging LLMs for personalization-related downstream applications, such as recommendation systems. In this work, we bridge the gap between these two separate main directions for the first time by introducing a taxonomy for personalized LLM usage and summarizing the key differences and challenges. We provide a formalization of the foundations of personalized LLMs that consolidates and expands notions of personalization of LLMs, defining and discussing novel facets of personalization, usage, and desiderata of personalized LLMs. We then unify the literature across these diverse fields and usage scenarios by proposing systematic taxonomies for the granularity of personalization, personalization techniques, datasets, evaluation methods, and applications of personalized LLMs. Finally, we highlight challenges and important open problems that remain to be addressed. By unifying and surveying recent research using the proposed taxonomies, we aim to provide a clear guide to the existing literature and different facets of personalization in LLMs, empowering both researchers and practitioners.
Authors: Zhilun Zhou, Jingyang Fan, Yu Liu, Fengli Xu, Depeng Jin, Yong Li
Abstract: The fast development of location-based social networks (LBSNs) has led to significant changes in society, resulting in popular studies of using LBSN data for socioeconomic prediction, e.g., regional population and commercial activity estimation. Existing studies design various graphs to model heterogeneous LBSN data, and further apply graph representation learning methods for socioeconomic prediction. However, these approaches heavily rely on heuristic ideas and expertise to extract task-relevant knowledge from diverse data, which may not be optimal for specific tasks. Additionally, they tend to overlook the inherent relationships between different indicators, limiting the prediction accuracy. Motivated by the remarkable abilities of large language models (LLMs) in commonsense reasoning, embedding, and multi-agent collaboration, in this work, we synergize LLM agents and knowledge graph for socioeconomic prediction. We first construct a location-based knowledge graph (LBKG) to integrate multi-sourced LBSN data. Then we leverage the reasoning power of LLM agent to identify relevant meta-paths in the LBKG for each type of socioeconomic prediction task, and design a semantic-guided attention module for knowledge fusion with meta-paths. Moreover, we introduce a cross-task communication mechanism to further enhance performance by enabling knowledge sharing across tasks at both LLM agent and KG levels. On the one hand, the LLM agents for different tasks collaborate to generate more diverse and comprehensive meta-paths. On the other hand, the embeddings from different tasks are adaptively merged for better socioeconomic prediction. Experiments on two datasets demonstrate the effectiveness of the synergistic design between LLM and KG, providing insights for information sharing across socioeconomic prediction tasks.
Authors: Donghoon Kim, Gusang Lee, Kyuhong Shim, Byonghyo Shim
Abstract: Recently, we have observed that Large Multi-modal Models (LMMs) are revolutionizing the way machines interact with the world, unlocking new possibilities across various multi-modal applications. To adapt LMMs for downstream tasks, parameter-efficient fine-tuning (PEFT) which only trains additional prefix tokens or modules, has gained popularity. Nevertheless, there has been little analysis of how PEFT works in LMMs. In this paper, we delve into the strengths and weaknesses of each tuning strategy, shifting the focus from the efficiency typically associated with these approaches. We first discover that model parameter tuning methods such as LoRA and Adapters distort the feature representation space learned during pre-training and limit the full utilization of pre-trained knowledge. We also demonstrate that prefix-tuning excels at preserving the representation space, despite its lower performance on downstream tasks. These findings suggest a simple two-step PEFT strategy called Prefix-Tuned PEFT (PT-PEFT), which successively performs prefix-tuning and then PEFT (i.e., Adapter, LoRA), combines the benefits of both. Experimental results show that PT-PEFT not only improves performance in image captioning and visual question answering compared to vanilla PEFT methods but also helps preserve the representation space of the four pre-trained models.
Authors: Danrun Cao (IRISA, EXPRESSION), Nicolas B\'echet (IRISA, UBS, EXPRESSION), Pierre-Fran\c{c}ois Marteau (IRISA, UBS, EXPRESSION)
Abstract: We address in this article the the quality of the WikiNER corpus, a multilingual Named Entity Recognition corpus, and provide a consolidated version of it. The annotation of WikiNER was produced in a semi-supervised manner i.e. no manual verification has been carried out a posteriori. Such corpus is called silver-standard. In this paper we propose WikiNER-fr-gold which is a revised version of the French proportion of WikiNER. Our corpus consists of randomly sampled 20% of the original French sub-corpus (26,818 sentences with 700k tokens). We start by summarizing the entity types included in each category in order to define an annotation guideline, and then we proceed to revise the corpus. Finally we present an analysis of errors and inconsistency observed in the WikiNER-fr corpus, and we discuss potential future work directions.
Authors: Herman Lassche (AFAS Software, University Groningen), Michiel Overeem (AFAS Software), Ayushi Rastogi (University Groningen)
Abstract: Companies support their customers using live chats and chatbots to gain their loyalty. AFAS is a Dutch company aiming to leverage the opportunity large language models (LLMs) offer to answer customer queries with minimal to no input from its customer support team. Adding to its complexity, it is unclear what makes a response correct, and that too in Dutch. Further, with minimal data available for training, the challenge is to identify whether an answer generated by a large language model is correct and do it on the fly. This study is the first to define the correctness of a response based on how the support team at AFAS makes decisions. It leverages literature on natural language generation and automated answer grading systems to automate the decision-making of the customer support team. We investigated questions requiring a binary response (e.g., Would it be possible to adjust tax rates manually?) or instructions (e.g., How would I adjust tax rate manually?) to test how close our automated approach reaches support rating. Our approach can identify wrong messages in 55\% of the cases. This work shows the viability of automatically assessing when our chatbot tell lies.
Authors: Yaran Fan, Jamie Pool, Senja Filipi, Ross Cutler
Abstract: Workplace meetings are vital to organizational collaboration, yet a large percentage of meetings are rated as ineffective. To help improve meeting effectiveness by understanding if the conversation is on topic, we create a comprehensive Topic-Conversation Relevance (TCR) dataset that covers a variety of domains and meeting styles. The TCR dataset includes 1,500 unique meetings, 22 million words in transcripts, and over 15,000 meeting topics, sourced from both newly collected Speech Interruption Meeting (SIM) data and existing public datasets. Along with the text data, we also open source scripts to generate synthetic meetings or create augmented meetings from the TCR dataset to enhance data diversity. For each data source, benchmarks are created using GPT-4 to evaluate the model accuracy in understanding transcription-topic relevance.
Authors: Yulong Wang, Chang Zuo, Yin Xuan, Hong Li, Ni Wei
Abstract: Fine-tuning large language models (LLMs) has become essential for adapting pretrained models to specific downstream tasks. In this paper, we propose Linear Chain Transformation (LinChain), a novel approach that introduces a sequence of linear transformations during fine-tuning to enrich optimization dynamics. By incorporating multiple linear transformations into the parameter update process, LinChain expands the effective rank of updates and enhances the model's ability to learn complex task-specific representations. We demonstrate that this method significantly improves the performance of LLM fine-tuning over state-of-the-art methods by providing more flexible optimization paths during training, while maintaining the inference efficiency of the resulting model. Our experiments on various benchmark tasks show that LinChain leads to better generalization, fewer learnable parameters, and improved task adaptation, making it a compelling strategy for LLM fine-tuning.
Authors: Parvez Zamil, Gollam Rabby, Md. Sadekur Rahman, S\"oren Auer
Abstract: The growing volume of biomedical scholarly document abstracts presents an increasing challenge in efficiently retrieving accurate and relevant information. To address this, we introduce a novel approach that integrates an optimized topic modelling framework, OVB-LDA, with the BI-POP CMA-ES optimization technique for enhanced scholarly document abstract categorization. Complementing this, we employ the distilled MiniLM model, fine-tuned on domain-specific data, for high-precision answer extraction. Our approach is evaluated across three configurations: scholarly document abstract retrieval, gold-standard scholarly documents abstract, and gold-standard snippets, consistently outperforming established methods such as RYGH and bio-answer finder. Notably, we demonstrate that extracting answers from scholarly documents abstracts alone can yield high accuracy, underscoring the sufficiency of abstracts for many biomedical queries. Despite its compact size, MiniLM exhibits competitive performance, challenging the prevailing notion that only large, resource-intensive models can handle such complex tasks. Our results, validated across various question types and evaluation batches, highlight the robustness and adaptability of our method in real-world biomedical applications. While our approach shows promise, we identify challenges in handling complex list-type questions and inconsistencies in evaluation metrics. Future work will focus on refining the topic model with more extensive domain-specific datasets, further optimizing MiniLM and utilizing large language models (LLM) to improve both precision and efficiency in biomedical question answering.
Authors: Amogh Akella
Abstract: In this paper, we explore how to optimize the usage of Large-Language Models to quickly and accurately solve mathematical problems. In particular, we show the effectiveness of using the classification of problems into different categories to facilitate problem-solving. Additionally, we optimize the classification of problems into categories by creating an accurate dataset. We believe that our technique for problem-solving works by helping mitigate hallucination in LLMs which is key to unlocking their ability to solve math problems.
Authors: B. D. Lam, S. Ma, I. Kovalenko, P. Wang, O. Jafari, A. Li, S. Horng
Abstract: Pulmonary embolism (PE) is a leading cause of preventable in-hospital mortality. Advances in diagnosis, risk stratification, and prevention can improve outcomes. There are few large publicly available datasets that contain PE labels for research. Using the MIMIC-IV database, we extracted all available radiology reports of computed tomography pulmonary angiography (CTPA) scans and two physicians manually labeled the results as PE positive (acute PE) or PE negative. We then applied a previously finetuned Bio_ClinicalBERT transformer language model, VTE-BERT, to extract labels automatically. We verified VTE-BERT's reliability by measuring its performance against manual adjudication. We also compared the performance of VTE-BERT to diagnosis codes. We found that VTE-BERT has a sensitivity of 92.4% and positive predictive value (PPV) of 87.8% on all 19,942 patients with CTPA radiology reports from the emergency room and/or hospital admission. In contrast, diagnosis codes have a sensitivity of 95.4% and PPV of 83.8% on the subset of 11,990 hospitalized patients with discharge diagnosis codes. We successfully add nearly 20,000 labels to CTPAs in a publicly available dataset and demonstrate the external validity of a semi-supervised language model in accelerating hematologic research.
Authors: Elena Kardanova (National Research University Higher School of Economics, Moscow, Russia), Alina Ivanova (National Research University Higher School of Economics, Moscow, Russia), Ksenia Tarasova (National Research University Higher School of Economics, Moscow, Russia), Taras Pashchenko (National Research University Higher School of Economics, Moscow, Russia), Aleksei Tikhoniuk (National Research University Higher School of Economics, Moscow, Russia), Elen Yusupova (National Research University Higher School of Economics, Moscow, Russia), Anatoly Kasprzhak (National Research University Higher School of Economics, Moscow, Russia), Yaroslav Kuzminov (National Research University Higher School of Economics, Moscow, Russia), Ekaterina Kruchinskaia (National Research University Higher School of Economics, Moscow, Russia), Irina Brun (National Research University Higher School of Economics, Moscow, Russia)
Abstract: The era of large language models (LLM) raises questions not only about how to train models, but also about how to evaluate them. Despite numerous existing benchmarks, insufficient attention is often given to creating assessments that test LLMs in a valid and reliable manner. To address this challenge, we accommodate the Evidence-centered design (ECD) methodology and propose a comprehensive approach to benchmark development based on rigorous psychometric principles. In this paper, we have made the first attempt to illustrate this approach by creating a new benchmark in the field of pedagogy and education, highlighting the limitations of existing benchmark development approach and taking into account the development of LLMs. We conclude that a new approach to benchmarking is required to match the growing complexity of AI applications in the educational context. We construct a novel benchmark guided by the Bloom's taxonomy and rigorously designed by a consortium of education experts trained in test development. Thus the current benchmark provides an academically robust and practical assessment tool tailored for LLMs, rather than human participants. Tested empirically on the GPT model in the Russian language, it evaluates model performance across varied task complexities, revealing critical gaps in current LLM capabilities. Our results indicate that while generative AI tools hold significant promise for education - potentially supporting tasks such as personalized tutoring, real-time feedback, and multilingual learning - their reliability as autonomous teachers' assistants right now remain rather limited, particularly in tasks requiring deeper cognitive engagement.
Authors: Harry Caufield, Carlo Kroll, Shawn T O'Neil, Justin T Reese, Marcin P Joachimiak, Harshad Hegde, Nomi L Harris, Madan Krishnamurthy, James A McLaughlin, Damian Smedley, Melissa A Haendel, Peter N Robinson, Christopher J Mungall
Abstract: Effective data-driven biomedical discovery requires data curation: a time-consuming process of finding, organizing, distilling, integrating, interpreting, annotating, and validating diverse information into a structured form suitable for databases and knowledge bases. Accurate and efficient curation of these digital assets is critical to ensuring that they are FAIR, trustworthy, and sustainable. Unfortunately, expert curators face significant time and resource constraints. The rapid pace of new information being published daily is exceeding their capacity for curation. Generative AI, exemplified by instruction-tuned large language models (LLMs), has opened up new possibilities for assisting human-driven curation. The design philosophy of agents combines the emerging abilities of generative AI with more precise methods. A curator's tasks can be aided by agents for performing reasoning, searching ontologies, and integrating knowledge across external sources, all efforts otherwise requiring extensive manual effort. Our LLM-driven annotation tool, CurateGPT, melds the power of generative AI together with trusted knowledge bases and literature sources. CurateGPT streamlines the curation process, enhancing collaboration and efficiency in common workflows. Compared to direct interaction with an LLM, CurateGPT's agents enable access to information beyond that in the LLM's training data and they provide direct links to the data supporting each claim. This helps curators, researchers, and engineers scale up curation efforts to keep pace with the ever-increasing volume of scientific data.
Authors: Albert N\"ossig, Tobias Hell, Georg Moser
Abstract: In this paper, we present an innovative iterative approach to rule learning specifically designed for (but not limited to) text-based data. Our method focuses on progressively expanding the vocabulary utilized in each iteration resulting in a significant reduction of memory consumption. Moreover, we introduce a Value of Confidence as an indicator of the reliability of the generated rules. By leveraging the Value of Confidence, our approach ensures that only the most robust and trustworthy rules are retained, thereby improving the overall quality of the rule learning process. We demonstrate the effectiveness of our method through extensive experiments on various textual as well as non-textual datasets including a use case of significant interest to insurance industries, showcasing its potential for real-world applications.
Authors: Ahmed Akib Jawad Karim, Kazi Hafiz Md. Asad, Md. Golam Rabiul Alam
Abstract: This work focuses on the efficiency of the knowledge distillation approach in generating a lightweight yet powerful BERT based model for natural language processing applications. After the model creation, we applied the resulting model, LastBERT, to a real-world task classifying severity levels of Attention Deficit Hyperactivity Disorder (ADHD)-related concerns from social media text data. Referring to LastBERT, a customized student BERT model, we significantly lowered model parameters from 110 million BERT base to 29 million, resulting in a model approximately 73.64% smaller. On the GLUE benchmark, comprising paraphrase identification, sentiment analysis, and text classification, the student model maintained strong performance across many tasks despite this reduction. The model was also used on a real-world ADHD dataset with an accuracy and F1 score of 85%. When compared to DistilBERT (66M) and ClinicalBERT (110M), LastBERT demonstrated comparable performance, with DistilBERT slightly outperforming it at 87%, and ClinicalBERT achieving 86% across the same metrics. These findings highlight the LastBERT model's capacity to classify degrees of ADHD severity properly, so it offers a useful tool for mental health professionals to assess and comprehend material produced by users on social networking platforms. The study emphasizes the possibilities of knowledge distillation to produce effective models fit for use in resource-limited conditions, hence advancing NLP and mental health diagnosis. Furthermore underlined by the considerable decrease in model size without appreciable performance loss is the lower computational resources needed for training and deployment, hence facilitating greater applicability. Especially using readily available computational tools like Google Colab. This study shows the accessibility and usefulness of advanced NLP methods in pragmatic world applications.
Authors: Andrew Estornell, Jean-Francois Ton, Yuanshun Yao, Yang Liu
Abstract: Large language models (LLMs) have demonstrated a remarkable ability to serve as general-purpose tools for various language-based tasks. Recent works have demonstrated that the efficacy of such models can be improved through iterative dialog between multiple models, frequently referred to as multi-agent debate (MAD). While debate shows promise as a means of improving model efficacy, most works in this area treat debate as an emergent behavior, rather than a learned behavior. In doing so, current debate frameworks rely on collaborative behaviors to have been sufficiently trained into off-the-shelf models. To address this limitation, we propose ACC-Debate, an Actor-Critic based learning framework to produce a two-agent team specialized in debate. We demonstrate that ACC-Debate outperforms SotA debate techniques on a wide array of benchmarks.
