Authors: Chetanya Rastogi, Prabhat Agarwal, Shreya Singh
In many applications, such as scientific literature management, researcher search, social network analysis and etc, Name Disambiguation (aiming at disambiguating WhoIsWho) has been a challenging problem. In addition, the growth of scientific literature makes the problem more difficult and urgent. Although name disambiguation has been extensively studied in academia and industry, the problem has not been solved well due to the clutter of data and the complexity of the same name scenario. In this work, we aim to explore models that can perform the task of name disambiguation using the network structure that is intrinsic to the problem and present an analysis of the models.
Authors: Sang Yun Kwon, Gagan Bhatia, El Moatez Billah Nagoudi, Muhammad Abdul-Mageed
Large language models (LLMs) finetuned to follow human instruction have recently exhibited significant capabilities in various English NLP tasks. However, their performance in grammatical error correction (GEC), especially on languages other than English, remains significantly unexplored. In this work, we evaluate the abilities of instruction finetuned LLMs in Arabic GEC, a complex task due to Arabic's rich morphology. Our findings suggest that various prompting methods, coupled with (in-context) few-shot learning, demonstrate considerable effectiveness, with GPT-4 achieving up to $65.49$ F$_{1}$ score under expert prompting (approximately $5$ points higher than our established baseline). Despite these positive results, we find that instruction finetuned models, regardless of their size, are still outperformed by fully finetuned ones, even if they are significantly smaller in size. This disparity highlights substantial room for improvements for LLMs. Inspired by methods used in low-resource machine translation, we also develop a method exploiting synthetic data that significantly outperforms previous models on two standard Arabic benchmarks. Our best model achieves a new SOTA on Arabic GEC, with $73.29$ and $73.26$ F$_{1}$ on the 2014 and 2015 QALB datasets, respectively, compared to peer-reviewed published baselines.
Authors: Veysel Kocaman, Hasham Ul Haq, David Talby
Recent research advances achieve human-level accuracy for de-identifying free-text clinical notes on research datasets, but gaps remain in reproducing this in large real-world settings. This paper summarizes lessons learned from building a system used to de-identify over one billion real clinical notes, in a fully automated way, that was independently certified by multiple organizations for production use. A fully automated solution requires a very high level of accuracy that does not require manual review. A hybrid context-based model architecture is described, which outperforms a Named Entity Recogniton (NER) - only model by 10% on the i2b2-2014 benchmark. The proposed system makes 50%, 475%, and 575% fewer errors than the comparable AWS, Azure, and GCP services respectively while also outperforming ChatGPT by 33%. It exceeds 98% coverage of sensitive data across 7 European languages, without a need for fine tuning. A second set of described models enable data obfuscation -- replacing sensitive data with random surrogates -- while retaining name, date, gender, clinical, and format consistency. Both the practical need and the solution architecture that provides for reliable & linked anonymized documents are described.
Authors: Lionel Wong, Jiayuan Mao, Pratyusha Sharma, Zachary S. Siegel, Jiahai Feng, Noa Korneev, Joshua B. Tenenbaum, Jacob Andreas
Effective planning in the real world requires not only world knowledge, but the ability to leverage that knowledge to build the right representation of the task at hand. Decades of hierarchical planning techniques have used domain-specific temporal action abstractions to support efficient and accurate planning, almost always relying on human priors and domain knowledge to decompose hard tasks into smaller subproblems appropriate for a goal or set of goals. This paper describes Ada (Action Domain Acquisition), a framework for automatically constructing task-specific planning representations using task-general background knowledge from language models (LMs). Starting with a general-purpose hierarchical planner and a low-level goal-conditioned policy, Ada interactively learns a library of planner-compatible high-level action abstractions and low-level controllers adapted to a particular domain of planning tasks. On two language-guided interactive planning benchmarks (Mini Minecraft and ALFRED Household Tasks), Ada strongly outperforms other approaches that use LMs for sequential decision-making, offering more accurate plans and better generalization to complex tasks.
Authors: Golnaz Shapurian, Michael J Kurtz, Alberto Accomazzi
The automatic identification of planetary feature names in astronomy publications presents numerous challenges. These features include craters, defined as roughly circular depressions resulting from impact or volcanic activity; dorsas, which are elongate raised structures or wrinkle ridges; and lacus, small irregular patches of dark, smooth material on the Moon, referred to as "lake" (Planetary Names Working Group, n.d.). Many feature names overlap with places or people's names that they are named after, for example, Syria, Tempe, Einstein, and Sagan, to name a few (U.S. Geological Survey, n.d.). Some feature names have been used in many contexts, for instance, Apollo, which can refer to mission, program, sample, astronaut, seismic, seismometers, core, era, data, collection, instrument, and station, in addition to the crater on the Moon. Some feature names can appear in the text as adjectives, like the lunar craters Black, Green, and White. Some feature names in other contexts serve as directions, like craters West and South on the Moon. Additionally, some features share identical names across different celestial bodies, requiring disambiguation, such as the Adams crater, which exists on both the Moon and Mars. We present a multi-step pipeline combining rule-based filtering, statistical relevance analysis, part-of-speech (POS) tagging, named entity recognition (NER) model, hybrid keyword harvesting, knowledge graph (KG) matching, and inference with a locally installed large language model (LLM) to reliably identify planetary names despite these challenges. When evaluated on a dataset of astronomy papers from the Astrophysics Data System (ADS), this methodology achieves an F1-score over 0.97 in disambiguating planetary feature names.
Authors: Xiaoxia Wu, Haojun Xia, Stephen Youn, Zhen Zheng, Shiyang Chen, Arash Bakhtiari, Michael Wyatt, Yuxiong He, Olatunji Ruwase, Leon Song, Zhewei Yao
This study examines 4-bit quantization methods like GPTQ in large language models (LLMs), highlighting GPTQ's overfitting and limited enhancement in Zero-Shot tasks. While prior works merely focusing on zero-shot measurement, we extend task scope to more generative categories such as code generation and abstractive summarization, in which we found that INT4 quantization can significantly underperform. However, simply shifting to higher precision formats like FP6 has been particularly challenging, thus overlooked, due to poor performance caused by the lack of sophisticated integration and system acceleration strategies on current AI hardware. Our results show that FP6, even with a coarse-grain quantization scheme, performs robustly across various algorithms and tasks, demonstrating its superiority in accuracy and versatility. Notably, with the FP6 quantization, \codestar-15B model performs comparably to its FP16 counterpart in code generation, and for smaller models like the 406M it closely matches their baselines in summarization. Neither can be achieved by INT4. To better accommodate various AI hardware and achieve the best system performance, we propose a novel 4+2 design for FP6 to achieve similar latency to the state-of-the-art INT4 fine-grain quantization. With our design, FP6 can become a promising solution to the current 4-bit quantization methods used in LLMs.
Authors: Jingxuan Wei, Linzhuang Sun, Xu Tan, Bihui Yu, Ruifeng Guo
Knowledge distillation, a technique for model compression and performance enhancement, has gained significant traction in Neural Machine Translation (NMT). However, existing research primarily focuses on empirical applications, and there is a lack of comprehensive understanding of how student model capacity, data complexity, and decoding strategies collectively influence distillation effectiveness. Addressing this gap, our study conducts an in-depth investigation into these factors, particularly focusing on their interplay in word-level and sequence-level distillation within NMT. Through extensive experimentation across datasets like IWSLT13 En$\rightarrow$Fr, IWSLT14 En$\rightarrow$De, and others, we empirically validate hypotheses related to the impact of these factors on knowledge distillation. Our research not only elucidates the significant influence of model capacity, data complexity, and decoding strategies on distillation effectiveness but also introduces a novel, optimized distillation approach. This approach, when applied to the IWSLT14 de$\rightarrow$en translation task, achieves state-of-the-art performance, demonstrating its practical efficacy in advancing the field of NMT.
