Authors: Kai Liu, Ze Chen, Zhihang Fu, Rongxin Jiang, Fan Zhou, Yaowu Chen, Yue Wu, Jieping Ye
Abstract: This paper presents a pioneering methodology, termed StructTuning, to efficiently transform foundation Large Language Models (LLMs) into domain specialists. It significantly minimizes the training corpus requirement to a mere 0.3% while achieving an impressive 50% of traditional knowledge injection performance. Our method is inspired by the educational processes for human students, particularly how structured domain knowledge from textbooks is absorbed and then applied to tackle real-world challenges through specific exercises. Based on this, we propose a novel two-stage knowledge injection strategy: Structure-aware Continual Pre-Training (SCPT) and Structure-aware Supervised Fine-Tuning (SSFT). In the SCPT phase, we organize the training data into an auto-generated taxonomy of domain knowledge, enabling LLMs to effectively memorize textual segments linked to specific expertise within the taxonomy's architecture. Subsequently, in the SSFT phase, we explicitly prompt models to reveal the underlying knowledge structure in their outputs, leveraging this structured domain insight to address practical problems adeptly. Our ultimate method has undergone extensive evaluations across model architectures and scales, using closed-book question-answering tasks on LongBench and MMedBench datasets. Remarkably, our method matches 50% of the improvement displayed by the state-of-the-art MMedLM2 on MMedBench, but with only 0.3% quantity of the training corpus. This breakthrough showcases the potential to scale up our StructTuning for stronger domain-specific LLMs. Code will be made public soon.
Authors: Sourabrata Mukherjee, Mateusz Lango, Zdenek Kasner, Ondrej Du\v{s}ek
Abstract: Text style transfer (TST) is an important task in controllable text generation, which aims to control selected attributes of language use, such as politeness, formality, or sentiment, without altering the style-independent content of the text. The field has received considerable research attention in recent years and has already been covered in several reviews, but the focus has mostly been on the development of new algorithms and learning from different types of data (supervised, unsupervised, out-of-domain, etc.) and not so much on the application side. However, TST-related technologies are gradually reaching a production- and deployment-ready level, and therefore, the inclusion of the application perspective in TST research becomes crucial. Similarly, the often overlooked ethical considerations of TST technology have become a pressing issue. This paper presents a comprehensive review of TST applications that have been researched over the years, using both traditional linguistic approaches and more recent deep learning methods. We discuss current challenges, future research directions, and ethical implications of TST applications in text generation. By providing a holistic overview of the landscape of TST applications, we hope to stimulate further research and contribute to a better understanding of the potential as well as ethical considerations associated with TST.
Authors: Zhuowan Li, Cheng Li, Mingyang Zhang, Qiaozhu Mei, Michael Bendersky
Abstract: Retrieval Augmented Generation (RAG) has been a powerful tool for Large Language Models (LLMs) to efficiently process overly lengthy contexts. However, recent LLMs like Gemini-1.5 and GPT-4 show exceptional capabilities to understand long contexts directly. We conduct a comprehensive comparison between RAG and long-context (LC) LLMs, aiming to leverage the strengths of both. We benchmark RAG and LC across various public datasets using three latest LLMs. Results reveal that when resourced sufficiently, LC consistently outperforms RAG in terms of average performance. However, RAG's significantly lower cost remains a distinct advantage. Based on this observation, we propose Self-Route, a simple yet effective method that routes queries to RAG or LC based on model self-reflection. Self-Route significantly reduces the computation cost while maintaining a comparable performance to LC. Our findings provide a guideline for long-context applications of LLMs using RAG and LC.
Authors: Fabrice Lamarche, Philippe Langlais
Abstract: Open Information Extraction (OIE) is a field of natural language processing that aims to present textual information in a format that allows it to be organized, analyzed and reflected upon. Numerous OIE systems are developed, claiming ever-increasing performance, marking the need for objective benchmarks. BenchIE is the latest reference we know of. Despite being very well thought out, we noticed a number of issues we believe are limiting. Therefore, we propose $\textit{BenchIE}^{FL}$, a new OIE benchmark which fully enforces the principles of BenchIE while containing fewer errors, omissions and shortcomings when candidate facts are matched towards reference ones. $\textit{BenchIE}^{FL}$ allows insightful conclusions to be drawn on the actual performance of OIE extractors.
Authors: Georgios Kollias, Payel Das, Subhajit Chaudhury
Abstract: Addressing the issue of hallucinations in large language models (LLMs) is a critical challenge. As the cognitive mechanisms of hallucination have been related to memory, here we explore hallucination for LLM that is enabled with explicit memory mechanisms. We empirically demonstrate that by simply scaling the readout vector that constrains generation in a memory-augmented LLM decoder, hallucination mitigation can be achieved in a training-free manner. Our method is geometry-inspired and outperforms a state-of-the-art LLM editing method on the task of generation of Wikipedia-like biography entries both in terms of generation quality and runtime complexity.
Authors: Yeongbin Seo, Dongha Lee, Jinyoung Yeo
Abstract: Previous studies on continual knowledge learning (CKL) in large language models (LLMs) have predominantly focused on approaches such as regularization, architectural modifications, and rehearsal techniques to mitigate catastrophic forgetting. However, these methods naively inherit the inefficiencies of standard training procedures, indiscriminately applying uniform weight across all tokens, which can lead to unnecessary parameter updates and increased forgetting. To address these shortcomings, we propose a novel CKL approach termed Train-Attention-Augmented Language Model (TAALM), which enhances learning efficiency by dynamically predicting and applying weights to tokens based on their usefulness. This method employs a meta-learning framework that optimizes token importance predictions, facilitating targeted knowledge updates and minimizing forgetting. Also, we observe that existing benchmarks do not clearly exhibit the trade-off between learning and retaining, therefore we propose a new benchmark, \textsc{LAMA-ckl}, to address this issue. Through experiments conducted on both newly introduced and established CKL benchmarks, TAALM proves the state-of-the-art performance upon the baselines, and also shows synergistic compatibility when integrated with previous CKL approaches.
Authors: Xiuying Chen, Tairan Wang, Taicheng Guo, Kehan Guo, Juexiao Zhou, Haoyang Li, Mingchen Zhuge, J\"urgen Schmidhuber, Xin Gao, Xiangliang Zhang
Abstract: Question Answering (QA) effectively evaluates language models' reasoning and knowledge depth. While QA datasets are plentiful in areas like general domain and biomedicine, academic chemistry is less explored. Chemical QA plays a crucial role in both education and research by effectively translating complex chemical information into readily understandable format. Addressing this gap, we introduce ScholarChemQA, a large-scale QA dataset constructed from chemical papers. This dataset reflects typical real-world challenges, including an imbalanced data distribution and a substantial amount of unlabeled data that can be potentially useful. Correspondingly, we introduce a QAMatch model, specifically designed to effectively answer chemical questions by fully leveraging our collected data. We first address the issue of imbalanced label distribution by re-weighting the instance-wise loss based on the inverse frequency of each class, ensuring minority classes are not dominated by majority ones during optimization. Next, we utilize the unlabeled data to enrich the learning process, generating a variety of augmentations based on a SoftMix operation and ensuring their predictions align with the same target, i.e., pseudo-labels. To ensure the quality of the pseudo-labels, we propose a calibration procedure aimed at closely aligning the pseudo-label estimates of individual samples with a desired ground truth distribution. Experiments show that our QAMatch significantly outperforms the recent similar-scale baselines and Large Language Models (LLMs) not only on our ScholarChemQA dataset but also on four benchmark datasets. We hope our benchmark and model can facilitate and promote more research on chemical QA.
Authors: Jaewoong Choi, Janghyeok Yoon, Changyong Lee
Abstract: Despite the usefulness of machine learning approaches for the early screening of potential breakthrough technologies, their practicality is often hindered by opaque models. To address this, we propose an interpretable machine learning approach to predicting future citation counts from patent texts using a patent-specific hierarchical attention network (PatentHAN) model. Central to this approach are (1) a patent-specific pre-trained language model, capturing the meanings of technical words in patent claims, (2) a hierarchical network structure, enabling detailed analysis at the claim level, and (3) a claim-wise self-attention mechanism, revealing pivotal claims during the screening process. A case study of 35,376 pharmaceutical patents demonstrates the effectiveness of our approach in early screening of potential breakthrough technologies while ensuring interpretability. Furthermore, we conduct additional analyses using different language models and claim types to examine the robustness of the approach. It is expected that the proposed approach will enhance expert-machine collaboration in identifying breakthrough technologies, providing new insight derived from text mining into technological value.
Authors: Huimin Lu, Masaru Isonuma, Junichiro Mori, Ichiro Sakata
Abstract: Large language models (LLMs) often inherit biases from vast amounts of training corpora. Traditional debiasing methods, while effective to some extent, do not completely eliminate memorized biases and toxicity in LLMs. In this paper, we study an unlearning-based approach to debiasing in LLMs by performing gradient ascent on hate speech against minority groups, i.e., minimizing the likelihood of biased or toxic content. Specifically, we propose a mask language modeling unlearning technique, which unlearns the harmful part of the text. This method enables LLMs to selectively forget and disassociate from biased and harmful content. Experimental results demonstrate the effectiveness of our approach in diminishing bias while maintaining the language modeling abilities. Surprisingly, the results also unveil an unexpected potential for cross-domain transfer unlearning: debiasing in one bias form (e.g. gender) may contribute to mitigating others (e.g. race and religion).
Authors: Sa\"uc Abadal Lloret, Shehzaad Dhuliawala, Keerthiram Murugesan, Mrinmaya Sachan
Abstract: We present ALT (ALignment with Textual feedback), an approach that aligns language models with user preferences expressed in text. We argue that text offers greater expressiveness, enabling users to provide richer feedback than simple comparative preferences and this richer feedback can lead to more efficient and effective alignment. ALT aligns the model by conditioning its generation on the textual feedback. Our method relies solely on language modeling techniques and requires minimal hyper-parameter tuning, though it still presents the main benefits of RL-based alignment algorithms and can effectively learn from textual feedback. We explore the efficacy and efficiency of textual feedback across different tasks such as toxicity reduction, summarization, and dialog response generation. We find that ALT outperforms PPO for the task of toxicity reduction while being able to match its performance on summarization with only 20% of the samples. We also explore how ALT can be used with feedback provided by an existing LLM where we explore an LLM providing constrained and unconstrained textual feedback. We also outline future directions to align models with natural language feedback.
Authors: Yujian Liu, Yang Zhang, Tommi Jaakkola, Shiyu Chang
Abstract: This paper investigates Who's Harry Potter (WHP), a pioneering yet insufficiently understood method for LLM unlearning. We explore it in two steps. First, we introduce a new task of LLM targeted unlearning, where given an unlearning target (e.g., a person) and some unlearning documents, we aim to unlearn only the information about the target, rather than everything in the unlearning documents. We further argue that a successful unlearning should satisfy criteria such as not outputting gibberish, not fabricating facts about the unlearning target, and not releasing factual information under jailbreak attacks. Second, we construct a causal intervention framework for targeted unlearning, where the knowledge of the unlearning target is modeled as a confounder between LLM input and output, and the unlearning process as a deconfounding process. This framework justifies and extends WHP, deriving a simple unlearning algorithm that includes WHP as a special case. Experiments on existing and new datasets show that our approach, without explicitly optimizing for the aforementioned criteria, achieves competitive performance in all of them. Our code is available at https://github.com/UCSB-NLP-Chang/causal_unlearn.git.
Authors: Anhao Zhao, Fanghua Ye, Jinlan Fu, Xiaoyu Shen
Abstract: Large language models (LLMs) exhibit remarkable in-context learning (ICL) capabilities. However, the underlying working mechanism of ICL remains poorly understood. Recent research presents two conflicting views on ICL: One attributes it to LLMs' inherent ability of task recognition, deeming label correctness and shot numbers of demonstrations as not crucial; the other emphasizes the impact of similar examples in the demonstrations, stressing the need for label correctness and more shots. In this work, we provide a Two-Dimensional Coordinate System that unifies both views into a systematic framework. The framework explains the behavior of ICL through two orthogonal variables: whether LLMs can recognize the task and whether similar examples are presented in the demonstrations. We propose the peak inverse rank metric to detect the task recognition ability of LLMs and study LLMs' reactions to different definitions of similarity. Based on these, we conduct extensive experiments to elucidate how ICL functions across each quadrant on multiple representative classification tasks. Finally, we extend our analyses to generation tasks, showing that our coordinate system can also be used to interpret ICL for generation tasks effectively.