Authors: Darian Rodriguez Vasquez, Afroditi Papadaki
Abstract: Despite the impressive performance of large language models across various tasks, they often struggle with reasoning under negated statements. Negations are important in real-world applications as they encode negative polarity in verb phrases, clauses, or other expressions. Nevertheless, they are underrepresented in current benchmarks, which mainly include basic negation forms and overlook more complex ones, resulting in insufficient data for training a language model. In this work, we propose NegVerse, a method that tackles the lack of negation datasets by producing a diverse range of negation types from affirmative sentences, including verbal, non-verbal, and affixal forms commonly found in English text. We provide new rules for masking parts of sentences where negations are most likely to occur, based on syntactic structure and use a frozen baseline LLM and prompt tuning to generate negated sentences. We also propose a filtering mechanism to identify negation cues and remove degenerate examples, producing a diverse range of meaningful perturbations. Our results show that NegVerse outperforms existing methods and generates negations with higher lexical similarity to the original sentences, better syntactic preservation and negation diversity. The code is available in https://github.com/DarianRodriguez/NegVerse
Authors: Ziyu Ye, Rishabh Agarwal, Tianqi Liu, Rishabh Joshi, Sarmishta Velury, Quoc V. Le, Qijun Tan, Yuan Liu
Abstract: Current RLHF frameworks for aligning large language models (LLMs) typically assume a fixed prompt distribution, which is sub-optimal and limits the scalability of alignment and generalizability of models. To address this, we introduce a general open-ended RLHF framework that casts alignment as an asymmetric game between two players: (i) a creator that generates increasingly informative prompt distributions using the reward model, and (ii) a solver that learns to produce more preferred responses on prompts produced by the creator. This framework of Evolving Alignment via Asymmetric Self-Play (eva), results in a simple and efficient approach that can utilize any existing RLHF algorithm for scalable alignment. eva outperforms state-of-the-art methods on widely-used benchmarks, without the need of any additional human crafted prompts. Specifically, eva improves the win rate of Gemma-2-9B-it on Arena-Hard from 51.6% to 60.1% with DPO, from 55.7% to 58.9% with SPPO, from 52.3% to 60.7% with SimPO, and from 54.8% to 60.3% with ORPO, surpassing its 27B version and matching claude-3-opus. This improvement is persistent even when new human crafted prompts are introduced. Finally, we show eva is effective and robust under various ablation settings.
Authors: Eunji Kim, Sriya Mantena, Weiwei Yang, Chandan Singh, Sungroh Yoon, Jianfeng Gao
Abstract: Recent large language models (LLMs) have excelled across a wide range of tasks, but their use in high-stakes and compute-limited settings has intensified the demand for interpretability and efficiency. We address this need by proposing Induction-head ngram models (Induction-Gram), a method that builds an efficient, interpretable LM by bolstering modern ngram models with a hand-engineered "induction head". This induction head uses a custom neural similarity metric to efficiently search the model's input context for potential next-word completions. This process enables Induction-Gram to provide ngram-level grounding for each generated token. Moreover, experiments show that this simple method significantly improves next-word prediction over baseline interpretable models (up to 26%p) and can be used to speed up LLM inference for large models through speculative decoding. We further study Induction-Gram in a natural-language neuroscience setting, where the goal is to predict the next fMRI response in a sequence. It again provides a significant improvement over interpretable models (20% relative increase in the correlation of predicted fMRI responses), potentially enabling deeper scientific investigation of language selectivity in the brain. The code is available at https://github.com/ejkim47/induction-gram.
Authors: Zhenbiao Cao, Yuanlei Zheng, Zhihao Fan, Xiaojin Zhang, Wei Chen
Abstract: Text-to-SQL generation aims to translate natural language questions into SQL statements. In large language models (LLMs) based Text-to-SQL, schema linking is a widely adopted strategy to streamline the input for LLMs by selecting only relevant schema elements, therefore reducing noise and computational overhead. However, schema linking faces risks that requires caution, including the potential omission of necessary elements and disruption of database structural integrity. To address these challenges, we propose a novel framework called RSL-SQL that combines bidirectional schema linking, contextual information augmentation, binary selection strategy, and multi-turn self-correction. Our approach improves the recall of schema linking through forward and backward pruning and hedges the risk by voting between full schema and contextual information augmented simplified schema. Experiments on the BIRD and Spider benchmarks demonstrate that our approach achieves state-of-the-art execution accuracy among open-source solutions, with 67.2% on BIRD and 87.9% on Spider using GPT-4o. Furthermore, our approach outperforms a series of GPT-4 based Text-to-SQL systems when adopting DeepSeek (much cheaper) with same intact prompts. Extensive analysis and ablation studies confirm the effectiveness of each component in our framework. The codes are available at https://github.com/Laqcce-cao/RSL-SQL.
Authors: Tong Niu, Shafiq Joty, Ye Liu, Caiming Xiong, Yingbo Zhou, Semih Yavuz
Abstract: Accurate document retrieval is crucial for the success of retrieval-augmented generation (RAG) applications, including open-domain question answering and code completion. While large language models (LLMs) have been employed as dense encoders or listwise rerankers in RAG systems, they often struggle with reasoning-intensive tasks because they lack nuanced analysis when judging document relevance. To address this limitation, we introduce JudgeRank, a novel agentic reranker that emulates human cognitive processes when assessing document relevance. Our approach consists of three key steps: (1) query analysis to identify the core problem, (2) document analysis to extract a query-aware summary, and (3) relevance judgment to provide a concise assessment of document relevance. We evaluate JudgeRank on the reasoning-intensive BRIGHT benchmark, demonstrating substantial performance improvements over first-stage retrieval methods and outperforming other popular reranking approaches. In addition, JudgeRank performs on par with fine-tuned state-of-the-art rerankers on the popular BEIR benchmark, validating its zero-shot generalization capability. Through comprehensive ablation studies, we demonstrate that JudgeRank's performance generalizes well across LLMs of various sizes while ensembling them yields even more accurate reranking than individual models.
Authors: Christopher Richardson, Roshan Sharma, Neeraj Gaur, Parisa Haghani, Anirudh Sundar, Bhuvana Ramabhadran
Abstract: Zero-shot domain adaptation for dialogue state tracking (DST) remains a challenging problem in task-oriented dialogue (TOD) systems, where models must generalize to target domains unseen at training time. Current large language model approaches for zero-shot domain adaptation rely on prompting to introduce knowledge pertaining to the target domains. However, their efficacy strongly depends on prompt engineering, as well as the zero-shot ability of the underlying language model. In this work, we devise a novel data augmentation approach, Schema Augmentation, that improves the zero-shot domain adaptation of language models through fine-tuning. Schema Augmentation is a simple but effective technique that enhances generalization by introducing variations of slot names within the schema provided in the prompt. Experiments on MultiWOZ and SpokenWOZ showed that the proposed approach resulted in a substantial improvement over the baseline, in some experiments achieving over a twofold accuracy gain over unseen domains while maintaining equal or superior performance over all domains.
Authors: Haritz Puerto, Martin Gubri, Sangdoo Yun, Seong Joon Oh
Abstract: Membership inference attacks (MIA) attempt to verify the membership of a given data sample in the training set for a model. MIA has become relevant in recent years, following the rapid development of large language models (LLM). Many are concerned about the usage of copyrighted materials for training them and call for methods for detecting such usage. However, recent research has largely concluded that current MIA methods do not work on LLMs. Even when they seem to work, it is usually because of the ill-designed experimental setup where other shortcut features enable "cheating." In this work, we argue that MIA still works on LLMs, but only when multiple documents are presented for testing. We construct new benchmarks that measure the MIA performances at a continuous scale of data samples, from sentences (n-grams) to a collection of documents (multiple chunks of tokens). To validate the efficacy of current MIA approaches at greater scales, we adapt a recent work on Dataset Inference (DI) for the task of binary membership detection that aggregates paragraph-level MIA features to enable MIA at document and collection of documents level. This baseline achieves the first successful MIA on pre-trained and fine-tuned LLMs.
Authors: John Wu, David Wu, Jimeng Sun
Abstract: Medical coding, the translation of unstructured clinical text into standardized medical codes, is a crucial but time-consuming healthcare practice. Though large language models (LLM) could automate the coding process and improve the efficiency of such tasks, interpretability remains paramount for maintaining patient trust. Current efforts in interpretability of medical coding applications rely heavily on label attention mechanisms, which often leads to the highlighting of extraneous tokens irrelevant to the ICD code. To facilitate accurate interpretability in medical language models, this paper leverages dictionary learning that can efficiently extract sparsely activated representations from dense language model embeddings in superposition. Compared with common label attention mechanisms, our model goes beyond token-level representations by building an interpretable dictionary which enhances the mechanistic-based explanations for each ICD code prediction, even when the highlighted tokens are medically irrelevant. We show that dictionary features can steer model behavior, elucidate the hidden meanings of upwards of 90% of medically irrelevant tokens, and are human interpretable.
Authors: Keivan Rezaei, Khyathi Chandu, Soheil Feizi, Yejin Choi, Faeze Brahman, Abhilasha Ravichander
Abstract: Large language models trained on web-scale corpora can memorize undesirable datapoints such as incorrect facts, copyrighted content or sensitive data. Recently, many machine unlearning methods have been proposed that aim to 'erase' these datapoints from trained models -- that is, revert model behavior to be similar to a model that had never been trained on these datapoints. However, evaluating the success of unlearning algorithms remains challenging. In this work, we propose the RESTOR framework for machine unlearning based on the following dimensions: (1) a task setting that focuses on real-world factual knowledge, (2) a variety of corruption scenarios that emulate different kinds of datapoints that might need to be unlearned, and (3) evaluation metrics that emphasize not just forgetting undesirable knowledge, but also recovering the model's original state before encountering these datapoints, or restorative unlearning. RESTOR helps uncover several novel insights about popular unlearning algorithms, and the mechanisms through which they operate -- for instance, identifying that some algorithms merely emphasize forgetting the knowledge to be unlearned, and that localizing unlearning targets can enhance unlearning performance. Code/data is available at github.com/k1rezaei/restor.
Authors: Yubin Kim, Chanwoo Park, Hyewon Jeong, Cristina Grau-Vilchez, Yik Siu Chan, Xuhai Xu, Daniel McDuff, Hyeonhoon Lee, Marzyeh Ghassemi, Cynthia Breazeal, Hae Won Park
Abstract: Medical Decision-Making (MDM) is a multi-faceted process that requires clinicians to assess complex multi-modal patient data patient, often collaboratively. Large Language Models (LLMs) promise to streamline this process by synthesizing vast medical knowledge and multi-modal health data. However, single-agent are often ill-suited for nuanced medical contexts requiring adaptable, collaborative problem-solving. Our MDAgents addresses this need by dynamically assigning collaboration structures to LLMs based on task complexity, mimicking real-world clinical collaboration and decision-making. This framework improves diagnostic accuracy and supports adaptive responses in complex, real-world medical scenarios, making it a valuable tool for clinicians in various healthcare settings, and at the same time, being more efficient in terms of computing cost than static multi-agent decision making methods.
Authors: Kazi Ahmed Asif Fuad, Lizhong Chen
Abstract: Large Language Models (LLMs) excel in data synthesis but can be inaccurate in domain-specific tasks, which retrieval-augmented generation (RAG) systems address by leveraging user-provided data. However, RAGs require optimization in both retrieval and generation stages, which can affect output quality. In this paper, we present LLM-Ref, a writing assistant tool that aids researchers in writing articles from multiple source documents with enhanced reference synthesis and handling capabilities. Unlike traditional RAG systems that use chunking and indexing, our tool retrieves and generates content directly from text paragraphs. This method facilitates direct reference extraction from the generated outputs, a feature unique to our tool. Additionally, our tool employs iterative response generation, effectively managing lengthy contexts within the language model's constraints. Compared to baseline RAG-based systems, our approach achieves a $3.25\times$ to $6.26\times$ increase in Ragas score, a comprehensive metric that provides a holistic view of a RAG system's ability to produce accurate, relevant, and contextually appropriate responses. This improvement shows our method enhances the accuracy and contextual relevance of writing assistance tools.
Authors: Jiwoong Sohn, Yein Park, Chanwoong Yoon, Sihyeon Park, Hyeon Hwang, Mujeen Sung, Hyunjae Kim, Jaewoo Kang
Abstract: Large language models (LLM) hold significant potential for applications in biomedicine, but they struggle with hallucinations and outdated knowledge. While retrieval-augmented generation (RAG) is generally employed to address these issues, it also has its own set of challenges: (1) LLMs are vulnerable to irrelevant or incorrect context, (2) medical queries are often not well-targeted for helpful information, and (3) retrievers are prone to bias toward the specific source corpus they were trained on. In this study, we present RAG$^2$ (RAtionale-Guided RAG), a new framework for enhancing the reliability of RAG in biomedical contexts. RAG$^2$ incorporates three key innovations: a small filtering model trained on perplexity-based labels of rationales, which selectively augments informative snippets of documents while filtering out distractors; LLM-generated rationales as queries to improve the utility of retrieved snippets; a structure designed to retrieve snippets evenly from a comprehensive set of four biomedical corpora, effectively mitigating retriever bias. Our experiments demonstrate that RAG$^2$ improves the state-of-the-art LLMs of varying sizes, with improvements of up to 6.1\%, and it outperforms the previous best medical RAG model by up to 5.6\% across three medical question-answering benchmarks. Our code is available at https://github.com/dmis-lab/RAG2.
Authors: Sajad Sotudeh, Nazli Goharian
Abstract: This study examines the potential of integrating Learning-to-Rank (LTR) with Query-focused Summarization (QFS) to enhance the summary relevance via content prioritization. Using a shared secondary decoder with the summarization decoder, we carry out the LTR task at the segment level. Compared to the state-of-the-art, our model outperforms on QMSum benchmark (all metrics) and matches on SQuALITY benchmark (2 metrics) as measured by Rouge and BertScore while offering a lower training overhead. Specifically, on the QMSum benchmark, our proposed system achieves improvements, particularly in Rouge-L (+0.42) and BertScore (+0.34), indicating enhanced understanding and relevance. While facing minor challenges in Rouge-1 and Rouge-2 scores on the SQuALITY benchmark, the model significantly excels in Rouge-L (+1.47), underscoring its capability to generate coherent summaries. Human evaluations emphasize the efficacy of our method in terms of relevance and faithfulness of the generated summaries, without sacrificing fluency. A deeper analysis reveals our model's superiority over the state-of-the-art for broad queries, as opposed to specific ones, from a qualitative standpoint. We further present an error analysis of our model, pinpointing challenges faced and suggesting potential directions for future research in this field.
Authors: Anish Pahilajani, Devasha Trivedi, Jincen Shuai, Khin S. Yone, Samyak Rajesh Jain, Namyong Park, Ryan A. Rossi, Nesreen K. Ahmed, Franck Dernoncourt, Yu Wang
Abstract: Large Language Models (LLMs) have excelled in multi-hop question-answering (M-QA) due to their advanced reasoning abilities. However, the impact of the inherent reasoning structures on LLM M-QA performance remains unclear, largely due to the absence of QA datasets that provide fine-grained reasoning structures. To address this gap, we introduce the Graph Reasoning-Structured Question Answering Dataset (GRS-QA), which includes both semantic contexts and reasoning structures for QA pairs. Unlike existing M-QA datasets, where different reasoning structures are entangled together, GRS-QA explicitly captures intricate reasoning pathways by constructing reasoning graphs, where nodes represent textual contexts and edges denote logical flows. These reasoning graphs of different structures enable a fine-grained evaluation of LLM reasoning capabilities across various reasoning structures. Our empirical analysis reveals that LLMs perform differently when handling questions with varying reasoning structures. This finding facilitates the exploration of textual structures as compared with semantics.
Authors: Jiaru Zou, Qing Wang, Pratyush Thakur, Nickvash Kani
Abstract: Advances in large language models (LLMs) have spurred research into enhancing their reasoning capabilities, particularly in math-rich STEM documents. While LLMs can generate equations or solve math-related queries, their ability to fully understand and interpret abstract mathematical symbols in long, math-rich documents remains limited. In this paper, we introduce STEM-PoM, a comprehensive benchmark dataset designed to evaluate LLMs' reasoning abilities on math symbols within contextual scientific text. The dataset, sourced from real-world ArXiv documents, contains over 2K math symbols classified as main attributes of variables, constants, operators, and unit descriptors, with additional sub-attributes including scalar/vector/matrix for variables and local/global/discipline-specific labels for both constants and operators. Our extensive experiments show that state-of-the-art LLMs achieve an average of 20-60% accuracy under in-context learning and 50-60% accuracy with fine-tuning, revealing a significant gap in their mathematical reasoning capabilities. STEM-PoM fuels future research of developing advanced Math-AI models that can robustly handle math symbols.
Authors: David Anugraha, Garry Kuwanto, Lucky Susanto, Derry Tanti Wijaya, Genta Indra Winata
Abstract: We present MetaMetrics-MT, an innovative metric designed to evaluate machine translation (MT) tasks by aligning closely with human preferences through Bayesian optimization with Gaussian Processes. MetaMetrics-MT enhances existing MT metrics by optimizing their correlation with human judgments. Our experiments on the WMT24 metric shared task dataset demonstrate that MetaMetrics-MT outperforms all existing baselines, setting a new benchmark for state-of-the-art performance in the reference-based setting. Furthermore, it achieves comparable results to leading metrics in the reference-free setting, offering greater efficiency.
Authors: Arjun Ramesh Kaushik, Sunil Rufus R P, Nalini Ratha
Abstract: The increasing prevalence of AI-generated content alongside human-written text underscores the need for reliable discrimination methods. To address this challenge, we propose a novel framework with textual embeddings from Pre-trained Language Models (PLMs) to distinguish AI-generated and human-authored text. Our approach utilizes Embedding Fusion to integrate semantic information from multiple Language Models, harnessing their complementary strengths to enhance performance. Through extensive evaluation across publicly available diverse datasets, our proposed approach demonstrates strong performance, achieving classification accuracy greater than 96% and a Matthews Correlation Coefficient (MCC) greater than 0.93. This evaluation is conducted on a balanced dataset of texts generated from five well-known Large Language Models (LLMs), highlighting the effectiveness and robustness of our novel methodology.
