Authors: Sander Schulhoff, Jeremy Pinto, Anaum Khan, Louis-François Bouchard, Chenglei Si, Svetlina Anati, Valen Tagliabue, Anson Liu Kost, Christopher Carnahan, Jordan Boyd-Graber
Large Language Models (LLMs) are increasingly being deployed in interactive contexts that involve direct user engagement, such as chatbots and writing assistants. These deployments are increasingly plagued by prompt injection and jailbreaking (collectively, prompt hacking), in which models are manipulated to ignore their original instructions and instead follow potentially malicious ones. Although widely acknowledged as a significant security threat, there is a dearth of large-scale resources and quantitative studies on prompt hacking. To address this lacuna, we launch a global prompt hacking competition, which allows for free-form human input attacks. We elicit 600K+ adversarial prompts against three state-of-the-art LLMs. We describe the dataset, which empirically verifies that current LLMs can indeed be manipulated via prompt hacking. We also present a comprehensive taxonomical ontology of the types of adversarial prompts.
Authors: Hui Yin, Amir Aryani, Gavin Lambert, Marcus White, Luis Salvador-Carulla, Shazia Sadiq, Elvira Sojli, Jennifer Boddy, Greg Murray, Wing Wah Tham
The number of publications related to the Sustainable Development Goals (SDGs) continues to grow. These publications cover a diverse spectrum of research, from humanities and social sciences to engineering and health. Given the imperative of funding bodies to monitor outcomes and impacts, linking publications to relevant SDGs is critical but remains time-consuming and difficult given the breadth and complexity of the SDGs. A publication may relate to several goals (interconnection feature of goals), and therefore require multidisciplinary knowledge to tag accurately. Machine learning approaches are promising and have proven particularly valuable for tasks such as manual data labeling and text classification. In this study, we employed over 82,000 publications from an Australian university as a case study. We utilized a similarity measure to map these publications onto Sustainable Development Goals (SDGs). Additionally, we leveraged the OpenAI GPT model to conduct the same task, facilitating a comparative analysis between the two approaches. Experimental results show that about 82.89% of the results obtained by the similarity measure overlap (at least one tag) with the outputs of the GPT model. The adopted model (similarity measure) can complement GPT model for SDG classification. Furthermore, deep learning methods, which include the similarity measure used here, are more accessible and trusted for dealing with sensitive data without the use of commercial AI services or the deployment of expensive computing resources to operate large language models. Our study demonstrates how a crafted combination of the two methods can achieve reliable results for mapping research to the SDGs.
Authors: Changnan Xiao, Bing Liu
Reasoning is a fundamental capability of AI agents. Recently, large language models (LLMs) have shown remarkable abilities to perform reasoning tasks. However, numerous evaluations of the reasoning capabilities of LLMs have also showed some limitations. An outstanding limitation is length generalization, meaning that when trained on reasoning problems of smaller lengths or sizes, the resulting models struggle with problems of larger sizes or lengths. This potentially indicates some theoretical limitations of generalization in learning reasoning skills. These evaluations and their observations motivated us to perform a theoretical study of the length generalization problem. This work focused on reasoning tasks that can be formulated as Markov dynamic processes (MDPs) and/or directed acyclic graphs (DAGs). It identifies and proves conditions that decide whether the length generalization problem can be solved or not for a reasoning task in a particular representation. Experiments are also conducted to verify the theoretical results.
Authors: Yuetian Chen, Mei Si
When dealing with text data containing subjective labels like speaker emotions, inaccuracies or discrepancies among labelers are not uncommon. Such discrepancies can significantly affect the performance of machine learning algorithms. This study investigates the potential of identifying and addressing outliers in text data with subjective labels, aiming to enhance classification outcomes. We utilized the Deep SVDD algorithm, a one-class classification method, to detect outliers in nine text-based emotion and sentiment analysis datasets. By employing both a small-sized language model (DistilBERT base model with 66 million parameters) and non-deep learning machine learning algorithms (decision tree, KNN, Logistic Regression, and LDA) as the classifier, our findings suggest that the removal of outliers can lead to enhanced results in most cases. Additionally, as outliers in such datasets are not necessarily unlearnable, we experienced utilizing a large language model -- DeBERTa v3 large with 131 million parameters, which can capture very complex patterns in data. We continued to observe performance enhancements across multiple datasets.
Authors: Yuhui Zhang, Brandon McKinzie, Zhe Gan, Vaishaal Shankar, Alexander Toshev
Recent advances in image tokenizers, such as VQ-VAE, have enabled text-to-image generation using auto-regressive methods, similar to language modeling. However, these methods have yet to leverage pre-trained language models, despite their adaptability to various downstream tasks. In this work, we explore this gap by adapting a pre-trained language model for auto-regressive text-to-image generation, and find that pre-trained language models offer limited help. We provide a two-fold explanation by analyzing tokens from each modality. First, we demonstrate that image tokens possess significantly different semantics compared to text tokens, rendering pre-trained language models no more effective in modeling them than randomly initialized ones. Second, the text tokens in the image-text datasets are too simple compared to normal language model pre-training data, which causes the catastrophic degradation of language models' capability.
Authors: Chengyang Zhang, Yong Zhang, Qitan Shao, Bo Li, Yisheng Lv, Xinglin Piao, Baocai Yin
Traffic prediction is one of the most significant foundations in Intelligent Transportation Systems (ITS). Traditional traffic prediction methods rely only on historical traffic data to predict traffic trends and face two main challenges. 1) insensitivity to unusual events. 2) poor performance in long-term prediction. In this work, we explore how generative models combined with text describing the traffic system can be applied for traffic generation and name the task Text-to-Traffic Generation (TTG). The key challenge of the TTG task is how to associate text with the spatial structure of the road network and traffic data for generating traffic situations. To this end, we propose ChatTraffic, the first diffusion model for text-to-traffic generation. To guarantee the consistency between synthetic and real data, we augment a diffusion model with the Graph Convolutional Network (GCN) to extract spatial correlations of traffic data. In addition, we construct a large dataset containing text-traffic pairs for the TTG task. We benchmarked our model qualitatively and quantitatively on the released dataset. The experimental results indicate that ChatTraffic can generate realistic traffic situations from the text. Our code and dataset are available at https://github.com/ChyaZhang/ChatTraffic.
Authors: Samuele Poppi, Tobia Poppi, Federico Cocchi, Marcella Cornia, Lorenzo Baraldi, Rita Cucchiara
Vision-and-Language models such as CLIP have demonstrated remarkable effectiveness across a wide range of tasks. However, these models 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 concern in their adoption. To overcome these limitations, we introduce a methodology to make Vision-and-Language models safer by removing their sensitivity to not-safe-for-work concepts. We show how this can be done by distilling from a large language model which converts between safe and unsafe sentences and which is fine-tuned starting from just 100 manually-curated pairs. We conduct extensive experiments on the resulting embedding space for both retrieval and text-to-image generation, where we show that our model can also be properly employed with pre-trained image generators. Our source code and trained models are available at: https://github.com/aimagelab/safe-clip.
Authors: Andreas Opedal, Eleftheria Tsipidi, Tiago Pimentel, Ryan Cotterell, Tim Vieira
The left-corner transformation (Rosenkrantz and Lewis, 1970) is used to remove left recursion from context-free grammars, which is an important step towards making the grammar parsable top-down with simple techniques. This paper generalizes prior left-corner transformations to support semiring-weighted production rules and to provide finer-grained control over which left corners may be moved. Our generalized left-corner transformation (GLCT) arose from unifying the left-corner transformation and speculation transformation (Eisner and Blatz, 2007), originally for logic programming. Our new transformation and speculation define equivalent weighted languages. Yet, their derivation trees are structurally different in an important way: GLCT replaces left recursion with right recursion, and speculation does not. We also provide several technical results regarding the formal relationships between the outputs of GLCT, speculation, and the original grammar. Lastly, we empirically investigate the efficiency of GLCT for left-recursion elimination from grammars of nine languages.