Authors: Kaiqiang Song, Xiaoyang Wang, Sangwoo Cho, Xiaoman Pan, Dong Yu
This paper introduces a novel approach to enhance the capabilities of Large Language Models (LLMs) in processing and understanding extensive text sequences, a critical aspect in applications requiring deep comprehension and synthesis of large volumes of information. Recognizing the inherent challenges in extending the context window for LLMs, primarily built on Transformer architecture, we propose a new model architecture, referred to as Zebra. This architecture efficiently manages the quadratic time and memory complexity issues associated with full attention in the Transformer by employing grouped local-global attention layers. Our model, akin to a zebra's alternating stripes, balances local and global attention layers, significantly reducing computational requirements and memory consumption. Comprehensive experiments, including pretraining from scratch, continuation of long context adaptation training, and long instruction tuning, are conducted to evaluate the Zebra's performance. The results show that Zebra achieves comparable or superior performance on both short and long sequence benchmarks, while also enhancing training and inference efficiency.
Authors: Yu Ji, Wen Wu, Yi Hu, Hong Zheng, Liang He
Few-shot prompting elicits the remarkable abilities of large language models by equipping them with a few demonstration examples in the input. However, the traditional method of providing large language models with all demonstration input-output pairs at once may not effectively guide large language models to learn the specific input-output mapping relationship. In this paper, inspired by the regulatory and supportive role of metacognition in students' learning, we propose a novel metacognition-enhanced few-shot prompting, which guides large language models to reflect on their thought processes to comprehensively learn the given demonstration examples. Furthermore, considering that positive reinforcement can improve students' learning motivation, we introduce positive reinforcement into our metacognition-enhanced few-shot prompting to promote the few-shot learning of large language models by providing response-based positive feedback. The experimental results on two real-world datasets show that our metacognition-enhanced few-shot prompting with positive reinforcement surpasses traditional few-shot prompting in classification accuracy and macro F1.
Authors: Junjie Li, Yiwei Guo, Xie Chen, Kai Yu
Zero-shot voice conversion (VC) aims to transfer the source speaker timbre to arbitrary unseen target speaker timbre, while keeping the linguistic content unchanged. Although the voice of generated speech can be controlled by providing the speaker embedding of the target speaker, the speaker similarity still lags behind the ground truth recordings. In this paper, we propose SEF-VC, a speaker embedding free voice conversion model, which is designed to learn and incorporate speaker timbre from reference speech via a powerful position-agnostic cross-attention mechanism, and then reconstruct waveform from HuBERT semantic tokens in a non-autoregressive manner. The concise design of SEF-VC enhances its training stability and voice conversion performance. Objective and subjective evaluations demonstrate the superiority of SEF-VC to generate high-quality speech with better similarity to target reference than strong zero-shot VC baselines, even for very short reference speeches.
Authors: Ye Chen, Wei Cai, Liangmin Wu, Xiaowei Li, Zhanxuan Xin, Cong Fu
We release and introduce the TigerBot family of large language models (LLMs), consisting of base and chat models, sized from 7, 13, 70 and 180 billion parameters. We develop our models embarking from Llama-2 and BLOOM, and push the boundary further in data, training algorithm, infrastructure, and application tools. Our models yield meaningful performance gain over SOTA open-source models, e.g., Llama-2, specifically 6\% gain in English and 20\% gain in Chinese. TigerBot model family also achieves leading performance in major academic and industrial benchmarks and leaderboards. We believe that TigerBot represents just a snapshot of lightning-fast progression in LLM open-source community. Therefore, we are thrilled to give back by publicly releasing our models and reporting our approach behind, with additional emphases on building SOTA LLMs in a democratized way and making LLMs of use in real-world applications.
Authors: Sorouralsadat Fatemi, Yuheng Hu
Financial sentiment analysis plays a crucial role in uncovering latent patterns and detecting emerging trends, enabling individuals to make well-informed decisions that may yield substantial advantages within the constantly changing realm of finance. Recently, Large Language Models (LLMs) have demonstrated their effectiveness in diverse domains, showcasing remarkable capabilities even in zero-shot and few-shot in-context learning for various Natural Language Processing (NLP) tasks. Nevertheless, their potential and applicability in the context of financial sentiment analysis have not been thoroughly explored yet. To bridge this gap, we employ two approaches: in-context learning (with a focus on gpt-3.5-turbo model) and fine-tuning LLMs on a finance-domain dataset. Given the computational costs associated with fine-tuning LLMs with large parameter sizes, our focus lies on smaller LLMs, spanning from 250M to 3B parameters for fine-tuning. We then compare the performances with state-of-the-art results to evaluate their effectiveness in the finance-domain. Our results demonstrate that fine-tuned smaller LLMs can achieve comparable performance to state-of-the-art fine-tuned LLMs, even with models having fewer parameters and a smaller training dataset. Additionally, the zero-shot and one-shot performance of LLMs produces comparable results with fine-tuned smaller LLMs and state-of-the-art outcomes. Furthermore, our analysis demonstrates that there is no observed enhancement in performance for finance-domain sentiment analysis when the number of shots for in-context learning is increased.
Authors: Bo Li, Wei Ye, Quansen Wang, Wen Zhao, Shikun Zhang
Textual label names (descriptions) are typically semantically rich in many natural language understanding (NLU) tasks. In this paper, we incorporate the prompting methodology, which is widely used to enrich model input, into the label side for the first time. Specifically, we propose a Mask Matching method, which equips an input with a prompt and its label with another, and then makes predictions by matching their mask representations. We evaluate our method extensively on 8 NLU tasks with 14 datasets. The experimental results show that Mask Matching significantly outperforms its counterparts of fine-tuning and conventional prompt-tuning, setting up state-of-the-art performances in several datasets. Mask Matching is particularly good at handling NLU tasks with large label counts and informative label names. As pioneering efforts that investigate the label-side prompt, we also discuss open issues for future study.
Authors: Thinh Pham, Dat Quoc Nguyen
Profile-based intent detection and slot filling are important tasks aimed at reducing the ambiguity in user utterances by leveraging user-specific supporting profile information. However, research in these two tasks has not been extensively explored. To fill this gap, we propose a joint model, namely JPIS, designed to enhance profile-based intent detection and slot filling. JPIS incorporates the supporting profile information into its encoder and introduces a slot-to-intent attention mechanism to transfer slot information representations to intent detection. Experimental results show that our JPIS substantially outperforms previous profile-based models, establishing a new state-of-the-art performance in overall accuracy on the Chinese benchmark dataset ProSLU.
Authors: Dat Thanh Nguyen
In recent years, the availability of large-scale annotated datasets, such as the Stanford Natural Language Inference and the Multi-Genre Natural Language Inference, coupled with the advent of pre-trained language models, has significantly contributed to the development of the natural language inference domain. However, these crowdsourced annotated datasets often contain biases or dataset artifacts, leading to overestimated model performance and poor generalization. In this work, we focus on investigating dataset artifacts and developing strategies to address these issues. Through the utilization of a novel statistical testing procedure, we discover a significant association between vocabulary distribution and text entailment classes, emphasizing vocabulary as a notable source of biases. To mitigate these issues, we propose several automatic data augmentation strategies spanning character to word levels. By fine-tuning the ELECTRA pre-trained language model, we compare the performance of boosted models with augmented data against their baseline counterparts. The experiments demonstrate that the proposed approaches effectively enhance model accuracy and reduce biases by up to 0.66% and 1.14%, respectively.
Authors: Daiki Asami, Saku Sugawara
What makes a presupposition of an utterance -- information taken for granted by its speaker -- different from other pragmatic inferences such as an entailment is projectivity (e.g., the negative sentence the boy did not stop shedding tears presupposes the boy had shed tears before). The projectivity may vary depending on the combination of presupposition triggers and environments. However, prior natural language understanding studies fail to take it into account as they either use no human baseline or include only negation as an entailment-canceling environment to evaluate models' performance. The current study attempts to reconcile these issues. We introduce a new dataset, projectivity of presupposition (PROPRES, which includes 12k premise-hypothesis pairs crossing six triggers involving some lexical variety with five environments. Our human evaluation reveals that humans exhibit variable projectivity in some cases. However, the model evaluation shows that the best-performed model, DeBERTa, does not fully capture it. Our findings suggest that probing studies on pragmatic inferences should take extra care of the human judgment variability and the combination of linguistic items.
