Authors: Arpit Aggarwal
Abstract: There are several improvements proposed over the baseline Absolute Positional Encoding (APE) method used in original transformer. In this study, we aim to investigate the implications of inadequately representing positional encoding in higher dimensions on crucial aspects of the attention mechanism, the model's capacity to learn relative positional information, and the convergence of models, all stemming from the choice of sinusoidal basis functions. Through a combination of theoretical insights and empirical analyses, we elucidate how these challenges extend beyond APEs and may adversely affect the performance of Relative Positional Encoding (RPE) methods, such as Rotatory Positional Encoding (RoPE). Subsequently, we introduce an innovative solution termed Orthogonal Polynomial Based Positional Encoding (PoPE) to address some of the limitations associated with existing methods. The PoPE method encodes positional information by leveraging Orthogonal Legendre polynomials. Legendre polynomials as basis functions offers several desirable properties for positional encoding, including improved correlation structure, non-periodicity, orthogonality, and distinct functional forms among polynomials of varying orders. Our experimental findings demonstrate that transformer models incorporating PoPE outperform baseline transformer models on the $Multi30k$ English-to-German translation task, thus establishing a new performance benchmark. Furthermore, PoPE-based transformers exhibit significantly accelerated convergence rates. Additionally, we will present novel theoretical perspectives on position encoding based on the superior performance of PoPE.
Authors: Zhan Su, Yuqin Zhou, Fengran Mo, Jakob Grue Simonsen
Abstract: We propose a novel tensor network language model based on the simplest tensor network (i.e., tensor trains), called `Tensor Train Language Model' (TTLM). TTLM represents sentences in an exponential space constructed by the tensor product of words, but computing the probabilities of sentences in a low-dimensional fashion. We demonstrate that the architectures of Second-order RNNs, Recurrent Arithmetic Circuits (RACs), and Multiplicative Integration RNNs are, essentially, special cases of TTLM. Experimental evaluations on real language modeling tasks show that the proposed variants of TTLM (i.e., TTLM-Large and TTLM-Tiny) outperform the vanilla Recurrent Neural Networks (RNNs) with low-scale of hidden units. (The code is available at https://github.com/shuishen112/tensortrainlm.)
Authors: Siqi Shen, Lajanugen Logeswaran, Moontae Lee, Honglak Lee, Soujanya Poria, Rada Mihalcea
Abstract: Large language models (LLMs) have demonstrated substantial commonsense understanding through numerous benchmark evaluations. However, their understanding of cultural commonsense remains largely unexamined. In this paper, we conduct a comprehensive examination of the capabilities and limitations of several state-of-the-art LLMs in the context of cultural commonsense tasks. Using several general and cultural commonsense benchmarks, we find that (1) LLMs have a significant discrepancy in performance when tested on culture-specific commonsense knowledge for different cultures; (2) LLMs' general commonsense capability is affected by cultural context; and (3) The language used to query the LLMs can impact their performance on cultural-related tasks. Our study points to the inherent bias in the cultural understanding of LLMs and provides insights that can help develop culturally aware language models.
Authors: Emre Can Acikgoz, Mete Erdogan, Deniz Yuret
Abstract: Large Language Models (LLMs) are becoming crucial across various fields, emphasizing the urgency for high-quality models in underrepresented languages. This study explores the unique challenges faced by low-resource languages, such as data scarcity, model selection, evaluation, and computational limitations, with a special focus on Turkish. We conduct an in-depth analysis to evaluate the impact of training strategies, model choices, and data availability on the performance of LLMs designed for underrepresented languages. Our approach includes two methodologies: (i) adapting existing LLMs originally pretrained in English to understand Turkish, and (ii) developing a model from the ground up using Turkish pretraining data, both supplemented with supervised fine-tuning on a novel Turkish instruction-tuning dataset aimed at enhancing reasoning capabilities. The relative performance of these methods is evaluated through the creation of a new leaderboard for Turkish LLMs, featuring benchmarks that assess different reasoning and knowledge skills. Furthermore, we conducted experiments on data and model scaling, both during pretraining and fine-tuning, simultaneously emphasizing the capacity for knowledge transfer across languages and addressing the challenges of catastrophic forgetting encountered during fine-tuning on a different language. Our goal is to offer a detailed guide for advancing the LLM framework in low-resource linguistic contexts, thereby making natural language processing (NLP) benefits more globally accessible.
Authors: Canaan Breiss, Alexis Ross, Amani Maina-Kilaas, Roger Levy, Jacob Andreas
Abstract: We propose an interactive approach to language learning that utilizes linguistic acceptability judgments from an informant (a competent language user) to learn a grammar. Given a grammar formalism and a framework for synthesizing data, our model iteratively selects or synthesizes a data-point according to one of a range of information-theoretic policies, asks the informant for a binary judgment, and updates its own parameters in preparation for the next query. We demonstrate the effectiveness of our model in the domain of phonotactics, the rules governing what kinds of sound-sequences are acceptable in a language, and carry out two experiments, one with typologically-natural linguistic data and another with a range of procedurally-generated languages. We find that the information-theoretic policies that our model uses to select items to query the informant achieve sample efficiency comparable to, and sometimes greater than, fully supervised approaches.
Authors: Chu Fei Luo, Ahmad Ghawanmeh, Xiaodan Zhu, Faiza Khan Khattak
Abstract: Modern large language models (LLMs) have a significant amount of world knowledge, which enables strong performance in commonsense reasoning and knowledge-intensive tasks when harnessed properly. The language model can also learn social biases, which has a significant potential for societal harm. There have been many mitigation strategies proposed for LLM safety, but it is unclear how effective they are for eliminating social biases. In this work, we propose a new methodology for attacking language models with knowledge graph augmented generation. We refactor natural language stereotypes into a knowledge graph, and use adversarial attacking strategies to induce biased responses from several open- and closed-source language models. We find our method increases bias in all models, even those trained with safety guardrails. This demonstrates the need for further research in AI safety, and further work in this new adversarial space.
Authors: Mahyar Abbasian, Iman Azimi, Mohammad Feli, Amir M. Rahmani, Ramesh Jain
Abstract: Agents represent one of the most emerging applications of Large Language Models (LLMs) and Generative AI, with their effectiveness hinging on multimodal capabilities to navigate complex user environments. Conversational Health Agents (CHAs), a prime example of this, are redefining healthcare by offering nuanced support that transcends textual analysis to incorporate emotional intelligence. This paper introduces an LLM-based CHA engineered for rich, multimodal dialogue-especially in the realm of mental health support. It adeptly interprets and responds to users' emotional states by analyzing multimodal cues, thus delivering contextually aware and empathetically resonant verbal responses. Our implementation leverages the versatile openCHA framework, and our comprehensive evaluation involves neutral prompts expressed in diverse emotional tones: sadness, anger, and joy. We evaluate the consistency and repeatability of the planning capability of the proposed CHA. Furthermore, human evaluators critique the CHA's empathic delivery, with findings revealing a striking concordance between the CHA's outputs and evaluators' assessments. These results affirm the indispensable role of vocal (soon multimodal) emotion recognition in strengthening the empathetic connection built by CHAs, cementing their place at the forefront of interactive, compassionate digital health solutions.
Authors: Zheyan Qu, Lu Yin, Zitong Yu, Wenbo Wang, Xing zhang
Abstract: Large language models (LLMs) have demonstrated astonishing capabilities in natural language processing (NLP) tasks, sparking interest in their application to professional domains with higher specialized requirements. However, restricted access to closed-source LLMs via APIs and the difficulty in collecting massive high-quality datasets pose obstacles to the development of large language models in education fields of various courses. Given these challenges, we propose CourseGPT-zh, a course-oriented education LLM that supports customization and low-cost deployment. To address the comprehensiveness and diversity requirements of course-specific corpora, we design a high-quality question-answering corpus distillation framework incorporating prompt optimization, which effectively mines textbook knowledge and enhances its diversity. Moreover, considering the alignment of LLM responses with user needs, a novel method for discrete prompt optimization based on LLM-as-Judge is introduced. During optimization, this framework leverages the LLM's ability to reflect on and exploit error feedback and patterns, allowing for prompts that meet user needs and preferences while saving response length. Lastly, we obtain CourseGPT-zh based on the open-source LLM using parameter-efficient fine-tuning. Experimental results show that our discrete prompt optimization framework effectively improves the response quality of ChatGPT, and CourseGPT-zh exhibits strong professional capabilities in specialized knowledge question-answering, significantly outperforming comparable open-source models.
Authors: Amrita Bhattacharjee, Raha Moraffah, Joshua Garland, Huan Liu
Abstract: Counterfactual examples are frequently used for model development and evaluation in many natural language processing (NLP) tasks. Although methods for automated counterfactual generation have been explored, such methods depend on models such as pre-trained language models that are then fine-tuned on auxiliary, often task-specific datasets. Collecting and annotating such datasets for counterfactual generation is labor intensive and therefore, infeasible in practice. Therefore, in this work, we focus on a novel problem setting: \textit{zero-shot counterfactual generation}. To this end, we propose a structured way to utilize large language models (LLMs) as general purpose counterfactual example generators. We hypothesize that the instruction-following and textual understanding capabilities of recent LLMs can be effectively leveraged for generating high quality counterfactuals in a zero-shot manner, without requiring any training or fine-tuning. Through comprehensive experiments on various downstream tasks in natural language processing (NLP), we demonstrate the efficacy of LLMs as zero-shot counterfactual generators in evaluating and explaining black-box NLP models.