Authors: Chenghua Huang, Zhizhen Fan, Lu Wang, Fangkai Yang, Pu Zhao, Zeqi Lin, Qingwei Lin, Dongmei Zhang, Saravan Rajmohan, Qi Zhang
Abstract: Reinforcement Learning from Human Feedback (RLHF) is a crucial technique for aligning language models with human preferences, playing a pivotal role in the success of conversational models like GPT-4, ChatGPT, and Llama 2. A core challenge in employing RLHF lies in training a reliable reward model (RM), which relies on high-quality labels typically provided by human experts or advanced AI system. These methods can be costly and may introduce biases that affect the language model's responses. As language models improve, human input may become less effective in further enhancing their performance. In this paper, we propose Self-Evolved Reward Learning (SER), a novel approach where the RM generates additional training data to iteratively improve itself. We conducted extensive experiments on multiple datasets such as HH-RLHF and UltraFeedback, using models like Mistral and Llama 3, and compare SER against various baselines. Our results demonstrate that even with limited human-annotated data, learning from self-feedback can robustly enhance RM performance, thereby boosting the capabilities of large language models (LLMs).
Authors: Aman Gupta, Anirudh Ravichandran, Ziji Zhang, Swair Shah, Anurag Beniwal, Narayanan Sadagopan
Abstract: Task-oriented dialogue systems are essential for applications ranging from customer service to personal assistants and are widely used across various industries. However, developing effective multi-domain systems remains a significant challenge due to the complexity of handling diverse user intents, entity types, and domain-specific knowledge across several domains. In this work, we propose DARD (Domain Assigned Response Delegation), a multi-agent conversational system capable of successfully handling multi-domain dialogs. DARD leverages domain-specific agents, orchestrated by a central dialog manager agent. Our extensive experiments compare and utilize various agent modeling approaches, combining the strengths of smaller fine-tuned models (Flan-T5-large & Mistral-7B) with their larger counterparts, Large Language Models (LLMs) (Claude Sonnet 3.0). We provide insights into the strengths and limitations of each approach, highlighting the benefits of our multi-agent framework in terms of flexibility and composability. We evaluate DARD using the well-established MultiWOZ benchmark, achieving state-of-the-art performance by improving the dialogue inform rate by 6.6% and the success rate by 4.1% over the best-performing existing approaches. Additionally, we discuss various annotator discrepancies and issues within the MultiWOZ dataset and its evaluation system.
Authors: Yun Jiang, Zilong Xie, Wei Zhang, Yun Fang, Shuai Pan
Abstract: Retrieval-augmented generation methods often neglect the quality of content retrieved from external knowledge bases, resulting in irrelevant information or potential misinformation that negatively affects the generation results of large language models. In this paper, we propose an end-to-end model with adaptive filtering for retrieval-augmented generation (E2E-AFG), which integrates answer existence judgment and text generation into a single end-to-end framework. This enables the model to focus more effectively on relevant content while reducing the influence of irrelevant information and generating accurate answers. We evaluate E2E-AFG on six representative knowledge-intensive language datasets, and the results show that it consistently outperforms baseline models across all tasks, demonstrating the effectiveness and robustness of the proposed approach.
Authors: Yang Janet Liu, Tatsuya Aoyama, Wesley Scivetti, Yilun Zhu, Shabnam Behzad, Lauren Elizabeth Levine, Jessica Lin, Devika Tiwari, Amir Zeldes
Abstract: Work on shallow discourse parsing in English has focused on the Wall Street Journal corpus, the only large-scale dataset for the language in the PDTB framework. However, the data is not openly available, is restricted to the news domain, and is by now 35 years old. In this paper, we present and evaluate a new open-access, multi-genre benchmark for PDTB-style shallow discourse parsing, based on the existing UD English GUM corpus, for which discourse relation annotations in other frameworks already exist. In a series of experiments on cross-domain relation classification, we show that while our dataset is compatible with PDTB, substantial out-of-domain degradation is observed, which can be alleviated by joint training on both datasets.
Authors: Do Xuan Long, Duong Ngoc Yen, Anh Tuan Luu, Kenji Kawaguchi, Min-Yen Kan, Nancy F. Chen
Abstract: We present Multi-expert Prompting, a novel enhancement of ExpertPrompting (Xu et al., 2023), designed to improve the large language model (LLM) generation. Specifically, it guides an LLM to fulfill an input instruction by simulating multiple experts, aggregating their responses, and selecting the best among individual and aggregated responses. This process is performed in a single chain of thoughts through our seven carefully designed subtasks derived from the Nominal Group Technique (Ven and Delbecq, 1974), a well-established decision-making framework. Our evaluations demonstrate that Multi-expert Prompting significantly outperforms ExpertPrompting and comparable baselines in enhancing the truthfulness, factuality, informativeness, and usefulness of responses while reducing toxicity and hurtfulness. It further achieves state-of-the-art truthfulness by outperforming the best baseline by 8.69% with ChatGPT. Multi-expert Prompting is efficient, explainable, and highly adaptable to diverse scenarios, eliminating the need for manual prompt construction.
Authors: Anbang Wang
Abstract: This paper presents ReverseNER, a framework aimed at overcoming the limitations of large language models (LLMs) in zero-shot Named Entity Recognition (NER) tasks, particularly in cases where certain entity types have ambiguous boundaries. ReverseNER tackles this challenge by constructing a reliable example library with the reversed process of NER. Rather than beginning with sentences, this method uses an LLM to generate entities based on their definitions and then expands them into full sentences. During sentence generation, the LLM is guided to replicate the structure of a specific 'feature sentence', extracted from the task sentences by clustering. This results in well-annotated sentences with clearly labeled entities, while preserving semantic and structural similarity to the task sentences. Once the example library is constructed, the method selects the most semantically similar example labels for each task sentence to support the LLM's inference. We also propose an entity-level self-consistency scoring mechanism to improve NER performance with LLMs. Experiments show that ReverseNER significantly outperforms traditional zero-shot NER with LLMs and surpasses several few-shot methods, marking a notable improvement in NER for domains with limited labeled data.
Authors: Zhili Feng, Tanya Marwah, Lester Mackey, David Alvarez-Melis, Nicolo Fusi
Abstract: Modern large language models use a fixed tokenizer to effectively compress text drawn from a source domain. However, applying the same tokenizer to a new target domain often leads to inferior compression, more costly inference, and reduced semantic alignment. To address this deficiency, we introduce Sparse Sinkhorn Token Translation (S2T2). S2T2 trains a tailored tokenizer for the target domain and learns to translate between target and source tokens, enabling more effective reuse of the pre-trained next-source-token predictor. In our experiments with finetuned English language models, S2T2 improves both the perplexity and the compression of out-of-domain protein sequences, outperforming direct finetuning with either the source or target tokenizer. In addition, we find that token translations learned for smaller, less expensive models can be directly transferred to larger, more powerful models to reap the benefits of S2T2 at lower cost.
Authors: Po-Chuan Chen, Mahdin Rohmatillah, You-Teng Lin, Jen-Tzung Chien
Abstract: Student mental health is a sensitive issue that necessitates special attention. A primary concern is the student-to-counselor ratio, which surpasses the recommended standard of 250:1 in most universities. This imbalance results in extended waiting periods for in-person consultations, which cause suboptimal treatment. Significant efforts have been directed toward developing mental health dialogue systems utilizing the existing open-source mental health-related datasets. However, currently available datasets either discuss general topics or various strategies that may not be viable for direct application due to numerous ethical constraints inherent in this research domain. To address this issue, this paper introduces a specialized mental health dataset that emphasizes the active listening strategy employed in conversation for counseling, also named as ConvCounsel. This dataset comprises both speech and text data, which can facilitate the development of a reliable pipeline for mental health dialogue systems. To demonstrate the utility of the proposed dataset, this paper also presents the NYCUKA, a spoken mental health dialogue system that is designed by using the ConvCounsel dataset. The results show the merit of using this dataset.
Authors: Houjing Wei, Hakaze Cho, Yuting Shi, Naoya Inoue
Abstract: Vision Large Language Models (VLLMs) usually take input as a concatenation of image token embeddings and text token embeddings and conduct causal modeling. However, their internal behaviors remain underexplored, raising the question of interaction among two types of tokens. To investigate such multimodal interaction during model inference, in this paper, we measure the contextualization among the hidden state vectors of tokens from different modalities. Our experiments uncover a four-phase inference dynamics of VLLMs against the depth of Transformer-based LMs, including (I) Alignment: In very early layers, contextualization emerges between modalities, suggesting a feature space alignment. (II) Intra-modal Encoding: In early layers, intra-modal contextualization is enhanced while inter-modal interaction is suppressed, suggesting a local encoding within modalities. (III) Inter-modal Encoding: In later layers, contextualization across modalities is enhanced, suggesting a deeper fusion across modalities. (IV) Output Preparation: In very late layers, contextualization is reduced globally, and hidden states are aligned towards the unembedding space.
Authors: Sho Yokoi, Han Bao, Hiroto Kurita, Hidetoshi Shimodaira
Abstract: The word embedding space in neural models is skewed, and correcting this can improve task performance. We point out that most approaches for modeling, correcting, and measuring the symmetry of an embedding space implicitly assume that the word frequencies are uniform; in reality, word frequencies follow a highly non-uniform distribution, known as Zipf's law. Surprisingly, simply performing PCA whitening weighted by the empirical word frequency that follows Zipf's law significantly improves task performance, surpassing established baselines. From a theoretical perspective, both our approach and existing methods can be clearly categorized: word representations are distributed according to an exponential family with either uniform or Zipfian base measures. By adopting the latter approach, we can naturally emphasize informative low-frequency words in terms of their vector norm, which becomes evident from the information-geometric perspective, and in terms of the loss functions for imbalanced classification. Additionally, our theory corroborates that popular natural language processing methods, such as skip-gram negative sampling, WhiteningBERT, and headless language models, work well just because their word embeddings encode the empirical word frequency into the underlying probabilistic model.
Authors: Minki Kang, Sung Ju Hwang, Gibbeum Lee, Jaewoong Cho
Abstract: As Large Language Models (LLMs) are increasingly deployed in specialized domains with continuously evolving knowledge, the need for timely and precise knowledge injection has become essential. Fine-tuning with paraphrased data is a common approach to enhance knowledge injection, yet it faces two significant challenges: high computational costs due to repetitive external model usage and limited sample diversity. To this end, we introduce LaPael, a latent-level paraphrasing method that applies input-dependent noise to early LLM layers. This approach enables diverse and semantically consistent augmentations directly within the model. Furthermore, it eliminates the recurring costs of paraphrase generation for each knowledge update. Our extensive experiments on question-answering benchmarks demonstrate that LaPael improves knowledge injection over standard fine-tuning and existing noise-based approaches. Additionally, combining LaPael with data-level paraphrasing further enhances performance.
Authors: Qingfei Zhao, Ruobing Wang, Xin Wang, Daren Zha, Nan Mu
Abstract: Retrieval-Augmented Generation (RAG) has emerged as a reliable external knowledge augmentation technique to mitigate hallucination issues and parameterized knowledge limitations in Large Language Models (LLMs). Existing Adaptive RAG (ARAG) systems struggle to effectively explore multiple retrieval sources due to their inability to select the right source at the right time. To address this, we propose a multi-source ARAG framework, termed MSPR, which synergizes reasoning and preference-driven retrieval to adaptive decide "when and what to retrieve" and "which retrieval source to use". To better adapt to retrieval sources of differing characteristics, we also employ retrieval action adjustment and answer feedback strategy. They enable our framework to fully explore the high-quality primary source while supplementing it with secondary sources at the right time. Extensive and multi-dimensional experiments conducted on three datasets demonstrate the superiority and effectiveness of MSPR.
Authors: Linda Zeng
Abstract: Code-mixing (CM), where speakers blend languages within a single expression, is prevalent in multilingual societies but poses challenges for natural language processing due to its complexity and limited data. We propose using a large language model to generate synthetic CM data, which is then used to enhance the performance of task-specific models for CM sentiment analysis. Our results show that in Spanish-English, synthetic data improved the F1 score by 9.32%, outperforming previous augmentation techniques. However, in Malayalam-English, synthetic data only helped when the baseline was low; with strong natural data, additional synthetic data offered little benefit. Human evaluation confirmed that this approach is a simple, cost-effective way to generate natural-sounding CM sentences, particularly beneficial for low baselines. Our findings suggest that few-shot prompting of large language models is a promising method for CM data augmentation and has significant impact on improving sentiment analysis, an important element in the development of social influence systems.
Authors: Armin Pournaki, Tom Willaert
Abstract: Narratives are key interpretative devices by which humans make sense of political reality. As the significance of narratives for understanding current societal issues such as polarization and misinformation becomes increasingly evident, there is a growing demand for methods that support their empirical analysis. To this end, we propose a graph-based formalism and machine-guided method for extracting, representing, and analyzing selected narrative signals from digital textual corpora, based on Abstract Meaning Representation (AMR). The formalism and method introduced here specifically cater to the study of political narratives that figure in texts from digital media such as archived political speeches, social media posts, political manifestos and transcripts of parliamentary debates. We conceptualize these political narratives as a type of ontological narratives: stories by which actors position themselves as political beings, and which are akin to political worldviews in which actors present their normative vision of the world, or aspects thereof. We approach the study of such political narratives as a problem of information retrieval: starting from a textual corpus, we first extract a graph-like representation of the meaning of each sentence in the corpus using AMR. Drawing on transferable concepts from narratology, we then apply a set of heuristics to filter these graphs for representations of 1) actors, 2) the events in which these actors figure, and 3) traces of the perspectivization of these events. We approach these references to actors, events, and instances of perspectivization as core narrative signals that initiate a further analysis by alluding to larger political narratives. By means of a case study of State of the European Union addresses, we demonstrate how the formalism can be used to inductively surface signals of political narratives from public discourse.
Authors: Hamees Sayed, Advait Joglekar, Srinivasan Umesh
Abstract: We develop a robust translation model for four low-resource Indic languages: Khasi, Mizo, Manipuri, and Assamese. Our approach includes a comprehensive pipeline from data collection and preprocessing to training and evaluation, leveraging data from WMT task datasets, BPCC, PMIndia, and OpenLanguageData. To address the scarcity of bilingual data, we use back-translation techniques on monolingual datasets for Mizo and Khasi, significantly expanding our training corpus. We fine-tune the pre-trained NLLB 3.3B model for Assamese, Mizo, and Manipuri, achieving improved performance over the baseline. For Khasi, which is not supported by the NLLB model, we introduce special tokens and train the model on our Khasi corpus. Our training involves masked language modelling, followed by fine-tuning for English-to-Indic and Indic-to-English translations.
Authors: Carl Edwards, Ziqing Lu, Ehsan Hajiramezanali, Tommaso Biancalani, Heng Ji, Gabriele Scalia
Abstract: Bridging biomolecular modeling with natural language information, particularly through large language models (LLMs), has recently emerged as a promising interdisciplinary research area. LLMs, having been trained on large corpora of scientific documents, demonstrate significant potential in understanding and reasoning about biomolecules by providing enriched contextual and domain knowledge. However, the extent to which LLM-driven insights can improve performance on complex predictive tasks (e.g., toxicity) remains unclear. Further, the extent to which relevant knowledge can be extracted from LLMs also remains unknown. In this study, we present Molecule Caption Arena: the first comprehensive benchmark of LLM-augmented molecular property prediction. We evaluate over twenty LLMs, including both general-purpose and domain-specific molecule captioners, across diverse prediction tasks. To this goal, we introduce a novel, battle-based rating system. Our findings confirm the ability of LLM-extracted knowledge to enhance state-of-the-art molecular representations, with notable model-, prompt-, and dataset-specific variations. Code, resources, and data are available at github.com/Genentech/molcap-arena.
Authors: Yiwen Ding, Zhiheng Xi, Wei He, Zhuoyuan Li, Yitao Zhai, Xiaowei Shi, Xunliang Cai, Tao Gui, Qi Zhang, Xuanjing Huang
Abstract: Self-improvement methods enable large language models (LLMs) to generate solutions themselves and iteratively train on filtered, high-quality rationales. This process proves effective and reduces the reliance on human supervision in LLMs' reasoning, but the performance soon plateaus. We delve into the process and find that models tend to over-sample on easy queries and under-sample on queries they have yet to master. As iterations proceed, this imbalance in sampling is exacerbated, leading to a long-tail distribution where solutions to difficult queries almost diminish. This phenomenon limits the performance gain of self-improving models. A straightforward solution is brute-force sampling to balance the distribution, which significantly raises computational costs. In this paper, we introduce Guided Self-Improvement (GSI), a strategy aimed at improving the efficiency of sampling challenging heavy-tailed data. It leverages Socratic-style guidance signals to help LLM reasoning with complex queries, reducing the exploration effort and minimizing computational overhead. Experiments on four models across diverse mathematical tasks show that GSI strikes a balance between performance and efficiency, while also being effective on held-out tasks.
Authors: Liang He, Peiran Jin, Yaosen Min, Shufang Xie, Lijun Wu, Tao Qin, Xiaozhuan Liang, Kaiyuan Gao, Yuliang Jiang, Tie-Yan Liu
Abstract: Proteins, essential to biological systems, perform functions intricately linked to their three-dimensional structures. Understanding the relationship between protein structures and their amino acid sequences remains a core challenge in protein modeling. While traditional protein foundation models benefit from pre-training on vast unlabeled datasets, they often struggle to capture critical co-evolutionary information, which evolutionary-based methods excel at. In this study, we introduce a novel pre-training strategy for protein foundation models that emphasizes the interactions among amino acid residues to enhance the extraction of both short-range and long-range co-evolutionary features from sequence data. Trained on a large-scale protein sequence dataset, our model demonstrates superior generalization ability, outperforming established baselines of similar size, including the ESM model, across diverse downstream tasks. Experimental results confirm the model's effectiveness in integrating co-evolutionary information, marking a significant step forward in protein sequence-based modeling.