Authors: Seungyoon Kim, Seungone Kim
Abstract: Large language model (LLM)-based evaluation pipelines have demonstrated their capability to robustly evaluate machine-generated text. Extending this methodology to assess human-written text could significantly benefit educational settings by providing direct feedback to enhance writing skills, although this application is not straightforward. In this paper, we investigate whether LLMs can effectively assess human-written text for educational purposes. We collected 100 texts from 32 Korean students across 15 types of writing and employed GPT-4-Turbo to evaluate them using grammaticality, fluency, coherence, consistency, and relevance as criteria. Our analyses indicate that LLM evaluators can reliably assess grammaticality and fluency, as well as more objective types of writing, though they struggle with other criteria and types of writing. We publicly release our dataset and feedback.
Authors: Sara Vera Marjanovi\'c, Haeun Yu, Pepa Atanasova, Maria Maistro, Christina Lioma, Isabelle Augenstein
Abstract: Knowledge-intensive language understanding tasks require Language Models (LMs) to integrate relevant context, mitigating their inherent weaknesses, such as incomplete or outdated knowledge. Nevertheless, studies indicate that LMs often ignore the provided context as it can conflict with the pre-existing LM's memory learned during pre-training. Moreover, conflicting knowledge can already be present in the LM's parameters, termed intra-memory conflict. Existing works have studied the two types of knowledge conflicts only in isolation. We conjecture that the (degree of) intra-memory conflicts can in turn affect LM's handling of context-memory conflicts. To study this, we introduce the DYNAMICQA dataset, which includes facts with a temporal dynamic nature where a fact can change with a varying time frequency and disputable dynamic facts, which can change depending on the viewpoint. DYNAMICQA is the first to include real-world knowledge conflicts and provide context to study the link between the different types of knowledge conflicts. With the proposed dataset, we assess the use of uncertainty for measuring the intra-memory conflict and introduce a novel Coherent Persuasion (CP) score to evaluate the context's ability to sway LM's semantic output. Our extensive experiments reveal that static facts, which are unlikely to change, are more easily updated with additional context, relative to temporal and disputable facts.
Authors: Yixiu Liu, Yuxiang Zheng, Shijie Xia, Yuan Guo, Jiajun Li, Yi Tu, Chaoling Song, Pengfei Liu
Abstract: The deployment of Large Language Models (LLMs) in content generation raises significant safety concerns, particularly regarding the transparency and interpretability of content evaluations. Current methods, primarily focused on binary safety classifications, lack mechanisms for detailed critique, limiting their utility for model improvement and user trust. To address these limitations, we introduce SAFETY-J, a bilingual generative safety evaluator for English and Chinese with critique-based judgment. SAFETY-J utilizes a robust training dataset that includes diverse dialogues and augmented query-response pairs to assess safety across various scenarios comprehensively. We establish an automated meta-evaluation benchmark that objectively assesses the quality of critiques with minimal human intervention, facilitating scalable and continuous improvement. Additionally, SAFETY-J employs an iterative preference learning technique to dynamically refine safety assessments based on meta-evaluations and critiques. Our evaluations demonstrate that SAFETY-J provides more nuanced and accurate safety evaluations, thereby enhancing both critique quality and predictive reliability in complex content scenarios. To facilitate further research and application, we will open-source SAFETY-J's training protocols, datasets, and code.
Authors: Jinghong Li, Wen Gu, Koichi Ota, Shinobu Hasegawa
Abstract: With the exponential growth in the number of papers and the trend of AI research, the use of Generative AI for information retrieval and question-answering has become popular for conducting research surveys. However, novice researchers unfamiliar with a particular field may not significantly improve their efficiency in interacting with Generative AI because they have not developed divergent thinking in that field. This study aims to develop an in-depth Survey Forest Diagram that guides novice researchers in divergent thinking about the research topic by indicating the citation clues among multiple papers, to help expand the survey perspective for novice researchers.
Authors: Anastasiia Sedova, Robert Litschko, Diego Frassinelli, Benjamin Roth, Barbara Plank
Abstract: One of the major aspects contributing to the striking performance of large language models (LLMs) is the vast amount of factual knowledge accumulated during pre-training. Yet, many LLMs suffer from self-inconsistency, which raises doubts about their trustworthiness and reliability. In this paper, we focus on entity type ambiguity and analyze current state-of-the-art LLMs for their proficiency and consistency in applying their factual knowledge when prompted for entities under ambiguity. To do so, we propose an evaluation protocol that disentangles knowing from applying knowledge, and test state-of-the-art LLMs on 49 entities. Our experiments reveal that LLMs perform poorly with ambiguous prompts, achieving only 80% accuracy. Our results further demonstrate systematic discrepancies in LLM behavior and their failure to consistently apply information, indicating that the models can exhibit knowledge without being able to utilize it, significant biases for preferred readings, as well as self inconsistencies. Our study highlights the importance of handling entity ambiguity in future for more trustworthy LLMs
Authors: Bernardo Consoli, Xizhi Wu, Song Wang, Xinyu Zhao, Yanshan Wang, Justin Rousseau, Tom Hartvigsen, Li Shen, Huanmei Wu, Yifan Peng, Qi Long, Tianlong Chen, Ying Ding
Abstract: Extracting social determinants of health (SDoH) from unstructured medical notes depends heavily on labor-intensive annotations, which are typically task-specific, hampering reusability and limiting sharing. In this study we introduced SDoH-GPT, a simple and effective few-shot Large Language Model (LLM) method leveraging contrastive examples and concise instructions to extract SDoH without relying on extensive medical annotations or costly human intervention. It achieved tenfold and twentyfold reductions in time and cost respectively, and superior consistency with human annotators measured by Cohen's kappa of up to 0.92. The innovative combination of SDoH-GPT and XGBoost leverages the strengths of both, ensuring high accuracy and computational efficiency while consistently maintaining 0.90+ AUROC scores. Testing across three distinct datasets has confirmed its robustness and accuracy. This study highlights the potential of leveraging LLMs to revolutionize medical note classification, demonstrating their capability to achieve highly accurate classifications with significantly reduced time and cost.
Authors: Fufangchen Zhao, Guoqiang Jin, Rui Zhao, Jiangheng Huang, Fei Tan
Abstract: In this work, we report our efforts to advance the standard operation procedure of developing Large Language Models (LLMs) or LLMs-based systems or services in industry. We introduce the concept of Large Language Model Development Lifecycle (LDLC) and then highlight the importance of consistency test in ensuring the delivery quality. The principled solution of consistency test, however, is usually overlooked by industrial practitioners and not urgent in academia, and current practical solutions are insufficiently rigours and labor-intensive. We thus propose a simple yet effective consistency test protocol, named SimCT. SimCT is mainly to proactively check the consistency across different development stages of "bare metal" LLMs or associated services without accessing the model artifacts, in an attempt to expedite the delivery by reducing the back-and-forth alignment communications among multiple teams involved in different development stages. Specifically, SimCT encompasses response-wise and model-wise tests. We implement the protocol with LightGBM and Student's t-test for two components respectively, and perform extensive experiments to substantiate the effectiveness of SimCT and the involved components.
Authors: Jan Lehe\v{c}ka, Josef V. Psutka, Lubo\v{s} \v{S}m\'idl, Pavel Ircing, Josef Psutka
Abstract: In this paper, we are comparing monolingual Wav2Vec 2.0 models with various multilingual models to see whether we could improve speech recognition performance on a unique oral history archive containing a lot of mixed-language sentences. Our main goal is to push forward research on this unique dataset, which is an extremely valuable part of our cultural heritage. Our results suggest that monolingual speech recognition models are, in most cases, superior to multilingual models, even when processing the oral history archive full of mixed-language sentences from non-native speakers. We also performed the same experiments on the public CommonVoice dataset to verify our results. We are contributing to the research community by releasing our pre-trained models to the public.
Authors: Hamad Zogan, Imran Razzak, Shoaib Jameel, Guandong Xu
Abstract: Social media posts provide valuable insight into the narrative of users and their intentions, including providing an opportunity to automatically model whether a social media user is depressed or not. The challenge lies in faithfully modelling user narratives from their online social media posts, which could potentially be useful in several different applications. We have developed a novel and effective model called \texttt{NarrationDep}, which focuses on detecting narratives associated with depression. By analyzing a user's tweets, \texttt{NarrationDep} accurately identifies crucial narratives. \texttt{NarrationDep} is a deep learning framework that jointly models individual user tweet representations and clusters of users' tweets. As a result, \texttt{NarrationDep} is characterized by a novel two-layer deep learning model: the first layer models using social media text posts, and the second layer learns semantic representations of tweets associated with a cluster. To faithfully model these cluster representations, the second layer incorporates a novel component that hierarchically learns from users' posts. The results demonstrate that our framework outperforms other comparative models including recently developed models on a variety of datasets.
Authors: Abhijith R. Beeravolu, Mirjam Jonkman, Sami Azam, Friso De Boer
Abstract: Recent advancements in natural language processing (NLP) have led to automation in various domains. However, clinical NLP often relies on benchmark datasets that may not reflect real-world scenarios accurately. Automatic ICD coding, a vital NLP task, typically uses outdated and imbalanced datasets like MIMIC-III, with existing methods yielding micro-averaged F1 scores between 0.4 and 0.7 due to many false positives. Our research introduces an enhanced approach to ICD coding that improves F1 scores by using chapter-based named entities and attentional models. This method categorizes discharge summaries into ICD-9 Chapters and develops attentional models with chapter-specific data, eliminating the need to consider external data for code identification. For categorization, we use Chapter-IV to de-bias and influence key entities and weights without neural networks, creating accurate thresholds and providing interpretability for human validation. Post-validation, we develop attentional models for three frequent and three non-frequent codes from Chapter-IV using Bidirectional-Gated Recurrent Units (GRUs) with Attention and Transformer with Multi-head Attention architectures. The average Micro-F1 scores of 0.79 and 0.81 from these models demonstrate significant performance improvements in ICD coding.
Authors: Ke Bao, Chonghuan Yang
Abstract: Named entity recognition on the in-domain supervised and few-shot settings have been extensively discussed in the NLP community and made significant progress. However, cross-domain NER, a more common task in practical scenarios, still poses a challenge for most NER methods. Previous research efforts in that area primarily focus on knowledge transfer such as correlate label information from source to target domains but few works pay attention to the problem of label conflict. In this study, we introduce a label alignment and reassignment approach, namely LAR, to address this issue for enhanced cross-domain named entity recognition, which includes two core procedures: label alignment between source and target domains and label reassignment for type inference. The process of label reassignment can significantly be enhanced by integrating with an advanced large-scale language model such as ChatGPT. We conduct an extensive range of experiments on NER datasets involving both supervised and zero-shot scenarios. Empirical experimental results demonstrate the validation of our method with remarkable performance under the supervised and zero-shot out-of-domain settings compared to SOTA methods.
Authors: Yuyang Ding, Hanglei Hu, Jie Zhou, Qin Chen, Bo Jiang, Liang He
Abstract: With the introduction of large language models (LLMs), automatic math reasoning has seen tremendous success. However, current methods primarily focus on providing solutions or using techniques like Chain-of-Thought to enhance problem-solving accuracy. In this paper, we focus on improving the capability of mathematics teaching via a Socratic teaching-based LLM (\texttt{SocraticLLM}), which guides learners toward profound thinking with clarity and self-discovery via conversation. We collect and release a high-quality mathematical teaching dataset, named \texttt{SocraticMATH}, which provides Socratic-style conversations of problems with extra knowledge. Also, we propose a knowledge-enhanced LLM as a strong baseline to generate reliable responses with review, guidance/heuristic, rectification, and summarization. Experimental results show the great advantages of \texttt{SocraticLLM} by comparing it with several strong generative models. The codes and datasets are available on \url{https://github.com/ECNU-ICALK/SocraticMath}.
Authors: Amirreza Naziri, Hossein Zeinali
Abstract: Writing, as an omnipresent form of human communication, permeates nearly every aspect of contemporary life. Consequently, inaccuracies or errors in written communication can lead to profound consequences, ranging from financial losses to potentially life-threatening situations. Spelling mistakes, among the most prevalent writing errors, are frequently encountered due to various factors. This research aims to identify and rectify diverse spelling errors in text using neural networks, specifically leveraging the Bidirectional Encoder Representations from Transformers (BERT) masked language model. To achieve this goal, we compiled a comprehensive dataset encompassing both non-real-word and real-word errors after categorizing different types of spelling mistakes. Subsequently, multiple pre-trained BERT models were employed. To ensure optimal performance in correcting misspelling errors, we propose a combined approach utilizing the BERT masked language model and Levenshtein distance. The results from our evaluation data demonstrate that the system presented herein exhibits remarkable capabilities in identifying and rectifying spelling mistakes, often surpassing existing systems tailored for the Persian language.