Authors: Ana Brassard, Benjamin Heinzerling, Keito Kudo, Keisuke Sakaguchi, Kentaro Inui
Abstract: Evaluating free-text explanations is a multifaceted, subjective, and labor-intensive task. Large language models (LLMs) present an appealing alternative due to their potential for consistency, scalability, and cost-efficiency. In this work, we present ACORN, a new dataset of 3,500 free-text explanations and aspect-wise quality ratings, and use it to gain insights into how LLMs evaluate explanations. We observed that replacing one of the human ratings sometimes maintained, but more often lowered the inter-annotator agreement across different settings and quality aspects, suggesting that their judgments are not always consistent with human raters. We further quantified this difference by comparing the correlation between LLM-generated ratings with majority-voted human ratings across different quality aspects. With the best system, Spearman's rank correlation ranged between 0.53 to 0.95, averaging 0.72 across aspects, indicating moderately high but imperfect alignment. Finally, we considered the alternative of using an LLM as an additional rater when human raters are scarce, and measured the correlation between majority-voted labels with a limited human pool and LLMs as an additional rater, compared to the original gold labels. While GPT-4 improved the outcome when there were only two human raters, in all other observed cases, LLMs were neutral to detrimental when there were three or more human raters. We publicly release the dataset to support future improvements in LLM-in-the-loop evaluation here: https://github.com/a-brassard/ACORN.
Authors: Dawei Li, Shu Yang, Zhen Tan, Jae Young Baik, Sunkwon Yun, Joseph Lee, Aaron Chacko, Bojian Hou, Duy Duong-Tran, Ying Ding, Huan Liu, Li Shen, Tianlong Chen
Abstract: Recent advancements in large language models (LLMs) have achieved promising performances across various applications. Nonetheless, the ongoing challenge of integrating long-tail knowledge continues to impede the seamless adoption of LLMs in specialized domains. In this work, we introduce DALK, a.k.a. Dynamic Co-Augmentation of LLMs and KG, to address this limitation and demonstrate its ability on studying Alzheimer's Disease (AD), a specialized sub-field in biomedicine and a global health priority. With a synergized framework of LLM and KG mutually enhancing each other, we first leverage LLM to construct an evolving AD-specific knowledge graph (KG) sourced from AD-related scientific literature, and then we utilize a coarse-to-fine sampling method with a novel self-aware knowledge retrieval approach to select appropriate knowledge from the KG to augment LLM inference capabilities. The experimental results, conducted on our constructed AD question answering (ADQA) benchmark, underscore the efficacy of DALK. Additionally, we perform a series of detailed analyses that can offer valuable insights and guidelines for the emerging topic of mutually enhancing KG and LLM. We will release the code and data at https://github.com/David-Li0406/DALK.
Authors: Yikuan Xia, Jiazun Chen, Xinchi Li, Jun Gao
Abstract: Generalized Entity Matching (GEM), which aims at judging whether two records represented in different formats refer to the same real-world entity, is an essential task in data management. The prompt tuning paradigm for pre-trained language models (PLMs), including the recent PromptEM model, effectively addresses the challenges of low-resource GEM in practical applications, offering a robust solution when labeled data is scarce. However, existing prompt tuning models for GEM face the challenges of prompt design and information gap. This paper introduces an augmented prompt tuning framework for the challenges, which consists of two main improvements. The first is an augmented contextualized soft token-based prompt tuning method that extracts a guiding soft token benefit for the PLMs' prompt tuning, and the second is a cost-effective information augmentation strategy leveraging large language models (LLMs). Our approach performs well on the low-resource GEM challenges. Extensive experiments show promising advancements of our basic model without information augmentation over existing methods based on moderate-size PLMs (average 5.24%+), and our model with information augmentation achieves comparable performance compared with fine-tuned LLMs, using less than 14% of the API fee.
Authors: Xiaomin Zhuang, Yufan Jiang, Qiaozhi He, Zhihua Wu
Abstract: In this report, we present ChuXin, an entirely open-source language model with a size of 1.6 billion parameters. Unlike the majority of works that only open-sourced the model weights and architecture, we have made everything needed to train a model available, including the training data, the training process, and the evaluation code. Our goal is to empower and strengthen the open research community, fostering transparency and enabling a new wave of innovation in the field of language modeling. Furthermore, we extend the context length to 1M tokens through lightweight continual pretraining and demonstrate strong needle-in-a-haystack retrieval performance. The weights for both models are available at Hugging Face to download and use.
Authors: Sankalp Bahad, Pruthwik Mishra, Karunesh Arora, Rakesh Chandra Balabantaray, Dipti Misra Sharma, Parameswari Krishnamurthy
Abstract: Named Entity Recognition (NER) is a useful component in Natural Language Processing (NLP) applications. It is used in various tasks such as Machine Translation, Summarization, Information Retrieval, and Question-Answering systems. The research on NER is centered around English and some other major languages, whereas limited attention has been given to Indian languages. We analyze the challenges and propose techniques that can be tailored for Multilingual Named Entity Recognition for Indian Languages. We present a human annotated named entity corpora of 40K sentences for 4 Indian languages from two of the major Indian language families. Additionally,we present a multilingual model fine-tuned on our dataset, which achieves an F1 score of 0.80 on our dataset on average. We achieve comparable performance on completely unseen benchmark datasets for Indian languages which affirms the usability of our model.
Authors: Yitian Li, Jidong Tian, Hao He, Yaohui Jin
Abstract: Prompt-based methods have gained increasing attention on NLP and shown validity on many downstream tasks. Many works have focused on mining these methods' potential for knowledge extraction, but few explore their ability to make logical reasoning. In this work, we focus on the effectiveness of the prompt-based methods on first-order logical reasoning and find that the bottleneck lies in logical negation. Based on our analysis, logical negation tends to result in spurious correlations to negative answers, while propositions without logical negation correlate to positive answers. To solve the problem, we propose a simple but effective method, Negation Augmenting and Negation Debiasing (NAND), which introduces negative propositions to prompt-based methods without updating parameters. Specifically, these negative propositions can counteract spurious correlations by providing "not" for all instances so that models cannot make decisions only by whether expressions contain a logical negation. Experiments on three datasets show that NAND not only solves the problem of calibrating logical negation but also significantly enhances prompt-based methods of logical reasoning without model retraining.
Authors: Paul Jideani, Aurona Gerber
Abstract: Recent social media posts on the cholera outbreak in Hammanskraal have highlighted the diverse range of emotions people experienced in response to such an event. The extent of people's opinions varies greatly depending on their level of knowledge and information about the disease. The documented re-search about Cholera lacks investigations into the classification of emotions. This study aims to examine the emotions expressed in social media posts about Chol-era. A dataset of 23,000 posts was extracted and pre-processed. The Python Nat-ural Language Toolkit (NLTK) sentiment analyzer library was applied to deter-mine the emotional significance of each text. Additionally, Machine Learning (ML) models were applied for emotion classification, including Long short-term memory (LSTM), Logistic regression, Decision trees, and the Bidirectional En-coder Representations from Transformers (BERT) model. The results of this study demonstrated that LSTM achieved the highest accuracy of 75%. Emotion classification presents a promising tool for gaining a deeper understanding of the impact of Cholera on society. The findings of this study might contribute to the development of effective interventions in public health strategies.
Authors: Yan Liu, Yazheng Yang, Xiaokang Chen
Abstract: Long text understanding is important yet challenging for natural language processing. A long article or document usually contains many redundant words that are not pertinent to its gist and sometimes can be regarded as noise. With recent advances of abstractive summarization, we propose our \emph{Gist Detector} to leverage the gist detection ability of a summarization model and integrate the extracted gist into downstream models to enhance their long text understanding ability. Specifically, Gist Detector first learns the gist detection knowledge distilled from a summarization model, and then produces gist-aware representations to augment downstream models. We evaluate our method on three different tasks: long document classification, distantly supervised open-domain question answering, and non-parallel text style transfer. The experimental results show that our method can significantly improve the performance of baseline models on all tasks.
Authors: Guochao Jiang, Zepeng Ding, Yuchen Shi, Deqing Yang
Abstract: In recent years, the rise of large language models (LLMs) has made it possible to directly achieve named entity recognition (NER) without any demonstration samples or only using a few samples through in-context learning (ICL). However, standard ICL only helps LLMs understand task instructions, format and input-label mapping, but neglects the particularity of the NER task itself. In this paper, we propose a new prompting framework P-ICL to better achieve NER with LLMs, in which some point entities are leveraged as the auxiliary information to recognize each entity type. With such significant information, the LLM can achieve entity classification more precisely. To obtain optimal point entities for prompting LLMs, we also proposed a point entity selection method based on K-Means clustering. Our extensive experiments on some representative NER benchmarks verify the effectiveness of our proposed strategies in P-ICL and point entity selection.