Authors: Yu-Chen Lin, Akhilesh Kumar, Wen-Liang Zhang, Norman Chang, Muhammad Zakir, Rucha Apte, Chao Wang, Jyh-Shing Roger Jang
Our paper investigates effective methods for code generation in "specific-domain" applications, including the use of Large Language Models (LLMs) for data segmentation and renewal, as well as stimulating deeper thinking in LLMs through prompt adjustments. Using a real company product as an example, we provide user manuals, API documentation, and other data. The ideas discussed in this paper help segment and then convert this data into semantic vectors to better reflect their true positioning. Subsequently, user requirements are transformed into vectors to retrieve the most relevant content, achieving about 70% accuracy in simple to medium-complexity tasks through various prompt techniques. This paper is the first to enhance specific-domain code generation effectiveness from this perspective. Additionally, we experiment with generating more scripts from a limited number using llama2-based fine-tuning to test its effectiveness in professional domain code generation. This is a challenging and promising field, and once achieved, it will not only lead to breakthroughs in LLM development across multiple industries but also enable LLMs to understand and learn any new knowledge effectively.
Authors: Kevin Wang, Seth Akins, Abdallah Mohammed, Ramon Lawrence
Generative AI systems such as ChatGPT have a disruptive effect on learning and assessment. Computer science requires practice to develop skills in problem solving and programming that are traditionally developed using assignments. Generative AI has the capability of completing these assignments for students with high accuracy, which dramatically increases the potential for academic integrity issues and students not achieving desired learning outcomes. This work investigates the performance of ChatGPT by evaluating it across three courses (CS1,CS2,databases). ChatGPT completes almost all introductory assessments perfectly. Existing detection methods, such as MOSS and JPlag (based on similarity metrics) and GPTzero (AI detection), have mixed success in identifying AI solutions. Evaluating instructors and teaching assistants using heuristics to distinguish between student and AI code shows that their detection is not sufficiently accurate. These observations emphasize the need for adapting assessments and improved detection methods.
Authors: Nikhil Anand, Joshua Tan, Maria Minakova
Modern ML systems ingest data aggregated from diverse sources, such as synthetic, human-annotated, and live customer traffic. Understanding \textit{which} examples are important to the performance of a learning algorithm is crucial for efficient model training. Recently, a growing body of literature has given rise to various "influence scores," which use training artifacts such as model confidence or checkpointed gradients to identify important subsets of data. However, these methods have primarily been developed in computer vision settings, and it remains unclear how well they generalize to language-based tasks using pretrained models.
In this paper, we explore the applicability of influence scores in language classification tasks. We evaluate a diverse subset of these scores on the SNLI dataset by quantifying accuracy changes in response to pruning training data through random and influence-score-based sampling. We then stress-test one of the scores -- "variance of gradients" (VoG) from Agarwal et al. (2022) -- in an NLU model stack that was exposed to dynamic user speech patterns in a voice assistant type of setting. Our experiments demonstrate that in many cases, encoder-based language models can be finetuned on roughly 50% of the original data without degradation in performance metrics. Along the way, we summarize lessons learned from applying out-of-the-box implementations of influence scores, quantify the effects of noisy and class-imbalanced data, and offer recommendations on score-based sampling for better accuracy and training efficiency.
Authors: Anusha Sabbineni, Nikhil Anand, Maria Minakova
While data selection methods have been studied extensively in active learning, data pruning, and data augmentation settings, there is little evidence for the efficacy of these methods in industry scale settings, particularly in low-resource languages. Our work presents ways of assessing prospective training examples in those settings for their "usefulness" or "difficulty". We also demonstrate how these measures can be used in selecting important examples for training supervised machine learning models. We primarily experiment with entropy and Error L2-Norm (EL2N) scores. We use these metrics to curate high quality datasets from a large pool of \textit{Weak Signal Labeled} data, which assigns no-defect high confidence hypotheses during inference as ground truth labels. We then conduct training data augmentation experiments using these de-identified datasets and demonstrate that score-based selection can result in a 2% decrease in semantic error rate and 4%-7% decrease in domain classification error rate when compared to the baseline technique of random selection.
Authors: Rob Grzywinski, Joshua D'Arcy, Rob Naidoff, Ashish Shukla, Alex Browne, Ren Gibbons, Brinnae Bent
Instruction-following language models demand robust methodologies for information retrieval to augment instructions for question-answering applications. A primary challenge is the resolution of coreferences in the context of chunking strategies for long documents. The critical barrier to experimentation of handling coreferences is a lack of open source datasets, specifically in question-answering tasks that require coreference resolution. In this work we present our Coreference Resolution in Question-Answering (CRaQAn) dataset, an open-source dataset that caters to the nuanced information retrieval requirements of coreference resolution in question-answering tasks by providing over 250 question-answer pairs containing coreferences. To develop this dataset, we developed a novel approach for creating high-quality datasets using an instruction-following model (GPT-4) and a Recursive Criticism and Improvement Loop.
Authors: Ranjita Naik, Spencer Rarrick, Vishal Chowdhary
Recent advances in neural methods have led to substantial improvement in the quality of Neural Machine Translation (NMT) systems. However, these systems frequently produce translations with inaccurate gender (Stanovsky et al., 2019), which can be traced to bias in training data. Saunders and Byrne (2020) tackle this problem with a handcrafted dataset containing balanced gendered profession words. By using this data to fine-tune an existing NMT model, they show that gender bias can be significantly mitigated, albeit at the expense of translation quality due to catastrophic forgetting. They recover some of the lost quality with modified training objectives or additional models at inference. We find, however, that simply supplementing the handcrafted dataset with a random sample from the base model training corpus is enough to significantly reduce the catastrophic forgetting. We also propose a novel domain-adaptation technique that leverages in-domain data created with the counterfactual data generation techniques proposed by Zmigrod et al. (2019) to further improve accuracy on the WinoMT challenge test set without significant loss in translation quality. We show its effectiveness in NMT systems from English into three morphologically rich languages French, Spanish, and Italian. The relevant dataset and code will be available at Github.
Authors: Yuhang Wang, Yanxu Zhu, Chao Kong, Shuyu Wei, Xiaoyuan Yi, Xing Xie, Jitao Sang
As the scaling of Large Language Models (LLMs) has dramatically enhanced their capabilities, there has been a growing focus on the alignment problem to ensure their responsible and ethical use. While existing alignment efforts predominantly concentrate on universal values such as the HHH principle, the aspect of culture, which is inherently pluralistic and diverse, has not received adequate attention. This work introduces a new benchmark, CDEval, aimed at evaluating the cultural dimensions of LLMs. CDEval is constructed by incorporating both GPT-4's automated generation and human verification, covering six cultural dimensions across seven domains. Our comprehensive experiments provide intriguing insights into the culture of mainstream LLMs, highlighting both consistencies and variations across different dimensions and domains. The findings underscore the importance of integrating cultural considerations in LLM development, particularly for applications in diverse cultural settings. Through CDEval, we aim to broaden the horizon of LLM alignment research by including cultural dimensions, thus providing a more holistic framework for the future development and evaluation of LLMs. This benchmark serves as a valuable resource for cultural studies in LLMs, paving the way for more culturally aware and sensitive models.
Authors: Takehiko Ohkawa, Takuma Yagi, Taichi Nishimura, Ryosuke Furuta, Atsushi Hashimoto, Yoshitaka Ushiku, Yoichi Sato
We propose a novel benchmark for cross-view knowledge transfer of dense video captioning, adapting models from web instructional videos with exocentric views to an egocentric view. While dense video captioning (predicting time segments and their captions) is primarily studied with exocentric videos (e.g., YouCook2), benchmarks with egocentric videos are restricted due to data scarcity. To overcome the limited video availability, transferring knowledge from abundant exocentric web videos is demanded as a practical approach. However, learning the correspondence between exocentric and egocentric views is difficult due to their dynamic view changes. The web videos contain mixed views focusing on either human body actions or close-up hand-object interactions, while the egocentric view is constantly shifting as the camera wearer moves. This necessitates the in-depth study of cross-view transfer under complex view changes. In this work, we first create a real-life egocentric dataset (EgoYC2) whose captions are shared with YouCook2, enabling transfer learning between these datasets assuming their ground-truth is accessible. To bridge the view gaps, we propose a view-invariant learning method using adversarial training in both the pre-training and fine-tuning stages. While the pre-training is designed to learn invariant features against the mixed views in the web videos, the view-invariant fine-tuning further mitigates the view gaps between both datasets. We validate our proposed method by studying how effectively it overcomes the view change problem and efficiently transfers the knowledge to the egocentric domain. Our benchmark pushes the study of the cross-view transfer into a new task domain of dense video captioning and will envision methodologies to describe egocentric videos in natural language.