Authors: Tony T. Wang, Miles Wang, Kaivu Hariharan, Nir Shavit
LLMs often face competing pressures (for example helpfulness vs. harmlessness). To understand how models resolve such conflicts, we study Llama-2-chat models on the forbidden fact task. Specifically, we instruct Llama-2 to truthfully complete a factual recall statement while forbidding it from saying the correct answer. This often makes the model give incorrect answers. We decompose Llama-2 into 1000+ components, and rank each one with respect to how useful it is for forbidding the correct answer. We find that in aggregate, around 35 components are enough to reliably implement the full suppression behavior. However, these components are fairly heterogeneous and many operate using faulty heuristics. We discover that one of these heuristics can be exploited via a manually designed adversarial attack which we call The California Attack. Our results highlight some roadblocks standing in the way of being able to successfully interpret advanced ML systems. Project website available at https://forbiddenfacts.github.io .
Authors: Navreet Kaur, Monojit Choudhury, Danish Pruthi
As corporations rush to integrate large language models (LLMs) to their search offerings, it is critical that they provide factually accurate information that is robust to any presuppositions that a user may express. In this work, we introduce UPHILL, a dataset consisting of health-related queries with varying degrees of presuppositions. Using UPHILL, we evaluate the factual accuracy and consistency of InstructGPT, ChatGPT, and BingChat models. We find that while model responses rarely disagree with true health claims (posed as questions), they often fail to challenge false claims: responses from InstructGPT agree with 32% of the false claims, ChatGPT 26% and BingChat 23%. As we increase the extent of presupposition in input queries, the responses from InstructGPT and ChatGPT agree with the claim considerably more often, regardless of its veracity. Responses from BingChat, which rely on retrieved webpages, are not as susceptible. Given the moderate factual accuracy, and the inability of models to consistently correct false assumptions, our work calls for a careful assessment of current LLMs for use in high-stakes scenarios.
Authors: Chaoya Jiang, Wei ye, Haiyang Xu, Qinghao Ye, Ming Yan, Ji Zhang, Shikun Zhang
Self-supervised Multi-modal Contrastive Learning (SMCL) remarkably advances modern Vision-Language Pre-training (VLP) models by aligning visual and linguistic modalities. Due to noises in web-harvested text-image pairs, however, scaling up training data volume in SMCL presents considerable obstacles in terms of computational cost and data inefficiency. To improve data efficiency in VLP, we propose Text-aware Image Mixing (TiMix), which integrates mix-based data augmentation techniques into SMCL, yielding significant performance improvements without significantly increasing computational overhead. We provide a theoretical analysis of TiMixfrom a mutual information (MI) perspective, showing that mixed data samples for cross-modal contrastive learning implicitly serve as a regularizer for the contrastive loss. The experimental results demonstrate that TiMix exhibits a comparable performance on downstream tasks, even with a reduced amount of training data and shorter training time, when benchmarked against existing methods. This work empirically and theoretically demonstrates the potential of data mixing for data-efficient and computationally viable VLP, benefiting broader VLP model adoption in practical scenarios.
Authors: Zhiyue Liu, Jinyuan Liu, Fanrong Ma
Although image captioning models have made significant advancements in recent years, the majority of them heavily depend on high-quality datasets containing paired images and texts which are costly to acquire. Previous works leverage the CLIP's cross-modal association ability for image captioning, relying solely on textual information under unsupervised settings. However, not only does a modality gap exist between CLIP text and image features, but a discrepancy also arises between training and inference due to the unavailability of real-world images, which hinders the cross-modal alignment in text-only captioning. This paper proposes a novel method to address these issues by incorporating synthetic image-text pairs. A pre-trained text-to-image model is deployed to obtain images that correspond to textual data, and the pseudo features of generated images are optimized toward the real ones in the CLIP embedding space. Furthermore, textual information is gathered to represent image features, resulting in the image features with various semantics and the bridged modality gap. To unify training and inference, synthetic image features would serve as the training prefix for the language decoder, while real images are used for inference. Additionally, salient objects in images are detected as assistance to enhance the learning of modality alignment. Experimental results demonstrate that our method obtains the state-of-the-art performance on benchmark datasets.
Authors: Xijie Huang, Li Lyna Zhang, Kwang-Ting Cheng, Mao Yang
Large language models (LLMs) have shown impressive capabilities in various tasks, yet they still struggle with math reasoning. Despite efforts to optimize Chain-of-Thoughts (CoT) prompts and fine-tune LLMs, the potential of few-shot learning remains unexplored. In this work, we propose CoT-Max, a novel approach pushing the boundaries of few-shot CoT learning to improve LLM math reasoning capabilities. CoT-Max addresses the challenges of the selection of useful examples and limited number of examples due to restricted context window length. Inspired by our observation that natural language inputs contain many redundancy, we propose a coarse-to-fine pruner as a plug-and-play module for LLMs, which first identifies crucial CoT examples from a large batch and then further prunes unimportant tokens. To train the pruner, we collect a math reasoning dataset with diverse difficulty and steps, introduce a reward to measure both the input's effectiveness for math reasoning and token length constraints, and propose a novel training approach with reinforcement learning. As a result, CoT-Max significantly outperforms CoT and few-shot prompting baselines across various LLMs (LLaMA2-7B, 13B, 70B) and 5 mathematical datasets, achieving up to 4.55% absolute improvements. Remarkably, without any fine-tuning, LLaMA2-70B with CoT-Max surpasses GPT-3.5 and a wide range of larger LLMs (PaLM, Minerva, etc.) on the GSM8K.
Authors: Zichao Wei, Yewei Qin
Linguistic landscape is an important field in sociolinguistic research. Eye tracking technology is a common technology in psychological research. There are few cases of using eye movement to study linguistic landscape. This paper uses eye tracking technology to study the actual fixation of the linguistic landscape and finds that in the two dimensions of fixation time and fixation times, the fixation of native Chinese speakers to the linguistic landscape is higher than that of the general landscape. This paper argues that this phenomenon is due to the higher information density of linguistic landscapes. At the same time, the article also discusses other possible reasons for this phenomenon.
Authors: Haoran Liao, Qinyi Du, Shaohua Hu, Hao He, Yanyan Xu, Jidong Tian, Yaohui Jin
Large language models (LLMs) face challenges in solving complex mathematical problems that require comprehensive capacities to parse the statements, associate domain knowledge, perform compound logical reasoning, and integrate the intermediate rationales. Tackling all these problems once could be arduous for LLMs, thus leading to confusion in generation. In this work, we explore the potential of enhancing LLMs with agents by meticulous decomposition and modeling of mathematical reasoning process. Specifically, we propose a formal description of the mathematical solving and extend LLMs with an agent-based zero-shot framework named $\bf{P}$lanner-$\bf{R}$easoner-$\bf{E}$xecutor-$\bf{R}$eflector (PRER). We further provide and implement two MathAgents that define the logical forms and inherent relations via a pool of actions in different grains and orientations: MathAgent-M adapts its actions to LLMs, while MathAgent-H aligns with humankind. Experiments on miniF2F and MATH have demonstrated the effectiveness of PRER and proposed MathAgents, achieving an increase of $12.3\%$($53.9\%\xrightarrow{}66.2\%$) on the MiniF2F, $9.2\%$ ($49.8\%\xrightarrow{}59.0\%$) on MATH, and $13.2\%$($23.2\%\xrightarrow{}35.4\%$) for level-5 problems of MATH against GPT-4. Further analytical results provide more insightful perspectives on exploiting the behaviors of LLMs as agents.