Authors: Yunjia Qi, Hao Peng, Xiaozhi Wang, Bin Xu, Lei Hou, Juanzi Li
Abstract: Large language models (LLMs) usually fall short on information extraction (IE) tasks and struggle to follow the complex instructions of IE tasks. This primarily arises from LLMs not being aligned with humans, as mainstream alignment datasets typically do not include IE data. In this paper, we introduce ADELIE (Aligning large language moDELs on Information Extraction), an aligned LLM that effectively solves various IE tasks, including closed IE, open IE, and on-demand IE. We first collect and construct a high-quality alignment corpus IEInstruct for IE. Then we train ADELIE_SFT using instruction tuning on IEInstruct. We further train ADELIE_SFT with direct preference optimization (DPO) objective, resulting in ADELIE_DPO. Extensive experiments on various held-out IE datasets demonstrate that our models (ADELIE_SFT and ADELIE_DPO) achieve state-of-the-art (SoTA) performance among open-source models. We further explore the general capabilities of ADELIE, and experimental results reveal that their general capabilities do not exhibit a noticeable decline. We will release the code, data, and models to facilitate further research.
Authors: Lasse Hyldig Hansen, Nikolaj Andersen, Jack Gallifant, Liam G. McCoy, James K Stone, Nura Izath, Marcela Aguirre-Jerez, Danielle S Bitterman, Judy Gichoya, Leo Anthony Celi
Abstract: Background Advancements in Large Language Models (LLMs) hold transformative potential in healthcare, however, recent work has raised concern about the tendency of these models to produce outputs that display racial or gender biases. Although training data is a likely source of such biases, exploration of disease and demographic associations in text data at scale has been limited. Methods We conducted a large-scale textual analysis using a dataset comprising diverse web sources, including Arxiv, Wikipedia, and Common Crawl. The study analyzed the context in which various diseases are discussed alongside markers of race and gender. Given that LLMs are pre-trained on similar datasets, this approach allowed us to examine the potential biases that LLMs may learn and internalize. We compared these findings with actual demographic disease prevalence as well as GPT-4 outputs in order to evaluate the extent of bias representation. Results Our findings indicate that demographic terms are disproportionately associated with specific disease concepts in online texts. gender terms are prominently associated with disease concepts, while racial terms are much less frequently associated. We find widespread disparities in the associations of specific racial and gender terms with the 18 diseases analyzed. Most prominently, we see an overall significant overrepresentation of Black race mentions in comparison to population proportions. Conclusions Our results highlight the need for critical examination and transparent reporting of biases in LLM pretraining datasets. Our study suggests the need to develop mitigation strategies to counteract the influence of biased training data in LLMs, particularly in sensitive domains such as healthcare.
Authors: Aylin Gunal, Baihan Lin, Djallel Bouneffouf
Abstract: Given the increasing demand for mental health assistance, artificial intelligence (AI), particularly large language models (LLMs), may be valuable for integration into automated clinical support systems. In this work, we leverage a decision transformer architecture for topic recommendation in counseling conversations between patients and mental health professionals. The architecture is utilized for offline reinforcement learning, and we extract states (dialogue turn embeddings), actions (conversation topics), and rewards (scores measuring the alignment between patient and therapist) from previous turns within a conversation to train a decision transformer model. We demonstrate an improvement over baseline reinforcement learning methods, and propose a novel system of utilizing our model's output as synthetic labels for fine-tuning a large language model for the same task. Although our implementation based on LLaMA-2 7B has mixed results, future work can undoubtedly build on the design.
Authors: Weijia Zhang, Vaishali Pal, Jia-Hong Huang, Evangelos Kanoulas, Maarten de Rijke
Abstract: Table summarization is a crucial task aimed at condensing information from tabular data into concise and comprehensible textual summaries. However, existing approaches often fall short of adequately meeting users' information and quality requirements and tend to overlook the complexities of real-world queries. In this paper, we propose a novel method to address these limitations by introducing query-focused multi-table summarization. Our approach, which comprises a table serialization module, a summarization controller, and a large language model (LLM), utilizes textual queries and multiple tables to generate query-dependent table summaries tailored to users' information needs. To facilitate research in this area, we present a comprehensive dataset specifically tailored for this task, consisting of 4909 query-summary pairs, each associated with multiple tables. Through extensive experiments using our curated dataset, we demonstrate the effectiveness of our proposed method compared to baseline approaches. Our findings offer insights into the challenges of complex table reasoning for precise summarization, contributing to the advancement of research in query-focused multi-table summarization.
Authors: Peiqin Lin, Andr\'e F. T. Martins, Hinrich Sch\"utze
Abstract: Recent studies have shown that leveraging off-the-shelf or fine-tuned retrievers, capable of retrieving high-quality in-context examples, significantly improves in-context learning of English. However, adapting these methods to other languages, especially low-resource ones, presents challenges due to the scarcity of available cross-lingual retrievers and annotated data. In this paper, we introduce XAMPLER: Cross-Lingual Example Retrieval, a method tailored to tackle the challenge of cross-lingual in-context learning using only annotated English data. XAMPLER first trains a retriever with positive/negative English samples, which are constructed based on the predictions of the multilingual large language model for in-context learning. Then, the trained retriever is directly employed to retrieve English examples as few-shot examples for in-context learning of target languages. Experiments on the massively multilingual text classification benchmark of SIB200 with 176 languages demonstrate that XAMPLER substantially improves the in-context learning performance across languages. Our code is available at https://github.com/cisnlp/XAMPLER.
Authors: Julia Krebs, Evie Malaia, Ronnie B. Wilbur, Isabella Fessl, Hans-Peter Wiesinger, Hermann Schwameder, Dietmar Roehm
Abstract: Across a number of sign languages, temporal and spatial characteristics of dominant hand articulation are used to express semantic and grammatical features. In this study of Austrian Sign Language (\"Osterreichische Geb\"ardensprache, or \"OGS), motion capture data of four Deaf signers is used to quantitatively characterize the kinematic parameters of sign production in verbs and adjectives. We investigate (1) the difference in production between verbs involving a natural endpoint (telic verbs; e.g. arrive) and verbs lacking an endpoint (atelic verbs; e.g. analyze), and (2) adjective signs in intensified vs. non-intensified (plain) forms. Motion capture data analysis using linear-mixed effects models (LME) indicates that both the endpoint marking in verbs, as well as marking of intensification in adjectives, are expressed by movement modulation in \"OGS. While the semantic distinction between verb types (telic/atelic) is marked by higher peak velocity and shorter duration for telic signs compared to atelic ones, the grammatical distinction (intensification) in adjectives is expressed by longer duration for intensified compared to non-intensified adjectives. The observed individual differences of signers might be interpreted as personal signing style.
Authors: Tommaso Pasini, Alejo L\'opez-\'Avila, Husam Quteineh, Gerasimos Lampouras, Jinhua Du, Yubing Wang, Ze Li, Yusen Sun
Abstract: Composing poetry or lyrics involves several creative factors, but a challenging aspect of generation is the adherence to a more or less strict metric and rhyming pattern. To address this challenge specifically, previous work on the task has mainly focused on reverse language modeling, which brings the critical selection of each rhyming word to the forefront of each verse. On the other hand, reversing the word order requires that models be trained from scratch with this task-specific goal and cannot take advantage of transfer learning from a Pretrained Language Model (PLM). We propose a novel fine-tuning approach that prepends the rhyming word at the start of each lyric, which allows the critical rhyming decision to be made before the model commits to the content of the lyric (as during reverse language modeling), but maintains compatibility with the word order of regular PLMs as the lyric itself is still generated in left-to-right order. We conducted extensive experiments to compare this fine-tuning against the current state-of-the-art strategies for rhyming, finding that our approach generates more readable text and better rhyming capabilities. Furthermore, we furnish a high-quality dataset in English and 12 other languages, analyse the approach's feasibility in a multilingual context, provide extensive experimental results shedding light on good and bad practices for lyrics generation, and propose metrics to compare methods in the future.
Authors: Inderjeet Nair, Lu Wang
Abstract: We study the task of conducting structured reasoning as generating a reasoning graph from natural language input using large language models (LLMs). Previous approaches have explored various prompting schemes, yet they suffer from error propagation due to the autoregressive nature and single-pass-based decoding, which lack error correction capability. Additionally, relying solely on a single sample may result in the omission of true nodes and edges. To counter this, we draw inspiration from self-consistency (SC), which involves sampling a diverse set of reasoning chains and taking the majority vote as the final answer. To tackle the substantial challenge of applying SC on generated graphs, we propose MIDGARD (MInimum Description length Guided Aggregation of Reasoning in Directed acyclic graph) that leverages Minimum Description Length (MDL)-based formulation to identify consistent properties among the different graph samples generated by an LLM. This formulation helps reject properties that appear in only a few samples, which are likely to be erroneous, while enabling the inclusion of missing elements without compromising precision. Our method demonstrates superior performance than comparisons across various structured reasoning tasks, including argument structure extraction, explanation graph generation, inferring dependency relations among actions for everyday tasks, and semantic graph generation from natural texts.