Authors: Harsha Nori, Yin Tat Lee, Sheng Zhang, Dean Carignan, Richard Edgar, Nicolo Fusi, Nicholas King, Jonathan Larson, Yuanzhi Li, Weishung Liu, Renqian Luo, Scott Mayer McKinney, Robert Osazuwa Ness, Hoifung Poon, Tao Qin, Naoto Usuyama, Chris White, Eric Horvitz
Generalist foundation models such as GPT-4 have displayed surprising capabilities in a wide variety of domains and tasks. Yet, there is a prevalent assumption that they cannot match specialist capabilities of fine-tuned models. For example, most explorations to date on medical competency benchmarks have leveraged domain-specific training, as exemplified by efforts on BioGPT and Med-PaLM. We build on a prior study of GPT-4's capabilities on medical challenge benchmarks in the absence of special training. Rather than using simple prompting to highlight the model's out-of-the-box capabilities, we perform a systematic exploration of prompt engineering. We find that prompting innovation can unlock deeper specialist capabilities and show that GPT-4 easily tops prior leading results for medical benchmarks. The prompting methods we explore are general purpose, and make no specific use of domain expertise, removing the need for expert-curated content. Our experimental design carefully controls for overfitting during the prompt engineering process. We introduce Medprompt, based on a composition of several prompting strategies. With Medprompt, GPT-4 achieves state-of-the-art results on all nine of the benchmark datasets in the MultiMedQA suite. The method outperforms leading specialist models such as Med-PaLM 2 by a significant margin with an order of magnitude fewer calls to the model. Steering GPT-4 with Medprompt achieves a 27% reduction in error rate on the MedQA dataset over the best methods to date achieved with specialist models and surpasses a score of 90% for the first time. Beyond medical problems, we show the power of Medprompt to generalize to other domains and provide evidence for the broad applicability of the approach via studies of the strategy on exams in electrical engineering, machine learning, philosophy, accounting, law, nursing, and clinical psychology.
Authors: Minkyu Shin, Jin Kim
Although large language models (LLMs) are reshaping various aspects of human life, our current understanding of their impacts remains somewhat constrained. Here we investigate the impact of LLMs on human communication, in the context of consumer complaints in the financial industry. Employing an AI detection tool on more than 780K complaints gathered by the Consumer Financial Protection Bureau (CFPB), we find evidence of LLM usage in the writing of complaints - shortly after the release of ChatGPT. Our analyses reveal that LLM usage is positively correlated with the likelihood of obtaining desirable outcomes (i.e., offer of relief from financial firms) and suggest that this positive correlation may be partly due to the linguistic features improved by LLMs. We test this conjecture with a preregistered experiment, which reveals results consistent with those from observational studies: Consumer complaints written with ChatGPT for improved linguistic qualities were more likely to receive hypothetical relief offers than the original consumer complaints, demonstrating the LLM's ability to enhance message persuasiveness in human communication. Being some of the earliest empirical evidence on LLM usage for enhancing persuasion, our results highlight the transformative potential of LLMs in human communication.
Authors: Pingyi Chen, Honglin Li, Chenglu Zhu, Sunyi Zheng, Lin Yang
Whole slide images are the foundation of digital pathology for the diagnosis and treatment of carcinomas. Writing pathology reports is laborious and error-prone for inexperienced pathologists. To reduce the workload and improve clinical automation, we investigate how to generate pathology reports given whole slide images. On the data end, we curated the largest WSI-text dataset (TCGA-PathoText). In specific, we collected nearly 10000 high-quality WSI-text pairs for visual-language models by recognizing and cleaning pathology reports which narrate diagnostic slides in TCGA. On the model end, we propose the multiple instance generative model (MI-Gen) which can produce pathology reports for gigapixel WSIs. We benchmark our model on the largest subset of TCGA-PathoText. Experimental results show our model can generate pathology reports which contain multiple clinical clues. Furthermore, WSI-text prediction can be seen as an approach of visual-language pre-training, which enables our model to be transferred to downstream diagnostic tasks like carcinoma grading and phenotyping. We observe that simple semantic extraction from the pathology reports can achieve the best performance (0.838 of F1 score) on BRCA subtyping without adding extra parameters or tricky fine-tuning. Our collected dataset and related code will all be publicly available.
Authors: Yucheng Han, Chi Zhang, Xin Chen, Xu Yang, Zhibin Wang, Gang Yu, Bin Fu, Hanwang Zhang
Multi-modal large language models have demonstrated impressive performances on most vision-language tasks. However, the model generally lacks the understanding capabilities for specific domain data, particularly when it comes to interpreting chart figures. This is mainly due to the lack of relevant multi-modal instruction tuning datasets. In this article, we create a high-quality instruction-tuning dataset leveraging GPT-4. We develop a multi-step data generation process in which different steps are responsible for generating tabular data, creating chart figures, and designing instruction tuning data separately. Our method's flexibility enables us to generate diverse, high-quality instruction-tuning data consistently and efficiently while maintaining a low resource expenditure. Additionally, it allows us to incorporate a wider variety of chart and task types not yet featured in existing datasets. Next, we introduce ChartLlama, a multi-modal large language model that we've trained using our created dataset. ChartLlama outperforms all prior methods in ChartQA, Chart-to-text, and Chart-extraction evaluation benchmarks. Additionally, ChartLlama significantly improves upon the baseline in our specially compiled chart dataset, which includes new chart and task types. The results of ChartLlama confirm the value and huge potential of our proposed data generation method in enhancing chart comprehension.
Authors: Xiang Yue, Yuansheng Ni, Kai Zhang, Tianyu Zheng, Ruoqi Liu, Ge Zhang, Samuel Stevens, Dongfu Jiang, Weiming Ren, Yuxuan Sun, Cong Wei, Botao Yu, Ruibin Yuan, Renliang Sun, Ming Yin, Boyuan Zheng, Zhenzhu Yang, Yibo Liu, Wenhao Huang, Huan Sun, Yu Su, Wenhu Chen
We introduce MMMU: a new benchmark designed to evaluate multimodal models on massive multi-discipline tasks demanding college-level subject knowledge and deliberate reasoning. MMMU includes 11.5K meticulously collected multimodal questions from college exams, quizzes, and textbooks, covering six core disciplines: Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering. These questions span 30 subjects and 183 subfields, comprising 30 highly heterogeneous image types, such as charts, diagrams, maps, tables, music sheets, and chemical structures. Unlike existing benchmarks, MMMU focuses on advanced perception and reasoning with domain-specific knowledge, challenging models to perform tasks akin to those faced by experts. Our evaluation of 14 open-source LMMs and the proprietary GPT-4V(ision) highlights the substantial challenges posed by MMMU. Even the advanced GPT-4V only achieves a 56% accuracy, indicating significant room for improvement. We believe MMMU will stimulate the community to build next-generation multimodal foundation models towards expert artificial general intelligence.
Authors: Kazuki Yamauchi, Yusuke Ijima, Yuki Saito
We propose StyleCap, a method to generate natural language descriptions of speaking styles appearing in speech. Although most of conventional techniques for para-/non-linguistic information recognition focus on the category classification or the intensity estimation of pre-defined labels, they cannot provide the reasoning of the recognition result in an interpretable manner. As a first step towards an end-to-end method for generating speaking-style prompts from speech, i.e., automatic speaking-style captioning, StyleCap uses paired data of speech and natural language descriptions to train neural networks that predict prefix vectors fed into a large language model (LLM)-based text decoder from a speech representation vector. We explore an appropriate text decoder and speech feature representation suitable for this new task. The experimental results demonstrate that our StyleCap leveraging richer LLMs for the text decoder, speech self-supervised learning (SSL) features, and sentence rephrasing augmentation improves the accuracy and diversity of generated speaking-style captions. Samples of speaking-style captions generated by our StyleCap are publicly available.
