CS@CU At ACL 2024

CS researchers presented their work at the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024).

Parallel Structures in Pre-training Data Yield In-Context Learning
Yanda Chen Columbia University, Chen Zhao New York University, Zhou Yu Columbia University, Kathleen McKeown Columbia University, He He New York University

Abstract:
Pre-trained language models (LMs) are capable of in-context learning (ICL): they can adapt to a task with only a few examples given in the prompt without any parameter update. However, it is unclear where this capability comes from as there is a stark distribution shift between pre-training text and ICL prompts. In this work, we study what patterns of the pretraining data contribute to ICL. We find that LMs’ ICL ability depends on parallel structures in the pre-training data—pairs of phrases following similar templates in the same context window. Specifically, we detect parallel structures by checking whether training on one phrase improves prediction of the other, and conduct ablation experiments to study their effect on ICL. We show that removing parallel structures in the pre-training data reduces LMs’ ICL accuracy by 51% (vs 2% from random ablation). This drop persists even when excluding common patterns such as n-gram repetitions and long-range dependency, showing the diversity and generality of parallel structures. A closer look at the detected parallel structures indicates that they cover diverse linguistic tasks and span long distances in the data.

Getting Serious about Humor: Crafting Humor Datasets with Unfunny Large Language Models
Zachary Horvitz Columbia University, Jingru Chen Columbia University, Rahul Aditya Columbia University, Harshvardhan Srivastava Columbia University, Robert West EPFL, Zhou Yu Columbia University, Kathleen McKeown Columbia University

Abstract:
Humor is a fundamental facet of human cognition and interaction. Yet, despite recent advances in natural language processing, humor detection remains a challenging task that is complicated by the scarcity of datasets that pair humorous texts with similar non-humorous counterparts. We investigate whether large language models (LLMs) can generate synthetic data for humor detection via editing texts. We benchmark LLMs on an existing human dataset and show that current LLMs display an impressive ability to “unfun” jokes, as judged by humans and as measured on the downstream task of humor detection. We extend our approach to a code-mixed English-Hindi humor dataset where we find that GPT-4’s synthetic data is highly rated by bilingual annotators and provides challenging adversarial examples for humor classifiers.

ICLEF: In-Context Learning with Expert Feedback for Explainable Style Transfer
Arkadiy Saakyan Columbia University and Smaranda Muresan Columbia University

Abstract:
While state-of-the-art large language models (LLMs) can excel at adapting text from one style to another, current work does not address the explainability of style transfer models. Recent work has explored generating textual explanations from larger teacher models and distilling them into smaller student models. One challenge with such approach is that LLM outputs may contain errors that require expertise to correct, but gathering and incorporating expert feedback is difficult due to cost and availability. To address this challenge, we propose ICLEF, a novel human-AI collaboration approach to model distillation that incorporates scarce expert human feedback by combining in-context learning and model self-critique. We show that our method leads to generation of high-quality synthetic explainable style transfer datasets for formality (E-GYAFC) and subjective bias (EWNC). Via automatic and human evaluation, we show that specialized student models finetuned on our datasets outperform generalist teacher models on the explainable style transfer task in one-shot settings, and perform competitively compared to few-shot teacher models, highlighting the quality of the data and the role of expert feedback. In an extrinsic task of authorship attribution, we show that explanations generated by smaller models fine-tuned on E-GYAFC are more predictive of authorship than explanations generated by few-shot teacher models.

ProLex: A Benchmark for Language Proficiency-oriented Lexical Substitution
Xuanming Zhang Columbia University, Zixun Chen Columbia University, and Zhou Yu Columbia University

Abstract:
Lexical Substitution discovers appropriate substitutes for a given target word in a context sentence. However, the task fails to consider substitutes that are of equal or higher proficiency than the target, an aspect that could be beneficial for language learners looking to improve their writing. To bridge this gap, we propose a new task — language proficiencyoriented lexical substitution. We also introduce ProLex, a novel benchmark designed to assess systems’ ability to generate not only appropriate substitutes but also substitutes that demonstrate better language proficiency. Besides the benchmark, we propose models that can automatically perform the new task. We show that our best model, a Llama2-13B model fine-tuned with task-specific synthetic data, outperforms ChatGPT by an average of 3.2% in F-score and achieves comparable results with GPT-4 on ProLex.

LOCALRQA: From Generating Data to Locally Training, Testing, and Deploying Retrieval-Augmented QA Systems
Xiao Yu Columbia University, Yunan Lu Columbia University, Zhou Yu Columbia University

Abstract:
Retrieval-augmented question-answering systems combine retrieval techniques with large language models to provide answers that are more accurate and informative. Many existing toolkits allow users to quickly build such systems using off-the-shelf models, but they fall short in supporting researchers and developers to customize the model training, testing, and deployment process. We propose LOCALRQA1 , an open-source toolkit that features a wide selection of model training algorithms, evaluation methods, and deployment tools curated from the latest research. As a showcase, we build QA systems using online documentation obtained from Databricks and Faire’s websites. We find 7B-models trained and deployed using LOCALRQA reach a similar performance compared to using OpenAI’s text-ada-002 and GPT-4-turbo.

Large Language Models are Few-Shot Training Example Generators: A Case Study in Fallacy Recognition
Tariq Alhindi Mohamed bin Zayed University of Artificial Intelligence, Smaranda Muresan Columbia University, Preslav Nakov Mohamed bin Zayed University of Artificial Intelligence

Abstract:
Recognizing fallacies is crucial for ensuring the quality and validity of arguments across various domains. However, computational fallacy recognition faces challenges due to the diverse genres, domains, and types of fallacies found in datasets. This leads to a highly multiclass, and even multi-label, setup with substantial class imbalance. In this study, we aim to enhance existing models for fallacy recognition by incorporating additional context and by leveraging large language models to generate synthetic data, thus increasing the representation of the infrequent classes. We experiment with GPT-3.5 to generate synthetic examples and we examine the impact of prompt settings for this. Moreover, we explore zero-shot and few-shot scenarios to evaluate the effectiveness of using the generated examples for training smaller models within a unified fallacy recognition framework. Furthermore, we analyze the overlap between the synthetic data and existing fallacy datasets. Finally, we investigate the usefulness of providing supplementary context for detecting fallacy types that need such context, e.g., diversion fallacies. Our evaluation results demonstrate consistent improvements across fallacy types, datasets, and generators. The code and the synthetic datasets are all publicly available.