Authors: Drew Walker, Annie Thorne, Sudeshna Das, Jennifer Love, Hannah LF Cooper, Melvin Livingston III, Abeed Sarker
Abstract: Objective: To detect and classify features of stigmatizing and biased language in intensive care electronic health records (EHRs) using natural language processing techniques. Materials and Methods: We first created a lexicon and regular expression lists from literature-driven stem words for linguistic features of stigmatizing patient labels, doubt markers, and scare quotes within EHRs. The lexicon was further extended using Word2Vec and GPT 3.5, and refined through human evaluation. These lexicons were used to search for matches across 18 million sentences from the de-identified Medical Information Mart for Intensive Care-III (MIMIC-III) dataset. For each linguistic bias feature, 1000 sentence matches were sampled, labeled by expert clinical and public health annotators, and used to supervised learning classifiers. Results: Lexicon development from expanded literature stem-word lists resulted in a doubt marker lexicon containing 58 expressions, and a stigmatizing labels lexicon containing 127 expressions. Classifiers for doubt markers and stigmatizing labels had the highest performance, with macro F1-scores of .84 and .79, positive-label recall and precision values ranging from .71 to .86, and accuracies aligning closely with human annotator agreement (.87). Discussion: This study demonstrated the feasibility of supervised classifiers in automatically identifying stigmatizing labels and doubt markers in medical text, and identified trends in stigmatizing language use in an EHR setting. Additional labeled data may help improve lower scare quote model performance. Conclusions: Classifiers developed in this study showed high model performance and can be applied to identify patterns and target interventions to reduce stigmatizing labels and doubt markers in healthcare systems.
Authors: Sean Noh, Ho-Chun Herbert Chang
Abstract: Powered by large language models (LLMs), AI agents have become capable of many human tasks. Using the most canonical definitions of the Big Five personality, we measure the ability of LLMs to negotiate within a game-theoretical framework, as well as methodological challenges to measuring notions of fairness and risk. Simulations (n=1,500) for both single-issue and multi-issue negotiation reveal increase in domain complexity with asymmetric issue valuations improve agreement rates but decrease surplus from aggressive negotiation. Through gradient-boosted regression and Shapley explainers, we find high openness, conscientiousness, and neuroticism are associated with fair tendencies; low agreeableness and low openness are associated with rational tendencies. Low conscientiousness is associated with high toxicity. These results indicate that LLMs may have built-in guardrails that default to fair behavior, but can be "jail broken" to exploit agreeable opponents. We also offer pragmatic insight in how negotiation bots can be designed, and a framework of assessing negotiation behavior based on game theory and computational social science.
Authors: Charles Koutcheme, Nicola Dainese, Sami Sarsa, Arto Hellas, Juho Leinonen, Paul Denny
Abstract: Large language models (LLMs) have shown great potential for the automatic generation of feedback in a wide range of computing contexts. However, concerns have been voiced around the privacy and ethical implications of sending student work to proprietary models. This has sparked considerable interest in the use of open source LLMs in education, but the quality of the feedback that such open models can produce remains understudied. This is a concern as providing flawed or misleading generated feedback could be detrimental to student learning. Inspired by recent work that has utilised very powerful LLMs, such as GPT-4, to evaluate the outputs produced by less powerful models, we conduct an automated analysis of the quality of the feedback produced by several open source models using a dataset from an introductory programming course. First, we investigate the viability of employing GPT-4 as an automated evaluator by comparing its evaluations with those of a human expert. We observe that GPT-4 demonstrates a bias toward positively rating feedback while exhibiting moderate agreement with human raters, showcasing its potential as a feedback evaluator. Second, we explore the quality of feedback generated by several leading open-source LLMs by using GPT-4 to evaluate the feedback. We find that some models offer competitive performance with popular proprietary LLMs, such as ChatGPT, indicating opportunities for their responsible use in educational settings.
Authors: Yutao Sun, Li Dong, Yi Zhu, Shaohan Huang, Wenhui Wang, Shuming Ma, Quanlu Zhang, Jianyong Wang, Furu Wei
Abstract: We introduce a decoder-decoder architecture, YOCO, for large language models, which only caches key-value pairs once. It consists of two components, i.e., a cross-decoder stacked upon a self-decoder. The self-decoder efficiently encodes global key-value (KV) caches that are reused by the cross-decoder via cross-attention. The overall model behaves like a decoder-only Transformer, although YOCO only caches once. The design substantially reduces GPU memory demands, yet retains global attention capability. Additionally, the computation flow enables prefilling to early exit without changing the final output, thereby significantly speeding up the prefill stage. Experimental results demonstrate that YOCO achieves favorable performance compared to Transformer in various settings of scaling up model size and number of training tokens. We also extend YOCO to 1M context length with near-perfect needle retrieval accuracy. The profiling results show that YOCO improves inference memory, prefill latency, and throughput by orders of magnitude across context lengths and model sizes. Code is available at https://aka.ms/YOCO.
URLs: https://aka.ms/YOCO.
Authors: Won-Gi Paeng, Daesuk Kwon
Abstract: This short note is written for rapid communication of long context training and to share the idea of how to train it with low memory usage. In the note, we generalize the attention algorithm and neural network of Generative Pre-Trained Transformers and reinterpret it in Path integral formalism. First, the role of the transformer is understood as the time evolution of the token state and second, it is suggested that the all key-token states in the same time as the query-token can attend to the attention with the query token states. As a result of the repetitive time evolution, it is discussed that the token states in the past sequence meats the token states in the present sequence so that the attention between separated sequences becomes possible for maintaining infinite contextual information just by using low memory for limited size of sequence. For the experiment, the $12$ input token window size was taken and one GPU with $24$GB memory was used for the pre-training. It was confirmed that more than $150$ length context is preserved. The sampling result of the training, the code and the other details will be included in the revised version of this note later.
Authors: Roelien C. Timmer, David Liebowitz, Surya Nepal, Salil S. Kanhere
Abstract: Honeyfiles are a particularly useful type of honeypot: fake files deployed to detect and infer information from malicious behaviour. This paper considers the challenge of naming honeyfiles so they are camouflaged when placed amongst real files in a file system. Based on cosine distances in semantic vector spaces, we develop two metrics for filename camouflage: one based on simple averaging and one on clustering with mixture fitting. We evaluate and compare the metrics, showing that both perform well on a publicly available GitHub software repository dataset.
Authors: Yunxin Li, Baotian Hu, Haoyuan Shi, Wei Wang, Longyue Wang, Min Zhang
Abstract: Large Multimodal Models (LMMs) have achieved impressive success in visual understanding and reasoning, remarkably improving the performance of mathematical reasoning in a visual context. Yet, a challenging type of visual math lies in the multimodal graph theory problem, which demands that LMMs understand the graphical structures accurately and perform multi-step reasoning on the visual graph. Additionally, exploring multimodal graph theory problems will lead to more effective strategies in fields like biology, transportation, and robotics planning. To step forward in this direction, we are the first to design a benchmark named VisionGraph, used to explore the capabilities of advanced LMMs in solving multimodal graph theory problems. It encompasses eight complex graph problem tasks, from connectivity to shortest path problems. Subsequently, we present a Description-Program-Reasoning (DPR) chain to enhance the logical accuracy of reasoning processes through graphical structure description generation and algorithm-aware multi-step reasoning. Our extensive study shows that 1) GPT-4V outperforms Gemini Pro in multi-step graph reasoning; 2) All LMMs exhibit inferior perception accuracy for graphical structures, whether in zero/few-shot settings or with supervised fine-tuning (SFT), which further affects problem-solving performance; 3) DPR significantly improves the multi-step graph reasoning capabilities of LMMs and the GPT-4V (DPR) agent achieves SOTA performance.
Authors: Mohamad Fazelnia, Viktoria Koscinski, Spencer Herzog, Mehdi Mirakhorli
Abstract: We investigate the use of Natural Language Inference (NLI) in automating requirements engineering tasks. In particular, we focus on three tasks: requirements classification, identification of requirements specification defects, and detection of conflicts in stakeholders' requirements. While previous research has demonstrated significant benefit in using NLI as a universal method for a broad spectrum of natural language processing tasks, these advantages have not been investigated within the context of software requirements engineering. Therefore, we design experiments to evaluate the use of NLI in requirements analysis. We compare the performance of NLI with a spectrum of approaches, including prompt-based models, conventional transfer learning, Large Language Models (LLMs)-powered chatbot models, and probabilistic models. Through experiments conducted under various learning settings including conventional learning and zero-shot, we demonstrate conclusively that our NLI method surpasses classical NLP methods as well as other LLMs-based and chatbot models in the analysis of requirements specifications. Additionally, we share lessons learned characterizing the learning settings that make NLI a suitable approach for automating requirements engineering tasks.
Authors: Zhaoxing Li, Jujie Yang, Jindi Wang, Lei Shi, Sebastian Stein
Abstract: The field of Knowledge Tracing aims to understand how students learn and master knowledge over time by analyzing their historical behaviour data. To achieve this goal, many researchers have proposed Knowledge Tracing models that use data from Intelligent Tutoring Systems to predict students' subsequent actions. However, with the development of Intelligent Tutoring Systems, large-scale datasets containing long-sequence data began to emerge. Recent deep learning based Knowledge Tracing models face obstacles such as low efficiency, low accuracy, and low interpretability when dealing with large-scale datasets containing long-sequence data. To address these issues and promote the sustainable development of Intelligent Tutoring Systems, we propose a LSTM BERT-based Knowledge Tracing model for long sequence data processing, namely LBKT, which uses a BERT-based architecture with a Rasch model-based embeddings block to deal with different difficulty levels information and an LSTM block to process the sequential characteristic in students' actions. LBKT achieves the best performance on most benchmark datasets on the metrics of ACC and AUC. Additionally, an ablation study is conducted to analyse the impact of each component of LBKT's overall performance. Moreover, we used t-SNE as the visualisation tool to demonstrate the model's embedding strategy. The results indicate that LBKT is faster, more interpretable, and has a lower memory cost than the traditional deep learning based Knowledge Tracing methods.