Authors: Oggi Rudovic, Pranay Dighe, Yi Su, Vineet Garg, Sameer Dharur, Xiaochuan Niu, Ahmed H. Abdelaziz, Saurabah Adya, Ahmed Tewfik
Abstract: Follow-up conversations with virtual assistants (VAs) enable a user to seamlessly interact with a VA without the need to repeatedly invoke it using a keyword (after the first query). Therefore, accurate Device-directed Speech Detection (DDSD) from the follow-up queries is critical for enabling naturalistic user experience. To this end, we explore the notion of Large Language Models (LLMs) and model the first query when making inference about the follow-ups (based on the ASR-decoded text), via prompting of a pretrained LLM, or by adapting a binary classifier on top of the LLM. In doing so, we also exploit the ASR uncertainty when designing the LLM prompts. We show on the real-world dataset of follow-up conversations that this approach yields large gains (20-40% reduction in false alarms at 10% fixed false rejects) due to the joint modeling of the previous speech context and ASR uncertainty, compared to when follow-ups are modeled alone.
Authors: Andre Niyongabo Rubungo, Kangming Li, Jason Hattrick-Simpers, Adji Bousso Dieng
Abstract: Large language models (LLMs) are increasingly being used in materials science. However, little attention has been given to benchmarking and standardized evaluation for LLM-based materials property prediction, which hinders progress. We present LLM4Mat-Bench, the largest benchmark to date for evaluating the performance of LLMs in predicting the properties of crystalline materials. LLM4Mat-Bench contains about 1.9M crystal structures in total, collected from 10 publicly available materials data sources, and 45 distinct properties. LLM4Mat-Bench features different input modalities: crystal composition, CIF, and crystal text description, with 4.7M, 615.5M, and 3.1B tokens in total for each modality, respectively. We use LLM4Mat-Bench to fine-tune models with different sizes, including LLM-Prop and MatBERT, and provide zero-shot and few-shot prompts to evaluate the property prediction capabilities of LLM-chat-like models, including Llama, Gemma, and Mistral. The results highlight the challenges of general-purpose LLMs in materials science and the need for task-specific predictive models and task-specific instruction-tuned LLMs in materials property prediction.
Authors: Beyazit Yalcinkaya, Niklas Lauffer, Marcell Vazquez-Chanlatte, Sanjit A. Seshia
Abstract: Goal-conditioned reinforcement learning is a powerful way to control an AI agent's behavior at runtime. That said, popular goal representations, e.g., target states or natural language, are either limited to Markovian tasks or rely on ambiguous task semantics. We propose representing temporal goals using compositions of deterministic finite automata (cDFAs) and use cDFAs to guide RL agents. cDFAs balance the need for formal temporal semantics with ease of interpretation: if one can understand a flow chart, one can understand a cDFA. On the other hand, cDFAs form a countably infinite concept class with Boolean semantics, and subtle changes to the automaton can result in very different tasks, making them difficult to condition agent behavior on. To address this, we observe that all paths through a DFA correspond to a series of reach-avoid tasks and propose pre-training graph neural network embeddings on "reach-avoid derived" DFAs. Through empirical evaluation, we demonstrate that the proposed pre-training method enables zero-shot generalization to various cDFA task classes and accelerated policy specialization without the myopic suboptimality of hierarchical methods.
Authors: Bohan Lyu, Yadi Cao, Duncan Watson-Parris, Leon Bergen, Taylor Berg-Kirkpatrick, Rose Yu
Abstract: Large Language Models (LLMs) demonstrate promising capabilities in solving simple scientific problems but often produce hallucinations for complex ones. While integrating LLMs with tools can increase reliability, this approach typically results in over-reliance on tools, diminishing the model's ability to solve simple problems through basic reasoning. In contrast, human experts first assess problem complexity using domain knowledge before choosing an appropriate solution approach. Inspired by this human problem-solving process, we propose a novel two-component fine-tuning method. In the first component World Knowledge Distillation (WKD), LLMs learn directly from solutions generated using tool's information to internalize domain knowledge. In the second component Tool Usage Adaptation (TUA), we partition problems into easy and hard categories based on the model's direct answering accuracy. While maintaining the same alignment target for easy problems as in WKD, we train the model to intelligently switch to tool usage for more challenging problems. We validate our method on six scientific benchmark datasets, spanning mathematics, climate science and epidemiology. On average, our models demonstrate a 28.18% improvement in answer accuracy and a 13.89% increase in tool usage precision across all datasets, surpassing state-of-the-art models including GPT-4o and Claude-3.5.
Authors: Evan Miller
Abstract: Evaluations are critical for understanding the capabilities of large language models (LLMs). Fundamentally, evaluations are experiments; but the literature on evaluations has largely ignored the literature from other sciences on experiment analysis and planning. This article shows researchers with some training in statistics how to think about and analyze data from language model evaluations. Conceptualizing evaluation questions as having been drawn from an unseen super-population, we present formulas for analyzing evaluation data, measuring differences between two models, and planning an evaluation experiment. We make a number of specific recommendations for running language model evaluations and reporting experiment results in a way that minimizes statistical noise and maximizes informativeness.
Authors: Nikolaos Flemotomos, Roger Hsiao, Pawel Swietojanski, Dogan Can, Xiaodan Zhuang
Abstract: Neural contextual biasing allows speech recognition models to leverage contextually relevant information, leading to improved transcription accuracy. However, the biasing mechanism is typically based on a cross-attention module between the audio and a catalogue of biasing entries, which means computational complexity can pose severe practical limitations on the size of the biasing catalogue and consequently on accuracy improvements. This work proposes an approximation to cross-attention scoring based on vector quantization and enables compute- and memory-efficient use of large biasing catalogues. We propose to use this technique jointly with a retrieval based contextual biasing approach. First, we use an efficient quantized retrieval module to shortlist biasing entries by grounding them on audio. Then we use retrieved entries for biasing. Since the proposed approach is agnostic to the biasing method, we investigate using full cross-attention, LLM prompting, and a combination of the two. We show that retrieval based shortlisting allows the system to efficiently leverage biasing catalogues of several thousands of entries, resulting in up to 71% relative error rate reduction in personal entity recognition. At the same time, the proposed approximation algorithm reduces compute time by 20% and memory usage by 85-95%, for lists of up to one million entries, when compared to standard dot-product cross-attention.
Authors: Srikumar Sastry, Subash Khanal, Aayush Dhakal, Adeel Ahmad, Nathan Jacobs
Abstract: We present TaxaBind, a unified embedding space for characterizing any species of interest. TaxaBind is a multimodal embedding space across six modalities: ground-level images of species, geographic location, satellite image, text, audio, and environmental features, useful for solving ecological problems. To learn this joint embedding space, we leverage ground-level images of species as a binding modality. We propose multimodal patching, a technique for effectively distilling the knowledge from various modalities into the binding modality. We construct two large datasets for pretraining: iSatNat with species images and satellite images, and iSoundNat with species images and audio. Additionally, we introduce TaxaBench-8k, a diverse multimodal dataset with six paired modalities for evaluating deep learning models on ecological tasks. Experiments with TaxaBind demonstrate its strong zero-shot and emergent capabilities on a range of tasks including species classification, cross-model retrieval, and audio classification. The datasets and models are made available at https://github.com/mvrl/TaxaBind.
Authors: Aashiq Muhamed, Mona Diab, Virginia Smith
Abstract: Understanding and mitigating the potential risks associated with foundation models (FMs) hinges on developing effective interpretability methods. Sparse Autoencoders (SAEs) have emerged as a promising tool for disentangling FM representations, but they struggle to capture rare, yet crucial concepts in the data. We introduce Specialized Sparse Autoencoders (SSAEs), designed to illuminate these elusive dark matter features by focusing on specific subdomains. We present a practical recipe for training SSAEs, demonstrating the efficacy of dense retrieval for data selection and the benefits of Tilted Empirical Risk Minimization as a training objective to improve concept recall. Our evaluation of SSAEs on standard metrics, such as downstream perplexity and $L_0$ sparsity, show that they effectively capture subdomain tail concepts, exceeding the capabilities of general-purpose SAEs. We showcase the practical utility of SSAEs in a case study on the Bias in Bios dataset, where SSAEs achieve a 12.5\% increase in worst-group classification accuracy when applied to remove spurious gender information. SSAEs provide a powerful new lens for peering into the inner workings of FMs in subdomains.
Authors: Ziting Wang, Haitao Yuan, Wei Dong, Gao Cong, Feifei Li
Abstract: Large Language Models (LLMs) have demonstrated remarkable generation capabilities but often struggle to access up-to-date information, which can lead to hallucinations. Retrieval-Augmented Generation (RAG) addresses this issue by incorporating knowledge from external databases, enabling more accurate and relevant responses. Due to the context window constraints of LLMs, it is impractical to input the entire external database context directly into the model. Instead, only the most relevant information, referred to as chunks, is selectively retrieved. However, current RAG research faces three key challenges. First, existing solutions often select each chunk independently, overlooking potential correlations among them. Second, in practice the utility of chunks is non-monotonic, meaning that adding more chunks can decrease overall utility. Traditional methods emphasize maximizing the number of included chunks, which can inadvertently compromise performance. Third, each type of user query possesses unique characteristics that require tailored handling, an aspect that current approaches do not fully consider. To overcome these challenges, we propose a cost constrained retrieval optimization system CORAG for retrieval-augmented generation. We employ a Monte Carlo Tree Search (MCTS) based policy framework to find optimal chunk combinations sequentially, allowing for a comprehensive consideration of correlations among chunks. Additionally, rather than viewing budget exhaustion as a termination condition, we integrate budget constraints into the optimization of chunk combinations, effectively addressing the non-monotonicity of chunk utility.
Authors: Xiong Wang, Yangze Li, Chaoyou Fu, Lei Xie, Ke Li, Xing Sun, Long Ma
Abstract: The rapid development of large language models has brought many new smart applications, especially the excellent multimodal human-computer interaction in GPT-4o has brought impressive experience to users. In this background, researchers have proposed many multimodal LLMs that can achieve speech-to-speech dialogue recently. In this paper, we propose a speech-text multimodal LLM architecture called Freeze-Omni. Our main contribution is the speech input and output modalities can connected to the LLM while keeping the LLM frozen throughout the training process. We designed 3-stage training strategies both for the modeling of speech input and output, enabling Freeze-Omni to obtain speech-to-speech dialogue ability using text-speech paired data (such as ASR and TTS data) and only 60,000 multi-round text Q&A data on 8 GPUs. Moreover, we can effectively ensure that the intelligence of the Freeze-Omni in the speech modality is at the same level compared with that in the text modality of its backbone LLM, while the end-to-end latency of the spoken response achieves a low level. In addition, we also designed a method to achieve duplex dialogue ability through multi-task training, making Freeze-Omni have a more natural style of dialogue ability between the users. Freeze-Omni mainly provides a possibility for researchers to conduct multimodal LLM under the condition of a frozen LLM, avoiding various impacts caused by the catastrophic forgetting of LLM caused by fewer data and training resources.
Authors: Abhilash Nandy, Manav Nitin Kapadnis, Sohan Patnaik, Yash Parag Butala, Pawan Goyal, Niloy Ganguly
Abstract: In this paper, we propose $FastDoc$ (Fast Continual Pre-training Technique using Document Level Metadata and Taxonomy), a novel, compute-efficient framework that utilizes Document metadata and Domain-Specific Taxonomy as supervision signals to continually pre-train transformer encoder on a domain-specific corpus. The main innovation is that during domain-specific pretraining, an open-domain encoder is continually pre-trained using sentence-level embeddings as inputs (to accommodate long documents), however, fine-tuning is done with token-level embeddings as inputs to this encoder. We perform such domain-specific pre-training on three different domains namely customer support, scientific, and legal domains, and compare performance on 6 different downstream tasks and 9 different datasets. The novel use of document-level supervision along with sentence-level embedding input for pre-training reduces pre-training compute by around $1,000$, $4,500$, and $500$ times compared to MLM and/or NSP in Customer Support, Scientific, and Legal Domains, respectively. The reduced training time does not lead to a deterioration in performance. In fact we show that $FastDoc$ either outperforms or performs on par with several competitive transformer-based baselines in terms of character-level F1 scores and other automated metrics in the Customer Support, Scientific, and Legal Domains. Moreover, reduced training aids in mitigating the risk of catastrophic forgetting. Thus, unlike baselines, $FastDoc$ shows a negligible drop in performance on open domain.
Authors: Yonchanok Khaokaew, Hao Xue, Flora D. Salim
Abstract: In recent years, predicting mobile app usage has become increasingly important for areas like app recommendation, user behaviour analysis, and mobile resource management. Existing models, however, struggle with the heterogeneous nature of contextual data and the user cold start problem. This study introduces a novel prediction model, Mobile App Prediction Leveraging Large Language Model Embeddings (MAPLE), which employs Large Language Models (LLMs) and installed app similarity to overcome these challenges. MAPLE utilises the power of LLMs to process contextual data and discern intricate relationships within it effectively. Additionally, we explore the use of installed app similarity to address the cold start problem, facilitating the modelling of user preferences and habits, even for new users with limited historical data. In essence, our research presents MAPLE as a novel, potent, and practical approach to app usage prediction, making significant strides in resolving issues faced by existing models. MAPLE stands out as a comprehensive and effective solution, setting a new benchmark for more precise and personalised app usage predictions. In tests on two real-world datasets, MAPLE surpasses contemporary models in both standard and cold start scenarios. These outcomes validate MAPLE's capacity for precise app usage predictions and its resilience against the cold start problem. This enhanced performance stems from the model's proficiency in capturing complex temporal patterns and leveraging contextual information. As a result, MAPLE can potentially improve personalised mobile app usage predictions and user experiences markedly.
Authors: Bach Pham, JuiHsuan Wong, Samuel Kim, Yunting Yin, Steven Skiena
Abstract: Dictionary definitions are historically the arbitrator of what words mean, but this primacy has come under threat by recent progress in NLP, including word embeddings and generative models like ChatGPT. We present an exploratory study of the degree of alignment between word definitions from classical dictionaries and these newer computational artifacts. Specifically, we compare definitions from three published dictionaries to those generated from variants of ChatGPT. We show that (i) definitions from different traditional dictionaries exhibit more surface form similarity than do model-generated definitions, (ii) that the ChatGPT definitions are highly accurate, comparable to traditional dictionaries, and (iii) ChatGPT-based embedding definitions retain their accuracy even on low frequency words, much better than GloVE and FastText word embeddings.
Authors: Do Xuan Long, Kenji Kawaguchi, Min-Yen Kan, Nancy F. Chen
Abstract: Reasoning and predicting human opinions with large language models (LLMs) is essential yet challenging. Current methods employ role-playing with personae but face two major issues: LLMs are sensitive to even a single irrelevant persona, skewing predictions by up to 30%, and LLMs fail to reason strategically over personae. We propose Chain-of-Opinion (COO), a simple four-step solution modeling which and how to reason with personae, inspired by the Value--Belief--Norm (VBN) theory. COO differentiates between explicit personae (demographics and ideology) and implicit personae (historical opinions), involves: (1) filtering irrelevant attributes from explicit personae, (2) ranking implicit personae into a preferential list for selecting top-k, (3) applying novel VBN reasoning to extract user environmental and personal value, belief, and norm variables for accurate and reliable predictions, and (4) iterating VBN reasoning with progressively larger lists of implicit personae to handle potential persona insufficiency. COO efficiently achieves new state-of-the-art opinion prediction via prompting with only 5 inference calls, improving prior techniques by up to 4%. Notably, fine-tuning LMs with COO data results in significantly better opinion-aligned models, by up to 23%.
Authors: Yongliang Shen, Kaitao Song, Xu Tan, Wenqi Zhang, Kan Ren, Siyu Yuan, Weiming Lu, Dongsheng Li, Yueting Zhuang
Abstract: In recent years, the remarkable progress of large language models (LLMs) has sparked interest in task automation, which involves decomposing complex tasks described by user instructions into sub-tasks and invoking external tools to execute them, playing a central role in autonomous agents. However, there is a lack of systematic and standardized benchmarks to promote the development of LLMs in task automation. To address this, we introduce TaskBench, a comprehensive framework to evaluate the capability of LLMs in task automation. Specifically, task automation can be divided into three critical stages: task decomposition, tool selection, and parameter prediction. To tackle the complexities inherent in these stages, we introduce the concept of Tool Graph to represent decomposed tasks and adopt a back-instruct method to generate high-quality user instructions. We propose TaskEval, a multi-faceted evaluation methodology that assesses LLM performance across these three stages. Our approach combines automated construction with rigorous human verification, ensuring high consistency with human evaluation. Experimental results demonstrate that TaskBench effectively reflects the capabilities of various LLMs in task automation. It provides insights into model performance across different task complexities and domains, pushing the boundaries of what current models can achieve. TaskBench offers a scalable, adaptable, and reliable benchmark for advancing LLM-based autonomous agents.
Authors: Lena Held, Ivan Habernal
Abstract: Why does an argument end up in the final court decision? Was it deliberated or questioned during the oral hearings? Was there something in the hearings that triggered a particular judge to write a dissenting opinion? Despite the availability of the final judgments of the European Court of Human Rights (ECHR), none of these legal research questions can currently be answered as the ECHR's multilingual oral hearings are not transcribed, structured, or speaker-attributed. We address this fundamental gap by presenting LaCour!, the first corpus of textual oral arguments of the ECHR, consisting of 154 full hearings (2.1 million tokens from over 267 hours of video footage) in English, French, and other court languages, each linked to the corresponding final judgment documents. In addition to the transcribed and partially manually corrected text from the video, we provide sentence-level timestamps and manually annotated role and language labels. We also showcase LaCour! in a set of preliminary experiments that explore the interplay between questions and dissenting opinions. Apart from the use cases in legal NLP, we hope that law students or other interested parties will also use LaCour! as a learning resource, as it is freely available in various formats at https://huggingface.co/datasets/TrustHLT/LaCour.
