Stance detection automatically detects the stance in a text towards a target, vital for content analysis in web and social media research. Despite their promising capabilities, LLMs encounter challenges when directly applied to stance detection. First, stance detection demands multi-aspect knowledge, from deciphering event-related terminologies to understanding the expression styles in social media platforms. Second, stance detection requires advanced reasoning to infer authors' implicit viewpoints, as stance are often subtly embedded rather than overtly stated in the text. To address these challenges, we design a three-stage framework COLA (short for Collaborative rOle-infused LLM-based Agents) in which LLMs are designated distinct roles, creating a collaborative system where each role contributes uniquely. Initially, in the multidimensional text analysis stage, we configure the LLMs to act as a linguistic expert, a domain specialist, and a social media veteran to get a multifaceted analysis of texts, thus overcoming the first challenge. Next, in the reasoning-enhanced debating stage, for each potential stance, we designate a specific LLM-based agent to advocate for it, guiding the LLM to detect logical connections between text features and stance, tackling the second challenge. Finally, in the stance conclusion stage, a final decision maker agent consolidates prior insights to determine the stance. Our approach avoids extra annotated data and model training and is highly usable. We achieve state-of-the-art performance across multiple datasets. Ablation studies validate the effectiveness of each design role in handling stance detection. Further experiments have demonstrated the explainability and the versatility of our approach. Our approach excels in usability, accuracy, effectiveness, explainability and versatility, highlighting its value.

The advent of the Web has brought about a paradigm shift in traditional economics, particularly in the digital economy era, enabling the precise recording and analysis of individual economic behavior. This has led to a growing emphasis on data-driven modeling in macroeconomics. In macroeconomic research, Agent-based modeling (ABM) emerged as an alternative, evolving through rule-based agents, machine learning-enhanced decision-making, and, more recently, advanced AI agents. However, the existing works are suffering from three main challenges when endowing agents with human-like decision-making, including agent heterogeneity, the influence of macroeconomic trends, and multifaceted economic factors. Large language models (LLMs) have recently gained prominence in offering autonomous human-like characteristics. Therefore, leveraging LLMs in macroeconomic simulation presents an opportunity to overcome traditional limitations. In this work, we take an early step in introducing a novel approach that leverages LLMs in macroeconomic simulation. We design prompt-engineering-driven LLM agents to exhibit human-like decision-making and adaptability in the economic environment, with the abilities of perception, reflection, and decision-making to address the abovementioned challenges. Simulation experiments on macroeconomic activities show that LLM-empowered agents can make realistic work and consumption decisions and emerge more reasonable macroeconomic phenomena than existing rule-based or AI agents. Our work demonstrates the promising potential to simulate macroeconomics based on LLM and its human-like characteristics.

Large language models (LLMs) are increasingly applied for tabular tasks using in-context learning. The prompt representation for a table may play a role in the LLMs ability to process the table. Inspired by prior work, we generate a collection of self-supervised structural tasks (e.g. navigate to a cell and row; transpose the table) and evaluate the performance differences when using 8 formats. In contrast to past work, we introduce 8 noise operations inspired by real-world messy data and adversarial inputs, and show that such operations can impact LLM performance across formats for different structural understanding tasks.

Recent studies have demonstrated that large language models (LLMs) store massive factual knowledge within their parameters. But existing LLMs are prone to hallucinate unintended text due to false or outdated knowledge. Since retraining LLMs is resource intensive, there has been a growing interest in the concept of model editing. Despite the emergence of benchmarks and approaches, these unidirectional editing and evaluation have failed to explore the reversal curse. Intuitively, if "The capital of France is" is edited to be a counterfact "London" within a model, then it should be able to naturally reason and recall the reverse fact, i.e., "London is the capital of" followed by "France" instead of "England". In this paper, we study bidirectional language model editing, aiming to provide rigorous model editing evaluation to assess if edited LLMs can recall the editing knowledge bidirectionally. A new evaluation metric of reversibility is introduced, and a benchmark dubbed as Bidirectional Assessment for Knowledge Editing (BAKE) is constructed to evaluate the reversibility of edited models in recalling knowledge in the reverse direction of editing. We surprisingly observe that while current editing methods and LLMs can effectively recall editing facts in the direction of editing, they suffer serious deficiencies when evaluated in the reverse direction. To mitigate the reversal curse, a method named Bidirectionally Inversible Relationship moDeling (BIRD) is proposed. A set of editing objectives that incorporate bidirectional relationships between subject and object into the updated model weights are designed. Experiments show that BIRD improves the performance of four representative LLMs of different sizes via question answering and judgement.

Transformer-based models, even though achieving super-human performance on several downstream tasks, are often regarded as a black box and used as a whole. It is still unclear what mechanisms they have learned, especially their core module: multi-head attention. Inspired by functional specialization in the human brain, which helps to efficiently handle multiple tasks, this work attempts to figure out whether the multi-head attention module will evolve similar function separation under multi-tasking training. If it is, can this mechanism further improve the model performance? To investigate these questions, we introduce an interpreting method to quantify the degree of functional specialization in multi-head attention. We further propose a simple multi-task training method to increase functional specialization and mitigate negative information transfer in multi-task learning. Experimental results on seven pre-trained transformer models have demonstrated that multi-head attention does evolve functional specialization phenomenon after multi-task training which is affected by the similarity of tasks. Moreover, the multi-task training strategy based on functional specialization boosts performance in both multi-task learning and transfer learning without adding any parameters.