Authors: Eugene Bagdasaryan, Ren Yi, Sahra Ghalebikesabi, Peter Kairouz, Marco Gruteser, Sewoong Oh, Borja Balle, Daniel Ramage
Abstract: The growing use of large language model (LLM)-based conversational agents to manage sensitive user data raises significant privacy concerns. While these agents excel at understanding and acting on context, this capability can be exploited by malicious actors. We introduce a novel threat model where adversarial third-party apps manipulate the context of interaction to trick LLM-based agents into revealing private information not relevant to the task at hand. Grounded in the framework of contextual integrity, we introduce AirGapAgent, a privacy-conscious agent designed to prevent unintended data leakage by restricting the agent's access to only the data necessary for a specific task. Extensive experiments using Gemini, GPT, and Mistral models as agents validate our approach's effectiveness in mitigating this form of context hijacking while maintaining core agent functionality. For example, we show that a single-query context hijacking attack on a Gemini Ultra agent reduces its ability to protect user data from 94% to 45%, while an AirGapAgent achieves 97% protection, rendering the same attack ineffective.
Authors: Zihan Wang, Naoki Yoshinaga
Abstract: Esports, a sports competition on video games, has become one of the most important sporting events. Although esports play logs have been accumulated, only a small portion of them accompany text commentaries for the audience to retrieve and understand the plays. In this study, we therefore introduce the task of generating game commentaries from esports' data records. We first build large-scale esports data-to-text datasets that pair structured data and commentaries from a popular esports game, League of Legends. We then evaluate Transformer-based models to generate game commentaries from structured data records, while examining the impact of the pre-trained language models. Evaluation results on our dataset revealed the challenges of this novel task. We will release our dataset to boost potential research in the data-to-text generation community.
Authors: Linyuan Gong, Jiayi Wang, Alvin Cheung
Abstract: We propose the Adversarial DEep Learning Transpiler (ADELT), a novel approach to source-to-source transpilation between deep learning frameworks. ADELT uniquely decouples code skeleton transpilation and API keyword mapping. For code skeleton transpilation, it uses few-shot prompting on large language models (LLMs), while for API keyword mapping, it uses contextual embeddings from a code-specific BERT. These embeddings are trained in a domain-adversarial setup to generate a keyword translation dictionary. ADELT is trained on an unlabeled web-crawled deep learning corpus, without relying on any hand-crafted rules or parallel data. It outperforms state-of-the-art transpilers, improving pass@1 rate by 17.4 pts and 15.0 pts for PyTorch-Keras and PyTorch-MXNet transpilation pairs respectively. We provide open access to our code at https://github.com/gonglinyuan/adelt.
Authors: Chong Ma, Zihao Wu, Jiaqi Wang, Shaochen Xu, Yaonai Wei, Fang Zeng, Zhengliang Liu, Xi Jiang, Lei Guo, Xiaoyan Cai, Shu Zhang, Tuo Zhang, Dajiang Zhu, Dinggang Shen, Tianming Liu, Xiang Li
Abstract: The 'Impression' section of a radiology report is a critical basis for communication between radiologists and other physicians, and it is typically written by radiologists based on the 'Findings' section. However, writing numerous impressions can be laborious and error-prone for radiologists. Although recent studies have achieved promising results in automatic impression generation using large-scale medical text data for pre-training and fine-tuning pre-trained language models, such models often require substantial amounts of medical text data and have poor generalization performance. While large language models (LLMs) like ChatGPT have shown strong generalization capabilities and performance, their performance in specific domains, such as radiology, remains under-investigated and potentially limited. To address this limitation, we propose ImpressionGPT, which leverages the in-context learning capability of LLMs by constructing dynamic contexts using domain-specific, individualized data. This dynamic prompt approach enables the model to learn contextual knowledge from semantically similar examples from existing data. Additionally, we design an iterative optimization algorithm that performs automatic evaluation on the generated impression results and composes the corresponding instruction prompts to further optimize the model. The proposed ImpressionGPT model achieves state-of-the-art performance on both MIMIC-CXR and OpenI datasets without requiring additional training data or fine-tuning the LLMs. This work presents a paradigm for localizing LLMs that can be applied in a wide range of similar application scenarios, bridging the gap between general-purpose LLMs and the specific language processing needs of various domains.
Authors: Song Jiang, Zahra Shakeri, Aaron Chan, Maziar Sanjabi, Hamed Firooz, Yinglong Xia, Bugra Akyildiz, Yizhou Sun, Jinchao Li, Qifan Wang, Asli Celikyilmaz
Abstract: Chain-of-thought (CoT) prompting, which offers step-by-step problem-solving rationales, has impressively unlocked the reasoning potential of large language models (LLMs). Yet, the standard CoT is less effective in problems demanding multiple reasoning steps. This limitation arises from the complex reasoning process in multi-step problems: later stages often depend on the results of several steps earlier, not just the results of the immediately preceding step. Such complexities suggest the reasoning process is naturally represented as a graph. The almost linear and straightforward structure of CoT prompting, however, struggles to capture this complex reasoning graph. To address this challenge, we propose Residual Connection Prompting (RESPROMPT), a new prompting strategy that advances multi-step reasoning in LLMs. Our key idea is to reconstruct the reasoning graph within prompts. We achieve this by integrating necessary connections-links present in the reasoning graph but missing in the linear CoT flow-into the prompts. Termed "residual connections", these links are pivotal in morphing the linear CoT structure into a graph representation, effectively capturing the complex reasoning graphs inherent in multi-step problems. We evaluate RESPROMPT on six benchmarks across three diverse domains: math, sequential, and commonsense reasoning. For the open-sourced LLaMA family of models, RESPROMPT yields a significant average reasoning accuracy improvement of 12.5% on LLaMA-65B and 6.8% on LLaMA2-70B. Breakdown analysis further highlights RESPROMPT particularly excels in complex multi-step reasoning: for questions demanding at least five reasoning steps, RESPROMPT outperforms the best CoT based benchmarks by a remarkable average improvement of 21.1% on LLaMA-65B and 14.3% on LLaMA2-70B. Through extensive ablation studies and analyses, we pinpoint how to most effectively build residual connections.
Authors: Ming Zhong, Chenxin An, Weizhu Chen, Jiawei Han, Pengcheng He
Abstract: Large Language Models (LLMs) inherently encode a wealth of knowledge within their parameters through pre-training on extensive corpora. While prior research has delved into operations on these parameters to manipulate the underlying implicit knowledge (encompassing detection, editing, and merging), there remains an ambiguous understanding regarding their transferability across models with varying scales. In this paper, we seek to empirically investigate knowledge transfer from larger to smaller models through a parametric perspective. To achieve this, we employ sensitivity-based techniques to extract and align knowledge-specific parameters between different LLMs. Moreover, the LoRA module is used as the intermediary mechanism for injecting the extracted knowledge into smaller models. Evaluations across four benchmarks validate the efficacy of our proposed method. Our findings highlight the critical factors contributing to the process of parametric knowledge transfer, underscoring the transferability of model parameters across LLMs of different scales. Project website: https://maszhongming.github.io/ParaKnowTransfer.
Authors: Rafael Rivera Soto, Kailin Koch, Aleem Khan, Barry Chen, Marcus Bishop, Nicholas Andrews
Abstract: The advent of instruction-tuned language models that convincingly mimic human writing poses a significant risk of abuse. However, such abuse may be counteracted with the ability to detect whether a piece of text was composed by a language model rather than a human author. Some previous approaches to this problem have relied on supervised methods by training on corpora of confirmed human- and machine- written documents. Unfortunately, model under-specification poses an unavoidable challenge for neural network-based detectors, making them brittle in the face of data shifts, such as the release of newer language models producing still more fluent text than the models used to train the detectors. Other approaches require access to the models that may have generated a document in question, which is often impractical. In light of these challenges, we pursue a fundamentally different approach not relying on samples from language models of concern at training time. Instead, we propose to leverage representations of writing style estimated from human-authored text. Indeed, we find that features effective at distinguishing among human authors are also effective at distinguishing human from machine authors, including state-of-the-art large language models like Llama-2, ChatGPT, and GPT-4. Furthermore, given a handful of examples composed by each of several specific language models of interest, our approach affords the ability to predict which model generated a given document. The code and data to reproduce our experiments are available at https://github.com/LLNL/LUAR/tree/main/fewshot_iclr2024.
URLs: https://github.com/LLNL/LUAR/tree/main/fewshot_iclr2024.