Authors: Peiyi Wang, Lei Li, Zhihong Shao, R.X. Xu, Damai Dai, Yifei Li, Deli Chen, Y.Wu, Zhifang Sui
Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks. However, even the most advanced open-source LLMs, such as the LLaMA family models, still face challenges when it comes to accurately solving complex multi-step mathematical problems. In this paper, we present an innovative process-oriented math verifier called \textbf{Math-Shepherd}, which assigns a reward score to each step of the LLM's outputs on math problems. The training of Math-Shepherd is achieved using automatically constructed process-wise supervision data, breaking the bottleneck of heavy reliance on manual annotation in existing work. With the guidance of Math-Shepherd, a series of open-source LLMs demonstrate exceptional performance. Among them, DeepSeek 67B \citep{DeepSeek-llm} stands out by achieving accuracy rates of 93.3\% on the GSM8K dataset and 48.1\% on the MATH dataset, without external enhancement such as tool usage. Our Math-Shepherd also outperforms the self-consistency method and other existing verification models. We believe that automatic process supervision holds significant potential for the future evolution of LLMs.
Authors: Maria Milkova, Maksim Rudnev, Lidia Okolskaya
Basic values are concepts or beliefs which pertain to desirable end-states and transcend specific situations. Studying personal values in social media can illuminate how and why societal values evolve especially when the stimuli-based methods, such as surveys, are inefficient, for instance, in hard-to-reach populations. On the other hand, user-generated content is driven by the massive use of stereotyped, culturally defined speech constructions rather than authentic expressions of personal values. We aimed to find a model that can accurately detect value-expressive posts in Russian social media VKontakte. A training dataset of 5,035 posts was annotated by three experts, 304 crowd-workers and ChatGPT. Crowd-workers and experts showed only moderate agreement in categorizing posts. ChatGPT was more consistent but struggled with spam detection. We applied an ensemble of human- and AI-assisted annotation involving active learning approach, subsequently trained several LLMs and selected a model based on embeddings from pre-trained fine-tuned rubert-tiny2, and reached a high quality of value detection with F1 = 0.75 (F1-macro = 0.80). This model provides a crucial step to a study of values within and between Russian social media users.
Authors: Markus Reiter-Haas, Beate Klösch, Markus Hadler, Elisabeth Lex
Revealing the framing of news articles is an important yet neglected task in information seeking and retrieval. In the present work, we present FrameFinder, an open tool for extracting and analyzing frames in textual data. FrameFinder visually represents the frames of text from three perspectives, i.e., (i) frame labels, (ii) frame dimensions, and (iii) frame structure. By analyzing the well-established gun violence frame corpus, we demonstrate the merits of our proposed solution to support social science research and call for subsequent integration into information interactions.
Authors: Dang Van Thin, Duong Ngoc Hao, Ngan Luu-Thuy Nguyen
The ComOM shared task aims to extract comparative opinions from product reviews in Vietnamese language. There are two sub-tasks, including (1) Comparative Sentence Identification (CSI) and (2) Comparative Element Extraction (CEE). The first task is to identify whether the input is a comparative review, and the purpose of the second task is to extract the quintuplets mentioned in the comparative review. To address this task, our team proposes a two-stage system based on fine-tuning a BERTology model for the CSI task and unified multi-task instruction tuning for the CEE task. Besides, we apply the simple data augmentation technique to increase the size of the dataset for training our model in the second stage. Experimental results show that our approach outperforms the other competitors and has achieved the top score on the official private test.
Authors: Tong Guo
In industry NLP application, our manually labeled data has a certain number of noisy data. We present a simple method to find the noisy data and relabel them manually, meanwhile we collect the correction information. Then we present novel method to incorporate the human correction information into deep learning model. Human know how to correct noisy data. So the correction information can be inject into deep learning model. We do the experiment on our own text classification dataset, which is manually labeled, because we need to relabel the noisy data in our dataset for our industry application. The experiment result shows that our learn-on-correction method improve the classification accuracy from 91.7% to 92.5% in test dataset. The 91.7% accuracy is trained on the corrected dataset, which improve the baseline from 83.3% to 91.7% in test dataset. The accuracy under human evaluation achieves more than 97%.
Authors: Max Glockner, Ieva Staliūnaitė, James Thorne, Gisela Vallejo, Andreas Vlachos, Iryna Gurevych
Automated fact-checking systems verify claims against evidence to predict their veracity. In real-world scenarios, the retrieved evidence may not unambiguously support or refute the claim and yield conflicting but valid interpretations. Existing fact-checking datasets assume that the models developed with them predict a single veracity label for each claim, thus discouraging the handling of such ambiguity. To address this issue we present AmbiFC, a fact-checking dataset with 10k claims derived from real-world information needs. It contains fine-grained evidence annotations of 50k passages from 5k Wikipedia pages. We analyze the disagreements arising from ambiguity when comparing claims against evidence in AmbiFC, observing a strong correlation of annotator disagreement with linguistic phenomena such as underspecification and probabilistic reasoning. We develop models for predicting veracity handling this ambiguity via soft labels and find that a pipeline that learns the label distribution for sentence-level evidence selection and veracity prediction yields the best performance. We compare models trained on different subsets of AmbiFC and show that models trained on the ambiguous instances perform better when faced with the identified linguistic phenomena.
Authors: Jared Edward Reser
This article provides an analytical framework for how to simulate human-like thought processes within a computer. It describes how attention and memory should be structured, updated, and utilized to search for associative additions to the stream of thought. The focus is on replicating the dynamics of the mammalian working memory system, which features two forms of persistent activity: sustained firing (preserving information on the order of seconds) and synaptic potentiation (preserving information from minutes to hours). The article uses a series of over 40 original figures to systematically demonstrate how the iterative updating of these working memory stores provides functional structure to behavior, cognition, and consciousness.
In an AI implementation, these two memory stores should be updated continuously and in an iterative fashion, meaning each state should preserve a proportion of the coactive representations from the state before it. Thus, the set of concepts in working memory will evolve gradually and incrementally over time. This makes each state a revised iteration of the preceding state and causes successive states to overlap and blend with respect to the information they contain. Transitions between states happen as persistent activity spreads activation energy throughout the hierarchical network searching long-term memory for the most appropriate representation to be added to the global workspace. The result is a chain of associatively linked intermediate states capable of advancing toward a solution or goal. Iterative updating is conceptualized here as an information processing strategy, a model of working memory, a theory of consciousness, and an algorithm for designing and programming artificial general intelligence.
Authors: Hammad A. Ayyubi, Christopher Thomas, Lovish Chum, Rahul Lokesh, Long Chen, Yulei Niu, Xudong Lin, Xuande Feng, Jaywon Koo, Sounak Ray, Shih-Fu Chang
Events describe happenings in our world that are of importance. Naturally, understanding events mentioned in multimedia content and how they are related forms an important way of comprehending our world. Existing literature can infer if events across textual and visual (video) domains are identical (via grounding) and thus, on the same semantic level. However, grounding fails to capture the intricate cross-event relations that exist due to the same events being referred to on many semantic levels. For example, in Figure 1, the abstract event of "war" manifests at a lower semantic level through subevents "tanks firing" (in video) and airplane "shot" (in text), leading to a hierarchical, multimodal relationship between the events.
In this paper, we propose the task of extracting event hierarchies from multimodal (video and text) data to capture how the same event manifests itself in different modalities at different semantic levels. This reveals the structure of events and is critical to understanding them. To support research on this task, we introduce the Multimodal Hierarchical Events (MultiHiEve) dataset. Unlike prior video-language datasets, MultiHiEve is composed of news video-article pairs, which makes it rich in event hierarchies. We densely annotate a part of the dataset to construct the test benchmark. We show the limitations of state-of-the-art unimodal and multimodal baselines on this task. Further, we address these limitations via a new weakly supervised model, leveraging only unannotated video-article pairs from MultiHiEve. We perform a thorough evaluation of our proposed method which demonstrates improved performance on this task and highlight opportunities for future research.
Authors: Nianlong Gu, Richard H.R. Hahnloser
Citation generation aims to generate a citation sentence that refers to a chosen paper in the context of a manuscript. However, a rigid citation generation process is at odds with an author's desire to control specific attributes, such as 1) the citation intent, e.g., either introducing background information or comparing results, and 2) keywords that should appear in the citation text. To provide these degrees of controllability during citation generation, we propose to integrate the manuscript context, the context of the referenced paper, and the desired control attributes into a structured template and use it to fine-tune a language model (LM) via next-token prediction. We then utilize Proximal Policy Optimization to directly optimize the LM in favor of a high score of our proposed controllability metric. The proposed workflow harmoniously combines citation attribute suggestion and conditional citation generation into one LM, allowing for better user control.