Authors: Guangji Bai, Yijiang Li, Chen Ling, Kibaek Kim, Liang Zhao
Abstract: The transformative impact of large language models (LLMs) like LLaMA and GPT on natural language processing is countered by their prohibitive computational demands. Pruning has emerged as a pivotal compression strategy, introducing sparsity to enhance both memory and computational efficiency. Yet, traditional global pruning is impractical for LLMs due to scalability issues, while local pruning, despite its efficiency, leads to suboptimal solutions. Addressing these challenges, we propose SparseLLM, a novel framework that redefines the global pruning process into manageable, coordinated subproblems, allowing for resource-efficient optimization with global optimality. SparseLLM's approach, which conceptualizes LLMs as a chain of modular functions and leverages auxiliary variables for problem decomposition, not only facilitates a pragmatic application on LLMs but also demonstrates significant performance improvements, particularly in high-sparsity regimes where it surpasses current state-of-the-art methods.
Authors: Michal Lukasik, Harikrishna Narasimhan, Aditya Krishna Menon, Felix Yu, Sanjiv Kumar
Abstract: Large language models (LLMs) have shown strong results on a range of applications, including regression and scoring tasks. Typically, one obtains outputs from an LLM via autoregressive sampling from the model's output distribution. We show that this inference strategy can be sub-optimal for common regression and scoring evaluation metrics. As a remedy, we build on prior work on Minimum Bayes Risk decoding, and propose alternate inference strategies that estimate the Bayes-optimal solution for regression and scoring metrics in closed-form from sampled responses. We show that our proposal significantly improves over baselines across datasets and models.
Authors: Chang Zong, Yuyan Chen, Weiming Lu, Jian Shao, Yueting Zhuang
Abstract: Large Language Models (LLMs) have demonstrated efficacy in various linguistic applications, including text summarization and controlled text generation. However, studies into their capacity of switching between styles via instruction tuning remain underexplored. This study concentrates on the style-switching abilities of LLMs and introduces a novel approach, named ProSwitch, which enables a language model to switch between professional and non-professional answers, by tuning and evaluating through the guidance of domain and style knowledge. ProSwitch unfolds across three phases: LLM-augmented preparation to collect domain knowledge and QA pairs, instruction tuning to optimize LLMs with multiple levels of knowledge, and comprehensive evaluation to assess both style discrimination and reference-based quality of generated text. Comparative analysis of ProSwitch against general and specialized LLMs reveals that our approach outperforms baselines in switching between professional and non-professional answers.
Authors: Jo\~ao Monteiro, \'Etienne Marcotte, Pierre-Andr\'e No\"el, Valentina Zantedeschi, David V\'azquez, Nicolas Chapados, Christopher Pal, Perouz Taslakian
Abstract: In-context learning (ICL) approaches typically leverage prompting to condition decoder-only language model generation on reference information. Just-in-time processing of a context is inefficient due to the quadratic cost of self-attention operations, and caching is desirable. However, caching transformer states can easily require almost as much space as the model parameters. When the right context isn't known in advance, caching ICL can be challenging. This work addresses these limitations by introducing models that, inspired by the encoder-decoder architecture, use cross-attention to condition generation on reference text without the prompt. More precisely, we leverage pre-trained decoder-only models and only train a small number of added layers. We use Question-Answering (QA) as a testbed to evaluate the ability of our models to perform conditional generation and observe that they outperform ICL, are comparable to fine-tuned prompted LLMs, and drastically reduce the space footprint relative to standard KV caching by two orders of magnitude.
Authors: Xin Qiu, Risto Miikkulainen
Abstract: With the widespread application of Large Language Models (LLMs) to various domains, concerns regarding the trustworthiness of LLMs in safety-critical scenarios have been raised, due to their unpredictable tendency to hallucinate and generate misinformation. Existing LLMs do not have an inherent functionality to provide the users with an uncertainty/confidence metric for each response it generates, making it difficult to evaluate trustworthiness. Although several studies aim to develop uncertainty quantification methods for LLMs, they have fundamental limitations, such as being restricted to classification tasks, requiring additional training and data, considering only lexical instead of semantic information, and being prompt-wise but not response-wise. A new framework is proposed in this paper to address these issues. Semantic density extracts uncertainty/confidence information for each response from a probability distribution perspective in semantic space. It has no restriction on task types and is "off-the-shelf" for new models and tasks. Experiments on seven state-of-the-art LLMs, including the latest Llama 3 and Mixtral-8x22B models, on four free-form question-answering benchmarks demonstrate the superior performance and robustness of semantic density compared to prior approaches.
Authors: Yongxin Zhu, Dan Su, Liqiang He, Linli Xu, Dong Yu
Abstract: While recent advancements in speech language models have achieved significant progress, they face remarkable challenges in modeling the long acoustic sequences of neural audio codecs. In this paper, we introduce \textbf{G}enerative \textbf{P}re-trained \textbf{S}peech \textbf{T}ransformer (GPST), a hierarchical transformer designed for efficient speech language modeling. GPST quantizes audio waveforms into two distinct types of discrete speech representations and integrates them within a hierarchical transformer architecture, allowing for a unified one-stage generation process and enhancing Hi-Res audio generation capabilities. By training on large corpora of speeches in an end-to-end unsupervised manner, GPST can generate syntactically consistent speech with diverse speaker identities. Given a brief 3-second prompt, GPST can produce natural and coherent personalized speech, demonstrating in-context learning abilities. Moreover, our approach can be easily extended to spoken cross-lingual speech generation by incorporating multi-lingual semantic tokens and universal acoustic tokens. Experimental results indicate that GPST significantly outperforms the existing speech language models in terms of word error rate, speech quality, and speaker similarity. The code is available at \url{https://github.com/youngsheen/GPST}.
Authors: Yangruibo Ding, Jinjun Peng, Marcus J. Min, Gail Kaiser, Junfeng Yang, Baishakhi Ray
Abstract: Code Large Language Models (Code LLMs) have excelled at tasks like code completion but often miss deeper semantics such as execution effects and dynamic states. This paper aims to bridge the gap between Code LLMs' reliance on static text data and the need for semantic understanding for complex tasks like debugging and program repair. We introduce a novel strategy, monologue reasoning, to train Code LLMs to reason comprehensive semantics, encompassing high-level functional descriptions, local execution effects of individual statements, and overall input/output behavior, thereby linking static code text with dynamic execution states. We begin by collecting PyX, a clean Python corpus of fully executable code samples with functional descriptions and test cases. We propose training Code LLMs not only to write code but also to understand code semantics by reasoning about key properties, constraints, and execution behaviors using natural language, mimicking human verbal debugging, i.e., rubber-duck debugging. This approach led to the development of SemCoder, a Code LLM with only 6.7B parameters, which shows competitive performance with GPT-3.5-turbo on code generation and execution reasoning tasks. SemCoder achieves 79.3% on HumanEval (GPT-3.5-turbo: 76.8%), 63.6% on CRUXEval-I (GPT-3.5-turbo: 50.3%), and 63.9% on CRUXEval-O (GPT-3.5-turbo: 59.0%). We also study the effectiveness of SemCoder's monologue-style execution reasoning compared to concrete scratchpad reasoning, showing that our approach integrates semantics from multiple dimensions more smoothly. Finally, we demonstrate the potential of applying learned semantics to improve Code LLMs' debugging and self-refining capabilities. Our data, code, and models are available at: https://github.com/ARiSE-Lab/SemCoder.
Authors: Haokun Lin, Haobo Xu, Yichen Wu, Jingzhi Cui, Yingtao Zhang, Linzhan Mou, Linqi Song, Zhenan Sun, Ying Wei
Abstract: Quantization of large language models (LLMs) faces significant challenges, particularly due to the presence of outlier activations that impede efficient low-bit representation. Traditional approaches predominantly address Normal Outliers, which are activations across all tokens with relatively large magnitudes. However, these methods struggle with smoothing Massive Outliers that display significantly larger values, which leads to significant performance degradation in low-bit quantization. In this paper, we introduce DuQuant, a novel approach that utilizes rotation and permutation transformations to more effectively mitigate both massive and normal outliers. First, DuQuant starts by constructing the rotation matrix, using specific outlier dimensions as prior knowledge, to redistribute outliers to adjacent channels by block-wise rotation. Second, We further employ a zigzag permutation to balance the distribution of outliers across blocks, thereby reducing block-wise variance. A subsequent rotation further smooths the activation landscape, enhancing model performance. DuQuant simplifies the quantization process and excels in managing outliers, outperforming the state-of-the-art baselines across various sizes and types of LLMs on multiple tasks, even with 4-bit weight-activation quantization. Our code is available at https://github.com/Hsu1023/DuQuant.
Authors: Namgyu Ho, Sangmin Bae, Taehyeon Kim, Hyunjik Jo, Yireun Kim, Tal Schuster, Adam Fisch, James Thorne, Se-Young Yun
Abstract: We introduce the Block Transformer which adopts hierarchical global-to-local modeling to autoregressive transformers to mitigate the inference bottlenecks associated with self-attention. Self-attention requires the key-value (KV) cache of all previous sequences to be retrieved from memory at every decoding step to retrieve context information, leading to two primary bottlenecks during batch inference. First, there is a significant delay in obtaining the first token, as the information of the entire prompt must first be processed to prefill the KV cache. Second, computation of subsequent tokens is bottlenecked by the high memory I/O demand of fetching the entire KV cache, which grows linearly with sequence length, incurring quadratic memory reads overall. We design the Block Transformer to strategically mitigate these costs, by incorporating coarsity and locality into an integrated global-to-local architecture. At the lower layers, we aggregate tokens into fixed size blocks to apply attention across the entire sequence at coarse-grained detail, to capture the global context while minimizing KV cache overhead. At upper layers, we apply attention within each block to decode individual tokens, to model fine-grained details with a lightweight local KV cache. We pretrain vanilla and Block Transformers from scratch and demonstrate that Block Transformers reach 10--20x inference throughput compared to vanilla transformers with equivalent perplexity and zero-shot task performance. Code is available at https://github.com/itsnamgyu/block-transformer.
Authors: Xiao Yang, Kai Sun, Hao Xin, Yushi Sun, Nikita Bhalla, Xiangsen Chen, Sajal Choudhary, Rongze Daniel Gui, Ziran Will Jiang, Ziyu Jiang, Lingkun Kong, Brian Moran, Jiaqi Wang, Yifan Ethan Xu, An Yan, Chenyu Yang, Eting Yuan, Hanwen Zha, Nan Tang, Lei Chen, Nicolas Scheffer, Yue Liu, Nirav Shah, Rakesh Wanga, Anuj Kumar, Wen-tau Yih, Xin Luna Dong
Abstract: Retrieval-Augmented Generation (RAG) has recently emerged as a promising solution to alleviate Large Language Model (LLM)'s deficiency in lack of knowledge. Existing RAG datasets, however, do not adequately represent the diverse and dynamic nature of real-world Question Answering (QA) tasks. To bridge this gap, we introduce the Comprehensive RAG Benchmark (CRAG), a factual question answering benchmark of 4,409 question-answer pairs and mock APIs to simulate web and Knowledge Graph (KG) search. CRAG is designed to encapsulate a diverse array of questions across five domains and eight question categories, reflecting varied entity popularity from popular to long-tail, and temporal dynamisms ranging from years to seconds. Our evaluation of this benchmark highlights the gap to fully trustworthy QA. Whereas most advanced LLMs achieve <=34% accuracy on CRAG, adding RAG in a straightforward manner improves the accuracy only to 44%. State-of-the-art industry RAG solutions only answer 63% of questions without any hallucination. CRAG also reveals much lower accuracy in answering questions regarding facts with higher dynamism, lower popularity, or higher complexity, suggesting future research directions. The CRAG benchmark laid the groundwork for a KDD Cup 2024 challenge and attracted thousands of participants and submissions. We commit to maintaining CRAG to serve research communities in advancing RAG solutions and general QA solutions. CRAG is available at https://github.com/facebookresearch/CRAG/.
Authors: Yoo Yeon Sung, Maharshi Gor, Eve Fleisig, Ishani Mondal, Jordan Lee Boyd-Graber
Abstract: Adversarial datasets should ensure AI robustness that matches human performance. However, as models evolve, datasets can become obsolete. Thus, adversarial datasets should be periodically updated based on their degradation in adversarialness. Given the lack of a standardized metric for measuring adversarialness, we propose AdvScore, a human-grounded evaluation metric. AdvScore assesses a dataset's true adversarialness by capturing models' and humans' varying abilities, while also identifying poor examples. AdvScore then motivates a new dataset creation pipeline for realistic and high-quality adversarial samples, enabling us to collect an adversarial question answering (QA) dataset, AdvQA. We apply AdvScore using 9,347 human responses and ten language model predictions to track the models' improvement over five years (from 2020 to 2024). AdvScore assesses whether adversarial datasets remain suitable for model evaluation, measures model improvements, and provides guidance for better alignment with human capabilities.
Authors: Deokyeong Kang, Ki Jung Seo, Taeuk Kim
Abstract: Modular programming, which aims to construct the final program by integrating smaller, independent building blocks, has been regarded as a desirable practice in software development. However, with the rise of recent code generation agents built upon large language models (LLMs), a question emerges: is this traditional practice equally effective for these new tools? In this work, we assess the impact of modularity in code generation by introducing a novel metric for its quantitative measurement. Surprisingly, unlike conventional wisdom on the topic, we find that modularity is not a core factor for improving the performance of code generation models. We also explore potential explanations for why LLMs do not exhibit a preference for modular code compared to non-modular code.
Authors: Manar Abdelatty, Jacob Rosenstein, Sherief Reda
Abstract: Hardware design workflows often involve working with Process Design Kits (PDKs) from various fabrication labs, each containing its own set of standard cell libraries optimized for metrics such as speed, power, or density. These libraries include multiple views for information on timing and electrical properties of cells, cell layout details, and process design rules. Engineers typically navigate between the design and the target technology to make informed decisions on different design scenarios, such as selecting specific gates for area optimization or enhancing critical path speed. Navigating this complex landscape to retrieve specific information about gates or design rules is often time-consuming and error-prone. To address this, we present HDLCopilot, a multi-agent collaborative framework powered by large language models that enables engineers to streamline interactions with hardware design and PDKs through natural language queries. HDLCopilot enables engineers to quickly access relevant information on gates and design rules, evaluate tradeoffs related to area, speed, and power in order to make informed decisions efficiently and accurately. The framework achieves an execution accuracy of 96.33\% on a diverse set of complex natural language queries. HDLCopilot positions itself as a powerful assistant in hardware design workflows, enhancing productivity and reducing potential human errors.
Authors: Chaofan Tao, Qian Liu, Longxu Dou, Niklas Muennighoff, Zhongwei Wan, Ping Luo, Min Lin, Ngai Wong
Abstract: Research on scaling large language models (LLMs) has primarily focused on model parameters and training data size, overlooking the role of vocabulary size. We investigate how vocabulary size impacts LLM scaling laws by training models ranging from 33M to 3B parameters on up to 500B characters with various vocabulary configurations. We propose three complementary approaches for predicting the compute-optimal vocabulary size: IsoFLOPs analysis, derivative estimation, and parametric fit of the loss function. Our approaches converge on the conclusion that the optimal vocabulary size depends on the compute budget, with larger models requiring larger vocabularies. Most LLMs, however, use insufficient vocabulary sizes. For example, we predict that the optimal vocabulary size of Llama2-70B should have been at least 216K, 7 times larger than its vocabulary of 32K. We validate our predictions empirically by training models with 3B parameters across different FLOPs budgets. Adopting our predicted optimal vocabulary size consistently improves downstream performance over commonly used vocabulary sizes. By increasing the vocabulary size from the conventional 32K to 43K, we improve performance on ARC-Challenge from 29.1 to 32.0 with the same 2.3e21 FLOPs. Our work highlights the importance of jointly considering tokenization and model scaling for efficient pre-training. The code and demo are available at https://github.com/sail-sg/scaling-with-vocab and https://hf.co/spaces/sail/scaling-with-vocab-demo.
URLs: https://github.com/sail-sg/scaling-with-vocab, https://hf.co/spaces/sail/scaling-with-vocab-demo.
Authors: Ting-Ruen Wei, Yuan Wang, Yoshitaka Inoue, Hsin-Tai Wu, Yi Fang
Abstract: Missing data in tabular dataset is a common issue as the performance of downstream tasks usually depends on the completeness of the training dataset. Previous missing data imputation methods focus on numeric and categorical columns, but we propose a novel end-to-end approach called Table Transformers for Imputing Textual Attributes (TTITA) based on the transformer to impute unstructured textual columns using other columns in the table. We conduct extensive experiments on three datasets, and our approach shows competitive performance outperforming baseline models such as recurrent neural networks and Llama2. The performance improvement is more significant when the target sequence has a longer length. Additionally, we incorporate multi-task learning to simultaneously impute for heterogeneous columns, boosting the performance for text imputation. We also qualitatively compare with ChatGPT for realistic applications.