Authors: Louis Castricato, Nathan Lile, Rafael Rafailov, Jan-Philipp Fr\"anken, Chelsea Finn
Abstract: The rapid advancement of language models (LMs) necessitates robust alignment with diverse user values. However, current preference optimization approaches often fail to capture the plurality of user opinions, instead reinforcing majority viewpoints and marginalizing minority perspectives. We introduce PERSONA, a reproducible test bed designed to evaluate and improve pluralistic alignment of LMs. We procedurally generate diverse user profiles from US census data, resulting in 1,586 synthetic personas with varied demographic and idiosyncratic attributes. We then generate a large-scale evaluation dataset containing 3,868 prompts and 317,200 feedback pairs obtained from our synthetic personas. Leveraging this dataset, we systematically evaluate LM capabilities in role-playing diverse users, verified through human judges, and the establishment of both a benchmark, PERSONA Bench, for pluralistic alignment approaches as well as an extensive dataset to create new and future benchmarks. The full dataset and benchmarks are available here: https://www.synthlabs.ai/research/persona.
Authors: Itamar Trainin, Omri Abend
Abstract: This paper introduces CovScore, an automatic reference-less methodology for evaluating thematic title sets, extracted from a corpus of documents. While such extraction methods are widely used, evaluating their effectiveness remains an open question. Moreover, some existing practices heavily rely on slow and laborious human annotation procedures. Inspired by recently introduced LLM-based judge methods, we propose a novel methodology that decomposes quality into five main metrics along different aspects of evaluation. This framing simplifies and expedites the manual evaluation process and enables automatic and independent LLM-based evaluation. As a test case, we apply our approach to a corpus of Holocaust survivor testimonies, motivated both by its relevance to title set extraction and by the moral significance of this pursuit. We validate the methodology by experimenting with naturalistic and synthetic title set generation systems and compare their performance with the methodology.
Authors: Yida Zhao, Chao Lou, Kewei Tu
Abstract: Syntactic Transformer language models aim to achieve better generalization through simultaneously modeling syntax trees and sentences. While prior work has been focusing on adding constituency-based structures to Transformers, we introduce Dependency Transformer Grammars (DTGs), a new class of Transformer language model with explicit dependency-based inductive bias. DTGs simulate dependency transition systems with constrained attention patterns by modifying attention masks, incorporate the stack information through relative positional encoding, and augment dependency arc representation with a combination of token embeddings and operation embeddings. When trained on a dataset of sentences annotated with dependency trees, DTGs achieve better generalization while maintaining comparable perplexity with Transformer language model baselines. DTGs also outperform recent constituency-based models, showing that dependency can better guide Transformer language models. Our code is released at https://github.com/zhaoyd1/Dep_Transformer_Grammars.
Authors: T. Ben Thompson (Confirm Labs), Michael Sklar (Confirm Labs)
Abstract: Many publicly available language models have been safety tuned to reduce the likelihood of toxic or liability-inducing text. Users or security analysts attempt to jailbreak or redteam these models with adversarial prompts which cause compliance with requests. One attack method is to apply discrete optimization techniques to the prompt. However, the resulting attack strings are often gibberish text, easily filtered by defenders due to high measured perplexity, and may fail for unseen tasks and/or well-tuned models. In this work, we improve existing algorithms (primarily GCG and BEAST) to develop powerful and fluent attacks on safety-tuned models like Llama-2 and Phi-3. Our technique centers around a new distillation-based approach that encourages the victim model to emulate a toxified finetune, either in terms of output probabilities or internal activations. To encourage human-fluent attacks, we add a multi-model perplexity penalty and a repetition penalty to the objective. We also enhance optimizer strength by allowing token insertions, token swaps, and token deletions and by using longer attack sequences. The resulting process is able to reliably jailbreak the most difficult target models with prompts that appear similar to human-written prompts. On Advbench we achieve attack success rates $>93$% for Llama-2-7B, Llama-3-8B, and Vicuna-7B, while maintaining model-measured perplexity $<33$; we achieve $95$% attack success for Phi-3, though with higher perplexity. We also find a universally-optimized single fluent prompt that induces $>88$% compliance on previously unseen tasks across Llama-2-7B, Phi-3-mini and Vicuna-7B and transfers to other black-box models.
Authors: Jiawei Gu, Zacc Yang, Chuanghao Ding, Rui Zhao, Fei Tan
Abstract: Large Language Models (LLMs) excel in diverse tasks but often underperform in specialized fields due to limited domain-specific or proprietary corpus. Continual pre-training (CPT) enhances LLM capabilities by imbuing new domain-specific or proprietary knowledge while replaying general corpus to prevent catastrophic forgetting. The data mixture ratio of general corpus and domain-specific corpus, however, has been chosen heuristically, leading to sub-optimal training efficiency in practice. In this context, we attempt to re-visit the scaling behavior of LLMs under the hood of CPT, and discover a power-law relationship between loss, mixture ratio, and training tokens scale. We formalize the trade-off between general and domain-specific capabilities, leading to a well-defined Critical Mixture Ratio (CMR) of general and domain data. By striking the balance, CMR maintains the model's general ability and achieves the desired domain transfer, ensuring the highest utilization of available resources. Therefore, if we value the balance between efficiency and effectiveness, CMR can be consider as the optimal mixture ratio.Through extensive experiments, we ascertain the predictability of CMR, and propose CMR scaling law and have substantiated its generalization. These findings offer practical guidelines for optimizing LLM training in specialized domains, ensuring both general and domain-specific performance while efficiently managing training resources.
Authors: Wenting Zhao, Tanya Goyal, Yu Ying Chiu, Liwei Jiang, Benjamin Newman, Abhilasha Ravichander, Khyathi Chandu, Ronan Le Bras, Claire Cardie, Yuntian Deng, Yejin Choi
Abstract: While hallucinations of large language models (LLMs) prevail as a major challenge, existing evaluation benchmarks on factuality do not cover the diverse domains of knowledge that the real-world users of LLMs seek information about. To bridge this gap, we introduce WildHallucinations, a benchmark that evaluates factuality. It does so by prompting LLMs to generate information about entities mined from user-chatbot conversations in the wild. These generations are then automatically fact-checked against a systematically curated knowledge source collected from web search. Notably, half of these real-world entities do not have associated Wikipedia pages. We evaluate 118,785 generations from 15 LLMs on 7,919 entities. We find that LLMs consistently hallucinate more on entities without Wikipedia pages and exhibit varying hallucination rates across different domains. Finally, given the same base models, adding a retrieval component only slightly reduces hallucinations but does not eliminate hallucinations.
Authors: Wenting Zhao, Ge Gao, Claire Cardie, Alexander M. Rush
Abstract: When seeking information from unfamiliar documents, users frequently pose questions that cannot be answered by the documents. While existing large language models (LLMs) identify these unanswerable questions, they do not assist users in reformulating their questions, thereby reducing their overall utility. We curate CouldAsk, an evaluation benchmark composed of existing and new datasets for document-grounded question answering, specifically designed to study reformulating unanswerable questions. We evaluate state-of-the-art open-source and proprietary LLMs on CouldAsk. The results demonstrate the limited capabilities of these models in reformulating questions. Specifically, GPT-4 and Llama2-7B successfully reformulate questions only 26% and 12% of the time, respectively. Error analysis shows that 62% of the unsuccessful reformulations stem from the models merely rephrasing the questions or even generating identical questions. We publicly release the benchmark and the code to reproduce the experiments.
Authors: Michael Saxon, Ari Holtzman, Peter West, William Yang Wang, Naomi Saphra
Abstract: Modern language models (LMs) pose a new challenge in capability assessment. Static benchmarks inevitably saturate without providing confidence in the deployment tolerances of LM-based systems, but developers nonetheless claim that their models have generalized traits such as reasoning or open-domain language understanding based on these flawed metrics. The science and practice of LMs requires a new approach to benchmarking which measures specific capabilities with dynamic assessments. To be confident in our metrics, we need a new discipline of model metrology -- one which focuses on how to generate benchmarks that predict performance under deployment. Motivated by our evaluation criteria, we outline how building a community of model metrology practitioners -- one focused on building tools and studying how to measure system capabilities -- is the best way to meet these needs to and add clarity to the AI discussion.
Authors: Rabiul Awal, Saba Ahmadi, Le Zhang, Aishwarya Agrawal
Abstract: Fine-grained understanding of objects, attributes, and relationships between objects is crucial for visual-language models (VLMs). Existing benchmarks primarily focus on evaluating VLMs' capability to distinguish between two very similar \textit{captions} given an image. In this paper, we introduce a new, challenging benchmark termed \textbf{Vis}ual \textbf{Min}imal-Change Understanding (VisMin), which requires models to predict the correct image-caption match given two images and two captions. The image pair and caption pair contain minimal changes, i.e., only one aspect changes at a time from among the following: \textit{object}, \textit{attribute}, \textit{count}, and \textit{spatial relation}. These changes test the models' understanding of objects, attributes (such as color, material, shape), counts, and spatial relationships between objects. We built an automatic framework using large language models and diffusion models, followed by a rigorous 4-step verification process by human annotators. Empirical experiments reveal that current VLMs exhibit notable deficiencies in understanding spatial relationships and counting abilities. We also generate a large-scale training dataset to finetune CLIP and Idefics2, showing significant improvements in fine-grained understanding across benchmarks and in CLIP's general image-text alignment. We release all resources, including the benchmark, training data, and finetuned model checkpoints, at \url{https://vismin.net/}.
URLs: https://vismin.net/
Authors: Jihyung Kil, Zheda Mai, Justin Lee, Zihe Wang, Kerrie Cheng, Lemeng Wang, Ye Liu, Arpita Chowdhury, Wei-Lun Chao
Abstract: The ability to compare objects, scenes, or situations is crucial for effective decision-making and problem-solving in everyday life. For instance, comparing the freshness of apples enables better choices during grocery shopping, while comparing sofa designs helps optimize the aesthetics of our living space. Despite its significance, the comparative capability is largely unexplored in artificial general intelligence (AGI). In this paper, we introduce CompBench, a benchmark designed to evaluate the comparative reasoning capability of multimodal large language models (MLLMs). CompBench mines and pairs images through visually oriented questions covering eight dimensions of relative comparison: visual attribute, existence, state, emotion, temporality, spatiality, quantity, and quality. We curate a collection of around 40K image pairs using metadata from diverse vision datasets and CLIP similarity scores. These image pairs span a broad array of visual domains, including animals, fashion, sports, and both outdoor and indoor scenes. The questions are carefully crafted to discern relative characteristics between two images and are labeled by human annotators for accuracy and relevance. We use CompBench to evaluate recent MLLMs, including GPT-4V(ision), Gemini-Pro, and LLaVA-1.6. Our results reveal notable shortcomings in their comparative abilities. We believe CompBench not only sheds light on these limitations but also establishes a solid foundation for future enhancements in the comparative capability of MLLMs.
Authors: Pontus Strimling, Joel Krueger, Simon Karlsson
Abstract: Prior research demonstrates that Open AI's GPT models can predict variations in moral opinions between countries but that the accuracy tends to be substantially higher among high-income countries compared to low-income ones. This study aims to replicate previous findings and advance the research by examining how accuracy varies with different types of moral questions. Using responses from the World Value Survey and the European Value Study, covering 18 moral issues across 63 countries, we calculated country-level mean scores for each moral issue and compared them with GPT-4's predictions. Confirming previous findings, our results show that GPT-4 has greater predictive success in high-income than in low-income countries. However, our factor analysis reveals that GPT-4 bases its predictions primarily on a single dimension, presumably reflecting countries' degree of conservatism/liberalism. Conversely, the real-world moral landscape appears to be two-dimensional, differentiating between personal-sexual and violent-dishonest issues. When moral issues are categorized based on their moral domain, GPT-4's predictions are found to be remarkably accurate in the personal-sexual domain, across both high-income (r = .77) and low-income (r = .58) countries. Yet the predictive accuracy significantly drops in the violent-dishonest domain for both high-income (r = .30) and low-income (r = -.16) countries, indicating that GPT-4's one-dimensional world-view does not fully capture the complexity of the moral landscape. In sum, this study underscores the importance of not only considering country-specific characteristics to understand GPT-4's moral understanding, but also the characteristics of the moral issues at hand.
Authors: Qishuai Zhong, Yike Yun, Aixin Sun
Abstract: Our study aims to identify behavior patterns in cultural values exhibited by large language models (LLMs). The studied variants include question ordering, prompting language, and model size. Our experiments reveal that each tested LLM can efficiently behave with different cultural values. More interestingly: (i) LLMs exhibit relatively consistent cultural values when presented with prompts in a single language. (ii) The prompting language e.g., Chinese or English, can influence the expression of cultural values. The same question can elicit divergent cultural values when the same LLM is queried in a different language. (iii) Differences in sizes of the same model (e.g., Llama2-7B vs 13B vs 70B) have a more significant impact on their demonstrated cultural values than model differences (e.g., Llama2 vs Mixtral). Our experiments reveal that query language and model size of LLM are the main factors resulting in cultural value differences.