Authors: Jianfei Gao, Yangze Zhou, Jincheng Zhou, Bruno Ribeiro
The task of inductive link prediction in knowledge graphs (KGs) generally focuses on test predictions with solely new nodes but not both new nodes and new relation types. In this work, we formally define the concept of double permutation-equivariant representations that are equivariant to permutations of both node identities and edge relation types. We then show how double-equivariant architectures are able to self-supervise pre-train on distinct KG domains and zero-shot predict links on a new KG domain (with completely new entities and new relation types). We also introduce the concept of distributionally double equivariant positional embeddings designed to perform the same task. Finally, we empirically demonstrate the capability of the proposed models against baselines on a set of novel real-world benchmarks. More interestingly, we show that self-supervised pre-training on more KG domains increases the zero-shot ability of our model to predict on new relation types over new entities on unseen KG domains.
Authors: Zhou Yu, Xuecheng Ouyang, Zhenwei Shao, Meng Wang, Jun Yu
Knowledge-based visual question answering (VQA) requires external knowledge beyond the image to answer the question. Early studies retrieve required knowledge from explicit knowledge bases (KBs), which often introduces irrelevant information to the question, hence restricting the performance of their models. Recent works have resorted to using a powerful large language model (LLM) as an implicit knowledge engine to acquire the necessary knowledge for answering. Despite the encouraging results achieved by these methods, we argue that they have not fully activated the capacity of the blind LLM as the provided textual input is insufficient to depict the required visual information to answer the question. In this paper, we present Prophet -- a conceptually simple, flexible, and general framework designed to prompt LLM with answer heuristics for knowledge-based VQA. Specifically, we first train a vanilla VQA model on a specific knowledge-based VQA dataset without external knowledge. After that, we extract two types of complementary answer heuristics from the VQA model: answer candidates and answer-aware examples. Finally, the two types of answer heuristics are jointly encoded into a formatted prompt to facilitate the LLM's understanding of both the image and question, thus generating a more accurate answer. By incorporating the state-of-the-art LLM GPT-3, Prophet significantly outperforms existing state-of-the-art methods on four challenging knowledge-based VQA datasets. To demonstrate the generality of our approach, we instantiate Prophet with the combinations of different VQA models (i.e., both discriminative and generative ones) and different LLMs (i.e., both commercial and open-source ones).
Authors: Marc Marone, Benjamin Van Durme
Foundation models are trained on increasingly immense and opaque datasets. Even while these models are now key in AI system building, it can be difficult to answer the straightforward question: has the model already encountered a given example during training? We therefore propose a widespread adoption of Data Portraits: artifacts that record training data and allow for downstream inspection. First we outline the properties of such an artifact and discuss how existing solutions can be used to increase transparency. We then propose and implement a solution based on data sketching, stressing fast and space efficient querying. Using our tools, we document a popular language modeling corpus (The Pile) and a recently released code modeling dataset (The Stack). We show that our solution enables answering questions about test set leakage and model plagiarism. Our tool is lightweight and fast, costing only 3% of the dataset size in overhead. We release a live interface of our tools at https://dataportraits.org/ and call on dataset and model creators to release Data Portraits as a complement to current documentation practices.
Authors: Riccardo Rende, Federica Gerace, Alessandro Laio, Sebastian Goldt
Transformers are neural networks which revolutionised natural language processing and machine learning. They process sequences of inputs, like words, using a mechanism called self-attention, which is trained via masked language modelling (MLM). In MLM, a word is randomly masked in an input sequence, and the network is trained to predict the missing word. Despite the practical success of transformers, it remains unclear what type of data distribution self-attention can learn efficiently. Here, we show analytically that if one decouples the treatment of word positions and embeddings, a single layer of self-attention learns the conditionals of a generalised Potts model with interactions between sites and Potts colours. Moreover, we show that training this neural network is exactly equivalent to solving the inverse Potts problem by the so-called pseudo-likelihood method, well known in statistical physics. Using this mapping, we compute the generalisation error of self-attention in a model scenario analytically using the replica method.
Authors: Ning Lu, Shengcai Liu, Rui He, Qi Wang, Yew-Soon Ong, Ke Tang
Large language models (LLMs) have shown remarkable performance in various tasks and have been extensively utilized by the public. However, the increasing concerns regarding the misuse of LLMs, such as plagiarism and spamming, have led to the development of multiple detectors, including fine-tuned classifiers and statistical methods. In this study, we equip LLMs with prompts, rather than relying on an external paraphraser, to evaluate the vulnerability of these detectors. We propose a novel Substitution-based In-Context example Optimization method (SICO) to automatically construct prompts for evading the detectors. SICO is cost-efficient as it requires only 40 human-written examples and a limited number of LLM inferences to generate a prompt. Moreover, once a task-specific prompt has been constructed, it can be universally used against a wide range of detectors. Extensive experiments across three real-world tasks demonstrate that SICO significantly outperforms the paraphraser baselines and enables GPT-3.5 to successfully evade six detectors, decreasing their AUC by 0.5 on average. Furthermore, a comprehensive human evaluation as well as a validation experiment in the wild show that the SICO-generated text achieves human-level readability and task completion rates. Finally, the strong performance of SICO exhibits its potential as a reliable evaluation tool for future detectors. The codes and data are located on https://github.com/ColinLu50/Evade-GPT-Detector.
Authors: Hyundong Cho, Andrea Madotto, Zhaojiang Lin, Khyathi Raghavi Chandu, Satwik Kottur, Jing Xu, Jonathan May, Chinnadhurai Sankar
Dialogue systems are frequently updated to accommodate new services, but naively updating them by continually training with data for new services in diminishing performance on previously learnt services. Motivated by the insight that dialogue state tracking (DST), a crucial component of dialogue systems that estimates the user's goal as a conversation proceeds, is a simple natural language understanding task, we propose reformulating it as a bundle of granular example-guided question answering tasks to minimize the task shift between services and thus benefit continual learning. Our approach alleviates service-specific memorization and teaches a model to contextualize the given question and example to extract the necessary information from the conversation. We find that a model with just 60M parameters can achieve a significant boost by learning to learn from in-context examples retrieved by a retriever trained to identify turns with similar dialogue state changes. Combining our method with dialogue-level memory replay, our approach attains state of the art performance on DST continual learning metrics without relying on any complex regularization or parameter expansion methods.
Authors: Jiazheng Li, Runcong Zhao, Yongxin Yang, Yulan He, Lin Gui
The remarkable performance of pre-trained large language models has revolutionised various natural language processing applications. Due to huge parametersizes and extensive running costs, companies or organisations tend to transfer the models to the target task by zero-shot prompting techniques. However, the prohibitive costs of tokens and time have hindered their adoption in applications. We propose OverPrompt, leveraging the in-context learning capability of LLMs to handle multiple task inputs, thereby reducing token and time costs. This approach could potentially improve task performance during API queries due to better conditional distribution mapping. Evaluated across diverse classification datasets, our experiments show that OverPrompt can achieve cost-efficient zero-shot classification without causing significant detriment to task performance, and in some cases, even improving it. An ablation study conducted on various LLMs, along with an investigation into the robustness of our prompting strategy to different input ordering, offers valuable insights into the broader applicability of our method across diverse tasks. These findings also suggest a more seamless integration of our method with LLMs through an API.
Authors: Victor Gallego
In this work, we address the problem of directing the text generation of a language model (LM) towards a desired behavior, aligning the generated text with the preferences of the human operator. We propose using another, instruction-tuned language model as a critic reward model in a zero-shot way thanks to the prompt of a Yes-No question that represents the user preferences, without requiring further labeled data. This zero-shot reward model provides the learning signal to further fine-tune the base LM using Reinforcement Learning from AI Feedback (RLAIF); yet our approach is also compatible in other contexts such as quality-diversity search. Extensive evidence of the capabilities of the proposed ZYN framework is provided through experiments in different domains related to text generation, including detoxification; optimizing sentiment of movie reviews, or any other attribute; steering the opinion about a particular topic the model may have; and personalizing prompt generators for text-to-image tasks. Code available at \url{https://github.com/vicgalle/zero-shot-reward-models/}.