Authors: Hao Pei, Si Lin, Chuanfu Li, Che Wang, Haoming Chen, Sizhe Li
A novel method for detecting faults in power grids using a graph neural network (GNN) has been developed, aimed at enhancing intelligent fault diagnosis in network operation and maintenance. This GNN-based approach identifies faulty nodes within the power grid through a specialized electrical feature extraction model coupled with a knowledge graph. Incorporating temporal data, the method leverages the status of nodes from preceding and subsequent time periods to aid in current fault detection. To validate the effectiveness of this GNN in extracting node features, a correlation analysis of the output features from each node within the neural network layer was conducted. The results from experiments show that this method can accurately locate fault nodes in simulated scenarios with a remarkable 99.53% accuracy. Additionally, the graph neural network's feature modeling allows for a qualitative examination of how faults spread across nodes, providing valuable insights for analyzing fault nodes.
Authors: Xinhong Chen, Zongxi Li, Yaowei Wang, Haoran Xie, Jianping Wang, Qing Li
The study of causal relationships between emotions and causes in texts has recently received much attention. Most works focus on extracting causally related clauses from documents. However, none of these works has considered that the causal relationships among the extracted emotion and cause clauses can only be valid under some specific context clauses. To highlight the context in such special causal relationships, we propose a new task to determine whether or not an input pair of emotion and cause has a valid causal relationship under different contexts and extract the specific context clauses that participate in the causal relationship. Since the task is new for which no existing dataset is available, we conduct manual annotation on a benchmark dataset to obtain the labels for our tasks and the annotations of each context clause's type that can also be used in some other applications. We adopt negative sampling to construct the final dataset to balance the number of documents with and without causal relationships. Based on the constructed dataset, we propose an end-to-end multi-task framework, where we design two novel and general modules to handle the two goals of our task. Specifically, we propose a context masking module to extract the context clauses participating in the causal relationships. We propose a prediction aggregation module to fine-tune the prediction results according to whether the input emotion and causes depend on specific context clauses. Results of extensive comparative experiments and ablation studies demonstrate the effectiveness and generality of our proposed framework.
Authors: Rui Yang, Qingcheng Zeng, Keen You, Yujie Qiao, Lucas Huang, Chia-Chun Hsieh, Benjamin Rosand, Jeremy Goldwasser, Amisha D Dave, Tiarnan D.L. Keenan, Emily Y Chew, Dragomir Radev, Zhiyong Lu, Hua Xu, Qingyu Chen, Irene Li
This study introduces MedGen, a comprehensive natural language processing (NLP) toolkit designed for medical text processing. MedGen is tailored for biomedical researchers and healthcare professionals with an easy-to-use, all-in-one solution that requires minimal programming expertise. It includes (1) Generative Functions: For the first time, MedGen includes four advanced generative functions: question answering, text summarization, text simplification, and machine translation; (2) Basic NLP Functions: MedGen integrates 12 essential NLP functions such as word tokenization and sentence segmentation; and (3) Query and Search Capabilities: MedGen provides user-friendly query and search functions on text corpora. We fine-tuned 32 domain-specific language models, evaluated them thoroughly on 24 established benchmarks and conducted manual reviews with clinicians. Additionally, we expanded our toolkit by introducing query and search functions, while also standardizing and integrating functions from third-party libraries. The toolkit, its models, and associated data are publicly available via https://github.com/Yale-LILY/MedGen.
Authors: Itamar Zimerman, Lior Wolf
Despite their dominance in modern DL and, especially, NLP domains, transformer architectures exhibit sub-optimal performance on long-range tasks compared to recent layers that are specifically designed for this purpose. In this work, drawing inspiration from key attributes of long-range layers, such as state-space layers, linear RNN layers, and global convolution layers, we demonstrate that minimal modifications to the transformer architecture can significantly enhance performance on the Long Range Arena (LRA) benchmark, thus narrowing the gap with these specialized layers. We identify that two key principles for long-range tasks are (i) incorporating an inductive bias towards smoothness, and (ii) locality. As we show, integrating these ideas into the attention mechanism improves results with a negligible amount of additional computation and without any additional trainable parameters. Our theory and experiments also shed light on the reasons for the inferior performance of transformers on long-range tasks and identify critical properties that are essential for successfully capturing long-range dependencies.
Authors: Gaël Le Mens, Aina Gallego
We use GPT-4 to obtain position estimates of political texts in continuous spaces. We develop and validate a new approach by positioning British party manifestos on the economic, social, and immigration policy dimensions and tweets by members of the US Congress on the left-right ideological spectrum. For the party manifestos, the correlation between the positions produced by GPT-4 and experts is 93% or higher, a performance similar to or better than that obtained with crowdsourced position estimates. For individual tweets, the positions obtained with GPT-4 achieve a correlation of 91% with crowdsourced position estimates. For senators of the 117th US Congress, the positions obtained with GPT-4 achieve a correlation of 97% with estimates based on roll call votes and of 96% with those based on campaign funding. Correlations are also substantial within party, indicating that position estimates produced with GPT-4 capture within-party differences between senators. Overall, using GPT-4 for ideological scaling is fast, cost-efficient, and reliable. This approach provides a viable alternative to scaling by both expert raters and crowdsourcing.
Authors: Jiahuan Yan, Haojun Gao, Zhang Kai, Weize Liu, Danny Chen, Jian Wu, Jintai Chen
Deep learning approaches exhibit promising performances on various text tasks. However, they are still struggling on medical text classification since samples are often extremely imbalanced and scarce. Different from existing mainstream approaches that focus on supplementary semantics with external medical information, this paper aims to rethink the data challenges in medical texts and present a novel framework-agnostic algorithm called Text2Tree that only utilizes internal label hierarchy in training deep learning models. We embed the ICD code tree structure of labels into cascade attention modules for learning hierarchy-aware label representations. Two new learning schemes, Similarity Surrogate Learning (SSL) and Dissimilarity Mixup Learning (DML), are devised to boost text classification by reusing and distinguishing samples of other labels following the label representation hierarchy, respectively. Experiments on authoritative public datasets and real-world medical records show that our approach stably achieves superior performances over classical and advanced imbalanced classification methods.
Authors: Gioele Cadamuro, Marco Gruppo
We hereby present a solution to a semantic textual similarity (STS) problem in which it is necessary to match two sentences containing, as the only distinguishing factor, highly specific information (such as names, addresses, identification codes), and from which we need to derive a definition for when they are similar and when they are not. The solution revolves around the use of a neural network, based on the siamese architecture, to create the distributions of the distances between similar and dissimilar pairs of sentences. The goal of these distributions is to find a discriminating factor, that we call "threshold", which represents a well-defined quantity that can be used to distinguish vector distances of similar pairs from vector distances of dissimilar pairs in new predictions and later analyses. In addition, we developed a way to score the predictions by combining attributes from both the distributions' features and the way the distance function works. Finally, we generalize the results showing that they can be transferred to a wider range of domains by applying the system discussed to a well-known and widely used benchmark dataset for STS problems.
Authors: Dan Ma, Jun Xu, Zongyu Wang, Xuezhi Cao, Yunsen Xian
Product reviews often contain a large number of implicit aspects and object-attribute co-existence cases. Unfortunately, many existing studies in Aspect-Based Sentiment Analysis (ABSA) have overlooked this issue, which can make it difficult to extract opinions comprehensively and fairly. In this paper, we propose a new task called Entity-Aspect-Opinion-Sentiment Quadruple Extraction (EASQE), which aims to hierarchically decompose aspect terms into entities and aspects to avoid information loss, non-exclusive annotations, and opinion misunderstandings in ABSA tasks. To facilitate research in this new task, we have constructed four datasets (Res14-EASQE, Res15-EASQE, Res16-EASQE, and Lap14-EASQE) based on the SemEval Restaurant and Laptop datasets. We have also proposed a novel two-stage sequence-tagging based Trigger-Opinion framework as the baseline for the EASQE task. Empirical evaluations show that our Trigger-Opinion framework can generate satisfactory EASQE results and can also be applied to other ABSA tasks, significantly outperforming state-of-the-art methods. We have made the four datasets and source code of Trigger-Opinion publicly available to facilitate further research in this area.