Authors: Swati Swati, Arjun Roy, Eirini Ntoutsi
Abstract: Despite the large body of work on fairness-aware learning for individual modalities like tabular data, images, and text, less work has been done on multimodal data, which fuses various modalities for a comprehensive analysis. In this work, we investigate the fairness and bias implications of multimodal fusion techniques in the context of multimodal AI-based recruitment systems using the FairCVdb dataset. Our results show that early-fusion closely matches the ground truth for both demographics, achieving the lowest MAEs by integrating each modality's unique characteristics. In contrast, late-fusion leads to highly generalized mean scores and higher MAEs. Our findings emphasise the significant potential of early-fusion for accurate and fair applications, even in the presence of demographic biases, compared to late-fusion. Future research could explore alternative fusion strategies and incorporate modality-related fairness constraints to improve fairness. For code and additional insights, visit: https://github.com/Swati17293/Multimodal-AI-Based-Recruitment-FairCVdb
URLs: https://github.com/Swati17293/Multimodal-AI-Based-Recruitment-FairCVdb
Authors: Erik Johannes Husom, Arda Goknil, Lwin Khin Shar, Sagar Sen
Abstract: In the rapidly evolving realm of artificial intelligence, deploying large language models (LLMs) poses increasingly pressing computational and environmental challenges. This paper introduces MELODI - Monitoring Energy Levels and Optimization for Data-driven Inference - a multifaceted framework crafted to monitor and analyze the energy consumed during LLM inference processes. MELODI enables detailed observations of power consumption dynamics and facilitates the creation of a comprehensive dataset reflective of energy efficiency across varied deployment scenarios. The dataset, generated using MELODI, encompasses a broad spectrum of LLM deployment frameworks, multiple language models, and extensive prompt datasets, enabling a comparative analysis of energy use. Using the dataset, we investigate how prompt attributes, including length and complexity, correlate with energy expenditure. Our findings indicate substantial disparities in energy efficiency, suggesting ample scope for optimization and adoption of sustainable measures in LLM deployment. Our contribution lies not only in the MELODI framework but also in the novel dataset, a resource that can be expanded by other researchers. Thus, MELODI is a foundational tool and dataset for advancing research into energy-conscious LLM deployment, steering the field toward a more sustainable future.
Authors: Daniel Hopp
Abstract: Generative artificial intelligence (AI), and in particular Large Language Models (LLMs), have exploded in popularity and attention since the release to the public of ChatGPT's Generative Pre-trained Transformer (GPT)-3.5 model in November of 2022. Due to the power of these general purpose models and their ability to communicate in natural language, they can be useful in a range of domains, including the work of official statistics and international organizations. However, with such a novel and seemingly complex technology, it can feel as if generative AI is something that happens to an organization, something that can be talked about but not understood, that can be commented on but not contributed to. Additionally, the costs of adoption and operation of proprietary solutions can be both uncertain and high, a barrier for often cost-constrained international organizations. In the face of these challenges, United Nations Trade and Development (UNCTAD), through its Global Crisis Response Group (GCRG), has explored and developed its own open-source Retrieval Augmented Generation (RAG) LLM application. RAG makes LLMs aware of and more useful for the organization's domain and work. Developing in-house solutions comes with pros and cons, with pros including cost, flexibility, and fostering institutional knowledge. Cons include time and skill investments and gaps and application polish and power. The three libraries developed to produce the app, nlp_pipeline for document processing and statistical analysis, local_rag_llm for running a local RAG LLM, and streamlit_rag for the user interface, are publicly available on PyPI and GitHub with Dockerfiles. A fourth library, local_llm_finetune, is also available for fine-tuning existing LLMs which can then be used in the application.
Authors: Jake R. Watts, Joel Sokol
Abstract: This paper proposes a new method for preventing unsafe or otherwise low quality large language model (LLM) outputs, by leveraging the stochasticity of LLMs. We propose a system whereby LLM checkers vote on the acceptability of a generated output, regenerating it if a threshold of disapproval is reached, until sufficient checkers approve. We further propose estimators for cost and failure rate, and based on those estimators and experimental data tailored to the application, we propose an algorithm that achieves a desired failure rate at the least possible cost. We demonstrate that, under these models, failure rate decreases exponentially as a function of cost when voter count and threshold are chosen according to the algorithm, and that the models reasonably estimate the actual performance of such a system in action, even with limited data.
Authors: Binzhe Li, Shurun Wang, Shiqi Wang, Yan Ye
Abstract: In recent years, large visual language models (LVLMs) have shown impressive performance and promising generalization capability in multi-modal tasks, thus replacing humans as receivers of visual information in various application scenarios. In this paper, we pioneer to propose a variable bitrate image compression framework consisting of a pre-editing module and an end-to-end codec to achieve promising rate-accuracy performance for different LVLMs. In particular, instead of optimizing an adaptive pre-editing network towards a particular task or several representative tasks, we propose a new optimization strategy tailored for LVLMs, which is designed based on the representation and discrimination capability with token-level distortion and rank. The pre-editing module and the variable bitrate end-to-end image codec are jointly trained by the losses based on semantic tokens of the large model, which introduce enhanced generalization capability for various data and tasks. {Experimental results demonstrate that the proposed framework could efficiently achieve much better rate-accuracy performance compared to the state-of-the-art coding standard, Versatile Video Coding.} Meanwhile, experiments with multi-modal tasks have revealed the robustness and generalization capability of the proposed framework.
Authors: Jan Lehe\v{c}ka, Zden\v{e}k Hanzl\'i\v{c}ek, Jind\v{r}ich Matou\v{s}ek, Daniel Tihelka
Abstract: In this paper, we experimented with the SpeechT5 model pre-trained on large-scale datasets. We pre-trained the foundation model from scratch and fine-tuned it on a large-scale robust multi-speaker text-to-speech (TTS) task. We tested the model capabilities in a zero- and few-shot scenario. Based on two listening tests, we evaluated the synthetic audio quality and the similarity of how synthetic voices resemble real voices. Our results showed that the SpeechT5 model can generate a synthetic voice for any speaker using only one minute of the target speaker's data. We successfully demonstrated the high quality and similarity of our synthetic voices on publicly known Czech politicians and celebrities.
Authors: Tobias K\"assmann, Yining Liu, Danni Liu
Abstract: With the rise of video production and social media, speech editing has become crucial for creators to address issues like mispronunciations, missing words, or stuttering in audio recordings. This paper explores text-based speech editing methods that modify audio via text transcripts without manual waveform editing. These approaches ensure edited audio is indistinguishable from the original by altering the mel-spectrogram. Recent advancements, such as context-aware prosody correction and advanced attention mechanisms, have improved speech editing quality. This paper reviews state-of-the-art methods, compares key metrics, and examines widely used datasets. The aim is to highlight ongoing issues and inspire further research and innovation in speech editing.
Authors: Zijian Wu, Jiayu Wang, Dahua Lin, Kai Chen
Abstract: Recently, large language models have presented promising results in aiding formal mathematical reasoning. However, their performance is restricted due to the scarcity of formal theorem-proving data, which requires additional effort to be extracted from raw formal language corpora. Meanwhile, a significant amount of human-written formal language corpora remains underutilized. To address this issue, we propose LEAN-GitHub, a dataset consisting of large-scale formal data extracted from almost all Lean 4 repositories on GitHub. After fine-tuning InternLM-math-plus on this dataset, our model achieved accuracies of 48.8% with a single pass and 54.5% with 64 passes on the Lean 4 miniF2F test, surpassing state-of-the-art method at 52%. And it also achieves state-of-the-art on two other Lean 4 benchmarks (ProofNet and Putnam) targeting different fields/levels of math. These results demonstrate that our proposed dataset is beneficial for formal reasoning on a wide range of math topics. We open-source our model at https://GitHub. com/InternLM/InternLM-Math and our data at https://huggingface.co/ datasets/InternLM/Lean-GitHub
Authors: Leo Yu-Ho Lo, Huamin Qu
Abstract: In this study, we address the growing issue of misleading charts, a prevalent problem that undermines the integrity of information dissemination. Misleading charts can distort the viewer's perception of data, leading to misinterpretations and decisions based on false information. The development of effective automatic detection methods for misleading charts is an urgent field of research. The recent advancement of multimodal Large Language Models (LLMs) has introduced a promising direction for addressing this challenge. We explored the capabilities of these models in analyzing complex charts and assessing the impact of different prompting strategies on the models' analyses. We utilized a dataset of misleading charts collected from the internet by prior research and crafted nine distinct prompts, ranging from simple to complex, to test the ability of four different multimodal LLMs in detecting over 21 different chart issues. Through three experiments--from initial exploration to detailed analysis--we progressively gained insights into how to effectively prompt LLMs to identify misleading charts and developed strategies to address the scalability challenges encountered as we expanded our detection range from the initial five issues to 21 issues in the final experiment. Our findings reveal that multimodal LLMs possess a strong capability for chart comprehension and critical thinking in data interpretation. There is significant potential in employing multimodal LLMs to counter misleading information by supporting critical thinking and enhancing visualization literacy. This study demonstrates the applicability of LLMs in addressing the pressing concern of misleading charts.
Authors: Siwei Wu, Kang Zhu, Yu Bai, Yiming Liang, Yizhi Li, Haoning Wu, Jiaheng Liu, Ruibo Liu, Xingwei Qu, Xuxin Cheng, Ge Zhang, Wenhao Huang, Chenghua Lin
Abstract: Given the remarkable success that large visual language models (LVLMs) have achieved in image perception tasks, the endeavor to make LVMLs perceive the world like humans is drawing increasing attention. Current multi-modal benchmarks mainly focus on the objective fact or certain topic related potential knowledge within a image, but overlook the associative relations between multiple images. Therefore, we define a multi-image relation association task, and meticulously curate \textbf{MMRA} benchmark, a \textbf{M}ulti-granularity \textbf{M}ulti-image \textbf{R}elational \textbf{A}ssociation benchmark, consisted of \textbf{1026} samples. In order to systematically and comprehensively evaluate mainstream LVLMs, we establish an associational relation system among images that contain \textbf{11 subtasks} (e.g, UsageSimilarity, SubEvent, etc.) at two granularity levels (i.e., "\textbf{image}" and "\textbf{entity}") according to the relations in ConceptNet. Our experiments demonstrate that, on our MMRA benchmark, current mainstream LVLMs all have their own advantages and disadvantages across different subtasks. It is worth noting that, at the entity level, the performance of all models is worse than that of them at the image level, indicating that the fine-grained multi-image perception task is still challenging for LVLMs. The tasks related to spatial perception are relatively difficult for LVLMs to handle. Furthermore, we find that LVMLs exhibit a good ability to perceive image details, and the key to enhancing their multi-image association capability is to strengthen the reasoning ability of their language model component. All our codes and data are released at htt\url{https://github.com/Wusiwei0410/MMRA}.
Authors: Jesin James, Balamurali B. T., Binu Abeysinghe, Junchen Liu
Abstract: This study investigates discriminative patterns learned by neural networks for accurate speech classification, with a specific focus on vowel classification tasks. By examining the activations and features of neural networks for vowel classification, we gain insights into what the networks "see" in spectrograms. Through the use of class activation mapping, we identify the frequencies that contribute to vowel classification and compare these findings with linguistic knowledge. Experiments on a American English dataset of vowels showcases the explainability of neural networks and provides valuable insights into the causes of misclassifications and their characteristics when differentiating them from unvoiced speech. This study not only enhances our understanding of the underlying acoustic cues in vowel classification but also offers opportunities for improving speech recognition by bridging the gap between abstract representations in neural networks and established linguistic knowledge
Authors: Ivan Khoma, Solomia Fedushko, Zoryana Kunch
Abstract: This article examines the use of manipulation in the coverage of events of the Ukrainian Revolution of Dignity in the mass media, namely in the content of the online newspaper Ukrainian Truth (Ukrainska pravda), online newspaper High Castle (Vysokyi Zamok), and online newspaper ZIK during the public protest, namely during the Ukrainian Revolution of Dignity. Contents of these online newspapers the historical, linguistic, and psychological approaches are used. Also media manipulations in the coverage of events of the Ukrainian Revolution of Dignity are studied. Internet resources that cover news are analyzed. Current and most popular Internet resources are identified. The content of online newspapers is analyzed and statistically processed. Internet content of newspapers by the level of significance of data (very significant data, significant data and insignificant data) is classified. The algorithm of detection of the media manipulations in the highlighting the course of the Ukrainian revolutions based on historical, linguistic, and psychological approaches is designed. Methods of counteracting information attacks in online newspapers are developed.