Authors: Zilin Yuan, Borun Chen, Yimeng Dai, Yinghui Li, Hai-Tao Zheng, Rui Zhang
Citation field learning is to segment a citation string into fields of interest such as author, title, and venue. Extracting such fields from citations is crucial for citation indexing, researcher profile analysis, etc. User-generated resources like academic homepages and Curriculum Vitae, provide rich citation field information. However, extracting fields from these resources is challenging due to inconsistent citation styles, incomplete sentence syntax, and insufficient training data. To address these challenges, we propose a novel algorithm, CIFAL (citation field learning by anchor learning), to boost the citation field learning performance. CIFAL leverages the anchor learning, which is model-agnostic for any Pre-trained Language Model, to help capture citation patterns from the data of different citation styles. The experiments demonstrate that CIFAL outperforms state-of-the-art methods in citation field learning, achieving a 2.68% improvement in field-level F1-scores. Extensive analysis of the results further confirms the effectiveness of CIFAL quantitatively and qualitatively.
Authors: Arpita Vats, Zhe Liu, Peng Su, Debjyoti Paul, Yingyi Ma, Yutong Pang, Zeeshan Ahmed, Ozlem Kalinli
Model adaptation is crucial to handle the discrepancy between proxy training data and actual users data received. To effectively perform adaptation, textual data of users is typically stored on servers or their local devices, where downstream natural language processing (NLP) models can be directly trained using such in-domain data. However, this might raise privacy and security concerns due to the extra risks of exposing user information to adversaries. Replacing identifying information in textual data with a generic marker has been recently explored. In this work, we leverage large language models (LLMs) to suggest substitutes of masked tokens and have their effectiveness evaluated on downstream language modeling tasks. Specifically, we propose multiple pre-trained and fine-tuned LLM-based approaches and perform empirical studies on various datasets for the comparison of these methods. Experimental results show that models trained on the obfuscation corpora are able to achieve comparable performance with the ones trained on the original data without privacy-preserving token masking.
Authors: Wes Gurnee, Max Tegmark
The capabilities of large language models (LLMs) have sparked debate over whether such systems just learn an enormous collection of superficial statistics or a coherent model of the data generation process -- a world model. We find preliminary evidence for the latter by analyzing the learned representations of three spatial datasets (world, US, NYC places) and three temporal datasets (historical figures, artworks, news headlines) in the Llama-2 family of models. We discover that LLMs learn linear representations of space and time across multiple scales. These representations are robust to prompting variations and unified across different entity types (e.g. cities and landmarks). In addition, we identify individual ``space neurons'' and ``time neurons'' that reliably encode spatial and temporal coordinates. While further investigation is needed, our results suggest modern LLMs learn rich spatiotemporal representations of the real world and possess basic ingredients of a world model.
Authors: Haoyu Zhang, Yu Wang, Guanghao Yin, Kejun Liu, Yuanyuan Liu, Tianshu Yu
Though Multimodal Sentiment Analysis (MSA) proves effective by utilizing rich information from multiple sources (e.g., language, video, and audio), the potential sentiment-irrelevant and conflicting information across modalities may hinder the performance from being further improved. To alleviate this, we present Adaptive Language-guided Multimodal Transformer (ALMT), which incorporates an Adaptive Hyper-modality Learning (AHL) module to learn an irrelevance/conflict-suppressing representation from visual and audio features under the guidance of language features at different scales. With the obtained hyper-modality representation, the model can obtain a complementary and joint representation through multimodal fusion for effective MSA. In practice, ALMT achieves state-of-the-art performance on several popular datasets (e.g., MOSI, MOSEI and CH-SIMS) and an abundance of ablation demonstrates the validity and necessity of our irrelevance/conflict suppression mechanism.
Authors: Yunfan Shao, Linyang Li, Junqi Dai, Xipeng Qiu
Large language models (LLMs) can be used to serve as agents to simulate human behaviors, given the powerful ability to understand human instructions and provide high-quality generated texts. Such ability stimulates us to wonder whether LLMs can simulate a person in a higher form than simple human behaviors. Therefore, we aim to train an agent with the profile, experience, and emotional states of a specific person instead of using limited prompts to instruct ChatGPT API. In this work, we introduce Character-LLM that teach LLMs to act as specific people such as Beethoven, Queen Cleopatra, Julius Caesar, etc. Our method focuses on editing profiles as experiences of a certain character and training models to be personal simulacra with these experiences. To assess the effectiveness of our approach, we build a test playground that interviews trained agents and evaluates whether the agents \textit{memorize} their characters and experiences. Experimental results show interesting observations that help build future simulacra of humankind.
Authors: Bo Ni, David L. Kaplan, Markus J. Buehler
Through evolution, nature has presented a set of remarkable protein materials, including elastins, silks, keratins and collagens with superior mechanical performances that play crucial roles in mechanobiology. However, going beyond natural designs to discover proteins that meet specified mechanical properties remains challenging. Here we report a generative model that predicts protein designs to meet complex nonlinear mechanical property-design objectives. Our model leverages deep knowledge on protein sequences from a pre-trained protein language model and maps mechanical unfolding responses to create novel proteins. Via full-atom molecular simulations for direct validation, we demonstrate that the designed proteins are novel, and fulfill the targeted mechanical properties, including unfolding energy and mechanical strength, as well as the detailed unfolding force-separation curves. Our model offers rapid pathways to explore the enormous mechanobiological protein sequence space unconstrained by biological synthesis, using mechanical features as target to enable the discovery of protein materials with superior mechanical properties.
Authors: Qi Jia, Siyu Ren, Yizhu Liu, Kenny Q. Zhu
Despite tremendous improvements in natural language generation, summarization models still suffer from the unfaithfulness issue. Previous work evaluates faithfulness either using models trained on the other tasks or in-domain synthetic data, or prompting a large model such as ChatGPT. This paper proposes to do zero-shot faithfulness evaluation simply with a moderately-sized foundation language model. We introduce a new metric FFLM, which is a combination of probability changes based on the intuition that prefixing a piece of text that is consistent with the output will increase the probability of predicting the output. Experiments show that FFLM performs competitively with or even outperforms ChatGPT on both inconsistency detection and faithfulness rating with 24x fewer parameters. FFLM also achieves improvements over other strong baselines.
Authors: Sicheng Zhu, Ruiyi Zhang, Bang An, Gang Wu, Joe Barrow, Zichao Wang, Furong Huang, Ani Nenkova, Tong Sun
Safety alignment of Large Language Models (LLMs) can be compromised with manual jailbreak attacks and (automatic) adversarial attacks. Recent studies suggest that defending against these attacks is possible: adversarial attacks generate unlimited but unreadable gibberish prompts, detectable by perplexity-based filters; manual jailbreak attacks craft readable prompts, but their limited number due to the necessity of human creativity allows for easy blocking. In this paper, we show that these solutions may be too optimistic. We introduce AutoDAN, an interpretable, gradient-based adversarial attack that merges the strengths of both attack types. Guided by the dual goals of jailbreak and readability, AutoDAN optimizes and generates tokens one by one from left to right, resulting in readable prompts that bypass perplexity filters while maintaining high attack success rates. Notably, these prompts, generated from scratch using gradients, are interpretable and diverse, with emerging strategies commonly seen in manual jailbreak attacks. They also generalize to unforeseen harmful behaviors and transfer to black-box LLMs better than their unreadable counterparts when using limited training data or a single proxy model. Furthermore, we show the versatility of AutoDAN by automatically leaking system prompts using a customized objective. Our work offers a new way to red-team LLMs and understand jailbreak mechanisms via interpretability.