Authors: Mohamed Nejjar, Luca Zacharias, Fabian Stiehle, Ingo Weber
Large language models (LLMs) have been touted to enable increased productivity in many areas of today's work life. Scientific research as an area of work is no exception: the potential of LLM-based tools to assist in the daily work of scientists has become a highly discussed topic across disciplines. However, we are only at the very onset of this subject of study. It is still unclear how the potential of LLMs will materialise in research practice. With this study, we give first empirical evidence on the use of LLMs in the research process. We have investigated a set of use cases for LLM-based tools in scientific research, and conducted a first study to assess to which degree current tools are helpful. In this paper we report specifically on use cases related to software engineering, such as generating application code and developing scripts for data analytics. While we studied seemingly simple use cases, results across tools differ significantly. Our results highlight the promise of LLM-based tools in general, yet we also observe various issues, particularly regarding the integrity of the output these tools provide.
Authors: Amos Calamida, Farhad Nooralahzadeh, Morteza Rohanian, Koji Fujimoto, Mizuho Nishio, Michael Krauthammer
We propose a new automated evaluation metric for machine-generated radiology reports using the successful COMET architecture adapted for the radiology domain. We train and publish four medically-oriented model checkpoints, including one trained on RadGraph, a radiology knowledge graph. Our results show that our metric correlates moderately to high with established metrics such as BERTscore, BLEU, and CheXbert scores. Furthermore, we demonstrate that one of our checkpoints exhibits a high correlation with human judgment, as assessed using the publicly available annotations of six board-certified radiologists, using a set of 200 reports. We also performed our own analysis gathering annotations with two radiologists on a collection of 100 reports. The results indicate the potential effectiveness of our method as a radiology-specific evaluation metric. The code, data, and model checkpoints to reproduce our findings will be publicly available.
Authors: Andreas Madsen, Siva Reddy, Sarath Chandar
Neural networks for NLP are becoming increasingly complex and widespread, and there is a growing concern if these models are responsible to use. Explaining models helps to address the safety and ethical concerns and is essential for accountability. Interpretability serves to provide these explanations in terms that are understandable to humans. Additionally, post-hoc methods provide explanations after a model is learned and are generally model-agnostic. This survey provides a categorization of how recent post-hoc interpretability methods communicate explanations to humans, it discusses each method in-depth, and how they are validated, as the latter is often a common concern.
Authors: Hanwen Xu, Jiayou Zhang, Zhirui Wang, Shizhuo Zhang, Megh Manoj Bhalerao, Yucong Liu, Dawei Zhu, Sheng Wang
In the expansion of biomedical dataset, the same category may be labeled with different terms, thus being tedious and onerous to curate these terms. Therefore, automatically mapping synonymous terms onto the ontologies is desirable, which we name as biomedical synonym prediction task. Unlike biomedical concept normalization (BCN), no clues from context can be used to enhance synonym prediction, making it essential to extract graph features from ontology. We introduce an expert-curated dataset OBO-syn encompassing 70 different types of concepts and 2 million curated concept-term pairs for evaluating synonym prediction methods. We find BCN methods perform weakly on this task for not making full use of graph information. Therefore, we propose GraphPrompt, a prompt-based learning approach that creates prompt templates according to the graphs. GraphPrompt obtained 37.2\% and 28.5\% improvement on zero-shot and few-shot settings respectively, indicating the effectiveness of these graph-based prompt templates. We envision that our method GraphPrompt and OBO-syn dataset can be broadly applied to graph-based NLP tasks, and serve as the basis for analyzing diverse and accumulating biomedical data. All the data and codes are avalible at: https://github.com/HanwenXuTHU/GraphPrompt
Authors: Yiyang Li, Lei Li, Marina Litvak, Natalia Vanetik, Dingxin Hu, Yuze Li, Yanquan Zhou
The issue of factual consistency in abstractive summarization has received extensive attention in recent years, and the evaluation of factual consistency between summary and document has become an important and urgent task. Most of the current evaluation metrics are adopted from the question answering (QA) or natural language inference (NLI) task. However, the application of QA-based metrics is extremely time-consuming in practice while NLI-based metrics are lack of interpretability. In this paper, we propose a cloze-based evaluation framework called ClozE and show the great potential of the cloze-based metric. It inherits strong interpretability from QA, while maintaining the speed of NLI- level reasoning. We demonstrate that ClozE can reduce the evaluation time by nearly 96% relative to QA-based metrics while retaining their interpretability and performance through experiments on six human-annotated datasets and a meta-evaluation benchmark GO FIGURE (Gabriel et al., 2021). Finally, we discuss three important facets of ClozE in practice, which further shows better overall performance of ClozE compared to other metrics.
Authors: Sanket Vaibhav Mehta, Jai Gupta, Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Jinfeng Rao, Marc Najork, Emma Strubell, Donald Metzler
Differentiable Search Indices (DSIs) encode a corpus of documents in model parameters and use the same model to answer user queries directly. Despite the strong performance of DSI models, deploying them in situations where the corpus changes over time is computationally expensive because reindexing the corpus requires re-training the model. In this work, we introduce DSI++, a continual learning challenge for DSI to incrementally index new documents while being able to answer queries related to both previously and newly indexed documents. Across different model scales and document identifier representations, we show that continual indexing of new documents leads to considerable forgetting of previously indexed documents. We also hypothesize and verify that the model experiences forgetting events during training, leading to unstable learning. To mitigate these issues, we investigate two approaches. The first focuses on modifying the training dynamics. Flatter minima implicitly alleviate forgetting, so we optimize for flatter loss basins and show that the model stably memorizes more documents ($+12\%$). Next, we introduce a generative memory to sample pseudo-queries for documents and supplement them during continual indexing to prevent forgetting for the retrieval task. Extensive experiments on novel continual indexing benchmarks based on Natural Questions (NQ) and MS MARCO demonstrate that our proposed solution mitigates forgetting significantly. Concretely, it improves the average Hits@10 by $+21.1\%$ over competitive baselines for NQ and requires $6$ times fewer model updates compared to re-training the DSI model for incrementally indexing five corpora in a sequence.
Authors: Yejin Bang, Samuel Cahyawijaya, Nayeon Lee, Wenliang Dai, Dan Su, Bryan Wilie, Holy Lovenia, Ziwei Ji, Tiezheng Yu, Willy Chung, Quyet V. Do, Yan Xu, Pascale Fung
This paper proposes a framework for quantitatively evaluating interactive LLMs such as ChatGPT using publicly available data sets. We carry out an extensive technical evaluation of ChatGPT using 23 data sets covering 8 different common NLP application tasks. We evaluate the multitask, multilingual and multi-modal aspects of ChatGPT based on these data sets and a newly designed multimodal dataset. We find that ChatGPT outperforms LLMs with zero-shot learning on most tasks and even outperforms fine-tuned models on some tasks. We find that it is better at understanding non-Latin script languages than generating them. It is able to generate multimodal content from textual prompts, via an intermediate code generation step. Moreover, we find that ChatGPT is 63.41% accurate on average in 10 different reasoning categories under logical reasoning, non-textual reasoning, and commonsense reasoning, hence making it an unreliable reasoner. It is, for example, better at deductive than inductive reasoning. ChatGPT suffers from hallucination problems like other LLMs and it generates more extrinsic hallucinations from its parametric memory as it does not have access to an external knowledge base. Finally, the interactive feature of ChatGPT enables human collaboration with the underlying LLM to improve its performance, i.e, 8% ROUGE-1 on summarization and 2% ChrF++ on machine translation, in a multi-turn "prompt engineering" fashion. We also release codebase for evaluation set extraction.
Authors: Mengxia Yu, Zhihan Zhang, Wenhao Yu, Meng Jiang
Comparative reasoning is a process of comparing objects, concepts, or entities to draw conclusions, which constitutes a fundamental cognitive ability. In this paper, we propose a novel framework to pre-train language models for enhancing their abilities of comparative reasoning over texts. While there have been approaches for NLP tasks that require comparative reasoning, they suffer from costly manual data labeling and limited generalizability to different tasks. Our approach introduces a novel method of collecting scalable data for text-based entity comparison, which leverages both structured and unstructured data. Moreover, we present a framework of pre-training language models via three novel objectives on comparative reasoning. Evaluation on downstream tasks including comparative question answering, question generation, and summarization shows that our pre-training framework significantly improves the comparative reasoning abilities of language models, especially under low-resource conditions. This work also releases the first integrated benchmark for comparative reasoning.