Authors: Federico Belotti, Fabio Dadda, Marco Cremaschi, Roberto Avogadro, Matteo Palmonari
Abstract: Tables are crucial containers of information, but understanding their meaning may be challenging. Over the years, there has been a surge in interest in data-driven approaches based on deep learning that have increasingly been combined with heuristic-based ones. In the last period, the advent of \acf{llms} has led to a new category of approaches for table annotation. However, these approaches have not been consistently evaluated on a common ground, making evaluation and comparison difficult. This work proposes an extensive evaluation of four STI SOTA approaches: Alligator (formerly s-elbat), Dagobah, TURL, and TableLlama; the first two belong to the family of heuristic-based algorithms, while the others are respectively encoder-only and decoder-only Large Language Models (LLMs). We also include in the evaluation both GPT-4o and GPT-4o-mini, since they excel in various public benchmarks. The primary objective is to measure the ability of these approaches to solve the entity disambiguation task with respect to both the performance achieved on a common-ground evaluation setting and the computational and cost requirements involved, with the ultimate aim of charting new research paths in the field.
Authors: Shantanu Ghosh, Rayan Syed, Chenyu Wang, Clare B. Poynton, Shyam Visweswaran, Kayhan Batmanghelich
Abstract: Error slice discovery associates structured patterns with model errors. Existing methods discover error slices by clustering the error-prone samples with similar patterns or assigning discrete attributes to each sample for post-hoc analysis. While these methods aim for interpretability and easier mitigation through reweighting or rebalancing, they may not capture the full complexity of error patterns due to incomplete or missing attributes. Contrary to the existing approach, this paper utilizes the reasoning capabilities of the Large Language Model (LLM) to analyze complex error patterns and generate testable hypotheses. This paper proposes LADDER: Language Driven slice Discovery and Error Rectification. It first projects the model's representation into a language-aligned feature space (eg CLIP) to preserve semantics in the original model feature space. This ensures the accurate retrieval of sentences that highlight the model's errors. Next, the LLM utilizes the sentences and generates hypotheses to discover error slices. Finally, we mitigate the error by fine-tuning the classification head by creating a group-balanced dataset using the hypotheses. Our entire method does not require any attribute annotation, either explicitly or through external tagging models. We validate our method with \textbf{five} image classification datasets. The code is available (https://github.com/batmanlab/Ladder).
Authors: Wentao Liu, Qianjun Pan, Yi Zhang, Zhuo Liu, Ji Wu, Jie Zhou, Aimin Zhou, Qin Chen, Bo Jiang, Liang He
Abstract: Large language models (LLMs) have obtained promising results in mathematical reasoning, which is a foundational skill for human intelligence. Most previous studies focus on improving and measuring the performance of LLMs based on textual math reasoning datasets (e.g., MATH, GSM8K). Recently, a few researchers have released English multimodal math datasets (e.g., MATHVISTA and MATH-V) to evaluate the effectiveness of large multimodal models (LMMs). In this paper, we release a Chinese multimodal math (CMM-Math) dataset, including benchmark and training parts, to evaluate and enhance the mathematical reasoning of LMMs. CMM-Math contains over 28,000 high-quality samples, featuring a variety of problem types (e.g., multiple-choice, fill-in-the-blank, and so on) with detailed solutions across 12 grade levels from elementary to high school in China. Specifically, the visual context may be present in the questions or opinions, which makes this dataset more challenging. Through comprehensive analysis, we discover that state-of-the-art LMMs on the CMM-Math dataset face challenges, emphasizing the necessity for further improvements in LMM development. We also propose a Multimodal Mathematical LMM (Math-LMM) to handle the problems with mixed input of multiple images and text segments. We train our model using three stages, including foundational pre-training, foundational fine-tuning, and mathematical fine-tuning. The extensive experiments indicate that our model effectively improves math reasoning performance by comparing it with the SOTA LMMs over three multimodal mathematical datasets.
Authors: Xinyuan Lu, Liangming Pan, Yubo Ma, Preslav Nakov, Min-Yen Kan
Abstract: Current Large Language Models (LLMs) exhibit limited ability to understand table structures and to apply precise numerical reasoning, which is crucial for tasks such as table question answering (TQA) and table-based fact verification (TFV). To address these challenges, we introduce our Tool-Augmented Reasoning framework for Tables (TART), which integrates LLMs with specialized tools. TART contains three key components: a table formatter to ensure accurate data representation, a tool maker to develop specific computational tools, and an explanation generator to maintain explainability. We also present the TOOLTAB dataset, a new benchmark designed specifically for training LLMs in table-tool integration. Our experiments indicate that TART achieves substantial improvements over existing methods (e.g., Chain-of-Thought) by improving both the precision of data processing and the clarity of the reasoning process. Notably, TART paired with CodeLlama achieves 90.0% of the accuracy of the closed-sourced LLM GPT-3.5-turbo, highlighting its robustness in diverse real-world scenarios. All the code and data are available at https://github.com/XinyuanLu00/TART.
Authors: Wei-Cheng Tseng, Yi-Jen Shih, David Harwath, Raymond Mooney
Abstract: Audio-visual pre-trained models have gained substantial attention recently and demonstrated superior performance on various audio-visual tasks. This study investigates whether pre-trained audio-visual models demonstrate non-arbitrary associations between sounds and visual representations$\unicode{x2013}$known as sound symbolism$\unicode{x2013}$which is also observed in humans. We developed a specialized dataset with synthesized images and audio samples and assessed these models using a non-parametric approach in a zero-shot setting. Our findings reveal a significant correlation between the models' outputs and established patterns of sound symbolism, particularly in models trained on speech data. These results suggest that such models can capture sound-meaning connections akin to human language processing, providing insights into both cognitive architectures and machine learning strategies.
Authors: Genshun Wan, Mengzhi Wang, Tingzhi Mao, Hang Chen, Zhongfu Ye
Abstract: The transducer model trained based on sequence-level criterion requires a lot of memory due to the generation of the large probability matrix. We proposed a lightweight transducer model based on frame-level criterion, which uses the results of the CTC forced alignment algorithm to determine the label for each frame. Then the encoder output can be combined with the decoder output at the corresponding time, rather than adding each element output by the encoder to each element output by the decoder as in the transducer. This significantly reduces memory and computation requirements. To address the problem of imbalanced classification caused by excessive blanks in the label, we decouple the blank and non-blank probabilities and truncate the gradient of the blank classifier to the main network. Experiments on the AISHELL-1 demonstrate that this enables the lightweight transducer to achieve similar results to transducer. Additionally, we use richer information to predict the probability of blank, achieving superior results to transducer.
Authors: Joshua Ashkinaze, Emily Fry, Narendra Edara, Eric Gilbert, Ceren Budak
Abstract: Recent debates raised concerns that language models may favor certain viewpoints. But what if the solution is not to aim for a 'view from nowhere' but rather to leverage different viewpoints? We introduce Plurals, a system and Python library for pluralistic AI deliberation. Plurals consists of Agents (LLMs, optionally with personas) which deliberate within customizable Structures, with Moderators overseeing deliberation. Plurals is a generator of simulated social ensembles. Plurals integrates with government datasets to create nationally representative personas, includes deliberation templates inspired by democratic deliberation theory, and allows users to customize both information-sharing structures and deliberation behavior within Structures. Six case studies demonstrate fidelity to theoretical constructs and efficacy. Three randomized experiments show simulated focus groups produced output resonant with an online sample of the relevant audiences (chosen over zero-shot generation in 75% of trials). Plurals is both a paradigm and a concrete system for pluralistic AI. The Plurals library is available at https://github.com/josh-ashkinaze/plurals and will be continually updated.
Authors: Xuyuan Xiong, Simeng Han, Ziyue Zhou, Arman Cohan
Abstract: Large Language Models (LLMs) are commonly used to generate solutions for mathematical reasoning problems in the following formats: natural language, code, or a combination of both. In this paper, we explore fundamental questions related to solving mathematical reasoning problems using natural language and code with state-of-the-art LLMs, including GPT-4o-mini and LLama-3.1-8b-Turbo. Our findings show that LLMs are better at reasoning in natural language compared to code. Additionally, although natural language and code serve as complementary forms of reasoning, they can affect each other in a negative way in certain scenarios. These insights motivate our development of a new prompting method, INC-Math, which leverages an LLM to dynamically select the most appropriate reasoning form, resulting in improved performance over comparable baselines with GPT-4o-mini.
Authors: Haoyu Zhang, Wenbin Wang, Tianshu Yu
Abstract: The field of Multimodal Sentiment Analysis (MSA) has recently witnessed an emerging direction seeking to tackle the issue of data incompleteness. Recognizing that the language modality typically contains dense sentiment information, we consider it as the dominant modality and present an innovative Language-dominated Noise-resistant Learning Network (LNLN) to achieve robust MSA. The proposed LNLN features a dominant modality correction (DMC) module and dominant modality based multimodal learning (DMML) module, which enhances the model's robustness across various noise scenarios by ensuring the quality of dominant modality representations. Aside from the methodical design, we perform comprehensive experiments under random data missing scenarios, utilizing diverse and meaningful settings on several popular datasets (\textit{e.g.,} MOSI, MOSEI, and SIMS), providing additional uniformity, transparency, and fairness compared to existing evaluations in the literature. Empirically, LNLN consistently outperforms existing baselines, demonstrating superior performance across these challenging and extensive evaluation metrics.
Authors: Dingzirui Wang, Xuanliang Zhang, Qiguang Chen, Longxu Dou, Xiao Xu, Rongyu Cao, Yingwei Ma, Qingfu Zhu, Wanxiang Che, Binhua Li, Fei Huang, Yongbin Li
Abstract: In-context learning (ICL) is an effective approach to help large language models (LLMs) adapt to various tasks by providing demonstrations of the target task. Considering the high cost of labeling demonstrations, many methods propose synthesizing demonstrations from scratch using LLMs. However, the quality of the demonstrations synthesized from scratch is limited by the capabilities and knowledge of LLMs. To address this, inspired by transfer learning, we propose In-Context Transfer Learning (ICTL), which synthesizes target task demonstrations by transferring labeled demonstrations from similar source tasks. ICTL consists of two steps: source sampling and target transfer. First, we define an optimization objective, which minimizes transfer error to sample source demonstrations similar to the target task. Then, we employ LLMs to transfer the sampled source demonstrations to the target task, matching the definition and format of the target task. Experiments on Super-NI show that ICTL outperforms synthesis from scratch by 2.0% on average, demonstrating the effectiveness of our method.
Authors: Vy Nguyen, Chau Pham
Abstract: The increasing frequency of suicidal thoughts highlights the importance of early detection and intervention. Social media platforms, where users often share personal experiences and seek help, could be utilized to identify individuals at risk. However, the large volume of daily posts makes manual review impractical. This paper explores the use of Large Language Models (LLMs) to automatically detect suicidal content in text-based social media posts. We propose a novel method for generating pseudo-labels for unlabeled data by prompting LLMs, along with traditional classification fine-tuning techniques to enhance label accuracy. To create a strong suicide detection model, we develop an ensemble approach involving prompting with Qwen2-72B-Instruct, and using fine-tuned models such as Llama3-8B, Llama3.1-8B, and Gemma2-9B. We evaluate our approach on the dataset of the Suicide Ideation Detection on Social Media Challenge, a track of the IEEE Big Data 2024 Big Data Cup. Additionally, we conduct a comprehensive analysis to assess the impact of different models and fine-tuning strategies on detection performance. Experimental results show that the ensemble model significantly improves the detection accuracy, by 5% points compared with the individual models. It achieves a weight F1 score of 0.770 on the public test set, and 0.731 on the private test set, providing a promising solution for identifying suicidal content in social media. Our analysis shows that the choice of LLMs affects the prompting performance, with larger models providing better accuracy. Our code and checkpoints are publicly available at https://github.com/khanhvynguyen/Suicide_Detection_LLMs.
URLs: https://github.com/khanhvynguyen/Suicide_Detection_LLMs.
Authors: Zineddine Tighidet, Andrea Mogini, Jiali Mei, Benjamin Piwowarski, Patrick Gallinari
Abstract: Large Language Models (LLMs) often encounter conflicts between their learned, internal (parametric knowledge, PK) and external knowledge provided during inference (contextual knowledge, CK). Understanding how LLMs models prioritize one knowledge source over the other remains a challenge. In this paper, we propose a novel probing framework to explore the mechanisms governing the selection between PK and CK in LLMs. Using controlled prompts designed to contradict the model's PK, we demonstrate that specific model activations are indicative of the knowledge source employed. We evaluate this framework on various LLMs of different sizes and demonstrate that mid-layer activations, particularly those related to relations in the input, are crucial in predicting knowledge source selection, paving the way for more reliable models capable of handling knowledge conflicts effectively.
Authors: Wen Yang, Junhong Wu, Chen Wang, Chengqing Zong, Jiajun Zhang
Abstract: Large Language Models (LLMs) have achieved state-of-the-art performance across numerous tasks. However, these advancements have predominantly benefited "first-class" languages such as English and Chinese, leaving many other languages underrepresented. This imbalance, while limiting broader applications, generates a natural preference ranking between languages, offering an opportunity to bootstrap the multilingual capabilities of LLM in a self-improving manner. Thus, we propose $\textit{Language Imbalance Driven Rewarding}$, where the inherent imbalance between dominant and non-dominant languages within LLMs is leveraged as a reward signal. Iterative DPO training demonstrates that this approach not only enhances LLM performance in non-dominant languages but also improves the dominant language's capacity, thereby yielding an iterative reward signal. Fine-tuning Meta-Llama-3-8B-Instruct over two iterations of this approach results in continuous improvements in multilingual performance across instruction-following and arithmetic reasoning tasks, evidenced by an average improvement of 7.46% win rate on the X-AlpacaEval leaderboard and 13.9% accuracy on the MGSM benchmark. This work serves as an initial exploration, paving the way for multilingual self-improvement of LLMs.
Authors: Zihan Chen, Bike Xie, Jundong Li, Cong Shen
Abstract: Large Language Models (LLMs) have demonstrated remarkable success across a wide range of language tasks, but their deployment on edge devices remains challenging due to the substantial memory requirements imposed by their large parameter sizes. Weight-only quantization presents a promising solution to reduce the memory footprint of LLMs. However, existing approaches primarily focus on integer-bit quantization, limiting their adaptability to fractional-bit quantization tasks and preventing the full utilization of available storage space on devices. In this paper, we introduce Channel-Wise Mixed-Precision Quantization (CMPQ), a novel mixed-precision quantization method that allocates quantization precision in a channel-wise pattern based on activation distributions. By assigning different precision levels to different weight channels, CMPQ can adapt to any bit-width constraint. CMPQ employs a non-uniform quantization strategy and incorporates two outlier extraction techniques that collaboratively preserve the critical information, thereby minimizing the quantization loss. Experiments on different sizes of LLMs demonstrate that CMPQ not only enhances performance in integer-bit quantization tasks but also achieves significant performance gains with a modest increase in memory usage. CMPQ thus represents an adaptive and effective approach to LLM quantization, offering substantial benefits across diverse device capabilities.
Authors: Wei Jie Yeo, Ranjan Satapathy, Erik Cambria
Abstract: Large Language Models (LLMs) are capable of generating persuasive Natural Language Explanations (NLEs) to justify their answers. However, the faithfulness of these explanations should not be readily trusted at face value. Recent studies have proposed various methods to measure the faithfulness of NLEs, typically by inserting perturbations at the explanation or feature level. We argue that these approaches are neither comprehensive nor correctly designed according to the established definition of faithfulness. Moreover, we highlight the risks of grounding faithfulness findings on out-of-distribution samples. In this work, we leverage a causal mediation technique called activation patching, to measure the faithfulness of an explanation towards supporting the explained answer. Our proposed metric, Causal Faithfulness quantifies the consistency of causal attributions between explanations and the corresponding model outputs as the indicator of faithfulness. We experimented across models varying from 2B to 27B parameters and found that models that underwent alignment tuning tend to produce more faithful and plausible explanations. We find that Causal Faithfulness is a promising improvement over existing faithfulness tests by taking into account the model's internal computations and avoiding out of distribution concerns that could otherwise undermine the validity of faithfulness assessments. We release the code in \url{https://github.com/wj210/Causal-Faithfulness}
Authors: Qingfei Zhao, Ruobing Wang, Yukuo Cen, Daren Zha, Shicheng Tan, Yuxiao Dong, Jie Tang
Abstract: Long-Context Question Answering (LCQA), a challenging task, aims to reason over long-context documents to yield accurate answers to questions. Existing long-context Large Language Models (LLMs) for LCQA often struggle with the "lost in the middle" issue. Retrieval-Augmented Generation (RAG) mitigates this issue by providing external factual evidence. However, its chunking strategy disrupts the global long-context information, and its low-quality retrieval in long contexts hinders LLMs from identifying effective factual details due to substantial noise. To this end, we propose LongRAG, a general, dual-perspective, and robust LLM-based RAG system paradigm for LCQA to enhance RAG's understanding of complex long-context knowledge (i.e., global information and factual details). We design LongRAG as a plug-and-play paradigm, facilitating adaptation to various domains and LLMs. Extensive experiments on three multi-hop datasets demonstrate that LongRAG significantly outperforms long-context LLMs (up by 6.94%), advanced RAG (up by 6.16%), and Vanilla RAG (up by 17.25%). Furthermore, we conduct quantitative ablation studies and multi-dimensional analyses, highlighting the effectiveness of the system's components and fine-tuning strategies. Data and code are available at https://github.com/QingFei1/LongRAG.
Authors: Anthony Cui, Pranav Nandyalam, Ethan Cheung, Kevin Zhu
Abstract: Momentum-Aided Prompt Optimization (MAPO) enhances the efficiency and efficacy of prompt optimization for Large Language Models (LLMs). Building on ProTeGi, MAPO uses positive natural language "gradients" and a momentum-based extension to refine prompts effectively. By tracking gradient history, MAPO avoids local minima and oscillations. It also utilizes beam search and an Upper Confidence Bound (UCB) algorithm for balanced candidate expansion and selection. Benchmark testing shows that MAPO achieves faster convergence time with fewer API calls and higher F1 scores than ProTeGi, proving it as a robust and scalable solution for automated prompt engineering in LLMs.