Authors: Zheqi He, Xinya Wu, Pengfei Zhou, Richeng Xuan, Guang Liu, Xi Yang, Qiannan Zhu, Hua Huang
Abstract: Multi-modal large language models(MLLMs) have achieved remarkable progress and demonstrated powerful knowledge comprehension and reasoning abilities. However, the mastery of domain-specific knowledge, which is essential for evaluating the intelligence of MLLMs, continues to be a challenge. Current multi-modal benchmarks for domain-specific knowledge concentrate on multiple-choice questions and are predominantly available in English, which imposes limitations on the comprehensiveness of the evaluation. To this end, we introduce CMMU, a novel benchmark for multi-modal and multi-type question understanding and reasoning in Chinese. CMMU consists of 3,603 questions in 7 subjects, covering knowledge from primary to high school. The questions can be categorized into 3 types: multiple-choice, multiple-response, and fill-in-the-blank, bringing greater challenges to MLLMs. In addition, we propose an evaluation strategy called Positional Error Variance for assessing multiple-choice questions. The strategy aims to perform a quantitative analysis of position bias. We evaluate seven open-source MLLMs along with GPT4-V, Gemini-Pro, and Qwen-VL-Plus. The results demonstrate that CMMU poses a significant challenge to the recent MLLMs. The data and code are available at https://github.com/FlagOpen/CMMU.
Authors: Sahil Kale, Gautam Khaire, Jay Patankar
Abstract: Frequently Asked Questions (FAQs) refer to the most common inquiries about specific content. They serve as content comprehension aids by simplifying topics and enhancing understanding through succinct presentation of information. In this paper, we address FAQ generation as a well-defined Natural Language Processing task through the development of an end-to-end system leveraging text-to-text transformation models. We present a literature review covering traditional question-answering systems, highlighting their limitations when applied directly to the FAQ generation task. We propose a system capable of building FAQs from textual content tailored to specific domains, enhancing their accuracy and relevance. We utilise self-curated algorithms to obtain an optimal representation of information to be provided as input and also to rank the question-answer pairs to maximise human comprehension. Qualitative human evaluation showcases the generated FAQs as well-constructed and readable while also utilising domain-specific constructs to highlight domain-based nuances and jargon in the original content.
Authors: Saurabh Bhausaheb Zinjad, Amrita Bhattacharjee, Amey Bhilegaonkar, Huan Liu
Abstract: Crafting the ideal, job-specific resume is a challenging task for many job applicants, especially for early-career applicants. While it is highly recommended that applicants tailor their resume to the specific role they are applying for, manually tailoring resumes to job descriptions and role-specific requirements is often (1) extremely time-consuming, and (2) prone to human errors. Furthermore, performing such a tailoring step at scale while applying to several roles may result in a lack of quality of the edited resumes. To tackle this problem, in this demo paper, we propose ResumeFlow: a Large Language Model (LLM) aided tool that enables an end user to simply provide their detailed resume and the desired job posting, and obtain a personalized resume specifically tailored to that specific job posting in the matter of a few seconds. Our proposed pipeline leverages the language understanding and information extraction capabilities of state-of-the-art LLMs such as OpenAI's GPT-4 and Google's Gemini, in order to (1) extract details from a job description, (2) extract role-specific details from the user-provided resume, and then (3) use these to refine and generate a role-specific resume for the user. Our easy-to-use tool leverages the user-chosen LLM in a completely off-the-shelf manner, thus requiring no fine-tuning. We demonstrate the effectiveness of our tool via a video demo and propose novel task-specific evaluation metrics to control for alignment and hallucination. Our tool is available at https://job-aligned-resume.streamlit.app.
Authors: Peter Wallis
Abstract: Theory based AI research has had a hard time recently and the aim here is to propose a model of what LLMs are actually doing when they impress us with their language skills. The model integrates three established theories of human decision-making from philosophy, sociology, and computer science. The paper starts with the collective understanding of reasoning from the early days of AI research - primarily because that model is how we humans think we think, and is the most accessible. It then describes what is commonly thought of as "reactive systems" which is the position taken by many philosophers and indeed many contemporary AI researchers. The third component to the proposed model is from sociology and, although not flattering to our modern ego, provides an explanation to a puzzle that for many years has occupied those of us working on conversational user interfaces.
Authors: Yiming Huang, Xiao Liu, Yeyun Gong, Zhibin Gou, Yelong Shen, Nan Duan, Weizhu Chen
Abstract: Large language models (LLMs) have shown great potential in complex reasoning tasks, yet their performance is often hampered by the scarcity of high-quality and reasoning-focused training datasets. Addressing this challenge, we propose Key-Point-Driven Data Synthesis (KPDDS), a novel data synthesis framework that synthesizes question-answer pairs by leveraging key points and exemplar practices from authentic data sources. KPDDS ensures the generation of novel questions with rigorous quality control and substantial scalability. As a result, we present KPMath, an extensive synthetic dataset tailored for mathematical reasoning, comprising over 800K question-answer pairs. Utilizing KPMath and augmenting it with additional reasoning-intensive corpora, we create the comprehensive KPMath-Plus dataset. The Qwen1.5-72B model, fine-tuned on KPMath-Plus, achieves 87.0% PASS@1 accuracy on GSM8K and 58.3% on MATH, surpassing competitors in the 7B to 70B range and best commercial models like GPT-4 across multiple math reasoning datasets.
Authors: Ayushi Nirmal, Amrita Bhattacharjee, Paras Sheth, Huan Liu
Abstract: Although social media platforms are a prominent arena for users to engage in interpersonal discussions and express opinions, the facade and anonymity offered by social media may allow users to spew hate speech and offensive content. Given the massive scale of such platforms, there arises a need to automatically identify and flag instances of hate speech. Although several hate speech detection methods exist, most of these black-box methods are not interpretable or explainable by design. To address the lack of interpretability, in this paper, we propose to use state-of-the-art Large Language Models (LLMs) to extract features in the form of rationales from the input text, to train a base hate speech classifier, thereby enabling faithful interpretability by design. Our framework effectively combines the textual understanding capabilities of LLMs and the discriminative power of state-of-the-art hate speech classifiers to make these classifiers faithfully interpretable. Our comprehensive evaluation on a variety of English language social media hate speech datasets demonstrate: (1) the goodness of the LLM-extracted rationales, and (2) the surprising retention of detector performance even after training to ensure interpretability. All code and data will be made available at https://github.com/AmritaBh/shield.
Authors: Olga Golovneva, Zeyuan Allen-Zhu, Jason Weston, Sainbayar Sukhbaatar
Abstract: Large language models (LLMs) have a surprising failure: when trained on "A has a feature B", they do not generalize to "B is a feature of A", which is termed the Reversal Curse. Even when training with trillions of tokens this issue still appears due to Zipf's law - hence even if we train on the entire internet. This work proposes an alternative training scheme, called reverse training, whereby all words are used twice, doubling the amount of available tokens. The LLM is trained in both forward and reverse directions by reversing the training strings while preserving (i.e., not reversing) chosen substrings, such as entities. We show that data-matched reverse-trained models provide superior performance to standard models on standard tasks, and compute-matched reverse-trained models provide far superior performance on reversal tasks, helping resolve the reversal curse issue.
Authors: Junxiong Wang, Ali Mousavi, Omar Attia, Ronak Pradeep, Saloni Potdar, Alexander M. Rush, Umar Farooq Minhas, Yunyao Li
Abstract: Entity disambiguation (ED), which links the mentions of ambiguous entities to their referent entities in a knowledge base, serves as a core component in entity linking (EL). Existing generative approaches demonstrate improved accuracy compared to classification approaches under the standardized ZELDA benchmark. Nevertheless, generative approaches suffer from the need for large-scale pre-training and inefficient generation. Most importantly, entity descriptions, which could contain crucial information to distinguish similar entities from each other, are often overlooked. We propose an encoder-decoder model to disambiguate entities with more detailed entity descriptions. Given text and candidate entities, the encoder learns interactions between the text and each candidate entity, producing representations for each entity candidate. The decoder then fuses the representations of entity candidates together and selects the correct entity. Our experiments, conducted on various entity disambiguation benchmarks, demonstrate the strong and robust performance of this model, particularly +1.5% in the ZELDA benchmark compared with GENRE. Furthermore, we integrate this approach into the retrieval/reader framework and observe +1.5% improvements in end-to-end entity linking in the GERBIL benchmark compared with EntQA.
Authors: Bowen Ding, Qingkai Min, Shengkun Ma, Yingjie Li, Linyi Yang, Yue Zhang
Abstract: Based on Pre-trained Language Models (PLMs), event coreference resolution (ECR) systems have demonstrated outstanding performance in clustering coreferential events across documents. However, the existing system exhibits an excessive reliance on the `triggers lexical matching' spurious pattern in the input mention pair text. We formalize the decision-making process of the baseline ECR system using a Structural Causal Model (SCM), aiming to identify spurious and causal associations (i.e., rationales) within the ECR task. Leveraging the debiasing capability of counterfactual data augmentation, we develop a rationale-centric counterfactual data augmentation method with LLM-in-the-loop. This method is specialized for pairwise input in the ECR system, where we conduct direct interventions on triggers and context to mitigate the spurious association while emphasizing the causation. Our approach achieves state-of-the-art performance on three popular cross-document ECR benchmarks and demonstrates robustness in out-of-domain scenarios.