Authors: Dan Iter, Reid Pryzant, Ruochen Xu, Shuohang Wang, Yang Liu, Yichong Xu, Chenguang Zhu
Large language models (LLMs) can use in-context demonstrations to improve performance on zero-shot tasks. However, selecting the best in-context examples is challenging because model performance can vary widely depending on the selected examples. We present a cross-entropy difference (CED) method for selecting in-context demonstrations. Our method is based on the observation that the effectiveness of in-context demonstrations negatively correlates with the perplexity of the test example by a language model that was finetuned on that demonstration. We utilize parameter efficient finetuning to train small models on training data that are used for computing the cross-entropy difference between a test example and every candidate in-context demonstration. This metric is used to rank and select in-context demonstrations independently for each test input. We evaluate our method on a mix-domain dataset that combines 8 benchmarks, representing 4 text generation tasks, showing that CED for in-context demonstration selection can improve performance for a variety of LLMs.
Authors: L Siddharth, Jianxi Luo
Aimed at supporting knowledge-intensive tasks in the design process, populating design knowledge from text documents involves the extraction of triples - head entity :: relationship :: tail entity or h :: r :: t that could be combined into a knowledge graph representation. As relationships are largely chosen from ontological or common-sense alternatives, knowledge graphs built using these depict an approximation or restricted view of design knowledge, rather than what is explicated in text document. In this article, we present a data-driven approach to identify and explicate facts (h :: r :: t) from sentences in patent documents. We create a dataset of 44,227 sentences and facts, encompassing all patent classifications while also capturing the variations among patent document sections. Using this dataset, we train taggers that classify tokens to: 1) identify all entities (h) and relationships (r) and 2) specific relationships (r) for a pair of entities (h :: ___ :: t). While these taggers are built upon transformer-based sequence classification models, we evaluate our proposed method against edge classification approaches that use linear classifiers and graph neural networks, incorporating transformer-based token embeddings and linguistic features. The simplicity and coverage of the proposed method enable its application to patent documents at any scale and variety. Upon deploying an open-source python package, we apply our method to patent documents related to fan systems. From the knowledge graphs thus extracted, we explain how facts could be generalised to domain ontologies as well as be specified to subsystem levels. We also highlight the importance of knowledge graph representations by retrieving and explicating the knowledge of key issues in fan systems, while holding a comparative discussion against opinions from ChatGPT.
Authors: Aounon Kumar, Chirag Agarwal, Suraj Srinivas, Aaron Jiaxun Li, Soheil Feizi, Himabindu Lakkaraju
Large language models (LLMs) released for public use incorporate guardrails to ensure their output is safe, often referred to as "model alignment." An aligned language model should decline a user's request to produce harmful content. However, such safety measures are vulnerable to adversarial attacks, which add maliciously designed token sequences to a harmful prompt to bypass the model's safety guards. In this work, we introduce erase-and-check, the first framework to defend against adversarial prompts with verifiable safety guarantees. We defend against three attack modes: i) adversarial suffix, which appends an adversarial sequence at the end of the prompt; ii) adversarial insertion, where the adversarial sequence is inserted anywhere in the middle of the prompt; and iii) adversarial infusion, where adversarial tokens are inserted at arbitrary positions in the prompt, not necessarily as a contiguous block. Our experimental results demonstrate that this procedure can obtain strong certified safety guarantees on harmful prompts while maintaining good empirical performance on safe prompts. For example, against adversarial suffixes of length 20, it certifiably detects 92% of harmful prompts and labels 94% of safe prompts correctly using the open-source language model Llama 2 as the safety filter. We further improve the filter's performance, in terms of accuracy and speed, by replacing Llama 2 with a DistilBERT safety classifier fine-tuned on safe and harmful prompts. Additionally, we propose two efficient empirical defenses: i) RandEC, a randomized version of erase-and-check that evaluates the safety filter on a small subset of the erased subsequences, and ii) GradEC, a gradient-based version that optimizes the erased tokens to remove the adversarial sequence. The code for our experiments is available at https://github.com/aounon/certified-llm-safety.
Authors: Shiqi Chen, Yiran Zhao, Jinghan Zhang, I-Chun Chern, Siyang Gao, Pengfei Liu, Junxian He
Assessing factuality of text generated by large language models (LLMs) is an emerging yet crucial research area, aimed at alerting users to potential errors and guiding the development of more reliable LLMs. Nonetheless, the evaluators assessing factuality necessitate suitable evaluation themselves to gauge progress and foster advancements. This direction remains under-explored, resulting in substantial impediments to the progress of factuality evaluators. To mitigate this issue, we introduce a benchmark for Factuality Evaluation of large Language Models, referred to as felm. In this benchmark, we collect responses generated from LLMs and annotate factuality labels in a fine-grained manner. Contrary to previous studies that primarily concentrate on the factuality of world knowledge (e.g.~information from Wikipedia), felm focuses on factuality across diverse domains, spanning from world knowledge to math and reasoning. Our annotation is based on text segments, which can help pinpoint specific factual errors. The factuality annotations are further supplemented by predefined error types and reference links that either support or contradict the statement. In our experiments, we investigate the performance of several LLM-based factuality evaluators on felm, including both vanilla LLMs and those augmented with retrieval mechanisms and chain-of-thought processes. Our findings reveal that while retrieval aids factuality evaluation, current LLMs are far from satisfactory to faithfully detect factual errors.
Authors: Abdul Karim Gizzini, Mustafa Shukor, Ali J. Ghandour
Current AI-based methods do not provide comprehensible physical interpretations of the utilized data, extracted features, and predictions/inference operations. As a result, deep learning models trained using high-resolution satellite imagery lack transparency and explainability and can be merely seen as a black box, which limits their wide-level adoption. Experts need help understanding the complex behavior of AI models and the underlying decision-making process. The explainable artificial intelligence (XAI) field is an emerging field providing means for robust, practical, and trustworthy deployment of AI models. Several XAI techniques have been proposed for image classification tasks, whereas the interpretation of image segmentation remains largely unexplored. This paper offers to bridge this gap by adapting the recent XAI classification algorithms and making them usable for muti-class image segmentation, where we mainly focus on buildings' segmentation from high-resolution satellite images. To benchmark and compare the performance of the proposed approaches, we introduce a new XAI evaluation methodology and metric based on "Entropy" to measure the model uncertainty. Conventional XAI evaluation methods rely mainly on feeding area-of-interest regions from the image back to the pre-trained (utility) model and then calculating the average change in the probability of the target class. Those evaluation metrics lack the needed robustness, and we show that using Entropy to monitor the model uncertainty in segmenting the pixels within the target class is more suitable. We hope this work will pave the way for additional XAI research for image segmentation and applications in the remote sensing discipline.
Authors: Liang Chen, Yichi Zhang, Shuhuai Ren, Haozhe Zhao, Zefan Cai, Yuchi Wang, Peiyi Wang, Tianyu Liu, Baobao Chang
In this study, we explore the potential of Multimodal Large Language Models (MLLMs) in improving embodied decision-making processes for agents. While Large Language Models (LLMs) have been widely used due to their advanced reasoning skills and vast world knowledge, MLLMs like GPT4-Vision offer enhanced visual understanding and reasoning capabilities. We investigate whether state-of-the-art MLLMs can handle embodied decision-making in an end-to-end manner and whether collaborations between LLMs and MLLMs can enhance decision-making. To address these questions, we introduce a new benchmark called PCA-EVAL, which evaluates embodied decision-making from the perspectives of Perception, Cognition, and Action. Additionally, we propose HOLMES, a multi-agent cooperation framework that allows LLMs to leverage MLLMs and APIs to gather multimodal information for informed decision-making. We compare end-to-end embodied decision-making and HOLMES on our benchmark and find that the GPT4-Vision model demonstrates strong end-to-end embodied decision-making abilities, outperforming GPT4-HOLMES in terms of average decision accuracy (+3%). However, this performance is exclusive to the latest GPT4-Vision model, surpassing the open-source state-of-the-art MLLM by 26%. Our results indicate that powerful MLLMs like GPT4-Vision hold promise for decision-making in embodied agents, offering new avenues for MLLM research. Code and data are open at https://github.com/pkunlp-icler/PCA-EVAL/.