Authors: Luping Wang, Sheng Chen, Linnan Jiang, Shu Pan, Runze Cai, Sen Yang, Fei Yang
Abstract: The large models, as predicted by scaling raw forecasts, have made groundbreaking progress in many fields, particularly in natural language generation tasks, where they have approached or even surpassed human levels. However, the unprecedented scale of their parameters brings significant computational and storage costs. These large models require substantial computational resources and GPU memory to operate. When adapting large models to specific downstream tasks, their massive parameter scale poses a significant challenge in fine-tuning on hardware platforms with limited computational power and GPU memory. To address this issue, Parameter-Efficient Fine-Tuning (PEFT) offers a practical solution by efficiently adjusting the parameters of large pre-trained models to suit various downstream tasks. Specifically, PEFT adjusts the parameters of pre-trained large models to adapt to specific tasks or domains, minimizing the introduction of additional parameters and the computational resources required. This review mainly introduces the preliminary knowledge of PEFT, the core ideas and principles of various PEFT algorithms, the applications of PEFT, and potential future research directions. By reading this review, we believe that interested parties can quickly grasp the PEFT methodology, thereby accelerating its development and innovation.
Authors: Hanshi Sun, Momin Haider, Ruiqi Zhang, Huitao Yang, Jiahao Qiu, Ming Yin, Mengdi Wang, Peter Bartlett, Andrea Zanette
Abstract: The safe and effective deployment of Large Language Models (LLMs) involves a critical step called alignment, which ensures that the model's responses are in accordance with human preferences. Prevalent alignment techniques, such as DPO, PPO and their variants, align LLMs by changing the pre-trained model weights during a phase called post-training. While predominant, these post-training methods add substantial complexity before LLMs can be deployed. Inference-time alignment methods avoid the complex post-training step and instead bias the generation towards responses that are aligned with human preferences. The best-known inference-time alignment method, called Best-of-N, is as effective as the state-of-the-art post-training procedures. Unfortunately, Best-of-N requires vastly more resources at inference time than standard decoding strategies, which makes it computationally not viable. In this work, we introduce Speculative Rejection, a computationally-viable inference-time alignment algorithm. It generates high-scoring responses according to a given reward model, like Best-of-N does, while being between 16 to 32 times more computationally efficient.
Authors: Jonas Becker
Abstract: In an era where single large language models have dominated the landscape of artificial intelligence for years, multi-agent systems arise as new protagonists in conversational task-solving. While previous studies have showcased their potential in reasoning tasks and creative endeavors, an analysis of their limitations concerning the conversational paradigms and the impact of individual agents is missing. It remains unascertained how multi-agent discussions perform across tasks of varying complexity and how the structure of these conversations influences the process. To fill that gap, this work systematically evaluates multi-agent systems across various discussion paradigms, assessing their strengths and weaknesses in both generative tasks and question-answering tasks. Alongside the experiments, I propose a taxonomy of 20 multi-agent research studies from 2022 to 2024, followed by the introduction of a framework for deploying multi-agent LLMs in conversational task-solving. I demonstrate that while multi-agent systems excel in complex reasoning tasks, outperforming a single model by leveraging expert personas, they fail on basic tasks. Concretely, I identify three challenges that arise: 1) While longer discussions enhance reasoning, agents fail to maintain conformity to strict task requirements, which leads to problem drift, making shorter conversations more effective for basic tasks. 2) Prolonged discussions risk alignment collapse, raising new safety concerns for these systems. 3) I showcase discussion monopolization through long generations, posing the problem of fairness in decision-making for tasks like summarization. This work uncovers both the potential and challenges that arise with multi-agent interaction and varying conversational paradigms, providing insights into how future research could improve the efficiency, performance, and safety of multi-agent LLMs.
Authors: Hajar Sakai, Sarah S. Lam, Mohammadsadegh Mikaeili, Joshua Bosire, Franziska Jovin
Abstract: Patient experience and care quality are crucial for a hospital's sustainability and reputation. The analysis of patient feedback offers valuable insight into patient satisfaction and outcomes. However, the unstructured nature of these comments poses challenges for traditional machine learning methods following a supervised learning paradigm. This is due to the unavailability of labeled data and the nuances these texts encompass. This research explores leveraging Large Language Models (LLMs) in conducting Multi-label Text Classification (MLTC) of inpatient comments shared after a stay in the hospital. GPT-4 Turbo was leveraged to conduct the classification. However, given the sensitive nature of patients' comments, a security layer is introduced before feeding the data to the LLM through a Protected Health Information (PHI) detection framework, which ensures patients' de-identification. Additionally, using the prompt engineering framework, zero-shot learning, in-context learning, and chain-of-thought prompting were experimented with. Results demonstrate that GPT-4 Turbo, whether following a zero-shot or few-shot setting, outperforms traditional methods and Pre-trained Language Models (PLMs) and achieves the highest overall performance with an F1-score of 76.12% and a weighted F1-score of 73.61% followed closely by the few-shot learning results. Subsequently, the results' association with other patient experience structured variables (e.g., rating) was conducted. The study enhances MLTC through the application of LLMs, offering healthcare practitioners an efficient method to gain deeper insights into patient feedback and deliver prompt, appropriate responses.
Authors: Dominic Sobhani, Amir Feder, David Blei
Abstract: Probabilistic topic models are a powerful tool for extracting latent themes from large text datasets. In many text datasets, we also observe per-document covariates (e.g., source, style, political affiliation) that act as environments that modulate a "global" (environment-agnostic) topic representation. Accurately learning these representations is important for prediction on new documents in unseen environments and for estimating the causal effect of topics on real-world outcomes. To this end, we introduce the Multi-environment Topic Model (MTM), an unsupervised probabilistic model that separates global and environment-specific terms. Through experimentation on various political content, from ads to tweets and speeches, we show that the MTM produces interpretable global topics with distinct environment-specific words. On multi-environment data, the MTM outperforms strong baselines in and out-of-distribution. It also enables the discovery of accurate causal effects.
Authors: Yuxiang Wei, Federico Cassano, Jiawei Liu, Yifeng Ding, Naman Jain, Zachary Mueller, Harm de Vries, Leandro von Werra, Arjun Guha, Lingming Zhang
Abstract: Instruction tuning is a supervised fine-tuning approach that significantly improves the ability of large language models (LLMs) to follow human instructions. We propose SelfCodeAlign, the first fully transparent and permissive pipeline for self-aligning code LLMs without extensive human annotations or distillation. SelfCodeAlign employs the same base model for inference throughout the data generation process. It first extracts diverse coding concepts from high-quality seed snippets to generate new tasks. It then samples multiple responses per task, pairs each with test cases, and validates them in a sandbox environment. Finally, passing examples are selected for instruction tuning. In our primary experiments, we use SelfCodeAlign with CodeQwen1.5-7B to generate a dataset of 74k instruction-response pairs. Finetuning on this dataset leads to a model that achieves a 67.1 pass@1 on HumanEval+, surpassing CodeLlama-70B-Instruct despite being ten times smaller. Across all benchmarks, this finetuned model consistently outperforms the original version trained with OctoPack, the previous state-of-the-art method for instruction tuning without human annotations or distillation. Additionally, we show that SelfCodeAlign is effective across LLMs of various sizes, from 3B to 33B, and that the base models can benefit more from alignment with their own data distribution. We further validate each component's effectiveness in our pipeline, showing that SelfCodeAlign outperforms both direct distillation from GPT-4o and leading GPT-3.5-based distillation methods, such as OSS-Instruct and Evol-Instruct. SelfCodeAlign has also led to the creation of StarCoder2-Instruct, the first fully transparent, permissively licensed, and self-aligned code LLM that achieves state-of-the-art coding performance.
Authors: Zhi Jing, Yongye Su, Yikun Han
Abstract: This survey explores the synergistic potential of Large Language Models (LLMs) and Vector Databases (VecDBs), a burgeoning but rapidly evolving research area. With the proliferation of LLMs comes a host of challenges, including hallucinations, outdated knowledge, prohibitive commercial application costs, and memory issues. VecDBs emerge as a compelling solution to these issues by offering an efficient means to store, retrieve, and manage the high-dimensional vector representations intrinsic to LLM operations. Through this nuanced review, we delineate the foundational principles of LLMs and VecDBs and critically analyze their integration's impact on enhancing LLM functionalities. This discourse extends into a discussion on the speculative future developments in this domain, aiming to catalyze further research into optimizing the confluence of LLMs and VecDBs for advanced data handling and knowledge extraction capabilities.
Authors: Chengxing Xie, Canyu Chen, Feiran Jia, Ziyu Ye, Shiyang Lai, Kai Shu, Jindong Gu, Adel Bibi, Ziniu Hu, David Jurgens, James Evans, Philip Torr, Bernard Ghanem, Guohao Li
Abstract: Large Language Model (LLM) agents have been increasingly adopted as simulation tools to model humans in social science and role-playing applications. However, one fundamental question remains: can LLM agents really simulate human behavior? In this paper, we focus on one critical and elemental behavior in human interactions, trust, and investigate whether LLM agents can simulate human trust behavior. We first find that LLM agents generally exhibit trust behavior, referred to as agent trust, under the framework of Trust Games, which are widely recognized in behavioral economics. Then, we discover that GPT-4 agents manifest high behavioral alignment with humans in terms of trust behavior, indicating the feasibility of simulating human trust behavior with LLM agents. In addition, we probe the biases of agent trust and differences in agent trust towards other LLM agents and humans. We also explore the intrinsic properties of agent trust under conditions including external manipulations and advanced reasoning strategies. Our study provides new insights into the behaviors of LLM agents and the fundamental analogy between LLMs and humans beyond value alignment. We further illustrate broader implications of our discoveries for applications where trust is paramount.
Authors: Weilin Cai, Juyong Jiang, Le Qin, Junwei Cui, Sunghun Kim, Jiayi Huang
Abstract: Expert parallelism has been introduced as a strategy to distribute the computational workload of sparsely-gated mixture-of-experts (MoE) models across multiple computing devices, facilitating the execution of these increasingly large-scale models. However, the All-to-All communication intrinsic to expert parallelism constitutes a significant overhead, diminishing the MoE models' efficiency. Current optimization approaches offer some relief, yet they are constrained by the sequential interdependence of communication and computation operations. To address this limitation, we present a novel shortcut-connected MoE (ScMoE) architecture with an overlapping parallel strategy, which effectively decouples communication from its conventional sequence, allowing for a substantial overlap of 70% to 100% with computation. When compared with the prevalent top-2 MoE architecture, ScMoE demonstrates training speed improvements of 30% and 11%, and inference improvements of 40% and 15%, in our distributed environments with PCIe and NVLink hardware, respectively, where communication constitutes 60% and 15% of the total MoE time consumption. Building on the ScMoE architecture, we further implement an expert offloading strategy to facilitate memory-limited inference, optimizing latency through the overlap of expert migration. Additionally, extensive experiments and theoretical analyses indicate that ScMoE not only achieves comparable but in some instances surpasses the model quality of existing approaches.
Authors: Yihao Zhang, Zeming Wei, Jun Sun, Meng Sun
Abstract: Since the rapid development of Large Language Models (LLMs) has achieved remarkable success, understanding and rectifying their internal complex mechanisms has become an urgent issue. Recent research has attempted to interpret their behaviors through the lens of inner representation. However, developing practical and efficient methods for applying these representations for general and flexible model editing remains challenging. In this work, we explore how to leverage insights from representation engineering to guide the editing of LLMs by deploying a representation sensor as an editing oracle. We first identify the importance of a robust and reliable sensor during editing, then propose an Adversarial Representation Engineering (ARE) framework to provide a unified and interpretable approach for conceptual model editing without compromising baseline performance. Experiments on multiple tasks demonstrate the effectiveness of ARE in various model editing scenarios. Our code and data are available at https://github.com/Zhang-Yihao/Adversarial-Representation-Engineering.
URLs: https://github.com/Zhang-Yihao/Adversarial-Representation-Engineering.
Authors: Hui Zheng, Hai-Teng Wang, Wei-Bang Jiang, Zhong-Tao Chen, Li He, Pei-Yang Lin, Peng-Hu Wei, Guo-Guang Zhao, Yun-Zhe Liu
Abstract: Invasive brain-computer interfaces with Electrocorticography (ECoG) have shown promise for high-performance speech decoding in medical applications, but less damaging methods like intracranial stereo-electroencephalography (sEEG) remain underexplored. With rapid advances in representation learning, leveraging abundant recordings to enhance speech decoding is increasingly attractive. However, popular methods often pre-train temporal models based on brain-level tokens, overlooking that brain activities in different regions are highly desynchronized during tasks. Alternatively, they pre-train spatial-temporal models based on channel-level tokens but fail to evaluate them on challenging tasks like speech decoding, which requires intricate processing in specific language-related areas. To address this issue, we collected a well-annotated Chinese word-reading sEEG dataset targeting language-related brain networks from 12 subjects. Using this benchmark, we developed the Du-IN model, which extracts contextual embeddings based on region-level tokens through discrete codex-guided mask modeling. Our model achieves state-of-the-art performance on the 61-word classification task, surpassing all baselines. Model comparisons and ablation studies reveal that our design choices, including (i) temporal modeling based on region-level tokens by utilizing 1D depthwise convolution to fuse channels in the ventral sensorimotor cortex (vSMC) and superior temporal gyrus (STG) and (ii) self-supervision through discrete codex-guided mask modeling, significantly contribute to this performance. Overall, our approach -- inspired by neuroscience findings and capitalizing on region-level representations from specific brain regions -- is suitable for invasive brain modeling and represents a promising neuro-inspired AI approach in brain-computer interfaces.
Authors: Minghao Chen, Yihang Li, Yanting Yang, Shiyu Yu, Binbin Lin, Xiaofei He
Abstract: Large Language Models (LLM) based agents have shown promise in autonomously completing tasks across various domains, e.g., robotics, games, and web navigation. However, these agents typically require elaborate design and expert prompts to solve tasks in specific domains, which limits their adaptability. We introduce AutoManual, a framework enabling LLM agents to autonomously build their understanding through interaction and adapt to new environments. AutoManual categorizes environmental knowledge into diverse rules and optimizes them in an online fashion by two agents: 1) The Planner codes actionable plans based on current rules for interacting with the environment. 2) The Builder updates the rules through a well-structured rule system that facilitates online rule management and essential detail retention. To mitigate hallucinations in managing rules, we introduce a *case-conditioned prompting* strategy for the Builder. Finally, the Formulator agent compiles these rules into a comprehensive manual. The self-generated manual can not only improve the adaptability but also guide the planning of smaller LLMs while being human-readable. Given only one simple demonstration, AutoManual significantly improves task success rates, achieving 97.4\% with GPT-4-turbo and 86.2\% with GPT-3.5-turbo on ALFWorld benchmark tasks. The code is available at https://github.com/minghchen/automanual.
Authors: ZeMing Gong, Austin T. Wang, Xiaoliang Huo, Joakim Bruslund Haurum, Scott C. Lowe, Graham W. Taylor, Angel X. Chang
Abstract: Measuring biodiversity is crucial for understanding ecosystem health. While prior works have developed machine learning models for taxonomic classification of photographic images and DNA separately, in this work, we introduce a multimodal approach combining both, using CLIP-style contrastive learning to align images, barcode DNA, and text-based representations of taxonomic labels in a unified embedding space. This allows for accurate classification of both known and unknown insect species without task-specific fine-tuning, leveraging contrastive learning for the first time to fuse DNA and image data. Our method surpasses previous single-modality approaches in accuracy by over 8% on zero-shot learning tasks, showcasing its effectiveness in biodiversity studies.
Authors: Jaehong Yoon, Shoubin Yu, Mohit Bansal
Abstract: Recent video generative models primarily rely on carefully written text prompts for specific tasks, like inpainting or style editing. They require labor-intensive textual descriptions for input videos, hindering their flexibility to adapt personal/raw videos to user specifications. This paper proposes RACCooN, a versatile and user-friendly video-to-paragraph-to-video generative framework that supports multiple video editing capabilities such as removal, addition, and modification, through a unified pipeline. RACCooN consists of two principal stages: Video-to-Paragraph (V2P) and Paragraph-to-Video (P2V). In the V2P stage, we automatically describe video scenes in well-structured natural language, capturing both the holistic context and focused object details. Subsequently, in the P2V stage, users can optionally refine these descriptions to guide the video diffusion model, enabling various modifications to the input video, such as removing, changing subjects, and/or adding new objects. The proposed approach stands out from other methods through several significant contributions: (1) RACCooN suggests a multi-granular spatiotemporal pooling strategy to generate well-structured video descriptions, capturing both the broad context and object details without requiring complex human annotations, simplifying precise video content editing based on text for users. (2) Our video generative model incorporates auto-generated narratives or instructions to enhance the quality and accuracy of the generated content. (3) RACCooN also plans to imagine new objects in a given video, so users simply prompt the model to receive a detailed video editing plan for complex video editing. The proposed framework demonstrates impressive versatile capabilities in video-to-paragraph generation, video content editing, and can be incorporated into other SoTA video generative models for further enhancement.