Authors: Ahmad Idrissi-Yaghir, Amin Dada, Henning Sch\"afer, Kamyar Arzideh, Giulia Baldini, Jan Trienes, Max Hasin, Jeanette Bewersdorff, Cynthia S. Schmidt, Marie Bauer, Kaleb E. Smith, Jiang Bian, Yonghui Wu, J\"org Schl\"otterer, Torsten Zesch, Peter A. Horn, Christin Seifert, Felix Nensa, Jens Kleesiek, Christoph M. Friedrich
Abstract: Recent advances in natural language processing (NLP) can be largely attributed to the advent of pre-trained language models such as BERT and RoBERTa. While these models demonstrate remarkable performance on general datasets, they can struggle in specialized domains such as medicine, where unique domain-specific terminologies, domain-specific abbreviations, and varying document structures are common. This paper explores strategies for adapting these models to domain-specific requirements, primarily through continuous pre-training on domain-specific data. We pre-trained several German medical language models on 2.4B tokens derived from translated public English medical data and 3B tokens of German clinical data. The resulting models were evaluated on various German downstream tasks, including named entity recognition (NER), multi-label classification, and extractive question answering. Our results suggest that models augmented by clinical and translation-based pre-training typically outperform general domain models in medical contexts. We conclude that continuous pre-training has demonstrated the ability to match or even exceed the performance of clinical models trained from scratch. Furthermore, pre-training on clinical data or leveraging translated texts have proven to be reliable methods for domain adaptation in medical NLP tasks.
Authors: Tianyu Cao, Natraj Raman, Danial Dervovic, Chenhao Tan
Abstract: As large language models (LLMs) expand the power of natural language processing to handle long inputs, rigorous and systematic analyses are necessary to understand their abilities and behavior. A salient application is summarization, due to its ubiquity and controversy (e.g., researchers have declared the death of summarization). In this paper, we use financial report summarization as a case study because financial reports not only are long but also use numbers and tables extensively. We propose a computational framework for characterizing multimodal long-form summarization and investigate the behavior of Claude 2.0/2.1, GPT-4/3.5, and Command. We find that GPT-3.5 and Command fail to perform this summarization task meaningfully. For Claude 2 and GPT-4, we analyze the extractiveness of the summary and identify a position bias in LLMs. This position bias disappears after shuffling the input for Claude, which suggests that Claude has the ability to recognize important information. We also conduct a comprehensive investigation on the use of numeric data in LLM-generated summaries and offer a taxonomy of numeric hallucination. We employ prompt engineering to improve GPT-4's use of numbers with limited success. Overall, our analyses highlight the strong capability of Claude 2 in handling long multimodal inputs compared to GPT-4.
Authors: Xiao Zhang, Chunliu Wang, Rik van Noord, Johan Bos
Abstract: The Parallel Meaning Bank (PMB) serves as a corpus for semantic processing with a focus on semantic parsing and text generation. Currently, we witness an excellent performance of neural parsers and generators on the PMB. This might suggest that such semantic processing tasks have by and large been solved. We argue that this is not the case and that performance scores from the past on the PMB are inflated by non-optimal data splits and test sets that are too easy. In response, we introduce several changes. First, instead of the prior random split, we propose a more systematic splitting approach to improve the reliability of the standard test data. Second, except for the standard test set, we also propose two challenge sets: one with longer texts including discourse structure, and one that addresses compositional generalization. We evaluate five neural models for semantic parsing and meaning-to-text generation. Our results show that model performance declines (in some cases dramatically) on the challenge sets, revealing the limitations of neural models when confronting such challenges.
Authors: Aleksi Huotala, Miikka Kuutila, Paul Ralph, Mika M\"antyl\"a
Abstract: Systematic review (SR) is a popular research method in software engineering (SE). However, conducting an SR takes an average of 67 weeks. Thus, automating any step of the SR process could reduce the effort associated with SRs. Our objective is to investigate if Large Language Models (LLMs) can accelerate title-abstract screening by simplifying abstracts for human screeners, and automating title-abstract screening. We performed an experiment where humans screened titles and abstracts for 20 papers with both original and simplified abstracts from a prior SR. The experiment with human screeners was reproduced with GPT-3.5 and GPT-4 LLMs to perform the same screening tasks. We also studied if different prompting techniques (Zero-shot (ZS), One-shot (OS), Few-shot (FS), and Few-shot with Chain-of-Thought (FS-CoT)) improve the screening performance of LLMs. Lastly, we studied if redesigning the prompt used in the LLM reproduction of screening leads to improved performance. Text simplification did not increase the screeners' screening performance, but reduced the time used in screening. Screeners' scientific literacy skills and researcher status predict screening performance. Some LLM and prompt combinations perform as well as human screeners in the screening tasks. Our results indicate that the GPT-4 LLM is better than its predecessor, GPT-3.5. Additionally, Few-shot and One-shot prompting outperforms Zero-shot prompting. Using LLMs for text simplification in the screening process does not significantly improve human performance. Using LLMs to automate title-abstract screening seems promising, but current LLMs are not significantly more accurate than human screeners. To recommend the use of LLMs in the screening process of SRs, more research is needed. We recommend future SR studies publish replication packages with screening data to enable more conclusive experimenting with LLM screening.
Authors: Haojie Zhang, Yimeng Zhuang
Abstract: Few-shot Named Entity Recognition (NER) aims to extract named entities using only a limited number of labeled examples. Existing contrastive learning methods often suffer from insufficient distinguishability in context vector representation because they either solely rely on label semantics or completely disregard them. To tackle this issue, we propose a unified label-aware token-level contrastive learning framework. Our approach enriches the context by utilizing label semantics as suffix prompts. Additionally, it simultaneously optimizes context-context and context-label contrastive learning objectives to enhance generalized discriminative contextual representations.Extensive experiments on various traditional test domains (OntoNotes, CoNLL'03, WNUT'17, GUM, I2B2) and the large-scale few-shot NER dataset (FEWNERD) demonstrate the effectiveness of our approach. It outperforms prior state-of-the-art models by a significant margin, achieving an average absolute gain of 7% in micro F1 scores across most scenarios. Further analysis reveals that our model benefits from its powerful transfer capability and improved contextual representations.
Authors: Xin Quan, Marco Valentino, Louise A. Dennis, Andr\'e Freitas
Abstract: Natural language explanations have become a proxy for evaluating explainable and multi-step Natural Language Inference (NLI) models. However, assessing the validity of explanations for NLI is challenging as it typically involves the crowd-sourcing of apposite datasets, a process that is time-consuming and prone to logical errors. To address existing limitations, this paper investigates the verification and refinement of natural language explanations through the integration of Large Language Models (LLMs) and Theorem Provers (TPs). Specifically, we present a neuro-symbolic framework, named Explanation-Refiner, that augments a TP with LLMs to generate and formalise explanatory sentences and suggest potential inference strategies for NLI. In turn, the TP is employed to provide formal guarantees on the logical validity of the explanations and to generate feedback for subsequent improvements. We demonstrate how Explanation-Refiner can be jointly used to evaluate explanatory reasoning, autoformalisation, and error correction mechanisms of state-of-the-art LLMs as well as to automatically enhance the quality of human-annotated explanations of variable complexity in different domains.
Authors: Qizhou Chen, Taolin Zhang, Xiaofeng He, Dongyang Li, Chengyu Wang, Longtao Huang, Hui Xue
Abstract: Model editing aims to correct outdated or erroneous knowledge in large language models (LLMs) without the need for costly retraining. Lifelong model editing is the most challenging task that caters to the continuous editing requirements of LLMs. Prior works primarily focus on single or batch editing; nevertheless, these methods fall short in lifelong editing scenarios due to catastrophic knowledge forgetting and the degradation of model performance. Although retrieval-based methods alleviate these issues, they are impeded by slow and cumbersome processes of integrating the retrieved knowledge into the model. In this work, we introduce RECIPE, a RetriEval-augmented ContInuous Prompt lEarning method, to boost editing efficacy and inference efficiency in lifelong learning. RECIPE first converts knowledge statements into short and informative continuous prompts, prefixed to the LLM's input query embedding, to efficiently refine the response grounded on the knowledge. It further integrates the Knowledge Sentinel (KS) that acts as an intermediary to calculate a dynamic threshold, determining whether the retrieval repository contains relevant knowledge. Our retriever and prompt encoder are jointly trained to achieve editing properties, i.e., reliability, generality, and locality. In our experiments, RECIPE is assessed extensively across multiple LLMs and editing datasets, where it achieves superior editing performance. RECIPE also demonstrates its capability to maintain the overall performance of LLMs alongside showcasing fast editing and inference speed.
Authors: DeepSeek-AI
Abstract: We present DeepSeek-V2, a strong Mixture-of-Experts (MoE) language model characterized by economical training and efficient inference. It comprises 236B total parameters, of which 21B are activated for each token, and supports a context length of 128K tokens. DeepSeek-V2 adopts innovative architectures including Multi-head Latent Attention (MLA) and DeepSeekMoE. MLA guarantees efficient inference through significantly compressing the Key-Value (KV) cache into a latent vector, while DeepSeekMoE enables training strong models at an economical cost through sparse computation. Compared with DeepSeek 67B, DeepSeek-V2 achieves significantly stronger performance, and meanwhile saves 42.5% of training costs, reduces the KV cache by 93.3%, and boosts the maximum generation throughput to 5.76 times. We pretrain DeepSeek-V2 on a high-quality and multi-source corpus consisting of 8.1T tokens, and further perform Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) to fully unlock its potential. Evaluation results show that, even with only 21B activated parameters, DeepSeek-V2 and its chat versions still achieve top-tier performance among open-source models.