Authors: Utsav Garg, Erhan Bas
Instruction-tuned large language models (LLMs) have demonstrated promising zero-shot generalization capabilities across various downstream tasks. Recent research has introduced multimodal capabilities to LLMs by integrating independently pretrained vision encoders through model grafting. These multimodal variants undergo instruction tuning, similar to LLMs, enabling effective zero-shot generalization for multimodal tasks. This study conducts a comparative analysis of different multimodal instruction tuning approaches and evaluates their performance across a range of tasks, including complex reasoning, conversation, image captioning, multiple-choice questions (MCQs), and binary classification. Through rigorous benchmarking and ablation experiments, we reveal key insights for guiding architectural choices when incorporating multimodal capabilities into LLMs. However, current approaches have limitations; they do not sufficiently address the need for a diverse multimodal instruction dataset, which is crucial for enhancing task generalization. Additionally, they overlook issues related to truthfulness and factuality when generating responses. These findings illuminate current methodological constraints in adapting language models for image comprehension and provide valuable guidance for researchers and practitioners seeking to harness multimodal versions of LLMs.
Authors: Golam Md Muktadir
This paper presents a comprehensive exploration of the evolution of prompt engineering and generation in the field of natural language processing (NLP). Starting from the early language models and information retrieval systems, we trace the key developments that have shaped prompt engineering over the years. The introduction of attention mechanisms in 2015 revolutionized language understanding, leading to advancements in controllability and context-awareness. Subsequent breakthroughs in reinforcement learning techniques further enhanced prompt engineering, addressing issues like exposure bias and biases in generated text. We examine the significant contributions in 2018 and 2019, focusing on fine-tuning strategies, control codes, and template-based generation. The paper also discusses the growing importance of fairness, human-AI collaboration, and low-resource adaptation. In 2020 and 2021, contextual prompting and transfer learning gained prominence, while 2022 and 2023 witnessed the emergence of advanced techniques like unsupervised pre-training and novel reward shaping. Throughout the paper, we reference specific research studies that exemplify the impact of various developments on prompt engineering. The journey of prompt engineering continues, with ethical considerations being paramount for the responsible and inclusive future of AI systems.
Authors: Letian Zhang, Xiaotong Zhai, Zhongkai Zhao, Yongshuo Zong, Xin Wen, Bingchen Zhao
Counterfactual reasoning, a fundamental aspect of human cognition, involves contemplating alternatives to established facts or past events, significantly enhancing our abilities in planning and decision-making. In light of the advancements in current multi-modal large language models, we explore their effectiveness in counterfactual reasoning. To facilitate this investigation, we introduce a novel dataset, C-VQA, specifically designed to test the counterfactual reasoning capabilities of modern multi-modal large language models. This dataset is constructed by infusing original questions with counterfactual presuppositions, spanning various types such as numerical and boolean queries. It encompasses a mix of real and synthetic data, representing a wide range of difficulty levels. Our thorough evaluations of contemporary vision-language models using this dataset have revealed substantial performance drops, with some models showing up to a 40% decrease, highlighting a significant gap between current models and human-like vision reasoning capabilities. We hope our dataset will serve as a vital benchmark for evaluating the counterfactual reasoning capabilities of models. Code and dataset are publicly available at https://bzhao.me/C-VQA/.
Authors: Linlu Qiu, Liwei Jiang, Ximing Lu, Melanie Sclar, Valentina Pyatkin, Chandra Bhagavatula, Bailin Wang, Yoon Kim, Yejin Choi, Nouha Dziri, Xiang Ren
The ability to derive underlying principles from a handful of observations and then generalize to novel situations -- known as inductive reasoning -- is central to human intelligence. Prior work suggests that language models (LMs) often fall short on inductive reasoning, despite achieving impressive success on research benchmarks. In this work, we conduct a systematic study of the inductive reasoning capabilities of LMs through iterative hypothesis refinement, a technique that more closely mirrors the human inductive process than standard input-output prompting. Iterative hypothesis refinement employs a three-step process: proposing, selecting, and refining hypotheses in the form of textual rules. By examining the intermediate rules, we observe that LMs are phenomenal hypothesis proposers (i.e., generating candidate rules), and when coupled with a (task-specific) symbolic interpreter that is able to systematically filter the proposed set of rules, this hybrid approach achieves strong results across inductive reasoning benchmarks that require inducing causal relations, language-like instructions, and symbolic concepts. However, they also behave as puzzling inductive reasoners, showing notable performance gaps between rule induction (i.e., identifying plausible rules) and rule application (i.e., applying proposed rules to instances), suggesting that LMs are proposing hypotheses without being able to actually apply the rules. Through empirical and human analyses, we further reveal several discrepancies between the inductive reasoning processes of LMs and humans, shedding light on both the potentials and limitations of using LMs in inductive reasoning tasks.
Authors: Yixiao Li, Yifan Yu, Chen Liang, Pengcheng He, Nikos Karampatziakis, Weizhu Chen, Tuo Zhao
Quantization is an indispensable technique for serving Large Language Models (LLMs) and has recently found its way into LoRA fine-tuning. In this work we focus on the scenario where quantization and LoRA fine-tuning are applied together on a pre-trained model. In such cases it is common to observe a consistent gap in the performance on downstream tasks between full fine-tuning and quantization plus LoRA fine-tuning approach. In response, we propose LoftQ (LoRA-Fine-Tuning-aware Quantization), a novel quantization framework that simultaneously quantizes an LLM and finds a proper low-rank initialization for LoRA fine-tuning. Such an initialization alleviates the discrepancy between the quantized and full-precision model and significantly improves generalization in downstream tasks. We evaluate our method on natural language understanding, question answering, summarization, and natural language generation tasks. Experiments show that our method is highly effective and outperforms existing quantization methods, especially in the challenging 2-bit and 2/4-bit mixed precision regimes. The code is available on https://github.com/yxli2123/LoftQ.
Authors: Hung Le, Hailin Chen, Amrita Saha, Akash Gokul, Doyen Sahoo, Shafiq Joty
Large Language Models (LLMs) have already become quite proficient at solving simpler programming tasks like those in HumanEval or MBPP benchmarks. However, solving more complex and competitive programming tasks is still quite challenging for these models - possibly due to their tendency to generate solutions as monolithic code blocks instead of decomposing them into logical sub-tasks and sub-modules. On the other hand, experienced programmers instinctively write modularized code with abstraction for solving complex tasks, often reusing previously developed modules. To address this gap, we propose CodeChain, a novel framework for inference that elicits modularized code generation through a chain of self-revisions, each being guided by some representative sub-modules generated in previous iterations. Concretely, CodeChain first instructs the LLM to generate modularized codes through chain-of-thought prompting. Then it applies a chain of self-revisions by iterating the two steps: 1) extracting and clustering the generated sub-modules and selecting the cluster representatives as the more generic and re-usable implementations, and 2) augmenting the original chain-of-thought prompt with these selected module-implementations and instructing the LLM to re-generate new modularized solutions. We find that by naturally encouraging the LLM to reuse the previously developed and verified sub-modules, CodeChain can significantly boost both modularity as well as correctness of the generated solutions, achieving relative pass@1 improvements of 35% on APPS and 76% on CodeContests. It is shown to be effective on both OpenAI LLMs as well as open-sourced LLMs like WizardCoder. We also conduct comprehensive ablation studies with different methods of prompting, number of clusters, model sizes, program qualities, etc., to provide useful insights that underpin CodeChain's success.