Authors: Anar Yeginbergen, Rodrigo Agerri
Abstract: Nowadays the medical domain is receiving more and more attention in applications involving Artificial Intelligence as clinicians decision-making is increasingly dependent on dealing with enormous amounts of unstructured textual data. In this context, Argument Mining (AM) helps to meaningfully structure textual data by identifying the argumentative components in the text and classifying the relations between them. However, as it is the case for man tasks in Natural Language Processing in general and in medical text processing in particular, the large majority of the work on computational argumentation has been focusing only on the English language. In this paper, we investigate several strategies to perform AM in medical texts for a language such as Spanish, for which no annotated data is available. Our work shows that automatically translating and projecting annotations (data-transfer) from English to a given target language is an effective way to generate annotated data without costly manual intervention. Furthermore, and contrary to conclusions from previous work for other sequence labelling tasks, our experiments demonstrate that data-transfer outperforms methods based on the crosslingual transfer capabilities of multilingual pre-trained language models (model-transfer). Finally, we show how the automatically generated data in Spanish can also be used to improve results in the original English monolingual setting, providing thus a fully automatic data augmentation strategy.
Authors: Shauli Ravfogel, Valentina Pyatkin, Amir DN Cohen, Avshalom Manevich, Yoav Goldberg
Abstract: Identifying texts with a given semantics is central for many information seeking scenarios. Similarity search over vector embeddings appear to be central to this ability, yet the similarity reflected in current text embeddings is corpus-driven, and is inconsistent and sub-optimal for many use cases. What, then, is a good notion of similarity for effective retrieval of text? We identify the need to search for texts based on abstract descriptions of their content, and the corresponding notion of \emph{description based similarity}. We demonstrate the inadequacy of current text embeddings and propose an alternative model that significantly improves when used in standard nearest neighbor search. The model is trained using positive and negative pairs sourced through prompting a LLM, demonstrating how data from LLMs can be used for creating new capabilities not immediately possible using the original model.
Authors: Yue Guo, Tal August, Gondy Leroy, Trevor Cohen, Lucy Lu Wang
Abstract: While there has been significant development of models for Plain Language Summarization (PLS), evaluation remains a challenge. PLS lacks a dedicated assessment metric, and the suitability of text generation evaluation metrics is unclear due to the unique transformations involved (e.g., adding background explanations, removing jargon). To address these questions, our study introduces a granular meta-evaluation testbed, APPLS, designed to evaluate metrics for PLS. We identify four PLS criteria from previous work -- informativeness, simplification, coherence, and faithfulness -- and define a set of perturbations corresponding to these criteria that sensitive metrics should be able to detect. We apply these perturbations to extractive hypotheses for two PLS datasets to form our testbed. Using APPLS, we assess performance of 14 metrics, including automated scores, lexical features, and LLM prompt-based evaluations. Our analysis reveals that while some current metrics show sensitivity to specific criteria, no single method captures all four criteria simultaneously. We therefore recommend a suite of automated metrics be used to capture PLS quality along all relevant criteria. This work contributes the first meta-evaluation testbed for PLS and a comprehensive evaluation of existing metrics. APPLS and our evaluation code is available at https://github.com/LinguisticAnomalies/APPLS.
Authors: Loukas Ilias, Ioannis Michail Kazelidis, Dimitris Askounis
Abstract: Although not all bots are malicious, the vast majority of them are responsible for spreading misinformation and manipulating the public opinion about several issues, i.e., elections and many more. Therefore, the early detection of bots is crucial. Although there have been proposed methods for detecting bots in social media, there are still substantial limitations. For instance, existing research initiatives still extract a large number of features and train traditional machine learning algorithms or use GloVe embeddings and train LSTMs. However, feature extraction is a tedious procedure demanding domain expertise. Also, language models based on transformers have been proved to be better than LSTMs. Other approaches create large graphs and train graph neural networks requiring in this way many hours for training and access to computational resources. To tackle these limitations, this is the first study employing only the user description field and images of three channels denoting the type and content of tweets posted by the users. Firstly, we create digital DNA sequences, transform them to 3d images, and apply pretrained models of the vision domain, including EfficientNet, AlexNet, VGG16, etc. Next, we propose a multimodal approach, where we use TwHIN-BERT for getting the textual representation of the user description field and employ VGG16 for acquiring the visual representation for the image modality. We propose three different fusion methods, namely concatenation, gated multimodal unit, and crossmodal attention, for fusing the different modalities and compare their performances. Finally, we present a qualitative analysis of the behavior of our best performing model. Extensive experiments conducted on the Cresci'17 and TwiBot-20 datasets demonstrate valuable advantages of our introduced approaches over state-of-the-art ones.
Authors: Mengkang Hu, Yao Mu, Xinmiao Yu, Mingyu Ding, Shiguang Wu, Wenqi Shao, Qiguang Chen, Bin Wang, Yu Qiao, Ping Luo
Abstract: This paper studies close-loop task planning, which refers to the process of generating a sequence of skills (a plan) to accomplish a specific goal while adapting the plan based on real-time observations. Recently, prompting Large Language Models (LLMs) to generate actions iteratively has become a prevalent paradigm due to its superior performance and user-friendliness. However, this paradigm is plagued by two inefficiencies: high token consumption and redundant error correction, both of which hinder its scalability for large-scale testing and applications. To address these issues, we propose Tree-Planner, which reframes task planning with LLMs into three distinct phases: plan sampling, action tree construction, and grounded deciding. Tree-Planner starts by using an LLM to sample a set of potential plans before execution, followed by the aggregation of them to form an action tree. Finally, the LLM performs a top-down decision-making process on the tree, taking into account real-time environmental information. Experiments show that Tree-Planner achieves state-of-the-art performance while maintaining high efficiency. By decomposing LLM queries into a single plan-sampling call and multiple grounded-deciding calls, a considerable part of the prompt are less likely to be repeatedly consumed. As a result, token consumption is reduced by 92.2% compared to the previously best-performing model. Additionally, by enabling backtracking on the action tree as needed, the correction process becomes more flexible, leading to a 40.5% decrease in error corrections.
Authors: Yun Zhu, Yaoke Wang, Haizhou Shi, Siliang Tang
Abstract: Rich textual and topological information of textual graphs need to be modeled in real-world applications such as webpages, e-commerce, and academic articles. Practitioners have been long following the path of adopting a shallow text encoder and a subsequent graph neural network (GNN) to solve this problem. In light of recent advancements in large language models (LLMs), it is apparent that integrating LLMs for enhanced textual encoding can substantially improve the performance of textual graphs. Nevertheless, the efficiency of these methods poses a significant challenge. In this paper, we propose ENGINE, a parameter- and memory-efficient fine-tuning method for textual graphs with an LLM encoder. The key insight is to combine the LLMs and GNNs through a tunable side structure, which significantly reduces the training complexity without impairing the joint model's capacity. Extensive experiments on textual graphs demonstrate our method's effectiveness by achieving the best model performance, meanwhile having the lowest training cost compared to previous methods. Moreover, we introduce two variants with caching and dynamic early exit to further enhance training and inference speed. Specifically, caching accelerates ENGINE's training by 12x, and dynamic early exit achieves up to 5x faster inference with a negligible performance drop (at maximum 1.17% relevant drop across 7 datasets). Our codes are available at: https://github.com/ZhuYun97/ENGINE
Authors: Zhibo Hu (Hye-Young), Chen Wang (Hye-Young), Yanfeng Shu (Hye-Young), Helen (Hye-Young), Paik, Liming Zhu
Abstract: The robustness of large language models (LLMs) becomes increasingly important as their use rapidly grows in a wide range of domains. Retrieval-Augmented Generation (RAG) is considered as a means to improve the trustworthiness of text generation from LLMs. However, how the outputs from RAG-based LLMs are affected by slightly different inputs is not well studied. In this work, we find that the insertion of even a short prefix to the prompt leads to the generation of outputs far away from factually correct answers. We systematically evaluate the effect of such prefixes on RAG by introducing a novel optimization technique called Gradient Guided Prompt Perturbation (GGPP). GGPP achieves a high success rate in steering outputs of RAG-based LLMs to targeted wrong answers. It can also cope with instructions in the prompts requesting to ignore irrelevant context. We also exploit LLMs' neuron activation difference between prompts with and without GGPP perturbations to give a method that improves the robustness of RAG-based LLMs through a highly effective detector trained on neuron activation triggered by GGPP generated prompts. Our evaluation on open-sourced LLMs demonstrates the effectiveness of our methods.
Authors: Jianhao Yan, Yun Luo, Yue Zhang
Abstract: The application scope of large language models (LLMs) is increasingly expanding. In practical use, users might provide feedback based on the model's output, hoping for a responsive model that can complete responses according to their feedback. Whether the model can appropriately respond to users' refuting feedback and consistently follow through with execution has not been thoroughly analyzed. In light of this, this paper proposes a comprehensive benchmark, RefuteBench, covering tasks such as question answering, machine translation, and email writing. The evaluation aims to assess whether models can positively accept feedback in form of refuting instructions and whether they can consistently adhere to user demands throughout the conversation. We conduct evaluations on numerous LLMs and find that LLMs are stubborn, i.e. exhibit inclination to their internal knowledge, often failing to comply with user feedback. Additionally, as the length of the conversation increases, models gradually forget the user's stated feedback and roll back to their own responses. We further propose a recall-and-repeat prompts as a simple and effective way to enhance the model's responsiveness to feedback.
Authors: Yiqiao Jin, Minje Choi, Gaurav Verma, Jindong Wang, Srijan Kumar
Abstract: Social media platforms are hubs for multimodal information exchange, encompassing text, images, and videos, making it challenging for machines to comprehend the information or emotions associated with interactions in online spaces. Multimodal Large Language Models (MLLMs) have emerged as a promising solution to these challenges, yet they struggle to accurately interpret human emotions and complex content such as misinformation. This paper introduces MM-Soc, a comprehensive benchmark designed to evaluate MLLMs' understanding of multimodal social media content. MM-Soc compiles prominent multimodal datasets and incorporates a novel large-scale YouTube tagging dataset, targeting a range of tasks from misinformation detection, hate speech detection, and social context generation. Through our exhaustive evaluation on ten size-variants of four open-source MLLMs, we have identified significant performance disparities, highlighting the need for advancements in models' social understanding capabilities. Our analysis reveals that, in a zero-shot setting, various types of MLLMs generally exhibit difficulties in handling social media tasks. However, MLLMs demonstrate performance improvements post fine-tuning, suggesting potential pathways for improvement. Our code and data are available at https://github.com/claws-lab/MMSoc.git.
Authors: Sinan Abdulhak, Wayne Hubbard, Karthik Gopalakrishnan, Max Z. Li
Abstract: Generative artificial intelligence (AI) and large language models (LLMs) have gained rapid popularity through publicly available tools such as ChatGPT. The adoption of LLMs for personal and professional use is fueled by the natural interactions between human users and computer applications such as ChatGPT, along with powerful summarization and text generation capabilities. Given the widespread use of such generative AI tools, in this work we investigate how these tools can be deployed in a non-safety critical, strategic traffic flow management setting. Specifically, we train an LLM, CHATATC, based on a large historical data set of Ground Delay Program (GDP) issuances, spanning 2000-2023 and consisting of over 80,000 GDP implementations, revisions, and cancellations. We test the query and response capabilities of CHATATC, documenting successes (e.g., providing correct GDP rates, durations, and reason) and shortcomings (e.g,. superlative questions). We also detail the design of a graphical user interface for future users to interact and collaborate with the CHATATC conversational agent.
Authors: Yifu Gao, Linbo Qiao, Zhigang Kan, Zhihua Wen, Yongquan He, Dongsheng Li
Abstract: Temporal knowledge graph question answering (TKGQA) poses a significant challenge task, due to the temporal constraints hidden in questions and the answers sought from dynamic structured knowledge. Although large language models (LLMs) have made considerable progress in their reasoning ability over structured data, their application to the TKGQA task is a relatively unexplored area. This paper first proposes a novel generative temporal knowledge graph question answering framework, GenTKGQA, which guides LLMs to answer temporal questions through two phases: Subgraph Retrieval and Answer Generation. First, we exploit LLM's intrinsic knowledge to mine temporal constraints and structural links in the questions without extra training, thus narrowing down the subgraph search space in both temporal and structural dimensions. Next, we design virtual knowledge indicators to fuse the graph neural network signals of the subgraph and the text representations of the LLM in a non-shallow way, which helps the open-source LLM deeply understand the temporal order and structural dependencies among the retrieved facts through instruction tuning. Experimental results on two widely used datasets demonstrate the superiority of our model.