Authors: Qidong Liu, Xian Wu, Yejing Wang, Zijian Zhang, Feng Tian, Yefeng Zheng, Xiangyu Zhao
Abstract: Sequential recommender systems (SRS) aim to predict users' subsequent choices based on their historical interactions and have found applications in diverse fields such as e-commerce and social media. However, in real-world systems, most users interact with only a handful of items, while the majority of items are seldom consumed. These two issues, known as the long-tail user and long-tail item challenges, often pose difficulties for existing SRS. These challenges can adversely affect user experience and seller benefits, making them crucial to address. Though a few works have addressed the challenges, they still struggle with the seesaw or noisy issues due to the intrinsic scarcity of interactions. The advancements in large language models (LLMs) present a promising solution to these problems from a semantic perspective. As one of the pioneers in this field, we propose the Large Language Models Enhancement framework for Sequential Recommendation (LLM-ESR). This framework utilizes semantic embeddings derived from LLMs to enhance SRS without adding extra inference load from LLMs. To address the long-tail item challenge, we design a dual-view modeling framework that combines semantics from LLMs and collaborative signals from conventional SRS. For the long-tail user challenge, we propose a retrieval augmented self-distillation method to enhance user preference representation using more informative interactions from similar users. To verify the effectiveness and versatility of our proposed enhancement framework, we conduct extensive experiments on three real-world datasets using three popular SRS models. The results show that our method surpasses existing baselines consistently, and benefits long-tail users and items especially. The implementation code is available at https://github.com/Applied-Machine-Learning-Lab/LLM-ESR.
URLs: https://github.com/Applied-Machine-Learning-Lab/LLM-ESR.
Authors: Songlin Yang, Bailin Wang, Yu Zhang, Yikang Shen, Yoon Kim
Abstract: Transformers with linear attention (i.e., linear transformers) and state-space models have recently been suggested as a viable linear-time alternative to transformers with softmax attention. However, these models still underperform transformers especially on tasks that require in-context retrieval. While more expressive variants of linear transformers which replace the additive update in linear transformers with the delta rule (DeltaNet) have been found to be more effective at associative recall, existing algorithms for training such models do not parallelize over sequence length and are thus inefficient to train on modern hardware. This work describes a hardware-efficient algorithm for training linear transformers with the delta rule, which exploits a memory-efficient representation for computing products of Householder matrices. This algorithm allows us to scale up DeltaNet to standard language modeling settings. We train a 1.3B model for 100B tokens and find that it outperforms recent linear-time baselines such as Mamba and GLA in terms of perplexity and zero-shot performance on downstream tasks. We also experiment with two hybrid models which combine DeltaNet layers with (1) sliding-window attention layers every other layer or (2) two global attention layers, and find that these hybrids outperform strong transformer baselines.
Authors: Waris Quamer, Ricardo Gutierrez-Osuna
Abstract: Speaker anonymization aims to conceal cues to speaker identity while preserving linguistic content. Current machine learning based approaches require substantial computational resources, hindering real-time streaming applications. To address these concerns, we propose a streaming model that achieves speaker anonymization with low latency. The system is trained in an end-to-end autoencoder fashion using a lightweight content encoder that extracts HuBERT-like information, a pretrained speaker encoder that extract speaker identity, and a variance encoder that injects pitch and energy information. These three disentangled representations are fed to a decoder that re-synthesizes the speech signal. We present evaluation results from two implementations of our system, a full model that achieves a latency of 230ms, and a lite version (0.1x in size) that further reduces latency to 66ms while maintaining state-of-the-art performance in naturalness, intelligibility, and privacy preservation.
Authors: Stefan K. Nielsen, Laziz U. Abdullaev, Rachel S. Y. Teo, Tan M. Nguyen
Abstract: Pairwise dot-product self-attention is key to the success of transformers that achieve state-of-the-art performance across a variety of applications in language and vision. This dot-product self-attention computes attention weights among the input tokens using Euclidean distance, which makes the model prone to representation collapse and vulnerable to contaminated samples. In this paper, we propose using a Mahalanobis distance metric for computing the attention weights to stretch the underlying feature space in directions of high contextual relevance. In particular, we define a hyper-ellipsoidal neighborhood around each query to increase the attention weights of the tokens lying in the contextually important directions. We term this novel class of attention Elliptical Attention. Our Elliptical Attention provides two benefits: 1) reducing representation collapse and 2) enhancing the model's robustness as Elliptical Attention pays more attention to contextually relevant information rather than focusing on some small subset of informative features. We empirically demonstrate the advantages of Elliptical Attention over the baseline dot-product attention and state-of-the-art attention methods on various practical tasks, including object classification, image segmentation, and language modeling across different data modalities.
Authors: Tianyu Fu, Haofeng Huang, Xuefei Ning, Genghan Zhang, Boju Chen, Tianqi Wu, Hongyi Wang, Zixiao Huang, Shiyao Li, Shengen Yan, Guohao Dai, Huazhong Yang, Yu Wang
Abstract: Sparse attention can effectively mitigate the significant memory and throughput demands of Large Language Models (LLMs) in long contexts. Existing methods typically employ a uniform sparse attention mask, applying the same sparse pattern across different attention heads and input lengths. However, this uniform approach fails to capture the diverse attention patterns inherent in LLMs, ignoring their distinct accuracy-latency trade-offs. To address this challenge, we propose the Mixture of Attention (MoA), which automatically tailors distinct sparse attention configurations to different heads and layers. MoA constructs and navigates a search space of various attention patterns and their scaling rules relative to input sequence lengths. It profiles the model, evaluates potential configurations, and pinpoints the optimal sparse attention compression plan. MoA adapts to varying input sizes, revealing that some attention heads expand their focus to accommodate longer sequences, while other heads consistently concentrate on fixed-length local contexts. Experiments show that MoA increases the effective context length by $3.9\times$ with the same average attention span, boosting retrieval accuracy by $1.5-7.1\times$ over the uniform-attention baseline across Vicuna-{7B,13B}, and Llama3-{8B,70B} models. Moreover, MoA narrows the capability gaps between sparse and dense models, reducing the maximum relative performance drop from $9\%-36\%$ to within $5\%$ across two long-context understanding benchmarks. MoA achieves a $1.2-1.4\times$ GPU memory reduction, boosting decode throughput by $6.6-8.2\times$ and $1.7-1.9\times$ compared to FlashAttention2 and vLLM, with minimal impact on performance. Our code is available at \url{https://github.com/thu-nics/MoA}.
Authors: Xuweiyi Chen, Ziqiao Ma, Xuejun Zhang, Sihan Xu, Shengyi Qian, Jianing Yang, David F. Fouhey, Joyce Chai
Abstract: Large vision language models (LVLMs) often suffer from object hallucination, producing objects not present in the given images. While current benchmarks for object hallucination primarily concentrate on the presence of a single object class rather than individual entities, this work systematically investigates multi-object hallucination, examining how models misperceive (e.g., invent nonexistent objects or become distracted) when tasked with focusing on multiple objects simultaneously. We introduce Recognition-based Object Probing Evaluation (ROPE), an automated evaluation protocol that considers the distribution of object classes within a single image during testing and uses visual referring prompts to eliminate ambiguity. With comprehensive empirical studies and analysis of potential factors leading to multi-object hallucination, we found that (1). LVLMs suffer more hallucinations when focusing on multiple objects compared to a single object. (2). The tested object class distribution affects hallucination behaviors, indicating that LVLMs may follow shortcuts and spurious correlations. (3). Hallucinatory behaviors are influenced by data-specific factors, salience and frequency, and model intrinsic behaviors. We hope to enable LVLMs to recognize and reason about multiple objects that often occur in realistic visual scenes, provide insights, and quantify our progress towards mitigating the issues.
Authors: Fengyu Cai, Xinran Zhao, Tong Chen, Sihao Chen, Hongming Zhang, Iryna Gurevych, Heinz Koeppl
Abstract: Recent studies show the growing significance of document retrieval in the generation of LLMs, i.e., RAG, within the scientific domain by bridging their knowledge gap. However, dense retrievers often struggle with domain-specific retrieval and complex query-document relationships, particularly when query segments correspond to various parts of a document. To alleviate such prevalent challenges, this paper introduces $\texttt{MixGR}$, which improves dense retrievers' awareness of query-document matching across various levels of granularity in queries and documents using a zero-shot approach. $\texttt{MixGR}$ fuses various metrics based on these granularities to a united score that reflects a comprehensive query-document similarity. Our experiments demonstrate that $\texttt{MixGR}$ outperforms previous document retrieval by 24.7%, 9.8%, and 6.9% on nDCG@5 with unsupervised, supervised, and LLM-based retrievers, respectively, averaged on queries containing multiple subqueries from five scientific retrieval datasets. Moreover, the efficacy of two downstream scientific question-answering tasks highlights the advantage of $\texttt{MixGR}$ to boost the application of LLMs in the scientific domain. The code and experimental datasets are available.
Authors: Fuzhao Xue, Yukang Chen, Dacheng Li, Qinghao Hu, Ligeng Zhu, Xiuyu Li, Yunhao Fang, Haotian Tang, Shang Yang, Zhijian Liu, Ethan He, Hongxu Yin, Pavlo Molchanov, Jan Kautz, Linxi Fan, Yuke Zhu, Yao Lu, Song Han
Abstract: Long-context capability is critical for multi-modal foundation models, especially for long video understanding. We introduce LongVILA, a full-stack solution for long-context visual-language models by co-designing the algorithm and system. For model training, we upgrade existing VLMs to support long video understanding by incorporating two additional stages, i.e., long context extension and long video supervised fine-tuning. However, training on long video is computationally and memory intensive. We introduce the long-context Multi-Modal Sequence Parallelism (MM-SP) system that efficiently parallelizes long video training and inference, enabling 2M context length training on 256 GPUs without any gradient checkpointing. LongVILA efficiently extends the number of video frames of VILA from 8 to 2048, improving the long video captioning score from 2.00 to 3.26 (out of 5), achieving 99.8% accuracy in 6,000-frame (more than 1 million tokens) video needle-in-a-haystack. LongVILA-7B demonstrates strong accuracy on the VideoMME benchmark, i.e., 61.8% with subtitle. Besides, MM-SP is 2.1x - 5.7x faster than ring style sequence parallelism and 1.1x - 1.4x faster than Megatron with a hybrid context and tensor parallelism. Moreover, it seamlessly integrates with Hugging Face Transformers.
Authors: Elvis Hsieh, Preston Fu, Jonathan Chen
Abstract: Language models (LMs) trained on web-scale datasets are largely successful due to their ability to memorize large amounts of training data, even if only present in a few examples. These capabilities are often desirable in evaluation on tasks such as question answering but raise questions about whether these models can exhibit genuine reasoning or succeed only at mimicking patterns from the training data. This distinction is particularly salient in forecasting tasks, where the answer is not present in the training data, and the model must reason to make logical deductions. We present Reasoning and Tools for Forecasting (RTF), a framework of reasoning-and-acting (ReAct) agents that can dynamically retrieve updated information and run numerical simulation with equipped tools. We evaluate our model with questions from competitive forecasting platforms and demonstrate that our method is competitive with and can outperform human predictions. This suggests that LMs, with the right tools, can indeed think and adapt like humans, offering valuable insights for real-world decision-making.
Authors: Guanwen Xie, Jingzehua Xu, Yiyuan Yang, Yimian Ding, Shuai Zhang
Abstract: Achieving the effective design and improvement of reward functions in reinforcement learning (RL) tasks with complex custom environments and multiple requirements presents considerable challenges. In this paper, we propose ERFSL, an efficient reward function searcher using LLMs, which enables LLMs to be effective white-box searchers and highlights their advanced semantic understanding capabilities. Specifically, we generate reward components for each numerically explicit user requirement and employ a reward critic to identify the correct code form. Then, LLMs assign weights to the reward components to balance their values and iteratively adjust the weights without ambiguity and redundant adjustments by flexibly adopting directional mutation and crossover strategies, similar to genetic algorithms, based on the context provided by the training log analyzer. We applied the framework to an underwater data collection RL task without direct human feedback or reward examples (zero-shot learning). The reward critic successfully corrects the reward code with only one feedback instance for each requirement, effectively preventing unrectifiable errors. The initialization of weights enables the acquisition of different reward functions within the Pareto solution set without the need for weight search. Even in cases where a weight is 500 times off, on average, only 5.2 iterations are needed to meet user requirements. The ERFSL also works well with most prompts utilizing GPT-4o mini, as we decompose the weight searching process to reduce the requirement for numerical and long-context understanding capabilities
Authors: Kuldeep Singh, Simerjot Kaur, Charese Smiley
Abstract: Financial decision-making hinges on the analysis of relevant information embedded in the enormous volume of documents in the financial domain. To address this challenge, we developed FinQAPT, an end-to-end pipeline that streamlines the identification of relevant financial reports based on a query, extracts pertinent context, and leverages Large Language Models (LLMs) to perform downstream tasks. To evaluate the pipeline, we experimented with various techniques to optimize the performance of each module using the FinQA dataset. We introduced a novel clustering-based negative sampling technique to enhance context extraction and a novel prompting method called Dynamic N-shot Prompting to boost the numerical question-answering capabilities of LLMs. At the module level, we achieved state-of-the-art accuracy on FinQA, attaining an accuracy of 80.6%. However, at the pipeline level, we observed decreased performance due to challenges in extracting relevant context from financial reports. We conducted a detailed error analysis of each module and the end-to-end pipeline, pinpointing specific challenges that must be addressed to develop a robust solution for handling complex financial tasks.
Authors: Fei Liu, Yiming Yao, Ping Guo, Zhiyuan Yang, Zhe Zhao, Xi Lin, Xialiang Tong, Mingxuan Yuan, Zhichao Lu, Zhenkun Wang, Qingfu Zhang
Abstract: Algorithm Design (AD) is crucial for effective problem-solving across various domains. The advent of Large Language Models (LLMs) has notably enhanced the automation and innovation within this field, offering new perspectives and promising solutions. Over the past three years, the integration of LLMs into AD (LLM4AD) has seen substantial progress, with applications spanning optimization, machine learning, mathematical reasoning, and scientific discovery. Given the rapid advancements and expanding scope of this field, a systematic review is both timely and necessary. This paper provides a systematic review of LLM4AD. First, we offer an overview and summary of existing studies. Then, we introduce a taxonomy and review the literature across four dimensions: the roles of LLMs, search methods, prompt methods, and application domains with a discussion of potential and achievements of LLMs in AD. Finally, we identify current challenges and highlight several promising directions for future research.
Authors: Mingan Lin, Fan Yang, Yanjun Shen, Haoze Sun, Tianpeng Li, Tao Zhang, Chenzheng Zhu, Tao Zhang, Miao Zheng, Xu Li, Yijie Zhou, Mingyang Chen, Yanzhao Qin, Youquan Li, Hao Liang, Fei Li, Yadong Li, Mang Wang, Guosheng Dong, Kun Fang, Jianhua Xu, Bin Cui, Wentao Zhang, Zenan Zhou, Weipeng Chen
Abstract: We introduce Nova, a suite of practical alignment techniques employed in a series of empirically validated high-performing models. This represents the first comprehensive account of alignment methodologies, offering valuable insights for advancing AI research. We investigate the critical components that enhance model performance during the alignment process, including optimization methods, data strategies, capability enhancements, and evaluation processes. The process spans three key stages: Prompt Augmentation System(PAS), Supervised Fine-Tuning(SFT), and Preference Alignment. The problems encountered, the solutions applied, and the improvements made are thoroughly recorded. Through comparisons across well-established benchmarks, we highlight the technological advancements enabled by Nova Alignment. Importantly, Qwen2-Nova-72B and Llama3-PBM-Nova-70B are instruct versions of the Qwen2-72B and Llama-3-70B base models, optimized through Nova. The Nova models show significant core improvements, with user experience gains of 17% to 28%, and excels on specialized benchmarks. In open-source benchmark evaluations, both Qwen2-Nova-72B and Llama3-PBM-Nova-70B consistently outperform their respective official instruct versions across nearly all datasets. This report aims to clarify the key technologies behind the alignment process, fostering a deeper understanding within the community. Llama3-PBM-Nova-70B model is available at https://huggingface.co/PKU-Baichuan-MLSystemLab/Llama3-PBM-Nova-70B.
URLs: https://huggingface.co/PKU-Baichuan-MLSystemLab/Llama3-PBM-Nova-70B.
Authors: Yining Hong, Beide Liu, Maxine Wu, Yuanhao Zhai, Kai-Wei Chang, Linjie Li, Kevin Lin, Chung-Ching Lin, Jianfeng Wang, Zhengyuan Yang, Yingnian Wu, Lijuan Wang
Abstract: Human beings are endowed with a complementary learning system, which bridges the slow learning of general world dynamics with fast storage of episodic memory from a new experience. Previous video generation models, however, primarily focus on slow learning by pre-training on vast amounts of data, overlooking the fast learning phase crucial for episodic memory storage. This oversight leads to inconsistencies across temporally distant frames when generating longer videos, as these frames fall beyond the model's context window. To this end, we introduce SlowFast-VGen, a novel dual-speed learning system for action-driven long video generation. Our approach incorporates a masked conditional video diffusion model for the slow learning of world dynamics, alongside an inference-time fast learning strategy based on a temporal LoRA module. Specifically, the fast learning process updates its temporal LoRA parameters based on local inputs and outputs, thereby efficiently storing episodic memory in its parameters. We further propose a slow-fast learning loop algorithm that seamlessly integrates the inner fast learning loop into the outer slow learning loop, enabling the recall of prior multi-episode experiences for context-aware skill learning. To facilitate the slow learning of an approximate world model, we collect a large-scale dataset of 200k videos with language action annotations, covering a wide range of scenarios. Extensive experiments show that SlowFast-VGen outperforms baselines across various metrics for action-driven video generation, achieving an FVD score of 514 compared to 782, and maintaining consistency in longer videos, with an average of 0.37 scene cuts versus 0.89. The slow-fast learning loop algorithm significantly enhances performances on long-horizon planning tasks as well. Project Website: https://slowfast-vgen.github.io