Authors: Xiang Huang, Sitao Cheng, Yuheng Bao, Shanshan Huang, Yuzhong Qu
While question answering over knowledge bases (KBQA) has shown progress in addressing factoid questions, KBQA with numerical reasoning remains relatively unexplored. In this paper, we focus on the complex numerical reasoning in KBQA and propose a new task, NR-KBQA, which necessitates the ability to perform both multi-hop reasoning and numerical reasoning. We design a logic form in Python format called PyQL to represent the reasoning process of numerical reasoning questions. To facilitate the development of NR-KBQA, we present a large dataset called MarkQA, which is automatically constructed from a small set of seeds. Each question in MarkQA is equipped with its corresponding SPARQL query, alongside the step-by-step reasoning process in the QDMR format and PyQL program. Experimental results of some state-of-the-art QA methods on the MarkQA show that complex numerical reasoning in KBQA faces great challenges.
Authors: Nitish Joshi, Javier Rando, Abulhair Saparov, Najoung Kim, He He
Large Language Models (LLMs) are trained on vast amounts of text from the internet, which contains both factual and misleading information about the world. Can language models discern truth from falsehood in this contradicting data? Expanding on the view that LLMs can model different communicative agents, we present the persona hypothesis: LLMs can cluster agents into personas using common features of their generations. For instance, a truthful persona is a group of agents that are likely to produce truthful text and that share similar features like formal writing styles and scientific references. By modeling this persona, LLMs can generalize truthfulness beyond the specific contexts in which each agent generated the training text. For example, the model can infer that the agent "Wikipedia" will behave truthfully on topics that were only generated by "Science" because they both belong to the truthful persona. We show evidence for the persona hypothesis via two observations: (1) we can probe whether a model's answer will be truthful before it is generated; (2) finetuning a model on a set of facts improves its truthfulness on unseen topics. Next, using arithmetics as a synthetic environment, we show that language models can separate true and false statements, and generalize truthfulness across agents; but only if agents in the training data share a truthful generative process that enables the creation of a truthful persona. Overall, our findings suggest that models can exploit hierarchical structures in the data to learn abstract concepts like truthfulness.
Authors: Dongfang Li, Zetian Sun, Xinshuo Hu, Zhenyu Liu, Ziyang Chen, Baotian Hu, Aiguo Wu, Min Zhang
Open-domain generative systems have gained significant attention in the field of conversational AI (e.g., generative search engines). This paper presents a comprehensive review of the attribution mechanisms employed by these systems, particularly large language models. Though attribution or citation improve the factuality and verifiability, issues like ambiguous knowledge reservoirs, inherent biases, and the drawbacks of excessive attribution can hinder the effectiveness of these systems. The aim of this survey is to provide valuable insights for researchers, aiding in the refinement of attribution methodologies to enhance the reliability and veracity of responses generated by open-domain generative systems. We believe that this field is still in its early stages; hence, we maintain a repository to keep track of ongoing studies at https://github.com/HITsz-TMG/awesome-llm-attributions.
Authors: Yoon Kyung Lee, Inju Lee, Minjung Shin, Seoyeon Bae, Sowon Hahn
We present a novel method, the Chain of Empathy (CoE) prompting, that utilizes insights from psychotherapy to induce Large Language Models (LLMs) to reason about human emotional states. This method is inspired by various psychotherapy approaches including Cognitive Behavioral Therapy (CBT), Dialectical Behavior Therapy (DBT), Person Centered Therapy (PCT), and Reality Therapy (RT), each leading to different patterns of interpreting clients' mental states. LLMs without reasoning generated predominantly exploratory responses. However, when LLMs used CoE reasoning, we found a more comprehensive range of empathetic responses aligned with the different reasoning patterns of each psychotherapy model. The CBT based CoE resulted in the most balanced generation of empathetic responses. The findings underscore the importance of understanding the emotional context and how it affects human and AI communication. Our research contributes to understanding how psychotherapeutic models can be incorporated into LLMs, facilitating the development of context-specific, safer, and empathetic AI.
Authors: Alessandra Grossi, Francesca Gasparini
In this paper, SER_AMPEL, a multi-source dataset for speech emotion recognition (SER) is presented. The peculiarity of the dataset is that it is collected with the aim of providing a reference for speech emotion recognition in case of Italian older adults. The dataset is collected following different protocols, in particular considering acted conversations, extracted from movies and TV series, and recording natural conversations where the emotions are elicited by proper questions. The evidence of the need for such a dataset emerges from the analysis of the state of the art. Preliminary considerations on the critical issues of SER are reported analyzing the classification results on a subset of the proposed dataset.
Authors: Alexander Sasha Vezhnevets, John P. Agapiou, Avia Aharon, Ron Ziv, Jayd Matyas, Edgar A. Duéñez-Guzmán, William A. Cunningham, Simon Osindero, Danny Karmon, Joel Z. Leibo
Agent-based modeling has been around for decades, and applied widely across the social and natural sciences. The scope of this research method is now poised to grow dramatically as it absorbs the new affordances provided by Large Language Models (LLM)s. Generative Agent-Based Models (GABM) are not just classic Agent-Based Models (ABM)s where the agents talk to one another. Rather, GABMs are constructed using an LLM to apply common sense to situations, act "reasonably", recall common semantic knowledge, produce API calls to control digital technologies like apps, and communicate both within the simulation and to researchers viewing it from the outside. Here we present Concordia, a library to facilitate constructing and working with GABMs. Concordia makes it easy to construct language-mediated simulations of physically- or digitally-grounded environments. Concordia agents produce their behavior using a flexible component system which mediates between two fundamental operations: LLM calls and associative memory retrieval. A special agent called the Game Master (GM), which was inspired by tabletop role-playing games, is responsible for simulating the environment where the agents interact. Agents take actions by describing what they want to do in natural language. The GM then translates their actions into appropriate implementations. In a simulated physical world, the GM checks the physical plausibility of agent actions and describes their effects. In digital environments simulating technologies such as apps and services, the GM may handle API calls to integrate with external tools such as general AI assistants (e.g., Bard, ChatGPT), and digital apps (e.g., Calendar, Email, Search, etc.). Concordia was designed to support a wide array of applications both in scientific research and for evaluating performance of real digital services by simulating users and/or generating synthetic data.
Authors: Xingtong Yu, Chang Zhou, Yuan Fang, Xinming Zhang
Graphs can inherently model interconnected objects on the Web, thereby facilitating a series of Web applications, such as web analyzing and content recommendation. Recently, Graph Neural Networks (GNNs) have emerged as a mainstream technique for graph representation learning. However, their efficacy within an end-to-end supervised framework is significantly tied to the availabilityof task-specific labels. To mitigate labeling costs and enhance robustness in few-shot settings, pre-training on self-supervised tasks has emerged as a promising method, while prompting has been proposed to further narrow the objective gap between pretext and downstream tasks. Although there has been some initial exploration of prompt-based learning on graphs, they primarily leverage a single pretext task, resulting in a limited subset of general knowledge that could be learned from the pre-training data. Hence, in this paper, we propose MultiGPrompt, a novel multi-task pre-training and prompting framework to exploit multiple pretext tasks for more comprehensive pre-trained knowledge. First, in pre-training, we design a set of pretext tokens to synergize multiple pretext tasks. Second, we propose a dual-prompt mechanism consisting of composed and open prompts to leverage task-specific and global pre-training knowledge, to guide downstream tasks in few-shot settings. Finally, we conduct extensive experiments on six public datasets to evaluate and analyze MultiGPrompt.
Authors: Zeyi Sun, Ye Fang, Tong Wu, Pan Zhang, Yuhang Zang, Shu Kong, Yuanjun Xiong, Dahua Lin, Jiaqi Wang
Contrastive Language-Image Pre-training (CLIP) plays an essential role in extracting valuable content information from images across diverse tasks. It aligns textual and visual modalities to comprehend the entire image, including all the details, even those irrelevant to specific tasks. However, for a finer understanding and controlled editing of images, it becomes crucial to focus on specific regions of interest, which can be indicated as points, masks, or boxes by humans or perception models. To fulfill the requirements, we introduce Alpha-CLIP, an enhanced version of CLIP with an auxiliary alpha channel to suggest attentive regions and fine-tuned with constructed millions of RGBA region-text pairs. Alpha-CLIP not only preserves the visual recognition ability of CLIP but also enables precise control over the emphasis of image contents. It demonstrates effectiveness in various tasks, including but not limited to open-world recognition, multimodal large language models, and conditional 2D / 3D generation. It has a strong potential to serve as a versatile tool for image-related tasks.