Authors: Chulin Xie, Zinan Lin, Arturs Backurs, Sivakanth Gopi, Da Yu, Huseyin A Inan, Harsha Nori, Haotian Jiang, Huishuai Zhang, Yin Tat Lee, Bo Li, Sergey Yekhanin
Abstract: Text data has become extremely valuable due to the emergence of machine learning algorithms that learn from it. A lot of high-quality text data generated in the real world is private and therefore cannot be shared or used freely due to privacy concerns. Generating synthetic replicas of private text data with a formal privacy guarantee, i.e., differential privacy (DP), offers a promising and scalable solution. However, existing methods necessitate DP finetuning of large language models (LLMs) on private data to generate DP synthetic data. This approach is not viable for proprietary LLMs (e.g., GPT-3.5) and also demands considerable computational resources for open-source LLMs. Lin et al. (2024) recently introduced the Private Evolution (PE) algorithm to generate DP synthetic images with only API access to diffusion models. In this work, we propose an augmented PE algorithm, named Aug-PE, that applies to the complex setting of text. We use API access to an LLM and generate DP synthetic text without any model training. We conduct comprehensive experiments on three benchmark datasets. Our results demonstrate that Aug-PE produces DP synthetic text that yields competitive utility with the SOTA DP finetuning baselines. This underscores the feasibility of relying solely on API access of LLMs to produce high-quality DP synthetic texts, thereby facilitating more accessible routes to privacy-preserving LLM applications. Our code and data are available at https://github.com/AI-secure/aug-pe.
Authors: Yupeng Li, Haorui He, Jin Bai, Dacheng Wen
Abstract: The prevalence of fake news across various online sources has had a significant influence on the public. Existing Chinese fake news detection datasets are limited to news sourced solely from Weibo. However, fake news originating from multiple sources exhibits diversity in various aspects, including its content and social context. Methods trained on purely one single news source can hardly be applicable to real-world scenarios. Our pilot experiment demonstrates that the F1 score of the state-of-the-art method that learns from a large Chinese fake news detection dataset, Weibo-21, drops significantly from 0.943 to 0.470 when the test data is changed to multi-source news data, failing to identify more than one-third of the multi-source fake news. To address this limitation, we constructed the first multi-source benchmark dataset for Chinese fake news detection, termed MCFEND, which is composed of news we collected from diverse sources such as social platforms, messaging apps, and traditional online news outlets. Notably, such news has been fact-checked by 14 authoritative fact-checking agencies worldwide. In addition, various existing Chinese fake news detection methods are thoroughly evaluated on our proposed dataset in cross-source, multi-source, and unseen source ways. MCFEND, as a benchmark dataset, aims to advance Chinese fake news detection approaches in real-world scenarios.
Authors: Siye Wu, Jian Xie, Jiangjie Chen, Tinghui Zhu, Kai Zhang, Yanghua Xiao
Abstract: By leveraging the retrieval of information from external knowledge databases, Large Language Models (LLMs) exhibit enhanced capabilities for accomplishing many knowledge-intensive tasks. However, due to the inherent flaws of current retrieval systems, there might exist irrelevant information within those retrieving top-ranked passages. In this work, we present a comprehensive investigation into the robustness of LLMs to different types of irrelevant information under various conditions. We initially introduce a framework to construct high-quality irrelevant information that ranges from semantically unrelated, partially related, and related to questions. Furthermore, our analysis demonstrates that the constructed irrelevant information not only scores highly on similarity metrics, being highly retrieved by existing systems, but also bears semantic connections to the context. Our investigation reveals that current LLMs still face challenges in discriminating highly semantically related information and can be easily distracted by these irrelevant yet misleading content. Besides, we also find that current solutions for handling irrelevant information have limitations in improving the robustness of LLMs to such distractions. All the resources are available on GitHub at https://github.com/Di-viner/LLM-Robustness-to-Irrelevant-Information.
URLs: https://github.com/Di-viner/LLM-Robustness-to-Irrelevant-Information.
Authors: Rickard Stureborg, Sanxing Chen, Ruoyu Xie, Aayushi Patel, Christopher Li, Chloe Qinyu Zhu, Tingnan Hu, Jun Yang, Bhuwan Dhingra
Abstract: One way to personalize chatbot interactions is by establishing common ground with the intended reader. A domain where establishing mutual understanding could be particularly impactful is vaccine concerns and misinformation. Vaccine interventions are forms of messaging which aim to answer concerns expressed about vaccination. Tailoring responses in this domain is difficult, since opinions often have seemingly little ideological overlap. We define the task of tailoring vaccine interventions to a Common-Ground Opinion (CGO). Tailoring responses to a CGO involves meaningfully improving the answer by relating it to an opinion or belief the reader holds. In this paper we introduce TAILOR-CGO, a dataset for evaluating how well responses are tailored to provided CGOs. We benchmark several major LLMs on this task; finding GPT-4-Turbo performs significantly better than others. We also build automatic evaluation metrics, including an efficient and accurate BERT model that outperforms finetuned LLMs, investigate how to successfully tailor vaccine messaging to CGOs, and provide actionable recommendations from this investigation. Code and model weights: https://github.com/rickardstureborg/tailor-cgo Dataset: https://huggingface.co/datasets/DukeNLP/tailor-cgo
URLs: https://github.com/rickardstureborg/tailor-cgo, https://huggingface.co/datasets/DukeNLP/tailor-cgo
Authors: Eden Avnat, Michal Levy, Daniel Herstain, Elia Yanko, Daniel Ben Joya, Michal Tzuchman Katz, Dafna Eshel, Sahar Laros, Yael Dagan, Shahar Barami, Joseph Mermelstein, Shahar Ovadia, Noam Shomron, Varda Shalev, Raja-Elie E. Abdulnour
Abstract: Clinical problem-solving requires processing of semantic medical knowledge such as illness scripts and numerical medical knowledge of diagnostic tests for evidence-based decision-making. As large language models (LLMs) show promising results in many aspects of language-based clinical practice, their ability to generate non-language evidence-based answers to clinical questions is inherently limited by tokenization. Therefore, we evaluated LLMs' performance on two question types: numeric (correlating findings) and semantic (differentiating entities) while examining differences within and between LLMs in medical aspects and comparing their performance to humans. To generate straightforward multi-choice questions and answers (QAs) based on evidence-based medicine (EBM), we used a comprehensive medical knowledge graph (encompassed data from more than 50,00 peer-reviewed articles) and created the "EBMQA". EBMQA contains 105,000 QAs labeled with medical and non-medical topics and classified into numerical or semantic questions. We benchmarked this dataset using more than 24,500 QAs on two state-of-the-art LLMs: Chat-GPT4 and Claude3-Opus. We evaluated the LLMs accuracy on semantic and numerical question types and according to sub-labeled topics. For validation, six medical experts were tested on 100 numerical EBMQA questions. We found that both LLMs excelled more in semantic than numerical QAs, with Claude3 surpassing GPT4 in numerical QAs. However, both LLMs showed inter and intra gaps in different medical aspects and remained inferior to humans. Thus, their medical advice should be addressed carefully.
Authors: Jiaxin Wen, Ruiqi Zhong, Pei Ke, Zhihong Shao, Hongning Wang, Minlie Huang
Abstract: When using language models (LMs) to solve complex problems, humans might struggle to understand the LM-generated solutions and repair the flawed ones. To assist humans in repairing them, we propose to automatically decompose complex solutions into multiple simpler pieces that correspond to specific subtasks. We introduce a novel objective for learning task decomposition, termed assistive value (AssistV), which measures the feasibility and speed for humans to repair the decomposed solution. We collect a dataset of human repair experiences on different decomposed solutions. Utilizing the collected data as in-context examples, we then learn to critique, refine, and rank decomposed solutions to improve AssistV. We validate our method under competitive programming problems: under 177 hours of human study, our method enables non-experts to solve 33.3\% more problems, speeds them up by 3.3x, and empowers them to match unassisted experts.
Authors: Jinghui Lu, Haiyang Yu, Yanjie Wang, Yongjie Ye, Jingqun Tang, Ziwei Yang, Binghong Wu, Qi Liu, Hao Feng, Han Wang, Hao Liu, Can Huang
Abstract: Recently, many studies have demonstrated that exclusively incorporating OCR-derived text and spatial layouts with large language models (LLMs) can be highly effective for document understanding tasks. However, existing methods that integrate spatial layouts with text have limitations, such as producing overly long text sequences or failing to fully leverage the autoregressive traits of LLMs. In this work, we introduce Interleaving Layout and Text in a Large Language Model (LayTextLLM)} for document understanding. In particular, LayTextLLM projects each bounding box to a single embedding and interleaves it with text, efficiently avoiding long sequence issues while leveraging autoregressive traits of LLMs. LayTextLLM not only streamlines the interaction of layout and textual data but also shows enhanced performance in Key Information Extraction (KIE) and Visual Question Answering (VQA). Comprehensive benchmark evaluations reveal significant improvements, with a 27.2% increase on KIE tasks and 12.0% on VQA tasks compared to previous state-of-the-art document understanding MLLMs, as well as a 15.1% improvement over other SOTA OCR-based LLMs on KIE tasks.
Authors: Darshan Prabhu, Yifan Peng, Preethi Jyothi, Shinji Watanabe
Abstract: Convolutions have become essential in state-of-the-art end-to-end Automatic Speech Recognition~(ASR) systems due to their efficient modelling of local context. Notably, its use in Conformers has led to superior performance compared to vanilla Transformer-based ASR systems. While components other than the convolution module in the Conformer have been reexamined, altering the convolution module itself has been far less explored. Towards this, we introduce Multi-Convformer that uses multiple convolution kernels within the convolution module of the Conformer in conjunction with gating. This helps in improved modeling of local dependencies at varying granularities. Our model rivals existing Conformer variants such as CgMLP and E-Branchformer in performance, while being more parameter efficient. We empirically compare our approach with Conformer and its variants across four different datasets and three different modelling paradigms and show up to 8% relative word error rate~(WER) improvements.
Authors: Rapha\"el Tinarrage, Henrique Ennes, Lucas E. Resck, Lucas T. Gomes, Jean R. Ponciano, Jorge Poco
Abstract: Binding precedents (S\'umulas Vinculantes) constitute a juridical instrument unique to the Brazilian legal system and whose objectives include the protection of the Federal Supreme Court against repetitive demands. Studies of the effectiveness of these instruments in decreasing the Court's exposure to similar cases, however, indicate that they tend to fail in such a direction, with some of the binding precedents seemingly creating new demands. We empirically assess the legal impact of five binding precedents, 11, 14, 17, 26 and 37, at the highest court level through their effects on the legal subjects they address. This analysis is only possible through the comparison of the Court's ruling about the precedents' themes before they are created, which means that these decisions should be detected through techniques of Similar Case Retrieval. The contributions of this article are therefore twofold: on the mathematical side, we compare the uses of different methods of Natural Language Processing -- TF-IDF, LSTM, BERT, and regex -- for Similar Case Retrieval, whereas on the legal side, we contrast the inefficiency of these binding precedents with a set of hypotheses that may justify their repeated usage. We observe that the deep learning models performed significantly worse in the specific Similar Case Retrieval task and that the reasons for binding precedents to fail in responding to repetitive demand are heterogeneous and case-dependent, making it impossible to single out a specific cause.
Authors: Xiuying Wei, Skander Moalla, Razvan Pascanu, Caglar Gulcehre
Abstract: State-of-the-art LLMs often rely on scale with high computational costs, which has sparked a research agenda to reduce parameter counts and costs without significantly impacting performance. Our study focuses on Transformer-based LLMs, specifically applying low-rank parametrization to the computationally intensive feedforward networks (FFNs), which are less studied than attention blocks. In contrast to previous works, (i) we explore low-rank parametrization at scale, up to 1.3B parameters; (ii) within Transformer language models rather than convolutional architectures; and (iii) starting from training from scratch. Experiments on the large RefinedWeb dataset show that low-rank parametrization is both efficient (e.g., 2.6$\times$ FFN speed-up with 32\% parameters) and effective during training. Interestingly, these structured FFNs exhibit steeper scaling curves than the original models. Motivated by this finding, we develop the wide and structured networks surpassing the current medium-sized and large-sized Transformer in perplexity and throughput performance. Our code is available at https://github.com/CLAIRE-Labo/StructuredFFN/tree/main.
URLs: https://github.com/CLAIRE-Labo/StructuredFFN/tree/main.