Dialogue summarization involves a wide range of scenarios and domains. However, existing methods generally only apply to specific scenarios or domains. In this study, we propose a new pre-trained model specifically designed for multi-scenario multi-domain dialogue summarization. It adopts a multi-stage pre-training strategy to reduce the gap between the pre-training objective and fine-tuning objective. Specifically, we first conduct domain-aware pre-training using large-scale multi-scenario multi-domain dialogue data to enhance the adaptability of our pre-trained model. Then, we conduct task-oriented pre-training using large-scale multi-scenario multi-domain "dialogue-summary" parallel data annotated by ChatGPT to enhance the dialogue summarization ability of our pre-trained model. Experimental results on three dialogue summarization datasets from different scenarios and domains indicate that our pre-trained model significantly outperforms previous state-of-the-art models in full fine-tuning, zero-shot, and few-shot settings.

As one of the most exciting features of large language models (LLMs), in-context learning is a mixed blessing. While it allows users to fast-prototype a task solver with only a few training examples, the performance is generally sensitive to various configurations of the prompt such as the choice or order of the training examples. In this paper, we for the first time theoretically and empirically identify that such a paradox is mainly due to the label shift of the in-context model to the data distribution, in which LLMs shift the label marginal $p(y)$ while having a good label conditional $p(x|y)$. With this understanding, we can simply calibrate the in-context predictive distribution by adjusting the label marginal, which is estimated via Monte-Carlo sampling over the in-context model, i.e., generation of LLMs. We call our approach as generative calibration. We conduct exhaustive experiments with 12 text classification tasks and 12 LLMs scaling from 774M to 33B, generally find that the proposed method greatly and consistently outperforms the ICL as well as state-of-the-art calibration methods, by up to 27% absolute in macro-F1. Meanwhile, the proposed method is also stable under different prompt configurations.

Interpretability methods in NLP aim to provide insights into the semantics underlying specific system architectures. Focusing on word embeddings, we present a supervised-learning method that, for a given domain (e.g., sports, professions), identifies a subset of model features that strongly improve prediction of human similarity judgments. We show this method keeps only 20-40% of the original embeddings, for 8 independent semantic domains, and that it retains different feature sets across domains. We then present two approaches for interpreting the semantics of the retained features. The first obtains the scores of the domain words (co-hyponyms) on the first principal component of the retained embeddings, and extracts terms whose co-occurrence with the co-hyponyms tracks these scores' profile. This analysis reveals that humans differentiate e.g. sports based on how gender-inclusive and international they are. The second approach uses the retained sets as variables in a probing task that predicts values along 65 semantically annotated dimensions for a dataset of 535 words. The features retained for professions are best at predicting cognitive, emotional and social dimensions, whereas features retained for fruits or vegetables best predict the gustation (taste) dimension. We discuss implications for alignment between AI systems and human knowledge.

In the intricate field of legal studies, the analysis of court decisions is a cornerstone for the effective functioning of the judicial system. The ability to predict court outcomes helps judges during the decision-making process and equips lawyers with invaluable insights, enhancing their strategic approaches to cases. Despite its significance, the domain of Arabic court analysis remains under-explored. This paper pioneers a comprehensive predictive analysis of Arabic court decisions on a dataset of 10,813 commercial court real cases, leveraging the advanced capabilities of the current state-of-the-art large language models. Through a systematic exploration, we evaluate three prevalent foundational models (LLaMA-7b, JAIS-13b, and GPT3.5-turbo) and three training paradigms: zero-shot, one-shot, and tailored fine-tuning. Besides, we assess the benefit of summarizing and/or translating the original Arabic input texts. This leads to a spectrum of 14 model variants, for which we offer a granular performance assessment with a series of different metrics (human assessment, GPT evaluation, ROUGE, and BLEU scores). We show that all variants of LLaMA models yield limited performance, whereas GPT-3.5-based models outperform all other models by a wide margin, surpassing the average score of the dedicated Arabic-centric JAIS model by 50%. Furthermore, we show that all scores except human evaluation are inconsistent and unreliable for assessing the performance of large language models on court decision predictions. This study paves the way for future research, bridging the gap between computational linguistics and Arabic legal analytics.

There are a number of diverging hypotheses about the neural text degeneration problem, i.e., generating repetitive and dull loops, which makes this problem both interesting and confusing. In this work, we aim to advance our understanding by presenting a straightforward and fundamental explanation from the data perspective. Our preliminary investigation reveals a strong correlation between the degeneration issue and the presence of repetitions in training data. Subsequent experiments also demonstrate that by selectively dropping out the attention to repetitive words in training data, degeneration can be significantly minimized. Furthermore, our empirical analysis illustrates that prior works addressing the degeneration issue from various standpoints, such as the high-inflow words, the likelihood objective, and the self-reinforcement phenomenon, can be interpreted by one simple explanation. That is, penalizing the repetitions in training data is a common and fundamental factor for their effectiveness. Moreover, our experiments reveal that penalizing the repetitions in training data remains critical even when considering larger model sizes and instruction tuning.