Authors: Nuo Chen, Ning Wu, Shining Liang, Ming Gong, Linjun Shou, Dongmei Zhang, Jia Li
This paper presents an in-depth analysis of Large Language Models (LLMs), focusing on LLaMA, a prominent open-source foundational model in natural language processing. Instead of assessing LLaMA through its generative output, we design multiple-choice tasks to probe its intrinsic understanding in high-order tasks such as reasoning and computation. We examine the model horizontally, comparing different sizes, and vertically, assessing different layers. We unveil several key and uncommon findings based on the designed probing tasks: (1) Horizontally, enlarging model sizes almost could not automatically impart additional knowledge or computational prowess. Instead, it can enhance reasoning abilities, especially in math problem solving, and helps reduce hallucinations, but only beyond certain size thresholds; (2) In vertical analysis, the lower layers of LLaMA lack substantial arithmetic and factual knowledge, showcasing logical thinking, multilingual and recognitive abilities, with top layers housing most computational power and real-world knowledge.
Authors: Xunjian Yin, Jin Jiang, Liming Yang, Xiaojun Wan
The imperative task of revising or updating the knowledge stored within large language models arises from two distinct sources: intrinsic errors inherent in the model which should be corrected and outdated knowledge due to external shifts in the real world which should be updated. Prevailing efforts in model editing conflate these two distinct categories of edits arising from distinct reasons and directly modify the original knowledge in models into new knowledge. However, we argue that preserving the model's original knowledge remains pertinent. Specifically, if a model's knowledge becomes outdated due to evolving worldly dynamics, it should retain recollection of the historical knowledge while integrating the newfound knowledge. In this work, we introduce the task of Temporal Knowledge Editing (TKE) and establish a benchmark AToKe (Assessment of TempOral Knowledge Editing) to evaluate current model editing methods. We find that while existing model editing methods are effective at making models remember new knowledge, the edited model catastrophically forgets historical knowledge. To address this gap, we propose a simple and general framework termed Multi-Editing with Time Objective (METO) for enhancing existing editing models, which edits both historical and new knowledge concurrently and optimizes the model's prediction for the time of each fact. Our assessments demonstrate that while AToKe is still difficult, METO maintains the effectiveness of learning new knowledge and meanwhile substantially improves the performance of edited models on utilizing historical knowledge.
Authors: Manish Nagireddy, Lamogha Chiazor, Moninder Singh, Ioana Baldini
Current datasets for unwanted social bias auditing are limited to studying protected demographic features such as race and gender. In this work, we introduce a comprehensive benchmark that is meant to capture the amplification of social bias, via stigmas, in generative language models. We start with a comprehensive list of 93 stigmas documented in social science literature and curate a question-answering (QA) dataset which involves simple social situations. Our benchmark, SocialStigmaQA, contains roughly 10K prompts, with a variety of prompt styles, carefully constructed to systematically test for both social bias and model robustness. We present results for SocialStigmaQA with two widely used open source generative language models and we demonstrate that the output generated by these models considerably amplifies existing social bias against stigmatized groups. Specifically, we find that the proportion of socially biased output ranges from 45% to 59% across a variety of decoding strategies and prompting styles. We discover that the deliberate design of the templates in our benchmark (e.g., by adding biasing text to the prompt or varying the answer that indicates bias) impact the model tendencies to generate socially biased output. Additionally, we report on patterns in the generated chain-of-thought output, finding a variety of problems from subtle bias to evidence of a lack of reasoning.
Warning: This paper contains examples of text which is toxic, biased, and harmful.
Authors: Wentao Liu, Hanglei Hu, Jie Zhou, Yuyang Ding, Junsong Li, Jiayi Zeng, Mengliang He, Qin Chen, Bo Jiang, Aimin Zhou, Liang He
In recent years, there has been remarkable progress in leveraging Language Models (LMs), encompassing Pre-trained Language Models (PLMs) and Large-scale Language Models (LLMs), within the domain of mathematics. This paper conducts a comprehensive survey of mathematical LMs, systematically categorizing pivotal research endeavors from two distinct perspectives: tasks and methodologies. The landscape reveals a large number of proposed mathematical LLMs, which are further delineated into instruction learning, tool-based methods, fundamental CoT techniques, and advanced CoT methodologies. In addition, our survey entails the compilation of over 60 mathematical datasets, including training datasets, benchmark datasets, and augmented datasets. Addressing the primary challenges and delineating future trajectories within the field of mathematical LMs, this survey is positioned as a valuable resource, poised to facilitate and inspire future innovation among researchers invested in advancing this domain.
Authors: Róbert Csordás, Piotr Piękos, Kazuki Irie, Jürgen Schmidhuber
The costly self-attention layers in modern Transformers require memory and compute quadratic in sequence length. Existing approximation methods usually underperform and fail to obtain significant speedups in practice. Here we present SwitchHead - a novel method that reduces both compute and memory requirements and achieves wall-clock speedup, while matching the language modeling performance of baseline Transformers with the same parameter budget. SwitchHead uses Mixture-of-Experts (MoE) layers for the value and output projections and requires 4 to 8 times fewer attention matrices than standard Transformers. Our novel attention can also be combined with MoE MLP layers, resulting in an efficient fully-MoE "SwitchAll" Transformer model. Our code is public.
Authors: Wenting Chen, Xiang Li, Linlin Shen, Yixuan Yuan
To address these issues, we propose a novel Adaptive patch-word Matching (AdaMatch) model to correlate chest X-ray (CXR) image regions with words in medical reports and apply it to CXR-report generation to provide explainability for the generation process. AdaMatch exploits the fine-grained relation between adaptive patches and words to provide explanations of specific image regions with corresponding words. To capture the abnormal regions of varying sizes and positions, we introduce the Adaptive Patch extraction (AdaPatch) module to acquire the adaptive patches for these regions adaptively. In order to provide explicit explainability for CXR-report generation task, we propose an AdaMatch-based bidirectional large language model for Cyclic CXR-report generation (AdaMatch-Cyclic). It employs the AdaMatch to obtain the keywords for CXR images and `keypatches' for medical reports as hints to guide CXR-report generation. Extensive experiments on two publicly available CXR datasets prove the effectiveness of our method and its superior performance to existing methods.
Authors: Songchi Zhou, Sheng Yu
Objective: To develop a high-throughput biomedical relation extraction system that takes advantage of the large language models' (LLMs) reading comprehension ability and biomedical world knowledge in a scalable and evidential manner. Methods: We formulate the relation extraction task as a simple binary classification problem for large language models such as ChatGPT. Specifically, LLMs make the decision based on the external corpus and its world knowledge, giving the reason for the judgment to factual verification. This method is tailored for semi-structured web articles, wherein we designate the main title as the tail entity and explicitly incorporate it into the context, and the potential head entities are matched based on a biomedical thesaurus. Moreover, lengthy contents are sliced into text chunks, embedded, and retrieved with additional embedding models, ensuring compatibility with the context window size constraints of available open-source LLMs. Results: Using an open-source LLM, we extracted 304315 relation triplets of three distinct relation types from four reputable biomedical websites. To assess the efficacy of the basic pipeline employed for biomedical relation extraction, we curated a benchmark dataset annotated by a medical expert. Evaluation results indicate that the pipeline exhibits performance comparable to that of GPT-4. Case studies further illuminate challenges faced by contemporary LLMs in the context of biomedical relation extraction for semi-structured web articles. Conclusion: The proposed method has demonstrated its effectiveness in leveraging the strengths of LLMs for high-throughput biomedical relation extraction. Its adaptability is evident, as it can be seamlessly extended to diverse semi-structured biomedical websites, facilitating the extraction of various types of biomedical relations with ease.