Authors: Hongyu Wang, Shuming Ma, Ruiping Wang, Furu Wei
Abstract: We introduce, Q-Sparse, a simple yet effective approach to training sparsely-activated large language models (LLMs). Q-Sparse enables full sparsity of activations in LLMs which can bring significant efficiency gains in inference. This is achieved by applying top-K sparsification to the activations and the straight-through-estimator to the training. We also introduce Block Q-Sparse for batch training and inference. The key results from this work are, (1) Q-Sparse can achieve results comparable to those of baseline LLMs while being much more efficient at inference time; (2) We present an inference-optimal scaling law for sparsely-activated LLMs; (3) Q-Sparse is effective in different settings, including training-from-scratch, continue-training of off-the-shelf LLMs, and finetuning; (4) Q-Sparse works for both full-precision and 1-bit LLMs (e.g., BitNet b1.58). Particularly, the synergy of BitNet b1.58 and Q-Sparse (can be equipped with MoE) provides the cornerstone and a clear path to revolutionize the efficiency, including cost and energy consumption, of future LLMs.
Authors: Shubham Vatsal, Harsh Dubey
Abstract: Large language models (LLMs) have shown remarkable performance on many different Natural Language Processing (NLP) tasks. Prompt engineering plays a key role in adding more to the already existing abilities of LLMs to achieve significant performance gains on various NLP tasks. Prompt engineering requires composing natural language instructions called prompts to elicit knowledge from LLMs in a structured way. Unlike previous state-of-the-art (SoTA) models, prompt engineering does not require extensive parameter re-training or fine-tuning based on the given NLP task and thus solely operates on the embedded knowledge of LLMs. Additionally, LLM enthusiasts can intelligently extract LLMs' knowledge through a basic natural language conversational exchange or prompt engineering, allowing more and more people even without deep mathematical machine learning background to experiment with LLMs. With prompt engineering gaining popularity in the last two years, researchers have come up with numerous engineering techniques around designing prompts to improve accuracy of information extraction from the LLMs. In this paper, we summarize different prompting techniques and club them together based on different NLP tasks that they have been used for. We further granularly highlight the performance of these prompting strategies on various datasets belonging to that NLP task, talk about the corresponding LLMs used, present a taxonomy diagram and discuss the possible SoTA for specific datasets. In total, we read and present a survey of 44 research papers which talk about 39 different prompting methods on 29 different NLP tasks of which most of them have been published in the last two years.
Authors: Jiayu Lin, Guanrong Chen, Bojun Jin, Chenyang Li, Shutong Jia, Wancong Lin, Yang Sun, Yuhang He, Caihua Yang, Jianzhu Bao, Jipeng Wu, Wen Su, Jinglu Chen, Xinyi Li, Tianyu Chen, Mingjie Han, Shuaiwen Du, Zijian Wang, Jiyin Li, Fuzhong Suo, Hao Wang, Nuanchen Lin, Xuanjing Huang, Changjian Jiang, RuiFeng Xu, Long Zhang, Jiuxin Cao, Ting Jin, Zhongyu Wei
Abstract: In this paper we present the results of the AI-Debater 2023 Challenge held by the Chinese Conference on Affect Computing (CCAC 2023), and introduce the related datasets. We organize two tracks to handle the argumentative generation tasks in different scenarios, namely, Counter-Argument Generation (Track 1) and Claim-based Argument Generation (Track 2). Each track is equipped with its distinct dataset and baseline model respectively. In total, 32 competing teams register for the challenge, from which we received 11 successful submissions. In this paper, we will present the results of the challenge and a summary of the systems, highlighting commonalities and innovations among participating systems. Datasets and baseline models of the AI-Debater 2023 Challenge have been already released and can be accessed through the official website of the challenge.
Authors: Shayne Longpre, Robert Mahari, Ariel Lee, Campbell Lund, Hamidah Oderinwale, William Brannon, Nayan Saxena, Naana Obeng-Marnu, Tobin South, Cole Hunter, Kevin Klyman, Christopher Klamm, Hailey Schoelkopf, Nikhil Singh, Manuel Cherep, Ahmad Anis, An Dinh, Caroline Chitongo, Da Yin, Damien Sileo, Deividas Mataciunas, Diganta Misra, Emad Alghamdi, Enrico Shippole, Jianguo Zhang, Joanna Materzynska, Kun Qian, Kush Tiwary, Lester Miranda, Manan Dey, Minnie Liang, Mohammed Hamdy, Niklas Muennighoff, Seonghyeon Ye, Seungone Kim, Shrestha Mohanty, Vipul Gupta, Vivek Sharma, Vu Minh Chien, Xuhui Zhou, Yizhi Li, Caiming Xiong, Luis Villa, Stella Biderman, Hanlin Li, Daphne Ippolito, Sara Hooker, Jad Kabbara, Sandy Pentland
Abstract: General-purpose artificial intelligence (AI) systems are built on massive swathes of public web data, assembled into corpora such as C4, RefinedWeb, and Dolma. To our knowledge, we conduct the first, large-scale, longitudinal audit of the consent protocols for the web domains underlying AI training corpora. Our audit of 14,000 web domains provides an expansive view of crawlable web data and how codified data use preferences are changing over time. We observe a proliferation of AI-specific clauses to limit use, acute differences in restrictions on AI developers, as well as general inconsistencies between websites' expressed intentions in their Terms of Service and their robots.txt. We diagnose these as symptoms of ineffective web protocols, not designed to cope with the widespread re-purposing of the internet for AI. Our longitudinal analyses show that in a single year (2023-2024) there has been a rapid crescendo of data restrictions from web sources, rendering ~5%+ of all tokens in C4, or 28%+ of the most actively maintained, critical sources in C4, fully restricted from use. For Terms of Service crawling restrictions, a full 45% of C4 is now restricted. If respected or enforced, these restrictions are rapidly biasing the diversity, freshness, and scaling laws for general-purpose AI systems. We hope to illustrate the emerging crises in data consent, for both developers and creators. The foreclosure of much of the open web will impact not only commercial AI, but also non-commercial AI and academic research.
Authors: Desta Haileselassie Hagos, Rick Battle, Danda B. Rawat
Abstract: The emergence of Generative Artificial Intelligence (AI) and Large Language Models (LLMs) has marked a new era of Natural Language Processing (NLP), introducing unprecedented capabilities that are revolutionizing various domains. This paper explores the current state of these cutting-edge technologies, demonstrating their remarkable advancements and wide-ranging applications. Our paper contributes to providing a holistic perspective on the technical foundations, practical applications, and emerging challenges within the evolving landscape of Generative AI and LLMs. We believe that understanding the generative capabilities of AI systems and the specific context of LLMs is crucial for researchers, practitioners, and policymakers to collaboratively shape the responsible and ethical integration of these technologies into various domains. Furthermore, we identify and address main research gaps, providing valuable insights to guide future research endeavors within the AI research community.
Authors: Yizhou Chi, Lingjun Mao, Zineng Tang
Abstract: Strategic social deduction games serve as valuable testbeds for evaluating the understanding and inference skills of language models, offering crucial insights into social science, artificial intelligence, and strategic gaming. This paper focuses on creating proxies of human behavior in simulated environments, with Among Us utilized as a tool for studying simulated human behavior. The study introduces a text-based game environment, named AmongAgents, that mirrors the dynamics of Among Us. Players act as crew members aboard a spaceship, tasked with identifying impostors who are sabotaging the ship and eliminating the crew. Within this environment, the behavior of simulated language agents is analyzed. The experiments involve diverse game sequences featuring different configurations of Crewmates and Impostor personality archetypes. Our work demonstrates that state-of-the-art large language models (LLMs) can effectively grasp the game rules and make decisions based on the current context. This work aims to promote further exploration of LLMs in goal-oriented games with incomplete information and complex action spaces, as these settings offer valuable opportunities to assess language model performance in socially driven scenarios.
Authors: Yongseung Jegal, Jaewoong Choi, Jiho Lee, Ki-Su Park, Seyoung Lee, Janghyeok Yoon
Abstract: Drug repositioning-a promising strategy for discovering new therapeutic uses for existing drugs-has been increasingly explored in the computational science literature using biomedical databases. However, the technological potential of drug repositioning candidates has often been overlooked. This study presents a novel protocol to comprehensively analyse various sources such as pharmaceutical patents and biomedical databases, and identify drug repositioning candidates with both technological potential and scientific evidence. To this end, first, we constructed a scientific biomedical knowledge graph (s-BKG) comprising relationships between drugs, diseases, and genes derived from biomedical databases. Our protocol involves identifying drugs that exhibit limited association with the target disease but are closely located in the s-BKG, as potential drug candidates. We constructed a patent-informed biomedical knowledge graph (p-BKG) by adding pharmaceutical patent information. Finally, we developed a graph embedding protocol to ascertain the structure of the p-BKG, thereby calculating the relevance scores of those candidates with target disease-related patents to evaluate their technological potential. Our case study on Alzheimer's disease demonstrates its efficacy and feasibility, while the quantitative outcomes and systematic methods are expected to bridge the gap between computational discoveries and successful market applications in drug repositioning research.
Authors: Hanlin Zhang, Benjamin L. Edelman, Danilo Francati, Daniele Venturi, Giuseppe Ateniese, Boaz Barak
Abstract: Watermarking generative models consists of planting a statistical signal (watermark) in a model's output so that it can be later verified that the output was generated by the given model. A strong watermarking scheme satisfies the property that a computationally bounded attacker cannot erase the watermark without causing significant quality degradation. In this paper, we study the (im)possibility of strong watermarking schemes. We prove that, under well-specified and natural assumptions, strong watermarking is impossible to achieve. This holds even in the private detection algorithm setting, where the watermark insertion and detection algorithms share a secret key, unknown to the attacker. To prove this result, we introduce a generic efficient watermark attack; the attacker is not required to know the private key of the scheme or even which scheme is used. Our attack is based on two assumptions: (1) The attacker has access to a "quality oracle" that can evaluate whether a candidate output is a high-quality response to a prompt, and (2) The attacker has access to a "perturbation oracle" which can modify an output with a nontrivial probability of maintaining quality, and which induces an efficiently mixing random walk on high-quality outputs. We argue that both assumptions can be satisfied in practice by an attacker with weaker computational capabilities than the watermarked model itself, to which the attacker has only black-box access. Furthermore, our assumptions will likely only be easier to satisfy over time as models grow in capabilities and modalities. We demonstrate the feasibility of our attack by instantiating it to attack three existing watermarking schemes for large language models: Kirchenbauer et al. (2023), Kuditipudi et al. (2023), and Zhao et al. (2023). The same attack successfully removes the watermarks planted by all three schemes, with only minor quality degradation.
Authors: Shengyin Sun, Yuxiang Ren, Chen Ma, Xuecang Zhang
Abstract: The latest advancements in large language models (LLMs) have revolutionized the field of natural language processing (NLP). Inspired by the success of LLMs in NLP tasks, some recent work has begun investigating the potential of applying LLMs in graph learning tasks. However, most of the existing work focuses on utilizing LLMs as powerful node feature augmenters, leaving employing LLMs to enhance graph topological structures an understudied problem. In this work, we explore how to leverage the information retrieval and text generation capabilities of LLMs to refine/enhance the topological structure of text-attributed graphs (TAGs) under the node classification setting. First, we propose using LLMs to help remove unreliable edges and add reliable ones in the TAG. Specifically, we first let the LLM output the semantic similarity between node attributes through delicate prompt designs, and then perform edge deletion and edge addition based on the similarity. Second, we propose using pseudo-labels generated by the LLM to improve graph topology, that is, we introduce the pseudo-label propagation as a regularization to guide the graph neural network (GNN) in learning proper edge weights. Finally, we incorporate the two aforementioned LLM-based methods for graph topological refinement into the process of GNN training, and perform extensive experiments on four real-world datasets. The experimental results demonstrate the effectiveness of LLM-based graph topology refinement (achieving a 0.15%--2.47% performance gain on public benchmarks).
Authors: Samuele Poppi, Tobia Poppi, Federico Cocchi, Marcella Cornia, Lorenzo Baraldi, Rita Cucchiara
Abstract: Large-scale vision-and-language models, such as CLIP, are typically trained on web-scale data, which can introduce inappropriate content and lead to the development of unsafe and biased behavior. This, in turn, hampers their applicability in sensitive and trustworthy contexts and could raise significant concerns in their adoption. Our research introduces a novel approach to enhancing the safety of vision-and-language models by diminishing their sensitivity to NSFW (not safe for work) inputs. In particular, our methodology seeks to sever "toxic" linguistic and visual concepts, unlearning the linkage between unsafe linguistic or visual items and unsafe regions of the embedding space. We show how this can be done by fine-tuning a CLIP model on synthetic data obtained from a large language model trained to convert between safe and unsafe sentences, and a text-to-image generator. We conduct extensive experiments on the resulting embedding space for cross-modal retrieval, text-to-image, and image-to-text generation, where we show that our model can be remarkably employed with pre-trained generative models. Our source code and trained models are available at: https://github.com/aimagelab/safe-clip.