Authors: Reem I. Masoud, Ziquan Liu, Martin Ferianc, Philip Treleaven, Miguel Rodrigues
Abstract: The deployment of large language models (LLMs) raises concerns regarding their cultural misalignment and potential ramifications on individuals and societies with diverse cultural backgrounds. While the discourse has focused mainly on political and social biases, our research proposes a Cultural Alignment Test (Hoftede's CAT) to quantify cultural alignment using Hofstede's cultural dimension framework, which offers an explanatory cross-cultural comparison through the latent variable analysis. We apply our approach to quantitatively evaluate LLMs, namely Llama 2, GPT-3.5, and GPT-4, against the cultural dimensions of regions like the United States, China, and Arab countries, using different prompting styles and exploring the effects of language-specific fine-tuning on the models' behavioural tendencies and cultural values. Our results quantify the cultural alignment of LLMs and reveal the difference between LLMs in explanatory cultural dimensions. Our study demonstrates that while all LLMs struggle to grasp cultural values, GPT-4 shows a unique capability to adapt to cultural nuances, particularly in Chinese settings. However, it faces challenges with American and Arab cultures. The research also highlights that fine-tuning LLama 2 models with different languages changes their responses to cultural questions, emphasizing the need for culturally diverse development in AI for worldwide acceptance and ethical use. For more details or to contribute to this research, visit our GitHub page https://github.com/reemim/Hofstedes_CAT/
Authors: Leixin Yang, Yu Xiang
Abstract: Mixup is an effective data augmentation method that generates new augmented samples by aggregating linear combinations of different original samples. However, if there are noises or aberrant features in the original samples, Mixup may propagate them to the augmented samples, leading to over-sensitivity of the model to these outliers . To solve this problem, this paper proposes a new Mixup method called AMPLIFY. This method uses the Attention mechanism of Transformer itself to reduce the influence of noises and aberrant values in the original samples on the prediction results, without increasing additional trainable parameters, and the computational cost is very low, thereby avoiding the problem of high resource consumption in common Mixup methods such as Sentence Mixup . The experimental results show that, under a smaller computational resource cost, AMPLIFY outperforms other Mixup methods in text classification tasks on 7 benchmark datasets, providing new ideas and new ways to further improve the performance of pre-trained models based on the Attention mechanism, such as BERT, ALBERT, RoBERTa, and GPT. Our code can be obtained at https://github.com/kiwi-lilo/AMPLIFY.
Authors: Songyun Qu, Shixin Zhao, Bing Li, Yintao He, Xuyi Cai, Lei Zhang, Ying Wang
Abstract: In recent years, various computing-in-memory (CIM) processors have been presented, showing superior performance over traditional architectures. To unleash the potential of various CIM architectures, such as device precision, crossbar size, and crossbar number, it is necessary to develop compilation tools that are fully aware of the CIM architectural details and implementation diversity. However, due to the lack of architectural support in current popular open-source compiling stacks, existing CIM designs either manually deploy networks or build their own compilers, which is time-consuming and labor-intensive. Although some works expose the specific CIM device programming interfaces to compilers, they are often bound to a fixed CIM architecture, lacking the flexibility to support the CIM architectures with different computing granularity. On the other hand, existing compilation works usually consider the scheduling of limited operation types (such as crossbar-bound matrix-vector multiplication). Unlike conventional processors, CIM accelerators are featured by their diverse architecture, circuit, and device, which cannot be simply abstracted by a single level if we seek to fully explore the advantages brought by CIM. Therefore, we propose CIM-MLC, a universal multi-level compilation framework for general CIM architectures. We first establish a general hardware abstraction for CIM architectures and computing modes to represent various CIM accelerators. Based on the proposed abstraction, CIM-MLC can compile tasks onto a wide range of CIM accelerators having different devices, architectures, and programming interfaces. More importantly, compared with existing compilation work, CIM-MLC can explore the mapping and scheduling strategies across multiple architectural tiers, which form a tractable yet effective design space, to achieve better scheduling and instruction generation results.
Authors: Z Liu, J Lou, W Bao, Y Hu, B Li, Z Qin, K Ren
Abstract: Fine-tuning on task-specific datasets is a widely-embraced paradigm of harnessing the powerful capability of pretrained LLMs for various downstream tasks. Due to the popularity of LLMs fine-tuning and its accompanying privacy concerns, differentially private (DP) fine-tuning of pretrained LLMs has been widely used to safeguarding the privacy of task-specific datasets. Lying at the design core of DP LLM fine-tuning methods is the satisfactory tradeoff among privacy, utility, and scalability. Most existing methods build upon the seminal work of DP-SGD. Despite pushing the scalability of DP-SGD to its limit, DP-SGD-based fine-tuning methods are unfortunately limited by the inherent inefficiency of SGD. In this paper, we investigate the potential of DP zeroth-order methods for LLM pretraining, which avoids the scalability bottleneck of SGD by approximating the gradient with the more efficient zeroth-order gradient. Rather than treating the zeroth-order method as a drop-in replacement for SGD, this paper presents a comprehensive study both theoretically and empirically. First, we propose the stagewise DP zeroth-order method (DP-ZOSO) that dynamically schedules key hyperparameters. This design is grounded on the synergy between DP random perturbation and the gradient approximation error of the zeroth-order method, and its effect on fine-tuning trajectory. We provide theoretical analysis for both proposed methods. We conduct extensive empirical analysis on both encoder-only masked language model and decoder-only autoregressive language model, achieving impressive results in terms of scalability and utility (compared with DPZero, DP-ZOPO improves 4.5% on SST-5, 5.5% on MNLI with RoBERTa-Large and 9.2% on CB, 3.9% on BoolQ with OPT-2.7B when $\epsilon=4$).
Authors: Vincent Freiberger, Erik Buchmann
Abstract: Privacy policies are expected to inform data subjects about their data protection rights and should explain the data controller's data management practices. Privacy policies only fulfill their purpose, if they are correctly interpreted, understood, and trusted by the data subject. This implies that a privacy policy is written in a fair way, e.g., it does not use polarizing terms, does not require a certain education, or does not assume a particular social background. We outline our approach to assessing fairness in privacy policies. We identify from fundamental legal sources and fairness research, how the dimensions informational fairness, representational fairness and ethics / morality are related to privacy policies. We propose options to automatically assess policies in these fairness dimensions, based on text statistics, linguistic methods and artificial intelligence. We conduct initial experiments with German privacy policies to provide evidence that our approach is applicable. Our experiments indicate that there are issues in all three dimensions of fairness. This is important, as future privacy policies may be used in a corpus for legal artificial intelligence models.
Authors: Yiqing Xie, Alex Xie, Divyanshu Sheth, Pengfei Liu, Daniel Fried, Carolyn Rose
Abstract: To facilitate evaluation of code generation systems across diverse scenarios, we present CodeBenchGen, a framework to create scalable execution-based benchmarks that only requires light guidance from humans. Specifically, we leverage a large language model (LLM) to convert an arbitrary piece of code into an evaluation example, including test cases for execution-based evaluation. We illustrate the usefulness of our framework by creating a dataset, Exec-CSN, which includes 1,931 examples involving 293 libraries revised from code in 367 GitHub repositories taken from the CodeSearchNet dataset. To demonstrate the complexity and solvability of examples in Exec-CSN, we present a human study demonstrating that 81.3% of the examples can be solved by humans and 61% are rated as "requires effort to solve". We conduct code generation experiments on open-source and proprietary models and analyze the performance of both humans and models. We provide the code at https://github.com/Veronicium/CodeBenchGen.
Authors: Sher Badshah, Hassan Sajjad
Abstract: Scale is often attributed as one of the factors that cause an increase in the performance of LLMs, resulting in models with billion and trillion parameters. One of the limitations of such large models is the high computational requirements that limit their usage, deployment, and debugging in resource-constrained scenarios. Two commonly used alternatives to bypass these limitations are to use the smaller versions of LLMs (e.g. Llama 7B instead of Llama 70B) and lower the memory requirements by using quantization. While these approaches effectively address the limitation of resources, their impact on model performance needs thorough examination. In this study, we perform a comprehensive evaluation to investigate the effect of model scale and quantization on the performance. We experiment with two major families of open-source instruct models ranging from 7 billion to 70 billion parameters. Our extensive zero-shot experiments across various tasks including natural language understanding, reasoning, misinformation detection, and hallucination reveal that larger models generally outperform their smaller counterparts, suggesting that scale remains an important factor in enhancing performance. We found that larger models show exceptional resilience to precision reduction and can maintain high accuracy even at 4-bit quantization for numerous tasks and they serve as a better solution than using smaller models at high precision under similar memory requirements.
Authors: Tyler Bikaun, Michael Stewart, Wei Liu
Abstract: This paper presents CleanGraph, an interactive web-based tool designed to facilitate the refinement and completion of knowledge graphs. Maintaining the reliability of knowledge graphs, which are grounded in high-quality and error-free facts, is crucial for real-world applications such as question-answering and information retrieval systems. These graphs are often automatically assembled from textual sources by extracting semantic triples via information extraction. However, assuring the quality of these extracted triples, especially when dealing with large or low-quality datasets, can pose a significant challenge and adversely affect the performance of downstream applications. CleanGraph allows users to perform Create, Read, Update, and Delete (CRUD) operations on their graphs, as well as apply models in the form of plugins for graph refinement and completion tasks. These functionalities enable users to enhance the integrity and reliability of their graph data. A demonstration of CleanGraph and its source code can be accessed at https://github.com/nlp-tlp/CleanGraph under the MIT License.