Authors: Eric J. Bigelow, Ekdeep Singh Lubana, Robert P. Dick, Hidenori Tanaka, Tomer D. Ullman
Large language models (LLMs) trained on huge corpora of text datasets demonstrate intriguing capabilities, achieving state-of-the-art performance on tasks they were not explicitly trained for. The precise nature of LLM capabilities is often mysterious, and different prompts can elicit different capabilities through in-context learning. We propose a framework that enables us to analyze in-context learning dynamics to understand latent concepts underlying LLMs' behavioral patterns. This provides a more nuanced understanding than success-or-failure evaluation benchmarks, but does not require observing internal activations as a mechanistic interpretation of circuits would. Inspired by the cognitive science of human randomness perception, we use random binary sequences as context and study dynamics of in-context learning by manipulating properties of context data, such as sequence length. In the latest GPT-3.5+ models, we find emergent abilities to generate seemingly random numbers and learn basic formal languages, with striking in-context learning dynamics where model outputs transition sharply from seemingly random behaviors to deterministic repetition.
Authors: Nuo Chen, Zinan Zheng, Ning Wu, Ming Gong, Yangqiu Song, Dongmei Zhang, Jia Li
Existing research predominantly focuses on developing powerful language learning models (LLMs) for mathematical reasoning within monolingual languages, with few explorations in preserving efficacy in a multilingual context. To bridge this gap, this paper pioneers exploring and training powerful Multilingual Math Reasoning (xMR) LLMs. Firstly, by utilizing translation, we construct the first multilingual math reasoning instruction dataset, MGSM8KInstruct, encompassing ten distinct languages, thus addressing the issue of training data scarcity in xMR tasks. Based on the collected dataset, we propose different training strategies to build powerful xMR LLMs, named MathOctopus, notably outperform conventional open-source LLMs and exhibit superiority over ChatGPT in few-shot scenarios. Notably, MathOctopus-13B reaches 47.6% accuracy which exceeds ChatGPT 46.3% on MGSM testset. Beyond remarkable results, we unearth several pivotal observations and insights from extensive experiments: (1) When extending the rejection sampling strategy to the multilingual context, it proves effective for model performances, albeit limited. (2) Employing parallel corpora for math Supervised Fine-Tuning (SFT) across multiple languages not only significantly enhances model performance multilingually but also elevates their monolingual performance. This indicates that crafting multilingual corpora can be regarded as a vital strategy for enhancing model performance in a specific language, especially in mathematical reasoning tasks. For instance, MathOctopus-7B improves its counterparts that trained on English from 42.2% to 50.8% on GSM8K testset.
Authors: Indira Sen, Dennis Assenmacher, Mattia Samory, Isabelle Augenstein, Wil van der Aalst, Claudia Wagner
NLP models are used in a variety of critical social computing tasks, such as detecting sexist, racist, or otherwise hateful content. Therefore, it is imperative that these models are robust to spurious features. Past work has attempted to tackle such spurious features using training data augmentation, including Counterfactually Augmented Data (CADs). CADs introduce minimal changes to existing training data points and flip their labels; training on them may reduce model dependency on spurious features. However, manually generating CADs can be time-consuming and expensive. Hence in this work, we assess if this task can be automated using generative NLP models. We automatically generate CADs using Polyjuice, ChatGPT, and Flan-T5, and evaluate their usefulness in improving model robustness compared to manually-generated CADs. By testing both model performance on multiple out-of-domain test sets and individual data point efficacy, our results show that while manual CADs are still the most effective, CADs generated by ChatGPT come a close second. One key reason for the lower performance of automated methods is that the changes they introduce are often insufficient to flip the original label.
Authors: Antonios Georgiou, Tankut Can, Mikhail Katkov, Misha Tsodyks
One of the most impressive achievements of the AI revolution is the development of large language models that can generate meaningful text and respond to instructions in plain English with no additional training necessary. Here we show that language models can be used as a scientific instrument for studying human memory for meaningful material. We developed a pipeline for designing large scale memory experiments and analyzing the obtained results. We performed online memory experiments with a large number of participants and collected recognition and recall data for narratives of different lengths. We found that both recall and recognition performance scale linearly with narrative length. Furthermore, in order to investigate the role of narrative comprehension in memory, we repeated these experiments using scrambled versions of the presented stories. We found that even though recall performance declined significantly, recognition remained largely unaffected. Interestingly, recalls in this condition seem to follow the original narrative order rather than the scrambled presentation, pointing to a contextual reconstruction of the story in memory.
Authors: Licheng Wen, Xuemeng Yang, Daocheng Fu, Xiaofeng Wang, Pinlong Cai, Xin Li, Tao Ma, Yingxuan Li, Linran Xu, Dengke Shang, Zheng Zhu, Shaoyan Sun, Yeqi Bai, Xinyu Cai, Min Dou, Shuanglu Hu, Botian Shi, Yu Qiao
The pursuit of autonomous driving technology hinges on the sophisticated integration of perception, decision-making, and control systems. Traditional approaches, both data-driven and rule-based, have been hindered by their inability to grasp the nuance of complex driving environments and the intentions of other road users. This has been a significant bottleneck, particularly in the development of common sense reasoning and nuanced scene understanding necessary for safe and reliable autonomous driving. The advent of Visual Language Models (VLM) represents a novel frontier in realizing fully autonomous vehicle driving. This report provides an exhaustive evaluation of the latest state-of-the-art VLM, GPT-4V(ision), and its application in autonomous driving scenarios. We explore the model's abilities to understand and reason about driving scenes, make decisions, and ultimately act in the capacity of a driver. Our comprehensive tests span from basic scene recognition to complex causal reasoning and real-time decision-making under varying conditions. Our findings reveal that GPT-4V demonstrates superior performance in scene understanding and causal reasoning compared to existing autonomous systems. It showcases the potential to handle out-of-distribution scenarios, recognize intentions, and make informed decisions in real driving contexts. However, challenges remain, particularly in direction discernment, traffic light recognition, vision grounding, and spatial reasoning tasks. These limitations underscore the need for further research and development. Project is now available on GitHub for interested parties to access and utilize: \url{https://github.com/PJLab-ADG/GPT4V-AD-Exploration}
Authors: Yuhan Li, Zhixun Li, Peisong Wang, Jia Li, Xiangguo Sun, Hong Cheng, Jeffrey Xu Yu
Graph plays a significant role in representing and analyzing complex relationships in real-world applications such as citation networks, social networks, and biological data. Recently, Large Language Models (LLMs), which have achieved tremendous success in various domains, have also been leveraged in graph-related tasks to surpass traditional Graph Neural Networks (GNNs) based methods and yield state-of-the-art performance. In this survey, we first present a comprehensive review and analysis of existing methods that integrate LLMs with graphs. First of all, we propose a new taxonomy, which organizes existing methods into three categories based on the role (i.e., enhancer, predictor, and alignment component) played by LLMs in graph-related tasks. Then we systematically survey the representative methods along the three categories of the taxonomy. Finally, we discuss the remaining limitations of existing studies and highlight promising avenues for future research. The relevant papers are summarized and will be consistently updated at: https://github.com/yhLeeee/Awesome-LLMs-in-Graph-tasks.
Authors: Sue Lim, Ralf Schmälzle
Advancements in artificial intelligence (AI) over the last decade demonstrate that machines can exhibit communicative behavior and influence how humans think, feel, and behave. In fact, the recent development of ChatGPT has shown that large language models (LLMs) can be leveraged to generate high-quality communication content at scale and across domains, suggesting that they will be increasingly used in practice. However, many questions remain about how knowing the source of the messages influences recipients' evaluation of and preference for AI-generated messages compared to human-generated messages. This paper investigated this topic in the context of vaping prevention messaging. In Study 1, which was pre-registered, we examined the influence of source disclosure on people's evaluation of AI-generated health prevention messages compared to human-generated messages. We found that source disclosure (i.e., labeling the source of a message as AI vs. human) significantly impacted the evaluation of the messages but did not significantly alter message rankings. In a follow-up study (Study 2), we examined how the influence of source disclosure may vary by the participants' negative attitudes towards AI. We found a significant moderating effect of negative attitudes towards AI on message evaluation, but not for message selection. However, for those with moderate levels of negative attitudes towards AI, source disclosure decreased the preference for AI-generated messages. Overall, the results of this series of studies showed a slight bias against AI-generated messages once the source was disclosed, adding to the emerging area of study that lies at the intersection of AI and communication.