Authors: Jian Ma, Chen Chen, Qingsong Xie, Haonan Lu
Abstract: Text-to-image diffusion models are well-known for their ability to generate realistic images based on textual prompts. However, the existing works have predominantly focused on English, lacking support for non-English text-to-image models. The most commonly used translation methods cannot solve the generation problem related to language culture, while training from scratch on a specific language dataset is prohibitively expensive. In this paper, we are inspired to propose a simple plug-and-play language transfer method based on knowledge distillation. All we need to do is train a lightweight MLP-like parameter-efficient adapter (PEA) with only 6M parameters under teacher knowledge distillation along with a small parallel data corpus. We are surprised to find that freezing the parameters of UNet can still achieve remarkable performance on the language-specific prompt evaluation set, demonstrating that PEA can stimulate the potential generation ability of the original UNet. Additionally, it closely approaches the performance of the English text-to-image model on a general prompt evaluation set. Furthermore, our adapter can be used as a plugin to achieve significant results in downstream tasks in cross-lingual text-to-image generation. Code will be available at: https://github.com/OPPO-Mente-Lab/PEA-Diffusion
Authors: Yunlong Tang, Jing Bi, Siting Xu, Luchuan Song, Susan Liang, Teng Wang, Daoan Zhang, Jie An, Jingyang Lin, Rongyi Zhu, Ali Vosoughi, Chao Huang, Zeliang Zhang, Pinxin Liu, Mingqian Feng, Feng Zheng, Jianguo Zhang, Ping Luo, Jiebo Luo, Chenliang Xu
Abstract: With the burgeoning growth of online video platforms and the escalating volume of video content, the demand for proficient video understanding tools has intensified markedly. Given the remarkable capabilities of large language models (LLMs) in language and multimodal tasks, this survey provides a detailed overview of recent advancements in video understanding that harness the power of LLMs (Vid-LLMs). The emergent capabilities of Vid-LLMs are surprisingly advanced, particularly their ability for open-ended multi-granularity (general, temporal, and spatiotemporal) reasoning combined with commonsense knowledge, suggesting a promising path for future video understanding. We examine the unique characteristics and capabilities of Vid-LLMs, categorizing the approaches into three main types: Video Analyzer x LLM, Video Embedder x LLM, and (Analyzer + Embedder) x LLM. Furthermore, we identify five sub-types based on the functions of LLMs in Vid-LLMs: LLM as Summarizer, LLM as Manager, LLM as Text Decoder, LLM as Regressor, and LLM as Hidden Layer. Furthermore, this survey presents a comprehensive study of the tasks, datasets, benchmarks, and evaluation methodologies for Vid-LLMs. Additionally, it explores the expansive applications of Vid-LLMs across various domains, highlighting their remarkable scalability and versatility in real-world video understanding challenges. Finally, it summarizes the limitations of existing Vid-LLMs and outlines directions for future research. For more information, readers are recommended to visit the repository at https://github.com/yunlong10/Awesome-LLMs-for-Video-Understanding.
URLs: https://github.com/yunlong10/Awesome-LLMs-for-Video-Understanding.
Authors: Vassilis Papadopoulos, J\'er\'emie Wenger, Cl\'ement Hongler
Abstract: We study the probabilistic modeling performed by Autoregressive Large Language Models (LLMs) through the angle of time directionality, addressing a question first raised in (Shannon, 1951). For large enough models, we empirically find a time asymmetry in their ability to learn natural language: a difference in the average log-perplexity when trying to predict the next token versus when trying to predict the previous one. This difference is at the same time subtle and very consistent across various modalities (language, model size, training time, ...). Theoretically, this is surprising: from an information-theoretic point of view, there should be no such difference. We provide a theoretical framework to explain how such an asymmetry can appear from sparsity and computational complexity considerations, and outline a number of perspectives opened by our results.
Authors: Nicholas Pochinkov, Nandi Schoots
Abstract: Understanding and shaping the behaviour of Large Language Models (LLMs) is increasingly important as applications become more powerful and more frequently adopted. This paper introduces a machine unlearning method specifically designed for LLMs. We introduce a selective pruning method for LLMs that removes neurons based on their relative importance on a targeted capability compared to overall network performance. This approach is a compute- and data-efficient method for identifying and removing neurons that enable specific behaviours. Our findings reveal that both feed-forward and attention neurons in LLMs are specialized; that is, for specific tasks, certain neurons are more crucial than others. Code from all experiments is available at https://github.com/nickypro/selective-pruning
Authors: Zhuowen Yuan, Zidi Xiong, Yi Zeng, Ning Yu, Ruoxi Jia, Dawn Song, Bo Li
Abstract: Recent advancements in Large Language Models (LLMs) have showcased remarkable capabilities across various tasks in different domains. However, the emergence of biases and the potential for generating harmful content in LLMs, particularly under malicious inputs, pose significant challenges. Current mitigation strategies, while effective, are not resilient under adversarial attacks. This paper introduces Resilient Guardrails for Large Language Models (RigorLLM), a novel framework designed to efficiently and effectively moderate harmful and unsafe inputs and outputs for LLMs. By employing a multi-faceted approach that includes energy-based training data augmentation through Langevin dynamics, optimizing a safe suffix for inputs via minimax optimization, and integrating a fusion-based model combining robust KNN with LLMs based on our data augmentation, RigorLLM offers a robust solution to harmful content moderation. Our experimental evaluations demonstrate that RigorLLM not only outperforms existing baselines like OpenAI API and Perspective API in detecting harmful content but also exhibits unparalleled resilience to jailbreaking attacks. The innovative use of constrained optimization and a fusion-based guardrail approach represents a significant step forward in developing more secure and reliable LLMs, setting a new standard for content moderation frameworks in the face of evolving digital threats.
Authors: Akshat Gupta, Dev Sajnani, Gopala Anumanchipalli
Abstract: ROME and MEMIT are largely believed to be two different model editing algorithms, with the major difference between them being the ability to perform batched edits. In this paper, we unify these two algorithms under a single conceptual umbrella, optimizing for the same goal, which we call the preservation-memorization objective. ROME uses an equality constraint to optimize this objective to perform one edit at a time, whereas MEMIT employs a more flexible least-square constraint that allows for batched edits. We generalize ROME and enable batched editing with equality constraint in the form of EMMET - an Equality-constrained Mass Model Editing algorithm for Transformers, a new batched memory-editing algorithm. EMMET can perform batched-edits up to a batch-size of 10,000, with very similar performance to MEMIT across multiple dimensions. With the introduction of EMMET, we truly unify ROME and MEMIT and show that both algorithms are equivalent in terms of their optimization objective, their abilities (singular and batched editing), their model editing performance and their limitations.
Authors: Yulin Luo, Ruichuan An, Bocheng Zou, Yiming Tang, Jiaming Liu, Shanghang Zhang
Abstract: The distribution of subpopulations is an important property hidden within a dataset. Uncovering and analyzing the subpopulation distribution within datasets provides a comprehensive understanding of the datasets, standing as a powerful tool beneficial to various downstream tasks, including Dataset Subpopulation Organization, Subpopulation Shift, and Slice Discovery. Despite its importance, there has been no work that systematically explores the subpopulation distribution of datasets to our knowledge. To address the limitation and solve all the mentioned tasks in a unified way, we introduce a novel concept of subpopulation structures to represent, analyze, and utilize subpopulation distributions within datasets. To characterize the structures in an interpretable manner, we propose the Subpopulation Structure Discovery with Large Language Models (SSD-LLM) framework, which employs world knowledge and instruction-following capabilities of Large Language Models (LLMs) to linguistically analyze informative image captions and summarize the structures. Furthermore, we propose complete workflows to address downstream tasks, named Task-specific Tuning, showcasing the application of the discovered structure to a spectrum of subpopulation-related tasks, including dataset subpopulation organization, subpopulation shift, and slice discovery. Furthermore, we propose complete workflows to address downstream tasks, named Task-specific Tuning, showcasing the application of the discovered structure to a spectrum of subpopulation-related tasks, including dataset subpopulation organization, subpopulation shift, and slice discovery.
Authors: Siyan Zhao, Tung Nguyen, Aditya Grover
Abstract: In-context learning is a key paradigm in large language models (LLMs) that enables them to generalize to new tasks and domains by simply prompting these models with a few exemplars without explicit parameter updates. Many attempts have been made to understand in-context learning in LLMs as a function of model scale, pretraining data, and other factors. In this work, we propose a new mechanism to probe and understand in-context learning from the lens of decision boundaries for in-context binary classification. Decision boundaries are straightforward to visualize and provide important information about the qualitative behavior of the inductive biases of standard classifiers. To our surprise, we find that the decision boundaries learned by current LLMs in simple binary classification tasks are often irregular and non-smooth, regardless of linear separability in the underlying task. This paper investigates the factors influencing these decision boundaries and explores methods to enhance their generalizability. We assess various approaches, including training-free and fine-tuning methods for LLMs, the impact of model architecture, and the effectiveness of active prompting techniques for smoothing decision boundaries in a data-efficient manner. Our findings provide a deeper understanding of in-context learning dynamics and offer practical improvements for enhancing robustness and generalizability of in-context learning.
Authors: Xuhong Wang, Haoyu Jiang, Yi Yu, Jingru Yu, Yilun Lin, Ping Yi, Yingchun Wang, Yu Qiao, Li Li, Fei-Yue Wang
Abstract: Large Language Models (LLMs) are increasingly integrated into diverse industries, posing substantial security risks due to unauthorized replication and misuse. To mitigate these concerns, robust identification mechanisms are widely acknowledged as an effective strategy. Identification systems for LLMs now rely heavily on watermarking technology to manage and protect intellectual property and ensure data security. However, previous studies have primarily concentrated on the basic principles of algorithms and lacked a comprehensive analysis of watermarking theory and practice from the perspective of intelligent identification. To bridge this gap, firstly, we explore how a robust identity recognition system can be effectively implemented and managed within LLMs by various participants using watermarking technology. Secondly, we propose a mathematical framework based on mutual information theory, which systematizes the identification process to achieve more precise and customized watermarking. Additionally, we present a comprehensive evaluation of performance metrics for LLM watermarking, reflecting participant preferences and advancing discussions on its identification applications. Lastly, we outline the existing challenges in current watermarking technologies and theoretical frameworks, and provide directional guidance to address these challenges. Our systematic classification and detailed exposition aim to enhance the comparison and evaluation of various methods, fostering further research and development toward a transparent, secure, and equitable LLM ecosystem.
Authors: Keisuke Sato
Abstract: This study investigates the influence of Japanese honorifics on the responses of large language models (LLMs) when explaining the law of conservation of momentum. We analyzed the outputs of six state-of-the-art AI models, including variations of ChatGPT, Coral, and Gemini, using 14 different honorific forms. Our findings reveal that honorifics significantly affect the quality, consistency, and formality of AI-generated responses, demonstrating LLMs' ability to interpret and adapt to social context cues embedded in language. Notable variations were observed across different models, with some emphasizing historical context and derivations, while others focused on intuitive explanations. The study highlights the potential for using honorifics to adjust the depth and complexity of AI-generated explanations in educational contexts. Furthermore, the responsiveness of AI models to cultural linguistic elements underscores the importance of considering cultural factors in AI development for educational applications. These results open new avenues for research in AI-assisted education and cultural adaptation in AI systems, with significant implications for personalizing learning experiences and developing culturally sensitive AI tools for global education.
Authors: Maud van Lier, Gorka Mu\~noz-Gil
Abstract: The arrival of Large Language Models (LLMs) has stirred up philosophical debates about the possibility of realizing agency in an artificial manner. In this work we contribute to the debate by presenting a theoretical model that can be used as a threshold conception for artificial agents. The model defines agents as systems whose actions and goals are always influenced by a dynamic framework of factors that consists of the agent's accessible history, its adaptive repertoire and its external environment. This framework, in turn, is influenced by the actions that the agent takes and the goals that it forms. We show with the help of the model that state-of-the-art LLMs are not agents yet, but that there are elements to them that suggest a way forward. The paper argues that a combination of the agent architecture presented in Park et al. (2023) together with the use of modules like the Coscientist in Boiko et al. (2023) could potentially be a way to realize agency in an artificial manner. We end the paper by reflecting on the obstacles one might face in building such an artificial agent and by presenting possible directions for future research.